Epigenetic Age Acceleration and Frailty Among People with HIV

Abstract

DNA methylation is a hallmark of aging; yet, our understanding of epigenetic age acceleration (EAA) in relationship to frailty in people with HIV (PWH) is poor. We conducted an observational study among PWH from the Veterans Aging Cohort Study (VACS) to test the hypothesis that EAA markers were associated with frailty. Epigenome-wide DNA methylation data from blood samples were used to derive EAA markers based on four established epigenetic clocks: Horvath, Hannum, PhenoAge, and GrimAge. Frailty was defined using a previously studied VACS frailty-related phenotype based on ≥1 survey item characterizing frailty factors: exhaustion, slowness, low physical activity, or weight loss. Logistic regression tested the association of participant characteristics and EAA markers with frailty. Adjusted models included each EAA marker as the independent variable, with significant participant characteristics as covariates. Among 1,076 PWH, frailty was evident in 397 (36.9%) individuals. The characteristics associated with frailty included chronological age, CD4⁺ T-cell count, HIV-1 RNA viral load, smoking, and age-related comorbid conditions. GrimAge acceleration (GrimAA), PhenoAge acceleration (PhenoAA), and HannumAge acceleration (HannumAA) were associated with frailty, but HorvathAge acceleration (HorvathAA) was not. The strength of the association was attenuated with adjustment for characteristics but remained significant for the three markers. Age acceleration based on GrimAA (values >0) was independently associated with a 45% increased odds of frailty (OR_adj: 1.45, 95% CI, 1.10, 1.93). In post hoc analyses, only GrimAA was associated with exercise frequency. In conclusion, select EAA markers were associated with frailty, independently of the traditional predictors of frailty. GrimAA, in particular, may be useful in future research to develop treatment strategies for frailty tailored to PWH.

Keywords

DNA methylation epigenomics aging VACS frailty HIV

Introduction

Frailty is a meaningful construct to study aging because it independently predicts lifespan (mortality) and health span (comorbidity and disability). Approaches to measure frailty usually use either an index based on deficits or a phenotype based on functional capacity.¹ Because there is a recognized phenomenon of accelerated aging in people with HIV (PWH), frailty has been well-studied in this high-risk group using both approaches.² Risk factors for frailty among PWH on antiretroviral therapy (ART) include smoking and age-related comorbidity.³ Strategies to prevent and reverse frailty in PWH would be advanced by prognostic biomarkers.⁴

Biomarkers associated with frailty are directly connected to several hallmarks of biological aging, including systemic inflammation, telomere shortening, mitochondrial dysfunction, and cellular senescence.¹ In particular, DNA methylation (DNAm) is an epigenetic change that occurs with aging and is linked to several biological pathways. These changes are sensitive to environmental factors and thus attractive targets for interventions.⁵ Epigenetic clocks are machine learning algorithms that use methylation changes at cytosine–phosphate–guanine (CpG) sites to produce an estimated DNAm Age in units of years that predicts, and thus is highly colinear with, chronologic age.⁶ Residuals from the regression of DNAm Age on chronological age produce an epigenetic marker that is independent of chronologic age and estimates relative age acceleration.⁶ These biomarkers are commonly referred to as epigenetic age acceleration (EAA) markers. While first-generation epigenetic clocks were developed to predict chronologic age (e.g., Horvath and Hannum), second-generation clocks (e.g., PhenoAge and GrimAge) were trained for morbidity and mortality⁷ and subsequently associated with physical activity.⁸

Age-associated epigenetic changes provide evidence that biological aging is accelerated in PWH and, further, is attenuated by ART.^9–12 Yet, in contrast to other biomarkers of frailty in PWH, our understanding of epigenetic markers to study frailty is limited. We previously reported¹³ in the Veterans Aging Cohort Study (VACS) that EAA markers were associated with the VACS Index, a deficit measure of frailty driven by laboratory measures of organ function and HIV progression.¹⁴ A recent epigenetic study corroborates these findings in another cohort of PWH.¹⁵ In contrast, epigenetic markers were not associated with a frailty phenotype in a small study of older PWH.¹⁶

The objective of the present study was to test the association of EAA markers with a functional frailty phenotype in PWH. We hypothesized that EAA was independently associated with frailty, and further, that the second-generation EAA markers PhenoAge and GrimAge would have the greatest strength of association.

Materials and Methods

Population

Participants with HIV infection in the VACS¹⁷ who provided a blood sample between 2005 and 2007 as a part of the VACS Biomarker Cohort¹⁸ were included in this study if the DNAm Age data were available (N = 1,110). Survey data used to derive the frailty outcome were missing in 34 (3%) participants, leading to the exclusion of these participants and a final analytical set of 1,076 participants. Written consent was obtained from all participants, as approved by each local institutional review board.

