the lethal phenotype of cancer: the molecular basis of ... · the lethal phenotype of cancer: the...

2007;57;225-241 CA Cancer J ClinJ. Pienta

Robert D. Loberg, Deborah A. Bradley, Scott A. Tomlins, Arul M. Chinnaiyan and Kenneth The Lethal Phenotype of Cancer: The Molecular Basis of Death Due to Malignancy

This information is current as of July 15, 2007

http://caonline.amcancersoc.org/cgi/content/full/57/4/225the World Wide Web at:

The online version of this article, along with updated information and services, is located on

http://caonline.amcancersoc.org/subscriptions/individuals only): , go to (USCA: A Cancer Journal for CliniciansTo subscribe to the print issue of

ISSN: 0007-9235. Online ISSN: 1542-4863. Clifton Road, NE, Atlanta, Georgia 30329. (©American Cancer Society, Inc.) All rights reserved. Print

is owned, published, and trademarked by the American Cancer Society, 1599CANovember 1950. Lippincott Williams & Wilkins. A bimonthly publication, it has been published continuously since

is published six times per year for the American Cancer Society byCA: A Cancer Journal for Clinicians

by guest on July 15, 2007 (©A

merican C

ancer Society, Inc.)

caonline.amcancersoc.org

Dow

nloaded from

http://caonline.amcancersoc.org/cgi/content/full/57/4/225

http://caonline.amcancersoc.org/subscriptions/

http://caonline.amcancersoc.org:80

225Volume 57 • Number 4 • July/August 2007

CA Cancer J Clin 2007;57:225–241

The Lethal Phenotype of Cancer: The Molecular Basis of DeathDue to MalignancyRobert D. Loberg, PhD; Deborah A. Bradley, MD; Scott A. Tomlins; Arul M. Chinnaiyan, MD, PhD; Kenneth J. Pienta, MD

ABSTRACT The last decade has seen an explosion in knowledge of the molecular basis and

treatment of cancer. The molecular events that define the lethal phenotype of various cancers—

the genetic and cellular alterations that lead to a cancer with a poor or incurable prognosis—

are being defined. While these studies describe the cellular events of the lethal phenotype of cancer

in detail, how these events result in the common clinical syndromes that kill the majority of can-

cer patients is not well understood. It is clear that the central step that makes most cancers incur-

able is metastasis. Understanding the traits that a cancer acquires to successfully grow and

metastasize to distant sites gives insight into how tumors produce multiple factors that result

in multiple different clinical syndromes that are lethal for the patient. (CA Cancer J Clin

2007;57:xxx–xxx.) © American Cancer Society, Inc., 2007.

To earn free CME credit for successfully completing the online quiz based on this article, go tohttp://CME.AmCancerSoc.org.

INTRODUCTION

In 2007, it is estimated that 559,650 people in the United States will die of can-cer.1 The last decade has seen an explosion in the amount of knowledge in themolecular basis and treatment of cancer. Multiple studies have been published describ-ing the molecular events that define the lethal phenotype of various cancers—the geneticand cellular alterations that lead to a cancer with a poor or incurable prognosis.While these studies describe the cellular events of the lethal phenotype of cancer in detail, how these events result inthe common clinical syndromes that kill the majority of cancer patients is not well understood. The majority of solid-tumor malignancies kill patients because they escape the primary site and metastasize (Figure 1). The traits that a can-cer acquires to successfully grow and metastasize to distant sites produce multiple factors that result in different clinicalsyndromes that are lethal for the patient.2–5 These syndromes can be broadly characterized into those related tocytokine overproduction and those related to organ failure. This paper describes how the molecular alterations ofmetastatic cancer result in the clinical lethal phenotype of cancer.

THE MOLECULAR BASIS OF CANCER

The process of carcinogenesis is the result of DNA damage that occurs in a normal cell and leads toward a growth andsurvival advantage (Figure 2).6–8 DNA damage is the result of gene-environment interactions on multiple levels, includ-ing the susceptibility for genetic damage inherited from parental genes.9,10 On their inherited genetic background, cellsare assaulted by a variety of gene-damaging environmental agents, including radiation, viruses and other microbes, andchemical carcinogens, as well as the free radicals that are byproducts of normal cellular processes that accumulate withage. These DNA-damaging agents are modulated by host defenses and intrinsic organ- and extrinsic nonorgan-specific

Dr. Loberg is Research Assistant Pro-fessor, Internal Medicine and Urology,University of Michigan, Ann Arbor, MI.

Dr. Bradley is Fellow, Departmentof Internal Medicine, Division of Hema-tology and Oncology, University ofMichigan, Ann Arbor, MI.

Mr. Tomlins is Graduate Student,Department of Pathology, Universityof Michigan, Ann Arbor, MI.

Dr. Chinnaiyan is S. P. Hicks En-dowed Professorship; Professor ofPathology and Urology; Director ofPathology Research Informatics; and Di-rector of Cancer Bioinformatics, Uni-versity of Michigan, Ann Arbor, MI.

Dr. Pienta is Professor, Internal Med-icine and Urology; and American Can-cer Society Clinical Research Professor,University of Michigan, Ann Arbor, MI.

This article is available online athttp://CAonline.AmCancerSoc.org

Disclosures: K.J.P. is supported by an American Cancer Society Clinical Research Professorship. A.M.C. is supported by a Burroughs Welcome FoundationAward in Clinical Translational Research. S.A.T. is supported by the Medical Scientist Training Program and a Rackham Predoctoral Award. This research wassupported in part by National Institutes of Health Grant RO1 CA102872 (to K.J.P.); RO1 CA97063 (to A.M.C.); U01 CA111275 (to A.M.C.); P50 CA69568(to K.J.P., A.M.C.); Department of Defense Grant PC051081 (to A.M.C.); Ralph Wilson Medical Research Foundation Award (K.J.P.); and a Prostate CancerFoundation Research Award (R.D.L.)


merican C



Dow

nloaded from


risk modulators. Host defenses include the stateof the patient’s immune system, nutritional status,and comorbid conditions. Intrinsic risk modula-tors are inherited traits that do not contributedirectly to DNA damage, but modulate the envi-ronment that the cells are exposed to (ie, howwell liver-metabolizing enzymes such as CYP3Afunction to modulate drug and hormone activity).11

Extrinsic risk modulators are best characterized bychemoprevention agents (ie, antioxidants such asselenium and vitamin E that remove damagingoxygen radicals from the intracellular environ-ment by facilitating their breakdown to water).12

Regardless of how damage to the genome orig-inates, cancers are the result of mutations thatresult in a group of common characteristics

or “hallmarks” that define the minimum set ofsurvival traits that a cancer cell must acquire toflourish (Figure 1). These hallmarks include thefollowing: (1) genetic instability; (2) limitlessreplicative potential (immortality); (3) anchorage-independent growth; (4) stimulation of angiogen-esis; (5) evasion of programmed cell death(apoptosis); and (6) ability to grow independ-ently of stimulation by growth factors.9

THE MOLECULAR BASIS OF METASTASIS

All of the above mutations, whether acquiredby chance accumulation or through clonal expan-sion of a cell population through selective pres-sure in a continued hostile environment, resultin successful growth of a cancer cell populationat the primary site. Only a small subset of thebillions of cells within a tumor accumulates thetraits of tissue invasion, extravasation, survivalin the circulation, and growth in secondary sitesthat characterize successful metastases.2–5,13–15

This subset of cells has characteristics heralded bya change in the cancer cell phenotype observedas an epithelial-mesenchymal shift and is theresult of reactivation and the loss of regulation ofcellular programs associated with wound healingand/or embryogenesis.16–19

A cell that does not acquire the genetic alter-ations necessary for invasion and metastasis doesnot acquire a lethal phenotype and only rarelycauses death. Several ongoing research effortsare aimed at differentiating/predicting whichtumors have acquired the necessary signaturethat correlates with metastasis and/or poor prog-nosis. By comparing the genes that are expressedbetween primary cancers and metastases, Rama-swamy and colleagues identified a 17-gene expres-sion signature that was able to distinguish primarytumors from metastases in several solid tumors andwas associated with poor prognosis (Table 1).20

Other investigators have identified unique genesets that function as metastasis signatures inmultiple solid tumors, including breast, renal,colon, oral, lung, and prostate cancers.23–29 Sim-ilarly, several disease-specific gene signaturesthat distinguish aggressive cancers (in general,those cancers that recurred, metastasized, orcaused death) from nonaggressive cancers (thosethat did not recur or metastasize) have been

The Lethal Phenotype of Cancer: The Molecular Basis of Death Due to Malignancy

226 CA A Cancer Journal for Clinicians

FIGURE 1 The Central Step in the Lethal Phenotype of Cancer IsMetastasis. Most cancers can be cured if they are treated while they are stillconfined to their primary site. Understanding the traits that a cancer acquiresto successfully grow and metastasize to distant sites gives insight into howtumors and the host produce cytokines and other factors that result in theclinical syndromes that are lethal for the patient.


merican C



Dow

nloaded from


reported.21,22,26,30–34 For example, Glinksy andcolleagues published an 11-gene signature panelwith which they demonstrated a significant asso-ciation between the expression pattern of the 11-gene signature and poor prognosis of patients witha wide variety of cancers (Table 1).21

