Comprehensive Genomic Analysis of Metastatic Non–Clear-Cell Renal Cell Carcinoma to Identify Therapeutic Targets

Abstract

PURPOSE

Non–clear-cell renal cell carcinoma (nccRCC) encompasses approximately 20% of renal cell carcinomas and includes subtypes that vary in clinical and molecular biology. Compared with clear cell renal cell carcinoma, nccRCC demonstrates limited sensitivity to conventional vascular endothelial growth factor– and mammalian target of rapamycin–directed agents, indicating a need for better therapies. Characterizing the genomic landscape of metastatic nccRCC variants may help define novel therapeutic strategies.

PATIENTS AND METHODS

We retrospectively analyzed tumor tissue from patients with metastatic nccRCC who consented to genomic analysis of their tumor and germline DNA. A hybridization capture–based assay was used to identify single nucleotide variants and small insertions and deletions across more than 340 cancer-associated genes with germline comparison. Clinical actionability of somatic mutations was assessed using OncoKB levels of evidence. Microsatellite instability (MSI) in the tumor was investigated.

RESULTS

Of 116 patients included in the analysis, 57 (49%) presented with de novo metastatic disease, and 59 (51%) presented with localized disease that later metastasized. Subtype classifications included unclassified (n = 41; 35%), papillary (n = 26; 22%), chromophobe (n = 17; 15%), translocation associated (n = 13; 11%), and other (n = 19; 16%). Of all tumors, 15 (13%) had putative driver somatic alterations amenable to targeted therapies, including alterations in MET, TSC1/2, and an ALK translocation. Of 45 patients who had germline testing, 11 (24%) harbored mutations, seven of which could potentially guide therapy. Of 115 available tumors for analysis, two (1.7%) had high and six (5%) had intermediate MSI status.

CONCLUSION

The mutation profiles of metastatic nccRCC vary by subtype. Comprehensive analysis of somatic mutations, germline mutations, and MSI, interpreted via an annotated precision oncology knowledge base, identified potentially targetable alterations in 22% of patients, which merits additional investigation.

CONTEXT

• Key Objective

• What is the frequency of clinically actionable mutations in patients with advanced non–clear-cell renal cell carcinoma (nccRCC)?

• Knowledge Generated

• Somatic mutation profiles vary by nccRCC subtype. Sequencing the tumor and germline and tumor microsatellite instability analysis reveal clinically actionable mutations in 22% of patients with advanced disease.

• Relevance

• Metastatic nccRCC can have a poor prognosis, and treatment is often extrapolated from clear cell renal cell carcinoma as a result of few studies that include this rare group of cancers. Routine sequencing of nccRCC yields alterations such as high microsatellite instability, which may affect clinical management.

INTRODUCTION

There are 63,000 estimated new patients diagnosed with renal cell carcinoma (RCC) annually in the United States and 14,000 deaths per year from advanced disease.¹ The most common subtype of RCC is clear cell RCC (ccRCC), which makes up approximately 75% to 80% of all RCCs, with defined roles for targeted systemic therapy in localized and metastatic disease.² The remaining histologic variants are collectively termed non–clear-cell RCC (nccRCC) and include papillary, chromophobe, collecting duct, translocation-associated, and unclassified types.³ Comparing outcomes to ccRCC, cancer-specific survival is more favorable in patients with nccRCC with localized disease but tends to be worse in the metastatic setting.^4,5

Better treatments for metastatic nccRCC are clearly needed. Dedicated data are sparse, and therapeutic strategies are largely extrapolated from clinical trials conducted in ccRCC. A limited number of phase II randomized trials have specifically studied nccRCC. The ESPN (Everolimus Versus Sunitinib Prospective Evaluation in Metastatic Non–Clear-Cell Renal Cell Carcinoma) and ASPEN (Everolimus Versus Sunitinib for Patients With Metastatic Non–Clear-Cell Renal Cell Carcinoma) trials compared the anti–vascular endothelial growth factor (VEGF) agent sunitinib to the mammalian target of rapamycin (mTOR) inhibitor everolimus as first-line treatment of patients with nccRCC.^6,7 Per primary end point analyses, everolimus was not superior to sunitinib, but the heterogeneity of the study populations and the small sizes of variant subgroups limit the applicability of these data. These studies have highlighted the need for novel therapeutic approaches and a better understanding of how RCC biology differs among subtypes.

