Changes in body mass index, weight, and waist-to-hip ratio over five years in HIV-positive individuals in the HIV Heart Aging Study compared to the general population

The prevalence of obesity and obesity-related comorbidities have been increasing worldwide over the last several decades [1, 2], including in people living with HIV (PLH) [3, 4]. However, in some cohort studies, PLH still have lower rates of obesity than age-matched HIV-negative controls [5]. Weight gain in PLH and the general population is multifactorial and associated with demographic factors, diet, exercise habits, psychological health, other comorbidities, and co-medications. Additionally, in PLH HIV disease parameters and the composition of antiretroviral therapies influence weight gain [6‐10]. HIV-associated wasting declined with the introduction of antiretroviral therapy (ART). Weight gain after initiation of ART has been associated with a reduced risk of mortality in individuals with underweight and individuals with normal weight [11]. While this increase in body weight with the initiation of ART is seen as a “return to health”, especially in patients with advanced disease [12], excessive weight and body-mass-index (BMI) increase in some individuals receiving contemporary ART may lead to obesity. So far, it has not been fully elucidated whether weight gain in aging PLH over time is substantially different from weight gain observed in the general population. Overweight and obesity are associated with several health issues, such as diabetes mellitus type II and cardiovascular disease [13].

Apart from body weight, BMI is generally used for evaluating overweight or obesity in the clinical setting, calculated from body weight and height. There are four categories of BMI that were established by the WHO in 1993: underweight, normal, overweight, and obese [14]. BMI increase can be associated with insulin resistance (IR), hyperglycemia, and new-onset diabetes mellitus type II, [15‐17]. Initiating ART with or switching to second generation integrase strand transfer inhibitors (INSTIs) and/or tenofovir alafenamide (TAF) has shown to be associated with more pronounced weight and BMI increases [8‐10, 18‐20]. Baseline ART was a predictor of post-switch weight gain. Using tenofovir disoproxil (TDF) was associated with reduced weight gain [21].

Nevertheless, BMI or body weight alone as an indicator of obesity does not differentiate between body lean mass and body fat mass since fat mass is not measured directly [22, 23]. Studies have shown that waist circumference correlates better with body fat than BMI, and thus, waist circumference alone or in combination with BMI has a stronger relation to specific health outcomes than BMI alone [24]. Other studies have shown associations between waist-to-hip ratio and cardiovascular events, metabolic risk factors, and death and suggested that it is superior to waist circumference alone, because it includes the hip circumference, too [25‐28].

Due to the various limitations of current analyses, more data and efforts are necessary to decipher the driving factors behind weight gain in PLH. It is important to define whether the general risk factors for weight gain and increased BMI have a similar impact on PLH compared to the general population. Most importantly, longitudinal data from clinical routines over more than one or two years are needed to describe the effects of weight changes.

The aim of our study was to compare changes in weight, BMI, and waist-to-hip-ratio (WHR) over a 5-year-follow-up period between PLH from the HIV Heart Aging (HIVH) cohort [29] and HIV-negative participants from the Heinz-Nixdorf-Recall (HNR) study [30], a cohort study from the same geographical area in Germany (Ruhr area) as the HIV Heart Aging study.

Within the analysis population, the matched participants in the HIVH and HNR study were predominantly male (~ 85%), with a mean age of 55 years. Further baseline characteristics of the HIVH and HNR study participants are shown in Table 1. Parameters such as blood lipids, weight, BMI, waist circumference, and hip circumference were lower in HIVH at baseline than in HNR. The prevalence of diabetes mellitus type II was also lower in HIVH compared to the HNR cohort (9.3% vs. 13.5% at baseline, respectively). The percentage of smoking participants in HIVH was higher than in HNR (44.5 vs. 28.8% at baseline, respectively). The absolute and relative differences after 5 years of weight, BMI, and waist-to-hip ratio (WHR) are also shown in Table 1. While HIVH participants had lower weight and BMI than the HNR participants at baseline, relative weight and BMI gain after 5 years was slightly higher in PLH (weight (kg):1.61 ± 6.7 vs. 1.00 ± 5.13; BMI (kg/m²): 0.59 ± 2.20 vs. 0.46 ± 1.72). However, WHR was higher in HIVH than HNR at baseline. The absolute WHR difference remained unchanged after 5 years in HIVH, but increased in HNR (0 ± 0.08 vs. 0.02 ± 0.06 units, respectively). The relative WHR difference after 5 years shows a decrease in WHR in PLH, while WHR in the general population increases (− 0.04 ± 8.12 vs. 2.08 ± 6.2%, respectively).

