Nivolumab for mismatch-repair-deficient or hypermutated gynecologic cancers: a phase 2 trial with biomarker analyses

Authors

Friedman CF

Manning-Geist BL

Zhou Q

Soumerai T

Holland A

Da Cruz Paula A

Green H

Ozsoy MA

Iasonos A

Hollmann T

Leitao MM Jr

Mueller JJ

Makker V

Tew WP

O’Cearbhaill RE

Liu YL

Rubinstein MM

Troso-Sandoval T

Lichtman SM

Schram A

Kyi C

Grisham RN

Causa Andrieu P

Wherry EJ

Aghajanian C

Weigelt B

Hensley ML

Zamarin D

Doi

10.1038/s41591-024-02942-7

PMID: 38653864 · DOI: 10.1038/s41591-024-02942-7 · Journal: Nature Medicine (2024)

TL;DR

This single-arm phase 2 trial evaluated nivolumab in 35 patients with dMMR/MSI-H or hypermutated endometrial or ovarian cancers at Memorial Sloan Kettering Cancer Center. The trial met both co-primary endpoints early: ORR was 58.8% (97.5% CI: 40.7–100%) and PFS at 24 weeks was 64.7%. Exploratory biomarker analyses identified dysfunctional CD8+PD-1+ and terminally dysfunctional CD8+PD-1+TOX+ T cells, as well as their spatial proximity to PD-L1+ cells, as independent predictors of clinical benefit. Somatic mutations in MEGF8 and SETD1B were significantly enriched in responders, while traditional biomarkers such as TMB and PD-L1 expression were not associated with response.

Cohort & data

  • Sample size: 35 patients enrolled (34 evaluable for response); single-center study at MSK (NCT03241745).
  • Cancer types: Recurrent or advanced endometrial cancer (UCEC, 83% endometrioid histology) and ovarian/fallopian tube/peritoneal cancers (OVT) with dMMR, MSI-H, or hypermutation (>=20 somatic mutations on MSK-IMPACT).
  • Dataset: ucec_msk_2024.
  • Assays: MSK-IMPACT for MSI/TMB assessment; whole-exome sequencing (33 of 34 evaluable tumors); multiplexed immunofluorescence microscopy (25 patients) using the Vectra system with a 7-marker panel (CD8, PD-1, TOX, PD-L1, PAX8, FoxP3, DAPI).
  • Median age: 64 years (range 36–87); 77% White, 11% Black, 6% Asian. Median follow-up 42.1 months.

Key findings

  • ORR: 58.8% (7 complete responses, 13 partial responses) in the evaluable cohort (n = 34) PMID:38653864.
  • PFS24: 64.7% (97.5% one-sided CI: 46.5–100%), exceeding the pre-specified threshold of 50%.
  • DCR: 73.5% (95% CI: 55.6–87.1%).
  • Median PFS: 21.6 months (95% CI: 4.9–NE); median OS: not reached; 1-year OS rate 79% (95% CI: 60.9–89.4%).
  • Safety: 91% had any-grade TRAE; 29% had grade 3/4 TRAEs. Most common TRAEs: arthralgia, fatigue, pain, and pruritis (each 29%). No grade 5 events.
  • TMB not predictive: No significant difference in TMB between responders and non-responders (18.1 vs. 14.4, P = 0.24).
  • PD-L1 not predictive: PD-L1 expression in tumor cells or combined tumor/immune cells was not associated with response.
  • T cell dysfunction predictive: Dysfunctional CD8+PD-1+ T cells (P = 0.006) and terminally dysfunctional CD8+PD-1+TOX+ T cells (P = 0.001) were strongly associated with PFS24.
  • Spatial proximity predictive: Interaction of CD8+PD-1+ T cells with PD-L1+ cells within 50 micrometers was associated with clinical benefit. A multivariate model combining CD8+PD-1+TOX+ percentage and PD-L1+ cell proximity achieved AUC = 0.897 (P = 0.0007).
  • MEGF8 mutations: Present in 32% of responders vs. 0% of non-responders (P = 0.027).
  • SETD1B mutations: Present in 58% of responders vs. 14% of non-responders (P = 0.015); also associated with increased CD8+PD-1+TOX+ T cell infiltration.
  • dMMR mechanism: No significant difference in response between genetic (germline/somatic MMR mutations) and epigenetic (MLH1 promoter hypermethylation) dMMR (P = 0.43).