Hundreds of large-scale DNA microarrayexperiments have been performed that havegenerated quantitative profiles of gene expres-sion in cancer, allowing types of cancer to bedistinguished by their gene expression patternsand, more importantly, to discover novel molec-ular subtypes of cancer that are associated witha variety of tumor properties, including mecha-nism of transformation, propensity to metasta-size, and sensitivity or resistance to particulartherapies.35,36 Oncomine™ is an online initia-tive that collects published cancer microarray dataand allows researchers to easily compare geneticexpression data across cancer types and subtypes(www.oncomine.com).37–39 We reviewed 14cDNA microarray data sets of primary versusmetastatic tumors within the Oncomine data setand found that no 2 cancers presented similargene signatures and that the number of statisticallysignificant (P � 0.01) genes that were differen-tially expressed varied from one study to another;thus, no consistent gene set has been identifiedthat predicts the lethal phenotype of cancer (ie,metastatic disease) across multiple organ sites. Itis likely that this is due to the fact that no inves-tigators have compared gene signature sets to theclinical syndromes such as cachexia, thrombosis,and bone metastases that are ultimately respon-sible for the death of the patient.40 Although nodirect research has been done to identify molec-ular signatures associated with these syndromes,when the different signatures are characterizedby ontological process rather than specific function, we found that they fall into generalcategories that include RNA processing, cell pro-liferation, cell cycle and cell division, extracel-lular matrix alteration, and differentiation (Table 1).Increased RNA processing leads to the increasedprotein synthesis necessary for the more meta-bolically active cancer cells; increased cellularproliferation and cellular division lead to increasedtumor burden; alterations in the extracellularmatrix are important for the establishment of theproper “fertile bed” of the microenvironment

that will support tumor growth; and loss of dif-ferentiation correlates with the activation ofembryonic genes necessary for cell move-ment.2–5,13–15 These studies are further complicatedby the fact that many of them analyze not onlythe cancer cells but also the supporting stromaltissues at the same time. Recent evidence sug-gests that the inherited genomic makeup of anindividual may predict the frequency and futuresites of metastasis by providing a favorablemicroenvironment for metastasizing cells to col-onize.41–43 Many investigators are now concen-trating on using laser-capture microdissection tocharacterize the individual cell populations of



FIGURE 2 The Process of Carcinogenesis Is a Result of Multiple ComplexInteractions Between the Host and the Environment. Carcinogenesis is theresult of DNA damage that occurs in a normal cell and leads toward a growthand survival advantage. DNA damage is the result of gene-environment inter-actions on multiple levels, including the susceptibility for genetic damageinherited from parental genes. On their inherited genetic background, cellsare assaulted by a variety of gene-damaging environmental agents, includingradiation, viruses, microbes, carcinogens, chemicals, and hormones, as wellas the free radicals that are byproducts of normal cellular processes thataccumulate with age. These DNA-damaging agents are modulated by hostdefenses and intrinsic organ- and extrinsic nonorgan-specific risk modulators.


merican C



Dow

nloaded from




TABLE 1 Gene Signatures from Three Independent Laboratories Defining Molecular Signatures of Metastasis,Poor Prognosis, and High- Versus Low-risk Patients

Ramaswamy S,Ross KN, Lander ES,Golub TR20 Metastatic Signature

Gene Name Function

UPSNRPF Small nuclear ribonucleoprotein polypeptide f Part of the pre-mRNA splicing machineryEIF4EL3 Eukaryotic translation initiation factor 4E member 2 Cap binding protein that stabilizes mRNA for processingHNRPAB Heterogeneous nuclear ribonucleoprotein A/B Pre-mRNA processingDHPS Deoxyhypusine synthase Vital for cell proliferationPTTG1 Pituitary tumor-transforming 1 Regulates cell divisionCOL1A1 Collagen, type I, α 1 Major component of type I collagenCOL1A2 Collagen, type I, α 2 Major component of type I collagenLMNB1 Lamin B1 Major component of nuclear matrix

DOWNACTG2 Actin, γ 2, smooth muscle, enteric Cytoskeletal structure and contractile appartusMYLK Myosin, light polypeptide kinase Facilitates myosin and actin interactionsMYH11 Myosin, heavy polypeptide 11, smooth muscle Major component of the contractile apparatusCNN1 Calponin 1, basic, smooth muscle Smooth muscle contractionHLA-DPB1 Major histocompatibility complex, class II, DP β 1 Immune responseRUNX1 Runt-related transcription factor 1 Lineage specification and homeostasisMT3 Metallothionein 3 (growth inhibitory factor) Inhibits growthNR4A1 Nuclear receptor subfamily 4, group A, member 1 Nuclear transcription factor, inducer of apoptosisRBM5 RNA-binding motif protein 5 Potential tumor-suppressor gene

Glinsky GV,Berezovska O,Glinskii AB21 Poor Prognosis

Gene Name Function

UPGBX2 Gastrulation brain homeobox 2 DevelopmentKI67 Antigen identified by monoclonal antibody Ki-67 Nuclear antigen present in proliferating cellsCCNB1 Cyclin B1 Regulatory protein involved in mitosisBUB1 BUB1 budding uninhibited by benzimidazoles 1 homolog Kinase involved in spindle checkpoint functionKNTC2 Kinetochore associated 2 Spindle checkpoint signalingUSP22 Ubiquitin-specific peptidase 22 Cell-cycle controlHCFC1 Host cell factor C1 (VP16-accessory protein) Regulation of cell cycle and transcriptional activityRNF2 Ring finger protein 2 Polycomb group protein involved in transcriptional regulationANK3 Ankyrin 3, node of Ranvier (ankyrin G) Integral membrane protein involved in motility, proliferation,

and activationFGFR2 Fibroblast growth factor receptor 2 GrowthCES1 Carboxylesterase 1 Hydrolize long-chain fatty acid esters

Varambally S,Yu J,Laxman B, et al22 High Risk Versus Low Risk

Gene Name Function

UPITGA5 Integrin, α 5 (fibronectin receptor, α polypeptide) Form a fibronectin receptorCIAP Baculoviral IAP repeat-containing 2 Inhibits apoptosisDRBP76 Interleukin enhancer binding factor 3, 90kDa Transcription factorKRIP-1 Tripartite motif-containing 28 Transcriptional controlAMACR α-Methylacyl-CoA racemase Conversion of pristanoyl-CoA and C27-bile acyl-CoAs to

their (S)-stereoisomersOCLN Occludin Integral membrane protein that is located at tight junctionsMCM2 MCM2 minichromosome maintenance deficient 2, mitotin Initiation of eukaryotic genome replicationNUP62 Nucleoporin p62 Components of the nuclear pore complex in eukaryotic cellsLAP2 Thymopoietin Regulation of nuclear architecture by binding lamin B1


merican C



Dow

nloaded from


the tumor microenvironment, including the can-cer cells, endothelial cells, and fibroblasts.44–46

THE BYPRODUCTS OF METASTATIC TUMOR CELLS AND THEIR INTERACTION WITH THE MICROENVIRONMENT

The multiple factors produced during theprocess of metastasis that contribute to themorbidity and mortality of patients come from

3 sources: the cancer cells themselves, normalcells that are trying to inhibit the growth andspread of the cancer, and the factors that arereleased by the local microenvironment of thetissue as these cells interact (Table 2).

A critical event occurs in a metastasis whenthe growth of the tumor cell mass reaches approx-imately 1 cubic millimeter in size.47 At this point,cells in the center of the tumor are beyond the dif-fusion distance of oxygen and other nutrientsnecessary for survival. Hypoxia, through theinduction of hypoxia-inducible factor-1α, causesthe production of multiple cytokines and growthfactors that increase the chance of cell survivaland turn on the cellular programs that promotegrowth, angiogenesis, and metastasis.2–5,9,13–15

These include autocrine motility factor, uroki-nase plasminogen activator (uPA), matrix met-alloproteinases (MMPs), cathepsins, endothelin-1



TABLE 2 Common Cytokines and Factors That Play a Role in the Production of the Lethal Phenotype

Selected Cytokinesand Factors Role in Production of the “Lethal Phenotype” of Metastatic Disease

ChemokinesCCL2/CCR2 Facilitates invasion and metastasis, promotes cancer cell growth by autocrine regulation, contributes to

regulation of angiogenesisCXCL12/CXCR4 Regulates stem cell homing and plays a crucial role in facilitating those tumors that metastasize to bone

CytokinesIL-1 Contributes to ability to metastasize; implicated as a tumor cell growth factor; stimulates angiogenic factors;

implicated in thrombosis, cachexia, and bone metastasesIL-6 Promotes cancer growth; implicated as a tumor cell growth factor; stimulates angiogenic factors;

implicated in thrombosis, cachexia, and bone metastasesNF-κB Key mediator and regulator of the inflammatory process, participates in feedback loop of proinflammatory

cytokines, suppresses apoptosis, promotes tumor invasion and metastasis, contributes to tumorproliferation by activating the expression of growth factor genes, contributes to genomic instability of thecancer cells

TNF-α Induces DNA damage and inhibits DNA repair, promotes tumor growth, induces angiogenic factors, key ininitiation of inflammatory cascade, regulates chemokines, contributes to ability for invasion, contributes tocachexia syndrome, implicated in thrombosis, contributes to bone metastases

TGF-β Contributes to angiogenesis, implicated in thrombosis, contributes to bone metastasesVEGF Induces tumor angiogenesis in solid tumors and promotes tumor growth and metastasis

ProteasesMMP Enzyme involved in degradation of extracellular matrix and is upregulated in most cancers, allowing tumor cell

invasion and metastasisuPA uPA levels in both resected tumor tissue and plasma are of independent prognostic significance for patient

survival in several types of human cancer

Coagulation cascadeThrombin Thrombin generation is crucial for metastasis through fibrin and platelet deposition; thrombin receptor

upregulation has been reported in a variety of malignant tissuesTF Advanced cancer is associated with a hypercoagulable state that is triggered by TF; TF significantly

participates in tumor-associated angiogenesis, and its expression levels have been correlated with themetastatic potential

Cell–cell interactions Cell–cell, cell–platelet, and platelet–platelet interactions appear to enhance metastasis

