research project-01

Peer-Reviewed ArticlesTheme 1 - General Aspects

Salvador J. Diaz-Cano, LMS, MD, PhD, FRCPathResearch Project • Clonal Evolution and Topographic Tumor Heterogeneity •

Salvador J. Diaz-Cano • Peer-Reviewed Articles • General Aspects 1

Contents General Aspects

Designing a Molecular Analysis of Clonality in Tumours. J Pathol 2000;191:343-344! 4

Salvador J. Diaz-Cano 4

PCR Techniques for Clonality Assays. Diagn Mol Pathol 2001;10(1):24-33! 4

Salvador J. Diaz–Cano, M.D., Ph.D., Alfredo Blanes, M.D., Ph.D., and Hubert J. Wolfe, M.D. 4

Clonal Origin and Expansions in Neoplasms: Biologic and Technical Aspects Must Be Considered Together. Am J Pathol 2003;162(1):353-355! 4

Lucia Pozo-Garcia, Salvador J. Diaz-Cano 4

Are PCR artifacts in microdissected samples preventable? Hum Pathol 2001;32(12):1415-1416! 4

Salvador J. Diaz-Cano, MD, PhD 4

DNA extractions from formalin-fixed paraffin-embedded tissues: Protein diges-tion as a limiting step for retrieval of high-quality DNA Diagn Mol Pathol 1997;6(6):342-346! 4

Salvador J. Diaz-Cano, MD, PhD, Stephen P. Brady, MD 4

Influence of intratumour heterogeneity in the interpretation of marker results in phaeochromocytomas. J Pathol 1999;189:627-629! 4

Salvador J. Diaz-Cano, Alfredo Blanes 4

DNA Mass Distribution

2c 4c

G1

G2M

SG0

G0/G1

G2+MS

Ki-67 +

Apoptoticcells


Clonality in Kaposi’s sarcoma. N Engl J Med 1997;337:570-572! 4

Salvador J. Diaz-Cano, M.D., Ph.D., Hubert J. Wolfe, M.D. 4

Analysis of clonality of atypical cutaneous lymphoid infiltrates associated with drug therapy by PCR/DGGE.Hum Pathol. 1999 Feb;30(2):130-6.! 5

Brady SP, Magro CM, Diaz-Cano SJ, Wolfe HJ. 5

PCR-based alternative for diagnosis of immunoglobulin heavy chain gene rear-rangement: principles, practice, and polemics.Diagn Mol Pathol. 1996 Mar;5(1):3-9. Review. ! 5

Diaz-Cano S. 5


DESIGNING A MOLECULAR ANALYSIS OF CLONALITY IN TU-MOURS. J Pathol 2000;191:343-344

Salvador J. Diaz-Cano

PCR TECHNIQUES FOR CLONALITY ASSAYS. Diagn Mol Pathol 2001;10(1):24-33

Salvador J. Diaz–Cano, M.D., Ph.D., Alfredo Blanes, M.D., Ph.D., and Hubert J. Wolfe, M.D.

CLONAL ORIGIN AND EXPANSIONS IN NEOPLASMS: BIO-LOGIC AND TECHNICAL ASPECTS MUST BE CONSIDERED TOGETHER. Am J Pathol 2003;162(1):353-355

Lucia Pozo-Garcia, Salvador J. Diaz-Cano

ARE PCR ARTIFACTS IN MICRODISSECTED SAMPLES PRE-VENTABLE? Hum Pathol 2001;32(12):1415-1416

Salvador J. Diaz-Cano, MD, PhD

DNA EXTRACTIONS FROM FORMALIN-FIXED PARAFFIN-EMBEDDED TISSUES: PROTEIN DIGESTION AS A LIMITING STEP FOR RETRIEVAL OF HIGH-QUALITY DNA Diagn Mol Pathol 1997;6(6):342-346

Salvador J. Diaz-Cano, MD, PhD, Stephen P. Brady, MD

INFLUENCE OF INTRATUMOUR HETEROGENEITY IN THE IN-TERPRETATION OF MARKER RESULTS IN PHAEOCHROMOCY-TOMAS. J Pathol 1999;189:627-629

Salvador J. Diaz-Cano, Alfredo Blanes

CLONALITY IN KAPOSI’S SARCOMA. N Engl J Med 1997;337:570-572

Salvador J. Diaz-Cano, M.D., Ph.D., Hubert J. Wolfe, M.D.


ANALYSIS OF CLONALITY OF ATYPICAL CUTANEOUS LYMPHOID INFILTRATES ASSOCIATED WITH DRUG THERAPY BY PCR/DGGE.HUM PATHOL. 1999 FEB;30(2):130-6.

Brady SP, Magro CM, Diaz-Cano SJ, Wolfe HJ.

PCR-BASED ALTERNATIVE FOR DIAGNOSIS OF IMMU-NOGLOBULIN HEAVY CHAIN GENE REARRANGEMENT: PRINCIPLES, PRACTICE, AND POLEMICS.DIAGN MOL PA-THOL. 1996 MAR;5(1):3-9. REVIEW.

Diaz-Cano S.


Editorial

Designing a molecular analysis of clonality in tumours

Salvador J. Diaz-CanoDepartment of Histopathology and Morbid Anatomy, St Bartholomew's and the Royal London School of Medicine and Dentistry, Whitechapel, LondonE1 1BB, UK

Abstract

Clonality analysis is used to test malignant transformation and tumour progression. X-chromosome

linked clonality assays have been employed for this purpose, but are subject to certain technical

limitations. This paper reviews the issues involved and the controls that are necessary to ensure

valid interpretation of such analyses.

Keywords: clonality; X-chromosome; lyonization; neoplasia

Clonality is an essential attribute of neoplasms and itsanalysis has been used to test malignant transforma-tion and tumour progression [1,2]. Concordant pat-terns of genetic markers (X-linked or not) in differenttumours suggest that a common progenitor contribu-ted to those lesions and favour, therefore, a multifocalrather than a multicentric origin. These shared geneticalterations also suggest a common cellular origin forbiphasic neoplasms [3]. Saxena et al. recently reporteda monoclonal pattern in smooth muscle cells and bloodvessels of sporadic angiomyolipoma, while the adiposetissue revealed a polyclonal pattern [4]. Based on these®ndings the authors concluded that the polyclonaladipose tissue is probably metaplastic or reactive.This represents a good example of the application ofclonality in tumour biology.

However, some biological and technical issues arisefrom this article. X-linked clonality assays are based onDNA polymorphism and random X-chromosomeinactivation (XCI) in females. Those features enableus to distinguish the maternally from the paternallyinherited X-chromosomes [1,5,6]. The mechanismsleading to XCI have not been fully characterized, butDNA methylation might maintain the inactive state,once it is established during early embryogenesis.These methylation patterns are then transmitted byclonal inheritance through the strong preference ofmammalian DNA (cytosine-5)-methyltransferase forhemimethylated DNA, involving the promoter regionsof alleles on the inactive X-chromosome only [7]. SinceXCI analysis is based on differential DNA methylationof one allele from X-chromosome genes (e.g. humanandrogen receptor gene), suboptimal enzymatic diges-tion and abnormal methylation can result in changesof clonality patterns.

According to Lyon's hypothesis, all but one X-chromosomes in a cell are randomly inactivated duringearly embryogenesis, when the primordial cell poolmay comprise as few as 16±30 cells [8]. Given thatsmall number of embryo-destined cells, it reasonable to

expect unequal numbers of paternally- and maternally-inherited inactive X-chromosomes, although the X-chromosome is randomly inactivated in each cell. Theaverage Lyonization ratio is close to 50 : 50 in large cellpopulations, although individual variation has beenfound [8]. Skewing towards one allele to an extent thatmeets the criteria for clonal derivation is consistentwith early XCI during embryogenesis (Figure 1).

This ®nding leads us to consider the selection ofappropriate controls to assess the Lyonization ratio ineach female. This ratio can also vary from tissue totissue in the same individual, due to unequal splittingof the cells derived from the primordial cell pool, or todifferent methylation patterns in different tissues [6,9].Controls for unequal Lyonization should thus ideallybe the most closely related tissue thought not to beinvolved in the disease process. An essential require-

Figure 1. Methylation pattern of androgen receptor alleles incontrol samples. Only polymorphic and polyclonal controls (twoallele bands in both undigested and digested samples) areconsidered informative for clonality assays (lanes 1 and 2). Theremaining possibilities (lanes 3±8) should be excluded fromclonality analyses, due to either monoclonal origin of controls(lanes 3±6) or absence of locus polymorphism (lanes 7and 8).U=undigested sample: D=digested sample

Journal of PathologyJ Pathol 2000; 191: 343±344.

Copyright # 2000 John Wiley & Sons, Ltd.

ment for clonality analysis is the identi®cation of apolymorphic locus in the normal control (Figure 1). Inevery case, the tumour sample must be compared withmatched controls from the same patient to test theheterozygosity for the marker. Additionally, the indi-vidual variability and tissue-related Lyonization ratiorequire samples of close embryological origin. Thisfeature must be maintained in the digested sample inthose tests based on XCI (Figure 1).

Positive allelic imbalances are determined case-by-case, using the skewed data normalized by the alleleratio in matched controls [1,6]. Allelic imbalanceanalysis is based on the allele ratio and requiresdensitometric analysis of both allele bands. Therefore,the allele ratio in the target DNA must be maintainedin the ampli®cation product, which has to avoid thePCR plateau phase. At this level, any PCR ampli®ca-tion bias should be considered, especially DNAdegradation of the larger allele in formalin-®xed,paraf®n-embedded tissues and defective ampli®cationof repetitive CG-rich sequences [10±12].

Early XCI occurs randomly and results in a chess-board pattern of cells descended from a commonprogenitor, which may grow together like a clone(patch size mosaicism). This pattern represents anexample of tissue heterogeneity that can also be presentin tumours. Sample size is a limiting factor; the lowerthe cell number, the higher the probability of mono-clonal patterns based on patch size mosaicism. Thisconcept becomes particularly important in mixedtumours, where multiple microdissected samples fromdifferent tumour areas (i100 cells) and from controlsare required to address the question.

Monoclonal patterns support a neoplastic ratherthan a reactive or hyperplastic process, but are notdiagnostic of it. Host cell contamination of tumoursamples could give false heterozygous results thatwould require careful microdissection and microscopiccontrol of the sample collection. However, the pitfallsmentioned above should be always excluded.

Some of these considerations do not appear to have

been addressed in the paper of Saxena et al. [4],especially those concerning tests for digestion comple-tion with restriction endonuclease; controls regardingboth tumour heterogeneity and their methylationpatterns; PCR bias in the ampli®cation of both alleles;tumour heterogeneity and patch size mosaicism; andthe meaning of monoclonal and polyclonal patterns.

References

1. Diaz-Cano SJ, Blanes A, Wolfe HJ. PCR-based techniques for

clonality analysis of neoplastic progression. Bases for its

appropriate application in paraf®n-embedded tissues. Diagn

Mol Pathol (in press).

2. Diaz-Cano SJ. Clonality studies in the analysis of adrenal

medullary proliferations: application principles and limitations.

Endocr Pathol 1998; 9: 301±316.

3. Zhuang Z, Lininger RA, Man YG, Albuquerque A, Merino

MJ, Tavassoli FA. Identical clonality of both components

of mammary carcinosarcoma with differential loss of hetero-

zygosity. Mod Pathol 1997; 10: 354±362.

4. Saxena A, Alport EC, Custead S, Skinnider LF. Molecular

analysis of clonality of sporadic angiomyolipoma. J Pathol 1999;

189: 79±84.

5. Sleddens HF, Oostra BA, Brinkmann AO, Trapman J. Tri-

nucleotide repeat polymorphism in the androgen receptor gene

(AR). Nucleic Acids Res 1992; 20: 1427.

6. Mutter GL, Boynton KA. X chromosome inactivation in the

normal female genital tract: implications for identi®cation of

neoplasia. Cancer Res 1995; 55: 5080±5084.

7. Lyon MF. Some milestones in the history of X-chromosome

inactivation. Annu Rev Genet 1992; 26: 16±28.

8. Fialkow PJ. Primordial cell pool size and lineage relationships of

®ve human cell types. Ann Hum Genet 1973; 37: 39±48.

9. Kappler JW. The 5-methylcytosine content of DNA: tissue

speci®city. J Cell Physiol 1971; 78: 33±36.

10. Mutter GL, Boynton KA. PCR bias in ampli®cation of

androgen receptor alleles, a trinucleotide repeat marker used in

clonality studies. Nucleic Acids Res 1995; 23: 1411±1418.

11. Diaz-Cano SJ, Brady SP. DNA extraction from formalin-®xed,

paraf®n-embedded tissues: protein digestion as a limiting step

for retrieval of high-quality DNA. Diagn Mol Pathol 1997; 6:

342±346.

12. Diaz-Cano SJ, de Miguel M, Blanes A, Tashjian R, Galera H,

Wolfe HJ. Clonality as expression of distinctive cell kinetics

patterns in nodular hyperplasias and adenomas of the adrenal

cortex. Am J Pathol 2000; 156: 311±319.

344 Editorial

Copyright # 2000 John Wiley & Sons, Ltd. J Pathol 2000; 191: 343±344.

PCR Techniques for Clonality Assays

Salvador J. Diaz–Cano, M.D., Ph.D., Alfredo Blanes, M.D., Ph.D., andHubert J. Wolfe, M.D.

Clonal overgrowths represent the hallmark of neoplastic pro-liferations, and their demonstration has been proved usefulclinically for the diagnosis of malignant lymphomas based onthe detection of specific and dominant immunoglobulin and/orT-cell receptor gene rearrangements. Nonrandom genetic alter-ations can also be used to test clonal expansions and the clonalevolution of neoplasms, especially analyzing hypervariable de-oxyribonucleic acid (DNA) regions from patients heterozygousfor a given marker. These tests rely basically on the demon-stration of loss of heterozygosity (LOH) resulting from eitherhemizygosity (nonrandom interstitial DNA deletions) or homo-zygosity of mutant alleles observed in neoplasms. LOH analy-ses identify clonal expansions of a tumor cell population, andpoint to monoclonal proliferation when multiple and consistentLOH are demonstrated. Based on the methylation-related inac-tivation of one X chromosome in female subjects, X-linkedmarkers (e.g., androgen receptor gene) will provide clonalityinformation using LOH analyses after DNA digestion withmethylation-sensitive restriction endonucleases. Therefore,both non-X-linked and X-linked analyses give complementaryinformation, related and not related to the malignant transfor-mation pathway respectively. Applied appropriately, thesetools can establish the clonal evolution of tumor cell popula-tions (tumor heterogeneity), identify early relapses, distinguishrecurrent tumors from other metachronic neoplasms, and dif-ferentiate field transformation from metastatic tumor growthsin synchronic and histologically identical neoplasms.Key Words: Clonality—X chromosome inactivation—Microsatellites—Tumor suppressor genes—Tumor progres-sion—Paraffin-embedded tissues.

Diagn Mol Pathol 10(1): 24–33, 2001.

Conceptually, a clone is a group of genetically iden-tical cells descended from one common ancestor. Cur-rently, clonal derivation of cells is the hallmark ofneoplasia and strongly implicates acquired somatic mu-tations giving survival advantage to a clonal cell popu-lation (22). Most somatic tissues are polyclonal unless amarked kinetic advantage within a subset of cells causesthem to proliferate at the expense of surrounding tissues,resulting in monoclonal tissues.

There are several tests to determine clonality. Themarkers for clone detection must define a readable andreliable feature linked to a particular cell type and mustinclude karyotypic alterations and single point mutations,among others. The presence of a common and nonran-dom alteration in all cells of a tumor confirm clonalorigin, but requires in many instances fresh, unfixed tis-sue—a condition that cannot be met with preneoplasia.Molecular demonstration of a genetic lesion, as small asa point mutation, within all cells is evidence that a com-mon progenitor cell contributed to that mutation. Thisapproach applies only to that fraction of cases in whicha “marker” mutation can be identified and fails specifi-cally to identify clonal proliferation that may have takenplace before the creation of a specific genetic lesion (Fig.1). In addition, those particular marker features must alsobe transmitted to descendant cells and must not induceapoptosis. Under the latter circumstance, any genetic al-teration does not result in a dominant clone (15).

A given clone should show proliferative advantagesover the remaining nonselected cell population to explaintumor growth. This proliferative advantage does not haveto be related to a higher proliferation rate. Apoptoticindices lower than those required normally for regularturnover can increase the total cell number. The mostimportant kinetic feature is the imbalance between cellproliferation and cell loss, ending in cellular over-growth (15,20). Genetic alterations in a kinetically activetumor cell population lead to progressive clone selectionand enhance genetic diversity in a given neoplasm aswell. Therefore, the inherent genetic instability within aneoplasm results in heterogeneity in tissues derived froma single cell. As soon as the neoplastic transformation

From the Department of Pathology (S.J.D.-C., H.J.W.), Tufts Uni-versity–New England Medical Center, Boston, Massachusetts, USA;St. Bartholomew’s and the Royal London School of Medicine andDentistry (S.J.D.-C.), London, UK; and the University of Malaga(A.B.), Spain.

Address correspondence and reprint requests to Dr. Salvador J.Diaz–Cano, Department of Histopathology & Morbid Anatomy, StBartholomew’s and the London NHS Trust, Whitechapel, London E11BB, UK (e-mail: [email protected]).

Diagnostic Molecular Pathology 10(1): 24–33, 2001 © 2001 Lippincott Williams & Wilkins, Inc., Philadelphia

24

takes place and neoplastic growth ensues, geneticallyidentical cells no longer exist. Nevertheless, certain ge-netic markers can be used to test clonal expansionswithin a tumor cell sample. These include gene rear-rangement analyses of immunoglobulin genes for B cellsand T-cell receptor genes for T cells (3,16).

We review the principles and requirements of theproper application of polymerase chain reaction (PCR)techniques in the detection of clonality using formalin-fixed and paraffin-embedded material. Two main groupsof techniques are presented: those based on the analysisof microsatellites (linked or not to X chromosome) andthose related to gene rearrangements. Their potential ap-plications in tumor pathology and pitfalls are summa-rized briefly. Lastly, some methodological recommenda-tions to obtain reliable results are given.

CLONALITY ASSAYS BASED ONMICROSATELLITE ANALYSIS

Currently, successful clonality assays using paraffin-embedded tissues are based mainly on analyses of widelydispersed, hypervariable regions composed of repetitivedeoxyribonucleic acid (DNA) sequences (microsatel-lites) (36). Several of these microsatellites have been usedextensively for DNA fingerprinting and are very usefulin genetic linkage analyses, based on their high percent-age of heterozygosity in the general population. The lossof heterozygosity (LOH) of a given genetic marker couldbe linked to loss of tumor suppressor genes (TSG) byDNA deletions (32–34), which would also contribute tothe multistep carcinogenesis process selecting cells withgrowth advantages (44). This progression of geneticevents has been found in most inherited cancer syn-dromes and plays an important role in sporadic cancerdevelopment.

Microsatellites have assumed an increasingly impor-tant role in this task because of their ubiquity, PCR“typability” (except for [dA] n multimers, whose sizepolymorphisms are difficult to type by PCR), Mendeliancodominant inheritance, and extreme polymor-phism (2,4). Microsatellites belong to the family ofhighly polymorphic and repetitive noncoding DNA se-quences. It must be noted that, although widely distrib-uted in the human genome, microsatellites are not uni-formly spaced. For example, they are underrepresentedin subtelomeric regions of chromosomes. Although theirorigin and function are not clear (36), their polymorphismhas been demonstrated to be very useful in delineatingcell lineage (2,4). The different length of these tandemlyarranged repeats in the paternally and maternally inher-ited alleles explains their high polymorphic informationcontent and obvious applications in genome mappingand positional cloning, personal identification, popula-tion genetic analysis, and the construction of human evo-lutionary trees. For clonality purposes, both X-linked andnon-X-linked clonality assays can be designed.

Tumor allelotyping (57,58)—the genotypic analysis ofall human chromosome pairs for regions of interstitialdeletion—is based conceptually on deletions of TSG al-leles. The term allelic imbalance is associated closelywith LOH and it is preferred when quantitation of minuteDNA amounts is not reliable (6,7,18,48), like in paraffin-embedded tissues after microdissection. Both findingsare evidence of clonal expansion in a tumor cell popu-lation, regardless of the DNA region deleted. Con-versely, if the selected polymorphic region is related toknown TSG, there is high probability of mutation orfunction dysregulation of the corresponding gene. Such asituation can explain the growth advantage of the tumorcells carrying that genetic alteration.

X-linked Clonality Analysis

X-linked clonality analysis can be used only forclonality analysis in informative female subjects, doesnot assess tumor heterogeneity, and does not provide anydata on the precise genetic alteration responsible forclonal proliferation (see Fig. 1). All methods of clonalityanalysis based on X chromosome inactivation (XCI) in-clude the ability to determine the paternally derived Xchromosome (Xp) from the maternally derived one (Xm).Many genes on the X chromosome are polymorphic andpermit the distinction between maternally and paternallyinherited X chromosomes. Their informativeness is re-lated directly to the frequency of their polymorphism ina population, from 29% heterozygosity for hypoxanthinephosphoribosyl transferase to more than 90% for the hu-man androgen receptor gene (26).

According to Lyon’s hypothesis (41), all X chromo-somes in a cell in excess of one are inactivated on a

FIG. 1. X-linked and non-X-linked clonality assays pro-vide complementary information. This polyclonal tissue byX-linked analysis would be termed monoclonal for a givenmarker (black star) after neoplastic-related clonal expan-sion takes place.

PCR-BASED TECHNIQUES FOR CLONALITY ANALYSIS OF NEOPLASTIC PROGRESSION 25

Diagn Mol Pathol, Vol. 10, No. 1, 2001

random basis during early embryogenesis. Inactivationof most, but not all, genes on one of the X chromosomesis mediated by a gene called Xist, and represents a grossimprinting of many genes to achieve a mammalian dosecompensation for X-linked genes, which renders mater-nal and paternal chromosomes nonequivalent function-ally. The mechanisms leading to XCI have not been char-acterized fully, but DNA methylation may maintain theinactive state once it is established. Approximately 60%of all Cytosine-phosphate-Guanidine (CpG) dinucleo-tides in the DNA of vertebrates are methylated at the C5position, but the frequency at particular sites varies be-tween cell types (31). These methylation patterns aretransmitted by clonal inheritance through the strong pref-erence of mammalian DNA (cytosine-5)—methyltrans-ferase for hemimethylated DNA. Methylation patternsare established during gametogenesis and early embryo-genesis (primordial cell pool of 16–30 cells) (23) andinvolve, among other DNA regions, promoter regions ofalleles on the inactive X chromosome (Xi), whereas al-leles on the active X chromosome (Xa) are normallyunmethylated (41). Given the small number of embryo-destined cells, it reasonable to expect unequal numbersof inactivated Xp and Xm, although X chromosome isinactivated randomly in each cell. Thus the Lyonizationratio (Xp

a/Xma) in the population follows a binomial dis-

tribution. Allelic variation is determined by the numberof progenitor cells at the moment of inactivation (vari-ance observed � pq/N, where p and q are probabilitiesof inactivating a particular X [both 0.5] and N is the stemcell pool size). The average Lyonization ratio is close to50:50 in large cell populations, although individualvariation has been found (23), resulting in skewing to-ward one allele. This is the reason for using as controlsfor unequal Lyonization the most closely related tissuethought not to be involved in the disease process (14).

The distinction of Xa from XI can be made by geneexpression analysis (messenger ribonucleic acid and pro-tein derive only from genes on Xa) or can be delineatedat the DNA level using methylation-sensitive restrictionendonucleases (Fig. 2). The last method can be per-formed in paraffin-embedded tissues, although it has theabsolute requirement of invariable and differential meth-ylation at the polymorphic locus in Xa and Xi. A perfectX-linked clonality assay would unite highly informativepolymorphism with an absolute differential methylationpattern between Xa and Xi. Therefore, an informativelocus amenable to PCR-based detection of nonrandomXCI must fulfill the following four criteria:

1. High frequency of heterozygosity2. A site that is methylated differentially in Xa and Xi,

and thus is subject to differential digestion by meth-ylation-sensitive restriction endonucleases

3. Polymorphic and methylated regions that are in suf-

ficient proximity for amplification by one PCR primerset

4. A target that amplifies efficiently and reliably (Fig.3). Noninvolved normal tissues related closely to thelesion must be tested simultaneously for unequalLyonization in each particular patient.

Several possible targets have been tested, but currentlythe best option is the human androgen receptorgene (55,56). This gene has a hypervariable CAG tri-nucleotide repeat in the coding region of its first exon,located less than 100 base pairs (bp) from four methyl-ation sites recognized by HhaI, and known to be meth-ylated on Xi but not on Xa (see Fig. 3).

Non-X-linked Clonality Analysis

These tests rely basically on the demonstration ofLOH, resulting from either hemizygosity (nonrandom in-terstitial DNA deletions) or homozygosity of mutant al-leles observed in neoplasms, but only provide informa-tion when the genetic marker represents or is linked to aTSG involved in the malignant transformation. Thesetests represent the first option to study tumor heteroge-neity if several samples are taken from a single neo-plasm (17), but two main problems must be considered.First, there is no specific sequence of genetic alterationsfor a given tumor, and therefore several genetic markersmust be tested to get information on clonality. Differentsequences of genetic alterations have been proposed forseveral tumor types and locations based on statistical

FIG. 2. Dual approach for the analysis of X-chromosomeinactivation in females. (Left) Gene expression analysesare based on messenger ribonucleic acid (mRNA) andprotein polymorphism from alleles located in the active Xchromosome (Xa). (Right) Conversely, deoxyribonucleicacid (DNA) analysis provides information through the am-plification of alleles located on inactive (methylated) Xchromosome (Xi) after DNA digestion with methylation-sensitive restriction endonucleases. PCR, polymerasechain reaction.

