bases moleculares de metastasis

Upload: pablo-paricahua

Post on 04-Apr-2018

217 views

Category:

Documents


0 download

TRANSCRIPT

  • 7/31/2019 Bases Moleculares de Metastasis

    1/10

    review article

    T h e n e w e n g l a n d j o u r n a l o f medicine

    n engl j med 359;26 www.nejm.org december 25, 20082814

    Molecular Origins of Cancer

    Molecular Basis of Metastasis

    Anne C. Chiang, M.D., Ph.D., and Joan Massagu, Ph.D.

    From the Department of Medicine (A.C.C.),the Cancer Biology and Genetics Program(J.M.), and the Howard Hughes MedicalInstitute (J.M.), Memorial Sloan-KetteringCancer Center, New York. Address reprintrequests to Dr. Massagu at Box 116, Me-morial Sloan-Kettering Cancer Center,1275 York Ave., New York, NY 10065, orat [email protected].

    N Engl J Med 2008;359:2814-23.Copyright 2008 Massachusetts Medical Society.

    Metastasis is the end product of an evolutionary process in

    which diverse interactions between cancer cells and their microenviron-ment yield alterations that allow these cells to transcend their programmed

    behavior. Tumor cells thus populate and flourish in new tissue habitats and, ulti-mately, cause organ dysfunction and death. Understanding the many molecular play-ers and processes involved in metastasis could lead to effective, targeted approachesto prevent and treat cancer metastasis.

    The tumornodemetastasis (TNM) staging system used for most solid tumors

    considers the tumor size and degree of local invasion (T), the number, size, andlocation of lymph nodes (N), and the presence or absence of distant metastases(M).1 Metastases of tumors originating in different sites, such as the breast or lung,are treated differently because they are thought to behave like the tissue of origin,with characteristic patterns and kinetics of spread, and distinct profiles of chemo-sensitivity. Lymph nodes are of paramount importance in current staging practices,but it is hard to interpret the clinical significance of the distance of metastasesfrom the primary site (e.g., a supraclavicular N3 vs. a mediastinal N2 lymph nodein lung cancer). Indeed, the distance from the primary tumor to the organ of metas-tasis does not affect staging. For this reason, the real value of staging is to serveas an indicator of the primary cancers composite capability to metastasize, ratherthan to ensure that the tumor lies within the prescribed limits of a local inter-vention. Recent advances bring hope for characterizing the metastatic behavior ofcancer cells beyond the simplistic TNM stage. In the future, staging could includeidentif ication of subpopulations of tumor cells that have different metastatic behav-ior. A deeper understanding of the molecular and genetic concepts and processesinvolved in metastasis may pave the way toward new prognostic models and waysof planning treatment.

    Basic Concepts of Metastasis

    Origins of Cellular Heterogeneity

    Primary tumors consist of heterogeneous populations of cells with genetic altera-

    tions that allow them to surmount physical boundaries, disseminate, and colonizea distant organ. Metastasis is a succession of these individual processes2-4 (Fig. 1),and fully metastatic cells are rare clones in the primary tumor. In animal models,0.01% or fewer of the cancer cells entering the circulation develop into metastases.5,6The intrinsic genomic instability of cancer cells increases the frequency of altera-tions necessary to acquire metastatic capacity. The genomic instability and hetero-geneity of tumor cells are apparent in the chromosomal gains, losses, and rearrange-ments associated with cancer. DNA integrity can be compromised by aberrantcell-cycle progression, telomeric crisis (i.e., telomere dysfunction characterized bycytogenetic abnormalities and chromosomal instability), inactivation of DNA repair

    The New England Journal of Medicine

    Downloaded from www.nejm.org on November 8, 2010. For personal use only. No other uses without permission.

    Copyright 2008 Massachusetts Medical Society. All rights reserved.

  • 7/31/2019 Bases Moleculares de Metastasis

    2/10

    Molecular Origins of Cancer

    n engl j med 359;26 www.nejm.org december 25, 2008 2815

    genes (see the Glossary), and altered epigeneticcontrol mechanisms. For example, 50% of cancershave lost the tumor-suppressor protein p53, whichresponds to DNA damage by inducing apoptosisor arresting cell growth.7 Loss of p53 allows theaccumulation of cells with DNA damage.8

    Selective Pressures of the Tumor

    Microenvironment

    Each tissue has a physical structure and an estab-lished functional anatomy complete with compart-mental boundaries, a vascular supply, and a char-acteristic extracellular milieu of nutrients andstroma. Cancer cells that circumvent this organi-zation become exposed to environmental stresses,including a lack of oxygen or nutrients, a low pH,reactive oxygen species, and mediators of the in-flammatory response. Such pressures can select

    tumor cells with the capability of growth despitethese challenges and in the process can causethem to acquire an aggressive phenotype. For ex-ample, hypoxia stabilizes hypoxia-inducible factor(HIF), which cues a program of gene expressionthat leads to changes in anaerobic metabolism,angiogenesis, invasion, and survival.9,10 HIF booststhe expression of lysyl oxidase; lysyl oxidase regu-

    lates the activity of focal adhesion kinase in a waythat enhances cell-matrix adhesion and invasion.11High levels of lysyl oxidase correlate with shortermetastasis-free survival and a poor prognosis inhead and neck cancer, as well as in estrogen-recep-tornegative breast cancer.12 Another product ofHIF-induced gene activation, the chemokine (C-X-Cmotif) receptor CXCR4, together with its ligand,the chemokine stromal-cellderived factor 1 (SDF-1,also called CXC chemokine ligand 12 [CXCL12]),

    Tumor initiation: unlimited growth potential, survival, genomic instability

    Genes: KRAS, BRAF, EGFR, HER2, P13K(suppressors:APC, p53, PTEN, BRCA1, VHL1)

    Genes: RHoC, LOX, VEGF, CSF-1, ID1, TWIST1, MET, FGFR, MMP-9, NEDD9

    Metastasis initiation: invasion, marrow mobilization, angiogenesis, epithelial-to-mesenchymal transition

    Genes: EREG, COX-2, MMP-1, CCL5, ANGPTL4

    Metastasis progression: vascular remodeling, immune evasion, extravasation

    Genes: CXCR4, RANKL, CTGF, interleukin-11, endothelin-1Metastasis virulence: organ-specific functions

    Primary tumor

    Metastasis

    :

    Figure 1. Tumor Initiation and Metastasis.

