Plasma Cell-Free Mitochondrial DNA as a Marker of Geriatric Syndromes in Older Adults with HIV

Abstract

Background:

Older people with HIV (PWH) experience more comorbidities and geriatric syndromes than their HIV-negative peers, perhaps due to residual inflammation despite suppressive antiretroviral therapy. Cell-free mitochondrial DNA (cfmtDNA) released during necrosis-mediated cell death potentially acts as both mediator and marker of inflammatory dysregulation. Thus, we evaluated plasma cfmtDNA as a potential biomarker of geriatric syndromes.

Methods:

Participants underwent the Montreal Cognitive Assessment (MoCA), frailty testing, and measurement of plasma cfmtDNA by qPCR and inflammatory markers including C-reactive protein (CRP), interleukin-6 (IL-6), interferon gamma (IFNγ), and tumor necrosis factor alpha (TNFα) in this cross-sectional study.

Results:

Across 155 participants, the median age was 60 years (Q1, Q3: 56, 64), one third were female and 92% had HIV-1 viral load <200 copies/ml. Median MoCA score was 24 (21, 27). Plasma cfmtDNA level was higher in those with cognitive impairment (MoCA <23) (p=0.02 by t-test) and remained significantly associated with cognitive impairment in a multivariable logistic regression model controlling for age, sex, race, CD4 T-cell nadir, HIV-1 viremia and depression. Two-thirds of participants met criteria for a pre-frail or frail state; higher plasma cfmtDNA was associated with slow walk and exhaustion but not overall frailty state. Cognitive dysfunction was not associated with CRP, IL-6, IFNγ, or TNFα, and frailty state was only associated with IL-6.

Conclusions:

Plasma cfmtDNA may have a role as a novel biomarker of cognitive dysfunction and key components of frailty. Longitudinal investigation of cfmtDNA is warranted to assess its utility as a biomarker of geriatric syndromes in older PWH.

Introduction:

Globally an estimated 7.5 million people over age 50 live with HIV¹. Although HIV-related mortality has decreased in individuals treated with antiretroviral therapy, older people with HIV (PWH) experience more comorbidities and geriatric syndromes than their age-matched HIV-negative peers^2,3, including neurocognitive impairment and frailty^2,4,5,6. Despite pharmacologic suppression of HIV viremia, comorbidities may be related to ongoing inflammatory pathophysiology. Persistent, low grade inflammatory changes with aging, termed “inflammaging”, are multifactorial and not fully understood⁷. Recent advances in understanding of cellular death processes, including necroptosis (necrosis-mediated cell death) have allowed for measurement of novel biomarkers of cellular death and inflammation, including cell-free mitochondrial DNA (cfmtDNA)^8,9. CfmtDNA has been demonstrated to be both a marker and mediator of systemic stress in the context of acute illnesses such as sepsis¹⁰, and it plays a significant role in cell signaling¹¹. Release of cfmtDNA due to cellular stress and/or necroptosis can lead to activation of innate immunity via multiple mechanisms, including activation of Toll-Like Receptor 9, the cGAS-STING pathway, and other innate immune pathways that stimulate pro-inflammatory genes to propagate the innate antiviral response^12,13. These inflammatory pathways have widespread physiologic effects and have been postulated to mediate geriatric syndromes including cognitive impairment and frailty^3,14.

In this study, we sought to assess cfmtDNA in plasma as a potential biomarker of geriatric syndromes including cognitive dysfunction and frailty in the setting of HIV and aging. We assessed the relationships between cfmtDNA and clinical laboratory markers of HIV/AIDS disease, cognitive function, frailty, and body composition. We hypothesized that older PWH with greater circulating cfmtDNA in plasma would have lower CD4 T-cell counts and CD4:CD8 ratios, lower cognitive performance, and higher prevalence of frailty.

