Recent approvals of immunotherapy-based regimens have revolutionized the treatment of patients with advanced renal cell carcinoma (aRCC).1 2 First-line aRCC treatments include the dual immune checkpoint inhibitor combination nivolumab plus ipilimumab (NIVO+IPI) as well as immunotherapy–tyrosine kinase inhibitor combinations, with numerous novel regimens under investigation.3–7 The extent to which response and survival benefits with immunotherapy-based regimens persist after long-term follow-up will inform optimal first-line treatment options.

NIVO+IPI combination therapy was the first to demonstrate superiority over sunitinib (SUN) in the first-line treatment of patients with International Metastatic Renal Cell Carcinoma Database Consortium (IMDC) intermediate-risk/poor-risk aRCC.3 In the primary analysis of the phase 3 CheckMate 214 trial (minimum follow-up, 17.5 months), overall survival (OS) was superior (HR, 0.63; p<0.001) and confirmed objective response rate (ORR) was higher (42% vs 27%; p<0.001 per independent radiology review committee (IRRC)).3 Similar efficacy benefits were observed with NIVO+IPI in intention-to-treat (ITT) patients (any IMDC risk) and while ORR was higher with SUN in the exploratory favorable-risk population, OS outcomes were immature as of the primary analysis.3 After 30 months’ minimum follow-up, efficacy benefits with NIVO+IPI over SUN were maintained, including improved OS in both intermediate-risk/poor-risk and ITT patients, while the difference in OS outcomes between treatment arms was inconclusive in the favorable-risk subgroup.8

Here, we report additional follow-up through a minimum of 42 months to better inform the long-term impact of NIVO+IPI on clinical outcomes in the primary, secondary, and exploratory efficacy populations in CheckMate 214. Response per IRRC, durability of response, health-related quality of life, and characterization of safety were assessed with extended follow-up. Additionally, post hoc landmark OS analyses were conducted in patients with treatment-related adverse events (AEs) of interest and by response outcomes. This ongoing, multicenter trial enrolled patients between October 16, 2014, and February 23, 2016. These data represent the longest follow-up from a phase 3 trial of a dual immune checkpoint inhibitor regimen for aRCC reported to date.

A total of 1096 patients were randomized to NIVO+IPI (425 with intermediate-risk/poor-risk and 125 with favorable-risk disease) or SUN (422 with intermediate-risk/poor-risk and 124 with favorable-risk disease). Efficacy analyses were conducted in intermediate-risk/poor-risk, ITT (any risk), and favorable-risk patients. Overall, 547 patients in the NIVO+IPI arm and 535 in the SUN arm received treatment and were included in the safety analyses. Patients were enrolled from October 16, 2014, through February 23, 2016. The database lock for this analysis was August 7, 2019. At a minimum study follow-up of 42 months, 60 (11%) of 547 patients in the NIVO+IPI arm and 27 (5%) of 535 patients in the SUN arm continued therapy (online supplementary figure S1). Median follow-up for OS was 43.6 months in the NIVO+IPI arm and 32.3 months in the SUN arm (median of 39.3 months for the total study population).

Key baseline characteristics were similar between treatment arms and across intermediate-risk/poor-risk, ITT, and favorable-risk patients, as reported previously (online supplementary table S1).3 8 Median duration of therapy (IQR) was 7.9 months (2.1–21.8) in the NIVO+IPI arm and 7.8 months (3.5–19.6) in the SUN arm. Treated patients in the NIVO+IPI arm received a median (range) of 14.0 doses (1–114) of nivolumab and 4.0 doses (1–4) of ipilimumab. Among all randomized patients, 51.8% (285/550) in the NIVO+IPI arm and 64.1% (350/546) in the SUN arm received subsequent systemic therapy. In the NIVO+IPI arm, subsequent systemic therapy included SUN (22.7%; 125/550), pazopanib (18.9%; 104/550), and axitinib (17.5%; 96/550). In the SUN arm, subsequent systemic therapy included nivolumab (38.6%; 211/546), axitinib (23.6%; 129/546), and cabozantinib (15.0%; 82/546).

