Real-World Comparative Effectiveness and Safety of Immune Checkpoint Inhibitors in Advanced Renal Cell Carcinoma: A Systematic Review Authors: Chin Hang Yiu, Gwyneth Yuet Yin Leung, Isabelle C. Lee, Christine Y. Lu Published: 2026 ABSTRACT BackgroundImmune checkpoint inhibitor (ICI)-based regimens are the standard first-line therapy for advanced renal cell carcinoma (RCC). However, their comparative effectiveness and safety in routine clinical practice remain incompletely characterised.ObjectiveThe aim of this study was to systematically synthesise comparative real-world evidence on the efficacy and safety of ICI-based regimens in patients with advanced RCC.MethodsA systematic search of MEDLINE, Embase, and Scopus was conducted from database inception to 21 January 2026, in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. Observational cohort studies evaluating ICI-based regimens in advanced RCC with an active comparator were included. Primary efficacy outcomes were overall survival (OS), progression-free survival (PFS), and objective response rate (ORR). Safety outcomes included treatment-related adverse events. Study quality was assessed using the Newcastle–Ottawa Scale, with efficacy and safety evaluated separately. Owing to substantial clinical and methodological heterogeneity, findings were synthesised narratively.ResultsOverall, 58 retrospective cohort studies comprising 35,215 patients were included. Most studies evaluated first-line therapy and involved populations with mixed histology and heterogeneous prognostic risk profiles. Overall, ICI-tyrosine kinase inhibitor (TKI) combinations were generally associated with more favourable survival outcomes compared with dual ICI therapy or TKI monotherapy, with the most consistent benefit observed for PFS. Evidence comparing dual ICI therapy with TKI monotherapy was mixed, with most studies reporting no statistically significant differences in survival outcomes. Comparative safety data were limited, with the majority of studies relying on unadjusted descriptive analyses, limiting robust comparative interpretation of safety outcomes.ConclusionsReal-world evidence suggests that ICI–TKI combinations are associated with more favourable survival outcomes compared with other treatment options in advanced RCC, while the comparative effectiveness of dual ICI therapy versus TKI monotherapy remains inconclusive. Substantial heterogeneity and methodological limitations, particularly in safety reporting, limit definitive interpretation. High-quality real-world studies incorporating robust confounding adjustment and clinically relevant subgroup analyses are needed to better inform treatment selection in routine practice.Trial RegistryInternational Prospective Register of Systematic Reviews (PROSPERO) registration number: CRD420251149614.Supplementary InformationThe online version contains supplementary material available at 10.1007/s11523-026-01227-y. Key Points In real-world clinical practice, immune checkpoint inhibitor–tyrosine kinase inhibitor (ICI–TKI) combinations were generally associated with more favourable survival outcomes compared with dual ICI therapy or TKI monotherapy.Comparative safety evidence for ICI-based regimens remains limited, as most real-world studies relied on unadjusted descriptive analyses, limiting robust assessment of relative toxicity.Substantial population heterogeneity and residual confounding across real-world studies highlight the need for methodologically rigorous comparative analyses to better inform treatment selection in routine clinical practice. In real-world clinical practice, immune checkpoint inhibitor–tyrosine kinase inhibitor (ICI–TKI) combinations were generally associated with more favourable survival outcomes compared with dual ICI therapy or TKI monotherapy.Comparative safety evidence for ICI-based regimens remains limited, as most real-world studies relied on unadjusted descriptive analyses, limiting robust assessment of relative toxicity.Substantial population heterogeneity and residual confounding across real-world studies highlight the need for methodologically rigorous comparative analyses to better inform treatment selection in routine clinical practice. Introduction Kidney cancer is the 14th most frequently diagnosed cancer globally, with an estimated 430,000 new cases and 155,000 deaths annually [1]. Prognosis remains poor for advanced disease, with 5-year relative survival rates of approximately 76% for regional disease and only 19% for distant metastases [2]. For more than a decade, tyrosine kinase inhibitors (TKIs), including sunitinib and sorafenib, represented the standard first-line treatment for advanced renal cell carcinoma (RCC) [3]. The advent of immune checkpoint inhibitors (ICIs) has substantially transformed the treatment landscape of advanced RCC [4]. Dual ICI therapy with nivolumab plus ipilimumab was the first ICI-based regimen approved for first-line treatment of metastatic clear cell RCC, based on the landmark CheckMate-214 trial, which demonstrated superior overall survival compared with sunitinib [5]. Subsequently, ICI–TKI combinations were approved following pivotal phase III trials, including KEYNOTE-581, which showed improved survival outcomes with pembrolizumab–lenvatinib compared with sunitinib monotherapy [6]. As a result, current clinical guidelines recommend ICI-based regimens as standard first-line therapy for patients with advanced RCC across all International Metastatic RCC Database Consortium (IMDC) risk categories, particularly for those with clear cell histology [7, 8]. Randomised controlled trials (RCTs) remain the gold standard for establishing comparative efficacy, but may have limited generalisability to routine clinical practice owing to eligibility criteria and controlled settings [9]. Real-world data (RWD) have therefore become increasingly important for evaluating treatment effectiveness in broader, more heterogeneous patient populations [10]. Emerging evidence suggests that up to 35% of patients with metastatic RCC treated in routine practice would be ineligible for participation in pivotal clinical trials and may experience poorer outcomes [11, 12]. In addition, several observational studies have reported shorter survival with ICI-based therapies in real-world settings than that observed in clinical trials [13–15]. To date, systematic reviews in advanced RCC have largely focused on synthesising evidence from RCTs and indirect treatment comparisons, including network meta-analyses, which primarily estimate comparative efficacy under controlled trial conditions [16–18]. Indirect comparisons from clinical trials also suggest potential heterogeneity in treatment effects across different ICI-based regimens [19]. However, there remains limited synthesis of comparative real-world evidence across ICI-based regimens in advanced RCC. Such evidence may better reflect treatment effectiveness in routine clinical practice, where patient populations are more heterogeneous than those included in clinical trials. Therefore, the aim of this systematic review was to synthesise comparative real-world evidence on the efficacy and safety of ICI-based regimens in patients with advanced RCC. These findings complement existing trial-based evidence and may inform interpretation of treatment effectiveness in routine clinical practice. Methods This systematic review was conducted and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [20]. The review protocol was prospectively registered in the International Prospective Register of Systematic Reviews (PROSPERO; CRD 420251149614). Any protocol amendments and corresponding justifications are documented in Supplementary Table S1. Search Strategy A comprehensive literature search was performed in MEDLINE (via Ovid), Embase, and Scopus from database inception to 21 January 2026. The search strategy combined three key concept domains using Boolean operators: (1) renal cell carcinoma; (2) immune checkpoint inhibitors (including specific agent names); and (3) real-world evidence and observational study designs (including terms such as registry, healthcare claims, and medical record database). The full search strategy for each database is detailed in Supplementary Table S2. Eligibility Criteria Eligible studies were observational cohort studies evaluating outcomes associated with ICI use in patients with advanced RCC, with no restrictions on line of therapy or prior treatment exposure. The inclusion of studies across all treatment lines was intended to capture the breadth of available real-world evidence. Studies were required to include a comparator group to enable evaluation of the comparative efficacy and/or safety of ICI-based versus non-ICI-based standard treatments (e.g. TKI-based regimens). Studies directly comparing different ICI-based regimens (e.g. dual ICIs versus ICI–TKI combinations) were also included. Included studies were required to report at least one efficacy outcome (e.g. survival endpoints) and/or one safety outcome (e.g. adverse events). Studies were excluded if the immunotherapy cohort did not clearly specify the ICI agents used, to avoid inclusion of non-ICI immunotherapies (e.g. interferon or interleukin-based regimens). Additional exclusion criteria included RCTs, case–control studies, commentaries, editorials, narrative or systematic reviews, non-English publications, and conference abstracts or posters without full-text availability. The English-language restriction was applied as previous empirical research has shown that language-restricted systematic reviews do not differ meaningfully from language-inclusive reviews in effect estimates in biomedical fields where English is the dominant language of publication [21]. Outcomes Consistent with a previous systematic review [22], primary efficacy outcomes were overall survival (OS), progression-free survival (PFS), and objective response rate (ORR). Additional efficacy outcomes included, but were not limited to, time on treatment (ToT) and time to next treatment (TTNT). Primary safety outcomes comprised adverse events (AEs), including treatment-related adverse events (TRAEs) and immune-related adverse events (irAEs), without restriction on severity. Additional safety outcomes included treatment discontinuation due to AEs and healthcare resource utilisation measures. Study Selection All records identified through the database searches were imported into Covidence (Veritas Health Innovation, Melbourne, Australia) [23], and duplicates were removed. Titles and abstracts were independently screened by at least two reviewers (C.H.Y., G.Y.Y.L., I.C.L.). Full-text articles of potentially eligible studies were assessed independently by the same reviewers to determine final inclusion. Discrepancies were resolved through discussion and, when necessary, consultation with a senior investigator (C.Y.L.) to achieve consensus. Data Extraction Data extraction of each included study was independently performed by at least two reviewers (C.H.Y., G.Y.Y.L., I.C.L.) using a prespecified data extraction template. Discrepancies were resolved through discussion and, if required, consultation with a senior investigator (C.Y.L.). Extracted data included study characteristics (author, year of publication, country and data source), sample size (overall and by study cohort), population characteristics (line of treatment, histology and IMDC risk classification), and reported efficacy and/or safety outcomes. Quality Assessment Methodological quality of the included non-randomised, non-interventional cohort studies was independently assessed by two investigators (C.H.Y. and G.Y.Y.L.) using the Newcastle–Ottawa Scale (NOS) [24]. The NOS tool has been applied in prior systematic reviews of real-world comparative effectiveness and safety of ICIs in colorectal cancer and small cell lung cancer [22, 25]. Studies were evaluated across three domains: selection, comparability and outcome assessment. Overall study quality was classified as good, fair or poor according to Agency for Healthcare Research and Quality (AHRQ) criteria. In line with a previous systematic review of real-world evidence [26], efficacy and safety outcomes were evaluated separately, reflecting differences in analytical approaches and outcome ascertainment across studies. This approach is consistent with the methodological principle that risk of bias may vary across outcomes within the same study depending on confounding control and outcome measurement methods [27]. Data Synthesis Given substantial clinical and methodological heterogeneity across studies, quantitative meta-analysis was not performed. Heterogeneity was evident across three domains: clinical heterogeneity (including histological subtypes, IMDC risk group, treatment line and comparator type); methodological heterogeneity (including unadjusted analyses, multivariable regression, propensity score matching and inverse probability weighting); and reporting heterogeneity (including inconsistent availability and format of effect estimates). Many studies did not report effect estimates in a standardised format suitable for quantitative pooling. Under these conditions, meta-analysis would risk generating summary estimates that combine fundamentally different patient populations, interventions and analytical frameworks. This approach is consistent with Cochrane guidance, which advises against meta-analysis when clinical and methodological heterogeneity is present [28]. Accordingly, findings were synthesised narratively. Results The database search identified 3589 records. After removal of duplicates, 2303 records remained for title and abstract screening. Of these, 120 full-text articles were assessed for eligibility. Applying the predefined inclusion and exclusion criteria resulted in 58 studies being included in this systematic review (Fig. 1) [29–86]. Two studies that otherwise met the inclusion criteria were excluded because it was unclear whether the immunotherapy cohorts consisted exclusively of ICI agents [87, 88]. Notably, a high-quality study by Chakiryan et al. employing 1:1:1 propensity score matching reported that patients in the immunotherapy cohorts may have received non-ICI immunotherapies (e.g. interferon-α or high-dose interleukin-2) [87], precluding definitive attribution of observed outcomes to ICI treatment.Fig. 1PRISMA flow diagram of the study selection process [20]. ICIs immune checkpoint inhibitors PRISMA flow diagram of the study selection process [20]. ICIs immune checkpoint inhibitors Study Characteristics Table 1 summarises the key characteristics of the included studies [29–86]. All studies were retrospective cohort designs and collectively included 35,215 patients with advanced or metastatic RCC, with individual study sample sizes ranging from 25 to 6297 (median 221) patients. Most studies were conducted in Japan (n = 18) [34, 37–39, 42, 60, 62, 64, 65, 71, 73, 77, 78, 80, 81, 84], followed by international studies (n = 13) [30, 35, 36, 41, 43, 47, 53, 55, 66, 69, 74, 76, 82], and the USA (n = 11) [32, 40, 48, 54, 56, 58, 61, 67, 72, 83, 85]. Only two studies used healthcare administrative or insurance claims data [40, 50], while other studies primarily used clinical data from electronic health records or cancer registries [29–39, 41–49, 51–86]. Table 1Baseline characteristics of all included studies (n = 58)Study (year)CountryData sourceSample sizePopulationLine of treatmentRCC histologyIMDC riskDe Vries-Brilland et al. (2025) [29]CanadaCanadian Kidney Cancer Information System197MetastaticFirst-lineNon-clear cell only (100% papillary)All risk groups included (favourable, intermediate, poor)Massari et al. (2025) [30]InternationalARON-1 global real-world database (across 56 centres in 17 countries)323MetastaticFirst-lineNon-clear cell only (predominantly papillary)Intermediate or poor onlyHilser et al. (2025) [31]GermanyGUARDIANS project (13 academic centres)121MetastaticFirst-lineNon-clear cell only (100% papillary)All risk groups included (favourable, intermediate, poor)Ostrowski et al. (2025) [32]USAFlatiron Health nationwide EHR-derived de-identified database1438MetastaticFirst-lineClear cell onlyIntermediate or poor onlyHaack et al. (2025) [33]GermanyUniversity Medical Center Mainz59Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Ishihara et al. (2025) [34]Japan5 academic centres (Tokyo Women’s Medical University and affiliates)194Advanced or metastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlySantoni et al. (2025) [35]InternationalARON-1 global real-world database (across 55 centres in 16 countries)760MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Gupta et al. (2025) [36]InternationalIMDC dataset3758MetastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyKikuta et al. (2025) [37]Japan3 academic institutions172MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Hara et al. (2025) [38]JapanMulticentre (Kobe University and affiliated centres)171MetastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyYanagisawa et al. (2024) [39]Japan6 university hospitals and 15 affiliated hospitals228MetastaticFirst-lineClear cell onlyIntermediate or poor onlyShah et al. (2024) [40]USAOptum Research Database (administrative medical and pharmacy claims)507MetastaticFirst-lineNot reported (not available in claims data)Not reported (not available in claims data)Roviello et al. (2025) [41]InternationalARON-1 global real-world database (across 55 centres in 18 countries)524MetastaticFirst-lineMixed (predominantly clear cell)Favourable onlyIshikawa et al. (2025) [42]JapanHamamatsu University School of Medicine125Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Ciccarese et al. (2024) [43]InternationalARON-1 global real-world database (across 55 centres in 18 countries)226Metastatic (with sarcomatoid features)First-lineMixed (predominantly clear cell)Intermediate or poor onlyYoshimura et al. (2024) [44]Japan9 institutions44MetastaticFirst-lineNon-clear cell only (predominantly papillary)All risk groups included (favourable, intermediate, poor)Bickley et al. (2024) [45]UKSt Bartholomew’s Hospital, London65MetastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyLiu et al. (2024) [46]ChinaZibo Central Hospital216Advanced or metastatic (older adults ≥ 60 years)First-lineClear cell onlyNot specifiedMassari et al. (2024) [47]InternationalARON-1 global real-world database (across 40 centres in 12 countries)200MetastaticFirst-lineNon-clear cell only (100% papillary)All risk groups included (favourable, intermediate, poor)Gebrael et al. (2024) [48]USAFlatiron Health nationwide EHR-derived de-identified database348MetastaticSecond-lineClear cell onlyAll risk groups included (favourable, intermediate, poor)Ishihara et al. (2024) [49]JapanTokyo Women’s Medical University and its 4 affiliated institutions175Advanced or metastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyAlbigès et al. (2023) [50]FranceThe French nationwide healthcare insurance system database (SNDS)872Advanced or metastaticFirst-lineNot specifiedNot specifiedSawaya et al. (2024) [51]CanadaCanadian Kidney Cancer information system1202MetastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor only Huang et al. 2024 [52] ChinaSun Yat-sen University Centre65MetastaticSecond-lineNon-clear cell only (predominantly papillary)All risk groups included (favourable, intermediate, poor)Esterberg et al. (2024) [53]InternationalChart review from 158 physicians across 6 countries498Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Blas et al. (2024) [54]USAFood and Drug Administration Adverse Event Reporting System (FAERS) database2921Advanced or metastaticNot specifiedNot specifiedNot specifiedSantoni et al. (2024) [55]InternationalARON-1 global real-world database (across 58 centres in 20 countries)930Advanced or metastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyLai et al. (2023) [56]USADatabase