Associations of leucocyte telomere length with cardio-metabolic risk profile in a South African HIV-infected population

Abstract Leukocyte Telomere length (LTL) is an independent predictor of cardio-metabolic diseases (CMDs) and Human Immuno Virus (HIV) infection. However, studies are lacking on the association between LTL with CMD profile in people with HIV. Accordingly, we investigated the association between LTL and CMD profile in HIV-infected adult South Africans. This cross-sectional study included 728 HIV patients (20.6% men; median age 38 years) recruited across 17 public healthcare facilities in Cape Town. CMD markers were compared across quartiles of LTL, and spearman correlations assessed the continuous association of LTL with CMD markers. Linear and logistic regressions were then used to relate LTL with CMD risk profile, with appropriate adjustment for confounders. The prevalence of obesity, hypertension and diabetes were 34.8%, 36.8%, and 8.4%, respectively. In age, sex and body mass index adjusted models, increasing Log10LTL was associated with decreasing systolic (β = −10.52) and diastolic (β = −6.74) blood pressures, HOMA-β (β = −70.72), increasing total cholesterol (β = 0.544), non-high-density lipoprotein cholesterol (β = 0.472), and waist-to-height-ratio > 0.5 (odds ratio [OR] = 5.67), all P < .05. Compared to those in the bottom quarter, those in the top LTL quarter had lower prevalence of hypertension (OR = 0.65), and higher prevalence of total cholesterol > 5 mmol/L (OR = 1.94), and low-density lipoprotein-cholesterol > 3 mmol/L (OR = 1.62), all P < .05. LTL was not associated with diabetes nor general obesity. It was associated with Alanine Transaminase (ALT) and heart rate in univariable analyses. LTL shortening was associated with some CMD risk factors in HIV-infected adults on anti-retroviral therapy in South Africa. Prospective research is needed to explore the direction and implications of these associations.


Introduction
Telomere is a region of repetitive nucleotide sequences at each end of a chromosome, which protects the end of the chromosome from deterioration or from fusion with neighboring chromosomes. [1] Another important function of telomere involves enabling linear deoxyribonucleic acid (DNA) to replicate completely. [1] Therefore, in the absence of telomere, important genes would be deleted during each cell division, negatively affecting the process of DNA replication which is essential for life. During each cell division, a small portion of telomeric DNA is lost resulting in a decline in telomere length (TL) [1] indicating that TL decreases with age.
In addition to this natural process, the decrease in TL can be exacerbated by infectious and non-infectious diseases. Consistent findings have reported associations between short TL and numerous conditions such as HIV infection, [2,3] stroke, [4] ischaemic heart disease, [5] type 2 diabetes mellitus (hereafter referred to as diabetes), [6] and hypertension. [7,8] However, little is known about the relationship of TL with cardio-metabolic diseases (CMDs) in the HIV-infected population in South Africa, the country with the greatest number of people living with HIV (PLWH) globally. The estimated number of HIV cases in the South African population was ∼7.97 million in 2019. [9] South Africa has the largest antiretroviral therapy (ART) programme worldwide with 4.8 million PLWH receiving ART in 2018. [10] Consequently, early morbidity and mortality associated with HIV infection has declined following the widespread uptake of ART and, with longevity, PLWH are now exposed to noncommunicable diseases, particularly CMD. The development of CMD in PLWH is linked to ageing and exposures to the traditional lifestyle/behavioural risk factors as in general populations [11][12][13] together with chronic inflammation caused by HIV infection and side effects of ART. [14][15][16] This is important in the South African context where, diabetes, cardiovascular diseases, and HIV infection are the 2nd, 4th, and 5th highest contributors to mortality in the country, respectively. [17] In view of the converging burdens of HIV infection and CMD in a large proportion of the South African population, it is important to examine the impact of these comorbidities on health status. Considering that TL is a good indicator of biological age vs chronological age and reflects an individual's health status, examining differences in TL in PLWH with and without CMD provides a unique opportunity to evaluate these associations. Therefore, this cross-sectional study aimed to investigate the association of LTL with CMD risk profile in a sample of adult South Africans living with HIV infection.

Study design and population
In this cross-sectional study, HIV-infected participants ≥18 years of age and receiving ART were recruited from primary health care facilities in the Western Cape province of South Africa between March 2014 and February 2015. Sixty-two health facilities, 42 in Cape Town and the others from the surrounding rural municipalities, that provided ART to at-least 325 patients per month, were included. Amongst these 62 health facilities, 17 were randomly selected (4 rural) with 15 to 60 participants recruited from each facility.
The South African Medical Research Council Ethics Committee approved the study, which was conducted in accordance with the principles of the Declaration of Helsinki. The Health Research Office of the Western Cape Department of Health, and the selected healthcare facilities granted permission for participants to be recruited from the health facilities.

