PICOS model was applied to perform the literature search: patients, intervention, comparator, objective and study design.10 Patients with dMMR or MSI-H diagnosed with advanced and/or metastatic EC who had previously received platinum-based chemotherapy were selected. All immunotherapies used as intervention and all comparators were included. Selected endpoints were objective response rate (ORR), progression-free survival (PFS) and overall survival (OS). In terms of design, clinical trials (CTs) with a minimum of 10 target patients were included. Conference communications were excluded due to a lack of information.

Preferred Reporting Items for Systematic reviews and Meta-Analyses (PRISMA)11 methodology was used to conduct a review in PubMed® and Embase® databases up to 28 May 2024. Filter ‘clinical trials’ was applied with the following terms: [microsatellite instability OR mismatch repair deficient] AND endometrial cancer. “PICO tool” was used on the Embase® database according to the PICOS model described above. The following search criteria were applied in Embase® database: (‘endometrium cancer’/exp. AND ‘microsatellite instability’ OR ‘mismatch repair deficient’/exp) AND ‘immunotherapy’/exp. AND (‘objective response rate’/exp. OR ‘progression free survival’/exp. OR ‘overall survival’/exp) AND ‘clinical trial’/exp. AND [<1966–2024]/py.

Two investigators conducted the search independently. Disagreements were resolved by discussion between study authors. Titles and abstracts were screened to exclude results that did not meet the study inclusion criteria. Full articles were reviewed in the eligibility process. Studies were accepted in English or Spanish.

Data from all studies were extracted and validated by two investigators. The following variables were collected from trials: authors and publication dates, study design, histology, stage, treatment lines, median follow-up, number of patients, treatments used as intervention and control. For efficacy endpoints, ORR values, median PFS and OS, confidence intervals and relative values in comparative CTs were collected. Regarding safety, the most frequent adverse events (AEs) of any grade, those of grade 3 or higher and immune-mediated AEs were recorded. Treatment reductions, discontinuations and deaths due to AEs were also obtained in those studies where available.

Importance of outcomes was assigned on the basis of clinical relevance for analysis of the efficacy of therapeutic alternatives. Final endpoint considered most important was OS (time from randomisation to death from any cause). PFS (time from randomisation to progression or death from any cause) was established as the surrogate endpoint of greatest value. ORR, defined as the proportion of patients with partial or complete response according to Response Evaluation Criteria in Solid Tumours (RECIST) criteria version 1.1,12 was analysed as the least relevant surrogate endpoint. The most interesting therapeutic alternatives were determined based on the above-mentioned efficacy criteria.

In terms of safety, AEs were assessed in patients who had received at least one dose of the drug during CT. AEs of grade 3 or higher were considered the most important due to their difficult management. The most frequent AEs (regardless of grade) and immune-mediated AEs were checked. Other data such as reductions, discontinuations or deaths due to treatment were reviewed.

The selected studies were grouped according to the treatment regimens used.

Availability of exclusive data on the target population (advanced and/or metastatic EC who had previously received platinum-based chemotherapy) or subgroup analyses were evaluated. Quality of data from the target population (previously treated patients) is superior to the quality of results from populations involving patients from different lines of treatment. Inappropriate interpretations of aggregate data could influence the conclusions. Sample sizes of CTs and follow-up periods were also checked. More reliable results are obtained from large sample sizes and long follow-up periods. Adequate sample sizes and patient follow-ups allow minimising unrealistic dispersion of results, as well as determining an acceptable number of events.

Cochrane risk of bias assessment tool was applied to the included studies,13 including eight domains: confounding bias; selection of participants into the study (or into the analysis) bias; classification of interventions bias; deviations from intended interventions bias; missing data bias; measure the outcome bias; selection of the reported results; and overall bias. Two investigators independently assessed the risk of bias.

