Advancing Clinical Trials for Rare Renal Cell Carcinoma Subtypes: Consensus Statements from the International Kidney Cancer Symposium North America 2025 Think Tank

Abstract

Rare renal cell carcinoma (RCC) subtypes present unique challenges for clinical trial design and drug development. This consensus initiative aimed to provide actionable expert guidance on advancing preclinical and clinical development of rational therapeutic strategies for non-clear cell RCC variants, including papillary, chromophobe, MiT family/translocation, collecting duct, renal medullary carcinoma, and fumarate hydratase-deficient RCC. A modified Delphi method was employed to develop consensus statements among a multidisciplinary panel of 46 experts in urologic oncology, medical oncology, radiation oncology, molecular biology, genetics, and biostatistics, with representatives from pharmaceutical industry, regulatory affairs, and patient advocacy. Over multiple rounds, including an in-person meeting on November 13, 2025, twenty initial statements were proposed, evaluated, refined, and voted on. Consensus was defined a priori as a median Likert score ≥8 out of 10. Twenty final consensus statements were endorsed across five thematic domains: 1) Trial Design and Endpoints; 2) Perioperative Trials; 3) Operations and Accrual; 4) Biology-driven and Histology/molecular Strategy Trials; and 5) Preclinical Efforts, Target Identification and Early Signal Testing. Key recommendations include prioritizing histology-specific trial designs over pooled ‘non-clear cell’ approaches, adopting innovative single-arm and adaptive designs for ultra-rare subtypes, leveraging patient advocacy collaborations and natural history registries, and pursuing mechanism-informed therapeutic development. These recommendations provide a framework to guide researchers, cooperative groups, regulatory bodies, and pharmaceutical industry partners in advancing evidence generation and therapeutic development for patients with rare kidney cancer variants.

1. Introduction

Rare renal cell carcinoma (RCC) subtypes collectively represent approximately 25% of all kidney cancers, yet patients with these diseases face substantial challenges in accessing evidence-based therapies [1–3]. Unlike clear cell RCC (ccRCC), which benefits from numerous randomized controlled trials and approved systemic therapies, more rare variants such as papillary RCC (pRCC), chromophobe RCC (ChRCC), MiT family/translocation RCC, collecting duct carcinoma, SMARCB1-deficient renal medullary carcinoma (RMC), and fumarate hydratase (FH)-deficient RCC remain largely understudied with limited clinical trial access and lack of dedicated US Food and Drug Administration (FDA)-approved treatment options in the advanced disease setting. The biological heterogeneity of these tumors, coupled with their low individual prevalence, creates significant barriers to traditional clinical trial design and accrual.

Historically, clinical trials in this space have pooled diverse rare RCC histologies under a single “non-clear cell” umbrella, an approach increasingly recognized as scientifically inadequate and potentially harmful to patients whose diseases carry distinct biological underpinnings and therapeutic susceptibilities [1, 4]. The 2022 World Health Organization classification of renal tumors reflects the evolving understanding of RCC as a collection of related, but molecularly distinct, entities rather than a monolithic disease, further underscoring the need for histology-specific research strategies [1, 5].

Recognizing these challenges, the Kidney Cancer Association (KCA) convened a multidisciplinary Think Tank during the International Kidney Cancer Symposium (IKCS) North America on November 13, 2025. This initiative brought together medical oncologists, urologic oncologists, radiation oncologists, geneticists, pathologists, translational scientists, biostatisticians, regulatory experts, pharmaceutical industry partners, and patient advocates. Building on the paradigm of the 2024 IKCS Think Tank, which generated consensus statements for biomarkers and risk stratification [6], this session focused specifically on clinical trial design and methodology of developing therapeutic strategies for rare RCC subtypes. The objectives were to: 1) examine current evidence and identify methodologic shortcomings of past and ongoing treatment development efforts for rare RCC variants, including the bench-to-bedside bottleneck; 2) discuss operational barriers for clinical trial conduct including challenges in patient access; 3) propose innovative study designs suited to rare cancers; and 4) develop consensus recommendations to guide future research. The outcome of this meeting was a set of 20 consensus statements reflecting collective expert opinion on how to most effectively advance clinical trials for patients with rare kidney cancer subtypes.

2. Materials and Methods

Study design and participants

A modified Delphi technique [7] was employed to develop consensus statements on clinical trials for rare RCC subtypes. The Delphi panel was composed of 59 experts in the field of kidney cancer invited by the International Kidney Cancer Symposium steering committee. The 46 participants represented diverse expertise including medical oncology, urology, molecular biology, radiation oncology, genetics, biostatistics, regulatory affairs, pharmaceutical industry partners, and patient advocates. This multidisciplinary composition ensured that scientific, clinical, regulatory, and patient-centered perspectives were incorporated throughout the consensus process.

Delphi process overview

The modified Delphi process was conducted in three structured rounds between October 2025 and December 2025, following methodology similar to the 2024 IKCS Think Tank consensus statement [6]. The modified Delphi process utilized iterative statement refinement coupled with anonymous scoring to minimize dominance effects and hierarchy-driven bias. Consensus assessment was determined using quantitative summaries (median and interquartile range [IQR]) to evaluate group agreement (Figure 1). In the first round, an online survey (SurveyMonkey) was sent in October 2025 containing 20 preliminary statements drafted by the co-chairs (M.V. and P.M.) and the KCA Chief Scientific Officer (S.L.R.) based on a literature review and organized across five thematic domains (Table 1): (1) Trial Design and Endpoints, (2) Perioperative Trials, (3) Operations and Accrual, (4) Biology-Driven and Histology/Molecular Strategy Trials, and (5) Preclinical Efforts, Target Identification, and Early Signal Testing. Participants rated statements on a 0–10 Likert scale (0 = strongly disagree, 10 = strongly agree). The ratings were compiled and analyzed, with median scores and IQRs calculated. Consensus was defined a priori as a median score ≥8 [6]. In round 2, an in-person meeting was held on November 13, 2025 for review and discussion of all five thematic domains and with two objectives: to refine statements that failed to reach initial consensus for subsequent revoting, and to discuss statements with high variability (IQR > 2) to clarify divergent viewpoints. Following the in-person meeting, a final online SurveyMonkey survey was sent to participants who responded to the initial survey, containing only the revised statements.

Figure 1.

Modified Delphi Process conducted between October 2025 and December 2025.

Table 1.

