Trends in body mass index in the pre-dolutegravir period in South Africa

Background Antiretroviral therapy (ART) is associated with weight gain, but this has been shown to be more marked with dolutegravir and other integrase strand transfer inhibitors. Objectives We studied weight gain in people living with HIV (PLWH) on ART compared to the general population in the period before dolutegravir was introduced in a rural South African cohort. Method Longitudinal analysis of the Ndlovu Cohort Study including 36–48 months’ follow-up data. From 2014 to 2019, data were collected annually in Limpopo, rural South Africa. Linear mixed models using HIV status, demographics, ART use and cardiovascular risk factors were used to estimate trends in body mass index (BMI) over time. Results In total, 1518 adult, non-pregnant participants were included, of whom 518 were PLWH on ART (79.8%), 135 PLWH not yet on ART (20.2%) and 865 HIV-negative. HIV-negative participants had significantly higher BMIs than PLWH on ART at all study visits. There was a significant increase in BMI in all subgroups after 36 months (PLWH on ART, BMI +1.2 kg/m2, P < 0.001; PLWH not on ART, BMI +1.8 kg/m2, P < 0.001 and HIV-negative, BMI +1.3 kg/m2, P < 0.001). Conclusion The increase in BMI in PLWH and HIV-negative participants is a serious warning signal as obesity results in morbidity and mortality.


Introduction
The world is facing an obesity epidemic.Since 1980, the prevalence of obesity has doubled in more than 70 countries, resulting in approximately 600 million adult people with obesity in 2015. 1 Overweight and obesity are risk factors for diabetes mellitus and cardiovascular diseases and result in an increased risk of premature mortality. 2,3Not only high-income countries suffer from this global pandemic.Trends in rising obesity rates in low-and middle-income countries have been reported as well. 4In 2016, 31% of men older than 15 years were overweight or obese in South Africa, and 68% of women. 5e obesity epidemic intersects with the HIV epidemic.At the end of 2019, 38 million people were living with an HIV infection globally. 6The most severely affected region is sub-Saharan Africa (SSA), where about two-thirds of all people living with HIV (PLWH) reside. 6Since the introduction of combined antiretroviral therapy (ART) in 1996, HIV-related morbidity and mortality has decreased and life expectancy increased. 7Together with an increased life expectancy, the risk of age-and lifestyle-related comorbidities, including obesity, also increased. 8 2018, the World Health Organization recommended to change first-line ART to an integrase strand transfer inhibitor (INSTI)-based regimen. 9The introduction of dolutegravir and other Background: Antiretroviral therapy (ART) is associated with weight gain, but this has been shown to be more marked with dolutegravir and other integrase strand transfer inhibitors.

Trends in body mass index in the pre-dolutegravir period in South Africa
Read online: Scan this QR code with your smart phone or mobile device to read online.

What this study adds:
Weight gain is a concern in HIV-positive people using integrase strand transfer inhibitors (INSTIs).This study shows that there was significant weight gain in both HIV-positive and HIV-negative people before the introduction of INSTIs, indicating that weight gain is a population-wide problem necessitating health interventions.

Measurements and definitions
Data were collected on demographics, medical history and medication use using standardised questionnaires.Blood samples were drawn to measure CD4 cell count and viral load (VL).All participants were invited for annual follow-up visits for up to 48 months, during which anthropometric measurements, including height and weight, CD4 cell count and VL were measured again.The same scales were used during the survey.Body mass index (BMI, kg/m 2 ) was calculated using height at baseline.BMI was classified as 'underweight' (BMI < 18.5 kg/m 2 ), 'normal weight' (BMI 18.5-24.9kg/m 2 ), 'overweight' (BMI 25.0-29.9kg/m 2 ) or 'obese' (BMI ≥ 30.0 kg/m 2 ).Intake of vegetables and fruit, used as a proxy for healthy food, was categorised as: 'poor' (< 2 servings/day), 'intermediate' (2-4 servings/day) or 'good' (≥ 5 servings/day).Monthly income was categorised in three categories: less than R648.00 (South African rand) (≈$46.00)(below the poverty line), between R648.00 and R992.00, and more than R992.00, as defined by Statistics South Africa in the period of study enrolment. 14Employment status was defined as: 'unemployed', 'self-employed' or 'other' (student, retired, volunteer).Relationship status was defined as 'stable', including married, cohabiting or having a life partner, or 'unstable', including divorced, single, widowed or multiple partners.The highest level of education level was categorised as 'none', 'primary', 'secondary and matric' and 'college and university'.Physical activity levels were measured with the International Physical Activity Questionnaire and categorised as 'low', 'moderate' or 'high'. 15,16

