Subgroup analysis of the AFTER I-O study: a retrospective study on the efficacy and safety of subsequent molecular targeted therapy after immune-oncology therapy in Japanese patients with metastatic renal cell carcinoma

Abstract Background We performed subgroup analyses of the AFTER I-O study to clarify the association of time-to-treatment failure (TTF) and discontinuation reason of prior immune-oncology (I-O) therapy, and molecular targeted therapy (TT) regimen with the outcomes of TT after I-O. Methods The data of Japanese metastatic renal cell carcinoma patients treated with TT after nivolumab (NIVO) (CheckMate 025) or NIVO + ipilimumab (IPI) (CheckMate 214) were retrospectively analyzed. The objective response rates (ORRs), progression-free survival (PFS) and overall survival (OS) of TT after I-O were analyzed by subgroups: TTF (<6 or ≥6 months) and discontinuation reason of prior I-O (progression or adverse events), and TT regimen (sunitinib or axitinib). We also analyzed PFS2 of prior I-O and OS from first-line therapy. Results The ORR and median PFS of TT after NIVO and NIVO+IPI among the subgroups was 17–36% and 20–44%, and 7.1–11.6 months and 16.3-not reached (NR), respectively. The median OS of TT after NIVO was longer in patients with longer TTF of NIVO and treated with axitinib. Conversely, median OS of TT after NIVO+IPI was similar among subgroups. The median PFS2 of NIVO and NIVO+IPI was 36.7 and 32.0 months, respectively. The median OS from first-line therapy was 70.5 months for patients treated with NIVO and NR with NIVO+IPI. The safety profile of each TT after each I-O was similar to previous reports. Conclusions The efficacy of TT after NIVO or NIVO+IPI was favorable regardless of the TTF and discontinuation reason of prior I-O, and TT regimen.


Introduction
The prognosis of patients with metastatic renal cell carcinoma (mRCC) has dramatically improved in the immuno-oncology (I-O) era than in the cytokine (1) and molecular targeted therapy (TT) eras (2)(3)(4). Many treatment options have been approved as firstline mRCC therapy, such as I-O combination therapies including nivolumab and ipilimumab (NIVO+IPI), pembrolizumab and axitinib, avelumab and axitinib, and classical vascular endothelial growth factor receptor tyrosine kinase inhibitor (VEGFR-TKI) monotherapies. Moreover, as for second-or later-line therapy, nivolumab (NIVO), cabozantinib, axitinib and everolimus have been approved. Among these treatment options, the design of sequential therapy is an important factor for the prognosis of mRCC patients; however, the lack of diversity of mechanisms of action for treating mRCC causes difficulties regarding switching the mechanism of action between treatment lines.
Iacovelli et al. (16) only reported the correlation between the outcomes of I-O treatment and efficacy of subsequent therapy. Furthermore, certain reports are available regarding the efficacy of specific TT after I-O regimens (10,13,15,16,(18)(19)(20).
In this study, we analyzed the outcome of the AFTER I-O study by subgroups: length of time-to-treatment failure (TTF), reason for discontinuation of NIVO or NIVO+IPI, and first TT regimen after NIVO or NIVO+IPI (sunitinib or axitinib). Moreover, we made additional analyses on progression-free survival 2 (PFS2) after NIVO and NIVO+IPI and overall survival (OS) from first-line therapy to clarify the long-term benefit of NIVO and NIVO+IPI.