Increased knowledge of genomics may help identify novel treatment strategies. Patients have shown marked responses to several targeted therapies when selected by molecular tumor features, rather than by tumor type.^8,9 Similarly, tumor microsatellite instability (MSI) status is predictive of response to the anti–programmed cell death protein 1 inhibitor pembrolizumab.¹⁰ Indeed, MSI status and defective mismatch repair are the first molecular features linked to a regulatory agency approval irrespective of underlying disease histology. Although The Cancer Genome Atlas (TCGA) and others have looked at the genomic landscape of some nccRCC entities, these studies have focused on mostly patients with nonmetastatic disease, and more advanced metastatic tumors arguably may harbor different genomics.^11,12 In this study, we determined whether tumor and germline genomic profiling was informative for patients with advanced nccRCC and defined the frequency of actionable alterations across the different subtypes.

PATIENTS AND METHODS

Patients

Patients were identified from an institutional database of patients with RCC who underwent next-generation sequencing (NGS) of archival tumor tissue under an institutional protocol. All patients with metastatic nccRCC who consented to tumor sequencing at Memorial Sloan Kettering Cancer Center (MSKCC) between April 2014 and January 2017 were included. Starting in October 2015, patients could also consent to additionally obtain the results of germline sequencing. Demographics, clinical characteristics, and tumor profiling results were collected via retrospective review of clinical records. This study was approved by the MSKCC Institutional Review Board.

NGS Analysis

Tumor samples from either primary or metastatic sites were reviewed by genitourinary pathologists and microdissected for maximum tumor content for DNA extraction. Blood samples were obtained as matched normal controls. Tumor and blood samples were sequenced using MSK-IMPACT, an NGS assay developed at MSKCC that achieves hybridization capture with target-specific probes from exons of at least 341 cancer-associated genes, as described previously.¹³ Mean depth of sequencing for the cohort was 660×. The gene panel is available in Appendix Table A1. MSI was quantified using MSIsensor.^14,15

Germline variants for 76 genes associated with cancer predisposition included in MSK-IMPACT were analyzed (Appendix Table A1). Variants with less than 1% frequency in the public Exome Aggregation Consortium database were interpreted. Variants were classified by a clinical molecular geneticist or molecular pathologist as pathogenic, likely pathogenic, of uncertain significance, likely benign, or benign, according to American College of Medical Genetics criteria.¹⁶ Only pathogenic or likely pathogenic variants (associated with disease causation and henceforth referred to as pathogenic variants) are included in this analysis.

Somatic Mutation Clinical Annotation

Somatic alterations were annotated through OncoKB (http://oncokb.org), a curated precision oncology knowledge base that stores information about the biologic function and therapeutic implications of individual gene alterations in a tumor type–specific manner. Gene alterations that act as predictive biomarkers of response to specific targeted therapies are assigned a specific OncoKB level of evidence.¹⁷ Levels 1 and 2A are US Food and Drug Administration (FDA)–recognized or National Comprehensive Cancer Network (NCCN)–listed biomarkers predictive of response to an FDA-approved targeted therapeutic. MSI-high (MSI-H) status in solid cancers is considered OncoKB level 1, based on the FDA approval of pembrolizumab.¹⁰ Level 2B is a standard of care biomarker in another indication (but not in RCC). Level 3A is compelling clinical evidence supporting the mutation acting as a predictive biomarker in the specific indication, on the basis of data from well-powered clinical trials or case series; if these data are in another tumor type (not RCC), the mutation is considered level 3B. We defined mutations with an assignment of level 1 to 3B as potentially clinically actionable. Mutations with level 4 assignment are those that are considered hypothetical biomarkers for which there is compelling biologic evidence that supports their being predictive of response to a drug and hence their potential use as eligibility criteria for clinical trials.

