Skip to main content


Keytruda vs. Opdivo: No Contest

You don’t always get to see a head-to-head showdown like this one: Merck’s Keytruda (pembrolizumab) and Bristol Myers-Squibb’s Opdivo (nivolumab) were both the subject of presentations at the European Society of Medical Oncology conference, reporting on non-small cell lung cancer trials, and there doesn’t seem to be any doubt that Keytruda looked better. Here’s the paper on it in the NEJM, and it’s hard to avoid Adam Feuerstein’s conclusions:

Keytruda and Opdivo are both highly effective and commercially successful drugs which harness a patient’s immune system to kill a growing list of cancer types. Both drugs are even effective in treating patients with lung cancer that no longer responds to prior therapies. 

However, in newly diagnosed lung cancer, a commercial market opportunity that could reach $8 billion in sales annually, Merck’s Keytruda works and Bristol’s Opdivo does not.

Why? Because Bristol gambled big by testing Opdivo in a broad population of new lung cancer patients. The bet blew up. Merck played conservative, choosing to target a smaller group of lung cancer patients with a biomarker suggesting Keytruda would be more effective. Merck’s strategy paid off.

Merck’s Phase III was restricted to enrolling patients who were expressing the PD-L1 biomarker in at least 50% of their cells (which still gives you about half of the total number of available NSCLC patients). Response rates were high compared to the standard chemotherapy, and adverse events were lower. From what I can tell, Keytruda not only improved progression-free survival compared to the standard, but also the real-world endpoint of overall survival. The paper has 6-month data, and you would hope and certainly expect the difference to continue to increase as you go further.

Bristol-Myers Squibb, for their part, were reporting the full results of the study that was such bad news for them recently, and the numbers are, if anything, worse than expected. They set a lower cutoff for PD-L1 expression, using their own test whose numbers are hard to compare to Merck’s, and at those levels Opdivo shows no benefit at all compared to the standard chemotherapy regime. At this point, based on the clinical data we have, the differences between the two drugs couldn’t be more stark – and for two antibodies that are probably quite similar, the lessons about what differences can show up in how you run your clinical trials couldn’t be more stark, either. . .

43 comments on “Keytruda vs. Opdivo: No Contest”

  1. Barry says:

    we’ve seen this scenario played out before. Genentech’s Tarceva showed clinical efficacy and Iressa (a very, very similar drug) showed inefficacy because Genentech aimed for a narrower indication, pre-selecting their clinical population with a fluorescence assay
    It’s a part of the larger lesson of Gleevec. In oncology, you should always target the narrower, better-characterized indication first. If you succeed there, you’ll have the resources to try more clinicals in broader indications.

    1. ScientistSailor says:

      Good point Barry, but the lesson is from Herceptin, not Gleevec…

    2. SedatedFMS says:

      I presume you mean OSI’s Tarceva?

    3. Stewart Lyman says:

      This does not always play out that way. Back in the early 90’s, Immunex went for the smaller indication for Leukine (GM-CSF), and Amgen went for the wider indication with Neupogen (G-CSF) for patients in need of white cell stimulators. Amgen went on to get over 90% of the market, and Leukine never recovered in the US marketplace. IF you go wide and win, then you can dominate the market from the get go.

  2. Me says:

    Big win for lung cancer patients.

    Makes the mind boggle – how many good drugs do we have with bad study designs?

    Vice versa?

    1. Janex says:

      The problem is marketing. I worked for a company once which insisted that marketing have a major say in clinical trial decisions, which meant that when the groups were sitting down and discussing alternative trial designs – that the broadest clinical trial on the table (aka the trial producing the largest customer base) was always chosen. When the trials bombed (they overshot and were too non-specific resulting in no significant effects), the marketing guys blamed it all on the clinical scientists, took their large bonuses and went to work for someone else.

      1. zero says:

        If I was an investor I’d be practically litigious with anger over marketing reps getting massive bonuses for a drug that didn’t get approved.

      2. IR says:

        Hmm, was that Pfizer by any chance?

        1. KM says:

          Sounds like AstraZenca.

  3. Vikash Peesapati says:

    Quoting from Adam Feuerstein’s article:

    “In an attempt to mitigate the study’s failure, Bristol examined patients containing tumors that expressed higher levels of PD-L1. The hope was that these patients would respond better to Opdivo, just like they did to Merck’s Keytruda. They did not. Even in patients with tumors expressing 50% PD-L1, there was no difference in tumor progression or survival between Opdivo and chemotherapy, said Fouad Namouni, Bristol’s oncology drug development chief. ”

    This points to basic differences in the drugs themselves. I don’t think the entire blame for the poor results lies in clinical trial design.

    1. PD-1 guy says:

      Not necessarily. I can tell you based on first-hand knowledge that there are profound differences in performance and reliability between the BMS anti-PD-L1 diagnostic mAb and Merck’s. I would encourage those who’re able and who have a vested interest to do so to compare them head to head.