Methods:

Older PWH age 50 and older were randomly selected (with oversampling of those age 60 and older) from a large urban academic medical center outpatient HIV clinical practice and invited to complete a detailed questionnaire focusing on health status, quality of life, psychosocial factors, and current substance use¹⁵. Participants age 55 and older who completed the questionnaire were then invited to participate in a substudy consisting of a cognitive evaluation using the Montreal Cognitive Assessment (MoCA)¹⁶, frailty testing, and bioelectric impedance analysis (BIA) to assess body composition [Figure 1]. The Veterans Aging Cohort Study (VACS) Index, a validated predictor of morbidity and mortality for people with HIV¹⁷, was calculated to assess the burden of comorbidities. Participants were asked to fast for at least 8 hours prior to study visit, blood samples were collected at the beginning of the study visit, and then participants were served a meal and water after the specimen collection prior to undergoing other assessments. All study participants provided written informed consent, and the study was approved by the Weill Cornell Medical College Institutional Review Board.

Figure 1:

Study Enrollment Diagram

Montreal Cognitive Assessment (MoCA) Testing:

The MoCA was administered to assess cognitive function. Scores on individual domains of the MoCA including visuospatial/executive function (5 points maximum), naming (3 points maximum), attention (6 points maximum), language (3 points maximum), abstraction (2 points maximum), delayed recall (5 points maximum), and orientation (6 points maximum). Scores were tabulated to compute the composite MoCA score, and one point was added to the score if the participant reported an education level of high school or less¹⁶. The MoCA was scored out of a total of 30 points, and scores <23 were categorized as cognitively impaired^16,18

Frailty Testing:

The frailty phenotype was assessed as previously described by Fried et al¹⁹. Participants completed a timed 4-meter walk, and slowness was determined by the average of two readings of the 4-meter walk: men ≤173 cm and women ≤159 cm in height who required >6.22 seconds, or men >173 cm and women >159 cm who required ≥5.33 seconds to complete the walk met criterion for slowness²⁰. Measurement of dominant-hand grip strength was assessed using dynamometer; participants completed three trials and the average was computed. Questions about exhaustion, level of physical activity and unintentional weight loss were completed in the questionnaire. Missing weight data were obtained from chart review. If 0 criteria were met the participant was classified as nonfrail, if 1-2 criteria were met they were classified as pre-frail, and if 3-5 criteria were met they were classified as frail, as previously described¹⁹.

Depression and Substance Use Assessment:

We considered depression and substance use as potential confounders of cognitive function. Depressive symptoms were assessed using the 10-item Center for Epidemiologic Studies Depression Scale (CES-D-10). Participants with scores of 10 or greater were dichotomized as having significant depressive symptoms²¹. Substance use was assessed in the psychosocial survey. If participants reported consuming alcohol, they were asked questions about alcohol use behavior with the AUDIT-C questionnaire²². Recreational drug use was considered to be active use if they reported use of marijuana, cocaine, crystal methamphetamine, poppers, or heroin within the three months prior to survey assessment.

Plasma CfmtDNA Measurement:

Fasting blood samples were drawn into chilled tubes and immediately placed onto ice. If phlebotomy was unsuccessful, or if participants were unwilling to provide a blood sample during the study visit, their data were excluded from analyses of cfmtDNA. Plasma was isolated from whole blood within 4 hours of being drawn by centrifuging samples at 490g at 4°C for 10 minutes. Samples were stored at −80°C prior to thawing and processing.

Samples were thawed on ice and vortexed briefly to mix, then 50 μL of plasma was diluted with 170 μL of PBS. Samples were then centrifuged at 700g at 4°C for 5 minutes, and for plasma samples, the supernatant was centrifuged again at 18,000g at 4°C for 15 minutes to remove cells and cellular debris. DNA was isolated using 180 μL of the supernatant with the QIAamp DSP DNA Blood Mini Kit (#61104; Qiagen), according to the manufacturer’s microcentrifuge protocol. Then 200 μL of the provided elution buffer was used to collect the DNA. Prior to quantification, the DNA solution was diluted 1:5 with nuclease-free deionized, distilled water. CfmtDNA levels were measured by SYBR Green dye-based qPCR assay using a PRISM 7500 sequence detection system (Applied Biosystems) as previously described¹⁰. Briefly, the following primer sequences were used: human NADH dehydrogenase 1 gene (forward 5’-ATACCCATGGCCAACCTC-3’, reverse 5’-GGGCCTTTGCGTAGTTGTAT-3’). For absolute quantitation of cfmtDNA, a standard curve was generated from a 125 base pair oligonucleotide containing the qPCR product in serial dilutions (Integrated DNA Technologies). Following this, 5 μL of sample was added to each well in triplicate along with a no-template control. The thermal profile was carried out with 2 minutes at 50°C, 10 minutes at 95°C, 40 cycles for 15 seconds at 95°C, and 1 minute at 60°C. The concentration of cfmtDNA in plasma was calculated in copies per μL.

