Subscribe
Share
Share
Embed
Life sciences, under the hood. Deep dives into the patent filings, regulatory gauntlets, and capital-risks that shape the business of biology.
--------------------------------------------------------------------------------
[00:00:00] INTRODUCTION: THE BEST-SELLING CANCER DRUG OF ALL TIME
--------------------------------------------------------------------------------
ALEX: Welcome to Approved — the definitive stories behind the world's most interesting biotech ventures. Each episode, we follow companies from founding through FDA approval and commercialization, and all of the drama that happens in between. I'm Alex Kesin.
MATT: And I'm Matthew Pech.
ALEX: And today, we are tracing the story of the best-selling cancer drug of all time. In 2025 alone, this single antibody generated $31.7 billion in revenue. We're of course talking about Merck's crown jewel, pembrolizumab, also known as Keytruda. It's a molecule that has completely reshaped modern oncology treatment and, in the process, saved Merck as a company.
MATT: But past those staggering numbers is a massive blind spot. The conventional narrative completely misses the messy behind-the-scenes pivots that transformed an unloved program into a world-leading franchise.
ALEX: In fact, as one of the drug's inventors told us, there is almost nobody that actually knows the full history. So today, backed by brand-new interviews with the scientists who drove the program through the valley of death, we are tearing up the PR script to piece together the unvarnished truth.
MATT: Specifically, we sat down with three people who lived this story from the inside: Greg Carven, the Organon scientist named on the Keytruda patent; Michel Streuli, who ran Organon's Cambridge research site; and Roy Baynes, the former chief medical officer of Merck Research Laboratories, who architected Keytruda's clinical strategy.
ALEX: We'll first trace the story of this molecule from an unsuccessful antibody campaign in the Netherlands…
MATT: To barely surviving the axe across multiple organizations…
ALEX: And then to the massive all-in clinical gambles that changed cancer care forever.
MATT: And by the way, if you haven't followed Approved yet, take two seconds to do that now on Apple Podcasts or Spotify. It's the single most useful thing you could do for us, and you'll get notified every time a new episode drops. All right, let's get into it.
ALEX: Today's episode of Approved is presented by JLL. More on them later.
--------------------------------------------------------------------------------
[00:02:15] PART ONE — A CENTURY OF FAILED CANCER IMMUNOTHERAPY
--------------------------------------------------------------------------------
MATT: It's easy to forget that if you rewind the clock just twenty years ago, the idea that immune therapy in particular would become a foundational pillar across such a massive range of human cancers was completely dismissed. The field had been an absolute graveyard.
ALEX: Exactly. A hundred years of accumulated, spectacular failure had essentially pushed the entire concept of immunotherapy to the fringes of mainstream oncology. The crazy thing is, the idea that you could turn the immune system against cancer is actually a surprisingly old concept. All the way back in the 1890s, there was a New York surgeon named William Coley who noticed some of his cancer patients experienced tumor regression after getting severe post-surgical infections. He ended up developing a mixture of killed bacteria and literally injected it into his patients. To his surprise, it ended up producing real responses, but they were maddeningly inconsistent. Some tumors shrank, others didn't, and nobody — no less Coley — knew why. Coley, at the time, was practicing at a cancer research hospital, which would later become Memorial Sloan Kettering. And it was at Memorial that his work was first sidelined by the advent of radiation therapy, which the medical establishment considered way more modern and quantifiable. To be honest, it also probably helped that the hospital's chief benefactor at the time was a mine owner who supplied them with radium. The hospital's hoard of eight grams reportedly included Marie Curie's original supplies, and at the time represented most of the known radium on Earth. How cool is that?
MATT: That's crazy. So let's fast-forward a little bit to the 1940s, and Memorial has transitioned from radium to chemical poisons — otherwise known as chemotherapy. Their new medical director, Cornelius Rhoads, had served as chief of research for the Chemical Warfare Service during World War II, which, by the way, is the exact same group that discovered mustard gas's potential as a cancer-killing agent. With Rhoads in charge, Memorial had become a chemotherapy-centric hospital. When Coley's daughter tried to get Rhoads interested in her father's immune system research, he just wasn't interested.
--------------------------------------------------------------------------------
[00:04:17] T CELLS, CD28, AND THE TWO-SIGNAL MODEL OF IMMUNE ACTIVATION
--------------------------------------------------------------------------------
ALEX: But the entire modern era of this story really hinges on modulating the function of T cells — which is crazy to think about now, because there was a time when just saying the word "T cell" in front of other scientists would get you laughed out of the room. Literally. At a 1968 immunology conference, a young researcher named Jacques Miller proposed that there might be two types of lymphocytes — B cells from the bone marrow that made antibodies, and T cells from the thymus that did something different. And he was publicly reminded by the audience that B and T are just the first and last letters of the word "bullshit."
MATT: That is brutal. Poor guy. But who gets the last laugh? Because Miller was right. Over the next couple of decades, researchers ended up figuring out that T cell biology is incredibly complicated. We know now the full picture: T cells are responsible for looking out for problems inside of cells — infections, mutations, cancerous changes. Every cell displays fragments of its internal proteins on its surface via a molecule called MHC, essentially giving T cells a window into what's happening inside. T cells patrol past, reading those displays, and if they spot something wrong — a viral protein, a mutated gene — and are given the go-ahead, they will kill the cell.
ALEX: But the central question for cancer particularly was this: T cells can recognize cancer cells, so why don't they kill them? By the early 1990s, they realized that turning on and activating a T cell required multiple signals — appropriate layers of control for a cell type capable of serially killing cells. It's like starting a car. First, you have to key the ignition — the T cell receptor has to recognize the right antigen on a sick cell. Second, you need the gas pedal, which is a co-stimulatory molecule called CD28.
--------------------------------------------------------------------------------
[00:06:25] JIM ALLISON'S CTLA-4 DISCOVERY AND THE PATH TO YERVOY
--------------------------------------------------------------------------------
MATT: There was a third signal as well, perhaps the most important of them all. Jim Allison, an immunologist at UC Berkeley, worked out that a molecule called CTLA-4 acted as a brake on T cells. So CD28 was the gas and CTLA-4 was the brake. And because the brake was inherently more powerful, flooring both meant the brake won. The T cell simply would not activate. And this makes sense as a safety mechanism. It's part of a system of checks and balances designed to keep the immune system from going into overdrive and killing healthy tissue. The immune system keeps asking, "Are you sure about this?" before letting T cells turn into killing machines. But Allison made a massive leap from his discovery of this mechanism. He wanted to know if cancer is exploiting this exact brake to avoid being targeted by the immune system. What if tumor cells survive because they can trigger CTLA-4 and the T cells just stop?
ALEX: In late 1995, Allison has his postdoc, Dana Leach, run an experiment to test exactly that. They induce tumors in a group of mice, inject half of them with an antibody blocking CTLA-4, and leave the other half alone. When the results come out in November of that year, they were unmistakably clear: every mouse that got the antibody was completely cured — no tumors at all — while the entire control group grew sicker and died. Literally a 100-to-zero split. Allison's reaction to that data is legendary. Quote: "According to the data, it was a perfect experiment, 100% alive versus 100% dead. Jesus. I mean, I was expecting something, but this was 100%. Either we just cured cancer or we'd really screwed up."
MATT: Yeah, that's a great quote, and I think the theme of large effect size is gonna keep coming up over and over again in this episode. As we know today, they hadn't screwed up — the experiment was correct. But at the time, translating that observation into a drug proved extremely difficult. Allison spent years trying to get pharmaceutical companies interested, and nobody would touch it. Because in their eyes, cancer immunotherapy had a long track record of being a dead end, and curing cancer in mice had a long track record of not translating in the clinic. And on top of that, he was boxed in on IP. In the years between CTLA-4 being discovered and Allison's lab figuring out what it actually did, BMS had already filed on the molecule. Peter Linsley's group at BMS got there in June of 1991, three years ahead of Allison. Here's the catch — BMS was going in the opposite direction. They wanted to turn CTLA-4 into a fusion protein and use it as an immunosuppressant to dampen T-cell responses in transplant rejection and autoimmune disease. That fusion protein, by the way, eventually became Orencia, a multi-billion dollar rheumatoid arthritis drug. So they weren't wrong about CTLA-4 having therapeutic value — they were just going hard at the opposite indication.
ALEX: Here's the thing, though. Allison knew from his own lab that CTLA-4 was a brake, and he knew the move for cancer was to block the brake. So he filed his own patent through Berkeley on exactly that — an antibody that took the brake off, for use against tumors. Two patents, same molecule, pointing in opposite directions. On the cancer story specifically, Allison had it right and BMS was backwards. And this is a pattern to file away because it's going to keep showing up: BMS keeps arriving first and making important discoveries in immunotherapy, and then at critical juncture points, making the wrong call about what to do with them.
MATT: But back to Allison. As we said before, he couldn't get big pharma to take a meeting with him. Between the BMS overhang and the field's long history of cancer-immunotherapy failures, every BD team politely passed on him. Berkeley ended up licensing the patent out to a small biotech called NeXstar in 1995, which itself got rolled up into Gilead, which sub-licensed it to a company called Medarex in 1999. Medarex used its HuMAb-Mouse platform to build the fully human anti-CTLA-4 antibody that became MDX-010 — later called ipilimumab. And by 2009, BMS buys Medarex for $2.4 billion. And in 2010, at a plenary session of the American Society of Clinical Oncology, they presented the data for ipilimumab — brand name Yervoy — in metastatic melanoma, which at the time was a cancer that was terminal in most patients. And the readout was striking. The median overall survival jumped from about six and a half months to ten.
ALEX: While an extra three or four months of overall survival might not sound that massive, the truly crazy part was the tail of the Kaplan-Meier curve. About 20% of patients treated with Yervoy were still alive after two years. That level of durability was just unheard of. Yervoy was approved in March 2011. It was a landmark — the first cancer drug that worked purely by releasing the immune system's brake. But it had severe limitations. Blocking CTLA-4 unleashed T cells everywhere in the body, not just at the tumor site. CTLA-4 operates upstream at the priming stage in the lymph nodes. Block it, and you get T cells attacking healthy organs — colitis, dermatitis, hypophysitis, a whole constellation of autoimmune-like toxicities. And while some patients responded dramatically, many did not.
--------------------------------------------------------------------------------
[00:12:03] TASUKU HONJO DISCOVERS PD-1: A BETTER BRAKE ON T CELLS
--------------------------------------------------------------------------------
MATT: There was, however, another brake — and it worked in a fundamentally different place. In 1991, a Japanese researcher named Tasuku Honjo and his lab discovered a gene they named PD-1, which stands for programmed death one.
ALEX: Which, we should probably mention, is the very gene featured on these limited-edition Uniqlo shirts we're both wearing today in Honjo's honor. As the viewer may guess, we are both big nerds.
MATT: Yes, very geeky on our part, but absolutely necessary.
ALEX: Very necessary. But here's the thing, Matthew. The name, funnily enough, is completely misleading. It's called this because initially, when it was identified in Honjo's lab, it was discovered in the context of programmed cell death — that's not actually how it works primarily. PD-1 is just another inhibitory receptor on T cells, just like CTLA-4, operating through a distinct mechanism. To get a real sense of why this was such a big deal, let's take a quick primer on how T cells work — specifically the difference between these two brakes, CTLA-4 and PD-1. CTLA-4 acts super early, back at the T cell's priming stage in the lymph nodes. You could call it the training camp for T cells. This makes blocking CTLA-4 like opening the gates to the barracks and letting every single T cell foot soldier run wild all over the body. You do get an immune response against cancer, but also a ton of collateral damage because they're attacking everywhere else too. But PD-1 acts late, right at the killing stage. Imagine a fully trained T cell has marched all the way to the tumor and has found a target in its sights. But right at the moment of contact, the tumor flashes a fake ID — a molecular badge called PD-L1. That badge binds to the T cell's PD-1 off switch and essentially says, "Hold your fire. Let's all calm down for a minute."
MATT: The simplistic story is that PD-1's only job is to wake up exhausted T cells that are present in the tumor site. As the field has generated more data on this subject, it's looking a lot more complex, and I think PD-1 blockade also seems to play a role in the initial priming and expansion phase of newly created T-cell clones.
--------------------------------------------------------------------------------
[00:14:26] LIEPING CHEN AND THE PD-L1 TUMOR EVASION HYPOTHESIS
--------------------------------------------------------------------------------
ALEX: I find it amazing how they figured that out — how they even start to differentiate the roles of CTLA-4 and PD-1.
MATT: Yeah, that's a good question. So in our story, we'll center some of the initial critical work that was done by Lieping Chen, who was working at the Mayo Clinic in the late 1990s. Chen was looking at the evolution of these tumors through a pretty Darwinian lens. He had a hunch that immune pressure early in tumor growth mainly selects for clones that are capable of immune evasion. Just like mutations can give a tumor a growth advantage, overexpressing an immunosuppressive protein like PD-L1 could potentially help it survive under immune pressure. To test this hypothesis, Chen and his team developed monoclonal antibodies to stain and measure PD-L1 levels in both healthy and cancerous tissues. And in 1997, they discovered that a ton of cancer tissues overexpress PD-L1, while normal, healthy tissues barely seem to express it at all. The exact detection method would eventually become the clinical standard for the predictive biomarker — a detail we'll revisit later in the story. The therapeutic implication of this was enormous. If tumors were indeed using PD-L1 to shut down T cells right at that moment of attack, then you could envision that blocking the interaction should unleash a more targeted immune response right at the tumor with a lot less collateral damage than via CTLA-4 blockade.
