Checkpoint inhibitors like pembrolizumab (Keytruda) have revolutionized treatment for many late-stage cancer patients, including Jimmy Carter. Recently, Keytruda stunned the world again with efficacy in metastatic non-small cell lung cancer, the leading cancer killer. Here’s a summary of its recent successes and the surprising path Keytruda took through two acquisitions and the Merck “out-license” list.

“Stage IV” and “metastatic” are phrases most hear as synonymous with “death sentence.” Untreated, the five-year survival rate for stage IV melanoma is a dismal 15-20%. For those with brain metastases, fewer than 6% survive beyond 2 years, even with surgery and radiation treatment.1 That’s why it’s incredible that Jimmy Carter is still alive and cancer-free at 93, nearly three years after it became public that he was battling melanoma with multiple liver and brain metastases. Carter is just one of the many now benefiting from treatment with pembrolizumab (Keytruda), a drug of the revolutionary checkpoint inhibitor class of immunotherapies.

(Left) Survival curves for a cohort of melanoma patients with brain metastases between 1985 and 2000. Reproduced from K.M. Fife et al.J. Clin. Onc. 200422, 1294- 1300. (Right) Jimmy having a good time in 2018.

After Carter’s treatment in 2015, Keytruda continued to stun the world with activity in metastatic non-small cell lung cancer (NSCLC), the leading cause of cancer death.  In a 2016 trial comparing pembro vs. chemotherapy in biomarker-selected, metastatic NSCLC patients, pembrolizumab showed such clear activity that the trial was stopped early so that the patients in the control chemotherapy arm could receive the experimental drug.2  This year, two Ph. III studies showed pembro plus chemo dramatically increased survival rates vs. chemo alone in metastatic NSCLC, cementing pembro’s place as part of the new standard of care for this disease.3  This is amazing given that the standard of care (chemotherapy) has not changed significantly in decades, and not so long ago, doctors wondered whether chemotherapy did anything at all in NSCLC!4

(Left) Progression-free survival in metastatic NSCLC patients with PD-L1+ tumor proportion score >50%, reproduced from data from KEYNOTE-024; (Right) Overall survival in metastatic nonsquamous NSCLC patients with PD-L1+ tumor proportion score >50%, reproduced from data from KEYNOTE-189

An Inflammatory Program

The durable responses observed with Keytruda are attributed to its ability to activate the immune system.  Ironically, it was actually discovered in an anti-inflammatory program seeking to suppress autoimmunity.5  Researchers in the lab of Andrea van Elsas at Organon were looking for activators of the PD-1 receptor, which might suppress inflammation for diseases like rheumatoid arthritis.  They didn’t find any PD-1 agonists – instead, they only found exceptional inhibitors of PD-1, which resulted in increased inflammation. At the time, there was still a lot of skepticism about the immuno-oncology field in general, though anti-CTLA-4 antibodies were just beginning to see some success.  Despite Organon’s non-oncology focus and relatively unimpressive preclinical data for PD-1 inhibitors (in syngeneic mouse models, PD-1 inhibition rarely lead to more than a delay in tumor growth, and did not affect growth of the B16 mouse melanoma tumor model6), Organon’s management team was persuaded to keep the program alive for oncology.

Unfortunately, while work to enable an IND for clinical development of pembro was on-going, Organon was acquired by Schering-Plough in 2007.  During the portfolio merger process, the PD1 program was ranked “low if not dead last because nothing in tumor immunology ever worked.”  Fortunately, the champions behind the PD-1 program including van Elsas were able to build support at Schering to move the PD-1 inhibitor into clinical development.

The PD-1 program was ranked “low if not dead last because nothing in tumor immunology ever worked.”

Just as the PD-1 program started to re-gain momentum, in 2009, Merck acquired Schering-Plough.  This time, the PD-1 program was officially shut down, and the molecule which would become Keytruda was put on the out-license list for a bargain.  The Organon/Schering/Merck PD-1 program was all but doomed when in 2010, Bristol-Myers Squibb published their positive Phase III results with ipilimumab (Yervoy).  “Ipi,” an anti-CTLA-4 checkpoint inhibitor acquired by BMS from Medarex, had shown a significant survival benefit in late-stage, metastatic melanoma.  With proof of concept for the checkpoint inhibitor class of molecules demonstrated, BMS was eagerly pushing a second Medarex checkpoint inhibitor through clinical trials, this one targeting PD-1.

Zero to Sixty

Significant and durable activity in late-stage cancer by a new class of molecules doesn’t appear often. Merck’s R&D leadership knew they were years behind BMS in checkpoint inhibitor development, but this was an opportunity that could not be missed.  In reaction to BMS’s data, the Merck PD-1 program was catapulted from “out-license” to patient enrollment in less than a year.

“Among the first 7 patients we enrolled (in an advanced melanoma cohort), 6 had objective responses. … I could not sleep if the trial closed and I would have suitable patients who could respond like the others and no slots for them.”

– Antoni Ribas, KEYNOTE-001 Investigator

Clinical investigators quickly noticed that pembro was different from other therapies they had used before.  Significant responses were already being observed in the phase I dose escalation process in the advanced melanoma cohort.  Antoni Ribas, a key investigator in the trial7 observed, “Among the first 7 patients we enrolled (in an advanced melanoma cohort), 6 had objective responses.” He urged the Merck team to share his data with upper management, worrying that Merck’s management might not move quickly enough to bring the drug to more patients. “I could not sleep if the trial closed and I would have suitable patients who could respond like the others and no slots for them.”