CCL2/CCR2 � monocyte cheomattractant protein-1 andits receptor.CXCL12/CXCR4 � stromal-derived factor-1 and itsreceptor.IL-1 � interleukin-1.IL-6 � interleukin-6.NF-κB � nuclear factor �B.TNF-α � tumor necrosis factor-�.TGF-β � transforming growth factor-�.VEGF � vascular endothelial growth factor.MMP � matrix metalloproteinase.uPA � uroplasminogen activator.TF � tissue factor.


merican C



Dow

nloaded from


(ET-1), vascular endothelial growth factor (VEGF),and transforming growth factor-β (TGF-β).48,49

The interaction of cancer cells with thenormal cells of the patient also results in theproduction of multiple factors that may be detri-mental to the host. A relationship between can-cer and the inflammation associated with hostresponse has been recognized since the 1860s,when Virchow observed leukocytes in neoplas-tic tissues.50 While some cancers are associatedwith an inflammatory response that is detri-mental to the tumor, there are several malig-nancies that appear to be facilitated by chronicstates of inflammation, including Helicobacterpylori and gastric cancer, acid reflux and esopha-geal cancer, and inflammatory bowel diseaseand colon cancer. These cancers are thought tobe the result, in part, of the production ofproinflammatory cytokines by the host immunecells as they try to destroy the cancer cells.50–52

Proinflammatory cytokines, including tumornecrosis factor-α (TNF-α), interleukin-1 (IL-1), interleukin-6 (IL-6), interleukin-11 (IL-11), and TGF-β, have been shown toinduce DNA damage and inhibit repair; inhibitapoptosis; facilitate tumor growth, invasion,and metastasis; induce production of angio-genic factors; and contribute to maintaining achronic state of inflammation by way of a self-activating feedback loop.50–52 These cytokinescan cause morbidity and mortality in patientsthrough activation of multiple signaling path-ways leading to clinical syndromes such ascachexia and coagulopathy.

Similarly, the interaction of other host cellswith cancer cells can lead to alterations in themicroenvironment. Perhaps the best character-ized example of this is metastases involving thebone.53–55 Prostate and breast cancer cells, forexample, are attracted to the bone by high lev-els of stromal-der ived factor-1 (SDF-1), achemokine secreted by bone stromal cells thathelps direct hematopoetic cell trafficking in andout of the bone marrow.2 Once there, they secreteseveral cytokines, including IL-6, IL-1, TNF-α,TGF-β, epidermal growth factor (EGF), andET-1, that stimulate the maturation and prolif-eration of osteoblasts.53–55 Osteoblasts in turnbuild up disorganized bone, as well as secretereceptor activator of nuclear factor κB ligand

(RANKL), which binds to receptor activatorof nuclear factor κB (RANK) on osteoclastprecursors, resulting in maturation and susbse-quent osteolysis of the bone matrix. This break-down of the bone matrix in turn releases growthfactors that stimulate the tumor cells to growfurther, resulting in a vicious cycle of bonedestruction and further tumor growth.

LINKING THE MOLECULAR BASIS OF METASTASISTO THE CYTOKINES INVOLVED IN THE LETHALCLINICAL SYNDROMES OF CANCER THROUGHMOLECULAR CONCEPT MODELING

To move beyond the analysis of a singlecytokine/chemokine as it relates to the lethalphenotype of cancer, we analyzed a cytokine-signature set (Kyoto Encyclopedia of Genes andGenomes cytokines-cytokine receptors) againstall advanced cancer and/or metastatic gene-signature data sets available in the OncomineDatabase using the Molecular Concepts Map(MCM).40,56 Using the MCM, we looked forenrichment of the cytokine concept in 20 DNA-microarray data sets representing advanced/metastatic disease. As shown in Figure 3A,advanced/metastatic signatures from distincttumor types shared significant overlap, likelyrepresenting common transcriptional programsactivated during metastasis (such as prolifera-tion). Analysis of this network also demonstratedenrichment of the cytokine-cytokine-receptorconcept in advanced/metastatic disease signa-tures. Further examination of the individualcytokines that are upregulated in the individ-ual gene expression arrays revealed that allcytokines (n � 258) were upregulated in �2gene expression arrays, with some cytokinesexpressed in 13 of the gene expression signa-tures (Figure 3B). The cytokines that are knownto be important in advanced/metastatic disease(Table 2) were shown to be expressed in multi-ple signatures from several cancer types. Thesedata demonstrate that cytokines (ie, IL-6, IL-1,CCL2, CXCL12, TGF-β, and TNF-α) areimportant mediators of advanced cancer andcontribute significantly to the lethal phenotype.Interestingly, no single cytokine or subset ofcytokines was upregulated in all advanced sig-natures. Many cytokines were identified by thisanalysis to be involved in metastasis and poorprognosis, suggesting multiple potential targets




merican C



Dow

nloaded from


that need to be explored at both the molecularand clinical levels. These analyses suggest thatmultiple cytokines/combinations of cytokinescause morbidity and mortality for cancer patientsand offer multiple avenues for therapeutic devel-opment that need to be addressed.

THE CLINICAL SYNDROMES RESULTING FROM THEGROWTH OF METASTATIC TUMOR CELLS

Clinically, the “lethal” phenotype of cancer isdefined by what kills the patient. Data from

autopsy series document where metastasis occurs,but rarely clearly document how cancer ends apatient’s life. We performed an extensive litera-ture search through PubMed and Google Scholarto identify published autopsy series that docu-mented sites of metastases in cancer patients atthe time of death (Table 3).56–70 The majority ofthe published autopsy data on cancer patientswas gathered between 1900 and the 1970s andis representative of deaths when little treatmentwas available beyond surgery and radiation. For



FIGURE 3 The Molecular Concepts Map. The Molecular Concepts Map (MCM) is a bioinformatics platform that analyzeshow genes, gene products (proteins), and biological concepts relate to each other through analysis of public-domain datasets and published studies. We looked for enrichment of the cytokine concept in 20 DNA microarray data sets representingadvanced/metastatic disease. (A) Signatures from different tumor types shared significant overlap, representing commontranscriptional programs activated during metastasis. (B) Examination of individual cytokines that are upregulated in theindividual gene expression arrays revealed that all cytokines (n � 258) were upregulated in 2 gene expression arrays, withsome cytokines expressed in 13 of the gene expression signatures. These data demonstrate that cytokines (ie, interleukin-6 [IL-6], interleukin-1 [IL-1], CCLs, CXCL12, transforming growth factor-β [TGF-β], and tumor necrosis factor-α [TNF-α])are important mediators of advanced cancer and contribute significantly to the lethal phenotype.


merican C



Dow

nloaded from


example, very few modern series are availablethat report metastases in patients receiving treat-ments such as chemotherapy that may alter the nat-ural history of the disease.60–62 Similarly, whilethese series report anatomic distribution, actualcause of death is rarely delineated. The majorsites of metastases in patients dying of cancer arethe lymph nodes, the lungs, the liver, and theskeleton (Table 3).56–70 Although how patientsdie from cancer depends on metastatic patternsof specific tumor types, the clinical syndromesby which patients succumb to cancer can beroughly divided into 2 categories: death due tospecific organ involvement with subsequent func-tional failure such as seen in many patients withmetastases to the brain, or death due to poorlydefined factors that lead to a variety of clinicalsyndromes that have a common theme of cytokineoverproduction (Figure 1).53,71–74

Cachexia

The incidence of cachexia varies by tumortype, with the highest frequency (83% to 87%)in patients with pancreatic and gastric cancer;intermediate frequency (48% to 61%) in patientswith colon, prostate, lung, and unfavorablenon-Hodgkin lymphoma; and lowest frequency(31% to 40%) in patients with breast cancer,sarcomas, leukemia, and favorable subtypes ofnon-Hodgkin lymphoma.75–77 Approximately20% of cancer deaths overall are attributable tocachexia, with death typically occurring whenweight loss approaches 30%.75–77

The inflammatory cascade set in place byhost and tumor results in an imbalance between

proinflammatory cytokines (including lipolyticfactor zinc α-2 protein (ZAG), proteolysis-inducing factor (PIF), TNF-α, IL-1, IL-6, andinterferon-γ) and anti-inflammatory cytokines(including interleukin-4, interleukin-12, andinterleukin-15) (Figure 4).50–52,75,76 Thesecytokines act on multiple targets, includingmyocytes, adipocytes, hepatocytes, bone mar-row, endothelial cells, and neurons, leading toa complex cascade of biological responses even-tually culminating in progressive weight loss,anorexia, anemia, metabolic alterations, asthe-nia, depletion of lipid stores, and severe loss ofskeletal muscle protein (Figure 3).75–77

In addition, patients often develop glucoseintolerance, insulin resistance, increased gluco-neogenesis from lactate and amino acids, increasedfat oxidation, and reduced lipogenesis as a resultof activation of futile and energy-inefficientcycles. Tumors consume a large amount of glu-cose and convert it to lactate, leading to an anaer-obic environment that does not provide a highenough oxygen tension for the Krebs cycle andmitochondrial oxidative phosphorylation to oper-ate. As a result, the Cori cycle, a much less energy-efficient cycle, is used for gluconeogenesis.75

There is an overall increase in lipolysis in patientswith cancer, resulting in glycerol and fatty acids,which can be utilized for gluconeogenesis withinhibition of lipogenesis contributing to deple-tion of fat stores.75 The muscle hypercatabolismobserved in cancer cachexia is thought to bedependent on hyperactivation of the calcium-dependent (calpains) and the ATP-ubiquitin-



TABLE 3 Frequent Sites of Metastases of Common Cancers

Total Cancer Frequency of Metastasis (%) (at Autopsy)