ALEX: Interesting. If I remember right, Chen also had problems with out-licensing the IP for this to pharma companies, because again, this time — early 2000s — was just really rough for immuno-oncology. It was really only around 2004 that he found a willing partner, and none other than what company, Matt?
MATT: Medarex once again.
ALEX: Yep. You'll hear that name again and again in this story.
--------------------------------------------------------------------------------
[00:16:26] PART TWO — ORGANON: THE DUTCH PAINT-COMPANY SUBSIDIARY BEHIND KEYTRUDA
--------------------------------------------------------------------------------
MATT: That brings us to an incredibly unlikely birthplace for Keytruda. Not at an elite oncology lab, but at a Dutch pharmaceutical outpost called Organon, not particularly interested in cancer.
ALEX: Alright, to set the scene a bit more — Organon is headquartered in the Netherlands. And at the time they were best known commercially for making contraceptives — Nexplanon in particular. To make things even weirder, Organon was a subsidiary of another company called AkzoNobel, a massive global conglomerate that was famous for making house paint. Crazy.
MATT: I mean, when you think of these sort of conglomerates, you think of the Japanese conglomerates or the South Korean ones, but here's the European equivalent of that — paint and contraceptives. Alright, so you have this paint company starting to make forays into antibodies. And the therapeutic research profile was literally the last place anyone would've looked for the next generation of cancer drugs. To understand what happens next, it really helps to get a feel for the sort of pressure Organon was under at that time. Greg Carven — ex-Organon employee, named inventor on the Keytruda patent, and a scientist who graciously sat down with Alex and I to tell us about his experience there — he told us that like a lot of pharma companies in the mid-2000s, Organon was staring down pretty major loss-of-exclusivity events on their legacy portfolio of drugs. Very standard situation. The patent cliff was upcoming and the company needed to pivot and refocus. Their solution was to innovate their way out of it. In 2005, they plant a flag right in Cambridge, Massachusetts, and launch a brand-new 30-person research site. To be clear, despite the deceptively small size of the team and the location far away from the mothership, this was a massive, high-stakes corporate priority for the company. You had Dutch executives constantly flying over the Atlantic to monitor this team because they were essentially the company's survival strategy. The mandate was very specific: build a new pipeline of biologics and autoimmunity drugs. An initial priority was — make an antibody to activate PD-1 as a treatment for rheumatoid arthritis.
ALEX: Wait, what? Activate PD-1?
MATT: That's right. The target indication wasn't cancer at all — it was rheumatoid arthritis. And the logic for picking this indication was pretty clean. Remember, BMS did this too with CTLA-4. If PD-1 acts as a brake on T cells, then developing an agonist that mimics the brake should dampen down an overactive immune system. And there is human genetic data to back up the logic too. In all fairness, certain people have a germline mutation in the PD-1 gene that predisposes them to develop RA. Back at the headquarters in the Netherlands, the initial antibody campaigns to make those PD-1 agonists had already been kicked off, and an immunologist named Hans van Eenennaam and his team had actually produced an antibody that looked like it did exactly that. In the early phases of their screening funnel — a plate-based screening assay — it looked like they were closing in on an agonist.
--------------------------------------------------------------------------------
[00:19:48] HOW MICHEL STREULI CAUGHT THE SOLID-PHASE SCREENING ARTIFACT
--------------------------------------------------------------------------------
ALEX: But here's where the sequence of events takes a turn. The Cambridge site gets a new head of research — Michel Streuli, who we had the pleasure of speaking with in addition to Greg. And Streuli is crucial here because he wasn't just an autoimmunity guy. He had spent the last 15 years working specifically on tumor immunology over at Dana-Farber. And when Streuli, and the new scientists like Greg, started looking at Hans's initial hits, they immediately want to stress-test them. The Dutch team had been relying on solid-phase plate screening, but Streuli knew from his Dana-Farber days that you have to be incredibly careful about how you analyze antibody activity. So the Cambridge team moves the antibodies off the engineered plates and tests them in more physiologically relevant conditions — like liquid-phase immunoassays. And almost immediately they realized the plate assay was giving HQ a completely false artifact.
MATT: Exactly. Streuli described that moment of realization to us.
MICHEL STREULI [interview clip]: "Hans and his team had developed an antibody that looked like an agonist. They were using a solid-phase screening assay. And so when I joined and saw the data, I said, well, show me that it works in solution — because, you know, I had spent 15 years at the Dana-Farber. You had to be careful about how you analyze activity of antibodies. And sure enough, when sort of the assays were repeating the stimulation assays, it turned out that in solution the antibody was an antagonist."
MATT: It's such a massive catch — because tethering antibodies to a solid surface can accidentally crosslink a receptor, making an antagonistic antibody read out like an agonist. Once it was in solution, the antibodies that were supposed to calm T cells were actually unleashing them. So while they were looking for a rheumatoid arthritis drug, they were finding the exact opposite. When we asked Carven how a screen built to find an agonist ends up producing the complete reverse, here's what he said.
GREG CARVEN [interview clip]: "I think because we were looking for agonists — because we were looking for things that could shut off the immune cells — we ended up finding things that looked a whole lot like PD-L1 in terms of the epitope and whatever. And that had all sorts of, like, millions of years of evolution to pick that particular epitope. Blocking that specific epitope ended up being maybe the best possible way to do that for an antagonist."
--------------------------------------------------------------------------------
[00:22:09] THE ACCIDENTAL ANTAGONIST: FROM RHEUMATOID ARTHRITIS DRUG TO CANCER DRUG
--------------------------------------------------------------------------------
ALEX: So because they were trying so hard to mimic the body's natural brake pedal, they accidentally built the perfect wedge to jam it open. And this is just wild to think about — this is the third episode in a row that we've seen scientists start off aiming for one effect, but end up working on a drug embodiment that works in the opposite direction of that target. First Amylin, then Geron, and now Organon. Pretty cool.
MATT: Yep. There's a lot of serendipity in drug discovery, that's for sure. And, you know, to be fair to the Organon team, Carven and others had anticipated that their screening campaign may produce a whole bunch of antagonists as they were running their antibody funnel. He had written up an internal contingency memo on what the program should potentially do if what they got out are antagonists. While he was writing that memo, that seemed like a far-off possibility. Now they knew what they actually had from this campaign. After screening thousands of clones, they got exquisitely potent antagonists and zero agonists. Remember here, they're an autoimmunity team. They were set out to find a drug for RA, and at almost any other large pharma company, this would've been the end of the road for the program.
ALEX: A playbook that's gonna come up with Keytruda is how many times this molecule came close to dying — and this is instance number one. Right? Like, if this campaign had been run in a traditional pharma organization that's organized around the standard therapeutic areas, you would be in a situation where the immunology team has been trying to get agonists, they don't get it, and they would've just thrown everything in the garbage can. And that would've been the end of it. The pivot to oncology would've been a thousand times harder. But the Organon site was built differently and organized differently. Remember that its mandate wasn't strictly to focus on autoimmunity — it was built as a Center of Excellence for Biologics. So it had this focus on antibody technology, and when the site proposed to pivot the program towards oncology, the site head — a man named Vibo Olijve — didn't shut it down. He deferred to the judgment of the scientists.
MATT: And who are those scientists? Well, it ended up being two critical people that advocated for the pivot. First, you have someone we already introduced — Michel Streuli at the Cambridge site. And the second was someone named Andrea van Elsas. And Andrea was an immunology leader back in the Netherlands who had actually championed the PD-1 program from the get-go. As it turned out, Andrea had the perfect background to recognize the massive potential of what they had accidentally discovered. Before he joined Organon, Andrea had actually done his postdoctoral training in…
ALEX: Jim Allison's lab!
MATT: Exactly. You know, the Nobel laureate who pioneered the entire concept of checkpoint inhibitors for tumor immunology.
ALEX: That's right. So when the Cambridge team finds these antagonists, Andrea immediately recognizes the potential, and there's this scientific weight behind the pivot.
MATT: Yeah. So with Andrea and Streuli feeling strongly about this sort of pivot, the Cambridge site head — Vibo — feels completely comfortable greenlighting the pivot to oncology. Streuli actually told us about this exact dynamic.
MICHEL STREULI [interview clip]: "You know, I still remember quite clearly the head of the site, Vibo Olijve, who was a fantastic site head, very charismatic and talented guy, you know, asked me — did I think we should continue developing it as an antagonist? Because it went into oncology, and Organon had pretty a strong mandate, pre-2006, to focus on RA, and that's why Hans was focusing on developing an agonist, and there was very good literature to support that. But my background was actually tumor immunology, so I was excited at the opportunity to developing an antagonist."
ALEX: Right as they get going, that theme of companies coming under stress from loss of exclusivity comes roaring back — because Organon's corporate situation is about to get incredibly messy.
--------------------------------------------------------------------------------
[00:26:34] SPONSOR: GRANT DETMER (JLL) ON BIOTECH REAL ESTATE STRATEGY
--------------------------------------------------------------------------------
ALEX: Alright, with Organon starting to build out a Massachusetts lab presence, now is the perfect time to introduce the sponsor for today — Grant Detmer at JLL. JLL is the largest global commercial real estate firm in the world. Grant Detmer and his team lead the company's West Coast biotech organization, specializing in helping biotech companies on all things real estate, both locally in the Bay Area as well as nationally and globally. Grant, some questions for you. When should a biotech company actually start thinking about real estate?
GRANT DETMER (JLL): Yeah, it's a really good question, and the answer is — earlier than you would think, but not necessarily in the way people would assume. You don't need to sign a lease early, but you want to have a strategy early. I tell founders, start thinking about real estate when you start thinking about your next scientific milestone, because your real estate should map your inflection points. IND filings, clinical phases, team growth. If you align those correctly, you avoid overcommitting capital and you maintain flexibility. The best-run companies treat real estate like a capital allocation decision, not just an operational one. And a general rule of thumb — for earlier-stage seed/Series A companies, you want to be at least nine to twelve months out from a need. And then for larger, more developed companies, twelve to twenty-four months out from a need to start planning.
ALEX: Thanks, Grant. How then should founders think about balancing flexibility versus cost?
GRANT DETMER (JLL): That's a really good question. There's a natural temptation to optimize for cost, especially in tougher fundraising environments. But in biotech, flexibility is often more valuable than saving a few dollars a square foot. If your science hits, you'll need to scale fast. If it doesn't, you'll need to pivot or shrink. A rigid lease can hurt you in both scenarios. So the question isn't "what's the cheapest deal?" It's "what structure gives us the most options over the next 24 to 36 months?" Sometimes that means shorter terms, expansion rights, or even paying a bit more upfront for flexibility. And in biotech, optionality is everything. That's some of the questions that we like to put in place when we're going through a process like this.
ALEX: Totally agree with you, Grant. I think we've all seen financing events or other aspects of a company's growth completely killed by having the wrong sort of real estate obligation on the books. What advice would you give to a first-time founder going into their first lab lease?
GRANT DETMER (JLL): Then — another great question. First piece of advice: don't go at it alone. Lab leases are complicated, and landlords do this for a living. Having someone who understands the market, the economics, and everything on the technical side can save you a meaningful amount of money, and more importantly, help you avoid bad decisions or roadblocks along the way. Secondly, I would focus on downside protection, not just the plan-A. Everyone signs a lease assuming everything goes right. The smart founders ask, "What happens if things don't go right? What happens if we need less space, more space, or need to exit?" If you answer those questions upfront, you're in a much stronger position.
ALEX: Thanks, Grant. At the end of the day, real estate won't make your science work, but it can absolutely impact whether your company has the time and capital to get there. Plan carefully and cautiously. Thank you again to JLL and Grant Detmer for sponsoring this episode. Now back to the story.
--------------------------------------------------------------------------------
[00:30:01] RUSSIAN NESTING-DOLL M&A: SCHERING-PLOUGH ACQUIRES ORGANON (2007)
--------------------------------------------------------------------------------
ALEX: So let's talk about how messy this corporate situation gets. In 2007, Schering-Plough acquires Organon for $14.4 billion in order to enhance their portfolio of late-stage clinical assets. Again, we're not gonna go into all the details here, but this is another pharmaceutical organization trying to optimally position themselves for loss of exclusivity in their pipeline. According to Greg Carven, this initial Schering-Plough acquisition wasn't too disruptive to the Cambridge team — they were basically left alone to operate as a satellite site. But when two pharmaceutical companies come together, one of the things that they do is something called a pipeline prioritization exercise, in which the new combined company decides which programs to keep and which ones to axe. And when these two companies combined their oncology portfolios, the PD-1 program did not exactly get a warm welcome.
MATT: Yeah, this is what Greg told us specifically about that time.
GREG CARVEN [interview clip]: "In that prioritization of all of the programs across Schering-Plough oncology, PD-1 was prioritized dead last. It was not killed — there were programs that were killed — but of the programs that were continuing, it was prioritized at the very, very bottom of the combined portfolio."
--------------------------------------------------------------------------------
[00:31:43] THE $41 BILLION MERCK–SCHERING-PLOUGH MEGA-MERGER OF 2009
--------------------------------------------------------------------------------
ALEX: So the molecule that will become Keytruda is barely hanging on in the new organization. And if that's bad, things are about to get a lot more complicated, because two years later the M&A cycle happens again. It's 2009. Merck and Schering-Plough agree to a massive $41 billion mega-merger.
MATT: Wow, that's huge.
ALEX: Yeah, absolutely huge merger of two 25,000-person organizations. And I think what's so funny here is, like, technically Schering-Plough acquires Merck, not vice versa.
MATT: Really?