Consensus statement development and final voting results

#	Consensus Statement	Median Score	IQR
1	In rare renal cell carcinoma subtypes, randomized trials should remain the preferred approach to establish substantial evidence of effectiveness when feasible (sample size, control selection), with single-arm designs reserved for truly infeasible settings.	8	2.75
2	The field should move away from ‘non-clear cell’ trials that pool various histologies as a composite population for primary endpoint analyses.	9	2
3	Basket, umbrella, or platform trials are appropriate when they increase relevance and efficiency for rare renal cell carcinomas, but histology-specific cohorts (e.g., RMC, ChRCC, pRCC) should be prospectively powered and analyzed as distinct groups.	9	2
4	Papillary RCC is a non-clear cell RCC variant that is sufficiently prevalent to justify histology-specific RCTs.	9	2
5	Initial Statement: Chromophobe RCC is a non-clear cell RCC variant that is sufficiently prevalent to justify histology-specific RCTs.	7	2
	Final Revised Statement: More work is needed to forge collaborations, explore targets and establish referral patterns for dedicated clinical trials in chromophobe RCC, a non-clear cell RCC variant that may be sufficiently prevalent to justify histology-specific RCT in the future.	9	2
6	For rarer RCC variants (e.g., RMC, FH-deficient RCC, collecting duct, MiT family/translocation RCC), we should strictly adopt innovative single-arm/adaptive designs with natural history/external controls, hierarchical borrowing, and explicit confirmatory strategies.	8	2
7	Central pathology review should be incorporated into all rare RCC trials.	9	2
8	Perioperative (adjuvant/neoadjuvant) trials in rare RCCs should be pursued when supported by strong biologic rationale and feasible endpoints. Pooled ‘non-clear cell’ approaches with secondary subgroup analyses should be avoided.	8	2
9	The preferred design for perioperative trials in rare RCCs is a randomized trial (e.g., vs placebo or observation); single-arm designs should be reserved for settings where the intervention is hypothesized to yield exceptionally large and durable effects.	8	2.75
10	For adjuvant papillary RCC randomized trials, the control arm should be observation and not pembrolizumab.	9	3
11	For adjuvant non-papillary RCC randomized trials, the control arm should be observation and not pembrolizumab.	9	2
12	To improve accrual and trial feasibility for rare RCC variants, clinical investigators should collaborate with patient advocacy groups and use natural history registries to identify and engage eligible patients.	10	1
13	Industry collaboration should be actively sought and leveraged to meet practical accrual targets and timelines in rare RCC subtypes.	10	1
14	Rare RCC protocols should include pre-specified adaptive fallback options (e.g., eligibility broadening or cohort expansion) to ensure study completion if initial accrual targets are not met.	9	2
15	Travel support and digital/remote trial operations (e.g., remote screening, eConsent, tele-PROs, decentralized specimen collection) are specifically important for rare RCC trials to reduce travel and access barriers.	9.5	1
16	Even if histology-agnostic labels exist in RCC, rare subtype trials should still pursue histology-specific evidence to refine indications and guide subtype-tailored standards, rather than over-extrapolating from clear cell populations.	9	2
17	For molecularly targeted agents with tumor-agnostic potential (e.g., SMARCB1 deficiency or NF2 loss), RCC-inclusive cohorts should follow methodological standards used in successful tissue-agnostic programs (robust ORR, durability, safety), with RCC-specific follow-up when signals emerge.	9	2
18	For dedicated, mechanism-directed trials in WHO recognized, molecularly-defined RCC populations (e.g., FH-deficient RCC), a “one-trial” paradigm should be pursued - explicitly planned a priori and consisting of an early ORR signal for accelerated approval paired with a built-in randomized confirmation component.	8	2
19	The generation of disease-specific murine models (including syngeneic models and organoids) should be a priority to the field since these are of central importance for biology-directed target discovery, trial design and justification.	9	3
20	Mechanism-informed testing of novel agents using single-patient IND (SPIND), particularly if based on preclinical data and paired with on-treatment pharmacodynamic biomarker acquisition, can be a helpful initial step in signal-testing prior to pursuit of histology/biology specific trials in rare-variant RCC.	8	3.75

Conflict of Interest Management

All panelists disclosed financial relationships in accordance with ICMJE standards (see Declaration of Competing Interest). Several features of the process were designed to minimize the influence of these relationships on the consensus. Scoring was conducted anonymously through independent online surveys, preventing any individual or affiliation from influencing others’ ratings and mitigating dominance effects. Consensus thresholds (median ≥8) were defined a priori. Industry-employed participants — including two panelists employed by the meeting sponsor (Exelixis) — constituted a small minority of the 46-member voting panel and participated without weighted influence. The sponsor had no role in drafting, selecting, discussing, or voting on statements, nor in the decision to submit for publication; its support was limited to logistical and financial facilitation of the in-person meeting. Initial statements drafted by the two academic co-chairs and the KCA Chief Scientific Officer were open to revision by the full multidisciplinary panel across iterative rounds. To confirm that industry affiliation did not drive the results, the analysis was repeated with the two sponsor-employed panelists excluded (Supplementary Table 1).

Statistical Analysis

Likert responses were summarized using median and IQR. Statistical summaries and data visualizations were generated using R v4.5.1.

3. Results

Surveys for the first round were completed by 46/59 invited panelists (78%). This comprised review of 20 consensus statements, of which 19/20 (95%) achieved initial consensus (median score ≥8). One statement regarding ChRCC did not achieve initial consensus (median score 7) and was revised based on panel discussion during the in-person meeting. The revised statement achieved consensus (median 9, IQR 2) in the subsequent vote, with responses from 38/46 panelists (83%). Although all final statements met the pre-specified threshold (median ≥8), the strength of agreement varied. Several met the threshold despite substantial dispersion—most notably Statement 20 (median 8, IQR 3.75; 61% of panelists scoring ≥8), Statement 1 (median 8, IQR 2.75; 57%), Statement 9 (median 8, IQR 2.75; 59%), and Statement 10 (median 9, IQR 3; 70%)—and these are best interpreted as reflecting general directional agreement rather than uniform endorsement. Full response distributions for all statements are provided in Supplementary Table 1. In a post hoc sensitivity analysis excluding the two industry-employed panelists (Exelixis), all statements retained their consensus classification: every statement meeting the threshold in the full panel continued to do so (median ≥8), medians were identical for 19 of 20 final statements and increased marginally for one (Statement 15 increased from 9.5 to 10.0), and interquartile ranges were essentially unchanged (Supplementary Table 1). Notably, the statement endorsing industry collaboration (Statement 13) was unchanged (median 10, IQR 1). The following are the individual consensus statements separated into five themes, followed by a brief rationale for each statement.

Trial Design and Endpoints

Statement 1.

In rare renal cell carcinoma subtypes, randomized trials should remain the preferred approach to establish substantial evidence of effectiveness when feasible (sample size, control selection), with single-arm designs reserved for truly infeasible settings (median score 8, IQR 2.75).