S. J. DIAN–CANO ET AL.26


analyses of case series (5,30,57), but the frequency ofeach genetic alteration is variable. Second, the back-ground level of LOH in normal tissues has been reportedto be between 4% and 20%, regardless of the detectionsystem used (8,13,51,59). A similar LOH frequency mustbe assumed as background in the evaluation of tumortissues (51). Considering the worst scenario of all geneticlesions being equally important and frequent (21), theprobability of randomly finding coexisting genetic alter-ations in normal tissues would be 0.22 � 4.0 10−2 fortwo genetic loci, 0.23 � 8.0 10−3 for three genetic loci,and so on. No single genetic alteration of TSG proves byitself that a given proliferation is monoclonal: The LOH

for that particular marker informs only on clonal expan-sion and cellular selection in genetically heterogeneoustumor cell populations. Only the accumulation of geneticlesions in TSG supports a monoclonal origin of tumors(15), especially if multiple samples from the same tumorshow concordant genetic alteration (14,18) (see Method-ological Aspects).

CLONALITY ANALYSIS BASED ON SPECIFICGENE REARRANGEMENTS

This application will not be covered in detail becauseit is better known and applicable to malignant lympho-mas only. It has been demonstrated to be useful in thediagnosis of malignant lymphomas, in which the pres-ence of homogeneous-appearing lymphoid cell over-growths are considered histologic evidence of clonal ex-pansion and malignancy (3,10,16).

Committed lymphoid precursors undergo unique se-quential assembly of the heavy and light chains of im-munoglobulin (B-cell precursors) and T-cell receptorchains (T-cell precursors) during their maturation (16).Their diversity is based on somatic DNA deletions, tem-plate-independent nucleotide additions, and specificsplicing (39). These processes involve, in the case ofimmunoglobulin heavy chain, three highly variable re-gions, complementarity-determining regions, whichseparate four framework regions (FR) (10). The locationsof DNA breaks are determined by short DNA sequences(7 and 9 bp) recognized by the recombination–activatingproteins. The presence of length-specific spacer se-quences provides the right splicing and determines theDNA rearrangement order (e.g., D-J and V-DJ in thecase of immunoglobulin). Template-independent nucleo-tide additions give the final DNA rearrangement of im-munoglobulin or T-cell receptor chains. All gene rear-rangements in lymphoid precursors take place only incells expressing terminal deoxynucleotidyl transferase(16).

The PCR design for gene rearrangement detection hasto consider the special situation of DNA sequence addi-tion and deletion. It is particularly important for theprimer binding regions, which must be located in lessvariable sequences (FR regions for immunoglobulingene rearrangement) to avoid failed amplification result-ing from their loss. This factor also helps to explainfalse-negative results in the detection of clonal rear-rangement in well-differentiated lymphoid neoplasms(related to gene hypermutability). Complete rearrange-ments may be associated with the loss of inner FR re-gions (in B cells) and with �-chain (in T cells), the mostsensitive primer binding regions to detect early rear-rangement. Therefore, a broad approach is recommendedfor detecting clonal rearrangement, including at least twodifferent set of primers (FR III–FR IV and FR I–FR IV)in B-cell lymphoid lesions (16), and several primer sets

FIG. 3. (A) Structure of the androgen receptor (AR) gene(exon 1). A highly polymorphic CAG repeat sequence lo-cation 100 base pairs (bp) downstream of a potentiallymethylated CG island allows a reliable yield of both com-ponents in a single polymerase chain reaction amplifica-tion (between 250 and 300 bp). Both alleles can be dif-ferentiated easily by size, and can be amplified only fromthe inactive X chromosome (Xi, digestion-resistant allele)in the HhaI-digested sample. (B) Allelic patterns of ARgene in monoclonal and polyclonal tissues. The presenceof a single allele in HhaI-digested samples (D) in informa-tive cases (two alleles in undigested samples [U]) definesmonoclonal tissues (left, right), whereas the retention ofboth alleles characterizes polyclonal ones (center). Eachcell shows the allelic structure after HhaI digestion to em-phasize that the polymerase chain reaction amplificationcomes from the intact allele only (methylated X chromo-some). HMW, high molecular weight; LMW, low molecularweight.



for the T-cell receptor �-chain in T-cell lymphoid le-sions (3).

All PCR-based analyses must be run with appropriatecontrols, including internal positive (provided by non-specific amplifications of locus-homologous sequences),positive (lymphoid proliferation with clonal rearrange-ment), negative (polyclonal lymphoid proliferation), andtechnique efficiency (mixture of monoclonally and poly-clonally rearranged DNA tested previously) (16). The lastcontrol is needed essentially for the clinical applicationof these tests when sensitivity is an absolute require-ment (16). Their interpretation must also consider thepresence of false-positive and false-negative cases, andtheir causes (3,16). It should be emphasized that clonalgene rearrangement does not mean malignancy, becauseeven some benign conditions can show it. Furthermore,well-defined monoclonal PCR bands can be observedwhen the lymphoid DNA template is present in traceamounts. Nonamplifiable DNA may result from the fail-ure of PCR amplification for technical reasons (contami-nation, highly fragmented DNA, etc.) or when the pres-ence of rearranged DNA is less than the sensitivity level(approximately 1% of the total cell population). Lastly,oligoclonal proliferation (especially in immunodeficien-cy-related lymphomas) can result in smear patterns fromoverlapped clonal bands.

GENERAL INTERPRETATION CRITERIA ANDQUALITY CONTROLS OF CLONALITY

MICROSATELLITE ANALYSES

Two complementary aspects are evaluated with X-linked and non-X-linked clonality analyses (see Fig. 1).XCI takes place early during embryologic development,usually before and unrelated to any genetic event in-volved in the malignant transformation. Therefore, it isreally informing about clonality in tumors and precan-cerous conditions, although it is not able to inform on thespecific molecular alterations. Different polymorphic re-gions have been related to TSG (recessive trait genes).Therefore, LOH analyses enable us to study the molecu-lar pathways involved in the malignant transformation,and to test clonal expansion and tumor heterogeneity ifvarious molecular markers are assessed in samples fromdifferent tumor areas. Moreover, the correlation of thesemolecular markers with other pathologic parameters(like tumor cell invasion, nuclear grade, proliferation in-dices, etc.) can help us to identify high-risk patients andto understand the process of multistep carcinogenesis.All nonrandom gene rearrangements, as markers of ma-lignancy, fulfill the criteria mentioned for LOH analyses.

Any molecular analysis must be run with appropriatecontrols, including known positive (monoclonal prolif-erations, homozygous for the marker) and negative(polyclonal proliferations, heterozygous for the marker)

controls, and from embryologically related tissues forXCI analysis to exclude a skewed Lyonization ratio (18,20,43). The PCR approach for microsatellite analysis (in-cluding both clonality assays) must amplify the rightlocus and accurately identify informative patients (twodifferent alleles present in control tissues; Fig. 4). Thisissue becomes especially important for microsatelliteanalysis when the presence of extra bands is not excep-tional, especially in cases of internal labeling. A sine quanon requirement for clonality analysis is the identifica-tion of a polymorphic locus in the normal control (seeFig. 4). In every case, the tumor sample must be com-pared with normal controls from the same patient to testpatient heterozygosity for the marker—a feature thatmust be maintained in HhaI-digested samples in testsbased on DNA methylation, such as XCI assays (Fig. 5).Patients with two identical alleles or showing skewedLyonization in control tissues should be considered non-informative and should be excluded from clonalityevaluation. Another cause of noninformative cases is theanomalous expansion or reduction of tandem repeats re-sulting from microsatellite instability that results in extrabands (see Fig. 4). True new bands, as true evidence ofmicrosatellite instability, are located normally in the ex-pected size range (usually approximately 100 bp), aboveor below the expected PCR product.

Therefore, LOH and allelic imbalances of gene locican be interpreted as evidence of monoclonal prolifera-tion (for X-linked assays of templates digested by meth-ylation-sensitive restriction enzyme, such as HhaI) orclonal expansion (for non-X-linked assays, includinggene rearrangements), if noninformative cases have beenexcluded previously (see Fig. 1 and 4). Positive allelic

FIG. 4. Polymerase chain reaction-based analysis of mi-crosatellites in clonality assays. Allelic polymorphism (P)in any given control identifies informative (I) patients (left).The tumor (T)–control (C) comparison allows case clas-sification as normal (retention of heterozygosity [ROH]) orabnormal (loss of heterozygosity [LOH] and microsatelliteinstability [MSI]). Cases showing MSI could be eithermonoclonal or polyclonal and should be excluded fromclonality assays. NP, no polymorphism; NI, noninforma-tive patient.



imbalances are determined case by case, in relation to thedensitometric allelic ratio in the normal control (43),which requires a threshold of at least 4:1 in skewed data.The host cell contamination of tumor samples could givefalse heterozygous results that would require careful mi-crodissection and microscopic control of the sample col-lection.

APPLICATIONS AND PITFALLS

Clonality assays can be useful in the analysis of dif-ferent biologic processes. They have been used mainly inthe study of malignant transformation and tumor pro-gression. At this level, the complementary informationprovided by X-linked and non-X-linked markers contrib-utes to the definition of the real nature of the lesion, asmentioned earlier. Monoclonal patterns would supportthe neoplastic nature, although they have been describedin other proliferative processes such as aggressive fibro-matosis (1) or focal nodular hyperplasia of the liver (25).Similarly, the polyclonal patterns reported in sacrococ-cygeal cystic teratomas suggest their hamartomatous na-ture, opposed to the monoclonal immature teratoma (54).The acquisition of additional genetic deletions in certainhistologic areas favors a molecular progression as re-ported for the adenoma–carcinoma sequence in co-lon (52), or sporadic neuroendocrine tumors of the pan-creas (45). The acquisition of genetic changes has beenconsidered evidence of molecular progression that alsoresults in tumor heterogeneity. Tests demonstrating thoseabnormalities only prove clonal expansions, but wouldsupport monoclonality if several markers from differenttumor areas show concordant genetic alterations (15, 18).

The identity of synchronic or metachronic tumors canbe tested at the molecular level using these clonalitymarkers. Both extremes have been reported in tumors.Some coexistent tumors have revealed the same patternof genetic markers (both X linked and non-X linked),suggesting that a common progenitor contributed tothose lesions, and thus supporting a multifocal ratherthan a multicentric origin. This particular situation hasbeen demonstrated in bladder tumors (18,53), human im-munodeficiency virus-associated Kaposi’s sarcoma (47),disseminated peritoneal leiomyomatosis (46), or multi-focal C-cell hyperplasias and nodular adrenal medullaryhyperplasias associated with multiple endocrine neopla-sia 2A (20a). In the case of malignant neoplasms, thisshared genetic alteration would support a metastatic ori-gin for the tumors (11), or a common cellular origin forbiphasic neoplasms (60). Opposite findings have beenreported for prostatic tumor foci when the heterogeneousgenetic composition suggests either an independent evo-lution of those foci from a common progenitor or a com-pletely different origin (28,29).

Lastly, any well-characterized genetic alteration canbe used for the early detection of recurrences, both localand systemic. These alterations pick clonal expansionfrom selected groups of cells, and they have been pro-posed as tools to study the resection margins in conser-vative surgery, as reported for head and neck squamouscell carcinomas (35). Likewise, the presence of minimalresidual disease can be defined better at the molecularlevel by detecting circulating tumor cells with specificgene alterations, more frequently gene rearrangements,as reported for malignant lymphoma (16) or sarcomas ofthe Ewing family (12).

These molecular studies should be interpreted withcaution. Some considerations should be made to set theproper value of these techniques. These considerationsinclude tumor cell heterogeneity, sample size, tissue con-trol, restriction enzyme digestion and abnormal methyl-ation, and artifactual allelic dropout.

Tumor Cell Heterogeneity

Tumor cell heterogeneity is linked to genetic instabil-ity and biologic progression. This genetic heterogeneityis often reflected in phenotypic expression. Examplesinclude the presence of a Ki-ras point mutation in carci-nomatous areas from adenomatous polyposis coli (APC)-mutated sporadic colorectal adenomas (52). Similarly,mutations in cell cycle regulators (e.g., tumor protein 53,retinoblastoma, cyclins, cyclin-dependent kinases, andcyclin-dependent kinase inhibitors) have been related toeither proliferative advantages or apoptotic dysregulation(9). So, it can explain their association with tumor gradeas far as nuclear atypia (pleomorphism, chromatin fea-tures, and size variability) is expression of both prolif-eration and apoptosis. Tumor heterogeneity must be

FIG. 5. Gel patterns of androgen receptor (AR) allelesfrom controls. Only polyclonal controls (two-band patternon HhaI-digested samples [D]) from tissues with AR–allelic polymorphism (two-band pattern on HhaI-undigested samples [U]) are informative for X-linkedclonality assays. Monoclonal controls (even from polymor-phic tissues) and tissues with AR–allelic monomorphismlack informativeness for clonality purposes.



studied using several tumor samples of appropriate sizefrom each tumor (see the following section).

Sample Size

Sample size is a limiting factor. To increase samplehomogeneity and to avoid normal cell contamination,very small samples (even single cells) have been used ingenetic analyses. However, the lower the number ofcells, the higher the probability of false monoclonal pat-terns based on inadequate sampling. Although early XCItakes place at random and usually gives a chessboardpattern, small cell populations descended from a com-mon stem cell may grow together like a clone (patch sizeconcept or contiguous cellular regions of the same lin-eage). For this reason, monoclonal XCI patterns are re-ported in breast lobules when studied from singlesamples. Multiple samplings from different areas andsample sizes larger than 100 cells or 0.25 mm2 can avoidthis problem.

The sample size in tumor cell analysis is, therefore, animportant parameter, particularly in light of tumor cellheterogeneity. Microdissection techniques allow us topick up very small samples selectively, which can showfalse cellular homogeneity, based on LOH or allelic im-balance. If the tumor cell populations selected for mo-lecular analysis are taken before they become a biologi-cally prominent component (with proliferative or inva-sive advantages), the results obtained may be confusingand irrelevant clinically, and need to be evaluated in theproper biologic context. This would be the case withmicroheterogeneity in tumors that tend to give disparateresults with meanings that remain essentially unknown.Except for intraepithelial proliferation, all cell sampleswith microdissection provide target cell-rich sampleswith a varying degree of host cell contamination (includ-ing stromal, inflammatory, and endothelial cells). Takingall these factors into consideration, multiple samplesfrom the same case should always be studied, and assaysshould be performed in duplicate before accepting theresults as relevant.

Tissue Control

Tissue control from the same patient is an absoluterequirement to test patient heterozygosity for a particularTSG marker (as mentioned earlier; see Fig. 5 and Fig. 6).Ideally, embryologically related tissues are the best toshow a reliable X-linked clonality pattern. Oncogene ge-netic changes such as RAS point mutations that are notfound in the germline are excluded from this approachand do not require heterozygosity control.

Restriction Enzyme Digestion andAbnormal Methylation

Every single step for all molecular tests must be con-trolled for completion to avoid false results. Appropriate

controls must be run, especially for restriction enzymedigestions. Because XCI analysis is based on differentialDNA methylation of one allele from X-chromosomegenes (e.g., human androgen receptor gene) and relies onendonuclease digestion by methylation-sensitive restric-tion enzymes, suboptimal enzymatic digestion provides achanged clonality pattern in the case of monoclonaltissues. Likewise, abnormal methylation provides poten-tially the same false result in the case of hypermethyl-ation, whereas hypomethylation could affect the clonal-ity pattern in polyclonal tissues giving a pseudomono-clonal one (see Fig. 6). Additionally, methylationabnormalities occur during the course of malignant trans-formation (37). Hypomethylation has been described inrelation to increased proliferation during early stages ofneoplasms, whereas hypermethylation has been linked to

FIG. 6. Methylation level and allelic patterns of androgenreceptor (AR). (A) Monoclonal tissues would only changethe clonality pattern if the amplified AR locus was hyper-methylated. A pseudopolyclonal pattern would be re-vealed on gel electrophoresis (right). (B) Polyclonal tis-sues would only change the clonality pattern if the ampli-fied AR locus was hypomethylated. A pseudomonoclonalpattern would be revealed on gel electrophoresis (B).PCR, polymerase chain reaction; U, HhaI-undigestedsample; D, HhaI-digested sample; HMW, high molecularweight; LMW, low molecular weight.



late stages associated with a higher mutation rate andtumor progression (42).

Artifactual Allelic Dropout

PCR bias against one allele (especially the larger one)can result in preferential amplification of the other allele(usually the smaller) (24,49). An appropriate extractionmethod providing DNA of enough quality (3,19), andPCR designs including both long denaturation and ex-tension during the first three cycles, and 7-deaza-deoxy-guanidine triphosphate (dGTP) in the amplification mix-ture to improve the amplification of CG-rich DNA re-gions reasonably avoids that bias (18,20,49).

METHODOLOGIC ASPECTS

Formalin-fixed, paraffin-embedded tissues are ana-lyzed more easily at the DNA level, resulting in partfrom the better preservation of this nucleic acid. There-fore, any PCR-based technique applied to this materialshould consider DNA extraction, DNA modification(such as restriction enzyme digestion), target amplifica-tion, and adequate gel resolution of the products.

DNA Extraction Process

Several chemical modifications are induced in tissuesby fixation and processing, including cross-linking be-tween basic amino acids of proteins and the aminogroups on DNA bases. Nonspecific amplifications arecaused mainly by primer-independent, but DNA poly-merase- and cycling-dependent, incorporation of nucleo-tides into DNA, possibly related to DNA repair and/orinternal priming (40). This is the reason for complete andintense protein digestion before DNA purification (19).Denaturing reagents can break the cross-linked strandsbut in turn provide short-length DNA strands, precludingtheir use in protocols that require DNA of 250 to 500 bpin length. The general DNA quality of the extractedDNA should be tested by gel electrophoresis of the pro-tein-digested sample, universal DNA amplification usingdegenerated oligonucleotide primer–PCR, or amplifying�-globin gene with primer sets at least 100 bp lengthierthan the final DNA target.

The easiest protocol giving the highest DNA quality is55 to 60°C prolonged proteinase K digestion (5–7 days,with every-day enzyme replacement) (19). The standardphenol–chloroform purification protocol results in thebest contaminant-free DNA for any PCR application.Negative amplification resulting from sample contami-nation (specific amplification lacking with no primerdimers) can be avoided by diluting the sample.

DNA Modifications

Certain applications need original DNA strands, suchas those based on the genomic imprinting of XCI. The

presence of methylated cytosine can be tested by sampledigestion by methylation-sensitive restriction endonucle-ases. In any case, appropriate internal control should beincluded to prove complete digestion. The samples al-ready show smear patterns, and even they may be unde-tectable using gel electrophoresis (especially for micro-dissected samples). A logical way to accomplish thisissue is to include DNA mimickers in every sample un-dergoing restriction enzyme digestion. These mimickersare normally viral DNA (such as phages) and shouldfulfill some requirements. They must be linear anddouble stranded (like human genomic DNA), and theymust contain base sequences recognized by the testedenzyme with reliable pre- and postdigestion patterns. Inaddition, no sequence similarity able to give nonspecificamplification in further PCR should be present. Xho-I-linearized �X174RII phage represents an ideal mimic forHhaI digestion used for XCI analysis.

XCI analysis tests the differential methylation level ina CpG island approximately 100 bp upstream of theCAG repeat (56). Different methylation-sensitive restric-tion enzymes have been used, especially HhaI and HpaII.The first provides more reliable results because of itsactivity with single-strand DNA (that activity has notbeen demonstrated for HpaII). We must keep in mindthat the embedding process partly denatures DNA and,therefore, single-strand DNA is a normal component inarchival material.

Target Amplification by PCR and Gel Resolution

All PCR methods should consider the appropriate con-ditions regarding Mg2+ and primer concentrations,nucleotide concentration, number of cycles in case ofPCR-based quantitative analyses, PCR product labeling,and detection methods. The standards for all PCR tech-nique need specific optimization for each set of primeraccording to Mg2+ (normally 1.5 mM) and nucleotideconcentration (in the 50-�M range for microsatelliteanalyses). PCR cycling conditions should always con-sider the number of cycles to avoid product saturationunacceptable for any quantitative PCR design; in gen-eral, between 25 and 30 cycles give adequate amplifica-tion in accordance with the initial DNA concentration.We optimized experimentally the conditions for PCR asfollows: the reactions were run with 1.5 mM MgCl2,using 0.3 �M each primer, 200 �M each dNTP (includ-ing 7-deaza-dGTP instead of dGTP) and 1 �L template.A long denaturation (4 minutes) was used in the firstthree cycles, the annealing temperature was 55°C, andthe number of cycles was optimized experimentally to28. A “hot start” (addition of primers to mixtures kept at85°C) should also be included to facilitate the completedenaturation of DNA strands with high CG content in theinitial amplifications (18, 20, 43).



DNA samples from microdissected tissues are notideal for reliable quantification and are generally runwith unknown target DNA concentration. So, relativequantification of both allelic bands must be taken intoconsideration to determine whether LOH (or allelic im-balance) is present. This issue brings us to consider thelabeling and detection method. 32P- and 33P-based radio-labeling represent the standard protocols, including bothexternal (one primer is 5� labeled) and internal (labelednucleotide in the PCR mixture) methods. Although thelatter usually gives more background, it permits the high-est sensitivity for microdissected paraffin-embeddedsamples. Allelic separation can be achieved by runningthe samples far enough into high-resolution denaturingpolyacrylamide gels (variable concentrations of formam-ide and urea). Some other detection methods have beenused, including fluorescent labeling (6,7) and silverstaining of PCR products (38). The highest sensitivity isachieved by radioisotopic methods, which remains thestandard for molecular detection of genetic alterations,especially in formalin-fixed, paraffin-embedded mate-rial. Additionally, the ratio between signal and initialDNA amount is highly variable for silver-stained gels,making the applications of this technique less reliable forquantification. Different technical approaches have beenused to detect interstitial DNA deletion and single basechanges (mutations/polymorphisms), including single-strand conformational polymorphism, denaturant gradi-ent gel electrophoresis, mutant allele-specific amplifica-tion, ribonuclease (RNase) protection, etc. (27,50). Al-though the final proof for any mutation must be directsequencing, one of the most sensitive methods for de-tecting single base changes is PCR/denaturant gradientgel electrophoresis, which is able to distinguish DNAstrands differing in only one base (3,18).

Lastly, the linear ratio between radioactive emissionand signal deposition can be maintained by film preflash-ing to get a 0.1 to 0.2-OD unit absorbance increase at 540nm in the preflashed film. In addition, signal stabilizationduring autoradiogram development requires −70°C stor-age. The allelic ratio has to be quantitated in normalizedsamples to exclude any potential contamination with nor-mal tissue (43). At that level, different computer soft-ware is available to aid in analysis. �

Acknowledgment: Presented in part at the Xth Interna-tional Congress of Histochemistry and Cytochemistry, Kyoto,Japan, August 18–23, 1996; and at the XXIst Interna-tional Congress of the International Academy of Pathology,Budapest, Hungary, October 20–25, 1996.

REFERENCES

1. Alman BA, Pajerski ME, Diaz–Cano S, Corboy K, Wolfe HJ.Aggressive fibromatosis (desmoid tumor) is a monoclonal disor-der. Diagn Mol Pathol 1997;6:98–101.

2. Boland CR. Setting microsatellites free. Nat Med 1996;2:972–4.3. Brady SP, Magro CM, Diaz–Cano SJ, Wolfe HJ. Analysis of

clonality of atypical cutaneous lymphoid infiltrates associated withdrug therapy by PCR/DGGE. Hum Pathol 1999;30:130–6.

4. Brentnall TA. Microsatellite instability. Shifting concepts in tu-morigenesis. Am J Pathol 1995;147:561–3.

5. Califano J, van der Riet P, Westra W, et al. Genetic progressionmodel for head and neck cancer: implications for field canceriza-tion. Cancer Res 1996;56:2488–92.

6. Cawkwell L, Bell SM, Lewis FA, Dixon MF, Taylor GR, QuirkeP. Rapid detection of allele loss in colorectal tumours using mic-rosatellites and fluorescent DNA technology. Br J Cancer 1993;67:1262–7.

7. Cawkwell L, Li D, Lewis FA, Martin I, Dixon MF, Quirke P.Microsatellite instability in colorectal cancer: improved assess-ment using fluorescent polymerase chain reaction. Gastroenterol-ogy 1995;109:465–71.

8. Chen LC, Kurisu W, Ljung BM, Goldman ES, Moore DI, SmithHS. Heterogeneity for allelic loss in human breast cancer. J NatlCancer Inst 1992;84:506–10.

9. Cordon–Cardo C. Mutations of cell cycle regulators. Biologicaland clinical implications for human neoplasia. Am J Pathol 1995;147:545–60.

10. Cossman J, Zehnbauer B, Garrett CT, et al. Gene rearrangementsin the diagnosis of lymphoma/leukemia. Guidelines for use basedon a multiinstitutional study. Am J Clin Pathol 1991;95:347–54.

11. Cuatrecasas M, Matias–Guiu X, Prat J. Synchronous mucinoustumors of the appendix and the ovary associated with pseudo-myxoma peritonei. A clinicopathologic study of six cases withcomparative analysis of c-Ki-ras mutations. Am J Surg Pathol1996;20:739–46.

12. de Alava E, Lozano MD, Patino A, Sierrasesumaga L, Pardo–Min-dan FJ. Ewing family tumors: potential prognostic value of re-verse–transcriptase polymerase chain reaction detection of mini-mal residual disease in peripheral blood samples. Diagn MolPathol 1998;7:152–7.

13. Deng G, Lu Y, Zlotnikov G, Thor AD, Smith HS. Loss of hetero-zygosity in normal tissue adjacent to breast carcinomas. Science1996;274:2057–9.

14. Diaz–Cano SJ. Designing a molecular analysis of clonality in tu-mours. J Pathol 2000;191:343–4.

15. Diaz–Cano SJ. Clonality studies in the analysis of adrenal medul-lary proliferations: application principles and limitations. EndocrPathol 1998;9:301–16.

16. Diaz–Cano S. PCR-based alternative for diagnosis of immuno-globulin heavy chain gene rearrangement: principles, practice, andpolemics. Diagn Mol Pathol 1996;5:3–9.

17. Diaz–Cano SJ, Blanes A. Influence of intratumor heterogeneity inthe interpretation of marker results in pheochromocytomas. JPathol 1999;189:627–8.

18. Diaz–Cano SJ, Blanes A, Rubio J, Matilla A, Wolfe HJ. Molecularevolution and intratumor heterogeneity by topographic compart-ments in muscle-invasive transitional cell carcinoma of the urinarybladder. Lab Invest 2000;80:279–89.

19. Diaz–Cano SJ, Brady SP. DNA extraction from formalin-fixed,paraffin-embedded tissues: protein digestion as a limiting step forretrieval of high-quality DNA. Diagn Mol Pathol 1997;6:342–6.