of TriNetX network6297MetastaticFirst-lineNot specifiedNot specifiedPoprach et al. (2023) [57]Czech RepublicRenIS (national registry of patients with metastatic RCC treated with non-immunological targeted therapy or with immunotherapy)1981MetastaticAnyMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Shah et al. (2023) [58]USAThe US Oncology Network (iKnowMed EHR)331MetastaticFirst-lineClear cell onlyAll risk groups included (favourable, intermediate, poor)Al-Mansour et al. (2024) [59]Saudi ArabiaPrincess Norah Oncology Centre in Jeddah and the Oncology Department at National Guard Hospital in Riyadh84MetastaticFirst-lineClear cell onlyAll risk groups included (favourable, intermediate, poor)Teishima et al. (2024) [60]JapanKobe University Hospital or an affiliated institute36MetastaticFirst-lineNon-clear cell only (predominantly papillary)All risk groups included (favourable, intermediate, poor)Shah et al. (2023) [61]USAMcKesson Specialty Health in the iKnowMed database (oncology-specific EHR system)1538MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Iinuma et al. (2023) [62]Japan9 Japanese institutions51Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)van Laar et al. (2022) [63]NetherlandsLeiden University Medical Center and the Haga Hospital138MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Kimura et al. (2020) [64]JapanOgaki Municipal Hospital25MetastaticSecond-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Ishihara et al. (2023) [65]JapanTokyo Women’s Medical University, Tokyo Women’s Medical University Adachi Medical Center, Saiseikai Kawaguchi General Hospital, Saiseikai Kazo Hospital, and Jyoban Hospital467Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Carril-Ajuria et al. (2023) [66]International14 institutions across 4 countries70MetastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyZarrabi et al. (2023) [67]USAFlatiron Health nationwide EHR-derived de-identified database1506MetastaticFirst-lineClear cell onlyAll risk groups included (favourable, intermediate, poor)Stühler et al. (2023) [68]GermanyDepartment of Urology of the University of Tuebingen201MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Santoni et al. (2022) [69]International11 international institutions343MetastaticSecond-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Hoeh et al. (2022) [70]Germany6 tertiary referral centres104MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Bando et al. (2022) [71]JapanKobe University Hopsital33MetastaticFirst-lineNon-clear cell only (predominantly papillary)All risk groups included (favourable, intermediate, poor)Bhanegaonkar et al. (2021) [72]USAUS Veterans Health Administration Database (172 VA medical centres and 1069 outpatient sites)759MetastaticFirst-lineNot specifiedNot specifiedKido et al. (2021) [73]Japan2 academic and 5 satellite centres278MetastaticFirst-lineClear cell onlyIntermediate or poor onlyStukalin et al. (2019) [74]InternationalIMDC dataset (38 international cancer centres, 10 countries)278MetastaticSecond-lineNon-clear cell onlyAll risk groups included (favourable, intermediate, poor)McGrane et al. (2024) [75]UK17 centres1286MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Santoni et al. (2023) [76]InternationalARON-1 global real-world database (across 47 centres in 16 countries)729MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Kato et al. (2022) [77]JapanIwate Medical University Hospital72Advanced or metastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyUeda et al. (2022) [78]JapanKurume University Hospital202Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Bai et al. (2024) [79]ChinaXijing Hospital55Metastatic/recurrentNot specifiedNon-clear cell onlyFavourable or intermediate onlyUeda et al. (2026) [80]Japan18 institutions in the Kyushu–Okinawa area88Advanced or metastatic (undergoing haemodialysis)First-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Mukae et al. (2026) [81]JapanNagasaki University Hospital58Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Massari et al. (2025) [82]InternationalARON-1 global real-world database (across 66 centres in 21 countries)350Metastatic (with sarcomatoid features)First-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Geynisman et al. (2025) [83]USARetrospective, multisite physician-abstracted medical chart review355Advanced or metastaticFirst-lineClear cell onlyAll risk groups included (favourable, intermediate, poor)Ikoma et al. (2025) [84]JapanNagoya City University Hospital and 7 affiliated institutions156MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Meng et al. (2025) [85]USAOhio State University Comprehensive Cancer Centre and Fred Hutichinson Cancer Centre69Metastatic (bone predominant)First-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Goebell et al. (2025) [86]GermanyRenal Cell Carcinoma Research Platform (160 study sites, hospital and office–based practices)936Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)EHR electronic health record, IMDC International Metastatic Renal Cell Carcinoma Database Consortium, RCC renal cell carcinoma Table 1Baseline characteristics of all included studies (n = 58)Study (year)CountryData sourceSample sizePopulationLine of treatmentRCC histologyIMDC riskDe Vries-Brilland et al. (2025) [29]CanadaCanadian Kidney Cancer Information System197MetastaticFirst-lineNon-clear cell only (100% papillary)All risk groups included (favourable, intermediate, poor)Massari et al. (2025) [30]InternationalARON-1 global real-world database (across 56 centres in 17 countries)323MetastaticFirst-lineNon-clear cell only (predominantly papillary)Intermediate or poor onlyHilser et al. (2025) [31]GermanyGUARDIANS project (13 academic centres)121MetastaticFirst-lineNon-clear cell only (100% papillary)All risk groups included (favourable, intermediate, poor)Ostrowski et al. (2025) [32]USAFlatiron Health nationwide EHR-derived de-identified database1438MetastaticFirst-lineClear cell onlyIntermediate or poor onlyHaack et al. (2025) [33]GermanyUniversity Medical Center Mainz59Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Ishihara et al. (2025) [34]Japan5 academic centres (Tokyo Women’s Medical University and affiliates)194Advanced or metastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlySantoni et al. (2025) [35]InternationalARON-1 global real-world database (across 55 centres in 16 countries)760MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Gupta et al. (2025) [36]InternationalIMDC dataset3758MetastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyKikuta et al. (2025) [37]Japan3 academic institutions172MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Hara et al. (2025) [38]JapanMulticentre (Kobe University and affiliated centres)171MetastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyYanagisawa et al. (2024) [39]Japan6 university hospitals and 15 affiliated hospitals228MetastaticFirst-lineClear cell onlyIntermediate or poor onlyShah et al. (2024) [40]USAOptum Research Database (administrative medical and pharmacy claims)507MetastaticFirst-lineNot reported (not available in claims data)Not reported (not available in claims data)Roviello et al. (2025) [41]InternationalARON-1 global real-world database (across 55 centres in 18 countries)524MetastaticFirst-lineMixed (predominantly clear cell)Favourable onlyIshikawa et al. (2025) [42]JapanHamamatsu University School of Medicine125Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Ciccarese et al. (2024) [43]InternationalARON-1 global real-world database (across 55 centres in 18 countries)226Metastatic (with sarcomatoid features)First-lineMixed (predominantly clear cell)Intermediate or poor onlyYoshimura et al. (2024) [44]Japan9 institutions44MetastaticFirst-lineNon-clear cell only (predominantly papillary)All risk groups included (favourable, intermediate, poor)Bickley et al. (2024) [45]UKSt Bartholomew’s Hospital, London65MetastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyLiu et al. (2024) [46]ChinaZibo Central Hospital216Advanced or metastatic (older adults ≥ 60 years)First-lineClear cell onlyNot specifiedMassari et al. (2024) [47]InternationalARON-1 global real-world database (across 40 centres in 12 countries)200MetastaticFirst-lineNon-clear cell only (100% papillary)All risk groups included (favourable, intermediate, poor)Gebrael et al. (2024) [48]USAFlatiron Health nationwide EHR-derived de-identified database348MetastaticSecond-lineClear cell onlyAll risk groups included (favourable, intermediate, poor)Ishihara et al. (2024) [49]JapanTokyo Women’s Medical University and its 4 affiliated institutions175Advanced or metastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyAlbigès et al. (2023) [50]FranceThe French nationwide healthcare insurance system database (SNDS)872Advanced or metastaticFirst-lineNot specifiedNot specifiedSawaya et al. (2024) [51]CanadaCanadian Kidney Cancer information system1202MetastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor only Huang et al. 2024 [52] ChinaSun Yat-sen University Centre65MetastaticSecond-lineNon-clear cell only (predominantly papillary)All risk groups included (favourable, intermediate, poor)Esterberg et al. (2024) [53]InternationalChart review from 158 physicians across 6 countries498Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Blas et al. (2024) [54]USAFood and Drug Administration Adverse Event Reporting System (FAERS) database2921Advanced or metastaticNot specifiedNot specifiedNot specifiedSantoni et al. (2024) [55]InternationalARON-1 global real-world database (across 58 centres in 20 countries)930Advanced or metastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyLai et al. (2023) [56]USADatabase of TriNetX network6297MetastaticFirst-lineNot specifiedNot specifiedPoprach et al. (2023) [57]Czech RepublicRenIS (national registry of patients with metastatic RCC treated with non-immunological targeted therapy or with immunotherapy)1981MetastaticAnyMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Shah et al. (2023) [58]USAThe US Oncology Network (iKnowMed EHR)331MetastaticFirst-lineClear cell onlyAll risk groups included (favourable, intermediate, poor)Al-Mansour et al. (2024) [59]Saudi ArabiaPrincess Norah Oncology Centre in Jeddah and the Oncology Department at National Guard Hospital in Riyadh84MetastaticFirst-lineClear cell onlyAll risk groups included (favourable, intermediate, poor)Teishima et al. (2024) [60]JapanKobe University Hospital or an affiliated institute36MetastaticFirst-lineNon-clear cell only (predominantly papillary)All risk groups included (favourable, intermediate, poor)Shah et al. (2023) [61]USAMcKesson Specialty Health in the iKnowMed database (oncology-specific EHR system)1538MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Iinuma et al. (2023) [62]Japan9 Japanese institutions51Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)van Laar et al. (2022) [63]NetherlandsLeiden University Medical Center and the Haga Hospital138MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Kimura et al. (2020) [64]JapanOgaki Municipal Hospital25MetastaticSecond-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Ishihara et al. (2023) [65]JapanTokyo Women’s Medical University, Tokyo Women’s Medical University Adachi Medical Center, Saiseikai Kawaguchi General Hospital, Saiseikai Kazo Hospital, and Jyoban Hospital467Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Carril-Ajuria et al. (2023) [66]International14 institutions across 4 countries70MetastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyZarrabi et al. (2023) [67]USAFlatiron Health nationwide EHR-derived de-identified database1506MetastaticFirst-lineClear cell onlyAll risk groups included (favourable, intermediate, poor)Stühler et al. (2023) [68]GermanyDepartment of Urology of the University of Tuebingen201MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Santoni et al. (2022) [69]International11 international institutions343MetastaticSecond-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Hoeh et al. (2022) [70]Germany6 tertiary referral centres104MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Bando et al. (2022) [71]JapanKobe University Hopsital33MetastaticFirst-lineNon-clear cell only (predominantly papillary)All risk groups included (favourable, intermediate, poor)Bhanegaonkar et al. (2021) [72]USAUS Veterans Health Administration Database (172 VA medical centres and 1069 outpatient sites)759MetastaticFirst-lineNot specifiedNot specifiedKido et al. (2021) [73]Japan2 academic and 5 satellite centres278MetastaticFirst-lineClear cell onlyIntermediate or poor onlyStukalin et al. (2019) [74]InternationalIMDC dataset (38 international cancer centres, 10 countries)278MetastaticSecond-lineNon-clear cell onlyAll risk groups included (favourable, intermediate, poor)McGrane et al. (2024) [75]UK17 centres1286MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Santoni et al. (2023) [76]InternationalARON-1 global real-world database (across 47 centres in 16 countries)729MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Kato et al. (2022) [77]JapanIwate Medical University Hospital72Advanced or metastaticFirst-lineMixed (predominantly clear cell)Intermediate or poor onlyUeda et al. (2022) [78]JapanKurume University Hospital202Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Bai et al. (2024) [79]ChinaXijing Hospital55Metastatic/recurrentNot specifiedNon-clear cell onlyFavourable or intermediate onlyUeda et al. (2026) [80]Japan18 institutions in the Kyushu–Okinawa area88Advanced or metastatic (undergoing haemodialysis)First-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Mukae et al. (2026) [81]JapanNagasaki University Hospital58Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Massari et al. (2025) [82]InternationalARON-1 global real-world database (across 66 centres in 21 countries)350Metastatic (with sarcomatoid features)First-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Geynisman et al. (2025) [83]USARetrospective, multisite physician-abstracted medical chart review355Advanced or metastaticFirst-lineClear cell onlyAll risk groups included (favourable, intermediate, poor)Ikoma et al. (2025) [84]JapanNagoya City University Hospital and 7 affiliated institutions156MetastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Meng et al. (2025) [85]USAOhio State University Comprehensive Cancer Centre and Fred Hutichinson Cancer Centre69Metastatic (bone predominant)First-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)Goebell et al. (2025) [86]GermanyRenal Cell Carcinoma Research Platform (160 study sites, hospital and office–based practices)936Advanced or metastaticFirst-lineMixed (predominantly clear cell)All risk groups included (favourable, intermediate, poor)EHR electronic health record, IMDC International Metastatic Renal Cell Carcinoma Database Consortium, RCC renal cell carcinoma Baseline characteristics of all included studies (n = 58) EHR electronic health record, IMDC International Metastatic Renal Cell Carcinoma Database Consortium, RCC renal cell carcinoma With respect to treatment line, 50 studies evaluated first-line therapy [29–47, 49–51, 53, 55, 56, 58–63, 65–68, 70–73, 75–78, 80–86], 5 evaluated second-line therapy [48, 52, 64, 69, 74], 1 included any treatment line [57], and 2 did not specify treatment line [54, 79]. Regarding histology, most studies (n = 34) included mixed histology, with patient populations predominantly composed of clear-cell RCC rather than papillary RCC [33–35, 37, 38, 41–43, 45, 49, 51, 53, 55, 57, 61–66, 68–70, 75, 76, 78, 80–82, 84–86]. In terms of IMDC risk classification, the majority of studies (n = 36) included patients across all risk groups (favourable, intermediate and poor) [29, 31, 33, 35, 37, 42, 44, 47, 48, 52, 53, 57–65, 67, 68, 70, 71, 74–76, 78, 80–86], with a higher proportion of intermediate- and poor-risk patients. This was followed by studies restricted to intermediate- and/or poor-risk patients only (n = 14) [30, 32, 34, 36, 38, 39, 43, 45, 49, 51, 55, 66, 73, 77]. The most frequently evaluated treatment cohorts were ICI plus TKI combinations (ICI–TKI), dual ICI (ICI–ICI), and TKI monotherapy. Specific agents investigated within these regimens were generally consistent with regulatory approvals and clinical practice guidelines, such as pembrolizumab–lenvatinib and nivolumab–cabozantinib (ICI–TKI), nivolumab–ipilimumab (the only ICI–ICI regimen investigated), and sunitinib monotherapy (TKI), supporting the real-world relevance of the included evidence. Tables 2 and 3 summarise the efficacy and safety outcomes reported by included studies, respectively, along with the corresponding quality ratings. Table 2Efficacy outcomes reported by included studies (n = 57)Study (year)Study cohorts, sample size (n)Efficacy outcome(s)Quality of evidenceDe Vries-Brilland et al. (2025) [29]Cohort 1: ICI-based (n = 70)• ICI–TKI• ICI–ICI• ICI monotherapyCohort 2: TKI monotherapy• Predominantly sunitinibICI-based (combined) versus TKI monotherapyOS: median 36.9 versus 26.5 months; aHR 0.54 (95% CI 0.30–0.90)TTF: median 9.9 versus 5.9 months; aHR 0.62 (95% CI 0.42–0.91)ORR: 37 versus 23%; aHR 2.2 (95% CI 0.95–5.2)ICI–TKI versus TKI monotherapyOS: aHR 0.26 (95% CI 0.08–0.83)TTF: aHR 0.47 (95% CI 0.26–0.85)ORR: 38 versus 23%ICI–ICI versus TKI monotherapyOS: aHR 0.90 (95% CI 0.47–1.72)TTF: aHR 2.04 (95% CI 1.2–3.4)ORR: 39 versus 23%GoodOstrowski et al. (2025) [32]Cohort 1: ICI–TKI (n = 659)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinib• Avelumab–axitinibCohort 2: ICI–ICI (n = 779)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIPS-matched OS: median 25.0 versus 28.8 months; aHR 1.01 (95% CI 0.86–1.19); sensitivity adjusted Cox HR 0.92 (95% CI 0.77–1.10)PS-matched TTNT: median 13.1 versus 7.8 months; aHR 0.78 (95% CI 0.68–0.89); sensitivity adjusted Cox HR 0.75 (95% CI 0.65–0.86)GoodHaack et al. (2025) [33]Cohort 1: ICI–TKI (n = 43)• Pembrolizumab–axitinib• Pembrolizumab–lenvatinib• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 16)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: median 33.0 versus 22.0 months; HR 0.74 (95% CI 0.36–1.51)PFS: median 14.0 versus 6.0 months; HR 0.53 (95% CI 0.28–0.99); multivariable Cox regression confirmed ICI–TKI as an independent predictor of improved PFS: aHR 0.47 (95% CI 0.24–0.92)ORR: 74% versus 44%, p = 0.035ToT: median 14.0 versus 6.0 months; HR 0.62 (95% CI 0.34–1.14), p = 0.109GoodIshihara et al. (2025) [34]Cohort 1: ICI–TKI (n = 85)• Pembrolizumab–lenvatinib• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 109)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIIPTW-adjusted OS: median not reached versus 56.0 months; aHR 0.63 (95% CI 0.32–1.26)IPTW-adjusted PFS: median 20.6 versus 9.0 months; aHR 0.52 (95% CI 0.34–0.79)IPTW-adjusted ORR: 74% versus 44%, p = 0.035IPTW-adjusted TTF: median 17.3 versus 4.8 months, IPTW-adjusted log-rank p < 0.001GoodSantoni et al. (2025) [35]Cohort 1: ICI–TKI (nivolumab–cabozantinib, n = 153)Cohort 2: ICI–TKI (pembrolizumab–axitinib, n = 607)Nivolumab–cabozantinib versus pembrolizumab–axitinibOS: median not reached versus 55.7 months; aHR 0.87 (95% CI 0.58–1.31)PFS: 27.6 versus 16.2 months; aHR 0.53 (95% CI 0.37–0.74)ORR: 59 versus 49%, p = 0.51GoodKikuta et al. (2025) [37]Cohort 1: ICI–TKI (n = 64)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 108)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIIPTW-adjusted OS: HR 0.74 (95% CI 0.35–1.56)IPTW-adjusted PFS: HR 0.72 (95% CI 0.40–1.29)IPTW-adjusted ORR: OR 2.34 (95% CI 1.14–4.82)GoodHara et al. (2025) [38]Cohort 1: ICI–TKI (n = 74)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 97)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: aHR 0.70 (95% CI 0.35–1.39)PFS: aHR 0.69 (95% CI 0.42–1.13)ORR: 54.9 versus 37.2%, p < 0.001GoodYanagisawa et al. (2024) [39]Cohort 1: ICI–TKI (n = 114)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 