Over the past two decades, systemic therapy for mRCC has undergone several revolutions, first with the introduction of targeted therapies such as TKIs, later with antibodies targeting the programmed death receptor axis (PD-1, IO therapies), and finally by combining them (Escudier 2019). The aim of the current study was to investigate the systemic therapies, their use over time and to assess PFS and OS rates in the real world setting of an academic institution. A retrospective study cohort of 201 mRCC patients treated with either IO-IO, TKI-IO combinations, or TKI monotherapy at our reference center is presented, starting in January 2006.
First, our heterogeneous cohort of patients treated with a variety of therapies over the past 2 decades reflects the current mRCC reality. It is evident that newly approved agents have been rapidly implemented into routine practice. As of 2017, 57.8% of patients have received ipilimumab plus nivolumab, 26.7% a TKI-IO based combination, and only 15.5% are still on TKI monotherapy for first-line mRCC treatment. An analysis of the 2018 German RCC clinical registry showed the rapid implementation of IO therapy with only 1% of patients treated with first-line IO-based therapy between 2015 and 2017, whereas nearly 20% received second-line IO-based therapy which was recently approved by that time (Goebell et al. 2018). With the plethora of first-line combination therapies, the choice of therapy is becoming increasingly difficult. As randomized phase III clinical trials directly comparing the spectrum of new first-line options are unlikely to be conducted, real-world data will be critical to guide clinical, policy, and financial decisions. However, patients in routine practice differ from patients treated in clinical trials primarily by lower performance status and a higher number of impairing comorbidities(Marschner et al. 2017, Mitchell et al. 2015, Heng et al. 2014).
Second, we showed that investigator assessed PFS was significantly improved for first-line TKI-IO combination compared to TKI monotherapy as well as versus IO-IO combination (p=0.034, p=0.012). Multivariate analysis confirmed the PFS benefit of TKI-IO versus IO-IO combination (p=0.023). The PFS of 23.9 months with TKI-IO combinations after a median follow-up of 24.1 months obtained in our study compared favorably with that observed in phase III clinical trials. To date, all phase III clinical trials of TKI-IO combinations have demonstrated improved PFS over comparator sunitinib across all IMDC risk groups, for axitinib plus pembrolizumab (Keynote 426: 15.7 months with median follow-up 42.8 months, HR 0.68, p<0.0001), for axitinib plus avelumab (Javelin renal 101: 13.8 months (PD-L1+ subgroup) with median follow-up 19 months, p<0.0001), for cabozantinib plus nivolumab (CheckMate 9ER: 17.0 months with median follow-up 23.5 months, HR 0.52, p<0.0001), and for lenvatinib plus pembrolizumab (CLEAR: 23.9 months with median follow-up 33.4 months, p<0.001) (Powles et al. 2020, Rini et al. 2021, Choueiri et al. 2020, Choueiri et al. 2021, Motzer, Choueiri, et al. 2021, Motzer, Alekseev, et al. 2021, Choueiri T 2021). The PFS of 6.1 months for ipilimumab plus nivolumab observed in our cohort is much shorter as the PFS observed in the Checkmate 214 trial (11.2 months) (Albiges et al. 2020). This could be on the one hand related to either the investigator assessment and not a blinded independent review as done in phase III clinical trials, on the other a different risk group balance between our real-world cohort and the clinical trial setting. In addition, a recent retrospective, multi-institutional cohort analysis also demonstrated a shorter median PFS after initiation of IO-based treatment of 9.8 months (Hoeh et al. 2022).
Third, in our cohort, a numerical improvement in OS was observed with first-line IO-based therapy over TKI monotherapy (NR vs. 2.64 years, HR 0.65, p=0.099). The lack of significant difference may be due to the difference in follow-up time between the study groups: 1.1 years for the IO-based group versus 2.5 years for the TKI monotherapy group. For the same reason, OS was numerically improved for the TKI-IO combination compared with TKI monotherapy, but failed to reach the threshold of statistical significance (NR vs. 2.64 years, HR 0.37, 95% CI 0.13-1.0, p=0.050). Median OS was not achieved for first-line therapy with IO-IO or TKI-IO combinations, and although there was a benefit for the TKI-IO combination with a HR of 0.45, this is likely related to the short follow-up period (p=0.183). Furthermore, exploratory multivariable analysis confirmed a trend toward better OS for first-line TKI-IO combination compared to TKI monotherapy (HR 0.38, p=0.065). Regarding phase III clinical trial results, TKI-IO combinations lead to an OS benefit compared to sunitinib with in Keynote-426 median 45.7 months (ITT population, median follow-up 42.8 months, HR 0.73, p=0.001); in CheckMate 9ER not achieved (ITT population, median follow-up 23.5 months, HR 0.66, p=0.0034); and in CLEAR not achieved (ITT population, median follow-up 33.4 months, HR 0.72, p=0.005) (Powles et al. 2020, Rini et al. 2021, Motzer, Alekseev, et al. 2021, Choueiri T 2021, Choueiri et al. 2021, Motzer, Choueiri, et al. 2021). For axitinib plus avelumab, there is still no significant impact on OS at a median follow-up of 19 months in the Javelin Renal 101 trial (NR, HR 0.83, p=0.1301) (Motzer et al. 2019, Choueiri et al. 2020).