20. Diaz–Cano SJ, de Miguel M, Blanes A, Tashjian R, Galera H,Wolfe HJ. Clonality as expression of distinctive cell kinetics pat-terns in nodular hyperplasias and adenomas of the adrenal cortex.Am J Pathol 2000;156:311–9.

20a. Diaz-Cano SJ, de Miguel M, Blanes A, Tashjian R, Galera H,Wolfe HJ. Clonal patterns in phaechromocytomas and MEN-2Aadrenal medullary hyperplasias: histologic and kinetic correlates. JPathol 2000;192:221–8.

21. Diaz–Cano SJ, Wolfe HJ. Clonality in Kaposi’s sarcoma. N Engl JMed 1997;337:571–2.

22. Fialkow PJ. Clonal origin of human tumors. Biochim Biophys Acta1976;458:283–321.

23. Fialkow PJ. Primordial cell pool size and lineage relationships offive human cell types. Ann Hum Genet 1973;37:39–48.

24. Findlay I, Matthews P, Quirke P. Multiple genetic diagnoses from



single cells using multiplex PCR: reliability and allele dropout.Prenat Diagn 1998;18:1413–21.

25. Gaffey MJ, Iezzoni JC, Weiss LM. Clonal analysis of focal nodularhyperplasia of the liver. Am J Pathol 1996;148:1089–96.

26. Gilliland DG, Blanchard KL, Levy J, Perrin S, Bunn HF. Clonalityin myeloproliferative disorders: analysis by means of the polymer-ase chain reaction. Proc Natl Acad Sci U S A 1991;88:6848–52.

27. Housman DE. DNA on trial—the molecular basis of DNA finger-printing. N Engl J Med 1995;332:534–5.

28. Hugel A, Wernert N. Loss of heterozygosity (LOH), malignancygrade and clonality in microdissected prostate cancer. Br J Cancer1999;79:551–7.

29. Jenkins RB, Qian J, Lieber MM, Bostwick DG. Detection of c-myconcogene amplification and chromosomal anomalies in metastaticprostatic carcinoma by fluorescence in situ hybridization. CancerRes 1997;57:524–31.

30. Jones PA, Droller MJ. Pathways of development and progressionin bladder cancer: new correlations between clinical observationsand molecular mechanisms. Semin Urol 1993;11:177–92.

31. Kappler JW. The 5-methylcytosine content of DNA: tissue speci-ficity. J Cell Physiol 1971;78:33–6.

32. Knudson AG. Antioncogenes and human cancer. Proc Natl AcadSci U S A 1993;90:10914–21.

33. Knudsen AG Jr. Mutation and cancer: statistical study of retino-blastoma. Proc Natl Acad Sci U S A 1971;68:820–3.

34. Knudsen AG Jr, Hethcote HW, Brown BW. Mutation and child-hood cancer: a probabilistic model for the incidence of retinoblas-toma. Proc Natl Acad Sci U S A 1975;72:5116–20.

35. Koch WM, Brennan JA, Zahurak M, et al. p53 Mutation andlocoregional treatment failure in head and neck squamous cellcarcinoma. J Natl Cancer Inst 1996;88:1580–6.

36. Koreth J, O’Leary JJ, McGee JOD. Microsatellites and PCR ge-nomic analysis. J Pathol 1996;178:239–48.

37. Laird PW, Jaenisch R. DNA methylation and cancer. Hum MolGenet 1994;3:1487–95.

38. Lau DH, Yang B, Hu R, Benfield JR. Clonal origin of multiplelung cancers: K-ras and p53 mutations determined by nonradio-isotopic single-strand conformation polymorphism analysis. DiagnMol Pathol 1997;6:179–84.

39. Lewis SM, Wu GE. The origin of V(D)J recombination. Cell 1997;88:159–62.

40. Long AA, Komminoth P, Lee E, Wolfe HJ. Comparison of indirectand direct in-situ polymerase chain reaction in cell preparationsand tissue sections. Detection of viral DNA, gene rearrangementsand chromosomal translocations. Histochemistry 1993;99:151–62.

41. Lyon MF. Some milestones in the history of X-chromosome in-activation. Annu Rev Genet 1992;26:16–28.

42. Magewu AN, Jones PA. Ubiquitous and tenacious methylation ofthe CpG site in codon 248 of the p53 gene may explain its frequentappearance as a mutational hot spot in human cancer. Mol Cell Biol1994;14:4225–32.

43. Mutter GL, Boynton KA. X chromosome inactivation in the nor-

mal female genital tract: implications for identification of neopla-sia. Cancer Res 1995;55:5080–4.

44. Nowell PC. The clonal evolution of tumor cell populations. Sci-ence 1976;194:23–8.

45. Perren A, Roth J, Muletta–Feurer S, et al. Clonal analysis of spo-radic pancreatic endocrine tumours. J Pathol 1998;186:363–71.

46. Quade BJ, McLachlin CM, Soto–Wright V, Zuckerman J, MutterGL, Morton CC. Disseminated peritoneal leiomyomatosis. Clonal-ity analysis by X chromosome inactivation and cytogenetics of aclinically benign smooth muscle proliferation. Am J Pathol 1997;150:2153–66.

47. Rabkin CS, Janz S, Lash A, et al. Monoclonal origin of multicen-tric Kaposi’s sarcoma lesions. N Engl J Med 1997;336:988–93.

48. Randerson J, Cawkwell L, Jack A, et al. Fluorescent polymerasechain reaction of a panel of CA repeats on chromosome 6 in theindolent phase of follicular centre cell lymphoma. Br J Cancer1996;74:942–6.

49. Ray PF, Handyside AH. Increasing the denaturation temperatureduring the first cycles of amplification reduces allele dropout fromsingle cells for preimplantation genetic diagnosis. Mol Hum Re-prod 1996;2:213–18.

50. Rosenthal N. Molecular medicine. Recognizing DNA. N Engl JMed 1995;333:925–7.

51. Sager R. Tumor suppressor genes: the puzzle and the promise.Science 1989;246:1406–12.

52. Shibata D, Schaeffer J, Li ZH, Capella G, Perucho M. Geneticheterogeneity of the c-K-ras locus in colorectal adenomas but notin adenocarcinomas. J Natl Cancer Inst 1993;85:1058–63.

53. Sidransky D, Frost P, Von Eschenbach A, Oyasu R, Preisinger AC,Vogelstein B. Clonal origin bladder cancer. N Engl J Med 1992;326:737–40.

54. Sinnock KL, Perez–Atayde AR, Boynton KA, Mutter GL. Clonalanalysis of sacrococcygeal “teratomas.” Pediatr Pathol Lab Med1996;16:865–75.

55. Sleddens HF, Oostra BA, Brinkmann AO, Trapman J. Trinucleo-tide (GGN) repeat polymorphism in the human androgen receptor(AR) gene. Hum Mol Genet 1993;2:493.

56. Sleddens HF, Oostra BA, Brinkmann AO, Trapman J. Trinucleo-tide repeat polymorphism in the androgen receptor gene (AR).Nucl Acid Res 1992;20:1427.

57. Vogelstein B, Fearon ER, Hamilton SR, et al. Genetic alterationsduring colorectal-tumor development. N Engl J Med 1988;319:525–32.

58. Vogelstein B, Fearon ER, Kern SE, et al. Allelotype of colorectalcarcinomas. Science 1989;244:207–11.

59. Wolman SR, Heppner GH. Genetic heterogeneity in breast cancer.J Natl Cancer Inst 1992;84:469–70.

60. Zhuang Z, Lininger RA, Man YG, Albuquerque A, Merino MJ,Tavassoli FA. Identical clonality of both components of mammarycarcinosarcoma with differential loss of heterozygosity. ModPathol 1997;10:354–62.



Correspondence

Clonal Origin and Expansions in Neoplasms:Biologic and Technical Aspects Must BeConsidered Together

To the Editor-in-Chief:

Microsatellite-based clonality assays include the analysisof X-chromosome inactivation (XCI) and loss of heterozy-gosity (LOH) of tumor suppressor genes, and have beenrarely applied to differentiate clonal origin from clonalexpansion in neoplasms. The key elements for that dis-tinction are: tumor natural history with particular attentionto the relative timing between test conversion and clonalexpansion, the lesion cell kinetic, and sample conditions.Studies based on allele ratio of genes involved in thetransformation pathway must validate technique condi-tions to obtain reliable quantification methods able todetect clonal growths. These aspects are relevant and,probably due to space restrictions, have not been con-sidered in detail in a recent paper on the clonality ofin-transit melanoma metastasis.1

Although clonality is considered the hallmark of neo-plasms, the distinction between clonal origin and clonalexpansion in tumors remains controversial. A priori, a mono-clonal proliferation is assumed to be neoplastic, whereas apolyclonal lesion is thought to be reactive. However, thereare many exceptions to this rule. Additionally, there is noconsensus on the application of clonality markers. Clonalityanalysis has been used to test malignant transformationand tumor progression,2,3 but the results must always beinterpreted in view of the natural history of the neoplasm.The relationship between the molecular marker and thepathway of neoplastic transformation is essential, in partic-ular, the relative timing between the positive conversion ofthe marker and the clonal expansion. Clonality results willsupport a clonal origin only if the clonal expansion occursafter the positive conversion (Figure 1). Positive conversionstaking place after the clonal expansion will result in hetero-geneous marker patterns, which do not support clonal ori-gin.4 This is a key element for studies based on the analysisof tumor suppressor genes, especially if reduced number ofcells (microdissected samples) are used.

Transformed cells result in neoplasms if geneticallydamaged cells are able to expand clonally. In contrast,extensive genetic damage triggering cell apoptosis willnot result in neoplasms. Therefore, it is artificial to sepa-rate the analysis of tumor clonality and cell kinetic (pro-liferation/apoptosis), as demonstrated by the close rela-tionship between them in benign adrenal corticalproliferative lesions.5 Expanding clones would also sug-

gest that somatic genetic alterations contribute to thekinetic advantage of those cells, which eventually out-number other cells and result in monoclonal patterns. Inthat sense, clonality would be the by-product of tumorcell selection, especially for advanced neoplasms, andone of the first alterations in early neoplasms as well(Figure 1). These situations are highlighted by LOH anal-ysis of tumor suppressor genes (advanced neoplasms)and XCI assays (early neoplasms). This combined anal-ysis of clonality and cell kinetics better defines the evo-lution and progression of neoplasms.2,3,5–8 LOH analysisof tumor suppressor genes in a given tumor will inform onclonal origin only if concordant patterns with several ge-netic markers are demonstrated.4,8,9 The interpretationmust consider that true monoclonal lesions retain theconstitutional heterozygosity before the conversion pointand that tumor heterogeneity and progressive cell selec-tion can result in discordant microsatellite patterns insamples from different areas within a single tumor (intra-tumoral heterogeneity).6,10,11 If the genetic abnormalitiesdetermine a kinetic advantage, tumor cells revealing LOHwill overgrow and become the predominant genotype(clonal expansion, Figure 1).

Among sample conditions, the size is the most im-portant limiting factor leading to misinterpretations dueto tumor heterogeneity.3,7 Microdissection techniquesallow very selective and homogeneous cell samples,but the sampling might not be representative of thetumor. Firstly, small cell groups descended from acommon progenitor may grow together like a clone(patch size concept), which can explain monoclonalpatterns in small-sized samples. Secondly, samplecells must be representative of tumor features (eg,kinetic and invasive capacities). If clonality is not eval-uated in the proper biological context, the results mightbe confusing or have unknown clinical meaning. Atypical example of this situation is microheterogeneityin tumors that tend to give disparate results whosemeaning remains unknown. Only multiple samplings ofenough size (�100 cells) from different tumor areasand running tests in duplicate can avoid this problem;this protocol should be systematically done before ac-cepting the results as relevant.

The intratumoral heterogeneity and the heteroge-neous cell composition of solid tumors make the quan-titative determination of the allele ratio (proportion ofeach allele in samples normalized by the correspond-ing control tissue) an absolute requirement to proveany clonal origin or expansion. Allele ratios greaterthan 3:1 or 4:1 in normalized samples are consideredevidence of monoclonal proliferation.3– 8,12 This thresh-

American Journal of Pathology, Vol. 162, No. 1, January 2003

Copyright © American Society for Investigative Pathology

353

old means concordant allele patterns for a givenmarker in 75% to 80% of the sample cells. However,reliable results require keeping linearity of the alleleratio between the amplified product and the targetDNA of the sample. For that purpose, technical as-pects of the PCR amplification are essential. The reac-tion should be maintained in the exponential phaseavoiding the plateau (product saturation) and theamount of tissue in control and test samples should besimilar, thus correcting biased patterns due to artifactsinduced by small target concentrations in tumor sam-ples only.3 In addition, microdissected samples nor-mally show a high incidence of PCR artifacts due to thesmall concentration of target DNA, fixation-inducedchanges of DNA, and conditions in the amplification ofrepetitive sequence (especially for those CG-rich se-quences) favoring misannealing and hairpin forma-tion.13 Appropriate modifications must be establishedto avoid these problems, thus improving the reproduc-ibility of LOH and MSI test in microdissected samples.

In conclusion, a proper interpretation of clonality testsrequires a combined knowledge of the tumor naturalhistory and technical aspects. Using the spectrum ofclonality tests available, a reliable distinction of clonalorigin/expansion can be made considering the relativetiming between test conversion and clonal expansion, thelesion cell kinetic, sample conditions, and the conditionsfor the allele ratio determination.

Lucia Pozo-GarciaHomerton University HospitalLondon, United Kingdom

Salvador J. Diaz-CanoBarts and the London School ofMedicine and DentistryLondon, United Kingdom

References

1. Nakayama T, Taback B, Turner R, Morton DL, Hoon DS: Molecularclonality of in-transit melanoma metastasis. Am J Pathol 2001, 158:1371–1378

2. Diaz-Cano SJ: Clonality studies in the analysis of adrenal medullaryproliferations: application principles and limitations. Endocr Pathol1998, 9:301–316

3. Diaz-Cano SJ, Blanes A, Wolfe HJ: PCR techniques for clonalityassays. Diagn Mol Pathol 2001, 10:24–33

4. Diaz-Cano SJ, de Miguel M, Blanes A, Tashjian R, Wolfe HJ: GermlineRET 634 mutation positive Men 2A-related C-cell hyperplasias havegenetic features consistent with intraepithelial neoplasia. J Clin En-docrinol Metab 2001, 86:3948–3957

5. Diaz-Cano SJ, de Miguel M, Blanes A, Tashjian R, Galera H, Wolfe HJ:Clonality as expression of distinct cell kinetics patterns in nodularhyperplasias and adenomas of the adrenal cortex. Am J Pathol 2000,156:311–319

6. Diaz-Cano SJ, Blanes A, Rubio J, Matilla A, Wolfe HJ: Molecularevolution and intratumor heterogeneity by topographic compartmentsin muscle-invasive transitional cell carcinoma of the urinary bladder.Lab Invest 2000, 80:279–289

7. Diaz-Cano SJ: Designing a molecular analysis of clonality in tumors.J Pathol 2000, 191:343–344

8. Diaz-Cano SJ, de Miguel M, Blanes A, Tashjian R, Galera H, Wolfe HJ:Clonal patterns in phaechromocytomas and MEN-2A adrenal medul-lary hyperplasias: histologic and kinetic correlates. J Pathol 2000,192:221–228

9. Zhuang Z, Lininger RA, Man YG, Albuquerque A, Merino MJ, Tavas-soli FA: Identical clonality of both components of mammary carcino-sarcoma with differential loss of heterozygosity. Mod Pathol 1997,10:354–362

10. Diaz-Cano SJ, Blanes A: Influence of intratumor heterogeneity in theinterpretation of marker results in pheochromocytomas. J Pathol1999, 189:627–628

11. Blanes A, Rubio J, Martinez A, Wolfe HJ, Diaz-Cano SJ: Kineticprofiles by topographic compartments in muscle-invasive transitionalcell carcinomas of the bladder: role of TP53 and NF1 genes. Am JClin Pathol 2002, 118:93–100

12. Mutter GL, Boynton KA: X chromosome inactivation in the normalfemale genital tract: implications for identification of neoplasia. Can-cer Res 1995, 55:5080–5084

13. Diaz-Cano SJ: Are PCR artifacts in microdissected samples prevent-able? Hum Pathol 2001, 32:1415

Authors’ Reply:

We appreciate the comments from the authors’ letter andwould like to respond by stating that the general purposeof this manuscript was to evaluate the genetic heteroge-neity among in-transit melanoma metastasis assessingmicrosatellites with loss of heterozygosity (LOH), an es-tablished methodology.1 In-transit melanoma is a rarephenomenon often occurring after removal of a primarytumor cutaneous melanoma tumor. This disease manifes-tation is clinically evident from where the site of originexisted as opposed to head and neck cancers where theprimary may be unknown or Barrett’s esophagus whereLOH has been used for clonal origin and delimiting fieldcancerization with clonal expansion.2–4 Thus, our intentwas to assess those LOH events associated with in-transit metastasis with further interest to determine

Figure 1. Cell kinetics and genetic changes during the clonal evolution ofneoplasms. The first genetic abnormality during the neoplastic transforma-tion is assumed to induce a clonal proliferation, the hallmark of neoplasms(bottom, left-hand bar). However, any other genetic changes will detect thelesion after the marker conversion only (eg, microsatellite instability or LOH,X2 in the diagram). Tumor tissue heterogeneity complicates the detectionbecause results supportive of monoclonal proliferation (right cell field)will be obtained only if the abnormal cells (dark nuclei) are prevalent inthe sample (gray cytoplasm). The expansion of genetically damaged cells(clone selection) is always due to disbalanced kinetic (1 proliferationand/or 2 apoptosis).

354 CorrespondenceAJP January 2003, Vol. 162, No. 1

whether a pattern of heterogeneity existed for the panel ofmarkers assessed. Furthermore, intratumoral heteroge-neity and congruity with the primary tumor was evaluatedfor concordance.

Multiple polymorphic microsatellite markers were cho-sen for their informativity and frequency in melanomatumors. In addition, many tumors were assessed to allowfor sufficient number of LOH events to occur to avoid aconclusion of homogeneity based on retention alone orheterogeneity due to nonlinear random occurrences.5 Tofurther determine whether results were consistent, intra-tumor heterogeneity was assessed. However, to avoidfalse monoclonality interpretation due to inadequate sam-pling from “patch size” clones, three separate regionswere chosen which were widely spaced and randomlyselected.6 Additionally, a large enough sample was mi-crodissected from each specimen to ensure a sufficientnumber of cells and DNA quantitated for conformity. Fi-nally, primary tumor blocks were assessed in a similarfashion to confirm the findings with concordant patternsof the genetic markers assessed. The fact that consis-tency was demonstrated through all three aspects of theinvestigation confirms the reliability of the methodology inthis study and leaves little doubt that these results oc-curred randomly for this specific disease entity.

We agree that good laboratory practice under rigorousstandard operating procedures must be strictly adheredto for any laboratory that is assessing LOH. Optimalsample conditions and repetitive assessments should beroutine for accurately assessing allelic imbalances (AI).As the biology and relevance of these AI is still develop-ing, optimal standardization has not been consistent inthe literature. Assessment for clonality in tumor speci-mens using assays previously reported as the authorssuggest does have limitations in interpretation. With re-gard to assessing molecular markers in relation to atumor’s natural history, all of the lesions were of thein-transit type, which is a unique model for this field ofinvestigation. However, an inherent problem with anystudy evaluating patients’ tumors is the inability to collectall specimens at identical time points in the disease pro-gression spectrum. The accumulation of genetic alter-ations is a continuum for each individual tumor cell andthus one can never obtain sufficient number of speci-mens all at the same time point during neoplastic trans-formation and progression to make an absolute conclu-sion with certainty. We recognize the authors’ concernregarding timing between test conversion and clonal ex-pansion as well as lesion cell kinetics but this methodprovides the most clinically relevant approach for evalu-ating the unique pathology and biology of in-transit mel-anoma disease. In-transit recurrence is consistent withdormant tumor cell clones from the primary tumortrapped in intervening lymphatics and our findings pro-vide a genetic association for this clinical experience.

Bret TabackDave S.B. Hoon

John Wayne Cancer InstituteSanta Monica, California

References

1. Califano J, Leong PL, Koch WM, Eisenberger CF, Sidransky D,Westra WH: Second esophageal tumors in patients with head andneck squamous cell carcinoma: an assessment of clonal relation-ships. Clin Cancer Res 1999, 5:1862–1867

2. Galipeau PC, Prevo LJ, Sanchez CA, Longton GM, Reid BJ: Clonalexpansion and loss of heterozygosity at chromosomes 9p and 17p inpremalignant esophageal (Barrett’s) tissue. J Natl Cancer Inst 1999,91:2087–2095

3. Wong DJ, Paulson TG, Prevo LJ, Galipeau PC, Longton G, Blount PL,Reid BJ: p16(INK4a) lesions are common, early abnormalities thatundergo clonal expansion in Barrett’s metaplastic epithelium. CancerRes 2001, 61:8284–8289

4. Califano J, Westra WH, Koch W, Meininger G, Reed A, Yip L, BoyleJO, Lonardo F, Sidransky D: Unknown primary head and neck squa-mous cell carcinoma: molecular identification of the site of origin.J Natl Cancer Inst 1999, 91:599–604

5. Barrett MT, Sanchez CA, Prevo LJ, Wong DJ, Galipeau PC, PaulsonTG, Rabinovitch PS, Reid BJ: Evolution of neoplastic cell lineages inBarrett esophagus. Nat Genet 1999, 22:106–109

6. Diaz-Cano SJ, Blanes A, Wolfe HJ: PCR techniques for clonalityassays. Diagn Mol Pathol 2001, 10:24–33

Interleukin-3 Receptors in Hodgkin’s Disease

To the Editor-in-Chief:

Growing evidence suggests that deregulated apoptosisis a frequent occurrence in a variety of human malignan-cies.1 The tumor cells in classical Hodgkin’s disease(HD), historically named Hodgkin and Reed-Sternberg(HRS) cells, derive from germinal center B cells and oftencontain “crippling” somatic mutations within rearrangedimmunoglobulin (Ig) heavy chain genes.2 Because such“crippling” mutations trigger apoptosis in germinal centerB cells, their detection in HRS tumor cells indicate thepresence of survival factors other than surface Ig.

In the February 2002 issue of The American Journal ofPathology, Aldinucci et al3 reported on the expression ofinterleukin-3 receptors (IL-3R) in HRS cells. As the au-thors pointed out correctly, it is quite surprising that IL-3Rexpression has not yet been investigated in the context ofHD because of the ligand for this receptor, IL-3, alsocalled multicolony-stimulating factor, is probably one ofthe least restricted growth factors, exerting its effects onhemopoietic stem cells and progenitors of numerous lin-eages,4 including the lymphoid lineage, as shown by invitro differentiation of IL-3-dependent B-cell precursorsinto mature B cells.5 In addition to describing IL-3R ex-pression in HRS cells, Aldinucci et al3 examined a num-ber of HD-derived cell lines to address the functionality ofthese receptors. Among the HD cell lines tested, IL-3Rexpression levels showed a remarkable variability. How-ever, the increased growth rates of cultured HD cells onstimulation with IL-3 did not strictly reflect the differencesin IL-3R expression. The fact that L1236 cells, whichexpress low levels of IL-3R, had a stronger response thanany other HD cell line to exogenous IL-3, should remindus to interpret data obtained from cultured HD cells withgreat caution. The relatively minor growth response of the

Correspondence 355AJP January 2003, Vol. 162, No. 1

CORRESPONDENCE

Are PCR Artifacts in Microdissected SamplesPreventable?

To the Editor:—In many instances, molecular results aredetermined by technical conditions and need validation. Lossof heterozygosity (LOH) and microsatellite instability (MSI)analyses share these limitations too, but only a few articlesdirectly study the technical aspects of molecular tests.

Sieben et al address technical issues of LOH and MSIanalysis using small amounts of DNA (microdissected tumorsamples).1 The study design is appropriate for the objectives,but the technique implications are quite extensive and abroader discussion on biological and technical aspects wouldhelp both interpreting and applying those analyses. The au-thors clearly show the inverse relationship between the con-centration of the target DNA and the presence of test arti-facts, but can those artifacts be explained by the DNAconcentration only? If that is the case, are the molecularresults based on single-cell analysis believable? Some thoughtstrying to answer these questions are provided.

The identification of extrabands in the amplificationproduct from a polymorphic DNA region will be consideredevidence of MSI if they are not present in the correspondingcontrol.2-5 To avoid misinterpretations, the amount of tissuein control and test samples should be similar, thus correctingbiased patterns caused by artifacts induced by small targetconcentrations in tumor samples only.5 Those additionalbands must be carefully distinguished from polymerase chainreaction (PCR) artifacts and different options must be inves-tigated and systematically excluded:

1. Target DNA modifications induced by fixation andprocessing in paraffin-embedded tissues. The fixa-tion process involves multiple cross-links betweenamino-groups of nucleic acid bases and polypeptideaminoacids to preserve tissue morphology.6 A sideeffect of the tissue processing is a variable nucleicacid fragmentation and denaturation that result inthe typical smear pattern of DNA extracted fromparaffin-embedded tissues. The covalent links be-tween different DNA fragments will determine thepresence of single-stranded sequences with 3�-OHfree that can act as primers during the polymeriza-tion step of PCR, the so-called primer-independentDNA amplification.6 A prolonged protein digestionin the appropriate conditions during the DNA ex-traction significantly reduces the number of cross-links, thus resulting in better definition of the am-plification product.

2. The repetitive sequence would result in a higherincidence of primer misannealing and hairpin for-mation. These possibilities can be avoided setting along denaturation in the first few cycles when thetemplate DNA is mainly genomic and including7-deaza-2�-dGTP in the reaction mixture if the targetis a CG-rich sequence. This nucleotide substitutionduring amplification reduces stability of intramolec-ular and intermolecular GC base pairing and avoidsbiased target amplification.7 These 2 technical mod-ifications favor proper annealing and amplificationof the specific target.7-9

3. The amplification conditions can also determine the

specificity of the reaction. Both nucleotide concen-tration and the labeling methods have been reportedlimiting factors.2,5 The probability of getting addi-tional bands decreases when nucleotide concentra-tion is reduced and an external labeling method(only one primer labeled) is used.