The panel acknowledged that randomized controlled trials (RCTs) remain the methodological gold standard for establishing substantial evidence of efficacy, providing the most robust inference and minimizing bias [8, 9]. When sample size, recruitment, and control selection are realistically achievable, RCTs are favored by regulatory agencies such as the U.S. FDA [10, 11]. However, the discussion highlighted the practical tension between methodological rigor and feasibility in ultra-rare RCC subtypes [12–14]. For diseases such as MiT family/translocation RCC, collecting duct carcinoma, and RMC, insistence on RCTs may inadvertently slow progress, delay patient access to promising agents, or make some trials effectively impossible. Many studies that have shaped management for rare RCC variants have been single-arm trials, reflecting both necessity and historical success [15–18]. The panel emphasized that feasibility must be assessed thoughtfully, incorporating rarity, available benchmarks, biological rationale, and patient need. The concept of the “relevance-robustness trade-off” was introduced (Table 2 and Figure 2), noting that while RCTs provide maximum robustness, relevance to individual patients and specific RCC subtypes may require flexibility. The consensus reflects a balanced view: RCTs as the ideal standard when feasible, accompanied by recognition that flexibility and judicious use of single-arm designs are essential tools for advancing therapies across the spectrum of rare RCC subtypes. This statement also drew wider dispersion than most (IQR 2.75): although the median met the consensus threshold, 43% of panelists scored below it and roughly one-fifth (22%) scored at or below the scale midpoint, reflecting a sizeable minority who favored greater latitude for single-arm designs than a strict “truly infeasible” standard would allow (Supplementary Table 1).

Table 2.

Glossary of Key Concepts in Rare RCC Trial Design.

Concept	Definition / Rationale in Rare RCC
Relevance-Robustness Trade-off	The unavoidable tension between using statistical methods tailored to a specific clinical question (relevance) versus methods that remain valid under a wider range of assumptions (robustness). More relevant models incorporate disease-specific knowledge and yield more tailored, interpretable answers — but only if their assumptions hold. More robust methods make fewer assumptions and can apply to broader populations but may provide less clinically actionable information tailored to specific contexts. Accordingly, randomized trials such as PAPMET may provide robust information across papillary RCC subtypes [30], whereas mechanism-based single-arm studies can provide relevant information in specific molecularly driven papillary RCC subtypes such as fumarate-hydratase deficient RCC [15].
Big Data Paradox	The statistical phenomenon where larger sample sizes from biased or heterogeneous populations (e.g., pooled non-clear cell) lead to narrow confidence intervals around a clinically incorrect or misleading estimate.
Rashomon Effect	The observation that multiple different models or biological interpretations can explain the same clinical dataset; in rare RCC, this necessitates causal mechanistic evidence to inform drug development.
Single Patient IND (SPIND)	A regulatory pathway allowing a single patient access to an investigational drug. In rare RCC, “programmatic” SPINDs with paired tissue can provide early proof-of-mechanism to justify larger trials.
One-Trial Paradigm	A regulatory pathway where a single pivotal study is sufficient for approval, rather than the standard two-trial requirement. The trial can be prospectively designed to provide both the initial signal for accelerated approval (e.g., ORR) and the subsequent randomized data for full confirmation.
Bayesian Borrowing	A statistical approach that uses information from external sources (e.g., historical controls or real-world data) to supplement the analysis of a small, prospectively enrolled rare disease cohort [27].

Figure 2.

The relevance robustness trade-off in clinical trial design for renal cell carcinoma (RCC) subtypes. The diagonal arrow represents the spectrum from complete conditioning (upper left), where inference is maximally tailored to individual patient characteristics, to complete randomization (lower right), where inference is maximally generalizable across populations. The dashed horizontal line denotes the approximate boundary below which traditional randomized controlled trials (RCTs) become feasible. RCC subtypes are positioned according to their practical trial design constraints: ultra-rare but molecularly well-defined variants such as renal medullary carcinoma (RMC) and fumarate hydratase (FH)-deficient RCC occupy the upper-left quadrant, where extreme rarity necessitates designs prioritizing patient relevance over population-level robustness. The greater clinical and biological heterogeneity of clear cell RCC and papillary RCC places them well within the zone where randomization is warranted to achieve population robustness. The horizontal dashed line represents the RCT boundary, the threshold at which a randomized controlled trial (RCT) is considered feasible based on disease prevalence and the availability of control arms. Chromophobe RCC is positioned near the RCT boundary with a question mark, reflecting the Think Tank’s consensus that while it is more prevalent than RMC, it currently lacks the established referral patterns and standardized control arms necessary to consistently justify histology-specific RCTs. The upper-right quadrant represents an idealized state of “deterministic inference” where both perfect relevance and robustness are achieved—a theoretical limit that cannot be attained in practice but serves as a conceptual anchor. This framework emphasizes that trial design choices for rare cancers inherently involve trade-offs, and that different RCC subtypes require different approaches along this spectrum rather than uniform application of traditional RCT methodology.

Statement 2.

The field should move away from ‘non-clear cell’ trials that pool various histologies as a composite population for primary endpoint analyses (median score 9, IQR 2).

Strong consensus emerged that while grouping diverse RCC subtypes under a single “non-clear cell” umbrella for primary endpoint analyses may have been a practical necessity in earlier eras of RCC research, this approach is increasingly recognized as inadequate given current understanding of the distinct molecular and clinical features of individual subtypes [1, 5]. Non-clear cell RCC encompasses biologically and clinically heterogeneous diseases each with distinct molecular features, natural histories, and therapeutic susceptibilities. Composite analyses risk masking clinically meaningful signals and may lead to erroneous conclusions about therapeutic activity. Patient-centered considerations further strengthened consensus; the term “non-clear cell” reduces a patient’s diagnosis to what it is not, rather than affirming its specific identity. The group agreed that primary endpoint analyses should be conducted within biologically coherent, histology-specific or molecularly defined cohorts, and that composite non-clear cell RCC populations should no longer serve as the basis for primary statistical inference.

The statistical phenomenon known as the “big data paradox” has direct implications for clinical trial design in rare RCC subtypes [12, 19]. This paradox describes how larger sample sizes drawn from biased or heterogeneous populations can produce narrower confidence intervals around estimates that are clinically misleading or incorrect (high precision but low accuracy). In the context of rare kidney cancers, pooling diverse histologies under a composite “non-clear cell” umbrella exemplifies this risk: the resulting trials may achieve statistical power and narrow confidence intervals, yet the point estimates may be irrelevant or even harmful when applied to any specific RCC subtype. Recognizing this paradox reinforces the consensus that histology-specific designs, even when statistically modest, may better serve patients than large composite trials that achieve narrow confidence intervals around clinically meaningless averages. The big data paradox also provides a statistical foundation for understanding the relevance-robustness trade-off (Figure 2): RCTs in large, heterogeneous populations maximize robustness (validity across a broad range of assumptions) but sacrifice relevance to the individual patient or specific disease subtype. Conversely, designs that condition heavily on histology, molecular features, or clinical context maximize relevance but depend on assumptions that may not hold universally.

