Sex, Timing, and Outcomes of Curative-Intent Immunotherapy Authors: Anne-Murielle Hollinger, Camille L. Gerard, Victoria J. Andreas, Zayne M. Roa Diaz, Silke Gillessen, Olivier Michielin, Miklos Pless, Solange Peters, Charlotte Micheloud, Stefanie Hayoz, Berna C. Özdemir, Sabine Schmid Published: 2026 ABSTRACT This systematic review and meta-analysis evaluates associations between immune checkpoint inhibitors and survival outcomes by patient sex and timing of administration among patients receiving curative treatment for solid tumors. Introduction Immune checkpoint inhibitors (ICIs) represent a cornerstone of therapy for various cancer types.1 In recent years, their use has expanded beyond the metastatic setting into earlier disease stages.2,3 Adjuvant and neoadjuvant ICIs, given alone or as combination therapies, have increasingly been integrated into treatment schemes across cancer types, including melanoma,4,5 non–small cell lung cancer (NSCLC),6,7 urothelial carcinoma (UC),8 and triple-negative breast cancer,9 among others. Differences in immune responses, hormone levels, tumor biology, and pharmacokinetics between female and male patients may contribute to variability in response to ICIs.10 Several studies in the metastatic setting have explored how sex might influence treatment outcomes, with inconsistent results but with some meta-analyses suggesting sex-related differences in treatment efficacy.11,12 To date, it remains unclear whether patient sex impacts outcomes of early-stage disease treated with ICIs. As the incorporation of ICI into treatment strategies expands, understanding possible sex-based differences in response is essential. The aim of adjuvant ICIs is to eradicate residual micrometastases and improve recurrence-free survival (RFS). However, the lack of reliable predictive biomarkers hampers the identification of patients who truly benefit, resulting in potential overtreatment of patients who may be cured by surgery alone. Moreover, it is not clear whether the improved RFS observed with adjuvant ICIs translates into improved overall survival (OS).13,14 These challenges have fueled the growing interest in neoadjuvant ICIs, which offer several advantages.15 Biologically, the intact primary tumor serves as a source of tumor-associated antigens, enhancing the priming and expansion of tumor-specific T cells and potentially eliciting a more durable systemic immune response.16 Preclinical data17 and early-phase clinical trials18 suggest that this strategy leads to better outcomes compared with adjuvant treatments. This systematic review and meta-analysis aimed to evaluate the associations of sex, cancer type, and treatment timing relative to surgery with the treatment outcomes of ICIs. Methods Search Strategy and Study Selection We conducted this systematic review and meta-analysis in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guideline.19,20 The protocol was registered in PROSPERO (CRD42023429373).21 To be eligible for inclusion, studies had to be randomized phase 3 clinical trials assessing the effect of ICIs in solid cancer types with a curative-intent strategy (ie, including a definite local therapy and the application of ICIs in a neoadjuvant, perioperative, adjuvant, or consolidation setting and/or concomitant with or prior to radiotherapy or chemoradiotherapy). Eligible interventions comprised anti–CTLA-4, anti–programmed cell death protein 1 (PD-1), or anti–programmed cell death 1 ligand 1 (PD-L1) agents, given as monotherapy, dual blockade, or in combination with other systemic therapies, provided they were approved by both the US Food and Drug Administration and the European Medicines Agency. We restricted inclusion to publications in English. We excluded sex-specific tumor entities (eg, breast, prostate cancer) and trials assessing use of ICIs in metastatic disease with noncurative intent. We systematically searched MEDLINE, Embase, the Cochrane Library, ClinicalTrials.gov, and the World Health Organization International Clinical Trials Registry Platform from January 1, 2010, to June 23, 2025, using controlled vocabulary combined with free-text search terms (eAppendix 2 in Supplement 1).22 In addition, a manual search in PubMed was conducted independently by 3 authors (A.M.H., C.L.G., V.J.A.) to ensure capturing of all studies and updates until August 10, 2025 (eAppendix 3 in Supplement 1). Each records was reviewed independently by 2 of 3 authors (A.M.H., C.L.G., V.J.A.) using Covidence software (Veritas Health Innovation Ltd).23 Discordances were assessed by a fourth author (S.S.). Data Extraction Data from each report were extracted independently by 2 of 3 authors (A.M.H., C.L.G., V.J.A.). Main information collected in the dataset included trial name, first author, year of publication, geographic region, tumor entity, type of local therapy, timing of ICIs relative to local therapy, type and name of ICIs, additional therapies, number of patients, sex distribution, median follow-up time, median time to event and hazard ratios (HRs) and 95% CIs for prespecified time-to-event end points, and rate of pathological complete response (pCR). Prespecified end points were disease-free survival (DFS), distant metastasis-free survival (DMFS) or time to death or distant metastasis (TDDM), event-free survival (EFS), OS, progression-free survival (PFS), RFS, and pCR in the intention-to-treat population and by sex (a list of study abbreviations is included in eAppendix 1 in Supplement 1). Outcomes were extracted if they reflected primary, secondary, or predefined exploratory end points (eAppendix 5 in Supplement 1). No formal study-level risk-of-bias assessment was performed for the included trials. Statistical Analysis The association of ICIs with the prespecified end points of interest was analyzed overall and stratified by sex and cancer type. Analyses were performed for each end point of interest separately. To analyze the association of ICIs with disease recurrence or progression across all trials, as well as to enable separate analyses and comparison of trials with a neoadjuvant or perioperative ICI component vs trials with an adjuvant ICI component solely, we conducted an analysis incorporating the primary end point reflecting disease control of each individual study, thereby generating a combined analysis of DFS, EFS, PFS, and RFS. The inclusion criteria allowed for between-trial heterogeneity regarding local treatment modality, control treatments, and the timing of ICIs. To better delineate the associations of sex and timing of ICIs with treatment outcomes, a sensitivity analysis was performed. This analysis was restricted to exclusively surgery trials to enhance the cross-study comparability and excluded studies with ICIs in the control arm to better define the treatment outcomes attributable to ICIs. Treatment outcomes were evaluated overall, by sex, and by timing of ICIs with respect to surgery. Owing to a limited number of purely neoadjuvant trials, treatment settings were classified as either adjuvant ICIs or neoadjuvant and perioperative ICIs combined. Subgroup analyses were conducted when at least 3 studies were available per category. HRs or risk ratios (RRs) and 95% CIs were calculated using a random-effects meta-analysis to account for between-study heterogeneity, and the results were illustrated using forest plots. Heterogeneity was assessed using the I2 statistic and Cochran Q test. Publication bias was evaluated using funnel plots and the Egger test. In the case of a suspected publication bias, trim and fill was applied as a sensitivity analysis to assess robustness. No adjustment for multiple testing was applied. All reported P values were 2-sided and a significance level of P < .05 was used for all analyses (eAppendix 5 in Supplement 1). Data were analyzed using R, version 4.5.0 (R Project for Statistical Computing), with the add-on packages meta, version 8.0-2, and forestplot, version 3.1.6. Results Study Characteristics We identified 13 356 records through our systematic search and included 49 publications originating from 32 trials4,6,7,8,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68 after exclusion of 16 publications at the full-text screening stage (eAppendix 4 in Supplement 1).41,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83 An additional 13 publications originating from 6 trials were retrieved through manual search84,85,86,87,88,89,90,91,92,93,94,95,96 for a total of 62 published reports from 38 trials (Figure 1). Two trials (CheckMate 91448,57 and E160941) included 2 investigational arms each, which were analyzed separately, resulting in 40 pairwise comparisons from the 38 trials.4,6,8,24,25,29,31,37,40,41,43,44,47,48,49,52,53,56,57,58,59,60,61,62,63,64,65,66,67,68,88,89,90,91,92,93,94,95,96 Flowchart of the Systematic Search and Selection Process ICI indicates immune checkpoint inhibitor; WHO, World Health