Maurie Markman, MD
It is difficult to overstate the tension that exists within the academic oncology arena regarding the development of acceptable approaches to evaluate the clinical utility of therapeutics that only involve a small subset of patients within any individual organ-based tumor site. Increasingly robust nonrandomized trial evidence of the benefits of a variety of such approaches is being published in the peer-reviewed oncology literature,1,2
but unfortunately, academic purists continue to argue that the value of molecularly targeted therapy can only be assessed through the conduct of traditional randomized trials.
For example, a recent publication of a randomized phase II trial comparing therapy selected based on matching a specific antineoplastic that was supposedly targeted to a suggested relevant molecular abnormality, compared with treatment that was not selected on this basis, contended that the negative trial outcome (no evidence of superiority for the molecularly targeted approach) demonstrated the current absence of benefit of precision cancer medicine.3
The fact that the drugs available today may have a minimal biological rationale to clinically impact the specific target being examined did not prevent the authors of this seriously conceptually flawed study from making their remarkably sweeping claims.
Similarly, a recent paper even challenges individual reports of exceptional or super responders to cancer drugs by suggesting that what has been observed may be no more than the natural history of the cancer.4
It is notable that the authors of this particular report appear to emphasize the “incompleteness in the reporting” for many of these previously published peer-reviewed cases. This analysis suggests that clinicians who report such N-of-1 experiences are somehow not at the same scientific level as clinical investigators who define evidence-based medicine as focusing solely, or at least extensively, on data from randomized clinical trials.
But the aim of this particular commentary is not to further challenge these manuscripts, although much more could be added to the preceding discussion. Rather, the goal is to briefly highlight one novel evaluation pathway that could be easily incorporated into published reports of these critically important N-of-1 experiences and that potentially could serve as a highly viable alternative to the randomized trial. This fresh approach has the realistic potential to effectively help neutralize arguments that all one is observing with a favorable N-of-1 outcome is the natural history of the malignancy in an individual patient.
As noted in several peer-reviewed publications, while the natural history of cancer progression in an individual patient will vary greatly regardless of the organ-site origin of the malignancy, it is biologically most unlikely that there will be a substantial slowing in the rate of disease progression over time in a given patient in the absence of a successful therapeutic intervention.1,5
Such interventions might include surgical removal of cancerous lesions, localized radiation, and a number of forms of antineoplastic drug therapy.
Therefore, it is objectively reasonable to suggest that if a molecularly targeted therapeutic administered to a specific patient extended the time to subsequent disease progression (as a prospectively defined endpoint), compared with the previously documented time to disease progression from the prior antineoplastic approach employed in that individual, then at a minimum this would represent a clear sign of biological activity of this agent and possibly a meaningful clinical outcome. This conclusion would only be appropriate if there were no other therapeutic interventions undertaken during this period of time (eg, local radiation of a single growing metastatic lesion) that could account for the extension of the time-to-progression interval. One report, which included a well-argued and well-presented statistical argument, required a 30% extension in the time to disease progression, compared with the prior progression-free interval, for an individual patient to be considered to have achieved a positive outcome.1 In the opinion of this commentator, it is reasonable to suggest this endpoint as being clinically and scientifically equivalent to the objective response rate (eg, RECIST criteria) as documented in a phase II or phase III clinical trial.
Further, this progression-free survival endpoint could serve as a rational and highly meaningful surrogate to data obtained in a traditional randomized trial, only in this situation the “study control” arm is the individual patient’s prior timeto- disease progression history rather than simply a different (randomized) patient.