Independent variable

Each EAA marker served as the independent variable in its respective model. To generate DNAm data, genome-wide CpG site profiling was performed using the Infinium Human Methylation 450 K BeadChip or the Illumina Methylation EPIC 850 BeadChip, depending on the era when the samples were tested.¹⁹ Raw beta values were generated by BeadStudio software with the proportion of the methylation intensity of a site (ranging from 0 to 1). As previously described,²⁰ CpG sites were removed from the analysis for the following reasons: >5% missing rate (927), probes contained single nucleotide polymorphisms within 10 base pairs of the CpG site (35,605), and their probes mapped to multiple locations in the genome (24,729). We also excluded samples with a >10% missing rate of methylation data and any results with a detection p value of >.01. We performed quantile normalization and batch effect corrections using the “minfi” package available in R.²¹

DNAm Age was calculated for the first-generation epigenetic clocks, Horvath and Hannum,^22,23 and the second-generation clocks, PhenoAge²⁴ and GrimAge,²⁵ using the online calculator (https://dnamage.genetics.ucla.edu) developed by Horvath and colleagues.²⁶ For each epigenetic clock, the residuals from the regression of DNAm Age on chronological age produced an epigenetic marker hereafter referred to as an EAA marker. This measure of relative age acceleration is independent of chronological age owing to its mathematical construction and is expressed in units of years.⁶ Positive values of EAA markers signify epigenetic age greater than chronological age; negative values (<0) signify epigenetic age less than chronological age. For this study, EAA was analyzed as present (>0) or absent (<0) as a dichotomous variable as well as a continuous variable. Individual markers were labeled by the individual name of the epigenetic clock (e.g., GrimAge acceleration [GrimAA]).

Dependent variable: Frailty phenotype

The frailty phenotype for this study was originally developed in VACS participants with and without HIV using VACS survey items to measure four factors: exhaustion, slowness, low physical activity, and weight loss,²⁷ that align with the components of the Fried frailty phenotype,²⁸ excluding only weakness. Due to the missing strength component and the reliance on self-reported measures, we termed this phenotype “frailty-related phenotype.” Despite these limitations, the VACS frailty-related phenotype predicted hospitalizations and mortality²⁷ and was independently associated with age-related comorbidities²⁹ and polypharmacy.³⁰ For this smaller VACS sample, we chose to combine prefrail and frail participants and defined frailty as the presence of ≥1 components, similar to another VACS substudy.³⁰ Survey data collected at the time point closest to the blood specimen collection were used.

Other covariates

VACS definitions for HIV and age-related comorbid conditions use validated methods based on electronic medical record diagnostic codes with additional qualifiers for hypertension, cardiovascular disease, and diabetes. Body mass index (BMI) and clinical laboratory values, including CD4⁺ T-cell count, HIV-1 plasma RNA viral load (VL), estimated glomerular filtration rate, and components of the Fibrosis-4 (FIB-4) index, were extracted from the electronic health record at the closest time point within 180 days of the blood specimen collection. Current ART was defined as prescribed ART within 180 days before or up to 7 days after blood specimen collection. Health behaviors, including smoking, alcohol use, and exercise frequency, were obtained from the VACS survey.

Statistical analysis

Initial tests for the association of EAA markers with frailty used chi-squared tests when EAA was categorized as present (>0) or absent (<0) and two-sample t-tests when continuous values of EAA markers were used. Logistic regression models were performed with the frailty-related phenotype as the dependent variable (frail vs. not frail) and EAA markers as the independent variable. Participant characteristics, which were previously established as predictors or mediators of frailty in PWH,^2,3 were tested for the association with the frailty-related phenotype by chi-squared tests. Significant characteristics (p < .05) were included in the adjusted logistic regression models with each biomarker as the independent variable. A sensitivity analysis was performed using the multinomial logistic regression of frailty expressed as the number of frailty components to test the odds of being frail (≥2 items) or prefrail (1 item) against not frail (0 items) as the reference group.

An exploratory analysis of the components of frailty was conducted to explore their individual relationship with EAA markers. Then, based on these results, a post hoc analysis of exercise frequency as a surrogate outcome for the component of physical activity was conducted. The difference between the mean EAA markers and exercise frequency was tested by one-way analysis of variance (ANOVA) and then adjusted for BMI in analysis of covariance (ANCOVA). Statistical significance was taken at the 0.05 level, and all analyses were conducted using SAS V9.4 (SAS Institute Inc.).