Inflammatory Marker Measurement:

Serum samples were stored at −80° C prior to assaying. Interleukin-6 (IL-6), interferon gamma (IFNγ), and tumor necrosis factor alpha (TNFα) were assayed using a multiplex kit (K15052G), and C-reactive protein (CRP) using a singleplex kit (K151STG) from MesoScale Discovery (Rockville, MD). Assays were run in duplicate with quality controls; 10% were repeated for confirmation. The intra-assay coefficients of variation (CV) for CRP, IL-6, IFN-γ, and TNF-α ranged from 2.4% to 6.4%, and inter-assay CVs ranged from 5.0%-10.2%. The detection limits of CRP, IL-6, IFN-γ, and TNF-α, were 0.01 ng/mL, 0.10 pg/mL, 0.40 pg/mL, and 0.20 pg/mL respectively.

Body Composition Assessment:

Body composition was assessed with tetrapolar bioelectrical impedance analysis (BIA, Quantum IV, RJL Systems, Inc., Clinton Township, MI) by trained study personnel. BIA was not conducted on participants with implanted electrical devices or implanted metal of unknown type. Electrical leads were placed on the right hand, wrist, ankle, and foot. Participants relaxed supine with their right arm at their side, not touching the rest of the body, and legs not touching one another for approximately one minute while the bioelectrical impedance data were collected. Raw data were analyzed and reported using the RJL Systems software (BC4 version 4.2.0). Skeletal muscle index and fat mass index were calculated by dividing the BIA results for these parameters by height (in meters) squared²³.

Statistical Methods:

Given the absence of prior data on cfmtDNA in this or a similar population, we calculated a generic sample size calculation. With a planned sample size of 200 participants there was 80% power to detect an effect size (Cohen’s d) of 0.28, using a t-test with a 0.05 two-sided significance level. This is generally considered to be a small to medium effect size. Accrual was halted at 164 participants, falling short of the goal of 200, due to funding constraints.

Statistical analysis was conducted using Stata/IC version 15.1 (StataCorp LLC, College Station, Texas). Comparisons of continuous variables between two groups were assessed using t-tests or the Wilcoxon-Rank Sum test as appropriate. Spearman correlations were used to evaluate associations between continuous variables. The nonparametric test of trend was used to assess for statistical significance with ordinal dependent variables (e.g., frailty category) and continuous independent variables (e.g., age). Missing data points were omitted, except in the case of CD4 T-cell nadir, which was imputed to the median value if nadir was not available.

Logistic and linear regression models were used to assess for relationships between cfmtDNA and dichotomous and continuous outcomes, respectively, with adjustment for potential confounders determined a priori. The outcome variable was MoCA score, dichotomized as “low” (<23) or “normal” (23 and above). Models that included body composition testing results were adjusted for age and sex²⁴. Reported p-values are 2-sided and not adjusted for multiple comparisons.

Results:

There were 164 participants in this study. As summarized in Figure 1, one participant did not complete MoCA testing and eight had a preferred language that was not English; these nine participants were excluded from the analysis, leaving 155 participants with data for the primary analysis focusing on cognitive function. Table 1 summarizes their demographic and clinical characteristics. The median age was 60 years old (Q1, Q3: 56, 64); the lower age bound for recruitment was 55 years old, however two participants age 54 who were turning 55 that calendar year were permitted to enroll. Overall, 92% had HIV-1 RNA levels < 200 copies/ml.

Table 1.

