EPAS1 (encoding HIF-2α) is a transcription factor involved in cellular hypoxia response. Under normoxic conditions, HIF-2α is hydroxylated and targeted for proteasomal degradation by the VHL tumor suppressor protein (15). Under hypoxic conditions or in VHL-inactivated tumors, HIF-2α dimerizes with HIF-1β, translocates to the nucleus and binds hypoxia-response elements (HREs) on target genes (16). This activates transcription of genes involved in angiogenesis, glucose metabolism and cell survival, contributing to the aggressive RCC phenotype (12).

In ccRCC, which constitutes 70–80% of all RCC cases, HIF-2α is frequently upregulated due to VHL mutations (12). This contributes to expression of downstream targets including VEGF, platelet-derived growth factor (PDGF) and glucose transporter 1 (GLUT1), supporting tumor growth and angiogenesis (16). Preclinical studies have shown that HIF-2α inhibition can reduce tumor growth and vascularization in ccRCC models (16–18). Associations between HIF-2α expression and clinicopathological prognosis have varied according to assay methodology and cellular localization. Accordingly, HIF-2α is currently more well established as a therapeutic target rather than a validated prognostic biomarker in ccRCC. Belzutifan demonstrated proof-of-concept efficacy in VHL disease-associated RCC (19), while in sporadic ccRCC after prior systemic therapy, belzutifan showed PFS benefit vs. everolimus (20). These data support HIF-2α as a therapeutic target, though patient populations and evidence levels differ.

The PTEN gene, located at 10q23.31, encodes a phosphatase that negatively regulates the PI3K/Akt signaling pathway by dephosphorylating PIP3 (21). PTEN mutations or deletions in ccRCC are associated with constitutive PI3K/Akt activation, which may promote cell proliferation, survival and resistance to apoptosis (21). PTEN alterations are associated with aggressive tumor features, poor prognosis, shorter OS and increased metastasis risk (21). Preclinical studies have suggested PTEN-deficient tumors may be more susceptible to PI3K/Akt/mTOR inhibitors (21,22) (Fig. 1).

The BAP1 gene, located at 3p21.1, encodes a nuclear deubiquitinase involved in gene expression regulation, DNA repair, cell cycle progression and apoptosis (31,32). BAP1 interacts with BRCA1 and other proteins to maintain genomic stability by removing ubiquitin modifications from histones and proteins, thereby modulating chromatin structure (31).

BAP1 dysfunction in ccRCC is associated with tumor development through increased genomic instability, chromatin remodeling defects and impaired DNA damage response (31,33). In genetically engineered mouse models, BAP1 loss has been shown to drive ccRCC development and is associated with high-grade tumor phenotypes and mTORC1 activation (34). Clinically, BAP1 mutations and protein loss have been associated with aggressive tumor features including high Fuhrman grade, advanced stage and poor prognosis (33,35). Patients with BAP1-altered tumors exhibit shorter OS and PFS compared with those with intact BAP1 (33,35). BAP1 loss disrupts DNA-damage responses and has provided a mechanistic rationale for exploring DNA-repair-directed strategies; however, whether BAP1 status predicts sensitivity to poly(ADP-ribose) polymerase inhibition or immune checkpoint blockade in RCC remains unproven (31,33).

SETD2, located at 3p21.31, encodes the principal histone methyltransferase responsible for H3K36 trimethylation and contributes to transcriptional fidelity, DNA-damage responses and genomic stability. Loss of SETD2 in RCC leads to broad epigenetic and genomic dysregulation (36–38). Preclinical RCC models further suggest that SETD2 loss may create therapeutic vulnerabilities, including increased sensitivity to PI3Kβ/Akt inhibition, despite no SETD2-based predictive strategy having been clinically validated (39).

The KDM5C gene, located at Xp11.22, encodes a histone lysine demethylase that specifically removes methyl groups from lysine 4 of histone H3, serving a key role in transcriptional regulation and chromatin remodeling. KDM5C contains numerous functional domains including a catalytic JmjC domain, zinc finger-C5HC2 domain, JmjN domain, Bright/ARID DNA-binding domain and two plant homeodomain fingers for histone binding (40). In ccRCC, KDM5C mutations are identified at relatively low frequency and are generally associated with loss of function and altered histone methylation patterns (40). Preclinical studies have suggested that KDM5C loss may produce context-dependent epigenetic effects and create therapeutic vulnerabilities in experimental models; however, its clinical prognostic and predictive importance remains investigational (40,41).