The presence of minute DNA amount in the reactionmixture results in more frequent PCR artifacts, especially ifthe target DNA is fragmented. In microsatellite analysis,Sieben et al report a significant increase of these PCR artifactswhen the amount of target DNA is lower than 5 ng (DNAisolated from frozen tissues) or 10 ng (DNA isolated fromformalin-fixed, paraffin-embedded tissues).1 Considering theDNA amount per cell (�7 pg), the limiting DNA amountwould represent between �700 cell equivalents (frozen tis-sues) and �1,400 cell equivalents (formalin-fixed, paraffin-embedded tissues). This issue is extremely important formicrodissected samples because they normally contain lessDNA. However, the relative incidence of these artifacts can becontrolled by applying the previously described principles,which bypass the most frequent causes of these artifacts. Thisalso emphasizes the importance of careful method design toobtain reliable results in molecular analyses.

The high incidence of PCR artifacts using microdissectedsamples is related to the small concentration of target DNA,fixation-induced changes of DNA, and conditions in the am-plification of repetitive sequence (especially for those CG-richsequences) favoring misannealing and hairpin formation. Ap-propriate modifications to avoid the previously mentionedconditions will significantly improve the reproducibility ofLOH and MSI test in microdissected samples.

SALVADOR J. DIAZ-CANO, MD, PHDDepartment of HistopathologyBart’s and The London Queen Mary’s

School of Medicine and DentistryUniversity of LondonLondon, England

1. Sieben NL, ter Haar NT, Cornelisse CJ, et al: PCR artifacts in LOH andMSI analysis of microdissected tumor cells. HUM PATHOL 31:1414-1419, 2000

2. Koreth J, O’Leary JJ, McGee JOD: Microsatellites and PCR genomicanalysis. J Pathol 178:239-248, 1996

3. Diaz-Cano SJ, Blanes A, Rubio J, et al: Molecular evolution and intra-tumor heterogeneity by topographic compartments in muscle-invasive transi-tional cell carcinoma of the urinary bladder. Lab Invest 80:279-289, 2000

4. Diaz-Cano SJ: Designing a molecular analysis of clonality in tumours.J Pathol 191:343-344, 2000

5. Diaz-Cano SJ, Blanes A, Wolfe HJ: PCR techniques for clonality assays.Diagn Mol Pathol 10:24-33, 2001

6. Diaz-Cano SJ, Brady SP: DNA extraction from formalin-fixed, paraffin-embedded tissues: Protein digestion as a limiting step for retrieval of high-quality DNA. Diagn Mol Pathol 6:342-346, 1997

7. Mutter GL, Boynton KA: PCR bias in amplification of androgen recep-tor alleles, a trinucleotide repeat marker used in clonality studies. Nucleic AcidsRes 23:1411-1418, 1995

8. Diaz-Cano SJ, de Miguel M, Blanes A, et al: Clonality as expression ofdistinctive cell kinetics patterns in nodular hyperplasias and adenomas of theadrenal cortex. Am J Pathol 156:311-319, 2000

9. Diaz-Cano SJ, de Miguel M, Blanes A, et al: Clonal patterns in phaechro-mocytomas and MEN-2A adrenal medullary hyperplasias: Histologic and ki-netic correlates. J Pathol 192:221-228, 2000

doi:10.1053/hupa.2001.29632

CORRESPONDENCE

1415

Reply

To the Editor:—We would like to thank Dr Diaz-Cano forhis comments on our recent article describing polymerasechain reaction (PCR) artifacts in loss of heterozygosity (LOH)and microsatellite instability (MSI) analysis of microdissectedtumor cells. It is becoming increasingly apparent that signif-icant problems may be encountered during attempts to un-ambiguously reproduce LOH and MSI data based on DNAobtained from microdissected tumor tissue. Although morethan one factor may play a role, the concentrations of tem-plate DNA used are seldom accurately quantitated and mayvary greatly between the individual tumor samples and corre-sponding normal tissue. To identify the necessary parametersfor facilitating reliable PCR-based analyses, we performedPCR with increasing dilutions of template DNA in nearly1,000 PCR reactions. By doing so, we showed a significantassociation between the concentration input DNA used andthe percentage of artifactual LOH or MSI observed. Based onthis study, a minimum of 5.0 ng for fresh frozen tissue and10.0 ng for formalin-fixed, paraffin-embedded tissue is set forreliable PCR analysis under our experimental conditions.

We agree with Dr Diaz-Cano that other factors such astissue-fixation method and processing influences DNA qualityand thus optimal PCR amplification. For example, tissuefixation in ethanol is less deleterious for DNA than formalinand less input DNA is necessary to obtain reproducible re-sults. We can fully subscribe the importance of optimal testconditions in microsatellite analysis, one of which should beDNA quantification and input standardization, and we do notseem to deviate from Dr Diaz-Cano’s conclusions.

NATHALIE L. SIEBEN, MDANNE-MARIE CLETON-JANSEN, PHDDepartment of PathologyLeiden University Medical CenterLeiden, The Netherlands

doi:10.1053/hupa.2001.29633

Tumor Invasion and Metastasis—Nature orNurture?

To the Editor:—In the article by Matias-Guiu,1 it is statedthat “Once the tumor has developed, several additional mo-lecular abnormalities occur in different neoplastic subclones;these new alterations are responsible for tumor heterogene-ity, tumor invasion, and metastasis.” In the accompanying edi-torial,2 it is stated that “Tumorigenesis is conceived as involv-ing a stepwise accumulation of genetic damage within a cellthat ultimately undergoes malignant transformation and, intime, the proliferation of clones having such phenotypicproperties as drug- or hormone-resistance, invasion, or metas-tasis.” Also in the editorial it is stated that “Additional molec-ular changes occurring in different neoplastic subclones areresponsible for the clinically apparent phenotypes of tumorinvasion and metastasis.”

Strongly implied but without proof is that some geneticalteration(s) transforms a dividing cell into a cell that candetach itself from its neighboring tumor cells, travel from onelocation to another, and enter into veins and/or lymphatics(invasion). Further, the genetic alteration(s) of the dividingcell transforms the cell into one that can now find nourish-ment, continue to divide, and avoid being destroyed by thebody’s immune defenses (metastasis) at some distant loca-tion.

There is nothing in the article by Matias-Guiu, nor in theeditorial, that would support any theories other than thatgenetic mutations occur and occur sequentially with time.The functional change brought about by any given alterationin DNA is speculative at best, when considering invasion andmetastasis. That such genetic events lead to a change in thecell’s ability to divide or not to divide seems to be an accept-able conclusion from what is known of cell division mecha-nisms. However, to conclude that genetic events are respon-sible for invasion and metastasis is premature.3

Elsewhere,3 I have reintroduced an old concept that willaccount for invasion and distant transport of tumors invokingonly the normal physiologic and mechanical processes oftraumatic separation of tumor cells from anchoring neigh-bors, traumatic disruption of basement membrane material,lymphatic flow dynamics, and the anatomic introduction oflymph-containing tumor cells and debris into the venoussystem to widely disseminate tumor cells throughout thebody. The mechanism of establishing a stable and/or growingmetastasis may as likely reside in the host defenses as in thetumor cell’s DNA.

Increasing numbers of genetic alterations may be impor-tant to increasing cell division and decreasing cell destruction(apoptosis), but there is no inherent biologic necessity toinvoke genetic events into the invasion and metastasis pro-cess.

ROBERT J. ROSSER, MDPathology DepartmentDesert Regional Medical CenterPalm Springs, CA

1. Matias-Guiu X, Catasus L, Bussaglia E, et al: Molecular pathology ofendometrial hyperplasia and carcinoma. HUM PATHOL 32:569-577, 2001

2. Foster CS, Gorstein F: Molecular basis of disease processes. HUM PATHOL

32:567-568, 2001 (editorial)3. Rosser RJ: A point of view: Trauma is the cause of occult micrometa-

static breast cancer in sentinel axillary lymph nodes. The Breast Journal 6:209-212, 2000

doi:10.1053/hupa.2001.29636

Reply

To the Editor:—We would like to express our appreciationto Dr Robert J Rosser for his interest in our review article,“Molecular Pathology of Endometrial Hyperplasia and Carci-noma.” There is a large body of evidence in the medicalliterature suggesting that some genetic alterations are associ-ated with the development and progression of human carci-nomas. Our article summarizes the results obtained by severalauthors, including ourselves, in understanding the role ofcertain genetic abnormalities in the appropriate morphologiccontext. Likewise, it is understood that the process of tumorprogression includes many other aspects such as host re-sponse and functional disruption of the relation betweenepithelial cells and their environment.

XAVIER MATIAS-GUIU, MDJAIME PRAT, MD, FRCPATH

Department of PathologyHospital de la Santa Creu i Sant PauBarcelona, Spain

doi:10.1053/hupa.2001.29637

HUMAN PATHOLOGY Volume 32, No. 12 (December 2001)

1416

Editors’ Reply

The editorial that accompanied the article by Matius-Guiu et al1 in the June issue of HUMAN PATHOLOGY was writtento challenge several currently held concepts and to highlighta series of issues presently being raised by a wide range ofdifferent molecular biological studies. Intentionally, the edi-torial was not restricted to a consideration of only thoseconcepts that were raised in the article concerning the mo-lecular pathology of endometrial hyperplasia and carcinoma.It is both unfortunate and unintentional if there was anyapparent ambiguity in the content of the editorial such that itseemed to imply that specific genetic alterations are requiredto transform a cell into one that can become metastatic, asindicated in the accompanying letter by Dr Rosser. If therehas been such an implication or interpretation, then weapologize that the sense of the editorial was not conveyed withgreater clarity.

Rather than being at variance, we agree with Dr Rosserthat there is no evidence for the existence of specific metastasisgenes or for specific genetic alterations that might be con-strued as metastasis genes. Although current evidence pointsto the initiation of neoplasia involving specific genetic events,the same is not true of the metastatic process. One difficultynow facing cancer researchers is the separation of thosemolecular and cellular events that are causal to the metastaticprocess from the myriad of epiphenomena that occur duringthe evolution and progression of each malignancy. Thus, weconcur with Dr Rosser that neither the article by Matius-Guiuet al nor our editorial supports any theories other than thetemporal occurrence of genetic mutations with tumor ad-vancement.

In his letter, Dr Rosser raises an important problem withrespect to the biology of the metastatic process that may beaddressed, in part, by consideration of events that occur innonmalignant tissues. During embryogenesis and tissue mor-phogenesis, the phenotypic properties of cellular diversion,migration, and invasion are both common and are character-istic attributes of benign cells. Even during postembryonicmaturation, some cell types (eg, interdigitating dendritic cellswithin the skin2 and normal B-lymphocyte subsets within theileum and salivary glands3) undergo migration, differentia-tion, homing, and tissue invasion. These observations relatedto mesenchymal cells within the reticuloendothelial systemmight be considered to be intrinsically migratory. However,recent independent observations have shown that cells ofadult hemopoietic origin can invade the human liver with theresult that they repopulate the epithelial compartment, ini-tially as hepatocytes.4 Further, migration and differentiationof these cells also result in the generation of cholangiocytes.5With respect to epithelial malignancy, additional evidence isaccumulating that several proteins essential to the metastatic

process are the normal products of nonmutated genes—butaberrently expressed or regulated by the metastatic cells.Amongst others that are now being described, these includethe Ca��-binding protein p9Ka,6 voltage-gated Na� chan-nels,7 C-FABP,8 and fascin,9 a molecule essential to the mi-gration and differentiation of interdigitating dendritic cellswithin the skin.

In contrast to oncogenic genes that are mutated and en-code abnormal products, evidence currently accumulatingsuggests that metastasis-promoting genes are nonmutated andyield normal proteins. Furthermore, there is no evidence toindicate that activity of metastasis-promoting proteins areobligated to any of the mutational events that are character-istic of or accompany tumor progression. Elucidation of thisrelationship is likely to provide key information concerningthe biological nature of the metastatic process.

CHRISTOPHER S. FOSTER, MD, PHD, FRCPATH

Editor, European Editorial OfficeUniversity of LiverpoolLiverpool, England

FRED GORSTEIN, MDEditorThomas Jefferson UniversityPhiladelphia, PA

1. Matias-Guiu X, Catasus L, Bussaglia E, et al: Molecular pathology ofendometrial hyperplasia and carcinoma. HUM PATHOL 32:569-577, 2001

2. Caux C, Ait-Yahia S, Chemin K, et al: Dendritic cell biology and regu-lation of dendritic cell trafficking by chemokines. Springer Semin Immuno-pathol 22:345-369, 2000

3. Woodruff JJ, Katz M, Lucas LE, et al: An in vitro model of lymphocytehoming II. Membrane and cytoplasmic events involved in lymphocyte adher-ence to specialized high-endothelial venules of lymph nodes. J Immunol 119:1603-1610, 1977

4. Alison MR, Poulsom R, Jeffery R, et al: Hepatocytes from non-hepaticadult stem cells. Nature 406:257, 2000

5. Theise ND, Nimmakayalu M, Gardner R, et al: Liver from bone marrowin humans. Hepatology 32:11-16, 2000

6. Ke Y, Jing C, Barraclough R, et al: Elevated expression of calcium-binding protein p9Ka is associated with increasing malignant characteristics ofrat prostate carcinoma cells. Int J Cancer 71:832-837, 1997

7. Smith P, Rhodes NP, Shortland AP, et al: Sodium channel proteinexpression enhabces the invasiveness of rat and human prostate cancer cells.FEBS Letters 423:19-24, 1998

8. Jing C, Beesley C, Foster CS, et al: Human cutaneous fatty acid-bindingprotein induces metastasis by up-regulating expression of vascular endothelialgrowth factor gene in rat RAMA 37 model cells. Cancer Res 61:4357-4364, 2001

9. Yamashiro S, Yamakita Y, Ono S, et al: Fascin, an actin-bundling protein,induces membrane protrusions and increases cell motility of epithelial cells.Mol Cell Biol 9:993-1006, 1998

doi:10.1053/hupa.2001.29638

CORRESPONDENCE

1417

J. Pathol. 189: 627–629 (1999)

LETTERS TO THE EDITOR

INFLUENCE OF INTRATUMOUR HETEROGENEITY IN THEINTERPRETATION OF MARKER RESULTS IN

PHAEOCHROMOCYTOMAS

In a recently published paper, Krijger et al.1 try topredict the clinical behaviour of a phaeochromocytomas(PCCs) on the basis of the immunoreactivity of threemarkers (p53, bcl-2, and c-erbB-2). They conclude thatthe co-expression of p53 and bcl-2 proteins may assistthat prediction. Our own preliminary results2,3 mainlyagree with theirs and point towards two interrelatedbiological issues: firstly, the clonal evolution of neo-plasms and intratumour heterogeneity of markers;and secondly, the importance of apoptosis in tumourinitiation and progression.

The results of Krijger et al. and other previousfindings indicate that several genes must be involved inPCC pathogenesis, regardless of the genetic background(sporadic or familial).4,5 Although the authors do notmention variability of the immunostaining, this mustbe assumed, because most PCCs are scored 1+ (10–50per cent positive cells). These relatively low values mustbe associated with heterogeneous expression of theprotein. Our preliminary results with p53 and pRB(both immuno-expression and the genetic evaluation ofpolymorphic DNA regions) support this point.

Tumour cell selection will determine progression,cellular heterogeneity, and clonal expansion.6 Loss ofheterozygosity (LOH) analyses have showed randomand non-tumour-related DNA deletions in 4–20 per centof normal tissues,7,8 confirming cellular heterogenity,which should be considered a limiting threshold in theinterpretation of any tumour marker. Intratumourheterogeneity for a given marker can represent either theexpression of selective tumour evolution, or a simplepassive by-product of other mechanisms, such as geneticinstability. In any case, the association of multiplegenetic alterations would become statistically less prob-able as the number of molecular markers increased9 andwould explain why the combination of p53 and bcl-2was strongly correlated with malignant PCC.1 Themechanism of selection of cell clones would allow us touse genetic markers to define tumour progression,6 butthese genetic changes are unpredictable and their hetero-geneity precludes their extensive clinical use for diagnos-tic and prognostic purposes. This heterogeneity alsocalls for caution in evaluating markers of malignancy;they must be extensively screened and validated in orderto avoid misinterpretations leading to false-positive andfalse-negative cases.

The results of Krijger et al. also stress the importanceof p53 in the pathogenesis of a subgroup of PCCs.1 Ourown experimental results (manuscript in preparation)

CCC 0022–3417/99/130627–03$17.50Copyright � 1999 John Wiley & Sons, Ltd.

support this point of view, based on the analysis of fivepolymorphic DNA regions (microsatellites) located onintrons of four tumour suppressor genes (p53, RB1,WT1, and NF1). The comparative study of peripheraland internal tumour areas confirmed that there was anincreased accumulation of genetic deletions in theperipheral tumour compartment, probably reflectingboth tumour progression and multistep tumourigenesis(�two loci showed LOH).

In addition, tumour cell selection is the expressionof cellular kinetics. Any genetic alteration leading tocellular ageing, differentiation, or activation of theapoptotic pathway will be non-tumour productive.6 Thepresence of a given marker points to an extensive kineticadvantage provided by the marker itself, or linked to it;it also represents the basic mechanism of cell selectionand tumour progression.10,11 In this framework, Krijgeret al.1 report the expression of bcl-2 in malignantPCC. bcl-2-expressing cells would be protected fromapoptosis, thus contributing to clonal expansion, andthe absence of normal p53 would protect geneticallydamaged cells from apoptosis.12 The associated expres-sion of abnormal p53 in malignant PCC would alsomaintain the proliferation of transformed cells.1

Krijger et al. clearly indicate the importance of p53and bcl-2 in the pathogenesis of a subgroup of PCCs.However, the variability of marker expression andthe potential relationship with kinetic features are alsohelpful in understanding the pathogenesis of PCC.

S J. Dí-C1 A B2

1Department of Histopathology and Morbid Anatomy,St Bartholomew’s and the Royal London School of Medicine and

Dentistry, Whitechapel, London E1 1BB, U.K.2Department of Pathology, University Hospital,

Campus Universitario Teatinos, s/n, 29071-Malaga, Spain

REFERENCES

1. Krijger RR, van der Harst E, van der Ham F, et al. Prognostic value of p53,bcl-2, and c-erbB-2 protein expression in phaeochromocytomas. J Pathol1999; 188: 51–55.

2. Dı́az-Cano SJ, de Miguel M, Galera-Davidson H, Wolfe HJ. Are locallyinvasive pheochromocytomas biologically distinct from benign chromaffinneoplasms? Pathol Int 1996; 46 (Suppl 1): 223.

3. Dı́az-Cano SJ, Tashjian R, de Miguel M, et al. Distinctive clonal andhistological patterns in locally invasive pheochromocytomas. Lab Invest1997; 76: 153A.

4. Khosla S, Patel VM, Hay ID, et al. Loss of heterozygosity suggests multiplegenetic alterations in pheochromocytomas and medullary thyroidcarcinomas. J Clin Invest 1991; 87: 1691–1699.

5. Shin E, Fujita S, Takami K, et al. Deletion mapping of chromosome 1p and22q in pheochromocytoma. Jpn J Cancer Res 1993; 84: 402–408.

628 LETTERS TO THE EDITOR

6. Dı́az-Cano SJ. Clonality studies in the analysis of the adrenal medullaryproliferations: application principles and limitations. Endocr Pathol 1998; 9:301–316.

7. Deng G, Lu Y, Zlotnikov G, Thor AD, Smith HS. Loss of heterozygosity innormal tissue adjacent to breast carcinomas. Science 1996; 274: 2057–2059.

8. Wolman SR, Heppner GH. Genetic heterogeneity in breast cancer. J NatlCancer Inst 1992; 84: 469–470.

AUTHORS

We would like to thank Dr Dı́az-Cano and Dr Blanesfor their comments on our paper,1 which they seem touse as an alibi for extensive reflections on tumourheterogeneity.

Human tumours are frequently genetically hetero-genous, as has recently been discussed in an excellentreview article.2 The differential acquisition of geneticabnormalities will lead to the formation of tumour cellclones with selective growth advantage, which will deter-mine tumour behaviour and thus patient prognosis.Research into the pathogenesis of several major cancertypes has benefitted from the comparative analysis oflesions in various stages of development, which hasled to the discovery of genetic abnormalities that areessential and/or specific for the development of particu-lar tumours. As regards PCCs, however, precursorlesions have not (yet) been defined. Potentially, adrenalmedullary hyperplasia may serve as some kind of pre-cursor lesion, although the difference with PCC is onlyquantitative, not qualitative.

As is the case in other human cancers, PCCs have(considerable) genetic heterogeneity. In previous work,we have shown that 4/27 (15 per cent) benign and 1/29 (3per cent) malignant sporadic PCCs have somatic RETmissense mutations, thus implying that there are othergenes involved in the pathogenesis of PCCs.3 At present,we are addressing VHL somatic mutations in benign andmalignant PCCs.

In our paper, we reported on p53 and bcl-2 immuno-reactivity. As Dı́az-Cano and Blanes noticed, thenumber of p53- and bcl-2-immunoreactive cells variedfrom one tumour to another, amounting to between 10

RE. SV40-LIKE DNA SEQUENCESBRONCHOPULMONARY CARCIN

PULMONARY

We read with interest the paper of Galateau-Salleet al. describing the presence of SV40 or SV40-like viralDNA in neoplastic, non-neoplastic mesothelium,and bronchial carcinoma.1 It is indeed a recognizedphenomenon that some SV40 primer sets result in ahigher percentage of positive PCR than others.2Preliminary data suggest that PCR primers consideredspecific for SV40 would, under certain conditions,amplify what appeared to be host DNA sequences,

Copyright � 1999 John Wiley & Sons, Ltd.

9. Smith HS. Stochastic model for interpreting the data on loss of hetero-zygosity in breast cancer. J Natl Cancer Inst 1990; 82: 793–794.

10. Nowell PC. The clonal evaluation of tumor cell populations. Science 1976;194: 23–28.

11. Hellman S. Darwin’s clinical relevance. Cancer 1997; 79: 2275–2281.12. Cordon-Cardo C. Mutations of cell cycle regulators. Biological and clinical

implications for human neoplasia. Am J Pathol 1995; 147: 545–560.

’ REPLY

and 50 per cent of the tumour cells. The immunoreactivecells had a somewhat patchy distribution in severalcases, which might be taken to reflect the presence ofdistinct cell clones. However, without microdissectionand clonal anlaysis, this is entirely speculative. Dı́az-Cano and Blanes point out that potential markers foradverse tumour behaviour must be extensively evaluatedand validated. Our agreement with this comment ismanifest in the discussion, where we have explicitlystated the practical implications of our analysis.

R R. K1, E H2,F H1, T S3,

W N. M. D1, J-W K4,H A. B2, S W. J. L4

F T. B1

1Department of Pathology, Erasmus University andUniversity Hospital, Rotterdam, The Netherlands2Department of Surgery, Erasmus University andUniversity Hospital, Rotterdam, The Netherlands

3Department of Biostatistics, Erasmus University andUniversity Hospital, Rotterdam, The Netherlands

4Department of Internal Medicine, Erasmus University andUniversity Hospital, Rotterdam, The Netherlands

REFERENCES

1. Krijger RR de, Harst E van der, et al. Prognostic value of p53 bcl-2, andc-erbB-2 protein expression in phaeochromocytomas. J Pathol 1999; 188:51–55.

2. Lenguaer C, Kinzler KW, Vogelstein B. Genetic instabilities in humancancers. Nature 1998; 396: 643–649.

3. Harst E van der, Krijger RR de, Bruining HA, et al. Prognostic value ofRET proto-onocogene point mutations in malignant and benign, sporadicpheochromocytomas. Int J Cancer 1998; 79: 537–540.

IN PLEURAL MESOTHELIOMA,OMA AND NON-MALIGNANT

DISEASE

indicating that PCR-based assays must be interpretedwith caution and correlated with viral antigenexpression.3

The authors used one monoclonal antibody, Pab 419(Ab-1 Oncogene Science), to investigate the presence ofviral SV40 antigens, as confirmation for their PCR-based assay. This antibody detects an epitope commonfor both the p94 large T- and the p21 small t-antigen. Itsmain applications have been described for immuno-

J. Pathol. 189: 627–629 (1999)

04/14/2006 02:32 PMNEJM -- Clonality in Kaposi's Sarcoma

Page 1 of 6http://content.nejm.org/cgi/content/full/337/8/570?andorexacttitle…96&fmonth=Apr&sendit=GO&searchid=1&FIRSTINDEX=0&resourcetype=HWCIT

Return to Search Result

Add to Personal ArchiveAdd to Citation ManagerNotify a FriendE-mail When Cited

Related Articleby Rabkin, C. S.

Find Similar ArticlesPubMed Citation

HOME | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | HELP

You are signed in as sjdiazcano at Subscriber level | Sign Out | Edit Your Information | CiteTrack Personal Alerts | Personal Archive

Previous Volume 337:570-572 August 21, 1997 Number 8 Next

Clonality in Kaposi's SarcomaTo the Editor: Rabkin et al. (April 3 issue)1 studied multiplebiopsy specimens from eight patients with Kaposi's sarcomaand claim that Kaposi's sarcoma begins as a clonal disease atsome specific site, circulates in the blood, and implants itself atmultiple sites in the skin to produce multicentric disease. Theunbalanced methylation pattern at the androgen-receptor locusin Kaposi's sarcoma lesions provides the evidence for their view.

We are concerned about some of the authors' technicalprocedures. They do not provide results of amplification by thepolymerase chain reaction with androgen-receptor primers andundigested DNA, to ensure that the observed pattern is trulydue to allelic methylation. Moreover, the HpaII restrictiondigestion was performed with heat-denatured samples, in which incorrect re-annealingcan lose the original HpaII site or generate additional sites. The use of proteinaseinhibitors, such as phenyl methyl sulfonyl fluoride, after digestion with proteinase K cancircumvent this problem. The authors do not give the age of the lesions before biopsy.Assuming that Kaposi's sarcoma begins as polyclonal disease that then evolves to a clonaldisease, lesions that have been present for many months are more likely to have clonalpatterns.

