Skip to Content

Bristol-Myers Squibb Sows Confusion

Yesterday was a weird day for Bristol-Myers Squibb. The company announced the results of a highly anticipated clinical trial, one combining Opdivo (nivolumab) and Yervoy (ipilimumab) in lung cancer patients. As those who follow this area know, BMS has been losing ground to Merck for the last year or two, as Merck’s Keytruda (pembrolizumab) makes gains in this area (a situation that goes back to the earlier clinical trials). This is a big result in the Big Immuno-oncology Hoedown, as everyone tries to figure out which types of cancer are best treated by these agents, how to identify such patients up front, and what kinds of combinations (IO and non-IO) work better than others. There are a huge number of possibilities, there are very large potential medical rewards, and many of the key questions are unanswered, so it’s a pretty fraught business.

Opdivo and Keytruda are both antibodies to the PD-1 receptor (as opposed to others that block its ligand, PD-L1), and Yervoy is an antibody to CTLA-4. There’s certainly room to think that simultaneous attack on these targets could be beneficial, but this was the first big controlled test of that idea. Matthew Herper has a good summary of what’s actually happened, though:

This study, called CK-227, was important because it aimed to establish that two immunotherapies are better than one. It was actually two different trials: one in patients who tested positive for PD-L1, and one in patients who tested negative. PD-L1 positive patients would receive one of three treatments: the combination of Opdivo and Yervoy, Opdivo alone, or chemotherapy. PD-L1 negative patients would receive the combination of Opdivo and Yervoy, Opdivo plus chemo, or chemo alone. Investors expected to find out whether the Opdivo and Yervoy beat chemo or Opdivo alone.

But Bristol decided to change the main measure of success. The company had already increased the size of the study – there were now 1200 PD-L1 positive patients, and 700 PD-L1 negative ones – and pooled the results. And it had changed the goal: instead of looking at PD-L1, the study would pool patients in both groups and pool them by another measure, tumor mutation burden, or TMB.

One problem is that no one knew about these changes until yesterday’s announcement. So the reaction was initially cheers for the successful trial, followed (in case of the sharper observers, followed within a minute or two) by “Wait, what?” The changes to the trial were one issue, the lack of transparency about how the trial had changed was another, and just the fact that a company this large, with such an important readout, would have done something like this was a third. The general feeling set in that you don’t mess around with a trial’s design lightly, and the fact that the company had done so in this case (and without saying a word) raised suspicions that the whole thing was an attempt to produce more positive results.

That may or may not be a fair criticism, but Bristol-Myers Squibb must have known that it was coming their way. One big problem is that tumor mutation burden has not (so far) been shown to be a predictor of overall survival in this area, so making that the big headline of your trial is going to raise unanswerable questions. And sure enough, the company got plenty of them. The press release was followed by a conference call with analysts, which was followed by a general press conference, but no one still knows the answer to “Is the Opdivo/Yervoy combination better than Opdivo alone?” The high PD-L1 portion of the trial is continuing to report on overall survival, but until that reads out (and more data are released in general), everyone watching the field is going to be somewhat more confused than they were last week. Which is not what you want your big clinical trial announcement to do for you, presumably.

The company’s stock jumped in the premarket and opened higher. But by 10 AM it was back to the previous close, and it spent the rest of the day down, with pretty much everything else. It’s sheer bad luck to have a big announcement on a day that the Dow drops over 1100, of course. But a lot of the other stuff was, perhaps, avoidable. (I’m writing this, as usual, well before the market opens on Tuesday, and won’t today be fun!) So the immuno-oncology field is (at least for now) as wide open as ever. Clarity will come – but not soon, and as is often the case,  it will arrive in pieces. Some assembly required.

19 comments on “Bristol-Myers Squibb Sows Confusion”

  1. Insilicoconsulting says:

    Assuming it wasn’t an adaptive trial, can companies change endpoints /protocols so late into the process?

    1. tnr says:

      Not for FDA registration purposes. For announcing to public, they can say anything they want.

  2. Mad Chemist says:

    Opdivo, Yervoy, and Keytruda, hmmm?
    Why do drug companies always name their drugs in Klingon/Zerg/Huttese/whatever? It doesn’t make sense.

    1. Another Guy named Dan says:

      You need a marketing name that can be remembered, a generic name such that your branded formulation is distinct, and neither of them can have any embedded secondary meanings in either English or other language (see the mostly mythical story of the Chevy Nova in Mexico). The easiest way to accomplish this is to combine recognizable syllables into word-like combinations.