Organization. The final analytic cohort comprised 9 cancer types (Table and eFigure 1 in Supplement 1). Esophageal cancer (EC), gastroesophageal junction cancer (GEJ), and gastric cancer (GC) were pooled and counted as 1 tumor entity. Table. Main Characteristics of the Included TrialsSource (trial name)Tumor siteTiming of ICIPatientsFollow-up, median (IQR), moExtracted primary end pointsHR (CI)Total, No.Sex, No. (%)MaleFemaleOverallMalesFemalesLarkin et al,40 2023 (CheckMate 238)MelanomaAdjuvant906527 (58)379 (42)61.5 (NR)/61.2 (NR)aRFS0.72 (95% CI, 0.60-0.86)0.74 (95% CI, 0.59-0.93)0.71 (95% CI, 0.54-0.95)Kirkwood et al,60 2023 (CheckMate 76K)MelanomaAdjuvant790483 (61)307 (39)15.8 (11.9-20.3)/15.9 (12.0-20.4)aRFS0.42 (95% CI, 0.30-0.59)0.33 (95% CI, 0.22-0.51)0.71 (95% CI, 0.39-1.29)Weber et al,25 2023 (CheckMate 915)MelanomaAdjuvant18441052 (57)792 (43)28.0 (NR)/28.1 (NR)aRFS0.92 (97.3% CI, 0.77-1.09)0.93 (95% CI, 0.76-1.14)0.94 (95% CI, 0.74-1.18)Tarhini et al,41 2020 (E1609 Ipi3 vs SOC)MelanomaAdjuvant1159723 (62)436 (38)57.4 (NR)RFS, OS0.85 (99.4% CI, 0.66-1.09)NRNRTarhini et al,41 2020 (E1609 Ipi10 vs SOC)MelanomaAdjuvant1147737 (64)410 (36)57.4 (NR)RFS, OS0.84 (99.4% CI, 0.65-1.09)NRNREggermont et al,29 2019 (EORTC 18071)MelanomaAdjuvant951589 (62)362 (38)79.2 (60.0-93.6)/85.2 (63.6-97.2)aRFS0.75 (95% CI, 0.63-0.88)0.76 (99% CI, 0.58-0.99)0.76 (99% CI, 0.54-1.07)Grossmann et al,24 2022 (SWOG S1404)MelanomaAdjuvant1301778 (60)523 (40)47.3 (NR)/43.3 (NR)aRFS, OS0.76 (99.62% CI, 0.59-0.99)0.76 (95% CI, 0.61-0.95)0.73 (95% CI, 0.55-0.96)Eggermont et al,31 2024 (KEYNOTE-054)MelanomaAdjuvant1019628 (62)391 (38)82.8 (NR)RFS0.63 (95% CI, 0.53-0.74)0.64 (99% CI, 0.49-0.84)0.60 (99% CI, 0.41-0.87)Luke et al,37 2025 (KEYNOTE-716)MelanomaAdjuvant976589 (60)387 (40)52.8 (NR)RFS0.62 (95% CI, 0.50-0.78)0.60 (95% CI, 0.46-0.78)0.67 (95% CI, 0.45-0.98)Blank et al,4 2024 (NADINA)MelanomaNeoadjuvant (response-guided: adjuvant)423276 (65)147 (35)10.6 (5.2-16.8)/9.9 (4.6-16.8)aEFS0.32 (99.9% CI, 0.15-0.66)0.33 (95% CI, 0.13-0.82)0.28 (95% CI, 0.08-0.99)Rischin et al,67 2025 (C-POST)Cutaneous SCCAdjuvant415348 (84)67 (16)24 (NR)DFS0.32 (95% CI, 0.20-0.51)0.34 (95% CI, 0.20-0.57)0.34 (95% CI, 0.12-0.95)Haddad et al,95 2025 (IMvoke010)HNSCCAdjuvant or consolidation406342 (84)64 (16)46.5 (NR)EFS0.94 (95% CI, 0.70-1.26)0.94 (95% CI, 0.69-1.29)0.99 (95% CI, 0.43-2.29)Lee et al,65 2021 (JAVELIN Head and Neck 100)HNSCCPeri-CRT697575 (82)122 (18)14.6 (8.5-19.6)/14.8 (11.6-18.8)aPFS1.21 (95% CI, 0.93-1.57)1.21 (95% CI, 0.91-1.61)1.16 (95% CI, 0.58-2.33)Machiels et al,93 2024 (KEYNOTE-412)HNSCCPeri-CRT804660 (82)144 (18)47.7 (42.1-52.3)EFS0.83 (95% CI, 0.68-1.03)0.85 (95% CI, 0.68-1.07)0.75 (95% CI, 0.44-1.27)Uppaluri et al,68 2025 (KEYNOTE-689)HNSCCPerioperative714563 (79)151 (21)38.3 (NR)EFS0.73 (95% CI, 0.58-0.92)0.67 (95% CI, 0.52-0.87)1.00 (95% CI, 0.57-1.76)Heymach et al,63 2023 (AEGEAN)NSCLCPerioperative740530 (72)210 (28)11.7 (NR)EFS, pCR0.68 (95% CI, 0.53-0.88)0.61 (95% CI, 0.44-0.82)0.95 (95% CI, 0.58-1.56)Cascone et al,6 2024 (CheckMate 77T)NSCLCPerioperative461327 (71)134 (29)25.4 (NR)EFS0.58 (97.36% CI, 0.42-0.81)0.53 (95% CI, 0.37-0.75)0.71 (95% CI, 0.41-1.20)Forde et al,89 2025 (CheckMate 816)NSCLCNeoadjuvant358255 (71)103 (29)68.4 (NR)EFS, pCR0.68 (95% CI, 0.51-0.91)0.66 (95% CI, 0.47-0.91)0.67 (95% CI, 0.36-1.27)Felip et al,44 2021 (IMpower010)bNSCLCAdjuvant882589 (67)293 (33)32.2 (27.4-38.3)DFS (in stage II-IIIA)0.79 (95% CI, 0.64-0.96)0.76 (95% CI, 0.59-0.99)0.80 (95% CI, 0.57-1.13)Spicer et al,43 2024 (KEYNOTE-671)NSCLCPerioperative797563 (71)234 (29)36.6 (27.6-47.8)EFS, OS0.59 (95% CI, 0.48-0.72)0.62 (95% CI, 0.49-0.78)0.52 (95% CI, 0.36-0.75)O’Brien et al,47 2022 (KEYNOTE-091)NSCLCAdjuvant1177804 (68)373 (32)35.6 (27.1-45.5)DFS0.76 (95% CI, 0.63-0.91)0.81 (95% CI, 0.65-1.01)0.73 (95% CI, 0.54-1.00)Lu et al,92 2024 (Neotorch)NSCLCPerioperative404370 (92)34 (8)18.3 (12.7-22.5)EFS0.40 (95% CI, 0.28-0.57)0.38 (95% CI, 0.26-0.54)0.54 (95% CI, 0.15-1.80)Spigel et al,88 2022 (PACIFIC)NSCLCConsolidation713500 (70)213 (30)34.2 (NR)PFS, OS0.55 (95% CI, 0.45-0.68)0.61 (95% CI, 0.48-0.76)0.52 (95% CI, 0.36-0.74)Yue et al,64 2025 (RATIONALE-315)NSCLCPerioperative453410 (91)43 (9)22.0 (15.5-28.0)EFS0.56 (95% CI, 0.40-0.79)0.57 (95% CI, 0.40-0.80)0.68 (95% CI, 0.19-2.40)Cheng et al,61 2024 (ADRIATIC)cSCLCConsolidation530366 (69)164 (31)27.4 (NR)/27.7 (NR)aPFS, OS0.76 (97.2% CI, 0.59-0.98)0.80 (95% CI, 