Association of Phenotypic Aging Marker with comorbidities, frailty and inflammatory markers in people living with HIV

Abstract

Background: Aging characteristics in people living with HIV (PLWH) are heterogeneous, and the identification of risk factors associated with aging-related comorbidities such as neurocognitive impairment (NCI) and frailty is important. We evaluated predictors of novel aging markers, phenotypic age (PhenoAge) and phenotypic age acceleration (PAA) and their association with comorbidities, frailty, and NCI.

Methods: In a cohort of PLWH and age- and sex-matched HIV-negative controls, we calculated PhenoAge using chronological age and 9 biomarkers from complete blood counts, inflammatory, metabolic-, liver- and kidney-related parameters. PAA was calculated as the difference between chronological age and PhenoAge. Multivariate logistic regression models were used to identify the factors associated with higher (>median) PAA. Area under the receiver operating characteristics curve (AUROC) was used to assess model discrimination for frailty.

Results: Among 333 PLWH and 102 HIV-negative controls (38% female), the median phenotypic age (49.4 vs. 48.5 years, p = 0.54) and PAA (- 6.7 vs. -7.5, p = 0.24) was slightly higher and PAA slightly less in PLWH although this did not reach statistical significance. In multivariate analysis, male sex (adjusted odds ratio = 1.68 [95%CI = 1.03-2.73]), current smoking (2.74 [1.30-5.79]), diabetes mellitus (2.97 [1.48-5.99]), hypertension (1.67 [1.02-2.72]), frailty (3.82 [1.33-10.93]), and higher IL-6 levels (1.09 [1.04-1.15]), but not HIV status and NCI, were independently associated with higher PAA. PhenoAge marker discriminated frailty better than chronological age alone (AUROC: 0.75 [0.66-0.85] vs. 0.65 [0.55-0.77], p = 0.04). In the analysis restricted to PLWH, PhenoAge alone predicted frailty better than chronological age alone (AUROC: 0.7412 vs. 0.6499, P = 0.09) and VACS index (AUROC: 0.7412 vs. 0.6811, P = 0.34) despite not statistically significant.

Conclusions: While PLWH did not appear to have accelerated aging in our cohort, the phenotypic aging marker was significantly associated with systemic inflammation, frailty, and cardiovascular disease risk factors. This simple aging marker could be useful to identify high-risk PLWH within a similar chronological age group.

Keywords: Aging; Comorbidities; Frailty; HIV/AIDS; Phenotypic age acceleration; Phenotypic aging marker; Thailand.

Fig. 1

Comparison of PhenoAge and PAA between frailty and NCI status in all population. A and B show the differences in PhenoAge and PAA with frailty status, respectively (P values < 0.001). PhenoAge and PAA were not statistically different between individuals with and without NCI in all population

Fig. 2

Comparison of PhenoAge and PAA between frailty and NCI status in PLWH only. A and B show the differences in PhenoAge and PAA with frailty status in PLWH, respectively (P < 0.001 for PhenoAge and P = 0.001 for PAA). PhenoAge and PAA were not statistically different between individuals with and without NCI in PLWH

Fig. 3

Receiver-operating characteristic curves for frailty in overall (3A) and among PLWH (3B). AUROC curves in A shows PhenoAge alone predicted frailty risk better than chronological age alone (AUROC: 0.7517 vs. 0.6528, P = 0.040) (A). It was also higher in the adjusted analysis (model 1) (AUROC: 0.8298 vs. 0.7934, P = 0.104) although statistically not significant. In the analysis restricting to PLWH, AUROC curves in B shows PhenoAge alone predicted frailty better than chronological age alone (AUROC: 0.7412 vs. 0.6499, P = 0.09) and VACS index (AUROC: 0.7412 vs. 0.6811, P = 0.34) although it was statistically significant. Model 2 + PhenoAge has significantly better discriminative ability for frailty compared with VACS index alone (AUROC: 0.835 vs. 0.6811, P = 0.02), but not with model 2 + VACS index (AUROC: 0.835 vs. 0.8231, P = 0.73). C and D show PAA has lower AUROC value compared to PhenoAge in predicting frailty risk among all participants and PLWH. In D, PAA predicted similarly as chronological age and VACS index in the adjusted models (i.e., P > 0.1 for AUROC of model 2 + chronological age or model 2 + VACS index compared to model 2 + PAA)/ Model 1 was adjusted for sex, BMI, smoking, alcohol drinking, education level, income, diabetes mellitus, hypertension, and statin use. Model 2 was adjusted for the variables in model 1 + current CD4 count and duration of ART