Allometric versus ratiometric normalization of left ventricular stroke volume by Doppler-echocardiography for outcome prediction in severe aortic stenosis with preserved ejection fraction

Background: Appropriate normalization methods to scale Doppler-derived stroke volume (SV) to body size in patients with aortic stenosis (AS) are poorly defined and reference values are lacking. We aim to establish reference values of normalized SV in adults, and to compare the prognostic value of SV normalized by different methods in AS patients.

Methods: In 2781 normotensive, non-obese adults without cardiovascular disease we defined normal relationships between SV and body size by nonlinear regression. Reference SV values were calculated by quantile regression. We subsequently analyzed by Cox analysis the prognostic performance of ratiometric and allometric normalized SV in 1450 patients with severe AS and preserved LVEF under medical and surgical management.

Results: Unlike ratiometric normalization, allometric indexation eliminated the residual relationships between normalized SV and body size. The allometric exponents that adequately described the SV-height (H) and SV-body surface area (BSA) relationships were 1.32, and respectively 0.88. In males, low-flow (LF) reference values were: <28 ml/m², <30 ml/m, <30ml/(m²)^0.88, and, respectively, <26 ml/m^1.32, and in females <27 ml/m², <28 ml/m, <29ml/(m²)^0.88, and, respectively, <24 ml/m^1.32. In patients with severe AS, SV/H^1.32 was most consistently associated with mortality and showed better prognostic performance than other normalized SV parameters (adjusted hazard ratios: 1.86 for SV/H^1.32, 1.72 for SV/H, 1.64 for SV/BSA, and 1.61 for SV/BSA^0.88). Compared to H-normalization, BSA-normalization markedly overestimated the frequency of LF (3% vs. 9%).

Conclusions: We provide normative reference values and appropriate normalization methods for SV by Doppler-echocardiography. In severe AS, SV/H^1.32 seems the most appropriate indexation method, especially in obese individuals.