Progress in pathology

a

bingiver

Keywords:

microsatellite stable (MSS) carcinomas (which represent both

Human Pathology (2010) xx, xxxxxx Department of Pathology, Fairview Southdale Hospital.1. Introduction

Colorectal carcinoma is a major health issue in the UnitedStates, representing the second most commonly fatalmalignancy after lung. Although it is often assumed thatalmost all colorectal carcinomas arise from conventionaladenomas via the suppressor pathway initiated with amutation of the APC gene (the Fearon-Vogelstein model),it is now clear that this pathway accounts for onlyapproximately 60% of colon carcinoma [1,2]. Most of the

remaining 40% is accounted for by the more recentlydescribed serrated pathway leading to CpG island methylatedphenotype (CIMP+) carcinoma (approximately 35%), withthe remaining 5% arising via the mutator pathway in Lynchsyndrome [2]. Trying to understand the multiplicity ofpathways is confused by a plethora of overlapping ways todescribe these pathways and cancers. Pathways have beendefined by the presumed initiating factor (ie, suppressor[chromosome instability] versus mutator [microsatelliteinstability] pathways) and by the presumed precursor lesions(ie, conventional adenoma-carcinoma sequence versus theserrated pathway). A commonly utilized subdivision into0dColon adenocarcinoma;Serrated pathway;Sessile serrated adenoma;Traditional serratedadenoma;

Hyperplastic polyp

Summary Adenocarcinoma of the large intestine can no longer be considered one disease but rather afamily of diseases with different precursor lesions, different molecular pathways, and different end-stagecarcinomas with varying prognoses. Approximately 60% of colorectal carcinomas arise fromconventional adenomas via the suppressor pathway leading to microsatellite stable carcinomas. Thesecarcinomas represent the pathway that has been the target of screening and prevention programs to date.However, approximately 35% of carcinomas arise along the serrated pathway developing from theprecursor lesion known as the sessile serrated adenoma (also referred to as the sessile serrated polyp).Sessile serrated adenomas/polyps lead to carcinomas with extensive CpG island promoter methylation(CpG island methylated phenotype positive carcinomas), which can be either microsatellite instable highor microsatellite stable. The remaining 5% of carcinomas arise from conventional adenomas in patientswith germ line mutations of mismatch repair genes (Lynch syndrome), leading to CpG islandmethylated phenotype negative microsatellite instable carcinomas. Carcinomas arising from sessileserrated adenomas/polyps are not prevented by removing conventional adenomas and hence may bemissed in routine screening programs. In addition, a subset of these lesions may potentially progressrapidly to carcinoma; hence, it is likely that these lesions will require a different screening strategy fromthat used for conventional adenomas. This article reviews the various pathways to colorectal carcinomawith emphasis on the serrated pathway and evaluates the implications of this pathway for colorectalcarcinomas screening programs. 2010 Published by Elsevier Inc.Received 7 May 2010; revised 11 June 2010; accepted 16 June 2010Update on the serrated pathwDale C. Snover MD

Department of Pathology, Fairview Southdale Hospital, HibDepartment of Laboratory Medicine and Pathology, The UnE-mail address: snoverd@umn.edu.

046-8177/$ see front matter 2010 Published by Elsevier Inc.oi:10.1016/j.humpath.2010.06.002y to colorectal carcinoma

, MN 55746, USAsity of Minnesota Medical School, Edina, MN 55435, USA

www.elsevier.com/locate/humpathsuppressor pathway carcinomas arising either sporadically or

in familial adenomatous polyposis as well as approximately60% of CIMP+ carcinomas arising along the serratedpathway) and microsatellite instable high (MSI-H) carcino-mas (by convention usually referred to as MSI carcinomawithout the high or -H descriptor), which represent allcarcinomas in Lynch syndrome arising via the mutatorpathway, as well as approximately 40% of sporadic CIMP+carcinomas, is useful for detecting patients with Lynchsyndrome but still represents a phenotypic end stage ofdiffering pathways. All of these different descriptions overlapto some degree. The conventional APC initiated adenoma-carcinoma pathway leads to FAP and sporadic CIMP coloncancers arising via the suppressor pathway as well as cancersin Lynch syndrome arising along the mutator pathway. Thealternate serrated pathway leads to CIMP+ carcinomas eitherby a mutator pathway (leading to sporadic MSI+CIMP+carcinomas) or a pathway (details of which are still unknown)that leads to MSS CIMP+ carcinomas. In addition, thereremain probable other pathways to colorectal carcinomas thathave not yet been well worked out, including MSI-low(MIS-L) carcinomas and carcinomas associated with muta-tion of the MYH gene, both of which may have somerelationship to the serrated pathway. These various pathwaysand their overlaps are summarized in Fig. 1. Current thinking

regarding colorectal pathways was recently summarized inreview by the late Jeremy Jass, and this article isrecommended reading for anyone wishing to understand themultiple pathways to colorectal carcinoma [2].

