Prediction of cardiovascular events by algorithm- and formula-based pulse wave velocity

Background: Carotid-femoral pulse wave velocity (PWV), a marker of arterial stiffness, is a recognized cardiovascular disease risk factor. As measuring PWV is time-consuming, reliable estimation methods have been developed, but their ability to inform cardiovascular risk prediction beyond what is achievable with current clinical risk tools is uncertain. Methods: This study includes participants aged between 40 and 69 years from the population-based CARTaGENE cohort. PWV estimations (ePWV) were obtained using published formulas (ePWV f ) or algorithmic transformation of pulse waveforms (ePWV algo ) and 10-year cardiovascular risk for each participant was computed using the ASCVD and the SCORE-2 risk equations. Participants were followed during 10 years for major adverse cardiovascular events occurrence (MACE: cardiovascular death, myocardial infarction, stroke). Associations of ePWV f and ePWV algo with MACE were obtained using Cox models adjusted for ASCVD or SCORE-2 in the overall population and in a subpopulation representative of the ePWV f derivation cohort. Results: Of 17 548 eligible participants, 2263 (12.9%) experienced a MACE during follow-up. Both ePWVf and ePWV algo were associated with MACE in unadjusted analyses, but only ePWV algo remained significant after adjustments for ASCVD [hazard ratio (HR) = 1.16 [1.09–1.22]] and SCORE-2 (HR = 1.07 [1.00–1.13]). In contrast, ePWV f was not associated with MACE after adjustment for either risk score, and only after adjustment with ASCVD when it was tested in the subpopulation representative of its derivation cohort. Conclusions: Algorithm-based PWV improved cardiovascular risk prediction beyond what is achievable from recognized risk equations, whereas the predictive ability of ePWV f may not be generalizable outside of its reference population.