Statistical Analysis

Clinical characteristics, somatic and germline results, and MSIsensor score are presented using descriptive statistics. Objective response rate was defined as the best response according to Response Evaluation Criteria in Solid Tumors (RECIST), version 1.1. Overall survival, defined as the time from diagnosis of metastatic disease to death or last known follow-up, was estimated using the Kaplan-Meier method. Statistical analyses were performed using R version 3.5.

RESULTS

Patient Characteristics

A total of 116 patients with metastatic nccRCC were included in this cohort; demographics are listed in Table 1. Median age was 53 years (range, 12 to 77 years), and 67% of patients were male. Sixty-five percent of patients self-identified as white, 21% as black, and 14% as other. Fifty percent presented with de novo metastatic disease. The most frequent subtypes were unclassified (n = 41; 35%), papillary (n = 26; 22%), chromophobe (n = 17; 15%), translocation associated (n = 13; 11%), and other (n = 19; 16%; Fig 1A). Patients with unclassified and papillary RCC had similar characteristics: most patients were male (78% and 89%, respectively) and were diagnosed initially with localized disease, followed by subsequent recurrence. In comparison, most patients with chromophobe and translocation-associated RCC were female (53% and 69%, respectively). The median overall survival time for the cohort from time of diagnosis of metastatic disease was 34 months (95% CI, 24 to 59 months), with a median follow-up of 18 months.

TABLE 1.

Patient Characteristics and Histologic Variants of Interest

FIG 1.

Subtypes of renal cell cancers, and oncogenic somatic mutations (Mt) with OncoKB level of evidence. (A) Subtypes of non–clear-cell renal cell carcinoma in the cohort. (B) Percentages of tumors with somatic alterations by OncoKB level of evidence, of tumors with oncogenic drivers without known potentially actionable alterations, and of tumors with no known driver. (C) Number of patients with specific alterations and stratification by OncoKB level of evidence. MSI, microsatellite instability.

Somatic Mutation Frequencies and Clinically Actionable Somatic Alterations

Most patients (57%) had sequencing of the primary renal tumor. Mutation frequencies across the various nccRCC subtypes included in our analysis are outlined in Figure 2 and Appendix Table A2. Findings varied by subtype. In unclassified tumors, the most frequent mutations were in NF2 (n = 12; 24%) and SETD2 (n = 9; 18%). In comparison, in papillary tumors, the most frequent mutations were TERT promoter mutations (n = 12; 50%), followed by alterations in MET (n = 9; 38%) and NF2 (n = 2; 17%). Chromophobe tumors had a distinct profile, with a high frequency of mutations in TP53 (n = 10; 63%). Translocation-associated tumors all had TFE3 translocations confirmed by fluorescence in situ hybridization or NGS but had a paucity of other somatic alterations.

FIG 2.

OncoPrint of the most frequently mutated genes overall and microsatellite instability (MSI) in tumors of 116 patients with advanced non–clear-cell renal cell carcinoma (RCC). Cancer subtype is indicated on the top line. MSI status is indicated in the next line. Genetic alteration type, including germline pathogenic mutations, is indicated at the bottom of the OncoPrint.