    The initiation and progression of tumors depend on the acquisition of specific functions by cancer cells at both the primary and meta-

    static sites. Functions associated with tumor initiation are provided by oncogenic mutations and inactivation of tumor-suppressor genes.Functions associated with the initiation of metastasis include functions to which tumor cells resort for local invasion and for circumvent-

    ing hypoxia and other limitations facing a growing tumor. Most functions for the initiation of both the tumor and metastasis remain es-sential for cancer cells to continue their metastatic development. Functions for metastasis progression provide a local advantage in a

    primary tumor and a distinct and sometimes organ-specific function during metastasis. Cancer cells that are endowed with these threesets of functions still depend on functions associated with metastasis virulence; these functions confer a selective advantage solely dur-

    ing the adaptation and takeover of a specific organ microenvironment. Genes associated with each of these functions have been identi-fied in recent years.

    The New England Journal of Medicine

    Downloaded from www.nejm.org on November 8, 2010. For personal use only. No other uses without permission.

    Copyright 2008 Massachusetts Medical Society. All rights reserved.

  • 7/31/2019 Bases Moleculares de Metastasis

    3/10

    T h e n e w e n g l a n d j o u r n a l o f medicine

    n engl j med 359;26 www.nejm.org december 25, 20082816

    facilitates the survival of cancer cells at sites ofmetastasis in breast cancer and renal-cell cancer.13

    Cancer Stem Cells and Metastasis

    The question of the extent to which self-renewingcancer stem cells initiate and sustain cancers ofdifferent types is a subject of intense investiga-tion, and there are probably different answers ac-cording to different tumor types. Such cells areenvisioned as a subpopulation of cancer cells that by one mechanism or another have the

    capacity to act as tumor-propagating cells.14

    These cells might resist apoptosis and DNA dam-age caused by drugs; they might also require aniche or specific microenvironment in order togrow.15 Such attributes would support the estab-lishment of both primary and metastatic tumors.The SDF-1CXCR4 axis is thought to function insupport of cancer cells and stem cells or precur-sor cells.16 A premetastatic niche has been de-scribed in animal models in which bone marrow

    derived progenitor cells home to specific distantsites before the formation of a metastasis.17,18 Theability of stem cells to evade destruction and sur-vive in distant sites, including the bone marrow,may explain why micrometastases can remain dor-mant after removal of the primary tumor, only torecur years later.

    The Environment

    of the Primary Tumor

    Invasion and Epithelial-to-Mesenchymal

    Transition

    In many primary tumors with invasive properties,intercellular adhesion is reduced, often becauseof a loss of E-cadherin, a direct mediator of cellcell adhesive interactions. The cytoplasmic tail ofE-cadherin is tethered, via -catenin and -catenin,to the actin cytoskeleton; one of actins proper-

    ties is to maintain cell junctions. The importanceof maintaining intercellular adhesion was shownin a mouse model of pancreatic cancer in whichdisruption of the expression of E-cadherin led toearly invasion and metastasis.19 Various mecha-nisms can cause a loss of E-cadherin: mutationsresulting in an inactive protein, gene silencing bypromoter methylation, or down-regulation stim-ulated by growth factor receptors (e.g., epidermalgrowth factor receptor [EGFR], fibroblast growthfactor receptor [FGFR], insulin-like growth fac-tor I [IGF-I] receptor, and MET) or SRC familykinases.19,20 Expression of the E-cadherin gene(CDH1) is also inhibited by several transcription-al repressors.21,22 Loss of E-cadherin function isnecessary, though not sufficient for epithelial-to-mesenchymal transition, a process wherebyepithelial cells switch to a mesenchymal progen-itor-cell phenotype, enabling detachment and re-organization of epithelial-cell sheets during em-bryonic development, as well as tumor invasionand metastasis.23

    Motility and Extracellular-Matrix

    Remodeling

    The extracellular matrix serves as a scaffold alongwhich cells attach and move by means of con-tacts between cell-surface receptors called integ-rins and extracellular-matrix components such asfibronectin, collagen, and laminin. Integrins alsointeract in a cytoplasmic complex consisting offocal adhesion kinases and SRC family kinasesto mediate attachment to the actin cytoskeleton.

    Glossary of Selected Genes

    ANGPTL4: Angiopoietin-like 4

    APC: Adenomatous polyposis coli

    BRAF: (Also known as V-raf murine sarcoma viral onco-

    gene homologue B1)

    BRCA1: Breast-cancer gene 1

    COX-2: Cyclooxygenase-2CSF-1: Colony-stimulating factor 1

    CTGF: Connective-tissue growth factor

    CXCR4: CXC chemokine receptor 4

    EGFR: Epidermal growth factor receptor

    EREG:Epiregulin

    FGFR: Fibroblast growth factor receptor

    HER2:Human epidermal growth factor receptor type 2

    ID1: Inhibitor of differentiation-1

    MMP-1:Matrix metalloproteinase 1

    MMP-9:Matrix metalloproteinase 9

    NEDD9: Neural precursor cell expressed, developmen-

    tally down-regulated 9P13K: Phosphoinositide 3-kinase

    PTEN: Phosphatase and tensin homologue

    RANKL: Ligand for the receptor activator of nuclearfactor-B

    RHoC: Ras homologue gene family, member C

    STK11: Serinethreonine kinase 11 (also known as LKB1)

    TWIST1: Twist homologue 1

    VEGF: Vascular endothelial growth factor

    VHL1: von HippelLindau 1

    The New England Journal of Medicine

    Downloaded from www.nejm.org on November 8, 2010. For personal use only. No other uses without permission.