Results

Table 1 shows the baseline characteristics of the study population (n = 1,076). The mean (SD) age was 52.4 (7.8) years. The majority of participants self-identified as men (97.7%) and non-Hispanic Black (82.8%). Frailty was present in 36.9% of the participants (n= 397), and the majority were positive with 1–2 components (33.8%). Poor health behaviors were common, including current smoking (55.7%) and no routine exercise (25.0%). Several participant characteristics were associated with frailty, including chronological age, CD4 count, VL, smoking, heart failure, chronic obstructive pulmonary disease, and liver fibrosis (Fig. 1).

FIG. 1.

Odds of frailty for participant characteristics and epigenetic age acceleration (EAA) markers. Forest plots show unadjusted odds ratios estimated from bivariate logistic regression models, where frailty (categorized as frail vs. not frail [reference]) is the dependent variable and each characteristic is included as the sole independent variable. Reference categories for variables: GrimAge acceleration (GrimAA), PhenoAge acceleration (PhenoAA), HannumAge acceleration (HannumAA), HorvathAge acceleration (HorvathAA) reference value = 0. Age: <50 years; Race: Black; Sex: male; Body mass index (BMI): Underweight/normal weight; CD4 count: ≥500 cells/µL; HIV viral load detectable (>75 copies/mL); Liver fibrosis: FIB-4 score <1.45; Kidney function: estimated glomerular filtration rate (eGFR) ≥60 mL/min/1.73 m²; Current smoking: prior or never smoker; and Absence of the following conditions: unhealthy alcohol use, AIDS-defining illness, current antiretroviral therapy, non-AIDS defining cancer, diabetes, congestive heart failure, coronary artery disease, stroke, chronic obstructive pulmonary disease, and hepatitis C virus.

Table 1.

Characteristics of the Study Population

CharacteristicsNo. (%) unless otherwise indicated	N = 1,076
Demographics
Age, years mean (SD)	51.37 (7.75)
Race
Black	891 (82.8%)
White	104 (9.7%)
Other	81 (7.5%)
Sex
Male	1,051 (97.7%)
Female	25 (2.3%)
Health behaviors
Current smoking	599 (55.7%)
Unhealthy alcohol use^a	624 (58.2%)
Frequency of exercise^a
Never/<1×/week	264 (25.0%)
1–2×/week	305 (28.9%)
3–4×/week	266 (25.2%)
>4×/week	221 (20.9%)
Body mass index^a
Underweight (<18.5 kg/m²)	36 (3.4%)
Normal weight (18.5–24.9 kg/m²)	468 (43.7%)
Overweight (25.0–29.9 kg/m²)	389 (36.3%)
Obese (≥30.0 kg/m²)	178 (16.6%)
HIV characteristics
AIDS-defining illness history	406 (37.7%)
Current antiretroviral therapy	901 (83.7%)
CD4 count^a
≤200 cells/µL	220 (20.5%)
200–500 cells/µL	501 (46.7%)
≥500 cells/µL	352 (32.8%)
Cells/µL, median (IQR)	394 (238, 575)
HIV viral load^a
≤75 copies/mL	576 (53.7%)
>75 copies/mL	497 (46.3%)
Log copies/mL, median (IQR)	1.87 (1.71, 3.65)
VACS Index score,^a median (IQR)	57.00 (46.00, 68.00)
Comorbid conditions
Diabetes	192 (17.8%)
Congestive heart failure	67 (6.2%)
Coronary artery disease	136 (12.6%)
Stroke	35 (3.3%)
Chronic obstructive pulmonary disease	171 (15.9%)
Anemia	408 (38.0%)
Any cancer	72 (6.7%)
Non-AIDS-defining cancer	54 (5.0%)
Hepatitis C infection^a	423 (43.1%)
Liver fibrosis based on FIB-4 score^a
Low, <1.45	568 (53.4%)
Intermediate, 1.45–3.25	389 (36.6%)
High, >3.25	107 (10.1%)
Kidney function based on eGFR, mL/min^a
≥60	995 (92.7%)
45–59.9	33 (3.1%)
30–44.9	14 (1.3%)
<30	31 (2.9%)
Epigenetic markers
GrimAge acceleration, years, mean (SD)	0.08 (4.96)
PhenoAge acceleration, years, median (IQR)	0.39 (−4.89, 5.70)
HannumAge acceleration, years, median (IQR)	0.05 (−3.62, 3.35)
HorvathAge acceleration, years, median (IQR)	−0.01 (−3.57, 3.98)
Frailty phenotype
Not frail	679 (63.1%)
Frail (≥1 components)	397 (36.9%)

Definitions: FIB-4 calculated as (years of age × AST)/(platelets in 109/L × square root of ALT); eGFR calculated as 186.3 × (serum creatinine)^−1.154 × (age^−0.203) × (0.742 for women) × (1.21 if black); HCV infection defined as positive antibody test or detectable virus; unhealthy alcohol use includes alcohol abuse/dependence and hazardous/at-risk/heavy episodes.

