OBJECTIVES: To estimate the long-term survival and the cost-effectiveness of relmacabtagene autoleucel (relma-cel) anti-CD19 autologous CAR-T cell immunotherapy product vs salvage chemotherapy for treating Chinese relapsed or refractory diffuse large B-cell lymphoma (r/r DLBCL) patients.

METHODS: A decision tree and a partitioned survival model were developed from the perspective of Chinese healthcare system with a lifetime horizon. A standard parametric survival model and a mixture cure model were built based on data from CORAL and RELIANCE trials to extrapolate the survival of patients with salvage chemotherapy and with relma-cel, respectively. The best fit model for each curve was determined according to AIC/BIC and the clinicians’ opinions. All utility values were sourced from published literature and the RELIANCE trial. Medical resource utilisations were estimated by experienced clinicians via an in-depth interview. Both one-way and probabilistic sensitivity analyses were conducted.

RESULTS: A log-logistic model for salvage chemotherapy and an exponential model for relma-cel were respectively selected to predict the long-term survival and progression-free survival of the two arms. The cure fraction for patients with relma-cel was estimated to be 38.1% representing those who will not experience disease progression or die from r/r DLBCL. The 5-year survival rates for chemotherapy and relma-cel were predicted to be 7.2% and 45.5%, respectively. Patients treated with relma-cel were estimated to gain 5.26 more QALYs and cost ¥1,067,430 more than those with chemotherapy. Hence, the incremental cost-effectiveness ratio was ¥203,137 per QALY. According to sensitivity analyses, the model was the most sensitive to the mixture cure model parameters, which were estimated based on the RELIANCE trial data, and the probability of relma-cel being cost-effective under a 3-time-GDP threshold was 74%.

CONCLUSIONS: For Chinese r/r DLBCL patients, relma-cel was associated with a significant QALY gained. The uncertainty exists, which however may be reduced by a longer follow-up of the RELIANCE trial.