Participant Demographics

Measure	MoCA <23 N=52	MoCA ≥23 N=103	Full Sample N=155
Age, median (years)	60 (57, 65)	60 (56, 63)	60 (56, 64)
Female sex	18 (35%)	34 (33%)	52 (34%)
Self- Reported Race
Black	30 (58%)	49 (48%)	79 (51%)
White	7 (13%)	37 (36%)	44 (28%)
Other	14 (27%)	14 (13%)	28 (18%)
Declined	1 (2%)	3 (3%)	4 (3%)
Ethnicity
Hispanic/Latinx	17 (33%)	18 (17%)	35 (25%)
Non-Hispanic/Latinx	29 (56%)	69 (67%)	98 (61%)
Declined	6 (11%)	16 (16%)	22 (14%)
Years with HIV	25 (22, 27)	25 (21, 29)	25 (22, 29)
HIV-1 viral load <200 copies/ml	47 (90%)	96 (93%)	143 (92%)
CD4 T-Cell count (cells/mm3)	594 (419, 829)	574 (342, 795)	583 (365, 811)
CD4 T-Cell nadir (cells/mm3)	124 (34, 253)	116 (46, 229)	118 (43, 240)
Plasma cfmtDNA	230 [202-324]	192 [171-219]	210 [189-237]
Veterans aging cohort study (VACS) score	31 (18, 44)	28 (18, 39)	28+ (18, 43)
Active tobacco smoking	9 (17%)	17 (16%)	26 (17%)
Depression	26 (50%)	52 (51%)	78 (51%)
High alcohol useϕ	7 (14%)	15 (15%)	47 (30%)
Recreational drug use within 3 months	15 (29%)	32 (31%)	47 (30%)

Continuous data are expressed as median (Quartile(Q)1, Q3). Plasma cfmtDNA was Ln-transformed and reported as geometric mean [95% Confidence Interval]. Depression was assessed with the CES-D-10 screen²¹.

⁺VACS of 28 corresponds to a 10.8% risk of mortality in 5 years¹⁷.

^ϕHigh alcohol use defined by sex-specific AUDIT-C score²². Data were missing for some participants for CD4 T-cell Nadir (n=6), plasma cfmtDNA (n=3), and depression (n=2).

Plasma cfmtDNA:

Plasma cfmtDNA was measured in 152 participants and natural log transformed due to rightward skew. Geometric mean cfmtDNA level in plasma was 210 copies/μl [95% CI: 189-237]. There were no statistically significant relationships between CD4 T-cell count, CD4:CD8 ratio, nor CD4 T-cell nadir, with plasma cfmtDNA (Spearman correlation rho (σ)=0.08, p=0.35; σ=0.05, p=0.50; σ=0.03, p=0.67, respectively). Plasma cfmtDNA not associated with HIV-1 viremia defined as detectable >200 cp/ul (p=0.23, Wilcoxon rank-sum test), or correlated with age (σ=0.00, p=0.96) or VACS index (σ= −0.08, p=0.33).

Cognitive Function:

The median MoCA score was 24 (Q1, Q3: 21, 27); 52 (34%) scored <23, indicating cognitive impairment. MoCA score was not correlated with age (σ= −0.07, p=0.36). Plasma cfmtDNA level was higher in those with MoCA <23 (p=0.02) by t-test [Figure 2]. There was no relationship between cognitive impairment and makers of inflammation, specifically CRP, IL-6, IFN-γ, or TNF-α (p=0.38, 0.27, 0.32, 0.39 by Wilcoxon rank-sum test).

Figure 2:

Plasma cfmtDNA by Cognitive Function Level

In a multivariable logistic regression model with cognitive dysfunction as the outcome, plasma cfmtDNA level was associated with cognitive impairment (OR 2.02 95% CI: 1.15-3.30] p=0.01) after adjustment for age, sex, CD4 T-cell nadir, HIV-1 viremia, and depression (Table 3).

