Efficacy and Safety of Nivolumab Plus Ipilimumab versus Sunitinib in First-line Treatment of Patients with Advanced Sarcomatoid Renal Cell Carcinoma

This article is featured in Highlights of This Issue, p. 1

Sarcomatoid features can coexist with any histologic subtype and stage of renal cell carcinoma (RCC), and are associated with rapid disease progression and mortality (1, 2). Sarcomatoid RCCs (sRCC) have pathologic features that are similar to sarcomas, with dense spindle-like cells and cellular atypia (3). About 5% of patients with RCC have sRCC, including 10%–20% of patients with advanced disease (3, 4). Most patients with sRCC have metastatic disease at diagnosis, and the presence of sarcomatoid features is an independent predictor of poor survival (1, 3, 5).

A significant unmet need remains for the management of patients with sRCC. To date, only limited efficacy with various cytotoxic chemotherapies has been reported (1, 3, 6–8). In addition, studies of molecular targeted therapies (e.g., VEGF signaling inhibitors, mTOR inhibitors) alone or in combination with chemotherapy showed limited benefit (3, 9–17). Clinical trials of chemotherapeutic regimens and targeted agents in sRCC have previously reported objective response rates (ORR) of 16%–33% with very few complete responses (CR), median progression-free survival (PFS) ranging from approximately 2 to 5 months, and median overall survival (OS) ranging from approximately 5 to 12 months (1, 7, 10, 11, 15, 16, 18).

Preliminary research with programmed death-1/programmed death ligand 1 (PD-1/PD-L1) immune checkpoint inhibitor–based monotherapy has shown promising efficacy in patients with sRCC (19–21). A small study reported that sRCCs expressed PD-L1 at higher rates [14/26 (54%) evaluable patients] than clear-cell RCC without sarcomatoid differentiation [5/29 (17%) evaluable patients; ref. 22]. A single-institution, retrospective study (N = 39) demonstrated that immune checkpoint inhibitors, including nivolumab (NIVO) alone or in combination with ipilimumab (IPI), were effective in treating patients with clear-cell sRCC, with five of 33 (15%) patients achieving a durable CR (23). However, clinical trial data assessing the role of immunotherapy in sRCC remain limited (24–26). Here, we present post hoc exploratory analyses in the subpopulation of patients with sRCC in the phase III CheckMate 214 trial, which demonstrated sustained superiority of first-line dual checkpoint inhibition with NIVO+IPI versus sunitinib in patients with advanced RCC (aRCC).

CheckMate 214 included adults with treatment-naïve aRCC with a clear-cell component; key inclusion and exclusion criteria were reported previously (27). Randomized patients were stratified according to geographical region and International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) risk category [favorable (score of 0) vs. intermediate (1 or 2) vs. poor (3–6); ref. 27]. Sufficient archival or recently acquired formalin-fixed, paraffin-embedded tumor tissue samples (block or minimum of 10 slides from core biopsy, punch biopsy, excisional biopsy, or surgical specimen) were required from all patients at screening before randomization; reporting of sarcomatoid features prospectively was not a protocol requirement.

This post hoc, exploratory analysis of CheckMate 214 identified patients with aRCC with sarcomatoid histology, and measured efficacy and safety outcomes in this subpopulation. Randomized patients were identified as having a sarcomatoid-positive RCC by two methods: (i) a manual, multilingual keyword search for “sarcomatoid” in local pathology reports accompanying preassessment tumor samples, and (ii) an independent central review of those patients with available tissue samples using hematoxylin and eosin–stained slides (Supplementary Fig. S1). For independent central review, sarcomatoid differentiation was defined according to criteria discussed at the International Society of Urological Pathology 2012 Consensus Conference (28). The percent of tumor area with sarcomatoid differentiation was independently assessed by three pathologists blinded to patient outcomes; discrepancies were resolved by consensus review. Tumors identified by either method as having any percent sarcomatoid-positive histology were considered to be sarcomatoid positive (28). Analyses described here focus on the combined population of patients with sarcomatoid-positive tumors identified by central or local pathology review. Outcomes were also assessed in the subpopulations of patients with sRCC identified by local pathology report alone and by central pathology review alone, which were described as sensitivity analyses (Supplementary Fig. S1). Most patients with sarcomatoid-positive RCC had IMDC intermediate/poor (I/P)-risk disease, although a very small number of patients with favorable-risk disease were identified as having sarcomatoid-positive tumors. Consistent with the primary CheckMate 214 efficacy population, this analysis is primarily focused on outcomes in patients with sRCC and I/P-risk disease. Efficacy and safety analyses for the entire CheckMate 214 patient population has been reported previously (29).