Primary Site Deaths (%)* Lymph Node Lung Pleura Liver Bone Brain

Lung 31†/26‡ 92–93§ 40 28 51–55 30–41 21–50¶Breast 15‡ 80–97 60–62 36–47 49–61 47–60 9–26Colon 10 25–77 12–54 14 36–81 1–18 1–8Prostate 9† 71–87 15–64 13–18 28–71 79–91 2–13Pancreas 6 50–88 25–49 18 75–78 16–18 2Ovary 6‡ 58–91 10–37 33 42–51 12–15 1–4All epithelial cancers 93†/91‡ 87 48 22 41 32 8

* Percent of estimated total cancer deaths in 2005 as reported by the American Cancer Society.1

† Male-specific percentage.‡ Female-specific percentage.§ Frequency range as reported from multiple autopsy series.57–70 Single-digit frequency as reported.59

¶ Histologic-subtype dependent.


merican C



Dow

nloaded from


dependent proteolytic pathways by cytokines.76

The progressive loss of muscle mass observed inpatients with cancer cachexia contributes signif-icantly to overall functional impairments, respi-ratory muscle weakness, and decreased immunity,ultimately culminating in death of the patient.72,74–77

Treatment of cancer cachexia was initiallyaimed at nutritional intervention.78 However,aggressive nutritional therapy did not show sig-nificant improvement in weight, lean bodymass, performance status, or quality of life.79–81

The understanding of the signal transductionand metabolic pathways associated with can-cer cachexia has opened several areas of poten-tial as well as active investigation to help patientssuffering from this syndrome (Table 4). Currenttherapies focus on affecting the hunger path-ways with goals of increasing appetite and inhibit-ing catabolic factors.94 One approach to increasingappetite is to modify hypothalamic-derived sig-nals to suppress cachexia. The best-known agentsof this type are megestrol and medroxyproges-terone acetate. Several randomized trials haveshown these agents to increase appetite andcaloric intake and stabilize weight; however, the

weight gain has been attributed to water and fatand not lean muscle tissue.82 It is unclear howwell these agents affect morbidity and mortal-ity.83 Other agents that affect central nervoussystem signaling are under active development,including melatonin receptor antagonists andagouti-related protein, as well as neuropeptide Ymimetics.84–86,95

In addition to central nervous system manip-ulations to treat cachexia, affecting hormonesthat act in periphery in muscle and fat cells alsoholds promise for cachexia treatment. Growthhormone, as well as growth hormone-releasinghormone, which stimulate increase in musclemass, have not been studied to ameliorate can-cer cachexia.85 Insulin resistance, although coun-terintuitive in a patient population with littleadiposity, occurs due to activation of adipocyteswith release of free fatty acids. Therefore, treat-ment of cancer cachexia with a class of drugsknown to enhance tissue insulin sensitivity, suchas the thiazolidinediones, may be of therapeuticbenefit.86 These drugs function as high-affinityligands for peroxisome proliferator-activatedreceptor-γ, which is the nuclear receptor in fat



FIGURE 4 The Cachexia Syndrome. The inflammatory cascade set in place by host and tumor result in an imbalancebetween proinflammatory cytokines that act on multiple targets, including myocytes, adipocytes, hepatocytes, bone mar-row, endothelial cells, and neurons, leading to production of a complex cascade of biological responses eventually culmi-nating in progressive weight loss, anorexia, anemia, and asthenia. TC � tumor cell; IL-1 � interleukin-1; IL-6 �interleukin-6; IL-11 � interleukin-11; LMF � lipid-mobilizing factor; IFNγ � interferon gamma; TNF-α � tumor necrosisfactor-α; PIF � proteolysis-inducing factor.


merican C



Dow

nloaded from


cells that is thought to be associated with weightgain in type II diabetes.86 This class of drugs mayalso suppress the hyperinsulinemia seen withcachexia that activates the hypothalamic axis,resulting in decreased orexigenic signaling.

NF-κB has also been implicated in playinga major role in cancer cachexia. By interactionwith proinflammatory cytokines, NF-κB acti-vation leads to suppression of myogenesis.Therefore, inhibition of NF-κB is postulatedto stimulate recovery of lost muscle mass.86

Several agents interfere with the synthesis andrelease of these cytokines by interfering withNF-κB, including eicosapentaenoic acid,dehydroepiandrosterone, pentoxifylline, cur-cumin, resvertrol, dehydroxymethylepoxy-quinomicin, and sodium salicylate.86–88 Fearonet al recently reported a trial comparing eicos-apentaenoic acid to placebo for treatment ofcancer cachexia in a double-blind, placebo-con-trolled trial of 518 patients with advanced gas-trointestinal and lung cancer that demonstratedno increase in survival. This trial may have beennegative because these types of agents may needto be utilized earlier at the onset of cachexia.

The major proinflammatory cytokines asso-ciated with cancer cachexia, TNF-α, IL-1, andIL-6, all offer potential targets for therapy.

Monoclonal antibodies that inhibit TNF-α havebeen utilized in small trials to treat cancer-asso-ciated cachexia, but have not demonstrated muchactivity.89,90 This may be because TNF-α levels varyin patients, and antibody therapy may need to betargeted to patients with high levels of particularcytokines. Other potential therapies include therecombinant interleukin-1 receptor (rIL-1r) antag-onist anakinra and antibodies to IL-6, both ofwhich are in clinical trials for rheumatoid dis-eases.96–98 It has become clear that cachexia is amultifactorial process that will likely need to beapproached from different angles. Much like thedisappointing results of single-agent therapy fortreating cancer itself, we should not be disap-pointed from trials of single interventions, as ulti-mately a combination approach will be needed.

Thrombotic Syndromes

The association between venous thromboem-bolism, coagulopathy, and malignancy was firstmade by Trousseau in 1877, with his descrip-tion of migratory thrombophlebitis and pancre-atic cancer.91 Since that time, thrombosis hasbecome recognized as a common complicationof cancer associated with significant morbidityand reduced survival.92,93,99 Although coagu-lopathy is only directly related to death in approx-imately 10% of cases, it has been demonstratedto be present in as high as 50% of patients at thetime of death.100–104

The characteristics that facilitate cancer cells’ability to invade locally and metastasize also resultin damage to endothelial cells and activation ofthe coagulation cascade, resulting in Virchow’striad of hypercoagulation, stasis, and endothe-lial cell damage (Figure 5). The procoagulant,



Target Examples of Potential Treatments

Metabolic pathwaysCentral nervous system Megestrol82,83

Peripheral rGrowth hormone, rGH-RH84,85

Thiazolidinediones86

CytokinesGeneral transcription of NF-κB Small molecules (EPA, curcumin, resvertrol, DHMEQ, sodium salicylate, DHEA)86–88

TNF-α mAbs (infliximab, adalimumab)Small molecules (etanercept, pentoxifylline)89,90

IL-1 rIL-1r antagonist (anakinra)91,92

IL-6 mAbs (tocilizumab, CNTO328)92,93

TABLE 4 Potential Therapeutic Approaches to Cancer Cachexia

NF-κB � nuclear factor �B.TNF-α � tumor necrosis factor-α.IL-1 � interleukin-1.IL-6 � interleukin-6.rGH-RH � recombinant growth hormone-releasinghormone.EPA � eicosapentaenoic acid.DHMEQ � dehydroxymethylepoxyquinomicin.DHEA � dehydroepiandrosterone.mAb � monoclonal antibody.rIL-1r � recombinant interleukin-1 receptor.


merican C



Dow

nloaded from


fibrinolytic, and proaggregating activities oftumor cells set up the perfect local environmentfor thrombosis. To break down the surround-ing microenvironment and allow the tumor massto grow, the cancer cells to move, and growingblood vessels to reach the tumor mass, the cellular programs that are used in wound heal-ing are activated, and cytokines and growth fac-tors are released that have local and systemiceffects. These factors include thrombin, VEGF,TNF-α, interleukin-1 β, uPA, MMPs, cathep-sins, and tissue factor (TF).16–19

TF, for example, is physiologically involvedin initiating molecular events leading to hemo-stasis by formation of a Factor VII/TF com-plex.105 The hemostatic process leads toactivation of thrombin and, therefore, conver-sion of fibrinogen to fibrin and formation ofclot at the site of vascular injury. In addition, theformation of new blood vessels associated withtumor growth results in changes in vascularpermeability, extravasation of plasma proteins,microhemorrhage, extravascular clotting, and fib-rinolysis, which contributes to the formation of

a scaffolding for new vessel development, butat the same time results in disruption of thenormal homestatic balance between coagulationand anticoagulation. Constitutive or excessiveproduction of TF by tumor cells, however, leadsto pathologic thrombosis and angiogenesis.

The potential to inhibit coagulopathies andthrombosis in cancer patients is enhanced bythe development of multiple agents for thetreatment of cardiovascular conditions.106

Multiple studies have suggested that treatmentwith anticoagulation via warfarin or variousheparins in addition to chemotherapy leads toincreased survival in patients with a variety ofcancers; however, the magnitude of the effectof anticoagulation on morbidity and mortalityfor cancer patients remains unclear.107–111

Multiple new agents that inhibit the clottingpathway are available for clinical trials in can-cer patients and include the direct thrombininhibitors, recombinant thrombomodulin, andinhibitors of TF.106–109,112–116

Factors secreted by the cancer cells, includ-ing MMPs, uPA, and cathepsins, break down



FIGURE 5 Thrombotic Syndromes and Coagulopathies. The characteristics that facilitate cancer cells’ ability to invadelocally and metastasize also result in damage to endothelial cells and activation of the coagulation cascade, resulting inVirchow’s triad of hypercoagulation, stasis, and endothelial cell damage. TC � tumor cell; L � lymphocyte; E � endothe-lial cell; TF � tissue factor; ECM � extracellular matrix; MMPs � matrix metalloproteinases; uPA � urokinase plasmino-gen activator; EGFR � epidermal growth factor receptor; IL-6 � interleukin-6; IL-1 � interleukin-1; TNF-α � tumornecrosis factor-α; TGF-β � transforming growth factor-β.


merican C



Dow

nloaded from


the extracellular matrix of the tumor microen-vironment and interrupt vascular integrity.Several small molecule inhibitors and antibod-ies to these molecules have been investigatedas single agents and in combination withchemotherapies for the treatment of multipletypes of cancer.117–123 These trials have focusedon tumor progression and/or survival, andtheir activity regarding decreasing morbidityand mortality as related to decreasing throm-bosis has not been investigated.