We classified somatic alterations as potentially actionable per definitions previously established through OncoKB.¹⁷ As summarized in Table 2, tumors of 15 patients (13%) harbored an actionable, putative driver somatic mutation assigned an OncoKB level of evidence of 2A, 2B, 3A, or 3B (Fig 1 and Appendix Table A3). Ten patients had mTOR pathway alterations, including in PIK3CA (level 3; predictive of response to phosphatidylinositol 3-kinase inhibitors), TSC1/2 (level 2B; predictive of response to NCCN-listed everolimus), and MTOR (hypothetical predictor of response to mTOR inhibitors).¹⁸ Of these, six patients received an mTOR inhibitor alone or in combination with bevacizumab or lenvatinib (Table 3).¹⁹ An additional patient with a PTEN mutation (level 4) was treated with bevacizumab and everolimus. Median duration on treatment with an mTOR inhibitor for the seven patients was 16 months (range, 1 to 42 months), with one patient still on treatment; three patients experienced a partial response (all with predicted driver mTOR pathway mutations), one had stable disease, and three were not evaluable.

TABLE 2.

Actionable Alterations Identified in Tumors and Germline of Patients With Non–Clear-Cell Renal Cell Carcinoma

TABLE 3.

Actionable Somatic Alterations and Response to Targeted Treatment

Three patients had amplifications in MET (level 3; possibly predictive of response to cabozantinib), and five patients had additional driver mutations; of these, two patients received the VEGF and MET inhibitor cabozantinib as second-line therapy and were evaluable for response. Both patients achieved stable disease as best response, and length on therapy ranged from 6 to 9 months. One patient had an ALK translocation (level 2B; predictive of response to ALK inhibitors in lung cancer), was enrolled onto a clinical trial with a multitarget ALK inhibitor, and, as previously reported, achieved a partial response with 19 months on therapy.²⁰ Twenty-eight patients (24%) had a level 4 alteration (hypothetical biomarker, not proven to be clinically actionable), including mutations in PTEN, SMARCB1, CDKN2A, and ATM (Figs 1B and 1C).

Fourteen patients had both the primary tumor and a metastatic lesion sequenced (and one additional patient had two metastatic sites sequenced). Of these, four patients (27%) had a primary or metastatic site tumor sample with an actionable somatic alteration (three with TSC2 and one with ALK translocation; Appendix Table A4). We analyzed the concordance between the primary tumor and the metastatic site of the actionable genomic alterations; all TSC2 mutations were shared in primary tumors and metastases. The ALK translocation was only present in the metastasis sample; this tumor did not share any alterations with the primary tumor.

MSI

Of 115 available tumors for analysis, two (1.7%) were MSI-H, including one chromophobe tumor (MSIsensor score = 17.18) and one medullary tumor (MSIsensor score = 15.97). An additional five tumors (5%) were considered MSI-intermediate (MSI-I; with MSIsensor scores ranging from 3.3 to 4.5). Chromophobe tumors had the highest frequency of MSI-H or MSI-I status (24%; Table 4).

TABLE 4.

Microsatellite Instability Status by Tumor Subtype

Germline Mutation Analysis

Of the 116 patients, 45 (39%) underwent germline mutation testing with a panel of 76 cancer-associated genes; some of these patients were included in a previous report.²¹ In total, 11 (24%) of 45 patients harbored germline events in cancer-associated genes. The most frequently germline mutated gene was FH (n = 6; 13%), followed by one patient each who carried mutations in MET, ATM, CHEK2, PALB2, and APC. An additional patient was known to have a germline FH mutation on the basis of prior testing. Of the patients with FH or MET mutations, none had a family history of RCC. Of FH carriers, all women had a history of uterine fibroids, and only one patient had a documented history of cutaneous leiomyomas. The patient with a germline MET mutation had bilateral renal lesions.

Three patients with FH germline mutations received treatment with combination bevacizumab and everolimus, which has been shown be associated with improved response in patients with nccRCC with papillary features.¹⁹ Two patients had a partial response, and one patient had stable disease; the median progression-free survival time was 11.6 months (range, 6.1 to 13.7 months).