CheckMate 214 was a randomized, open-label, phase III trial of NIVO+IPI followed by nivolumab monotherapy versus sunitinib monotherapy. The study design has been described previously (27). Briefly, patients were randomized 1:1 to either NIVO+IPI or sunitinib. NIVO+IPI was administered at doses of 3 mg/kg of nivolumab intravenously over a period of 60 minutes and 1 mg/kg of ipilimumab intravenously over 30 minutes, every 3 weeks for four doses (induction phase), followed by nivolumab monotherapy at 3 mg/kg every 2 weeks (maintenance phase). Sunitinib was administered at a dose of 50 mg orally once daily for 4 weeks on and 2 weeks off in each 6-week cycle. Treatment continued until disease progression or unacceptable toxicity (27).

This trial was approved by the institutional review board or independent ethics committee at each center and conducted in accordance with Good Clinical Practice guidelines defined by the International Conference on Harmonisation. Written informed consent was provided by all patients according to the principles of the Declaration of Helsinki, as reported previously (27).

Outcomes in patients with sRCC were exploratory, and not prespecified in the trial protocol. Similar to the primary analyses in the global CheckMate 214 trial population, this post hoc analysis in patients with sRCC and I/P-risk disease evaluated OS, PFS, and ORR (including response kinetics; ref. 27). Analyses in a small exploratory population of patients with sRCC and favorable-risk disease were descriptive. Outcomes were also assessed according to baseline tumor PD-L1 expression level (≥1% vs. <1%) in patients with sRCC and I/P-risk disease. Baseline tumor PD-L1 expression was assessed at a central laboratory using the Dako PD-L1 IHC 28-8 pharmDx test. Specimens that were missing, not optimally collected, or otherwise not categorized were classified as unknown. Safety was assessed in all treated patients with sRCC and I/P-risk disease. Treatment-free interval was defined as the time between protocol therapy discontinuation until subsequent therapy initiation or last known date alive.

The HRs between treatment arms in patients with sRCC and I/P-risk disease were calculated along with the two-sided 95% confidence intervals (CIs) for OS and PFS. OS, PFS, and duration of response were estimated using Kaplan–Meier methods (30). Confirmed ORRs, along with the exact two-sided 95% CI by the Clopper–Pearson method (31), were computed. Efficacy outcomes in patients with sRCC and favorable-risk disease as well as safety outcomes in all treated patients with sRCC and I/P-risk disease were reported using descriptive statistics.

Bristol Myers Squibb's policy on data sharing may be found at https://www.bms.com/researchers-and-partners/independent-research/data-sharing-request-process.html.

A total of 1,096 patients underwent randomization in CheckMate 214 (550 to NIVO+IPI and 546 to sunitinib in the intent-to-treat population; 425 to NIVO+IPI and 422 to sunitinib had I/P-risk disease; and 125 and 124, respectively) had favorable-risk disease (27). Of all randomized patients, 145 patients with aRCC were identified in total as having sarcomatoid-positive tumors; 139/847 (16.4%) had I/P-risk disease, and 6/249 (2.4%) had favorable-risk disease (Supplementary Fig. S1). Among the 139 patients with sRCC and I/P-risk disease, 73 of 74 (98.6%) patients randomized to NIVO+IPI and all 65 (100%) patients randomized to sunitinib received treatment. Baseline characteristics in the sRCC subpopulation were similar in the NIVO+IPI and sunitinib arms (Table 1).

As of the database lock (August 7, 2019), the minimum study follow-up was 42 months for the total study population in CheckMate 214 and among patients with sRCC (median, 47.7 months in patients with sRCC). The median duration of treatment (95% CI) in patients with sRCC and I/P-risk disease was 7.9 months (4.2–14.5) with NIVO+IPI and 4.7 months (2.9–6.4) with sunitinib. Of all patients who received treatment, 13 of 73 (18%) in the NIVO+IPI arm versus one of 65 (2%) in the sunitinib arm remained on treatment. The primary reason for treatment discontinuation was disease progression, observed in 27 of 73 (37.0%) treated patients in the NIVO+IPI arm and 46 of 65 (70.8%) in the sunitinib arm (Supplementary Fig. S1).