In the primary efficacy population of intermediate-risk/poor-risk patients, OS was superior with NIVO+IPI versus SUN after extended follow-up (HR, 0.66; 95% CI, 0.55–0.80). The 42-month OS probability was 52% with NIVO+IPI versus 39% with SUN (figure 1A). An improvement in PFS benefit with NIVO+IPI versus SUN was observed (HR, 0.75; 95% CI, 0.62–0.90), and 42-month PFS probabilities were 33% versus 16%, respectively (figure 2A). ORR (95% CI) was 42.1% (37.4%–47.0%) with NIVO+IPI versus 26.3% (22.2%–30.8%) with SUN (table 1). In addition, a higher proportion of patients achieved a complete response (CR) with NIVO+IPI versus SUN (10.1% vs 1.4%). Similar results were observed per investigator assessment for ORR (42.4% vs 29.4%) and CR (12.2% vs 1.4%) (online supplementary table S2). Time to response was shorter and duration of response was longer with NIVO+IPI versus SUN (table 1; figure 2B). In the NIVO+IPI arm, 36/43 (83.7%) responses were ongoing in patients with CR and 85/136 (62.5%) responses were ongoing in patients with partial response (PR; table 1).

Among ITT patients (secondary efficacy population), OS was also superior with NIVO+IPI versus SUN (HR, 0.72; 95% CI, 0.61–0.86), with 42-month OS probabilities of 56% versus 47%, respectively (figure 1B). A separation in the PFS curves was evident after ~24 months (HR, 0.88; 95% CI, 0.75–1.04), and 42-month PFS probabilities were 32% versus 20%, respectively (figure 2C). ORR (95% CI) was 39.1% (35.0%–43.3%) with NIVO+IPI versus 32.6% (28.7%–36.7%) with SUN, and a higher proportion of patients achieved CR with NIVO+IPI (10.7% vs 2.4%) (table 1). Investigator-assessed response was concordant in ITT patients (online supplementary table S2). Time to response was shorter and duration of response was longer with NIVO+IPI versus SUN (table 1; figure 2D). In the NIVO+IPI arm, 51/59 (86.4%) and 95/156 (60.9%) responses were ongoing in patients with CR and PR, respectively; in the SUN arm, 12/13 (92.3%) and 82/165 (49.7%) responses were ongoing in patients with CR and PR, respectively (table 1).

Among the exploratory efficacy population of favorable-risk patients, median OS was not reached in either arm and the HR for death was 1.19 (95% CI, 0.77–1.85); 42-month OS probabilities were comparable (70% with NIVO+IPI vs 73% with SUN; figure 1C). PFS benefits were observed with SUN (HR, 1.65; 95% CI, 1.16–2.35); the 42-month PFS probabilities were 27% with NIVO+IPI versus 33% with SUN (figure 2E). At the time of database lock, 48 of 125 versus 67 of 124 favorable-risk patients assessed for efficacy were progression free with NIVO+IPI versus SUN; 8 of 124 versus 11 of 119 treated patients remained on therapy in each arm, respectively. ORR (95% CI) was 28.8% (21.1–37.6) with NIVO+IPI versus 54.0% (44.9–63.0) with SUN, however, a higher proportion of patients achieved CR with NIVO+IPI versus SUN (12.8% vs 5.6%), and median (95% CI) duration of response was not reached (40.1–not estimable) versus 27.4 months (23.5–40.3), respectively (table 1; figure 2F). Responses were ongoing in 15/16 (93.8%) patients with CR and 10/20 (50.0%) patients with PR among favorable-risk patients in the NIVO+IPI arm (table 1). Some degree of discordance between response per IRRC and per investigator was observed, with ORR per investigator comparatively higher versus IRRC assessment in favorable-risk patients with NIVO+IPI (online supplementary table S2).

