Keytruda: One Molecule, 80 FDA Approvals

Seven years of quiet science

Keytruda was not invented at Merck. It was created at Organon, the pharmaceutical division of Dutch chemical conglomerate Akzo Nobel, in Oss, Netherlands. Starting around 2003, a team including Greg Carven, Hans van Eenennaam, and Joh Dulos built a blocking antibody against PD-1 as part of a broader immune regulation program. It was rational drug design aimed at a target whose clinical value was unclear. PD-1 biology had been identified academically - Honjo's lab in Kyoto had shown in 2005 that blocking PD-1 could stop tumor spread in mice - but human validation was nonexistent. There were no validated biomarkers, no clear indication strategy, no precedent for durable responses through immune modulation.

The pharma industry at the time was focused elsewhere: kinase inhibitors, targeted antibodies against HER2, EGFR, and VEGF. A PD-1 program was scientifically interesting but commercially ambiguous.

The work was filed as patent WO2008156712A1 with a priority date of June 2007 under Organon's name - the only public trace of the molecule before it had a brand name.

Then the molecule got buried in corporate integration. Schering-Plough acquired Organon in 2007. Merck acquired Schering-Plough in 2009. It is entirely normal for early-stage assets to get deprioritized in a single re-org without a champion at a senior level, let alone two where leadership turns over completely. MK-3475 had no human data and was competing with dozens of clearer assets. In most large pharma integrations, this is exactly where programs die.

What changed was not internal - it was external, and it traces back to one person's persistence. James Allison, a Texan immunologist who lost his mother to lymphoma at age 11, had shown in 1996 that blocking a different checkpoint - CTLA-4 - could cure cancer in mice. He spent years trying to convince pharma companies to develop it. All the major companies passed. Eventually a small biotech called Medarex took it on, and Bristol-Myers Squibb acquired Medarex in 2009. At ASCO 2010, BMS presented the ipilimumab Phase III data: the first therapy ever to show an overall survival benefit in metastatic melanoma. 10 months versus 6.4 months. It was not a huge effect. But it was real, and it was the first time checkpoint inhibition had worked in humans.

That single dataset changed the equation for every PD-1 program in the industry. Early PD-1 data was suggesting better tolerability with even stronger responses than CTLA-4 blockade. The shift was categorical: "immune therapy sometimes works" became "checkpoint inhibition is a viable drug class."

Merck moved fast. MK-3475 was promoted from a background exploratory asset into a serious oncology program. The IND was filed in 2010. The first-in-human trial - KEYNOTE-001 - started soon after. Merck did not invent PD-1, and they did not prove it first. But they recognized the inflection point and mobilized without hesitation. The drug had existed for years before the opportunity existed. Once validation arrived from outside, the company that reacted fastest won.

If you listen to Guy Raz's How I Built This, he asks every founder how much of their success is luck versus hard work. Drug development is the same question at a higher stakes table. Every step involves decisions made under enormous uncertainty - which targets to pursue, which indications to test, when to accelerate, when to wait. Merck made the right call on Keytruda in 2010. They also had the molecule in the building because of two acquisitions they did for entirely different reasons. Skill and luck are hard to separate when the outcome is $32 billion a year.

80 approvals in 11 years

Keytruda's first FDA approval came in September 2014, for advanced melanoma. Since then, the FDA has issued roughly 80 approval supplements for Keytruda - new indications, new combinations, new lines of therapy, expanded patient populations, and updated dosing regimens. Not all 80 are distinct cancer types. Many are expansions within the same cancer - moving from second-line to first-line treatment, adding a combination partner, or broadening the eligible patient population. The result is 20 named cancer types plus two tissue-agnostic biomarker approvals, built through relentless clinical execution.

Each new cancer type took a year or two to show up in the revenue. Oncologists adopt carefully. Guidelines need updating, payer coverage needs confirming, treatment protocols need rewriting. The 2017 approvals - seven in one year, including the first tissue-agnostic approval in FDA history - produced a revenue surge that hit in 2018 and 2019.