Statistical analysis
Demographics were reported as mean and standard deviation, median with interquartile range or count with percentage, as appropriate.At baseline, participants were divided in three groups: 'PLWH on ART', 'PLWH not on ART', and 'HIV-negative'.Differences in demographics and clinical characteristics between groups at baseline was presented using descriptive statistics.Participants on firstand second-line ART were combined, since the percentage of participants on second-line ART was too small to be analysed separately.We excluded participants with missing data on ART at baseline, participants with only a single visit, and female participants who reported themselves to be pregnant at any visit.
The trend in BMI over time was analysed with linear mixed models (estimated with maximum likelihood).We used two different approaches to categorise patients, both aiming to calculate estimated marginal means for BMI by HIV and ART status.In the first approach, assignment to a group was flexible over time, for example depending on HIV and ART status at a specific visit.Participants could possibly, therefore, change groups every follow-up visit.ART use was defined in two different ways.In model 1, HIV status and self-reported ART status was used to define the following three groups: 'PLWH on ART', 'PLWH not on ART' and 'HIV-negative'.In model 2, VL was used as a proxy for ART use.A VL < 1000 copies/mL was classified as 'PLWH on ART', a VL ≥ 1000 copies/mL as 'PLWH not on ART'.
In the second approach (model 3), assignment to a group was fixed.Participants were assigned to a group according to HIV status and VL at baseline and follow-up, so participants could not vary between groups during follow-up.PLWH on ART with VL < 1000 copies/mL at baseline, who remained virally suppressed (VL < 1000 copies/mL) during follow-up, were classified as 'PLWH on stable ART'.

Ethical considerations
Ethical clearance to conduct this study was obtained from the University of Pretoria, Faculty of Health Sciences Research Ethics Committee (No. 227/2014).

Results
The NCS included 1927 participants.We excluded three participants due to missing ART information at baseline, 342 participants without any follow-up visit and 64 female participants due to pregnancy at any point in the study.In total, 1518 participants were included ( ).The distribution of population characteristics at 36 months did not differ from the distribution of these characteristics at baseline (Table 2).On average, all groups, regardless of HIV and ART status, gained weight during study follow-up (Figure 1). Figure 2 shows the increase in percentage of PLWH on ART with overweight during study follow-up (18.9% at baseline versus 25.9% at 36 months) and obesity (13.5% at baseline versus 20.5% at 36 months).For HIV-negative participants, the percentage of overweight people did not change much (21.4% at baseline and 21.6% at 36 months), but obesity increased from 19.8% at baseline to 26.1% at 36 months.
Model 1 (Figure 3) shows the trends in BMI over time, where groups are defined using self-reported ART status.All subgroups had a significant weight gain at 36 months compared to the baseline visit (PLWH on ART, BMI +1.2 kg/m 2 , P < 0.001; PLWH not on ART, BMI +1.9 kg/m 2 , P < 0.001; HIV-negative, BMI +1.3 kg/m 2 , P < 0.001).
Weight gain was significant in the first 12 months in PLWH on ART (BMI +0.3 kg/m 2 , P = 0.001).HIV-negative participants had a significantly higher BMI compared to PLWH on ART at all study visits (after 36 months BMI +1.8 kg/m 2 , P < 0.001).Model 2 (Figure 1-A1), using VL as a proxy for ART use, showed similar trends in BMI compared to model 1.
In all models, female gender, higher educational status (university and college), currently not smoking, fruit and vegetable intake as dietary proxy and age were significantly associated with a higher BMI (P < 0.05).The initial BMI was inversely correlated with the duration of ART treatment (P < 0.05) (Appendix 1, Table 1-A1).