Data collection and sampling
Trained fieldworkers administered survey questionnaires, while clinical measurements and biochemical analyses were performed by clinicians and nurses. On the day of recruitment, sociodemographic information, physical assessments (anthropometry and blood pressure) and medical history of HIV infection (duration of diagnosed HIV infection, CD4 counts, and HIV treatment) were collected using a structured intervieweradministered questionnaire adapted from the WHO STEP-wise approach to Surveillance (STEPS) tool. Socio-demographic information was self-reported while medical history of HIV was obtained from medical records and by capturing medications brought to the clinics by participants.
Height, weight, waist circumference (WC), and hip circumference (HC) were measured in duplicate using standardised techniques. [18] Height to the nearest 0.1 cm and weight to the nearest 0.1 kg were taken using a Leicester Height Scale and A&D Personal Scaler, respectively, with the participants in light clothing and bare feet. WC and HC were measured to the nearest millimetre using a non-stretchable measuring tape. For each variable, the average of both measurements was used for analysis. Three blood pressure (BP) measurements were taken at 3-minute intervals using a digital automatic BP monitor after the participant had rested for at-least 5 minutes. The average of the second and third BP measurements was used for analysis.
Blood samples were collected the following day in Ethylene Diamine Tetra Acetate (EDTA) and dry tubes after participants had fasted for at least 8 hours. Serum and plasma were prepared from blood in dry tubes and a portion of blood in EDTA tubes and stored at À20°C for biochemical analyses. The other portion in the EDTA tube (5 mL) was stored at À80°C for LTL measurements. Biochemical parameters were measured at an ISO 15189 accredited pathology laboratory (PathCare, Reference Laboratory, Cape Town, South Africa) which had no access to participants' clinical information. Plasma glucose (hexokinase), serum creatinine (cayman chemical), gamma glutamyl transferase (GGT) (Abcam) were measured using colorimetric methods according to the manufacturer's protocol. Estimated Glomerular Filtration rate (eGFR) was calculated using the Modification of Diet in Renal Disease Study (MDRD) equation; eGFR = 175 Â Creatinine (mg/dL) À1.154 Â Age (years) À0.203 Â 0.742 (female). [19] Total cholesterol (TC), high-density lipoprotein cholesterol (HDL-C) and triglycerides were measured in serum using standard enzymatic techniques. [20][21][22] Low-density lipoprotein cholesterol (LDL-C) was calculated using the Friedewald equation, [23] while non-HDL-C was calculated using the formula: TC -HDL-C.
Plasma alanine transaminase (ALT) and aspartate transaminase (AST) levels were measured with ALT and AST reagent kits, respectively, according to the manufacturer's protocol (Thermo-Fisher Scientific). All colorimetric measurements were carried out using Beckman Coulter AU 500 spectrophotometer. Plasma insulin concentrations were measured by the Chemiluminescence Immunoassay method (Human insulin CLIA kit: Abnova), glycated haemoglobin (HbA1c) levels were determined using high-performance liquid chromatography technique (Thermo-Fisher Scientific) and highly sensitive c-reactive protein (hs-CRP) measured by Enzyme Linked Immuno-Sorbent Assay (ELISA) method (BIOMATIK) according to the manufacturer's protocol. For LTL assay, DNA was extracted from blood samples using the salt extraction technique. Briefly, 5 mL blood samples in EDTA tubes were defrosted to 4°C and poured into a 50 mL centrifuge tube. Thirty mL lysis buffer was added, and red blood cells were lysed by incubation on ice and vortexing. After lysis of red blood cells, the pellets were washed thrice with phosphate buffered saline which was later discarded. The pellets were then incubated with nuclear lysis buffer overnight at 60°C. The next day, the supernatant was collected, and the proteins precipitated using 1 mL saturated sodium chloride (6 M) solution. The supernatant containing the DNA was collected into new 15 mL centrifuge tubes and absolute ethanol added to precipitate the DNA by inversion. Precipitated DNA was removed and washed with 70% ethanol. After washing, the precipitate was dissolved in Tris-EDTA (TE) buffer and the concentration and quality of the DNA measured using a Nano drop. All samples with optical density from 1.7 to 2 were diluted to 5 mg/mL using polymerase chain reaction (PCR) grade water and TL measured by quantitative real time polymerase chain reaction (PCR) using the method described by O'Callaghan and Fenech. [24] Serial dilutions of the telomere standard and the single copy gene (36B4) standard were made. A master mix solution containing Power SYBR I (AmpliTaq Gold DNA polymerase, dNTPs, SYBR I Green Dye, optimised buffers and passive reference dye (ROX) (10 mL, 1Â), forward primer (1 mL, 0.1 m M), reverse primer (1 mL, 0.1 mM), ddH 2 O (4 uL) was prepared, mixed well and briefly centrifuged. Using a multichannel pipetted, 16 mL master mix solution were pipetted into each well of a 96 well plate. Into the corresponding wells were added 4 mL each of DNA sample, standards, positive and non-template control (distilled water). The plate was sealed with an optical clear film, centrifuged briefly and run using the following PCR conditions; 10 minutes at 95°C, followed by 40 cycles of 95°C for 15 seconds 60°C for 1 minute, followed by a dissociation (or melt) curve. At the end of the run, the plate was removed and discarded. Each sample was amplified twice, using telomere forward and reverse primers and the single copy gene forward and reverse primers. After amplification is completed the AB software produces a value for each reaction that is equivalent to kb/ reaction based on the telomere standard curve values. The kb/ reaction for telomere and genome copies/reaction for diploid genome copy values are exported as used to calculate the LTL as follows; LTL ¼ telomere kilobase per reaction value diploid genome copy number