Why one drug treats 20 cancers

Most cancer drugs target the tumor. Keytruda targets the immune system.

T-cells are the immune system's killers. They patrol the body looking for cells that display abnormal proteins - infected cells, damaged cells, cancer cells. But T-cells also carry safety brakes to prevent them from attacking healthy tissue. One of those brakes is a receptor called PD-1 (programmed death-1), discovered in 1992 by Tasuku Honjo at Kyoto University.

PD-1 exists for a good reason. Without it, the immune system would attack the body's own organs - which is exactly what happens in autoimmune disease. The problem is that tumors learned to exploit this brake. Cancer cells display a ligand called PD-L1 on their surface that binds to PD-1 on T-cells and sends a "don't attack me" signal. The T-cell that should be killing the tumor stands down instead.

Pembrolizumab is a monoclonal antibody that physically blocks PD-1 on the T-cell surface, preventing PD-L1 from binding. The brake is released. The T-cell recognizes the tumor and attacks.

The molecule

Pembrolizumab is a humanized IgG4 antibody. It started as a mouse antibody - clone J105, generated at Organon - and was then humanized by grafting the mouse binding regions (CDRs) onto a human antibody framework. That humanization work was done by LifeArc (then MRC Technology), a UK medical research nonprofit. A mouse antibody, humanized in the UK, owned by a Dutch company, eventually developed by an American one.

The IgG4 subclass was chosen deliberately. Most therapeutic antibodies are IgG1, which activates complement and recruits natural killer cells to destroy whatever the antibody binds to. That is useful when the target is a cancer cell. It is not useful when the target is a T-cell you want to keep alive. IgG4 binds its target without triggering cell destruction - it blocks PD-1 without killing the T-cell that carries it.

The antibody has a half-life of about 26 days, which allows dosing every three weeks (200mg) or every six weeks (400mg). This matters in oncology: fewer infusions means fewer clinic visits, less chair time, and better quality of life for patients who may be on treatment for years.

Why this explains 20 cancers

Because Keytruda targets the immune system rather than the tumor itself, it works wherever cancers use the PD-1/PD-L1 escape route. Melanoma, lung, bladder, liver, kidney - different organs, same immune evasion strategy. The drug doesn't need to know anything about the cancer's biology. It only needs the cancer to be hiding behind PD-L1.

This is what makes checkpoint inhibitors fundamentally different from targeted therapies like imatinib (which targets a specific mutation in a specific cancer) or trastuzumab (which targets HER2 overexpression). Those drugs are precise but narrow. Keytruda is broad because the mechanism it exploits - immune evasion - is universal across cancer types.

The cost of that breadth is side effects. Releasing the immune system's brakes on cancer also releases them on healthy tissue. Patients on Keytruda can develop immune-mediated inflammation of the thyroid, liver, lungs, skin, or gut. The same mechanism that treats 20 cancers can attack 20 organs. Managing these side effects is a significant part of oncology practice with checkpoint inhibitors.

The Nobel Prize

Tasuku Honjo and James Allison shared the 2018 Nobel Prize in Physiology or Medicine. They never collaborated. Honjo worked on PD-1 in Kyoto, Allison on CTLA-4 in Texas. Twenty-six years passed between Honjo's identification of PD-1 (1992) and the Nobel. Twenty-two years between Allison's mouse cure (1996) and the prize. Allison, who plays harmonica in a band called The Checkpoints, once said every pharma company turned him down. Honjo's discovery was pursued commercially by Ono Pharmaceutical in Japan, which licensed PD-1 antibody rights that eventually reached Merck.

Honjo and Merck ended up in a licensing dispute over the intellectual property behind the discovery. The terms of what Honjo and Ono were owed became contentious. The scientist whose work underpins $32 billion a year in revenue had to fight for his share of it.

May 2017: the approval that changed oncology

On May 23, 2017, the FDA did something it had never done before. It approved Keytruda for any solid tumor - regardless of where in the body it originated - that had a specific genetic feature: high microsatellite instability (MSI-high) or mismatch repair deficiency (dMMR).