The clinical experience with RMC and immune checkpoint therapy provides a concrete example of how dedicated, histology-specific investigations, even with small sample sizes, can detect clinically critical signals that remain undetectable in larger, pooled cohorts. A phase I trial of cabozantinib combined with nivolumab and ipilimumab enrolled 54 patients with diverse genitourinary malignancies, including patients with RMC, and reported an overall response rate of 30.6% with a median overall survival of 12.6 months [20]. However, the outcomes for the RMC patients within this composite cohort were not separately analyzed, precluding any subtype-specific inferences about efficacy or resistance patterns. In contrast, a dedicated phase II trial of nivolumab plus ipilimumab specifically designed for RMC enrolled 10 patients and was halted for futility at prespecified interim analysis when all patients experienced rapid progression, with 5 of 10 meeting radiological criteria for hyperprogression and a median progression-free survival of only 1.4 months [18]. Similarly, a dedicated phase II basket trial cohort of pembrolizumab in only 5 patients with RMC prospectively identified a striking signal of resistance, with all patients experiencing rapid disease progression (median time to progression 8.7 weeks) [21]. These histology-specific findings would have been impossible to discern in larger, pooled analyses and underscore how the big data paradox can manifest in rare RCC variants.

Statement 3.

Basket, umbrella, or platform trials are appropriate when they increase relevance and efficiency for rare renal cell carcinomas, but histology-specific cohorts (e.g., RMC, ChRCC, pRCC) should be prospectively powered and analyzed as distinct groups (median score 9, IQR 2).

The panel endorsed a hybrid approach: shared trial infrastructure with segregated, prospectively powered cohorts for each biologically coherent subtype [22, 23]. Innovative multi-arm trial structures can substantially accelerate progress by improving operational efficiency, enabling adaptive decision-making, and reducing the need to launch multiple parallel single-cohort studies [24–26]. However, these efficiencies must not obscure the biological distinctiveness of rare RCC subtypes. Modern statistical approaches such as hierarchical modeling provide a framework to balance this trade-off, allowing information sharing across related cohorts without conflating efficacy estimates across biologically unrelated diseases [14, 27–29]. The phase II basket trial of nivolumab plus cabozantinib in non-clear cell RCC subtypes was cited as an example, in which ChRCC was intentionally isolated into its own Simon optimal two-stage cohort based on preliminary evidence of limited checkpoint inhibitor sensitivity. With no radiographic responses observed during the first phase, the ChRCC cohort closed to accrual early, while enrollment of other RCC variants could continue [17]. The panel noted that basket designs can group more than one rare RCC subtype together only when they share a clear biological target or actionable pathway, making the basket biologically coherent rather than merely histology driven.

Statement 4.

Papillary RCC is a non-clear cell RCC variant that is sufficiently prevalent to justify histology-specific RCTs (median score 9, IQR 2).

The group reached strong consensus that pRCC is the rare RCC with the strongest track record and feasibility for histology-specific RCTs based on disease prevalence, existing therapeutic benchmarks, and the availability of rational comparator arms [30, 31]. pRCC occupies a unique position among non-clear cell entities: it is sufficiently common to support multi-institutional accrual, it has established activity signals from prior trials, and there is a clear scientific foundation for comparative testing. This has enabled prospective exploration of disease-specific, mechanism-based strategies: The substantial historical and contemporary evidence base includes MET-inhibitor trials [30, 31], immunotherapy studies, and combination regimens [16, 17, 32, 33], which provide both precedent and practical anchors for powering assumptions. Among rare RCC subtypes, pRCC stands alone in reliably supporting randomized clinical trials. It is important to note that the former ‘type 2 papillary’ designation has been largely abandoned, as it encompasses a heterogeneous collection of molecularly distinct subtypes—including FH-deficient RCC and ALK-rearranged RCC—that are now classified separately under the 2022 WHO system [1, 5]. This distinction is critical, as grouping these biologically disparate diseases under a single ‘papillary’ label would replicate the same pooling problem identified in Statement 2, sacrificing relevance to individual molecularly defined populations in exchange for the robustness of a larger but biologically incoherent sample.

Statement 5.

More work is needed to forge collaborations, explore targets, and establish referral patterns for dedicated clinical trials in chromophobe RCC, a non-clear cell RCC variant that may be sufficiently prevalent to justify histology-specific RCTs in the future (median score 9, IQR 2 – revised statement).

The original statement (“Chromophobe RCC is a non-clear cell RCC variant that is sufficiently prevalent to justify histology-specific RCTs”) was the only statement that did not reach initial consensus in round 1 (median 7, IQR 2), reflecting significant uncertainty regarding the feasibility and appropriateness of randomized trials specifically for ChRCC. During the in-person discussion, multiple contributors raised the concern that ChRCC remains too rare in the metastatic setting to feasibly support traditional RCTs. Biologically, ChRCC appears largely unresponsive to immune checkpoint inhibitors, has limited activity with antiangiogenic tyrosine kinase inhibitor (TKI) therapies, and lacks a well-established standard of care that could serve as a meaningful control arm [32, 34–39].

Unlike papillary RCC, limited mechanistic insight into actionable molecular underpinnings of chromophobe biology has constrained progress toward histology-specific clinical trial design. Panel members nonetheless highlighted several emerging preclinical leads that warrant maturation into trial-ready strategies: promising examples include KIT (CD117)-directed antibody–drug conjugates, which exploit the characteristic KIT expression of chromophobe tumors [40]; pharmacologic induction of ferroptosis through FSP1 inhibition, leveraging a metabolic vulnerability identified in chromophobe models [41]; and immune-based approaches informed by the tumor-intrinsic and microenvironmental determinants of the characteristically “cold” chromophobe immune microenvironment, including the interleukin-15/cytotoxic innate lymphoid cell axis [42]. The panel viewed these as promising but still preliminary targets requiring further validation and early-phase clinical testing before histology-specific randomized trials could be justified. Patient advocates emphasized the urgency of generating evidence for ChRCC, with one advocate describing patients who feel like “sitting ducks” when told immunotherapy does not work but there are no alternatives. The revised statement removes the assumption that RCTs are feasible now and instead emphasizes the foundational scientific and infrastructural work needed to potentially enable them in the future. This formulation achieved strong consensus in the second voting round.

Statement 6.

For rarer RCC variants (e.g., RMC, FH-deficient RCC, collecting duct, MiT family/translocation RCC), we should strictly adopt innovative single-arm/adaptive designs with natural history/external controls, hierarchical borrowing, and explicit confirmatory strategies (median score 8, IQR 2).

For the rarest and most biologically aggressive RCC variants, traditional RCTs are often not feasible and, in many cases, not methodologically appropriate. Subtypes such as RMC, FH-deficient RCC, collecting duct carcinoma, and MiT family/translocation RCC represent extremely low-prevalence diseases where adequate sample sizes for randomization are unattainable even across large cooperative groups [43]. Patients often deteriorate before they can enroll in conventional trials, and many academic centers see only a handful of cases per year. The panel endorsed methodological strategies capable of generating high-quality evidence despite small sample sizes: single-arm trials with predefined benchmarks, adaptive designs allowing early stopping or expansion, hierarchical or Bayesian borrowing enabling cautious information sharing when biologically justified, external controls derived from natural history cohorts, and explicit confirmatory pathways such as seamless phase II/III designs [13, 14, 27–29, 44–49]. The critical value of natural history registries and informatics infrastructure was highlighted; robust, well-curated registries can provide meaningful denominators to define expected outcomes, substitute for lack of prospective concurrent controls and support external control arms when randomization is impossible [50–53].