Genes & alterations

  • MLH1: Promoter hypermethylation in 66% of cohort; somatic mutations in a subset. dMMR mechanism (genetic vs. epigenetic) not associated with differential response to nivolumab PMID:38653864.
  • MSH2, MSH6, PMS2: Somatic and germline mutations detected; 30% of cases harbored pathogenic somatic MMR gene mutations.
  • PTEN: Mutated in 76% of cohort overall. Not associated with clinical benefit or resistance.
  • PIK3CA: Mutated in 48% of cohort. Not associated with clinical benefit.
  • ARID1A: Mutated in 82% of cohort (SWI/SNF complex). Not differentially associated with benefit.
  • JAK1: Mutated in 24% of cohort. Previously implicated in resistance to immunotherapy, but no significant association with benefit in this cohort.
  • CTNNB1: Mutated in 15%. Not associated with benefit or resistance.
  • MEGF8: Mutated in 18% overall; 32% in responders vs. 0% in non-responders (P = 0.027). Functions as a negative regulator of Hedgehog signaling, which can promote immunosuppressive TME PMID:38653864.
  • SETD1B: Histone lysine methyltransferase. Mutated in 58% of responders vs. 14% of non-responders (P = 0.015). Mutations also associated with increased CD8+PD-1+TOX+ T cell infiltration, suggesting a link between chromatin remodeling and immune recognition PMID:38653864.
  • PTCH1: Hedgehog pathway component, mutated in 18% overall (26% in responders vs. 7% in non-responders).
  • KRAS: Mutated in 24% overall (26% responders, 21% non-responders).
  • TP53: Mutated in 18% overall (11% responders, 29% non-responders).
  • B2M: Mutated in 6% overall. No enrichment of antigen-presenting machinery alterations in non-responders.
  • POLE: One case with POLE exonuclease domain hotspot mutation (p.F367S) had the highest TMB and was MSS.
  • PBRM1, ARID2: SWI/SNF complex members; no significant difference in mutation rates between responders and non-responders.

Clinical implications

  • Nivolumab achieves durable responses in dMMR/MSI-H gynecologic cancers with an acceptable safety profile, consistent with other anti-PD-1 agents in this setting.
  • Traditional biomarkers (TMB and PD-L1 expression) do not predict response within dMMR-selected populations, limiting their utility for patient selection beyond dMMR status itself.
  • T cell functional states – specifically dysfunctional and terminally dysfunctional CD8+ T cells and their spatial interaction with PD-L1+ cells – represent candidate predictive biomarkers that could be assessed clinically using a 3–4 marker immunohistochemistry panel (CD8, PD-1, TOX, PD-L1).
  • Somatic mutations in MEGF8 and SETD1B are candidate genomic biomarkers for response to PD-1 blockade in dMMR cancers, warranting validation in larger cohorts.
  • The mechanism of dMMR (genetic vs. epigenetic) does not appear to predict differential response to nivolumab, in contrast to some prior reports with pembrolizumab.

Limitations & open questions

  • Small sample size (n = 35) limits statistical power, particularly for genomic subgroup analyses.
  • Single-arm design without a control group; no randomization.
  • Limited archival tissue availability (25 of 35 patients for TME analysis; 33 of 34 for WES).
  • Cohort enriched for high-grade disease (endometrioid or undifferentiated/dedifferentiated), which differs from published dMMR endometrial cancer populations where only 29% have high-grade disease.
  • Only 5 patients with Lynch syndrome, precluding conclusions about differential response in this subgroup.
  • TME biomarker cutoffs (bottom tertiles) were derived post hoc from the same cohort and require prospective validation.
  • Whether the MEGF8 and SETD1B associations are specific to nivolumab/PD-1 blockade or generalizable across checkpoint inhibitors remains unknown.
  • Multiplexed immunofluorescence assay is not widely available clinically; translation to standard IHC panels needs demonstration.

Citations from this paper used in the wiki

  • “The ORR was 58.8% (97.5% confidence interval (CI): 40.7–100%), and the PFS24 rate was 64.7% (97.5% one-sided CI: 46.5–100%), meeting the pre-specified endpoints.” (Abstract)
  • “Increases in both dysfunctional (CD8+PD-1+) and terminally dysfunctional (CD8+PD-1+TOX+) T cells were strongly associated with clinical benefit (P = 0.006 and P = 0.001, respectively).” (TME analyses)
  • “MEGF8 (32% in patients with PFS >= 24 weeks versus 0% in patients with PFS < 24 weeks, P = 0.027) and SET domain-containing 1B (SETD1B) […] (58% in patients with PFS >= 24 weeks versus 14% in patients with PFS < 24 weeks, P = 0.015), were statistically significantly different between the tumors from patients who had clinical benefit versus those who did not.” (Genomic analyses)
  • “A receiver operating characteristic (ROC) curve […] resulting in an area under the curve (AUC) of 0.897 (P = 0.0007), demonstrating a strong ability of these variables to predict immunotherapy outcomes.” (TME analyses)
  • “No significant difference was observed in TMB between responders and non-responders, suggesting that, among patients with dMMR/MSI tumors, TMB has no additional benefit in predicting likely response.” (Discussion)

This page was processed by crosslinker on 2026-05-04.