This was the first tissue-agnostic approval in FDA history. The agency was no longer asking "what organ is the cancer in?" It was asking "what does the cancer's DNA look like?"

MSI-high tumors have a broken DNA repair mechanism that produces many mutations. More mutations mean more abnormal proteins on the tumor surface - more targets for the immune system to recognize. Keytruda releases the immune system's brakes, and MSI-high tumors give it plenty to attack.

The practical impact was enormous. A colorectal cancer patient, an endometrial cancer patient, and a gastric cancer patient could all receive the same drug if their tumors shared this biomarker. The approval compressed what would have been years of indication-by-indication clinical trials into a single regulatory decision.

In 2020, the FDA went further. It approved Keytruda for any solid tumor with high tumor mutational burden (TMB-high) - a second tissue-agnostic approval. Two biomarker-based approvals meant Keytruda could potentially treat dozens of cancer subtypes beyond its 20 named indications.

Every major oncology company redesigned its clinical trial strategy after 2017. Biomarker-selected, tissue-agnostic trials went from exotic to expected. Keytruda didn't just treat cancer differently - it changed how the industry develops cancer drugs.

The de facto standard

As of April 2026, there are roughly 2,800 clinical trials on ClinicalTrials.gov involving pembrolizumab. More than 1,350 are active. Over 730 are recruiting patients. Nearly 350 are Phase 3.

Merck sponsors roughly 240 of these. The other 80% are run by other companies and academic institutions - not because Merck asked them to, but because Keytruda has become the baseline that the rest of oncology measures itself against.

AstraZeneca, Lilly, Amgen, Daiichi Sankyo, Roche, and Seagen/Pfizer are all running active trials that use pembrolizumab - either testing their own drugs in combination with it, or using it as the comparator arm that a new therapy must beat. BMS, the company that pioneered checkpoint inhibition, sponsors trials that test against Keytruda as the standard of care.

Academic cancer centers drive over half of all pembrolizumab trial activity. MD Anderson, Memorial Sloan Kettering, Dana-Farber, Mayo Clinic, City of Hope - these institutions have built entire research programs around this molecule. The NCI alone sponsors 48 active trials.

Keytruda is no longer just Merck's product. It is infrastructure - the molecule the field runs on. That makes the 2028 patent expiry not just a corporate event but a systemic one. When the economics of pembrolizumab change, the economics of every trial designed around it change too.

Two drugs, one target, different paths

Bristol-Myers Squibb pioneered this entire field. BMS developed ipilimumab (Yervoy) - the first checkpoint inhibitor to show a survival benefit in cancer. Their PD-1 inhibitor, Opdivo (nivolumab), was approved for melanoma in December 2014, three months after Keytruda. Without BMS's early work on checkpoint inhibition, the category might not exist.

The two drugs diverged on first-line lung cancer - the largest oncology market in the world. Both companies ran pivotal trials. They made different design choices.

Merck's KEYNOTE-024 enrolled only patients whose tumors had high PD-L1 expression (TPS of 50% or greater) - a bet on biomarker selection. BMS's CheckMate-026 enrolled patients with PD-L1 expression of 1% or greater - a broader approach that could have captured a larger market.

KEYNOTE-024 met its primary endpoint. CheckMate-026 did not. The biomarker enrichment strategy paid off for Merck in this indication.

That trial result shaped how the two drugs were adopted in lung cancer. But Opdivo built its own franchise - combinations with Yervoy, strong positions in liver cancer, renal cell carcinoma, and other indications. At $9.3B in 2024, it is one of the most successful oncology drugs ever developed.

Eleven drugs target PD-1 or PD-L1. Keytruda holds roughly 70% of the market. That concentration - one drug dominating a class with ten alternatives - is unusual in oncology and makes the 2028 cliff even more significant.

Want the next brief? Subscribe — free, every week.

December 2028

Keytruda's US compound patent expires on December 31, 2028. European supplementary protection certificates extend to mid-2030. Japan extends to late 2031.