0.61-1.04)0.71 (95% CI, 0.47-1.08)Kang et al,66 2024 (ATTRACTION-5)GC or GEJAdjuvant755530 (70)225 (30)49.1 (43.1-56.7)RFS0.90 (95.72% CI, 0.69-1.18)0.88 (95% CI, 0.64-1.20)0.93 (95% CI, 0.60-1.44)Kelly et al,56 2021 (CheckMate 577)GEJ or ECAdjuvant794671 (85)123 (15)24.4 (NR)DFS0.69 (96.4% CI, 0.56-0.86)0.73 (95% CI, 0.59-0.91)0.59 (95% CI, 0.35-1.00)Zheng et al,96 2024 (HCHTOG1909)ECPerioperative252194 (77)58 (23)14.0 (NR)/11.4 (NR)aEFS0.62 (95% CI, 0.39-1.00)NRNRShitara et al,62 2024 (KEYNOTE-585)GC or GEJPerioperative804575 (72)229 (28)47.7 (38.0-54.8)EFS, OS, pCR0.81 (95% CI, 0.67-0.99)0.86 (95% CI, 0.68-1.09)0.70 (95% CI, 0.48-1.03)Janjigian et al,94 2025 (MATTERHORN)GC or GEJPerioperative948682 (72)266 (28)31.5 (26.7-36.6)EFS0.71 (95% CI, 0.58-0.86)0.65 (95% CI, 0.51-0.83)0.84 (95% CI, 0.58-1.24)Qin et al,91 2023 (IMbrave050)HCCAdjuvant668555 (83)113 (17)17.4 (13.8-22.1)/17.6 (14.0-22.1)aRFS0.72 (95% CI, 0.53-0.98)0.74 (95% CI, 0.56-0.98)0.73 (95% CI, 0.38-1.40)Motzer et al,48 2023 (CheckMate 914)RCCAdjuvant816580 (71)236 (29)37.0 (31.3-43.7)DFS0.92 (95% CI, 0.71-1.19)0.97 (95% CI, 0.72-1.31)0.86 (95% CI, 0.50-1.45)Motzer et al,57 2025 (CheckMate 914 Part B)RCCAdjuvant619446 (72)173 (28)27.0 (NR)DFS0.87 (95% CI, 0.62-1.21)0.80 (95% CI, 0.55-1.18)0.98 (95% CI, 0.51-1.91)Pal et al,49 2022 (IMmotion010)RCCAdjuvant778565 (73)213 (27)45.1 (39.4-51.6)/44.2 (38.6-50.8)aDFS0.93 (95% CI, 0.75-1.15)1.08 (95% CI, 0.84-1.39)0.61 (95% CI, 0.40-0.94)Choueiri et al,52 2024 (KEYNOTE-564)RCCAdjuvant994706 (71)288 (29)57.2 (NR)DFS0.72 (95% CI, 0.59-0.87)0.74 (95% CI, 0.59-0.94)0.68 (95% CI, 0.47-0.99)Allaf et al,58 2024 (PROSPER ECOG-ACRIN EA8143)RCCPerioperative819571 (70)248 (30)30.4 (21.5-42.4)/30.1 (21.9-41.8)aRFS0.94 (95% CI, 0.74-1.21)0.89 (95% CI, 0.66-1.20)1.03 (95% CI, 0.64-1.68)Galsky et al,90 2025 (CheckMate 274)UCAdjuvant709540 (76)169 (24)36.1 (NR)DFS0.71 (95% CI, 0.58-0.86)0.69 (95% CI, 0.56-0.87)0.79 (95% CI, 0.52-1.20)Bellmunt et al,53 2021 (IMvigor010)UCAdjuvant809638 (79)171 (21)21.9 (13.2-29.8)DFS0.89 (95% CI, 0.74-1.08)0.91 (95% CI, 0.73-1.13)1.00 (95% CI, 0.65-1.52)Apolo et al,59 2025 (KEYNOTE-123)UCAdjuvant702524 (75)178 (25)45.7 (NR)/40.5 (NR)aDFS, OS0.73 (95% CI, 0.59-0.90)0.70 (95% CI, 0.56-0.89)0.92 (95% CI, 0.58-1.47)Powles et al,8 2024 (NIAGARA)cUCPerioperative1063870 (82)193 (18)42.3 (NR)EFS, pCR0.68 (95% CI, 0.56-0.82)0.71 (95% CI, 0.58-0.88)0.56 (95% CI, 0.35-0.88)Abbreviations: CRT, chemoradiotherapy; DFS, disease-free survival; EC, esophageal cancer; EFS, event-free survival; GC, gastric cancer; GEJ, gastroesophageal junction cancer; HCC, hepatocellular carcinoma; HNSCC, head and neck squamous cell carcinoma; HR, hazard ratio; ICI, immune checkpoint inhibitor; Ipi 3, ipilimumab, 3 mg/kg; Ipi 10, ipilimumab, 10 mg/kg; NR, not reported; NSCLC, non–small cell lung cancer; OS, overall survival; pCR, pathological complete response; PFS, progression-free survival; RCC, renal cell carcinoma; RFS, recurrence-free survival; SCC, squamous cell carcinoma; SCLC, small cell lung cancer; SOC, standard of care; UC, urothelial carcinoma. a Follow-up times differed between the intervention and control group (intervention follow-up time is listed first). b For IMpower010, patient numbers and DFS for stage II-IIIA disease are reported (as this was defined as a primary end point and data per sex are available for the population with stage II-IIIA disease and not intention to treat). Data with longer follow-up for DFS in stage II-IIIA disease overall (without data per sex) are available and were used for the meta-analysis but are not presented here. c Median follow-up is reported for all patients with censored data. Main Characteristics of the Included Trials Abbreviations: CRT, chemoradiotherapy; DFS, disease-free survival; EC, esophageal cancer; EFS, event-free survival; GC, gastric cancer; GEJ, gastroesophageal junction cancer; HCC, hepatocellular carcinoma; HNSCC, head and neck squamous cell carcinoma; HR, hazard ratio; ICI, immune checkpoint inhibitor; Ipi 3, ipilimumab, 3 mg/kg; Ipi 10, ipilimumab, 10 mg/kg; NR, not reported; NSCLC, non–small cell lung cancer; OS, overall survival; pCR, pathological complete response; PFS, progression-free survival; RCC, renal cell carcinoma; RFS, recurrence-free survival; SCC, squamous cell carcinoma; SCLC, small cell lung cancer; SOC, standard of care; UC, urothelial carcinoma. Follow-up times differed between the intervention and control group (intervention