    Copyright 2008 Massachusetts Medical Society. All rights reserved.

  • 7/31/2019 Bases Moleculares de Metastasis

    4/10

    Molecular Origins of Cancer

    n engl j med 359;26 www.nejm.org december 25, 2008 2817

    Through calcium-dependent guanosine triphos-phatases (GTPases), extracellular-matrix signalscause cytoskeletal changes that form individualcytoplasmic extensions called filopodia, whichcoalesce into larger lamellipodia, structures thatare important in migratory movement. Expres-sion profiling of melanoma cell lines obtained by

    means of in vivo selection has shown that thecalcium-dependent GTPase RhoC is important inlung metastasis.24 Homozygous RhoC-deficientmice have normal formation of primary tumorsbut impaired cancer-cell mobility and almost nolung metastases.25 NEDD9, a scaffolding proteininvolved in cell adhesion, colocalizes in focal con-tacts with focal adhesion kinase and can promotecell motility and invasion.26 Various members ofthe matrix metalloproteinase (MMP) family (e.g.,MMP-2 and MMP-9) are also implicated in cancer-cell invasion.27-29 Independent screens for genes

    that mediate bone or lung metastasis in breastcancer have identified MMP-1 as being necessaryfor spread to the bone and lungs.30,31 The metas-tasis-suppressor microRNA miR335, which inhib-its metastasis to the lungs and bones in humanbreast-cancer xenografts, suppresses the expres-sion of two mediators of cancer-cell invasion, thetranscription factor SOX-4 and tenascin-C, a ma-trix glycoprotein.32 A low level of miR335 in breast-cancer cells is associated with relapse.

    Stromal Interactions

    Not only are cancer cells able to traverse thestructural boundaries of the primary tumor, butthey can also co-opt local and bone marrowderived stromal-cell responses to their advantage.At points of basement-membrane invasion inmouse tumors, tumor-associated macrophagesproliferate in response to tumor-derived colony-stimulating factor 1 and produce growth factors(e.g., fibroblast growth factor, EGFR ligands, andplatelet-derived growth factor [PDGF]) and pro-teases (e.g., MMPs and cathepsins).33,34 In addi-

    tion, tumor-associated macrophages activate aparticular type of carcinoma-associated mesen-chymal cell, the myofibroblast, which secretes thecytokine SDF-1; this cytokine enables the myofi-broblast to recruit endothelial progenitor cells.35Impaired metastases of breast-cancer cells to thelungs occur in mice with genetic defects in macro-phages.36 The stroma-derived cytokine, transform-ing growth factor (TGF-), induces the expres-sion of genes such as ANGPTL4 in breast-cancer

    cells; TGF- enhances metastatic activity and isassociated with increased metastases to the lungsin estrogen-receptornegative breast cancer.37 Inshort, several types of stromal cells and their se-creted factors provide selective prometastatic ad-vantages.

    Organ-Specific Metastasis

    Some types of cancers have a characteristic pro-clivity to metastasize to certain organs, but notto others (Fig. 2).38-42 Breast cancer spreads to

    Brain

    Breast

    Colorectal

    Renal-cell

    Stomach

    Bone

    Liver

    Lung

    Lung

    Melanoma

    Colorectal

    Renal-cell

    Melanoma

    Breast

    Sarcoma

    Colorectal

    Renal-cell

    Lung

    Breast

    Prostate

    Colorectal

    Pancreatic

    Lung

    Breast

    :

    Figure 2. Patterns of Metastatic Spread of Solid Tumors.

    Brain metastases may occur as a result of hematogenous spread late in thecourse of a widely metastatic tumor, or as a result of secondary metastasis

    from a primary or a metastatic tumor that can access the arterial circulationthrough the pulmonary venous circulation to seed the brain.38 Tumors with

    the highest incidence of brain metastases include lung carcinoma, breast

    carcinoma, melanoma, and to a lesser extent, renal-cell and colorectal car-cinomas. Leptomeningeal disease may develop through the spread of can-

    cer cells through perineural lymphatic vessels, and it is a sign of advanceddisease.38 Some tumors have a strong proclivity for dissemination to the

    lungs; for example, in one study, the rate of dissemination associated with

    sarcoma was 23%.

    39

    Other tumors that frequently spread to the lungs in-clude renal-cell, colorectal, melanoma, and breast carcinomas.40,41 Gastro-intestinal tumors easily access the liver circulation through the portal-vein

    system. The incidence of liver metastases is highest among patients withcolorectal or pancreatic cancer, followed by breast and lung cancers.40 Estro-

    gen-receptornegative breast-cancer tumors more often metastasize to vis-ceral organs, including the liver, whereas estrogen-receptorpositive breast

    cancer more often metastasizes to the bone.42 Bone metastasis occurs in

    patients with primary tumors associated with breast, lung, prostate, renal-cell, and colon cancer, in this order of frequency.40 Bone metastases may be

    primarily osteolytic or osteoblastic, depending on the tumor of origin.

    The New England Journal of Medicine

    Downloaded from www.nejm.org on November 8, 2010. For personal use only. No other uses without permission.

    Copyright 2008 Massachusetts Medical Society. All rights reserved.

  • 7/31/2019 Bases Moleculares de Metastasis

    5/10

    T h e n e w e n g l a n d j o u r n a l o f medicine

    n engl j med 359;26 www.nejm.org december 25, 20082818

    the bones, lungs, brain, and liver; distant metas-tases of prostate cancer occur most prominentlyin bone. Breast-cancer and prostate-cancer cellscan both spread to and colonize the bone, but theyform osteolytic or osteoblastic metastases, respec-tively. According to Pagets seed vs. soil hy-pothesis, perceived compatibilities between dis-

    seminated cancer cells (the seed) and certaindistant sites (the soil) have long influenced ourview of the metastatic process.43

    The formation of bone metastases alters bonehomeostasis the balance of action of osteo-clasts in degrading bone against the constant re-building of bone by osteoblasts. Breast-cancercells preferentially cause osteolytic lesions by in-ducing osteoclasts to secrete PTHrP (parathyroidhormonerelated peptide), tumor necrosis factor (TNF-), and cytokines such as interleukin-1,interleukin-6, interleukin-8, and interleukin-11.