The numbers are difficult to process. Keytruda generated $31.7B in 2025 - 49% of Merck's total company revenue of $65.0B, including pharmaceuticals, vaccines, and animal health. Merck's second-largest product is Gardasil at $8.6B - less than a third of Keytruda. No major pharmaceutical company has had this much revenue riding on a single molecule.

The biosimilar programs are already well underway. Our trials database tracks 14 active pembrolizumab biosimilar trials across eight companies. The race is global.

Sandoz has already completed its Phase 3. Samsung Bioepis and Formycon/Zydus expect primary completion in 2026. With FDA review taking roughly 12 months, the earliest US biosimilar approvals could come in 2027-2028 - but Merck holds 237+ patents on pembrolizumab, and the core patents don't expire until December 2028. No public patent settlements between Merck and any biosimilar developer have been announced. Most analysts expect actual US biosimilar launches in 2029 or 2030.

The first pembrolizumab biosimilar launched in Paraguay in August 2025. Vietnam followed in November. Keytruda's current US list price runs roughly $200,000 per patient per year. Biosimilar pricing is expected at a 40-60% discount, consistent with precedents set by oncology biosimilars like trastuzumab and bevacizumab.

The expectation across the industry is a 70-80% revenue loss by 2031. Opdivo's patent expires the same year. The two largest checkpoint inhibitors - over $40 billion in combined annual revenue - go biosimilar simultaneously. Nothing at this scale has happened before in oncology.

Merck's defense

In September 2025, the FDA approved Keytruda Qlex - a subcutaneous formulation of pembrolizumab combined with hyaluronidase. Same drug, different delivery. Instead of a 30-minute IV infusion in an oncology center, Qlex is a subcutaneous injection that takes 3-5 minutes.

This is lifecycle management. Merck is betting that oncologists and patients who switch to the subcutaneous version will stay on it even after IV biosimilars become available. The convenience gap matters - a 5-minute injection versus a 30-minute infusion plus chair time, scheduling, and clinic overhead. But there is a constraint: Qlex is currently approved for only 2 of Keytruda's 22 cancer types - melanoma and NSCLC. Each additional indication requires its own regulatory filing. Merck has to expand Qlex across 20 more indications before IV biosimilars arrive for all of them.

It is the same strategy that companies have used before patent cliffs - Enbrel extended its franchise for years with a combination patent strategy. Whether it works for Keytruda depends on how aggressively payers push physicians toward cheaper IV biosimilars.

Merck is also building beyond Keytruda. Winrevair (pulmonary arterial hypertension, launched 2024) reached $1.4B in its first full year. Lagevrio, Lynparza, and Welireg add smaller but meaningful revenue. But none of these are Keytruda-scale. Nothing is.

20 cancers after the cliff

The patent cliff conversation almost always focuses on what happens to the company. The stock price, the revenue gap, the M&A strategy. There is a different question.

Keytruda is used across 20 cancer types. Some of those indications - advanced melanoma, first-line NSCLC, MSI-high tumors - represent hundreds of thousands of patients per year worldwide. When biosimilar pembrolizumab becomes available at a fraction of the price, does access expand?

In theory, yes. Cheaper checkpoint inhibitors could open treatment in countries where Keytruda's pricing puts it out of reach. In sub-Saharan Africa, Southeast Asia, and parts of Latin America, PD-1 inhibitors are functionally unavailable. Biosimilar pricing could change that.

In practice, it is more complicated. Pembrolizumab biosimilars are IV-administered oncology drugs that require infusion infrastructure, oncologist oversight, and biomarker testing to identify eligible patients. The barriers to access are not just price. They are infrastructure, training, and diagnostic capacity. A cheaper drug does not help if there is no oncology center to administer it and no PD-L1 test to identify who should receive it.

Keytruda went from a discarded program valued at nothing to the highest-revenue drug in pharmaceutical history. It treated 20 cancers. It changed how the FDA approves drugs. It made Merck dependent on a single molecule for half its revenue.

December 2028 is not just a financial event for Merck. It is the beginning of an experiment in whether the most important cancer drug of this generation can reach the patients who have never had access to it.