      1. loupgarous says:

        A heads-up comparison between the two anti-PD-L1 assays is a very good idea. If nothing else, it’s valuable to know why Opdivo wasn’t better than standard chemo and Keytruda was.

        I know it’s post-hoc analysis, but it might be informative to work the Opdivo analysis backwards and see which subsets of BMS’s newly discovered small-cell lung cancer cohort responded better than standard chemo – then try to identify why that was.

        Won’t be easy, (a) because you’d be doing post-hoc analysis, always suspect and to be avoided, (b) it’ll be very hard to identify confounding factors, and (c) the study wouldn’t be controlling for the new factors. I wouldn’t make a safety or efficacy conclusion from this re-examination of the data, merely use it to gather possible useful information on why Opdivo performed as it did in this study, to guide design of future studies of Opdivo.

  4. D square says:

    I admit my ignorance here.
    Can someone clarify if there was any experimental confirmation of target engagement for either antibody in these trials?
    Also… since Derek mentions the two antibodies could be quite similar (I assume structurally), how are these compounds qualified? Besides binding affinity to the PD-L1 target, is their ensuing functional response established somehow?


  5. Chrispy says:

    I’m curious about the merits of going after the PD-1 side on the T cells, as these drugs do, rather than the PD-L1 side on the tumor. You’d think going after the tumor would have clear advantages; for one thing, the antibodies wouldn’t need to be immunologically silent. Rather than simply blocking the path they could actively induce an ADCC response. But perhaps PD-L1 is too widely expressed in normal tissue?

    1. PD-1 guy says:

      PD-L1 is highly expressed by normal spleen and liver and can be very highly expressed on tumor. Moreover, it’s rapidly turned over and internalized upon binding by antibody including the Roche drug. PD-1 expression is much more restricted and half-life of bound antibody on the surface quite long.

    2. Barry says:

      One advantage of going after the PD-1 side (on the T-cells) is that they’re genetically stable, while the PD-1L side (on the tumor) is genetically unstable/prone to developing resistance

  6. UK Chemist says:

    Scientist Sailor
    I think you’ll find it was Tarceva

    1. ScientistSailor says:

      No, the lesson of using a diagnostic to narrow the pt. population was done with Herceptin before Tarceva or Gleevec. The drug would not have been efficacious in an all-comers breast cancer trial.

  7. steve says:

    I take the point about the differences in trial design but that doesn’t necessarily mean it’s the reason for the difference in outcome. Different antibodies can have very different effects; one antibody might cause an antigen to be internalized, for example, whereas another antibody binding to the same antigen but a different epitope may not. Affinity/avidity, cross-reactivity and other parameters could differ as well. The epitope for Keytruda has been defined ( but to my knowledge the one for Optiva has not. PD-1 may differ in different cells through different glycosylation, expression levels, etc and it may be that there are differences between lung and melanoma that explain why Optiva worked in one rather than the other. It may well be just a trial design issue but it could also be reflecting something intrinsically different about the two antibodies,

    1. steve says:

      Correction: I meant that PDL-1 can differ on different cells, so the way it interacts with PD-1 could differ in different tumors

    2. PD-1 guy says:

      The epitopes are similar, but not identical. The two drugs are very similar with a slight edge in potency to Keytruda.

  8. Imaging guy says:

    This article by a patent attorney might answer some of the questions raised in the comments.
    “Intellectual property issues of immune checkpoint inhibitors”
    doi: 10.1080/19420862.2015.1107688

  9. Bruce Quinn says:

    There is this urban myth that Keytruda only worked in (+) patients in second line, but Opdivo worked in all patiens. No! If you go to the current 10-2016 Opdivo label, yes, it was approved for the whole population in second line lung cancer, because that’s how the trial was done. No biomarker up front. But at the back of the FDA label there are several data cuts by PDL1 expression. The impression is that zero to low PDL1 expresses had zero benefit. No doctor who read the Figure 8 Page 15 of the Opdivo label would have seen a reason to give Opdivo to “all” patients. Of course, no normal doctor would spend time reading page 15 of a label, but it’s there as a matter of public record and has been for a long time.

    1. PD-1 guy says:

      And imagine if your means to accurately assess PD-L1 levels in patient samples was flawed. Would further confound response rate interpretation, yes?

      1. Bruce Quinn says:

        I would say it means there was ONE method by which low or zero expressers had zero…zero… clinical benefit. That’s a powerful take home lesson. And it’s on the label, just on page 15 not page 1. It was no secret. You didn’t have to find some obscure journal article.

  10. anjampie says:

    “Opdivo shows no benefit at all compared to the standard chemotherapy regime”… or could have been phrased “Opdivo is as effective as the standard chemotherapy regime (but without most of the nasty side effects of chemo”. So there is activity in lower level biomarker population, just not as much as in the high population… which can’t be a super big surprise really.

    1. loupgarous says:

      And if the overall toxicity of Opdivo’s less than standard chemo, you’ll have been right.

      Speaking as someone whose bone marrow (and immune system) were annoyingly dinged by chemo and radiation, it’s worth having a different club with which to whale away at cancer.