The rheumatoid diseases have provided thecancer community with a paradigm for the treat-ment of diseases based on the inhibition ofproinflammatory cytokines. The prototypical agentsare the monoclonal antibodies that inhibit TNF-α.89,90 As noted above, anti-TNF-α strategies mayhave value in a subset of patients suffering fromcachexia. TNF-α, however, also plays a role ininflammation associated with vascular injury.Similarly, IL-1 is a proinflammatory molecule thatalso has a role in thrombosis, and an inhibitor usedin rheumatoid diseases, anakinra, is available for

clinical trials.91 Small-molecule inhibitors and anti-bodies directed against other cytokines such asIL-6 and TGF-β are also in clinical develop-ment.92,93,124–126 Trials need to be designed with aneye to their effect on morbidity and mortality asso-ciated with coagulaopathy (Table 5).

Bone Involvement

Bone involvement is the main cause of directcancer pain. Skeletal involvement is present in anaverage of 32% of cancer patients at autopsy, withmuch higher prevalence in patients with lung,breast, kidney, and prostate cancers. In recentcareful autopsy studies, 100% of men who die ofprostate cancer have bone involvement.61,62 Aspreviously described, the activation of osteoblastsand osteoclasts by cancer cells results in a viciouscycle of bone destruction and increased tumorgrowth, resulting in pain, fractures, and spinalcord compression (Figure 6). In a significant pro-portion of patients, this pain requires narcoticanalgesia. Patients require higher and higher dosesof opioid analgesics, resulting in somnolence,sometimes with subsequent aspirations and/orcoma. Review of the autopsy series literature didnot reveal what percentage of cancer patients diewith concurrent aspiration.56–70 In our currentautopsy series of 48 patients who died of metasta-tic prostate cancer, concurrent aspiration pneumo-nia was documented in fewer than 10% of cases.62



TABLE 5 Potential Treatments to Prevent Thrombotic Syndromes in Cancer Patients

Contributor toThrombosis Examples of Potential Treatments

Thrombin Warfarin, hirudin, argatroban, rThrombomodulin107–110,112,115

TF Heparins, pentasaccharide, mAb (6A6)109,114–116

MMPs Small molecules (BMS-27529, tanomastat)117–119

uPA Small peptide (A6)120,121

Cathepsins Small molecules (relacatib), AAE581122,123

CytokinesTGF-β mAbs (lerdelimumab, metelimumab), antisense (AP12009)124–126

TNF-α mAbs (etanercept, infliximab, adalimumab)89,90

IL-1 rIL-1r antagonist (anakinra)91


Vascular traumaTumor massesInflammatory responseSurgeryVascular devices Warfarin, heparins110,111

RadiationChemotherapyInfection

TF � tissue factor.MMPs � matrix metalloproteinases.uPA � uroplasminogen activator.TGF-β � transforming growth factor-β.TNF-α � tumor necrosis factor-α.IL-1 � interleukin-1.IL-6 � interleukin-6.mAb � monoclonal antibody.rIL-1r � recombinant interleukin-1 receptor.


merican C



Dow

nloaded from


Approximately one third of the patients were inan opioid-induced coma at the time of death andhad no other discernible cause of death at autopsy.

The bone microenvironment presents severalpotential targets that mediate the effects of thevicious cycle of bone destruction. The endothe-lial cells, osteoblasts, and osteoclasts that interactwith the tumor cells all present targets for mod-ulation (Figure 6, Table 6). Clinical trials of the ET-1 receptor antagonist atrasentan to inhibitosteoblasts have been completed and are ongo-ing.127 Osteoclast destruction can be inhibited byFood and Drug Administration-approved bispho-sphonates such as zoledronate and by radioactiveisotopes that bind to hydroxyapatite (samarium,strontium).128–130 Osteoclast function can also beinhibited by the src tyrosine kinase inhibitors, aswell as by targeting the osteoblast-osteoclast axisthrough the inhibition of RANKL.131–133 Multiplestudies are utilizing inhibitors of VEGF to targettumor-related endothelial cell proliferation inbone metastases, both as single agents, as well asin combination with chemotherapy.134–137

Soluble factors and cytokines secreted by thetumor cells, as well as cells in the bone microen-vironment, also provide an array of targets. Asnoted previously, delineating potential targetsfor inhibiting thrombosis, MMPs, and cathep-sins break down the extracellular matrix andpromote tumor cell growth, invasion, and metas-tasis. Small-molecule and antibody inhibitors ofthese enzymes have demonstrated activity in avariety of cancers, including metastatic prostatecancer.117–119,122,123 EGF antibodies have demon-strated antitumor activity, but also may inhibitstimulation of endothelial cells and osteo-blasts.140,141 Stromal-derived factor-1 is a cytokinethat has been implicated in the homing of can-cer cells to the bone. AMD3100 is a small-molecule inhibitor of stromal-derived factor-1(SDF-1, CXCL12), first developed for HIVinfection, that could potentially inhibit propa-gation of metastases to the bone microenviron-ment.138 Monocyte chemoattractant protein-1(MCP-1, CCL2) is a cytokine that attracts can-cer cells, as well as proinflammatory macrophages,



FIGURE 6 Skeletal Metastases. The activation of osteoblasts and osteoclasts by cancer cells results in a vicious cycleof bone destruction and increased tumor growth, resulting in pain, fractures, and spinal cord compression. TC � tumorcell; OC � osteoclast; OB � osteoblast; SDF-1 � stromal-derived factor-1; MCP-1 � monocyte chemoattractant protein-1; EGF � epidermal growth factor; IL-6 � interleukin-6; IL-1 � interleukin-1; TNF-α � tumor necrosis factor-α;TGF-β � transforming growth factor-β; ET-1 � endothelin-1; PTHrp � parathyroid hormone-related peptide;RANKL � receptor activator of nuclear factor κB ligand.


merican C



Dow

nloaded from


to the bone microenvironment. The inhibitionof this cytokine by the antibody CNTO888appears to have direct cytotoxic effects throughinhibition of tumor cell proliferation, as well asinhibition of the infiltration of macrophages intothe tumor microenvironment that promotestumor growth and angiogenesis.141 As previouslydetailed, antagonists to the cytokines TGF-β,TNF-α, IL-1, and IL-6 may all be important inameliorating the effects of the vicious cycle oftumor-microenvironment interactions that leadsto pain caused by bone destruction.89–93,96–98,124–126

Dyspnea

Dyspnea occurs in 20% to 80% of patientswith cancer and is severe is 10% to 60% of patients,especially in the last 6 weeks of life.142–145 Breathingis controlled by the respiratory center (integratesall peripheral and central afferent input andgenerates efferent activity resulting in respira-tion), chemoreceptors (sense small changes inpH and pCO2), and mechanoreceptors (respondto irritants and stretching of airways). The causeof dyspnea in a given patient is usually multi-factorial, stemming from direct lung involve-ment, local and systemic cytokine production,treatment-related causes, and underlying dis-eases such as congestive heart failure and chronicobstructive pulmonary disease (Table 7).

Tumor burden occupying the lung paren-chyma, pulmonary lymphangetic spread of



TABLE 6 Potential Therapeutic Targets in the Bone Microenvironment

Target Examples of Potential Treatments

Osteoblast Endothelin receptor (atrasentan)127

Osteoclast Hydroxyapatite (zoledronate, samarium, strontium)128–130

src tyrosine kinase (dasatinib)131

RANKL (mAb denosumab)132,133

Endothelial cells VEGF (mAb bevcizumab, VEGF Trap)134

VEGFr tyrosine kinase (BAY43–9006, PTK787, ZD6474)135–137

MMPs Small molecules (BMS-27529, tanomastat)117–119

Cathepsins Small molecules (relacatib), AAE581122,123

CytokinesTGF-β mAbs (lerdelimumab, metelimumab), antisense (AP12009)124–126

TNF-α mAbs (etanercept, infliximab, adalimumab)89,90

IL-1 rIL-1r antagonist (anakinra)91


CXCL12/CXCR4 (SDF1) Small molecule (AMD3100)138

CCL2/CCR2 (MCP1) mAb (CNTO888)139

Epidermal growth factor EGFr mAbs (gefitinib, cetuximab, erlotinib, laputinib, trastuzumab)140,141

MMPs � matrix metalloproteinases.TGF-β � transforming growth factor-β.TNF-α � tumor necrosis factor-α.IL-1 � interleukin-1.IL-6 � interleukin-6.SDF1 � stromal-derived factor-1.MCP1 � monocyte cheomattractant protein-1.RANKL � receptor activator of nuclear factor �B ligand.mAb � monoclonal antibody.VEGF � vascular endothelial growth factor.VEGFr � vascular endothelial growth factor receptor.rIL-1r � recombinant interleukin-1 receptor.EGFr � epidermal growth factor receptor.