      1. Jeff says:

        “Ask your doctor if Darnitol is right for you!”

        1. Louis says:

          When I was at Pfizer there was a (very tongue in cheek) suggestion to give sildenafil the brand name “Coxaphlopin”…

          …It was not received with enthusiasm.

          1. john adams says:

            Thanks for that one. Absolutely hilarious !!!!!

    2. Wavefunction says:

      Agreed. Opdivo sounds like an optometrist, Yervoy sounds like a Yiddish pastry and Keytruda sounds like a Harry Potter curse.

    3. b says:

      It’s not the drug companies, it’s the FDA, and it’s rather complicated. See:

      https://www.fda.gov/downloads/Drugs/GuidanceComplianceRegulatoryInformation/Guidances/ucm075068.pdf

      1. steve says:

        It actually is the drug companies. They want the generic to be as difficult to pronounce as possible because it makes their easy-to-pronounce brand name the one people ask for.

        1. Some Pharma Cog says:

          No. It’s the regulators. A contest we had at my company to come up with the generic name for our advanced drug candidate had to be abandoned because after meeting all the regulatory requirements, employees only had flexibility to change 3 or 4 of the 12 total letters in the name.

          In any case, if drug companies are supposed to be making brand names easier to pronounce than the generic names, somebody ought to tell Pfizer. I’d prefer to say enzalutamide over Xtandi any day.

  3. David Young MD says:

    Pembrozilumab (Keytruda) is FDA approved for first line lung cancer in patients who have PD-L1 positive tumors. So, an astute oncologist will check the tumor for PD-L1 expression and offer Keytruda to those patients whose tumors overexpress PD-L1. It is these patients who are the most likely to respond favorably. Opdivo is for second line or and can be given to PD-L1 negative tumors, in which the outcome is not as assured. So, you can see that Keytruda becomes the more attractive drug…. from this alone. In other regards, they are both good, safe medications.

  4. Ex-Pharma guy says:

    Merck’s anti-PD-L1 diagnostic mAb is a far better reagent than is the one used by BMS (or Roche). This difference surely impacts apparent correlation of PD-L1 status with response and outcome.

  5. David Borhani says:

    I think potentially very relevant here are the recent reports in JCI that it’s PD-L1 on *host myeloid/antigen presenting cells*, not the tumor cells per se, that drives response to therapy. So, as Ex-Pharma guy says, it may be not just the better diagnostic mAb (which help pick the correct patients), but also exactly which cells inside the “tumor” give rise the positive signal to that diagnostic mAb.

    https://www.jci.org/articles/view/96061 primary article 1; free
    https://www.jci.org/articles/view/96113 primary article 2; free
    https://www.jci.org/articles/view/99047 commentary; paywall

  6. Anon says:

    As a ex-Merck employee I have observed and lauded over many years that BMS has great chemists and visionary leaders whose tactic of bringing break through projects to light way ahead of their peers, including Merck. But, along the way to fruition, they wimp out and it keeps on giving! Compare Januvia/Onglyza and now Keytruda/Opdivo. May be there are others and their screw-up need not be related to science or research, but sheer clumsiness and it shows in $$$$.

    1. Mol Biologist says:

      As not a ex-Merck employee I am wondering if this type of “immunotherapy” is not useful for cancer treatment in general. And high level of PD-1 is just indications of adversed metabolism similar to elevated level of symmetric dimethylarginine (SDMA) and asymmetric dimethylarginine (ADMA) as risk markers of cardiovascular disease.
      https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5505150/
      Oppressed and deficient immune system can not be activated before it is restored. Same as pre-conditioning “drugs” did not work in diseased or failing myocardium.

  7. The companion and complementary diagnostic assays linked to the use of the different PD-1/PD-L1 inhibitors plays an important role for the efficacy in NSCLC. The ways that the individual PD-L1 expressions assays have been developed are different, especially with regards to the clinical cut-off selection, which seems to have major impact. This issue is discussed in article just published in published in Therapeutic Drug Monitor.

  8. Jan Trost Jorgensen says:

    Link to the article in Therapeutic Drug Monitor: https://www.ncbi.nlm.nih.gov/pubmed/29084031

  9. oliver says:

    Dear Derek,
    regarding the changing of the trial outcome measures

    Wasn’t it that what the PACE trial for CFS did?

    And look how that turned out to be?

Leave a Reply

Your email address will not be published. Required fields are marked *

Time limit is exhausted. Please reload CAPTCHA.