2. UroToday Home
  3. Video Lectures
  4. mCRPC Treatment

International Analysis of Anti-PD-1/PD-L1 Efficacy and Biomarker Selection in Mismatch Repair-Deficient mCRPC - Niven Mehra

November 7, 2025

Niven Mehra discusses a multi-institutional study examining PD-1/PD-L1 immunotherapy efficacy in DNA mismatch repair-deficient metastatic castration-resistant prostate cancer. The study enrolled 93 patients selected by mismatch repair deficiency, microsatellite instability, or pathogenic mutations in mismatch repair genes. Results demonstrated median progression-free survival of 7.7 months, overall survival of 27 months, and objective response rate of 46%, with approximately 25% of patients experiencing durable responses. Dr. Mehra emphasizes that patients selected using multiple biomarkers performed significantly better than those identified by single tests. MSH2 and MSH6 represent the most commonly altered mismatch repair genes in prostate cancer. The discussion highlights that while only approximately 5% of metastatic castration-resistant prostate cancer patients harbor these alterations, this subset demonstrates substantial benefit from checkpoint inhibitors. Dr. Mehra stresses early testing and treatment initiation, as heavily pretreated patients show diminished responses.

Biographies:

Niven Mehra, MD, PhD, Clinical Scientist, Medical Oncology Consultant, Department of Medical Oncology, Associate Professor, Precision Medicine in Genito-Urinary Cancer, Radboud University Medical Centre, Nijmegen, Netherlands

A. Oliver Sartor, MD, Director, Transformational Prostate Cancer Research Center, East Jefferson General Hospital Cancer Center, Tulane University Cancer Center, New Orleans, LA

Related Content:

Efficacy of Anti-PD-(L)1 Immunotherapy in Patients with DNA Mismatch Repair-deficient Metastatic Castration-resistant Prostate Cancer - Beyond the Abstract

Unlocking the Future of mCRPC Treatment: Exploring Immune Checkpoint Inhibitors for MMR Loss and MSI-High - Sumit Subudhi

ESMO 2025: Optimized Use of Immunotherapy and Targeted Therapies in mCRPC
Read the Full Video Transcript

Oliver Sartor: Hi, I'm Dr. Oliver Sartor, and it's my pleasure on UroToday to be able to welcome Niven Mehra. Niven is at Radboud University in the Netherlands, a medical oncologist there who has led a number of important studies, and today we're going to be listening and learning a little bit more about PD-1/PD-L1 treatment in prostate cancer. Welcome, Niven.

Niven Mehra: Thanks, Oliver. Thanks for having me. So I would like to talk about a paper we have just published on the efficacy of anti-PD-1 and anti-PD-L1 immunotherapy in patients with DNA mismatch repair-deficient mCRPC, and this was a multi-institutional international study. And as you may be aware, immune checkpoint inhibitors in monotherapy have shown very limited efficacy, like response rates of about 5% to 6%, in unselected patients with mCRPC. And also, recently, we have seen many phase III trials with anti-PD-1 or anti-PD-L1 in combination with ARPIs, taxanes and even TKIs, which have also shown very limited efficacy. So actually, there was very limited future, I think, for immune checkpoints in unselected patients.

But nevertheless, in all these trials, you could see that there were super-responders, patients who had PSA declines of 90% or 99%, who had complete responses. And many of these patients have been examined, and many of these patients appear to have microsatellite-unstable prostate cancer. So why do the other patients not respond? That's probably due to limited neoantigens in prostate cancer, shown here. So it's a relatively low mutational burden disease. But also, prostate cancer is believed to have an immunosuppressive microenvironment.

So those patients who responded really well with mismatch repair deficiency, we know in prostate cancer, this is infrequent. So up to 5% of the patients with mCRPC appear to have mutations in mismatch repair genes, leading to a hypermutated phenotype and unstable microsatellites. And this is also a phenotype which gives rise to neoantigens, like frameshift mutations, which are highly immunogenic. So these patients we'd like to focus on.