Missing data in <1%.

IQR, interquartile range; VACS, Veterans Aging Cohort Study; FIB-4, Fibrosis-4; eGFR, estimated glomerular filtration rate.

EAA was present (EAA > 0) in approximately half of the participants across all epigenetic clocks and was associated with frailty for GrimAA, PhenoAge acceleration (PhenoAA), and HannumAge acceleration (HannumAA), but not HorvathAge acceleration (HorvathAA) (Table 2). The results were unchanged when EAA was examined as a continuous measure of EAA in years (Supplementary Fig. S1). The presence of EAA based on the Grim epigenetic clock (GrimAA >0) had the greatest odds for frailty (OR: 1.60, 95% CI, 1.24, 2.05; Table 3). The association between epigenetic markers and frailty was slightly attenuated but remained significant after adjustment for HIV-related factors, smoking, and age-related comorbidities (Table 3). Sensitivity analyses with the outcome of the number of frailty components provided similar results, with the exception that some of the adjusted models for PhenoAA were no longer significant (Supplementary Table S1).

Table 2.

Association of EAA Markers with Frailty in People with HIV

Epigenetic marker, years	TotalN (%)	Frailty		p Value*
Epigenetic marker, years	TotalN (%)	PresentN = 397	AbsentN = 679	p Value*
Grim Age acceleration				<.001
>0	547 (50.8%)	231 (58.2%)	316 (46.5%)
<0	529 (49.2%)	166 (41.8%)	363 (53.5%)
Pheno Age acceleration				.003
>0	552 (51.3%)	227 (57.2%)	325 (47.9%)
<0	524 (48.7%)	170 (42.8%)	354 (52.1%)
Hannum Age acceleration				<.001
>0	545 (50.7%)	230 (57.9%)	315 (46.4%)
<0	531 (49.3%)	167 (42.1%)	364 (53.6%)
Horvath Age acceleration				.658
>0	538 (50.0%)	202 (50.9%)	336 (49.5%)
<0	538 (50.0%)	195 (49.1%)	343 (50.5%)

Data presented as N (%).

EAA present (>0) or absent (<0) based on the value of a marker.

*Chi-squared test.

EAA, epigenetic age acceleration.

Table 3.

Logistic Regression of Frailty on Each EAA Marker

EAA markers^a	Unadjusted modelOR (95% CI)	Age-adjusted model with HIV-related factors^bOR (95% CI)	Age-adjusted model with smoking, CHF, COPD, liver fibrosis^c OR (95% CI)
GrimAA > 0	1.60 (1.24, 2.05)^*	1.49 (1.15, 1.92)^**	1.45 (1.10, 1.92)^**
PhenoAA > 0	1.45 (1.13, 1.87)^**	1.32 (1.02, 1.71)^*	1.32 (1.02, 1.71)^*
HannumAA > 0	1.59 (1.24, 2.04)^***	1.49 (1.15, 1.92)^**	1.57 (1.21, 2.03)^***
HorvathAA > 0	1.06 (0.83, 1.35)	1.00 (0.78, 1.29)	1.00 (0.78, 1.29)

Reference group is value <0.

CD4 count, cells/µL; log₁₀-transformed HIV viral load, copies/mL.

Current smoking vs. former/never, liver fibrosis (FIB-4 score ≥ 1.45 vs. < 1.45).

*p < .05.

**p < .01.

***p < .001.

GrimAA, GrimAge acceleration; PhenoAA, PhenoAge acceleration; HannumAA, HannumAge acceleration, HorvathAA, HorvathAge acceleration; CHF, congestive heart failure; COPD, chronic obstructive pulmonary disease; EAA, epigenetic age acceleration.

Among the frailty components, low physical activity was the most prevalent and was significantly associated with EAA across GrimAA, PhenoAA, and HannumAA markers (Supplementary Table S2). Exercise frequency was inversely associated with GrimAA only; the results were unchanged when adjusted for BMI (Supplementary Table S3).

Discussion

We found that EAA was associated with a frailty-related phenotype in PWH, as evidenced by several epigenetic markers, GrimAA, PhenoAA, and HannumAA. The results are consistent with our prior work using a deficit-accumulation measure of frailty, the VACS Index, supporting a link between epigenetic markers and frailty in PWH. Our findings further address the gap in the literature by demonstrating a significant relationship between EAA markers and frailty after adjustment for smoking, HIV disease characteristics, and age-related comorbidity.