TABLE 7 Causes of Dyspnea in Malignancy

Dyspnea directly related to cancerParenchymal tumor massLymphangitic spreadPleural effusionSuperior vena cava syndromePericardial effusionAscites

Dyspnea indirectly related to cancerCachexiaAnemiaInfectionEmboliDeconditioning

Dyspnea related to cancer treatmentRadiation/chemotherapy-induced pnuemonitisRadiation/chemotherapy-induced percarditisSurgical resection of lung parenchyma

Dyspnea unrelated to cancer or cancer treatmentPulmonary disease (chronic obstructive pulmonary disease,

asthma)Cardiac disease (coronary artery disease, congestive heart

failure )AnxietyObesity


merican C



Dow

nloaded from


disease, malignant pleural effusions, and pul-monary embolism are common, well-recog-nized causes of dyspnea. Treatment of canceritself can contribute to the dyspnea experiencedby patients through radiation- or chemother-apy-induced pneumonitis and drug-related pleu-ral effusions, through pneumonia secondary toneutropenia, and through tachypnea due to ane-mia. Cachexia can result in respiratory muscleweakness. Inactivity can lead to decondition-ing, and decreased consciousness from pain con-trol can also lead to deconditioning, as well asaspiration pneumonia.

Cancer-related dyspnea is generally consid-ered to be a late event in the disease course, andsystematic approaches to treatment beyond tar-geting identifiable causes such as anemia havenot been undertaken. It is speculative, but likely,that inhibition of the proinflammatory cytokines

that have already been delineated above mayhave a role in decreasing the morbidity and mor-tality associated with dyspnea.89–93,96–98,124–126

CONCLUSION AND IMPLICATIONS

The disease cancer is the result of a complexinterplay between the growing tumor and localand systemic responses by the patient to the pres-ence of malignancy. Traditionally, cancer ther-apy has focused on cytotoxic agents rather thantherapies that ameliorate the effects of byprod-ucts of the cancer cells or the proinflammatoryhost response to their presence. Insight into themolecular events underlying the lethal clinicalsyndromes that contribute to the morbidity andmortality of cancer patients suggests avenues oftreatment, many of which have already beenexplored in other disease settings.



REFERENCES

1. Jemal A, Siegel R, Ward E, et al. Cancer sta-tistics, 2007. CA Cancer J Clin 2007;57:43–66.

2. Pienta KJ, Loberg R. The “emigration, migra-tion, and immigration”of prostate cancer. ClinProstate Cancer 2005;4:24–30.

3. Fidler IJ. The pathogenesis of cancer metasta-sis: the ‘seed and soil’ hypothesis revisited. Nat RevCancer 2003;3:453–458.

4. Paget S. The distribution of secondary growthsin cancer of the breast. Lancet 1889;1:571–573.

5. Ewing J. Metastasis, in Ewing J. Neoplastics.3rd ed. Philadelphia, PA: Saunders; 1928:77–89.

6. Radman M, Matic I, Taddei F. Evolution ofevolvability. Ann NY Acad Sci 1999;870:146–155.

7. Greaves M. Cancer causation: the Darwiniandownside of past success? Lancet Oncol 2002;3:244–251.

8. Nowell PC. The clonal evolution of tumor cellpopulations. Science 1976;194:23–28.

9. Hanahan D, Weinberg RA. The hallmarks ofcancer. Cell 2000;100:57–70.

10. Nesse RM, Williams GC. Evolution and theorigins of disease. Sci Am 1998;279:86–93.

11. Coffey DS. Similarities of prostate and breastcancer: Evolution, diet, and estrogens. Urology2001;57:31–38.

12. Farinati F, Cardin R, Della Libera G, et al.The role of anti-oxidants in the chemopreven-tion of gastric cancer. Eur J Cancer Prev 1994;3(suppl):93–97.

13. Sugarbaker PH. Metastatic inefficiency: the sci-entific basis for resection of liver metastases from col-orectal cancer. J Surg Oncol 1993;3(suppl):158–160.

14. Luzzi KJ, MacDonald IC, Schmidt EE, et al.Multistep nature of metastatic inefficiency: dor-mancy of solitary cells after successful extravasa-tion and limited survival of early micrometastases.Am J Pathol 1998;153:865–873.

15. Glaves D, Huben RP, Weiss L. Haematogenousdissemination of cells from human renal adeno-carcinomas. Br J Cancer 1988;57:32–35.

16. Yang J, Mani SA, Donaher JL, et al. Twist, amaster regulator of morphogenesis, plays an essen-tial role in tumor metastasis. Cell 2004;117:927–939.

17. Grego-Bessa J, Diez J, Timmerman L, de laPompa JL. Notch and epithelial-mesenchyme tran-sition in development and tumor progression: anotherturn of the screw. Cell Cycle 2004;3:718–721.

18. Liotta LA, Kohn EC. The microenvironment ofthe tumour-host interface. Nature 2001;411:375–379.

19. Bissell MJ, Radisky D. Putting tumours incontext. Nat Rev Cancer 2001;1:46–54.

20. Ramaswamy S, Ross KN, Lander ES, GolubTR. A molecular signature of metastasis in pri-mary solid tumors. Nat Genet 2003;33:49–54.

21. Glinsky GV, Berezovska O, Glinskii AB.Microarray analysis identifies a death-from-cancer signature predicting therapy failure inpatients with multiple types of cancer. J ClinInvest 2005;115:1503–1521.

22. Varambally S, Yu J, Laxman B, et al. Integrativegenomic and proteomic analysis of prostate can-cer reveals signatures of metastatic progression.Cancer Cell 2005;8:393–406.

23. Minn AJ, Kang Y, Serganova I, et al. Distinctorgan-specific metastatic potential of individualbreast cancer cells and primary tumors. J Clin Invest2005;115:44–55.

24. D’Arrigo A, Belluco C, Ambrosi A, et al.Metastatic transcriptional pattern revealed by geneexpression profiling in primary colorectal carci-noma. Int J Cancer 2005;115:256–262.

25. Budhu AS, Zipser B, Forgues M, et al. Themolecular signature of metastases of human hepato-cellular carcinoma. Oncology 2005;69(suppl):23–27.

26. Jones J, Otu H, Spentzos D, et al. Gene signa-tures of progression and metastasis in renal cell can-cer. Clin Cancer Res 2005;11:5730–5739.

27. O’Donnell RK, Kupferman M, Wei SJ, et al.Gene expression signature predicts lymphatic metas-tasis in squamous cell carcinoma of the oral cav-ity. Oncogene 2005;24:1244–1251.

28. Wang W, Wyckoff JB, Frohlich VC, et al. Singlecell behavior in metastatic primary mammarytumors correlated with gene expression patternsrevealed by molecular profiling. Cancer Res2002;62:6278–6288.

29. Pantel K, Woelfle U. Micrometastasis in breastcancer and other solid tumors. J Biol RegulHomeost Agents 2004;18:120–125.

30. van de Vijver MJ, He YD, van’t Veer LJ,et al. A gene-expression signature as a predictorof survival in breast cancer. N Engl J Med 2002;347:1999–2009.

31. Weigelt B, Hu Z, He X, et al. Molecular por-traits and 70-gene prognosis signature are preservedthroughout the metastatic process of breast can-cer. Cancer Res 2005;65:9155–9158.

32. Singh D, Febbo PG, Ross K, et al. Gene expres-sion correlates of clinical prostate cancer behavior.Cancer Cell 2002;1:203–209.

33. Dhanasekaran SM, Barrette TR, Ghosh D,et al. Delineation of prognostic biomarkers inprostate cancer. Nature 2001;412:822–826.

34. Woelfle U, Cloos J, Sauter G, et al. Molecularsignature associated with bone marrow micro-metastasis in human breast cancer. Cancer Res2003;63:5679–5684.

35. Cancer Bioinformatics, Comprehensive CancerCenter, University of Michigan. Cancer ProfilingDatabase: ONCOMINE™. Available at: www.oncomine.org. Accessed March 12, 2007.

36. Chung CH, Bernard PS, Perou CM. Molecularportraits and the family tree of cancer. Nat Genet2002;32(suppl):533–540.

37. Rhodes DR, Yu J, Shanker K, et al. Large-scale meta-analysis of cancer microarray data iden-tifies common transcriptional profiles of neoplastic


merican C



Dow

nloaded from




transformation and progression. Proc Natl AcadSci USA 2004;101:9309–9314.

38. Rhodes DR, Kalyana-Sundaram S, MahavisnoV, et al. Mining for regulatory programs in the can-cer transcriptome. Nat Genet 2005;37:579–583.

39. Rhodes DR, Chinnaiyan AM. Integrativeanalysis of the cancer transcriptome. Nat Genet2005;37(suppl):S31–S37.

40. Tomlins SA, Mehra R, Rhodes DR, et al.Integrative molecular concept modeling of prostatecancer progression. Nat Genet 2007;39:41–51.

41. Yang H, Crawford N, Lukes L, et al. Metastasispredictive signature profiles pre-exist in normaltissues. Clin Exp Metastasis 2005;22:593–603.

42. Hunter KW. Host genetics and tumour metas-tasis. Br J Cancer 2004;90:752–755.

43. Shen R, Ghosh D, Chinnaiyan AM. Prognosticmeta-signature of breast cancer developed by two-stage mixture modeling of microarray data. BMCGenomics 2004;5:94.

44. Li HR, Wang-Rodriguez J, Nair TM, et al.Two-dimensional transcriptome profiling: identi-fication of messenger RNA isoform signatures inprostate cancer from archived paraffin-embeddedcancer specimens. Cancer Res 2006;66:4079–4088.

45. Hoang CD, Guillaume TJ, Engel SC, et al.Analysis of paired primary lung and lymph nodetumor cells: a model of metastatic potential bymultiple genetic programs. Cancer Detect Prev2005;29:509–517.

