Healthy Aging and the Gut Microbiome in People With and Without HIV

Abstract

Background: Aging-related comorbidities are more common in people with human immunodeficiency virus (HIV) compared to people without HIV. The gut microbiome may play a role in healthy aging; however, this relationship remains unexplored in the context of HIV.

Methods: 16S rRNA gene sequencing was conducted on stool from 1409 women (69% with HIV; 2304 samples) and 990 men (54% with HIV; 1008 samples) in the MACS/WIHS Combined Cohort Study. Associations of age with gut microbiome diversity, uniqueness, and genus-level abundance were examined in women and men separately, followed by examining relationships of aging-related genera with frailty (Fried frailty phenotype) and mortality risk (Veterans Aging Cohort Study [VACS] index).

Results: Older age was associated with greater microbiome diversity and uniqueness, greater abundance of Akkermansia and Streptococcus, and lower abundance of Prevotella and Faecalibacterium, among others; findings were generally consistent by sex and HIV status. An aging-related microbiome score, generated via combination of 18 age-related genera, significantly increased with age in both women and men independently of demographic, behavioral, and cardiometabolic factors. In general, age was more strongly related to microbiome features (eg, diversity, microbiome score) in men without compared to with HIV, but age-microbiome associations were similar in women with and without HIV. Some age-related genera associated with healthy/unhealthy aging, such as Faecalibacterium (related to reduced frailty) and Streptococcus (related to higher VACS index).

Conclusions: Age is associated with consistent changes in the gut microbiome in both women and men with or without HIV. Some aging-related microbiota are associated with aging-related declines in health.

Keywords: HIV; age; frailty; gut microbiome; healthy aging.

Figure 1.

Associations of age with gut microbiome diversity and uniqueness in women and men with and without HIV. Linear mixed-effects models with random intercept were used to assess the association of age with (A) number of amplicon sequence variants, (B) the Shannon diversity index, and (C) uniqueness based on the Jensen Shannon divergence, among 1409 women (2304 person-visits) and 990 men (1008 person-visits). Model 1 includes weights from inverse probability weighting of age based on race/ethnicity, study site, income, educational attainment, employment, smoking status, alcohol use, drug use, hepatitis C virus serostatus, and HIV serostatus (and men who have sex with men status in men only). Model 2 includes aforementioned weights and is adjusted for body mass index, diastolic blood pressure, systolic blood pressure, diabetes, antidiabetic medication, lipid-lowering medication, antihypertensive medication, and estimated glomerular filtration rate. Abbreviations: CI, confidence interval; PWH, people with HIV; PWOH, people without HIV.

Figure 2.

Age and gut microbiome genera in women and men with and without HIV. A, LFC in genus abundance per year of age for women (x-axis) and men (y-axis), from unadjusted analysis of compositions of microbiomes with bias correction (ANCOM-BC) models among 1409 women (2304 person-visits) and 990 men (1008 person-visits): 169 tested genera are plotted; 18 genera shown in color represent those significantly associated with age in both women and men at false discovery rate-adjusted P < .10. B, Association of age with CLR-transformed abundance of 18 genera in linear mixed-effects models, stratified by sex and HIV serostatus. Models include weights from inverse probability weighting of age based on race/ethnicity, study site, income, educational attainment, employment, smoking status, alcohol use, drug use, hepatitis C virus serostatus, and HIV serostatus (and men who have sex with men status in men only). C, Scatter plot of age (x-axis) and age-related microbiome score (y-axis), derived for each sample by adding (or subtracting) CLR-transformed abundance for 18 genera positively (or inversely) associated with age. Linear regression lines are displayed stratified by sex and HIV serostatus. D, Association of age with age-related microbiome score in linear mixed-effects models. Model 1 includes weights from inverse probability weighting of age described in (B). Model 2 includes aforementioned weights and adjusts for body mass index, diastolic and systolic blood pressure, diabetes, antidiabetic, lipid-lowering, and antihypertensive medications, and estimated glomerular filtration rate. Abbreviations: CI, confidence interval; CLR, centered log ratio; LFC, log fold change; MWH, men with HIV; MWOH, men without HIV; PWH, people with HIV; PWOH, people without HIV; WWH, women with HIV; WWOH, women without HIV.

Figure 3.

Correlations among aging-related gut microbiome features in women and men with and without HIV. Spearman correlations are shown for (A) women (n = 1409) and (B) men (n = 990). Only 1 sample per participant (ie, the first stool sample) was included in correlation analysis. Abbreviation: ASV, amplicon sequence variant. *p < 0.05, **p < 0.01, ***p < 0.001, ****p < 0.0001.

Figure 4.

Association of aging-related microbiome score and genera with indicators of unhealthy aging in women and men with and without HIV. Relationship of the aging-related microbiome score or centered log ratio-transformed genus abundance with (A) frailty using binomial logistic regression models, or (B) the VACS index using linear mixed-effects models. For the outcome of frailty, we calculated cumulative visits with frailty out of total visits with available data over the 2–5 visits prior to/inclusive of the last person-visit with a stool sample. All models were adjusted for age. Color indicates whether the genus was found to increase or decrease with age. Abbreviations: CI, confidence interval; OR, odds ratio; VACS, Veterans Aging Cohort Study.

Figure 5.

Interaction of age and aging-related microbiome score on indicators of unhealthy aging in women and men with and without HIV. Relationship of age to frailty (A–D) and the VACS index (E–H) among participants with low or high age-related microbiome score, defined as below or above the sex- and HIV-specific median of the age-related microbiome score. For the outcome of frailty, we calculated cumulative visits with frailty out of total visits with available data over the 2–5 visits prior to/inclusive of the last person-visit with a stool sample. P values displayed on plots are for the interaction of high versus low age-related microbiome score with age, in binomial logistic regression models for the frailty outcome, and linear mixed-effects models for the VACS index outcome. Abbreviations: Int, interactive; VACS, Veterans Aging Cohort Study.