We have studied tumor tissue from 11 women with Kaposi's sarcoma, all of whom werepolymorphic for CAG repeats in exon 1 of the androgen receptor (Table 1). Tumorregions were microdissected and analyzed by a method similar to that of Rabkin et al.1 Insome cases more than one discrete tumor region from the biopsy specimen was assayed.Tumor-biopsy specimens from five of the women had evidence of clonality. Two women,



from whom one and nine specimens were obtained, had evidence of more than one clone(Table 2). In two of the five women with clonal disease, there were areas of the tumorthat were not clonal. We think that Kaposi's sarcoma is clonal in some cases but thatindependent clones may develop, either spontaneously or after undergoing apremalignant hyperplastic stage. To investigate these possibilities, a large study isneeded in which the age of cutaneous lesions is carefully recorded and specimensobtained from visceral tumors are included.

View this table:[in this window]

[in a new window]

Table 1. Methylation Patterns of the Tumor-Biopsy SpecimensStudied.

View this table:[in this window]

[in a new window]

Table 2. Two Patients with Discordance of the Methylated Allele,Indicating the Presence of at Least Two Clones.

Parkash Gill, M.D. Yvonne Tsai, Ph.D. Adupa P. Rao, M.D. Peter Jones, Ph.D. University of Southern California School of MedicineLos Angeles, CA 90033

References

. 1 Rabkin CS, Janz S, Lash A, et al. Monoclonal origin of multicentric Kaposi's sarcomalesions. N Engl J Med 1997;336:988-993.[Abstract/Full Text]

To the Editor: Rabkin et al. readdress the topic of the clonal origin of neoplasms thatfrequently present as multiple lesions, such as Kaposi's sarcoma in patients with humanimmunodeficiency virus infection.1,2 This issue has been raised before with regard toother synchronic tumors, such as transitional-cell neoplasms3 and head and neckcancers.4 Most of the information has been obtained by techniques that involve therandom inactivation of one X chromosome in female subjects, as exemplified in thearticle by Rabkin et al.1 This issue cannot be resolved in a straightforward fashion,especially by a technique in which each cell has only two possibilities for X-chromosomeinactivation: inactivation of the X chromosome inherited from the father or of that



inherited from the mother.

The authors used a statistical analysis that showed the low likelihood that their findingswould occur by chance. Their assumption of concordant patterns of allele methylation iscorrect with respect to cell–cell comparisons, but their findings are based on DNAextracted from many cells after the microdissection of tumor nodules. Under thesecircumstances, each tissue sample from a female subject with informative alleles can bepolyclonal, monoclonal with preferential methylation of the larger allele, or monoclonalwith the smaller allele predominating. Assuming that there is an equal probability of eachof these three patterns, the likelihood of a concordant pattern would be 2/3n (where 2 isthe number of alleles and n the number of tumors being compared); that is, 2/32 (orapproximately 22 percent) for two tumors, 2/33 (approximately 7.4 percent) for threetumors, 2/34 (approximately 2.5 percent) for four tumors, and 2/35 (approximately 0.8percent) for five tumors. Under these circumstances, the probability that different tumorsfrom the same patient have a monoclonal origin is greater than the authors calculate.

We suggest caution in the interpretation of these calculations because of the skewing ofX-chromosome methylation in different tissues. Synchronic or metachronic tumors canbe shown to be truly monoclonal (derived from the same clone) only if several molecularmarkers are concordant. An analysis of X-chromosome inactivation cannot distinguishmetastatic tumors (arising from a single clone) from synchronic tumors arising fromdifferent clones but expressing the same inactivated X chromosome. The results obtainedby analyzing a single molecular marker prove clonal expansion of a subgroup of tumorcells with proliferative advantages, but they do not prove actual clonality, especially whenthe assay is based on patterns of methylation that may be influenced by the functionalstatus of the cell, tumor progression, or both.5

Salvador J. Diaz-Cano, M.D., Ph.D. Hubert J. Wolfe, M.D. New England Medical Center HospitalsBoston, MA 02111

References

. 1 Rabkin CS, Janz S, Lash A, et al. Monoclonal origin of multicentric Kaposi's sarcomalesions. N Engl J Med 1997;336:988-993.[Abstract/Full Text]

. 2 Rabkin CS, Bedi G, Musaba E, et al. AIDS-related Kaposi's sarcoma is a monoclonalneoplasm. Clin Cancer Res 1995;1:257-260.[Abstract]

. 3 Jones PA, Droller MJ. Pathways of development and progression in bladder cancer:new correlations between clinical observations and molecular mechanisms. SeminUrol 1993;11:177-192.[Medline]



. 4 Califano J, van der Riet P, Westra W, et al. Genetic progression model for head andneck cancer: implications for field cancerization. Cancer Res 1996;56:2488-2492.[Abstract]

. 5 Laird PW, Jaenisch R. DNA methylation and cancer. Hum Mol Genet 1994;3:1487-1495.[Abstract]

The authors reply:

To the Editor: Gill et al. report balanced methylation patterns in 26 of 40 microdissectedspecimens of Kaposi's sarcoma (65 percent). They correctly note that balancedmethylation could reflect mixed clonality of tumor or stromal cells. However, only 4 ofthe 32 Kaposi's sarcoma tumors we studied (12 percent) had balanced methylation.Although the difference could be due to variation in the frequency of polyclonal disease,it could also be explained by varying degrees of success in obtaining sufficiently purepreparations of tumor cells.

The two patients described by Gill et al. who had discordant methylated alleles couldrepresent cases in which there are multiple primary lesions, but we found no discordancein a larger number of clonal samples. The technical issues they raise do not explain thediscrepancy. We observed balanced HUMARA amplification with undigested tumor DNAand balanced HpaII restriction after inactivation of normal-skin DNA by heat. Ifmulticentric Kaposi's sarcoma lesions evolve independently, we should also have seencases in which there was discordance of the methylated allele. Our model of a circulatingneoplastic progenitor is suggested by the statistical improbability that our resultsoccurred by chance alone.

Drs. Diaz-Cano and Wolfe correctly note that Kaposi's sarcoma tissue from a femalesubject with informative alleles can reveal either a polyclonal or a monoclonal pattern ofHUMARA allele methylation. However, we had no basis for assigning relative probabilitiesto these alternatives in our statistical analysis. Instead, we calculated the conditionalprobability of concordant allelic methylation given a monoclonal pattern. We disagreewith the writers' interpretation that polyclonal and monoclonal patterns in differenttumors indicate that the tumors have different clonal origins, since an admixture ofstromal and tumor DNA may appear to be polyclonal. Nevertheless, our Patients 1, 4, and5 each had monoclonal patterns in all the tumors that could be evaluated. Under theassumptions proposed by Drs. Diaz-Cano and Wolfe, the combined probability that thisresult would occur by chance is less than 0.00001 (2/35 x 2/35 x 2/33). Thus, even theirassumptions lead to a conclusion of monoclonality, at least in some patients.

We appreciate the potential problem of skewed methylation of normal tissue precursors.1



Return to Search Result

Add to Personal ArchiveAdd to Citation ManagerNotify a FriendE-mail When Cited

Normal skin from six of the eight patients we studied had balanced methylation patterns,but it is uncertain what is the most appropriate normal control for Kaposi's sarcoma.However, changes in DNA methylation that occur during tumorigenesis 2 cannot explainskewing of X-chromosome methylation in the absence of either clonal expansion or X-linked differences in the cellular propensity to Kaposi's sarcoma (for which there is noother evidence). We agree that detecting additional genetic changes would support theevidence of clonality provided by the HUMARA assay, which has contributed to thecurrent understanding of histiocytosis X,3 desmoid fibromatosis,4 and Rosai–Dorfmandisease.5

Charles S. Rabkin, M.D. Siegfried Janz, M.D. Zhengping Zhuang, M.D., Ph.D. National Cancer InstituteRockville, MD 20892

References

. 1 Gale RE, Wheadon H, Boulos P, Linch DC. Tissue specificity of X-chromosomeinactivation patterns. Blood 1994;83:2899-2905.[Abstract/Full Text]

. 2 Laird PW, Jaenisch R. DNA methylation and cancer. Hum Mol Genet 1994;3:1487-1495.[Abstract]

. 3 Willman CL, Busque L, Griffith BB, et al. Langerhans'-cell histiocytosis (histiocytosisX) -- a clonal proliferative disease. N Engl J Med 1994;331:154-160.[Abstract/Full Text]

. 4 Li M, Cordon-Cardo C, Gerald WL, Rosai J. Desmoid fibromatosis is a clonal process.Hum Pathol 1996;27:939-943.[Medline]

. 5 Paulli M, Bergamaschi G, Tonon L, et al. Evidence for a polyclonal nature of the cellinfiltrate in sinus histiocytosis with massive lymphadenopathy (Rosai-Dorfmandisease). Br J Haematol 1995;91:415-418.[Medline]

This article has been cited by otherarticles:

Malatos, S., Neubert, H., Kicman, A. T., Iles, R. K. (2005).Identification of Placental Transforming Growth Factor-{beta} and Bikunin Metabolites as Contaminants ofPharmaceutical Human Chorionic GonadotrophinPreparations by Proteomic Techniques. Mol Cell Proteomics



Related Articleby Rabkin, C. S.

Find Similar ArticlesPubMed Citation

4: 984-992 [Abstract] [Full Text] Butler, S. A., Iles, R. K. (2003). Ectopic Human ChorionicGonadotropin {beta} Secretion by Epithelial Tumors andHuman Chorionic Gonadotropin {beta}-Induced Apoptosisin Kaposi's Sarcoma: Is There a Connection?. Clin CancerRes 9: 4666-4673 [Abstract] [Full Text] Barillari, G., Ensoli, B. (2002). Angiogenic Effects of Extracellular HumanImmunodeficiency Virus Type 1 Tat Protein and Its Role in the Pathogenesis of AIDS-Associated Kaposi's Sarcoma. Clin. Microbiol. Rev. 15: 310-326 [Abstract] [Full Text] Judde, J.-G., Lacoste, V., Brière, J., Kassa-Kelembho, E., Clyti, E., Couppié, P.,Buchrieser, C., Tulliez, M., Morvan, J., Gessain, A. (2000). Monoclonality orOligoclonality of Human Herpesvirus 8 Terminal Repeat Sequences in Kaposi'sSarcoma and Other Diseases. J Natl Cancer Inst 92: 729-736 [Abstract] [Full Text] Moore, P. S., Chang, Y. (1998). Kaposi's Sarcoma-Associated Herpesvirus-EncodedOncogenes and Oncogenesis. J Natl Cancer I Monographs 1998: 65-71 [Abstract] [Full Text]

HOME | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | HELP

Comments and questions? Please contact us.

The New England Journal of Medicine is owned, published, and copyrighted © 2006 MassachusettsMedical Society. All rights reserved.

Analysis of Clonality of Atypical Cutaneous Lymphoid Infiltrates Associated With Drug Therapy By PCR/DGGE STEPHEN P. BRADY, MD, CYNTHIA M. MAGRO, MD, SALVADOR J. DIAZ-CANO, MD, PHD, AND HUBERT J. WOLFE, MD

Atypical lymphocytic infiltrates that mimic cutaneous lymphoma (ie, pseudolymphoma) are often observed in skin biopsy specimens from patients with altered immune function. The latter may reflect systemic immune dysregulatory states such as collagen vascular disease or human immunodeflciency virus infection. Among the iatrogenic causes are drug therapy with agents that abrogate lymphocyte function. These drugs encompass the anticonvulsants, antidepressants, phenothiazines, calcium channel blockers, and angiotensin- converting enzyme inhibitors. The appellation of lymphomatoid hypersensitivity reaction has been applied to cases of drug-associated pseudolymphoma. Pathologically and clinically, the distinction of such cases from cutaneous lymphoma is difficult. We employed the polymerase chain reaction (PCR) on archival material of proven drug-associated lymphomatoid hypersensitivity reactions both to ex- plore its utility as an adjunct in diagnosis and to investigate the genotypic aberrations induced by drug therapy. Formalin-fixed, paraffin-embedded biopsy spec imens from seven cutaneous T-cell lymphomas (CTCL), one nodal T-cell lymphoma, two cutaneous B-cell lymphomas, three typical hypersensitivity reactions, one tonsil, and 14 lymphomatoid hypersensitivity reactions were studied. Control cases for which DNA derived from fresh tissue was used include the Jurkat T-cell tumor line, placenta, one nodal B-cell lymphoma, and one case of reactive lymph node hyperplasia. DNA was obtained and purified by standard methods, then amplified with oligonucieotide primers specific for the T-cell receptor gamma locus and the immunoglobulin

heavy chain genes. T-cell ampllcons were analyzed by denaturing gradient gel electrophoresis (DGGE) and B-cell amplicons by either nondenaturing polyacrylamide or agarose gel electrophoresis. The nodal andJurkat T-cell lymphomas, six of seven CTCL, one cutaneous B-cell lymphoma, and 2 of 14 lymphomatoid hypersensitivity reactions showed dominant ("monoclonal") T-cell gene rearrangement patterns, and the remainder of cases were polyclonal. A causal relationship between drug therapy and skin eruption was ascertained in the two patients showing T-cell rearrangements, and both experienced complete and sustained lesional resolution on discontinuation of the implicated drug. The only immunoglobulin heavy chain gene rearrangements detected by PCR were in two of the three B-cell lymphomas. We conclude that PCR/DGGE is a powerful method for assaying T-cell clonality in archival tissue and can aid in the discrimination of reactive from malignant cutaneous infiltrates with appropriate clinicopathologic correlation. Recognition that a monoclonal TCR~/ rearrangement can be observed in cases of drug-associated lymphomatoid hypersensitivity may help in avoiding a misdiagnosis of malignant lymphoma. HUM PATnOL 30:130-136. Copyright © 1999 by W.B. Saunders Company

Key words: Clonality, T-cell receptor, PCR/DGGE, cutaneous pseudolymphoma, immunodysregulation.

Abbreviations: DGGE, denaturing gradient gel electrophoresis; IgH, Immunoglobulin heavy chain; PCR, polymerase chain reaction; TCR, T cell receptor.

Atypical cutaneous lymphoid infiltrates that clinically and light microscopically resemble lymphoma are problematic for the clinician and pathologist. Such cases fall under the broad designation of pseudolymphoma. It appears that the predisposed populace are those with underlying perturbations in immune function. 1,2 Systemic causes include autoimmune disease, atopy, and hematologic malignancy, while therapy with drugs known to alter lymphocyte function is a common iatrogenic cause. Several commonly prescribed agents, including angiotensin-converting enzyme inhibitors, calcium channel blockers, histamine antagonists, antidepressants, lipid-lowering agents, anticonvulsants, benzo- diazepines, and phenothiazines, 1-s can exert such effects. Among their effects are depression of T suppresser function or promotion of lymphocyte mitogenesis. It has been hypothesized that the aberrant immune re-

From the Department of Pathology, Tufts University School of Medicine, Boston, MA; Ameripath CPI Laboratories, Beachwood, OH; and the Department of Dermatology, University Hospitals, Case Western Reserve University, Cleveland, OH. Accepted for publication July 14, 1998.

Address correspondence and reprint requests to Stephen P. Brady, MD, Dermatopathology Department, Massachusetts General Hospital, Warren 526, 55 Fruit St, Boston, MA 02114.

Copyright © 1999 by W.B. Saunders Company 0046-8177/99/30024)004510.00/0

sponse in such cases either may be directed against the immune-dysregulating drug itself or may be triggered by an unrelated antigen. 1-3 In either scenario, the immune response triggered by the antigenic stimulus is abnormal. On histomorphologic grounds alone, the differentiation of these atypical lesions from cutaneous lymphoma often poses difficulty. We explored the utility of gene rearrangement studies as an aid in diagnosis. Such information also may facilitate our comprehen- sion at a molecular level of lesions of lymphomatoid hypersensitivity. The T-cell receptor gamma (TCR~/) locus was chosen for analysis because of both its relatively simple organization (Fig 1) and the fact that gamma rearrangements occur early in T lymphocyte ontogeny and persist after rearrangements of other TCR loci. 4,5 Rearrangements involving the immunoglobulin heavy chain genes were studied to assess B-cell clonality.

MATERIALS AND METHODS Selection of Cases/Criteria Used for Diagnosis of Lymphomatoid Hypersensitivity

The cases were provided by one o f the authors (C.M.M.) and inc luded three cases (cases 1, 4, 14) previously r epor t ed by Magro and Crowson. 1 In all of the cases, a diagnosis of drug-associated lymphomatoid hypersensitivity was made based

130

CLONALITY OF ATYPICAL CUTANEOUS LYMPHOID INFILTRATES (Brady et al)

Vy1-8 V?9 VT10 VyI| Jl C1 J2 C2

P r i m e r Bind ing ~ - - - Sites ~ :

o

V N 3 C o . ' " " ' " ' "

I ~ : . . ~ ~ Rearrat~ged DNA

FIGURE 1. The relative simplicity of the TCRy locus and its high frequency of conserved sequences permit the design of just a few primer pairs capable of detecting most T-cell gene rearrangements. N depicts the variable number of nucleotides randomly inserted or deleted during V and J joining. Arrows indicate primer binding sites for both V-y1-8 (nested PCR) and V@ (seminested PCR) genes.

on ingestion of drugs with potential immune dysregulating properties before the onset of the eruption, characteristic light microscopic findings alluded to below, lesional resolution in those cases in which there was cessation of one or more of the implicated drugs or response to other treatment modalities, and isolation of a definite antigenic trigger in the biopsy specimen (ie, an infective agent). By light microscopy, lymphomatoid hypersensitivity can be categorized as (1 and 2) interface dermatitis with variable epitheliotropism versus a psoriasiform and eczematous dermatitis with lymphoid atypia, hence recapitulating the morphology of pa tch /p laque stage mycosis fungoides, (3) nodular lymphoid hyperplasia produc- ing a morphology resembling B-cell lymphoma, (4) angiocentric atypical lymphocytic infiltrates (described as lymphomatoid vascular reactions) resembl ing angiocentr ic T-cell lymphoma, and (5) lymphomato id follicular mucinosis whereby the outer root sheath of the follicular epithelium is permeated by atypical lymphocytes and there is at tendant mucinosis. Such cases raise diagnostic consideration to follicu- locentric mycosis fungoides. Positive controls included one nodal T-cell lymphoma, seven cutaneous T-cell lymphomas inclusive of mycosis fungoides, and the Jurkat T-cell tumor line. Negative controls consisted of three typical cutaneous hypersensitivity reactions, one tonsil, one reactive lymph node, and placenta. Light microscopic features of typical cutaneous responses include vacuolar interface dermatitis, perivascular lymphocytic infiltrates with tissue eosinophilia, and eczematous dermatitis. Three B-cell lymphomas (two cutaneous) also were examined. All cases in the series were formalin fixed and paraffin embedded except for the Jurkat clone, nodal B-cell lymphoma, lymph node hyperplasia, and placenta, for which DNA derived from fresh-frozen tissue was available. Figure 2 illustrates the histopathology of representative cases.

Analysis o f T- a n d B-Cell C lona l i t y DNA Extraction

Five 20-pm unbaked unstained sections were prepared from each block and the epidermis microdissected using a fine needle to enrich for lymphocyte-derived DNA. These thick sections were preceded and followed by standard 5-1am hematoxylin and eosin sections to ensure lesional persistence through the leveling process. The tissue sections were deparaf-

finized in xylene, washed with ethanol, and subjected to prolonged proteinase K digestion (5 to 6 days), and the subsequently released DNA was purified by phenol-chloroform extraction as recently described 6 (Fig 3). Finally, the DNA was ethanol precipitated and resuspended in 40 taL Tris-ethylenediaminetetra-acetic acid buffer before use in the polymerase chain reaction (PCR).

TCR~, Ampflficafion and Criteria for DGGE Interpretation

Previously published nested consensus primers recognizing Vyl-8 genes and semi-nested consensus primers recognizing the Vy9 gene were employed to detect T-cell gene rearrangements. ~ A two-round amplification strategy was employed, with a "nested" approach used for Vyl-8 targets and a "semi-nested" for Vy9 (Fig 1). Reaction and thermocycling conditions as described by Wood et al 7 were followed, with the exception that 30 first-round cycles succeeded by 20 cycles in the second round was found to be optimal for DNA extracted from paraffin-embedded tissues. PCR amplification was performed using a Perkin-Elmer 9600 thermocycler (Norwalk, CT), and all runs included a previously established positive control and tubes to which no DNA was added. One-half microgram DNA isolated from fresh or frozen tissues was used as template in the first reaction round, and DNA derived from archival tissues was not quantitated (5 IlL of each Tris- ethylenediaminetetra-acetic acid resuspended sample was used). The presence of amplification product in the expected size range was verified by agarose gel electrophoresis (Fig 4), and then the amplicons were separated by denaturing gradient gel electrophoresis as previously described. 7,8 All cases were amplified in duplicate using both nondiluted and 1/20 diluted first-round DNA as template in the second round. Gels were stained 15 minutes with ethidium bromide and photographed under ultraviolet illumination.

Denaturing gradient gel electrophoresis (DGGE) patterns were interpreted as polyclonal if electrophoresis re- suited in either a diffuse smear or in multiple weak bands, and as clonally rearranged if a dominant banding pattern was achieved. Bands were classified as dominant if they were discrete, sharp, and persisted or intensified after dilution of the first-round product. Up to four such bands were permit ted because of the possibilities of bi-allelic rearrangements and

131

HUMAN PATHOLOGY Volume 30, No. 2 (February 1999)

widely variable size, such that clones generate unique banding patterns detectable by nondenaturing electrophoresis.

RESULTS Cl in ica l Features

The clinical information is summarized in Table 1.

Histology C a s e 1 (MF-like)

The biopsy specimen was remarkable for a superficial infiltrate, which focally assumed a band-like pat tern of infiltration within the epidermis. There was striking epitheliotropism with localization to the rete and acrosy- ringium. The infiltrate had a polymorphous composition comprising small, intermediate, and t ransformed hyperchromatic and irregularly con toured lymphocytes.

Case 2 (MF-like)

The biopsy specimen displayed a superficial band- like lymphocytic infiltrate. The epidermis showed variable hyperplasia. The lymphocytes were in the 7- to 9-gin range. A few scattered cells with a Sezary morphology were identified.

FIGURE 2. (A) Plaque-stage mycosis fungoides with hyperkeratosis, extensive epidermal colonization by atypical lymphocytes, and dermal fibrosis. (H&E, original magnification ×125 [same case as illustrated in Fig 5B, lanes 13-14].) (B) Case of lymphomatoid lichenoid hypersensitivity with striking similarity to the lymphoma depicted in A (patient no. 4). (H&E, original magnification ×125.) Note the hyperkeratosis and papillary dermal fibrosis, as well as focal epidermal permeation by small atypical lymphocytes. This single-plaque lesion was associated with a histamine antagonist and showed monoclonal TCR~ rearrangement by PCR/DGGE (Fig 5C, lane 17), Complete resolution occurred within several weeks of discontinuing the medication, and the patient remains free of disease 2 years later.

heteroduplex formation, but usually only one to two bands were present. Weakly staining bands were not infrequently noted, especially after amplification with the V~9 primers, and tended to disappear or lose intensity after template dilution-- such bands could not be considered representative of a clonal expansion.

Immunoglobulin Heavy Chain Amplification

Primers recognizing consensus immunoglobulin H ([gH) variable region framework (FR) I-IV and III-IV sequences and a consensus IgH joining region primer were used to study B-cell gene rearrangements as described by Segal et al. 9,1° The B-cell amplicons were separated by either 3% agarose (FR I-IV) or 12% nondenaturing polyacrylamide (FR III-IV) gel electrophoresis. The gels were stained and photographed in the same manner as for the T-cell studies. In contrast to TCR% rearrangements involving the IgH locus result in products of

Case 3 (MF-like)

The biopsy specimen showed a ban&like lymphocytic infiltrate with obscuration of the dermoepidermal junction. There was prominent infiltration of the epidermis by atypical lymphocytes with foci of frank architec- tural ablation of the epidermis. The cells within the dermis were dominated by small mature lymphocytes with minimal atypia, whereas those cells within the epidermis were significantly atypical and included cells with a Sezary morphology.

Case 4 (MF-like)

A band-like lymphocytic infiltrate in close apposi- tion to the epidermis was present, with colonization of the basal and parabasilar epidermis by small lymphocytes showing significant nuclear contour irregularity (Fig 2B). A coarse sclerosing dermal response within the papillary and superficial reticular dermis was additionally noted.

Case 5 (B-Cell Lymphoma Cutis-like Pattern)

The biopsy specimen was remarkable for a pander- mal dense nodular nonepi thel iotropic lymphocytic infiltrate. Centrally the dominant composit ion of the infiltrate was one of t ransformed lymphocytes, and the per iphery showed small mature lymphocytes with ad- mixed eosinophils.

Case 6 (Lymphomatoid Vascular Reaction)

The biopsy specimen was remarkable for dense superficial and deep angiocentric lymphocytic infil-

132

CLONALITY OF ATYPICAL CUTANEOUS LYMPHOID INFILTRATES (Brady et al)

T-cell Gene Rearrangement

p~&~ E ~ GE

B-Cell Gene Rearrangement

Epidermal Microdissection

Phenol - Chloroform

! { i ~ Extraction

1 Proteinase K

Digestion

FIGURE 3. Schematic illustration of DNA extraction process from archival tissue with subsequent amplification and product analysis.

trates predominated by immunoblasts with an admixture of other inflammatory cell elements, including small mature lymphocytes, plasma cells, and eosinophils. Scabetic mites were identified within the stratum corneum.


The biopsy specimen showed a striking vascular reaction involving the entire sampled cutaneous vasculature. The infiltrates comprised a monomorphic populace of small and intermediate-sized lymphocytes with significant nuclear contour irregularity. A concomitant

SM 12,216

1018 517

134

FIGURE 4. Agarose gel with typical second-round PCR products. Each sample was amplified in duplicate with both Vyl-8 and V~,9 primers, accounting for the four lanes/sample (representative samples indicated by white numbers within gel), The staggered appearance results from the larger product obtained after Vyl-8 amplification (-420 bp) versus V@ (-380 bp).

diffuse interstitial granulomatous component also was present.

Case 8 (Folliculotropic MF-Iike and Lymphoma Cuffs-like)

A dense lymphohistiocytic infiltrate involving the mid and deep dermis with extension into the subcutane- ous fat was present, as well as marked infiltration of sampled hair follicles by this infiltrate. One of the inflamed follicles showed extensive permeation by fungal forms consistent with an endothrix infection.

