Rate of Dyslipidemia Higher Among Women Living With HIV

A Comparison of Metabolic and Cardiovascular Health in a Cohort to Study Aging in HIV

EAB Russell; AYK Albert; HCF Côté; AYY Hsieh; A Nesbitt; AR Campbell; EJ Maan; J Brophy; N Pick; MCM Murray


HIV Medicine. 2020;21(7):418-428. 

In This Article


Study Design and Population

The CARMA study is an ongoing prospective cohort study of WLWH and HIV-negative controls that aims to investigate the effects of HIV and ART on cellular aging in women and children. Several studies have stemmed from this main study, including the CARMA-endocrine (ENDO) study, the aim of which was to examine the endocrine, metabolic and reproductive health of women and female youth living with HIV. The CARMA-ENDO study enrolled nonpregnant women and girls (age ≥ 12 years) who were either living with HIV or HIV-negative between January 2013 and August 2017. Enrolment occurred at two Canadian sites [the HIV Clinic at the Children's Hospital of Eastern Ontario in Ottawa, Ontario and the Oak Tree Clinic at British Columbia (BC) Women's Hospital in Vancouver, BC]. Control participants were recruited through Craigslist advertisements and through posters placed around BC Children's and Women's Hospital, the Vancouver Native Health Society, and the Downtown Eastside Women's Centre. All individuals who had participated previously in the CARMA cohort who had given consent to be contacted for future studies were invited to participate in the endocrine substudy when they returned for regularly scheduled clinic appointments, if HIV-positive, or were contacted by email or phone, if HIV-negative. The results presented herein represent a cross-sectional analysis of the data acquired during the CARMA-ENDO study visit. The study was approved by the University of British Columbia Research Ethics Board at the Children's and Women's Hospital (H08-02018) and the Children's Hospital of Eastern Ontario (CHEO) Ethics Board (#10/21E).

Data Collection and Testing

Demographic, clinical, anthropometric, laboratory and substance exposure data were collected prospectively during CARMA study visits. For WLWH, these visits took place during regular HIV care visits, at their CARMA-ENDO study visit. Data collected included important covariates and confounding variables such as ethnicity [African/Caribbean/black (ACB), white, indigenous or other/mixed/unknown], tobacco smoking (never, past or current), alcohol use (never, past or current), and hepatitis C virus (HCV) infection (never vs. past/current). HIV-specific covariates included peak HIV plasma viral load < 100 000 vs. ≥ 100 000 HIV-1 RNA copies/mL, CD4 count nadir, current CD4 count, and current nonnucleot(s)ide reverse transcriptase inhibitor (NNRTI), PI, integrase inhibitor or tenofovir use. Markers of cellular aging including relative LTL and mtDNA content as covariates were measured as described previously.[21,22]

Follicular stimulating hormone (FSH) was measured for those women reporting ≥ 12 months of amenorrhoea, either by enzyme-linked immunosorbent assay (ELISA) using the Enzo Life Sciences FSH kit (ENZO Life Sciences, Farmingdale, NY) or in the BC Children's Hospital lab (ELISA testing data had good correlation with hospital data). Laboratory testing included measurements of haemoglobin A1C (HbA1C), fasting blood sugar (FBS), random blood sugar (RBS), triglycerides, high-density lipoprotein (HDL) cholesterol, low-density lipoprotein (LDL) cholesterol, non-HDL cholesterol, apolipoprotein B, and total cholesterol. We collected fasting lipid values whenever possible. Triglyceride levels were only used if taken fasting; however, for all other lipid parameters, levels from fasting and nonfasting participants were included, commensurate with current guidelines.[23–26]

A significant proportion of WLWH in our cohort experienced prolonged amenorrhoea in the absence of menopause.[27] To account for this, we defined menopause based upon self-report of amenorrhoea for 1 year coupled with a measured FSH > 25 IU/L, as proposed previously.[28]


The prevalences of diabetes, medium/high Framingham risk score (FRS), hypertension, metabolic syndrome and dyslipidaemia were determined through a combination of self-report (medical history and medication profile), anthropometric measurements, chart review and laboratory testing. Calculation of FRS was performed as described elsewhere.[29] BP was measured according to Canadian Hypertension guidelines, using an automated BP cuff with the participant seated for at least 5 min prior to measurement.[30] Waist circumference was measured at end expiration, by placing a measuring tape around the abdomen at the level of the iliac crest, according to National Institute of Health guidelines.[31] Height and weight were measured in clinic; weight was measured on a calibrated scale, and height was measured on a calibrated stadiometer.

Diabetes, hypertension, metabolic syndrome and dyslipidaemia were considered present if the disorder had previously been diagnosed by a physician, if there was bloodwork evidence of the disorder based on current Canadian guidelines, or if the patient was taking medication consistent with the diagnosis (with the exception of those on metformin in the case of diabetes, as this medication can be used to treat other conditions). Table 1 shows detailed definitions of the cardiovascular risk factors. FRS was calculated for women ≥ 30 years old, as it has been validated for this age group.[29] Participants were stratified into low-, moderate- and high-risk groups based on their FRS, according to Canadian Cardiovascular Society guidelines.[26] Because of the effects of HIV infection on HDL cholesterol and triglycerides,[9,12] we used an expanded definition to determine the prevalence of dyslipidaemia in the sample, similar to that described previously.[23]

Statistical Methods

We compared demographic and clinical data between WLWH and HIV-negative women using Wilcoxon rank-sum and Fisher's exact tests for continuous and categorical variables, respectively. We investigated the relationship between cardiovascular and metabolic disorders (diabetes, moderate- to high-risk FRS, hypertension, metabolic syndrome and dyslipidaemia) and HIV status using logistic regressions while controlling for age. We also analysed relationships between the disorders and ethnicity, tobacco smoking, alcohol use, HCV infection, menopause, relative LTL, and blood mtDNA content, once again adjusting for age. Variables with a P < 0.1 univariably were considered in a multivariable model with HIV status to estimate the adjusted odds ratio (OR). Where subgroup sample sizes were sufficient, we tested for an interaction between HIV status and other significant variables to estimate potential moderation effects.

Within WLWH, we investigated the relationships between the outcomes and HIV-specific variables using logistic regression. Multivariable models were built as above and included age as a covariate. All analyses were carried out in R version 3.5.0 (2018-04-23).[34]