114)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIPS-matched OS: median not reached versus 53 months, p = 0.24PS-matched PFS: median 25 versus 13 months, p = 0.047PS-matched ORR: 55 versus 51%, p = 0.8GoodShah et al. (2024) [40]Cohort 1: ICI–TKI (n = 126)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 381)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: not estimable in either group due to short follow-up; 12-month survival 77.3 versus 66.9%ToT: 12.4 versus 4.1 months, p < 0.001GoodBickley et al. (2024) [45]Cohort 1: ICI–TKI (n = 14)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Ipilimumab–nivolumab–cabozantinib• Atezolizumab–bevacizumabCohort 2: ICI–ICI (n = 23)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 28)• Sunitinib• PazopanibICI–TKI versus ICI–ICI versus TKI monotherapyOS: median 40.0 versus 12.9 versus 15.6 months• ICI–TKI versus TKI monotherapy: aHR 0.61 (95% CI 0.27–1.36)• ICI–TKI versus ICI–ICI: aHR 0.82 (95% CI 0.31–2.17)PFS: median 14.7 versus 5.1 versus 6.7 months• ICI–TKI versus TKI monotherapy: aHR 0.47 (95% CI 0.24–0.92)• ICI–TKI versus ICI–ICI: aHR 0.38 (95% CI 0.19–0.78)GoodMassari et al. (2024) [47]Cohort 1: ICI-based (n = 73)• Pembrolizumab–axitinib (ICI–TKI)• Nivolumab–cabozantinib (ICI–TKI)• Pembrolizumab–lenvatinib (ICI–TKI)• Nivolumab–ipilimumab (ICI–ICI)Cohort 2: TKI monotherapy (n = 127)• Sunitinib• Cabozantinib• PazopanibICI-based (combined) versus TKI monotherapyOS: median 28.8 versus 22.5 months, p = 0.081PFS: median 17.4 versus 6.4 months; p < 0.001• aHR 2.31 (95% CI 1.55–3.44), p < 0.001 (TKI versus ICI-based)• Median PFS was significantly longer for ICI-based versus TKI monotherapy in the IMDC good-risk (39.7 versus 5.8 months, p < 0.001) and intermediate-risk (12.2 versus 7.3 months, p = 0.047) subgroups, but the difference did not reach significance in the poor-risk subgroup (12.2 versus 6.2 months, p = 0.093).ORR: 41 versus 27%GoodGebrael et al. (2024) [48]Cohort 1: ICI–TKI (n = 68)• Nivolumab–cabozantinib• Pembrolizumab–cabozantinib• Avelumab–cabozantinib• Nivolumab–pembrolizumab–cabozantinibCohort 2: TKI monotherapy (n = 280)• CabozantinibICI–TKI versus TKI monotherapyPS-matched OS: median 21 versus 20 months; HR 1.15 (95% CI 0.73–1.79)PS-matched TTNT: median 18.6 versus 10.1 months; HR 0.74 (95% CI 0.49–1.12)GoodIshihara et al. (2024) [49]Cohort 1: ICI–TKI (n = 72)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 103)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIIPTW-adjusted OS: median 46.7 versus 49.0 months; p = 0.465IPTW-adjusted PFS: median 15.6 versus 8.3 months; p = 0.04IPTW-adjusted ORR: 51.2 versus 43.9%, p = 0.36IPTW-adjusted PD rate (best response): 3.3 versus 27.4%, p < 0.0001GoodSawaya et al. (2024) [51]Cohort 1: ICI-based (n = 411)• ICI–TKI (unspecified)• ICI–ICI (nivolumab–ipilimumab)Cohort 2: TKI monotherapy (n = 791)• Mainly sunitinibICI-based versus TKI monotherapy (all patients):IPTW-adjusted OS: median 72 versus 36 months; HR 0.58 (95% CI 0.47–0.71)IPTW-adjusted ORR: 40.8 versus 21.6% (p < 0.0001)ICI-based versus TKI monotherapy (Sarcomatoid subgroup):IPTW-adjusted OS: median 48 versus 18 months; HR 0.41 (95% CI 0.26–0.64)IPTW-adjusted ORR: 49.5 versus 13.8% (p < 0.0001)ICI-based versus TKI monotherapy (non-sarcomatoid subgroup):IPTW-adjusted OS: median 72 versus 48 months; HR 0.63 (95% CI 0.50–0.80)IPTW-adjusted ORR: 38.5 versus 23.5% (p < 0.0001)ICI–TKI versus ICI–ICIIPTW-adjusted OS: median 48 versus 60 months; HR 1.28 (95% CI 0.89–1.84)IPTW-adjusted ORR: 42.9 versus 39.2% (p = 0.0003)Good Huang et al. 2024 [52] Cohort 1: ICI–TKI (n = 44)• Anti-PD-1 agent + axitinib/sunitinib/pazopanib/ cabozantinib/lenvatinib/apatinib/sorafenibCohort 2: TKI monotherapy (n = 21)• AxitinibICI–TKI versus TKI monotherapyIPTW-adjusted OS: median 31.9 versus 16.2 months, p = 0.133IPTW-adjusted PFS: median 9.2 versus 5.4 months, p = 0.002IPTW-adjusted ORR: 50 versus 14.3%, p = 0.006GoodEsterberg et al. (2024) [53]Cohort 1: ICI-based (n = 197)• ICI–TKI (pembrolizumab–axitinib)• ICI–TKI (nivolumab–cabozantinib)• ICI–TKI (avelumab–axitinib)• ICI–ICI (nivolumab–ipilimumab)Cohort 2: TKI monotherapy (n = 250)• Sunitinib• Pazopanib• Sorafenib• Cabozantinib• AxitinibICI-based versus TKI monotherapyOS: median 72 versus 36 months; aHR 0.66 (95% CI 0.39–1.11)PFS: median 27.1 versus 15.5 months; aHR 0.50 (95% CI 0.36–0.72)TTNT: median 30.1 versus 16.7 months; aHR 0.54 (95% CI 0.39–0.73)ICI–TKI versus ICI–ICIPFS: median not reached versus 23.0 months; aHR 0.50 (95% CI 0.28–0.97)GoodPoprach et al. (2023) [57]Cohort 1: ICI-based (n = 214)• ICI monotherapy (nivolumab)• ICI–ICI (nivolumab–ipilimumab)• ICI–TKI (avelumab–axitinib)Cohort 2: TT only (n = 1767)• Included predominantly TKIs (e.g. sunitinib, axitinib), as well as other targeted agents such as mTOR inhibitors (e.g. everolimus)ICI-based versus TT (overall)OS: median not reached versus 24.5 months; aHR 0.23 (p < 0.001)ICI-based versus TT (1L)OS: median not reached versus 24.5 months; aHR 0.23 (p = 0.003)PFS: median 11.9 versus 9.3 months; aHR 0.63 (p = 0.046)ICI-based versus TT (2L)OS: median 37.8 versus 16.6 months; aHR 0.35 (p < 0.001)PFS: median 13.9 versus 6.3 months; aHR 0.49 (p < 0.001)ICI-based versus TT (3L)OS: median 30.4 versus 18.3 months; aHR 0.70 (p = 0.101)PFS: median 10.1 versus 6.0 months; aHR 0.63 (p = 0.005)GoodShah et al. (2023) [58]Cohort 1: ICI–TKI (n = 145)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 186)• Nivolumab–ipilimumabICI–TKI versus ICI–ICI (overall)PFS: median 13.9 versus 9.8 months, aHR 0.70 (95% CI 0.49–0.99)ORR: 71 versus 45.2%, p < 0.001TTNT: median 21.3 versus 11.9 months, aHR 0.60 (95% CI 0.42–0.87)ToT: median 10.5 versus 4.2 months, aHR 0.53 (95% CI 0.40–0.71)ICI–TKI versus ICI–ICI (subgroup of IMDC intermediate- or poor-risk patients)PFS: aHR 0.70 (95% CI 0.48–1.03)TTNT: aHR 1.00 (95% CI 0.98–1.02)ToT: aHR 0.52 (95% CI 0.38–0.72)GoodShah et al. (2023) [61]Cohort 1: ICI–TKI (n = 279)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 618)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 641)• Cabozantinib• Sunitinib• Pazopanib• AxitinibICI–TKI versus ICI–ICI versus TKIOS: median not reached versus 27.6 versus 26.9 months, p = 0.24TTNT: median 16.4 versus 8.3 versus 8.4 months, p < 0.001ToT: median 13.6 versus 5.8 versus 3.4 months, p < 0.001GoodIshihara et al. (2023) [65]Cohort 1: ICI-based (n = 152)• Nivolumab–ipilimumab (ICI–ICI)• Pembrolizumab–axitinib (ICI–TKI)• Avelumab–axitinib (ICI–TKI)• Nivolumab–cabozantinib (ICI–TKI)• Pembrolizumab–lenvatinib (ICI–TKI)Cohort 2: TKI monotherapy (n = 315)• Sorafenib• Sunitinib• PazopanibICI-based versus TKI (overall population)OS: median not reached versus 26.1 months, p = 0.0072PFS: median 10.2 versus 9.86 months, p = 0.681ORR: 48.7 versus 27.6%, p < 0.0001ICI-based versus TKI (trial eligible)OS: median not reached versus 14.7 months, p = 0.091PFS: median 13.1 versus 14.7 months, p = 0.287ORR: 51.3 versus 34.0%, p = 0.0124ICI-based versus TKI (trial ineligible)OS: median 33.7 versus 17.8 months, p = 0.02; aHR 0.47 (95% CI 0.29–0.75)PFS: median 7.36 versus 6.15 months, p = 0.07ORR: 46.1 versus 21.2%, p < 0.001GoodZarrabi et al. (2023) [67]Cohort 1: ICI–TKI (n = 547)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 959)• Nivolumab–ipilimumabICI–ICI versus ICI–TKIUnadjusted OS: median 24.3 versus 28.9 months, p = 0.9Unadjusted PFS: median 6.9 versus 10.6 months, p = 0.04IPTW-adjusted OS:• Full cohort: HR 1.16 (95% CI 0.97–1.39)• IMDC favourable risk: HR 3.28 (95% CI 1.04–10.3)• IMDC intermediate/poor risk: HR 1.07 (95% CI 0.89–1.29)IPTW-adjusted PFS:• Full cohort: HR 1.17 (95% CI 1.00–1.36)• IMDC favourable risk: HR 2.30 (95% CI 1.20–4.43)• IMDC intermediate risk: HR 1.11 (95% CI 0.94–1.32)GoodKido et al. (2021) [73]Cohort 1: ICI–ICI (n = 52)• Nivolumab–ipilimumabCohort 2: TKI (n = 226)• Sunitinib• Axitinib• Sorafenib• PazopanibICI–ICI versus TKIIPTW-adjusted OS: HR 0.51 (95% CI 0.27–0.95)• IMDC intermediate risk: HR 0.87 (95% CI 0.35–2.17)• IMDC poor risk: HR 0.37 (95% CI 0.15–0.93)IPTW-adjusted PFS: HR 0.60 (95% CI 0.37–0.97)• IMDC intermediate risk: HR 0.88 (95% CI 0.35–2.20)• IMDC poor risk: HR 0.38 (95% CI 0.15–0.94)IPTW-adjusted ORR: 38.5 versus 22.9%, p = 0.018• IMDC intermediate risk: 52 versus 25%, p = 0.010• IMDC poor risk: 24 versus 19%, p = 0.567GoodKato et al. (2022) [77]Cohort 1: ICI–ICI (n = 33)• Nivolumab–ipilimumabCohort 2: TKI monotherapy (n = 39)• Sunitinib• PazopanibICI–ICI versus TKIIPTW-adjusted PFS: median 15.0 versus 13.8 months, p = 0.16IPTW-adjusted ORR: 46.2 versus 21.7%, p < 0.01GoodUeda et al. (2022) [78]Cohort 1: ICI-based (n = 58)• Nivolumab–ipilimumab• Avelumab–axitinib• Pembrolizumab–axitinibCohort 2: TT (n = 144)• Included predominantly TKIs (e.g. sunitinib, axitinib), as well as other targeted agents such as mTOR inhibitors (e.g. temsirolimus)TT versus ICI-basedPFS: median 6.4 versus 17.6 months, p = 0.004; aHR 1.96 (95% CI 1.29–2.99)OS: HR 1.42 (95% CI 0.80–2.53)• No significant differences in IMDC favourable/intermediate risk• Significantly worse for TT in IMDC poor risk, p = 0.03GoodUeda et al. (2026) [80]Cohort 1: ICI–TKI (n = 17)• Nivolumab–cabozantinib• Pembrolizumab–levantinib• Avelumab–axitinib• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 18)• Nivolumab–ipilimumabCohort 3: TT (n = 53)• Included predominantly TKIs (e.g. sunitinib, pazopanib), as well as other targeted agents such as mTOR inhibitors (e.g. temsirolimus)ICI–TKI versus ICI–ICI versus TKI monotherapyOS: no significant difference was observed between the three groups (p = 0.92)PFS: median 7.5 versus 5.4 versus 2.5 months, p = 0.003• ICI–TKI versus TKI: p = 0.02• ICI–ICI versus TKI: p = 0.02• ICI–TKI versus ICI–ICI: p = 1ORR: 26.7 versus 21.4% versus 20.0%, p = 0.68GoodGeynisman et al. (2025) [83]Cohort 1: ICI–TKI (n = 130)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 225)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIIPTW-adjusted OS: median 29.3 versus 35.0 months, p = 0.37IPTW-adjusted PFS: median 15.4 versus 15.9 months, p = 0.20IPTW-adjusted ORR: 77.3 versus 71.5%, p = 0.24GoodGoebell et al. (2025) [86]Cohort 1: ICI–TKI (n = 447)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 257)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 166)• Pazopanib• Cabozantinib• Sunitinib• Tivozanib• Axitinib• SorafenibICI–TKI versus ICI–ICI versus TKIIPTW-adjusted OS: median 29.0 versus 21.9 versus 31.7 months• ICI–ICI versus ICI–TKI: HR 1.25 (95% CI 0.95–1.63)• TKI versus ICI–TKI: HR 1.04 (95% CI 0.77–1.41)• TKI versus ICI–ICI: HR 0.83 (95% CI 0.60–1.16)IPTW-adjusted PFS: median 12.3 versus 8.3 versus 8.5 months• ICI–ICI versus ICI–TKI: HR 1.25 (95% CI 1.00–1.58)• TKI versus ICI–TKI: HR 1.51 (95% CI 1.19–1.91)• TKI versus ICI–ICI: HR 1.20 (95% CI 0.91–1.58)GoodMassari et al. (2025) [30]Cohort 1: ICI–TKI (pembrolizumab–axitinib, n = 109)Cohort 2: ICI–ICI (nivolumab–ipilimumab, n = 100)Cohort 3: ICI–TKI (nivolumab–cabozantinib, n = 75)Cohort 4: ICI–TKI (pembrolizumab–lenvatinib, n = 39)Pembrolizumab–axitinib versus nivolumab–ipilimumab versus nivolumab–cabozantinib versus pembrolizumab–lenvatinibOS: median 31.2 versus 24.6 versus 27.8 versus 31.9 months, p = 0.651PFS: median 13.4 versus 6.8 versus 18.4 versus 24.5 months, p = 0.038• Nivolumab–ipilimumab versus nivolumab–cabozantinib: HR of mortality 1.57 (95% CI 1.02–2.43)• Nivolumab–ipilimumab versus pembrolizumab–lenvatinib: HR of mortality 1.87 (95% CI 1.12–3.12)ORR: 35 versus 32% versus 40 versus 59%, p = 0.005PoorHilser et al. (2025) [31]Cohort 1: ICI–TKI* (n = 47)Cohort 2: ICI–ICI* (n = 24)Cohort 3: TKI monotherapy (n = 50)*Did not specify which ICI agentsICI–TKI versus ICI–ICI versus TKI monotherapyOS: median not reached versus 27.9 versus 30.8 months. No significant differences between groupsPFS: median 16.9 versus 5.4 versus 8.8 months. Significantly longer for ICI–TKI versus ICI–ICI (p = 0.02)ORR: 40 versus 33% versus 32%PoorGupta et al. (2025) [36]Cohort 1: ICI–TKI (n = 354)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 769)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 2635)• Sunitinib• PazopanibICI–TKI versus TKI monotherapy (IMDC favourable-risk cohort)Treatment-free survival (TFS): 3.7 versus 3.1 monthsOS: mean 32.1 versus 31.3 monthsICI–ICI versus ICI–TKI versus TKI monotherapy (IMDC intermediate/poor-risk cohort)TFS: 5.3 versus 3.7 versus 2.1 monthsOS: mean 25.7 versus 27.9 versus 21.2 monthsPoorRoviello et al. (2025) [41]Cohort 1: ICI–TKI (n = 266)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: TKI monotherapy (n = 258)• SunitinibICI–TKI versus TKI monotherapyOS: median not reached versus not reached, p = 0.761PFS: median 30.7 versus 22.9 months, p = 0.007ORR: 53% versus 50%, p = 0.551Duration of response: median not reached versus 31.6 months, p < 0.001PoorIshikawa et al. (2025) [42]Cohort 1: ICI–TKI (n = 62)• Pembrolizumab–axitinib• Pembrolizumab–lenvatinib• Nivolumab–cabozantinib• Avelumab–axitinibCohort 2: TKI monotherapy (n = 63)• Sunitinib• Pazopanib• Sorafenib• Axitinib• CabozantinibICI–TKI versus TKI monotherapyOS: median not reached versus 34.2 months; aHR 0.36 (95% CI 0.18–0.72)PFS: median 18.6 versus 10.1 months; aHR 0.49 (95% CI 0.31–0.78)ORR: 64.5 versus 22.2%, p < 0.001PoorCiccarese et al. (2024) [43]Cohort 1: ICI–TKI (n = 101)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 125)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: median 34.4 versus 26.4 months, p = 0.73PFS: median 12.4 versus 12.3 months, p = 0.61ORR: 48 versus 48%Duration of response: median not reached in either group, p = 0.76PoorYoshimura et al. (2024) [44]Cohort 1: ICI–TKI (n = 24)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• avelumab –axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 20)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: median 23.9 vs. 26.8 months, p = 0.60PFS: median 6.4 versus 8.8 months, p = 0.78ORR: 41.6 versus 30.0%PoorLiu et al. (2024) [46]Cohort 1: ICI–TKI (nivolumab–cabozantinib, n = 111)Cohort 2: ICI–TKI (nivolumab–sunitinib, n = 105)Nivolumab–cabozantinib (ICI–TKI) versus nivolumab–sunitinib (ICI–TKI)OS: median 20.0 versus 15.0 months; p < 0.001Disease control rate: 54.05 versus 39.05%; p = 0.027PoorAlbigès et al. (2023) [50]Cohort 1: ICI monotherapy (n = 381)• nivolumabCohort 2: TKI monotherapy (n = 368)• mainly axitinib, cabozantinibCohort 3: Other (n = 123)• aldesleukin, bevacizumab–interferon alfa-2a, everolimus, temsirolimusICI monotherapy versus TKI monotherapyOS: median 19.6 versus 15.2 monthsTNT-D: median 8.5 versus 6.8 monthsPoorSantoni et al. (2024) [55]Cohort 1: ICI–TKI (n = 485)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 445)• Nivolumab–ipilimumabICI–TKI versus ICI–ICI (IMDC intermediate risk)OS: median 55.7 versus 40.2 months, p = 0.047PFS: median 30.7 versus 13.2 months, p < 0.001ORR: 50 versus 43%, p = 0.322ICI–TKI versus ICI–ICI (IMDC poor risk)OS: median 18.4 versus 15.5 months, p = 0.613PFS: median 11.6 versus 10.0 months, p = 0.614ORR: 46 versus 38%, p = 0.253PoorLai et al. (2023) [56]Cohort 1: ICI–TKI (n = 559)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 1555)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 4183)• Sunitinib• PazopanibICI–TKI versus ICI–ICI versus TKI monotherapyOS: median not reached versus 39.7 versus 42.2 months• ICI–TKI versus TKI: HR 0.81 (95% CI 0.68–0.96)• ICI–ICI versus TKI: HR 1.05 (95% CI 0.95–1.15)• ICI–TKI versus ICI–ICI: HR 0.76 (95% CI 0.63–0.91)TNT: median 22.1 versus 16.6 versus 16.6 monthsToT: median 8.3 versus 5.4 versus 2.5 monthsPoorAl-Mansour et al. (2024) [59]Cohort 1: ICI–combination based (n = 39)• Nivolumab–ipilimumab (ICI–ICI)• Pembrolizumab–lenvatinib (ICI–ICI)• Pembrolizumab–axitinib (ICI–TKI)• Nivolumab–cabozantinib (ICI–TKI)Cohort 2: ICI monotherapy (n = 10)• Nivolumab• PembrolizumabCohort 3: TKI monotherapy (n = 35)• Sunitinib• Pazopanib• CabozantinibICI combination versus ICI monotherapy versus TKIOS: median 42.0 versus 23.0 versus 33.3 months, p = 0.728PFS: median 9.0 versus 21.2 versus 9.4 months, p = 0.591ORR: 41.0 versus 20.0% versus 17.1%PoorTeishima et al. (2024) [60]Cohort 1: ICI–TKI (n = 10)• Avelumab–axitinib• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 26)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: no significant difference, p = 0.398PFS: median 9.7 versus 4.6 months, p = 0.0499ORR: 30.0 versus 34.6%, p = 0.9433PoorIinuma et al. (2023) [62]Cohort 1: avelumab–axitinib (ICI–TKI, n = 19)Cohort 2: pembrolizumab–axitinib (ICI–TKI, n = 19)Cohort 3: nivolumab–cabozantinib (ICI–TKI, n = 8)Cohort 4: pembrolizumab–lenvatinib (ICI–TKI, n = 5)Efficacy by individual regimenAvelumab–axitinib:ORR: 52.6%DCR: 79.0%Pembrolizumab–axitinib:ORR: 84.2%DCR: 94.7%Nivolumab–cabozantinib:ORR: 75.0%DCR: 100%Pembrolizumab–lenvatinib:ORR: 60.0%DCR: 80.0%Poorvan Laar et al. (2022) [63]Cohort 1: ICI–ICI (n = 28)• Nivolumab–ipilimumabCohort 2: TKI monotherapy (n = 104)• Sunitinib• PazopanibICI–ICI versus sunitinib versus pazopanibOS: median not reached versus 28.1 versus 39.3 monthsPFS: median not reached versus 6.9 versus 16.3 monthsPoorKimura et al. (2020) [64]Cohort 1: ICI monotherapy (n = 9)• NivolumabCohort 2: TKI monotherapy (n = 16)• AxitinibICI monotherapy versus TKIOS: median 12.3 versus 9.2 months, p = 0.63Treatment duration: median 5.4 versus 3.4 months, p = 0.09PoorCarril-Ajuria et al. (2023) [66]Cohort 1: ICI–TKI (n = 25)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 45)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: median not reached versus 9.8 months, p = 0.286• No significant difference across IMDC risk groupsPFS: median 6.0 versus 3.8 months, p = 0.842• No significant difference across IMDC risk groupsORR: 42 versus 23%, p = 0.38• Brain metastases subgroup: 33 versus 0%, p = 0.051PoorStühler et al. (2023) [68]Cohort 1: ICI–TKI (n = 21)• Pembrolizumab–axitinib• Avelumab–axitinib• Pembrolizumab–lenvatinib• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 55)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 125)• Cabozantinib• Sunitinib• Pazopanib• AxitinibICI–ICI versus ICI–TKI versus TKI monotherapyPFS (overall): median 6.1 versus 23.9 versus 10.3 months• ICI–TKI versus ICI–ICI: HR 0.37 (95% CI 0.17–0.80)• ICI–TKI versus TKI monotherapy: HR 0.48 (95% CI 0.24–0.95)• ICI–ICI versus TKI monotherapy: HR 1.22 (95% CI 0.82–1.82)• TKI monotherapy versus ICI-based: median 10.29 versus 10.26 months; HR 1.08, (95% CI 0.75–1.55)OS (overall): median not reached versus not reached versus 2.64 years• ICI–TKI versus ICI–ICI: HR 0.45, p = 0.183• ICI–TKI versus TKI monotherapy: HR 0.37 (95% CI 0.13–1.0)• TKI monotherapy versus ICI-based: median 29.9 months versus not reached; HR 0.65 (95% CI 0.39–1.09)PoorSantoni et al. (2022) [69]Cohort 1: ICI monotherapy (n = 220)• NivolumabCohort 2: TKI monotherapy (n = 123)• CabozantinibICI monotherapy versus TKIOS (overall): median 22.7 versus 16.9 months, p = 0.309• IMDC good risk: 47.0 versus 15.5 months, p = 0.285• IMDC intermediate risk: 26.0 versus 25.7 months, p = 0.519• IMDC poor risk: 14.4 versus 11.0 months, p = 0.357PoorHoeh et al. (2022) [70]Cohort 1: ICI–TKI (n = 36)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 68)• Nivolumab–ipilimumabICI–ICI versus ICI–TKIOS: median not reached in either group; HR 0.36 (95% CI 0.10–1.24)• 12-month survival in subgroup of elderly patients (≥ 70 years): 57.0 versus 91.0%, p = 0.042PFS: median 9.8 versus 12.3 months; HR 0.69 (95% CI 0.38–1.25)• Subgroup of elderly patients (≥ 70 years): 11.9 versus 12.3 months, p = 0.83PoorBando et al. (2022) [71]Cohort 1: ICI–ICI (n = 10)• Nivolumab–ipilimumabCohort 1: TT (n = 23)• Sunitinib• Sorafenib• Pazopanib• TemsirolimusICI–ICI versus TTOS: median 19.6 versus 10.6 months, p = 0.23PFS: median 3.5 versus 4.7 months, p = 0.61ORR: 30.0 versus 4.5%, p = 0.04TTF: median 2.4 versus 5.0 months, p = 0.16PoorBhanegaonkar et al. (2021) [72]Cohort 1: ICI (n = 33)• Nivolumab• Ipilimumab• Atezolizumab• PembrolizumabCohort 