Case 9 (Folliculotropic MF-like)

A striking reaction involving sampled hair follicles characterized by dense perifollicular and intrafollicular lymphohistiocytic infiltrates with attendant follicular mucinosis was present, as were nodular mononuclear cell infiltrates surrounding and permeating perifollicular vessels.


The biopsy specimen was remarkable for nodular angiocentric mononuclear cell infiltrates involving the superficial and mid-dermal vasculature, with attendant luminal and mural fibrin deposition and red cell extravasation. The cytomorphology of the mononuclear cells encompassed intermediate and transformed lymphocytes with mild nuclear contour irregularity.

Case 11 (Folliculotropic MF-Iike and Lymphoma Cuffs-like)

The biopsy specimen was remarkable for a dense nodular lymphohistiocytic infiltrate, which had as its epicenter sampled follicles. Concomitant cytopathic alterations of the follicles were noted consistent with molluscum contagiosum infection.

133

HUMAN PATHOLOGY Volume 30, No. 2 (February 1999)

TABLE 1. Summary of Clinical Data and DGGE Results

Patient Age Clinical Nature of Therapy and DGGE No. (yr)/Sex Presentation Immunodysregulation Current Status Pattern

1 44F Chest papule H2A 2 63F Plaque on left leg CCB, SS 3 66M Plaque on palm H2A 4 48F Plaque on thigh H1A 5 59F Nodule on tip of nose SS 6 86F Nodules on back and left arm (scabetic) ACE, HIA × 2; 2 others

7 70M Maculopapular rash on extremities Prilosec

8 52F CCB, Anticonv X2 Papules and plaques on face associated with Endothrix

9 48M Papules on leg and torso AD, Anticonv 10 39F Multiple papules on distal extremities ACE, AD × 2 11 18F Papule on cheek SS; Atopic diathesis 12 54F Plaques on face, neck, and forearm CCB 13 43F Generalized papules H1A; 1 other 14 65M Patches & plaques on trunk BB, ACE

& extremities

Lesion Excised/NED P CCB discontinued/NED P H2A discontinued/NED M H1A discontinued/NED M Treated with x-ray therapy/NED P Continued drug therapy, lesions resolved P

with treatment of scabetic infesta- t ion/NED

Lesions resolved with steroid P therapy/NED

Lesions resolved with anti-fungal P therapy/NED

AD discontinued/NED P ACE discontinued/NED P Excision of lesion/NED P CCB discontinued/NED P H1A discontinued/NED P Drugs discontinued/NED P

Abbreviations: NED, no evidence or recurrence of disease after discontinuation of implicated drug or other therapy; AD, antidepressants (ie, Amitriptyline [case no. 9], Prozac [case no. 10], Klonopin [case no. 10]); Li, Lithium; ACE, angiotensin-converting enzyme inhibitor; CCB, calcium channel blocker; Bzp, benzodiazepine; LiLA, lipid-lowering agent; H2A, histamine 2A-receptor antagonist; H1A, histamine 1A receptor antagonist; SS, sex steroids (ie, estrogen and progesterone); anficonv, anticonvulsants (ie, carbamazepine, dilantin, phenobarbital); BB, beta-blocker; other, Nitroprusside, Dyazide, Flexeril, Relafan; ×, number of drugs of that class that the patient was receiving; P, polyclonal; M, monoclonal.

Case 12 (MF-like)

The biopsy specimen showed a vacuolar and lichenoid interface dermatitis with epitheliotropism.

Case 13 (MF-like)

Two biopsy specimens were available and appeared morphologically similar. In each a superficial interstitial and perivascular lymphocytic infiltrate was observed. There was haphazard infiltration of the epidermis by lymphocytes with lymphoid forms extending into the upper layers of the epidermis. The cytomorphology comprised small and intermediate-sized lymphocytes with nuclear contour irregularity.

(Fig 6). The latter also showed an apparent TCR~ rearrangement. Although rearrangements discordant with immunophenotype have been described, 11 it is more likely in this particular case that amplification of DNA from relatively rare infiltrating T lymphocytes resulted in a pseudoclonal pattern.

When the DGGE results were compared with tissue reaction pattern, a correlation between detectable clonal

2 4 6 8 10 12 14 16 18

Case 14 (MF-like)

The biopsy specimen was remarkable for a superficial perivascular and interstitial mononuclear cell infiltrate. There was slight exocytosis of lymphocytes. The vessels amidst the infiltrate showed injurious alterations as indicated by mural and luminal fibrin deposition. The infiltrate was dominated by intermediate and transformed atypical lymphocytes inclusive of cells with a Sezary morphology.

Molecular Studies The nodal and Jurkat T-cell lymphomas, six of

seven cutaneous T-cell lymphomas (86%), and 2 of 14 of the study cases (14%) displayed dominant TCR~ gene rearrangement patterns, whereas the remainder of cases were polyclonal (Fig 5). The only immunoglobn- lin heavy chain gene rearrangements detected by PCR were in two of the three B-cell lymphomas, including the nodal and one of the two cutaneous lymphomas

FIGURE 5. DGGE Results. (A) Lymph node hyperplasia shows a diffuse smear indicative of its polyclonal nature (lanes 1-2), whereas the Jurkat T-cell tumor line shows a V71-8 rearrangement (lanes 5-6; results of Jurkat amplification with V79 primers are seen in lanes 3-4). (B) Nodal T-cell lymphoma with monoclonal pattern (lanes 7-8), two cases of polyclonal LyHR (lanes 9-10 [patient no. 2] and 11-12 [patient no. 10]), MF showing a dominant banding pattern diagnostic of rearrangement (lanes 13-14, same case as in Fig 2A), and polyclonal case of typical hypersensitivity (lanes 15-16); (C) Monoclonal LyHR (lane 17, patient no. 4), monoclonal MF (lane 18), and polyclonal tonsil (lane 19).

134

CLONALITY OF ATYPICAL CUTANEOUS LYMPHOID INFILTRATES (Brady et at)

SM 2 4 6 8 10 12 14 16

SM 2 4 6 8 10 12 14 16 18

FIGURE 6. (A) FR I-IV, Multiplex PCR: 3% agarose gel shows IgH chain gene rearrangement in case of paraffin-embedded cutaneous B-cell lymphoma (arrow, lane 2) and fresh-frozen nodal B-cell lymphoma (arrow, lane 11). (B) FR Ill-IV, semi- nested PCR: 12% nondenaturing polyacrylamide gel with first- and second-round PCR products. Arrows point to clona[ gene rearrangement detected after first (lane 6) and second (lane 4) amplification rounds, Same case as depicted in lane 2 of (A).

expansion and a dense band-like (ie, lichenoid) morphology was apparent. Only 4 of 14 cases displayed this pattern (patients 1 through 4), yet they included both monoclonal cases. The nonl ichenoid patterns encompassed (1) a lower-density epitheliotropic, eczematous, and psoriasiform morphology, (2) lymphomatoid vascu- litis, (3) follicular mucinosis, and (4) lymphocytoma cutis, none of which manifested any T-cell clonal rearrangement.

DISCUSSION

We have investigated T-cell clonality of drug- associated lymphomatoid hypersensitivity reactions by

PCR/DGGE using formalin-fixed, paraffin-embedded tissues. Although many investigators employing similar techniques to study malignant and reactive lymphocytic infiltrates of skin have relied on fresh or frozen material, 12-14 there are increasing reports in the literature regarding the use of such archival tissue for TCR~ analysis. 15-1s Our 86% detection of TCR~/ rearrange- merits in our cutaneous T-cell lymphoma controls (89% detection of positives if the nodal lymphoma andJurkat samples are included) is in close agreement with that published by others. 7,1~,16 It also reaffirms PCR as a powerful method for assaying T-cell clonality, a characteristic hallmark of malignant T-cell lymphoma. Regard- ing B-cell clonality, none of our study cases showed IgH chain rearrangement. This contrasts with the findings of Ritter et al, 19 who not uncommonly identified such rearrangements in mixed B and T as well as predomi- nantly T-cell cutaneous lymphoid infiltrates. 19

The current series represents a collection of atypical reactive lymphocytic infiltrates that raised cutaneous lymphoma as a diagnostic possibility. The appellation of lymphomatoid hypersensitivity reaction has been applied to such cases. I When the antigenic trigger is isolated, the terminology is modified accordingly. Hence, in the setting of an abnormal immune response to a contact, the term lymphomatoid contact dermatitis is used. 2° Although previous studies of reactive cutaneous infiltrates have shown clonal rearrangements ranging from 0% 12"14,17,18 to 6%, 7 detectable rearrangements were present in 2 of 14 of our lymphomatoid hypersensitivity cases (14%). This finding is similar to that of Staib and Sterry 16 who also found dominant rearrangements in 14 % o f " pseudolymphoma." The association of atypical lymphoid hyperplasia with a variety of therapeutic agents has previously been reported by Magro and Crowson, ls who postulated that blocking or stimulation of receptor-mediated lymphocyte function may result in aberrant immune responses to a variety of antigenic triggers. A similar theory of systemic or iatrogenic immune dysregulation leading to atypical lymphoid hyperplasia has been postulated by other authors. 21,~2 In one of the reports, they isolated herpes as the antigenic trigger; the underlying immune dysregulatory state was malignant lymphoma. 21 The role of an infective antigenic trigger in the propagation of the infiltrate was similarly documented in three of our cases. It is possible that emergence of a dominant clone or clones is facilitated by dysregulated T-cell function or by direct mitogenic effects of pharmacological agents on lymphocytes, and that this accounts for our increased finding of monoclonality.

When the implicated drugs were discontinued in the two patients showing clonal rearrangement, both experienced complete resolution of their lesions and remain free of disease up to 2 years later (patients 3 and 4). Although both drug-associated pseudolymphoma progressing to lymphoma 23 and drug-associated lymphoma are described, 24 a diagnosis of malignant lymphoma is unwarranted in our two cases in light of the spontaneous and sustained resolution of lesions after drug withdrawal. It is increasingly recognized that

135

REVIEW

General morphological and biological features of neoplasms:integration of molecular findings

S J Diaz-CanoDepartment of Histopathology, King’s College Hospital and King’s College London School of Medicine, London, UK

Diaz-Cano S J

(2008) Histopathology 53, 1–19

General morphological and biological features of neoplasms: integration of molecularfindings

This review highlights the importance of morphology–molecular correlations for a proper implementationof new markers. It covers both general aspects oftumorigenesis (which are normally omitted in papersanalysing molecular pathways) and the general mech-anisms for the acquired capabilities of neoplasms. Themechanisms are also supported by appropriate dia-grams for each acquired capability that include over-looked features such as mobilization of cellularresources and changes in chromatin, transcriptionand epigenetics; fully accepted oncogenes and tumoursuppressor genes are highlighted, while the pathwaysare also presented as activating or inactivating withappropriate colour coding. Finally, the concepts andmechanisms presented enable us to understand thebasic requirements for the appropriate implementationof molecular tests in clinical practice. In summary, thebasic findings are presented to serve as a bridge toclinical applications. The current definition of neo-plasm is descriptive and difficult to apply routinely.Biologically, neoplasms develop through acquisition ofcapabilities that involve tumour cell aspects and modi-fied microenvironment interactions, resulting in un-restricted growth due to a stepwise accumulation ofcooperative genetic alterations that affect key mole-

cular pathways. The correlation of these molecularaspects with morphological changes is essential forbetter understanding of essential concepts as earlyneoplasms ⁄ precancerous lesions, progression ⁄ dedif-ferentiation, and intratumour heterogeneity. Theacquired capabilities include self-maintained repli-cation (cell cycle dysregulation), extended cell survival(cell cycle arrest, apoptosis dysregulation, and repli-cative lifespan), genetic instability (chromosomal andmicrosatellite), changes of chromatin, transcriptionand epigenetics, mobilization of cellular resources,and modified microenvironment interactions (tumourcells, stromal cells, extracellular, endothelium). Theacquired capabilities defining neoplasms are thehallmarks of cancer, but they also comprise usefultools to improve diagnosis and prognosis, as well aspotential therapeutic targets. The application of theseconcepts in oncological pathology leads to consider-ation of the molecular test requirements (MolecularTest Score System) for reliable implementation; theserequirements should cover biological effects, molecularpathway, biological validation, and technical valida-tion. Sensible application of molecular markers intumour pathology always needs solid morphologicalsupport.

Keywords: cell kinetics, differentiation, genetic instability, molecular pathology, neoplasia, pathogenesis,progression, tumour heterogeneity, tumour microenvironment

Abbreviations: AIF, apoptosis-inducing factor; ALT, alternative lengthening of telomeres; APE, apurinic ⁄apyrimidinic endonuclease; BRCA, Breast cancer; CAD, caspase-activated DNase; CDK, cyclin-dependent kinase;DFF, DNA fragmentation factor; EndoG, endonuclease G; HDAC, histone deacetylase; HIF, hypoxia inducible factor;

Address for correspondence: S J Diaz-Cano, MD, PhD, FRC Path, Department of Histopathology, King’s College Hospital, Denmark Hill, London

SE5 9RS, UK. e-mail: [email protected]

The protocol used in the study was approved by the Hospital Research Board and Ethical Committees and complied with their requirements.

� 2008 The Author. Journal compilation � 2008 Blackwell Publishing Limited.

Histopathology 2008, 53, 1–19. DOI: 10.1111/j.1365-2559.2007.02937.x

LOH, loss of heterozygosity; MAP, mitogen-activated protein; MBD, methyl CpG-binding domain; Mcm, mini-chromosome maintenance complex; MMP, matrix metalloproteinase; MSI, microsatellite instability; MTA,metastasis-associated protein; mTOR, mammalian target of rapamycin, NAD, nicotinamide adenine dinucleotide;NF, nuclear factor; ORC, origin recognition complex; PARP, polyADP-ribose polymerase; PAX, paired box; PDK1,3-phosphoinositide-dependent protein kinase-1; PIP3, phosphatidylinositol 3,4,5-triphosphate; PPAR, peroxisomeproliferator-activated receptor; PTC, patched; PTEN, phosphatase and tensin homologue; RAS, rat sarcoma; Rb,retinoblastoma; RET, rearranged during transfection; ROS, reactive oxygen species; TGF, transforming growthfactor; TMM, telomere maintenance mechanism; TNF, tumour necrosis factor; VEGF, vascular endothelial growthfactor

Introduction

Histopathology remains the gold standard of tumourdiagnosis, but it needs strong links to all new develop-ments such as molecular markers. Proper morpholog-ical evaluation has been demonstrated to be invaluablefor both establishing concepts and helping with tech-nical issues, especially in dealing with the hetero-geneous components in neoplasms; the application ofthese principles would allow more reliable comparisonof results and implementation of meaningful concepts.Any sustained progress in this area should refer to themorphological changes for verification and validation.Hanahan and Weinberg’s article on tumorigenesis isone with the greatest degree of impact,1 but does notpay much attention to molecular–morphological cor-relation in neoplasms. Nevertheless, key oncologicalfeatures need careful morphological correlation whennew markers are introduced in the diagnosis andmanagement of patients. In tumour pathology newmarkers have been incorporated at the tissue level tohelp defining subtypes, but there is an increaseddemand for molecular tests on tissue sections (in situtechniques) and on solid support (techniques afterextraction). Proper application of these techniquesrequires sensible selection of markers that should bebased on concepts of tumour biology (both general andtissue specific).

General aspects of tumorigenesis

A general definition of neoplasm, such as ‘cellulardisease characterized by abnormal growth regulatorymechanisms’, is descriptive and difficult to apply rou-tinely, working definitions being required. The intro-duction of new markers has improved diagnosticprecision, but can potentially result in big changes inprevalence and uncertainties for particular lesions. Thecurrent World Health Organization classifications oftumours incorporate new developments based onpathology and genetics, the leading criteria still being

morphological; in this context, molecular findings com-plement the histological evaluation without replacing it.Additionally, any new definition should be validatedagainst the accepted standard (specificity ⁄ sensitivity),should improve patient management, and should pro-vide a biological meaning for its application. The firstrequirement is normally met on the initial design, andthe second would be expected in any successful imple-mentation. The third criterion is more difficult to apply,but any new definition should be biologically meaning-ful and would incorporate core elements in tumourbiology (in particular, genetic and kinetic correlates).2–6

These elements need to be included in a score system.Examples include the paired box (PAX) 8 ⁄ peroxisomeproliferator-activated receptor (PPAR) c fusion genedescribed in follicular thyroid carcinomas and adeno-mas, and rearranged during transfection (RET) ⁄ patched(PTC) fusion genes reported in papillary thyroid car-cinomas and Hashimoto’s thyroiditis.

Biologically, neoplasms develop through acquisitionof capabilities that involve tumour cell aspects andmicroenvironment interactions, as explained in moredetail below (General molecular mechanisms).1 Theunrestricted growth observed in neoplasms is generallydue to a stepwise accumulation of cooperative geneticalterations in oncogenes and tumour-suppressor genes,the number being more important than the order ofchanges;7 the evidence available suggest that five toseven genetic alterations are required for clinicallydetectable tumours, correlating with morphologicalprogression in some locations. These capabilities arenot equally relevant at different stages during tumor-igenesis, as highlighted by careful morphological eval-uations. The markers should be selected consideringthe capability to test and the marker role duringtumour initiation and promotion. Tumour promotionmarkers would be more relevantly assessed duringprogression, which needs to be defined on clear clinicaland morphological grounds. These aspects are relevantfor three essential concepts: early neoplasms ⁄ pre-cancerous lesions, progression ⁄ dedifferentiation, and

2 S J Diaz-Cano

� 2008 The Author. Journal compilation � 2008 Blackwell Publishing Ltd, Histopathology, 53, 1–19.

intratumour heterogeneity (considering tumour cellsegregation and heterotypic biology ⁄ landscaper effect).

early neoplasms and precancerous les ions

These two concepts are closely related to tumourinitiation, have been developed for epithelial neoplasmsand corroborate the concept of multistep tumorigenesisand accumulation of cooperative genetic abnormalities(‘gatekeeper’ and ‘caretaker’ pathways).7 The para-digm is the concept of intraepithelial neoplasm ⁄ malig-nancy, it being more difficult to extrapolate the conceptto non-epithelial lesions. These lesions would bemeaningful when they are present in structures withanatomical boundaries and the cells do not recircu-late ⁄ migrate in physiological conditions, regardless oflesion size. Regarding anatomical considerations, thelimiting structure is the basement membrane, not thetumour capsule.

Although some genetic alterations are described asneoplasm-specific, the presence of a single geneticalteration cannot be considered diagnostic of malig-nancy, even for early stages. These problems precludeestablishing reliable diagnoses of follicular carcinomasin situ for encapsulated neoplasms carrying PAX8 ⁄PPARc fusion genes, even for lesions that initially carrymolecular changes reported in malignancy. The oppo-site situation is equally important: histologically con-firmed intraepithelial lesions are considered precursors,but they can accumulate genetic alterations and showkinetic features of malignancies, as reported for C-cellhyperplasias in multiple endocrine neoplasia 2A.8,9

Non-random genetic alterations can also be used totest clonal expansions and the clonal evolution ofneoplasms, especially analysing hypervariable DNAregions from patients heterozygous for a given marker.These tests rely basically on the demonstration of lossof heterozygosity (LOH) resulting from either hemi-zygosity (non-random interstitial DNA deletions) orhomozygosity of mutant alleles observed in neoplasms.LOH analyses identify clonal expansions of a tumour cellpopulation and point to monoclonal proliferation whenmultiple and consistent LOH is demonstrated (highfractional allelic loss).4 Applied appropriately, these toolscan establish the clonal evolution of tumour cell popu-lations (tumour heterogeneity) and differentiate fieldtransformation from metastatic tumour growths insynchronic and histologically identical neoplasms.

neoplastic progress ion and dedifferentiation

Neoplastic transformation evolves over a period oftime and involves the phenotypic progression of

tumour cells along with the interaction of theinitiated cell with its microenvironment. The elucida-tion of the steps of cancer progression (Figure 1)relates to the acquisition of invasive capability inintraepithelial lesions and metastatic potential ininvasive malignancies and is of utmost importancein the differential diagnosis of neoplasms and in theestablishment of more efficient therapeutic regimens.This functional characterization of the particularstage of tumour will certainly allow for betterdiagnosis, staging, prognostication and treatment ofcancers. Tumour cell metabolic activation also deter-mines the degree of differentiation and needs to becarefully coupled with kinetic features; any imbalancein the ribosomal activation ⁄ kinetics will result inapoptosis activation (see General molecular mecha-nisms). Changes in gene expression are commonfindings in neoplasms, due to general and capability-specific factors. On one hand, general mobilizationof resources will result in activation of the cellularmachinery needed for transcription (with correspond-ing chromatin changes) and translation, which isregulated by mammalian target of rapamycin) andeIF4E pathways, as well as for ribosomal activation.On the other hand, capability-specific markers acti-vate transcription factors that modified the geneexpression. These two collaborative aspects explaingene expression modifications that can be powerfulmarkers of neoplastic transformation and have beenproposed as diagnostic criteria for a molecular clas-sification of neoplasms.10–15

Histopathological examination of solid tumoursfrequently reveals pronounced tumour cell hetero-geneity, often demonstrating substantial diversitywithin a given tumour. The molecular mechanismsunderlying the phenotypic heterogeneity are verycomplex, with genetic, epigenetic and environmentalcomponents, such as shortage of oxygen. Hypoxiagreatly influences cellular phenotypes by altering theexpression of specific genes, makes the tumours moreaggressive, and is an important contributor to intra-and intertumour cell diversity, as revealed by thepronounced but non-uniform expression of hypoxia-driven genes in solid tumours.16 Hypoxic tumour cellslose their differentiated gene expression patterns anddevelop stem cell-like, immature or dedifferentiatedphenotypes.17 Not only will hypoxia-induced dediffer-entiation contribute to tumour heterogeneity, but itcould also be one mechanism behind increasedaggressiveness of hypoxic tumours. Intratumoralhypoxia is an independent indicator of poor patientoutcome, and increasing evidence supports a role forhypoxia in the development of metastatic disease.16

Molecular bases of tumorigenesis 3


Studies suggest that the acquisition of the metastaticphenotype is not simply the result of dysregulatedsignal transduction pathways, but instead is achievedthrough a stepwise selection process driven byhypoxia.18,19 Hypoxia facilitates disruption of tissueintegrity through repression of E-cadherin expression,with concomitant gain of N-cadherin expression,which allows cells to escape anoikis. Through up-regulation of urokinase-type plasminogen activatorreceptor expression, hypoxia enhances proteolyticactivity at the invasive front and alters the interac-tions between integrins and components of theextracellular matrix, thereby enabling cellular inva-sion through the basement membrane and theunderlying stroma. Cell motility is increased throughhypoxia-induced hepatocyte growth factor–MET recep-tor signalling, resulting in cell migration towards theblood or lymphatic microcirculation. Hypoxia-inducedvascular endothelial growth factor (VEGF) activity alsoplays a critical role in the dynamic tumour–stromal

interactions required for the subsequent stages ofmetastasis.

Metastasis, the final step in malignancy, is theresult of selected aspects of the complex tumour-progression process. Tumour cell dissemination isthe prerequisite of metastasis and is correlated withloss of epithelial differentiation and the acquisition ofa migratory phenotype, a hallmark of malignanttumour progression.17,20,21 A stepwise, irreversibleaccumulation of genetic alterations is considered tobe the responsible driving force, but, strikingly,metastases of most carcinomas recapitulate theorganization of their primary tumours. The loss ofepithelial characteristics (i.e. lack of intercellularadhesion molecules) results in breakdown of epithe-lial-cell homeostasis, correlates with the acquisition ofa migratory phenotype and leads to progression.22

This epithelial to mesenchymal transition is consid-ered to be a crucial event in malignancy.21 Theimportant steps that enable metastasis are reversible,

Figure 1. Molecular pathways involved in tumour cell progression.

4 S J Diaz-Cano


and therefore cannot be explained solely by irrevers-ible genetic alterations, indicating the existence of adynamic component to human tumour progression, inparticular a regulatory role for the tumour microenvi-ronment (Figure 2).23 Tumours are morphologicaland functionally heterogeneous and segregate intotumour cells with different capabilities: individualtumours show distinct sub-areas of proliferation andcell-cycle arrest, differentiation, cell adhesion anddissemination, some of them determined by topogra-phy (see below). VEGF promotes angiogenesis andlymphangiogenesis in the primary tumour, providingthe necessary routes for dissemination. VEGF-inducedchanges in vascular integrity and permeability pro-mote both intravasation and extravasation, whereasVEGF-induced angiogenesis in the secondary tissue isessential for cell proliferation and establishment ofmetastatic lesions. Through regulation of these criticalmolecular targets, hypoxia promotes each step of themetastatic cascade and selects tumour cell populations

that are able to escape the unfavourable microenvi-ronment of the primary tumour.

intratumour heterogeneity and tumour cell

segregation

Intratumour heterogeneity comprises both tumourcells and heterotypic components (immune ⁄ inflamma-tory cells, myofibroblasts, endothelial cells and extra-cellular matrix). Intratumour heterogeneity is assumedto occur randomly, but some factors such as topogra-phy control the segregation of tumour cells withinneoplasms.24–29 The topographic intratumour hetero-geneity suggests differential selection of tumour cells,but can also be expression of either selective clonalevolution or a simple passive by-product of geneticinstability. The differential kinetic profile by topo-graphic compartments has been related to lower cellturnover and apoptosis down-regulation in deep ⁄peripheral compartments, resulting in accumulation

Figure 2. Molecular pathways involved in microenvironment interactions (intercellular, and stromal, neoangiogenesis, and invasion).



of genetic alterations and segregation of tumour cellswith differential genetic backgrounds as demonstratedin adrenal gland, colon and bladder. This processhas been linked with mismatch repair protein down-regulation and it is unlikely to be related with hypoxia,which is more pronounced in central compartments.However, the coexistence of genetic alterations sup-ports a key role in tumorigenesis, the topographicheterogeneity resulting from the accumulation ofgenetic damage. This concept is central and supportsmultiple sampling to assess reliably the geneticabnormalities of neoplasms.