Because most screening and intervention programs forcolon carcinoma have been directed toward the eliminationof conventional adenomas, it would appear that at best a 65%reduction in colorectal carcinoma can be expected by ourcurrent approach. Eliminating the other 35% will require abetter understanding of the serrated pathway and precursorlesions of the serrated pathway, mainly the lesion known asthe sessile serrated adenoma (SSA; AKA sessile serratedpolyp) and the less common traditional serrated adenoma(TSA). This review will focus on these lesions and currentcontroversies related to them.

2. Molecular mechanisms of carcinomasdevelopment via the serrated pathway

Developing a management scheme for dealing withserrated lesions requires an understanding of the underlyingmechanism of molecular carcinogenesis. The basic

e themutaand sis; Ccrosa

2 D. C. SnoverFig. 1 Schematic represent several overlapping ways to describmechanisms based on putative initiating factors (the suppressor andprecursor lesion (the conventional adenoma-carcinoma sequencecharacterized pathways. (Key: FAP = familial adenomatous polyposisland methylator phenotype negative; MSI-H = high degree of miMSS = microsatellite stable; TSA = traditional serrated adenoma.)development of colorectal carcinoma. The red circles representtor pathways). The blue circles represent mechanisms based on theerrated pathways). The yellow circles represent currently poorlyIMP+ = CpG island methylator phenotype positive; CIMP- = CpGtellite instability; MSI-L = low degree of microsatellite instability;

mechanism for the suppressor pathway to colon carcinomabegins with a mutation of the APC tumor suppressor gene,which allows unregulated proliferation of epithelial cells withthe formation of the a lesion conventionally known as anadenoma (these APC-related adenomas will be referred toas conventional adenomas throughout this review; they arecommonly subdivided into tubular, tubulovillous, and villoustypes, and with these designations, the term conventional isunderstood and need not be applied in daily practice). Thisproliferative population then undergoes additional mutations,presumably in a relatively random order, with the eventualprogression of a small subset of conventional adenomas tocarcinoma with intermediate stages showing increasingdegrees of dysplasia. With the exception of conventionaladenomas arising in Lynch syndrome, which appear toprogress rapidly to carcinoma, the most conventionaladenomas never progress and those that do require manyyears (probably 10 or more) to develop into malignancy.Hence, screening programs periodically removing conven-tional adenomas can be very successful in preventing MSScarcinomas arising via the suppressor pathway.

The mechanism of the most common carcinomas arisingvia the serrated pathway appears to begin with an activatingmutation of the BRAF gene, a gene which when mutatedinhibits normal apoptosis of colonic epithelial cells. Lesionsbearing the BRAF mutation develop into serrated lesions thatare mainly either microvesicular hyperplastic polyps(MVHP) or sessile serrated adenomas/polyps (SSA/P, seeterminology, below) [3]. These lesions are prone tomethylation of the CpG island promoter regions resultingin the epigenetic silencing of a number of genes. It ispresumed that the specific genes silenced may be randomevents. The specific gene silenced determines what happenswith a lesion. The most well characterized epigeneticsilencing in these lesions is that of the hMLH1 gene,which is the gene silenced in sporadic MSI carcinomas. ThehMLH1 gene is one of the mismatch repair genes, and whensilencing occurs, these lesions become MSI and are prone tothe development of additional mutations at a rapid rate,similar to what happens to conventional adenomas in patientswith Lynch syndrome. Although in most (although unfortu-nately not all) lesions it appears that methylation of hMLH1

a. Thom n

3Serrated pathway to colorectal carcinomaFig. 2 Representation of the serrated pathway to MSI-H carcinomSSA/P remains debatable. It is possible that SSA/P arises directly frhence, the arrows for these steps are dotted.is sequential pathway involves slow and rapid steps. The origin oformal mucosa or SSA/P might develop from a preexisting MVHP;

is a late event, occurring in all likelihood many years after thelesion initially develops, once hMLH1 is inactivated, there israpid development of cytological dysplasia (see terminology,below) followed by potentially rapid development of frankmalignant transformation. The rate of progression of theselesions has an important impact on screening strategies,discussed below. Carcinomas arising via this pathway areCIMP+MSI carcinomas. The specific molecular eventsleading to the development of CIMP+MSS carcinomas arenot nearly as well characterized as that for CIMP+MSIcarcinomas but presumably involve epigenetic silencing of

and the molecular mechanisms evaluated [6].

