understand why the Phase II and III failure rates are so high in oncology, or why—despite the highly innovative and targeted nature of the latest therapies—that survival gains are relatively small.
Without understanding what disease(s) a patient has, no matter how powerful our medicinal tools may be, we will not be able to apply them to their full potential. Three things must change to realize this vision:
The first change: Diagnostics
The root of the problem is not oncologists, but rather the diagnostics tools available to them. The limitations of surgical biopsies are well known, in particular for the inability to repeatedly sample tumors for changes during the course of treatment and the likelihood of missing tumor regions with actionable subpopulations. These complications are exponentially increased in the metastatic state of the disease. The introduction of early CTC-based liquid biopsies afforded the requisite resolution over time, but their focus on counting and characterizing only a few classes of CTCs limited their clinical utility. Liquid biopsies based on cell-free DNA (cfDNA) from tumor cells provide greater sensitivity, but they still don’t identify the cellular origins or the combinatorial “stacking” of mutations.
Even these newer diagnostics (whether cfDNA or CTC analysis based on epithelial protein expression or size-based selection) analyze cancer as a “disease of averages.” This view precludes an understanding of how cancer actually progresses. It precludes detection of minor subpopulations of metastatic tumor cells that can (unexpectedly) drive drug resistance or that may have stacks of actionable mutations that require careful therapeutic triage.
Therefore, the most utilitarian diagnostics for cancer management must report the cellular drivers of disease by detecting the clones that make the patient susceptible or resistant to treatment. Since CTCs are a vital mode of metastasis and also reflect tumor heterogeneity, future diagnostics must be able to detect all CTC species present. Upon detection of the full range of CTCs present, each CTC can be interrogated by a multiplicity of proteomic and genomic biomarkers. Such tests must also account for the realities of clinical infrastructure and the costs payers can bear. Only then can clinicians account for the frequency and distribution of the subclones behind a patient’s active disease, and help guide singular or combination therapies.
The second change: Rational monitoring
Monitoring a patient’s disease based on subpopulations of CTCs will move oncology from a reactive to a proactive approach.
Health care systems must also prepare for the massive big data challenge that will come with prioritizing subpopulations of patients for treatment and constructing individualized combinations of therapeutics. The informatics needed to support oncologists in these analyses and the need to share individual outcomes will drive additional changes in therapies, medical institutions, and the healthcare industry.
The third change: Drug development and regulatory approval
An equally complex change will be how clinical trials are structured and how drugs are approved.
We are now learning that the success of a patient in a clinical trial is not just about the presence of a drug target within an undefined portion of a patient’s tumor cells. It’s equally, if not more important, to tease apart cellular diversity to check for resistant clones that may cause the patient to be refractory to the experimental drug in question. Having such a broad view of the patient’s cancer will help diversify risk by recruiting the right patients to the right trials, which can provide more meaningful conclusions. Such precision recruitment will also create new opportunities for collaboration between drug developers to reduce study start-up costs.
Understanding tumor heterogeneity also will change the regulatory pathway for approval. The indication will no longer be based on tumor stage or origin, but rather the mechanism of action and subpopulations affected. Approval and reimbursement will no longer be predicated in large part on survival endpoints, but rather on a candidate drug’s ability to combat specific cell populations of common phenotypes and genomic drivers.
How we treat cancer, diagnose cancer, and develop anti-cancer therapeutics today is all based on a reductionist approach to treating the disease as a monolith, as an average. As the New Year begins, let’s resolve to make 2016 the end of the average. The future of oncology is in heterogeneity.
