Commentary: The nature of cancer research

Cancer research reflects an implicit conflict. On the one hand, there is an overwhelming desire to control the disease. We all wish that. On the other hand, we would like to understand why cancer follows so many clearly defined yet puzzling patterns. Why is there such regularity in the rates of progression? Why do different tissues vary so much? There should, of course, be no conflict between control and understanding. But the history of cancer research seems to say that those different goals remain oddly estranged. Peto’s 1977 article locates the seeds of this conflict most clearly. He describes what is still the most powerful theoretical perspective for analyzing the causes of cancer. He presents many key unsolved puzzles within that context. He also says why most cancer researchers are not interested in these fundamental issues. The subsequent decades of research grew around this rift, blindly, in the way that research disciplines often grow. Let us revisit Peto, almost 40 years ago. We can learn much about the current nature of cancer research.

💡 Research Summary

The commentary opens by framing cancer research as a discipline caught between two seemingly opposing imperatives: the urgent desire to control the disease—through prevention, early detection, and therapy—and the intellectual drive to understand why cancer follows remarkably regular yet puzzling patterns. The author argues that, while there should be no inherent conflict between these goals, the history of the field suggests a persistent estrangement. Central to this argument is a re‑examination of Richard Peto’s seminal 1977 paper, which articulated the most powerful theoretical perspective on cancer causation that still dominates epidemiological thinking today.

Peto’s work highlighted three striking regularities that have remained largely unexplained. First, the incidence of most cancers rises with age in an almost perfect log‑linear fashion, a phenomenon often described as the “age‑incidence law.” Second, the absolute risk varies dramatically across tissue types; for example, the lifetime risk of colon cancer is orders of magnitude higher than that of pancreatic cancer, even after adjusting for age and known risk factors. Third, the same age‑incidence curve appears across diverse populations, suggesting that underlying biological processes, rather than specific environmental exposures, drive the pattern. Peto interpreted these observations through a multi‑stage model, proposing that cancer results from a series of 5–7 probabilistic events (mutations, epigenetic changes, selective expansions) that accumulate over a person’s lifespan.

The commentary then traces how, over the ensuing four decades, the research enterprise has gravitated toward the “control” side of the equation. The rise of molecular biology, high‑throughput sequencing, and targeted drug development created a research culture that rewards the identification of “driver” mutations (e.g., KRAS, TP53, BRAF) and the rapid translation of those findings into clinical trials. Funding agencies, pharmaceutical companies, and high‑impact journals have all reinforced this trajectory by prioritizing projects with clear therapeutic endpoints and short‑term translational potential. Consequently, the field has amassed an impressive catalog of genetic alterations and a growing arsenal of targeted agents, yet the deeper question of why those alterations appear in particular tissues at particular ages remains under‑explored.

The author points out that this focus on single‑gene drivers obscures the multi‑stage nature of carcinogenesis that Peto emphasized. In practice, many tumors display extensive intra‑tumoral heterogeneity, sub‑clonal evolution, and therapy‑induced resistance—all phenomena that are more naturally explained by a series of sequential, partially reversible steps rather than a single “hit.” The commentary argues that the failure to integrate the multi‑stage framework into modern experimental design limits our ability to predict which pre‑malignant lesions will progress, why certain cancers recur after apparently successful treatment, and how to design truly preventive interventions.

A second major theme is tissue specificity. The commentary reviews evidence that differences in stem‑cell turnover rates, microenvironmental cues, hormonal milieus, and immune surveillance contribute to the observed variation in cancer incidence across organs. Yet most contemporary studies treat each cancer type in isolation, applying the same driver‑centric paradigm without accounting for these organ‑level variables. This siloed approach hampers the development of cross‑cancer insights that could reveal universal principles of tumorigenesis.

Beyond scientific content, the author critiques the structural incentives that perpetuate the control‑centric bias. The pressure to publish high‑impact, clinically relevant papers, the availability of large‑scale “omics” datasets, and the allure of big‑data analytics have shifted attention toward correlation‑driven research. While such studies can generate hypotheses, they often lack the mechanistic depth required to answer Peto’s fundamental questions about the regularity of cancer rates. Moreover, the commentary warns that this trend may inadvertently reinforce a “black‑box” view of cancer, where statistical associations are treated as ends in themselves rather than stepping stones toward causal understanding.

To bridge the divide, the author proposes a set of strategic recommendations. First, reinvigorate epidemiological modeling by integrating modern molecular data (e.g., mutational signatures, epigenetic clocks) into the classic multi‑stage framework, thereby creating quantitative predictions that can be tested in longitudinal cohorts. Second, develop experimental platforms—such as organ‑oid systems and lineage‑tracing mouse models—that can recapitulate the sequential nature of tumor evolution and allow systematic perturbation of each stage. Third, restructure funding mechanisms to support “dual‑aim” projects that combine mechanistic basic research with translational objectives, ensuring that insights into why cancers arise are directly linked to preventive or therapeutic innovation. Fourth, promote data sharing standards and meta‑analysis pipelines that enable cross‑tissue comparisons, facilitating the discovery of shared vulnerabilities and the validation of Peto‑style regularities in contemporary datasets.

In conclusion, the commentary asserts that the apparent conflict between controlling cancer and understanding its underlying regularities is not a necessary dichotomy but a product of historical choices and institutional incentives. By revisiting Peto’s insights with today’s technological toolkit—and by realigning research priorities to value both control and comprehension—the field can move toward a more integrated paradigm. Such a paradigm would not only improve the efficacy of existing therapies but also pave the way for truly preventive strategies that anticipate and intercept the multi‑stage processes that give rise to cancer in the first place.