NIVO+IPI showed notable survival benefits over sunitinib. Median OS (95% CI) was not reached (25.2 months–not estimable) with NIVO+IPI versus 14.2 months (9.3–22.9) with sunitinib; HR for death was 0.45 (95% CI, 0.3–0.7; P = 0.0004; Fig. 1). The 42-month probability of survival (95% CI) was 50.1% (37.9–61.2) with NIVO+IPI versus 22.6% (13.3–33.4) with sunitinib. Median PFS (95% CI) was also significantly longer with NIVO+IPI versus sunitinib: 26.5 months (8.4–not estimable) versus 5.1 months (4.0–6.9); HR for disease progression or death was 0.54 (95% CI, 0.3–0.9; P = 0.0093; Fig. 2). The 36-month probability of PFS (95% CI) was 48.2% (35.7–59.7) with NIVO+IPI versus 20.3% (9.8–33.5) with sunitinib. In addition, patients had higher ORRs (95% CI) with NIVO+IPI versus sunitinib [60.8% (49–72) vs. 23.1% (14–35); P < 0.0001], with 18.9% and 3.1% of patients achieving CRs, respectively (Table 2).

With NIVO+IPI, most patients experienced either no increase or a reduction in target lesion size over time (Fig. 3). Median (range) time to response was similar between treatment arms: 2.8 months (0.9–18.1) for patients treated with NIVO+IPI and 2.8 months (2.4–23.5) for those treated with sunitinib. Median duration of response (95% CI) was not reached (22.5 months–not estimable) with NIVO+IPI versus 20.7 months (7.2–38.7) with sunitinib. In responders, ongoing response was observed in 31 of 45 (69%) patients (11 of 14 patients with CR) with NIVO+IPI and 8 of 15 (53%) patients with sunitinib (1 of 2 patients with CR). Eleven of 45 (24%) responders treated with NIVO+IPI and 1 of 15 (7%) responders treated with sunitinib remained on therapy at the time of database lock. In addition, 20 of 45 (44%) responders experienced a treatment-free interval without subsequent systemic therapy in the NIVO+IPI arm versus five of 15 (33%) responders treated with sunitinib (Supplementary Fig. S2).

In the sensitivity analyses, results for OS, PFS, and confirmed ORR were largely consistent between the combined pathology sRCC cohort and the sRCC cohorts identified by either central review or local review alone, with the exception of varying CR rates (Supplementary Table S1).

Baseline PD-L1 expression was quantifiable in 71 of 74 (96%) patients with sRCC treated with NIVO+IPI and 62 of 65 (95%) patients treated with sunitinib. Of these, 36 (51%) patients in the NIVO+IPI arm and 33 (53%) patients in the sunitinib arm had baseline tumor PD-L1 expression ≥1% (Table 1).

Efficacy outcomes were notably better with NIVO+IPI versus sunitinib, regardless of PD-L1 expression. In all patients with sRCC and evaluable baseline tumor PD-L1 expression ≥1%, the median OS (95% CI) was not reached (29.9 months–not estimable) with NIVO+IPI and 20.9 months (9.3–41.2) with sunitinib; HR for death was 0.42 (0.20–0.91; P = 0.0260). Median OS among patients with baseline tumor PD-L1 expression <1% was 40.4 months (18.5–not estimable) with NIVO+IPI and 13.8 months (5.5–20.3) with sunitinib, with an HR for death of 0.38 (0.20–0.71; P = 0.0020; Supplementary Fig. S3A). Median PFS (95% CI) in patients with baseline tumor PD-L1 expression ≥1% was not reached (9.1 months–not estimable) with NIVO+IPI and 4.4 months with sunitinib (2.8–6.9); HR for PFS was 0.31 (0.16–0.62; P = 0.0006). The median PFS among patients with baseline tumor PD-L1 expression <1% was 10.9 months (3.3–not estimable) with NIVO+IPI and 5.1 months (2.9–17.0) with sunitinib, with an HR for PFS of 0.65 (0.34–1.23; P = 0.1852; Supplementary Fig. S3B). Furthermore, the ORR among patients with ≥1% tumor PD-L1 expression reached 69.4% with NIVO+IPI (n = 36) versus 24.2% with sunitinib (n = 33), with an odds ratio (OR) of 7.1 (95% CI, 2.2–23.9); of these, 22.2% versus 3.0% of patients had CRs, respectively. Ongoing response was observed in 18 of 25 (72%) responders treated with NIVO+IPI and in five of eight (63%) responders treated with sunitinib. For patients with baseline tumor PD-L1 <1%, the ORRs for NIVO+IPI (n = 35) versus sunitinib (n = 29) were 54.3% and 20.7% (OR 4.6; 95% CI, 1.3–16.8). Of these patients, 17.1% treated with NIVO+IPI and 3.4% treated with sunitinib had CRs (Table 2). In addition, ongoing response was observed in 12 of 19 (63%) responders treated with NIVO+IPI and three of six (50%) responders treated with sunitinib.