Patients and methods
The 'AFTER I-O study' was a multicenter, retrospective, observational study conducted in Japan. This study analyzed patients that participated in the CheckMate 025 or CheckMate 214 trials and were treated with TT as a subsequent therapy before 31 March 2019, and after the discontinuation of NIVO or NIVO+IPI. The primary endpoints were the objective response rates (ORRs) to the first TT after discontinuation of NIVO or NIVO+IPI. The secondary endpoints included the efficacy of TT after NIVO or NIVO+IPI, such as PFS, OS and safety. Additionally, we analyzed PFS2 after NIVO and NIVO+IPI and OS from first-line therapy. PFS was defined as the time from the first TT dose after NIVO or NIVO+IPI to disease progression (PD) or death. PFS2 was defined as the time from the first dose of NIVO or NIVO+IPI to PD or death during subsequent therapy with the first TT after NIVO or NIVO+IPI. The OS of TT was defined as the time from the first TT dose after NIVO or NIVO+IPI to death and the OS of first-line therapy as the time from first-line therapy to death.
The AFTER I-O study was approved by the Ethics Committee of Niigata University (Approval Number: 2018-0416, Date: 10 April 2019) and other independent institutional review boards and was conducted according to the Declaration of Helsinki and Ethical Guidelines for Medical and Health Research Involving Human Subjects. This study is registered in the University hospital Medical Information Network under number UMIN000036063. This retrospective study used medical records for analysis, and thus, informed consent from patients was not required. This paper does not disclose any personally identifiable information of any of the participants in any form. Hence, consent for publication is not applicable.

Statistical analyses
OS, PFS and PFS2 were estimated by the Kaplan-Meier method, and the 95% confidence intervals for each subgroup were determined using hierarchical Bayesian survival analysis and Cox's proportional hazards model. SAS (SAS Institute Japan Ltd., version 9.4) was used for all analyses. The efficacy of the first TT after discontinuation of NIVO or NIVO+IPI was analyzed in the next subgroups: TTF of NIVO or NIVO+IPI, cutoff at 6 months; reason for discontinuation of NIVO or NIVO+IPI, PD or adverse event; regimen of TT after discontinuation of NIVO or NIVO+IPI, sunitinib or axitinib, and safety were analyzed in subgroups of TT regimens, sunitinib or axitinib. Statistical differences were not tested between any subgroups due to the small size of the study.

Results
The patient characteristics are summarized in Table 1, and patient characteristics of intermediate/poor risks of CheckMate 214 are summarized in Table S1. A total of 45 mRCC patients from 20 Japanese centers were retrospectively analyzed, including 26 out of 37 Japanese patients treated with NIVO in CheckMate 025 (23,24) and 19 out of 38 Japanese patients (all risks) treated with NIVO+IPI in CheckMate 214 (25). The median follow-up period from the start of the first TT after discontinuation of NIVO or NIVO+IPI to the date of analysis or death was 22 The ORR and BOR of all the patients and their subgroups are summarized in Table 2 for the patients from CheckMate 025 and in Table 3 for the patients from CheckMate 214 (all risks). Surgery after ICI discontinuation, n (%) Intermediate 14     (Table 3). The median PFS of TT after NIVO varied from 7.1 to 11.6 months among subgroups (Fig. 1a-c) and from 16.3 months to not reached (NR) for TT after NIVO+IPI (Fig. 1d-f). The median OS of TT after NIVO was longer in patients who had a longer TTF of NIVO (47.1 vs. 20.8 months) and in patients treated with axitinib vs. sunitinib (37.9 vs. 22.1 months). Conversely, the median OS of TT after NIVO+IPI was relatively similar among subgroups (Fig. 2d-f). The median PFS2 of NIVO and NIVO+IPI was 36.7 and 32.0 months, respectively (Fig. 3). The median OS from first-line therapy was 70.5 months for patients treated with NIVO and NR for patients treated with NIVO+IPI (Fig. 4). The median PFS2 and OS from first-line therapy of patients treated with NIVO were longer in patients treated with axitinib after NIVO than in patients treated with sunitinib (39.8 vs. 14.7 months and 92.0 vs. 54.4 months , Figs 3c and 4c).
The safety data are summarized in Table 4. All patients treated with sunitinib after NIVO or NIVO+IPI and almost all patients treated with axitinib experienced treatment-related adverse events, of which grade 3-4 events were more common in patients treated with sunitinib than in those treated with axitinib. Finally, no treatmentrelated deaths and no new safety signals were reported.