ALEX: Yeah. So it was structured as a reverse merger, specifically to avoid triggering a change-in-control clause. Essentially, Schering had a blockbuster drug called Remicade, which they had co-developed with J&J, and they were trying to get cutesy and avoid having to give the drug back to J&J. As you can imagine, this sort of legal cleverness led to all sorts of lawsuits that took a long time to work themselves out. In any event, Schering-Plough and Merck are merging. They are slamming together two massive 25,000-person organizations. These are two companies each that had a good amount of overlap between the two pipelines. So when we talk about a pipeline prioritization exercise, leadership is not just thinking about, "Hey, what programs should we pursue?" They're also thinking, "Hey, we have two duplicative programs. Which of the two should we keep, and which one should we kill?"
MATT: Just to make sure I'm not getting too lost in the weeds here — why exactly did this merger happen between Merck and Schering to begin with?
ALEX: Great question. At this time, Merck was a cardiovascular- and respiratory-focused company. Their big blockbuster drug was an asthma drug called Singulair, and they were about to lose market exclusivity. They needed to diversify their portfolio to survive. To be clear, whenever a massive merger like this happens in pharma, early-stage programs get killed all the time. Budgets shift, priorities change. If a molecule doesn't have a really fierce internal champion, it often ends up shelved or on the chopping block. The PD-1 antibody, which has now been named MK-3475, was buried so deep in this list of acquired assets that it was practically invisible. Here's what Greg had to say about going through a second pipeline prioritization exercise in the span of two years.
GREG CARVEN [interview clip]: "We had to start the whole thing all over again. And there was a year-long wait from the announcement of the Schering acquisition to day one. While things were still moving forward, we actually could write the IND and the investigator brochure and all of the parts that go along with it, but we kind of had to wait for Merck to make up its mind on, like, alright, now they've gotta prioritize all of the programs again. And went through the program priority — because by that point, Merck and Schering looked a lot more similar than Organon and Schering did. And there were two of the HGF programs and two IGF programs and two PCSK9s and right — so there was a lot of, which programs do we do at all? And then for the ones that we have two of, which one do we keep? And that took a really, really long time to work its way through. And politically, I think Merck disliked immune oncology even more than Schering, if that was possible."
MATT: Yeah, it's insane to think about that now — this program, which had gone through, like, basically a Russian nesting doll of M&A. Right? First acquired by Schering and then by Merck in such little time, and now is on the chopping block. Not dead yet, but pretty close.
--------------------------------------------------------------------------------
[00:34:58] CORPORATE GUERRILLA WARFARE: FOUR SCRAPPY STUNTS THAT SAVED PEMBROLIZUMAB
--------------------------------------------------------------------------------
ALEX: Yeah, it was incredibly close to dying. But Michel Streuli and Greg Carven and others really refuse to let it die. The team starts to run really a masterclass in sort of corporate guerrilla warfare, right? So they are trying to keep this program alive and scrape together the resources to be able to file an IND, and they really pull off a rapid-fire series of kind of insane and super creative and inspiring stunts. So let's just rattle some of those off.
MATT: Yeah, let's definitely talk about these. So stunt number one goes something like this. Imagine you're Greg, or you're Michel — your own oncology department kills your cancer drug. Who's left to pitch it to? Well, the two of them ended up driving down to New Jersey from Cambridge to pitch it to the Infectious Disease unit of Merck. Here's what Streuli had to say about how they pulled this off.
MICHEL STREULI [interview clip]: "But I remember going down with Greg Carven to New Jersey to speak with the head of the infectious disease unit there, Daria Hazuda, and she was very gracious because she gave us the go-ahead to keep continuing developing the PD-1 antibody. In part because Merck had a legacy PD-1 program in the infectious-disease program, particularly for things like HIV cure — there was real interest. So they had a scientific rationale to moving that forward."
ALEX: It's really kind of a brilliant maneuver. So they pitch it as a potential HIV cure, and essentially gave them just enough bureaucratic air cover to keep the lights on. Unfortunately, just 'cause the lights are on doesn't mean you have the budget to do anything. So enter stunt number two: clinical planning. As a zero-priority program, the internal clinical organization wouldn't give them the time of day. So they literally went off the books.
GREG CARVEN [interview clip]: "Being under the radar, being at the satellite site, we actually used sort of non-traditional resources. Schering-Plough had a large clinical organization, they couldn't give us the time of day. We hired our own external consulting clinicians to help us build a, uh, early clinical plan."
MATT: Yeah, it's kind of crazy to think how they were operating as a fiefdom within Merck. Perhaps one of the most consequential stunts for the later development of Keytruda was stunt number three: getting the actual clinical material manufactured. And this one's really funny. The manufacturing group within Merck wanted to test out a new flow cytometry sorting technique, but they weren't allowed to experiment on a high-priority program. So who's there to be the guinea pig? The PD-1 team. And they volunteered their "worthless" molecule as the test dummy for this.
GREG CARVEN [interview clip]: "We needed to make material and we needed to test that in GLP studies. And the ways to get that done more efficiently were to do that via some technology development work that the manufacturing group had wanted to do. They wanted to sort cells for higher expression, I think, using flow cytometry, and they wouldn't have had the opportunity to do that on a high-priority program. But hey, we can absolutely do this on a low-priority program, sure, right? And then all of a sudden we have a manufacturing cell line that's made clinical material, and it's like, this is great."
ALEX: And finally, stunt number four: non-clinical tox studies. One of the more expensive parts of preclinical development is making sure that your molecules are safe. And for antibodies, this is typically done in monkeys. Right around this time, the industry was completely freaked out by the TeGenero disaster, where a trial drug caused catastrophic immune activation in healthy volunteers. That trial tested an antibody designed to activate CD28, which unexpectedly triggered a massive, life-threatening cytokine storm in six healthy participants within hours of the very first dose.
MATT: Yeah. Scary stuff. So Schering and Merck more broadly had to put together a special task force to figure out predicting immunotoxicity in monkeys. And naturally — what program better suited than PD-1 to jump on the opportunity to get more free work done?
GREG CARVEN [interview clip]: "They were looking, actually, for programs to test in monkeys and to look at ways to do this. And timing was perfect. It's like, here you go — here's a program where we can go and run an immunotox program and show that antagonizing an inhibitory checkpoint is different than agonizing and activating a checkpoint. And built out a safety package that ultimately was supportive of the IND."
--------------------------------------------------------------------------------
[00:39:32] BIONOVION'S SPITE-COMPANY BID TO BUY PEMBRO BACK
--------------------------------------------------------------------------------
ALEX: Yep. Peak scrappiness. But even as all aspects of the IND filing were completed and the IND was ready to go, the drug was officially put out on the out-licensing list — to be sold off for parts. We sound like a dead record around the program being dead, but at this point the program is officially dead, dead, dead. And here is the most incredible, ironic part of the whole story — with the program for sale, who came up to offer to buy this program?
MATT: Andrea van Elsas.
ALEX: Insane. So remember, this was the same guy who originally championed the program at Organon. It turned out he had gotten laid off during the Merck merger, started his own company called BioNovion with other ex-Organon people — so Hans and Vibo. Oh my God. And they just try to buy Keytruda back from the company that made it.
MATT: You can't make this stuff up if you tried. Sometimes reality is really stranger than fiction. And this is the second time in a row we've had, like, a spite company being created.
ALEX: Right. Yeah. I mean, it's just — I think at the end of the day, scientists who work deeply on a program have a very deep understanding of the science, and they can often see potential where sort of other folks cannot. So, you know, it's completely understandable that they made this shot at trying to get their baby back.
MATT: Agreed, agreed. Michel Streuli, who was still again within Merck at the time, told us what this dynamic was like — seeing on one hand his former compatriots at Organon at this new company trying to acquire the asset, and his own very strong advocacy for the molecule being developed within Merck. Here's what he had to say about that.
MICHEL STREULI [interview clip]: "No, it was rather awkward, 'cause I was very fond of Andrea and Hans and Vibo, who started BioNovion. But I felt responsible to my employer to keep the program at Merck at that time. But I think drug development needs — particularly when you're in the clinic and then to commercialize, you know — you need deep pockets and a lot of expertise, and Merck had that."
ALEX: What a crazy position to be in — basically having to play corporate defense against the very people who built the drug from scratch, just to make sure it survived because Merck had the pockets. So Streuli and the holdovers win this tug-of-war.
MATT: So how exactly does Pembro go from being on the chopping block to being a high priority for Merck and moving back even into the land of oncology?
--------------------------------------------------------------------------------
[00:42:10] BMS AT ASCO 2010: THE DATA PRINT THAT REVIVED MERCK'S PD-1 PROGRAM
--------------------------------------------------------------------------------
ALEX: Well, it involves none other than BMS. The summer of 2010, BMS shows up at ASCO with a clinical data print for their anti-PD-1 antibody, nivolumab. This molecule, as it turns out, came from none other than Medarex. As we know, BMS acquired Medarex — it was just another rat in the bow of the ship until now. This was the data print heard round the world. At ASCO, they presented data on 25 to 30% response rates in melanoma and kidney cancer — numbers just completely unheard of in the pre-checkpoint era. Suddenly, Merck's leadership looks at the PD-1 program in a completely different light. Peter Kim, who was running Merck Research Labs at the time, picks up the phone and calls Streuli.
MICHEL STREULI [interview clip]: "I think in 2010, I got a call from Peter Kim, asking me to sort of lead the PD-1 program for oncology. And that was really on the heels of the Medarex BMS data being presented at ASCO. It was quite strong data, and people got excited. And so I think they realized, ah, we have this PD-1 program. And so at that point, I became head of the program and was able to get it into the clinic, and then starting to see the first patients dosed, which was pretty amazing."
MATT: So the molecule miraculously survives the mergers, but it's incredibly far behind BMS. We're talking years behind here.
ALEX: So yeah, at this point, the Merck PD-1 program is absolutely years and years behind the BMS program. And again, ironically, this wasn't because the drug discovery program was moving particularly slowly — I think they moved quite quickly from target to development candidate, all things considered. It was just tremendously slowed down from all of these organizational issues.
MATT: Right. You know, why does it matter that this program is far behind BMS? There are situations in drug development where being second to market isn't just like a small disadvantage — it can fundamentally change the math of your entire clinical strategy, right? When you are first to market, to get regulatory approval, you have to prove that your drug is better than whatever the old standard of care was. So in the case of nivo, it would be proving you're better than some sort of chemotherapy regimen that didn't work particularly well. However, if you're second to market, your regulatory and commercial bar gets quite a bit higher. You actually have to prove that you're better than the new competing molecule. And this is a much harder, bigger, riskier, more expensive clinical trial to run. And then even if it is successful, you're now in a position of facing the commercial hurdle of convincing physicians to start changing their newly-formed prescribing habits. So Merck is sitting there staring down this massive chronological disadvantage, trying to figure out if it's even worth the fight.
ALEX: And after years of delay and bureaucratic fighting, they finally managed to get the trial off the ground. In 2011, they initiate a small 30-patient phase one study focused on melanoma. They call it KEYNOTE-001. It's a start, but at this exact moment, a 30-patient safety trial is not gonna catch up with BMS. To actually close that multi-year gap, they needed a leader willing to take a massive, unprecedented bet on this molecule. Which brings us to April 2013. There's a new sheriff in town at Merck Research Labs — Roger Perlmutter — and he's going to change the fate of this program forever.
MATT: Yeah. Here's where the pedal hits the metal for Keytruda.
--------------------------------------------------------------------------------
[00:45:55] PART THREE — ROGER PERLMUTTER JOINS A BLEEDING MERCK (APRIL 2013)
--------------------------------------------------------------------------------
ALEX: To understand what Roger is walking into in April 2013, you have to know just how badly Merck is bleeding. Remember the Singulair patent cliff we mentioned earlier? Well, at this point, it's not a future problem anymore. Merck is in full crisis mode. Singulair peaked at over $5 billion in annual sales in 2011. Then its patent expired. By the first quarter of 2013, worldwide sales had collapsed 75% year-over-year. One of the most profitable drugs in pharmaceutical history, just evaporating off the P&L with no cushioning. And the pipeline that's supposed to replace it — it's blowing up in real time. In February of that same year, just a few months before Perlmutter joins, Merck is forced to delay its osteoporosis drug, odanacatib. Then the FDA rejects their insomnia drug, suvorexant, over safety concerns about the recommended dose. If that wasn't bad enough, the FDA hands them a CRL for sugammadex, their anesthesia reversal drug — which by the way came from Organon. These were all supposed to be the post-Singulair lifelines, but now they're all stumbling to the finish line at the exact same moment. The analyst community was not pulling any punches about Merck's state of affairs as a company. Bernard Munos, one of the most respected voices on pharma R&D productivity, basically said Merck's innovation problem was, quote, "self-inflicted" — that a stifling process culture had smothered what had been for decades the most innovative company in the industry.
MATT: And so CEO Ken Frazier himself, promoted to the post around 2011, decides that Merck Research Labs needs some fresh blood. Peter Kim, who had run MRL since 2003, is out. The person Frazier brings in to replace him — in what is effectively a turnaround hire for the entire organization — is a guy named Roger Perlmutter, our man of the episode. The first thing to know about Perlmutter is that he's an immunologist by training. He founded the Department of Immunology at the University of Washington, and then he ran Amgen's research organization for a decade. One of the key people he pulls with him to Merck is Roy Baynes, who had worked under him at Amgen. When Perlmutter calls him, the pitch was basically this.