The heterotypic biology of neoplasms is an essentialelement in understanding tumour growth. The under-lying defect (clonal genetic alteration) may reside instromal and not tumour cells, as reported in juvenilepolyposis syndrome and ulcerative colitis hamartoma-tous polyps.30 This finding suggests that, at leastinitially, the stromal cells are the neoplastic cells,whereas secreting factors drive the epithelial prolifer-ation, and might thus eventually also be responsible forthe induction of epithelial malignancy. This bystanderrole, mutations inducing stromal abnormalities that inturn induce epithelial neoplasia, has been called alandscaper effect: the microenvironment surroundingepithelial cells as a major determinant of the disturbedepithelial architecture, differentiation and proliferation(Figures 1 and 2).

General molecular mechanisms intumorigenesis

The main tumorigenesis molecular pathways mustbe evaluated according to the acquired capabilities:self-maintained replication (cell-cycle dysregulation),extended cell survival (cell-cycle arrest, apoptosis dys-regulation, and replicative lifespan), genetic instability(chromosomal and microsatellite), changes of chro-matin, transcription and epigenetics, mobilization ofcellular resources, and modified microenvironmentinteractions (tumour cells, stromal cells, extracellular,endothelium). All these aspects must finally be inte-grated in the mechanisms of tumour initiation(including clonality) and progression (see above).Knowledge of these pathways in each acquired capa-bility is also essential to plan any sensible molecularevaluation of neoplasms: it will allow marker selectionbased on biological features and it will allow preciseselection of surrogate ⁄ secondary markers to validatethe results. In addition, some pathways are mutuallyexclusive [i.e. rat sarcoma (RAS) and B-RAF muta-tions or epidermal growth factor receptor (EGFR) and

RAS analyses] and have to be evaluated simulta-neously.

self-maintained proliferation (f igure 3)

The cell-cycle transition from G1 to S phase is a keyregulatory point in the cell cycle. The G1 ⁄ S cell cyclecheckpoint controls the passage of eukaryotic cells fromthe first ‘gap’ phase (G1) into the DNA synthesis phase(S). Two main cyclin-dependent kinases (CDK) com-plexes, CDK4 ⁄ 6–cyclin D and CDK2–cyclin E, and thetranscription complex that includes retinoblastoma(Rb) and E2F are pivotal in controlling this check-point.31–33 During G1 phase, the Rb–HDAC repressorcomplex binds to the E2F-DP1 transcription factors,inhibiting downstream transcription.34 Phosphoryla-tion of Rb by CDK4 ⁄ 6 and CDK2 dissociates the Rb–repressor complex, permitting transcription of S-phasegenes encoding for proteins that amplify the G1 to Sphase switch and that are required for DNA replication.Many different stimuli exert checkpoint control, includ-ing transforming growth factor (TGF)-b, DNA damage,contact inhibition, replicative senescence, and growthfactor withdrawal. The first four act by inducingmembers of the INK4 or Kip ⁄ Cip families of cell cyclekinase inhibitors.31,32 TGFb additionally inhibits thetranscription of Cdc25A, a phosphatase that activatesCDKs.35,36 Growth factor withdrawal activates GSK3b,which phosphorylates cyclin D, leading to its rapidubiquitylation and proteosomal degradation. Ubiquity-lation, nuclear export and degradation are mechanismscommonly used to rapidly reduce the concentration ofcell-cycle control proteins.

Other pathways acting through Ras, Rac and Rho alsoregulate the G1 to S transition.37,38 Ras regulates cyclinD1 expression to affect the G1 to S transition. Trans-forming forms of Ras or Raf induce cyclin D1 expressionand cause early entry into S phase. Signalling from Rasto Raf to MEK to ERKs induces cyclin D1 expression,allowing cyclin D1 to complex with Cdk4 and Cdk6 andphosphorylate Rb. Rac-1 and PAK appear to inducecyclin D1 expression and induce G1 to S transitionprimarily through activation of nuclear factor (NF)-jB toactivate the cyclin D1 promoter. Rho activates cdk2 andalso inhibits p21 and p27 to induce cyclin D1 andstimulate G1 to S transition. Rho represses p21 expres-sion to block p21 induction by Ras and to allow Ras-induced progression from G1 to S. Cells that lack p21 donot require Rho for Ras to induce cell cycle progressionfrom G1 to S phase. The cooperative action of Ras, Racand Rho to induce cyclin D1 expression is a keycomponent of oncogenic transformation.

6 S J Diaz-Cano


TP53 is a transcription factor whose activity isregulated by phosphorylation.39–41 The function of p53is to keep the cell from progressing through the cellcycle if there is damage to DNA present. It may do thisin multiple ways, from holding the cell at a checkpointuntil repairs can be made to causing the cell to enterapoptosis if the damage cannot be repaired. The criticalrole of p53 is evidenced by the fact that it is mutatedin a very large fraction of tumours from nearly allsources.

Phosphatase and tensin homologue (PTEN) is atumour suppressor gene capable of dephosphorylatingphosphatidylinositol 3,4,5-triphosphate (PIP3), theproduct of phosphatidylinositol 3-kinase. Many ofthe cancer-related mutations have been mapped to thephosphatase catalytic domain, and it has been sug-gested that the phosphatase activity of PTEN is requiredfor its tumour suppressor function.42–44 The activationof PKB ⁄ AKT is regulated in a complex manner viaphosphorylation of AKT by 3-phosphoinositide-depen-dent protein kinase-1 (PDK1) and integrin-linked

kinase (ILK), respectively.45 Inactivation of PTEN willconstitutively activate the PKB ⁄ AKT pathway. Inaddition to its role in regulating the PI3–K ⁄ AKT cellsurvival pathway, PTEN also inhibits growth factor-induced Shc phosphorylation and suppresses themitogen-activated protein (MAP) kinase signallingpathway.

extended cell survival (f igure 4 )

This capability is due to cell-cycle arrest (which isopposed to the capability above), apoptosis dysregula-tion, and replicative lifespan.

Apoptosis dysregulationApoptosis can be triggered by many different stimulithat result in activation of caspase signalling pathwaysfrom extracellular [i.e. tumour necrosis factor (TNF) orFAS pathway] and intracellular (mitochondria) signals.There are also signal pathways (i.e. AKT pathway)regulating these mechanisms.45

Figure 3. Molecular pathways involved in the tumour cell acquired capability of self-maintained proliferation.



TNFR1 is the receptor for TNF-a and will also bindTNF-b.46,47 Upon binding TNF-a, a TNFR1+ cell istriggered to undergo apoptosis by activating theproteolytic caspase cascade that results in the degra-dation of many critical cellular proteins, and byactivating Bid, a Bcl-2 family member, which activatesmitochondria-mediated apoptosis.48 Mitochondriaparticipate in apoptotic signalling pathways throughthe release of mitochondrial proteins [cytochrome c,Apaf-1, Smac ⁄ DIABLO, or apoptosis-inducing factor(AIF)] into the cytoplasm.49–51 Cytochrome c activatesthe protease Apaf-1, which then activates caspase-9and the rest of its pathway. Smac ⁄ DIABLO inhibits IAPproteins that normally interact with caspase-9 toinhibit apoptosis. AIF released into the cytoplasminduces a non-caspase-dependent apoptosis. Apoptosisregulation by Bcl-2 family proteins occurs as familymember complexes enter the mitochondrial membrane,regulating the release of cytochrome c and otherproteins. Activated Bax, a Bcl-2 family member,localizes to the mitochondrial membrane and increase

its permeability, whereas Bcl-2 and Bcl-xL prevent poreformation, blocking apoptosis.

PIP3 conveys signals to the cytoplasm that activatethe kinase PDK1, which in turn activates the kinaseAKT, also known as protein kinase B.45 Proteinsphosphorylated by activated AKT promote cell survival:phosphorylation of Ij-B kinase leads to activation ofthe transcription factor NF-jB to oppose apoptosis;phosphorylation of Bad blocks anti-apoptotic activity topromote cell survival; phosphorylation of caspase 9 orforkhead transcription factors block their apoptosisinduction. AKT promotes cell survival and opposesapoptosis by a variety of routes.52

The characteristic cellular response of apoptosis isthe internucleosomal fragmentation of the nucleargenome to create a DNA ladder pattern, due toactivation of multiple nucleases.53 One nucleaseinvolved in apoptosis is DNA fragmentation factor(DFF), a caspase-activated DNase (CAD).54–56 DFF ⁄CAD is activated through cleavage of its associatedinhibitor ICAD by caspase proteases during apoptosis.

Figure 4. Molecular pathways involved in the tumour cell acquired capability of extended cell survival.

8 S J Diaz-Cano


DFF ⁄ CAD interacts with chromatin components suchas topo II and histone H1 to condense chromatinstructure and perhaps recruit CAD to chromatin.Another apoptosis-activated protease is endonucleaseG (EndoG). EndoG is encoded in the nuclear genome,but is localized to mitochondria in normal cells.57,58

EndoG may play a role in the replication of themitochondrial genome, as well as in apoptosis. Apop-totic signalling causes the release of EndoG frommitochondria. Mitochondria are involved in apoptoticsignalling in other ways as well, through the release ofcytochrome c induced by Bid to activate the caspaseprotease cascade. These pathways are independent,since the EndoG pathway still occurs in cells lackingDFF.

Replicative lifespanTelomeres, which define the ends of chromosomes,consist of short, tandemly repeated DNA sequencesloosely conserved in eukaryotes. Human telomeresconsist of many kilobases of (TTAGGG)n together withvarious associated proteins. Small amounts of theseterminal sequences are lost from the chromosome tipsduring each S phase because of incomplete DNAreplication, but de novo addition of TTAGGG repeatsby the enzyme telomerase compensates for this loss.Many human cells progressively lose terminal baseswith each cell division, a loss that correlates with theapparent absence of telomerase in these cells. There hasbeen considerable interest in the possible relationshipbetween human telomeres and cellular senescence,immortalization, and cancer.59–61 The activation of atelomere maintenance mechanism (TMM) is indispens-able for cellular immortalization, a hallmark of humancancer.

Telomerase is a ribonucleoprotein complex, whichin vitro recognizes a single-stranded G-rich telomereprimer and adds multiple telomeric repeats to its3-prime end by using an RNA template. Telomerasemay also have a role in de novo formation of telomeres.Telomerase has been identified in actively dividing celltypes. The active reverse transcriptase component hasbeen identified in the TERT protein. The presence ofthis factor determines the availability of the telomerasefunction.62 The TERT protein has a high turnover rateand its expression is regulated by factors that promotegrowth (c-MYC, v-k-ras, Bcl-2 and E6) and inhibitingfactors (RB and p53) that promote cell death or thatblock cell division. It appears that the regulation ofactive telomerase has many levels and can be inhib-ited by TEP1 not releasing TERT or by TRF1 whichbinds the end repeats and prevents access to thechromosome ends.63 Additional modulation is due to

phosphorylation by PKC and AKT or dephosphoryla-tion by PP2A.

Some human cancers with complex karyotypes, suchas specific subtypes of soft tissue sarcomas, astrocyticbrain tumours and osteosarcomas, use an alternativelengthening of telomeres (ALT) mechanism as theirTMM.64 Some ALT cells have atypical features, sug-gesting the possibility that there is more than one ALTmechanism. ALT cells are characterized by specificminisatellite instability with stable microsatellites andby high rates of telomeric recombinational exchange.In ALT cells, asymmetrical chromosome segregationand unequal telomeric exchange contribute to telomerelength maintenance. In at least some ALT cells, TMMrequires the integrity of the MRN (MRE11-RAD50-NBS1) recombination complex and is efficientlyrepressed by its sequestration. Microsatellite instability(MSI) often results in disruption of MRN, so ALT mayusually be incompatible with MSI. We suggest that ALTin human tumours is a dysregulated version of anaspect of normal mammalian telomere homeostasis,which may be a vestige of the TMM used by ancienteukaryotes.

genetic instabil ity – dna damage and repair

( F I G U R E 5 )

The accumulation of genetic alterations is the result ofthe balance between DNA damage and repair. Themain damaging signals are dysfunctional telomeres,replication stress, and the reactive oxygen species (ROS,in particular H2O2), which finally target proteasomes,kinases, phosphatases, cytoskeleton, histones, tran-scription factors, telomerase and DNA. The DNAdamage is variable and ranges from base ⁄ nucleotideabnormalities, mismatch, single- and double-strandbreaks. These changes will be detected by sensormechanisms that through kinases and adaptors acti-vate the effector mechanisms, which restore genomicintegrity.65,66 The cellular response to the geneticdamage will be apoptosis (when the damage cannot berepaired), cell-cycle arrest (to provide time for repair)and changes in gene expression. One of the mostimportant consequences of this damage ⁄ repair disbal-ance is tumour genomic instability,40 which has beennormally classified in chromosomal and microsatelliteinstability.67

As regards the signalling pathway, one of the mostimportant systems is the G2 ⁄ M DNA damage check-point that prevents the cell from entering mitosis (Mphase) if the genome is damaged. The Cdc2-cyclin Bkinase is pivotal in regulating this transition. DuringG2 phase, Cdc2 is maintained in an inactive state by



the kinases Wee1 and Mt1. As cells approach Mphase, the phosphatase Cdc25 is activated, perhaps bythe polo-kinase Pik1. Cdc25 then activates Cdc2,establishing a feedback amplification loop that effi-ciently drives the cell into mitosis. DNA damageactivates the DNA-PK ⁄ ATM ⁄ ATR kinases, initiatingtwo parallel cascades that inactivate Cdc2-cyclinB.68,69 The first cascade rapidly inhibits progressioninto mitosis: the CHK kinases phosphorylate andinactivate Cdc25, which can no longer activateCdc2. The second cascade is slower. Phosphorylationof p53 dissociates it from MDM2, activating its DNAbinding activity. Acetylation by p300 ⁄ PCAF furtheractivates its transcriptional activity. The genes thatare turned on by p53 constitute effectors of thissecond cascade. They include 14-3-3s, which binds tothe phosphorylated Cdc2-cyclin B kinase and exportsit from the nucleus; GADD45, which apparently bindsto and dissociates the Cdc2-cyclin B kinase; andp21Cip1, an inhibitor of a subset of the cyclin-dependent kinases including Cdc2 (CDK1).

Breast cancer (BRCA) 1 and BRCA2 are involved inthe cellular response to DNA damage, includingblocking cell cycle progression and inducing DNArepair to preserve the integrity of the genome duringcell division.70 BRCA1 and BRCA2 induce double-stranded repair of breaks using homologous recombi-nation, in part through activation of RAD51. BRCA1acts as a ubiquitin ligase targeting the protein FancD2that activates checkpoint control, integrating the ATMresponse to ionizing radiation and the Fanconi anaemia(FA) response to cross-linking agents such as mitomy-cin C. One member of the FA complex has recently beenidentified as BRCA2. Another related factor involved inthe response of cells to DNA damage is the kinase ATM.Like ATM, ATR serves as a checkpoint kinase that haltscell-cycle progression and induces DNA repair whenDNA is damaged. Loss of ATR results in a loss ofcheckpoint control in response to DNA damage, leadingto cell death, and deletion of the ATR gene in mice isembryonic lethal. ATRIP is a protein that interacts withATR and is a substrate for its kinase activity. ATRIP is

Figure 5. Molecular pathways involved in the tumour cell acquired capability of genetic instability – DNA damage and repair.

10 S J Diaz-Cano


required for ATR function, and removal of ATRIP alsoleads to loss of checkpoint control of the cell cycle. ATRand ATM kinase targets include repair enzymes such asRad51, and the checkpoint kinases Chk1 and Chk2, aswell as BRCA1 and BRCA2.

modif ications in chromatin, transcription

and epigenetic changes (f igure 6 )

Histone and non-histone proteins package the eukary-otic genome to form chromatin, which is compacted innucleosomal arrays. Their assembly into higher-orderchromatin structures creates a highly restrictive envi-ronment for nuclear processes that require access toDNA, resulting in the repression of transcription,replication and repair.41 As counterbalance, a varietyof ATP-dependent chromatin remodelling factors facil-itate the interaction of proteins such as replication andtranscription factors with nucleosomal DNA. ATP-dependent chromatin remodelling complexes are char-

acterized by the presence of an ATPase subunit fromSNF2-like family of the DEAD ⁄ H (SF2) DNA-stimulatedATPases. The highly conservative hSWI ⁄ SNF multi-subunit complexes contain hBRM or BRG1 ATPases,which alter the histone–DNA contacts, enabling accessof general transcription factors to promoter regions.71

Remodelling complexes are targeted to promotersvia interactions with sequence-specific transcriptionfactors.

The chromatin packaging of the genome is dynamic,changing with transcriptional regulation and with thecell cycle, whereas the nuclear matrix network pro-vides structure and regulates chromatin condensation.Chromatin is condensed for chromosome segregationduring mitosis, whereas chromatin is more open fortranscription. Regulated interactions of matrix pro-teins, DNA and other factors in different cell-cyclephases alter the structure and function of chromatin.

Transcription factors that can interact with therepressive chromatin structure and remodel the

Figure 6. Molecular pathways involved in the tumour cell acquired capability of modifications of chromatin, transcription and epigenetic

changes.



chromatin to allow other transcription factors to bind.In addition to nuclear receptors, transcription acti-vators as c-Myc, HSF1, or EBNA2 have been found torecruit SWI ⁄ SNF to specific promoters. Nuclear recep-tors recruit an ATP-dependent remodelling hSWI ⁄ SNFcomplex (BRG1–BAF) to remodel the chromatin.71 Theremodelling process converts the closed conformationof the promoter to an open one without altering thenucleosomal positioning. The remodelling of the pro-moter permits NF1 binding and the assembly of atranscription initiation complex.

Initiation of DNA replication in eukaryotes is ahighly conserved, multi-step process (replication licens-ing) designed to restrict initiation events to once perreplication origin per S phase. Its control has beenuncovered by the discovery of CDKs as master regula-tors of the cell cycle and the initiator proteins ofDNA replication, such as the origin recognition com-plex (ORC), Cdc6 ⁄ 18, Cdt1 and the mini-chromo-some maintenance complex (Mcm). The proteins andthe sequence of events involved in this process areconserved throughout the eukaryotic kingdom. First,the ORC comprising six proteins binds to replicationorigins in the chromosomal DNA. At the end of mitosis,ORC, Cdc6 ⁄ 18 and Cdt1 assist the binding of Mcmproteins 2–7 to chromatin, and chromatin becomeslicensed for replication. The activated Mcm complexfunctions as a replicating helicase and moves alongwith the replication fork to bring the origins to theunlicensed state. The cycling of CDK activity in the cellcycle regulates the two states of replication origins, thelicensed state in G1 phase and the unlicensed state forthe rest of the cell cycle. The restriction on licensing isrelieved when CDK falls off at the completion of mitosisto allow a new round of replication.

Nuclear matrix associated proteins such as AKAP95bind to PKA (cAMP-dependent protein kinase) througha PKA RII regulatory subunit, an interaction thatrequires PKA phosphorylation by Cdk1. PKA activityand cAMP are reduced during entry into mitosis, butrecruited PKA to condensed chromosomes is essentialto maintain the condensed state.72 Other proteinsrecruited during mitotic chromatin condensation suchas Eg7 form part of a multiprotein condensin complex,recruiting another key component of mitotic chroma-tin condensation.73 Modification of the core histonesthrough phosphorylation regulates chromatin conden-sation. Histone H3 interacts with the condensincomplex and is phosphorylated during mitosis. HistoneH3 phosphorylation by Aurora-2 induces chromatincondensation, and dephosphorylation by PP1 pro-motes chromatin decondensation for re-entry intointerphase. Nuclear matrix-associated proteins may

play a role during the regulation of chromatin struc-ture for transcription during interphase also. Theinteraction with the p68 RNA helicase recruits thisenzyme to the nuclear matrix during interphase.74

Other nuclear RNA helicases interact with transcrip-tion factors and cofactors, suggesting that the p68RNA helicase also may regulate interactions of tran-scription complexes.

epigenetics

Tumorigenesis is known to be a multistep process thataccumulates defects in various cancer genes. Epigeneticmodifications, most importantly DNA methylationevents, are frequently involved in transcriptionalchanges in both tumour suppressor genes and onco-genes.75 DNA methylation of gene promoter regions(generally at CpG dinucleotides) is generally correlatedwith gene silencing due to two underlying mech-anisms.41 First, binding of transcription factors orenhancer blocking elements, such as CTCF, may beinhibited by DNA methylation. The second and moregeneral mechanism involves proteins that detectmethylated DNA through methyl CpG-binding domains(MBDs). These proteins mediate recruitment of repres-sor complexes that include histone deacetylases(HDACs). HDACs remove acetyl groups from lysineresidues of histones H3 and H4, which results incondensation of chromatin and thus limits access oftranscription factors to promoter regions of geneslocalized nearby. Co-repressor complexes specificallybind methylated DNA, and copurify with the Sin3A/HDAC corepressor complex. The two main proteincomplexes share four polypeptides (HDAC1, HDAC2,RbAp46 and RbAp48) and contain unique polypep-tides (Sin3A, SAP30 and SAP18 in the Sin3 complex,and Mi2, metastasis-associated protein (MTA) 1, MTA2and MBD3 in the NuRD complex). The NuRD complexpossesses nucleosome-remodelling activity because ofthe presence of Mi2, a member of the SWI2 ⁄ SNF2helicase ⁄ ATPase family.76 This complex preferentiallybinds, remodels and deacetylates methylated nucleo-somes. MTA1 or MTA2 expression levels are ele-vated in metastatic cancer cells; MTA2 modulates theenzymatic activity of the histone deacetylase corecomplex.77,78

Transcription repression is also related to nuclearreceptors (such as RXR and RAR). Once retinoic acidbinds the receptors, a receptor conformational changecauses the dissociation of the corepressors and thebinding of coactivators with histone acetylase activity.Following ligand binding by the heterodimer, thereceptors and proteins in the basal transcription

12 S J Diaz-Cano


machinery (such as TBP and TAF135) are degraded bythe proteasome.

mobil ization of cellular resources (f igures 7

and 8)

Neoplastic transformation requires adequate machin-ery for protein synthesis and the provision of energy forthe active process of signalling and control of theacquired cell capabilities. In this sense, protein synthe-sis–ribosomes (Figure 7) and mitochondria (Figure 8)are going to play a central role in neoplastic transfor-mation.

Protein synthesis–ribosomesmTOR and MAP kinase signalling pathways modulatethe phosphorylation of translation factors, and theassociation of RNA-binding proteins with specificmRNAs.79,80 These effects contribute both to overallcontrol of protein synthesis and to modulation of thetranslation or stability of specific mRNAs. eIF4E plays

an important role in human cancers, being expressedat high levels in many cancers, which correlatepositively with tumour aggressiveness. eIF4E repressesapoptosis and promotes the translation of the mRNAsfor proteins with roles in cell-cycle progression (e.g.cyclin D1), cell transformation (e.g. ornithine decar-boxylase), tumour vascularization (e.g. VEGF) ormetastasis [e.g. matrix metalloproteinase (MMP)-9].eIF4E increases the cytoplasmic levels of the cyclinD1mRNA, apparently involving a feature in the 3¢-UTRof the cyclin D1 mRNA, rather than the cap-bindingfunction of eIF4E, which is thought to allow eIF4E topromote transport of the cyclin D1 mRNA from thenucleus to the cytoplasm. Such a function wouldrequire eIF4E to spend some time in the nucleus, andindeed a proportion of the total cellular eIF4E is foundin the nuclear fraction.

The existence of the S6K-based negative-feedbackloop means that mTOR blocking will probably lead toenhanced PI(3)K–AKT activation in some tumours.Depending on other mutations found in the tumour,

Figure 7. Molecular pathways involved in the tumour cell acquired capability of mobilization of cellular resources – ribosomes (protein

synthesis).



this hyperactivation of PI(3)K–AKT signalling couldmake the tumour more aggressive. The critical func-tion of the mTORC1 complex required for cell growth isactivation of eIF4E. Because eIF4E transcription iscontrolled by multiple signalling inputs, resistance torapamycin might develop by a variety of paths that willdepend on the genetic background of the tumour.

A critical target for tumour cell growth and survivalis the activation of eIF4E and hypoxia inducible factor(HIF). Tumours with initiating mutations in RTKs, Rasor Raf have multiple routes to signal to eIF4E and HIF.In contrast, tumours with initiating lesions in PI(3)K ormore direct regulators of mTOR (such as LKB1 andTSC) do not have alternative routes to activate eIF4Eand HIF. Hence, these tumours show greater responseto rapamycin. Similarly, the expression and use ofspecific adaptor proteins that enhance certain armsof pathway signalling will dictate the therapeuticresponse.

More broadly, identifying the primary mechanismsby which the PI(3)K pathway is activated in a given

tumour should facilitate the choice of potentialinterventions. For tumours bearing Ras mutations,defining the critical downstream effector in eachgiven tissue context is essential. In some settingswhere Raf has been assumed to be critical, itseffectors must be evaluated: Raf mutations are com-mon and mutually exclusive with N-ras mutations orloss of PTEN. So in these tissues testing these criticaleffectors may be critical in tumorigenesis. Conversely,there is also clear evidence for both synergy andredundancy between Raf and PI(3)K-mediated signal-ling on specific biochemical effectors. Activation ofmTORC1 activity can be mediated by distinct phos-phorylation events on tuberin by ERK and RSK, aswell as by AKT. The cell-death effector BAD is alsoinhibited by both of these pathways. In addition,given the close similarity of the RSK and AKT kinasedomains, it is possible that both can phosphorylatethe same sites on some target proteins, such as GSK-3or even tuberin. The availability of eIF4E controlledby signalling through mTOR provides one probable

Figure 8. Molecular pathways involved in the tumour cell acquired capability of mobilization of cellular resources – mitochondria

(ATP synthesis, apoptosis, reactive oxygen species and damage).

14 S J Diaz-Cano


link between mTOR signalling and tumorigenesis ⁄ cellproliferation.81,82

ATP synthesis–mitochondriaMutations in nuclear DNA or mtDNA OXPHOS genesthat impede electron flow increase mitochondrial ROSproduction.83–87 The resulting H2O2, which is rela-tively stable, diffuses out of the mitochondrion, throughthe cytosol and into the nucleus, activating normallyinactive genes and inducing nuclear DNA mutations.The resulting nuclear DNA damage activates thenuclear DNA repair systems, including the polyADP-ribose polymerase (PARP). The activated PARP de-grades the nuclear nicotinamide adenine dinucleotide(NAD)+ in the process of adding poly ADP-ribosechains to histones and other nuclear proteins.