Few patients with sRCC and favorable-risk disease were identified (three each in the NIVO+IPI and sunitinib arms; one patient with favorable-risk disease randomized to receive sunitinib was not treated). In patients with favorable-risk disease who received NIVO+IPI, best overall responses included one patient with stable disease and two with progressive disease. Of the two patients with favorable-risk disease who received sunitinib, both had best overall responses of stable disease. The primary reason for treatment discontinuation was disease progression, observed in two of three patients in the NIVO+IPI arm and in one of two treated patients in the sunitinib arm.

Treatment-related adverse events (AEs) of any grade were reported in 71 of 73 (97%) patients treated with NIVO+IPI and in 63 of 65 (97%) patients treated with sunitinib. Grade 3 or 4 AEs occurred in 36 (49%) and 29 (45%) patients, respectively (Table 3). Treatment-related AEs leading to discontinuation occurred in 15 (21%) patients in the NIVO+IPI arm and eight (12%) patients in the sunitinib arm. Treatment-related select AEs (potentially immune-mediated) were generally infrequent in the pulmonary (NIVO+IPI, 10%; sunitinib, 0%) and renal (NIVO+IPI, 11%; sunitinib, 6%) categories and more marked with regard to any-grade skin-related events (NIVO+IPI, 52%; sunitinib, 42%), any-grade gastrointestinal-related events (NIVO+IPI, 25%; sunitinib, 37%), any-grade hepatic-related events (NIVO+IPI, 21%; sunitinib, 11%), and any-grade endocrine-related events (NIVO+IPI, 37%; sunitinib, 18%; Table 3). Of the 73 patients with sRCC treated with NIVO+IPI, 15 (21%) received corticosteroids [≥40 mg prednisone daily or equivalent (PDE)] to manage treatment-related select AEs. Nine (12%) patients received corticosteroid at a dose of ≥40 mg PDE continuously for ≥2 weeks and 4 (5%) received ≥40 mg PDE continuously for ≥30 days. One treatment-related death, reported previously, occurred in a patient with sRCC and I/P-risk disease treated with NIVO+IPI (a 79-year-old man with grade 4 hepatic failure who later developed grade 3 febrile neutropenia; ref. 27). No deaths were reported among patients with sRCC and I/P-risk disease treated with sunitinib. In addition, the safety profile observed with NIVO+IPI in patients with sRCC was consistent with the trial population as a whole (32).

A marked efficacy benefit in OS, PFS, and ORR outcomes was observed with first-line NIVO+IPI over sunitinib in patients with sRCC and IMDC I/P-risk disease in this post hoc exploratory analysis of CheckMate 214. After an extended minimum follow-up of 42 months, ORR and CR rates were considerably higher, and long-term OS and PFS benefits favored treatment with NIVO+IPI over sunitinib. Efficacy was better in patients treated with NIVO+IPI versus sunitinib regardless of tumor PD-L1 expression level; however, the magnitudes of OS, PFS, and ORR benefits with NIVO+IPI observed in this patient subset were greater for those with tumor PD-L1 expression ≥1% versus those with tumor PD-L1 expression <1%. Interestingly, 22% of patients with sRCC and tumor PD-L1 expression ≥1% achieved CR with NIVO+IPI. Though baseline tumor PD-L1 expression in patients with sRCC and I/P-risk disease in this analysis was higher than that in all patients with I/P-risk disease in the overall CheckMate 214 population, this is anticipated given that PD-L1 is a prognostic factor for poor outcomes and is typically associated with more advanced disease (22, 27). Altogether, these data support the use of NIVO+IPI dual combination immunotherapy for the treatment of patients with sRCC regardless of tumor PD-L1 expression level. Molecular analyses are underway to further characterize the immune response in sRCC and to assess the biological rationale for the apparent enriched response to NIVO+IPI observed in patients with sRCC and I/P-risk disease in this study.