not be repeated here except to provide a brief review andintroduction into current issues.

mucin being designated mucin poor HPs (MPHP). GCHPs

Table 1 Classification of serrated lesions of the large intestine

HPMicrovesicular type (MVHP)Goblet cellrich type (GCHP)Mucin poor type (MPHP)SSA/PWithout cytological dysplasiaWith cytological dysplasiaTSAWithout conventional dysplasiaWith conventional dysplasia

lopment of serrated lesions based on the location of the proliferative zone.e (indicated by the thick line) is located at the lower third of the crypt, andts the muscularis mucosae. A, HP. The proliferative zone remains in therypt. Cells continue to mature toward the surface (arrow), but because ofSA/P. The proliferative zone is located at various locations throughout theut also toward the base of the crypt (arrow), which leads to distorted andwth of the crypts. C, TSA. The proliferative zone is also variably located,These miniature crypts create a complex growth pattern. (adapted from

4 D. C. Snover3. Precursor lesions of the serrated pathway

As recently as 2003, all of the various serrated lesionswere considered hyperplastic or metaplastic polyps, andtheir malignant potential was not well recognized, but it isnow generally recognized that these lesions represent acomplex family of lesions differing in their specificunderlying mutation and degree of methylation [3-5,7-11].As yet unknown factors lead to alterations in the location ofthe proliferative zone as well as differences in the anchoringof crypts within the lesions, leading to a wide range ofhistological appearances [10]. It is mainly architecturalalterations resulting from these abnormalities of proliferationthat allow distinction of the different categories of polyps. Asummary of the current classification of serrated polyps

Fig. 3 Graphic representation of the putative sequence for the deveAll 3 lesions begin with normal mucosa in which the proliferative zoncells mature toward the lumen (arrow). The double thin line represenlower portion of the crypt, occupying up to half of the length of the cdelayed apoptosis, the cells tend to pile up, leading to serrations. B, Scrypt length. This allows maturing cells to move toward the lumen bdilated crypts since the muscularis mucosae blocks the downward grobut small ectopic crypts develop from the side of the original crypt.Reference [10]).other genes leading to carcinoma via as yet undefinedpathways. A simplified diagram of the serrated pathway toCIMP+MSI carcinoma is shown in Fig. 2.

There may be several more serrated pathways not relatedto the BRAF gene mutation. Several types of serrated polyps,including goblet cellrich hyperplastic polyps (GCHP) andTSA, are mutated at the Kras gene rather than the BRAFgene (4,5). There is currently no known direct role of GCHPin carcinogenesis; however, it is clear that TSA may developinto carcinoma, although there has been some controversy asto the type of carcinoma. It has been suggested that MSI-Lcarcinomas may develop in these lesions after methylation ofthe MGMT gene or possibly secondary to partial methylationof hMLH1. Finally, there is a recent report of multipleserrated polyps in patients with MYH polyposis associatedwith Kras mutations; however, this remains to be confirmedAlthough the terminology is not perfect because theselesions are not truly hyperplastic or metaplastic, theterm hyperplastic polyp (HP) has been retained for the mostindolent form of serrated lesions [11]. These lesions arecharacterized by variably prominent serrations occurring incrypts that are generally straight and demonstrate more orless normal localization of the proliferative zone to the baseof the crypts (Figs. 3A and 4). HPs are subdivided based onthe mucin content, with those demonstrating exclusivelygoblet cell mucin being termed GCHP, those with smalldroplet (microvesicular) mucin either alone or intermixedwith goblet cells being called MVHP, and those with no(based on the fourth edition of the World Health Organiza-tion [WHO] blue book to be published in 2010 [11]) isshown in Table 1. A diagrammatic representation of theproliferative zone abnormalities and resultant architecturalchanges of the various serrated lesions is shown in Fig. 3.

Although there is now general consensus on theimportance of the serrated lesions in the development ofcarcinomas, and the general categories of lesions areaccepted, challenges remain. Included among these areissues related to terminology, diagnostic criteria, and naturalhistory, which affects management decisions.

