KEYNOTE-B49 Points to a PD-L1-Defined Subset Benefiting From Immunotherapy in Breast Cancer

CancerNetwork® spoke with Hope S. Rugo, MD, director of the Women’s Cancers Program, division chief of Breast Medical Oncology, and professor in the Department of Medical Oncology & Therapeutics Research at City of Hope Comprehensive Cancer Center, about interim and final analysis data from the phase 3 KEYNOTE-B49 trial (NCT04895358), which she presented at the 2026 European Society for Medical Oncology (ESMO) Congress.¹

The trial evaluated pembrolizumab (Keytruda) plus chemotherapy vs placebo plus chemotherapy in patients with endocrine-resistant, PD-L1–positive, hormone receptor (HR)–positive, HER2-negative metastatic breast cancer. Rugo discussed the trial’s design and the enrollment challenges that led to a reduced sample size, the progression-free survival benefit observed with pembrolizumab and its enrichment in patients with a PD-L1 combined positive score of 10 or greater, the factors that may have affected overall survival, and safety. She also discussed next steps, including the phase 3 TroFuse-010 trial (NCT06312176) of the investigational TROP2-directed antibody-drug conjugate sacituzumab tirumotecan plus pembrolizumab, considerations for combining immunotherapy with trastuzumab deruxtecan (Enhertu), and pending event-free survival data from the phase 3 KEYNOTE-756 trial (NCT03725059) in early-stage disease.

CancerNetwork: Could you provide some background on KEYNOTE-B49, its design, and why it is important for patients with breast cancer?

I presented data from both the interim and final analyses of the KEYNOTE-B49 trial.¹ This was a phase 3, randomized, double-blind trial that evaluated the addition of pembrolizumab to standard chemotherapy vs placebo plus chemotherapy in patients with endocrine-resistant, centrally confirmed PD-L1–positive, HR-positive, HER2-negative metastatic breast cancer.

The definitions of hormone therapy resistance were strictly defined in the protocol. Patients had to have received a CDK4/6 inhibitor or, if they hadn’t received a CDK4/6 inhibitor, have relapsed within the first 2 years of adjuvant endocrine therapy or progressed within 6 months of first-line endocrine therapy. The trial was designed to see whether we could define a population of patients with HR-positive, PD-L1–positive disease who benefited from the addition of pembrolizumab once they needed chemotherapy. PD-L1 positivity was defined as a combined positive score (CPS) of 1 or greater.

How did enrollment challenges shape the trial and its statistical power?

When we started screening and randomizing patients, it turned out that about 59% of screened patients had submitted tissue that was not PD-L1 positive, even using a cutoff of just 1. Of the rest, some patients didn’t have adequate tissue, and some had PD-L1–positive disease but didn’t fit other eligibility criteria. We ended up with very slow accrual. Overall, of all the samples we tested, 38% had a CPS of 1 or greater, and if you take away the patients who weren’t eligible, you end up with a pretty small number.

Because of this slow accrual, and because of competing interests, such as the idea that it might be better to give immunotherapy with an antibody-drug conjugate (ADC), the decision was made to reduce patient accrual from the originally planned 800 to [340]. That’s a big drop. What happened to the statistics was that we reduced the statistical power for both progression-free survival (PFS) and overall survival (OS). Power for PFS went from a little over 90% to a little over 80%, which is still adequate, but statistical power for OS dropped to [approximately 65%].¹

We then analyzed both an interim and a final analysis of the primary end point, which was PFS by blinded independent central review in patients with a CPS of 1 or greater. The secondary end points were OS in the same population, and PFS and OS in patients with a PD-L1 CPS of 10 or greater, which is, of course, the criterion used in triple-negative breast cancer.

What did the enrolled patient population look like?

The patient population is important. Almost all patients, 99%, had received a CDK4/6 inhibitor, and 85% had visceral metastases. When we looked at CPS, 79% had a CPS of 1 to 9, and only 21% had a CPS of 10 or greater. It’s an interesting population. Overall, we saw about half the rate of PD-L1 positivity compared with triple-negative breast cancer.

What did the primary PFS analysis show?

In the overall PD-L1–positive group, PFS was statistically significantly superior in patients who received pembrolizumab vs placebo in combination with chemotherapy. The hazard ratio (HR) was 0.74, which is great, but the difference in median PFS was small, going from 6.0 to 6.7 months. When we looked at the curves in both the interim and final analyses, we could see that there was a steep drop-off in the early part of the Kaplan-Meier curve, and then the curves separated over time.

We looked at the 24-month analysis of patients who were free from progression, and more than 3 times as many patients who received pembrolizumab were free from progression vs those who received placebo along with chemotherapy, in both the interim and final analyses. In the final analysis, that was a little over 3% in the placebo arm and about 14% in the pembrolizumab arm.¹ That identified a clear subset of patients whose disease is sensitive to pembrolizumab and who could benefit from this with a long duration of disease control.

How did patients with a CPS of 10 or greater fare?

Patients with a CPS of 10 or greater represented 36 patients in each arm, for a total of 72. In that population, we saw an enhancement of the PFS benefit, with the HR dropping to [0.59].¹ That’s an interesting difference in this small population. The proportion of patients free from progression more than doubled [at 12 months], so you’re seeing almost a quarter of the patients in this small subset who are free from progression at the [1-year] mark. We don’t normally see that with chemotherapy, so that further identified this sensitive group.