    These factors cue osteoblasts to release RANKL(the ligand for the receptor activator of nuclearfactor-B [RANK]), which stimulates osteoclastdifferentiation (Fig. 3). Osteoclasts demineralizebone, thereby causing the release of growth fac-tors such as bone morphogenetic proteins, IGF-I,and TGF- from the exposed bone matrix; allthese growth factors support cancer-cell prolif-eration and induce further release of PTHrP. In abreast-cancer xenograft model, breast-cancer cellsthat preferentially colonized bone had up-regu-lated expression of genes encoding CXCR4, osteo-pontin, CTGF, MMP-1, and interleukin-11.30 Bycontrast, prostate-cancer cells secrete osteoblast-stimulating factors such as Wnt family ligands,bone morphogenetic proteins, PDGF, and endo-thelin-1; these factors stimulate formation of thehallmark osteoblastic metastases of prostate can-cer. Tumor-derived signals suppress the ability ofosteoblasts to secrete osteoprotegerin, a RANKLantagonist that blocks RANKLRANK interactionand resulting osteoclast activation. Thus, factorssecreted by cancer cells, or seeds, can influence

    the type of metastases formed.Cancer cells may regulate the expression ofother molecules to target colonization in otherorgans.44 Such molecules include the gene encod-ing ezrin (an intracellular protein needed forearly survival of metastatic osteosarcoma cells inthe lung), serinethreonine kinase 11 (STK11,also known as LKB1) (a metastasis-suppressorgene regulating NEDD9 in lung cancer45), andgenes in an 18-gene breast-to-lung metastatic

    gene-expression signature including the EGFRligand EREG, COX-2, MMP-1, ANGPTL4, and othermediators of infiltration and colonization by can-cer cells in the lung.46

    The soil, or distant metastatic site, is a large-ly nonpermissive environment, as evidenced bythe rarity of metastatic clones arising after inject-

    ing millions of cells into circulation in experi-ments in animals. In humans, also, thousandsof circulating tumor cells have been found in theabsence of metastases. Certain seedsoil inter-actions can support the cancer cells ability tosurvive in the metastatic microenvironment, in-cluding the RANKLRANK interaction. Anotherexample involves the SDF-1 chemokine in thebone marrow, which recruits breast-cancer andprostate-cancer cells and enhances their surviv-al.47 Whereas the mechanisms of metastasis tobone and lung have been extensively studied and

    are partially understood, there is a dearth of in-formation about the molecular basis for metasta-sis to other organs, such as the liver and brain.

    An Integrated Model

    of Metastasis

    In the past decade, our view of metastasis haschanged from snapshots detailing specific bio-logic processes to a moving picture of how vari-ous cancer cells acquire functions and co-optstromal signals for spread and encampment in adistant site. Random genetic and epigenetic alter-ations in cancer cells in combination with a plas-tic and responsive microenvironment support themetastatic evolution of tumors. Moreover, genesneeded at individual steps along the metastaticprocess have been identified.

    These genes have been classified into threecategories: initiation, progression, and virulence48(Fig. 1). Genes that are associated with metastaticprogression give the cancer cell particular advan-tages at multiple points during its sojourn to a dis-

    tant site. These advantages can influence the cellsmetastatic destination. Genes associated with theinitiation of metastasis and virulence operate in theearliest and latest stages of invasion and growthin the primary tumor and different metastatichabitats, respectively. The use of such a frameworkto organize specific genes and their functions al-lows a multidimensional picture (including localeand time) of metastasis and may aid in the devel-opment of rational antimetastatic strategies.

    The New England Journal of Medicine

    Downloaded from www.nejm.org on November 8, 2010. For personal use only. No other uses without permission.

    Copyright 2008 Massachusetts Medical Society. All rights reserved.

  • 7/31/2019 Bases Moleculares de Metastasis

    6/10

    Molecular Origins of Cancer

    n engl j med 359;26 www.nejm.org december 25, 2008 2819

    Models of Metastasis

    and Tumor Progres sion

    Early theories of metastasis pitted models of ge-netic predetermination against those of orderlyanatomic progression. The advent of molecular

    genetics has refashioned the model of tumor pro-gression in which somatic mutations were thoughtto accumulate sequentially, resulting in rare cellscapable of metastasis.49 Other models emphasizedynamic heterogeneity and clonal selection, prin-ciples that suggest that an unstable metastatic

    Primary tumor

    Lung metastasis

    Bone metastasis

    ANGPTL4

    EREG

    COX-2

    MMP-1

    ID1

    Niche cells

    Stromalcells

    Bonelysis

    BMPTGF-IGF-I

    SDF-1

    Osteoblasts

    Osteoclasts

    RANKLOsteo-

    protegerin

    PTHrP, TNF-,interleukin-6,

    interleukin-11

    Cancercells

    Cancercells

    Myeloidprogenitor

    cells

    :

    Figure 3. Genes, Functions, and Cellular Players in Organ-Specific Metastasis.

