Abstract Body

Some people living with HIV under long-term successful antiretroviral therapy (PLWH) suffer from cardiometabolic comorbidities, among other aging-related disorders. Though case-control studies comparing PLWH and HIV-negative control (HC) have been done previously, the heterogeneity among PLWH related to the incidence of metabolic diseases has not been investigated. To predict metabolic changes in PLWH, we aim to determine clusters of patients based on three omics layers integration and the underlying mechanisms separating these clusters using advanced network analysis.

Samples from 97 patients from the Copenhagen Comorbidity in HIV-infection (COCOMO) study and 20 age, BMI, and gender-matched HC were used for this study. The clinical data were collected from the COCOMO database, untargeted plasma metabolomics and lipidomics were performed using ultra-high-performance liquid chromatography/mass spectrometry (UHPLC/MS/MS) and microbiome profiling by bacterial 16S rRNA analyses. PLWH were clustered based on three layers integration (metabolome, lipidome and microbiome) using similarity network fusion (SNF). The clusters were then characterized by clinical parameters and the individual omics levels using advanced statistics.

Three clusters (C1-C3) of PLWH were identified based on SNF (Fig 1A-C). Lipids were shown to have the most influence in the cluster repartition (Fig 1B). C1 and C3 include patients with the healthiest profile compared to C2. C2 represented patients with high BMI, metabolic syndrome, high visceral adipose tissue, subcutaneous adipose tissue (SAT) and hypertension (all p<0.05) but interestingly also had a higher CD4 count (p<0.05). C3 showed similar metabolomics and lipidomics profiles like HC. At clinical level, patients from C3 have higher BMI and SAT than C1 but lower than C2. C2 showed a drastic increase in triglycerides and diglycerides compared to C1 and C3 (FDR<0.1). At microbiome level, alpha diversity of HC, C2 and C3 were almost similar while C1 was significantly lower. C2 was also enriched in Prevotella genus compared to C1 and C3 and have patients with men who have sex with men.

In conclusion, we have shown that the molecular heterogeneity among the patients should be considered while defining immunological and virological success to determine the healthy state as PLWH displayed different profiles at -omics and clinical levels. Certain patients have dysregulated metabolic profiles despite having suppressed viral load and high CD4