DISCUSSION

We performed a detailed investigation of somatic and germline mutations, as well as tumor MSI status, to further our understanding of disease heterogeneity and identify potentially targetable pathways in patients with nccRCC. Unlike most other studies of nccRCC, this study focused exclusively on patients with metastatic disease, which may have intrinsically different genomic mutation profiles. Overall, 22% of patients had a potentially actionable genomic alteration (13% somatic alteration, 7% germline, and 2% MSI-H) that could guide standard or, perhaps more importantly, investigational therapy selection.

Previously, the development of effective targeted therapies across nccRCC variants has been hindered by small patient numbers and the heterogeneity of diseases.^6,7,19,22-25 Recently, several large-scale sequencing efforts have attempted to delineate the genomic landscape of different nccRCC subtypes. The TCGA analyzed papillary and chromophobe tumors in two separate efforts.^11,12 In contrast with the TCGA analyses, where only primary tumors were processed, the patients in our study had primary and metastatic sites sequenced, perhaps better representing the limitations seen in clinic, where primary tumor tissue is either not available or not sampled. Pal et al²⁶ investigated the frequency of genomic alterations in advanced papillary tumors. Similar to these authors,²⁶ we found a predominance of MET, TERT promoter, and NF2 mutations in papillary tumors (although unlike the Pal et al²⁶ study, we did not distinguish between type 1 and type 2 papillary tumors). The current study demonstrates that papillary and unclassified RCC had shared genomic features, such as a comparatively high mutation frequency of NF2 in metastatic sites. This is consistent with a previous study that showed that NF2 loss in unclassified RCC may be a driver of tumorigenesis and associated with worse clinical outcome.²⁷ The comparatively high proportion of unclassified patients, many with molecular similarities to patients with papillary RCC, reflects the evolving definitions of RCC variant, with RCCs previously termed high-grade papillary RCC now termed unclassified.

Using the OncoKB classification to annotate individual mutations for their therapeutic potential, we found that 15% of tumors harbored potentially actionable somatic mutations. Unlike in other cancers such as melanoma or non–small-cell lung cancer, in RCC, there are no FDA-recognized or NCCN-listed biomarkers predictive of response to therapies (ie, level 1 OncoKB); the only exception is high tumor MSI, which is predictive of response to pembrolizumab in any cancer type. We did find 10 patient tumors (9%) with a level 2 alteration, indicating a standard care biomarker predictive of response in an FDA-approved drug in RCC or another indication. These alterations included an ALK translocation, MET amplifications, and TSC1 or TSC2 alterations. Five patients had tumors with level 3 alterations, predictive of clinical response on the basis of evidence in nccRCC or other tumors, including alterations in MTOR and PIK3CA. Most alterations were level 4, which are hypothetical biomarkers potentially predictive of response on the basis of preclinical data. Although these are not considered clinically actionable, they may be biomarkers and may be used as eligibility criteria for early-phase clinical trials.

We explored the efficacy of targeted therapies in patients with alterations considered potentially actionable who had received matched treatment. Although the number of patients was small, 33% of patients had an objective response, and an additional 25% had stable disease, which is notable in a group of diseases for which there are few effective therapies. Patients with MET alterations (amplifications or driver missense mutations) who received cabozantinib and patients with mTOR pathway driver alterations who received everolimus seemed to benefit from therapy, such as a patient with a PIK3CA driver mutation who had a partial response for more than 42 months on the combination of everolimus and bevacizumab. One patient with medullary RCC and ALK translocation who was enrolled onto a basket trial with an ALK inhibitor experienced a partial response and was on therapy for 19 months.²⁰ There have been previous reports of ALK translocations in patients with the rare but aggressive medullary subtype.²⁸ Given the lack of other effective therapies for patients with medullary RCC, molecular profiling would be an appropriate consideration for this subgroup.

