Abstract
Nivolumab and cabozantinib are approved agents in mRCC patients after sunitinib/pazopanib (TKI) failure. However, the optimal sequence, cabozantinib then nivolumab (CN) or nivolumab then cabozantinib (NC), is still unknown. The CABIR study aimed to identify the optimal sequence between CN and NC after frontline VEGFR-TKI. In this multicenter retrospective study, we collected data from mRCC pts receiving CN or NC, after frontline VEGFR-TKI. A propensity score (PrS) was calculated to manage bias selection, and sequence comparisons were carried out with a cox model on a matched sample 1:1. The primary endpoint was progression-free survival (PFS) from the start of second line to progression in third line (PFS2-3 ). Key secondary endpoints included overall survival from second line (OS2 ). Out of 139 included mRCC patients, 38 (27%) and 101 (73%) received CN and NC, respectively. Overlap in PrS allowed 1:1 matching for each CN pts, with characteristics well balanced. For both PFS2-3 and OS2 , NC sequence was superior to CN (PFS2-3 : HR = 0.58 [0.34-0.98], P = .043; OS2 : 0.66 [0.42-1.05], P = .080). Superior PFS2-3 was in patients treated between 6 and 18 months with prior VEGFR-TKI (P = .019) and was driven by a higher PFSL3 with cabozantinib when given after nivolumab (P < .001). The CABIR study shows a prolonged PFS of the NC sequence compared to CN in mRCC after first line VEGFR-TKI failure. The data suggest that cabozantinib may be more effective than nivolumab in the third-line setting, possibly related to an ability of cabozantinib to overcome resistance to PD-1 blockade.