Real-world and randomized clinical trial comparison in patients with m <i>IDH1</i> CCA treated with ivosidenib: ProvIDHe vs ClarIDHy.

Abstract

e16173 Background: Cholangiocarcinoma (CCA) is a rare disease, and most patients are diagnosed at an advanced stage. Ivosidenib is approved in many geographies for the treatment of IDH1 mutated CCA in second and later lines, based on ClarIDHy, a phase 3 randomized trial. More recently, ivosidenib has been investigated in the real-world phase 3b, open-label ProvIDHe study. This work aims to compare the effectiveness of ivosidenib in the RCT and real-world settings of these studies. Methods: Heterogeneity between the studies was assessed and characteristics likely to impact patient outcomes or treatment effect were validated by independent experts. Only patients with metastatic disease were included to homogenize patient populations across the two studies. To mitigate potential sources of bias in the comparison, a propensity score weighting model was implemented to align the ProvIDHe population with ClarIDHy with respect to the following characteristics: ECOG 0; one prior regimen of therapy; prior surgery; and prior cisplatin and gemcitabine. These were informed by clinician input and assessment of data availability. Using this adjusted data, overall survival (OS) and investigator-defined progression-free survival (PFS) were compared by estimating a hazard ratio (HR) using a weighted Cox proportional hazards (PH) model. Due to violations in the PH assumption for OS and PFS, Kaplan-Meier estimates of survival rates and restricted mean survival times (RMST) were compared. Results: After reweighting, 6- and 12-month PFS rates were 35.9% and 26.4% in the weighted ProvIDHe vs 29.8% and 10.0% in ivosidenib arm from ClarIDHy. The HR in PFS between ProvIDHe and ClarIDHy was 0.63 (95% confidence interval [CI]: 0.45, 0.90) indicating a marked improvement in PFS for those treated with ivosidenib in real-world. RMST estimates at 6 and 12 months for PFS were 3.90 and 5.69 months in the weighted ProvIDHe population, compared with 3.24 and 4.37 months in ClarIDHy. While no statistically significant difference in OS was identified (HR 0.75; 95% CI: 0.47, 1.21), OS was numerically superior in the weighted ProvIDHe population particularly at later timepoints, with 6-, 12- and 18-month OS rates of 74.0%, 57.9% and 42.8% compared to 67.5%, 42.9%, and 28.5% in ClarIDHy (RMST for OS after 18-months follow-up: 11.85 months vs 10.48 months). However, the lack of statistical significance surrounding this result may be due to the limited follow-up available for OS in the ongoing ProvIDHe study. Conclusions: Following alignment of the ProvIDHe and ClarIDHy populations, patients treated with ivosidenib in the real-world setting of ProvIDHe had significantly improved PFS and numerically improved OS compared to those who received ivosidenib in the RCT setting of ClarIDHy. This suggests that ivosidenib may provide greater benefit in the real-world than previously demonstrated in the RCT setting of the ClarIDHy study. Clinical trial information: NCT02989857 ; NCT05876754 .