Kallman-Price J, VanderNoot N, Escheik C, Austin P, Estep JM, et al. 2021. Principal components analysis (PCA) in chronic liver disease (CLD) reveals three component model highlighting discrete contributions by sarcopenia criteria measures and serum myokines. Gastroenterology 160(6):S-762; doi: 10.1016/S0016-5085(21)0235-x.
Abstract
Introduction: Sarcopenia secondary to NAFLD or other CLDs is not well understood. Activity level (AAS METs), relative grip strength (GS), and % lean mass (%LM) can be used pragmatically to assess for sarcopenia. Aim: To look at indicators of secondary sarcopenia with a NAFLD population and associated myokines Methods: Data from subjects with CLD (NAFLD, HCV) and Controls across an initial 19 continuous variables pertaining to sarcopenia were assessed for missing data, outliers, normality, linearity, homoscedasticity, and homogeneity of variance and met requirements for inclusion in multivariate analyses. Data included a myokine panel [apelin, fractalkine, erythropoetin, LIF, IL-15, BDNF, irisin, FSTL-1, FABP3, IL-6, osteonectin, oncostatin, and osteocrin] on all subjects. Possible sarcopenia (PS) as operationally defined in this analysis incorporated pre-sarcopenia (at least one criteria indicative of risk), sarcopenia (2 criteria), and severe (3 criteria met). PS was compared to subjects with no criteria indicative of sarcopenia. Pearson correlations were used to preserve demographic and self-report variable ranges for comparison to myokine values. Only items with significant correlations were used to generate linear regressions, which informed variable inclusion in the principal components analysis (PCA). Results: 75 subjects [48.0 ± 12.8 years, 37.3% female, 25 NAFLD (15 PS-4 Pre/9 Sarcopenia/2 Severe), 30 HCV (13 PS-5/5/ 3), 20 Control (4 PS-1/3/0), 30.1 ± 5.75 BMI, 72.2 ± 9.3 % lean mass (%LM), 8.2 ± 1.3 AAS METs, 3.4 ± 1.6 FSS]. PCA yielded a three component model of 7 variables accounting for a cumulative percentage of 69.6%: 1) myokines FSTL-1 (0.865), IL-6 (0.829), and relative grip strength (GS, 0.480); 2) % lean mass (-0.911), PS (0.893); 3) AAS METs (- 0.815), FSS (0.788). Each component is associated with a separate criteria for PS. PS loaded strongly with %LM, but not with components containing additional PS criteria measures. The inverse association of AAS METs and FSS is consistent with prior research. GS scored highest for unique contribution to the model, and its FSTL-1 and IL-6 associations require further exploration. Conclusions: Criteria measures for possible sarcopenia define discrete model components and contribute independent fatigue and myokine associations. Possible sarcopenia is highly significantly associated with the % lean mass