ROY BAYNES [interview clip]: "Also, he gave me this somewhat amazing exhortation. He said, come and help Merck — a once iconic biopharmaceutical company — to its former greatness and rightful place in history."
ALEX: It's such a great detail. Because around this time in 2013, Merck's research organization was struggling a lot. Baynes told us that it was heavily matrixed with a consensus-driven culture, where big bets simply couldn't be made because no single person had enough authority to pull the trigger. And it wasn't just issues with decision-making. He also told us that they had completely lost their ability to execute clinical trials because they were outsourcing everything.
ROY BAYNES [interview clip]: "There'd been tremendous loss of operational competence. You know, Merck was, in its heyday, probably the best clinical operations entity in the business. And essentially, again, consultant-led outsourcing had led them to completely lose their way in terms of how to actually operate."
MATT: So when Perlmutter takes over, he inherits this paralyzed, outsourced organization. He knows almost nothing about the PD-1 program beyond what was publicly available. And remember, BMS already has Opdivo — nivolumab — further along in development. By any reasonable estimate, BMS at this point has a two-year head start. But within mere months of arriving, Perlmutter looks at the early pembrolizumab data from KEYNOTE-001 and he's completely stunned. Baynes told us that the phones were literally ringing off the hook with responders. These were the types of patients who had failed literally every other therapy in line. And these patients in the study saw their tumors shrinking. It wasn't just melanoma patients that saw these responses, which was expected. Remember, Yervoy treated melanoma and was approved for it a mere two years before — it was one of the easier target diseases for immunotherapy, since melanoma cancer cells are highly mutated and highly immunogenic. The real surprise was seeing responses in lung cancer, a much harder disease to target. That meant PD-1 was not just a melanoma drug. Here's what Baynes remembers about those initial trial numbers.
ROY BAYNES [interview clip]: "It was very clear, literally from 001, that this was gonna be a remarkable drug. I wasn't there when 001 started, but I'm told, you know, they studied this in patients, and the first group of patients were melanoma patients. At first imaging, the phone was ringing off the hook with responders, and these were patients who had failed everything. So these were salvage patients. So it was clear that this was an inordinately apt drug in immunologically sensitive malignancies. The lung cancer data takes a group where indeed not quite as immunologically active, but the results were also knock your socks off. So it was clear this was gonna be an important drug."
--------------------------------------------------------------------------------
[00:50:48] "LET ME MANAGE THE TIGERS": KEN FRAZIER BACKS THE ALL-IN BET ON PEMBROLIZUMAB
--------------------------------------------------------------------------------
ALEX: So as you can imagine, there was a lot of excitement in the company. But Perlmutter then makes a decision that defines the next decade of Merck. He tells his organization to stop what they're working on and focus entirely on pembrolizumab. He was quoted in the Wall Street Journal saying, quote: "Whatever the other projects you're working on, you can stop now, because we're gonna be doing this and we're gonna put a lot of muscle behind this."
MATT: "Put a lot of muscle" indeed. Ultimately, more than half of Merck's entire drug-development budget gets redirected towards pembrolizumab, ultimately fueling over 700 clinical trials across dozens of cancer types. Unsurprisingly, analysts were quite concerned initially. Merck was betting a great deal of their future on the fate of a single drug. And, you know, I think that concern is reasonable. It's this exact same sort of question that people ask today about Novo and their weight loss drugs, or other pharmaceutical companies that are pretty reliant on a single drug line. What happens when your whole company is a single molecule?
ALEX: The best counter-argument to this would be something like — the business of pharma is fundamentally a hits-driven business. Only a small set of blockbuster molecules are gonna be driving most of the returns in even the biggest companies. It's kind of the name of the game that pharmas are typically reliant on only a handful of molecules at any given moment in time. Basically, the Keytruda/Merck story isn't really an anomaly, but par for course in this narrative.
MATT: So redirecting half of a massive company's R&D budget is super risky. The only reason Perlmutter and Baynes survived the internal politics of that decision was ultimately because Ken Frazier acted as their internal champion. Baynes told us that Frazier made them a very specific promise, which was this.
ROY BAYNES [interview clip]: "So Ken had a tremendous amount of trust in Roger, and Roy Vagelos, who, you know, had been a previous head at Merck, also had a lot of confidence in Roger. So, and the company was, you know, in really a tough spot. It was a tough spot for many reasons. You know, they'd been through the Vioxx issue, they had been through the issues related to Fosamax — so, again, there wasn't a whole lot obvious it was gonna save the company. You may remember the Schering-Plough acquisition really went after a lot of cardiovascular assets, most of which didn't pan out. And really this little sleeper, you know, was there in plain sight, and it was gonna, you know, really turn the company around. Ken, I think, bought into that idea really quickly. Pretty much was completely aligned all the time. Ken's statement always to me was, let me manage the tigers and give you guys peace to get your work done. And that, that's how he approached it."
ALEX: "Let me manage the tigers." That's a great line from a CEO. Definitely one of the jobs of a pharma executive is to give the research team enough time to execute while Wall Street is desperate for the next thing.
MATT: Yeah. And thank God he did, because Perlmutter fully acknowledged the risk that they were taking. He later said in an interview: "This is a very easy business to make mistakes in. I do feel we have a responsibility with Keytruda to ensure that we fully demonstrate what this drug can do."
ALEX: Yeah, I think looking back on it years later, even Perlmutter was incredibly candid about how crazy this whole bet seemed in retrospect. He said, quote: "If you had come to me 10 years ago and said, do you realize that we can block a single receptor-ligand pair and reveal preexisting immunity in a substantial fraction of cancer patients, causing their tumors to disappear? I would've said, you're out of your mind."
MATT: And, you know, this wasn't just hindsight by an executive. The original bench scientists who were actually discovering and building this molecule felt the exact same way too. Greg Carven told us that really nobody saw this coming.
GREG CARVEN [interview clip]: "I'll absolutely say nobody predicted it would work as well as it does. And anybody who says otherwise is lying. We thought maybe it would be second or third line in a small subset of immunogenic tumor settings — melanoma, maybe renal or liver. I don't think anybody would've ever said, hey, this will get into whatever it is in 30 different labels, in like 18 different tumor types."
ALEX: Yeah, what a wild retrospective. "Out of your mind" is certainly a way to phrase it. But Perlmutter made the bet anyway. He took a hollowed-out company staring down a catastrophic patent cliff, pushed half of their R&D budget into a drug that had literally been on the out-licensing clearance rack, and completely changed the history of medicine. Got to give him a little bit of kudos for that.
MATT: Definitely kudos to everyone involved here.
--------------------------------------------------------------------------------
[00:55:22] BREAKTHROUGH THERAPY DESIGNATION AND ERIC RUBIN'S ADAPTIVE TRIAL DESIGN
--------------------------------------------------------------------------------
MATT: So Roger was busy betting the entire company on this molecule — completely re-polarizing the clinical R&D effort towards it. But Merck had a couple aces up its sleeve before Roger even walked in the door. The first was a perfectly timed new regulatory path that was being spearheaded by the FDA. Back in 2012, the head of Merck's Non-Clinical Safety Group, Joseph DeGeorge, had spent 13 years at the FDA. At a conference in Japan, he caught wind from former colleagues about a brand-new regulatory pathway the agency was launching called Breakthrough Therapy Designation, or BTD. BTD was designed specifically to speed up the review of drugs for serious conditions if early evidence showed a massive improvement over existing therapies. Off of that early melanoma data with pembrolizumab, Merck applied for and got the breakthrough designation in January 2013, a few months before Perlmutter arrived on site. Getting that designation was huge — it gave them rolling reviews, much faster FDA interactions, formal organizational priority at the FDA. Most importantly, their competitors at BMS did not have breakthrough designation.
ALEX: Their second hidden advantage was Eric Rubin, the leader of the phase one early-development team on Keytruda. Remember how small KEYNOTE-001 was initially — just 30 patients? Well, even though it was just supposed to be this standard, small dose-escalation study, Rubin pushed to turn it into a massive adaptive trial. He built multiple expansion cohorts into it right from the start. So the second a signal appeared in any tumor type, Merck could instantly pivot into a larger cohort without having to stop and write an entirely new protocol. When we asked Roy Baynes about this trial design, he told us that the FDA held it up as the absolute gold standard for modern drug development — before immediately regretting it. Here's what he had to say.
ROY BAYNES [interview clip]: "Dr. Pazdur, when he was at FDA, held up Keynote 001 as the poster child, if you will, for adaptive design. He also then rapidly said, you know, created absolute chaos at the agency, and we're probably not gonna let anyone do this again."
ALEX: God, that's hilarious. I think in practical terms, sort of chaos of adaptively adding cohorts let Merck continuously amend the trial — right? Really let them fully leverage this trial design.
MATT: Again, just to really re-emphasize the point that Alex was making, because this is really important. Instead of having the standard, relatively small phase one trial, analyze the data, then set up phase two and phase three trials — they just kept amending the existing phase one trial. And the moment they saw efficacy, they just would add an expansion cohort on the fly. You know, we talked about KEYNOTE-001 being a small 30-person trial. By adding new arms to it, it ballooned to a huge 1,000-patient study.
--------------------------------------------------------------------------------
[00:58:25] KEYTRUDA'S 2014 FDA APPROVAL ERASES BMS'S FOUR-YEAR LEAD
--------------------------------------------------------------------------------
ALEX: Yeah, with this approach, Merck compressed a lot of years of clinical development into a single, relatively quick protocol. And God, this paid off more than anyone, I think, could have predicted, because Perlmutter and Baynes both joined in 2013, staring down this huge headstart by BMS. But with this adaptive design that gets 1,200 people into a phase one trial, and because of the BTD, they managed to close that gap. On September 4th, 2014, the FDA grants accelerated approval to pembrolizumab — Keytruda — for advanced melanoma. They erased a multi-year advantage in only 18 months.
ROY BAYNES [interview clip]: "I joined Merck in 13, as did Roger, and we actually got Keytruda approved in 14, and it was the first PD-1 approved in the US. So it erased this greater than four-year advantage that nivolumab had. So it's a pretty remarkable story of acceleration."
MATT: It's the ultimate comeback. Keytruda is officially the first PD-1 inhibitor approved in the United States. So they actually beat out nivo here.
ALEX: That's right — only by a couple of months, but they managed to pull it off.
MATT: Wow, that's incredible. Maybe one thing I would add here is a part of drug development that isn't particularly sexy, but it is pretty mind-blowing — and that is the manufacturing scale-up required to support this trial. You gotta remember the initial CMC work was a hacked-together workaround that was used because the pembro program was a zero-priority program. And so you now have a situation where suddenly you have to produce an enormous quantity of clinical-grade antibody to support a 1,200-person clinical trial. Not only that, the drug is now approved and you gotta meet the CMC standards of a commercially sold drug. It's a real logistical miracle that the manufacturing team pulled this off, and really kudos to everyone involved in that endeavor.
--------------------------------------------------------------------------------
[01:00:24] THE LUNG CANCER BATTLEFIELD AND THE PD-L1 BIOMARKER BET
--------------------------------------------------------------------------------
ALEX: Kudos indeed. But as monumental as this approval is, melanoma is kind of the starting ground for PD-1. The real war — the one that's going to eventually decide the future of Merck and BMS — that's lung cancer. This is a fundamentally different battlefield. Melanoma, relatively speaking, is a smaller indication than lung cancer. Lung cancer is literally the leading cause of cancer death worldwide. We're talking about an addressable market size multiples larger than melanoma — the crown jewel of oncology, some call it. And with that massive market to win, Merck has to make one final, completely unprecedented bet on biomarkers.
MATT: Right. So Roger's predecessor at MRL, Peter Kim, had always championed using molecular markers to identify exactly who was most likely to respond to a drug. That instinct carried right into the pembro program. Gary Gilliland, who was head of oncology at the time, insisted on building biomarker analysis into every single cohort from the very beginning. Extremely prescient. So as that 1,200-person KEYNOTE-001 trial enrolled, they were collecting biomarker data on every single patient. And their internal conviction was starting to build. Not only did the drug work, but it worked disproportionately well in a specific subset of patients who could be identified in advance with a simple immunohistochemistry stain. The question from there was more strategic than scientific. What do you do with that information? Is that signal clean enough to bet an entire pivotal lung cancer study on? Meanwhile, BMS is two years up the road, staring at the exact same underlying signal in their own data.
--------------------------------------------------------------------------------
[01:02:19] BMS VS. MERCK: ALL-COMERS VS. BIOMARKER-ENRICHED TRIAL STRATEGY
--------------------------------------------------------------------------------
ALEX: So as we go into the lung cancer battle between Merck and BMS, let's wind back the clock a bit and set the stage. It's June 2012. Right around this time is when BMS publishes a landmark paper in the New England Journal of Medicine. It's the first major human dataset on their anti-PD-1 drug, Opdivo. They have clinical data, they have buzz from ASCO, and they have a multi-year headstart on Merck. At this point, Merck's pembrolizumab hadn't even produced public phase one trial results yet.
MATT: But with Merck's blitz ahead with KEYNOTE-001, both companies end up facing the exact same strategic decision. They're both designing and running pivotal trials for first-line non-small cell lung cancer — the single biggest commercial prize in oncology.
ALEX: Just to quickly clarify why we emphasize first-line here — in oncology, treatment essentially works like a funnel. First-line is the very first systemic therapy a patient receives after diagnosis. It's the absolute largest pool of eligible patients. They stay on the drug the longest, and their immune systems are usually at their strongest.
