Cardiometabolic risk among HIV-POSITIVE Ugandan adults: prevalence, predictors and effect of long-term antiretroviral therapy

Introduction We investigated the prevalence, predictors of and effect of Antiretroviral Therapy (ART) regimen on cardiometabolic risk among HIV-positive Ugandan adults at enrolment into a prospective cohort to study the Complications of Long-Term ART (CoLTART). Methods We collected data on cardiometabolic risk factors including dyslipidemia, hypertension, hyperglycemia, obesity and calculated the mean atherogenic index for Plasma (AIP) and 10 year Framingham risk score (FHS). Exposures were: ART regimen, duration on ART, demographic, socio-economic, behavioral, and life-style factors including smoking, physical activity and diet (including fruit and vegetables consumption). Results We enrolled 1024 participants, 65% female, mean age was 44.8 years (SD 8.0) and median duration on ART was 9.4 years (IQR 6.1-9.8). The prevalence of abdominal obesity was 52.6%, BMI≥25 kg/m2 -26.1%, hypertension-22.6%, high AIP-31.3% and FHS above 10% was 16.6%. The prevalence of low High Density Lipoprotein (HDL) was 37.5%, high Total cholesterol (Tc)-30.2%, high Low Density Lipoprotein (LDL) -23.6%, high Triglycerides (TG)-21.2%, low physical activity-46.4% and alcohol consumption-26.4%. In multivariate linear regression analyses, increasing age was associated with higher mean Tc, HDL, LDL, FHS (P<0.001) and hyperglycemia (p<0.005). In multivariate logistic regression analyses, Protease Inhibitor (PI) containing regimens were significantly associated with higher risks of abnormal: Tc, LDL, TG, AIP, abdominal obesity, hypertension, low HDL and lower risk of a FHS >10% compared to the non PI regimen. Conclusion ART increases cardiometabolic risk. Integration of routine assessment for cardiometabolic risk factors and preventive interventions into HIV care programs in resource-limited settings is recommended.


Introduction
among PLWH may be similar to the general population and include family history, age, male gender, hypertension, smoking, obesity, Diabetes mellitus and hyperlipidemia [4]. PLWH on ART have an estimated 10-year CVD risk greater than 20% [5]. The expected deaths attributable to CVD are projected to double to 2.4 million in 2030 relative to reports from 2000 [6]. Mortality due to CVD in SSA including Uganda is estimated to be threefold higher than in Western Europe [7]. In industrialised countries where ART has been provided for longer periods, ART agents particularly Protease Inhibitors (PI), have been associated with cardiometabolic risk factors, including impaired glucose metabolism, dyslipidemia, obesity and hypertension [8] but this has not been demonstrated in African settings. There is evidence to suggest that baseline metabolic profiles and associations between HIV, ART and cardiometabolic risk factors may differ between populations [9]. In Uganda, about 31700 deaths in 2002 were due to CVD and there is evidence of increasing CVD-related mortality from verbal autopsy data from among HIV infected people [10]. Conversely, a homebased cross-sectional study in eastern Uganda reported no effect of ART on CVD, but that study had few participants on ART, for a shorter duration and did not examine the specific effects of a PI regimen [11]. HIV prevalence in Uganda in 2014 was 7.4% and about 0.75 million (50%) of the 1.5 million PLWH were on ART of whom about 10% had HIV viral load above 1000 copies/ml [12].
The roll out of HIV viral load testing in Uganda and other African countries will increase detection of individuals with virological failure needing second line PI based ART [13]. In this study, we investigated the prevalence and predictors of, and the effect of ART on cardiometabolic risk among HIV-positive individuals who have been receiving ART for up to a decade. We specifically compared the adjusted mean differences and the risk of the various cardiometabolic risk factors between participants on a PI containing ART regimen and non PI ART regimen against the confounding effect of demographic, socio economic and behavioral factors.