A total of 54 records were found in the systematic search, of which 40 publications were excluded. Reasons for exclusion of studies were as follows: 12 were developed in a different clinical context, 10 did not show efficacy outcome data for pharmacological treatments, 6 were conference communications, 4 included less than ten target patients, 3 were not a CT, 2 were duplicated, 2 did not evaluate immunotherapy and 1 showed different efficacy endpoints. Finally, 14 studies were included.5,14–20,22,23,25,26,28,30Fig. 1 shows the systematic review conducted according to PRISMA methodology. Two of the included studies14,17 were extension studies of Marabelle et al. trial.16 The designs of included studies were: 8 non-randomised phase II studies, 3 non-randomised phase I studies, 1 randomised phase II study, 1 randomised phase III study and 1 randomised phase Ib/II study. All trials included patients who had received at least one first-line platinum-based treatment. Among CTs, 14.3% included patients with advanced EC,22,26 14.3% metastatic disease,15,18 and 71.4% both.5,14,16,17,19,20,23,25,28,30 All studies reported patient follow-up data (median values between 6 and 42.6 months). Sample size of included patients ranged from 11 to 143. The regimens analysed were: pembrolizumab monotherapy, pembrolizumab plus lenvatinib combination, durvalumab, durvalumab plus tremelimumab, dostarlimab, nivolumab and avelumab. Only two trials included a comparator arm. Table 1 shows the data and results of selected studies.

Figure 1.

Systematic review of literature.

The largest numerical effect was achieved for pembrolizumab after a median follow-up of 25.8 months (OS = 40.0 months [95% CI, 25.3-Not Reached]); (PFS = 23.5 months [95% CI, 10.7-NR]); (ORR = 58% [95% CI, 37–78]).15 Pembrolizumab plus lenvatinib was positioned as another interesting alternative after 12.2 months of follow-up (OS = NR [95% CI, NR-NR]); (PFS = 10.7 months, [95% CI 5.6-NR]); (ORR = 40% [95% CI, 28–53]).20 No OS data were found in 2 trials (14.3%).18,23

Pembrolizumab is an IgG4 anti-PD-1 monoclonal antibody, blocking the interaction between PD-1 and its ligands PD-L1 and PD-L2.14.15 Multicohort KEYNOTE-158 study showed results on the use of pembrolizumab in patients diagnosed with cancer with MSI-H/dMMR, excluding colorectal cancer.16 This was a phase II, nonrandomized, multicenter, single-arm CT. Subsequently, two extensions of this study were published including results from D and K cohorts with patients diagnosed of EC with MSI-H/dMMR.14,17O'Malley et al.14 included 90 patients from D (n = 11) and K (n = 79) cohorts. With a median follow-up of 42.6 months (95% CI, 6.4–56.1), ORR was achieved by 48.0% (95% CI, 37.0–60.0) of patients according to an independent radiological committee. The median PFS observed was 13.1 months (95% CI, 4.3–34.4), while the median OS was not reached. Bellone et al.15 developed an investigation with similar characteristics to KEYNOTE-158 trial. Median follow-up was 25.8 months. An ORR of 58% (95% CI, 37–78) was achieved according to investigator analysis. Median PFS was 23.5 months (95% CI, 10.7–NR), and median OS was 40.0 months (95% CI, 25.3–NR).

Other trials combined pembrolizumab with lenvatinib, such as Makker et al.18 and Makker et al.19 Both studies had similar designs: single-arm phase II trial. Patients with MSI-H/dMMR and MSS/MMRp were enrolled. MSS/MMRp was detected in 94 patients by Makker et al.,19 while 11 patients presented MSI-H/dMMR. ORR based on investigator assessment in MSI-H/dMMR group was 63.6% (95% CI, 30.8–89.1). Median PFS was 18.9 months (95% CI, 4.0-NR), and median OS was not achieved in this MSI-H/dMMR group. Makker et al.20 was a phase III, randomised, open-label, comparative CT. Pembrolizumab-lenvatinib regimen was evaluated versus chemotherapy. ORR achieved in MSI-H/dMMR population was 40% (95% CI, 28–53) in pembrolizumab plus lenvatinib group and 12% (95% CI, 5–23) in the control group. Median PFS in group receiving pembrolizumab in combination with lenvatinib was 10.7 months (95% CI, 5.6–NR) and 3.7 months (95% CI, 3.1–4.4) in control group. Median OS was not reached in intervention group and 8.6 months (95% CI, 5.5–12.9) in the group receiving chemotherapy.