2: TKI (n = 645)• Pazopanib• Sunitinib• Axitinib• Cabozantinib• Sorafenib• Lenvatinib• ErlotinibCohort 3: mTOR inhibitors (n = 61)• Everolimus• TemsirolimusCohort 4: Other (n = 20)• Interleukin• InterferonICI versus TKI versus mTOR inhibitorsOS: median 9.6 versus 25.1 versus 10.8 monthsPoorStukalin et al. (2019) [74]Cohort 1: ICI monotherapy (n = 225)• NivolumabCohort 2: TKI monotherapy (n = 53)• CabozantinibICI monotherapy versus TKIOS: median 22.1 versus 23.7 months; aHR 1.3 (95% CI 0.73–2.31)ORR: 21 versus 20%TTF: median 6.90 versus 7.39 months (p = 0.20)PoorMcGrane et al. (2024) [75]Cohort 1: ICI–TKI (n = 197)• Pembrolizumab–axitinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 311)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 778)• Cabozantinib• Sunitinib• Pazopanib• Axitinib• TivozanibICI–TKI versus ICI–ICI versus TKI monotherapy (overall cohort)OS: median not reached versus 25.0 versus 23.8 months, p = 0.02• ICI–TKI versus TKI: HR 0.68 (95% CI 0.52–0.90)• ICI–ICI versus TKI: HR 0.96 (95% CI 0.80–1.16)PFS: median 11.3 versus 7.6 versus 7.8 months, p < 0.0001• ICI–TKI versus TKI: HR 0.61 (95% CI 0.50–0.75)• ICI–ICI versus TKI: HR 0.93 (95% CI 0.80–1.08)ICI–TKI versus TKI monotherapy (IMDC favourable group)OS: median not reached versus 41.1 months, p = 0.04; HR 0.42 (95% CI 0.18–0.99)PFS: median 25.0 versus 14.6 months, p = 0.02; HR 0.60 (95% CI 0.39–0.91)PoorSantoni et al. (2023) [76]Cohort 1: ICI–TKI (n = 416)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 313)• Nivolumab–ipilimumabICI–TKI versus ICI–ICI (IMDC intermediate/poor risk)OS: median 55.7 versus 29.7 months, p = 0.045PFS: median 15.9 versus 11.1 months, p = 0.011PoorBai et al. (2024) [79]Cohort 1: ICI–TKI (n = 9)Cohort 2: TKI (n = 12)ICI–TKI versus TKIPFS: median 13.0 versus 7.0 monthsORR: 33.3 versus 25%PoorMukae et al. (2026) [81]Cohort 1: ICI–TKI (n = 22)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 36)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: median 51.0 versus 49.0 months, p = 0.75PFS: median 25 versus 30 months, p = 0.88ORR: 59.1 versus 55.6%, p = 0.79Massari et al. (2025) [82]Cohort 1: pembrolizumab–axitinib (n = 91)Cohort 2: nivolumab–ipilimumab (n = 168)Cohort 3: nivolumab–cabozantinib (n = 53)Cohort 4: pembrolizumab–lenvatinib (n = 38)Favourable-risk features:OS: median was not reached with all three combinations (pembrolizumab–axitinib, nivolumab–cabozantinib, pembrolizumab–lenvatinib)Intermediate-risk features:OS: median not reached (pembrolizumab–axitinib) versus not reached (nivolumab–cabozantinib) versus 25.0 months (nivolumab–ipilimumab) versus 29.4 months (pembrolizumab–lenvatinib), p = 0.681Poor-risk features:OS: median 15.2 (pembrolizumab–axitinib) versus 20.9 (nivolumab–cabozantinib) versus 8.5 months (nivolumab–ipilimumab) versus 7.2 months (pembrolizumab–lenvatinib), p = 0.597PoorIkoma et al. (2025) [84]Cohort 1: ICI–TKI (n = 66)Cohort 2: ICI–ICI (n = 90)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS:• Older adult: not reached in either group, p = 0.65• Non-older adult: not reached versus 52.2 months, p = 0.65PFS:• Older adult: 15.5 versus 7.5 months, p = 0.38• Non-older adult: 41.0 versus 12.8 months, p = 0.03ORR:• Older adult: 55 versus 39%, p = 0.38PoorMeng et al. (2025) [85]Cohort 1: ICI-based (n = 29)• Single–agent ICIs (nivolumab or pembrolizumab) or combination ICI therapy with nivolumab plus ipilimumabCohort 2: TKI monotherapy (n = 40)• Sunitinib, pazopanib, cabozantinib, sorafenib, axitinib, and tivozanibICI-based versus TKIOS: median 19.3 versus 41.3 months; HR 1.96 (95% CI 1.03–3.71)PFS: median 4.9 versus 7.9 monthsORR: 12.0 versus 22.9%, p = 0.332PooraHR adjusted hazard ratio, CI confidence interval, DCR disease control rate, HR hazard ratio, ICI immune checkpoint inhibitor, IMDC International Metastatic Renal Cell Carcinoma Database Consortium, IPTW inverse probability of treatment weighting, OR odds ratio, ORR objective response rate, OS overall survival, PD progressive disease, PFS progression-free survival, PS propensity score, TFS treatment-free survival, TKI tyrosine kinase inhibitor, TNT-D time to next treatment or death, ToT time on treatment, TTF time to treatment failure, TTNT time to next treatmentTable 3Safety outcomes reported by included studies (n = 24)Study (year)Study cohorts, sample size (n)Safety outcome(s)Quality of evidenceIshihara et al. (2025) [34]Cohort 1: ICI–TKI (n = 85)• Pembrolizumab–lenvatinib• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 109)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIAny-grade AEs: 97.4 versus 78.8%, p < 0.0001Grade 3 or above AEs: 55.2 versus 44.1%, p = 0.141Treatment interruption: 70.5 versus 51.1%, p = 0.007Treatment discontinuation: 31.8 versus 20.0%, p = 0.076Glucocorticoid requirement: 43.8 versus 27.5%, p = 0.023GoodYanagisawa et al. (2024) [39]Cohort 1: ICI–TKI (n = 114)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 114)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIAny-grade TRAEs: 85 versus 71%, p = 0.016Grade ≥ 3 TRAEs: 36 versus 39%, p = 0.8GoodShah et al. (2024) [40]Cohort 1: ICI–TKI (n = 126)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 381)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIInpatient stays (first 90 days): adjusted IRR 0.42 (95% CI 0.27–0.66), p < 0.001Inpatient stays (full follow-up): adjusted IRR 0.57 (95% CI 0.41–0.80), p = 0.001Time to first inpatient stay: median 9.0 versus 5.6 months, p = 0.02Inpatient stays with ICU visit (first 90 days): adjusted IRR 0.68 (95% CI 0.36–1.32), p = 0.27Inpatient stays with ICU visit (full follow-up): adjusted IRR 0.70 (95% CI 0.43–1.14), p = 0.15GoodIshihara et al. (2024) [49]Cohort 1: ICI–TKI (n = 72)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 103)• Nivolumab–ipilimumabICI–TKI versus ICI–ICI (after IPTW)Any AE: 100 versus 80.5%, p < 0.0001Grade ≥ 3 AE: 53.9 versus 46.2%, p = 0.336Treatment interruption: 70.3 versus 49.2%, p = 0.005Corticosteroid use: 20.3 versus 43.3%, p = 0.001GoodGeynisman et al. (2025) [83]Cohort 1: ICI–TKI (n = 130)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 225)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOverall number of AEs: 25.3 versus 29.1%, significance not testedGrade 1–2 irAE: 21.9 versus 20.3%Grade 3 irAE: 2.1 versus 7.3%Steroid use: 5.9 versus 16.8%GoodDe Vries-Brilland et al. (2025) [29]Cohort 1: ICI-based (n = 70)• ICI–TKI• ICI–ICI• ICI monotherapyCohort 2: TKI monotherapy• Predominantly sunitinibICI-based regimens (combined) versus TKI monotherapyGrade 3–5 TRAEs: 30.8 versus 27.4%, p = 0.51PoorHilser et al. (2025) [31]Cohort 1: ICI–TKI* (n = 47)Cohort 2: ICI–ICI* (n = 24)Cohort 3: TKI monotherapy (n = 50)*Did not specify which ICI agentsICI–TKI versus ICI–ICI versus TKI monotherapyAny AEs: 87 versus 75% versus 52%Grade 3–5 TRAEs: 55 versus 17% versus 24%PoorHaack et al. (2025) [33]Cohort 1: ICI–TKI (n = 43)• Pembrolizumab–axitinib• Pembrolizumab–lenvatinib• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 16)• Nivolumab–ipilimumabICI–TKI versus ICI–ICITRAEs (any): similar between groups, p = 0.5TRAEs (grade 3 or above): similar between groups, p = 0.9Grade 4 TRAEs: 19 versus 0%Treatment-related discontinuation: 14 versus 6%PoorIshikawa et al. (2025) [42]Cohort 1: ICI–TKI (n = 62)• Pembrolizumab–axitinib• Pembrolizumab–lenvatinib• Nivolumab–cabozantinib• Avelumab–axitinibCohort 2: TKI monotherapy (n = 63)• Sunitinib• Pazopanib• Sorafenib• Axitinib• CabozantinibICI–TKI versus TKI monotherapyGrade ≥ 3 AEs: 66.1 versus 39.7%, p = 0.004Treatment discontinuation due to AEs: 14.5 versus 28.6%, p = 0.081PoorYoshimura et al. (2024) [44]Cohort 1: ICI–TKI (n = 24)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab –axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 20)• Nivolumab–ipilimumabICI–TKI versus ICI–ICITreatment discontinuation due to AEs: 25.0 versus 15.0%, p = 0.413PoorLiu et al. (2024) [46]Cohort 1: ICI–TKI (nivolumab–cabozantinib, n = 111)Cohort 2: ICI–TKI (nivolumab–sunitinib, n = 105)Nivolumab–cabozantinib (ICI–TKI) versus nivolumab–sunitinib (ICI–TKI)Total incidence of AEs: 80.18 versus 96.19%, p < 0.001Poor Huang et al. 2024 [52] Cohort 1: ICI–TKI (n = 44)• Anti-PD-1 agent + axitinib/sunitinib/pazopanib/ cabozantinib/lenvatinib/apatinib/sorafenibCohort 2: TKI monotherapy (n = 21)• AxitinibICI–TKI versus TKI monotherapyGrade ≥ 3 TRAEs: 56.8 versus 52.4%, no significance reportedPoorBlas et al. (2024) [54]Cohort 1: ICI–TKI (n = 627)• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 2294)• Nivolumab–ipilimumabICI–TKI versus ICI–ICISAEs: 68.6 versus 94.4%; ROR 0.12 (95% CI 0.10–0.16)Incidence rates of cardiac, endocrine, renal/urinary disorders were lower in the ICI–TKI groupPoorTeishima et al. (2024) [60]Cohort 1: ICI–TKI (n = 10)• Avelumab–axitinib• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 26)• Nivolumab–ipilimumabICI–TKI versus ICI–ICINo significant difference for any AEs below grade 2 (p = 0.85) or above grade 3 (p = 0.38)PoorKimura et al. (2020) [64]Cohort 1: ICI monotherapy (n = 9)• NivolumabCohort 2: TKI monotherapy (n = 16)• AxitinibIncidence of TRAEs were similar, but no significance reportedPoorCarril-Ajuria et al. (2023) [66]Cohort 1: ICI–TKI (n = 25)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 45)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIGrade 3–4 AEs: 25 versus 20%, p = 0.636PoorHoeh et al. (2022) [70]Cohort 1: ICI–TKI (n = 36)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 68)• Nivolumab–ipilimumabICI–ICI versus ICI–TKIAny-grade AEs: 57.4 versus 83.1%, p = 0.058Grade 3 AEs: 20.5 versus 33.3%, p = 0.23Grade 4–5 AEs: 7.3 versus 0%, p = 0.54PoorBando et al. (2022) [71]Cohort 1: ICI–ICI (n = 10)• Nivolumab–ipilimumabCohort 1: TT (n = 23)• Sunitinib• Sorafenib• Pazopanib• TemsirolimusICI–ICI versus TTAEs: 60 versus 91.3%Grade 3–4 AEs: 30 versus 8.7%PoorBhanegaonkar et al. (2021) [72]Cohort 1: ICI (n = 33)• Nivolumab• Ipilimumab• Atezolizumab• PembrolizumabCohort 2: TKI (n = 645)• Pazopanib• Sunitinib• Axitinib• Cabozantinib• Sorafenib• Lenvatinib• ErlotinibCohort 3: mTOR inhibitors (n = 61)• Everolimus• TemsirolimusCohort 4: other (n = 20)• Interleukin• InterferonICI versus mTOR inhibitorsTreatment interruption: 10.1 versus 2.5 per 100 person-yearsICI versus TKIPermanent discontinuation: 62.0 versus 25.5 per 100 person-yearsPoorKido et al. (2021) [73]Cohort 1: ICI–ICI (n = 52)• Nivolumab–ipilimumabCohort 2: TKI (n = 226)• Sunitinib• Axitinib• Sorafenib• PazopanibICI–ICI versus TKIAny AEs: 71 versus 81%, p = 0.131Grade 3 or above AEs: 37 versus 50%, p = 0.091AE-related treatment discontinuation: 69 versus 17%, p < 0.001PoorKato et al. (2022) [77]Cohort 1: ICI–ICI (n = 33)• Nivolumab–ipilimumabCohort 2: TKI monotherapy (n = 39)• Sunitinib• PazopanibICI–ICI versus TKITreatment discontinuation due to AEs: 26.7 versus 20.5%Ueda et al. (2026) [80]Cohort 1: ICI–TKI (n = 17)• Nivolumab–cabozantinib• Pembrolizumab–levantinib• Avelumab–axitinib• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 18)• Nivolumab–ipilimumabCohort 3: TT (n = 53)• Included predominantly TKIs (e.g. sunitinib, pazopanib), as well as other targeted agents such as mTOR inhibitors (e.g. temsirolimus)ICI–TKI versus ICI–ICI versus TKI monotherapyAny-grade TRAEs: 70.6 versus 61.1% versus 88.7%Grade ≥ 3 TRAEs: 41.2 versus 33.3% versus 30.2%, p = 0.71PoorMukae et al. (2026) [81]Cohort 1: ICI–TKI (n = 22)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 36)• Nivolumab–ipilimumabICI–TKI versus ICI–ICITreatment discontinuation: 40.9 versus 30.6%, p = 0.42• The frequency of corticosteroid administration was not significantly different between the two groupsPoorIkoma et al. (2025) [84]Cohort 1: ICI–TKI (n = 66)Cohort 2: ICI–ICI (n = 90)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIAny-grade AEs: 86 versus 66%Grade 3 or above AEs: 29 versus 41%Discontinuation due to AEs: 42 versus 21%PoorAEs adverse events, CI confidence interval, ICI immune checkpoint inhibitor, ICU intensive care unit, IPTW inverse probability of treatment weighting, irAE immune-related adverse event, IRR incidence rate ratio, ROR reporting odds ratio, SAEs serious adverse events, TKI tyrosine kinase inhibitor, TRAEs treatment-related adverse events Table 2Efficacy outcomes reported by included studies (n = 57)Study (year)Study cohorts, sample size (n)Efficacy outcome(s)Quality of evidenceDe Vries-Brilland et al. (2025) [29]Cohort 1: ICI-based (n = 70)• ICI–TKI• ICI–ICI• ICI monotherapyCohort 2: TKI monotherapy• Predominantly sunitinibICI-based (combined) versus TKI monotherapyOS: median 36.9 versus 26.5 months; aHR 0.54 (95% CI 0.30–0.90)TTF: median 9.9 versus 5.9 months; aHR 0.62 (95% CI 0.42–0.91)ORR: 37 versus 23%; aHR 2.2 (95% CI 0.95–5.2)ICI–TKI versus TKI monotherapyOS: aHR 0.26 (95% CI 0.08–0.83)TTF: aHR 0.47 (95% CI 0.26–0.85)ORR: 38 versus 23%ICI–ICI versus TKI monotherapyOS: aHR 0.90 (95% CI 0.47–1.72)TTF: aHR 2.04 (95% CI 1.2–3.4)ORR: 39 versus 23%GoodOstrowski et al. (2025) [32]Cohort 1: ICI–TKI (n = 659)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinib• Avelumab–axitinibCohort 2: ICI–ICI (n = 779)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIPS-matched OS: median 25.0 versus 28.8 months; aHR 1.01 (95% CI 0.86–1.19); sensitivity adjusted Cox HR 0.92 (95% CI 0.77–1.10)PS-matched TTNT: median 13.1 versus 7.8 months; aHR 0.78 (95% CI 0.68–0.89); sensitivity adjusted Cox HR 0.75 (95% CI 0.65–0.86)GoodHaack et al. (2025) [33]Cohort 1: ICI–TKI (n = 43)• Pembrolizumab–axitinib• Pembrolizumab–lenvatinib• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 16)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: median 33.0 versus 22.0 months; HR 0.74 (95% CI 0.36–1.51)PFS: median 14.0 versus 6.0 months; HR 0.53 (95% CI 0.28–0.99); multivariable Cox regression confirmed ICI–TKI as an independent predictor of improved PFS: aHR 0.47 (95% CI 0.24–0.92)ORR: 74% versus 44%, p = 0.035ToT: median 14.0 versus 6.0 months; HR 0.62 (95% CI 0.34–1.14), p = 0.109GoodIshihara et al. (2025) [34]Cohort 1: ICI–TKI (n = 85)• Pembrolizumab–lenvatinib• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 109)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIIPTW-adjusted OS: median not reached versus 56.0 months; aHR 0.63 (95% CI 0.32–1.26)IPTW-adjusted PFS: median 20.6 versus 9.0 months; aHR 0.52 (95% CI 0.34–0.79)IPTW-adjusted ORR: 74% versus 44%, p = 0.035IPTW-adjusted TTF: median 17.3 versus 4.8 months, IPTW-adjusted log-rank p < 0.001GoodSantoni et al. (2025) [35]Cohort 1: ICI–TKI (nivolumab–cabozantinib, n = 153)Cohort 2: ICI–TKI (pembrolizumab–axitinib, n = 607)Nivolumab–cabozantinib versus pembrolizumab–axitinibOS: median not reached versus 55.7 months; aHR 0.87 (95% CI 0.58–1.31)PFS: 27.6 versus 16.2 months; aHR 0.53 (95% CI 0.37–0.74)ORR: 59 versus 49%, p = 0.51GoodKikuta et al. (2025) [37]Cohort 1: ICI–TKI (n = 64)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 108)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIIPTW-adjusted OS: HR 0.74 (95% CI 0.35–1.56)IPTW-adjusted PFS: HR 0.72 (95% CI 0.40–1.29)IPTW-adjusted ORR: OR 2.34 (95% CI 1.14–4.82)GoodHara et al. (2025) [38]Cohort 1: ICI–TKI (n = 74)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 97)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: aHR 0.70 (95% CI 0.35–1.39)PFS: aHR 0.69 (95% CI 0.42–1.13)ORR: 54.9 versus 37.2%, p < 0.001GoodYanagisawa et al. (2024) [39]Cohort 1: ICI–TKI (n = 114)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 114)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIPS-matched OS: median not reached versus 53 months, p = 0.24PS-matched PFS: median 25 versus 13 months, p = 0.047PS-matched ORR: 55 versus 51%, p = 0.8GoodShah et al. (2024) [40]Cohort 1: ICI–TKI (n = 126)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 381)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: not estimable in either group due to short follow-up; 12-month survival 77.3 versus 66.9%ToT: 12.4 versus 4.1 months, p < 0.001GoodBickley et al. (2024) [45]Cohort 1: ICI–TKI (n = 14)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Ipilimumab–nivolumab–cabozantinib• Atezolizumab–bevacizumabCohort 2: ICI–ICI (n = 23)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 28)• Sunitinib• PazopanibICI–TKI versus ICI–ICI versus TKI monotherapyOS: median 40.0 versus 12.9 versus 15.6 months• ICI–TKI versus TKI monotherapy: aHR 0.61 (95% CI 0.27–1.36)• ICI–TKI versus ICI–ICI: aHR 0.82 (95% CI 0.31–2.17)PFS: median 14.7 versus 5.1 versus 6.7 months• ICI–TKI versus TKI monotherapy: aHR 0.47 (95% CI 0.24–0.92)• ICI–TKI versus ICI–ICI: aHR 0.38 (95% CI 0.19–0.78)GoodMassari et al. (2024) [47]Cohort 1: ICI-based (n = 73)• Pembrolizumab–axitinib (ICI–TKI)• Nivolumab–cabozantinib (ICI–TKI)• Pembrolizumab–lenvatinib (ICI–TKI)• Nivolumab–ipilimumab (ICI–ICI)Cohort 2: TKI monotherapy (n = 127)• Sunitinib• Cabozantinib• PazopanibICI-based (combined) versus TKI monotherapyOS: median 28.8 versus 22.5 months, p = 0.081PFS: median 17.4 versus 6.4 months; p < 0.001• aHR 2.31 (95% CI 1.55–3.44), p < 0.001 (TKI versus ICI-based)• Median PFS was significantly longer for ICI-based versus TKI monotherapy in the IMDC good-risk (39.7 versus 5.8 months, p < 0.001) and intermediate-risk (12.2 versus 7.3 months, p = 0.047) subgroups, but the difference did not reach significance in the poor-risk subgroup (12.2 versus 6.2 months, p = 0.093).ORR: 41 versus 27%GoodGebrael et al. (2024) [48]Cohort 1: ICI–TKI (n = 68)• Nivolumab–cabozantinib• Pembrolizumab–cabozantinib• Avelumab–cabozantinib• Nivolumab–pembrolizumab–cabozantinibCohort 2: TKI monotherapy (n = 280)• CabozantinibICI–TKI versus TKI monotherapyPS-matched OS: median 21 versus 20 months; HR 1.15 (95% CI 0.73–1.79)PS-matched TTNT: median 18.6 versus 10.1 months; HR 0.74 (95% CI 0.49–1.12)GoodIshihara et al. (2024) [49]Cohort 1: ICI–TKI (n = 72)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 103)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIIPTW-adjusted OS: median 46.7 versus 49.0 months; p = 0.465IPTW-adjusted PFS: median 15.6 versus 8.3 months; p = 0.04IPTW-adjusted ORR: 51.2 versus 43.9%, p = 0.36IPTW-adjusted PD rate (best response): 3.3 versus 27.4%, p < 0.0001GoodSawaya et al. (2024) [51]Cohort 1: ICI-based (n = 411)• ICI–TKI (unspecified)• ICI–ICI (nivolumab–ipilimumab)Cohort 2: TKI monotherapy (n = 791)• Mainly sunitinibICI-based versus TKI monotherapy (all patients):IPTW-adjusted OS: median 72 versus 36 months; HR 0.58 (95% CI 0.47–0.71)IPTW-adjusted ORR: 40.8 versus 21.6% (p < 0.0001)ICI-based versus TKI monotherapy (Sarcomatoid subgroup):IPTW-adjusted OS: median 48 versus 18 months; HR 0.41 (95% CI 0.26–0.64)IPTW-adjusted ORR: 49.5 versus 13.8% (p < 0.0001)ICI-based versus TKI monotherapy (non-sarcomatoid subgroup):IPTW-adjusted OS: median 72 versus 48 months; HR 0.63 (95% CI 0.50–0.80)IPTW-adjusted ORR: 38.5 versus 23.5% (p < 0.0001)ICI–TKI versus ICI–ICIIPTW-adjusted OS: median 48 versus 60 months; HR 1.28 (95% CI 0.89–1.84)IPTW-adjusted ORR: 42.9 versus 39.2% (p = 0.0003)Good Huang et al. 2024 [52] Cohort 1: ICI–TKI (n = 44)• Anti-PD-1 agent + axitinib/sunitinib/pazopanib/ cabozantinib/lenvatinib/apatinib/sorafenibCohort 2: TKI monotherapy (n = 21)• AxitinibICI–TKI versus TKI monotherapyIPTW-adjusted OS: median 31.9 versus 16.2 months, p = 0.133IPTW-adjusted PFS: median 9.2 versus 5.4 months, p = 0.002IPTW-adjusted ORR: 50 versus 14.3%, p = 0.006GoodEsterberg et al. (2024) [53]Cohort 1: ICI-based (n = 197)• ICI–TKI (pembrolizumab–axitinib)• ICI–TKI (nivolumab–cabozantinib)• ICI–TKI (avelumab–axitinib)• ICI–ICI (nivolumab–ipilimumab)Cohort 2: TKI monotherapy (n = 250)• Sunitinib• Pazopanib• Sorafenib• Cabozantinib• AxitinibICI-based versus TKI monotherapyOS: median 72 versus 36 months; aHR 0.66 (95% CI 0.39–1.11)PFS: median 27.1 versus 15.5 months; aHR 0.50 (95% CI 0.36–0.72)TTNT: median 30.1 versus 16.7 months; aHR 0.54 (95% CI 0.39–0.73)ICI–TKI versus ICI–ICIPFS: median not reached versus 23.0 