Monoclonal tumor origin is an underlying misconception of the RESIC approach

Attolini et al. (1) recently reported a computational approach named retracing the evolutionary steps in cancer (RESIC), the purpose of which is to “deduce the temporal sequence of genetic events during tumorigenesis from cross-sectional data of tumors at their fully transformed stage” (1). The authors (1) indicated that they chose colon cancer as a model in which to validate their approach because there is an “established sequence of genetic events during the multistep process of colon carcinogenesis” (1). The basic premise underlying the RESIC approach (that initial events can be ordered based on the frequency of events in fully transformed tumors) is flawed. In developing the RESIC approach, the authors (1) have not considered a critical aspect of carcinogenesis, specifically, that many, or perhaps most, tumors are polyclonal in origin (2). Ironically, definitive evidence that tumors are polyclonal comes from the colon cancer model (2–4). Despite convincing evidence from human studies (3), consistent findings in rodent models (4), and multiple reviews on polyclonal tumor origin (ref. 2 has additional citations), monoclonal tumor origin too often remains an unacknowledged underlying premise in the field of oncology.

The reason that early genetic lesions cannot be identified based on their prevalence in fully developed malignancy has been presented graphically (figure 2a in ref. 2). Basically, if tumor initiation involves the interaction of two or more clones of cells, then the different clones need not have the same fitness during tumor development, and their population sizes may expand or contract accordingly. Two or more cooperating phenotypes might be needed for initial progression, but subsequent accumulation of additional lesions in one clone could eliminate phenotypic reliance on a second clone as the tumor progresses. Evidence for the polyclonality of tumors is provided by the colon cancer model, where KRAS codon 12 GAT and GTT mutations were analyzed using a sensitive and quantitative approach called ACB-PCR (5). KRAS mutations are present in normal colonic mucosa and at some level in all colon tumors. Furthermore, a statistically significant decrease in KRAS codon 12 GTT mutation level was shown to occur as tumors progress from adenoma to adenocarcinoma. These data argue that the order of mutation acquisition cannot be predicted accurately from mutation prevalence at the fully transformed stage.

The common assumption that all clinically relevant tumor mutations can be identified by DNA sequencing (the input data for RESIC) also is based on tacit acceptance of monoclonal tumor origin. Attolini et al. (1) mentioned that cancer genomes are being analyzed with “unprecedented throughput and resolution.” However, given that most cancer genomes are being analyzed by DNA sequencing, with a sensitivity of <10% on a DNA copy number basis, it becomes obvious that significant proportions of tumor mutations are being missed (5). In my opinion, the clinical significance of mutant tumor subpopulations is an area that will require careful investigation in the future if the use of personalized cancer therapeutics is to achieve its promise.