  11. cooky emerson says:

    so does this mean someone with metastic liver cancer will not benefit from opdivo and would benefit from keytruda

    1. Franchelle Richardson says:

      Cooky, did you receive an answer to your question “so does this mean someone with metastatic liver cancer will not benefit from Opdivo and would benefit from Keytruda”? I have metastatic liver cancer from lung primary and am scheduled to begin Opdivo on Thursday, February 2, 2017. I only read about Keytruda today. Do you think I should question my oncologist about his decision to start me on Opdivo?

  12. Joan says:

    I believe that Merck has been misleading patients and oncologists about keytruda in lung cancer trials. The keytruda lung cancer trial was combined with chemotherapy, and 50 percent PD-L1 positive ; the Opdivo lung cancer trial was Opdivo ONLY, without chemotherapy. The reason that Bristol did not choose a narrower patient population with PD-L1 positive, because the PD-L1 test would take about a month to get result and is very inconvenience for oncology clinic. I don’t believe Merck’s marketing spin because oncology drugs have been Bristol’s specialty for three decades. Merck had no oncology drug before and Keytruda was their first one.

    1. Andy II says:

      Joan: It does not seem to be a combined treatment of Keytruda and Chemo. At least, in KEYNOTE-024 study, the patients received either pembrolizumab (at a fixed dose of 200 mg every 3 weeks) or the investigator’s choice of platinum-based chemotherapy. Crossover from the chemotherapy group to the pembrolizumab group was permitted in the event of disease progression from NEJM 2016.

  13. Theresa Delano says:

    I have had lung cancer since 2006. Had my left lower lobe removed in 2006, they caught it a the very beginning🙏🏻 2010 I had my right upper lobe removed. But I skipped my cat scan, because I was feeling so good!! Well bad mistake on my part!! 2011 went from nothing to stage 4, non small cell lung cancer! It spread to both lungs and my scapula bone on my back!! So I’ve been on and off chemo since then. I’m not sure how many different chemo’ I’ve been on?? December 2015 my doctor put me on Opdivo. Was on it for a year, it grew a little bit with every scan!! Well December’s scan showed a lot more than usual!! Stopped it, the opdivo. Did not work at all on me😢 So I’m back on another chemo, which my Doctor says is very new and promising🙏🏻🙏🏻 It was great to be on nothing for a year🎉🎉🎉. So in a nutshell, I’ve had stage 4 lung cancer for 5 years❤❤❤

    1. Theresa Delano says:

      I’m on a mission!! With so many people Praying for me, I’m gonna kick cancer❤

      1. loupgarous says:

        God bless you, Theresa, and keep you strong!

  14. Theresa Delano says:

    The name of my new chemo is navelbine🙏🏻🙏🏻🙏🏻

    1. Karl says:

      Nivolumab? Theresa? i start it tomorrow for a base of tongue mass. now im second guessing.

      1. loupgarous says:

        You, too, Karl, God be with you!

  15. Grampy says:

    I have been on Opdivo for about a year now. My small cell lung cancer has continued to grow at a slow pace and now occupies the entire lower lobe of my right lung. I am wondering if any studies have looked at patients on Opdivo who switched to Keytruda (or vice versa) and if there were any changes in growth rate of the cancer.

  16. Tim says:

    It’s so sad to hear this author talk about the drugs like they are a dog race. He obviously hasn’t lost or had family with terminal cancer and I forgive him for his ignorance and insensitivity.

    If Bristol Meyers didn’t gamble gallantly on that trial then we would never know the outcome of low PDL1 protein patients. Where Merck was being safe, Bristol was advancing through raw results.

    Where you see profit and loss, I see advancement and transparency to it’s effect and results. If Bristol played it safe we wouldn’t know the effect on Kate stage low protein patients. So I applaud them for taking that chance and being human and not greedy robots like some people….

    1. Mol Biologist says:

      Agreed it is not dog race, especially according to Reuters: Merck & Co said on Monday that its Keytruda immunotherapy failed to extend survival in previously treated patients with advanced head and neck cancer more than the standard combination therapy in a late-stage trial.
      IMO recent Forbes publication sounds at least overoptimistic since Keytruda immunotherapy trials on melanoma and lymphatic cancer were stopped again due to high numbers of unexpected death.

  17. Gargi Take says:

    I need help with this exact topic for my Mother’s treatment
    Which one of the 2 medications is more effective and has lesser side effects for advanced metastatic malignant melanoma? Pembrolizumab Keytruda vs Nivolumab?

    1. Pennpenn says:

      Honestly that’d probably be a question to put to whoever is responsible for managing her treatment and not a bunch of randoms on a blog (who yes, almost certainly have some medical knowledge mixed in, but still it would not be even remotely responsible for anyone here to reccomend treatment options).

      1. Derek Lowe says:

        This is true. I’m not even sure that there’s a firm answer to your question, so even the MDs who come by here might not be able to help. And this would be for a patient that they haven’t even seen.

Comments are closed.