MATT: Right. If patients fail that therapy and continue to be alive, the patient moves down the funnel to second-line therapy. Of course, at that point, the patient pool has shrunk. The cancer is more resistant and aggressive, and the patient's body is usually even more battered. So winning the first-line setting doesn't just mean more revenue — it effectively crowns you the absolute standard of care for that disease.
ALEX: Exactly. And to win that crown, BMS has this one intriguing data point buried in their phase one results. Out of 42 patients they evaluated, zero of the 17 PD-L1 negative patients responded, but nine out of the 25 PD-L1 positive patients did. That's a 36% response rate.
MATT: Here's the catch. It's a tiny sample size — only 42 patients — and it's done as an exploratory endpoint using an unstandardized, research-grade antibody. So at this point, you can basically imagine two paths forward for both companies. Path one is the all-comers approach. Enroll every single patient into the study regardless of whether their tumor expresses PD-L1 or not. You get the broadest possible market and fastest enrollment. Physicians don't need to run a test before prescribing the drug. But the massive risk is that if the drug primarily works in a PD-L1 positive patient, you're diluting the treatment effect across the whole population and your trial may fail.
ALEX: Path two, on the other hand, is biomarker enrichment. You restrict enrollment to only PD-L1 high patients — those with expression over 50%. You do get higher response rates, a cleaner signal, and a smaller trial. The risk is a much narrower label, delays with the companion diagnostic, and you might just be totally wrong about where to draw that cutoff line. Ultimately, BMS chooses path A, all comers, while Merck chooses path B, the biomarker enrichment strategy. And this decision has been analyzed ever since. There's this great piece on Substack by Liang Chang who essentially sent today's most advanced AI models back in time to June 2012. He gave those models only the information available at the exact moment and asked them to make the call. As it turns out, both Claude and ChatGPT independently recommend the all-comers approach, which is the one BMS chose. And the reason makes, frankly, total sense. Biomarker data was relatively immature and based off of 42 patients with an unvalidated assay. The FDA, after all, had approved other immunotherapies without a companion diagnostic. The commercial math heavily favored the broadest possible label.
MATT: And at this point you need to remember that Merck wasn't making this decision in a vacuum. They were making it two years behind the market leader. So when you're two years behind, copying the leader's strategy doesn't necessarily make sense. Merck needed a different bet. Merck knew that they had something BMS lacked at this decision point. As Merck's own KEYNOTE-001 expansion dataset was maturing, the PD-L1 signal was hardening from an initial small dataset into something that was much clearer. They discovered that patients with expression above 50% had a roughly 45% response rate, while patients below 1% PD-L1 responded at only about a 10% rate. That data print ultimately gave them the conviction that they needed to make the biomarker leap.
ROY BAYNES [interview clip]: "The 001 study, again, provided good validation for the biomarker — that for monotherapy, tumor proportional score greater than 50 predicted a good outcome with monotherapy."
ALEX: So to summarize: the science gave them conviction, the competitive position gave them the right motivation, and so they enriched. But before those massive first-line phase three trials actually read out, these two companies had already clashed in a second-line setting. And this is where Merck's strategy initially looks like a huge commercial mistake.
MATT: So at this point, we have to introduce a second KEYNOTE trial — KEYNOTE-010. While KEYNOTE-001 was their massive phase one discovery trial, KEYNOTE-010 was their dedicated trial for second-line lung cancer. And it's crucial to remember that this second-line battle was essentially the opening skirmish happening right as both companies were gearing up for the massive first-line showdown.
ALEX: Yeah. And in that second-line setting, Merck's drug was approved with a strict companion diagnostic. Physicians legally had to run a test to ensure the patient was eligible. On the other hand, BMS got approved in second-line for all comers. They were out there telling physicians, "You don't need a test with our drug, but you do need a test with the other company's."
MATT: Yeah. And that's a huge difference when you're selling a drug to physicians — with that lower barrier of entry. Roy Baynes told us Merck was getting killed in market share. The industry was telling them that their science was cute, but their commercial strategy was a disaster. BMS was eating their lunch in the second line.
ROY BAYNES [interview clip]: "The promotional focus oftentimes was on, you don't need to test with our drug, but you do need to test for their drug. And the net result of that is in second line, you know, we were well behind the competitor in terms of market share. And we got a lot of heat for it. We were told that, you know, this is elegant science, et cetera, but, you know, essentially your competition's eating your lunch."
--------------------------------------------------------------------------------
[01:08:18] KEYNOTE-024 VS. CHECKMATE-026: THE TRIAL THAT DECIDED THE CATEGORY
--------------------------------------------------------------------------------
ALEX: But then the first-line trials read out — hugely consequential for the destiny of both drugs and both companies too.
MATT: That's right. This is where most of the lung cancer patients are in clinical settings. So they're dueling phase threes. They come out, and it's a bloodbath. BMS's all-comers trial, CheckMate-026, produced a hazard ratio of 1.02. It showed absolutely zero benefit and completely failed. Meanwhile, Merck's enriched KEYNOTE-024 trial hits its mark perfectly. Bullseye.
ALEX: And the commercial calculus flips completely overnight, because Merck is now the only one with a first-line approval. All of the criticism about biomarkers instantly vanishes. Here is Baynes describing that exact whiplash.
ROY BAYNES [interview clip]: "Half a year later, KEYNOTE-024 read out strikingly positive, and the competitor trial failed miserably. Still not, I'm still not sure why their, um, post analysis of the PD-L1 positive population was not successful, and I'm not sure anyone really has a good explanation for that. There's lots of hypotheses, but don't really know why. But that has changed the whole calculus completely. Essentially, you know, Keytruda became the market leader. Then when the combination with chemo came in, it really took off, and obviously criticism about biomarkers, you know, disappeared."
MATT: Meanwhile, over at BMS, the mood was the exact opposite. Nate Lonberg, a BMS scientist who helped develop the drug, described the aftermath, quote: "That result destroyed BMS. We had huge pressure from investors. It radically changed the company. It was just one bad trial design. But there was no clawing our way back."
ALEX: Yeah, what a turnaround. So for whoever is keeping track, this is the second time BMS gets the mark directionally wrong on immuno-oncology. Even though the data pointed toward enrichment in their phase one, BMS chose the broad, safe, consensus path both times — and both times it cost 'em the category.
MATT: The financial fallout of this data differential is staggering. If you look at graphs of revenue for Keytruda versus Opdivo, Keytruda is essentially a straight line up since 2016. And Opdivo is more or less a flat line since 2016. To be fair, that flat line is at $10 billion — that is mega blockbuster category and nothing to be sneezed at. But when you look at what could have been, Keytruda is now pulling in $30 billion a year.
ALEX: This set of decisions around clinical trial design is one of the reasons that Keytruda is now the best-selling drug in the world. But there's a final twist here. This biomarker that won Merck the prize of first-in-line lung cancer — the original reasoning for why it even worked at all was not even right. So Perlmutter's starting model for harnessing the biomarker companion diagnostic in clinical trials was something like this: PD-L1 on tumor cells is a sign of active immune evasion, so PD-L1 overexpression in tumors marks exactly where the blockade should pay off. Not wrong, exactly, but not complete either. And in 2019, in a Nature interview, Perlmutter makes exactly this quote: "PD-L1, we later discovered, is transcriptionally regulated by gamma interferon. So wherever there is a high level of gamma interferon locally, PD-L1 is expressed. So instead, what PD-L1 told us was something slightly different, which is just that there was evidence of inflammation in the environment."
MATT: Crazy. Yeah, I mean, essentially the biomarker strategy was correct clinically, but for the scientifically wrong reasons.
ALEX: Yeah, exactly. They pretty much got lucky. Merck's clinical team picked this correct enrichment strategy based on a biological model that was flawed from the beginning, but empirically whose consequences happened to be right anyway.
MATT: That's biology for you, man. Hard and confusing.
--------------------------------------------------------------------------------
[01:12:03] LUIS DIAZ, MSI-H, AND THE FAILED BMS TRIAL THAT MADE KEYTRUDA TISSUE-AGNOSTIC
--------------------------------------------------------------------------------
ALEX: So as we've learned, PD-L1 was a messy biomarker, but Merck managed to leverage it just enough to win the lung cancer war. To truly solidify their dominance, they had to capitalize on a completely different biomarker, discovered in a totally different setting, under even stranger circumstances. And the irony, again, is the whole thing started with a BMS trial. To understand how this happened, we have to move down to Johns Hopkins University. Unlike most cancer centers at the time, Hopkins actually had decades of immunotherapy infrastructure. They had spent years developing a cancer vaccine called GVAX. GVAX ultimately failed as a standalone drug, but it built something very valuable for Johns Hopkins. It helped create a major academic hub of immunotherapy experts and clinical trial infrastructure at the university.
MATT: It was within this specific environment that Luis Diaz, a gastrointestinal oncologist, was working with Bert Vogelstein, the legendary cancer geneticist. They had been studying patients with Lynch syndrome. These patients had defects in the genes that repair DNA, which made them highly prone to mutations and ultimately highly prone to colorectal cancer. But Diaz and Vogelstein noticed something really weird — tumors from Lynch patients were always heavily infiltrated with immune cells, and these patients lived longer with their cancers than other colorectal cancer patients did.
ALEX: Diaz was essentially sensing a Planet X. He couldn't see it directly, but he could sense that because these tumors couldn't repair their DNA well, they were mutating uncontrollably. More mutations meant more neoantigens — basically telling the immune system to attack. This hunch leads to a fateful hallway conversation with a fellow clinical oncologist. A conversation about a recently failed PD-1 trial led none other by BMS. In this chitchat, Diaz finds out that BMS had tested their PD-1 drug, Opdivo, in 33 colorectal cancer patients, and only one patient responded. In standard clinical development, a 1-in-33 response rate means the drug is a total dud in that cancer type. But Diaz doesn't throw it out, because his brain is already primed by this Planet X that he suspects is going on. Did that one patient happen to have a mismatch repair deficiency? Somehow he manages to track down the clinical data for that single responder to get their tumor tested for its mutational load. And years later, he explained exactly what they'd found.
LUIS DIAZ [archival]: "Colon cancer cells typically only have a few dozen mutations. But we were thinking, maybe that patient's tumors had a mismatch repair deficiency and would harbor thousands of mutations per cell. And lo and behold, that turned out to be the case."
MATT: Surprise, surprise, he was right. But it didn't stop there. Encouraged by this clinical anecdote, Diaz and a young oncologist named Dung Le set up their own small 41-patient trial to test Keytruda in both patients who had DNA-repair-deficient tumors and in patients with normal tumors. The idea here being that they expected for a difference in outcomes between the two groups. The results were a total knockout. While normal tumor patients had zero responses, the DNA-repair-deficient patients — who had completely run out of other options — posted response rates between 40 and 71%. The results were so dramatic that even as the trial was still enrolling, they published the results in the New England Journal of Medicine.
ALEX: Absolutely crazy. This was a tiny 41-patient academic study, but enough of a signal to fully vindicate Diaz's hunch, for real. So from here, it obviously went up the clinical totem pole at Merck. And when we asked Roy Baynes about it, he credited Diaz directly for giving them the roadmap to prove this at scale.
ROY BAYNES [interview clip]: "When it came to the histology-agnostic approvals, that was another interesting discussion. You know, it was very clear there were a number of biomarkers that were informative. PD-L1 was informative. Obviously, MSI — we had learned from Dr. Diaz about his data. And then we also had the observation that it's likely that tumor mutational burden would correlate with neoantigens, and that it's possible that TMB might predict, you know, important results."
MATT: And in 2017, the FDA approved pembrolizumab for the treatment of any solid tumor with that specific DNA-repair defect, regardless of where in the body the cancer originated.
ALEX: Yeah, this is a really important milestone. This was the first time in the history of oncology that a drug had gotten approval based off of a genetic biomarker, completely independent of what the tissue of origin was for that specific tumor. I think the upshot of that is that there have been lots of patients with relatively rare tumor types who, if they do happen to be microsatellite-instable, are able to get access to this drug and potentially get some clinical benefit.
MATT: Yeah. And to think that this, again, came from something that happened in a BMS-run trial — it's really interesting to see how the two companies handled these findings in such small patient populations.
ALEX: Yep. This is one of the key advantages that Merck has behind BMS — they have the luxury to let BMS be the pioneer to a certain extent, and then look at that data with fresh eyes and potentially make a nicely informed decision with it.
--------------------------------------------------------------------------------
[01:17:18] KEYNOTE-189: PERLMUTTER'S BET ON COMBINING KEYTRUDA WITH CHEMOTHERAPY
--------------------------------------------------------------------------------
MATT: Exactly. Okay, let's set the stage here. So at this point, Merck has decisively won the biomarker war. The reality is, it only gave them a beachhead. Restricting the drug to PD-L1 high patients in frontline lung cancer meant that they were still only partially capturing the big prize. So to get the rest, Perlmutter had to make another bet that the rest of the industry thought was insane. He decided to combine Keytruda with the standard chemotherapy used in lung cancer — which on its face sounds completely counterintuitive. Chemotherapy famously suppresses and destroys the immune system. So why would you pair it with a drug that literally relies on the immune system to work? Will patients even be able to handle these two relatively toxic drugs?
ALEX: It was definitely a big, unorthodox risk to take. When we talked to Roy Baynes, he told us that other pharma companies flat-out refused to even entertain the idea.
ROY BAYNES [interview clip]: "He was also very intrigued by the concept of combinations, and particularly the addition of agents which caused antigenic cell death — such as chemotherapy. And you can imagine at the time, this was pretty much a heretical view. I mean, there were companies absolutely no way combining IO with chemo."