Notably, several ultra-rare RCC variants—including RMC, FH-deficient RCC, and MiT family/translocation RCC—occur at clinically significant frequencies in adolescents where dedicated pediatric trials are impractical due to extremely limited patient numbers [54]. The FDA’s guidance on “Considerations for the Inclusion of Adolescent Patients in Adult Oncology Clinical Trials” recommends enrolling patients aged 12 and older in adult oncology trials when the histology and biologic behavior of the cancer are shared across age groups [55]. Incorporating adolescent eligibility into adult rare RCC trials is therefore both scientifically justified and operationally essential, as it provides these patients earlier access to investigational therapies while simultaneously accelerating accrual in disease histologies that suffer from very infrequent diagnoses [56].

Registry-based external controls will only be credible if the underlying data are sufficiently rigorous, as regulators are unlikely to accept heterogeneous, unstandardized retrospective datasets [57]. At a minimum, outcomes should be assessed using standardized, blinded procedures, ideally with independent central RECIST 1.1 review and harmonized definitions of progression and censoring [58–61]. A prospectively specified common data model should capture the key prognostic variables needed for adjustment, including performance status, sites of disease, and centrally confirmed molecular subtype [62]. Eligibility criteria should closely mirror those of the trial, with a clearly defined treatment-initiation “time zero” consistent with target-trial emulation, and controls should come from the same treatment era [49, 63–65]. The statistical analysis should be prespecified in a finalized protocol and statistical analysis plan and should incorporate propensity-score, outcome-model, or other covariate-adjustment methods to address measured confounding, together with quantitative bias or sensitivity analyses to assess the potential impact of residual unmeasured confounding [48, 57, 66, 67]. These safeguards can be complemented by dynamic borrowing approaches, such as robust meta-analytic-predictive priors, commensurate priors, or dynamic/normalized power priors, which adaptively reduce the influence of external data when they are inconsistent with concurrently enrolled patients [27–29, 47, 48, 68–70]. Because this level of harmonization is resource-intensive, the required rigor should reflect how heavily the external data will be used: a preliminary benchmark for an early single-arm study may justify a less intensive approach than an external control intended to support registration. Ideally, these expectations should be agreed upon prospectively with regulators.

Statement 7.

Central pathology review should be incorporated into all rare RCC trials (median score 9, IQR 2).

Given the biological heterogeneity of rare RCC subtypes and the complexity of histopathological classification, the panel unanimously agreed that central pathology review is essential for all rare RCC trials. Accurate histological classification directly impacts eligibility determination, endpoint interpretation, and the validity of subtype-specific conclusions [1, 2, 71]. Misclassification can dilute treatment effects, introduce noise into efficacy analyses, and undermine the scientific value of histology-specific investigations. Central review by expert genitourinary pathologists ensures consistent application of current WHO classification criteria and molecular definitions [5].

Perioperative Trials

Statement 8.

Perioperative (neoadjuvant/adjuvant) trials in rare RCCs should be pursued when supported by strong biologic rationale and feasible endpoints. Pooled ‘non-clear cell’ approaches with secondary subgroup analyses should be avoided (median score 8, IQR 2).

The panel agreed that perioperative clinical trials are both important and appropriate in rare RCC subtypes when guided by compelling biologic rationale and when meaningful endpoints can be identified. The fundamental challenge is that as of 2025, no therapy has demonstrated sufficient benefit to justify routine perioperative use in rare RCC subtypes. Rare subtype perioperative trials should not simply mirror strategies proven effective in ccRCC; rather, they must be grounded in subtype-specific biology, pathways, and vulnerabilities. Patient advocates expressed interest in adjuvant trials for rare RCC subtypes, underscoring that patients, frustrated by the lack of data-driven guidance, are motivated and willing to participate provided the scientific justification is sound. The group concluded that perioperative strategies will likely become more relevant as therapies become more mechanistically tailored, but pooled non-clear cell RCC approaches should be avoided.

Statement 9.

The preferred design for perioperative trials in rare RCCs is a randomized controlled trial (placebo/observation control), with single-arm designs reserved only for settings hypothesizing large, durable effects and supported by robust natural history comparators (median score 8, IQR 2.75).

The panel agreed in principle that RCTs are preferred for perioperative research, as they minimize bias and enable clearest interpretation of perioperative endpoints such as event-free survival, recurrence-free survival, or pathologic response [8, 9]. However, the wider IQR for this statement reflected reservations. Some respondents expressed concern that the nuances of different subtypes make it difficult to create an overarching recommendation. For many rare RCC variants, there may be no biological or pathological rationale to support perioperative therapy, particularly in the neoadjuvant setting [72]. Without such rationale, even a well-designed RCT may be uninformative. Single-arm perioperative trials should be rare but may be appropriate when a very large, durable effect is biologically plausible and high-quality natural history datasets exist for comparison.

Statement 10.

For adjuvant papillary RCC randomized trials, the control arm should be observation and not pembrolizumab (median score 9, IQR 3).

Consensus emerged that observation, rather than pembrolizumab, should serve as the control arm for any randomized adjuvant pRCC trial, although the wider dispersion of scores (IQR 3) indicates this view was not unanimous. There is no biological, clinical, or trial evidence that pembrolizumab provides benefit in pRCC in the adjuvant setting [73, 74]. The adjuvant KEYNOTE-564 trial of pembrolizumab leading to regulatory approval was conducted in ccRCC [75], and its results cannot be reliably extrapolated to pRCC—a disease driven by markedly different genetic and oncogenic pathways. Because no adjuvant therapy has demonstrated benefit in pRCC, observation remains the standard of care, and using pembrolizumab as a control arm would create therapeutic noise and undermine interpretability.

Statement 11.

For adjuvant non-papillary rare RCC randomized trials, the control arm should be observation and not pembrolizumab (median score 9, IQR 2).

This statement extends the logic of Statement 10 to all non-papillary rare RCC subtypes (e.g., chromophobe, MiT family/translocation, collecting duct carcinoma, FH-deficient RCC). Rare RCC subtypes do not share the immune biology, antigenicity, or tumor microenvironment features that appear to underlie the benefit of pembrolizumab in clear cell RCC [2, 18, 34, 35, 71, 76–78] and accordingly were not included in KEYNOTE-564 [75]. For many subtypes, available biological or clinical evidence suggests low expected responsiveness to adjuvant PD-1 blockade. Using pembrolizumab as a control arm would be biologically unjustified and potentially misleading for statistical interpretation. The narrower IQR compared to Statement 10 indicates even tighter agreement, likely reflecting that in the non-papillary setting there is even less biological plausibility for pembrolizumab efficacy.