The degradation of the nuclear NAD+, together withthe high NADH ⁄ NAD+ ratio, inactivates histonedeacetylase, and nuclear transcription is repressed bythe deacetylation of histones and activated by histoneacetylation. Therefore, histone acetylation turns on thetranscription of normally inactive genes. In post-mitotic tissues, histone acetylation permits the activa-tion of the genes that regulate cell replication anddifferentiation, the proto-oncogenes. Nuclear H2O2 canmutate proto-oncogenes, converting them into func-tional oncogenes. Moreover, increased cytosolic andnuclear H2O2 activates a variety of cellular signaltransduction factors, including NF-jB, apurinic ⁄ apyri-midinic endonuclease (APE)-1, Fos, Jun and tyrosinekinases. This drives the cell into replication. Con-sequently, mutations in mitochondrial genes thatinhibit electron flow result in chronically increasedmitochondrial ROS production, which can act as both atumour initiator (mutation of proto-oncogenes) andtumour promoter (activation of transcription andreplication).85,88,89

Gene defects in certain nuclear DNA-encoded mito-chondrial genes have been directly linked to somehereditary cancers.83–87 Mutations in mitochondrialfumarate hydratase have been associated with uterineleiomyomas and renal cell carcinomas, and mutationsin three of the four nuclear DNA-encoded subunits ofcomplex II have been linked to paragangliomas: SDHD,SDHC and SDHB subunit. In addition, mutations in theSDHB subunit are also associated with pheochromo-cytoma and early-onset renal cell carcinoma. Thestriking difference in clinical effects of SDHA subunitmutations versus SDHB, C and D mutations stronglyimplicates mitochondrial ROS production in the aetiol-ogy of cancer. Transformation of certain tumours withthe MnSOD cDNA can reverse the malignant pheno-type, and a cluster of three mutations in the MnSOD

gene promoter region that alter AP-2 binding andpromoter efficiency is found in a number of tumours.Moreover, ROS production in association with theinactivation of p16ink4a has been hypothesized to beone of the two main mechanisms for tumorigenesis; theother is p53 deficiency.

Mitochondrial DNA mutations that inhibit OXPHOSand impede electron flow should increase ROS produc-tion and contribute to cancer.83–87 Mitochondrial ROScould contribute to neoplastic transformation, both as atumour initiator by causing nuclear DNA mutations inproto-oncogenes and tumour-suppressor genes, and asa tumour promoter through driving cellular prolifera-tion. At low levels, ROS has been found to be an activemitogen, thought to act though interaction withvarious kinases (Src kinase, protein kinase C, MAPK,and receptor tyrosine kinases), as well as with differenttranscription factors (Fos, Jun, NF-jB). Furthermore,the dual-function APE-1 functions not only in the DNAbase excision pathway but also in the redox regulationof the transcription factors Fos, Jun, NF-jB, PAX,HIF-1a and p53.

Moreover, the role of mitochondrial defects in thepathophysiology of cancer would appear to be thegeneration of increased ROS, which acts as both anuclear DNA mutagen and cellular mitogen. Sincemitochondrial mutations that increase ROS wouldresult in the accumulation within the cell of unoxidizedNADH and pyruvate, the excess NADH and pyruvatewould then be converted to lactate by lactate dehydro-genase, also described as aerobic-glycolysis. Mitochon-drial damage is the result of nuclear or mitochondrialDNA damage that affects the electron transport chainand ATP production, which express morphologically asmitochondrial hyperplasia (oncocytic changes ⁄ meta-plasia).90–92 In tumour pathology, the timing betweenthe neoplastic-inducing damage and the mitochondrialdamage is important. Mitochondrial damage precedingneoplastic-inducing damage will result in oncocyticneoplasms, whereas the opposite scenario gives neo-plasms with oncocytic differentiation (normally focal).

modif ied microenvironment interactions –

invasion and angiogenesis ( F I G U R E S 1 A N D 2 )

Cellular transformation is accompanied by many cel-lular changes, including loss of the differentiated cellmorphology and invasion of the extracellular matrix.These processes are dependent on cellular and stromalinteractions and on extracellular matrix degradation.20

Cellular interactions responsible for cell–cell adhe-sion also can communicate signals, often involvinginteractions with cytoskeletal elements, to produce



changes in cell motility, migration, proliferation andshape.21,93,94 The cadherins are cell-surface adhesionmolecules that help form tight junctions between cells,such as formation of epithelial cell layers. In addition tomediating adhesion with other cells, cadherins trans-duce signals into cells through interactions with thecatenins.23 Catenins probably affect actin cytoskeletalfunction through interactions with proteins such asactinin and vinculin. Catenins also probably triggerchanges in cell shape and motility with signals throughthe Rho small GTPases. Paxillin acts as an adaptorprotein between proteins involved in adhesion signal-ling such as FAK and src and cytoskeletal elements. Inaddition to signals created by adhesion molecules toalter cellular responses, other signalling pathways canalter adhesion through components of the focal adhe-sion complex.

Extracellular matrix interactions depend on integrins,cell-surface receptors that mediate intracellular signalsto control cellular shape, motility and progressionthrough the cell cycle.95,96 Integrins do not themselvespossess a kinase domain or enzymatic activity, but relyon association with other signalling molecules totransmit signals. Interactions between the extracellularmatrix and the actin cytoskeleton commonly take placeat focal adhesions on the cell surface that containlocalized concentrations of integrins, signalling mole-cules and cytoskeletal elements. Talin forms a directinteraction with the integrin cytoplasmic domain, andinteracts with cytoskeletal elements (actin) and signal-ling factors. Paxillin and Crk-associated substrate (CAS)also localize in focal adhesions and may serve as ascaffold for other integrin signalling components suchas FAK and src. Interaction of FAK, CAS and src may berequired for integrin regulation of cell-cycle progression.The CrkL adaptor protein may regulate downstreamintegrin signalling. Growth factor signalling pathwaysand the caveolin receptor exhibit important crosstalkwith integrin receptors in cellular responses such asactivation of map kinase, proliferation and motility.PTEN also interact with FAK, a key molecule implicatedin integrin signalling pathways, and it directly dep-hosphorylates tyrosine-phosphorylated FAK. PTENdown-regulation of p130CAS through FAK results ininhibition of cell migration and spreading.

Extracellular matrix degradation is a key componentof tumour cell invasion into surrounding tissues. MMPsare a class of proteases secreted by tumour cells,degrading extracellular matrix proteins and allowingmetastasis.97,98 RECK is a membrane-anchored inhib-itor of MMPs, inhibiting MMP-2, MMP-9 and MT1-MMP.99 The processing of MMPs to their active formoccurs at the plasma membrane, making the localiza-

tion of RECK at the membrane a key to its potentactivity as an inhibitor of MMP activity.100 Solublesecreted MMP inhibitors have also been identified,TIMPs, which appear to be less active at inhibitingMMPs and even perhaps to be essential for MMPmaturation. The inhibition of MMPs inhibits tissueinvasion, metastasis and tumour angiogenesis, and isessential for angiogenesis, inflammation and normaldevelopment. RECK expression is inhibited by ras,suggesting one component by which ras inducestransformation. High levels of RECK expression intumours is correlated with cancer patient survival, andoverexpression of RECK may offer a therapeutic strat-egy for the control of cancer.

Conclusions

The application of these concepts in oncologicalpathology leads to consideration of the molecular testrequirements (Molecular Test Score System) for areliable implementation, which covers biological effects(1–3), molecular pathway (4, 5), biological validation(6–8) and technical validation (9, 10).1 As tumours are the result of multiple and coopera-tive genetic abnormalities, tests should be assessingmore than one acquired capability.2 Genetic targets are more likely to be useful if they areinvolved in more than one acquired capability.3 Selected targets must be testing both initiation(including tests for clonality and fractional allelic loss)and promotion ⁄ progression of tumours in a givenlocation.4 Molecular pathways are redundant and overlap inseveral markers that will potentially be the mostinformative.5 Molecular markers can be pleiotropic and theireffects will be observed in different pathways (biologicalconsequences).6 The effects of each marker ⁄ target must be evaluatedin view of the pathway (both upstream and down-stream markers), including the appropriate surrogatemarkers to validate the expected results.7 Markers from mutually exclusive pathways must betested simultaneously for consistency of results.8 Molecular tumour assessment must be based on theselection of sensible markers, single tests being gener-ally insufficient.9 Tumours must be tested using samples from at leasttwo areas that should be covering any topographicheterogeneity demonstrated in neoplasms.10 Samples should be run in duplicate to checkconsistency of results and must include positive,negative and sensitivity controls.

16 S J Diaz-Cano


Competing interests

None to declare.

References

1. Hanahan D, Weinberg RA. The hallmarks of cancer. Cell 2000;

100; 57–70.

2. Blanes A, Diaz-Cano SJ. DNA and kinetic heterogeneity during

the clonal evolution of adrenocortical proliferative lesions. Hum.

Pathol. 2006; 37; 1295–1303.

3. Blanes A, Rubio J, Martinez A, Wolfe HJ, Diaz-Cano SJ. Kinetic

profiles by topographic compartments in muscle-invasive tran-

sitional cell carcinomas of the bladder: role of TP53 and NF1

genes. Am. J. Clin. Pathol. 2002; 118; 93–100.

4. Diaz-Cano SJ, Blanes A, Wolfe HJ. PCR techniques for clonality

assays. Diagn. Mol. Pathol. 2001; 10; 24–33.

5. Diaz-Cano SJ, de Miguel M, Blanes A, Tashjian R, Galera H,

Wolfe HJ. Clonality as expression of distinctive cell kinetics

patterns in nodular hyperplasias and adenomas of the adrenal

cortex. Am. J. Pathol. 2000; 156; 311–319.

6. Pozo-Garcia L, Diaz-Cano SJ. Clonal origin and expansions in

neoplasms: biologic and technical aspects must be considered

together. Am. J. Pathol. 2003; 162; 353–354; author reply

354–355.

7. Kinzler KW, Vogelstein B. Lessons from hereditary colorectal

cancer. Cell 1996; 87; 159–170.

8. Diaz-Cano SJ, de Miguel M, Blanes A, Tashjian R, Wolfe HJ.

Germline RET 634 mutation positive MEN 2A-related C-cell

hyperplasias have genetic features consistent with intraepithe-

lial neoplasia. J. Clin. Endocrinol. Metab. 2001; 86; 3948–3957.

9. Mears L, Diaz-Cano SJ. Difference between familial and sporadic

medullary thyroid carcinomas. Am. J. Surg. Pathol. 2003; 27;

266–267.

10. Alizadeh AA, Eisen MB, Davis RE et al. Distinct types of diffuse

large B-cell lymphoma identified by gene expression profiling.

Nature 2000; 403; 503–511.

11. Bittner M, Meltzer P, Chen Y et al. Molecular classification of

cutaneous malignant melanoma by gene expression profiling.

Nature 2000; 406; 536–540.

12. Dhanasekaran SM, Barrette TR, Ghosh D et al. Delineation of

prognostic biomarkers in prostate cancer. Nature 2001; 412;

822–826.

13. Ahr A, Holtrich U, Solbach C et al. Molecular classification of

breast cancer patients by gene expression profiling. J. Pathol.

2001; 195; 312–320.

14. Alizadeh AA, Ross DT, Perou CM, van de Rijn M. Towards a

novel classification of human malignancies based on gene

expression patterns. J. Pathol. 2001; 195; 41–52.

15. Carvajal-Carmona LG, Howarth KM, Lockett M et al. Molecular

classification and genetic pathways in hyperplastic polyposis

syndrome. J. Pathol. 2007; 212; 378–385.

16. Harris AL. Hypoxia – a key regulatory factor in tumour growth.

Nat. Rev. Cancer 2002; 2; 38–47.

17. Li F, Tiede B, Massague J, Kang Y. Beyond tumorigenesis: cancer

stem cells in metastasis. Cell Res. 2007; 17; 3–14.

18. Keith B, Simon MC. Hypoxia-inducible factors, stem cells, and

cancer. Cell 2007; 129; 465–472.

19. Semenza GL. HIF-1, O(2), and the 3 PHDs: how animal cells

signal hypoxia to the nucleus. Cell 2001; 107; 1–3.

20. Gupta GP, Massague J. Cancer metastasis: building a frame-

work. Cell 2006; 127; 679–695.

21. Kang Y, Massague J. Epithelial–mesenchymal transitions: twist

in development and metastasis. Cell 2004; 118; 277–279.

22. Brabletz T, Jung A, Spaderna S, Hlubek F, Kirchner T.

Opinion: migrating cancer stem cells – an integrated concept

of malignant tumour progression. Nat. Rev. Cancer 2005; 5;

744–749.

23. Nguyen DX, Massague J. Genetic determinants of cancer

metastasis. Nat. Rev. Genet. 2007; 8; 341–352.

24. Diaz-Cano SJ, Blanes A, Rubio J, Matilla A, Wolfe HJ. Molecular

evolution and intratumor heterogeneity by topographic com-

partments in muscle-invasive transitional cell carcinoma of the

urinary bladder. Lab. Invest. 2000; 80; 279–289.

25. Blanes A, Sanchez-Carrillo JJ, Diaz-Cano SJ. Topographic

molecular profile of pheochromocytomas: role of somatic

down-regulation of mismatch repair. J. Clin. Endocrinol. Metab.

2006; 91; 1150–1158.

26. Arif S, Patel J, Blanes A, Diaz-Cano SJ. Cytoarchitectural and

kinetic features in the histological evaluation of follicular

thyroid neoplasms. Histopathology 2007; 50; 750–763.

27. Diaz-Cano SJ. Kinetic topographical heterogeneity in follicular

thyroid neoplasms and growth patterns. Histopathology 2007;

51; 416–418.

28. Pozo L, Sanchez-Carrillo J, Martinez A, Blanes A, Diaz-Cano S.

Differential kinetic features by tumour topography in cutaneous

small-cell neuroendocrine (Merkel cell) carcinomas. J. Eur. Acad.

Dermatol. Venereol. 2007; 21; 1220–1228.

29. Rubio J, Blanes A, Sanchez-Carrillo JJ, Diaz-Cano SJ. Microsat-

ellite abnormalities and somatic down-regulation of mismatch

repair characterize nodular-trabecular muscle-invasive urothe-

lial carcinoma of the bladder. Histopathology 2007; 51; 458–

467.

30. Kinzler KW, Vogelstein B. Landscaping the cancer terrain.

Science 1998; 280; 1036–1037.

31. Kaldis P. Another piece of the p27Kip1 puzzle. Cell 2007; 128;

241–244.

32. Kim WY, Sharpless NE. The regulation of INK4 ⁄ ARF in cancer

and aging. Cell 2006; 127; 265–275.

33. Campisi J. Senescent cells, tumor suppression, and organismal

aging: good citizens, bad neighbors. Cell 2005; 120; 513–522.

34. Sherr CJ. Cancer cell cycles. Science 1996; 274; 1672–1677.

35. Derynck R, Zhang Y, Feng XH. Smads: transcriptional activators

of TGF-beta responses. Cell 1998; 95; 737–740.

36. Shi Y, Massague J. Mechanisms of TGF-beta signaling from cell

membrane to the nucleus. Cell 2003; 113; 685–700.

37. Bar-Sagi D, Hall A. Ras and Rho GTPases: a family reunion.

Cell 2000; 103; 227–238.

38. Yaswen P, Campisi J. Oncogene-induced senescence pathways

weave an intricate tapestry. Cell 2007; 128; 233–234.

39. Levine AJ. p53, the cellular gatekeeper for growth and division.

Cell 1997; 88; 323–331.

40. Lombard DB, Chua KF, Mostoslavsky R, Franco S, Gostissa M,

Alt FW. DNA repair, genome stability, and aging. Cell 2005;

120; 497–512.

41. Sherr CJ. Principles of tumor suppression. Cell 2004; 116;

235–246.

42. Baker SJ. PTEN enters the nuclear age. Cell 2007; 128; 25–28.

43. Di Cristofano A, Pandolfi PP. The multiple roles of PTEN in

tumor suppression. Cell 2000; 100; 387–390.

44. Rossi DJ, Weissman IL. Pten, tumorigenesis, and stem cell self-

renewal. Cell 2006; 125; 229–231.

45. Manning BD, Cantley LC. AKT ⁄ PKB signaling: navigating

downstream. Cell 2007; 129; 1261–1274.



46. Locksley RM, Killeen N, Lenardo MJ. The TNF and TNF receptor

superfamilies: integrating mammalian biology. Cell 2001; 104;

487–501.

47. Varfolomeev EE, Ashkenazi A. Tumor necrosis factor: an

apoptosis JuNKie? Cell 2004; 116; 491–497.

48. Harris MH, Thompson CB. The role of the Bcl-2 family in the

regulation of outer mitochondrial membrane permeability.

Cell Death Differ. 2000; 7; 1182–1191.

49. Joza N, Susin SA, Daugas E et al. Essential role of the

mitochondrial apoptosis-inducing factor in programmed cell

death. Nature 2001; 410; 549–554.

50. Shi Y. A structural view of mitochondria-mediated apoptosis.

Nat. Struct. Biol. 2001; 8; 394–401.

51. Srinivasula SM, Hegde R, Saleh A et al. A conserved XIAP-

interaction motif in caspase-9 and Smac ⁄ DIABLO regulates

caspase activity and apoptosis. Nature 2001; 410; 112–116.

52. Brazil DP, Park J, Hemmings BA. PKB binding proteins. Getting

in on the Akt. Cell 2002; 111; 293–303.

53. Zhang J, Xu M. Apoptotic DNA fragmentation and tissue

homeostasis. Trends Cell Biol. 2002; 12; 84–89.

54. Enari M, Sakahira H, Yokoyama H, Okawa K, Iwamatsu A,

Nagata S. A caspase-activated DNase that degrades DNA during

apoptosis, and its inhibitor ICAD. Nature 1998; 391; 43–50.

55. Sakahira H, Enari M, Nagata S. Cleavage of CAD inhibitor in

CAD activation and DNA degradation during apoptosis. Nature

1998; 391; 96–99.

56. Durrieu F, Samejima K, Fortune JM, Kandels-Lewis S, Osheroff

N, Earnshaw WC. DNA topoisomerase IIalpha interacts with

CAD nuclease and is involved in chromatin condensation during

apoptotic execution. Curr. Biol. 2000; 10; 923–926.

57. Li LY, Luo X, Wang X. Endonuclease G is an apoptotic DNase

when released from mitochondria. Nature 2001; 412; 95–99.

58. Parrish J, Li L, Klotz K, Ledwich D, Wang X, Xue D.

Mitochondrial endonuclease G is important for apoptosis in C.

elegans. Nature 2001; 412; 90–94.

59. Chang S, DePinho RA. Telomerase extracurricular activities.

Proc. Natl Acad Sci USA 2002; 99; 12520–12522.

60. Shay JW, Zou Y, Hiyama E, Wright WE. Telomerase and cancer.

Hum. Mol. Genet. 2001; 10; 677–685.

61. Wang J, Hannon GJ, Beach DH. Risky immortalization by

telomerase. Nature 2000; 405; 755–756.

62. Stewart SA, Hahn WC, O’Connor BF et al. Telomerase contrib-

utes to tumorigenesis by a telomere length-independent mech-

anism. Proc. Natl Acad Sci USA 2002; 99; 12606–12611.

63. Okabe J, Eguchi A, Masago A, Hayakawa T, Nakanishi M. TRF1 is

a critical trans-acting factor required for de novo telomere

formation in human cells. Hum. Mol. Genet. 2000; 9; 2639–2650.

64. Muntoni A, Reddel RR. The first molecular details of ALT in

human tumor cells. Hum. Mol. Genet. 2005; 2; R191–R196.

65. Rouse J, Jackson SP. Interfaces between the detection, signaling,

and repair of DNA damage. Science 2002; 297; 547–551.

66. Zhou BB, Elledge SJ. The DNA damage response: putting

checkpoints in perspective. Nature 2000; 408; 433–439.

67. Blanes A, Diaz-Cano SJ. Complementary analysis of microsatel-

lite tumor profile and mismatch repair defects in colorectal

carcinomas. World J. Gastroenterol. 2006; 12; 5932–5940.

68. Cortez D, Guntuku S, Qin J, Elledge SJ. ATR and ATRIP: partners

in checkpoint signaling. Science 2001; 294; 1713–1716.

69. Taniguchi T, Garcia-Higuera I, Xu B et al. Convergence of the

fanconi anemia and ataxia telangiectasia signaling pathways.

Cell 2002; 109; 459–472.

70. Venkitaraman AR. Cancer susceptibility and the functions of

BRCA1 and BRCA2. Cell 2002; 108; 171–182.

71. Gibbons RJ. Histone modifying and chromatin remodelling

enzymes in cancer and dysplastic syndromes. Hum. Mol. Genet.

2005; 14 Spec No 1; R85–R92.

72. Murnion ME, Adams RR, Callister DM, Allis CD, Earnshaw WC,

Swedlow JR. Chromatin-associated protein phosphatase 1 reg-

ulates aurora-B and histone H3 phosphorylation. J. Biol. Chem.

2001; 276; 26656–26665.

73. Ball AR Jr, Schmiesing JA, Zhou C et al. Identification of a

chromosome-targeting domain in the human condensin subunit

CNAP1 ⁄ hCAP-D2 ⁄ Eg7. Mol. Cell. Biol. 2002; 22; 5769–5781.

74. Akileswaran L, Taraska JW, Sayer JA, Gettemy JM, Coghlan VM.

A-kinase-anchoring protein AKAP95 is targeted to the nuclear

matrix and associates with p68 RNA helicase. J. Biol. Chem.

2001; 276; 17448–17454.

75. Plass C. Cancer epigenomics. Hum. Mol. Genet. 2002; 11;

2479–2488.

76. Dilworth FJ, Chambon P. Nuclear receptors coordinate the

activities of chromatin remodeling complexes and coactivators

to facilitate initiation of transcription. Oncogene 2001; 20;

3047–3054.

77. Feng Q, Zhang Y. The MeCP1 complex represses transcription

through preferential binding, remodeling, and deacetylating

methylated nucleosomes. Genes Dev. 2001; 15; 827–832.

78. Nan X, Ng HH, Johnson CA et al. Transcriptional repression by

the methyl-CpG-binding protein MeCP2 involves a histone

deacetylase complex. Nature 1998; 393; 386–389.

79. Schmelzle T, Hall MN. TOR, a central controller of cell growth.

Cell 2000; 103; 253–262.

80. Proud CG. Signalling to translation: how signal transduction

pathways control the protein synthetic machinery. Biochem. J.

2007; 403; 217–234.

81. Shaw RJ, Cantley LC. Ras, PI(3)K and mTOR signalling controls

tumour cell growth. Nature 2006; 441; 424–430.

82. West MJ, Stoneley M, Willis AE. Translational induction of the

c-myc oncogene via activation of the FRAP ⁄ TOR signalling

pathway. Oncogene 1998; 17; 769–780.

83. Broker LE, Kruyt FA, Giaccone G. Cell death independent of

caspases: a review. Clin. Cancer Res. 2005; 11; 3155–3162.

84. Henry-Mowatt J, Dive C, Martinou JC, James D. Role of

mitochondrial membrane permeabilization in apoptosis and

cancer. Oncogene 2004; 23; 2850–2860.

85. Modica-Napolitano JS, Kulawiec M, Singh KK. Mitochondria

and human cancer. Curr. Mol. Med. 2007; 7; 121–131.

86. Singh KK. Mitochondrial dysfunction is a common phenotype in

aging and cancer. Ann. NY Acad. Sci. 2004; 1019; 260–264.

87. Wallace DC. A mitochondrial paradigm of metabolic and

degenerative diseases, aging, and cancer: a dawn for evolution-

ary medicine. Annu. Rev. Genet. 2005; 39; 359–407.

88. Irminger-Finger I. Science of cancer and aging. J. Clin. Oncol.

2007; 25; 1844–1851.

89. Neuzil J, Stantic M, Zobalova R et al. Tumour-initiating cells vs.

cancer ‘stem’ cells and CD133: What’s in the name? Biochem.

Biophys. Res. Commun. 2007; 355; 855–859.

90. Maximo V, Botelho T, Capela J et al. Somatic and germline

mutation in GRIM-19, a dual function gene involved in

mitochondrial metabolism and cell death, is linked to mito-

chondrion-rich (Hurthle cell) tumours of the thyroid. Br. J.

Cancer 2005; 92; 1892–1898.

91. Maximo V, Soares P, Lima J, Cameselle-Teijeiro J, Sobrinho-

Simoes M. Mitochondrial DNA somatic mutations (point muta-

tions and large deletions) and mitochondrial DNA variants in

human thyroid pathology: a study with emphasis on Hurthle

cell tumors. Am. J. Pathol. 2002; 160; 1857–1865.

18 S J Diaz-Cano


92. Sobrinho-Simoes M, Maximo V, Castro IV et al. Hurthle

(oncocytic) cell tumors of thyroid: etiopathogenesis, diagnosis

and clinical significance. Int. J. Surg. Pathol. 2005; 13; 29–35.

93. Arias AM. Epithelial mesenchymal interactions in cancer and

development. Cell 2001; 105; 425–431.

94. Gumbiner BM. Cell adhesion: the molecular basis of tissue

architecture and morphogenesis. Cell 1996; 84; 345–357.

95. Muthuswamy SK. ErbB2 makes beta 4 integrin an accomplice in

tumorigenesis. Cell 2006; 126; 443–445.

96. Burridge K, Wennerberg K. Rho and Rac take center stage.

Cell 2004; 116; 167–179.

97. Blobel CP. Metalloprotease-disintegrins: links to cell adhesion

and cleavage of TNF alpha and Notch. Cell 1997; 90; 589–592.

98. Montell DJ. Anchors away! Fos fosters anchor-cell invasion.

Cell 2005; 121; 816–817.

99. Oh J, Takahashi R, Kondo S et al. The membrane-anchored

MMP inhibitor RECK is a key regulator of extracellular matrix

integrity and angiogenesis. Cell 2001; 107; 789–800.

100. Takahashi C, Sheng Z, Horan TP et al. Regulation of matrix

metalloproteinase-9 and inhibition of tumor invasion by the

membrane-anchored glycoprotein RECK. Proc. Natl Acad Sci

USA 1998; 95; 13221–13226.



research project-01

Documents

diazcano sj

j pathol

hum pathol

diazcanopcr techniques

diazcanoare pcr artifacts

clonality assays

technical aspects

peerreviewed articlestheme