MATT: Yet there was some biological rationale to these combinations. Yes, chemotherapy is toxic, but it blows up tumor cells. When those cells die, they can spill all of their contents and proteins out into the surrounding tissue, creating inflammation — a source of new antigens for immune cells to pick up. So basically the rationale from Merck was: things that would increase local inflammation will work in combination with pembro somewhat better.
ALEX: In 2018, Merck publishes the results of their combination trial — KEYNOTE-189. They gave Keytruda and standard-of-care chemotherapy to all comers regardless of PD-L1 status, and compared that to chemotherapy alone. That combination cut the risk of death in half compared to chemotherapy. It was an absolute home run. Where the biomarker trial had given Merck that high PD-L1 population, KEYNOTE-189 gave them literally everyone else. That combination has now become the global first-in-line standard of care for lung cancer.
MATT: Yep. And in many ways, since the moment where the war with BMS ends, Keytruda was the drug you started everyone on in lung cancer.
--------------------------------------------------------------------------------
[01:19:21] MERCK'S CLINICAL DEVELOPMENT PLAYBOOK: BASKET TRIALS, BACKWARDS MARCH, EXTERNAL COLLABS
--------------------------------------------------------------------------------
MATT: The biomarker strategy we just described — using PD-L1 positivity or DNA mismatch repair status — definitely gets all the headlines. But Merck's broader clinical development strategy was just as critical to the success of Keytruda. Understanding this playbook explains exactly why Keytruda's lead over the rest of the industry has kept widening instead of narrowing over time. I'd also highlight that we're at a time where a new generation of potentially broad-acting anti-cancer drugs are about to enter the clinic — the so-called KRAS inhibitors — and so the playbooks that worked for Keytruda are potentially quite relevant to that class of drugs.
ALEX: Alright, playbook number one. The first piece of this playbook was the so-called parsimonious basket trials Merck ran. Instead of making one massive, slow bet, Merck ran incredibly small single-arm studies across dozens of different cancer types simultaneously. They would select patients based on biomarkers and just look for any signal big enough to justify hitting the gas. Here's how Roy described it.
ROY BAYNES [interview clip]: "That was the outcome from the basket trials, where, you know, we took tumors that were either PD-L1 strongly positive, were highly mutated, or both. And I looked at monotherapy activity in, you know, a small number of patients — about 30 patients per study. So a parsimonious trial. Lo and behold, virtually every tumor type reacted, and reacted amazingly positively. So immediately you knew that this was gonna be huge, and you also knew that it was gonna be an industrial-strength program to develop."
MATT: Then, once a signal started to come in, Merck would race to grab an accelerated approval in the most advanced relapsed/refractory patients, often based on single-arm data. With the beachhead approval in place, the drug could start getting sold into that market. At the same time, a confirmatory trial would be run to more rigorously prove clinical benefit and convert that initial accelerated approval into a full normal FDA approval.
ALEX: Playbook number two. With an initial toehold in relapsed/refractory patients, Merck would systematically walk the drug backwards into earlier and earlier lines of treatment. From a financial perspective, that's bigger and bigger total addressable markets. To quote Roy again.
ROY BAYNES [interview clip]: "The strategy really was, let's get to market as quickly as possible with accelerated approvals. And those were obviously gonna be in advanced patient populations without options, and based upon single-arm data. That's gotten harder to do these days. But that was, at the time, you know, very well received by the agency, partly because, you know, the results, the effect sizes were so big. Then we had to deploy the confirmatory trials, and what we did there was to move into a slightly earlier line of treatment. We used the biomarkers to help us in all of these monotherapy trials — it turns out that the biomarker is quite predictive in monotherapy. We then moved, you know, forward with a number of combination trials. In addition, we moved into earlier lines of treatment. So we went from salvage to second line, to first line, to adjuvant, then into neoadjuvant, and then ultimately into the perioperative space."
MATT: The third dimension of their playbook — and a brilliant one at that — was how they handled combinations. As we mentioned earlier, Merck at this point had a relatively small internal pipeline of their own molecules to pair with Keytruda. But outside Merck's walls, there were hundreds of experimental candidates at other companies. So Baynes's team decided to turn that asymmetry into an asset. They essentially created an open-door policy for other pharma companies. If you had an experimental drug and wanted to see how it worked alongside Keytruda, Merck would supply the Keytruda for free, as long as your drug was safe and your company paid for the clinical trial.
ROY BAYNES [interview clip]: "Externally, there are many. Internally, we were few. You know, having collaborators — provided they met certain bars. So generally when a company came forward to look for a collaboration, we would look quite stringently at their efficacy data, their safety data. We wouldn't take on projects which looked as though they could pose risk to our molecule. But provided, you know, a safety bar was met, the idea of providing Keytruda to these folk — they go and they fund the study and give us the data afterwards — that's a pretty cost-effective way of generating a heck of a lot of data. And to us, we didn't see much downside to that. Plus, those types of interactions led us to certain conclusions — for example, in the business development realm, where positive combinations or positive interactions may have driven certain business development decisions. So as I mentioned, at peak, you know, we had well over 100 collaborations going. I don't know what the number is today, but I'm sure it's still a big number. And we also had a huge number of trials ongoing, which essentially someone else was funding."
ALEX: The result was a massive expansion of the Keytruda clinical program without Merck having to bear the full cost. And several of those external collaborations delivered massive blockbusters. For example, a bladder cancer win — Keytruda paired with the Pfizer-Astellas antibody-drug conjugate, Padcev — came directly out of the permissive collaboration structure. Here's Roy again.
ROY BAYNES [interview clip]: "We combined with a number of ADCs, and, you know, I think you look at the data with enfortumab vedotin, for example, from Seagen, Astellas, and Pfizer, in the bladder space — that has transformed the treatment of bladder cancer. And essentially, that came out of the extramural collaborations."
MATT: Keytruda essentially became the backbone molecule that every other oncology company needed to partner with. It stopped being just a Merck asset and became the substrate that the rest of the industry's innovation was built on top of. And in doing so, Merck cemented the place of Keytruda into the standard of care for a huge number of different cancers.
--------------------------------------------------------------------------------
[01:25:29] PART FOUR — THE IO GRAVEYARD: TIGIT, CD47, IDO1, LAG-3 AND TENS OF BILLIONS INCINERATED
--------------------------------------------------------------------------------
ALEX: By 2018, just four years after the first PD-1 approval of Keytruda, the industry had completely lost its collective mind around immuno-oncology. It was a whole bonanza, really. There were suddenly over 160 different PD-1 and PD-L1 drugs in the pipeline worldwide.
MATT: I mean, in many ways, this is totally understandable. There were so many different clinical readouts with huge effect size. This drug was working, and it was working all over the place. It was starting to be a running joke among researchers that there were really only two types of drug companies left in the world: those that were deep into cancer IO, and those that desperately wanted to be. And because PD-1 had been working so miraculously well, the industry made a massive multi-billion-dollar bet. They figured that the immune synapse must contain other, maybe even more important brakes than PD-1. Maybe there are multiple redundant brakes on T-cell function. Based on that assumption — and frankly, just chasing the absolute financial bonanza of Keytruda — tens of billions of dollars suddenly flooded into entirely new targets. Things like TIGIT, TIM-3, CD47, and a dozen others. Everyone was looking for the next Keytruda.
ALEX: Honestly, in retrospect, it was completely understandable. What are the odds that the second checkpoint inhibitor the industry ever really tested turned out to be the absolute best way to activate the immune system? Right? It's just reasonable pattern matching. You see a massive success with Keytruda, and you try to replicate it. But on the sidelines, out of all of the people to be saying this, Roger Perlmutter saw this massive gold rush happening and was incredibly skeptical. He didn't think these other targets were going to be foundational the way PD-1 was. In fact, in this excellent 2019 Nature interview, which we keep referring to, he basically called out the entire industry for just blindly chasing hype. Quote: "We have systems in which people tend to pursue simultaneously and repetitively all around the world the same few observations, following the same kind of herd instinct everywhere."
MATT: Herd instinct. Oof. Well, to be fair to all those people who are in the herd, his claim was kind of unfalsifiable at the time. There wasn't really the clinical data yet to prove that other targets would fail. I think this was just a bet on his own biological intuition — and frankly, very well aligned with the sort of financial bets that Merck was making, which was essentially: plow every single marginal dollar into another Keytruda trial. At the same time, though, standing here as we record in 2026, he should get credit for being patient. Right? He called it. PD-1, PD-L1 has ended up being kind of the one-hit wonder of this axis. When we asked Roy Baynes to look back from the vantage point of today and all the money that was spent trying to find the next big checkpoint inhibitor, his verdict was pretty definitive.
ROY BAYNES [interview clip]: "When we look at the other IO agents, it's been really, uh, largely one of disappointment. They haven't added very much."
ALEX: Yeah, "disappointment" is putting it politely. Honestly, if you look at the actual graveyard of clinical trials over the last number of years, it's just been an absolute bloodbath. We can do a quick speed run of the craters. First up, IDO1 inhibitors. The industry was convinced that this was the next big combo target after PD-1. Merck even ran a massive phase three trial pairing Keytruda with an IDO1 drug called epacadostat. It bombed. Baynes told us the data was, quote, a "spectacularly neutral finding," which is a very polite way of saying a failure.
MATT: Then you have the TIGIT saga. Roche and Genentech essentially bet the farm on their massive Skyscraper trials — clinical failures.
ALEX: Then there's CD47. Gilead spent nearly $5 billion acquiring Forty Seven Inc. to get their flagship CD47 drug, magrolimab. They had to completely scrap the entire program, with some of the trials showing hints that it could even worsen patient survival. TIM-3, OX40, the list goes on and on. Tens of billions of dollars incinerated.
MATT: Basically, as we've already said, it's turned out that Perlmutter's biological intuition is exactly right. PD-1 wasn't just one brake on the immune system that happened to be discovered very early on — for the vast majority of solid tumors, it's the brake. And, you know, to be fair to the industry, and as someone who has worked on trying to find novel ways to unleash T-cell function, I do think that there was a real probability that there were more checkpoint-like drugs out there waiting to be discovered. And that investment needs to be understood in the context of what the opportunity was, which was for a broadly acting oncology drug. We know those drugs can do $10 to $30 billion a year in annual sales. That's a level where a very healthy level of investment does make sense.
--------------------------------------------------------------------------------
[01:30:23] WHY PD-1 WAS THE ONLY CHECKPOINT THAT WORKED (LIEPING CHEN REVISITED)
--------------------------------------------------------------------------------
ALEX: But the obvious follow-up question here is why? Why did tens of billions of dollars get incinerated, and why did all of these other targets fail? One interesting and plausible explanation is found in a paper published last year by Lieping Chen. If you remember the name, it's because he's actually the scientist who first discovered PD-L1 upregulation on tumors way back in 1997. And his argument cuts to the root of the problem. It goes something like this: Tumors are incredibly efficient machines. Once they establish a dominant immune-evasion mechanism, like overexpressing PD-L1, there is absolutely zero evolutionary pressure to develop a backup system for immune escape. A tumor has no incentive to build a second brake if the first one already works. But then he points to a massive safety component. PD-1 is uniquely druggable because PD-L1 is specifically upregulated right at the tumor site. Targets like TIM-3 or CTLA-4, however, bind to ligands that are broadly expressed all over the normal healthy tissues in the body. If you block them systemically, you risk indiscriminate, widespread toxicity, similar to what we saw with CTLA-4 and ipilimumab. So potentially the biological reality is, inhibiting the PD-1 axis has slightly more tumor-selective activity than other forms of checkpoint inhibition.
MATT: I think the other thing to really think about is the fact that each of these checkpoint inhibitory pathways all seem to have their own flavor of toxicity, and potentially would have the possibility of working if it weren't for an impossibly small therapeutic window. Right — 4-1BB targeted molecules suffer from TI issues. I think they were having liver signals at doses that were seeing anti-tumor activity. Attempts to agonize co-stimulatory pathways like ICOS and OX40 had a therapeutic-index issue where they were modulating regulatory T cells, not what you want to do in the context of a cancer. It's turned out to be incredibly difficult to drug additional members of the tumor immune synapse. The way this industry works — it could be that it will eventually work, and it'll just take another 10 or 20 years to perfect the drug embodiment.
ALEX: Since the PD-1 boom has started, there's really only been one truly new checkpoint inhibitor that has managed to win an FDA approval, and that's an agent that targets LAG-3. BMS got it approved for melanoma. Even that, though — there's a pretty big asterisk next to it. It's not a monotherapy. It's a fixed-dose combination of the LAG-3 targeted drug with BMS's old drug nivolumab. So not only does it not sell super well, the sales that it does have arguably are just cannibalizing the sales they would've had for the nivo franchise. It's definitely not been this groundbreaking new combination therapy.
--------------------------------------------------------------------------------
[01:33:32] PART FIVE — INSIDE THE BEST-SELLING DRUG OF ALL TIME
--------------------------------------------------------------------------------
ALEX: Before we talk about the future of Merck and Keytruda, let's just baseline what the state of play is in 2026. I really wanna emphasize here how hard it is to overstate how much a single drug has completely rewritten the reality of cancer treatment. As we mentioned earlier, it has 42 approved indications across 18 different cancer types. It has essentially become the foundational first-line treatment for a massive chunk of solid tumor cancers.
MATT: Yeah. And, again, we're not talking about delaying tumor growth here. It's really driving significant, meaningful improvements in overall survival in many different cancer subtypes. The long-term survival curves in things like metastatic melanoma and lung cancer have started to really become decoupled from their historical baselines because of these drugs. These numbers are staggering.
