Elon Musk, the founder of Tesla Motors, once said, "It’s okay to have your eggs in one basket, as long as you control what happens to that basket." This was brought to mind by a recent article in the New England Journal of Medicine entitled “Vemurafenib in Multiple Nonmelanoma Cancers with BRAF V600 Mutation”. The study tested whether genetic pre-screening could provide high response rates for Vemurafenib in patients who were found BRAF-mutation positive. That is, the target of the drug.
As Vemurafenib, FDA-approved for BRAF mutated malignant melanoma provided response rates up to 50% in this disease, it was reasoned that this could be replicated in other cancers that carried the same BRAF gene target. Using a study design known as a “basket trial”, 122 patients with a variety of cancers who were found (+) for BRAF mutation were offered the drug. Again, only patients found (+) for BRAF could accrue. This enriched the pool, ensuring the highest probability of clinical success. It was hoped, I dare say expected, that response rates similar to melanoma would be observed.
The results were underwhelming. Much to the chagrin of the authors, only 18 of the 75 patients treated had objective responses (24%) While better responses were observed in lung cancers (42%) and a very rare form of cancer known as histiocytosis (43%), the response rate in all other cancers, even after careful preselection was only 9.5%. More troubling was the fact that only 1/27 (4%) colorectal cancers showed objective response, even after Erbitux was added to the mix. Similarly only 1 of 8 biliary and 1 of 7 thyroid cancers responded, again after careful genomic pre-screening.
This prompted a certain amount of navel-gazing on the part of the authors, who stated "an important implication is that conventional tumor nosology based on organ site cannot be entirely replaced by molecular nosology." They continued, these data present a challenge to clinicians who want to make treatment decisions on the basis of tumor genomic profiling."
While human genomics has progressed rapidly, providing insights into many diseases and some treatments, the field has suffered from over exuberance manifested in the widely held belief that genomic analyses hold the key to all medical problems, big and small.
These favorable results in lung cancer and histiocytosis are encouraging but will need confirmation. Meanwhile, the numerous commercial entities, universities, and national institutions that tout genomic profiling as the panacea for cancer must continue to confront the painful reality that human tumor biology is more complex than the sum of its genes.
Cancer is a disease of context and as the accompanying editorial points out "One bio-informaticist's driver mutation is another's passenger mutation." This speaks to the fact that the presence of a mutation is but one factor in the broader disease process that drives cancers. Functional analyses based on phenotype (the observable properties of an organism that are produced by the interaction of the genotype and the environment) can interrogate the importance of a given genetic abnormality. Many alterations present in tumors may have no impact on their behavior. Only through the careful study of cancer phenomics and cellular responses will targeted agent activities be accurately determined. We believe this establishes the pressing need to incorporate phenotypic, as well as genotypic, analyses in the study of human cancer.