Previously reported outcomes for patients with sRCC treated with traditional therapies were suboptimal, with most clinical studies reporting OS medians of <1 year from the time of diagnosis (1, 3, 5, 7, 10–14, 16, 33, 34). In this setting, NIVO+IPI appears particularly efficacious, yet emerging phase III clinical trial data show that immuno-oncology (I-O) combinations with VEGF tyrosine kinase inhibitors (TKIs) may also provide some benefit relative to older treatment options (1, 3, 5, 7, 10–14, 16, 24, 25, 33–35). Clinical trial data from KEYNOTE-426 and JAVELIN Renal 101 appear to show that such combinations are more effective than sunitinib, although the long-term benefits in patients with sRCC treated with these regimens are not yet known (24, 25). A subgroup analysis of 115 previously untreated patients with sRCC included in the phase III KEYNOTE-426 trial assessed outcomes with the combination of anti–PD-1 checkpoint inhibitor pembrolizumab plus axitinib versus sunitinib and reported data with a 12-month median follow-up. Median OS was not reached for pembrolizumab plus axitinib or sunitinib (HR, 0.70; 95% CI, 0.28–1.73); median PFS was 12.6 months with this I-O+TKI combination versus 8.4 months with sunitinib (HR, 0.56; 95% CI, 0.32–0.99); an ORR of 54.9% versus 31.5% was reported with CR rates of 13.7% versus 0%, respectively (25). The phase III, JAVELIN Renal 101 trial assessed the PD-L1 checkpoint inhibitor avelumab plus axitinib versus sunitinib in 108 patients with untreated sRCC, reporting a median PFS of 7.0 months versus 4.0 months, an ORR of 46.8% versus 21.3%, and CR rates of 4.3% versus 0% in patients with sRCC, respectively (24).

The limitations of the current analysis include its exploratory post hoc nature: patients with sRCC were not prospectively randomized in CheckMate 214, although the distribution of patients with sRCC was balanced between treatment arms. This analysis was not powered to detect differences according to baseline PD-L1 expression, therefore, analyses of outcomes by PD-L1 status in patients with sRCC should only be considered as hypothesis generating. In addition, this analysis identified patients with aRCC and sarcomatoid histology via two distinct methodologies: central pathology review, which allowed for analysis of archived patient tissue samples, and local pathology reports, which reduced the risk of identifying false negatives due to limited tissue availability for central review. Despite the differing methodologies, sensitivity analyses of patients with aRCC identified to be sarcomatoid positive by central pathology review alone or local pathology report alone showed similar efficacy outcomes, with the exception of CR rates (Supplementary Table S1). The reason for this difference, which reflects only a small number of patients, is unclear. Finally, individual tumor samples were derived from both nephrectomy and metastatic biopsy in CheckMate 214, and a definitive determination of sarcomatoid status was not possible in biopsy specimens.

In conclusion, dual checkpoint inhibition with NIVO+IPI showed substantial long-term clinical benefit in patients with sRCC, a particularly aggressive form of RCC that traditionally carries a much worse prognosis and has a significant unmet medical need. The data presented here show that with 42 months' minimum follow-up, NIVO+IPI offers the potential to achieve durable responses, high ORR and CR rates, and OS benefit, supporting this combination as a standard of care for first-line treatment in patients with clear-cell sRCC and I/P-risk disease. In light of these new data, a recently updated Society for Immunotherapy of Cancer consensus statement has recommended NIVO+IPI combination immunotherapy as a first-line treatment option for patients with sRCC based on sarcomatoid subset analyses of several trials enrolling patients with clear-cell RCC (36).