ALEX: Yeah. I was reading in research for this episode — a great book by a journalist named Charles Graeber, "The Breakthrough." I remember there was a mention of how the marker of efficacy in cancer drugs changed from progression-free survival to overall survival in some cancers, which is just a testament to how well these drugs worked. So one, for example, that we've been talking about a lot is metastatic non-small cell lung cancer. Before checkpoint inhibitors came into the scene, the five-year overall survival rate with the standard of care of the time — platinum chemotherapy — was 5%. That's basically nothing. Metastatic NSCLC diagnosis was pretty much a death sentence. Then after KEYNOTE-024, first-line Keytruda monotherapy in PD-L1 high patients, the five-year overall survival rate is 30% compared to just 16% on chemo. And the chemo arm is already inflated, by the way, because most of those patients crossed over to Keytruda after progressing.
MATT: Yeah, I mean, it's undeniable here that the effect sizes are large and that real progress is being made in the treatment of lung cancer and other indications. You know, I think arguably the improvements in metastatic melanoma survival are even more dramatic than lung cancer, just because the baseline was even worse. I think pre-2011, you didn't see too many long-term survivors with metastatic melanoma, and ipi was the first drug that started to bend that curve. Ironically, ipi is now the comparator arm in the KEYNOTE trials. And in KEYNOTE-006, 34% of Keytruda patients are alive after ten-year follow-up. You're doubling median overall survival for some of these drugs. We may even be getting to the point where you need to think about — is the limit now just standard overall life expectancy? It tends to be older people; they get cancer. Are you getting to the point where their lifespan is just getting limited by other factors associated with old age?
ALEX: Exactly. I think the cleanest articulation of this is actually in the KEYNOTE-024 paper itself, where the investigators wrote that pembrolizumab may be transforming metastatic NSCLC into a treatable chronic disease. Those were the words they used. The long-term survivors of metastatic lung cancer used to be a nearly null set, but now oncologists are dealing with survivorship questions that did not even exist a decade ago.
MATT: Yeah. You know, I don't think we're quite where we're at with multiple myeloma. That's a disease where there's pretty much no change in overall life expectancy if you're diagnosed with that. We're not there with lung cancer, but we've definitely made progress.
ALEX: As we talk about these vignettes and different cancer indications, you multiply that across 42 different indications, and you can start to understand why this is one of the best-selling drugs ever in pharmaceutical history.
--------------------------------------------------------------------------------
[01:37:03] "BUILD A WALL, HIGH AND WIDE": MERCK’S COMMERCIAL STRATEGY FOR KEYTRUDA
--------------------------------------------------------------------------------
MATT: Yeah. So when we asked Roy Baynes to describe Merck's commercial strategy for Keytruda, he put it like this in a very specific way.
ROY BAYNES [interview clip]: "The strategy, as Roger described it, was: we're gonna build a wall. The wall is gonna be very high and very wide. And those coming after us are either gonna have to go over the wall or through the wall. And that's what we executed."
ALEX: "High and wide" is the perfect way to describe this strategy. It's just so good. Look at the scale of the label — you have 42 approved indications, 18 different cancer types. That is a really wide wall. And in many of those indications, patients will start on Keytruda from the time of diagnosis. That's a very high wall. And you gotta remember here, this is not a cheap daily pill. This is getting priced the way that oncology drugs get priced, and a year's course of infusions with pembrolizumab will easily cost in the $200,000 range.
MATT: Yeah, so I mean, taking these two things into account combined — a $200,000 price tag. Is this list price or negotiated? My simple brain can't handle that.
ALEX: It's all good. It gives you roughly what the ballpark we're talking about. And again, this is a per-year price, not an overall price.
MATT: That's right. $200,000 in price with this huge patient population. Net-net, that starts to make a lot of sense how this becomes the best-selling cancer drug of all time. Cancer in aggregate is very common, and patients have been very eager to try anything that works for their cancer.
ALEX: One of the ironies here is that, like, if you rewind the clock on the pharmaceutical industry, and you go back to the era before, let's say, the 1990s, the pharma industry did not view oncology as a particularly attractive therapeutic area. It was an area that was really dominated by academic investigator-initiated trials. And the wisdom of pharma at the time was, like, you couldn't make a lot of money there. The real money was in more of primary care, cardiometabolic conditions, et cetera.
MATT: Yeah, it's crazy to think of a time where the biggest franchises for pharma were not cancer-related. They used to be cardiovascular-related — pills for cholesterol, blood pressure, asthma, something that tens of millions of people would take for decades or for their entire life, like Lipitor or even Merck with Singulair. That patent loss of exclusivity that started this Keytruda story — inadvertently, cancer was an incredibly fragmented, almost niche commercial market compared to these. You gotta remember too that, like, the worse a cancer drug actually works — given that pricing is usually on a per-dose or per-month basis — the duration of therapy was, is gonna be pretty short. And it's hard to scale revenue the way that a cardiovascular drug, which may get prescribed to millions of people for many years, does.
ALEX: Yeah, exactly. As we've learned, Keytruda has completely flipped that math upside down. It's an immunotherapy — much more tolerable than traditional chemo — and more importantly, it works. So it's creating these long-term survivors that we've been talking about, where the duration of therapy suddenly stretches from a few months to, in many cases, years. And Merck has been relentlessly executing in their clinical trials to increase the aperture, to increase the eligible patient population for this drug.
MATT: Right. We talked about this playbook of a relentless march back to the frontline and neoadjuvant. As they've executed on that strategy, the pool of eligible patients for their drugs just keeps getting bigger and bigger.
ALEX: Yeah. It's the ultimate compounder, and it rewired how the pharma industry thinks about oncology. But it's so easy to look at this from 2026 and say that this was all inevitable. The truth is, in 2014, even when Keytruda first got its accelerated approval, there was no idea that this was gonna be a multi-deca-billion-dollar drug every year in annual revenue. Do you wanna know what the consensus estimate was at the time for analysts?
MATT: You know, I bet it was, like — I bet analysts thought, look, it's gonna get approved in melanoma and some other boutique cancers. Maybe it does a billion or two.
ALEX: Exactly. Yeah. They thought that within three years of launch, it would generate one and a half billion dollars. Instead, that year it generated $5 billion. So they were off by more than threefold. And that tells you everything about even analysts on the forefront, seeing this drug get approved, how badly they predicted where this drug would go.
MATT: Yeah. One of my favorite anecdotes about people underestimating the revenue potential here actually connects back to a company called MRC Technology, which was a British medical research charity that was involved in the humanization process of pembrolizumab way back in the Organon days. So humanization is the process of taking an antibody that was made in mice and then changing the protein sequence around a little bit so that it looks more like a human antibody. And the crux of the matter is doing that in a way that doesn't alter the functional properties that you started with. When you wind back to the early 2000s, this process of humanization was not trivial. And so places like the MRC had real leverage. You know, when they did this for Keytruda and other massive blockbuster antibodies of the era, they were paid not only the amount of money for their services, but they got a royalty stake in the drug. MRC, for better or for worse, as Keytruda was really showing its legs, decided to sell a chunk of their royalty rights off to a private equity fund for about $150 million.
ALEX: Sounds good. I guess.
MATT: Unfortunately, they sold that back before the rocket ship went completely vertical. I mean, I bet they left billions of dollars on the table from early sale of their royalty stake. I guess only MRC knows what the true scale is there.
ALEX: Yeah, that's true. And even people currently running Merck recognize how unique this rocket ship has been over the past decade-ish. At ASCO 2024, a couple years ago now, Merck's current CEO Robert Davis made this really interesting comment on the record. He said, "We're not out there trying to find the next Keytruda. That was lightning in a bottle. Most of what we're looking at are therapies that are very specific." So basically, the company that built the world's biggest drug flat-out admits that they don't expect to do it again. Honestly, that tells you exactly how Merck's leadership views the Keytruda franchise: a once-in-a-generation program, and by no means a replicable template.
MATT: Yeah, I see where he's coming from. It's not a template. At the same time, I think the reality of the pharma industry is that they're pretty much always dependent on a handful of mega-blockbuster drugs. Will Merck ever find another drug that works as well and as broadly as Keytruda? Maybe. Maybe not. At the same time, I don't see a world where Merck as a corporation doesn't eventually become completely dependent on some other major franchise.
ALEX: Fair assessment. I agree.
--------------------------------------------------------------------------------
[01:43:44] PART SIX — THE PATENT CLIFF AND LOSS OF EXCLUSIVITY IN PHARMA
--------------------------------------------------------------------------------
ALEX: Keytruda today is a $30 billion-per-year juggernaut, but the clock is literally ticking. The original patents for Keytruda begin expiring in 2028, and there are already eight different pembro biosimilars in development ready to jump in when it finally does. For a drug this big, that is an existential threat to Merck's revenue base.
MATT: Which brings us full circle to a concept at the very core of the pharma business model: the never-ending issue of loss of exclusivity, or LOE. In almost any other industry, if you have an incumbent that is highly reliant on a single, aging blockbuster product, they're a sitting duck. There's the ability for startups to come in and try to disrupt them via counter-positioning. What's very interesting about pharma is that every single drug has a well-defined, legally mandated date at which you no longer have either regulatory exclusivity or patent exclusivity. And so companies know pretty far in advance when essentially the cash faucet is gonna get turned off. And because of that, pharma has evolved a very explicit set of lifecycle management strategies that get baked in pretty much from the moment that a drug gets approved — to start thinking about how to handle that LOE event.
--------------------------------------------------------------------------------
[01:45:35] KEYTRUDA QLEX (SUBCUTANEOUS) AND THE LIFECYCLE MANAGEMENT PLAYBOOK
--------------------------------------------------------------------------------
ALEX: Exactly. So as we're recording this today, Merck is currently executing a massive two-pronged lifecycle management strategy to survive the Keytruda cliff. The first prong is to extend the life of the molecule itself through new formulations. In late 2025, last year, Merck launched Keytruda QLEX, which is a subcutaneous version of the drug.
MATT: So the logic here is, if you're a sales rep for Merck, you are trying to shift as many of your physicians to start prescribing the new patent-protected subcutaneous formulation of Keytruda before the old IV formulation goes biosimilar. Right now you'll see estimates out there that estimate peak sales for Keytruda QLEX around $7 billion a year. We just talked about how inaccurate some of these estimates can be, but that's what's floating out there right now. As I've thought about whether Merck will actually be able to transition their sales to the new formulation, I think this is gonna be a bigger hurdle than people appreciate, and I'm not really sure what's gonna happen here. As a reminder, the cancer patients are not transitioning from being forced to get an infusion of the drug — sort of coming into the hospital or the clinic every couple of weeks — to being able to take a drug at home. In both scenarios, the patients have to interact with the healthcare system to take this drug. And I think the second really important thing is that when people are on Keytruda, 60 to 70% of the time they're also getting chemotherapy, and that is administered intravenously.
ALEX: Yeah. How much value is there if you're already having the patient in clinic with a drip in their arm to then go in with the subq shot? They're already there, ready to have IV. It makes you think the clinical economics itself are gonna be kind of weird too. Right? So we've talked about this a little bit before in our last episode. Clinics get reimbursed roughly twice as much for administering an IV infusion as they do for a subq injection. So the financial incentives lie in keeping patients on IV Keytruda instead of going to QLEX. The other option being to switch them to a cheaper IV biosimilar, which in itself will probably be close to current brand-name Keytruda in price.
MATT: You also kind of wonder how much insurers are gonna have a problem with this. I know for in-clinic, it's a little bit different than Medicare Part D in terms of how they aggressively prior auth these things. But for something that has these features — being injected in clinic alongside chemotherapy — they could use these utilization management strategies, or step requirements, physician justifications, et cetera, to get people on QLEX. It's really complicated, and there's very complex financial motivations at play for Merck versus the insurer versus the physician versus the hospital versus the patient. And we'll see how it all plays out. I think it's sort of too early to tell. In any event, despite trying to transition folks over to the new formulation, Merck is executing on a second prong of their strategy, which is, you know, diversify the pipeline. They're investing heavily in antibody-drug conjugates, new targeted therapies. A lot's going on there.
ALEX: Yeah. We have this great quote from Alex Snyder, who's Merck's current head of translational medicine for oncology, and has been very vocal about the one-pipeline approach. This is the quote: "We have a one-pipeline approach, by which I mean we consider both external and internal candidates to be part of the same pipeline. From my experience in pharma, that's unique. If you would've asked me 15 to 20 years ago if we would've had an approved drug to treat KRAS tumors, I would've said no way. Now we have approved drugs, and we're looking at next-generation therapies in combinations." Hard to miss both the echo of Perlmutter and Roy Baynes in that quote.
MATT: Yeah. I think there's a deeper question here, though, which is whether Merck can replicate the amazing clinical execution and strategy that made Keytruda a blockbuster in the first place. We're talking about the cleverness with biomarkers, the permissive collaboration model, the insane speed of execution. Can they do that with a whole new set of molecules in a different competitive landscape? It awaits to be seen. Sometimes I wonder whether the insane effect sizes and clinical benefit of Keytruda was a really necessary component, enabling them to move so aggressively and execute so well.
ALEX: It's a great point. If the efficacy were only as strong as CTLA-4 blockade when Perlmutter strolled in, in 2013, maybe they wouldn't have jumped on it as aggressively.
MATT: Yeah, exactly. I don't see a molecule like CTLA-4 being strong enough to completely restructure Merck as an organization.