4. Terminology and diagnostic criteria

Details of the histological appearance of these lesionshave been reviewed extensively elsewhere [7-11] and will

5Serrated pathway to colorectal carcinoma

SSA, tubular adenoma, tubulovillous adenoma, and villousadenoma once they are familiar with the concept. Manyauthors believe that SSA is the best term because it reflectsthe premalignant potential of this lesion in the serratedpathway and is likely to lead to more appropriatemanagement that other more noncommittal terms. Nonethe-less, consensus has not been reached, and the upcomingfourth edition of the WHO classification of tumors of thedigestive trace will reflect that at the current time either termis acceptable as long as it is recognized that SSA and SSP arereferring to the same lesion [11]. For clarity sake, SSA/P willbe used throughout this article. Unfortunately, there are stillsome pathologists who continue to use the term hyperplasticpolyp or variant hyperplastic polyp for these lesions, a mostunfortunate fact given the considerable management differ-ences between HP and SSA/Ps. The use of the hyperplasticpolyp for these lesions should be avoided.

SSA/P is characterized mechanistically by movement ofthe proliferative zone away from its usual location in the baseof the crypts, resulting in maturation, which may developtoward the base of the crypts leading to distortion of the cryptarchitecture, commonlywith dilated, L, inverted T, or anchor-

6 D. C. Snoverare found almost exclusively in the left colon andcommonly demonstrate Kras mutations, whereas MVHP,which is still predominantly left-sided but with more right-sided lesions than GCHP, tends to be BRAF mutated [4].MPHP appears to represent a damaged MVHP with reactivechange, but this category is rare and has not been wellstudied. It has been suggested that MVHP may act as aprecursor to the more advanced and clearly premalignantSSA/P, in large part because they share the BRAF mutationand a propensity to methylation, although this has not beenproven and perhaps is of more theoretical than practicalsignificance. Very rarely will one encounter a MVHP thatdemonstrates cytological dysplasia resembling conventionaladenoma, and hence, it is possible that MVHP mayoccasionally become methylated at a gene critical forcarcinogenesis, as occurs in SSA/P, and hence may progressto malignancy without an intervening SSA/P.

Fig. 4 Microvesicular hyperplastic polyp. Note that the crypts arerelatively straight with narrow bases (40). Also, see Fig. 3A.Perhaps the most persistent and perplexing issueregarding serrated lesions is the fact that there remains noconsensus on the best terminology for the lesion knownvariably in the literature as sessile serrated adenoma (SSA),sessile serrated polyp (SSP), or serrated polyp with abnormalproliferation . The term SSA was the first term used for thislesion, and a majority of the current literature is written usingthe term SSA; however, there is a second large group whoprefers SSP. The term serrated polyp with abnormalproliferation is rapidly losing favor. The rationale for theuse of polyp rather than adenoma for this lesion revolvesaround a concern that use of adenoma will cause confusionwith conventional adenomas arising via the APC mutationpathway, despite the fact that most authors in the field acceptthat these lesions are functionally premalignant and henceneoplastic. In the practice of this author, who has used theterm SSA since 1996, confusion with conventional adenomashas not been an issue with our gastroenterologists, who seemto have no difficulty understanding the difference betweenshaped crypts with mature cells where the proliferative zonenormally is located (as evidenced bymucinous differentiationas well as marking for cytokeratin 20 (Figs. 3B and 5) [10]).The crypts still retain their attachment or orientation towardto the muscularis mucosae, however, and are arrangedperpendicular to the muscularis mucosae. SSA/P in itsearliest stage does not manifest dysplasia resemblingconventional adenoma, although with progression towardcarcinomas, such dysplasia develops.

There has been variation regarding the best term forlesions that combine the features of SSA/P with areas offrank cytological dysplasia that resemble conventionaladenomas (Fig. 6). These lesions have in most of the older

Fig. 5 SSA/P without cytological dysplasia. Compared to theHP, the overall architecture is distorted with variably shaped cryptsand mature cells at the base of the crypts (40). Also, see Fig. 3B.

known to be 3 or possibly 4 different lesions (SSA/P, SSA/Pith cytological dysplasia, TSA, and conventional APCdenomas with a serrated growth pattern). Recognizing this,2003, it was suggested that the most unique of this group,

nd the one member of this group which did not have a betterame, should be recognized as a separate entity, and the termaditional serrated adenoma was recommended (because itas thought that this lesion was the one member of thisroup that was most easily recognized by most pathologists,nd the one that at the time was probably most commonlyeing called serrated adenoma, in large part because at theme SSA/P was still being called HP by most pathologists,nd SSA/P with cytological dysplasia was being calledixed HP-TA.) [3]. It is of some interest along these linesat a recent reproducibility study identified this lesion as thene with the best kappa statistic of the various serratedsions, supporting our contention that most pathologistslready know how to recognize this lesion [13].There does continue to be some variation in the diagnosis

f TSA, however, in large part because of recent suggestionsor expansion of the diagnostic criteria [10]. The originalefinition of TSA was a very simplistic one, with the most