To better characterize long-term outcomes in complete responders, treatment-free interval and subsequent therapy were assessed among intermediate-risk/poor-risk and favorable-risk patients in both arms (figure 3). Among all 59 complete responders in the NIVO+IPI arm, 20 (33.9%) were still on therapy, 28 (47.5%) discontinued therapy with no subsequent systemic therapy, and 11 (18.6%) discontinued and then received subsequent systemic therapy. Among all complete responders, median (range) time to response was 2.8 months (0.9–9.8) and median duration of response was not reached with NIVO+IPI. Median (range) duration of study therapy was 47.5 months (40.5–53.2) among the 20 complete responders who remained on protocol therapy. Median (range) treatment-free interval was 34.6 months (0.5–49.7) among 28 complete responders who discontinued without subsequent systemic therapy.

Exploratory post hoc analyses were conducted to assess long-term OS outcomes in ITT patient subgroups categorized by early response and treatment-related AEs at a landmark of 6 months. More evaluable ITT patients achieved a depth of response ≥50% maximal tumor reduction at 6 months with NIVO+IPI versus SUN (156/493 (31.6%) vs 65/472 (13.8%)). Similarly, more patients achieved a greater RECIST-defined response (CR or PR) with NIVO+IPI versus SUN at 6 months (198/481 (41.2%) vs 134/449 (29.8%) patients, respectively). A positive association was seen between RECIST-defined response at 6 months and OS in both treatment arms (figure 4). In the NIVO+IPI arm, OS probabilities at 42 months from randomization were 97% for patients with CR, 75% for patients with PR, 61% for patients with stable disease, and 27% for patients with progressive disease (figure 4A). Additionally, the association of AEs at 6 months with long-term survival was assessed in the NIVO+IPI arm among 493 ITT patients at risk. OS outcomes were similar between patients with immune-related AEs versus those without and were similar between patients who discontinued therapy due to any-grade treatment-related AEs versus those who did not, indicating that these AEs did not negatively affect long-term OS (figure 5).

Consistent with previous reports,3 8 similar overall rates of treatment-related AEs of any grade occurred in the NIVO+IPI and SUN arms with extended follow-up (514/547 (94.0%) vs 521/535 (97.4%) patients). Yet, there were fewer grade 3–4 treatment-related AEs with NIVO+IPI versus SUN (47.3% vs 64.1%; online supplementary table S3). Treatment-related AEs leading to discontinuation within 30 days of last dose occurred in 121 (22.1%) patients in the NIVO+IPI arm and in 69 (12.9%) patients in the SUN arm. No additional treatment-related deaths were reported since the primary analysis: 8 (1.5%) in the NIVO+IPI arm and 4 (0.7%) in the SUN arm. The incidence of any-grade and grade 3–4 treatment-related AEs by 6-month interval was consistently lower with NIVO+IPI versus SUN over time (figure 6A). The overall incidence of treatment-related select AEs with NIVO+IPI was similar to previous reports3 8 (online supplementary table S4). The temporal patterns of treatment-related select AE incidence and corticosteroid use (≥40 mg PDE) both peaked within the first 6 months of treatment with NIVO+IPI (figure 6B,C). In total, 157 (28.7%) of 547 patients treated with NIVO+IPI received corticosteroids (≥40 mg PDE) to manage any-grade treatment-related select AEs; 104 (19.0%) patients received ≥40 mg PDE continuously for ≥2 weeks and 54 (9.9%) received ≥40 mg PDE continuously for ≥30 days.

Kidney cancer–specific health-related quality of life benefits were observed with NIVO+IPI over SUN after extended follow-up, as measured using FKSI-19 questionnaires. The questionnaire completion rate was adjusted for study attrition and exceeded 80% for those who remained on protocol therapy for the duration of the study. Analyses of time to confirmed deterioration per FKSI-19 total score among intermediate-risk/poor-risk and ITT patients showed that NIVO+IPI significantly reduced the risk of worsening quality of life compared with SUN (intermediate/poor risk, HR, 0.64 (95% CI, 0.54–0.77); ITT, HR 0.64 (95% CI, 0.55–0.74)). Similarly, the risk was significantly reduced with NIVO+IPI in disease-related symptoms, physical disease-related symptoms, treatment side effects, and functional well-being domain scores (online supplementary figure S2).

