Metastasis of cancer: when and how?

Despite encouraging advances in both local and systemic treatment over the past two decades, cancer remains the leading cause of death worldwide. Metastasis, a process of cancer cells spreading from the primary tumor to surrounding tissues and to distant organs is the primary cause of cancer mortality. It is estimated that metastasis is responsible for ∼90% of cancer deaths [1.Chaffer C.L. Weinberg RA. A perspective on cancer cell metastasis.Science. 2011; 331: 1559-1564Crossref PubMed Scopus (3331) Google Scholar]. This has not changed much in the past half century [2.Sporn MB. The war on cancer.Lancet. 1996; 347: 1377-1381Abstract PubMed Scopus (540) Google Scholar]. Understanding metastases is critical to improving cancer patient outcomes. Although metastasis equates to late stage cancer clinically, when the process begins and how metastasis occurs is largely unknown. Cancer is a genetic disease resulting from accumulation of genomic aberrations that may drive cancer progression including metastasis. Pioneering studies have attempted to dissect this process at the genetic level by comparing the genetic similarities and differences between primary and metastatic tumors to understand how cancer cells acquire the invasive ability and migrate into the secondary metastatic sites [3.Yoshida B.A. Sokoloff M.M. Welch D.R. Rinker-Schaeffer CW. Metastasis-suppressor genes: a review and perspective on an emerging field.J Natl Cancer Inst. 2000; 92: 1717-1730Crossref PubMed Scopus (235) Google Scholar, 4.Kang Y. Siegel P.M. Shu W. et al.A multigenic program mediating breast cancer metastasis to bone.Cancer Cell. 2003; 3: 537-549Abstract Full Text Full Text PDF PubMed Scopus (2036) Google Scholar, 5.Minn A.J. Gupta G.P. Siegel P.M. et al.Genes that mediate breast cancer metastasis to lung.Nature. 2005; 436: 518-524Crossref PubMed Scopus (2269) Google Scholar, 6.Bos P.D. Zhang X.H.-F. Nadal C. et al.Genes that mediate breast cancer metastasis to the brain.Nature. 2009; 459: 1005-1009Crossref PubMed Scopus (1291) Google Scholar]. However, these studies have been confined to a limited number of genes investigated while cancer is a genetically complex disease. The emergence of next-generation sequencing technologies has enabled researchers to detect somatic mutations and large-scale genomic rearrangements at the whole-genome or whole-exome level. Comparative studies of primary tumors and paired metastases across a number of tumor types showed a varying degree of genetic divergence. Both linear (metastasis founding clone emerges late) and parallel (metastasis develops early in tumorigenesis and continues to evolve) metastatic progression patterns have been observed in many tumor types, and even in the same patient with multiple metastases [7.Turajlic S. Swanton C. Metastasis as an evolutionary process.Science. 2016; 352: 169-175Crossref PubMed Scopus (364) Google Scholar]. In this issue of Annals of Oncology, Wei et al. [8.Wei Q. Ye Z. Zhong X. et al.Multiregion whole-exome sequencing of matched primary and metastatic tumors revealed genomic heterogeneity and suggested polyclonal seeding in colorectal cancer metastasis.Ann Oncol. 2017; 28: 2135-2141Abstract Full Text Full Text PDF PubMed Scopus (77) Google Scholar] compared the subclonal architectures of paired primary colorectal cancer (CRC) and metastasis by multiregion WES (>150×) in 28 regions of primary and matched metastases from four CRC patients. Primary tumors were resected before any systemic treatment, eliminating the cofounding effect that may result from systemic therapeutic agents. The majority of mutations were shared between primary tumors and paired metastases suggesting that metastases were relatively late molecular events in these patients. And as expected, varying degree of intra-individual heterogeneity was demonstrated although to a much lesser extent compared with inter-individual heterogeneity. Furthermore, they inferred the subclonal architectures of all tumor regions by leveraging cancer cell fraction and demonstrated that the subclonal architectures of metastases were substantially different between those of primary tumors, indicating ongoing mutation in both primary tumors and metastases. Interestingly, metastases exhibited less intra-tumor heterogeneity than did primary tumors. It is still under debate whether metastasis follows the monoclonal seeding model, where metastasis results from one particular cancer cell or polyclonal seeding model, where multiple genetically different cancer cell clones give rise to metastases. In monoclonal seeding model, metastasis only carries mutations in the lineage that single cell represents, while in the polyclonal model, mutations from multiple primary tumor lineages would be found in the metastases. The monoclonal seeding model has been widely accepted [9.Poste G. Fidler IJ. The pathogenesis of cancer metastasis.Nature. 