MSI-H and mismatch repair deficiency have been identified as important predictors of response to checkpoint blockade across solid tumors.²⁹ To our knowledge, there are no prior reports of comprehensive MSI testing across cohorts of metastatic nccRCC, despite patients with this advanced disease being considered candidates for immunotherapy. Two patients in our study, one with medullary RCC and one with chromophobe RCC, had high MSI scores; neither had received immunotherapy at the time of analysis. Of interest, 29% of chromophobe tumors were MSI-H (MSIsensor score of 10 or greater) or MSI-I (MSIsensor score between 3 and 9). The association of MSI-H or MSI-I status with response to checkpoint inhibitors in patients with nccRCC is unknown. Previous analyses of prevalence of MSI-H in limited subtypes of nccRCC showed a 1% or lower prevalence of MSI-H; however, in these cohorts, less than 10% of patients had metastatic disease.^11,12,30,31 Additional investigation to characterize larger cohorts of chromophobe, medullary, and other nccRCC tumors, especially in patients with metastatic disease, is warranted. The overall low rate of MSI-H in nccRCC should not preclude the possibility of benefit from checkpoint inhibitors; despite MSI-H being rare in ccRCC, these tumors are clearly responsive to checkpoint inhibitors.^32,33 In addition, retrospective series in nccRCC have suggested benefit from anti–programmed cell death protein 1 therapy.³⁴

Notably, consistent with previous reports, there was a high proportion of patients with germline mutations, some which could direct therapy.²¹ Of 45 patients receiving germline genetic testing through our protocol, 13% had mutations in FH, which is diagnostic of the syndrome hereditary leiomyomatosis and RCC (HLRCC), which in turn is associated with nccRCC and cutaneous and uterine leiomyomas.^21,35 Identification of patients with HLRCC has become particularly relevant for treatment decisions since the 2018 NCCN guidelines added two bevacizumab-containing combination therapies (bevacizumab plus everolimus and bevacizumab plus erlotinib) as potential treatment for patients with HLRCC.³⁶ In our exploratory analysis of response to bevacizumab and everolimus, most patients had an objective response, with a median progression-free survival nearing 12 months. Although not currently an OncoKB level 2 or 3 alteration, germline and somatic activating MET mutations may also predict response to MET inhibitors. In a phase II study of a dual MET/VEGF receptor 2 inhibitor in patients with papillary RCC, the presence of a germline MET mutation was highly predictive of response.³⁷ One patient in our cohort harbored a MET germline mutation, but although MET-directed therapy with cabozantinib was initiated, the patient was subsequently lost to follow-up.

Of note, there was a high percentage of black patients with nccRCC (21%) in this cohort. In comparison, in our previously published cohort of patients with metastatic ccRCC who underwent sequencing under the same institutional protocol, only 1% were black.²¹ Although this may be a result of referral patterns, racial differences in RCC subtype prevalence have been noted by others, as well as potential tumor genomic differences.^38,39 These observations merit additional investigation in larger cohorts.

Although this study represents a large cohort of patients with several subtypes of nccRCC, it is important to recognize its limitations. Patient numbers limit the degree of certainty with which predictive and prognostic effects of specific genetic alterations can be determined. This is also a single-institution experience, and prevalence of RCC subtypes may vary by geography or ancestry. There may also be referral bias to a specialized cancer center. Finally, our analysis was limited to targeted exome sequencing, and this approach could have missed other nonexonic as well as epigenetic changes

In summary, in this detailed genomic analysis of a wide array of metastatic nccRCC tumors, matched germline and tumor analyses revealed recurrent genomic events across RCC variants and suggest that a relevant proportion of patients harbor potentially actionable alterations. There is a need for additional collaborative studies to characterize metastatic nccRCC further so that targeted therapies can be pursued rationally, exploiting such information.

Appendix

TABLE A1.

Genes on MSK-IMPACT and Syndromes Associated With Germline Mutations

TABLE A2.

Frequency of Mutations by Subtype

TABLE A3.

Actionable Alterations by Highest OncoKB Level

TABLE A4.