Discussion
This subgroup analysis of the AFTER I-O study revealed that the efficacy of TT after NIVO or NIVO+IPI was promising, regardless of the TTF of NIVO or NIVO+IPI, reason for discontinuation of I-O therapy, or TT regimen, sunitinib or axitinib.
To date, three studies have reported that patients with short PFS or TTF of first-line VEGFR-TKI had poorer prognoses (4,26,27). In this subgroup analysis, PFS and OS of TT after first-line NIVO+IPI were similar among subgroups of TTF cut-off at 6 months (Figs 1d and 2d). However, the median OS of TT after NIVO+IPI with a short TTF of NIVO+IPI (30.5 months) was longer than the OS of the RECORD-1 (everolimus, median OS: 14.8 months), AXIS (axitinib after subitinib, median OS: 15.2 months) and METEOR (cabozantinib, median OS: 21.4 months) trials (28)(29)(30), and real world data of axitinib in Japan reported by Miyake et al. (median OS: 27.0 months) (31).
As for TT-TT sequential therapy, many studies have reported patients who discontinued TT due to PD had poorer PFS and OS than those that discontinued due to adverse events (32)(33)(34)(35). Such a trend was not present in our current analysis of either TT after NIVO ( Table 2, Fig. 1b and e) or NIVO+IPI (Table 3, Fig. 2b and e), indicating the advantage of changing the mechanism of action between treatment lines.
Ishihara et al. (18) reported the efficacy of third-line axitinib after second-line NIVO (ORR: 29.4%, median PFS: 12.8 months, median OS: NR), and Yasuoka et al. (20) also reported the efficacy of third-or fourth-line axitinib after NIVO (ORR: 56.3%, median PFS: 7.9 months, median OS: NR). This analysis is the third that reports favorable efficacy of axitinib after NIVO. There are no other available reports regarding the efficacy of sunitinib or axitinib after NIVO+IPI.
Although the median OS of sunitinib was shorter than that of axitinib after NIVO (22.1 vs. 37.9 months, Fig. 2c), the median OS of sunitinib after NIVO was comparable or longer than that in the pivotal studies, such as RECORD-1, AXIS and METEOR   (28)(29)(30). The median PFS of axitinib after NIVO (10.1 months) and NIVO+IPI (16.3 months) was similar to the Japanese subgroup analysis of the AXIS study which included first-line cytokine and TT therapy (median PFS: 12.1 months) (36).
The PFS2 and OS from the first-line therapies of the RECORD-3, SWICH, SWICH II and CROSS-J-RCC trials were previously reported for TT sequential therapy (37)(38)(39)(40). The JAVELIN Renal 101 study was the only previous report of PFS2 in the I-O era (41). In the present study, the median PFS2 (36.7 and 32.0 months for patients from CheckMate 025 and CheckMate 214, respectively) and the median OS from first-line therapy (70.5 months for patients from CheckMate 025) were also longer than the previous reports on TT sequential therapy (median PFS2: 8.6-27.8 months; median OS from first-line therapy: 22.4-38.9 months). The median PFS2 was longer for patients from CheckMate 025 than CheckMate 214, probably because the patients from CheckMate 214 with long PFS after NIVO+IPI were not treated with subsequent TT by the time of analysis. The median OS from first-line therapy of patients treated with axitinib after NIVO exceeded 7 years, which is a remarkable data. In the I-O era, systemic therapy may result in OS of more than 7 years, which will be encouraging for the patients of mRCC.
The safety profile of sunitinib or axitinib after NIVO and NIVO+IPI was consistent with previous reports and without new safety signals.
The AFTER I-O study had some limitations: first the nature of the retrospective study, and second, the small sample size; as each subgroup was too small, comparisons between subgroups should be performed with caution.
In conclusion, the efficacy of TT after NIVO or NIVO+IPI was favorable independent of the TTF and the reason for discontinuation of NIVO or NIVO+IPI, and TT regimen.

Data availability statement
The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Supplementary Material
Supplementary material can be found at Japanese Journal of Clinical Oncology online.

Author contributions
Tomita, Kaneko, and Tajima have full access to all the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis. Study concept and design: Tomita, Kaneko, and Tajima. Provision of study materials or patients: Tomita, Kimura, Fukasawa, Numakura, Sugiyama, Yamana, Naito, and Oya.
Collection and assembly of data: Tomita, Kaneko, and Tajima. Data analysis and interpretation: All authors. Drafting of the manuscript: Tomita, Kaneko, and Tajima. Critical revision of the manuscript for important intellectual content: All authors.
Final approval of manuscript: All authors.