follow-up time is listed first). For IMpower010, patient numbers and DFS for stage II-IIIA disease are reported (as this was defined as a primary end point and data per sex are available for the population with stage II-IIIA disease and not intention to treat). Data with longer follow-up for DFS in stage II-IIIA disease overall (without data per sex) are available and were used for the meta-analysis but are not presented here. Median follow-up is reported for all patients with censored data. The 38 included trials enrolled a total of 31 721 patients, of whom 9407 (29.7%) were female and 22 314 (70.3%) were male. The proportion of female patients per trial ranged from 8%92 to 43%25 (Table). Female participants accounted for less than one-third of the total study population in 29 trials (76.3%).6,7,8,42,44,46,47,48,49,50,53,55,56,57,58,59,61,62,63,64,65,66,67,68,91,92,93,94,95,96 Sex was used as a stratification factor during randomization in only 3 trials (7.9%).7,44,46 Local treatment consisted of surgery in 32 trials (84.2%)4,6,7,8,24,25,26,32,34,41,42,44,47,48,49,50,53,55,56,57,58,59,60,62,63,64,66,67,68,92,94,96 and chemoradiotherapy in 4 trials (10.5%).46,61,65,93 In 1 trial (2.6%), both treatment options were permitted95 and in 1 trial (2.6%), surgery and ablation were allowed.91 Four trials (10.5%) included control arms using ICI treatment.4,24,25,38 In 23 trials (60.5%), an ICI was given after local treatment (as adjuvant or consolidation therapy).24,25,26,32,34,38,41,44,46,47,48,49,50,53,55,56,57,59,60,61,66,67,91,95 ICIs were administered in a neoadjuvant or perioperative setting in 13 trials (34.2%)4,6,7,8,42,58,62,63,64,68,92,94,96 and as a short lead-in treatment before chemoradiotherapy in 2 trials (5.3%),65,93 followed by a consolidation phase (eFigure 1 and eTable in Supplement 1). Overall Analysis Individual Time-to-Event End Points In the overall analysis of all trials and treatment settings, OS was the most reported end point and was extracted from 30 trials (78.9%): in 7 (18.4%) as the primary end point24,41,43,59,61,62,88 and in 23 (60.5%) as the secondary end point.8,25,29,40,45,47,49,52,54,58,64,65,66,67,68,89,90,91,92,93,94,95,96 In comparison with the control arms, ICI therapy in the intervention group was associated with reduced mortality in the overall population (HR, 0.83; 95% CI, 0.79-0.89; P < .001) (eFigure 2 in Supplement 1). There was moderate heterogeneity between the trials (I2 = 28%; Q-test P = .08). The funnel plot showed no evidence for publication bias (Egger test P = .71) (eFigure 23 in Supplement 1). In 13 of those 30 trials (43.3%), OS was reported by sex.8,24,29,40,43,52,61,62,66,88,89,90,95 Benefit of ICIs was shown within each sex group, with an HR of 0.76 (95% CI, 0.67-0.86) in female and 0.78 (95% CI, 0.72-0.85) in male patients (P = .74 for subgroup differences) (Figure 2). Forest Plot Showing Associations of Experimental Immune Checkpoint Inhibitor (ICI) Administration With Overall Survival, Stratified by Sex Squares indicate study-specific hazard ratios (HRs) with square size proportional to study weight, and horizontal lines indicate 95% CIs. Diamonds represent pooled HRs, with outer points indicating 95% CIs. Two trials had control arms using ICIs.24,40 The second most frequently analyzed end point was EFS (15 trials [39.5%]),4,6,8,43,49,62,63,64,68,89,92,93,94,95,96 followed by RFS (11 trials [28.9%]),24,25,29,31,37,40,41,58,60,66,91 DMFS-TDDM (10 trials [26.3%]),7,29,31,37,49,56,60,65,88,90 DFS (10 trials [26.3%]),44,45,47,48,49,52,53,56,57,59,67,90 and PFS (3 trials [7.9%]).61,65,88 Except for 1 EFS end point,49 all EFS, DFS, RFS, and PFS end points reflected primary end points. DMFS-TDDM end points were secondary or exploratory end points. Identically named end points could differ in their definitions, while differently named end points could be conceptually equivalent (definitions and analyses for each end point are presented in eAppendix 6 and eFigures 10-19 in Supplement 1). Combined DFS, EFS, PFS, and RFS By combining the analysis of DFS, EFS, PFS, and RFS, we generated an overall and sex-specific outcome analysis across all 38 trials. The combined primary end points primarily captured time from randomization to disease recurrence, progression, or death, and in trials with a neoadjuvant ICI component, events precluding surgery were generally included (eAppendix 6 in Supplement 1). Owing to variation in end point selection and in