N.M. Tannir reports grants from Bristol Myers Squibb during the conduct of the study; grants and personal fees from Exelixisa and Nektar Therapeutics; grants from Calithera Bioscience, Takeda, Arrowhead Pharmaceuticals, and Epizyme, Inc.; personal fees from Bristol Myers Squibb, Pfizer, Lilly Oncology, Neolukin Therapeutics, Eisai, Novartis, Oncorena, Ono Pharmaceutical, Surface Oncology, and Ipsen outside the submitted work. S. Signoretti reports grants and personal fees from Bristol Myers Squibb during the conduct of the study; and grants and personal fees from AstraZeneca; grants from Exelixis and Novartis; personal fees from Merck, CRISPR Therapeutics AG, AACR, and NCI outside the submitted work, and received royalties from Biogenex related to their CDX2 antibody. T.K. Choueiri reports grants, personal fees, nonfinancial support, and other from BMS, Merck, Roche, Pfizer, EMD, Exelixis, Novartis, and AstraZeneca [advisory board, consultancy, honorarium, payments (personal and for institution)], and a patent for IO biomarkers pending, issued, and licensed to DFCI (related to IO). D.F. McDermott reports personal fees from BMS (advisory board participant) during the conduct of the study; personal fees from Merck, Genentech, and Pfizer outside the submitted work. R.J. Motzer reports grants and other from Bristol Myers Squibb (travel expense to author and grant support to employers) during the conduct of the study; grants and personal fees from Genentech/Roche (paid consultancy to self and grant support to employer), Merck (paid consultancy to self and grant support to employer), Pfizer (paid consultancy to self and grant support to employer), Eisai (paid consultancy to self and grant support to employer), and Exelixis (paid consultancy to self and grant support to employer); grants from AstraZeneca (paid consultancy to self) and grants and personal fees from Novartis (paid consultancy to self and grant support to employer) outside the submitted work. J.-C. Pignon reports personal fees from Bristol Myers Squibb (current consultant for BMS) outside the submitted work. S. George reports grants and personal fees from Bayer (institutional grant), BMS (institutional grant), Corvus (institutional grant), Eisai (institutional grant), Merck (institutional grant), Pfizer (institutional grant), and Seattle Genetics (institutional grant); personal fees from Sanofi, Exelixis, Genentech, and EMD Serono; grants from Agensys (institutional grant), Immunomedics (institutional grant), Calithera (institutional grant), and Novartis (institutional grant) outside the submitted work. T. Powles reports grants and personal fees from AstraZeneca and Roche; personal fees from BMS, MSD, Novartis, Pfizer, Exelixis, and Merck Serono during the conduct of the study. F. Donskov reports grants from Pfizer (to institution), MSD (to institution), and Ipsen (to institution) outside the submitted work. M.R. Harrison reports grants from Bristol Myers Squibb (research funding to institution) during the conduct of the study; grants from Acerta, Astellas, Clovis Oncology, and Merck; grants and personal fees from AstraZeneca, Bayer, Bristol Myers Squibb, Exelixis, Genentech, and Pfizer; personal fees from FujiFilm and Janssen, and grants from Seattle Genetics outside the submitted work. P. Barthélémy reports other from BMS (consulting, travel) during the conduct of the study; other from MSD (consulting, travel), Pfizer (consulting, travel), Ipsen (consulting, travel), Janssen Cilag (consulting, travel), AstraZeneca (consulting), Amgen (consulting, travel), Novartis (consulting), Eusapharma (consulting), Roche (consulting, travel), and Astellas (consulting, travel) outside the submitted work. S.S. Tykodi reports other from Bristol Myers Squibb (clinical trial support received on behalf of my institution) and nonfinancial support from Bristol Myers Squibb (manuscript preparation) during the conduct of the study; other from Merck (clinical trial support received on behalf of my institution), Merck (consultant), Nektar Therapeutics (clinical trial support received on behalf of my institution), Jounce Therapeutics (clinical trial support received on behalf of my institution), Pfizer (clinical trial support received on behalf of my institution), Clinigen (clinical trial support received on behalf of my institution), Exelixis (clinical trial support received on behalf of my institution), Genentech (clinical trial support received on behalf of my institution), and Calithera Biosciences (clinical trial support received on behalf of my institution) outside the submitted work. A. Ravaud reports personal fees and nonfinancial support from Pfizer, Merck, MSD, Ipsen, BMS, and Roche during the conduct of the study; grants, personal fees, and nonfinancial support from Pfizer and Merck; and personal fees and nonfinancial support from AstraZeneca outside the submitted work. J.R. Rodriguez-Cid reports other from Bristol Myers Squibb (trial funding) during the conduct of the study; and other Bristol Myer Squibb (trials funding, speaker, advisory), Roche (trials funding, speaker, advisory), MSD (trials funding, speaker, advisory), Novartis (trials funding, speaker, advisory), Boehinger Ingelheim (trials funding, speaker, advisory), Takeda (trials funding, speaker, advisory), Pfizer (speaker, advisory), Bayer (speaker, advisory), Celltrion (trials funding), Sanofi Aventis (speaker, advisory), Janssen (trials funding), and AstraZeneca (trials funding, speaker, advisory) outside the submitted work. S.K. Pal reports personal fees from Novartis, Aveo, Genentech, Bristol Myers Squibb, Astellas Pharma, Eisai, Roche, Ipsen, and Medivation outside the submitted work. Y. Ishii reports being a BMS employee. S.S. Saggi reports other from Bristol Myers Squibb (employee) outside the submitted work. M.B. McHenry reports being a BMS employee and owning BMS stock. B.I. Rini reports grants and personal fees from BMS during the conduct of the study; grants, personal fees, and nonfinancial support from Merck and Pfizer; grants and personal fees from Genentech and Aveo; personal fees from GSK, 3D Medicines, and Surface Oncology; grants from AstraZeneca, Exelixis, Alkermes, Arrowhead, and Arravive outside the submitted work. No disclosures were reported by the other authors.