Concerning safety analysis, KEYNOTE-158 study showed that the most frequent AEs of pembrolizumab monotherapy were fatigue (14.6%), pruritus (12.9%) and diarrhoea (12.0%).16 With respect to grade 3 or higher AEs, pembrolizumab was associated with increased transaminases (1.7%) and pneumonitis (1.3%). A total of 9.4% of patients in KEYNOTE-158 study discontinued treatment due to AEs. No patients died from AEs associated with treatment. In the study by Bellone15, pembrolizumab frequently caused diarrhoea (56.0%), fatigue (48.0%) and skin disorders (44.0%). In this trial, the most frequent grade 3 or higher AEs were hyperglycaemia (16.0%) and diarrhoea (12.0%). No information on discontinuations, interruptions, delays or deaths was reported. In Makker et al.,20 the most common AEs of pembrolizumab-lenvatinib combination were hypertension (64.0%), hypothyroidism (57.4%) and diarrhoea (54.2%). Likewise, pembrolizumab-lenvatinib regimen showed hypertension (37.9%), weight loss (10.3%) and decreased appetite (7.9%) as predominant grade 3 or higher AEs. Furthermore, 69.2% of the population discontinued combination therapy. Two deaths occurred in the group of patients receiving pembrolizumab-lenvatinib.

This IgG1 κ-type monoclonal antibody selectively blocks the interaction of PD-L1 with PD-1 and CD80.21 Two studies with durvalumab were included in this review, one of which used durvalumab-tremelimumab combination. The study by Antill22 evaluated durvalumab in monotherapy for patients with dMMR (n = 36) and MMRp (n = 35). However, 21 dMMR patients had not received any previous line of treatment. This scientific publication was a phase II, non-randomised, single-arm trial. Follow-up of dMMR population was 19 months. Investigator-assessed ORR in dMMR group reached 47% (95% CI, 32–63). Median PFS was 8.3 months (95% CI 2.4–NR), and median OS was not reached. Durvalumab plus tremelimumab regimen was evaluated in a phase II, randomised, open-label, active comparator study.23 Thirty-eight patients were assigned to the durvalumab monotherapy arm and 39 patients to the durvalumab-tremelimumab combination. dMMR was determined in 13.2% of patients treated with durvalumab monotherapy and 10.3% in population with a combination of immunotherapeutic agents. ORR at 6 months was 10.8% (90% CI, 4.8–100) in durvalumab arm and 5.3% (90% CI, 1.4–100) in durvalumab-tremelimumab scheme arm. Median PFS was similar in both arms: 1.9 months (90% CI, 1.75–Inf) for durvalumab and 2.0 months (90% CI, 1.75–Inf) for durvalumab plus tremelimumab. No OS data were found.

Immune-mediated AEs were collected in Antill et al.22 highlighting thyroid disorders: hypothyroidism (14.0%) and hyperthyroidism (9.0%). Three patients discontinued treatment due to AEs. No other safety data were reported in this trial. In Rubinstein et al.,23 the most frequent AEs were hyperglycemia (95% in monotherapy arm and 92% in durvalumab-tremelimumab group) and anaemia (82.0% and 87.0%, respectively). The most common grade 3 or higher AEs were: anaemia (29.0% for durvalumab and 28.0% for durvalumab-tremelimumab regimen) and decreased lymphocyte count (26.0% and 38.0%, respectively). Data on treatment reductions, discontinuations, or deaths due to AEs were not reported.

The mechanism of action of this IgG4 monoclonal antibody is based on binding to PD-1 receptors and blocking their association with PD-L1 and PD-L2.24 A single-arm phase I trial was found. In the first published results, Oaknin et al.25 included patients with MSI-H/dMMR profile. In an update by Oaknin et al.,5 two cohorts were established: cohort A1 (MSI-H/dMMR, n = 108) and cohort A2 (MMRp/MSS, n = 161). ORR was similar in both publications: 42.3% (95% CI, 30.6–54.6) from Oaknin et al.25vs. 43.5% (95% CI, 34.0–53.4) in the A1 cohort.5 In both papers, ORR was assessed by a blinded independent central committee. On the other hand, PFS was 8.1 months (95% CI, 3.0–18.0) in Oaknin et al.25versus 12.2 months (no CI data) in the most updated publication. Median OS was not reached in either study. Subsequently, a phase I single-arm trial by André et al.26 enrolled patients diagnosed with different solid tumours. This trial included a post-hoc analysis presenting the results disaggregated by tumour type. For 143 patients diagnosed of EC with dMMR or mutated POLE, an independent committee determined an ORR of 45.5% (95% CI, 37.1–54.0). Median PFS was 6.0 (95% CI, 4.1–18.0) months, and median OS was not reached (95% CI, 25.7–NR).

