Distinct patterns of fasting plasma glucose and lipid profile levels over time in adults tested positive for HIV on HAART in Shanghai, China, revealed using growth mixture models

OBJECTIVES: This study sought to identify potential change patterns and predictors of fasting plasma glucose (FPG) and lipid levels after initiating highly active antiretroviral therapy (HAART). METHODS: A retrospective cohort study was conducted on 1,572 patients tested positive for HIV who initiated HAART between January 2010 and October 2020 in Shanghai, China. The growth mixture models (GMM) were used for capturing subgroups of FPG trajectories as well as triglyceride (TG) and total cholesterol (TC) dual-trajectories. Multinomial logistic regression models identified correlates of given trajectories. RESULTS: The median follow-up time was 2.0 years (IQR 1.0–4.7). Three FPG trajectory subgroups were identified as FPG low-stable (62.3%), medium-stable (30.5%), and high-increasing (7.2%). Furthermore, three subgroups of TG and TC dual-trajectories were identified as TG and TC high-slight increasing (13.7%), low-rapid increasing (27.6%), and a subgroup of medium-stable TC and slight-decreasing TG (58.7%). Older age, high TG, FPG, BMI, CD4 count of <200 at baseline, and initial use of zidovudine (AZT) and protease inhibitors (PIs) helped to identify the class with increasing glucose or lipid metabolism trajectories. CONCLUSION: The change patterns of plasma glucose and lipid in patients tested positive for HIV were heterogeneous and tailored interventions should be considered in specific subgroups.