Our understanding of epigenetic markers and frailty is limited in PWH.^13,15,16 In the current study, we found that EAA was present (>0) in about half of the PWH for all EAA markers studied. Frailty was present in about a third of the individuals as a combination of frail and prefrail states and was significantly associated with EAA for GrimAA, PhenoAA, and HannumAA, but not HorvathAA. The relationship between EAA markers with frailty remained significant, although attenuated, with adjustment for smoking and clinical factors. The largest effect size was found for GrimAA; individuals with EAA based on GrimAge had a 60% increased odds of frailty compared with those without age acceleration. Similarly, relative years of EAA was higher in frail participants compared with nonfrail participants for GrimAA, PhenoAA, and HannumAA. Our results are consistent with previous reports in the general population but provide novel results among PWH. A previous study in PWH failed to find a significant association between these EAA markers and the Fried frailty phenotype but was limited by a small sample.¹⁶ Using the VACS Index as a deficit-accumulation measure of frailty, we previously reported that GrimAA, PhenoAA, and HannumAA, but not HorvathAA, were significantly associated with frailty in VACS participants with HIV.¹³ The consistency with our current findings across EAA markers using a frailty-related phenotype supports the possibility that vulnerability is related to age-associated epigenetic changes and, further, this process can be captured by specific EAA markers. In addition, exploratory analysis of the frailty components led us to show that GrimAA is associated with low physical activity and infrequent exercise. This novel result in PWH is consistent with exercise research in adults without HIV³¹ and corroborates physical activity studies in adults with³² and without HIV.⁸

In contrast to our prior research in epigenetic markers and the VACS Index, in this study, we sought to examine the relationship between EAA and frailty in the context of smoking and comorbidity, known epigenetic modifiers³³ that are associated with frailty in PWH.^2,3 Smoking is a major modifier of DNAm. DNAm signatures, which predict smoking in PWH, are also associated with the VACS Index.³⁴ Recent work shows that GrimAA is associated with current smoking in PWH,³⁵ consistent with the architecture of GrimAge, which includes a DNAm-based estimator of smoking pack-years.²⁵ Our results show that GrimAA remains associated with frailty despite adjustment for smoking. Furthermore, we found that GrimAA, PhenoAA, and HannumAA were associated with frailty after adjustment for age-related comorbid conditions. These findings suggest that methylation changes in these EAA markers are not dependent on smoking and disease effects. However, further research is needed to assess the impact of other lifestyle factors, such as illicit drug use, physical activity, and nutrition.

Putative methylation modifications are associated with HIV replication and attenuated by ART,^11,12 which was recently confirmed in longitudinal studies using EAA markers.^9,10 Our findings demonstrate that GrimAA, PhenoAA, and HannumAA are associated with frailty, with and without adjustment for HIV-related factors. Our result on liver fibrosis raises questions about the potential underlying mechanisms related to dual infection with viral hepatitis³⁶ and metabolic-dysfunction-associated fatty liver disease,³⁷ both of which are linked to putative epigenetic changes among PWH. Although beyond the scope of a cross-sectional study, these findings inform future research to understand the pathogenesis, which is being captured by EAA markers among PWH.

Our results are consistent with the larger body of epigenetic research of frailty in the general population. Among first-generation clocks, several studies showed that the Hannum and Horvath epigenetic clocks are associated with frailty.³⁸ However, others, like our results, found that only HannumAA is associated with a Fried phenotype.³⁹ However, a large cohort of 3,200 individuals, which included both a frailty index and a phenotype measure, demonstrated that neither first-generation clock predicted frailty despite significant findings for mortality.⁴⁰ In contrast with first-generation clocks, the second-generation clocks GrimAge and PhenoAge are consistently associated with frailty.^31,40,41 This is not surprising given that frailty as a geriatric syndrome is meant to capture the loss of physiological reserve, analogous to GrimAA and PhenoAA that were trained on a combination of physiological processes to predict mortality.⁴² It is notable that only second-generation epigenetic clocks are associated with physical activity.⁸ GrimAge, in particular, is associated with aging phenotypes that include a physical component, such as functional performance and self-reported exercise frequency.^31,43 Furthermore, GrimAA has demonstrated its value as a prognostic indicator by independently predicting incident frailty.^44,45 Our results are based on a self-reported frailty-related phenotype, supporting the value of GrimAA as a biomarker of the broad measures of reduced physiological reserve among PWH.