7Serrated pathway to colorectal carcinomaliterature been called mixed HP-TA, and with the recognitionof SSA/P, it turned out that the serrated portion of most ofthese mixed lesions was in reality SSA/P, leading to the useof the term mixed SSA/P-TA for these lesions [7]. However,it is currently felt that this term is misleading because thecytologically dysplastic part of these polyps is not aconventional TA from a molecular perspective and probablyhas a more aggressive course than conventional TA. Asnoted above, conventional TA is a lesion that starts with anAPC mutation, acquires addition mutations over time, andeventually, in a small number of cases and generally over the

Fig. 6 SSA/P with cytological dysplasia. This lesion hascharacteristics of an SSA/P on the right, but on the left, theepithelium is cytologically dysplastic demonstrating hyperchro-matic elongated nuclei with pseudostratification. This may indicateconversion to a MSI state. There is an underlying invasiveadenocarcinoma as well (40).course of 10 years or more, becomes malignant. SSA/P withcytological dysplasia, on the other hand, in some cases (ie, inthe case of lesions developing into MSI carcinomas)represents the development of microsatellite instability inthe lesion and hence is creating a lesion with a propensity todevelop into carcinoma with high frequency and rapidity (seeFig. 2 and discussion above). These lesions are not APCmutated and probably require considerably more aggressivefollow-up than conventional adenomas. Hence, SSA/P withcytological dysplasia is the preferred term for this lesion,and the term mixed SSA/P-TA or HP-TA is no longerconsidered acceptable.

The other term that has had some controversy is that of thetraditional serrated adenoma (TSA). This lesion is one ofseveral forms of serrated lesions that fit the definition ofserrated adenoma proposed in the seminal article byLongacre and Fenoglio-Prieser in 1990 [12]. The definitionprovided initially for serrated adenomawas that a serratedadenomawas a lesion with dysplasia (and hence adenomaby conventional criteria in use at the time) but which showedan overall serrated configuration. Unfortunately, this defini-tion is overly broad and can be used to describe what are nownotable feature being the presence of a peculiar type ofatypical cell characterized by relatively abundant eosino-philic cytoplasm resembling that of an oncocyte, withelongated nuclei (typically not hyperchromatic) often locatedin the mid portion of the cell and with little or nopseudostratification and with little proliferative activity asdetermined by mitotic count or Ki67 immunoreactivity[3,7,10]. This cell was considered by many as beingdysplastic and therefore was often equated with thedysplasia of conventional adenomas. Given the lack ofproliferative activity (as well as the lack of APC mutations)in these cells, they are not likely to have the same neoplasticpotential as conventional dysplasia and are better considered

Fig. 7 TSA. This lesion is characterized by an overall somewhatvilliform configuration with a very complex overall architecture(40). Also, see Fig. 3C.wainantrwgabtiamtholea

ofd

a metaplastic or possibly senescent cell. Although this celltype was original considered a sine qua non for the diagnosisof TSA, it is now apparent that this cell type can be seen in a

Table 2 Proposal for a semimolecular classification of colorectal polyps

Traditional terminology(pre-2003)

Recently revised terminology(post-2003)

P

Adenoma Conventional adenoma ATubular adenoma Tubular adenomaTubulovillous adenoma Tubulovillous adenomasVillous adenoma Villous adenomaSerrated adenomaHyperplastic polyp Serrated adenomas CHyperplastic polyp SSA/P without cytological dysplasia

SSA/P with cytological dysplasiaC

TSA without conventional dysplasiaTSA with conventional dysplasiaHyperplastic polyps HMicrovesicular HP

HGCHP

HMucin poor HP

8 D. C. Snovervariety of circumstances including in small areas ofotherwise typical SSA/P, in some conventional adenomas(APC-related) and sometimes even in reactive mucosa. Animmunohistochemical analysis of proliferation and matura-tion in serrated lesions has pointed out that a more likelydiagnostic abnormality in TSA is the presence of smallectopic crypts created by apparent loss of anchoring of theFig. 8 TSA with conventional dysplasia. This TSA hascytological dysplasia similar to that seen in conventional adenomasin the right portion of the photo. In this particular case, the dysplasiais high grade with a cribriform pattern of growth. There wasinvasive carcinoma elsewhere in this polyp (40).crypt bases to the muscularis mucosae [10] (Figs. 3C and 7).These ectopic crypts demonstrated strong CK20 staining inthe luminal portions and Ki67 staining in the deeper aspects,hence recapitulating small crypts except that the base of thecrypts was free floating in the lamina propria of villi ratherthan sitting on (or being anchored to) the muscularismucosae. We now feel that these ectopic crypts are probablythe best defining feature of TSA, although most TSAs willalso have the atypical eosinophilic cell noted above. Theseectopic crypts are not present in all areas of all TSAs butshould be present at least focally. Often TSA will have largeareas with the metaplastic eosinophilic cells described above,but using these diagnostic criteria, a TSA can also be