These results demonstrate long-term survival benefit and durable responses with NIVO+IPI over SUN after extended follow-up of greater than 42 months. OS and ORR benefits were maintained with NIVO+IPI over SUN in intermediate-risk/poor-risk patients and in the ITT population comprising all patients, regardless of risk category. Additionally, a PFS plateau emerged after 36 months at ~33% with NIVO+IPI in both intermediate-risk/poor-risk and ITT patients, further supporting the unique durable response seen with this dual checkpoint inhibitor regimen. Among favorable-risk patients, ORR was higher and median PFS was longer with SUN; yet, the differences in OS outcomes between treatment arms were not conclusive. Additionally, the CR rate was higher in the NIVO+IPI arm, and a separation in the duration of response curves emerged after ~18 months in favor of NIVO+IPI.

More responses were durable with NIVO+IPI versus SUN across all IMDC risk categories. A greater proportion of all patients achieved a CR with NIVO+IPI, and most of these CRs were durable at the time of database lock. Notably, almost half of all complete responders experienced a treatment-free interval without initiating subsequent therapy in the NIVO+IPI arm. Additionally, a greater proportion of patients had a deep response (≥50% maximal tumor shrinkage), and RECIST-defined ORR was higher with NIVO+IPI versus SUN among evaluable ITT patients at 6 months. Favorable RECIST-defined response at 6 months was positively associated with long-term OS with NIVO+IPI.

Neither incidence of immune-related AEs nor discontinuation of therapy due to treatment-related AEs at 6 months negatively impacted long-term OS with NIVO+IPI. Interestingly, a positive trend was observed between these AEs and OS, suggesting that these early events may be indicative of immune activation and could potentially be prognostic of response and long-term survival with NIVO+IPI. Previous studies have shown that immune-related events occurring early in the course of immune checkpoint inhibitor treatment may correlate with clinical benefit in several malignancies.14–17

No new safety signals were observed, and the overall incidence of treatment-related events with longer follow-up was similar to previous rates.3 8 Looking at treatment-related AEs by 6-month interval, the overall incidence decreased over time in both arms, yet consistently higher rates of any-grade and grade 3–4 AEs occurred with SUN versus NIVO+IPI. The incidence of treatment-related select (potentially immune-mediated) AEs and corticosteroid use (≥40 mg PDE) was highest within the first 6 months of treatment with NIVO+IPI before decreasing over time, and overall rates of both were similar to previous reports.8 Additionally, intermediate-risk/poor-risk and ITT patients (comprising all IMDC risk, including favorable-risk patients) reported health-related quality of life benefits with NIVO+IPI versus SUN, as measured by time to deterioration of FKSI-19 total and multiple domain scores.

CheckMate 214 was not designed to assess outcomes in each efficacy population with equal power nor was it designed to assess outcomes within treatment arms across individual-risk groups. The analyses of the IMDC favorable-risk subgroup were limited by the comparatively small number of patients and wide 95% CIs observed. Additionally, response outcomes in this analysis were aligned with the primary analysis and focused on IRRC assessment, which differed from the most recent interim analyses after 30 months’ minimum follow-up.8

Taken together, after an extended minimum follow-up of 42 months, OS benefits were maintained with NIVO+IPI versus SUN in both intermediate-risk/poor-risk and ITT patients, and OS results were not conclusive in the favorable-risk subgroup. Responses with NIVO+IPI were generally deep and durable regardless of IMDC risk-based prognosis. CR rates were higher, and a comparatively high proportion of ITT patients remained progression-free at 42 months with NIVO+IPI versus SUN. In summary, these results support NIVO+IPI as a first-line treatment option with the potential for durable response in patients with aRCC.