Trial Operations and Accrual

Statement 12.

Collaboration with patient advocacy groups and use of consented natural history registries should be pursued to accelerate identification of eligible patients and increase trial access (median score 10, IQR 1).

This statement achieved near-unanimous consensus, reflecting universal agreement that patient advocacy organizations and natural history registries are essential infrastructure for advancing rare RCC trials. Traditional referral pathways and single-center accrual models are insufficient for ultra-rare histologies, where each institution may see only a few patients per year. Advocacy networks often serve as the first point of contact for newly diagnosed patients, enabling rapid dissemination of trial opportunities and education about eligibility criteria. By establishing disease denominators and tracking patients through predictable clinical transitions, registries can enable earlier identification of trial-eligible patients, reducing the interval between disease progression and enrollment. Integration with molecular data further enhances the utility of registries by enabling eligibility pre-screening and providing external control cohorts for single-arm and Bayesian-borrowing designs [27, 50–53]. When registries are used as external-control cohorts rather than simply to identify eligible patients, the harmonization requirements outlined in Statement 6 become critical. In particular, standardized blinded RECIST 1.1 assessment and complete capture of the main prognostic covariates are necessary for the data to be considered acceptable for regulatory purposes [57].

Statement 13.

Industry collaboration should be actively sought to meet accrual targets and timelines that are otherwise impractical for rare RCC (median score 10, IQR 1).

Strategic industry collaboration should be actively sought whenever possible for rare RCC research. Given the limited patient numbers, trials often require multicenter, international, and/or multi-sponsor collaboration to succeed. Recent successful models were discussed, including larger single-arm studies such as KEYNOTE-427 and KEYNOTE-B61 [16, 79], and randomized trials such as SAVOIR [31]. These examples demonstrate that industry engagement can enable feasible timelines through adequate global site activation, provide access to investigational agents, integrate unbiased early-signal stop/go decision making, and deliver high-quality correlative science. Early co-design with industry is beneficial, as aligned incentives and coordinated recruitment strategies can help address the significant feasibility challenges often encountered in rare RCC trials and ultimately would serve to pave a feasible registration pathway, ideally with integration of companion diagnostics.

Statement 14.

Rare RCC protocols should include adaptive fallback options (e.g., broaden eligibility criteria) to avoid study closure if real-world accrual underperforms projections (median score 9, IQR 2).

Trial fragility is a defining operational challenge in rare RCC, where narrowly defined entry criteria are often scientifically justified yet automatically limit enrollment and increase risk of premature closure, lost investment, and lost scientific opportunity. Proactive, pre-specified fallback mechanisms can sustain trial viability while preserving scientific integrity even when an ambitious original accrual strategy proves infeasible or when treatment landscapes change with approval of other agents. The COSMIC-021 RCC cohort (cabozantinib plus atezolizumab in RCC) was cited as an example where prior therapy was allowed for the non-clear cell RCC expansion cohort but not the clear cell cohort, recognizing added accrual challenges in the non-clear cell population, thus improving feasibility without compromising scientific objectives [80]. The multi-arm randomized phase II PAPMET study incorporated a flexible design that allowed closure of some of its arms early for lack of efficacy: while more specific MET-targeting arms (crizotinib, savolitinib) were discontinued due to futility, the study continued with cabozantinib (which inhibits MET but also possesses a broader receptor TKI spectrum) compared to the control therapy of sunitinib, demonstrating how platform structures can absorb arm-level failure without premature termination [30].

Statement 15.

Travel support and digital/remote trial operations (e.g., remote screening, eConsent, tele-PROs, decentralized specimen collection) are specifically important for rare RCC trials to reduce travel and access barriers (median score 9.5, IQR 1).

Decentralized and remote trial operations emerged as a strongly endorsed strategy for improving feasibility in rare RCC trials. Travel burden, logistical constraints, and limited access to specialized centers disproportionately hinder enrollment for patients with rare RCC subtypes [81–86]. Patients with rare RCC histologies have few trial options, and those that exist tend to be smaller with limited geographic reach, placing the burden on patients to locate and travel to participating sites. Nevertheless, motivated patients and caregivers frequently accept substantial logistical and financial burdens to pursue investigational options, weighing hope against uncertainty. This raises both operational and ethical considerations for referring oncologists and clinical study teams. Approaches such as remote screening, eConsent, tele-PROs, virtual safety assessments, and decentralized specimen collection were viewed as essential tools to expand access and reduce inequities in trial participation [85, 87–91]. Decentralized trial workflows can also be combined with newer statistical approaches to maintain rigor when conventional trial designs are not feasible. In this context, external control cohorts would be drawn from structured, quality-controlled real-world data, primarily electronic health records and disease registries collected through decentralized or remote systems [48, 92–95]. When these data are curated using standardized, harmonized elements, as outlined under Statement 6, and appropriately adjusted for differences from the prospectively enrolled trial population, they may supplement or, in selected settings, replace a concurrent control group when randomization is not feasible. Once patients are identified, consented, and screened, travel reimbursement and housing support are essential to minimize financial hardship and to ensure treatment continuity for those who derive benefit.

Biology-Driven and Histology/Molecular Strategy Trials

Statement 16.

While histology-agnostic labels exist in RCC, rare subtype trials should still pursue histology-specific evidence to refine indications and guide subtype-tailored standards, rather than over-extrapolating from clear cell populations (median score 9, IQR 2).

The panel reaffirmed that rare subtype trials must generate histology-specific and molecular evidence to refine indications and standards, rather than extrapolating from clear cell RCC. To prevent the distinct therapeutic responses of rare variants from being lost in aggregate data, trials that recruit broader RCC populations should incorporate histology-specific subgroup assessments as a standard component of the study design. Participants additionally highlighted the need to harmonize therapeutic with diagnostic advances and move toward molecularly classified renal tumors when choosing target populations for drug development, as traditional pathology labels are no longer adequate and current diagnostic standards increasingly put emphasis on integrating molecular features [1, 5]. While tumor-agnostic approaches are possible when strong molecular drivers are shared across cancers, examples were given where inadequate molecular awareness led to missed enrollment of RCC patients in pan-cancer trials, underscoring the need for histology-specific and biology-specific frameworks in parallel. The NF2/Hippo/YAP-TEAD axis exemplifies how lack of awareness in trial design can lead to missed opportunities for rare RCC patients. NF2 loss occurs at clinically significant frequencies across several aggressive RCC subtypes, including pRCC, collecting duct RCC, unclassified RCC, and RMC [77, 96–98], yet the first-in-human YAP/TEAD inhibitor trials (VT3989, IK-930, IAG933) did not provide subtype-specific data in these populations due to limited enrollment of patients with RCC [99–101]. Mechanism-defined, histology-agnostic trials typically operate within early drug development programs that can be geographically and institutionally separate from the genitourinary oncology centers where rare RCC patients receive care. Closing this operational gap requires deliberate collaboration between RCC specialists and first-in-human (phase 1) investigators.