46. Niklaus AL, Pollard JW. Mining the mousetranscriptome of receptive endometrium revealsdistinct molecular signatures for the luminal andglandular epithelium. Endocr inology 2006;147:3375–3390.

47. Folkman J. Toward an understanding of angio-genesis: search and discovery. Perspect Biol Med1985;29:10–36.

48. Giaccia A, Siim BG, Johnson RS. HIF-1 as atarget for drug development. Nat Rev Drug Discov2003;2:803–811.

49. Semenza GL. Targeting HIF-1 for cancer ther-apy. Nat Rev Cancer 2003;3:721–732.

50. Balkwill F, Mantovani A. Inflammation andcancer: back to Virchow? Lancet 2001;357:539–545.

51. Coussens LM, Werb Z. Inflammation and can-cer. Nature 2002;420:860–867.

52. Lu H, Ouyang W, Huang C. Inflammation, akey event in cancer development. Mol Cancer Res2006;4:221–233.

53. Loberg RD, Logothetis CJ, Keller ET, PientaKJ. Pathogenesis and treatment of prostate cancerbone metastases: targeting the lethal phenotype.J Clin Oncol 2005;23:8232–8241.

54. Kozlow W, Guise TA. Breast cancer metasta-sis to bone: mechanisms of osteolysis and implica-tions for therapy. J Mammary Gland Biol Neoplasia2005;10:169–180.

55. Kakonen SM, Mundy GR. Mechanisms ofosteolytic metastases in breast carcinoma. Cancer2003;97(suppl):834–839.

56. Cancer Bioinformatics, Comprehensive CancerCenter, University of Michigan. Molecular ConceptMap. Available at: http://www.molecularconcepts.org/main/review.jsp. Accessed March 12, 2007.

57. Willis RA. The Pathology of Tumours. 4thed. London, UK: Butterworths; 1967.

58. Welch JP, Donaldson GA. The clinical corre-lation of an autopsy study of recurrent colorectalcancer. Ann Surg 1979;189:496–502.

59. Berge T, Lundberg S. Cancer in Malmo1958–1969. An autopsy study. Acta Pathol MicrobiolScand 1977;260(suppl):1–235.

60. Hassan I. eMedicine from WebMD. Lung,Metastases. Available at: http://www.emedicine.com/radio/topic404.htm. Accessed March 13,2007.

61. Roudier MP, Vesselle H, True LD, et al. Bonehistology at autopsy and matched bone scintigra-phy findings in patients with hormone refractoryprostate cancer: the effect of bisphosphonate ther-apy on bone scintigraphy results. Clin Exp Metastasis2003;20:171–180.

62. Shah RB, Mehra R, Chinnaiyan AM, et al.Androgen-independent prostate cancer is a het-erogeneous group of diseases: lessons from a rapidautopsy program. Cancer Res 2004;64:9209–9216.

63. Pickren JW, Tsukada Y, Lane WW. Liver metas-tasis, in Weiss L, Gilbert HA (eds). Analysis ofAutopsy Data. Boston, MA: GK Hall and Company;1982:2–18.

64. Weiss L, Grundmann E, Torhorst J, et al.Haematogenous metastatic patterns in colonic car-cinoma: an analysis of 1541 necropsies. J Pathol1986;150:195–203.

65. Reed E, Zerbe CS, Brawley OW, et al. Analysisof autopsy evaluations of ovarian cancer patientstreated at the National Cancer Institute, 1972–1988.Am J Clin Oncol 2000;23:107–116.

66. Lee YT, Tatter D. Carcinoma of the pan-creas and periampullary structures. Pattern ofmetastasis at autopsy. Arch Pathol Lab Med1984;108:584–587.

67. Lee YT. Breast carcinoma: pattern of metas-tasis at autopsy. J Surg Oncol 1983;23:175–180.

68. Cox JD, Yesner RA. Causes of treatment fail-ure and death in carcinoma of the lung. Yale J BiolMed 1981;54:201–207.

69. Cho SY, Choi HY. Causes of death andmetastatic patterns in patients with mammarycancer. Ten-year autopsy study. Am J Clin Pathol1980;73:232–234.

70. Viadana E, Au K-L. Patterns of metastases inadenocarcinomas of man. An autopsy study of4,728 cases. J Med 1975;6:1–14.

71. Dewys WD, Begg C, Lavin PT, et al. Prognosticeffect of weight loss prior to chemotherapy in can-cer patients. Eastern Cooperative Oncology Group.Am J Med 1980;69:491–497.

72. Inagaki J, Rodriguez V, Bodey GP. Proceedings:Causes of death in cancer patients. Cancer 1974;33:568–573.

73. Prandoni P, Falanga A, Piccioli A. Cancer andvenous thromboembolism. Lancet Oncol 2005;6:401–410.

74. Dudgeon DJ, Kristjanson L, Sloan JA, et al.Dyspnea in cancer patients: prevalence and associ-ated factors. J Pain Symptom Manage 2001;21:95–102.

75. Tisdale MJ. Molecular pathways leading to can-cer cachexia. Physiology (Bethesda) 2005;20:340–348.

76. Argiles JM, Busquets S, Felipe A, Lopez-SorianoFJ. Molecular mechanisms involved in muscle wast-ing in cancer and ageing: cachexia versus sarcope-nia. Int J Biochem Cell Biol 2005;37:1084–1104.

77. Muscaritoli M, Bossola M, Aversa Z, et al.Prevention and treatment of cancer cachexia:

new insights into an old problem. Eur J Cancer2006;42:31–41.

78. Capra S, Ferguson M, Ried K. Cancer: impactof nutrition intervention outcome—nutrition issuesfor patients. Nutrition 2001;17:769–772.

79. Evans WK, Nixon DW, Daly JM, et al. A ran-domized study of oral nutritional support versusad lib nutritional intake during chemotherapy foradvanced colorectal and non-small-cell lung caner.J Clin Oncol 1987;5:113–124.

80. Ovesen L, Allingstrup L, Hannibal J, et al.Effect of dietary counseling on food intake, bodyweight, response rate, survival, and quality of lifein cancer patients undergoing chemotherapy: aprospective, randomized study. J Clin Oncol1993;11:2043–2049.

81. Klein S, Kinney J, Jeejeebhoy K, et al. Nutritionsupport in clinical practice: review of publisheddata and recommendations for future research direc-tions. Summary of a conference sponsored by theNational Institutes of Health, American Societyfor Parenteral and Enteral Nutrition, and AmericanSociety for Clinical Nutrition. Am J Clin Nutr1997;66:683–706.

82. Pascual Lopez A, Roque i Figuls M, UrrutiaCuchi G, et al. Systematic review of megestrolacetate in the treatment of anorexia-cachexia syn-drome. J Pain Symptom Manage 2004;27:360–369.

83. Jatoi A, Rowland K, Loprinzi CL, et al. Aneicosapentaenoic acid supplement versus mege-strol acetate versus both for patients with cancer-associated wasting: a North Central CancerTreatment Group and National Cancer Instituteof Canada collaborative effort. J Clin Oncol2004;22:2469–2476.

84. Nagaya N, Kojima M, Kangawa K. Ghrelin, anovel growth hormone-releasing peptide, in thetreatment of cardiopulmonary-associated cachexia.Intern Med 2006;45:127–134.

85. Nagaya N, Itoh T, Murakami S, et al. Treatmentof cachexia with ghrelin in patients with COPD.Chest 2005;128:1187–1193.

86. Hamerman D. Molecular-based therapeuticapproaches in treatment of anorexia of aging andcancer cachexia. J Gerontol A Biol Sci Med Sci2002;57:M511-M518.

87. Boddaert MS, Gerritsen WR, Pinedo HM.On our way to targeted therapy for cachexia incancer? Curr Opin Oncol 2006;18:335–340.

88. Fearon KC, Barber MD, Moses AG, et al.Double-blind, placebo-controlled, randomizedstudy of eicosapentaenoic acid diester in patientswith cancer cachexia. J Clin Oncol 2006;24:3401–3407.

89. Bongartz T, Sutton AJ, Sweeting MJ, et al.Anti-TNF antibody therapy in rheumatoid arthri-tis and the risk of serious infections and malignan-cies: systematic review and meta-analysis of rareharmful effects in randomized controlled trials.JAMA 2006;295:2275–2285.

90. Chen YF, Jobanputra P, Barton P, et al. A sys-tematic review of the effectiveness of adalimumab,etanercept and infliximab for the treatment ofrheumatoid arthritis in adults and an economicevaluation of their cost-effectiveness. Health TechnolAssess 2006;10:iii-iv,xi-xiii,1-229.

91. Trousseau A. Phlegmasia alba dolens, inClinique Medicale de L’Hotel Dieu de Paris. 2nded. Paris, France: Bailliere; 1865:695–739.


merican C



Dow

nloaded from




92. Sallah S, Wan JY, Nguyen NP. Venous throm-bosis in patients with solid tumors: determinationof frequency and characteristics. Thromb Haemost2002;87:575–579.

93. Levitan N, Dowlati A, Remick SC, et al. Ratesof initial and recurrent thromboembolic diseaseamong patients with malignancy versus those with-out malignancy. Risk analysis using Medicare claimsdata. Medicine (Baltimore) 1999;78:285–291.

94. Elamin EM, Glass M, Camporesi E. Pharma-cological approaches to ameliorating catabolicconditions. Curr Opin Clin Nutr Metab Care2006;9:449–454.

