Immunotherapy in a/rEC after failure of platinum-based chemotherapy

One of the pivotal studies in immunotherapy for a/rEC was the KEYNOTE-158 trial. The trial confirmed that pembrolizumab monotherapy provided durable responses, with an objective response rate (ORR) of 42.6% in dMMR EC patients [14]. This breakthrough paved the way for additional trials evaluating the efficacy of ICIs.

The GARNET trial examined the efficacy of dostarlimab, an anti-PD-1 agent, in patients with dMMR/MSI-H and MMRp/MSS EC. In the dMMR subgroup, the results demonstrated a 45.5% ORR with long-lasting responses, supporting the clinical benefit of immunotherapy in this molecular subgroup. The trial provided a strong basis for dostarlimab’s approval in dMMR a/rEC, reinforcing the role of ICIs as a standard treatment for these tumors. Notably, minimal clinical benefit was observed in patients with MMRp/MSS tumors, with an ORR of only 15.4% and generally less durable responses, further emphasizing the importance of molecular profiling in guiding immunotherapy use [15]. For MMRp/MSS EC, where response rates to ICIs as monotherapy were limited, KEYNOTE-775 evaluated a combination approach using pembrolizumab plus lenvatinib, a multitarget antiangiogenic. The study showed a significant improvement in progression-free survival (median PFS 7.2 vs. 3.8 months, HR 0.56, p < 0.001) and overall survival (median OS 18.3 vs. 11.4 months, HR 0.62, p < 0.001) compared to chemotherapy, establishing this combination as the preferred second-line treatment for MMRp/MSS a/rEC [23].

Immunotherapy in first-line a/rEC

Given the promising results in pretreated patients, first-line immunotherapy strategies have been explored in large-scale phase III trials, where ICIs were integrated into platinum-paclitaxel chemotherapy regimens.

The MITO END-3 trial, a phase II randomized, open-label study conducted in 31 Italian Centers, compared carboplatin-paclitaxel with or without avelumab in first-line treatment of a/r EC. In the intention-to-treat (ITT) population, no significant difference in PFS was observed (9.9 vs. 9.6 months; HR 0.78). However, a clinically relevant benefit of chemoimmunotherapy was reported in the dMMR/MSI-H subgroup, with improved 12-month PFS (60% vs. 35%) and 24-month OS rates (76% vs. 55%), and a significant interaction by MMR status for both endpoints (p interaction = 0.015 for PFS and 0.029 for OS). The study suggests a benefit from adding avelumab in patients with dMMR/MSI-H tumors. Notably, MMR status was not used as a stratification factor, and the proportion of dMMR patients enrolled (47%) was higher than typically reported in the literature (25–30%) [26, 27].

The RUBY Part 1 trial assessed dostarlimab combined with chemotherapy in first-line a/rEC. Results showed a significant PFS improvement in dMMR/MSI-H patients (HR 0.28), along with a notable benefit in the ITT population (HR 0.64) [28]. Updated data showed a substantial OS advantage, particularly in dMMR/MSI-H tumors (HR 0.30), and a clinically relevant OS benefit also in the MMRp/MSS population (HR 0.79). These results supported the approval of dostarlimab plus chemotherapy in an all-comers population [29, 30]. Consistently, post hoc analyses from RUBY Part 1 showed that dostarlimab plus carboplatin-paclitaxel achieved the longest OS in MMRp/MSS patients, irrespective of subsequent therapies, reinforcing its role as a standard first-line option [30].

Similarly, the NRG-GY018 trial, evaluating pembrolizumab as first-line therapy, confirmed a strong PFS benefit in dMMR/MSI-H patients (HR 0.30), and a more modest yet statistically significant benefit in MMRp/MSS patients (HR 0.54) [31]. Updated OS analyses further demonstrated a survival benefit with the addition of immunotherapy, both in dMMR/MSI-H (HR 0.55) and in the MMRp/MSS subgroup (HR 0.79), reinforcing the role of chemo-immunotherapy as a new standard of care irrespective of MMR status [32]. In addition, updated data submitted to the EMA within the pembrolizumab regulatory dossier reported a PFS HR of approximately 0.74 and an OS HR of 0.80 in MMRp/MSS EC.