So what we tried to do is we tried to fill a data gap, because there were only anecdotal patients published with prostate cancer who did really well on anti-PD-1, but there had not been a large study to show what the efficacy was. And we did have the KEYNOTE-158 study in non-colorectal cancers with MSI-high, there were 350 patients, of which only eight patients with prostate cancer, so I think this was indeed a data gap. So we tried to identify sufficient patients through an international study. So we identified 93 patients who were treated with checkpoint inhibitors; most of them were anti-PD-1. All of these patients were considered to be microsatellite-unstable. Many of these patients had multiple tests, so they needed to be either mismatch repair-deficient by immunohistochemistry, or they needed to have a likely or pathogenic mutation in a somatic or a germline mismatch repair gene, or have microsatellite instability by pentaplex PCR or larger NGS panels. Our primary endpoint was progression-free survival; our secondary endpoints were overall survival, PSA responses and objective responses.

So let's dive into the results. So this was our main endpoint. Overall, we saw that the median progression-free survival was about 7.7 months, and you could see a flattening of the curve. The one-year PFS rate was 39%—patients free of progression—and two years was 27% and three years 26%. And patients with visceral metastases appeared to do a bit worse. And for overall survival, you could see that the median overall survival was 27 months. Also, here, patients with less favorable performance status or a lower mutational burden also did a bit less well.

So in this cohort, our objective response rate was about 46%. As said, the progression-free survival was 7.7 months. And this is something what I would like to discuss also in relation to the KEYNOTE-158 study, which was actually quite good. This is better than we expected for prostate cancer, knowing that prostate cancer is—we expected it to have an immunosuppressive phenotype, maybe a worse response than the non-colorectal mismatch repair patients in the KEYNOTE study, so these are actually quite good results, I think. And these responses were also durable in about 25% of patients.

What I didn't show here is that the selection of the patients is important. So the results appeared best if you selected patients with multiple tests. So immunohistochemistry did best, but if you combined that with NGS data or even tumor mutational burden, you could identify patients with multiple tests—if they were positive—to do even much better than if you only selected patients with one test. So if you only select patients by NGS on ctDNA, those patients appear to do less well than patients who were selected on tissue. So I think there are still some uncertainties to discover.

As I said, this study has been published recently; it's available online in European Urology Oncology.

Oliver Sartor: Niven, I appreciate your overview. Let me ask you a couple of questions, because remember, we have a broad practitioner audience here, so you're an expert in prostate cancer, but not everybody who's listening is. Now, there are three biomarkers: there's going to be MSI-high, high TMB, and there's going to be mismatch repair deficiency. Now, let's talk about each one. What is MSI-high for those people who may not understand what it is?

Niven Mehra: Yeah, okay. So patients who are mismatch repair-deficient, so let me start there—patients who are deficient in repairing single bases, they get a hypermutated phenotype, and especially repetitive regions on the DNA, which are microsatellites, they get hypermutated, because those are areas which are very prone to accumulating mutations. So those areas get unstable, and that leads to microsatellite instability, and that's something which you can analyze by looking at these microsatellites. So you can do that, for instance, with PCR, looking at five microsatellites, but you can also do it by looking at whole-exome sequencing, looking at hundreds of microsatellite sites. Looking at more sites makes it more accurate in identifying if a patient is really truly mismatch repair-deficient. So that's MSI-high—that's something which you test with preferably next-generation sequencing, so DNA sequencing.

Oliver Sartor: That needs to be on tissue, or can you do it in ctDNA?

Niven Mehra: Yeah, you can also do it in ctDNA; this can be done—

Oliver Sartor: Is it reliable in ctDNA?

Niven Mehra: Yeah, so that was one of the things we looked at. So most of our patients were selected based on tissue; MSI was tested on tissue. But I think we had about eight to 10 patients who were selected by MSI on liquid biopsy. So what we saw is that the patients who were selected based on MSI by ctDNA appeared to do less well than the patients who were selected by tissue. So I don't know for sure, but it could be that there are different thresholds, or that for the patients with MSI detected in blood, it could also be a later event. Sometimes, we have seen also in databases where we looked at patients longitudinally over time, that you see in some patients who are very late-stage, end-stage prostate cancer, that they develop MSI and that could be subclonal. So I'm not sure why the patients who were selected by liquid biopsy did less well, but that was something that we saw, so we need to look into that in larger series.

Oliver Sartor: All right. You have mismatch repair; there are four mismatch repair genes, and we have PMS2, we have MLH1, MSH2 and MSH6. Which ones are the ones that are altered in prostate cancer? Because not all of them are. Help the readers understand, or listeners understand, a little bit more about what genes are actually involved.