The HIF-1 signaling pathway is involved in RCC pathogenesis. HIF-1α accumulation drives transcription of angiogenic and metabolic genes (44,47), with PI3K/Akt/mTOR crosstalk enhancing oncogenic signaling (48). Direct HIF-1α targeting has not yet yielded an established therapeutic strategy in RCC, whereas HIF-2α has proved more pharmacologically tractable; efforts have therefore concentrated on downstream VEGF targets and HIF-2α (49,50).

The PI3K/Akt pathway is initiated by growth factor binding to receptor tyrosine kinases, activating PI3K to convert PIP2 to PIP3, which recruits and activates Akt (21). Akt phosphorylates downstream effectors including mTOR, GSK3β and FOXO transcription factors, which may promote cell survival and proliferation. In ccRCC, PIK3CA mutations and PTEN loss-of-function mutations are associated with constitutive pathway activation (21). Although PI3K/Akt signaling remains biologically relevant and selective PI3K or Akt inhibitors continue to be investigated, neither class has established a clinically validated single-agent role in RCC. By contrast, the downstream mTOR inhibitors everolimus and temsirolimus have demonstrated efficacy in defined clinical settings, although their roles have become more limited in the contemporary ICI/TKI era (9,21,51).

Despite canonical KRAS mutations being uncommon in ccRCC, MAPK pathway activation may occur through upstream receptor tyrosine kinase signaling and pathway crosstalk (54). The Ras pathway intersects with the VHL-HIF pathway, with hypoxia-induced HIFs enhancing Ras-mediated oncogenic signaling (47). MEK inhibitors such as trametinib have shown promise in preclinical models (54), though clinical application remains limited by pathway redundancy and compensatory mechanisms. To the best of our knowledge, no MEK inhibitor has demonstrated RCC-specific clinical activity sufficient to advance to phase III development, illustrating the challenge of targeting downstream signaling nodes in a pathway-driven tumor.

RCC exhibits profound metabolic reprogramming, including enhanced glycolysis (Warburg effect), glutaminolysis and lipid synthesis, driven by VHL-HIF pathway activation (44,55). While these metabolic alterations are well characterized biologically, to the best of our knowledge, no metabolism-targeted therapy has established clinical efficacy or a standard therapeutic role in RCC and metabolic interventions remain investigational. The primary therapeutic relevance of metabolic reprogramming currently lies in its role as a resistance mechanism to VEGF-TKIs and ICIs, rather than as a direct drug target.

The IL-6/JAK/STAT3 axis is a well-characterized oncogenic and immunomodulatory pathway across cancers (58,59). In metastatic ccRCC, elevated circulating IL-6 has been associated with poorer clinical outcomes and reduced response to sunitinib (60). Accordingly, the IL-6/JAK/STAT3 pathway remains an investigational therapeutic axis rather than an established treatment target in RCC.

The VEGF pathway is an important driver of angiogenesis and tumor growth in RCC. First-generation VEGF receptor TKIs established the foundation for targeted therapy. In a pivotal phase III trial, sunitinib demonstrated a median PFS of 11 months compared with 5 months with interferon-α and an objective response rate (ORR) of 31 vs. 6% (11,69). Sorafenib improved median PFS to 5.5 months compared with 2.8 months with placebo in the TARGET trial (70). Axitinib, a second-generation selective TKI, exhibited improved PFS (6.7 months) compared with sorafenib (4.7 months) in previously treated advanced RCC (71).

Combination therapies are being explored to address therapeutic resistance. A Phase Ib trial of axitinib + pembrolizumab reported a 58% ORR and a median PFS of 20.9 months in the dose-expansion cohort, illustrating the complementary mechanisms of VEGF inhibition and immune checkpoint blockade (72). Next-generation VEGF inhibitors such as cabozantinib target mesenchymal-epithelial transition (MET) and AXL in addition to VEGFR. The Cabozantinib vs. Everolimus in Advanced Renal-Cell Carcinoma Trial (METEOR) demonstrated that cabozantinib significantly improved PFS [7.4 vs. 3.9 months; hazard ratio (HR) 0.51; P<0.0001] and OS (21.4 vs. 17.1 months; HR 0.70; P=0.0002) compared with everolimus in previously treated patients (73,74). The Cabozantinib vs. Sunitinib in Metastatic Renal Cell Carcinoma Trial (CABOSUN) demonstrated that cabozantinib improved median PFS compared with sunitinib as first-line therapy in intermediate- and poor-risk patients (8.6 vs. 5.3 months by independent review; HR 0.48), with updated median OS of 26.6 vs. 21.2 months (HR 0.80) (75,76).