Direct comparisons across trials should be interpreted cautiously due to differences in inclusion criteria, histological composition, platinum-free interval definitions, and molecular subgroup distribution.

The RUBY Part 2 trial evaluated dostarlimab plus niraparib as maintenance after first-line chemotherapy in a/rEC. In the overall population, the combination reduced the risk of progression or death by 40% (HR 0.60, 95% CI 0.43–0.82), with a median PFS of 14.5 vs. 8.3 months. A benefit was also seen in the MMRp/MSS subgroup, with a 37% risk reduction (HR 0.63, 95% CI 0.44–0.91) and a 6.0-month gain in median PFS (14.3 vs. 8.3 months) [33]. In contrast, the DUO-E trial included separate treatment arms with durvalumab alone and durvalumab plus olaparib, allowing a more direct assessment of the potential contribution of PARPi to clinical outcomes. Thus, it is not possible to understand the value of the addition of a PARPi to the chemo-immunotherapy combination.

The AtTEnd trial investigated atezolizumab in combination with chemotherapy, demonstrating that dMMR/MSI-H patients achieved a marked reduction in risk of progression (HR 0.36), though benefits in the ITT population were more moderate (HR 0.74). Notably, no significant benefit was observed in the MMRp/MSS subgroup (HR 0.92) [34]. Updated overall survival data did not show a statistically significant OS benefit, confirming the limited impact of this approach beyond the dMMR/MSI-H population [35].

The DUO-E trial, a randomized three-arm study, evaluated chemotherapy alone versus chemotherapy combined with durvalumab, with or without olaparib, followed by maintenance therapy with durvalumab alone or durvalumab plus olaparib. The study met its dual primary endpoints for PFS in the ITT population, demonstrating an HR of 0.71 for the durvalumab arm and 0.55 for the durvalumab plus olaparib arm compared with chemotherapy alone. The greatest benefit of CP plus durvalumab versus CP was observed in the population (HR 0.41). In the MMRp/MSS subgroup, the addition of olaparib maintenance to durvalumab further enhanced PFS outcomes (HR 0.57) [24]. The MMRp/MSS population was highly heterogeneous, with frequent overlap of biomarkers and histological subtypes; notably, 84% of patients in the biomarker-known population were positive for at least one biomarker, and 79% of ctDNA-evaluable patients had detectable ctDNA at baseline. Importantly, the PFS benefit associated with the addition of olaparib maintenance was consistently observed across a range of biomarker-defined and histological subgroups, including patients with detectable ctDNA at baseline, supporting a broader antitumor activity of the chemo-immunotherapy plus PARPi strategy in pMMR/MSS disease [24, 36, 37]. Consequently, the independent contribution of PARP inhibition remains difficult to isolate. Several clinical trials have evaluated the role of ICIs in platinum-naïve a/rEC, with key studies summarized in Table 2.

The phase III LEAP-001 trial was the first to evaluate a chemoth-free combination, pembrolizumab plus lenvatinib, as first-line treatment for patients with a/rEC, compared to standard carboplatin-paclitaxel. The study did not meet the prespecified statistical criteria for PFS or OS, neither in the MMRp/MSS subgroup nor in the overall population. In the MMRp/MSS population, median PFS was 9.6 months with pembrolizumab-lenvatinib vs. 10.2 months with chemotherapy (HR 0.99), and median OS was 30.9 vs. 29.4 months (HR 1.02). In the ITT population, PFS was 12.5 vs. 10.2 months (HR 0.91), and OS was 37.7 vs. 32.1 months (HR 0.93). The trial did not meet the criteria for superiority in PFS at interim analyses (IA1 or IA2), nor for non-inferiority in OS at final analysis [38]. These findings also underline challenges associated with chemotherapy-free approaches, including patient selection and toxicity management. The results of LEAP-001 further suggest that, despite the established efficacy of ICI plus anti-angiogenic therapy in pretreated MMRp/MSS EC, chemotherapy may represent a critical immunogenic backbone in the treatment-naïve setting, potentially enhancing immune priming and improving responsiveness to immunotherapy.

