Correspondence on 'Cost-effectiveness of transcatheter edge-to-edge repair in secondary mitral regurgitation does need confirmation' by Cohen et al

To the Editor: we have read with considerable interest the paper by Cohen et al estimating the costeffectiveness of the Mitraclip system in patients with secondary mitral regurgitation (SMR). Like other published works that adopted different healthcare perspectives, including the one by Baron et al, the costeffectiveness analysis conducted by Cohen et al was based on 2year data from the Coapt randomised controlled trial (RCT). Generating lifetime estimates of survival gain (1.57 years here) from the 2year data of Coapt requires extensive extrapolation of about 13 years beyond observed data and >95% of benefit reported in the percutaneous repair (PR) arm accrues in the extrapolation phase rather than the observation phase. The observed source data used for extrapolation can therefore exert a profound influence on estimation of gained benefit from PR. We were surprised that Cohen et al chose not to consider in their economic model the 3year data from Coapt which were released as oral presentation in 2019 and fully published in late 2020. Indeed, the 3year allcause mortality curve for the Mitraclip arm of Coapt reported by Mack et al 5 indicates a doubling in mean mortality rate in years 2–3 relative to years 1–2 (estimated using area under the curve as is done in costeffectiveness analysis). The upturn in mortality during years 2–3 in the PR arm is similarly reflected in the cumulative percentage mortality reported at years 1, 2 and 3 of 19%, 28.2% and 42.8%, respectively, that translates to a crude estimate of annual mortality rates of 19% over the first year, 9.2% over years 1–2, and 14.6% over years 2–3; an increase of 59% in rate for years 2–3 relative to years 1–2. It is therefore evident that using 3year mortality data instead of 2year will considerably influence estimated survival gains accrued in economic models. Consequently, we believe that the lifetime extrapolation beyond the observed 2year data is highly optimistic in the work by Cohen et al, particularly for the intervention arm, and that this has potential to impact the costeffectiveness in favour of the intervention. As a general principle, it would be expected that using longer rather than shorter followup results from trials is likely to reduce uncertainty in costeffectiveness estimates. This can optimise decisionmaking in territories such as the UK where costeffectiveness is a key criterion to judge recommendation of new technologies. In consequence, in our opinion, the 2year mortality data from Stone et al are likely to be unsuitable for reliable costeffectiveness analysis. Further prespecified analyses from Coapt at 4 and 5 years are eagerly awaited. Additionally, the generalisability of Cohen et al findings using US Coapt population as source of clinical inputs may be questionable since, as acknowledged by authors in the Limitation section, the inputs from the MitraFr RCT (undertaken in a French population and showing no advantage of PR relative to medical treatment alone) were not considered as an option; Cohen et al suggest in their conclusion that their results apply for patients with similar characteristics to those enrolled in Coapt, which postulates that the two trials had divergent findings due to the differences in patients’ characteristics. However, posthoc analyses from MitraFr looking for subgroups that most resembled those in Coapt failed to identify a benefit from PR. 8 These considerations indicate further investigations are desirable to identify the population most likely to derive a benefit from PR. This will help to optimise allocation of resources. The work presented by Cohen et al represents a valuable contribution; however, we respectfully disagree with the editorial conclusion drawn by Dr Garbi and Dr Mariani where they posit the costeffectiveness of PR in SMR does not need confirmation.