HIV-specific characteristics are shown in Table 2. Among the HIVH participants, the mean age at the diagnosis of the HIV infection was ~ 41 years, with men having sex with men (MSM) as the main HIV transmission risk category (61%). At baseline, more than one-third had at least one documented AIDS-defining event in their medical history. The vast majority of the participants (> 99%) received ART at baseline. During the observation period of five years, the use of antiretroviral drug classes was changing: There was an increase in the use of INSTIs (14.2–44.4%), while the use of NNRTIs (47.4–39.1%) and PIs decreased (47.4–29.0%). Among the NRTIs, at baseline only 1% of the PLH used tenofovir alafenamide (TAF) in their ART regimen. After 5 years, this had increased to 40.1%. Also, the use of single tablet regimens increased (16.5% to 40.6%). CD4/CD8 ratio improved over the 5 years of observation among HIVH (0.8 ± 0.43–1.01 ± 1.98). The proportion of participants with HIV RNA below the level of detection (< 50 copies/ml) was 93.3% at baseline and did not change within the five years of follow-up (93.9%). Lipodystrophy syndrome is divided into lipodystrophy, lipoatrophy and lipohypertrophy and were acquired in HIVH. At baseline 22.8% of the PLH from HIVH had lipodystrophy, and 22% had lipoatrophy. Furthermore, 10.3% of the HIVH participants had lipohypertrophy. During the observation time of 5 years, the percentages of lipoatrophy and lipohypertrophy reduced to 20.7% and 8.4%, respectively. Lipodystrophy increased by less than 1%.

Histograms for the absolute and relative weight, BMI, and WHR differences after 5 years in the two study populations are shown in Online Resource 2. The histograms for weight and BMI (Online Resource 2 a + b and c + d, respectively) showed a wider differences distribution after 5 years in HIVH than in HNR. For WHR the histograms reflected a reduction in WHR in HIVH in comparison to HNR (Online Resource 2 e + f).

Sankey bar diagrams in Fig. 1 visualize the shift between BMI categories during 5 years in the two study populations. HIVH and HNR differed regarding their percentage BMI categories distributions at baseline. While the HIVH study had more participants with a normal BMI (51%) and 2% participants in the underweight category, the HNR study had only 23% participants with normal BMI and no underweight participants. More than 50% of the HNR participants had overweight, and almost one-quarter had obesity at baseline. 36% of the HIVH participants were in the overweight BMI category, and 11% had obesity. After 5 years, the changes within the BMI categories were small in both HIVH and HNR, and we observed similar trends regarding the group shifts. The normal BMI category decreased after 5 years (HIVH: from 51 to 46% vs. HNR from 23 to 21%), while the group with obesity increased during the observation period (HIVH 15% vs. HNR 28%). BMI changes of more than one category rarely occurred (< 1%). Despite higher weight gains over time, the proportion of overweight and obesity in the HIVH cohort was still lower compared to the HNR participants. Excessive weight gain or loss of 10% or more occurred in ~ 20% of the participants in HIVH and ~ 10% of participants in HNR (Online Resource 3). 13.5% of the HIVH vs 6.2% of the HNR participants had ≥ 10% weight gain after 5 years, while 5.8% in the HIVH vs 2.8% in the HNR cohort lost ≥ 10% of their weight.