N.M. Tannir: Resources, formal analysis, investigation, writing-original draft, writing-review and editing. S. Signoretti: Resources, data curation, formal analysis, validation, investigation, visualization, writing-original draft, writing-review and editing, independently/blindly assessed percent of tumor area with sarcomatoid differentiation. T.K. Choueiri: Formal analysis, writing-review and editing. D.F. McDermott: Formal analysis, writing-review and editing. R.J. Motzer: Formal analysis, writing-review and editing. A. Flaifel: Resources, data curation, formal analysis, validation, investigation, visualization, writing-review and editing, independently/blindly assessed percent of tumor area with sarcomatoid differentiation. J.-C. Pignon: Resources, data curation, formal analysis, validation, investigation, visualization, writing-review and editing. M. Ficial: Resources, data curation, formal analysis, validation, investigation, visualization, writing-review and editing, independently/blindly assessed percent of tumor area with sarcomatoid differentiation. O. Arén Frontera: Formal analysis, writing-review and editing. S. George: Formal analysis, writing-review and editing. T. Powles: Formal analysis, writing-review and editing. F. Donskov: Formal analysis, writing-review and editing. M.R. Harrison: Formal analysis, writing-review and editing. P. Barthélémy: Formal analysis, writing-review and editing. S.S. Tykodi: Formal analysis, writing-review and editing. J. Kocsis: Formal analysis, writing-review and editing. A. Ravaud: Formal analysis, writing-review and editing. J.R. Rodriguez-Cid: Formal analysis, writing-review and editing. S.K. Pal: Formal analysis, writing-review and editing. A.M. Murad: Formal analysis, writing-review and editing. Y. Ishii: Formal analysis, writing-review and editing. S.S. Saggi: Resources, formal analysis, writing-review and editing. M.B. McHenry: Resources, formal analysis, writing-review and editing. B.I. Rini: Formal analysis, investigation, writing-original draft, writing-review and editing.

This work was supported by Bristol Myers Squibb (Princeton, NJ) and ONO Pharmaceutical Company Ltd. (Osaka, Japan). Authors received no financial support or compensation for publication of this article. The University of Texas MD Anderson Cancer Center is supported by the NIH (grant no. P30 CA016672). Patients treated at Memorial Sloan Kettering Cancer Center were supported in part by Memorial Sloan Kettering Cancer Center Support Grant (Core grant number P30 CA008748).

The patients and families who made this study possible. The clinical study teams who participated in the study. The study was supported by Bristol Myers Squibb. All authors contributed to and approved the presentation; writing and editorial assistance was provided by Rachel Lieberman, PhD, of Parexel, funded by Bristol Myers Squibb. Dako, an Agilent Technologies, Inc. company, for collaborative development of the PD-L1 IHC 28-8 pharmDx assay (Santa Clara, CA). Bristol Myers Squibb (Princeton, NJ) and ONO Pharmaceutical Company Ltd. (Osaka, Japan).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.