Immunotherapy alone beat chemo for a common type of womb cancer

Pembrolizumab on its own, with no chemotherapy, kept a common type of advanced womb cancer under control for far longer than standard chemotherapy in a phase 3 trial reported on 26 October 2026. In KEYNOTE-C93, people with mismatch repair deficient (dMMR) endometrial cancer who had the immunotherapy drip went a median of 38.2 months before their cancer grew or they died. With carboplatin and paclitaxel chemotherapy, that figure was 8.4 months.

The results were presented by Brian Slomovitz of Mount Sinai Medical Center in Miami Beach in a Presidential Symposium at the ESMO Congress 2026 in Madrid, and published the same day in The Lancet. Severe side effects were much less common with pembrolizumab. There is a catch, though. The trial did not show that people lived longer, at least not yet, and it compared pembrolizumab with chemotherapy alone. Today's usual first treatment for this cancer combines chemotherapy with an immunotherapy drug.

What is dMMR womb cancer?

Endometrial cancer starts in the lining of the womb (the endometrium). It is the most common type of womb cancer. Many cases are found early and treated with surgery, but some people are diagnosed when it has already spread, and in others it comes back after treatment.

Mismatch repair is the set of proteins a cell uses to fix spelling mistakes in its DNA when it copies itself. When one of these proteins is missing, the tumour builds up thousands of mutations. That is what dMMR means. The mutated proteins look foreign to the immune system, which is why dMMR cancers often respond strongly to checkpoint inhibitors such as pembrolizumab (Keytruda). About 25% to 30% of advanced or recurrent endometrial cancers are dMMR, according to OncLive's report of the ESMO presentation.

Some dMMR womb cancers are caused by Lynch syndrome, an inherited condition that also raises the risk of bowel cancer. In England, NICE diagnostics guidance DG42, published on 28 October 2020, recommends testing every endometrial cancer for Lynch syndrome. In practice that means a laboratory stain for the mismatch repair proteins on the tumour sample, so most people with womb cancer in the UK should already know their MMR status or be able to ask for it.

The same biology sits behind other results we have covered. Nivolumab plus ipilimumab in MSI-high bowel cancer and dostarlimab for dMMR rectal cancer both rely on these tumours being easy for the immune system to spot.

Who took part in KEYNOTE-C93?

The trial (NCT05173987) enrolled 299 people at 122 hospitals in 24 countries between February 2022 and December 2023. Everyone had endometrial carcinoma or carcinosarcoma that could not be removed by surgery, with dMMR status confirmed by a central laboratory. Nobody had received drug treatment for advanced disease. People who had chemotherapy around the time of earlier surgery could join if their cancer came back at least six months after the last dose.

The median age was 62 in the pembrolizumab group and 61 in the chemotherapy group. In both groups, 91.3% had endometrioid cancer, the most common microscopic type. Just over half (55.7% and 53.3%) had cancer that had come back after earlier treatment, and the rest were newly diagnosed with advanced disease. All were fully active or only mildly limited in daily life.

People were split at random into two groups:

It was an open-label trial, so patients and doctors knew which treatment each person was getting. Scans were judged by independent radiologists who did not know the group. Our guide to open-label and blinded trials explains how that can affect results like these. The two primary endpoints were progression-free survival (time until the cancer grows or the person dies) and overall survival.

How much longer was cancer held back?

At the data cutoff on 24 April 2026, median follow-up was 36.7 months. The Lancet paper reports these estimates:

The hazard ratio for progression or death was 0.57 (95% CI 0.42 to 0.79). Merck describes this as a 43% lower risk of progression or death. That relative figure hides an important pattern.

For the first six months, the two curves sat on top of each other, and if anything chemotherapy did slightly better early on. Slomovitz said the curves crossed at about six months. After that, the people on pembrolizumab pulled clearly ahead, because their responses lasted much longer. Responses to chemotherapy were just as common but faded, with a median duration of 10.2 months. For pembrolizumab, the median duration had not been reached.

Christian Marth of Innsbruck Medical University, who discussed the trial at ESMO, read the early part of the curve as a warning. He said it suggests roughly 20% of patients progress by the first scan, according to CancerNetwork's report. dMMR status predicts a better chance of response. It does not guarantee one.

A worked example for 100 people

Relative risks are hard to picture, so here is the same data turned into counts. Our guide to reading trial results uses the same method.

Imagine 100 people starting each treatment. At 12 months, 67 on pembrolizumab and 40 on chemotherapy would still be free of progression. That is 27 extra people. At three years, the numbers are 53 and 30, which is 23 extra people out of every 100.

Side effects run the other way. Serious (grade 3 or higher) treatment-related side effects hit 18 people per 100 on pembrolizumab and 51 per 100 on chemotherapy. So for every 100 people treated, about 33 fewer would have a severe treatment-related side effect. Hair loss shows the gap most clearly: 2 in 100 with pembrolizumab, 50 in 100 with chemotherapy.

There is a cost on the other side of that ledger. Pembrolizumab is given for up to about two years, against about four months for chemotherapy, so it means many more hospital visits for infusions and blood tests.

Did people live longer?

