Merck Already Survived Its Worst Drug. The Best One Is the Real Danger.

In 2004, a single drug nearly broke Merck. Vioxx was pulled from the U.S. market on September 30 of that year, after a trial showed it raised cardiovascular risk past the eighteen-month mark.³ More than 80 million people worldwide had taken it.⁴ The lawsuits ran into the tens of thousands; the eventual personal-injury settlement reached $4.85 billion, and the Justice Department later extracted another $950 million.⁵⁶ It is remembered as one of the great corporate near-death experiences in pharma. And Merck walked away from it. That is the comforting part of the story — and it is the part that misleads.

The official lesson runs like this: Merck endured Vioxx, therefore Merck can endure anything — including the patent cliff bearing down on its blockbuster, Keytruda. It's a reassuring analogy. It is also backwards. Vioxx was survivable precisely because, in revenue terms, it was small. Keytruda is not small. Keytruda is the company.

Here is the thesis, stated plainly: Merck's dependence on Keytruda is a structurally larger threat than Vioxx ever was, because a scandal you settle is a one-time wound, but a patent cliff under your single largest product is the permanent loss of the engine. The crisis that made headlines was the easy one. The crisis that doesn't make headlines — yet — is the dangerous one.

The drug that almost sank Merck was a rounding error next to today's

Vioxx's annual sales topped $2.5 billion at peak.⁴ That is a real number, but it was never the whole company. The damage from Vioxx came not from losing the revenue line but from the liabilities attached to it — the injury claims, the criminal probe, the reputational hit. Those were enormous, but they were bounded. Merck recorded the $4.85 billion settlement charge in a single year and moved on.⁵ A settlement is a check you write once. Keytruda's exposure is a different category of problem entirely.

In 2024, Keytruda generated $29.482 billion — an 18% jump — against total Merck revenue of $64.168 billion.² That is 46% of everything the company sold, and Merck's own 10-K states that the oncology portfolio Keytruda leads accounted for substantially all of its revenue growth.¹ One molecule is now nearly half the business and effectively all of its forward momentum. Vioxx at its worst was a scandal layered on top of a diversified company. Keytruda is the floor the company is standing on.

A cliff is not a lawsuit — and that's exactly why it's worse

The mechanism of the Vioxx crisis was a one-time accounting event: discover the harm, withdraw the drug, settle the claims, take the charge. Painful, finite, over. The mechanism of a patent cliff is the opposite. Keytruda's core U.S. composition-of-matter patent expires in 2028, and biosimilar makers — Samsung Bioepis, Amgen — are already building toward entry.⁷ When biosimilars arrive, they don't sue Merck for damages. They simply take the market, year after year, at a lower price. There is no settlement that ends it, because there was no wrongdoing to settle — just the ordinary expiry of a monopoly the law always intended to be temporary.

Merck knows this, which is why it is doing exactly what an incumbent does when the clock runs out: it is building a wall around the date. Watchdog group I-MAK counts nearly 300 patent filings on Keytruda and over 100 granted patents — a thicket designed to make biosimilar entry slower and more contested than the single 2028 headline suggests.⁸ The headline-grabbing piece of that strategy is the subcutaneous version, Keytruda Qlex, approved in September 2025 across 38 solid-tumor indications.⁹ A shot you can give in minutes, rather than an IV infusion, is more convenient — and, not incidentally, governed by its own newer patents. The IV biosimilars can flood in at 2028; the convenient version they can't yet copy is meant to keep the franchise breathing past it.

But didn't Merck rebuild before — and won't the thicket hold?

The fair objection is that this is too neat. Merck has done this before: it lost Zocor, it lost Fosamax and Singulair and Cozaar, and the patent cliff is the oldest¹², most-anticipated event in the pharmaceutical playbook. Companies don't sleepwalk into 2028 — they spend the intervening years acquiring, licensing, and reformulating to soften the drop. The Qlex subcutaneous approval and the patent thicket are evidence the planning is already underway, not evidence the company is helpless.⁸⁹ All of that is true, and none of it dissolves the problem. The honest counter is one of scale, not of competence.

No company in pharmaceutical history has had to replace $29.5 billion of high-margin revenue concentrated in one product on a known deadline. A 46% dependency means that even a strong replacement pipeline has to do extraordinary work just to stand still. The patent thicket can delay biosimilars; it cannot repeal the cliff, and the same I-MAK filings that may buy time are simultaneously inviting the regulatory and political scrutiny that 'product-hopping' to a subcutaneous shot tends to attract.⁸ Merck surviving Vioxx proved it could absorb a bounded shock. It proved nothing about absorbing an unbounded one. The analogy people reach for is comforting precisely because it lets them skip the arithmetic.

There's a tidy irony in the comparison. The Vioxx crisis was a company doing something it shouldn't have — promoting a drug for unapproved use, as the DOJ later established when it extracted that $950 million.⁶ The Keytruda problem is a company doing everything right: building the world's best-selling cancer drug, winning approval after approval, 40 indications and counting.¹¹⁹ Success this concentrated is its own form of fragility. Merck didn't make a mistake with Keytruda. It made the opposite mistake's mirror image — it built something so good that the company now depends on a single patent's expiry date the way a town depends on one factory.

Merck survived Vioxx because Vioxx was a wound. Keytruda is not a wound; it is the heart. And in 2028 the law that built the heart's monopoly does exactly what it was always going to do — it lets go. The lesson the company learned in 2004 was how to absorb a one-time blow. The lesson 2028 is about to teach is harder, and entirely new: that the most dangerous number on your books isn't the one in the lawsuit. It's the one in the revenue line you can't replace — sitting there at 46%, ticking.