The Sankey diagrams in Fig. 2 show the shift in WHR categories over 5 years between HIVH and HNR using the categorization of the World Health Organization (WHO) Expert Consultation on Waist Circumference and Waist–Hip Ratio. They revealed that in both study groups the majority of the participants have an “increased risk of metabolic complications” at baseline according to their WHR (> 80%), which increased further after 5 years. The percentage of the HNR study participants with “normal risk of metabolic complications” according to their WHR changed from 17 to 12% over 5 years. “Normal risk for metabolic complications” in HIVH changed from 8 to 6%, therefore, a percentage twice as high as in HNR had an “increased risk of metabolic complications”, despite the lower baseline and post-5 years weight in HIVH.

The results of the linear regression models for assessing associations of HIV status with weight, BMI, and WHR at baseline and their changes after 5 years as outcome are shown in Table 3. A group difference in weight between HIVH and HNR at baseline of ~ 6 kg reflects the lower body weight in HIVH participants at baseline (95% Cl − 7.5 to − 4.6). The absolute change in weight after 5 years of follow-up was 0.61 kg (95% Cl 0.1 to 1.2), and the relative change was 0.85% (95% Cl 0.2 to 1.5) higher in HIVH compared to HNR. When adjusting for smoking the relative and absolute weight difference after five years was even more distinct (0.71 kg and 0.93%, respectively), demonstrating the influence of smoking on weight gain and the differences in smoking status between HIVH and HNR.

HIVH had a 2.6 kg/m² lower baseline BMI compared to HNR (95% Cl − 3.0 to − 2.2). The absolute change in BMI after 5 years of follow-up was 0.13 kg/m² (95% Cl − 0.1 to 0.3) higher in HIVH compared to HNR, while the relative change was 0.6% kg/m² (95% Cl − 0.1 to 1.3) higher in HIVH compared to HNR. After adjusting for smoking beta estimates were slightly higher.

We observed a difference in WHR between HIVH and HNR at baseline of ~ 0.03 units (95% Cl 0.02 to 0.03). The absolute and relative WHR change after 5 years of follow-up was lower by 0.02 units and 2.1% in HIVH compared to HNR, respectively. Beta estimates differed only slightly after adjustment for smoking.

After adjusting all regression models for parameters potentially affecting body weight change (i.e., diabetes mellitus type II, LDL, HDL, and systolic blood pressure) in addition to smoking, beta estimates for assessing the strength of association between HIV status and differences in anthropometric measures over 5 years were not affected, demonstrating that the observed differences between HIVH and HNR were independent of the parameters adjusted for (Table 3). This was also reflected by the results displayed in Online Resource 4. Here, absolute differences of parameters potentially affected by body weight change after 5 years were presented stratified by different categories of relative weight change. No strong differences in these parameters were observed between HIVH and HNR across weight change categories.