1980; 283: 139-146Crossref PubMed Scopus (1357) Google Scholar, 10.Fidler IJ. The pathogenesis of cancer metastasis: the "seed and soil" hypothesis revisited.Nat Rev Cancer. 2003; 3: 453-458Crossref PubMed Scopus (3532) Google Scholar, 11.Talmadge J.E. Fidler IJ. AACR centennial series: the biology of cancer metastasis: historical perspective.Cancer Res. 2010; 70: 5649-5669Crossref PubMed Scopus (795) Google Scholar] although the evidence for the polyclonal seeding model has started to emerge recently in many tumor types. In a study on prostate cancer, for example, multiple metastases were found to share the same mutation clusters that were present subclonally in each region in 5 of the 10 patients, suggestive of possible polyclonal seeding mechanism in these patients [12.Gundem G. Van Loo P. Kremeyer B. et al.The evolutionary history of lethal metastatic prostate cancer.Nature. 2015; 520: 353-357Crossref PubMed Scopus (929) Google Scholar]. Using breast cancer mouse models, Aceto et al. [13.Aceto N. Bardia A. Miyamoto D.T. et al.Circulating tumor cell clusters are oligoclonal precursors of breast cancer metastasis.Cell. 2014; 158: 1110-1122Abstract Full Text Full Text PDF PubMed Scopus (1516) Google Scholar] showed that circulating tumor cell (CTC) clusters have 23- to 50-fold increased metastatic potential than single CTCs. The distinction between monoclonal versus polyclonal seeding model may have important clinical implications [14.Beltran H. Demichelis F. Prostate cancer: Intrapatient heterogeneity in prostate cancer.Nat Rev Urol. 2015; 12: 430-431Crossref PubMed Scopus (16) Google Scholar]. If a single clone leads to the metastasis, then identification of the most lethal clone in the primary tumor could guide the treatment of metastasis. However, if metastasis results from polyclonal seeding, it will be much more challenging to identify all clones with metastatic potential. In the current study, the authors correlated subclones in primary tumors versus metastases and deduced the evolutional route from primary to metastasis. Interestingly, all patients in this study showed evidence of polyclonal seeding. The lymphatic spread is a common route of metastasis. Many epithelial cancers first progress to regional lymph nodes and circulate through the lymphatic system [15.Karaman S. Detmar M. Mechanisms of lymphatic metastasis.J Clin Invest. 2014; 124: 922-928Crossref PubMed Scopus (334) Google Scholar] before hematogenous metastasis occurs. In the patient with metastatic involvement in three lymph nodes and a lung metastasis, the lung metastasis demonstrated a unique subclonal architecture compared with that of lymph node metastases. With a caveat of only one patient analyzed, this is indicative of parallel polyclonal seeding in which subclones of primary tumor directly spread to the metastatic site without the mediation of lymph nodes. With the caveat of small sample size, the current study attempts to address a clinically and biologically important question of molecular mechanisms underlying metastasis of CRC. The sophisticated analyses and interesting findings affirmed the value of high-depth multiregion sequencing in delineating subclonal architectures and evolutionary patterns from primary to metastatic tumors. While congratulating the authors for successfully conducting this study, several points are worth discussing for future work comparing primary and metastatic tumors. First, even with the limited number of samples available, it is desirable to compare matched primary and metastatic tumors because of the high level of inter-tumor heterogeneity shown in the recent efforts of large-scale genomic profiling of tumors such as The Cancer Genome Atlas and The International Cancer Genome Consortium. Second, intra-tumor heterogeneity is a possible cofounding factor. It has been reported in many cancers that different regions of the same tumor can show distinct molecular characteristics [16.Gerlinger M. Rowan A.J. Horswell S. et al.Intratumor heterogeneity and branched evolution revealed by multiregion sequencing.N Engl J Med. 2012; 366: 883-892Crossref PubMed Scopus (5730) Google Scholar, 17.Gerlinger M. Horswell S. Larkin J. et al.Genomic architecture and evolution of clear cell renal cell carcinomas defined by multiregion sequencing.Nat Genet. 2014; 46: 225-233Crossref PubMed Scopus (893) Google Scholar, 18.Zhang J. Fujimoto J. Zhang J. et al.Intratumor heterogeneity in localized lung adenocarcinomas delineated by multiregion sequencing.Science. 2014; 346: 256-259Crossref PubMed Scopus (696) Google Scholar, 19.Kim T.-M. Jung S.