    Organ-specific metastasis of breast-cancer cells involves different molecular players during colonization of the lungs and the bones. Inthe lung, cancer cells producing EREG, COX-2, MMP-1, and ANGPTL4 are better equipped to exit the pulmonary vasculature, since these

    factors alter the integrity of lung microcapillary endothelia; this function is less important for infiltration into the bone marrow becauseof the naturally fenestrated structure of the bone marrow sinusoid vasculature. In the lung parenchyma, the activity of the antidifferenti-

    ation gene ID1 and interactions with still unknown niche factors promote tumor reinitiation. In the bone marrow, stromal-cellderivedfactor 1 (SDF-1), acting through its chemokine (C-X-C motif) receptor (CXCR4) on cancer cells, is thought to provide cell-survival func-

    tions. The secretion of parathyroid hormonerelated peptide (PTHrP), interleukin-6, tumor necrosis factor (TNF-), interleukin-11,and other factors by cancer cells stimulates osteoblasts to release the ligand for the receptor activator of nuclear factor-B (RANKL),

    which in turn stimulates osteoclast differentiation from myeloid progenitor cells. Other cancer cellderived factors suppress the produc-

    tion of the RANKL antagonist osteoprotegerin, augmenting the efficacy of RANKL. The lytic action of osteoclasts releases bone matrixassociated growth factors, including transforming growth factor (TGF-), insulin-like growth factor I (IGF-I), and bone morphogenetic

    proteins (BMPs). IGF-I is a survival factor, and TGF- incites cancer cells to further release PTHrP, interleukin-11, and other prometa-static factors, establishing a vicious cycle.

    The New England Journal of Medicine

    Downloaded from www.nejm.org on November 8, 2010. For personal use only. No other uses without permission.

    Copyright 2008 Massachusetts Medical Society. All rights reserved.

  • 7/31/2019 Bases Moleculares de Metastasis

    7/10

    T h e n e w e n g l a n d j o u r n a l o f medicine

    n engl j med 359;26 www.nejm.org december 25, 20082820

    variant can expand and prevail in the populationof cells.50,51 The presence of metastasis genes ingene-expression signatures of primary tumorswould seem to challenge the traditional tumor-progression model of somatic evolution in whichmetastatic cells would be too rare to inf luence apopulation-averaged gene-expression profile of the

    primary tumor. This finding, however, probablyreflects an abundance of partially competent can-cer cells that have accumulated a sufficient num-ber of malignant functions to promote expansionof the primary tumor, and which may be neces-sary but not sufficient for forming metastases.By contrast, genes associated with metastatic vir-ulence provide an aggressive edge in survival andproliferation solely during colonization of themetastatic site (Fig. 1). Many of these genes do notgive the primary tumor a selective advantage,and thus they would not be represented in gene

    signatures of the primary tumor.

    Metastasis-Progression Genes

    Genes that are necessary for certain functionssuch as vascular remodeling can participate inboth the primary tumor and the metastatic envi-ronment; these genes are metastasis-progressiongenes, and they could be enriched in primary tu-mors (Fig. 1). An 18-gene lung-metastatic signa-ture derived from selected in vivo breast-cancercells that efficiently spread to the lungs includesEREG, COX-2, and MMP-1. These genes cooperatein remodeling the vasculature in sites of mam-mary tumors and lung metastasis. In the breast,they allow neoangiogenesis and intravasation ofcancer cells; in the lung, they mediate extravasa-tion of circulating cancer cells from capillariesinto the parenchyma.52 Breast cancers with thelung-metastatic signature have a high risk of lungmetastases, but not of metastases to the bones orliver. A likely explanation is that extravasation isnot essential for passage through the fenestrated

    vasculature of the bone marrow and liver sinu-soids. Thus, metastasis-progression genes maycouple the tissue-specific features of the micro-environment in a particular organ to a matchingrole in the progression of a primary tumor. Ex-pression of the lung-metastatic signature geneANGPTL4 is a bystander event in mammary tu-mors, but when cancer cells expressingANGPTL4reach lung capillaries, its role in mediating extrav-

    asation by disrupting endothelial cellcell con-tacts becomes manifest.37

    Both the cells of primary tumors and metastat-ic cells require the ability to initiate self-renewaland bypass senescence. ID1 (inhibitor of differ-entiation-1) is the sole transcriptional regulator inthe lung-metastatic signature, and it can be found

    in clusters of cancer cells within breast tumors ofthe basal or triple-negative (i.e., estrogen-recep-tornegative, progesterone-receptornegative, andhuman epidermal growth factor receptor type 2[HER2]negative) subtype. Suppression of ID1expression inhibits the initiation of mammarytumors and metastases in the lungs.53 Thus, ID1may promote micrometastatic outgrowth fromdormancy at the metastatic site. Related to thisfunction, in mouse models of metastatic breastcancer, ID1 cooperates with activated RAS onco-genes to avert cell senescence.54

    Metastatic Dissemination

    Cancer cells can disseminate from a tumor veryearly in the life of a tumor. They have been detect-ed in the bone marrow of patients with breast can-cer with early-stage disease. Such cancer cells weregenetically distinct from the matched primary tu-mors, but bone marrowderived cancer cells inpatients with overt metastatic disease were lessgenetically disparate.55,56 This finding may reflectdifferences between the departure time from theprimary neoplasm and the duration of exposureto selective pressures. Dormant cancer cells iso-lated from the bone marrow of transgenic micewith preinvasive breast cancer and patients withductal carcinoma in situ became activated whentransplanted into the bone marrow and causedthe growth of lethal tumors.57 Many mechanismsfor metastatic dormancy have been postulated.58In patients with advanced metastatic disease,breast cancer cells that are competent in vascularentry can efficiently exit at a distant organ and

    perhaps reenter to repeat the process. Tumor in-filtration by means of its own circulating progenyof metastatic cancer cells has been raised as apossible mechanism for the later rapid expansionof tumor growth.59 According to this hypothesis,large primary tumors may also be the end productof aggressive reseeding. This would be a new per-spective on the long-standing observation thatmetastatic relapse correlates with tumor size.60

    The New England Journal of Medicine

    Downloaded from www.nejm.org on November 8, 2010. For personal use only. No other uses without permission.

    Copyright 2008 Massachusetts Medical Society. All rights reserved.