the definition of individual end points across trials, the combined analysis is best characterized as an approximate measure of disease control. Incorporation of ICIs was associated with improved outcomes in this combined analysis (HR, 0.73; 95% CI, 0.68-0.77; P < .001) (Figure 3), with evidence of substantial heterogeneity (I2 = 69%; Q-test P < .001) and no statistically significant sex-specific differences (female patients: HR, 0.74 [95% CI, 0.69-0.79]; male patients: HR, 0.73 [95% CI, 0.67-0.78]; P = .66 for subgroup differences) (eFigure 3 in Supplement 1). The funnel plot showed asymmetric distribution and evidence of potential publication bias (Egger test P = .01) (eFigure 23 in Supplement 1). A trim-and-fill sensitivity analysis showed that accounting for potential missing studies did not materially alter the conclusions, supporting the robustness of the pooled estimate. Forest Plot Showing Associations of Experimental Immune Checkpoint Inhibitor (ICI) Administration With Combined Disease-Free, Event-Free, Progression-Free, and Recurrence-Free Survival Squares indicate study-specific hazard ratios (HRs) with square size proportional to study weight, and horizontal lines indicate 95% CIs. Diamonds represent pooled HRs, with outer points indicating 95% CIs. Four trials had control arms using ICIs.4,24,25,40 Ipi 3 indicates ipilimumab, 3 mg/kg; Ipi 10, ipilimumab, 10 mg/kg; SOC, standard of care. By Cancer Type In the cancer-specific analysis of OS and the combined end points, the ICI intervention was associated with improved outcomes in NSCLC,6,7,43,45,47,63,64,88,89,92 melanoma,4,24,25,29,31,37,40,41,60 EC-GEJ-GC,56,62,66,94,96 and UC,8,53,54,59,90 with numerically largest benefit for NSCLC followed by melanoma. Studies of renal cell carcinoma (RCC)48,49,52,57,58 showed a benefit regarding the combined end points but not regarding OS, and the only RCC study incorporating a neoadjuvant ICI component did not meet its primary end point of RFS.58 No significant benefit in either analysis was found for head and neck squamous cell carcinoma (HNSCC)65,68,93,95 (eFigures 20 and 21 in Supplement 1). Sensitivity Analysis in Surgery Trials Overall survival in exclusively surgery trials was analyzed for 21 trials (55.3%).8,29,41,43,45,47,49,52,54,58,59,62,64,66,67,68,89,90,92,94,96 The HR for OS was 0.81 (95% CI, 0.77-0.87; P < .001), without a statistically significant difference between sexes (eFigures 4 and 5 in Supplement 1). Benefit of ICI therapy was numerically greater if given in the combined neoadjuvant and perioperative setting,8,43,58,62,64,68,89,92,94,96 with an HR of 0.77 (95% CI, 0.69-0.86; P < .001), as compared with 0.84 (95% CI, 0.78-0.91; P < .001) in the adjuvant setting29,41,45,47,49,52,54,59,66,67,90 (eFigures 6 and 7 in Supplement 1). The combined end points analysis in strictly surgery trials (29 trials [76.3%])4,6,8,29,31,37,41,43,45,47,48,49,52,53,56,57,58,59,60,62,63,64,66,67,68,89,90,92,94,96 showed an HR of 0.70 (95% CI, 0.65-0.76; P < .001) with the addition of ICIs (eFigure 8 in Supplement 1). Benefit was observed for both sexes without a between-group difference (eFigure 9 in Supplement 1). Benefit of the addition of ICIs could be shown regardless of treatment setting but was greater for the group of trials of neoadjuvant or perioperative ICIs4,6,8,43,58,62,63,64,68,89,92,94,96 (HR, 0.64; 95% CI, 0.57-0.73; P < .001) than in the adjuvant ICI trials29,31,37,41,45,47,48,49,52,53,56,57,59,60,66,67,90 (HR, 0.74; 95% CI, 0.68-0.81; P < .001) (Figure 4). As no formal statistical comparison between the treatment approaches was performed, these between-group differences reflect exploratory findings. Forest Plots Showing Associations of Immune Checkpoint Inhibitor (ICI) Administration With Combined Disease-Free, Event-Free, Progression-Free, and Recurrence-Free Survival by Treatment Timing in Surgery Trials Squares indicate hazard ratios (HRs) with square size proportional to study weight, and horizontal lines indicate 95% CIs. Diamonds represent pooled HRs, with outer points indicating 95% CIs. Trials containing ICIs as a control treatment were excluded except NADINA,4 which used ICIs as adjuvant therapy in the control arm. Ipi 3 indicates ipilimumab, 3 mg/kg; Ipi 10, ipilimumab, 10 mg/kg; SOC, standard of care. We analyzed 11 trials (28.9%) for pCR6,7,8,42,62,63,64,68,92,94,96 and found evidence of increased response with the addition of ICIs (RR, 4.88; 95% CI, 3.11-7.66; P < .001). No sex-specific differences were observed (eFigure 22 in Supplement 1). Discussion In our systematic review and meta-analysis of published phase 3 trials assessing the benefit of ICIs in curative settings in non–sex-dependent cancer types, we found evidence of improved outcomes with ICIs across treatment settings and most tumor types. We did not observe any statistically significant difference in survival outcomes between sexes. Sex-specific reporting for OS was available in 13 of 38 trials, with an HR of 0.76 for female and 0.78 for male patients. The pooled estimate of the combined end points of DFS, EFS, PFS, and RFS (available for all 38 trials), representing a combined analysis of variably defined primary end points reflecting disease control, also revealed no sex-specific differences (HR of 0.74 for females and 0.73 for males). This contrasts with the advanced-disease setting, where meta-analyses have suggested sex differences in treatment outcomes, particularly in melanoma and NSCLC.11,97 In metastatic melanoma, in a nationwide Danish cohort, females undergoing first-line ICI therapy had greater 5-year OS rates than males (47% vs 38%).98 It is not possible to distinguish whether such outcome disparities are due to differences in tumor biology, treatment patterns, or sex-specific patient characteristics or prognostic factors. Of note, a meta-analysis of ICIs for advanced RCC and UC did not find any sex differences in outcome.99 In early-stage disease, potential sex-related differences in ICI performance may be less pronounced than in metastatic cancers, in which ICI efficacy depends more directly on host immune responses and tumor biology,100,101 factors that may amplify underlying sex-related differences in treatment efficacy. Moreover, recurrence-based end points primarily reflect tumor control and are less affected by sex-specific comorbidities or noncancer mortality, whereas sex-related differences in OS might be attenuated by these factors.102 However, sex-specific reporting in this study was largely restricted to primary end points, with clear underreporting for secondary end points. Across all studies, 9407 female and 22 314 male patients were enrolled, with female participants constituting less than one-third of the total study population in over 75% of the trials (29 of 38 studies). The overall female-to-male ratio of approximately 30 to 70 in our study is even lower than the reported ratio of 40 to 60 for global cancer trials for major cancer types between 2000 and 2020,103 a disparity that urgently needs to be addressed. We did not formally assess subgroup credibility using the criteria of Sun et al.104 However, most studies failed to meet the specification of an a priori hypothesis regarding potential sex differences. Additionally, sex was used as a stratification factor during randomization in only 3 out of the 38 trials. These limitations preclude drawing definitive conclusions from the reported absence of significant sex differences in our analysis. Therefore, adequately powered, prospectively designed, sex-based subgroup analyses remain an unmet need. In the analysis of surgery-only trials, the difference in benefit from ICIs with respect to timing was more pronounced for neoadjuvant or perioperative settings than for strictly adjuvant treatment (HRs of 0.64 and 0.74, respectively) for the analysis of combined early end points. For OS, a numerical difference in HR for neoadjuvant and perioperative trials (HR, 0.77) vs adjuvant studies (HR, 0.84) was also seen. Therefore, timing of ICI therapy in the curative setting may matter, at least in the context of surgery, and inclusion of a neoadjuvant component might lead to improved outcomes compared with purely adjuvant treatment strategies. However, these findings derived from separate groups of trials rather than direct randomized comparisons and should therefore be regarded as exploratory. Of note, not all neoadjuvant strategies demonstrated superiority on the individual study level: for RCC, the only study incorporating a neoadjuvant ICI component did not meet its primary end point of RFS,58 thereby limiting the generalizability of superiority of neoadjuvant approaches. Our results are in line with previous reports and analyses. The importance of timing of ICI therapy has already been acknowledged15 and is supported by a clear biological rationale and preclinical evidence.105 ICI treatment induces de novo priming