Concordance of Genomic Alterations in Matched Primary and Metastasis Samples

AUTHOR CONTRIBUTIONS

Conception and design: Maria I. Carlo, Nabeela Khan, Mark Robson, A. Ari Hakimi, Robert J. Motzer, Martin H. Voss

Administrative support: Almedina Redzematovic, Devyn T. Coskey

Provision of study materials or patients: Maria I. Carlo, Chung-Han Lee, Darren R. Feldman, Robert J. Motzer, Martin H. Voss

Collection and assembly of data: Maria I. Carlo, Nabeela Khan, Ahmet Zehir, Yasser Ged, Almedina Redzematovic, Devyn T. Coskey, David M. Hyman, Marc Ladanyi, Ying-Bei Chen, Mark Robson, Chung-Han Lee, Martin H. Voss

Data analysis and interpretation: Maria I. Carlo, Nabeela Khan, Sujata Patil, Yasser Ged, David M. Hyman, Marc Ladanyi, Ying-Bei Chen, Mark Robson, Chung-Han Lee, Darren R. Feldman, Jianjiong Gao, Debyani Chakravarty, Robert J. Motzer, Martin H. Voss

Manuscript writing: All authors

Final approval of manuscript: All authors

Accountable for all aspects of the work: All authors

AUTHORS' DISCLOSURES OF POTENTIAL CONFLICTS OF INTEREST

The following represents disclosure information provided by authors of this manuscript. All relationships are considered compensated. Relationships are self-held unless noted. I = Immediate Family Member, Inst = My Institution. Relationships may not relate to the subject matter of this manuscript. For more information about ASCO's conflict of interest policy, please refer to www.asco.org/rwc or ascopubs.org/jco/site/ifc.

Maria I. Carlo

Consulting or Advisory Role: Pfizer

David M. Hyman

Consulting or Advisory Role: Atara Biotherapeutics, Chugai Pharma, CytomX Therapeutics, Boehringer Ingelheim, AstraZeneca, Pfizer, Bayer, Debiopharm Group, Genentech

Research Funding: AstraZeneca, Puma Biotechnology, Loxo

Travel, Accommodations, Expenses: Genentech, Chugai Pharma

Marc Ladanyi

Honoraria: Merck (I)

Consulting or Advisory Role: National Comprehensive Cancer Network/AstraZeneca Tagrisso RFP Advisory Committee, Takeda, Bristol-Myers Squibb, Bayer, Merck (I)

Research Funding: Loxo (Inst), Helsinn Therapeutics

Mark Robson

Honoraria: AstraZeneca, Pfizer

Consulting or Advisory Role: McKesson, AstraZeneca, Merck

Research Funding: AstraZeneca (Inst), Myriad Genetics (Inst), InVitae (Inst), Pfizer (Inst), Abbvie (Inst), Tesaro (Inst), Medivation (Inst)

Travel, Accommodations, Expenses: AstraZeneca

Chung-Han Lee

Consulting or Advisory Role: Exelixis, Eisai

Research Funding: Pfizer (Inst), Eisai (Inst), Bristol-Myers Squibb (Inst), Calithera Biosciences (Inst), Exelixis (Inst)

Travel, Accommodations, Expenses: Eisai

Darren R. Feldman

Research Funding: Novartis, Seattle Genetics, Decibel Therapeutics (Inst)

Other Relationship: UpToDate

Robert J. Motzer

Consulting or Advisory Role: Pfizer, Novartis, Eisai, Exelixis, Merck, Genentech, Incyte

Research Funding: Pfizer (Inst), Bristol-Myers Squibb (Inst), Eisai (Inst), Novartis (Inst), Genentech (Inst)

Martin H. Voss

Honoraria: Novartis

Consulting or Advisory Role: Novartis, Calithera Biosciences, GlaxoSmithKline, Exelixis, Pfizer, Alexion Pharmaceuticals, Eisai, Corvus Pharmaceuticals

Research Funding: Bristol-Myers Squibb, Genentech

Travel, Accommodations, Expenses: Takeda, MedImmune, Eisai

No other potential conflicts of interest were reported.