There are several limitations to our study that warrant discussion. The EAA markers were based on blood samples from a single time point, limiting conclusions on causal inferences, especially relevant for epigenetic markers given the potential for the differential rate of change in longitudinal studies.⁶ Thus, the directionality of the relationship between EAA and frailty cannot be concluded based on our results. While we attempted to adjust for the factors known to be associated with both epigenetic modifications and frailty, we were limited to prevalent conditions and self-reported measures. In particular, for PWH, the epigenetic impact from other viral infections and illicit drug use needs to be considered. Because low body weight is a component of the frailty phenotype, we did not include BMI in our adjusted models. Yet, there is evidence for accelerated epigenetic aging of the liver in individuals with obesity.⁴⁶ Finally, although our blood samples were collected within 6 months of the frailty ascertainment, DNAm in leukocytes does not necessarily reflect aging in other tissues, which may be especially relevant for the functional phenotypes of aging.

Conclusions

Our study found that several markers of EAA were associated with a frailty-related phenotype in PWH, adjusted for smoking and both HIV-related and age-related comorbid conditions. The findings suggest that specific epigenetic markers may help clarify the heterogeneity of accelerated aging in PWH, including contributing environmental and disease factors. GrimAA, in particular, may be a valuable marker to study frailty and associated phenotypes among PWH. Epigenetic clocks may provide a quantifiable estimate of aging, but longitudinal measures with well-defined lifestyle factors and aging phenotypes are needed to investigate the contributing factors and causal pathways of accelerated aging in PWH.

Authors’ Contributions

K.K.O., V.C.M., and Y.V.S. designed the study. A.C.J. obtained the funding and specimens. A.J.L. conducted the statistical analysis with input from K.K.O., V.C.M., and A.C.J. K.X. contributed to DNA methylation data. K.S. contributed to the phenotypic data. All authors provided critical review and revision to the article and approved the final article.

Footnotes

Acknowledgments

The authors recognize the Veterans Aging Cohort Study participants and clinical research sites.

Author Disclosure Statement

V.C.M. has received investigator-initiated research grants (to the institution) and consultation fees (all unrelated to the current work) from Eli Lilly, Bayer, Gilead Sciences, Merck, and ViiV. The remaining authors have no disclosures.

Funding Information

This study was supported by the U.S. Department of Veterans Affairs [I01 RX002790 to K.K.O. and V.C.M.] and the National Institutes of Health [P30 AI050409 to V.C.M. and Y.V.S.]; [R01DK125187 to Y.V.S.]; [R01DA047820 to K.X.]; [R01DA047063 to K.X.]; [U24AA020794 to A.C.J.]; [U01AA020790 to A.C.J.]; [U24-AA022001 to A.C.J.]; [U10 AA013566 (completed) to A.C.J.]; and [K01HL134147 to K.S.].

Supplemental Material

References

Kane

, Sinclair

. Frailty biomarkers in humans and rodents: Current approaches and future advances. Mech Ageing Dev, 2019; 180:117–128; doi: 10.1016/j.mad.2019.03.007

Kehler

, Milic

, Guaraldi

, et al. Frailty in older people living with HIV: Current status and clinical management. BMC Geriatr, 2022; 22(1):919; doi: 10.1186/s12877-022-03477-7

Morgello

, Gensler

, Sherman

, et al.; NNTC. Frailty in medically complex individuals with chronic HIV. AIDS, 2019; 33(10):1603–1611; doi: 10.1097/QAD.0000000000002250

Erlandson

, Piggott

. Frailty and HIV: Moving from characterization to intervention. Curr HIV/AIDS Rep, 2021; 18(3):157–175; doi: 10.1007/s11904-021-00554-1

Wang

, Liu

, Hu

, et al. Epigenetic regulation of aging: Implications for interventions of aging and diseases. Signal Transduct Target Ther, 2022; 7(1):374; doi: 10.1038/s41392-022-01211-8

Teschendorff

, Horvath

. Epigenetic ageing clocks: Statistical methods and emerging computational challenges. Nat Rev Genet, 2025; 26(5):350–368; doi: 10.1038/s41576-024-00807-w

Martínez-Magaña

, Hurtado-Soriano

, Rivero-Segura

, et al. Towards a novel frontier in the use of epigenetic clocks in epidemiology. Arch Med Res, 2024; 55(5):103033; doi: 10.1016/j.arcmed.2024.103033

Oblak

, van der Zaag

, Higgins-Chen

, et al. A systematic review of biological, social and environmental factors associated with epigenetic clock acceleration. Ageing Res Rev, 2021; 69:101348; doi: 10.1016/j.arr.2021.101348

Schoepf

, Esteban-Cantos

, Thorball

, et al.; Swiss HIV Cohort Study. Epigenetic ageing accelerates before antiretroviral therapy and decelerates after viral suppression in people with HIV in Switzerland: A longitudinal study over 17 years. Lancet Healthy Longev, 2023; 4(5):e211–e218; doi: 10.1016/S2666-7568(23)00037-5

10.

