The Cancer Genome Atlas: Moving Forward

Ryan Chow

The Cancer Genome Atlas (TCGA) began in 2006, envisioned as the next frontier in the “War on Cancer”.1 Newly armed with the foundational framework of the Human Genome Project,2 the scientific community saw great promise in deeply analyzing the underlying genomic abnormalities of cancer. Though the project had its skeptics, its reductionist logic was immensely compelling: what if cancer, just like cystic fibrosis and other genetic disorders, could be easily reduced to a few key mutations? Thus, to the tune of $100 million from the National Institutes of Health (NIH), the TCGA took off in the form of a three-year pilot project.

Nearly a decade later, the project has officially come to an end. In total, around 10,000 tumors have been analyzed across 31 cancer subtypes.3 With an astounding combination of public available gene expression analyses, somatic mutation profiles, copy number variations, and several other genomic characterizations, the TCGA has produced petabytes of data that nearly anyone with an internet connection can access. The question is: now what?

Early on in the project, scientists realized that finding mutations was hardly the problem. Figuring out which mutations actually mattered, however, quickly proved to be more difficult.4 While the development of powerful statistical algorithms for ferreting out the most relevant alterations would grow to play a major role in the TCGA, as the data continued to accumulate, so too did the complexity of the problem grow.5 To tackle these complexities, in more recent years some scientists have shifted their focus away from subtype-specific analyses and into so-called “Pan-Cancer” analyses.6 Instead of drilling into the details of, say, lung adenocarcinoma or clear cell kidney carcinoma, a few research teams have begun to look at the data from a macroscopic level in the hopes of deriving true biological insights that are universal to all cancers; since cancers generally rely on the same cellular processes to grow and avoid destruction, it follows that there could be much to learn from looking at the similarities between subtypes, rather than analyzing them in isolation. While relatively few in number, these studies have been quite successful thus far. For instance, a combined analysis of 12 cancer subtypes revealed one surprisingly distinct subtype that spanned across traditional modes of classification.7 As another example, numerous research groups have demonstrated that the same mutational networks are at work in most cancers.8-12 Recognizing the importance of such analyses, high-profile journals such as Nature Genetics have specifically called for “re-analysis” papers that comb through published databases in order to derive novel biological insights.13 As more research teams learn to utilize publicly available information on the TCGA, the hope is that more labs will bring their personal fields of expertise into the realm of cancer genomics.

But paradoxically, although an overwhelming number of mutations have been found as a result of the TCGA, the core group of cancer-driving mutations has not changed significantly since the project’s inception. Contrary to what many had expected, as the TCGA project progressed, the vast majority of the “newly” identified oncogenic mutations had already been known for years prior. In retrospect, perhaps this was to be expected: scientists certainly did not require a multimillion-dollar sequencing project to tell them that mutations in DNA repair proteins would cause cancer. At its core, the true bottleneck had always been, and still is, the development of clinically viable drugs that can target key malignant pathways.14 Making matters worse, even for the few patients with mutations that can actually be targeted by next-generation drugs, it is already well-established that tumors are highly prone to developing drug resistance.15 Though the molecular portraits of cancer are coming into higher and higher resolution, the path to clinical translation still remains incredibly blurry.

That is not to say that cancer genomics has reached a complete dead end, however. One particularly promising avenue for future research is that of intratumoral heterogeneity – in essence, the study of diversity not between different tumors, but within a single tumor. The biological relevance of this approach is readily apparent, for within any individual tumor, there is actually tremendous phenotypic and genotypic diversity. Indeed, recent research has demonstrated that the various cell types and clonal subpopulations that coexist in a single tumor can all contribute to the survival of the tumor.16,17 Perhaps most importantly, these different populations can have wildly varying responses to clinical intervention.18

Unfortunately, the samples that were used for the TCGA did not lend themselves to intratumoral analysis. In most cases, whole tumor biopsies were quite literally mashed into a pulp without consideration of the heterogeneity inherent to each of these tumors. Moving forward, scientists must pay closer attention to the details, for the effectiveness of personalized cancer therapy depends on it. Taken together, the TCGA has certainly revolutionized the world of cancer research. Though the translation of these genomic findings to the clinic has admittedly been slow, the foundations have been set for the next battle in the War on Cancer. An emphasis on the similarities between different cancers, together with careful examination of the differences within a single tumor, will likely yield valuable biological insights.

About the Author

Ryan Chow is currently a junior at Harvard College, concentrating in Human Developmental and Regenerative Biology. By pursuing a career as a physician-scientist, he hopes to deepen our basic understanding of human disease, with the ultimate goal of clinical translation.


  1. Ledford, H. (2010). Big science: The cancer genome challenge. Nature News 464, 972–974.
  2. International Human Genome Sequencing Consortium. (2004). Finishing the euchromatic sequence of the human genome. Nature 431, 931-945.
  3. Ledford, H. (2015). End of cancer-genome project prompts rethink. Nature 517, 128–129.
  4. Cilbulskis, K., et al. (2013). Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nature Biotechnology 31, 213-219.
  5. Check, H.E. (2008). Cancer complexity slows quest for cure. Nature 455, 148.
  6. The Cancer Genome Atlas Research Network, et al. (2013). The Cancer Genome Atlas Pan-Cancer analysis project. Nature Genetics 45, 1113-1120.
  7. Hoadley, K.A., et al. (2014). Multiplatform Analysis of 12 Cancer Types Reveals Molecular Classification within and across Tissues of Origin. Cell 158, 929–944.
  8. Fredriksson, N.J., Ny, L., Nilsson, J.A., and Larsson, E. (2014). Systematic analysis of noncoding somatic mutations and gene expression alterations across 14 tumor types. Nat Genet 46, 1258–1263.
  9. Kandoth, C., et al. (2013). Mutational landscape and significance across 12 major cancer types. Nature 502, 333–339.
  10. Lawrence, M.S., et al. (2014). Discovery and saturation analysis of cancer genes across 21 tumour types. Nature 505, 495–501.
  11. Leiserson, M.D.M., et al. (2015). Pan-cancer network analysis identifies combinations of rare somatic mutations across pathways and protein complexes. Nat Genet 47, 106–114.
  12. Martínez, E., et al. (2014). Comparison of gene expression patterns across 12 tumor types identifies a cancer supercluster characterized by TP53 mutations and cell cycle defects. Oncogene, 4 August 2014.
  13. (2014). Second call for pan-cancer analysis. Nat Genet 46, 1251–1251.
  14. Verdine, G.L., Walensky, L.D. (2007). The Challenge of Drugging Undruggable Targets in Cancer: Lessons Learned from Targeting BCL-2 Family Members. Clin Cancer Res 13, 7264-7270.
  15. Crystal, A.S., et al. (2014). Patient-derived models of acquired resistance can identify effective drug combinations for cancer. Science 346, 1480-1486.
  16. De Bruin, E.C., et al. (2014). Spatial and temporal diversity in genomic instability processes defines lung cancer evolution. Science 346, 251–256.
  17. Zhang, J., et al. (2014). Intratumor heterogeneity in localized lung adenocarcinomas delineated by multiregion sequencing. Science 346, 256–259.
  18. (2013). Targeting molecular tumor types. Nat Genet 45, 1103–1103.