  • 7/31/2019 Bases Moleculares de Metastasis

    8/10

    Molecular Origins of Cancer

    n engl j med 359;26 www.nejm.org december 25, 2008 2821

    Clinical I mplications

    Molecular Signatures of Metastasis

    Gene-expression signatures of primary breastcancers that predict clinical outcome61-65 gener-ally do not overlap and range from a 70-genepoor-prognosis signature (detected with the use

    of the MammaPrint test) to a hand-picked set of21 recurrence genes (detected with the use ofthe Oncotype Dx test) that includes estrogen-receptor, HER2, and proliferation markers. Othersignatures consist of genes with expressions thatare associated with a process or pathway, such asthe response to serum mitogens,66 hypoxia,67 ac-tivation of specific oncogenes (e.g., RAS, MYC,and SRC),68,69 stimulation with a growth factor(e.g., TGF-),37 or treatment with specific chemo-therapeutic drugs to establish a drug-sensitivityprofile.70 To specifically identify genes that me-

    diate metastasis, animal models have been usedto select in vivo for highly metastatic and organ-specific derivatives of human cancer cell lines.3Such signatures can correlate with bone-specificand lung-specific spread.30,46 The lung-metastasissignature further correlates with clinical outcome,including the recurrence of disease in the lungs,in primary breast-cancer tumors.60 Functionalvalidation approaches (e.g., overexpression orknockdown experiments in culture or xenograftexperiments) have confirmed that these genes,particularly in combination, are critical for meta-static functions.52

    Targets of Therapy

    In principle, each metastasis-specific gene is apotential target for a treatment. Ongoing clinicaltrials target the metastatic initiation gene c-MET(e.g., the small-molecule inhibitor ARQ 197, inphase 12 trials) and two metastatic virulencegenes, RANK ligand (e.g., denosumab, in phase 3trials) and TGF- (e.g., monoclonal antibodyGC1008, in phase 1 trials). Combination therapy

    may be needed to overcome the intrinsic biologicredundancy in metastasis and to target sequen-tial steps in metastasis. In one series of preclini-cal experiments, only combination (not single-agent) therapy with the drugs celecoxib andcetuximab, meant to target two metastatic pro-gression genes, was effective in blocking lungmetastases by highly lung-metastatic breast-can-cer cells.52 If cancer cells are constantly on themove between sites of metastasis in the lung,

    treatment with these drugs could prevent furtherreseeding and growth of metastatic sites. Cancertreatment may need to combine multiple anti-metastatic drugs with cytotoxic chemotherapy.For example, bevacizumab, an antibody targetingvascular endothelial growth factor, is being stud-ied in combination with chemotherapy in the ad-

    juvant setting for colorectal, ovarian, and nonsmall-cell lung cancers. Therapies that target themechanisms that keep dormant micrometastasesalive are also needed.

    Clinical Translation

    Clinical trials involving antimetastatic agents facea number of obstacles. Any adjuvant trial to as-sess the recurrence of metastatic disease requiresmany patients because of the infrequency and lagtime to progression of metastatic disease in manytypes of cancer. Measuring response rates be-

    yond stable disease will further increase the num-ber of patients in a trial. Moreover, correlative stud-ies of tissue obtained from metastatic sites areessential to understand the results of such trials.

    These barriers are sobering, and they perhapsunderscore the conceptual shifts that will beneeded for the development of new cancer thera-pies. What changes can we envision? The profileof a tumor could include not only histopatho-logical or genetic determinants, or both, but alsoa molecular snapshot that would indicate a me-tastasis quotient. The metastasis quotient couldbe a measure of how adept the cells are with re-spect to metastatic functions, and it could serveas a prognostic framework. By focusing on meta-static progression and virulence functions, can-cer therapy might be dictated by the metastaticsite and not only by the specific tissue of origin.A current example of a treatment targeting ametastatic organ is the use of bisphosphonatesor denosumab (an anti-RANK antibody), or both,to treat bone metastases originating from thebreast, lung, and even multiple myeloma. Drug

    regimens for patients with cancer might includemultiple drugs targeting different metastatic sitesand seeding among sites. There is now hope forachieving the ultimate goal curing metastaticdisease.

    Dr. Massagu reports receiving consulting fees from Accele-ron Pharmaceuticals, lecture fees from Eli Lilly, Genzyme, andEisai. No other potential conf lict of interest relevant to this arti-cle was reported.

    We thank D. Nguyen, G. Riely, K. Politi, and D. Spriggs forinsightful discussions.

    The New England Journal of Medicine

    Downloaded from www.nejm.org on November 8, 2010. For personal use only. No other uses without permission.

    Copyright 2008 Massachusetts Medical Society. All rights reserved.

  • 7/31/2019 Bases Moleculares de Metastasis

    9/10

    T h e n e w e n g l a n d j o u r n a l o f medicine

    n engl j med 359;26 www.nejm.org december 25, 20082822

    References

    DeVita VT Jr, Lawrence TS, Rosenberg1.SA, eds. DeVita, Hellman, and RosenbergsCancer: principles & practice of oncology.8th ed. Philadelphia: Wolters Kluwer/Lip-pincott Williams & Wilkins, 2008.

    Weinberg RA. The biology of cancer.2.New York: Garland Science, 2007.

    Fidler IJ. The pathogenesis of cancer3.

    metastasis: the seed and soil hypothesisrevisited. Nat Rev Cancer 2003;3:453-8.

    Gupta GP, Massague J. Cancer metas-4.tasis: building a framework. Cell 2006;127:679-95.

    Chambers AF, Groom AC, MacDonald5.IC. Dissemination and growth of cancercells in metastatic sites. Nat Rev Cancer2002;2:563-72.

    Luzzi KJ, MacDonald IC, Schmidt EE,6.et al. Multistep nature of metastatic inef-ficiency: dormancy of solitary cells aftersuccessful extravasation and limited sur-vival of early micrometastases. Am JPathol 1998;153:865-73.

    Sykes SM, Mellert HS, Holbert MA, et7.

    al. Acetylation of the p53 DNA-bindingdomain regulates apoptosis induction.Mol Cell 2006;24:841-51.

    Halazonetis TD, Gorgoulis VG, Bar-8.tek J. An oncogene-induced DNA damagemodel for cancer development. Science2008;319:1352-5.

    Pouyssgur J, Dayan F, Mazure NM.9.Hypoxia signalling in cancer and approach-es to enforce tumour regression. Nature2006;441:437-43.