roposed semimolecular terminology

PC-related neoplastic polypsNonserrated neoplastic polyp without villous componentNonserrated neoplastic polyp with villous component

IMP-adenomas, BRAF typeSessile serrated neoplastic polyp without cytological dysplasiaSessile serrated neoplastic polyp with cytological dysplasiaIMP-adenomas, KRAS typeFiliform serrated neoplastic polyp without conventional dysplasiaFiliform serrated neoplastic polyp with conventional dysplasiayperplastic polyp, BRAF typeMicrovesicular HPyperplastic polyp, Kras typeGCHPyperplastic polyp, otherMucin poor HPcomposed mainly of goblet cells. Whether these different celltypes have any molecular or prognostic implications is alsonot known. The lesion described as filiform serratedadenoma is probably synonymous with TSA or represents asubset of TSA and may provide less confusing terminology(see Table 2) [14].

Although eosinophilic metaplastic cells often make up aconsiderable portion of TSAs, atypical cells with many ofthe morphological attributes of conventional dysplasia (ie,hyperchromasia, pseudostratification and increased prolifer-ative activity) do occur in TSA, probably as part ofconversion toward malignancy (Fig. 8). The molecularcharacteristics of these cells (eg, mutation and methylationstatus) have not been well characterized, however, and therehas been essentially no discussion of the terminology of thisconverted lesion in the literature. Trying to be analogous toSSA/P, it would probably be appropriate to refer to theselesions as TSA with conventional cytological dysplasia,despite the fact that this is somewhat cumbersome. Thespecific mechanism of carcinogenesis in TSA is currentlynot well known and is discussed briefly above under

molecular mechanisms of carcinomas development via the

that CIMP+MSS development has the same potential rapid

underlying lesion. Regardless of the reason, this apparentenhancement of interval cancers with CIMP+ carcinomas

on the appropriate interval for rescreening patients with a

9Serrated pathway to colorectal carcinomaarising via the serrated pathway has significant implicationsprogression as seen with CIMP+MSI carcinoma, thisfinding would bolster the theory that some of the issuerelated to interval cancers involves failure to identify theserrated pathway.

5. The role of SSA/P in carcinogenesis and itseffect on management of these lesions

As noted in the introduction, it is now generallyrecognized that colon cancer comprises a family ofdiseases with various molecular pathways. There is now noquestion that SSA/P is the immediate precursor lesion tosporadic CIMP+MSI carcinomas. SSA/P is likely theprecursor lesion to CIMP+MSS carcinomas as well.Combined, CIMP+MSI and CIMP+MSS carcinomas ac-count for approximately 35% of all colorectal carcinoma;hence, these are very important lesions to recognize anddeal with if one wants to reduce the risk of colon carcinomathrough screening. Unfortunately, we are still somewhatimpaired in our understanding of the natural history ofserrated lesions because of a lack of longitudinal studies,and current recommendations for management are based inlarge part on our understanding of the molecular mechan-isms of the pathway and on observational studies includingrecent data on the frequency of serrated pathwayassociated carcinomas presenting as interval cancers inscreening programs.

One important difference in screening for serratedlesions (as opposed to screening for conventional adeno-mas) results from the fact that the CIMP+MSI pathwaymay be a rapidly progressive pathway, much more rapidthan carcinogenesis via the conventional APC pathway, asdescribed above. Recent analysis of interval carcinomasoccurring in screening populations has indicated that thesecancers are disproportionally MSI [15]. One possiblereason for this is that some of these lesions may becomemalignant very rapidly, with their entire lifespan fitting inthe 5 or 10 years between screening examinations. Anotherpotential reason that should not be overlooked, however, isthat SSA/P is a very subtle lesion and may be difficult toidentify on endoscopic examination, especially for anendoscopist who is not experienced with recognizing thesubtle changes that might indicate the presence of a lesion(especially adherent stool or mucin that must be washed offto visualize the lesion). Therefore, interval MSI cancersmight also occur because SSA/Ps are being missed. It turnsout that interval cancers are not only disproportionallyCIMP+MSI, but they are also disproportionally CIMP+MSS [16]. Because there is no a priori reason to suspectknown SSA/P.Our past recommendations had suggested that the most