--------------------------------------------------------------------------------
[01:50:23] PD-1/VEGF BISPECIFICS: IVONESCIMAB, SUMMIT THERAPEUTICS, AND THE NEXT THREAT
--------------------------------------------------------------------------------
ALEX: As we look forward, one of the competitive challenges to Keytruda's dominance actually — it may not just come from biosimilars, it may come from a new class of drugs called bispecific molecules.
MATT: There's a lot of irony here. We spent some time discussing how the industry had spent years and billions of dollars trying to find the next important checkpoint molecule — TIGIT, LAG-3, TIM-3, et cetera, et cetera — and this data is all maturing right now. But it turns out that the next big thing beyond PD-1 may be — drumroll — a slightly different version of PD-1. You can make your own vanilla PD-1 antibody in a new indication and easily do a billion dollars. You can launch a subcutaneous version of PD-1, maybe make $7 billion. It's looking like you can also, in a something called a bispecific format, take a PD-1 binder and slightly alter its behavior or pharmacology, and that there may be some marginal benefit to that.
ALEX: If you all haven't caught the hint by now, we're referring to the recent story about PD-1 and VEGF — or as I like to call it, PD-1 wearing a fake mustache. Let's dig into this narrative. Over the past couple of years, we've been seeing data come out from a company called Summit Therapeutics and their Chinese partner, Akeso. They developed this PD-1 × VEGF bispecific antibody called ivonescimab, and it recently produced incredibly strong phase three results in China in a head-to-head lung cancer trial — improved progression-free survival compared to Keytruda plus chemo, which marks the very first time a drug has beaten Keytruda on its own home turf.
MATT: Yeah. There's tremendous BD excitement right now around this target, but I think it should be tempered. For one, we're talking about PFS changes rather than OS, which is important to clarify. Secondly, the global phase three trial for ivo, I think, missed its OS endpoint, and so there's that piece as well. And then I think something that the entire industry is struggling with right now to triangulate from is the degree to which clinical data that's generated in China translates to Western markets. We're still figuring that out.
ALEX: Yeah, exactly. We, of course, couldn't resist but ask Roy Baynes about this, and he had a pretty measured opinion on whether this story of PD-1 × VEGF in a bispecific is a really fundamentally new thing or just a retread of the greatest hits, so to speak.
ROY BAYNES [interview clip]: "The combination of a VEGF together with a PD-1 is really a two-part question. The first one is, is this sum of the inhibition of those two agents beneficial? In other words, if you had just a PD-1 and just a VEGF, would the results be meaningful? And actually that experiment's been done many times — you know, with TKIs or with Avastin or with regorafenib. And the short answer is usually you see an improvement in PFS, but it generally doesn't translate into an improvement in overall survival. On first principles, sum of the parts probably is not gonna add a great deal."
MATT: So that was an interesting perspective. And here's another perspective that doesn't fully agree and gives this story the benefit of the doubt. A biological rationale for why a bispecific could be stronger than just giving the two drugs separately is — VEGF is broadly immunosuppressive in the tumor microenvironment. It induces PD-1 on those tumor-infiltrating T cells and promotes regulatory T cells. So blocking VEGF by itself doesn't just choke off the tumor's blood supply, but also could potentially help lift that brake on the immune response. And that only happens when they're tethered together for that activity.
ALEX: Yeah. I agree with Roy that it's telling that the combination treatment of the monospecific agents doesn't really do anything, but, like you, I am open intellectually to the notion that a PD-1 bispecific could act in ways that are unexpected, and that small changes to the pharmacology of PD-1 inhibition or the biodistribution of the molecule could ultimately translate into a situation where you've improved over a PD-1 monospecific — i.e., Keytruda. Time will tell. I think what's undeniable is that pharma is incredibly interested in these PD-1 × VEGF bispecifics. There's been a tremendous number of deals done in this space, and even Merck is starting to place bets here. They have their own Chinese in-licensed PD-1 × VEGF program that's been progressing.
MATT: Right. But from an objective point of view, everybody here is still reading the tea leaves, no pun intended. And it's still early days for this strategy of drugging PD-1.
--------------------------------------------------------------------------------
[01:55:31] THE SCORECARD: PATIENT, ACADEMIC, AND FINANCIAL IMPACT
--------------------------------------------------------------------------------
ALEX: Alright, we're getting to the end of the episode here. Just like for episode two on Geron, let's score Keytruda across three distinct dimensions: patient impact, financial impact, and academic impact. Alright, let's start with patient impact.
MATT: So yeah, I don't wanna beat a dead horse here. I think the term "revolutionary" and "game changing" is really overused in our industry, but this is a term that is completely appropriate for this drug. It really has changed the landscape of cancer care, and it really has increased life expectancy for people with certain cancers. I think the second piece here is 10-out-of-10 execution from Merck in terms of bringing this drug to patients at insane speed. I think a lot of credit should be given to Merck, should be given to the FDA, and should be given to the molecule itself for having, like, insane effect sizes, making it possible to move really quickly and get this drug to patients really quickly. So I give it an A-plus.
ALEX: I do as well. It feels like this drug crash-landed on Earth from outer space. I mean, as you said yourself, beating a dead horse here to say how much it's changed cancer care — I think this is, dare I say, A-plus-plus.
MATT: Yep. Academic impact. So this is a very interesting category. If you think about it, if anyone was really a pioneer here in the clinic, it wasn't Merck, right? Like, they were drafting off a lot of the learnings from BMS that was ahead of them. They benefited a lot from that. You know, PD-1 was not some completely esoteric target that Merck discovered and brought forward. Right? Like, in the case of Geron, we talked about them cloning the gene and going all the way from the basic science forward. This was not one of those situations. I think too, the act of making the drug Keytruda — a blocking antibody to a surface receptor — is, like, a pretty standard thing in the 2000s. Keytruda is a very good drug. So from, like, an academic science impact, giving them a lot of credit for that doesn't seem fair. What am I missing here?
ALEX: Well, I think to be generous to the contributions that Merck did — for the target is tested in so many patient populations, like basically throwing the kitchen sink at it to see where it would do well. And in that sense, the novelty in settings in which PD-1 has been approved in is in part Merck's responsibility, right? In microsatellite-instable patients, or taking contrarian bets on patient populations that weren't expected to work, like with Perlmutter betting on chemo being combined with pembro in NSCLC. I think it was also a bold bet, and there's reason to believe that took some intellectual academic courage.
MATT: Yeah, you make a good point that they really contributed a lot with the biomarker strategy. I'm sort of surprised the degree to which they have not led the charge to better understand why PD-1 inhibition works. It seems like they would have a lot of motivation to do that. And the other thing to highlight is the financial success of Keytruda has revitalized and mainstreamed the entire field of cancer immunotherapy research, and that all the money that has flowed into it is downstream of the success of this drug. So all in all, I don't know — I gave it a B, maybe.
ALEX: I think I'm on board with that, for reasons that are really pedantic — that being that they technically didn't contribute to a lot of the early understanding. I would give more credit to BMS in that regard. Right. Alright. The final one — financial impact. This is another really interesting one where we can think about financial impact on a couple different vectors and the answer's pretty different. Maybe to start off with the easy one — for Merck, this is another A-plus-plus sort of situation, right? This drug completely revitalized and revamped the organization. The merger with Schering-Plough totally paid off from just this one drug. It turned them into an oncology powerhouse. When you think about subcutaneous Keytruda, if we just accept that it does $7 billion a year in sales, that is pure margin with almost no technical risk there. The ROI on a drug like that is gonna be enormous, right?
MATT: Yep. Just to double-click on how good the Merck/Schering-Plough purchase was — we had talked about how Merck almost sold Keytruda for nothing. They actually had another mega-blockbuster drug out of that merger that they did in fact sell for almost nothing, right?
ALEX: Which was that?
MATT: It was, you know, another Organon drug. They sold it for a thousand dollars upfront.
ALEX: Oh my God.
MATT: It was the BTK inhibitor, Calquence. I think, you know, they do a couple billion a year.
ALEX: Oh, no kidding.
MATT: They talk about this in "For Blood and Money."
ALEX: Yeah. Like, even throwing away really good stuff from that merger, it's still totally paid off. And if you look at the stock, the stock has been in a straight line up since 2012. It's tripled. And that's despite — like many pharmas, it pays out a very healthy dividend every year, and they've increased their dividend nicely. So Merck and the shareholders are very happy.
MATT: Agreed. But in all fairness, I think everybody in this space has, by definition, been a winner because the pie is so big that this has opened up. So the drug that it famously beat — right, nivolumab — that does $10 billion a year. So BMS isn't exactly getting pennies for every dollar that Merck makes. It's still a huge category with a lot of winners. Another couple of PD-1, PD-L1s — Tecentriq and Imfinzi — those do blockbuster numbers. There are a lot of second, third place entrants that are doing really well in revenue.
--------------------------------------------------------------------------------
[02:01:19] WHO ACTUALLY PROFITED: MERCK VS. ORGANON VS. THE SCIENTISTS WHO BUILT THE DRUG
--------------------------------------------------------------------------------
ALEX: Yeah, exactly. This is just such a monster category that being really late to the party — you're not making $30 billion a year, but you're doing pretty well. And again, from a risk-adjusted basis, you're dealing with pretty de-risked biology there. It's interesting to think about — like, the biggest loser of this financially was, like, kind of Schering-Plough and Organon, right? Like, these were incredible drug R&D organizations that within the walls of their labs, they discovered Keytruda, they discovered Calquence. We talked about that anesthesia reversal drug that was also discovered at Organon. So they had all of these really good drugs, and a subset of these were effectively valued at zero when they were sold to Merck. In many ways, Schering-Plough/Organon — they should be the big surviving pharma, right? Not Merck.
MATT: So it's just funny to think about — well, all things considered, the entity is called Merck-Schering-Plough, is that right? The legal — it's called Merck-Schering-Plough's… No more, man.
ALEX: Yeah, no, you're right. Yeah. I mean, that's the thing — like, Merck is this hundred-year-old organization, but as we heard the story of from Roy, it changed so much from Keytruda — a drug that was not even discovered there. It's like you sort of wonder if it's just — this organization is very long-lived in name, but it's probably tremendously changed many, many different times as the pipeline changes. Sure. The Ship of Theseus of Merck.
MATT: So do you wanna talk a little bit about the interesting dynamic between being one of the early inventors and developers on this program compared to being at Merck in the late-stage clinical development of this drug, in terms of the payoff?
ALEX: Yeah. I think it's worth talking about that, especially as in the news right now you're hearing about fundamental researchers in AI getting paid tens or hundreds of millions of dollars to work for specific companies on effectively pretty early research projects. The work on pembro has driven to date hundreds of billions of dollars of profit and market cap. And the scientists who pushed that all forward, speaking quite generally, they got paid the standard low-to-mid six-figure salaries that R&D professionals make, and from an equity perspective, participated very marginally in the upside that came from these drugs. We have a nice little vignette — which was, one of the scientists involved here, when Keytruda got to the development candidate stage, can you guess how much money they got from Schering-Plough for sort of reaching that milestone?
MATT: I could only guess — not that much.
ALEX: A couple hundred dollars worth of, uh, Schering-Plough stock. That is the reward for pushing forward a $100 billion drug. I think in the past, when these conversations have come up and people have said, like, how come the R&D scientists are not participating more in the spoils of drug discovery? It's usually an argument along the lines of, well, you know, drug discovery is really hard. It takes a lot of money. It's really long, and it's risky, especially compared to tech companies. And I think what's interesting in this story is, like — Keytruda went warp speed from discovery to approved drug. Right? I'd need to do the math, but the antibody campaign to approval went faster than it took OpenAI to go from founding to first revenue, let alone first product, right?
MATT: I think it's interesting that there's these AI companies spending billions now in R&D costs and lots of years to develop products, and are nowhere near positive cash flows. It's interesting that those companies, for a variety of different reasons, are sort of, like, sharing the generational wealth with the employees, and that for a variety of different reasons, it's just not what happens in drug discovery.
ALEX: So anyways, yeah, it's an unfortunate thing that just so happens in this industry. The value inflection where the revenue is determined is essentially on the clinical studies themselves, and that's where the spoils of war are earned, I suppose. But I think that sets it for this playbook section. Then, A-plus for patient impact, academic impact — we give a B with some nuance, and then financial impact, we give A-plus-plus, again with some nuance, depending who you're asking.
MATT: Yep. I agree with that.
--------------------------------------------------------------------------------
[02:05:54] EPILOGUE: THE JIMMY CARTER DRUG
--------------------------------------------------------------------------------
ALEX: In August 2015, former president Jimmy Carter disclosed that he had metastatic melanoma that had spread to his brain. He was 90 years old at the time. At a press conference, Carter famously told the world he was "surprisingly at ease" with whatever came next. But his medical team didn't just send him home and hope for the best. They treated him with a combination of radiation therapy and a newly approved revolutionary drug — pembrolizumab. And the results were nothing short of historic. By December of that exact same year, his brain lesions were completely gone. He continued his treatment and went on to live for another entire decade, celebrating his 100th birthday.
MATT: The media dubbed pembrolizumab "the Jimmy Carter drug." The nickname stuck. It did more for public awareness of immunotherapy than any scientific paper or clinical trial or podcast episode ever could.
ALEX: It was a single anecdote, but it was the perfect anecdote at the exact right time. You had a beloved public figure handed a grim, terminal prognosis who essentially walked away from it because a drug worked simply by unleashing his own immune system. It's the ultimate testament to the work of everyone we talked about today — from Lieping Chen to the Organon holdovers to the clinical teams at Merck.
MATT: Thanks for listening to Approved. We'll see you next time.
ALEX: We'll see you next time.
================================================================================
END OF TRANSCRIPT
================================================================================