Venetoclax is licensed to treat relapsed or refractory (R/R) chronic lymphocytic leukaemia (CLL). As part of the Single Technology Appraisal (STA) ID944, the National Institute for Health and Care Excellence (NICE) invited AbbVie, the manufacturer, to submit evidence on the use of venetoclax, within its licensed indication. The Evidence Review Group (ERG), Warwick Evidence, was asked to provide an independent and critical review of the submitted evidence. Evidence came from three single-arm trials in CLL patients with or without 17p deletion [del(17p])/TP53 chromosomal abnormalities. The anticipated licensed indication specified that venetoclax-eligible del(17p)/TP53 patients should have not responded to, or be deemed unsuitable for, B-cell receptor inhibitor (BCRi) therapy, and that non-del(17p)/TP53 patients should have not responded to both chemoimmunotherapy and BCRi therapy. The three trials were heterogeneous in terms of both del(17p)/TP53 status and previous exposure to BCRi therapy. The M13-982 study investigated 158 R/R CLL patients with the 17p deletion, but only a small number had received previous BCRi therapy; the M12-175 study investigated 67 patients with CLL or small lymphocytic lymphoma, some with the 17p deletion, but very few previously treated with BCRi therapy; and the M14-032 study included 105 patients previously treated with BCRi therapy (either idelalisib or ibrutinib), some of whom had unknown mutation status. The ERG concluded that the study populations did not directly conform to those specified in the licensed indication or in the NICE scope. Outcomes reported included overall response rate (ORR), duration of response, progression-free survival (PFS) and overall survival (OS); adverse events were reported for the pooled population of all three studies, as well as separately for each study. The median PFS was 41.4 and 27.2 months among patients in the M12-175 and M13-982 trials, respectively, whereas the median PFS was not reached in the M14-032 trial. Some results were designated academic in confidence and cannot be reported here. The submission provided a de novo partitioned survival cost-effectiveness model with three health states: pre-progression, post-progression and dead. Transition probabilities between health states were estimated using Weibull models for PFS and OS. The ERG judged the model structure to be appropriate. Venetoclax was compared with best supportive care (BSC) in patients with or without del(17p)/TP53 mutation status, and with palliative care (PC). To populate the del(17p)/TP53 venetoclax arm, the submission pooled del(17p)/TP53 patients from all three studies and fitted Weibull models for PFS and OS. PFS and OS models for non-del(17p)/TP53 venetoclax patients were obtained by applying hazard ratios (HRs) to the del(17p)/TP53 OS and PFS models, derived using Cox's regression analysis comparing del(17p)/TP53 and non-del(17p)/TP53 patients pooled from the M14-032 and M12-175 studies. The ERG expressed reservations about the company's pooling procedure, but acknowledged its expedience given the small evidence base. For the BSC comparator arm, the submission used the rituximab + placebo arm from a randomised controlled trial comparing idelalisib + rituximab versus placebo + rituximab ('study 116'). Weibull regression data for OS and PFS were taken from the idelalisib STA (ID764) submitted by Gilead to NICE. The ERG considered the use of the study 116 rituximab arm to be inconsistent with the licensed indication for venetoclax because these patients had neither not responded to nor were inappropriate for BCRi therapy, being eligible to be randomised to idelalisib. Another difficulty was the requirement for a technical correction in survival analysis because of considerable switching from rituximab to idelalisib. The ERG considered that post-progression survival of patients from the idelalisib arm of study 116 provided a more appropriate representation of BSC since these patients had not responded to BCRi therapy, consistent with venetoclax's licensed indication. For PC, the company submission used data from the UK CLL Forum. The company's base-case analysis indicated that venetoclax was clinically effective, but the resulting incremental cost-effectiveness ratios (ICERs) for del(17p)/TP53 (A 39,940 pound/quality-adjusted life-year [QALY] gained) and non-del(17p)/TP53 (A 47,370 pound/QALY gained) patients were well above the NICE threshold of A 20,000-30,000 pound/QALY. The ERG identified two errors in the implementation of the company's parametric models-one related to the implementation of HRs, and the other to the derivation of the Weibull shape parameters obtained from the Gilead idelalisib submission. The ERG made plausible adjustments to the company's base-case and corrected errors, resulting in a reduced estimate of the cost effectiveness of venetoclax in non-del(17p)/TP53 and del(17p)/TP53 indications; in the ERG's preferred base case, using post-progression survival of patients in the idelalisib arm of study 116 as the BSC comparator, deterministic ICERs were higher than the company's base-case for both indications: A 57,476 pound/QALY gained for del(17p)/TP53 and A 77,779 pound/QALY gained for non-del(17p)/TP53. The NICE Appraisal Committee's preliminary recommendation was that venetoclax used within its licensed indication should not be recommended for use in the National Health Service (NHS). In response to the preliminary recommendation, the company submitted new analyses; however, at a subsequent appraisal committee meeting, the original recommendation was upheld and the committee concluded there were large uncertainties around the clinical effectiveness of venetoclax and BSC, and that under the committee's preferred assumptions, the ICERs were higher than those generally considered cost effective, even when end-of-life criteria were taken into account. The company submitted further evidence, and the final guidance recommended venetoclax for use with the Cancer Drugs Fund for the two populations in this technology appraisal.