Esteban-Cantos

, Rodríguez-Centeno

, Barruz

, et al.; NEAT001/ANRS143 Study Group. Epigenetic age acceleration changes 2 years after antiretroviral therapy initiation in adults with HIV: A substudy of the NEAT001/ANRS143 randomised trial. Lancet HIV, 2021; 8(4):e197–e205; doi: 10.1016/S2352-3018(21)00006-0

11.

Titanji

, Gwinn

, Marconi

, et al. Epigenome-wide epidemiologic studies of human immunodeficiency virus infection, treatment, and disease progression. Clin Epigenetics, 2022; 14(1):8; doi: 10.1186/s13148-022-01230-w

12.

Nelson

, Hui

, Rimland

, et al. Identification of HIV infection-related DNA methylation sites and advanced epigenetic aging in HIV-positive, treatment-naive U.S. veterans. AIDS, 2017; 31(4):571–575; doi: 10.1097/QAD.0000000000001360

13.

Oursler

, Marconi

, Wang

, et al. Epigenetic age acceleration markers are associated with physiologic frailty and all-cause mortality in people with human immunodeficiency virus. Clin Infect Dis, 2023; 76(3):e638–e644; doi: 10.1093/cid/ciac656

14.

Justice

, Tate

. Strengths and limitations of the Veterans Aging Cohort Study Index as a measure of physiologic frailty. AIDS Res Hum Retroviruses, 2019; 35(11–12):1023–1033; doi: 10.1089/AID.2019.0136

15.

Woods

, Teixeira

, Martins

, et al. Accelerated epigenetic aging in older adults with HIV disease: Associations with serostatus, HIV clinical factors, and health literacy. Geroscience, 2023; 45(4):2257–2265; doi: 10.1007/s11357-023-00759-9

16.

Shiau

, Arpadi

, Shen

, et al. Epigenetic aging biomarkers associated with cognitive impairment in older African American adults with human immunodeficiency virus (HIV). Clin Infect Dis, 2021; 73(11):1982–1991; doi: 10.1093/cid/ciab563

17.

Justice

, Dombrowski

, Conigliaro

, et al. Veterans Aging Cohort Study (VACS): Overview and description. Med Care, 2006; 44(8 Suppl 2):S13–S24.

18.

Armah

, McGinnis

, Baker

, et al. HIV status, burden of comorbid disease, and biomarkers of inflammation, altered coagulation, and monocyte activation. Clin Infect Dis, 2012; 55(1):126–136.

19.

Chen

, Hui

, Titanji

, et al. A multi-trait epigenome-wide association study identified DNA methylation signature of inflammation among men with HIV. Clin Epigenetics, 2024; 16(1):152; doi: 10.1186/s13148-024-01763-2

20.

Mathur

, Hui

, Huang

, et al. DNA methylation markers of type 2 diabetes among HIV-Positive and -Negative male veterans. J Infect Dis, 2019; doi: 10.1093/infdis/jiz023

21.

Aryee

, Jaffe

, Corrada-Bravo

, et al. Minfi: A flexible and comprehensive Bioconductor package for the analysis of Infinium DNA methylation microarrays. Bioinformatics, 2014; 30(10):1363–1369; doi: 10.1093/bioinformatics/btu049

22.

Marioni

, Shah

, McRae

, et al. DNA methylation age of blood predicts all-cause mortality in later life. Genome Biol, 2015; 16(1):25; doi: 10.1186/s13059-015-0584-6

23.

Perna

, Zhang

, Mons

, et al. Epigenetic age acceleration predicts cancer, cardiovascular, and all-cause mortality in a German case cohort. Clin Epigenetics, 2016; 8:64; doi: 10.1186/s13148-016-0228-z

24.

Levine

, Lu

, Quach

, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging (Albany NY), 2018; 10(4):573–591; doi: 10.18632/aging.101414

25.

, Quach

, Wilson

, et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging (Albany NY), 2019; 11(2):303–327; doi: 10.18632/aging.101684

26.

Horvath

. DNA methylation age of human tissues and cell types. Genome Biol, 2013; 14(10):R115; doi: 10.1186/gb-2013-14-10-r115

27.

Akgün

, Tate

, Crothers

, et al. An adapted frailty-related phenotype and the VACS index as predictors of hospitalization and mortality in HIV-infected and uninfected individuals. J Acquir Immune Defic Syndr, 2014; 67(4):397–404; doi: 10.1097/QAI.0000000000000341

28.

Fried

, Tangen

, Walston

, et al.; Cardiovascular Health Study Collaborative Research Group. Frailty in older adults: Evidence for a phenotype. J Gerontol A Biol Sci Med Sci, 2001; 56(3):M146–M156.

29.