These studies collectively confirm that ICIs significantly improve PFS and OS in dMMR/MSI-H EC, establishing them as a key component of first-line treatment. In the MMRp/MSS tumors, despite an improvement in patients’ outcomes having been reported, the magnitude of benefit is lower. Combination strategies, such as ICIs with chemotherapy and targeted agents, are being explored to enhance immunogenicity in these less responsive tumors.

MMR/MSI: a key but complex predictive biomarker

MMR deficiency is a well-established predictive biomarker for immunotherapy response in EC (Figure 2 — green arrow). The loss of function in MMR proteins (MLH1, PMS2, MSH2, MSH6) leads to MSI-H, resulting in a high TMB and increased neoantigen load, which enhances immune recognition and response to PD-1 inhibitors. This has been confirmed in clinical trials such as GARNET and KEYNOTE-158, supporting the approval of dostarlimab and pembrolizumab [25, 39].

Display full size

Figure 2. Biomarkers of response to immune checkpoint inhibitors (ICIs) in endometrial cancer (EC) with or without poly-ADP-ribose polymerase inhibitors (PARPi). MSI-H EC cells (blue cell) and high tumor mutational burden (TMB, blue points) are predictive biomarkers of ICIs treatment due to their higher neoantigen load. In contrast, MSI-H EC cells (blue cell) and high TMB (blue points) are predictive biomarkers of ICIs due to their higher neoantigen load. In contrast, homologous recombination deficiency (HRD) status, BRCA alterations, and possibly TP53 status may represent emerging stratification factors for combination strategies involving ICIs and PARPi, whereas their predictive role for ICIs alone remains unproven. PD-L1 expression currently lacks sufficient evidence as a validated predictive biomarker for ICIs in EC. Arrows represent predictive (green) and Inconsistent/Non-standardized Biomarker (red dashed) of response to ICI monotherapy. Created in BioRender. Colombo, I. (2026) https://BioRender.com/gik2ooa

However, emerging evidence indicates that not all dMMR/MSI-H tumors behave identically. The underlying mechanism of MMR deficiency, whether due to germline mutations (e.g., Lynch syndrome), somatic mutations, or more commonly, epigenetic MLH1 promoter hypermethylation (MLH1ph), may influence immunotherapy outcomes [40, 41]. Tumors with MLH1ph tend to present at more advanced stages, with higher rates of lymphovascular space invasion (LVSI) and fewer tumor-infiltrating lymphocytes (TILs), potentially explaining their reduced sensitivity to PD-1 blockade compared to tumors with genetic MMR alterations [41, 42]. Chow et al. [42] found that among ICI responders, patients with genetic mutations (Mut-dMMR responders) exhibited a strong CD8+ T-cell response, which was further enhanced by PD-1 inhibition. In contrast, responders with epigenetic alterations (epi-dMMR responders) relied more on natural killer (NK) cells, particularly the CD16+ NK cell population. These findings suggest that immune responses to immunotherapy may differ between these groups, yet both can benefit from treatment due to the distinct immune profiles of their tumors.

Several clinical trials have explored whether these biological differences translate into differences in patient outcomes. The NRG-GY018 trial showed a slightly better 12-month PFS in patients with MMR gene mutations (85%) compared to those with MLH1ph (75%), although not statistically significant [43]. Likewise, a post-hoc analysis of the GARNET trial found no meaningful difference in ORR across dMMR subgroups, suggesting that dostarlimab is effective regardless of the cause of MMR deficiency [44]. Consistently, exploratory analyses from the RUBY trial did not show clinically relevant differences in efficacy between dMMR tumors with MLH1 hypermethylation and those with MMR gene mutations, supporting a class effect of immune checkpoint inhibition across dMMR subtypes [45].