Table 3:

Adjusted Association Between Plasma Cell-Free Mitochondrial DNA and Cognitive Dysfunction

Outcome: Cognitive Dysfunction
Predictor Variable	Odds Ratio	95% Confidence Interval	p-value
Plasma cfmtDNA*	2.02	1.17-3.47	0.01
Age (per year)	1.00	0.94-1.07	0.97
Male Sex	0.68	0.30-1.58	0.37
Race
Black (ref)	1.00
White	0.30	0.11-0.80	0.02
Other	1.83	0.72-4.64	0.20
CD4 T-cell Nadir (per 50 cells/mm3)	1.02	0.90-1.16	0.72
HIV-1 Viral Load >200 copies/ml	1.83	0.44-7.64	0.40
Depression	0.83	0.39-1.77	0.62

^*cfmtDNA is Ln-transformed.

In a subanalysis, MoCA domain scores were examined, including visuospatial/executive function, naming, attention, language, abstraction, delayed recall and orientation. Of the domains, abstraction had the strongest inverse correlation to plasma cfmtDNA, which was statistically significant (Spearman correlation rho=−0.19, p=0.02). The remaining domains did not demonstrate statistically significant correlations with plasma cfmtDNA levels [Table 2].

Table 2:

MoCA Subset Score Correlation with Plasma cfmtDNA

MoCA Subset	Spearman’s Rho	p-value
Visuospatial/Executive	−0.13	0.10
Naming	−0.05	0.52
Attention	−0.08	0.30
Language	−0.07	0.39
Abstraction	−0.19	0.02
Delayed Recall	−0.02	0.84
Orientation	0.07	0.40

Subsets of Montreal Cognitive Assessment testing in relation to plasma cfmtDNA level using Spearman Correlation.

Frailty:

Frailty testing data was available for 149 of 155 participants. Missing data were due to participants opting out of the Fried Frailty testing procedure because of pain in hands using the dynamometer (n = 5), or unanswered survey questions about physical activity and exhaustion (n = 1). Missing data, including composite frailty score for these participants, were excluded from the analysis of frailty data.

Frailty assessment revealed 52 (35%) were nonfrail, 80 (54%) were prefrail, and 17 (11%) were frail. Age was associated with greater frailty status by nonparametric test of trend (p<0.01). Those with slow walk and exhaustion had higher plasma cfmtDNA levels than those who did not meet these component criteria by t-test (geometric mean cfmtDNA 270 copies/μl [95% CI: 198-372] for slow walk compared to 198 copies/μl [95% CI: 176-224] without slow walk (p=0.03), and 252 copies/μl [95% CI: 204-308] for those with exhaustion compared to 192 copies/μl [95% CI: 169-224] without exhaustion (p=0.03). There were no statistically significant differences in mean plasma cfmtDNA levels according to weak grip (p=0.15), weight loss (p=0.83), low physical activity (p=0.33), or composite frailty score of non-frail as compared to pre-frail and frail (p=0.98) by t-test. Higher IL-6 level was related to greater frailty status (p<0.01 by nonparametric test of trend). Plasma cfmtDNA was not correlated with IL-6 level (rho=−0.04 p=0.58). Additionally, CRP, IFN-γ, and TNF-α were not associated with frailty status nor correlated with plasma cfmtDNA (data not shown).

Median Body mass index (BMI) was in the overweight range 27.0 (Q1, Q3: 24.1, 31.1). Body composition analysis with BIA was measured in a subset of 131 participants. In separate linear regression models adjusted for age and sex, skeletal muscle (SMI) and fat mass indices (FMI) were not associated with plasma cfmtDNA level (β=0.30, p=0.13; and β =−0.15, p=0.73, respectively). Those with greater SMI had a trend towards lower odds of slow walk in an unadjusted logistic regression model (OR 0.79 [95% CI: 0.61-1.01], p=0.07). In a multivariable logistic regression model, the trend between higher levels of plasma cfmtDNA and slow walk persisted after adjusting for SMI and age (OR 1.71 [CI: 0.92-3.19], p=0.09). BMI was correlated with SMI (Spearman’s rho=0.49, p<0.01) and FMI (rho=0.76, p<0.01)

Discussion:

In this study of older PWH, a substantial portion of study participants exhibited cognitive and physical vulnerabilities, as over one-third of study participants had a MoCA score consistent with cognitive impairment, and two-thirds met criteria for a pre-frail or frail state. Age was associated with greater frailty state in our study, as expected⁴, but not with cognitive impairment. The latter finding may be due to our use of a conservative MoCA cut-off of 23 to define cognitive impairment¹⁸. Higher levels of plasma cfmtDNA were associated with lower cognitive function and key components of frailty, specifically slow walk and exhaustion. These findings add to our prior data associating higher urine cfmtDNA concentrations with involuntary weight loss and lower skeletal muscle and fat mass indices²⁵.

Previous studies have demonstrated that older PWH develop more comorbid conditions and frailty than their HIV-negative peers, despite control of HIV viremia^3,4,26. The pathogenesis of geriatric syndromes in older adults with HIV is complex and remains incompletely understood, although it is likely due to multiple factors including chronic inflammation^27,28, mitochondrial dysfunction^29,30 and cellular death pathways associated with stimulating ongoing inflammation³¹. Recent data from the general population suggest the presence of cfmtDNA may stimulate pathways of immune activation that can lead to inflammation and autoimmune disease processes³², hence studying cfmtDNA in plasma and other body fluids and tissues is warranted to define the pathophysiology of inflammation in the setting of chronic disease.

While our study did not obtain cerebrospinal fluid (CSF) samples, prior findings indicate cfmtDNA levels in CSF were correlated with global deficit scores and inflammation in adults with HIV and neurocognitive impairment, but interestingly, not in those without cognitive impairment³³. Furthermore, in a recent study of post-mortem samples from adults without HIV, but with neurodegenerative disease, ventricular CSF cfmtDNA levels were higher in the more neuropathologically severe cases, suggesting a possible connection between neurodegeneration and release of cfmtDNA³⁴. Further research is needed to investigate the relationships between cfmtDNA in CSF and plasma in the setting of neuroinflammation.

Additionally, there was no correlation observed between plasma cfmtDNA and conventional serum markers of inflammation in our study. A previous study of women, and children with HIV of both sexes, demonstrated a relationship between cfmtDNA in plasma and IL-6 level³⁵, and a murine study demonstrated a relationship between oxidized cfmtDNA and TNF-α level³⁶. Our study may have been underpowered to detect what could be small to moderate differences in levels of serum inflammatory markers in relation to cfmtDNA.

Our study is unique in that it conducted a novel investigation of plasma cfmtDNA as a potential biomarker of geriatric syndromes in older PWH. A strength of our study is the diverse participant population, with half identifying as Black, and over one-third women. We also attempted to reduce selection bias by randomly selecting individuals from the entire patient pool to invite to the study, instead of generally recruiting from the practice.

There are several limitations to our study. Its population was derived from a single academic medical center clinic in New York City. While our random selection process was intended to mitigate selection bias, not everyone chose to participate, and it is possible that those who did may have been more health-conscious or more concerned about aging-related complications. Additionally, our data are limited by a lack of HIV-negative controls, which would help to determine whether plasma cfmtDNA shows similar links among those without HIV. Thirdly, our data represent a cross-sectional analysis and cannot address longitudinal relationships with cognitive function and frailty. Lastly, given that our analyses were based on an a priori hypothesis, we did not adjust for multiple comparisons.

In summary, our data suggest that plasma cfmtDNA may have a role as a potential novel biomarker in understanding the pathophysiology of geriatric syndromes in older PWH including cognitive dysfunction and key components of frailty. Due to the challenges in predicting geriatric syndromes such as frailty and cognitive dysfunction, there is considerable clinical and translational utility in identifying and validating biomarkers to prognosticate and aid in these diagnoses¹⁴. Predicting and managing cognitive dysfunction and frailty in older PWH remains challenging in the clinical setting, and the burden of co-morbidities including cognitive impairment and frailty is likely to grow as the population of older PWH continues to increase and age. Further research is warranted to investigate cfmtDNA as a biomarker of pathogenic inflammation and its relationship to cognitive impairment and frailty in older PWH.