Tivozanib is a highly selective VEGFR-1/2/3 inhibitor with a narrower kinase inhibition profile compared with broader-spectrum TKIs. Its clinical activity was initially demonstrated in the phase III TIVO-1 trial (77), while the phase III TIVO-3 trial further supported its role in previously treated advanced RCC by showing improved PFS compared with sorafenib (78). Therefore, tivozanib is best discussed as a VEGFR-selective option with particular relevance in the later-line setting rather than as a broadly preferred first-line standard (77,78). Despite these advancements, challenges persist. Biomarkers to predict VEGF inhibitor response remain limited and angiogenesis pathway redundancy contributes to treatment failure (66). The broader toxicity profile of cabozantinib (including palmar-plantar erythrodysesthesia and gastrointestinal adverse events) requires experience in management (79) (Table III).

The mTOR pathway regulates cell proliferation and angiogenesis in RCC. The Renal Cell Cancer Treatment with Oral RAD001 Given Daily Trial (RECORD-1) trial established everolimus as a commonly used second-line therapy after TKI failure, demonstrating a median PFS of 4.9 vs. 1.9 months for placebo (P<0.001) (9). Subsequent phase II studies further evaluated single-agent everolimus in metastatic RCC. The phase II trial by Amato et al (80) reported disease stabilization and prolonged progression-free survival in patients with predominantly ccRCC, supporting the antitumor activity of mTOR inhibition in this setting. Temsirolimus demonstrated superior survival benefits in patients with poor prognosis. The Advanced Renal Cell Carcinoma Trial revealed a median OS of 10.9 months for temsirolimus vs. 7.3 months for interferon-α (P=0.008) in untreated high-risk patients with metastatic RCC (51).

Biomarkers are key for personalizing mTOR-targeted therapy. Combined subgroup analyses of the RECORD-1 and REACT trials demonstrated that hyperglycemia and hypercholesterolemia during everolimus treatment associated with improved PFS (7.4 vs. 5.3 months, P=0.01), suggesting metabolic toxicity as a potential surrogate for drug activity (81). Proactive management of everolimus-associated adverse events, including stomatitis, noninfectious pneumonitis and metabolic abnormalities, is important for maintaining dose intensity (82).

The PD-1/PD-L1 axis is key in tumor immune evasion. Nivolumab, a monoclonal antibody targeting PD-1, demonstrated notable efficacy in advanced RCC after prior VEGF-targeted therapy. In the phase III CheckMate 025 trial, nivolumab showed superior OS compared with everolimus (25.0 vs. 19.6 months; HR 0.73; P=0.002) with an ORR of 25 vs. 5% (83).

Ipilimumab + nivolumab combination therapy has reshaped first-line treatment. In an 8-year analysis of CheckMate 214, nivolumab + ipilimumab maintained an OS benefit vs. sunitinib in both the intention-to-treat population (median OS, 52.7 vs. 37.8 months; HR 0.72, 95% CI 0.62–0.83) and the intermediate/poor-risk population (median OS, 46.7 vs. 26.0 months; HR 0.69, 95% CI 0.59–0.81) (84). In the intent-to-treat (ITT) population, median PFS was similar (12.4 vs. 12.3 months; HR 0.88), whereas greater PFS separation was observed in the intermediate/poor-risk subgroup. At the final analysis with 9.3 years of median follow-up, the OS benefit remained durable in the ITT population (HR 0.71; 95% CI 0.62–0.82), with estimated 108-month OS probabilities of 31.4 vs. 19.5% (85).

The Pembrolizumab + Axitinib vs. Sunitinib Trial (KEYNOTE-426) demonstrated efficacy over sunitinib in the first-line setting, with a median OS of 47.2 vs. 40.8 months (HR 0.84; 95% CI 0.71–0.99) and median PFS of 15.7 vs. 11.1 months (HR 0.69; 95% CI 0.59–0.81) (5). The ORR was 60.6 vs. 39.6%, with complete responses in 11.6 vs. 4.0% of patients. In addition, 5-year follow-up data demonstrated durable benefit (5).

The Avelumab + Axitinib vs. Sunitinib Trial (JAVELIN Renal 101) exhibited significant PFS improvement compared with sunitinib in the primary analysis (13.8 vs. 8.4 months; HR 0.69; 95% CI 0.574–0.825; P<0.0001) (57). In the final analysis with a minimum follow-up of 68 months, median PFS was 13.9 vs. 8.5 months (HR 0.66) and the median OS was 44.8 vs. 38.9 months in the overall population (HR 0.88; 95% CI 0.749–1.039; P=0.0669) and 43.2 vs. 36.2 months in the PD-L1-positive population (HR 0.86; 95% CI 0.701–1.057; P=0.0755), favoring the avelumab + axitinib arm but not reaching statistical significance (86). The ORR was 59.7 vs. 32.0% (86). Nivolumab + cabozantinib also demonstrated durable first-line efficacy in CheckMate 9ER. In the final analysis (median follow-up, 5.6 years), median PFS was 16.4 vs. 8.3 months (HR 0.58) and median OS was 46.5 vs. 35.5 months (HR 0.79) compared with sunitinib (8).