worrisome lesions of this group, and the lesion with the mostpressing need for close surveillance, were the SSA/Ps withcytological dysplasia, since these lesions have the potentialto rapidly progress to carcinoma [7,8]. We still maintain thatany lesion with this histology should be removed entirely bywhatever means necessary and that the patient undergorepeat endoscopy no longer than a year after initial excision(if the endoscopist is comfortable that the lesion has beenentirely removed) or sooner if there is any question aboutadequacy of excision. TSAs, either with or without changesof conventional dysplasia, are probably best treated asconventional adenomas because they do not lead to thedevelopment of MSI carcinomas, and hence, there is noreason to suspect that they might be more aggressive thanother lesions. In addition, because they tend to occur in therectosigmoid region and are usually well-circumscribedprotruding lesions, surveillance and removal tend to beeasier. The real management issue is for SSA/P withoutcytological dysplasia because these are the most commonlesions and are the ones that are most difficult to visualizeand remove.

Our original recommendations were relatively leisurelyregarding this group, with recommendations to remove thelesion if possible or, if the lesion was not easily resected viathe endoscopy, to do surveillance at annual intervals toliberally biopsy the lesion and look for cytological dysplasiathat would then accelerate the need to remove the lesion[7,8]. If the lesion was completely removed, then wesuggested reversion to standard adenoma surveillanceintervals. Given the recent data regarding CIMP+ andinterval cancers, however, this may not be an adequatestrategy. Personal observations by myself and others(Kenneth Batts, MD, personal communications) wouldindicate that often patients with one SSA/P will harborothers, either synchronously or metachronously. This seemsespecially true for large lesions found in the right colon.When the criteria for serrated polyposis (formerly knownas hyperplastic polyposis) are met ([1] at least 5 serratedpolyps proximal to the sigmoid colon with 2 or more ofthem larger than 10 mm, [2] any number of serrated polypsproximal to the sigmoid colon in an individual who has afirst-degree relative with serrated polyposis, or [3] morethan 20 serrated polyps of any size, but distributedthroughout the colon ), the situation is more defined, andclose surveillance with possible colectomy is considered[11]. However, the criteria for serrated polyposis are at thispoint still entirely arbitrary, and therefore, it is possible ifnot probable that the finding of multiple SSA/Ps of any sizeor location may be adequate to recommend more frequentsurveillance. As noted above, the issue with SSA/P may be2-foldpotential rapid progression to carcinoma andincomplete removal. To prevent interval cancers, therefore,shortened interval surveillance may be appropriate.

Reasonable risk factors suggesting the need for shortinterval reendoscopy would include size of lesions (lesionssmaller than 1 cm may not be as important), location oflesions (right sided lesions may be more important), and age

carcinomas arising via MMR mutations in Lynch syndrome);however, finding a convenient shorthand substitute will bedifficult. Nonetheless, as a starting point, a suggestion for asemimolecular classification is proposed in Table 2.

10 D. C. SnoverA reasonable strategy, and one we are currently recom-mending, would be to repeat surveillance at 1 year foranyone with 2 or more completely removed SSA/Ps greaterthan 1 cm and determine future screening based on thesefindings. If there is any significant risk that a lesion was notcompletely removed, repeat endoscopy at a shorter intervalmay well be indicated. If additional SSA/Ps are identified at1 year, annual surveillance (at least until there is a negativecolonoscopy) would seem reasonable. If screening at 1 yearis clear of lesions, then extension of the interval to 2 to 3years would seem appropriate.

6. Proposed solutions to ongoing issues

As noted above, perhaps the more important issuesregarding serrated lesions involve appropriate management,although terminology remains an issue as well. Themanagement issue is amenable to further clinicopathologicalstudies focusing on natural history, not only for the morecommon SSA/P but also for TSA, since we know the leastabout that lesion. Additional molecular characterization ofspecific genetic changes resulting in CIMP+MSS carcino-mas and for carcinomas arising in TSA will also be valuablein this regard, since this information may help determine thebest surveillance and prevention strategies.