Despite these findings, although all dMMR EC tumors generally benefit from immunotherapy, the magnitude of the response varies, and the precise reasons remain unclear [41]. Emerging evidence also suggests that developmental signaling pathways, including Sonic Hedgehog (SHH), may contribute to shaping the tumor immune microenvironment in EC. Recent bioinformatic analyses have associated SHH pathway activation with increased CD8+ T-cell and NK-cell infiltration and improved survival outcomes, supporting its potential role as an emerging biomarker for immune contexture and patient stratification. However, further validation is required before clinical application. Moreover, although several biological mechanisms have been proposed, the determinants of primary resistance to ICIs in dMMR EC remain incompletely understood and are not yet clinically actionable [46]. Further research is needed to determine whether refining dMMR classifications could enhance treatment personalization and optimize outcomes [41].

TMB and its role in immunotherapy response

The TMB has been investigated as a potential biomarker for response to ICIs, as it reflects the neoantigen load, which enhances immune recognition. However, its predictive value is not consistent across all cancer types. The KEYNOTE-158 trial demonstrated that EC with high TMB had an ORR four times higher than those with low TMB, supporting its role as a predictor of immunotherapy response. This led to the FDA approval of pembrolizumab for TMB-high tumors, though no significant correlation was found between TMB and PD-L1 expression [39]. Higher TMB was associated with increased CD8+ T-cell infiltration, neoantigen load, and improved ORR in the GARNET trial regardless of MSS status (86.5% of MMRd/MSI-H tumors were TMB-high, compared to only 7% of MMRp/MSS tumors), suggesting a possible association with response to ICIs, although its independent predictive value remains uncertain (Figure 2 — green arrow) [25]. The substantial variability in assays, cut-offs, and scoring systems limits reproducibility and currently prevents PD-L1 implementation as a validated predictive biomarker in EC.

In this regards, TMB assessment is challenged by a lack of standardization. Different assays and thresholds, such as the ≥ 10 mutations/megabase cutoff from Foundation Medicine, result in inconsistent outcomes across trials and Centers. Several studies highlight differences in gene coverage, bioinformatics pipelines, and significant variability in defining TMB positivity affecting comparability in clinical trials [47–49]. These discrepancies underscore the need for standardized assays and well-defined thresholds to effectively incorporate TMB as a stratification criterion in clinical trials. This will be critical for clearly defining its role in EC and optimizing its use as an already established predictive biomarker for immunotherapy [50].

PD-L1 expression: variability in cut-offs, assays, and possible role in EC

The role of PD-L1 and PD-1 as prognostic and predictive biomarkers in EC is still uncertain. Although PD-1 plays a role in suppressing autoimmunity and promoting self-tolerance, variability in assays and scoring systems has made it difficult to define PD-L1 as a reliable predictive biomarker, with expression levels ranging from 5% to 76% across studies [51, 52]. A post-hoc analysis of the GARNET trial used the combined positive score (CPS, cut-off = 1) to evaluate PD-L1 expression in both dMMR/MSI-H and MMRp/MSS tumors. There was no significant difference in response to ICIs between PD-L1-positive and negative groups. However, patients with both high TMB and PD-L1 positivity showed better response rates, suggesting that TMB-high may be a stronger predictive marker than PD-L1 alone [25]. The DUO-E trial used the Tumor Area Positivity (TAP) score to assess PD-L1 status and observed that patients with TAP ≥ 1 had improved PFS in both the immunotherapy alone and immunotherapy + olaparib arms compared with patients with TAP < 1 [24]. The AtTEnd trial showed that patients with PD-L1-positive IC had a better response to atezolizumab + chemotherapy than chemotherapy alone [34].

These trials used different assays and scoring systems (CPS, IC, tumor proportion score [TPS], TAP), highlighting the lack of a standardized approach for PD-L1 assessment [53]. Among them, CPS, IC, and TAP scores appear to be more reliable, as they account for both IC and TPS, thereby reflecting the broader role of the tumor microenvironment and its interaction with the immune system in therapeutic response. From a molecular perspective, dMMR tumors tend to have high PD-L1 expression, supporting their sensitivity to immunotherapy. Conversely, p53-mutated and POLE-mutated subgroups show no clear association with PD-L1 positivity [53].