The Lenvatinib + Pembrolizumab or Everolimus for Advanced Renal Cell Carcinoma Trial (CLEAR) demonstrated that lenvatinib + pembrolizumab is an active first-line option. In the final prespecified OS analysis, median OS was 53.7 months with lenvatinib + pembrolizumab vs. 54.3 months with sunitinib (HR 0.79; 95% CI 0.63–0.99), while median PFS remained markedly longer with the combination (23.9 vs. 9.2 months; HR 0.47), with an ORR of 71.3 vs. 36.7% (6,87). However, the toxicity profile of lenvatinib + pembrolizumab (including hypertension, diarrhea and proteinuria) requires careful patient selection and monitoring.

FGF/FGFR signaling has been implicated as a bypass pathway contributing to resistance to VEGF-directed therapy in RCC (66). However, FGFR-targeted agents have not established a validated clinical role in RCC and RCC-specific prospective evidence remains limited. c-MET signaling is dysregulated in RCC, particularly in the papillary subtype (88). Cabozantinib, a multi-kinase inhibitor targeting c-MET, VEGFR and AXL, demonstrated efficacy in the METEOR trial in previously treated patients (73). Belzutifan, a HIF-2α inhibitor, demonstrated proof-of-concept efficacy in VHL disease-associated RCC (19) and in post-treatment sporadic ccRCC (20).

Combination therapies integrating ICIs with TKIs or dual checkpoint blockade have emerged as clinically impactful approaches. Table IV presents verified efficacy data from pivotal phase II and III trials. Furthermore, the heterogeneity and complexity of RCC necessitate multifaceted treatment strategies. Combination therapies targeting distinct pathways can synergistically inhibit tumor growth and angiogenesis while mitigating resistance. Pairing ICIs with VEGF-TKIs leverages immune recognition complemented by anti-angiogenic effects, creating a more favorable tumor microenvironment.

The CheckMate 214 trial established nivolumab + ipilimumab as a commonly used option for intermediate- and poor-risk patients, demonstrating improved long-term OS and ORR compared with sunitinib (Table IV) (84). However, nivolumab + ipilimumab has a distinct immune-mediated toxicity profile. In long-term follow-up, overall grade 3–4 treatment-associated adverse events were less frequent compared with sunitinib, although immune-mediated adverse events may require corticosteroid treatment and can lead to treatment discontinuation (85,89). KEYNOTE-426, JAVELIN Renal 101 and CLEAR each demonstrated improved PFS for their respective ICI-TKI combinations compared with sunitinib (Table IV) (5,6,57).

A critical efficacy finding was demonstrated through the Belzutifan vs. Everolimus for Advanced Renal-Cell Carcinoma Trial (LITESPARK)-005, which evaluated belzutifan vs. everolimus in heavily pretreated patients (post-ICI and post-VEGF-TKI). While median PFS was identical between arms (5.6 vs. 5.6 months), PFS was significantly improved with belzutifan (HR 0.75, 95% CI 0.63–0.90; two-sided log-rank P=0.002) and ORR favored belzutifan (21.9 vs. 3.5%), OS did not reach statistical significance (HR 0.88; 95% CI 0.73–1.07; P=0.200), representing an important limitation in this refractory population (20). The absence of a statistically significant OS advantage at the reported analysis tempers enthusiasm for belzutifan monotherapy in this population and combination strategies may further define its optimal clinical role. Other belzutifan-based combination strategies remain investigational, whereas LITESPARK-011 has now reported phase III efficacy results (90).

Building upon these findings, the phase III LITESPARK-011 trial (NCT04586231) evaluated belzutifan + lenvatinib vs. cabozantinib in patients with advanced clear cell RCC after progression on anti-PD-L1 therapy; at the second interim analysis, the combination significantly improved median PFS (14.8 vs. 10.7 months; HR 0.70; 95% CI 0.59–0.84; one-sided P<0.0001) and ORR (52.6 vs. 40.2%), while median OS (34.9 vs. 27.6 months; HR 0.85) did not reach statistical significance at the time of this interim analysis (90). In addition, the phase III LITESPARK-022 study (NCT05239728) evaluated adjuvant pembrolizumab + belzutifan vs. pembrolizumab plus placebo in patients with clear cell RCC at increased risk of recurrence after nephrectomy, demonstrating significantly improved disease-free survival (HR 0.72; 95% CI 0.59–0.87; two-sided P<0.001), although OS data remain immature (91). These trials support an expanding role for HIF-2α-directed therapy across sequential, combination and adjuvant settings, although optimal patient selection, treatment sequencing and mature OS benefit remain to be defined.

