OBJECTIVES :

To assess the quality and risk of bias (ROB) in the overall survival (OS) estimates generated from the pivotal trials supporting European Medicines Agency (EMA) authorisation of CAR T-cell therapies, tisagenlecleucel and axicabtagene ciloleucel.

METHODS :

The EMA Public Assessment Report (EPAR) of each drug was consulted to identify the pivotal clinical trial informing the clinical efficacy of each drug and indication. Due to the single arm nature of the trials, evidence was graded as non-randomised. The ROB associated with OS outcomes was assessed using the GRADE study limitations in observational studies criteria and the ROBINS-I tool.

RESULTS :

Three pivotal trials were identified. Two trials assessed the efficacy of tisagenlecleucel: JULIET for relapsed/refractory diffuse large B-cell lymphoma (r/r DLBCL) and ELIANA for paediatric acute lymphoblastic leukaemia (pALL). One trial assessed the efficacy of axicabtagene ciloleucel: ZUMA-1 for r/r DLBCL. Applying the GRADE criteria resulted in a grade of low for the quality of OS evidence of all three trials. This grade is warranted by the non-randomised nature of the evidence. This was further downgraded due to the high ROB of the studies, owing mainly to the immaturity of the data. When assessed using the ROBINS-I tool, OS from all three trials was deemed to have a serious ROB. This is attributable to bias due to confounding and bias in selection of participants into the study.

CONCLUSIONS :

OS data supporting EMA authorisation of CAR T-cell therapies is considered to be of low quality and has a serious ROB when assessed using the GRADE study limitations in observational studies criteria and the ROBINS-I tool, respectively. Using these studies to inform health technology assessment leads to considerable uncertainty in the cost-effectiveness estimates.