Merck’s leadership was keenly aware of how important pembro was. Merck, which was not known for its oncology portfolio or oncology development, couldn’t find enough resources to support all of the clinical studies they wanted to do to support Keytruda. Hard decisions were made to de-prioritize many non-oncology programs in order to support Keytruda development. Roger Perlmutter, head of Merck R&D recalled, “I said to everyone, ‘Whatever other projects you’re working on, you can stop now, because we’re going to be doing this, and we’re going to put a lot of muscle behind this.’” 

In 2013, oncology products made up less than 4% of Merck’s total pharmaceutical sales. Despite this, Merck’s leadership recognized the opportunity presented by checkpoint inhibitors and poured resources into the development of pembrolizumab.

A Battle Won by the Biomarker

Though pembro was clearly efficacious in early studies with melanoma, it did not work in all cancer patients. In most colorectal cancers, for example there were virtually no responses.8  A key question in clinical development was how to identify the patients that would benefit from pembro treatment apart from all others?  One hypothesis was that tumors which expressed PD-L1, the ligand for the PD-1 receptor, were more likely to have pre-existing immune responses against them, and therefore more dependent on the PD-L1/PD-1 mechanism of immune evasion.  Though the science behind whether PD-L1 expression should predict responses was still hotly debated,9 the Merck team decided to narrow their focus to patients whose tumors expressed high levels of PD-L1, the ligand for the PD-1 receptor.

Merck’s decision to focus on patients with high levels (>50%) of the PD-L1 biomarker was a critical fork in the road.  BMS had decided to use a less stringent PD-L1 cutoff (>5%) in their development of their PD-1 inhibitor, nivolumab (Opdivo). If BMS’s drug, Opdivo, was approved with this lower cutoff, their drug would likely be more widely used.  The decisions to start these trials were made around 2013, but the answer to which camp was right became clear only last year (2017).  BMS’s PD-1 inhibitor, nivolumab, failed to show any improvement over chemotherapy in a large NSCLC study using their >5% PD-L1 cutoff.10  Pembro on the other hand, was so stunningly effective in a >50% PD-L1 NSCLC group that its trial was stopped early to let the chemo control patients take Keytruda instead.11

In NSCLC trials, nivolumab did not outperform chemotherapy in patients with PD-L1 tumor proportion score >5%, whereas pembrolizumab significantly outperformed chemotherapy in patients with PD-L1 tumor proportion score >50%. Reproduced from data from CHECKMATE-026 and KEYNOTE-024.

It was a long journey from Organon to Jimmy Carter, but all the work it took by the Merck organization to get Keytruda to the right patients paid off. Though Merck’s development team started years behind BMS’s, the aggressive push to test Keytruda clinically and in biomarker-selected patients has helped equalize the playing field between the two companies’ PD-1 inhibitors. Sales of Keytruda have nearly approached those of Opdivo today, and with hundreds of clinical trials on-going for each molecule, the number of people benefiting from PD-1 checkpoint immunotherapy can only increase. Though Merck did not start with a heavy presence in oncology in 2010, today the Jimmy Carter drug is its largest single stream of pharmaceutical revenue.

(Left) Sales of Keytruda vs. Opdivo over time – Keytruda is closing the gap. (Right) Sales of Merck pharmaceutical product lines in Q1 2018.

Lessons for Drug Hunters

First, the story of pembrolizumab is yet another example suggesting that in drug discovery, we need to let ourselves be surprised.  The original goal of a drug discovery program is often not what the resulting drug ends up being known for (the Organon PD-1 program was meant to find agonists, not antagonists; Viagra was meant to treat high blood pressure; Lyrica was originally intended to treat seizures, not pain).  Novel drugs need to be given a chance to be discovered.

Second, in the words of Warren Buffett: opportunities come rarely, so when it’s raining gold, reach for a bucket.  Roger Perlmutter is rumored to have said, “I don’t want to wake up in five years and wish I blew another $1 billion on this drug.”  When the competition is fiercest because the trophy is in sight, the whole team needs to give it everything to get across the finish line.  Despite Merck not being an oncology powerhouse, it knew it couldn’t miss on PD-1, threw everything it had into pembro’s clinical development, and executed historically complex but carefully designed clinical trials in a short amount of time.

Finally, the role of people and leadership in this story also cannot be understated.  A senior leader I respect highly recently told me, “Programs die without champions.”  Without champions like Andrea van Elsas, Antoni Ribas, and Roger Perlmutter promoting Organon’s PD-1 program from early development through commercialization, through two mergers, difficult science, and heavy competition, pembrolizumab wouldn’t be the blockbuster Keytruda is today.

Explore for more.

  1. Fife, K. M. et al., J. Clin. Oncol. 2004, 22, 1293-1300.
  2. KEYNOTE-024: N. Engl. J. Med. 2016375, 1823-1833.
  3. KEYNOTE-409 in squamous NSCLC; KEYNOTE-189 in non-squamous NSCLC: N. Engl. J. Med., 2018378, 2078-2092.
  4. BMJ 1995, 311, 899.
  5. An excellent, brief history of Keytruda’s development by David Shaywitz can be found on Forbes
  6. see for example (a) Iwai, Y., Terawaki, S., Honjo, T. Int. Immunology200417, 133-144. (b) Korman, A. J. et al. PLoS One201611, e016177.
  7. And a very helpful member of FLX Bio’s Scientific Advisory Board!
  8. Only a subset with defects in mismatch repair responded: N. Engl. J. Med. 2015372, 2509-2520
  9. Vareki, S. M.; Garrigos, C.; Duran, I. Crit. Rev. Oncol. Hematol. 2017116, 116-124 and ref. therein
  10. N. Engl. J. Med. 2017376, 2415-2426.
  11. N. Engl. J. Med. 2016375, 1823-1833.