Statistical analysis
Participants' characteristics were summarized as medians (25th-75th percentiles) for continuous variables and as counts (percentages) for categorical variables. Comparison of baseline characteristics by sex was carried out using Mann-Whitney U test. TL was categorised into four quartiles and the linear trend in CMD profile (continuous variables) across the different quartiles of TL was computed using the median test. Similarly, chi square test was computed and the Cochrane trend test was used to assess the linear trend in proportions across the quartiles of TL. Spearman correlation was used to assess the association between TL and cardio-metabolic parameters. Linear and logistic regressions were then used to investigate the independent association of TL with CMD risk profile, while accounting for the possible effects of confounders. All analyses were carried out using statistical packages for social science (SPSS) Version 21.

Clinical characteristic of the study sample
The clinical characteristics of the study participants are shown in Table 1. Amongst the 833 participants included in the study, TL data were available for 728 of whom 93% provided information on ART. The participants' average age was 36 years with men significantly older than women (41 years vs 37 years, P < .001). The median CD4 count was 400 cells/mm 3 with counts significantly higher in women (410 cells/mm 3 ) compared to men (272 cells/mm 3 ), P = .001. However, LTL was not significantly different between men and women (P = .214). Amongst the participants receiving ART (n = 675), the majority were receiving 1st line ART (56.3%) with the distribution of ART regimens differing in men and women (P = .005). All measures of obesity were significantly higher in women compared with men (P < .001). However, the prevalence of hypertension (36.8%) and diabetes (8.4%) were similar in men and women. Among participants with hypertension (n = 268) and diabetes (n = 61), the majority (60.8% and 54.1%, respectively) were newly diagnosed.

Clinical characteristics across TL quartiles
TL was categorised into four quartiles with the first quartile being the shortest and the 4th quartile the longest. SBP, DBP, hypertension prevalence, and HOMA-b decreased significantly across increasing quartiles of TL (P < .05 for all) ( Table 2). Moreover, correlation analysis showed a significant negative correlation between TL and all BP parameters (SBP, DBP, hypertension), ALT and HOMA-b. Surprisingly, heart rate was higher in the two longer TL quartiles compared with the shorter TL quartiles. Lipid profile, markers of liver function (ALT and AST), diabetes, HbA1c, and obesity determined by BMI, WC, WHR, and WHtR across quartiles of TL were not significantly different (Table 2).

Table 3
Linear regression models (coefficients and standard errors) for the associations of Log 10 telomere length with cardio-metabolic variables.
Age  Table 5). Quartiles of TL were associated with the prevalence of hypertension and dyslipidaemia (hypercholesterolaemia and raised LDL-C and non-HDL-C) ( Table 6). TL was not significantly associated with diabetes or obesity defined by BMI, WC, and WHR.

