Predicting Overall Survival in Patients (pts) with Treatment-Naive Metastatic Merkel Cell Carcinoma (mmcc) Treated with Avelumab.

e21620 Background: Data from the JAVELIN Merkel 200 (JM 200) trial of avelumab in mMCC with a minimum follow-up of 18 mo for 88 pts who had received prior chemotherapy demonstrated robust survival estimates. Available data for treatment-naive pts are less mature due to partial enrolment (Mar 24, 2017 data cut: 6 wk min follow-up, 39 pts enrolled). To enable treatment reimbursement decisions, health technology assessment agencies require long-term survival data. To provide this, the current analysis used novel statistical extrapolation methods (based on JM 200 data) to predict overall survival for treatment-naive pts with mMCC treated with avelumab. Methods: Flexible parametric models using data from both the prior treatment and treatment-naive populations were fitted, and a stitching analysis was used to combine extrapolations. Resulting survival estimates were compared with observed survival data in both populations, and the extrapolated mean survival time was calculated. Results: Extrapolation using early data from the treatment-naive cohort demonstrated that the hazard of death was greater for prior treatment pts than for treatment-naive pts after 21 mo. Assuming the same extrapolation for treatment-naive pts as for prior treatment pts produced more realistic estimates of the hazard of death after 21 mo but underestimated survival for treatment-naive pts before this time. The stitching analysis avoided the aforementioned issues, although it may still underestimate survival given the reliance on longer-term prior treatment data. Survival estimates are in the table. Conclusions: The use of novel extrapolation methods quantifies the expected long-term survival associated with avelumab in treatment-naive pts, which is expected to be at least as good as the prior treatment population despite the lack of long-term data. While survival extrapolations are promising, maturing data from the ongoing JM 200 trial will aid validation of these estimates. Extrapolation Method Survival Estimate Mean (2.5th, 97.5th percentiles), mo 1-y, % 5-y, % Treatment-naive data 49.9 (6.3, 179.4) 66 23 Prior treatment data 42.3 (28.4, 77.4) 51 19 Stitching analysis 54.9 (27.4, 70.6) 66 25