The combination of lenvatinib + everolimus (Lenvatinib + Everolimus Combination Trial (Study 205)], a phase II three-arm trial in previously treated patients, demonstrated a median PFS of 14.6 vs. 5.5 months (HR 0.40; P=0.0005) and median OS of 25.5 vs. 15.4 months (HR 0.51; P=0.024) compared with everolimus alone (92). This supported subsequent first-line combination development.

Ongoing trials are exploring novel combinations, including the addition of C-X-C chemokine receptor type 4 (CXCR4) inhibitors to immunotherapy regimens. A phase 1b trial of the CXCR4 inhibitor mavorixafor in combination with nivolumab in patients with advanced ccRCC unresponsive to nivolumab monotherapy demonstrated potential antitumor activity and a manageable safety profile (93). In addition, the search for reliable biomarkers to guide treatment selection and predict responses remains a key area of investigation.

The transition from empiric to precision therapy in RCC requires integrating molecular profiling, validated biomarkers and resistance-aware treatment sequencing, objectives that remain incompletely realized. Precision medicine in RCC has progressed beyond single-biomarker approaches toward integrated, multi-dimensional decision models. Molecular profiling at baseline establishes the biological rationale for subsequent therapeutic decisions. VHL mutation status represents a central biological axis: VHL inactivation activates the HIF-VEGF axis and provides a strong biological rationale for sensitivity to anti-angiogenic therapy (50); however, to the best of our knowledge, VHL status has not been prospectively validated as a predictive biomarker for treatment selection. Given the important role of VHL-HIF axis dysregulation in ccRCC pathogenesis, HIF-2α inhibition with belzutifan is a mechanistically rational approach. However, translation from VHL-disease biology to sporadic ccRCC efficacy requires careful population-specific interpretation, as the clinical activity in VHL disease does not fully predict outcomes in sporadic metastatic RCC (20).

International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) risk stratification integrates clinical and laboratory parameters (Karnofsky performance status, time from diagnosis to treatment, hemoglobin, calcium, neutrophil count and platelet count) to categorize patients into favorable, intermediate and poor risk groups (94). IMDC is an established prognostic model (94) and has subsequently been incorporated into multiple phase III RCC trials (4–6). However, whether IMDC risk independently predicts differential benefit from specific therapeutic classes (ICI combinations vs. TKI monotherapy) remains less certain, as cross-trial comparisons are inherently limited by population differences.

Key chromatin remodeling gene mutations provide additional prognostic information. BAP1 mutations associate with aggressive disease features and shorter OS (33,35). PBRM1 mutations show variable prognostic impact depending on co-occurring alterations; the association between PBRM1 alterations and immunotherapy response remains controversial, with conflicting data across different cohorts that likely reflect co-occurring mutations (especially BAP1), patient heterogeneity and assay differences (28–30). SETD2 mutations are associated with higher tumor grade and increased metastasis risk (37). PD-L1 expression has been evaluated as an adjunctive biomarker in several RCC trials; however, its predictive utility is inconsistent and it is not validated for stand-alone treatment selection (57).

First-line treatment selection is guided primarily by IMDC risk category and histological features. For patients of IMDC favorable-risk, TKI monotherapy, including sunitinib (11) or pazopanib (95), remains an active treatment option based on phase III evidence, particularly for patients with low disease burden, indolent disease behavior or contraindications to immunotherapy (7,11,95). Alternatively, ICI + TKI combinations (such as pembrolizumab + axitinib or nivolumab + cabozantinib) may be considered for patients of favorable-risk with high disease burden or rapid progression potential (5,8). The choice incorporates patient comorbidities, performance status and toxicity preferences. For patients of IMDC intermediate/poor-risk, dual ICI with nivolumab + ipilimumab offers durable long-term benefit; at 8-year follow-up, median OS was 46.7 months vs. 26.0 months with sunitinib (HR 0.69; 95% CI 0.59–0.81) (84). ICI + TKI combinations such as pembrolizumab + axitinib, nivolumab + cabozantinib and lenvatinib + pembrolizumab have produced high objective response rates in their respective phase III trials and may be considered when rapid tumor shrinkage is clinically important (5–8,57). However, to the best of our knowledge, no head-to-head trial directly compares dual ICI with ICI + TKI regimens and indirect response-rate comparisons across trials should therefore be interpreted cautiously. For non-clear-cell RCC, treatment selection should be histology-specific. In papillary RCC, cabozantinib has randomized phase II evidence from the SWOG 1500/PAPMET trial (96), whereas evidence for other non-clear-cell histologies is more heterogeneous and enrollment in clinical trials is encouraged.