When we looked at duration of response, we saw a longer duration of response in patients with a CPS of 10 or greater, whereas it was relatively similar between the arms in patients with a CPS of 1 or greater.¹

How should clinicians interpret the OS findings?

OS was not different in either the CPS of 1 or greater or the CPS of 10 or greater populations. The power to detect a difference, even in the overall population, was very small, and it was much lower in the population with a CPS of 10 or greater.

We then looked at an adjusted analysis of OS, because it turned out that patients in the experimental arm were much less likely to receive subsequent therapy and less likely to receive an ADC. The difference was quite striking: there was almost a 13% difference in subsequent therapy favoring the placebo arm, and fewer patients received ADCs in the pembrolizumab arm. In the adjusted analysis, which again was not well powered, the HR went to just over 0.9, so it wasn’t 1. That suggests that the difference in subsequent therapies and the underpowering for OS impacted our ability to tell a difference.

What did the safety data show, and what is the overall takeaway from KEYNOTE-B49?

When we looked at toxicity, there was no difference from what we would expect when giving pembrolizumab with chemotherapy in any patient population, including triple-negative breast cancer. There were no new safety signals.

Overall, it tells us that there is a subpopulation of patients with HR-positive metastatic breast cancer who now have endocrine-resistant disease and who can benefit from immunotherapy. It’s a small population, and I’m hoping that we’ll see an enhanced benefit when we look at combining ADCs with immunotherapy vs chemotherapy in this population. That’s now being tested in the TroFuse-010 trial, with sacituzumab tirumotecan (sac-TMT), a TROP2 ADC, along with pembrolizumab, in both the overall population and in a PD-L1–positive population as a subanalysis.² It will be very interesting to see where we’re going in the future.

What role might immunotherapy play earlier in the course of HR-positive disease?

It’s really the same as in triple-negative breast cancer. Our greatest benefit from immunotherapy is in the early-stage setting, where we see enhanced host immunity, which gets suppressed by the burden of disease in the metastatic setting. We also see much higher rates of PD-L1 positivity in early-stage disease. The benefit of pembrolizumab was seen in the KEYNOTE-756 trial in the neoadjuvant setting in estrogen receptor (ER)–positive disease, regardless of PD-L1 expression.³ We also see enhanced benefit in patients who have more triple-negative–like disease, with ER expression of less than 10%, so that’s another patient population of great interest.

We’re waiting for event-free survival data from the neoadjuvant study, but we all have high hopes that it will show a benefit, even with the receipt of adjuvant endocrine therapy and CDK4/6 inhibitors. We’re slowly moving the needle forward in this patient population and in understanding the potential benefit of immunotherapy.

Now that we have the interim and final analyses of KEYNOTE-B49, where do we go next, and what questions remain unanswered?

We build on the data with the analysis of immunotherapy in combination with an ADC in patients in the first-line chemotherapy, endocrine-resistant setting. That’s being tested in TroFuse-010 with sacituzumab tirumotecan.² It’s very interesting to consider this with fam-trastuzumab deruxtecan-nxki (T-DXd; Enhertu), because it has such great efficacy in HER2-low, HR-positive disease. The complication there is understanding the pulmonary toxicity. In KEYNOTE-B49, we saw a rate of pneumonitis of 4.7%.¹ What do you do when you see pulmonary toxicity? You don’t know how to separate T-DXd from immunotherapy, so it complicates the issue.

It may be that in the future, we’re able to evaluate newer ADCs that have less pulmonary toxicity, or that there’s a subset of patients in whom we could look at T-DXd and carefully evaluate the pulmonary toxicity. For now, we’re sticking with TROP2 ADCs, and TroFuse-010 will be the first study to really give us that information.

Then, of course, we’re waiting for the event-free survival data from the neoadjuvant trial evaluating pembrolizumab in high-grade, all grade 3, high-risk, HR-positive early-stage breast cancer, where an improvement in pathologic complete response was seen with the addition of pembrolizumab.³

References

1. Rugo HS, Kok M, Goncalves A, et al. KEYNOTE-B49: a phase 3, randomized, double-blind study of pembrolizumab vs placebo plus chemotherapy in participants with HR+/HER2− PD-L1–positive advanced breast cancer. Presented at: 2026 European Society for Medical Oncology Congress; October 23-27, 2026; Madrid, Spain.

2. A study of sacituzumab tirumotecan (MK-2870) as a single agent and in combination with pembrolizumab (MK-3475) versus treatment of physician’s choice in participants with HR+/HER2- unresectable locally advanced or metastatic breast cancer (MK-2870-010). ClinicalTrials.gov. Accessed October 28, 2026. https://clinicaltrials.gov/study/NCT06312176

3. Cardoso F, O’Shaughnessy J, Liu Z, et al. Pembrolizumab and chemotherapy in high-risk, early-stage, ER+/HER2− breast cancer: a randomized phase 3 trial. Nat Med. 2025;31(2):442-448. doi:10.1038/s41591-024-03415-7