95. DeBoer MD, Marks DL. Cachexia: lessonsfrom melanocortin antagonism. Trends EndocrinolMetab 2006;17:199–204.

96. Lewis AM, Varghese S, Xu H, Alexander HR.Interleukin-1 and cancer progression: the emerg-ing role of interleukin-1 receptor antagonist as anovel therapeutic agent in cancer treatment. J TranslMed 2006;4:48.

97. Nishimoto N, Kishimoto T. Interleukin 6:from bench to bedside. Nat Clin Pract Rheumatol2006;2:619–626.

98. Wallner L, Dai J, Escara-Wilke J, et al. Inhibitionof interleukin-6 with CNTO328, an anti-inter-leukin-6 monoclonal antibody, inhibits conver-sion of androgen-dependent prostate cancer to anandrogen-independent phenotype in orchiec-tomized mice. Cancer Res 2006;66:3087–3095.

99. Sorensen HT, Mellemkjaer L, Olsen JH, BaronJA. Prognosis of cancers associated with venous throm-boembolism. N Engl J Med 2000;343:1846–1850.

100. Svendsen E, Karwinski B. Prevalence of pul-monary embolism at necropsy in patients with can-cer. J Clin Pathol 1989;42:805–809.

101. Ogren M, Bergqvist D, Wahlander K, et al.Trousseau’s syndrome—what is the evidence? Apopulation-based autopsy study. Thromb Haemost2006;95:541–545.

102. Mousa SA. Role of current and emergingantithrombotics in thrombosis and cancer. DrugsToday (Barc) 2006;42:331–350.

103. Deitcher SR. Cancer and thrombosis: mech-anisms and treatment. J Thromb Thrombolysis2003;16:21–31.

104. Lee AY. Epidemiology and management ofvenous thromboembolism in patients with cancer.Thromb Res 2003;110:167–172.

105. Rak J, Milsom C, May L, et al. Tissue factorin cancer and angiogenesis: the molecular linkbetween genetic tumor progression, tumor neo-vascularization, and cancer coagulopathy. SeminThromb Hemost 2006;32:54–70.

106. Hiller E. Cancer and thrombosis: managingthe risks and approaches to thromboprophylaxis.Onkologie 2006;29:474–478.

107. Nakchbandi W, Muller H, Singer MV, et al.Effects of low-dose warfarin and regional chemother-apy on survival in patients with pancreatic carci-noma. Scand J Gastroenterol 2006;41:1095–1104.

108. Saito H, Maruyama I, Shimazaki S, et al.Efficacy and safety of recombinant human solublethrombomodulin (ART-123) in disseminatedintravascular coagulation: results of a phase III, ran-domized, double-blind clinical trial. J ThrombHaemost 2007;5:31–41.

109. Kakkar AK, Levine MN, Kadziola Z, et al.Low molecular weight heparin, therapy with dal-

teparin, and survival in advanced cancer: the frag-min advanced malignancy outcome study(FAMOUS). J Clin Oncol 2004;22:1944–1948.

110. Fagnani D, Franchi R, Porta C, et al.Thrombosis-related complications and mortalityin cancer patients with central venous devices: anobservational study on the effect of antithromboticprophylaxis. Ann Oncol 2007;18:551-555.

111. Linenberger ML. Catheter-related throm-bosis: risks, diagnosis, and management. J NatlCompr Canc Netw 2006;4:889–901.

112. Schwienhorst A. Direct thrombin inhibitors—a survey of recent developments. Cell Mol Life Sci2006;63:2773–2791.

113. Yeh RW, Jang IK. Argatroban: update. AmHeart J 2006;151:1131–1138.

114. Lee AY, Levine MN, Baker RI, et al. Low-molecular-weight heparin versus a coumarin forthe prevention of recurrent venous thromboem-bolism in patients with cancer. N Engl J Med2003;349:146–153.

115. Mousa SA. Arixtra (fondaparinux sodium).Methods Mol Med 2004;93:239–246.

116. Eigenbrot C, Meng YG, Krishnamurthy R,et al. Structural insight into how an anti-idiotypicantibody against D3H44 (anti-tissue factor anti-body) restores normal coagulation. J Mol Biol2003;331:433–446.

117. Hirte H, Vergote IB, Jeffrey JR, et al. A phaseIII randomized trial of BAY 12–9566 (tanomas-tat) as maintenance therapy in patients with advancedovarian cancer responsive to primary surgery andpaclitaxel/platinum containing chemotherapy: aNational Cancer Institute of Canada Clinical TrialsGroup Study. Gynecol Oncol 2006;102:300–308.

118. Lara PN Jr, Stadler WM, Longmate J, et al.A randomized phase II trial of the matrix metallo-proteinase inhibitor BMS-275291 in hormone-refractory prostate cancer patients with bonemetastases. Clin Cancer Res 2006;12:1556–1563.

119. Overall CM, Kleifeld O. Towards third gen-eration matrix metalloproteinase inhibitors for can-cer therapy. Br J Cancer 2006;94:941–946.

120. Berkenblit A, Matulonis UA, Kroener JF,et al. A6, a urokinase plasminogen activator (uPA)-derived peptide in patients with advanced gyne-cologic cancer: a phase I trial. Gynecol Oncol2005;99:50–57.

121. Duffy MJ. The urokinase plasminogen acti-vator system: role in malignancy. Curr Pharm Des2004;10:39–49.

122. Leung-Toung R, Zhao Y, Li W, et al. Thiolproteases: inhibitors and potential therapeutic tar-gets. Curr Med Chem 2006;13:547–581.

123. Bierie B, Moses HL. Tumour microenviron-ment: TGFbeta: the molecular Jekyll and Hyde ofcancer. Nat Rev Cancer 2006;6:506–520.

124. Iyer S, Wang ZG, Akhtari M, et al. TargetingTGFbeta signaling for cancer therapy. Cancer BiolTher 2005;4:261–266.

125. Yingling JM, Blanchard KL, Sawyer JS.Development of TGF-beta signaling inhibitorsfor cancer therapy. Nat Rev Drug Discov 2004;3:1011–1022.

126. Kumar S, Dare L, Vasko-Moser JA, et al.A highly potent inhibitor of cathepsin K (rela-catib) reduces biomarkers of bone resorptionboth in vitro and in an acute model of elevated

bone turnover in vivo in monkeys. Bone 2007;40:122–131.

127. Carducci MA, Jimeno A. Targeting bonemetastasis in prostate cancer with endothelin recep-tor antagonists. Clin Cancer Res 2006;12:6296s-6300s.

128. Santini D, Fratto ME, Vincenzi B, et al.Zoledronic acid in the management of metastaticbone disease. Expert Opin Biol Ther 2006;6:1333–1348.

129. Lin A, Ray ME. Targeted and systemic radio-therapy in the treatment of bone metastasis. CancerMetastasis Rev 2006;25:669–675.

130. Bauman G, Charette M, Reid R, Sathya J.Radiopharmaceuticals for the palliation of painfulbone metastasis—a systemic review. RadiotherOncol 2005;75:258–270.

131. Cortes J, Rousselot P, Kim DW, et al. Dasatinibinduces complete hematologic and cytogeneticresponses in patients with imatinib-resistant or -intolerant chronic myeloid leukemia in blast cri-sis. Blood 2007;109:3207–3213.

132. Close P, Neuprez A, Reginster JY. Develop-ments in the pharmacotherapeutic managementof osteoporosis. Expert Opin Pharmacother2006;7:1603–1615.

133. Lewiecki EM. RANK ligand inhibition withdenosumab for the management of osteoporosis.Expert Opin Biol Ther 2006;6:1041–1050.

134. Tonra JR, Hicklin DJ. Targeting the vascu-lar endothelial growth factor pathway in the treat-ment of human malignancy. Immunol Invest2007;36:3–23.

135. Baka S, Clamp AR, Jayson GC. A review ofthe latest clinical compounds to inhibit VEGF inpathological angiogenesis. Expert Opin Ther Targets2006;10:867–876.

136. Lierman E, Lahortiga I, Van Miegroet H,et al. The ability of sorafenib to inhibit oncogenicPDGFRbeta and FLT3 mutants and overcomeresistance to other small molecule inhibitors.Haematologica 2007;92:27–34.

137. Herbst RS, Heymach JV, O’Reilly MS, et al.Vandetanib (ZD6474): an orally available receptortyrosine kinase inhibitor that selectively targetspathways critical for tumor growth and angiogen-esis. Expert Opin Investig Drugs 2007;16:239–249.

138. Cashen AF, Nervi B, DiPersio J. AMD3100:CXCR4 antagonist and rapid stem cell-mobilizingagent. Future Oncol 2007;3:19–27.

139. Loberg RD, Day LL, Harwood J, et al. CCL2is a potent regulator of prostate cancer cell migra-tion and proliferation. Neoplasia 2006;8:578–586.

140. Reid A, Vidal L, Shaw H, de Bono J. Dualinhibition of ErbB1 (EGFR/HER1) and ErbB2(HER2/neu). Eur J Cancer 2007;43:481–489.

141. Astsaturov I, Cohen RB, Harari PM. EGFR-targeting monoclonal antibodies in head and neckcancer. Curr Cancer Drug Targets 2006;6:691–710.

142. Ripamonti C. Management of dyspnea inadvanced cancer patients. Support Care Cancer1999;7:233–243.

143. Ripamonti C, Bruera E. Dyspnea: patho-physiology and assessment. J Pain Symptom Manage1997;13:220–232.

144. Thomas JR, Von Gunten CF. Treatment ofdyspnea in cancer patients. Oncology (WillistonPark) 2002;16:745–750.

145. Reuben DB, Mor V. Dyspnea in terminallyill cancer patients. Chest 1986;89:234–236.


merican C



Dow

nloaded from


the lethal phenotype of cancer: the molecular basis of ... · the lethal phenotype of cancer: the...

Documents