months; aHR 0.50 (95% CI 0.28–0.97)GoodPoprach et al. (2023) [57]Cohort 1: ICI-based (n = 214)• ICI monotherapy (nivolumab)• ICI–ICI (nivolumab–ipilimumab)• ICI–TKI (avelumab–axitinib)Cohort 2: TT only (n = 1767)• Included predominantly TKIs (e.g. sunitinib, axitinib), as well as other targeted agents such as mTOR inhibitors (e.g. everolimus)ICI-based versus TT (overall)OS: median not reached versus 24.5 months; aHR 0.23 (p < 0.001)ICI-based versus TT (1L)OS: median not reached versus 24.5 months; aHR 0.23 (p = 0.003)PFS: median 11.9 versus 9.3 months; aHR 0.63 (p = 0.046)ICI-based versus TT (2L)OS: median 37.8 versus 16.6 months; aHR 0.35 (p < 0.001)PFS: median 13.9 versus 6.3 months; aHR 0.49 (p < 0.001)ICI-based versus TT (3L)OS: median 30.4 versus 18.3 months; aHR 0.70 (p = 0.101)PFS: median 10.1 versus 6.0 months; aHR 0.63 (p = 0.005)GoodShah et al. (2023) [58]Cohort 1: ICI–TKI (n = 145)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 186)• Nivolumab–ipilimumabICI–TKI versus ICI–ICI (overall)PFS: median 13.9 versus 9.8 months, aHR 0.70 (95% CI 0.49–0.99)ORR: 71 versus 45.2%, p < 0.001TTNT: median 21.3 versus 11.9 months, aHR 0.60 (95% CI 0.42–0.87)ToT: median 10.5 versus 4.2 months, aHR 0.53 (95% CI 0.40–0.71)ICI–TKI versus ICI–ICI (subgroup of IMDC intermediate- or poor-risk patients)PFS: aHR 0.70 (95% CI 0.48–1.03)TTNT: aHR 1.00 (95% CI 0.98–1.02)ToT: aHR 0.52 (95% CI 0.38–0.72)GoodShah et al. (2023) [61]Cohort 1: ICI–TKI (n = 279)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 618)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 641)• Cabozantinib• Sunitinib• Pazopanib• AxitinibICI–TKI versus ICI–ICI versus TKIOS: median not reached versus 27.6 versus 26.9 months, p = 0.24TTNT: median 16.4 versus 8.3 versus 8.4 months, p < 0.001ToT: median 13.6 versus 5.8 versus 3.4 months, p < 0.001GoodIshihara et al. (2023) [65]Cohort 1: ICI-based (n = 152)• Nivolumab–ipilimumab (ICI–ICI)• Pembrolizumab–axitinib (ICI–TKI)• Avelumab–axitinib (ICI–TKI)• Nivolumab–cabozantinib (ICI–TKI)• Pembrolizumab–lenvatinib (ICI–TKI)Cohort 2: TKI monotherapy (n = 315)• Sorafenib• Sunitinib• PazopanibICI-based versus TKI (overall population)OS: median not reached versus 26.1 months, p = 0.0072PFS: median 10.2 versus 9.86 months, p = 0.681ORR: 48.7 versus 27.6%, p < 0.0001ICI-based versus TKI (trial eligible)OS: median not reached versus 14.7 months, p = 0.091PFS: median 13.1 versus 14.7 months, p = 0.287ORR: 51.3 versus 34.0%, p = 0.0124ICI-based versus TKI (trial ineligible)OS: median 33.7 versus 17.8 months, p = 0.02; aHR 0.47 (95% CI 0.29–0.75)PFS: median 7.36 versus 6.15 months, p = 0.07ORR: 46.1 versus 21.2%, p < 0.001GoodZarrabi et al. (2023) [67]Cohort 1: ICI–TKI (n = 547)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 959)• Nivolumab–ipilimumabICI–ICI versus ICI–TKIUnadjusted OS: median 24.3 versus 28.9 months, p = 0.9Unadjusted PFS: median 6.9 versus 10.6 months, p = 0.04IPTW-adjusted OS:• Full cohort: HR 1.16 (95% CI 0.97–1.39)• IMDC favourable risk: HR 3.28 (95% CI 1.04–10.3)• IMDC intermediate/poor risk: HR 1.07 (95% CI 0.89–1.29)IPTW-adjusted PFS:• Full cohort: HR 1.17 (95% CI 1.00–1.36)• IMDC favourable risk: HR 2.30 (95% CI 1.20–4.43)• IMDC intermediate risk: HR 1.11 (95% CI 0.94–1.32)GoodKido et al. (2021) [73]Cohort 1: ICI–ICI (n = 52)• Nivolumab–ipilimumabCohort 2: TKI (n = 226)• Sunitinib• Axitinib• Sorafenib• PazopanibICI–ICI versus TKIIPTW-adjusted OS: HR 0.51 (95% CI 0.27–0.95)• IMDC intermediate risk: HR 0.87 (95% CI 0.35–2.17)• IMDC poor risk: HR 0.37 (95% CI 0.15–0.93)IPTW-adjusted PFS: HR 0.60 (95% CI 0.37–0.97)• IMDC intermediate risk: HR 0.88 (95% CI 0.35–2.20)• IMDC poor risk: HR 0.38 (95% CI 0.15–0.94)IPTW-adjusted ORR: 38.5 versus 22.9%, p = 0.018• IMDC intermediate risk: 52 versus 25%, p = 0.010• IMDC poor risk: 24 versus 19%, p = 0.567GoodKato et al. (2022) [77]Cohort 1: ICI–ICI (n = 33)• Nivolumab–ipilimumabCohort 2: TKI monotherapy (n = 39)• Sunitinib• PazopanibICI–ICI versus TKIIPTW-adjusted PFS: median 15.0 versus 13.8 months, p = 0.16IPTW-adjusted ORR: 46.2 versus 21.7%, p < 0.01GoodUeda et al. (2022) [78]Cohort 1: ICI-based (n = 58)• Nivolumab–ipilimumab• Avelumab–axitinib• Pembrolizumab–axitinibCohort 2: TT (n = 144)• Included predominantly TKIs (e.g. sunitinib, axitinib), as well as other targeted agents such as mTOR inhibitors (e.g. temsirolimus)TT versus ICI-basedPFS: median 6.4 versus 17.6 months, p = 0.004; aHR 1.96 (95% CI 1.29–2.99)OS: HR 1.42 (95% CI 0.80–2.53)• No significant differences in IMDC favourable/intermediate risk• Significantly worse for TT in IMDC poor risk, p = 0.03GoodUeda et al. (2026) [80]Cohort 1: ICI–TKI (n = 17)• Nivolumab–cabozantinib• Pembrolizumab–levantinib• Avelumab–axitinib• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 18)• Nivolumab–ipilimumabCohort 3: TT (n = 53)• Included predominantly TKIs (e.g. sunitinib, pazopanib), as well as other targeted agents such as mTOR inhibitors (e.g. temsirolimus)ICI–TKI versus ICI–ICI versus TKI monotherapyOS: no significant difference was observed between the three groups (p = 0.92)PFS: median 7.5 versus 5.4 versus 2.5 months, p = 0.003• ICI–TKI versus TKI: p = 0.02• ICI–ICI versus TKI: p = 0.02• ICI–TKI versus ICI–ICI: p = 1ORR: 26.7 versus 21.4% versus 20.0%, p = 0.68GoodGeynisman et al. (2025) [83]Cohort 1: ICI–TKI (n = 130)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 225)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIIPTW-adjusted OS: median 29.3 versus 35.0 months, p = 0.37IPTW-adjusted PFS: median 15.4 versus 15.9 months, p = 0.20IPTW-adjusted ORR: 77.3 versus 71.5%, p = 0.24GoodGoebell et al. (2025) [86]Cohort 1: ICI–TKI (n = 447)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 257)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 166)• Pazopanib• Cabozantinib• Sunitinib• Tivozanib• Axitinib• SorafenibICI–TKI versus ICI–ICI versus TKIIPTW-adjusted OS: median 29.0 versus 21.9 versus 31.7 months• ICI–ICI versus ICI–TKI: HR 1.25 (95% CI 0.95–1.63)• TKI versus ICI–TKI: HR 1.04 (95% CI 0.77–1.41)• TKI versus ICI–ICI: HR 0.83 (95% CI 0.60–1.16)IPTW-adjusted PFS: median 12.3 versus 8.3 versus 8.5 months• ICI–ICI versus ICI–TKI: HR 1.25 (95% CI 1.00–1.58)• TKI versus ICI–TKI: HR 1.51 (95% CI 1.19–1.91)• TKI versus ICI–ICI: HR 1.20 (95% CI 0.91–1.58)GoodMassari et al. (2025) [30]Cohort 1: ICI–TKI (pembrolizumab–axitinib, n = 109)Cohort 2: ICI–ICI (nivolumab–ipilimumab, n = 100)Cohort 3: ICI–TKI (nivolumab–cabozantinib, n = 75)Cohort 4: ICI–TKI (pembrolizumab–lenvatinib, n = 39)Pembrolizumab–axitinib versus nivolumab–ipilimumab versus nivolumab–cabozantinib versus pembrolizumab–lenvatinibOS: median 31.2 versus 24.6 versus 27.8 versus 31.9 months, p = 0.651PFS: median 13.4 versus 6.8 versus 18.4 versus 24.5 months, p = 0.038• Nivolumab–ipilimumab versus nivolumab–cabozantinib: HR of mortality 1.57 (95% CI 1.02–2.43)• Nivolumab–ipilimumab versus pembrolizumab–lenvatinib: HR of mortality 1.87 (95% CI 1.12–3.12)ORR: 35 versus 32% versus 40 versus 59%, p = 0.005PoorHilser et al. (2025) [31]Cohort 1: ICI–TKI* (n = 47)Cohort 2: ICI–ICI* (n = 24)Cohort 3: TKI monotherapy (n = 50)*Did not specify which ICI agentsICI–TKI versus ICI–ICI versus TKI monotherapyOS: median not reached versus 27.9 versus 30.8 months. No significant differences between groupsPFS: median 16.9 versus 5.4 versus 8.8 months. Significantly longer for ICI–TKI versus ICI–ICI (p = 0.02)ORR: 40 versus 33% versus 32%PoorGupta et al. (2025) [36]Cohort 1: ICI–TKI (n = 354)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 769)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 2635)• Sunitinib• PazopanibICI–TKI versus TKI monotherapy (IMDC favourable-risk cohort)Treatment-free survival (TFS): 3.7 versus 3.1 monthsOS: mean 32.1 versus 31.3 monthsICI–ICI versus ICI–TKI versus TKI monotherapy (IMDC intermediate/poor-risk cohort)TFS: 5.3 versus 3.7 versus 2.1 monthsOS: mean 25.7 versus 27.9 versus 21.2 monthsPoorRoviello et al. (2025) [41]Cohort 1: ICI–TKI (n = 266)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: TKI monotherapy (n = 258)• SunitinibICI–TKI versus TKI monotherapyOS: median not reached versus not reached, p = 0.761PFS: median 30.7 versus 22.9 months, p = 0.007ORR: 53% versus 50%, p = 0.551Duration of response: median not reached versus 31.6 months, p < 0.001PoorIshikawa et al. (2025) [42]Cohort 1: ICI–TKI (n = 62)• Pembrolizumab–axitinib• Pembrolizumab–lenvatinib• Nivolumab–cabozantinib• Avelumab–axitinibCohort 2: TKI monotherapy (n = 63)• Sunitinib• Pazopanib• Sorafenib• Axitinib• CabozantinibICI–TKI versus TKI monotherapyOS: median not reached versus 34.2 months; aHR 0.36 (95% CI 0.18–0.72)PFS: median 18.6 versus 10.1 months; aHR 0.49 (95% CI 0.31–0.78)ORR: 64.5 versus 22.2%, p < 0.001PoorCiccarese et al. (2024) [43]Cohort 1: ICI–TKI (n = 101)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 125)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: median 34.4 versus 26.4 months, p = 0.73PFS: median 12.4 versus 12.3 months, p = 0.61ORR: 48 versus 48%Duration of response: median not reached in either group, p = 0.76PoorYoshimura et al. (2024) [44]Cohort 1: ICI–TKI (n = 24)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• avelumab –axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 20)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: median 23.9 vs. 26.8 months, p = 0.60PFS: median 6.4 versus 8.8 months, p = 0.78ORR: 41.6 versus 30.0%PoorLiu et al. (2024) [46]Cohort 1: ICI–TKI (nivolumab–cabozantinib, n = 111)Cohort 2: ICI–TKI (nivolumab–sunitinib, n = 105)Nivolumab–cabozantinib (ICI–TKI) versus nivolumab–sunitinib (ICI–TKI)OS: median 20.0 versus 15.0 months; p < 0.001Disease control rate: 54.05 versus 39.05%; p = 0.027PoorAlbigès et al. (2023) [50]Cohort 1: ICI monotherapy (n = 381)• nivolumabCohort 2: TKI monotherapy (n = 368)• mainly axitinib, cabozantinibCohort 3: Other (n = 123)• aldesleukin, bevacizumab–interferon alfa-2a, everolimus, temsirolimusICI monotherapy versus TKI monotherapyOS: median 19.6 versus 15.2 monthsTNT-D: median 8.5 versus 6.8 monthsPoorSantoni et al. (2024) [55]Cohort 1: ICI–TKI (n = 485)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 445)• Nivolumab–ipilimumabICI–TKI versus ICI–ICI (IMDC intermediate risk)OS: median 55.7 versus 40.2 months, p = 0.047PFS: median 30.7 versus 13.2 months, p < 0.001ORR: 50 versus 43%, p = 0.322ICI–TKI versus ICI–ICI (IMDC poor risk)OS: median 18.4 versus 15.5 months, p = 0.613PFS: median 11.6 versus 10.0 months, p = 0.614ORR: 46 versus 38%, p = 0.253PoorLai et al. (2023) [56]Cohort 1: ICI–TKI (n = 559)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 1555)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 4183)• Sunitinib• PazopanibICI–TKI versus ICI–ICI versus TKI monotherapyOS: median not reached versus 39.7 versus 42.2 months• ICI–TKI versus TKI: HR 0.81 (95% CI 0.68–0.96)• ICI–ICI versus TKI: HR 1.05 (95% CI 0.95–1.15)• ICI–TKI versus ICI–ICI: HR 0.76 (95% CI 0.63–0.91)TNT: median 22.1 versus 16.6 versus 16.6 monthsToT: median 8.3 versus 5.4 versus 2.5 monthsPoorAl-Mansour et al. (2024) [59]Cohort 1: ICI–combination based (n = 39)• Nivolumab–ipilimumab (ICI–ICI)• Pembrolizumab–lenvatinib (ICI–ICI)• Pembrolizumab–axitinib (ICI–TKI)• Nivolumab–cabozantinib (ICI–TKI)Cohort 2: ICI monotherapy (n = 10)• Nivolumab• PembrolizumabCohort 3: TKI monotherapy (n = 35)• Sunitinib• Pazopanib• CabozantinibICI combination versus ICI monotherapy versus TKIOS: median 42.0 versus 23.0 versus 33.3 months, p = 0.728PFS: median 9.0 versus 21.2 versus 9.4 months, p = 0.591ORR: 41.0 versus 20.0% versus 17.1%PoorTeishima et al. (2024) [60]Cohort 1: ICI–TKI (n = 10)• Avelumab–axitinib• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 26)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: no significant difference, p = 0.398PFS: median 9.7 versus 4.6 months, p = 0.0499ORR: 30.0 versus 34.6%, p = 0.9433PoorIinuma et al. (2023) [62]Cohort 1: avelumab–axitinib (ICI–TKI, n = 19)Cohort 2: pembrolizumab–axitinib (ICI–TKI, n = 19)Cohort 3: nivolumab–cabozantinib (ICI–TKI, n = 8)Cohort 4: pembrolizumab–lenvatinib (ICI–TKI, n = 5)Efficacy by individual regimenAvelumab–axitinib:ORR: 52.6%DCR: 79.0%Pembrolizumab–axitinib:ORR: 84.2%DCR: 94.7%Nivolumab–cabozantinib:ORR: 75.0%DCR: 100%Pembrolizumab–lenvatinib:ORR: 60.0%DCR: 80.0%Poorvan Laar et al. (2022) [63]Cohort 1: ICI–ICI (n = 28)• Nivolumab–ipilimumabCohort 2: TKI monotherapy (n = 104)• Sunitinib• PazopanibICI–ICI versus sunitinib versus pazopanibOS: median not reached versus 28.1 versus 39.3 monthsPFS: median not reached versus 6.9 versus 16.3 monthsPoorKimura et al. (2020) [64]Cohort 1: ICI monotherapy (n = 9)• NivolumabCohort 2: TKI monotherapy (n = 16)• AxitinibICI monotherapy versus TKIOS: median 12.3 versus 9.2 months, p = 0.63Treatment duration: median 5.4 versus 3.4 months, p = 0.09PoorCarril-Ajuria et al. (2023) [66]Cohort 1: ICI–TKI (n = 25)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 45)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: median not reached versus 9.8 months, p = 0.286• No significant difference across IMDC risk groupsPFS: median 6.0 versus 3.8 months, p = 0.842• No significant difference across IMDC risk groupsORR: 42 versus 23%, p = 0.38• Brain metastases subgroup: 33 versus 0%, p = 0.051PoorStühler et al. (2023) [68]Cohort 1: ICI–TKI (n = 21)• Pembrolizumab–axitinib• Avelumab–axitinib• Pembrolizumab–lenvatinib• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 55)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 125)• Cabozantinib• Sunitinib• Pazopanib• AxitinibICI–ICI versus ICI–TKI versus TKI monotherapyPFS (overall): median 6.1 versus 23.9 versus 10.3 months• ICI–TKI versus ICI–ICI: HR 0.37 (95% CI 0.17–0.80)• ICI–TKI versus TKI monotherapy: HR 0.48 (95% CI 0.24–0.95)• ICI–ICI versus TKI monotherapy: HR 1.22 (95% CI 0.82–1.82)• TKI monotherapy versus ICI-based: median 10.29 versus 10.26 months; HR 1.08, (95% CI 0.75–1.55)OS (overall): median not reached versus not reached versus 2.64 years• ICI–TKI versus ICI–ICI: HR 0.45, p = 0.183• ICI–TKI versus TKI monotherapy: HR 0.37 (95% CI 0.13–1.0)• TKI monotherapy versus ICI-based: median 29.9 months versus not reached; HR 0.65 (95% CI 0.39–1.09)PoorSantoni et al. (2022) [69]Cohort 1: ICI monotherapy (n = 220)• NivolumabCohort 2: TKI monotherapy (n = 123)• CabozantinibICI monotherapy versus TKIOS (overall): median 22.7 versus 16.9 months, p = 0.309• IMDC good risk: 47.0 versus 15.5 months, p = 0.285• IMDC intermediate risk: 26.0 versus 25.7 months, p = 0.519• IMDC poor risk: 14.4 versus 11.0 months, p = 0.357PoorHoeh et al. (2022) [70]Cohort 1: ICI–TKI (n = 36)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 68)• Nivolumab–ipilimumabICI–ICI versus ICI–TKIOS: median not reached in either group; HR 0.36 (95% CI 0.10–1.24)• 12-month survival in subgroup of elderly patients (≥ 70 years): 57.0 versus 91.0%, p = 0.042PFS: median 9.8 versus 12.3 months; HR 0.69 (95% CI 0.38–1.25)• Subgroup of elderly patients (≥ 70 years): 11.9 versus 12.3 months, p = 0.83PoorBando et al. (2022) [71]Cohort 1: ICI–ICI (n = 10)• Nivolumab–ipilimumabCohort 1: TT (n = 23)• Sunitinib• Sorafenib• Pazopanib• TemsirolimusICI–ICI versus TTOS: median 19.6 versus 10.6 months, p = 0.23PFS: median 3.5 versus 4.7 months, p = 0.61ORR: 30.0 versus 4.5%, p = 0.04TTF: median 2.4 versus 5.0 months, p = 0.16PoorBhanegaonkar et al. (2021) [72]Cohort 1: ICI (n = 33)• Nivolumab• Ipilimumab• Atezolizumab• PembrolizumabCohort 2: TKI (n = 645)• Pazopanib• Sunitinib• Axitinib• Cabozantinib• Sorafenib• Lenvatinib• ErlotinibCohort 3: mTOR inhibitors (n = 61)• Everolimus• TemsirolimusCohort 4: Other (n = 20)• Interleukin• InterferonICI versus TKI versus mTOR inhibitorsOS: median 9.6 versus 25.1 versus 10.8 monthsPoorStukalin et al. (2019) [74]Cohort 1: ICI monotherapy (n = 225)• NivolumabCohort 2: TKI monotherapy (n = 53)• CabozantinibICI monotherapy versus TKIOS: median 22.1 versus 23.7 months; aHR 1.3 (95% CI 0.73–2.31)ORR: 21 versus 20%TTF: median 6.90 versus 7.39 months (p = 0.20)PoorMcGrane et al. (2024) [75]Cohort 1: ICI–TKI (n = 197)• Pembrolizumab–axitinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 311)• Nivolumab–ipilimumabCohort 3: TKI monotherapy (n = 778)• Cabozantinib• Sunitinib• Pazopanib• Axitinib• TivozanibICI–TKI versus ICI–ICI versus TKI monotherapy (overall cohort)OS: median not reached versus 25.0 versus 23.8 months, p = 0.02• ICI–TKI versus TKI: HR 0.68 (95% CI 0.52–0.90)• ICI–ICI versus TKI: HR 0.96 (95% CI 0.80–1.16)PFS: median 11.3 versus 7.6 versus 7.8 months, p < 0.0001• ICI–TKI versus TKI: HR 0.61 (95% CI 0.50–0.75)• ICI–ICI versus TKI: HR 0.93 (95% CI 0.80–1.08)ICI–TKI versus TKI monotherapy (IMDC favourable group)OS: median not reached versus 41.1 months, p = 0.04; HR 0.42 (95% CI 0.18–0.99)PFS: median 25.0 versus 14.6 months, p = 0.02; HR 0.60 (95% CI 0.39–0.91)PoorSantoni et al. (2023) [76]Cohort 1: ICI–TKI (n = 416)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 313)• Nivolumab–ipilimumabICI–TKI versus ICI–ICI (IMDC intermediate/poor risk)OS: median 55.7 versus 29.7 months, p = 0.045PFS: median 15.9 versus 11.1 months, p = 0.011PoorBai et al. (2024) [79]Cohort 1: ICI–TKI (n = 9)Cohort 2: TKI (n = 12)ICI–TKI versus TKIPFS: median 13.0 versus 7.0 monthsORR: 33.3 versus 25%PoorMukae et al. (2026) [81]Cohort 1: ICI–TKI (n = 22)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 36)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS: median 51.0 versus 49.0 months, p = 0.75PFS: median 25 versus 30 months, p = 0.88ORR: 59.1 versus 55.6%, p = 0.79Massari et al. (2025) [82]Cohort 1: pembrolizumab–axitinib (n = 91)Cohort 2: nivolumab–ipilimumab (n = 168)Cohort 3: nivolumab–cabozantinib (n = 53)Cohort 4: pembrolizumab–lenvatinib (n = 38)Favourable-risk features:OS: median was not reached with all three combinations (pembrolizumab–axitinib, nivolumab–cabozantinib, pembrolizumab–lenvatinib)Intermediate-risk features:OS: median not reached (pembrolizumab–axitinib) versus not reached (nivolumab–cabozantinib) versus 25.0 months (nivolumab–ipilimumab) versus 29.4 months (pembrolizumab–lenvatinib), p = 0.681Poor-risk features:OS: median 15.2 (pembrolizumab–axitinib) versus 20.9 (nivolumab–cabozantinib) versus 8.5 months (nivolumab–ipilimumab) versus 7.2 months (pembrolizumab–lenvatinib), p = 0.597PoorIkoma et al. (2025) [84]Cohort 1: ICI–TKI (n = 66)Cohort 2: ICI–ICI (n = 90)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOS:• Older adult: not reached in either group, p = 0.65• Non-older adult: not reached versus 52.2 months, p = 0.65PFS:• Older adult: 15.5 versus 7.5 months, p = 0.38• Non-older adult: 41.0 versus 12.8 months, p = 0.03ORR:• Older adult: 55 versus 39%, p = 0.38PoorMeng et al. (2025) [85]Cohort 1: ICI-based (n = 29)• Single–agent ICIs (nivolumab or pembrolizumab) or combination ICI therapy with nivolumab plus ipilimumabCohort 2: TKI monotherapy (n = 40)• Sunitinib, pazopanib, cabozantinib, sorafenib, axitinib, and tivozanibICI-based versus TKIOS: median 19.3 versus 41.3 months; HR 1.96 (95% CI 1.03–3.71)PFS: median 4.9 versus 7.9 monthsORR: 12.0 versus 22.9%, p = 0.332PooraHR adjusted hazard ratio, CI confidence interval, DCR disease control rate, HR hazard ratio, ICI immune checkpoint inhibitor, IMDC International Metastatic Renal Cell Carcinoma Database Consortium, IPTW inverse probability of treatment weighting, OR odds ratio, ORR objective response rate, OS overall survival, PD progressive disease, PFS progression-free survival, PS propensity score, TFS treatment-free survival, TKI tyrosine kinase inhibitor, TNT-D time to next treatment or death, ToT time on treatment, TTF time to treatment failure, TTNT time to next treatment Efficacy outcomes reported by included studies (n = 57) Cohort 1: ICI-based (n = 70) • ICI–TKI • ICI–ICI • ICI monotherapy Cohort 2: TKI monotherapy • Predominantly sunitinib ICI-based (combined) versus TKI monotherapy OS: median 36.9 versus 26.5 months; aHR 0.54 (95% CI 0.30–0.90) TTF: median 9.9 versus 5.9 months; aHR 0.62 (95% CI 0.42–0.91) ORR: 37 versus 23%; aHR 2.2 (95% CI 0.95–5.2) ICI–TKI versus TKI monotherapy OS: aHR 0.26 (95% CI 0.08–0.83) TTF: aHR 0.47 (95% CI 0.26–0.85) ORR: 38 versus 23% ICI–ICI versus TKI monotherapy OS: aHR 0.90 (95% CI 0.47–1.72) TTF: aHR 2.04 (95% CI 1.2–3.4) ORR: 39 versus 23% Cohort 1: ICI–TKI (n = 659) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib • Avelumab–axitinib Cohort 2: ICI–ICI (n = 779) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI PS-matched OS: median 25.0 versus 28.8 months; aHR 1.01 (95% CI 0.86–1.19); sensitivity adjusted Cox HR 0.92 (95% CI 0.77–1.10) PS-matched TTNT: median 13.1 versus 7.8 months; aHR 0.78 (95% CI 0.68–0.89); sensitivity adjusted Cox HR 0.75 (95% CI 0.65–0.86) Cohort 1: ICI–TKI (n = 43) • Pembrolizumab–axitinib • Pembrolizumab–lenvatinib • Nivolumab–cabozantinib Cohort 2: ICI–ICI (n = 16) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI OS: median 33.0 versus 22.0 months; HR 0.74 (95% CI 0.36–1.51) PFS: median 14.0 versus 6.0 months; HR 0.53 (95% CI 0.28–0.99); multivariable Cox regression confirmed ICI–TKI as an independent predictor of improved PFS: aHR 0.47 (95% CI 0.24–0.92) ORR: 74% versus 44%, p = 0.035 ToT: median 14.0 versus 6.0 months; HR 0.62 (95% CI 0.34–1.14), p = 0.109 Cohort 1: ICI–TKI (n = 85) • Pembrolizumab–lenvatinib • Nivolumab–cabozantinib Cohort 2: ICI–ICI (n = 109) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI IPTW-adjusted OS: median not reached versus 56.0 months; aHR 0.63 (95% CI 0.32–1.26) IPTW-adjusted PFS: median 20.6 versus 9.0 months; aHR 0.52 (95% CI 0.34–0.79) IPTW-adjusted ORR: 74% versus 44%, p = 0.035 IPTW-adjusted TTF: median 17.3 versus 4.8 months, IPTW-adjusted log-rank p < 0.001 Cohort 1: ICI–TKI (nivolumab–cabozantinib, n = 153) Cohort 2: ICI–TKI (pembrolizumab–axitinib, n = 607) Nivolumab–cabozantinib versus pembrolizumab–axitinib OS: median not reached versus 55.7 months; aHR 0.87 (95% CI 0.58–1.31) PFS: 27.6 versus 16.2 months; aHR 0.53 (95% CI 0.37–0.74) ORR: 59 versus 49%, p = 0.51 Cohort 1: ICI–TKI (n = 64) • Pembrolizumab–axitinib • Avelumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 108) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI IPTW-adjusted OS: HR 0.74 (95% CI 0.35–1.56) IPTW-adjusted PFS: HR 0.72 (95% CI 0.40–1.29) IPTW-adjusted ORR: OR 2.34 (95% CI 1.14–4.82) Cohort 1: ICI–TKI (n = 74) • Pembrolizumab–axitinib • Avelumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 97) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI OS: aHR 0.70 (95% CI 0.35–1.39) PFS: aHR 0.69 (95% CI 0.42–1.13) ORR: 54.9 versus 37.2%, p < 0.001 Cohort 1: ICI–TKI (n = 114) • Pembrolizumab–axitinib • Avelumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 114) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI PS-matched OS: median not reached versus 53 months, p = 0.24 PS-matched PFS: median 25 versus 13 months, p = 0.047 PS-matched ORR: 55 versus 51%, p = 0.8 Cohort 1: ICI–TKI (n = 126) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 381) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI OS: not estimable in either group due to short follow-up; 12-month survival 77.3 versus 66.9% ToT: 12.4 versus 4.1 months, p < 0.001 Cohort 1: ICI–TKI (n = 14) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Ipilimumab–nivolumab–cabozantinib • Atezolizumab–bevacizumab Cohort 2: ICI–ICI (n = 23) • Nivolumab–ipilimumab Cohort 3: TKI monotherapy (n = 28) • Sunitinib • Pazopanib ICI–TKI versus ICI–ICI versus TKI monotherapy OS: median 40.0 versus 12.9 versus 15.6 months • ICI–TKI versus TKI monotherapy: aHR 0.61 (95% CI 0.27–1.36) • ICI–TKI versus ICI–ICI: aHR 0.82 (95% CI 0.31–2.17) PFS: median 14.7 versus 5.1 versus 6.7 months • ICI–TKI versus TKI monotherapy: aHR 0.47 (95% CI 0.24–0.92) • ICI–TKI versus ICI–ICI: aHR 0.38 (95% CI 0.19–0.78) Cohort 1: ICI-based (n = 73) • Pembrolizumab–axitinib (ICI–TKI) • Nivolumab–cabozantinib (ICI–TKI) • Pembrolizumab–lenvatinib (ICI–TKI) • Nivolumab–ipilimumab (ICI–ICI) Cohort 2: TKI monotherapy (n = 127) • Sunitinib • Cabozantinib • Pazopanib ICI-based (combined) versus TKI monotherapy OS: median 28.8 versus 22.5 months, p = 0.081 PFS: median 17.4 versus 6.4 months; p < 0.001 • aHR 2.31 (95% CI 1.55–3.44), p < 0.001 (TKI versus ICI-based) • Median PFS was significantly longer for ICI-based versus TKI monotherapy in the IMDC good-risk (39.7 versus 5.8 months, p < 0.001) and intermediate-risk (12.2 versus 7.3 months, p = 0.047) subgroups, but the difference did not reach significance in the poor-risk subgroup (12.2 versus 6.2 months, p = 0.093). ORR: 41 versus 27% Cohort 1: ICI–TKI (n = 68) • Nivolumab–cabozantinib • Pembrolizumab–cabozantinib • Avelumab–cabozantinib • Nivolumab–pembrolizumab–cabozantinib Cohort 2: TKI monotherapy (n = 280) • Cabozantinib ICI–TKI versus TKI monotherapy PS-matched OS: median 21 versus 20 months; HR 1.15 (95% CI 0.73–1.79) PS-matched TTNT: median 18.6 versus 10.1 months; HR 0.74 (95% CI 0.49–1.12) Cohort 1: ICI–TKI (n = 72) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Avelumab–axitinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 103) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI IPTW-adjusted OS: median 46.7 versus 49.0 months; p = 0.465 IPTW-adjusted PFS: median 15.6 versus 8.3 months; p = 0.04 IPTW-adjusted ORR: 51.2 versus 43.9%, p = 0.36 IPTW-adjusted PD rate (best response): 3.3 versus 27.4%, p < 0.0001 Cohort 1: ICI-based (n = 411) • ICI–TKI (unspecified) • ICI–ICI (nivolumab–ipilimumab) Cohort 2: TKI monotherapy (n = 791) • Mainly sunitinib ICI-based versus TKI monotherapy (all patients): IPTW-adjusted OS: median 72 versus 36 months; HR 0.58 (95% CI 0.47–0.71) IPTW-adjusted ORR: 40.8 versus 21.6% (p < 0.0001) ICI-based versus TKI monotherapy (Sarcomatoid subgroup): IPTW-adjusted OS: median 48 versus 18 months; HR 0.41 (95% CI 0.26–0.64) IPTW-adjusted ORR: 49.5 versus 13.8% (p < 0.0001) ICI-based versus TKI monotherapy (non-sarcomatoid subgroup): IPTW-adjusted OS: median 72 versus 48 months; HR 0.63 (95% CI 0.50–0.80) IPTW-adjusted ORR: 38.5 versus 23.5% (p < 0.0001) ICI–TKI versus ICI–ICI IPTW-adjusted OS: median 48 versus 60 months; HR 1.28 (95% CI 0.89–1.84) IPTW-adjusted ORR: 42.9 versus 39.2% (p = 0.0003) Cohort 1: ICI–TKI (n = 44) • Anti-PD-1 agent + axitinib/sunitinib/pazopanib/ cabozantinib/lenvatinib/apatinib/sorafenib Cohort 2: TKI monotherapy (n = 21) • Axitinib ICI–TKI versus TKI monotherapy IPTW-adjusted OS: median 31.9 versus 16.2 months, p = 0.133 IPTW-adjusted PFS: median 9.2 versus 5.4 months, p = 0.002 IPTW-adjusted ORR: 50 versus 14.3%, p = 0.006 Cohort 1: ICI-based (n = 197) • ICI–TKI (pembrolizumab–axitinib) • ICI–TKI (nivolumab–cabozantinib) • ICI–TKI (avelumab–axitinib) • ICI–ICI (nivolumab–ipilimumab) Cohort 2: TKI monotherapy (n = 250) • Sunitinib • Pazopanib • Sorafenib • Cabozantinib • Axitinib ICI-based versus TKI monotherapy OS: median 72 versus 36 months; aHR 0.66 (95% CI 0.39–1.11) PFS: median 27.1 versus 15.5 months; aHR 0.50 (95% CI 0.36–0.72) TTNT: median 30.1 versus 16.7 months; aHR 0.54 (95% CI 0.39–0.73) ICI–TKI versus ICI–ICI PFS: median not reached versus 23.0 months; aHR 0.50 (95% CI 0.28–0.97) Cohort 1: ICI-based (n = 214) • ICI monotherapy (nivolumab) • ICI–ICI (nivolumab–ipilimumab) • ICI–TKI (avelumab–axitinib) Cohort 2: TT only (n = 1767) • Included predominantly TKIs (e.g. sunitinib, axitinib), as well as other targeted agents such as mTOR inhibitors (e.g. everolimus) ICI-based versus TT (overall) OS: median not reached versus 24.5 months; aHR 0.23 (p < 0.001) ICI-based versus TT (1L) OS: median not reached versus 24.5 months; aHR 0.23 (p = 0.003) PFS: median 11.9 versus 9.3 months; aHR 0.63 (p = 0.046) ICI-based versus TT (2L) OS: median 37.8 versus 16.6 months; aHR 0.35 (p < 0.001) PFS: median 13.9 versus 6.3 months; aHR 0.49 (p < 0.001) ICI-based versus TT (3L) OS: median 30.4 versus 18.3 months; aHR 0.70 (p = 0.101) PFS: median 10.1 versus 6.0 months; aHR 0.63 (p = 0.005) Cohort 1: ICI–TKI (n = 145) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 186) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI (overall) PFS: median 13.9 versus 9.8 months, aHR 0.70 (95% CI 0.49–0.99) ORR: 71 versus 45.2%, p < 0.001 TTNT: median 21.3 versus 11.9 months, aHR 0.60 (95% CI 0.42–0.87) ToT: median 10.5 versus 4.2 months, aHR 0.53 (95% CI 0.40–0.71) ICI–TKI versus ICI–ICI (subgroup of IMDC intermediate- or poor-risk patients) PFS: aHR 0.70 (95% CI 0.48–1.03) TTNT: aHR 1.00 (95% CI 0.98–1.02) ToT: aHR 0.52 (95% CI 0.38–0.72) Cohort 1: ICI–TKI (n = 279) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 618) • Nivolumab–ipilimumab Cohort 3: TKI monotherapy (n = 641) • Cabozantinib • Sunitinib • Pazopanib • Axitinib ICI–TKI versus ICI–ICI versus TKI OS: median not reached versus 27.6 versus 26.9 months, p = 0.24 TTNT: median 16.4 versus 8.3 versus 8.4 months, p < 0.001 ToT: median 13.6 versus 5.8 versus 3.4 months, p < 0.001 Cohort 1: ICI-based (n = 152) • Nivolumab–ipilimumab (ICI–ICI) • Pembrolizumab–axitinib (ICI–TKI) • Avelumab–axitinib (ICI–TKI) • Nivolumab–cabozantinib (ICI–TKI) • Pembrolizumab–lenvatinib (ICI–TKI) Cohort 2: TKI monotherapy (n = 315) • Sorafenib • Sunitinib • Pazopanib ICI-based versus TKI (overall population) OS: median not reached versus 26.1 months, p = 0.0072 PFS: median 10.2 versus 9.86 months, p = 0.681 ORR: 48.7 versus 27.6%, p < 0.0001 ICI-based versus TKI (trial eligible) OS: median not reached versus 14.7 months, p = 0.091 PFS: median 13.1 versus 14.7 months, p = 0.287 ORR: 51.3 versus 34.0%, p = 0.0124 ICI-based versus TKI (trial ineligible) OS: median 33.7 versus 17.8 months, p = 0.02; aHR 0.47 (95% CI 0.29–0.75) PFS: median 7.36 versus 6.15 months, p = 0.07 ORR: 46.1 versus 21.2%, p < 0.001 Cohort 1: ICI–TKI (n = 547) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 959) • Nivolumab–ipilimumab ICI–ICI versus ICI–TKI Unadjusted OS: median 24.3 versus 28.9 months, p = 0.9 Unadjusted PFS: median 6.9 versus 10.6 months, p = 0.04 IPTW-adjusted OS: • Full cohort: HR 1.16 (95% CI 0.97–1.39) • IMDC favourable risk: HR 3.28 (95% CI 1.04–10.3) • IMDC intermediate/poor risk: HR 1.07 (95% CI 0.89–1.29) IPTW-adjusted PFS: • Full cohort: HR 1.17 (95% CI 1.00–1.36) • IMDC favourable risk: HR 2.30 (95% CI 1.20–4.43) • IMDC intermediate risk: HR 1.11 (95% CI 0.94–1.32) Cohort 1: ICI–ICI (n = 52) • Nivolumab–ipilimumab Cohort 2: TKI (n = 226) • Sunitinib • Axitinib • Sorafenib • Pazopanib ICI–ICI versus TKI IPTW-adjusted OS: HR 0.51 (95% CI 0.27–0.95) • IMDC intermediate risk: HR 0.87 (95% CI 0.35–2.17) • IMDC poor risk: HR 0.37 (95% CI 0.15–0.93) IPTW-adjusted PFS: HR 0.60 (95% CI 0.37–0.97) • IMDC intermediate risk: HR 0.88 (95% CI 0.35–2.20) • IMDC poor risk: HR 0.38 (95% CI 0.15–0.94) IPTW-adjusted ORR: 38.5 versus 22.9%, p = 0.018 • IMDC intermediate risk: 52 versus 25%, p = 0.010 • IMDC poor risk: 24 versus 19%, p = 0.567 Cohort 1: ICI–ICI (n = 33) • Nivolumab–ipilimumab Cohort 2: TKI monotherapy (n = 39) • Sunitinib • Pazopanib ICI–ICI versus TKI IPTW-adjusted PFS: median 15.0 versus 13.8 months, p = 0.16 IPTW-adjusted ORR: 46.2 versus 21.7%, p < 0.01 Cohort 1: ICI-based (n = 58) • Nivolumab–ipilimumab • Avelumab–axitinib • Pembrolizumab–axitinib Cohort 2: TT (n = 144) • Included predominantly TKIs (e.g. sunitinib, axitinib), as well as other targeted agents such as mTOR inhibitors (e.g. temsirolimus) TT versus ICI-based PFS: median 6.4 versus 17.6 months, p = 0.004; aHR 1.96 (95% CI 1.29–2.99) OS: HR 1.42 (95% CI 0.80–2.53) • No significant differences in IMDC favourable/intermediate risk • Significantly worse for TT in IMDC poor risk, p = 0.03 Cohort 1: ICI–TKI (n = 17) • Nivolumab–cabozantinib • Pembrolizumab–levantinib • Avelumab–axitinib • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 18) • Nivolumab–ipilimumab Cohort 3: TT (n = 53) • Included predominantly TKIs (e.g. sunitinib, pazopanib), as well as other targeted agents such as mTOR inhibitors (e.g. temsirolimus) ICI–TKI versus ICI–ICI versus TKI monotherapy OS: no significant difference was observed between the three groups (p = 0.92) PFS: median 7.5 versus 5.4 versus 2.5 months, p = 0.003 • ICI–TKI versus TKI: p = 0.02 • ICI–ICI versus TKI: p = 0.02 • ICI–TKI versus ICI–ICI: p = 1 ORR: 26.7 versus 21.4% versus 20.0%, p = 0.68 Cohort 1: ICI–TKI (n = 130) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 225) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI IPTW-adjusted OS: median 29.3 versus 35.0 months, p = 0.37 IPTW-adjusted PFS: median 15.4 versus 15.9 months, p = 0.20 IPTW-adjusted ORR: 77.3 versus 71.5%, p = 0.24 Cohort 1: ICI–TKI (n = 447) • Pembrolizumab–axitinib • Avelumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 257) • Nivolumab–ipilimumab Cohort 3: TKI monotherapy (n = 166) • Pazopanib • Cabozantinib • Sunitinib • Tivozanib • Axitinib • Sorafenib ICI–TKI versus ICI–ICI versus TKI IPTW-adjusted OS: median 29.0 versus 21.9 versus 31.7 months • ICI–ICI versus ICI–TKI: HR 1.25 (95% CI 0.95–1.63) • TKI versus ICI–TKI: HR 1.04 (95% CI 0.77–1.41) • TKI versus ICI–ICI: HR 0.83 (95% CI 0.60–1.16) IPTW-adjusted PFS: median 12.3 versus 8.3 versus 8.5 months • ICI–ICI versus ICI–TKI: HR 1.25 (95% CI 1.00–1.58) • TKI versus ICI–TKI: HR 1.51 (95% CI 1.19–1.91) • TKI versus ICI–ICI: HR 1.20 (95% CI 0.91–1.58) Cohort 1: ICI–TKI (pembrolizumab–axitinib, n = 109) Cohort 2: ICI–ICI (nivolumab–ipilimumab, n = 100) Cohort 3: ICI–TKI (nivolumab–cabozantinib, n = 75) Cohort 4: ICI–TKI (pembrolizumab–lenvatinib, n = 39) Pembrolizumab–axitinib versus nivolumab–ipilimumab versus nivolumab–cabozantinib versus pembrolizumab–lenvatinib OS: median 31.2 versus 24.6 versus 27.8 versus 31.9 months, p = 0.651 PFS: median 13.4 versus 6.8 versus 18.4 versus 24.5 months, p = 0.038 • Nivolumab–ipilimumab versus nivolumab–cabozantinib: HR of mortality 1.57 (95% CI 1.02–2.43) • Nivolumab–ipilimumab versus pembrolizumab–lenvatinib: HR of mortality 1.87 (95% CI 1.12–3.12) ORR: 35 versus 32% versus 40 versus 59%, p = 0.005 Cohort 1: ICI–TKI* (n = 47) Cohort 2: ICI–ICI* (n = 24) Cohort 3: TKI monotherapy (n = 50) *Did not specify which ICI agents ICI–TKI versus ICI–ICI versus TKI monotherapy OS: median not reached versus 27.9 versus 30.8 months. No significant differences between groups PFS: median 16.9 versus 5.4 versus 8.8 months. Significantly longer for ICI–TKI versus ICI–ICI (p = 0.02) ORR: 40 versus 33% versus 32% Cohort 1: ICI–TKI (n = 354) • Pembrolizumab–axitinib • Avelumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 769) • Nivolumab–ipilimumab Cohort 3: TKI monotherapy (n = 2635) • Sunitinib • Pazopanib ICI–TKI versus TKI monotherapy (IMDC favourable-risk cohort) Treatment-free survival (TFS): 3.7 versus 3.1 months OS: mean 32.1 versus 31.3 months ICI–ICI versus ICI–TKI versus TKI monotherapy (IMDC intermediate/poor-risk cohort) TFS: 5.3 versus 3.7 versus 2.1 months OS: mean 25.7 versus 27.9 versus 21.2 months Cohort 1: ICI–TKI (n = 266) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: TKI monotherapy (n = 258) • Sunitinib ICI–TKI versus TKI monotherapy OS: median not reached versus not reached, p = 0.761 PFS: median 30.7 versus 22.9 months, p = 0.007 ORR: 53% versus 50%, p = 0.551 Duration of response: median not reached versus 31.6 months, p < 0.001 Cohort 1: ICI–TKI (n = 62) • Pembrolizumab–axitinib • Pembrolizumab–lenvatinib • Nivolumab–cabozantinib • Avelumab–axitinib Cohort 2: TKI monotherapy (n = 63) • Sunitinib • Pazopanib • Sorafenib • Axitinib • Cabozantinib ICI–TKI versus TKI monotherapy OS: median not reached versus 34.2 months; aHR 0.36 (95% CI 0.18–0.72) PFS: median 18.6 versus 10.1 months; aHR 0.49 (95% CI 0.31–0.78) ORR: 64.5 versus 22.2%, p < 0.001 Cohort 1: ICI–TKI (n = 101) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 125) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI OS: median 34.4 versus 26.4 months, p = 0.73 PFS: median 12.4 versus 12.3 months, p = 0.61 ORR: 48 versus 48% Duration of response: median not reached in either group, p = 0.76 Cohort 1: ICI–TKI (n = 24) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • avelumab –axitinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 20) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI OS: median 23.9 vs. 26.8 months, p = 0.60 PFS: median 6.4 versus 8.8 months, p = 0.78 ORR: 41.6 versus 30.0% Cohort 1: ICI–TKI (nivolumab–cabozantinib, n = 111) Cohort 2: ICI–TKI (nivolumab–sunitinib, n = 105) Nivolumab–cabozantinib (ICI–TKI) versus nivolumab–sunitinib (ICI–TKI) OS: median 20.0 versus 15.0 months; p < 0.001 Disease control rate: 54.05 versus 39.05%; p = 0.027 Cohort 1: ICI monotherapy (n = 381) • nivolumab Cohort 2: TKI monotherapy (n = 368) • mainly axitinib, cabozantinib Cohort 3: Other (n = 123) • aldesleukin, bevacizumab–interferon alfa-2a, everolimus, temsirolimus ICI monotherapy versus TKI monotherapy OS: median 19.6 versus 15.2 months TNT-D: median 8.5 versus 6.8 months Cohort 1: ICI–TKI (n = 485) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 445) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI (IMDC intermediate risk) OS: median 55.7 versus 40.2 months, p = 0.047 PFS: median 30.7 versus 13.2 months, p < 0.001 ORR: 50 versus 43%, p = 0.322 ICI–TKI versus ICI–ICI (IMDC poor risk) OS: median 18.4 versus 15.5 months, p = 0.613 PFS: median 11.6 versus 10.0 months, p = 0.614 ORR: 46 versus 38%, p = 0.253 Cohort 1: ICI–TKI (n = 559) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 1555) • Nivolumab–ipilimumab Cohort 3: TKI monotherapy (n = 4183) • Sunitinib • Pazopanib ICI–TKI versus ICI–ICI versus TKI monotherapy OS: median not reached versus 39.7 versus 42.2 months • ICI–TKI versus TKI: HR 0.81 (95% CI 0.68–0.96) • ICI–ICI versus TKI: HR 1.05 (95% CI 0.95–1.15) • ICI–TKI versus ICI–ICI: HR 0.76 (95% CI 0.63–0.91) TNT: median 22.1 versus 16.6 versus 16.6 months ToT: median 8.3 versus 5.4 versus 2.5 months Cohort 1: ICI–combination based (n = 39) • Nivolumab–ipilimumab (ICI–ICI) • Pembrolizumab–lenvatinib (ICI–ICI) • Pembrolizumab–axitinib (ICI–TKI) • Nivolumab–cabozantinib (ICI–TKI) Cohort 2: ICI monotherapy (n = 10) • Nivolumab • Pembrolizumab Cohort 3: TKI monotherapy (n = 35) • Sunitinib • Pazopanib • Cabozantinib ICI combination versus ICI monotherapy versus TKI OS: median 42.0 versus 23.0 versus 33.3 months, p = 0.728 PFS: median 9.0 versus 21.2 versus 9.4 months, p = 0.591 ORR: 41.0 versus 20.0% versus 17.1% Cohort 1: ICI–TKI (n = 10) • Avelumab–axitinib • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 26) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI OS: no significant difference, p = 0.398 PFS: median 9.7 versus 4.6 months, p = 0.0499 ORR: 30.0 versus 34.6%, p = 0.9433 Cohort 1: avelumab–axitinib (ICI–TKI, n = 19) Cohort 2: pembrolizumab–axitinib (ICI–TKI, n = 19) Cohort 3: nivolumab–cabozantinib (ICI–TKI, n = 8) Cohort 4: pembrolizumab–lenvatinib (ICI–TKI, n = 5) Efficacy by individual regimen Avelumab–axitinib: ORR: 52.6% DCR: 79.0% Pembrolizumab–axitinib: ORR: 84.2% DCR: 94.7% Nivolumab–cabozantinib: ORR: 75.0% DCR: 100% Pembrolizumab–lenvatinib: ORR: 60.0% DCR: 80.0% Cohort 1: ICI–ICI (n = 28) • Nivolumab–ipilimumab Cohort 2: TKI monotherapy (n = 104) • Sunitinib • Pazopanib ICI–ICI versus sunitinib versus pazopanib OS: median not reached versus 28.1 versus 39.3 months PFS: median not reached versus 6.9 versus 16.3 months Cohort 1: ICI monotherapy (n = 9) • Nivolumab Cohort 2: TKI monotherapy (n = 16) • Axitinib ICI monotherapy versus TKI OS: median 12.3 versus 9.2 months, p = 0.63 Treatment duration: median 5.4 versus 3.4 months, p = 0.09 Cohort 1: ICI–TKI (n = 25) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 45) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI OS: median not reached versus 9.8 months, p = 0.286 • No significant difference across IMDC risk groups PFS: median 6.0 versus 3.8 months, p = 0.842 • No significant difference across