of CD8+ T cells in tumors and in tumor-draining lymph nodes by using the in situ tumor as a reservoir of tumor antigens.106 Intratumoral responses to ICIs seem to be mediated by long-lived resident memory T cells retained in tissue with a specificity for a wide range of tumor antigens, which have been shown to be substantially increased by neoadjuvant ICIs.17,107,108 In melanoma, the treatment paradigm has clearly shifted to a neoadjuvant ICI strategy based on the phase 2 SWOG 1801 trial of anti-PD1 monotherapy5 and the phase 3 NADINA trial4 combining ipilimumab and nivolumab. In both trials, neoadjuvant or perioperative ICI therapy significantly improved EFS compared with adjuvant ICI treatment alone, although with immature OS data. For NSCLC, no direct comparison between neoadjuvant and adjuvant ICI strategies is currently available to our knowledge. Differences in treatment effect across tumor types likely reflect variations in tumor biology and the immune microenvironment. The pronounced benefit in NSCLC and melanoma aligns with their high burden of tumor allelic variation, strong immunogenicity, and established sensitivity to immune checkpoint inhibition.109 However, our subgroup analyses by tumor entity should be interpreted with caution. Only a limited number of trials were available for analysis, and the respective subgroups differed substantially in treatment settings and numbers of trials. Difference in treatment outcomes might reflect methodologic issues and may not necessarily indicate cancer-dependent susceptibility to ICI treatment. The lack of benefit found in RCC could be due to heterogenous trial populations encompassing different risk profiles and consecutive limitations in trial design.110 In HNSCC, the concomitant use of chemoradiotherapy could modulate immune activation and blunt the effect of ICIs. Limitations This study has limitations. The included studies encompassed heterogenous interventional therapies (ICI monotherapy or combination therapy) and different modalities of local therapy, cancer types, and stages. In addition, 4 trials included ICI-containing comparator treatments, which possibly affected the observed treatment outcomes.4,24,25,38 The heterogeneity of this trial landscape may have precluded the detection of sex-specific differences within specific clinical contexts, and outcomes observed in the overall population may therefore not be generalizable across all the included trials. We addressed this by restricting the sensitivity analysis to surgery trials and excluding comparator arms containing ICIs. We also performed a pooling of primary end points primarily reflecting disease recurrence or progression, which might have contributed to heterogeneity of our analysis. Furthermore, the low number of trials with sex as a stratification factor was a methodologic limitation, thus limiting the interpretability of the absence of a statistically significant treatment difference between sexes. While we found an association of ICI implementation in curative treatment settings with OS and combined end points, we did not perform an adjustment for multiple testing, which may have increased the risk of type I error and should be particularly considered when interpreting treatment outcomes in subgroups with respect to ICI timing or tumor entity. Additionally, a formal study-level risk-of-bias assessment was not performed for the included trials, which represents a methodologic limitation of the analysis. Our meta-analysis also was not based on individual patient data, which limits the ability to perform uniform subgroup analyses and adjust for potential confounders across studies. Conclusions In this systematic review and meta-analysis, the addition of ICIs to the management of early-stage cancer was associated with improved outcomes across most cancer types and treatment settings. We detected no statistically significant difference in treatment outcomes by sex. However, none of the included trials were designed or powered to assess sex differences, and female patients were substantially underrepresented. Future studies should ensure adequate female representation and prespecified, adequately powered sex-based subgroup analyses. Timing of ICI therapy appeared to be clinically relevant: greater benefit was observed with the incorporation of a neoadjuvant component. While exploratory, this finding supports the increasing adoption of neoadjuvant ICI therapy approaches.