    Bristow RG, Hill RP. Hypoxia and me-10.tabolism: hypoxia, DNA repair and geneticinstability. Nat Rev Cancer 2008;8:180-92.

    Higgins DF, Kimura K, Bernhardt WM,11.et al. Hypoxia promotes fibrogenesis invivo via HIF-1 stimulat ion of epithelial-to-mesenchymal transition. J Clin Invest2007;117:3810-20.

    Erler JT, Bennewith KL, Nicolau M,12.et al. Lysyl oxidase is essential for hypoxia-induced metastasis. Nature 2006;440:1222-6.

    Staller P, Sulitkova J, Lisztwan J, Moch13.H, Oakeley EJ, Krek W. Chemokine recep-tor CXCR4 downregulated by von Hippel-Lindau tumour suppressor pVHL. Nature2003;425:307-11.

    Clarke MF, Fuller M. Stem cells and14.cancer: two faces of Eve. Cell 2006;124:1111-5.

    Croker AK, Allan AL. Cancer stem15.cells: implications for the progression andtreatment of metastat ic disease. J Cell MolMed 2008;12:374-90.

    Kucia M, Reca R, Miekus K, et al.16.Trafficking of normal stem cells and me-tastasis of cancer stem cells involve simi-lar mechanisms: pivota l role of the SDF-1-CXCR4 axis. Stem Cells 2005;23:879-94.

    Kaplan RN, Riba RD, Zacharoulis S,17.et al. VEGFR1-positive haematopoieticbone marrow progenitors initiate the pre-metastat ic niche. Nature 2005;438:820-7.

    Wels J, Kaplan RN, Rafii S, Lyden D.18.Migratory neighbors and distant invaders:tumor-associated niche cells. Genes Dev2008;22:559-74.

    Perl AK, Wilgenbus P, Dahl U, Semb H,19.Christofori G. A causal role for E-cadherinin the transition from adenoma to carci-noma. Nature 1998;392:190-3.

    Thiery JP. Epithelial-mesenchymal tran-20.sitions in tumour progression. Nat RevCancer 2002;2:442-54.

    Cano A, Prez-Moreno MA, Rodrigo I,21.et al. The transcription factor snail con-trols epithelial-mesenchymal transitionsby repressing E-cadherin expression. NatCell Biol 2000;2:76-83.

    Yang J, Mani SA, Donaher JL, et al.22.Twist, a master regulator of morphogen-esis, plays an essential role in tumor me-tastasis. Cell 2004;117:927-39.

    Yang MH, Wu MZ, Chiou SH, et al.23.Direct regulat ion of TWIST by HIF-1alphapromotes metastasis. Nat Cell Biol 2008;10:295-305.

    Clark EA, Golub TR, Lander ES, Hynes24.RO. Genomic analysis of metastasis revea lsan essential role for RhoC. Nature 2000;406:532-5. [Erratum, Nature 2001;411:974.]

    Hakem A, Sanchez-Sweatman O, You-25.Ten A, et al. RhoC is dispensable for em-bryogenesis and tumor initiation but es-sential for metastasis. Genes Dev 2005;19:1974-9.

    Kim M, Gans JD, Nogueira C, et al.26.Comparative oncogenomics identifiesNEDD9 as a melanoma metastasis gene.Cell 2006;125:1269-81.

    Egeblad M, Werb Z. New functions for27.the matrix metalloproteinases in cancerprogression. Nat Rev Cancer 2002;2:161-74.

    Lpez-Otin C, Matrisian LM. Emerg-28.ing roles of proteases in tumour suppres-sion. Nat Rev Cancer 2007;7:800-8.

    Martin MD, Matrisian LM. The other29.side of MMPs: protective roles in tumorprogression. Cancer Metastasis Rev 2007;26:717-24.

    Kang Y, Siegel PM, Shu W, et al. A mul-30.tigenic program mediating breast cancermetastasis to bone. Cancer Cell 2003;3:537-49.

    Minn AJ, Kang Y, Serganova I, et al.31.Distinct organ-specific metastatic poten-

    tial of individual breast cancer cells andprimary tumors. J Clin Invest 2005;115:44-55.

    Tavazoie SF, Alarcn C, Oskarsson T,32.et al. Endogenous human microRNAs thatsuppress breast cancer metastasis. Nature2008;451:147-52.

    Lewis CE, Pollard JW. Distinct role of33.macrophages in different tumor microen-vironments. Cancer Res 2006;66:605-12.

    Joyce JA. Therapeutic targeting of the34.tumor microenvironment. Cancer Cell2005;7:513-20.

    Orimo A, Gupta PB, Sgroi DC, et al.35.Stromal fibroblasts present in invasive hu-man breast carcinomas promote tumorgrowth and angiogenesis through elevatedSDF-1/CXCL12 secretion. Cell 2005;121:335-48.

    Condeelis J, Polla rd JW. Macrophages:36.obligate partners for tumor cell migra-

    tion, invasion, and metastasis. Cell 2006;124:263-6.

    Padua D, Zhang XH, Wang Q, et al.37.TGFbeta primes breast tumors for lungmetastasis seeding through angiopoietin-like 4. Cell 2008;133:66-77.

    Gavrilovic IT, Posner JB. Brain metas-38.tases: epidemiology and pathophysiology.J Neurooncol 2005;75:5-14.

    Billingsley KG, Burt ME, Jara E, et al.39.Pulmonary metastases from soft tissuesarcoma: analysis of patterns of diseasesand postmetastasis survival. Ann Surg1999;229:602-10.

    Hess KR, Varadhachary GR, Taylor SH,40.et al. Metastatic patterns in adenocarci-

    noma. Cancer 2006;106:1624-33.Leiter U, Meier F, Schittek B, Garbe C.41.

    The natural course of cutaneous melano-ma. J Surg Oncol 2004;86:172-8.

    Lacroix M. Significance, detection and42.markers of disseminated breast cancercells. Endocr Relat Cancer 2006;13:1033-67.