Our analysis points out that the OS benefit of TKI-IO combinations as observed in clinical trial cohorts also applies to this cohort of patients treated in real-world clinical practice, confirming the applicability of these findings on a broader basis. However, we were unable to confirm the significant improvement of median OS of 48.1 months (HR 0.65, p<0.0001) observed in the Checkmate 214 trial of ipilimumab plus nivolumab in intermediate/low-risk IMDC patients in our cohort (HR 0.84, p=0.557), so far (Albiges et al. 2020).
Regarding differences in OS within IO-based therapies, our data show that median OS for first-line IO-IO or TKI-IO combinations is not yet mature and needs long follow time to draw definitive conclusions. Although a non-significant improvement was demonstrated for the TKI-IO combination with a HR of 0.45 (p=0.183), interpretation should be made with caution due to the limited number of events in the IO-based group. Our results are similar to findings from recently published studies. In a retrospective cohort study from the National Cancer Database, it was found that in patients with clear cell mRCC, both IO-IO and TKI-IO combinations were associated with significantly better OS than TKI monotherapy (for IO-IO group: HR 0.60, p<0.001; for the TKI-IO combination group: HR 0.74, p=0.005), with no survival difference observed between the TKI-IO and IO-IO combination (HR 1.24, 95% CI, 0.98-1.56, p=0.08) (Chakiryan et al. 2021). Moreover, in the aforementioned retrospective multi-institutional cohort analysis with real-world data, OS rates were comparable between first-line IO-IO and TKI-IO treatment approaches, with OS rates at 12 months of 73.9 versus 90.0% (p=0.089), respectively (Hoeh et al. 2022). Results from the International Metastatic Renal-cell Carcinoma Database Consortium (n=113 for TKI-IO combinations, n=75 for ipilimumab plus nivolumab) also confirmed the non-significant differences in OS between first-line IO-IO and TKI-IO treatment (median OS NR vs. NR, after adjustment for IMDC risk factors p=0.14) (Dudani et al. 2019).
Forth, given the routine use of nivolumab in mRCC patients from second-line onward, we also provide a qualitative assessment of the efficacy of this IO monotherapy after first-line in our cohort. Our collective reflects past reality as all patients received prior TKI monotherapies or mTOR inhibitors followed by nivolumab from second-line. Our data show a median PFS of 5.5 months for second-line treatment with nivolumab compared to 7.9 months for second-line treatment with TKI monotherapy (p=0.91). The addition of nivolumab either in the second-line or beyond significantly improved OS compared to a TKI or mTOR therapy alone (≥2 TKIs; median 6.13 vs. 2.61 years, HR 0.46, p=0.003). Regarding the associated pivotal study, the 5-year analysis of CheckMate 025 comparing nivolumab to everolimus in mRCC patients previously treated with 1-2 antiangiogenic therapies confirmed the superior efficacy of nivolumab over everolimus. At a minimum follow-up of 64 months (median 72 months), PFS favored nivolumab with 4.2 months (HR 0.84, p=0.0331). In addition, nivolumab maintained an OS benefit over everolimus (median 25.8 vs. 19.7 months; HR 0.73, p<0.0001), with 5-year OS probabilities of 26% and 18%, respectively. Both PFS and OS in our analysis were comparable to those in the Checkmate 025 trial, although patients in our analysis received a second-line TKI rather than the mTOR inhibitor everolimus as in the Checkmate 025 trial (Motzer et al. 2020). Real-world data on the use of nivolumab in 228 mRCC patients from the non-interventional NORA study showed that efficacy and safety in this real-world population were consistent with the pivotal clinical trial after a median follow-up of 37 months, with median PFS of 5.3 months and OS of 24 months (Grimm et al. 2021). These findings support the use of nivolumab after prior TKI monotherapy in mRCC patients.
Our analysis has several limitations. There is an inherent risk of selection bias in retrospective comparative effectiveness studies. In addition, sample size is limited and follow-up time, especially for the IO-based therapies, is variable and in some cases limited. Consequently, clinically meaningful differences for the end points in the IO combination groups cannot be excluded by the present analysis. Further reports with more mature data are warranted. Other important limitations include the lack of collecting toxicity data and response rates, which can only be indirectly measured by the PFS time interval from the start of systemic therapy. With regard to PFS, it should also be mentioned that tumor assessment in routine care was partly not performed according to the Response Evaluation Criteria In Solid Tumors (RECIST) used in clinical trials.
Another potential selection bias may occur as no standardized protocol was used for decision making regarding treatment regimens. However, this may make these data more reflective of real-world experience. Overall, the IO-based and TKI monotherapy cohorts are quite homogeneous in several patient and tumor characteristics, but there are still some differences.
Finally, the novelty of the study is limited, as numerous similar and larger studies have been published in recent years. Nevertheless, the data have value as they reflect the reality of the rapidly evolving therapeutic landscape for mRCC at a major urologic center and provide comparable results to pivotal trials.