Discussion
TL shortening was significantly associated with some CMD risk factors in HIV infected participants in this study, even after accounting for the confounding effects of age, gender, and BMI. The significant associations with shorter TL included 1. BP variables and prevalent hypertension, 2. hypercholesterolaemia and levels of TC and LDL-C, and 3. insulin secretion defects (HOMA-b).
Furthermore, elevated ALT, a liver enzyme, was inversely and significantly correlated with shorter TL.
Our results accord with other studies that have shown associations between TL and CMD in diverse populations. TL shortening was associated with hypertension and/or increased systolic and/or diastolic BPs in cross-sectional studies in general populations of Lebanon, [8] China, [32] Taiwan [33] and USA adults, [34,35] and children. [36] The possible mechanism of action is bi-directional: short TL may induce hypertension, while factors that contribute to hypertension may enhance telomere shortening. Short TL may induce endothelial cells and vascular smooth muscle cells dysfunction, and insulin resistance which contribute to the development of raised BP and hypertension. [32,37] On the other hand, oxidative stress and inflammation, which are risk factors for hypertension, also lead to telomere shortening. [38] Although causality and the pathway for TL shortening cannot be established in this cross-sectional study, the significant negative correlation between TL and ALT shown may suggest the influence of oxidative stress on TL shortening. High levels of ALT, a marker of abnormal liver function and indicative of hepatocyte damage, often precede oxidative stress. [39] This suggests that oxidative stress may have contributed to telomere shortening in this study.
The association of short TL with hypercholesterolaemia and rising LDL-C in this study may be via the substantial contribution of altered serum lipids to systemic inflammation and oxidative stress, which in turn induces telomere shortening. [38] Moreover, hypercholesterolaemia is associated with cellular damage, chronic subclinical inflammation, and cell replication, which lead to telomere shortening. [40] Similar findings have been reported in diverse populations in cross-sectional studies from China, [41] Iran [42] and United States. [34] Our study reported a linear inverse association between TL and HOMA-b, which increases with beta cell dysfunction and precedes progression to diabetes. [43] Short telomeres may lead to premature b-cell death, resulting in reduced b-cell mass, impaired insulin secretion, [44] and diabetes. However, no association was found between short TL and diabetes in this study, which may likely be due to the relatively few participants with diabetes (n = 61) in the sample. Nevertheless, the association between short TL and diabetes has consistently been reported in the literature. [45] The association between TL and obesity is controversial as reveal by a systematic review with 38% of the studies showing no association. [46] Most of the studies used BMI to define obesity and all studies with association reported a significant negative association between TL and obesity. In our study, the association between TL and obesity was found only with WHtR > 0.5. [46] Regression analysis adjusting for age and sex showed an association between short TL and WHtR > 0.5. WHtR may be a better predictor for obesity and CMDs such as raised BP, Table 4 Linear regression for the associations of telomere length quartiles with cardio-metabolic variables. Legend: DBP = diastolic blood pressure, HOMA-b = Homeostasis model assessment of beta cell function, LDL-C = low-density lipoprotein cholesterol, Non-HDL-C = non-high-density lipoprotein cholesterol, SBP = Systolic blood pressure, T C = Total cholesterol. * * * = P < .001 * * = P < .01 * = P < .05.

Table 5
Logistic regression models (odds ratios and 95% confidence intervals) for the associations of Log 10 telomere length with cardiometabolic conditions. dyslipidaemia and altered glucose metabolism in this population than BMI or WC, which were not associated with short TL. [47] The association of telomere shortening with WHtR, but not WC or BMI, likely add to the body of literature that supports the use of WHtR to identify increased cardiometabolic risk.
Our study did not find any association between TL and age probably because of the narrow age range (median age 38 years, IQR = 32-45). However, TL has been widely reported as a marker of ageing characterised by TL shortening with older age. [48][49][50][51][52] Nevertheless, the cardio-metabolic abnormalities of obesity (WHtR > 0.5), HOMA-b, lipids levels (rising LDL-C and rising TC), higher systolic and diastolic BPs and hypertension were significantly associated with short TL and ageing in our study.

Strength and limitations
The main limitation of this study is the cross-sectional design which prevents inferring causal associations between TL and CMD risk profile. Moreover, the absence of a control group precludes controlling for the effects of HIV and ART which influence both TL and CMDs. The underlying mechanisms are only hypothesised since markers of oxidative stress and inflammation, and telomerase activity were not measured. Considering that most HIV-related studies in Africa are singleclinic-based, a major strength of this study is the inclusion of multiple healthcare facilities. Moreover, the included healthcare facilities, selected using random sampling methods, were based in both urban and rural areas, which allows for the generalizability of the results to other South African HIVinfected populations.
Including a control group in future studies to control for the effect of HIV and ART is essential. The direction of the association, if established in longitudinal studies, will enable the use of TL as a possible biomarker for the early identification hypertension, dyslipidaemia and diabetes in PLWH on ART in South Africa.

Conclusions
The association of TL shortening with hypertension, dyslipidaemia and defects in insulin secretion in PLWH on ART in South Africa suggests that TL may be used as an early biomarker of these CMDs in this population. More research is needed to explore the direction of these associations in longitudinal studies, and to examine differences in TL shortening in HIV-infected ART-naïve and HIV-uninfected populations with CMDs. Such data will provide greater insights on the contributors to TL shortening and may perhaps lead to new therapies in future.