The ballad of TIGIT

There exist drug classes that seem, in retrospect, cursed. As these chemicals worm their way through the clinical trial system, they consume billions of dollars along the way, and squelch through thousands of sick patients. When finally it dawns on everyone how useless the whole endeavour was, the drugs life is at last cut short, nothing useful left in its destructive wake. The prototype here are amyloid-beta drugs. These are Alzheimer’s treatments that are widely perceived as immense disappointments, with the negative sentiment even leaking to the broader public. To be fair to these chemicals, the story here is a bit more complicated than the tabloids let on. Lots of amyloid research was not fake, and the drugs may genuinely be useful for early-stage Alzheimer’s. But they remain, regardless, disappointments. Beyond amyloid-beta, which has been steadily disappointing for awhile now, there is one other such category of drug whose particular dance has just recently wrapped up. It may very well someday gets its chance in the spotlight, but it will take time. Because it—just like every other chemical in this class—shares a searing, burning radioactivity. You should not touch them. You should not suggest touching them. In fact, no serious person should touch them for years to come, because to do so will be to receive the scorn of other serious people. What I am talking about are, of course, TIGIT drugs. TIGIT emerged in the wake of boundless enthusiasm from over a century of grueling cancer immunotherapy research. Much of this work went nowhere, but a small fragment of it helped produce the most valuable molecule in existence: Keytruda (pembrolizumab). This drug was so astonishingly, grossly successful that it would be barely an exaggeration to credit Keytruda with creating a Big Pharma. Since its approval in 2014, it has saved millions of years’ worth of patient lives, and will likely continue to save millions more. So, if you worked in pharma R&D in the mid-2010’s, and you were on the hunt for the next big thing, “something like Keytruda” was the most attractive thing on the board. And TIGIT drugs were supposed to be that.

An explanation of what TIGIT actually is would require you to hold roughly seven concepts in your mind at the same time, the names of which—in characteristic immunology fashion—are not helpful in the slightest. What is important to understand is that TIGIT is a particular protein, and theorized to be another immune-system brake. The aforementioned century of immunology research had already proven that these brakes mattered: Keytruda worked by blocking a different brake and allowing immune cells to attack tumors again. TIGIT seemed to offer the same promise, since tumors appeared to exploit it to quiet nearby immune cells. But this story was set to be even more intriguing. Unlike Keytruda’s target, TIGIT sat at an especially busy intersection of immune regulation. Blocking it might not merely release one brake, but rather two brakes and one accelerator, tilting the local immune environment towards such an absurdly anticancer direction that it was unthinkable that it wouldn’t be clinically effective. So, the theory went: block TIGIT, or create an ‘anti-TIGIT’ drug, and you’ve got something even better than Keytruda on your hands. Dollar signs appeared in the eyes of nearly every pharmaceutical executive upon learning this. Roche was the first here, their group establishing the above scientific observations, publishing them in a 2014 paper. "The immunoreceptor TIGIT regulates antitumor and antiviral CD8+ T cell effector function”. The molecule that emerged from this work was something called tiragolumab; the first anti-TIGIT drug to exist. Its initial clinical debut was at ASCO 2020, a major oncology conference. There, Roche discussed the results of a 135-patient phase 2 trial in metastatic non-small-cell lung cancer, randomizing patients one-to-one to the standard-of-care, plus either tiragolumab or placebo. The combination produced a response rate of 31% versus 16% in the placebo. Tiragolumab seemed to work. Yes, it wasn’t a cure for cancer, but neither was Keytruda, which still managed to rake in nearly ten billion dollars a year. The FDA granted tiragolumab a breakthrough designation in January 2021 on the basis of that study, and, within a year, Roche began to spin up phase 3 trials.