Despite extensive research, PD-L1 remains an inconsistent biomarker due to variability in expression, assay techniques, and cut-off definitions (Figure 2 — red dashed arrow) [52]. Current data suggest that evaluating PD-L1 in both tumor and ICs is more informative than tumor expression alone. These findings reinforce the idea that the tumor microenvironment plays a central role in immunotherapy response, emphasizing the need for a standardized approach for PD-L1 assessment before it can be widely used as a predictive marker.

Emerging biomarkers of immune response in EC: TP53, homologous recombination repair (HRR), homologous recombination deficiency (HRD), and BRCA

HRR is a DNA double-strand break repair mechanism involving proteins encoded by BRCA1/2 (germline and somatic mutations, and promoter methylation), RAD51, RAD51C, RAD51D, PALB2, ATM, and ATR genes [54]. HRD represents genomic instability resulting from HRR pathway dysfunction, but also detectable through specific genomic scar signatures including telomeric allelic imbalance, large-scale transitions, and loss of heterozygosity [54, 55]. In EC, dMMR and HRD present distinct molecular signatures [56], though emerging data reveal biological intersections between the HRR and MMR pathways, with potential synergistic implications [56]. HRR genes such as ATM and ATR, which regulate the cell cycle, can upregulate PD-L1 expression. Additionally, defective DNA repair mechanisms and cytosolic DNA accumulation increase neoantigen load and activate type I immunity, respectively, supporting the combined application of PARPi and ICIs [57].

TP53 mutations critically impact tumor biology through immune activation, cell cycle regulation, apoptosis, and DNA repair [22]. Notably, copy-number high, serous-like EC and High Grade Serous Ovarian Cancer exhibit significant molecular similarities, suggesting shared pro-oncogenic pathways like alteration in TP53. Moreover, increasing evidence suggested a role of p53 mutations in DNA damage response [58]. Approximately 34% of ECs frequently display alterations in genes such as ARID1A and BRCA, leading to HRD status [59]. De Jonge et al. [60] showed that HRD was significantly associated with non-endometrioid histology and TP53-mutated ECs, with 100% of HRD-positive ECs harboring TP53 mutations, while only 46% of TP53-mutated tumors are HRD-positive, indicating a strong, though not exclusive, association between the two biomarkers. Additionally, HRD status is more frequently observed in MMRp ECs with TP53 mutations and high-grade histology, further reinforcing the link between TP53 alterations and HRD [61]. Given the established sensitivity of HRD-positive tumors to platinum-based chemotherapy and DDR-targeted agents in other malignancies, it is plausible that HRD-positive EC, particularly the serous subtype and p53-mutated, may also benefit from these therapies [60]. In the post-hoc subgroup analysis of the RUBY trial part 1, dostarlimab improved overall survival in MMRp ECs with TP53 mutations [29]. In contrast, the MITO-END trial found no clinical benefit from Avelumab in MSS tumors with TP53 mutations, though the analysis was limited by sample size [27]. Finally, the DUO-E trial demonstrated that combining chemotherapy plus durvalumab, with or without olaparib, improved clinical outcomes regardless of HRR status. Notably, subgroup analyses suggested a greater benefit from the addition of olaparib in patients with p53-abnormal tumors [24]. These findings suggest that HRD status and TP53 alterations may represent biologically relevant stratification factors in combination strategies involving ICIs and PARPi, rather than validated predictive biomarkers of response to immunotherapy alone [24].

Despite variability in study designs and clinical endpoints, current evidence supports HRD status and BRCA alterations as biologically relevant biomarkers for sensitivity to DDR-targeted therapies rather than validated predictive biomarkers of immunotherapy response (Figure 2 — red dashed arrows). Although exploratory analyses suggest potential interactions between DNA repair defects and immune activation, their predictive value for ICIs remains uncertain and is currently unsupported by robust prospective clinical evidence. Similarly, the role of TP53 alterations in predicting immunotherapy response remains controversial and requires further validation (Figure 2 — red dashed arrows). These findings underscore the need for cautious interpretation and additional biomarker-driven studies before clinical implementation.