This study compared body weight, BMI, and WHR in PLH and individuals from the general population to show the changes in these parameters after 5 years of follow-up. We observed lower baseline body weight and BMI in PLH compared to the general population like in another trial with age-matched controls [5]. The absolute and relative differences after 5 years in weight and BMI were slightly higher in PLH as well. Smoking has been correlated with lower body weight in several studies [34]. After adjusting for the potential confounder “smoking”, the linear regression models revealed the underlying effect: HIV is associated with weight gain, with an increase of 0.7 kg weight (relative weight gain of ~ 1%) within 5 years in PLH compared to the general population. Other studies demonstrated similar findings, although the amount of weight gain was variable [5, 35‐37]. This effect is also observed for BMI, although not as strongly as for weight (absolute/relative BMI gain in PLH by 0.15 units/0.7%). According to prior studies, living with HIV has been associated with lower weight and body constitution at the time of diagnosis [5, 38, 39]. Due to improved ART medication and effective suppression of the virus, the weight gain is observed over time in ART-naive PLH and is considered part of the normal return to health following successful antiretroviral treatment [3, 40]. The weight gain in the first year after ART initiation was associated with survival benefits among initially underweight and normal-weight patients [41]. Prior studies could not clearly show if weight gain in PLH over time was substantially different from the general population. Excessive weight change (≥ 10%) was twice as frequent in HIVH compared to the HNR cohort after 5 years. Thereby, excessive weight gain was twice as frequent compared to excessive weight loss in both cohorts. Both excessive weight gain and weight loss occurred twice as frequently in HIVH participants compared to HNR participants during the observation period. Recently, a study by Garcia et al. showed that weight gain was primarily observed in PLH with BMI < 25 at diagnosis, suggesting that baseline weight, time since HIV diagnosis, and past weight history should be taken into account [5]. The comparison of PLH with the general population of the same age and the same geographic area shows that the general population in the German Ruhr area already has high rates of overweight and obesity, which are associated with health issues like diabetes mellitus type II and cardiovascular diseases [13]. Our results show that the proportion of overweight and obesity in PLH was still much lower than in the general population, suggesting that the BMI shift among PLH reflected the “return-to-health” respectively “coming-back-to-normal” effects of the ART. The observed BMI shifts after 5 years as a result of the weight gain in PLH shown in the Sankey bar charts might present the aligning of the BMI of PLH with the general population, but does not inevitably mean a “return to health”. In the cohort of the US veterans, the same trends in veterans living with HIV compared to HIV-negative veterans were observed during a 5-year follow-up period but the weight gain was described as transient [5].

The BMI allows the classification of patients as having overweight or having obesity and the comparability across different populations and areas. But the BMI is only of limited use for predicting cardio-metabolic risk because relevant factors such as muscle mass and fat distribution are not taken into account. Intra-abdominal visceral adiposity is associated with an elevated cardio-metabolic risk but does not always present with an increased BMI [42]. An Italian study has even demonstrated improved metabolic profiles in PLH in recent years compared with previous, despite increasing weight and BMI [43]. The WHR measures central adiposity more accurately and might be a more appropriate predictor of cardio-metabolic risk [42]. WHR in HIVH was higher at baseline than in participants from HNR representing the general population. HIV infection, some protease inhibitors, and first-generation NRTIs were associated with lipodystrophy syndrome. Metabolic syndrome, insulin resistance, and liver steatosis were common in PLH and associated with abdominal obesity [40, 44], which might explain the less favorable WHR in HIVH vs. HNR participants. The absolute and relative WHR differences within 5 years were lower in HIVH. Thus, increases in WHR were less pronounced in HIVH participants, which may reflect the already high WHR in HIVH compared to HNR. Moreover, we observed a decrease in lipohypertrophy as part of the lipodystrophy syndrome, which also reflects the decrease in WHR change after 5 years in PLH. The gain of fat in PLH treated with modern ART regimens including INSTI is generalized in most patients and not defined as lipodystrophy [40]. Previous studies observed no weight gain in PLH with baseline CD4 > 200 cells/µL and viral load < 100,000 copies/mL [45]. In our analysis, the majority of the participants in HIVH (> 93%) had a viral load of < 50 copies/ml and therefore the association with viral load was not analyzed here. We could, however, observe that younger age was associated with more weight gain [5] when stratifying for weight gain differences > 10% (see Online Resource 5).

While PLH had lower weight and BMI than the general population, weight, and BMI gain after 5 years was slightly higher in PLH. This effect is more pronounced after adjusting for smoking, which has a major impact on weight gain. This increasing weight difference after 5 years in PLH compared to the general population could be the result of the “return-to-health”, respectively, “coming-back-to-normal” effects of the treatment with ART. Moreover, we observed that WHR differences after 5 years are lower in PLH due to a higher WHR at baseline compared to the general population and thus increased risk of metabolic and cardiovascular complications according to WHO recommendations. During the 5 years of observation the lipodystrophy syndrome did not increase, but rather trends toward a decrease, which also affects WHR in PLH. Longitudinal data from clinical cohorts over more than one or two years are needed to describe the effects of weight changes in PLH.