Multidimensional profiling of serum biomarker nutritional scores in TAVR using principal component analysis

Pre-procedural nutritional status has been proposed as a prognostic marker in transcatheter aortic valve replacement (TAVR), although most studies rely on isolated nutrition scores. This study examined how multiple serum-based nutritional scores influence post-TAVR survival.

Consecutive patients undergoing TAVR between December 2021 and December 2023 were retrospectively analyzed. Eight serum biomarker-based scores were calculated for each patient (GNRI, PNI, CONUT, PALBI, TCBI, PACI, AGR, ALBI). Associations with overall survival were assessed with Kaplan-Meier curves and adjusted multivariable Cox regression. Principal component analysis (PCA) evaluated relationships among scores.

A total of 459 patients were included (mean age 81.7 ± 7.1 years, 45.8% female, median STS 2.8 [1.8-4.4]). The eight scores frequently classified the same patients into different risk strata. Significant survival differences were observed for all scores' most severe stratification except CONUT (p = 0.083) and PNI (p = 0.133). In adjusted Cox models, most scores showed a statistically significant association with overall mortality, except for GNRI (p = 0.237), PALBI (p = 0.378) and TCBI (p = 0.125). PCA demonstrated substantial collinearity among several scores and highlighted a dominant first principal component (PC), PC1 = 54%, without improved discrimination compared to STS score. In the adjusted Cox model, PC1 was also significantly associated with overall mortality (p < 0.001).

Despite capturing overlapping biological domains, serum-based nutritional scores inconsistently classify TAVR patients. PCA suggests that a single underlying biomarker dimension explains most of the variability among scores, highlighting overlap and inconsistency across them and supporting the need for a more integrated approach to nutritional assessment rather than single metrics.