Solving Clinical Trial Equations

Biotech investors are often left to solve a variation of the following equation:

a(OR) + b(PFS) + c(DOR) = OS


  • OR is the magical objective response rate, in %*
  • PFS is the mystical progression free survival, in months
  • DOR is the mercurial duration of response, in months
  • OS is the elusive overall survival, in months (or years if you’re brave)


No problem! A few hard numbers to define coefficients a through c, some reliable OR, PFS, DOR values, and you can triumphantly tap Return to cue Excel’s internal canine to fetch your answer. Unfortunately, none of the cases that matter are ever so easy. And like your high school physics teacher, biotech companies know to present the easiest examples in class, but send you home with problems that bear no relation.

And so we’re left with relatively confusing situations, like Incyte’s epacadostat development program. This program got off the ground with OR data in combination with pembrolizumab for the treatment of advanced melanoma, presented in late 2015. It was quickly signaled that a phase 3 trial would soon begin, hoping to extend PFS in patients receiving combination therapy. But as the data from this small trial matured, its fit within the comparator landscape did not necessarily become clearer.

Screen Shot 2017-08-22 at 10.13.02 AM.png

The above table attempts to provide some clarity to the comparisons that await epacadostat in this disease setting. Although the OR for the epacadostat combinations appear numerically higher than PD-1 monotherapy, the epacadostat phase 3 trial is measuring PFS, not OR. PFS for the combination therapy has been updated only once, at SITC 2016 (right). Inexplicably, the company did not update this melanoma dataset at all for ASCO 2017. 

That leaves us with the immature data presented in 2016 as the bellwether for the phase 3 ECHO-301 trial. But due to the small patient numbers in the ECHO-202 dataset, neither the OR or PFS values are comfortably clear of the PD-1 monotherapy ranges. Attempting to better estimate the PFS data through DOR is not helpful either, as the latter are too immature to provide any meaningful insight. Incyte has hinted these data may be updated by year end, but will that provide the necessary confidence? Given their reluctance to be transparent with the accrued data, it’s fair for investors to be skeptical about the quality of the developing dataset, and the extent of curation it seems to require.

But fear not: Incyte has suggested the consistently high OR across disease indications should give investors all the confidence they need. Below are a stack of tables that attempt to position the epacadostat combination data in various indications versus the results from PD-1 and PD-L1 monotherapy trials. One theme is common: although the OR trend higher in the epacadostat combinations, there are no PFS data available, and the DOR data are too immature to remove any uncertainty about outcomes.

This uncertainty around these data have been compounded by Incyte’s strategy towards their release. At ASCO, the company provided a data snapshot that was months out of date. A convoluted explanation was weaved that included a desire to focus on more “mature” datasets, as well as calls for consistency. In effect, they admitted to having more data, but chose not to disclose it. This triggered a cardinal rule for assessing company communications: If the data aren’t disclosed, they’re likely not favourable.

Although this rule is not infallible, it does capture the essence of most public company disclosure strategies. A company is much more likely to keep the veil in front of poor datasets than to obscure strong data they have in hand. That places Incyte’s ASCO disclosure in a rather poor light, and tempered initial enthusiasm garnered at the time of abstract release.

So are Incyte’s actions around their IDO data disclosures necessarily a bad omen? Not entirely. The immunotherapy doublet landscape is far from formed, and it would not be unusual for a company to limit public dissemination of data on competitive grounds. For instance, pembrolizumab and nivolumab are engaged in a head-to-head battle for NSCLC, and the competitive / strategic choices by their sponsors have had large consequences. In that regard, it was notable Bristol-Myers did not release data from the NSCLC cohort of the ECHO-204 study testing nivolumab plus epacadostat. It’s not controversial to suggest there were NSCLC enrollees in the trial, since Incyte and Bristol-Myers have already announced intentions for a phase 3 trial. Therefore, the lack of disclosure in this instance was likely for competitive reasons.**

On the other hand, it is very likely that Incyte management does feel some pressure to diversify and increase revenues. Although ruxolitinib sales continue to grow, the difficulty with baricitinib’s FDA application has definitely increased the focus on epacadostat. And although the series of phase 3 trial announcements for epacadostat may signal confidence, it is not unreasonable for companies to take steps to more quickly increase their revenues. In that way, the phase 3 trial announcements could be viewed as a calculated risk in a competitive landscape rather than an outward show of confidence. After all, phase 3 trials can always be stopped.

Regardless of the hopes investors may have for this entire development program, a dispassionate look at the totality of the data doesn’t reveal unequivocal promises of success. Risk minimization and protection are prudent, as it wouldn’t surprise if the phase 3 readout next year was the first of multiple disappointments.



* Why is this termed a “rate” when it is presented as a percentage? Nonsensical conventions are maddening. Someone needs to rename it.

** This could be the ultimate bluff by Bristol-Myers to goad Merck into spending its money. At which point, well played.