Upon disease progression, treatment selection is guided by the prior therapeutic context. After progression on VEGF-TKI monotherapy, cabozantinib has shown activity in clinical trials (multi-target TKI with MET/AXL inhibition) (73), nivolumab (ICI switch) (83), lenvatinib + everolimus (TKI + mTOR combination) (92) or tivozanib (selective VEGFR TKI) (77,78). The choice depends on prior TKI exposure, residual toxicity, and molecular features. After progression on ICI-based therapy, VEGF-directed TKIs remain important treatment options. Cabozantinib served as the active comparator in the phase III CONTACT-03 trial after prior ICI therapy, whereas the addition of atezolizumab to cabozantinib did not improve outcomes (97). Tivozanib represents another later-line VEGFR-directed option supported by phase III TIVO-3 data (78). Belzutifan (HIF-2α inhibitor) is a mechanistically rational option after prior ICI and VEGF-TKI therapy, based on LITESPARK-005 results showing PFS benefit (20), though the absence of a statistically significant OS advantage at the reported analysis tempers enthusiasm for belzutifan monotherapy in this population.

Continuous monitoring enables treatment modification throughout the disease course. Among candidate biomarkers, PBRM1 remains investigational because associations with ICI response have been inconsistent across cohorts and may be influenced by co-occurring alterations such as BAP1 (28–30). BAP1 is more established as a prognostic marker associated with aggressive tumor biology than as a validated predictive marker for treatment selection (33,35). PD-L1 expression should not be used as a stand-alone treatment-selection biomarker in RCC because its predictive associations vary across therapeutic regimens, assay platforms and cut-off definitions (4,57). Circulating biomarkers, including circulating tumor (ctDNA) and circulating tumor cells (CTCs), also remain investigational; relatively low ctDNA shedding and the lack of prospectively validated treatment-adaptation algorithms currently limit routine clinical implementation. Systemic inflammatory markers such as NLR and CRP provide prognostic information in ICI-treated RCC cohorts, with elevated values generally associated with poorer outcomes (68). Standard radiographic assessment per RECIST remains the cornerstone of response evaluation, performed at regular intervals (typically every 8–12 weeks during active therapy) (7,98).

RCC continues to pose challenges in oncology despite key therapeutic advances. The integration of multi-omics data, including genomics, transcriptomics, proteomics, metabolomics and spatial profiling, holds promise for unraveling the complex molecular landscape of RCC (67,99). Recent studies have highlighted the importance of genomic classification in RCC, emphasizing the need for personalized treatment strategies (67,99). Notably, a multi-omic analysis of 100 ccRCC cases integrating genomic, transcriptomic, proteomic, metabolomic and spatial data identified metabolic reprogramming associated with disease progression (99). Metabolomic analyses have revealed potential therapeutic targets and biomarkers. Single-cell omics approaches are emerging but remain underexplored in RCC; their application resolves intra-tumor heterogeneity and clarifies differential therapeutic responses at the cellular level. By combining multi-omics datasets, including prospective single-cell profiling, researchers can develop more precise treatment plans tailored to individual patients.

Artificial intelligence and radiomics are emerging as complementary tools for precision oncology in RCC. Machine learning algorithms applied to radiomic features extracted from routine imaging may non-invasively predict molecular subtypes, treatment response and resistance patterns, though these approaches remain investigational and require prospective validation before clinical implementation (100,101).

Drug resistance in RCC follows distinct evolutionary patterns depending on the therapeutic class. Understanding these mechanisms informs rational treatment sequencing and the design of clinical trials evaluating novel combination approaches.

Resistance to VEGF-TKIs involves primary mechanisms (alternative pro-angiogenic signaling through FGF/PDGF, MET pathway activation and MDSC recruitment) (66) and acquired mechanisms (EMT, alternative angiogenic factor upregulation and vessel co-option). Collectively, VEGF resistance reflects adaptive vascular rewiring rather than linear pathway escape, explaining why sequential VEGF-directed strategies and combination approaches targeting parallel axes (ICI, HIF-2α) have become necessary.

Resistance to ICIs reflects both tumor-intrinsic and microenvironmental factors. An important challenge is that RCC employs multiple redundant immune evasion mechanisms, antigen loss, T-cell exhaustion and alternative checkpoint upregulation, meaning no single intervention is likely to restore immune control in resistant disease. Primary resistance involves an immunologically ‘cold’ phenotype with defective antigen presentation and an immunosuppressive microenvironment (56). Acquired resistance develops through T-cell exhaustion, antigen loss through immunoediting, and compensatory checkpoint upregulation (56). Strategies intended to limit or circumvent primary immune resistance include dual checkpoint blockade and ICI-TKI combinations (4,6). However, these first-line trials do not establish reversal of acquired ICI resistance and to the best of our knowledge, no strategy has shown consistent efficacy in reversing established resistance. Novel targets such as lymphocyte activation gene 3 (LAG-3) and T-cell immunoreceptor with Ig and ITIM domains (TIGIT) therefore remain under investigation.