Statement 17.

For molecularly targeted agents with tumor-agnostic potential (e.g., SMARCB1 deficiency or NF2 loss), RCC-inclusive cohorts should follow methodological standards used in successful tissue-agnostic programs (robust ORR, durability, safety), with RCC-specific follow-up when signals emerge (median score 9, IQR 2).

RCC-inclusive cohorts within tumor-agnostic development programs should adopt the same methodological rigor that has enabled successful tissue-agnostic approvals. Examples such as the successful development of belzutifan in von Hippel-Lindau disease demonstrate how a molecularly targeted agent can show consistent activity across multiple organ systems when eligibility is anchored in a shared pathogenic mechanism rather than histology [102–104]. Basket trial frameworks such as VE-BASKET for BRAF V600E-mutant tumors demonstrate that robust efficacy signals within individual histologic cohorts, even if limited in size, can support tumor type-specific regulatory approvals within a molecularly unified, robustly designed study, as exemplified by the accelerated approval of vemurafenib for Erdheim-Chester disease based on basket trial data [105, 106]. RCC-specific validation is therefore required when signals emerge, because pharmacodynamics, resistance biology, immune microenvironment, and natural history may differ from other tissues even when driven by the same molecular alteration.

Statement 18.

For dedicated, mechanism-directed trials in WHO-recognized, molecularly-defined RCC populations (e.g., FH-deficient RCC), a “one-trial” paradigm should be pursued—explicitly planned a priori and consisting of an early ORR signal for accelerated approval paired with a built-in randomized confirmation component (median score 8, IQR 2).

Discussion emphasized both the promise and practical challenges of a “one-trial” paradigm for molecularly-defined RCC populations, in which a single adequately powered trial with robust efficacy data may suffice for FDA approval rather than the traditional requirement for two adequate and well-controlled studies. This approach has precedent in rare cancers—for example, the approval of larotrectinib for tropomyosin receptor kinase (TRK) fusion-positive solid tumors was based primarily on a single basket trial demonstrating a 75% overall response rate across histologies [107]. A prospectively planned design that includes an early ORR-based accelerated-approval component paired with a confirmation phase is conceptually attractive and consistent with regulatory trends [108–111]. However, several experts raised feasibility concerns, noting that a built-in randomized confirmation stage may be unrealistic for many —if not most— RCC molecular subsets, where patient numbers are extremely limited and no meaningful control arm may exist. The belzutifan development program in VHL disease was cited as an example where randomization was never feasible or plausible, and where the agent moved forward successfully through a non-randomized mechanism-driven strategy [102, 103]. The group supported the principle but acknowledged that in the rarest molecular entities, randomized confirmation may not be achievable, requiring flexibility in how the paradigm is operationalized.

Preclinical Efforts, Target Identification, and Early Signal Testing

Statement 19.

The generation of disease-specific murine models (including syngeneic models and organoids) should be a priority to the field since these are of central importance for biology-directed target discovery, trial design, and justification (median score 9, IQR 3).

While there are efforts nationally to reduce the unnecessary use of mice, murine models have enabled significant advances in our understanding of RCC and have proven faithful for drug development. For example, tumorgraft models of ccRCC provided the first evidence that HIF2 inhibitors would have activity against ccRCC in humans [104, 112]. In addition, they identified mechanisms of resistance, which were subsequently ratified in patients [112, 113]. However, it is important that models be appropriately selected, which depends on the particular application they will be used for. Models should be shown to faithfully reproduce the process that is being evaluated so that robust inferences can be made. While murine systems remain indispensable for functional exploration and commonly lay the groundwork for mechanism-based strategies, several experts cautioned that their limitations must be explicitly acknowledged. Murine models do not always reflect the relevant features of human kidney cancer biology; for example, somatic models of ccRCC do not respond to belzutifan, despite its proven activity in human VHL-associated tumors [114–116]. Mouse systems excel where causal manipulation is required—for pathways such as SMARCB1, murine and GEMM systems permit on/off switching, enabling mechanistic validation that human data can only correlate [18, 117]. Mice also provide the whole-organism context needed for pharmacokinetic (PK) and pharmacodynamic (PD) studies, earliest determination of dosing strategies, immune interactions, stromal cues, and schedule dependence. In this context, it is important that drug studies in mice consider differences in drug metabolism and that drug regimens are adjusted based on murine PK data so as to reproduce human exposures [118]. However, mouse immunity, stromal cues, metabolism, and microbiome differences can mispredict human responses [119–122]. Emerging alternative platforms including microphysiological systems (kidney-on-chip, tumor-immune-on-chip), co-culture ecosystems, and digital twins/quantitative systems pharmacology were highlighted as essential complements to improve mechanistic inference [123–129].

These preclinical tools also address a limitation inherent to trials themselves. A randomized trial can establish whether a treatment works, and how well, but not why [130, 131]—a gap that becomes critical when predicting whether a result will hold in patients whose biology, comorbidities, or tumor microenvironment differ from those enrolled [8, 132, 133]. Because several distinct mechanisms can explain the same trial data equally well (the Rashomon effect), a positive result rarely reveals which one is responsible [134–136]. Disease-specific murine models and organoids provide the means to test these mechanisms directly. Without that grounding, judgments about generalizability rest on assumption rather than evidence [73], and rational follow-on therapies become far harder to design. In rare RCC, these models are therefore as important for interpreting and extending trials as they are for identifying targets in the first place.

Statement 20.

Mechanism-informed testing of novel agents using single-patient IND, particularly if based on preclinical data and paired with on-treatment pharmacodynamic biomarker acquisition, can be a helpful initial step in signal-testing prior to pursuit of histology/biology-specific trials in rare-variant RCC (median score 8, IQR 3.75).

Single-patient INDs (SPINDs) occupy a pragmatic niche for early signal generation in rare-variant RCC when promising mechanistic hypotheses exist but sponsor hesitancy, small populations, and operational barriers limit activation of formal clinical trials. Notably, this statement generated the widest IQR among the panel, reflecting a divide regarding the risks of ad hoc SPINDs, which often lack dedicated institutional infrastructure, strain regulatory resources due to real-time activation demands, and yield only single case-level outcomes. Nonetheless, SPINDS are low risk for sponsors hesitant to explore orphan indications despite compelling preclinical data. They provide patients access to novel compounds and offer investigators flexibility to test hypotheses on short timelines. One example was the evaluation of a glycolysis inhibitor in a patient with a germline mutation in the FH gene and metastatic FH-deficient RCC [137]. The study was supported by a strong rationale and preclinical data showing that FH-deficient RCC cells rely on glycolysis for ATP generation, and the single patient IND enabled adaptive changes in the treatment approach that led to successful pathway inhibition. When paired with in-depth biomarker assessments, SPINDs can generate invaluable preliminary data for securing third-party funding and industry partnerships, ultimately supporting the launch of formal clinical trials. The panel reached consensus that SPINDs are most valuable when programmatic—pursued deliberately in partnership with sponsors and laboratory collaborators rather than as ad hoc efforts—and when accompanied by paired pre- and post-treatment tissue allowing pharmacodynamic confirmation of mechanism engagement.