Akgun

, Tate

, Oursler

, et al. Association of chronic obstructive pulmonary disease with frailty measurements in HIV-infected and uninfected Veterans. AIDS, 2016; 30(14):2185–2193; doi: 10.1097/QAD.0000000000001162

30.

Sung

, Gordon

, Edelman

, et al. Polypharmacy and frailty among persons with HIV. AIDS Care, 2021; 33(11):1492–1499; doi: 10.1080/09540121.2020.1813872

31.

McCrory

, Fiorito

, Hernandez

, et al. GrimAge outperforms other epigenetic clocks in the prediction of age-related clinical phenotypes and all-cause mortality. J Gerontol A Biol Sci Med Sci, 2021; 76(5):741–749; doi: 10.1093/gerona/glaa286

32.

Dye

, Wu

, Jackson

, et al. Epigenetic aging in older people living with HIV in Eswatini: A pilot study of HIV and lifestyle factors and epigenetic aging. Clin Epigenetics, 2024; 16(1):32; doi: 10.1186/s13148-024-01629-7

33.

Salameh

, Bejaoui

, El Hajj

. DNA methylation biomarkers in aging and age-related diseases. Front Genet, 2020; 11:171; doi: 10.3389/fgene.2020.00171

34.

Zhang

, Hu

, Aouizerat

, et al. Machine learning selected smoking-associated DNA methylation signatures that predict HIV prognosis and mortality. Clin Epigenetics, 2018; 10(1):155; doi: 10.1186/s13148-018-0591-z

35.

Esteban-Cantos

, Montejano

, Rodríguez-Centeno

, et al. Longitudinal changes in epigenetic age acceleration in Aviremic human immunodeficiency virus-infected recipients of long-term antiretroviral treatment. J Infect Dis, 2022; 225(2):287–294; doi: 10.1093/infdis/jiab338

36.

Zhang

, Hu

, Justice

, et al. DNA methylation signatures of illicit drug injection and hepatitis C are associated with HIV frailty. Nat Commun, 2017; 8(1):2243; doi: 10.1038/s41467-017-02326-1

37.

Moreno

, Martínez-Sanz

, Martín-Mateos

, et al. Global DNA methylation and telomere length as markers of accelerated aging in people living with HIV and non-alcoholic fatty liver disease. BMC Genomics, 2023; 24(1):567; doi: 10.1186/s12864-023-09653-2

38.

Ryan

, Wrigglesworth

, Loong

, et al. A systematic review and meta-analysis of environmental, lifestyle, and health factors associated with DNA methylation age. J Gerontol A Biol Sci Med Sci, 2020; 75(3):481–494; doi: 10.1093/gerona/glz099

39.

Gale

, Marioni

, Harris

, et al. DNA methylation and the epigenetic clock in relation to physical frailty in older people: The Lothian Birth Cohort 1936. Clin Epigenetics, 2018; 10(1):101; doi: 10.1186/s13148-018-0538-4

40.

Kuiper

, Polinder-Bos

, Bizzarri

, et al. Epigenetic and metabolomic biomarkers for biological age: A comparative analysis of mortality and frailty risk. J Gerontol A Biol Sci Med Sci, 2023; 78(10):1753–1762; doi: 10.1093/gerona/glad137

41.

Seligman

, Berry

, Lipsitz

, et al. Epigenetic age acceleration and change in frailty in MOBILIZE Boston. J Gerontol A Biol Sci Med Sci, 2022; 77(9):1760–1765; doi: 10.1093/gerona/glac019

42.

Duan

, Fu

, Sun

, et al. Epigenetic clock: A promising biomarker and practical tool in aging. Ageing Res Rev, 2022; 81:101743; doi: 10.1016/j.arr.2022.101743

43.

Peterson

, Collins

, Meier

HCS

, et al. Grip strength is inversely associated with DNA methylation age acceleration. J Cachexia Sarcopenia Muscle, 2023; 14(1):108–115; doi: 10.1002/jcsm.13110

44.

Phyo

AZZ

, Espinoza

, Murray

, et al. Epigenetic age acceleration and the risk of frailty, and persistent activities of daily living (ADL) disability. Age Ageing, 2024; 53(6):afae127; doi: 10.1093/ageing/afae127

45.

Verschoor

, Lin

DTS

, Kobor

, et al. Epigenetic age is associated with baseline and 3-year change in frailty in the Canadian Longitudinal Study on Aging. Clin Epigenetics, 2021; 13(1):163; doi: 10.1186/s13148-021-01150-1

46.

Horvath

, Erhart

, Brosch

, et al. Obesity accelerates epigenetic aging of human liver. Proc Natl Acad Sci U S A, 2014; 111(43):15538–15543; doi: 10.1073/pnas.1412759111