Resistance to HIF-2α inhibition remains incompletely characterized. Translational studies with the first-generation HIF-2α inhibitor PT2385 identified an acquired EPAS1/HIF-2α G323E gatekeeper mutation that interfered with drug binding and preserved HIF-2 complex formation, providing evidence for on-target acquired resistance in ccRCC (102). Additional HIF-independent or alternative clonal escape mechanisms remain under investigation. Belzutifan-based combinations are being explored clinically. Notably, LITESPARK-011 has now reported phase III results demonstrating improved PFS with belzutifan + lenvatinib vs. cabozantinib after prior anti-PD-L1 therapy (90), whereas other belzutifan-based combination strategies remain investigational.

Potential cross-class interactions merit consideration. VEGF-pathway inhibition can exert immunomodulatory effects and may alter the tumor microenvironment in ways that could influence subsequent ICI activity (103,104). Conversely, VEGF-directed TKIs retain clinical activity in selected patients after ICI progression, as illustrated by the cabozantinib comparator arm in CONTACT-03 (97). These observations are hypothesis-generating and prospective evidence defining optimal cross-class sequencing remains lacking (Fig. 3).

A number of therapeutic approaches that once appeared promising have not yet established durable roles in RCC clinical practice. Everolimus was commonly used as second-line therapy after the RECORD-1 trial demonstrated PFS benefit vs. placebo (9), but nivolumab (83) and cabozantinib (73) have largely replaced it in this setting based on improved OS data. Everolimus monotherapy now has a limited role except in specific contexts where immunotherapy is contraindicated. Although TIVO-1 did not establish tivozanib as a broadly adopted first-line standard, subsequent TIVO-3 data supported its role as a later-line option in previously treated advanced RCC (77,78). Thus, tivozanib is better viewed as a VEGFR-selective later-line therapy rather than as a failed therapeutic strategy. FGFR inhibitors, including erdafitinib (105), have demonstrated activity in urothelial carcinoma but lack RCC-specific phase III evidence; extrapolation from other tumor types is not supported by current evidence without dedicated trials. Imatinib was evaluated directly in metastatic RCC, but a small phase II study reported no objective responses among 12 evaluable patients, illustrating the limited clinical translation of KIT/PDGF receptor-directed therapy in unselected RCC populations (106).

The biomarker landscape in RCC remains marked by inconsistency and limited clinical utility. PBRM1 mutations illustrate this challenge: A number of studies suggest enhanced ICI benefit while others demonstrate resistance (28–30). This controversy likely reflects co-occurring mutations (especially BAP1), patient heterogeneity, differences in assay platforms and varying analytical approaches across cohorts. No consensus exists on the clinical utility of PBRM1 as a predictive biomarker and its use in routine treatment selection is not supported by prospective data. Overall, no single biomarker has achieved sufficient consistency and prospective validation to guide routine clinical decision-making in RCC. PD-L1 expression was incorporated into prespecified analyses in trials such as JAVELIN Renal 101 (57), but its predictive value has not been consistent across RCC trials. CheckMate 214 showed benefit regardless of PD-L1 status (4) and it is not validated as a stand-alone treatment-selection biomarker. These discordant results limit routine clinical use and PD-L1 remains an adjunctive rather than definitive predictive marker. VHL is a central pathway-defining alteration in ccRCC, but its clinical predictive value for selecting among anti-angiogenic therapies remains unvalidated; routine VHL testing to guide TKI selection has not been adopted (12,50). ctDNA holds promise in other malignancies for response monitoring and resistance detection, but RCC-specific validation is lacking due to lower tumor shedding rates and no prospective data support ctDNA-guided treatment adaptation in RCC.

From a clinical implementation perspective and consistent with the aforementioned biomarker-specific assessment, none of these molecular biomarkers currently provide sufficient prospective evidence for routine treatment selection in RCC. PBRM1, BAP1, PD-L1 and circulating biomarkers such as ctDNA and CTCs all remain investigational for treatment selection and their predictive utility requires prospective validation in dedicated clinical trials.