The panel did not endorse a specific governance structure. One possible model would place these programs under a disease-focused organization, such as a rare-variant RCC consortium or a patient advocacy group like the Kidney Cancer Association, working with academic and industry partners. Rather than relying on repeated one-off submissions, single-patient INDs could be managed through an intermediate-size expanded access protocol under 21 CFR 312.315, led by a coordinating sponsor-investigator and overseen by a central IRB. A multidisciplinary committee could review the mechanistic rationale for each case, require paired pre- and post-treatment biospecimens, and define criteria for advancing promising signals into formal clinical trials. This approach would be consistent with existing FDA expanded access pathways and the emerging Plausible Mechanism Pathway (PMP) [138]. As per its current description, in the absence of a draft guidance, the PMP is designed for diseases where the biologic cause is known, the product targets the underlying molecular alteration, there is a well-characterized natural history, and clinical data demonstrate improvement in outcomes. For ultra-rare RCC subtypes, the PMP may eventually permit drugs and biologics to obtain marketing authorization in the US through a phased operational model beginning with consecutive patients treated with bespoke therapies, using patients as their own controls when supported by robust natural history data, thus formalizing the “programmatic SPIND” approach endorsed by the panel.

4. Discussion

This Delphi-based consensus is the first to focus specifically on clinical trial design and methodology for rare RCC subtypes. Given the persistent lack of evidence-based therapies and the ongoing challenges in conducting trials for these diseases, the panel’s 20 endorsed statements provide a practical framework to guide future research, regulatory interactions, and clinical practice. Key recommendations include several foundational principles. First, the field should abandon pooled “non-clear cell” trial designs in favor of histology-specific approaches focusing ever more on molecular drivers that respect the biological heterogeneity of rare RCC subtypes [1, 2, 5]. Second, randomized controlled trials should remain the preferred design when feasible, particularly for papillary RCC, but innovative single-arm and adaptive designs are essential for ultra-rare variants where traditional RCTs are impossible [13, 14, 22, 24–29]. Third, patient advocacy collaborations, natural history registries, and industry partnerships are not optional but indispensable infrastructure for rare RCC research. Fourth, decentralized trial operations and adaptive fallback mechanisms are critical to maintaining trial viability in the face of accrual challenges [48, 81–85, 87–95]. Fifth, mechanism-informed therapeutic development, anchored in robust preclinical science and potentially initiated through single-patient IND programs, can accelerate progress while maintaining scientific rigor.

The relevance-robustness trade-off emerged as a unifying conceptual framework throughout the discussion (Figure 2 and Table 2) [8, 12, 14]. This trade-off is not unique to RCC but is a shared struggle across the global orphan disease landscape, where small patient populations routinely force researchers to choose between statistical stability (robustness) and biological specificity (relevance) [8, 14]. RCTs achieve high robustness by remaining valid under a wide range of assumptions, but their inference targets a broad, heterogeneous population that may not reflect the specific biology, clinical context, or treatment-effect modifiers of an individual patient with a rare RCC subtype. Conversely, single-arm studies with external controls can be designed to maximize relevance by conditioning on histology, molecular features, and clinical characteristics—but this relevance depends on assumptions (e.g., comparability of external controls, absence of unmeasured confounding) that may not hold. The panel emphasized that navigating this trade-off requires explicit acknowledgment of the assumptions underlying each design choice and transparent reporting of limitations.

The only statement that did not achieve initial consensus concerned ChRCC, highlighting genuine uncertainty in the field about the feasibility of histology-specific RCTs for this subtype. The revised statement, which achieved strong consensus, reflects a more nuanced position retaining RCTs as the ultimate goal while acknowledging that infrastructure and collaboration must be strengthened before chromophobe-specific RCTs can be realistically pursued. This example illustrates the value of the Delphi process in surfacing disagreement and facilitating constructive revision.

Several limitations should be acknowledged. First, all 20 statements were drafted a priori by the two co-chairs (P.M. and M.V.) and the KCA Chief Scientific Officer (S.L.R.) based on a literature review. Because the initial framing and wording of statements necessarily reflects the perspectives and priorities of this small drafting group—the two co-chairs, both of whom also report industry relationships (see Declaration of Competing Interest and Conflict of Interest Management)—this process may have introduced framing bias, constraining the scope of issues considered and the direction of consensus. Several features were intended to mitigate but ultimately cannot eliminate this risk: statements were open to revision by the full 46-member multidisciplinary panel across rounds, scoring was anonymous to limit dominance and hierarchy effects, the in-person round allowed open challenge and revision (as occurred for the chromophobe RCC statement), and a post hoc sensitivity analysis confirmed that consensus classifications were unchanged when the two industry-employed panelists were excluded (Supplementary Table 1). Second, participation declined across rounds: 46 of 59 invited experts (78%) completed the initial survey, and 38 of 46 (83%) completed the round-3 revote. Because only the revised chromophobe RCC statement was re-voted in round 3, this later attrition affects the strength of consensus for that single statement, whereas the remaining 19 statements reflect scoring by the full round-1 panel; nonetheless, we cannot exclude that non-responders differed systematically from responders in ways that could affect consensus estimates. Third, the Delphi process inherently relies on expert opinion, which may be influenced by individual experience and institutional context. Several statements met the consensus threshold despite meaningful dispersion in individual scores (e.g., Statements 1, 9, 10, and 20). For these, the median-based threshold reflects central tendency rather than uniform agreement, and the corresponding recommendations should be regarded as provisional and weighted accordingly. Finally, the rapidly evolving landscape of clinical trial methodology means that some recommendations may require updating as new evidence and regulatory guidance emerge.

5. Conclusion

The 2025 IKCS Think Tank provides a practical framework for the next generation of rare renal cell carcinoma research, marking a necessary departure from the historical reliance on heterogeneous “non-clear cell” pooling. By prioritizing histology-specific cohorts and biology-driven trial targets, the oncology community can move toward a model of drug development that is suitable for distinct molecular landscapes of variants such as pRCC, ChRCC, MiT family/translocation RCC, and RMC. The integration of innovative statistical tools—including Bayesian hierarchical modeling and the use of external natural history controls—offers a viable pathway to generate high-quality evidence without the prohibitive sample size requirements of traditional ccRCC trials. Furthermore, the emphasis on decentralized operations and programmatic SPINDs ensures that research remains both equitable and responsive to early mechanistic signals. Ultimately, these recommendations underscore that clinical progress in rare RCC depends on achieving a causal understanding that transcends mere descriptive trial results. By implementing this consensus-driven approach, stakeholders can bridge the gap between aggregate study data and the transportable, personalized insights required to treat individual patients in the clinic.