Not proven yet. Overall survival was the second primary endpoint, and at this interim analysis it missed the bar for statistical significance. The hazard ratio was 0.77 (95% CI 0.52 to 1.15), with p=0.10. Median survival had not been reached in either group. At three years, 70.7% were alive in the pembrolizumab group and 62.0% in the chemotherapy group.

A big reason is that most people on chemotherapy got immunotherapy later. Of the 88 people in the chemotherapy group whose cancer grew, 85.2% went on to receive an immunotherapy drug, many by switching to pembrolizumab within the trial. "In effect, the [OS] comparison is largely pembrolizumab first vs pembrolizumab second," Slomovitz said. Three extra analyses that tried to adjust for this switching gave hazard ratios between 0.55 and 0.66. These are exploratory and should not be read as proof.

The trial is still running. Merck says survival follow-up continues, and the registry lists an estimated completion date of 27 May 2027. Our explainer on what it means when a trial misses a goal covers how to read a result like this.

What were the side effects?

Treatment-related side effects of any grade affected 71.1% of people on pembrolizumab and 97.9% on chemotherapy. The most common with pembrolizumab were an underactive thyroid and tiredness. With chemotherapy they were hair loss, anaemia and nausea.

The Lancet paper lists these treatment-related side effects, each seen in at least a quarter of one group:

Immune-related side effects are the trade-off. Pembrolizumab releases the brakes on the immune system, which can then attack healthy organs such as the thyroid, bowel, liver or lungs. Immune-related side effects and infusion reactions were counted in 41.6% of the pembrolizumab group and 17.4% of the chemotherapy group, and were severe in 7.4% and 3.5%. Two people on pembrolizumab died from treatment-related causes (one from liver failure and one from bacterial meningitis), as did one person on chemotherapy (from a blood clot in the lung). Treatment was stopped for side effects in 9.4% and 14.6% of people.

Quality of life moved in opposite directions. At week 12, people on pembrolizumab reported better general health and quality of life than at the start, while those on chemotherapy reported worse, a 7-point gap between the groups. Physical and social functioning also favoured pembrolizumab.

How does this compare with treatment in the UK now?

This is where the result needs the most care. For newly diagnosed advanced or recurrent dMMR endometrial cancer, NICE already recommends immunotherapy combined with chemotherapy. The options include dostarlimab with platinum-based chemotherapy (guidance published 22 May 2025), durvalumab with platinum-based chemotherapy (9 September 2026) and pembrolizumab with carboplatin and paclitaxel (27 August 2025), which covers all mismatch repair types.

KEYNOTE-C93 did not test pembrolizumab alone against any of those combinations. Both the trial team and the discussant were clear on this. "[KEYNOTE-C93] does not show that pembrolizumab alone is equivalent to chemotherapy and immunotherapy," Slomovitz said in his presentation. Marth added that immunotherapy plus chemotherapy "results in a better oncological outcome than [pembrolizumab] alone, but with the price of a higher toxicity."

So the likely use is narrower. Both men described pembrolizumab alone as an option for people who want to avoid chemotherapy or who put tolerability and quality of life first, accepting that a combination may control the cancer better. That could include people who are especially worried about hair loss or nerve damage from chemotherapy.

One more detail is worth knowing. The benefit looked bigger in people whose cancer had come back after earlier treatment (hazard ratio 0.45) than in people newly diagnosed with advanced disease (hazard ratio 0.77, with a confidence interval that crosses 1). Subgroup results like this can be chance findings, so they need confirming.

Is pembrolizumab alone available for this?

Not as a first treatment, as far as we can find. In the US, pembrolizumab alone is approved for dMMR or MSI-high endometrial cancer only after earlier drug treatment. Merck says the KEYNOTE-C93 results "will be shared with regulatory authorities worldwide", and we found no public filing with the FDA, the MHRA or the European Medicines Agency for first-line use on its own.

UK hospitals took part. The trial registry lists five UK sites: The Christie in Manchester, the Beatson West of Scotland Cancer Centre in Glasgow, and St Bartholomew's, the Royal Marsden and Hammersmith hospitals in London. The trial is no longer recruiting. If you want to look for related studies, our guide on how to find and join a clinical trial covers the UK registries.

Questions to take to your cancer team

If you have advanced or recurrent womb cancer and are weighing up a first treatment, these questions may help:

1. Has my tumour been tested for mismatch repair, and is it dMMR or pMMR?
2. If it is dMMR, which immunotherapy and chemotherapy combination would you recommend, and why?
3. How would my chances compare if I had immunotherapy alone, given that KEYNOTE-C93 did not compare it with a combination?
4. Was my cancer newly diagnosed as advanced, or has it come back after earlier treatment?
5. Should I or my family be tested for Lynch syndrome?

This article is general information about a clinical trial. It is not medical advice, and treatment decisions should be made with your own specialist team, who know your full history.

Merck funded the trial, which was run with the GOG Foundation and the European Network of Gynaecological Oncological Trial Groups (ENGOT). Slomovitz reported consulting fees from MSD and other companies. The next firm answer on survival should come from the final analysis, expected after the trial's estimated completion in May 2027.