Blood was in the water for TIGIT, and though Roche was first to bite, others followed. Merck had vibostolimab. BMS had BMS-986207. BeiGene had ociperlimab, for which Novartis paid $300 million in early 2021 for co-development rights. Arcus had domvanalimab, for which Gilead in late 2020 paid $175 million up front plus a $200 million equity position in the company. iTeos, a little Belgian immuno-oncology outfit, had something called EOS-448, which GSK licensed in mid-2021 for $625 million upfront. Everyone wanted a bite and was willing to pay for it. Typically, with drug classes that have as much buzz as TIGIT did, companies like to run multiple trials in parallel, each one focused on a different cancer or patient subpopulation. This is to avoid a situation where your drug works spectacularly, it gets approved for the cancer subtype you tested it on, and then you have to watch on as your competitors’ drugs flood the remaining subtypes with their copycat chemicals. And the theoretical evidence for TIGIT was so strong, so overwhelming, that when combined with the promising phase 2 results, pushed Roche to go all in. At one point, they were running twelve concurrent Phase 2 and Phase 3 trials, each focused on a slightly different patient population, altogether covering ~5,000 human lives. This effort, which was branded ‘SKYSCRAPER’, represented one of the largest parallel-indication programs in modern immuno-oncology, its total costs likely running into the multiple billions. In May 2022, the first crack showed. Roche reported that its first major Phase 3 trial, in first-line small-cell lung cancer (SKYSCRAPER-02), had missed on progression-free survival, or PFS. But this was not the end of the world. Small-cell lung cancer is a rather miserable disease. Relatively little works here anyway. This subtype has swallowed a long procession of drugs that excelled in other settings, so this was not viewed so much as a failure as it was an admirable, Hail Mary attempt that was almost assuredly not going to work out anyway.

But a few weeks later came a bigger problem: Roche’s flagship lung-cancer trial (SKYSCRAPER-01), tested on an ostensibly curable type of lung cancer, also missed on PFS. To be clear: they did not miss it by a lot. Roche would spend the next two years insisting that the values were in the right direction, just not at statistical significance. Either way, the company demurred, the PFS metric is not what matters most. They were not wrong. PFS means something quite specific: from the start of the trial, how long did it take for a patient’s cancer to either grow on imaging or kill them. It is a useful data point, but it is ultimately a fuzzy surrogate of the metric that people actually care about: overall survival, or OS. How long did this patient live? Unfortunately, this metric takes years to read out and is confounded by whatever subsequent lines of therapy the patient picks up after the trial, so PFS is an often relied-on proxy metric. And Roche believed that OS would ultimately exonerate tiragolumab. And in August 2023, Roche ‘accidentally’ leaked data suggesting that overall survival had indeed improved on the drug. The stock ticked up on what in retrospect was the last uncomplicated moment of optimism in the TIGIT race. On November 26, 2024, Roche reported the final OS analysis. The flagship trial had missed. The survival trend had narrowed to the point of insignificance, and the trial that was supposed to anchor the entire program—the indication on which Breakthrough Therapy Designation had been granted, the signal on which ten other trials had been launched—could no longer be the anchor. But the flagship’s collapse wasn’t even the nadir. The nadir, really, was Roche’s worse-than-nothing readout in July 2024 (SKYSCRAPER-06), in the interim between the flagship’s PFS miss and its OS miss. It did not merely fail to show superiority to the standard of care, but was actively worse. Patients on tiragolumab died faster than the control group. Everything began to unwind from here on out for Roche. A planned follow-up was canceled before it had really begun, and another was deprioritized.

Over the subsequent year, the GI indications collapsed one after another, a locally advanced esophageal-cancer trial failed, a head-and-neck study was abandoned, and the last major Roche hope, in first-line liver cancer, missed PFS with no trend toward OS. The only success, awkwardly, was a trial in esophageal squamous-cell carcinoma (SKYSCRAPER-08), which produced statistically significant survival results. But by the time the full paper appeared in early 2026, Roche had already removed tiragolumab from its pipeline.The TIGIT game, for Roche, had ended. But what of the other players? Could it be that tiragolumab was the problem, and not the TIGIT hypothesis? Perhaps a different molecule, one still targeting TIGIT, would have worked. After Roche, Merck was the second biggest believer in TIGIT. Remember when I said Keytruda had almost single-handedly created a Big Pharma? Merck is that pharma. And with their patent over Keytruda set to expire in 2028, they were the ones most interested—and best positioned—to own its successor. In their exuberance, they decided to match Roche: twelve parallel trials of their own, each one running an anti-TIGIT drug called vibostolimab. The same pattern repeated. In May 2023, Merck’s melanoma trial was halted because vibostolimab was causing such a high rate of immune-related adverse events that patients were discontinuing therapy faster than any efficacy signal could accumulate. In August 2024, a small-cell lung-cancer trial was halted for OS futility, with the combination arm running worse than the control on both efficacy and safety. In December 2024, two more lung-cancer studies were abandoned halfway through the trial. And by 2025, Merck announced the discontinuation of the entire vibostolimab program. But there was one last hope. What if the biological story here wasn’t complete? What if the original Roche paper, a decade back at this point, had gotten something wrong? Every anti-TIGIT drug was structured like an antibody, a protein shaped like a ‘Y’.