부산대학교 도서관

Introduction: Wide variability in LDL-C change is observed with statins, yet determinants of statin response are uncertain.Methods: Participants were selected from the primary prevention cohort of the Pravastatin Inflammation/CRP Evaluation double-blind trial that randomized participants to pravastatin 40 mg/d or placebo over 24 weeks. Baseline and 24-week levels of LDL-C and 15 other biomarkers were measured in 495 participants. We defined optimal statin response as >=30% LDL-C reduction and suboptimal response as <30% reduction. Sub-optimal hs-CRP response was defined as >=median (14%) decline in hs-CRP from baseline to 24 weeks and non-response as no decrease or an increase in hs-CRP. χ, t-tests and ANOVA were used to compare variables across optimal statin response (N=166) and suboptimal response (N=287). Multivariable logistic regression models evaluated associations of determinants of statin response. Forward selection identified variables that associated with response. Xgboost was used to train and validate the models using 2/3 and 1/3 of the data respectively.Results: Significant determinants of optimal statin response included older age, and higher baseline levels of LDL-C and triglyceride-rich lipoproteins. By contrast, female sex, alcohol intake >=1 drink/day, diabetes, higher baseline levels of apo B and lipoprotein(a) were associated with decreased response, as was hs-CRP non-response (Table). Race, baseline hs-CRP and sub-optimal hs-CRP response, smoking, HDL-C and BMI had no significant effect on statin LDL-C response. Training and validation of models predicted suboptimal LDL-C response with an AUC of 0.71. Similarly, training and validation of models using Xgboost yielded an AUC of 0.85.Conclusion: This study identified discordant lipid phenotype and other determinants of moderate-intensity statin response and suggests other pathways of CVD risk beyond those addressed by statin treatment that require further investigation.