    Paget S. The distribution of secondary43.growths in cancer of the breast. Lancet1889;1:571-3.

    Khanna C, Wan X, Bose S, et al. The44.membrane-cytoskeleton linker ezrin isnecessary for osteosarcoma metastasis.Nat Med 2004;10:182-6.

    Ji H, Ramsey MR, Hayes DN, et al.45.LKB1 modulates lung cancer differentia-tion and metastasis. Nature 2007;448:807-10.

    Minn AJ, Gupta GP, Siegel PM, et al.46.Genes that mediate breast cancer metas-tasis to lung. Nature 2005;436:518-24.

    Wang J, Loberg R, Taichman RS. The47.pivotal role of CXCL12 (SDF-1)/CXCR4axis in bone metastasis. Cancer Metasta-sis Rev 2006;25:573-87.

    Nguyen DX, Massague J. Genetic de-48.terminants of cancer metastasis. Nat RevGenet 2007;8:341-52.

    Vogelstein B, Fearon ER, Hamilton SR,49.et al. Genetic alterations during colorec-

    tal-tumor development. N Engl J Med 1988;319:525-32.Ling V, Chambers AF, Harris JF, Hill50.

    RP. Quantitative genetic analysis of tumorprogression. Cancer Metastasis Rev 1985;4:173-92.

    Waghorne C, Thomas M, Lagarde A,51.Kerbel RS, Breitman ML. Genetic evidencefor progressive selection and overgrowthof primary tumors by metastatic cell sub-populat ions. Cancer Res 1988;48:6109-14.

    Gupta GP, Nguyen DX, Chiang AC, et52.al. Mediators of vascular remodelling co-

    The New England Journal of Medicine

    Downloaded from www.nejm.org on November 8, 2010. For personal use only. No other uses without permission.

    Copyright 2008 Massachusetts Medical Society. All rights reserved.

  • 7/31/2019 Bases Moleculares de Metastasis

    10/10

    Molecular Origins of Cancer

    n engl j med 359;26 www.nejm.org december 25, 2008 2823

    opted for sequential steps in lung metas-tasis. Nature 2007;446:765-70.

    Gupta GP, Perk J, Achary ya S, et al. ID53.genes mediate tumor reinitiation duringbreast cancer lung metastasis. Proc NatlAcad Sci U S A 2007;104:19506-11.

    Swarbrick A, Roy E, Allen T, Bishop54.JM. Id1 cooperates with oncogenic Ras toinduce metastatic mammary carcinoma by

    subversion of the cellular senescence re-sponse. Proc Natl Acad Sci U S A 2008;105:5402-7.

    Klein CA, Blankenstein TJ, Schmidt-55.Kittler O, et al. Genetic heterogeneity ofsingle disseminated tumour cells in mini-mal residual cancer. Lancet 2002;360:683-9.

    Schmidt-Kittler O, Ragg T, Daskalakis56.A, et al. From latent disseminated cells toovert metastasis: genetic analysis of sys-temic breast cancer progression. Proc NatlAcad Sci U S A 2003;100:7737-42.

    Hsemann Y, Geigl JB, Schubert F, et57.al. Systemic spread is an early step inbreast cancer. Cancer Cell 2008;13:58-

    68.Aguirre-Ghiso JA. The problem of can-58.

    cer dormancy: understanding the basicmechanisms and identifying therapeuticopportunit ies. Cell Cycle 2006;5:1740-3.

    Norton L, Massagu J. Is cancer a dis-59.ease of self-seeding? Nat Med 2006;12:875-8.

    Minn AJ, Gupta GP, Padua D, et al.60.Lung metastasis genes couple breast tu-mor size and metastatic spread. Proc Natl

    Acad Sci U S A 2007;104:6740-5.Bertucci F, Cervera N, Birnbaum D.61.A gene signature in breast cancer. N EnglJ Med 2007;356:1887-8.

    Buyse M, Loi S, vant Veer L, et al.62.Validation and clinical ut ility of a 70-geneprognostic signature for women with node-negative breast cancer. J Natl Cancer Inst2006;98:1183-92.

    Fan C, Oh DS, Wessels L, et al. Con-63.cordance among gene-expressionbasedpredictors for breast cancer. N Engl J Med2006;355:560-9.

    van de Vijver MJ, He YD, vant Veer LJ,64.et al. A gene-expression signature as a pre-dictor of survival in breast cancer. N Engl

    J Med 2002;347:1999-2009.vant Veer LJ, Dai H, van de Vijver MJ,65.

    et al. Gene expression profiling predictsclinical outcome of breast cancer. Nature2002;415:530-6.

    Chang HY, Sneddon JB, Alizadeh AA,66.et al. Gene expression signature of fibro-blast serum response predicts humancancer progression: similarities betweentumors and wounds. PLoS Biol 2004;2(2):e7.

    Chi JT, Wang Z, Nuyten DS, et al. Gene67.expression programs in response to hy-poxia: cell type specificity and prognosticsignif icance in human cancers. PLoS Med2006;3(3):e47.

    Bild AH, Potti A, Nevins JR. Linking68.oncogenic pathways with therapeutic op-portunities. Nat Rev Cancer 2006;6:735-41.

    Bild AH, Yao G, Chang JT, et al. Onco-69.genic pathway signatures in human can-cers as a guide to targeted therapies. Na-ture 2006;439:353-7.

    Nevins JR, Potti A. Mining gene ex-70.pression profiles: expression signatures ascancer phenotypes. Nat Rev Genet 2007;8:

    601-9.Copyright 2008 Massachusetts Medical Society.

    personalarchivesinthejournalonline

    Individual subscribers can store articles and searches using a featureon the Journals Web site (www.nejm.org) called Personal Archive.Each article and search result links to this feature. Users can create

    personal folders and move articles into them for convenient retrieval later.

    The New England Journal of Medicine

    Downloaded from www.nejm.org on November 8, 2010. For personal use only. No other uses without permission.