Discussion
Weight-gain trajectories in PLWH on or initiating ART compared to HIV-negative participants were similar and significant.This suggests that the weight gain seen in various observational studies and randomised controlled trials (RCTs) is, at least in part, a return to an obesogenic population trajectory.This observation aligns with the findings of other observational studies, affirming a sustained increase in BMI over time in patients, regardless of their HIV status. 17,18,19ere is ample evidence that PLWH starting INSTI gain significantly more weight compared to PLWH using a firstline ART regimen without INSTI. 10,20,21,22However, most studies addressing weight gain in people on INSTI-based ART lack comparison with HIV-negative controls, which makes it hard to address excess weight gain compared to the non-HIV-positive population.RCTs conducted in ART-naïve PLWH in Johannesburg, South Africa, and in Yaoundé, Cameroon, showed a significant increase in weight over 96 weeks in people on a dolutegravir-emtricitabine-tenofovircontaining regimen compared to the standard care group receiving a tenofovir disoproxil fumarate and efavirenz-based regimen. 22,23The trend in weight gain observed in people on non-INSTI regiments during 96 weeks follow-up in the South African RCT is comparable to our study results (at 96 weeks, participants on non-INSTI containing ART increased 2.3 kg versus 2.4 kg in PLWH initiating ART in our study after 24 months). 23The Swiss Cohort study, an observational study between 1990 and 2012 with 1601 PLWH (80% male, predominantly men who have sex with men), found that BMI increased most steeply within the first year of ART use (BMI +0.92 kg/m 2 , 95% CI: 0.8-1.0kg/m 2 ) whereafter BMI continued to increase, but at a lower rate (BMI +0.31 kg/m 2 per year, 95% CI: 0.29-0.34kg/m 2 ). 24The steep increase in weight in the first 12 months is in line with what we found in PLWH initiating ART (BMI +0.52 kg/m 2 per year, 95% CI: 0.06 -0.98 kg/m 2 ), but in our study the curve did not flatten after 12 months, unlike in the Swiss Cohort study.The continued increase in BMI observed in our study could be related to the obesogenic environment, differences in population demographics such as ethnicity and gender, as well as agerelated increase in BMI. 25 The North American AIDS Cohort Collaboration on Research and Design study was another observational study which analysed BMI of 14 084 PLWH (83% male, 57% non-Caucasian) between 1998 and 2010. 26rends in BMI were compared to BMI trends in the general United States population.During the study period, the prevalence of obesity increased more in PLWH on ART compared to HIV-negative participants, although weight increased in both groups (in 1998, prevalence of obesity was 9% in PLWH on ART versus 22% in HIV-negative participants and in 2010, 18% of PLWH on ART were obese versus 27% controls).In line with these findings, we found an increased percentage in obesity rates in both PLWH on ART and HIVnegative participants over time, namely a 7.0% increase in In our second model (Figure 1-A1), VL was used as a proxy for ART use to account for therapy non-adherence and ART failure.As the trends in BMI were similar to the trends seen in model 1 (Figure 3), the use of self-reported ART information seems to be reliable.
Although we did not find excessive weight gain in PLWH compared to HIV-negative participants, the increase in BMI in general was significant.As the prevalence of obesity is rising, the risk of comorbidities like cardiovascular diseases, type 2 diabetes mellitus, hypertension, sleep apnoea and some malignancies is increasing. 27As a result, overweight or obese people have an increased risk of all-cause mortality. 28ight gain in our study was most pronounced in women, regardless of HIV status.Other studies on PLWH, including studies addressing INSTI-based ART regiments, also observed excessive weight gain in women compared to men. 22,23,29,30,31In HIV-negative women in South Africa, poverty in childhood and the lack of access to resources in adulthood life are associated with higher obesity rates compared to HIV-negative men. 32Other factors influencing the increase in body weight are urbanisation, unhealthy diets linked to availability of fast food, and poverty and social perceptions. 33

Strengths and limitations
Our study is the largest study in rural SSA investigating BMI over time, including both PLWH and HIV-negative controls.Another strength is the implementation of a linear    on small numbers from the first follow-up visit.This is, however, not the case in the model where ART use is defined based on VL, as that model provides insight into the group of PLWH that is either therapy non-compliant or resistant to ART.Interpretation of both models provides comprehensive insight into the effect of continued viraemia on BMI.Finally, we used models in which participants' allocation to a group could change based on self-reported ART and VL per visit, and a model where allocation to a group was fixed over time.Therefore, our models are suitable to make both general statements about the trend in BMI over time, as well as individualised statements (e.g., the expected trajectory of a participant based on viral suppression during study follow-up).The main limitation in our study is the loss to follow-up percentage of 25.7% at 36 months.However, loss to follow-up is regarded to be missing at random, and this is supported by the finding that the distribution of population characteristics at 36 months is more or less the same as at baseline.The second limitation in our study is the self-reported ART use during follow-up.Although we used VL as proxy for ART use, more accurate data on ART use as well as information on specific ART regimens would possibly have added valuable data.A third limitation is the low prevalence of second-line ART use in our study.Therefore, we could not study the impact of different ART regimens on BMI.

Conclusion
Pre-INSTI efavirenz-based ART regimens did not result in excessive weight gain in PLWH compared to HIV-negative participants.Among PLWH on or initiating ART, BMI increased significantly in the first 12 months.Over the course of 36 months, there was a significant increase in BMI in the whole population, similar across groups.This is a serious warning signal as obesity results in morbidity and mortality.
Policymakers and healthcare workers should prioritise awareness and intervention campaigns to combat the increasing prevalence of unhealthy body weight.Future studies should focus on INSTI-related weight gain in PLWH compared to HIVnegative controls in SSA to seek out whether an increase in BMI is more significant in PLWH than in HIV-negative controls.

Note:
The figure illustrates the relative percentage of participants having underweight, normal weight, overweight or obesity during 48 months of study follow up.PLWH, people living with HIV; ART, antiretroviral therapy.

FIGURE 3 :
FIGURE 3: Body mass index over time by self-reported antiretroviral therapy (model 1).

TABLE 1 :
Distribution of population characteristics at baseline.
2, P = 0.001; HIV-negative; BMI +1.3 kg/m 2 , P < 0.001).PLWH either ART non-adherence or therapy resistant (BMI +0.3 kg/m 2 , P = 0.229) and seroconverters (BMI +0.7 kg/m 2 , P = 0.147) had no significant weight gain in 36 months.In the first 12 months, weight gain was most pronounced in PLWH who http://www.sajhivmed.org.zaOpen Access Note: Mean body mass per study year stratified by HIV status and ART used in the Ndlovu Cohort Study.PLWH, people living with HIV; ART, antiretroviral therapy.Body mass index over time in the Ndlovu Cohort Study.