Up to 93.9% and 95.3% of patients in Oaknin et al.25 and Oaknin et al.,5 respectively, experienced an AE. In patients diagnosed of EC and dMMR, the most frequent AEs were diarrhoea (15.4% and 16.3% in 2020 and 2022 studies, respectively) and asthenia (15.4% and 14.0%). Considering AEs of grade 3 or higher, anaemia was the most common in both studies (2.9% and 3.9%). A total of 23.1% of patients in Oaknin et al.25 presented to discontinue treatment due to AEs. Nevertheless, in Oaknin et al.,5 such discontinuations occurred in 3.9%. No deaths were recorded in dostarlimab groups in both studies.

Safety analysis in André et al.26 included the overall population without differentiation by solid tumour type. The most frequent AEs were diarrhoea (15.4%) and asthenia (14.3%). Anaemia (2.5%) and increased transaminases (1.9%) were found as AEs of grade 3 or higher. Treatment discontinuations were led by AEs in 6.9% of subjects. There were two deaths attributed to treatment, but none in the population with EC and dMMR.

Nivolumab is an IgG4 monoclonal antibody that prevents the interaction between the PD-1 receptor and PD-L1.27 The study found was a phase II basket single-arm CT with several types of tumours.28 Median patient follow-up was 17.3 months. The efficacy outcomes achieved by the global population were: ORR = 36% (90% CI, 23.5–49.5), median PFS = 6.3 months and median OS = 17.3 months (both outcomes without reported CIs).

Regarding safety, the most frequent AEs were fatigue (40%), anaemia (33%) and skin rash (17%). Likewise, anaemia (18.4%), dehydration (5.3%), fatigue (5.3%), maculo-papular rash (5.3%), sepsis (5.3%) and skin infections (5.3%) were the most common grade 3 or higher AEs. No data were available on treatment reductions, discontinuations or deaths due to AEs.

Avelumab is an IgG1 class monoclonal antibody that interacts with PD-L1 by preventing binding to PD-1 and B7.1.29 Konstantinopoulos et al.30 consisted of a phase II single-arm CT with two cohorts. The selected population was distributed according to the presence of dMMR/POLE mutated (n = 15) or MMRp/POLE without mutations (n = 16). The co-primary endpoints were ORR and PFS at 6 months. OS was also measured as a secondary endpoint. With a median follow-up of 18.6 months, ORR in dMMR/POLE cohort was 26.7% (95% CI, 7.8–55.1). PFS at 6 months was 40.0% (95% CI, 16.3–66.7), while the median was 4.4 (95% CI, 1.7–NR). Median OS was not reached.

Concerning safety, 71.0% of patients presented at least one AE. Fatigue (35.5%) followed by nausea (16%) were the most frequent AEs. Grade 3 or higher AEs were registered in 19.4% of cases, most frequently being: anaemia (6.5%), diarrhoea (6.5%), bradycardia (3.2%), hypothyroidism (3.2%), myositis (3.2%) and rash acneiform (3.2%). No data were available on treatment reductions, discontinuations or deaths due to AEs.

Durvalumab presented global data from a population involving previously untreated and pre-treated patients.22 Sample sizes of CTs ranged from 11 (pembrolizumab-lenvatinib regimen, Makker et al.19 to 143 (dostarlimab)26 patients. CTs of schemes with the smallest sample sizes were: pembrolizumab-lenvatinib combination (n = 11),19 nivolumab (n = 13),28 avelumab (n = 15),30 pembrolizumab monotherapy (n = 24)15 and durvalumab (n = 36).22 Median follow-up of estudies ranged from 6 (durvalumab-tremelimumab combination)23 to 42.6 (pembrolizumab)14 months. Follow-up period was reported by all CTs.

All of the included studies had an overall assessment of a low risk of bias when the Cochrane risk of bias assessment tool was applied. Fig. 2 shows the analysis with the risk of bias assessment tool.

Figure 2.

Cochrane risk of bias assessment tool.