Regarding the terminology issues, at the current time, thereappears to be a stalemate between the proponents of SSA andthose for SSP. Much of this problem exists because of thetraditional use of the term adenoma not as a generic family oflesions (ie, polyps with premalignant potential) but rather for aspecific subset of adenomas, that is, adenomas predominantlyarising along the suppressor pathway initiated with APCmutations. Perhaps it is time for gastrointestinal pathologists toconsider reclassification of adenomas on a semimolecularbasis, realizing that it is not practical to perform moleculartesting on all lesions removed as adenomas. We could,however, alter our terminology to indicate the likely molecularaspects of adenomas, something that would point out thepotential molecular difference and would at a minimum havean educational effect if not a directmanagement effect. It mightalso be advisable to eliminate the term adenoma fromterminology given the long history of using this word onlyfor precursor lesions to non-CIMP carcinomas (ie, APC-related carcinomas arising via the suppressor pathway andMSIReferences[1] Fearon ER, Vogelstein B. A genetic model for colorectal tumorigen-

esis. Cell 1990;61:759-67.[2] Jass JR. Classification of colorectal cancer based on correlation of

clinical, morphological and molecular features. Histopathology2007;50:113-30.

[3] Torlakovic E, Skovland E, Snover DC, et al. Morphologicreappraisal of serrated colorectal polyps. Am J Surg Pathol2003;27:65-81.

[4] O'Brien MJ, Yang S, Mack C, et al. Comparison of microsatelliteinstability, CpG island methylation phenotype, BRAF and KRASstatus in serrated polyps and traditional adenomas indicates separatepathways to distinct colorectal carcinoma end points. Am J Surg Pathol2006;30:1491-501.

[5] Yachida S,Mudali S,Martin SA,Montgomery EA, Iacobuzio-DonahueCA. Beta-catenin nuclear labeling is a common feature of sessileserrated adenomas and correlates with early neoplastic progression afterBRAF activation. Am J Surg Pathol 2009;33:1823-32.

[6] Boparai KS, Dekker E, Van Eeden S, et al. Hyperplastic polyps andsessile serrated adenomas as a phenotypic expression of MYH-associated polyposis. Gastroenterology 2008;135:2014-8.

[7] Snover DC, Jass JR, Fenoglio-Preiser C, Batts KP. Serrated polyps ofthe large intestine: a morphologic and molecular review of an evolvingconcept. Am J Clin Pathol 2005;124:380-91.

[8] Snover DC. Serrated polyps of the large intestine. Sem Diag Pathol2005;22:301-8.

[9] East JE, Saunders BP, Jass JR. Sporadic and syndromic hyperplasticpolyps and serrated adenomas of the colon: classification, moleculargenetics, natural history, and clinical management. Gastroenterol ClinNorth Am 2008;37:25-46.

[10] Torlakovic EE, Gomez JD, Driman DK, et al. Sessile serrated adenoma(SSA) vs. traditional serrated adenoma (TSA). Am J Surg Pathol2008;32:21-9.

[11] Snover DC, Ahnen DJ, Burt RW, Odze RD. Serrated polyps of thecolon and rectum and serrated (hyperplastic) polyposis. In: BozmanFT, Carneiro F, Hruban RH, Theise N, editors. WHO classication oftumours. Pathology and genetics. Tumours of the digestive system.Berlin: Springer-Verlag (4th ed. [in press]).

[12] Longacre TA, Fenoglio-Preiser CF. Mixed hyperplastic adenomatouspolyps/serrated adenomas. A distinct form of colorectal neoplasia. AmJ Surg Pathol 1990;14:524-37.

[13] Farris AB, Misdraji J, Srivastava A, et al. Sessile serrated adenoma:challenging discrimination from other serrated colonic polyps. Am JSurg Pathol 2008;32:30-5.

[14] Yantiss RK, Oh KY, Chen YT, et al. Filiform serrated adenomas: aclinicopathologic and immunophenotypic study of 18 cases. Am JSurg Pathol 2007;31:1238-45.

[15] Sawhney MS, Farrar WD, Gudiseva S, et al. Microsatellite instabilityin interval colon cancers. Gastroenterology 2006;131:1700-5.

[16] Arain MA, Sawhney M, Sheikh S, et al. CIMP status of interval coloncancers: another piece to the puzzle. Am J Gastroenterol 2010;105:1189-95.(older patients being more prone to SSA/P as well as tosporadic MSI carcinomas), in addition to number of lesions.

Update on the serrated pathway to colorectal carcinomaIntroductionMolecular mechanisms of carcinomas development via the serrated pathwayPrecursor lesions of the serrated pathwayTerminology and diagnostic criteriaThe role of SSA/P in carcinogenesis and its effect on management of these lesionsProposed solutions to ongoing issuesReferences