IMDC risk groups ORR: 42 versus 23%, p = 0.38 • Brain metastases subgroup: 33 versus 0%, p = 0.051 Cohort 1: ICI–TKI (n = 21) • Pembrolizumab–axitinib • Avelumab–axitinib • Pembrolizumab–lenvatinib • Nivolumab–cabozantinib Cohort 2: ICI–ICI (n = 55) • Nivolumab–ipilimumab Cohort 3: TKI monotherapy (n = 125) • Cabozantinib • Sunitinib • Pazopanib • Axitinib ICI–ICI versus ICI–TKI versus TKI monotherapy PFS (overall): median 6.1 versus 23.9 versus 10.3 months • ICI–TKI versus ICI–ICI: HR 0.37 (95% CI 0.17–0.80) • ICI–TKI versus TKI monotherapy: HR 0.48 (95% CI 0.24–0.95) • ICI–ICI versus TKI monotherapy: HR 1.22 (95% CI 0.82–1.82) • TKI monotherapy versus ICI-based: median 10.29 versus 10.26 months; HR 1.08, (95% CI 0.75–1.55) OS (overall): median not reached versus not reached versus 2.64 years • ICI–TKI versus ICI–ICI: HR 0.45, p = 0.183 • ICI–TKI versus TKI monotherapy: HR 0.37 (95% CI 0.13–1.0) • TKI monotherapy versus ICI-based: median 29.9 months versus not reached; HR 0.65 (95% CI 0.39–1.09) Cohort 1: ICI monotherapy (n = 220) • Nivolumab Cohort 2: TKI monotherapy (n = 123) • Cabozantinib ICI monotherapy versus TKI OS (overall): median 22.7 versus 16.9 months, p = 0.309 • IMDC good risk: 47.0 versus 15.5 months, p = 0.285 • IMDC intermediate risk: 26.0 versus 25.7 months, p = 0.519 • IMDC poor risk: 14.4 versus 11.0 months, p = 0.357 Cohort 1: ICI–TKI (n = 36) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 68) • Nivolumab–ipilimumab ICI–ICI versus ICI–TKI OS: median not reached in either group; HR 0.36 (95% CI 0.10–1.24) • 12-month survival in subgroup of elderly patients (≥ 70 years): 57.0 versus 91.0%, p = 0.042 PFS: median 9.8 versus 12.3 months; HR 0.69 (95% CI 0.38–1.25) • Subgroup of elderly patients (≥ 70 years): 11.9 versus 12.3 months, p = 0.83 Cohort 1: ICI–ICI (n = 10) • Nivolumab–ipilimumab Cohort 1: TT (n = 23) • Sunitinib • Sorafenib • Pazopanib • Temsirolimus ICI–ICI versus TT OS: median 19.6 versus 10.6 months, p = 0.23 PFS: median 3.5 versus 4.7 months, p = 0.61 ORR: 30.0 versus 4.5%, p = 0.04 TTF: median 2.4 versus 5.0 months, p = 0.16 Cohort 1: ICI (n = 33) • Nivolumab • Ipilimumab • Atezolizumab • Pembrolizumab Cohort 2: TKI (n = 645) • Pazopanib • Sunitinib • Axitinib • Cabozantinib • Sorafenib • Lenvatinib • Erlotinib Cohort 3: mTOR inhibitors (n = 61) • Everolimus • Temsirolimus Cohort 4: Other (n = 20) • Interleukin • Interferon ICI versus TKI versus mTOR inhibitors OS: median 9.6 versus 25.1 versus 10.8 months Cohort 1: ICI monotherapy (n = 225) • Nivolumab Cohort 2: TKI monotherapy (n = 53) • Cabozantinib ICI monotherapy versus TKI OS: median 22.1 versus 23.7 months; aHR 1.3 (95% CI 0.73–2.31) ORR: 21 versus 20% TTF: median 6.90 versus 7.39 months (p = 0.20) Cohort 1: ICI–TKI (n = 197) • Pembrolizumab–axitinib • Avelumab–axitinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 311) • Nivolumab–ipilimumab Cohort 3: TKI monotherapy (n = 778) • Cabozantinib • Sunitinib • Pazopanib • Axitinib • Tivozanib ICI–TKI versus ICI–ICI versus TKI monotherapy (overall cohort) OS: median not reached versus 25.0 versus 23.8 months, p = 0.02 • ICI–TKI versus TKI: HR 0.68 (95% CI 0.52–0.90) • ICI–ICI versus TKI: HR 0.96 (95% CI 0.80–1.16) PFS: median 11.3 versus 7.6 versus 7.8 months, p < 0.0001 • ICI–TKI versus TKI: HR 0.61 (95% CI 0.50–0.75) • ICI–ICI versus TKI: HR 0.93 (95% CI 0.80–1.08) ICI–TKI versus TKI monotherapy (IMDC favourable group) OS: median not reached versus 41.1 months, p = 0.04; HR 0.42 (95% CI 0.18–0.99) PFS: median 25.0 versus 14.6 months, p = 0.02; HR 0.60 (95% CI 0.39–0.91) Cohort 1: ICI–TKI (n = 416) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Avelumab–axitinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 313) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI (IMDC intermediate/poor risk) OS: median 55.7 versus 29.7 months, p = 0.045 PFS: median 15.9 versus 11.1 months, p = 0.011 Cohort 1: ICI–TKI (n = 9) Cohort 2: TKI (n = 12) ICI–TKI versus TKI PFS: median 13.0 versus 7.0 months ORR: 33.3 versus 25% Cohort 1: ICI–TKI (n = 22) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Avelumab–axitinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 36) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI OS: median 51.0 versus 49.0 months, p = 0.75 PFS: median 25 versus 30 months, p = 0.88 ORR: 59.1 versus 55.6%, p = 0.79 Cohort 1: pembrolizumab–axitinib (n = 91) Cohort 2: nivolumab–ipilimumab (n = 168) Cohort 3: nivolumab–cabozantinib (n = 53) Cohort 4: pembrolizumab–lenvatinib (n = 38) Favourable-risk features: OS: median was not reached with all three combinations (pembrolizumab–axitinib, nivolumab–cabozantinib, pembrolizumab–lenvatinib) Intermediate-risk features: OS: median not reached (pembrolizumab–axitinib) versus not reached (nivolumab–cabozantinib) versus 25.0 months (nivolumab–ipilimumab) versus 29.4 months (pembrolizumab–lenvatinib), p = 0.681 Poor-risk features: OS: median 15.2 (pembrolizumab–axitinib) versus 20.9 (nivolumab–cabozantinib) versus 8.5 months (nivolumab–ipilimumab) versus 7.2 months (pembrolizumab–lenvatinib), p = 0.597 Cohort 1: ICI–TKI (n = 66) Cohort 2: ICI–ICI (n = 90) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI OS: • Older adult: not reached in either group, p = 0.65 • Non-older adult: not reached versus 52.2 months, p = 0.65 PFS: • Older adult: 15.5 versus 7.5 months, p = 0.38 • Non-older adult: 41.0 versus 12.8 months, p = 0.03 ORR: • Older adult: 55 versus 39%, p = 0.38 Cohort 1: ICI-based (n = 29) • Single–agent ICIs (nivolumab or pembrolizumab) or combination ICI therapy with nivolumab plus ipilimumab Cohort 2: TKI monotherapy (n = 40) • Sunitinib, pazopanib, cabozantinib, sorafenib, axitinib, and tivozanib ICI-based versus TKI OS: median 19.3 versus 41.3 months; HR 1.96 (95% CI 1.03–3.71) PFS: median 4.9 versus 7.9 months ORR: 12.0 versus 22.9%, p = 0.332 aHR adjusted hazard ratio, CI confidence interval, DCR disease control rate, HR hazard ratio, ICI immune checkpoint inhibitor, IMDC International Metastatic Renal Cell Carcinoma Database Consortium, IPTW inverse probability of treatment weighting, OR odds ratio, ORR objective response rate, OS overall survival, PD progressive disease, PFS progression-free survival, PS propensity score, TFS treatment-free survival, TKI tyrosine kinase inhibitor, TNT-D time to next treatment or death, ToT time on treatment, TTF time to treatment failure, TTNT time to next treatment Table 3Safety outcomes reported by included studies (n = 24)Study (year)Study cohorts, sample size (n)Safety outcome(s)Quality of evidenceIshihara et al. (2025) [34]Cohort 1: ICI–TKI (n = 85)• Pembrolizumab–lenvatinib• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 109)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIAny-grade AEs: 97.4 versus 78.8%, p < 0.0001Grade 3 or above AEs: 55.2 versus 44.1%, p = 0.141Treatment interruption: 70.5 versus 51.1%, p = 0.007Treatment discontinuation: 31.8 versus 20.0%, p = 0.076Glucocorticoid requirement: 43.8 versus 27.5%, p = 0.023GoodYanagisawa et al. (2024) [39]Cohort 1: ICI–TKI (n = 114)• Pembrolizumab–axitinib• Avelumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 114)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIAny-grade TRAEs: 85 versus 71%, p = 0.016Grade ≥ 3 TRAEs: 36 versus 39%, p = 0.8GoodShah et al. (2024) [40]Cohort 1: ICI–TKI (n = 126)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 381)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIInpatient stays (first 90 days): adjusted IRR 0.42 (95% CI 0.27–0.66), p < 0.001Inpatient stays (full follow-up): adjusted IRR 0.57 (95% CI 0.41–0.80), p = 0.001Time to first inpatient stay: median 9.0 versus 5.6 months, p = 0.02Inpatient stays with ICU visit (first 90 days): adjusted IRR 0.68 (95% CI 0.36–1.32), p = 0.27Inpatient stays with ICU visit (full follow-up): adjusted IRR 0.70 (95% CI 0.43–1.14), p = 0.15GoodIshihara et al. (2024) [49]Cohort 1: ICI–TKI (n = 72)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 103)• Nivolumab–ipilimumabICI–TKI versus ICI–ICI (after IPTW)Any AE: 100 versus 80.5%, p < 0.0001Grade ≥ 3 AE: 53.9 versus 46.2%, p = 0.336Treatment interruption: 70.3 versus 49.2%, p = 0.005Corticosteroid use: 20.3 versus 43.3%, p = 0.001GoodGeynisman et al. (2025) [83]Cohort 1: ICI–TKI (n = 130)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 225)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIOverall number of AEs: 25.3 versus 29.1%, significance not testedGrade 1–2 irAE: 21.9 versus 20.3%Grade 3 irAE: 2.1 versus 7.3%Steroid use: 5.9 versus 16.8%GoodDe Vries-Brilland et al. (2025) [29]Cohort 1: ICI-based (n = 70)• ICI–TKI• ICI–ICI• ICI monotherapyCohort 2: TKI monotherapy• Predominantly sunitinibICI-based regimens (combined) versus TKI monotherapyGrade 3–5 TRAEs: 30.8 versus 27.4%, p = 0.51PoorHilser et al. (2025) [31]Cohort 1: ICI–TKI* (n = 47)Cohort 2: ICI–ICI* (n = 24)Cohort 3: TKI monotherapy (n = 50)*Did not specify which ICI agentsICI–TKI versus ICI–ICI versus TKI monotherapyAny AEs: 87 versus 75% versus 52%Grade 3–5 TRAEs: 55 versus 17% versus 24%PoorHaack et al. (2025) [33]Cohort 1: ICI–TKI (n = 43)• Pembrolizumab–axitinib• Pembrolizumab–lenvatinib• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 16)• Nivolumab–ipilimumabICI–TKI versus ICI–ICITRAEs (any): similar between groups, p = 0.5TRAEs (grade 3 or above): similar between groups, p = 0.9Grade 4 TRAEs: 19 versus 0%Treatment-related discontinuation: 14 versus 6%PoorIshikawa et al. (2025) [42]Cohort 1: ICI–TKI (n = 62)• Pembrolizumab–axitinib• Pembrolizumab–lenvatinib• Nivolumab–cabozantinib• Avelumab–axitinibCohort 2: TKI monotherapy (n = 63)• Sunitinib• Pazopanib• Sorafenib• Axitinib• CabozantinibICI–TKI versus TKI monotherapyGrade ≥ 3 AEs: 66.1 versus 39.7%, p = 0.004Treatment discontinuation due to AEs: 14.5 versus 28.6%, p = 0.081PoorYoshimura et al. (2024) [44]Cohort 1: ICI–TKI (n = 24)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab –axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 20)• Nivolumab–ipilimumabICI–TKI versus ICI–ICITreatment discontinuation due to AEs: 25.0 versus 15.0%, p = 0.413PoorLiu et al. (2024) [46]Cohort 1: ICI–TKI (nivolumab–cabozantinib, n = 111)Cohort 2: ICI–TKI (nivolumab–sunitinib, n = 105)Nivolumab–cabozantinib (ICI–TKI) versus nivolumab–sunitinib (ICI–TKI)Total incidence of AEs: 80.18 versus 96.19%, p < 0.001Poor Huang et al. 2024 [52] Cohort 1: ICI–TKI (n = 44)• Anti-PD-1 agent + axitinib/sunitinib/pazopanib/ cabozantinib/lenvatinib/apatinib/sorafenibCohort 2: TKI monotherapy (n = 21)• AxitinibICI–TKI versus TKI monotherapyGrade ≥ 3 TRAEs: 56.8 versus 52.4%, no significance reportedPoorBlas et al. (2024) [54]Cohort 1: ICI–TKI (n = 627)• Nivolumab–cabozantinibCohort 2: ICI–ICI (n = 2294)• Nivolumab–ipilimumabICI–TKI versus ICI–ICISAEs: 68.6 versus 94.4%; ROR 0.12 (95% CI 0.10–0.16)Incidence rates of cardiac, endocrine, renal/urinary disorders were lower in the ICI–TKI groupPoorTeishima et al. (2024) [60]Cohort 1: ICI–TKI (n = 10)• Avelumab–axitinib• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 26)• Nivolumab–ipilimumabICI–TKI versus ICI–ICINo significant difference for any AEs below grade 2 (p = 0.85) or above grade 3 (p = 0.38)PoorKimura et al. (2020) [64]Cohort 1: ICI monotherapy (n = 9)• NivolumabCohort 2: TKI monotherapy (n = 16)• AxitinibIncidence of TRAEs were similar, but no significance reportedPoorCarril-Ajuria et al. (2023) [66]Cohort 1: ICI–TKI (n = 25)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 45)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIGrade 3–4 AEs: 25 versus 20%, p = 0.636PoorHoeh et al. (2022) [70]Cohort 1: ICI–TKI (n = 36)• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 68)• Nivolumab–ipilimumabICI–ICI versus ICI–TKIAny-grade AEs: 57.4 versus 83.1%, p = 0.058Grade 3 AEs: 20.5 versus 33.3%, p = 0.23Grade 4–5 AEs: 7.3 versus 0%, p = 0.54PoorBando et al. (2022) [71]Cohort 1: ICI–ICI (n = 10)• Nivolumab–ipilimumabCohort 1: TT (n = 23)• Sunitinib• Sorafenib• Pazopanib• TemsirolimusICI–ICI versus TTAEs: 60 versus 91.3%Grade 3–4 AEs: 30 versus 8.7%PoorBhanegaonkar et al. (2021) [72]Cohort 1: ICI (n = 33)• Nivolumab• Ipilimumab• Atezolizumab• PembrolizumabCohort 2: TKI (n = 645)• Pazopanib• Sunitinib• Axitinib• Cabozantinib• Sorafenib• Lenvatinib• ErlotinibCohort 3: mTOR inhibitors (n = 61)• Everolimus• TemsirolimusCohort 4: other (n = 20)• Interleukin• InterferonICI versus mTOR inhibitorsTreatment interruption: 10.1 versus 2.5 per 100 person-yearsICI versus TKIPermanent discontinuation: 62.0 versus 25.5 per 100 person-yearsPoorKido et al. (2021) [73]Cohort 1: ICI–ICI (n = 52)• Nivolumab–ipilimumabCohort 2: TKI (n = 226)• Sunitinib• Axitinib• Sorafenib• PazopanibICI–ICI versus TKIAny AEs: 71 versus 81%, p = 0.131Grade 3 or above AEs: 37 versus 50%, p = 0.091AE-related treatment discontinuation: 69 versus 17%, p < 0.001PoorKato et al. (2022) [77]Cohort 1: ICI–ICI (n = 33)• Nivolumab–ipilimumabCohort 2: TKI monotherapy (n = 39)• Sunitinib• PazopanibICI–ICI versus TKITreatment discontinuation due to AEs: 26.7 versus 20.5%Ueda et al. (2026) [80]Cohort 1: ICI–TKI (n = 17)• Nivolumab–cabozantinib• Pembrolizumab–levantinib• Avelumab–axitinib• Pembrolizumab–axitinibCohort 2: ICI–ICI (n = 18)• Nivolumab–ipilimumabCohort 3: TT (n = 53)• Included predominantly TKIs (e.g. sunitinib, pazopanib), as well as other targeted agents such as mTOR inhibitors (e.g. temsirolimus)ICI–TKI versus ICI–ICI versus TKI monotherapyAny-grade TRAEs: 70.6 versus 61.1% versus 88.7%Grade ≥ 3 TRAEs: 41.2 versus 33.3% versus 30.2%, p = 0.71PoorMukae et al. (2026) [81]Cohort 1: ICI–TKI (n = 22)• Pembrolizumab–axitinib• Nivolumab–cabozantinib• Avelumab–axitinib• Pembrolizumab–lenvatinibCohort 2: ICI–ICI (n = 36)• Nivolumab–ipilimumabICI–TKI versus ICI–ICITreatment discontinuation: 40.9 versus 30.6%, p = 0.42• The frequency of corticosteroid administration was not significantly different between the two groupsPoorIkoma et al. (2025) [84]Cohort 1: ICI–TKI (n = 66)Cohort 2: ICI–ICI (n = 90)• Nivolumab–ipilimumabICI–TKI versus ICI–ICIAny-grade AEs: 86 versus 66%Grade 3 or above AEs: 29 versus 41%Discontinuation due to AEs: 42 versus 21%PoorAEs adverse events, CI confidence interval, ICI immune checkpoint inhibitor, ICU intensive care unit, IPTW inverse probability of treatment weighting, irAE immune-related adverse event, IRR incidence rate ratio, ROR reporting odds ratio, SAEs serious adverse events, TKI tyrosine kinase inhibitor, TRAEs treatment-related adverse events Safety outcomes reported by included studies (n = 24) Cohort 1: ICI–TKI (n = 85) • Pembrolizumab–lenvatinib • Nivolumab–cabozantinib Cohort 2: ICI–ICI (n = 109) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI Any-grade AEs: 97.4 versus 78.8%, p < 0.0001 Grade 3 or above AEs: 55.2 versus 44.1%, p = 0.141 Treatment interruption: 70.5 versus 51.1%, p = 0.007 Treatment discontinuation: 31.8 versus 20.0%, p = 0.076 Glucocorticoid requirement: 43.8 versus 27.5%, p = 0.023 Cohort 1: ICI–TKI (n = 114) • Pembrolizumab–axitinib • Avelumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 114) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI Any-grade TRAEs: 85 versus 71%, p = 0.016 Grade ≥ 3 TRAEs: 36 versus 39%, p = 0.8 Cohort 1: ICI–TKI (n = 126) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 381) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI Inpatient stays (first 90 days): adjusted IRR 0.42 (95% CI 0.27–0.66), p < 0.001 Inpatient stays (full follow-up): adjusted IRR 0.57 (95% CI 0.41–0.80), p = 0.001 Time to first inpatient stay: median 9.0 versus 5.6 months, p = 0.02 Inpatient stays with ICU visit (first 90 days): adjusted IRR 0.68 (95% CI 0.36–1.32), p = 0.27 Inpatient stays with ICU visit (full follow-up): adjusted IRR 0.70 (95% CI 0.43–1.14), p = 0.15 Cohort 1: ICI–TKI (n = 72) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Avelumab–axitinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 103) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI (after IPTW) Any AE: 100 versus 80.5%, p < 0.0001 Grade ≥ 3 AE: 53.9 versus 46.2%, p = 0.336 Treatment interruption: 70.3 versus 49.2%, p = 0.005 Corticosteroid use: 20.3 versus 43.3%, p = 0.001 Cohort 1: ICI–TKI (n = 130) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 225) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI Overall number of AEs: 25.3 versus 29.1%, significance not tested Grade 1–2 irAE: 21.9 versus 20.3% Grade 3 irAE: 2.1 versus 7.3% Steroid use: 5.9 versus 16.8% Cohort 1: ICI-based (n = 70) • ICI–TKI • ICI–ICI • ICI monotherapy Cohort 2: TKI monotherapy • Predominantly sunitinib ICI-based regimens (combined) versus TKI monotherapy Grade 3–5 TRAEs: 30.8 versus 27.4%, p = 0.51 Cohort 1: ICI–TKI* (n = 47) Cohort 2: ICI–ICI* (n = 24) Cohort 3: TKI monotherapy (n = 50) *Did not specify which ICI agents ICI–TKI versus ICI–ICI versus TKI monotherapy Any AEs: 87 versus 75% versus 52% Grade 3–5 TRAEs: 55 versus 17% versus 24% Cohort 1: ICI–TKI (n = 43) • Pembrolizumab–axitinib • Pembrolizumab–lenvatinib • Nivolumab–cabozantinib Cohort 2: ICI–ICI (n = 16) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI TRAEs (any): similar between groups, p = 0.5 TRAEs (grade 3 or above): similar between groups, p = 0.9 Grade 4 TRAEs: 19 versus 0% Treatment-related discontinuation: 14 versus 6% Cohort 1: ICI–TKI (n = 62) • Pembrolizumab–axitinib • Pembrolizumab–lenvatinib • Nivolumab–cabozantinib • Avelumab–axitinib Cohort 2: TKI monotherapy (n = 63) • Sunitinib • Pazopanib • Sorafenib • Axitinib • Cabozantinib ICI–TKI versus TKI monotherapy Grade ≥ 3 AEs: 66.1 versus 39.7%, p = 0.004 Treatment discontinuation due to AEs: 14.5 versus 28.6%, p = 0.081 Cohort 1: ICI–TKI (n = 24) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Avelumab –axitinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 20) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI Treatment discontinuation due to AEs: 25.0 versus 15.0%, p = 0.413 Cohort 1: ICI–TKI (nivolumab–cabozantinib, n = 111) Cohort 2: ICI–TKI (nivolumab–sunitinib, n = 105) Nivolumab–cabozantinib (ICI–TKI) versus nivolumab–sunitinib (ICI–TKI) Total incidence of AEs: 80.18 versus 96.19%, p < 0.001 Cohort 1: ICI–TKI (n = 44) • Anti-PD-1 agent + axitinib/sunitinib/pazopanib/ cabozantinib/lenvatinib/apatinib/sorafenib Cohort 2: TKI monotherapy (n = 21) • Axitinib ICI–TKI versus TKI monotherapy Grade ≥ 3 TRAEs: 56.8 versus 52.4%, no significance reported Cohort 1: ICI–TKI (n = 627) • Nivolumab–cabozantinib Cohort 2: ICI–ICI (n = 2294) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI SAEs: 68.6 versus 94.4%; ROR 0.12 (95% CI 0.10–0.16) Incidence rates of cardiac, endocrine, renal/urinary disorders were lower in the ICI–TKI group Cohort 1: ICI–TKI (n = 10) • Avelumab–axitinib • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 26) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI No significant difference for any AEs below grade 2 (p = 0.85) or above grade 3 (p = 0.38) Cohort 1: ICI monotherapy (n = 9) • Nivolumab Cohort 2: TKI monotherapy (n = 16) • Axitinib Cohort 1: ICI–TKI (n = 25) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 45) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI Grade 3–4 AEs: 25 versus 20%, p = 0.636 Cohort 1: ICI–TKI (n = 36) • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 68) • Nivolumab–ipilimumab ICI–ICI versus ICI–TKI Any-grade AEs: 57.4 versus 83.1%, p = 0.058 Grade 3 AEs: 20.5 versus 33.3%, p = 0.23 Grade 4–5 AEs: 7.3 versus 0%, p = 0.54 Cohort 1: ICI–ICI (n = 10) • Nivolumab–ipilimumab Cohort 1: TT (n = 23) • Sunitinib • Sorafenib • Pazopanib • Temsirolimus ICI–ICI versus TT AEs: 60 versus 91.3% Grade 3–4 AEs: 30 versus 8.7% Cohort 1: ICI (n = 33) • Nivolumab • Ipilimumab • Atezolizumab • Pembrolizumab Cohort 2: TKI (n = 645) • Pazopanib • Sunitinib • Axitinib • Cabozantinib • Sorafenib • Lenvatinib • Erlotinib Cohort 3: mTOR inhibitors (n = 61) • Everolimus • Temsirolimus Cohort 4: other (n = 20) • Interleukin • Interferon ICI versus mTOR inhibitors Treatment interruption: 10.1 versus 2.5 per 100 person-years ICI versus TKI Permanent discontinuation: 62.0 versus 25.5 per 100 person-years Cohort 1: ICI–ICI (n = 52) • Nivolumab–ipilimumab Cohort 2: TKI (n = 226) • Sunitinib • Axitinib • Sorafenib • Pazopanib ICI–ICI versus TKI Any AEs: 71 versus 81%, p = 0.131 Grade 3 or above AEs: 37 versus 50%, p = 0.091 AE-related treatment discontinuation: 69 versus 17%, p < 0.001 Cohort 1: ICI–ICI (n = 33) • Nivolumab–ipilimumab Cohort 2: TKI monotherapy (n = 39) • Sunitinib • Pazopanib ICI–ICI versus TKI Treatment discontinuation due to AEs: 26.7 versus 20.5% Cohort 1: ICI–TKI (n = 17) • Nivolumab–cabozantinib • Pembrolizumab–levantinib • Avelumab–axitinib • Pembrolizumab–axitinib Cohort 2: ICI–ICI (n = 18) • Nivolumab–ipilimumab Cohort 3: TT (n = 53) • Included predominantly TKIs (e.g. sunitinib, pazopanib), as well as other targeted agents such as mTOR inhibitors (e.g. temsirolimus) ICI–TKI versus ICI–ICI versus TKI monotherapy Any-grade TRAEs: 70.6 versus 61.1% versus 88.7% Grade ≥ 3 TRAEs: 41.2 versus 33.3% versus 30.2%, p = 0.71 Cohort 1: ICI–TKI (n = 22) • Pembrolizumab–axitinib • Nivolumab–cabozantinib • Avelumab–axitinib • Pembrolizumab–lenvatinib Cohort 2: ICI–ICI (n = 36) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI Treatment discontinuation: 40.9 versus 30.6%, p = 0.42 • The frequency of corticosteroid administration was not significantly different between the two groups Cohort 1: ICI–TKI (n = 66) Cohort 2: ICI–ICI (n = 90) • Nivolumab–ipilimumab ICI–TKI versus ICI–ICI Any-grade AEs: 86 versus 66% Grade 3 or above AEs: 29 versus 41% Discontinuation due to AEs: 42 versus 21% AEs adverse events, CI confidence interval, ICI immune checkpoint inhibitor, ICU intensive care unit, IPTW inverse probability of treatment weighting, irAE immune-related adverse event, IRR incidence rate ratio, ROR reporting odds ratio, SAEs serious adverse events, TKI tyrosine kinase inhibitor, TRAEs treatment-related adverse events Quality Assessment The quality assessment for efficacy outcomes is summarised in Supplementary Table S3. Of the 57 studies reporting efficacy outcomes [29–53, 55–86], 27 were assessed as good quality. These studies demonstrated sufficient comparability between treatment cohorts, with potential confounders appropriately addressed through multivariable Cox regression analyses, propensity score matching, or inverse probability of treatment weighting (IPTW). The remaining 30 studies were rated as poor quality, primarily due to inadequate adjustment for confounding. For safety outcomes [29, 31, 33, 34, 39, 40, 42, 44, 46, 52, 54, 60, 64, 66, 70–73, 77, 80, 81, 83, 84], 5 were assessed as good quality, while the remaining 19 were rated as poor quality. Most safety analyses were descriptive and lacked adjustment for treatment differences or potential confounders, limiting the validity of comparative safety conclusions (Supplementary Table S4). ICI-Based Versus Non-ICI Based Regimens ICI–TKI Versus TKI Monotherapy The most common comparison was ICI–TKI versus TKI monotherapy (n = 15) [29, 31, 36, 41, 42, 45, 48, 52, 56, 61, 68, 75, 79, 80, 86], of which seven studies rated as good quality in terms of efficacy outcomes [29, 45, 48, 52, 61, 80, 86]. Among good-quality studies, ICI–TKI was associated with improved PFS compared with TKI monotherapy, while OS findings were less consistent across studies [45, 52, 80, 86]. Using stabilised IPTW, Goebell et al. reported that TKI monotherapy was associated with a significantly shorter median PFS compared with ICI–TKI (8.5 versus 12.3 months; hazard ratio [HR] 1.51, 95% confidence interval [CI] 1.19–1.91), whereas no statistically significant difference in OS was observed (median 31.7 versus 29.0 months; HR 1.04, 95% CI 0.77–1.41) [86]. Similarly, in the second-line setting, Huang et al. used propensity score matching and found a longer median PFS with ICI–TKI compared with TKI monotherapy (9.2 versus 5.4 months; p = 0.002) [52]. Findings from poor-quality studies generally showed similar directional trends, although interpretation is limited by lack of confounding adjustment. Regarding safety, outcomes were generally comparable between ICI–TKI and TKI monotherapy [29, 31, 52, 80]. One study reported a higher incidence of grade ≥ 3 AEs with ICI–TKI compared with TKI monotherapy (66.1% versus 39.7%, p = 0.004) [42]. However, all these studies were considered poor quality, as safety outcomes were reported descriptively without adjustment for treatment differences or potential confounders. ICI–ICI Versus TKI Monotherapy A total of 13 studies compared dual ICI with TKI monotherapy [29, 31, 36, 45, 56, 61, 63, 68, 73, 75, 77, 80, 86]. Overall, evidence from good-quality studies did not demonstrate a consistent survival advantage of dual ICI over TKI monotherapy. Among the seven good-quality studies [29, 45, 61, 73, 77, 80, 86], results were largely neutral, with only Kido et al. reporting significant improvements in OS (HR 0.51, 95% CI 0.27–0.95), PFS (HR 0.60, 95% CI 0.37–0.97) and ORR (38.5 versus 22.9%, p = 0.018) in favour of dual ICI using IPTW-adjusted analyses [73]. Within this study, OS and PFS benefits were limited to the IMDC poor-risk subgroup and were not observed in the intermediate-risk subgroup. Two additional good-quality studies reported higher ORR with dual ICI but without corresponding survival benefits [29, 77], while the remaining good-quality studies showed no significant differences in survival endpoints. Findings from poor-quality studies were consistent with this overall pattern, with all reporting no significant survival differences between the two regimens [31, 36, 56, 63, 68, 75]. Regarding safety, all five studies reporting comparative safety data were rated as poor quality [29, 31, 73, 77, 80]. No clinically meaningful differences were reported between dual ICI and TKI monotherapy [29, 31, 73, 77, 80]. In Kido et al., no significant differences were observed in any or grade ≥ 3 AEs; however, AE-related treatment discontinuation was higher with dual ICI (69% versus 17%) [73]. Importantly, although IPTW was applied to efficacy analyses, safety outcomes were unadjusted, limiting comparative interpretation. Overall, safety data were descriptive and lacked confounding adjustment. ICI Monotherapy Versus TKI Monotherapy Across five studies comparing ICI monotherapy with TKI monotherapy [50, 59, 64, 69, 74], all were considered poor quality, primarily due to lack of adequate confounding adjustment. Across these studies, no consistent differences in survival outcomes were observed. Santoni et al. and Stukalin et al. both conducted studies in the second-line setting and reported no statistically significant differences in survival outcomes between nivolumab and cabozantinib [69, 74]. In addition, Santoni et al. stratified patients by IMDC risk category (favourable, intermediate and poor) and found no significant differences in survival outcomes within any risk subgroup. However, given the lack of confounding control, these findings should be interpreted cautiously. Other Three good-quality studies compared ICI-based regimens with targeted therapy, which predominantly consisted of TKIs but also included other targeted agents such as mTOR inhibitors (e.g. temsirolimus) [57, 71, 78]. Across these studies, ICI-based regimens were associated with improved efficacy outcomes compared with targeted therapy. Additional studies compared ICI-based regimens with TKI monotherapy and reported similar directional findings favouring ICI-based regimens [47, 51, 53, 59, 65]; however, interpretation is limited by heterogeneity in study quality and confounding adjustment. Notably, Sawaya et al. reported improved median OS and ORR with ICI-based regimens compared with TKI monotherapy in patients with sarcomatoid features (48 versus 18 months; 49.5% versus 13.8%) as well as those without sarcomatoid differentiation (72 versus 48 months; 42.9% versus 39.2%) [51]. In the only study specifically evaluating bone-predominant metastatic RCC, Meng et al. reported worse outcomes with ICI-based regimens compared with TKI monotherapy [85]; however, these findings were limited by poor baseline comparability and small sample size, and should be interpreted cautiously. Different ICI-Based Regimens ICI–TKI Versus ICI–ICI A total of 34 studies compared outcomes between ICI in combination with TKI and dual ICI [30–34, 36–40, 43–45, 47, 49, 51, 53–56, 58, 60, 61, 66–68, 70, 75, 76, 80–84, 86]. Across good-quality studies, ICI–TKI was generally associated with improved efficacy outcomes compared with dual ICI therapy. Several high-quality studies employing propensity score matching or IPTW demonstrated improved outcomes across multiple efficacy endpoints, including OS, PFS, ORR, TTNT and time to treatment failure (TTF) [32, 34, 38, 49, 51, 67, 86]. Notably, Zarrabi et al. reported significantly worse OS (HR 3.28, 95% CI 1.04–10.3) and PFS (HR 2.30, 95% CI 1.20–4.43) with ICI–ICI compared with ICI–TKI in the IMDC favourable-risk subgroup, while no significant differences were observed in the intermediate- or poor-risk subgroups [67]. Findings from lower-quality studies were broadly consistent, although interpretation is limited by lack of confounding adjustment. Regarding safety, 15 studies reported comparative safety data [31, 33, 34, 39, 40, 44, 49, 54, 60, 66, 70, 80, 81, 83, 84]. Of these, five studies were rated as high quality owing to good baseline comparability between treatment groups [34, 39, 40, 49, 83]. Notably, these were the only studies across the entire systematic review to achieve good comparability ratings in safety outcomes. Three studies reported a higher incidence of any adverse events in the ICI–TKI cohort [34, 39, 49]. Shah et al. examined healthcare resource utilisation using multivariable-adjusted analyses and found no significant differences between ICI–TKI and ICI–ICI in inpatient stays involving intensive care unit admission (adjusted incidence rate ratio 0.70, 95% CI 0.43–1.14) [40]. The remaining studies were predominantly unadjusted and generally reported no significant safety differences between regimens; however, as with other treatment comparisons in this review, the reliance on unadjusted analyses and lack of confounding control limit the robustness of comparative safety conclusions. Different ICI–TKI Regimens Comparisons between different ICI–TKI regimens showed heterogeneous efficacy outcomes, with no single regimen demonstrating consistent superiority across studies [30, 35, 46, 62, 82]. Direct head-to-head comparisons between specific ICI–TKI regimens were limited. Santoni et al. compared nivolumab–cabozantinib with pembrolizumab–axitinib and reported no significant difference in OS (not reached versus 55.7 months, adjusted HR 0.87, 95% CI 0.58–1.31), while median PFS favoured nivolumab–cabozantinib (27.6 versus 16.2 months, adjusted HR 0.53, 95% CI 0.37–0.74) [35]. Similarly, Liu et al. reported improved OS and disease control with nivolumab–cabozantinib compared with nivolumab–sunitinib [46]. Overall, evidence comparing different ICI–TKI regimens was limited by small sample sizes, heterogeneity in study design and inconsistent adjustment for confounders, precluding definitive conclusions. Discussion To the best of our knowledge, this is the first systematic review to synthesise comparative real-world observational evidence on the efficacy and safety of ICI-based regimens in advanced RCC. Overall, the available real-world evidence suggests that ICI-based regimens are associated with improved outcomes compared with non-ICI-based regimens. In particular, ICI–TKI combinations are associated with improved survival outcomes compared with TKI monotherapy, findings that are consistent with results from pivotal trials such as KEYNOTE-581 [6]. ICI–TKI regimens also appeared to provide more consistent benefits than dual ICI therapy. Network meta-analyses of clinical trial data have similarly suggested superior efficacy of ICI–TKI regimens relative to other first-line options [17, 89], and the findings from our review are broadly consistent with these trial-based findings. Although recent trial evidence from COSMIC-313 suggests PFS benefits from adding a TKI to dual ICI therapy (nivolumab–ipilimumab plus cabozantinib versus nivolumab–ipilimumab) [90], this strategy was not evaluated in the included real-world studies. Moreover, while indirect trial evidence indicates potential heterogeneity in effectiveness across individual ICI–TKI regimens [19], the available real-world data were limited and insufficient for robust comparative assessment. Taken together, these real-world findings are consistent with trial-derived evidence suggesting improved survival outcomes with ICI–TKI combinations relative to alternative regimens. However, given the retrospective and observational nature of the included studies, these findings should be interpreted as associative rather than causal and as supportive of existing trial evidence rather than as a basis for independent treatment recommendations. Treatment selection between ICI–TKI and dual ICI therapy should continue to be individualised on the basis of trial evidence, patient risk profile, comorbidities and therapeutic goals. Further high-quality real-world studies evaluating contemporary ICI–TKI regimens are therefore needed to inform optimal regimen selection in routine clinical practice. In contrast to ICI–TKI combinations, dual ICI therapy with nivolumab–ipilimumab was not consistently associated with improved survival or treatment response outcomes compared with TKI monotherapy in real-world settings. Most studies reported no statistically significant differences [31, 36, 45, 56, 61, 63, 68, 75, 86], highlighting a potential divergence between trial efficacy and real-world effectiveness. These inconsistencies may reflect substantial population heterogeneity, as most studies enrolled patients with mixed histology and heterogeneous IMDC risk profiles, as well as residual confounding inherent to observational designs [91]. Differences in outcome definitions and assessment frequency between real-world studies and clinical trials, particularly for PFS and ORR, may further contribute to discrepancies in observed effectiveness and should be considered when interpreting results. Future real-world studies with more robust confounding adjustment, longer follow-up and clinically meaningful subgroup analyses—including granular stratification by histology, IMDC risk, disease burden and patient characteristics—are warranted to better delineate the comparative efficacy of dual ICI therapy in routine clinical practice. Until such data are available, these findings should be interpreted cautiously, and treatment decisions should continue to be individualised on the basis of trial evidence, patient risk profile and comorbidities. This review also highlights the limited availability of high-quality real-world evidence comparing the safety of ICI-based regimens. Only five studies in this systematic review were assessed as good quality, all evaluating ICI–TKI versus ICI–ICI comparisons [34, 39, 40, 49, 83]. Most safety analyses relied solely on descriptive statistics without adjustment for treatment differences, limiting the ability to draw robust comparative safety conclusions. AEs may also be under-captured in routine clinical datasets compared with clinical trials, further limiting interpretation of these findings. These findings highlight the need for future real-world studies employing rigorous analytical methods, such as propensity score matching or IPTW, to better account for confounding and reduce bias in safety assessments. For instance, IPTW can reduce confounding by balancing measured baseline characteristics between treatment groups, thereby approximating a randomised setting and improving the reliability of comparative estimates [92]. Strengths and Limitations A key strength of this systematic review is its comprehensive search strategy, which captured relevant real-world studies evaluating ICI-based regimens in advanced RCC. The included studies offer high external validity by capturing treatment outcomes in patient populations that are more representative of routine clinical practice than those typically enrolled in clinical trials. Minimal restrictions on patient characteristics, treatment line, and ICI-based regimens allowed inclusion of a wide spectrum of real-world clinical scenarios. Importantly, the included studies largely reflect therapies routinely used in clinical practice or approved by regulatory agencies, enhancing the clinical relevance and generalisability of the findings [93]. A broad range of efficacy and safety outcomes was evaluated, and inclusion of studies from diverse geographic regions—including international real-world datasets such as ARON-1—further strengthens the generalisability of the findings. Several limitations should be acknowledged. Substantial heterogeneity across studies in terms of patient characteristics, sample sizes, treatment regimens, follow-up duration and analytical approaches precluded quantitative synthesis and robust subgroup or meta-analytic comparisons. Subgroup observations in this review, including apparent differences by IMDC risk group and histological subtype, should be interpreted as exploratory as these analyses were typically conducted in small subsets. Publication bias is also possible, as small single-centre studies reporting null or unfavourable results may be less likely to be published, potentially leading to an over-representation of positive findings in literature. A formal assessment of publication bias (e.g. funnel plots or Egger’s test) was not feasible given the narrative synthesis approach. Furthermore, the observational nature of the included studies makes them susceptible to residual confounding and bias, and therefore reduced internal validity [91]. Unadjusted analyses are susceptible to confounding by indication, whereby patients with more favourable baseline characteristics may be preferentially assigned to certain treatments, potentially biasing comparative effectiveness estimates. Propensity score-based methods and IPTW are also limited to measured variables and cannot address unmeasured confounders such as physician preference, patient frailty or reasons for treatment selection that are not routinely captured in medical records. Accordingly, even adjusted estimates from the included studies should be interpreted cautiously, as residual confounding cannot be excluded in any observational comparison. These findings should therefore be interpreted in the context of existing randomised trial evidence. Real-world data can help contextualise whether treatment effects observed in trials are reproduced in broader populations but do not provide the same level of causal inference as well-conducted RCTs. Conclusions Real-world evidence from this systematic review suggests that ICI–TKI combinations are associated with favourable survival outcomes compared with other first-line options in advanced RCC, whereas the comparative effectiveness of dual ICI therapy versus TKI remains inconsistent. Substantial heterogeneity and methodological limitations across studies, particularly regarding comparative safety, limit definitive conclusions. Clinicians should interpret current real-world evidence cautiously and continue to individualise treatment decisions according to patient risk stratification, comorbidities and therapeutic goals. High-quality real-world studies with robust confounding adjustment and clinically meaningful subgroup analyses are needed to guide optimal treatment selection in routine clinical practice. Supplementary Information Below is the link to the electronic supplementary material.Supplementary file1 (PDF 295 KB) Supplementary file1 (PDF 295 KB)