The I-SPY 2 Clinical Trial

For those of you who read the Wall Street Journal, an article appeared in the Friday, October 1 issue that described the I-SPY 2 (investigation of serial studies to predict your therapeutic response with imaging and molecular analysis 2) clinical trial. This is an adaptive phase II trial designed to facilitate the introduction of new forms of therapy into clinical practice.

The reporter presents the trial as a dramatic advance, suggesting that the era of “personalized care” is finally upon us. I applaud the intent of a trial to apply “window therapy” (i.e. using the window of time before definitive intervention to introduce and test new therapies) to facilitate drug introduction. However, despite the author’s enthusiasm, the design and application of this trial is demonstrably less than meets the eye.

I-SPY2 uses several molecular markers and established prognostics in conjunction with a new molecular profile (mammaprint) to subgroup candidates prior to randomization. The randomization then allows patients to receive either a standard treatment, or one of five investigational drugs combined with standard agents. Sophisticated imaging technologies are used as surrogates for clinical response, while additional biopsies will provide insights into genomic events.

What this trial does not do is utilize molecular markers (beyond those already available to most clinicians) to select patients for therapy. As such, despite the WSJ author’s glowing review, the trial is, at its core, a randomized selection of candidates. While it may enable the investigators to interrogate the tissue biopsies to answer scientific questions of interest, it does so with no immediate benefit to the patients who participate. Indeed, patients who gain benefit (after being randomized to the investigational arm and then receiving a new combination that actually works) receive said benefit by what could best be described as blind luck. The suggestion that this is “personalized care” falls flat when one realizes that a good outcome is nothing more than a chance event!

Truly personalized care represents the application of validated predictive models to select candidates for specific therapies. Good outcomes can then be ascribed to the intelligent selection and application of effective treatments. The cancer research community’s single-minded focus on genomic platforms, to the exclusion of functional platforms, forces patients to continue to participate in “randomized” trials to test hypotheses of interest to the investigators, largely at the expense of the patients in need. These types of advances could be more rapidly made utilizing functional profiles, such as the one offered at Rational Therapeutics.

What these genomic investigators are expecting their patients to say to them is “You may not be able to treat me any better, but I like the way you think.” What informed patients should be saying instead is, “I don’t care how you think. I want you to treat me better!”