-H. An C.H. et al.Subclonal genomic architectures of primary and metastatic colorectal cancer based on intratumoral genetic heterogeneity.Clin. Cancer Res. 2015; 21: 4461-4472Crossref PubMed Scopus (132) Google Scholar]. Therefore, without multiregion profiling, it is difficult to ascertain whether genomic alterations identified exclusively in primary or metastasis are truly 'primary- or metastasis-specific' or if they are actually spatially confined genomic events that are missed because of sampling bias. As an example, a study [20.Ding L. Ellis M.J. Li S. et al.Genome remodelling in a basal-like breast cancer metastasis and xenograft.Nature. 2010; 464: 999-1005Crossref PubMed Scopus (988) Google Scholar] in basal-like breast cancer using WES approach identified two de novo mutations and a large deletion that were exclusively in the metastasis, but not in the primary tumor. However, all three genomic aberrations were identified in a xenograft model derived from the primary tumor suggesting that these 'metastasis-specific aberrations' were due to the intra-tumor heterogeneity of primary tumor. Third, because all subsequent statistical and mathematical analyses are highly dependent on large amount of sequencing data, a major challenge of such studies is the quality of sequencing reads, particularly from small formalin-fixed paraffin embedded (FFPE) samples [21.Chen L. Liu P. Evans T.C. Ettwiller LM. DNA damage is a pervasive cause of sequencing errors, directly confounding variant identification.Science. 2017; 355: 752-756Crossref PubMed Scopus (133) Google Scholar]. FFPE is known to introduce artifacts, particularly C:G>T:A transitions. For certain tumors, melanoma, for example, where C:G>T:A transitions are the dominant form of somatic mutations, complete elimination of these artifacts is almost impossible. Therefore, extreme cautions should be taken when interpreting the data from FFPE samples. Fourth, our current mathematical and statistical models are primitive and may not be able to be applied to all datasets. For example, in the current study, a comprehensive subclonal architecture could not be unambiguously constructed for some regions. Tremendous efforts are needed to improve our analytic tools and in the meanwhile, we should be conscious of limitations in interpreting data. Although we have made substantial progress, our understanding of the precise mechanisms underlying metastasis is still rudimentary. The metastasis is a multi-step process in which cancer cells leave their original tumor legion, circulate through the bloodstream, and finally settle and start growing in other sites or tissues. During the metastatic cascade, tumor cells may accumulate molecular alterations that change the phenotypic features of the cells by interacting with surrounding microenvironment. The fate of cancer cells is a product of complex interaction between cancer cells of certain molecular and phonotypical features and their surrounding microenvironment. In 1889, Stephen Paget proposed the 'seed and soil' hypothesis [22.Paget S. The distribution of secondary growths in cancer of the breast.Lancet. 1889; 133: 571-573Abstract Scopus (2264) Google Scholar] that secondary tumors do not arise randomly. Cancer cells (seed) can be theoretically disseminated to every tissue in our body, but seeding alone does not guarantee that they will successfully develop a secondary tumor; instead, some property of the recipient tissue (soil) must support the new growth. Indeed, several studies reported that decreased presence of lymphatic vessel, reduced immune cytotoxicity and recruitment of immunosuppresive cells are the hallmarks representing metastasis-permissive microenvironment [23.Mlecnik B. Bindea G. Kirilovsky A. et al.The tumor microenvironment and immunoscore are critical determinants of dissemination to distant metastasis.Sci Transl Med. 2016; 8: 327ra26.Crossref PubMed Scopus (317) Google Scholar, 24.Kitamura T. Qian B.-Z. Pollard JW. Immune cell promotion of metastasis.Nat Rev Immunol. 2015; 15: 73-86Crossref PubMed Scopus (749) Google Scholar]. Majority of previous studies have mainly focused on the genetic changes that may enable cancer cells with metastatic potential, but other molecular alterations, for example, somatic epigenetic changes may also impact neoplastic transformation and fitness. Future studies are required to depict the overall molecular (genetic, epigenetic, gene expression) landscape of cancer cells as well as the tumor microenvironment, particularly immune contexture in order to systematically understand when and how cancer metastasis occurs. We thank Dr P. Andrew Futreal for his critical review and discussion. MD Anderson Physician Scientist Program; Khalifa Scholar Award; the Cancer Prevention and Research Institute of Texas Multi-Investigator Research Award grant (grant number RP160668); T.J. Martell Foundation.