Numerous mechanistic assumptions underlying RCC therapy have been challenged or tempered by clinical evidence. The assumption that ‘VEGF inhibition alone is sufficient’ has been challenged by the rapid emergence of resistance to single-agent VEGF-TKIs and the subsequent need for multi-target TKIs and combination approaches, though even these more intensive strategies demonstrate limited long-term durability, highlighting the adaptive capacity of RCC. The assumption that ‘immunotherapy benefit is independent of VEGF’ has been challenged by ICI + TKI combination data showing improved outcomes to historical ICI monotherapy data (5,6,8), though to the best of our knowledge, no direct ICI vs. ICI + TKI head-to-head trials exist to conclusively address this question. The assumption that ‘HIF-2α inhibition translates directly from VHL-disease to sporadic RCC’ has been tempered by LITESPARK-005, which demonstrated a PFS benefit, whereas OS did not meet the prespecified statistical-significance criterion at the reported analysis in a heavily pretreated population (20), suggesting that the mechanistic rationale, while sound, may not fully capture the complexity of treatment-resistant sporadic disease. Finally, the concept of ‘EMT as a therapeutic target’, despite extensive preclinical literature (66,107), lacks sufficient prospective validation in RCC, EMT is a descriptive biological process rather than a validated therapeutic target in RCC.

Evidence grading limitations should be acknowledged when interpreting the RCC literature. Numerous biomarker claims that are often cited as high-level evidence derive from post hoc subgroup analyses of phase III trials rather than from prospectively designed biomarker-stratified studies. This distinction is key, as post-hoc analyses are hypothesis-generating rather than hypothesis-confirming and their findings require prospective validation before clinical implementation. The IMDC prognostic model was established prospectively and validated across multiple trials, but its predictive value for guiding treatment selection (as opposed to prognosis) is less certain. Cross-trial comparisons are inherently limited by differences in patient populations, eligibility criteria, assessment methodologies and follow-up durations, yet such comparisons frequently appear in the literature and should be interpreted with caution. Real-world patient populations are often excluded from pivotal trials, including those with brain metastases, poor performance status and non-clear cell histology, limiting the generalizability of trial findings to the broader RCC population. These limitations underscore the need for prospective biomarker-embedded trials and real-world evidence studies to complement the existing randomized controlled trial data.

Liquid biopsy technology (108), involving analysis of CTCs, cell-free DNA and exosomes (109), offers a non-invasive alternative for RCC diagnosis, prognosis and treatment monitoring. Cell-free DNA analysis can detect actionable mutations, monitor treatment response and identify emerging resistance mechanisms in real time. However, RCC exhibits lower baseline ctDNA shedding rates compared with malignancies such as lung or colorectal cancer, which may limit the sensitivity of ctDNA-based monitoring in this disease. Numerous studies have demonstrated the feasibility of liquid biopsy in RCC (108–110), though prospective validation and standardization in RCC remain ongoing challenges. Integration of liquid biopsy into clinical practice could enable dynamic treatment adaptation and early intervention at progression, pending prospective RCC-specific validation.

The RCC therapeutic landscape continues to evolve with investigation of novel targets. HIF-2α inhibition with belzutifan may represent an emerging targeted approach directed at the important VHL-HIF axis (19,111). Additional belzutifan-based combination strategies remain investigational and belzutifan + lenvatinib has now demonstrated a PFS benefit over cabozantinib in the phase III LITESPARK-011 trial (90). In the adjuvant setting, LITESPARK-022 further extends this strategy by combining HIF-2α inhibition with PD-1 blockade; pembrolizumab + belzutifan notably improved disease-free survival compared with pembrolizumab alone, although OS data remain immature (91). Other investigational approaches include novel immunotherapy combinations such as LAG-3, TIGIT and T-cell immunoglobulin and mucin-domain containing-3 inhibitors, tumor-infiltrating lymphocyte therapy and chimeric antigen-T cell approaches; antibody-drug conjugates targeting RCC-specific antigens; and metabolic pathway inhibitors targeting glycolysis, glutaminolysis and lipid synthesis. The integration of these novel approaches with established targeted and immunotherapy agents may further improve outcomes for patients with advanced RCC.

The RCC therapeutic landscape continues to evolve rapidly. Numerous areas warrant particular attention in future research. First, the development of predictive biomarkers capable of guiding treatment selection beyond IMDC risk stratification remains a key unmet need. Prospective validation of circulating tumor DNA, radiomic signatures and multi-omic integrative platforms may enable dynamic treatment adaptation. Second, further phase III validation, mature OS follow-up and prospective sequencing studies are needed for HIF-2α-based combination strategies, while novel checkpoint targets such as LAG-3 and TIGIT require dedicated prospective clinical evaluation. Third, treatment approaches for non-clear cell RCC histologies and patient populations underrepresented in clinical trials demand dedicated investigation and careful prospective trial design will be essential to validate these emerging strategies (112).