Study population:
The study sample selection was based on nonprobability sampling of all HIV-positive adults aged 18 years and above, from two former HIV cohorts; the DART trial and RCC, who were receiving ART and consenting to undergo all study procedures.
Study information was given to potential participants by a study clerk. Individuals who were too sick to undergo study procedures or unable to give consent were excluded. Participants were either taking a PI containing ART regimen or Non PI containing regimen.
Non PI regimen included a standard two Nucleoside Reverse Transcriptase inhibitor (NRTI) and one Non-NRTI. Some participants from the DART trial cohort were taking a triple nucleoside therapy consisting of 3 NRTI agents. The PI containing regimen consisted of a ritonavir boosted PI mainly Lopinavir in combination with one or two NRTI. proportions of participants with a metabolic abnormality among those on a PI-containing ART regimen assuming that 10% of the participants in the non-PI containing ART regimen group have this abnormality. We also assumed a within group standard deviation of:

Measurements
(a) 1.0 mmol/L for Tc, (b) 0.8 mmol/L for LDL, (c). 0.5 mmol/L for HDL (d) 0.8 mmol/L for TG, (e) 1.6 mmol/L for FBG (f) 19 mm Hg for SBP and (g) 12 mm Hg for DBP. We assumed having at least 200 participants on a PI-containing ART regimen and at least 800 participants on a non PI ART regimen would provide sufficient power with which a given between group mean difference will be detected as statistically significant at the 5% level. were females (65%), most were aged 40 years and above (mean age 44.8, SD 8.0), 76% were on ART for 5 to 10 years (median 9.4 years, IQR 6.1-9.8). Participants were mainly self-employed or small scale business owners (40%), 45% had incomplete primary or no education and 42% were of low SES. The prevalence of current tobacco consumption was (7.8%), current alcohol consumption (26.4%), consumption of animal protein>3 days a week ( Figure 1 (C and D).

Discussion
In our Ugandan cohort of people on long-term ART, more than a half of participants had abdominal obesity, about one third had high BMI, high Tc, low HDL, high AIP, and almost a quarter had high TG, FHS > 10% and hypertension. Few participants had hyperglycemia.
Age, gender, physical activity, high viral load and PI-based ART regimen significantly affected cardiometabolic risk factors. The prevalence of hypertension compares with that reported in a South African study with similar median age [21], but higher than that reported from an ART cohort of younger adults in Rakai-Uganda (8.0%), Mbarara-Uganda (5.2%) and Kenya (8.8%) albeit lower than what was reported from Urban Uganda (27%)(24-26).
Although the differences might be attributed to variation in population characteristics and settings, the prevalence exceeded that from the background general population cohort (GPC)-13% [10]. The risk of hypertension was higher among individuals on non Age induced dyslipidemia is thought to be due to increased inflammation due to high levels of Tumor necrosis alpha (TNF-α) and interleukin-6 (IL-6) which interfere with lipid metabolism. Our men had higher mean TG consistent with studies in South Africa and the US [24,38]. We chose AIP as cardiovascular risk stratification to assess the influence of atherogenic dyslipidemia (TG and LDL) in predicting coronary atherosclerosis and cardiovascular risk in both hypertensive and normotensive individuals [22]. Higher mean AIP in men has also been reported among HIV patients in Nigeria [39], probably due to the elevated TG among men.
Consumption of animal protein was associated with elevation of both TG and AIP. The PI regimen was associated with higher risk of abnormal AIP probably due to PI associated dyslipidemia.
Participants in the urban study site had significantly higher mean LDL and Tc: HDL SBP, DBP, WC and BMI than those from the rural study site. This is in line with findings in the general population linking increasing urbanicity to higher prevalence of CVD risk factors

Competing interests
The authors declare no competing interest.       Data was missing on; SES 1 -(27 women and 10 men), occupation 1 (2 women and 1 man), Level of education 1 (1 woman and 3 men), consumption of Alcohol 1 and animal protein 1 (7 women and 5 men). Data was missing for; Tobacco consumption 2 (2 women and 2 men) and Physical activity 2 (2 women and 3 men). Data was Missing on; Serum Tc 3 , HDL 3 LDL 3 , TG 3 Glucose (6 women and 4 men). Calculation of AIP 4 excluded 6 women and 3 men with missing data. There was missing data on; HT 5 (10 Women and 4 men) BMI 5 (15 women and 7 men), WHR 5 and WC (11 women and 5 men), Baseline VL 6 (23 women and 11 men) and Baseline CD4 cell count 6 ( 53 women, 7 men)