Niven Mehra: Yeah. So this varies a little bit between different cancer types, and in prostate cancer, MSH2 and MSH6 are the most mutated genes. MLH1 is frequently inactivated by epigenetic mechanisms. And PMS2 was the least altered gene; only eight of our patients in the series had PMS2 alterations. So mainly, MSH2 and MSH6—those are the most predominant genes.

Oliver Sartor: Good. Now, you did not mention polymerase epsilon or POLE. Were there any of the POLE patients within this cohort?

Niven Mehra: No. So we did not select patients with high TMB only without mismatch repair deficiency. So I know in the US, there's a reimbursement for high TMB or mismatch repair deficiency, but in this paper, we only looked at the patients who were mismatch repair-deficient, so there were no POLE patients in there.

Oliver Sartor: Now, TMB, tumor mutational burden, there are different cutoffs for tissue versus ctDNA. Which ones were tissue, which ones were ctDNA, and where were the cutoffs for high TMB in your study?

Niven Mehra: Yeah. So we didn't select patients for TMB in our study, so we only selected our patients based on mismatch repair deficiency or MSI-high. But we did analyze the tumor mutational burden in relation to response in patients who were mismatch repair-deficient or MSI-high. So what we saw is that patients who had—if you use a threshold of 20 nonsynonymous mutations per megabase, we clearly could see that there was a difference in response rate and durability of response. So the patients who have mismatch repair deficiency and a low TMB do less well. I know there's a pan-cancer agnostic threshold of 10 mutations per megabase, so patients with prostate cancer can be treated with a checkpoint inhibitor. But if those patients are mismatch repair-deficient and have a TMB a bit above 10 to 20, probably these patients have a lesser chance of having a long-term durable response. That's what we saw.

Oliver Sartor: So in the end, we actually have three biomarkers: we have the MSI-high, the mismatch repair deficiency and the TMB. And you alluded to the fact that when you have all three, you do really quite well. If you just have one, you're less likely to do well. And if you have two, you're somewhere in the middle. Is that a fair take-home message when you're looking—

Niven Mehra: Yeah, yeah, yeah. We saw if you had all three features, your ability to respond was twice as high, so yes. What I would stress is that in our cohort, we also looked at patients who had mutations in mismatch repair genes, germline mutations or somatic. Some of these patients probably did not have a second hit which led to microsatellite instability. So I think that's one take-home message which is important. So if you have a Lynch syndrome carrier who develops prostate cancer, you need to test their tissue for MSI by immunohistochemistry, for instance, to make sure that this patient is truly deficient. So that's an important take-home message, because in our cohort, we did see in this real-world cohort that patients were treated with checkpoints with only a mutation, but when there's a mutation, we need to check if their phenotype also is mismatch repair-deficient.

Oliver Sartor: Got it, very important point. Well, listen, Niven, I appreciate you going over this important work. We all know that there's a subset of men who are going to respond nicely to PD-1 or PD-L1 inhibitors; you show that clearly in your study. This is not for all-comers, this is not for everybody who walks in the door. But for patients who have an MSI-high, a high tumor mutational burden and a mismatch repair deficiency, or any of the above, you're going to be looking at a pretty good chance of responding, and this is where the PD-1/PD-L1 inhibitors should be used in prostate cancer. Is that a fair statement?

Niven Mehra: Yes. So it's a small group, one in 20 patients, but these patients have a pretty good chance to have a durable response and may even be cured from their metastatic disease. So yeah, it is infrequent, but you need to test your patients. So that's something which I stress: these patients should be tested, and I think even early on. NCCN guidelines say you should test CRPC patients, and I think in the US, you can treat these patients after ARPI or taxane, but try to treat them as early as possible. We have seen also in post-hoc analyses that the more treated patients are, the less likely they are to benefit from checkpoint immunotherapy, so try to treat them as early as possible.

Oliver Sartor: Thank you. Very important take-home messages. Niven Mehra from the Netherlands, thank you for your work, thank you for your contributions in helping men with prostate cancer be able to be treated better with biomarkers.

Niven Mehra: Thank you, Oliver, and thanks UroToday for having me.