Effects of exercise training on bone mineral density in adults living with HIV: a retrospective study

Abstract

Background: Evidence on the effects of exercise training on the bone health of men and women living with HIV (MLHIV and WLHIV) is limited.

Objective: To investigate the effects of a long-term multimodal exercise program on the bone mineral density (BMD) of MLHIV and WLHIV.

Methods: A retrospective cohort of 39 patients (13 women; 48.4 ± 7.6 y; HIV-infection for 15.5 ± 6.5 y; combined antiretroviral therapy for 12.2 ± 7.0 y) performed a multimodal exercise program (60-min sessions of aerobic, resistance, and flexibility exercises performed 3 times/week for 9–106 months). MLHIV and WLHIV were allocated into groups showing either advanced osteopenia/osteoporosis or normal BMD (+ or −).

Results: MLHIV+ increased BMD at the femoral neck, total femur, and lumbar spine (∼3–4%) compared to MLHIV− (p ≤ 0.03). Changes in whole-body BMD were similar between MLHIV groups (p = 0.55). WLHIV+ exhibited higher loss of BMD at the femoral neck (∼6%) than WLHIV− (p = 0.04), whereas reductions in the whole-body, total femur, and lumbar spine (∼3–5%) were similar between groups (p ≥ 0.25). Among men, changes in femoral neck BMD were inversely correlated to femoral neck T-score (r = −0.62; p < 0.001), but not to the time of follow-up, appendicular skeletal muscle mass (ASM) index, or age (p ≥ 0.08). In women, these changes were inversely correlated with time of follow-up (r = −0.58) and age (r = −0.70) and positively correlated with femoral neck T-score (r = 0.46) and ASM index (r = 0.47) (p < 0.01).

Conclusion: Multimodal exercise training may improve the BMD in people living with HIV, especially men with advanced osteopenia/osteoporosis. Adjuvant therapies to exercise should be considered to counteract losses in WLHIV.

Keywords:

• AIDS
• osteopenia
• osteoporosis
• physical exercise

Introduction

The Acquired Immunodeficiency Syndrome (AIDS), caused by Human Immunodeficiency Virus (HIV) was first described in 1983 and persists as a public health problem.^1,2 The advent of combined antiretroviral therapy (cART) has significantly increased the life span of people living with HIV (PLHIV).³ On the other hand, the prolonged exposure to both cART and HIV infection has been shown to increase the risk of non-AIDS-related illness.⁴ In this scenario, aging comorbidities appear to affect PLHIV earlier vs. non-infected populations.⁵

Accumulated evidence demonstrates that reductions in bone mineral density (BMD) among PLHIV are greater than in healthy counterparts, which may result in early osteoporosis.^6,7 A prior meta-analysis⁶ indicated an osteoporosis prevalence three times greater in PLHIV vs. uninfected controls, and a strong association between HIV infection and incidence of hip fracture.⁷ This is important given the relationship between osteoporotic fractures and disability/mortality.⁸ On the other hand, treating low BMD in PLHIV may be quite challenging, since regular pharmacological therapies may not be recommended due to their adverse effects.⁹ Therefore, non-pharmacological strategies for preventing or treating bone demineralization and fracture risk in PLHIV are warranted.¹⁰

Physical exercise plays a key role in the preservation of bone health, representing the only clinically relevant intervention capable to stimulate bone mineralization at low cost and without adverse effects.^9,11–13 Despite this, research on the impact of exercise training on bone health in PLHIV is limited.^14,15 We could locate only six experimental trials with this purpose.^16–21 Of these, three did not detect benefits on BMD or bone turnover in non-osteoporotic individuals that underwent short-term resistance training (≤16 weeks).^16–18 Three studies reported slight improvements (3–8%) in the lumbar spine, femoral neck, and proximal radius,^19,21 after resistance^19,21 or multimodal²⁰ training in relatively young patients (18–50 y) with osteopenia. Multimodal exercise training has been indicated for PLHIV given its potential to improve several aspects of physical fitness and cardiometabolic risk.^22,23 This modality of training is also acknowledged to preserve²⁴ or improve²⁵ the BMD in postmenopausal women. Despite this, we could find a single trial investigating the impact of combined resistance and aerobic training on the BMD of young patients (∼30–40 y)²⁰ – after 6 months, slight improvements vs. baseline but not vs. controls were detected.

In short, the effects of exercise training on the BMD of PLHIV are not consensual, which probably results from methodological limitations of the current literature. In general, trials applied exclusive resistance training of relatively short duration and included small samples not stratified by sex, which is a major factor influencing bone turnover.^26,27 Further research with multimodal exercise training performed for prolonged periods, and including patients of both sexes with advanced osteopenia is therefore warranted. Thus, we retrospectively investigated the effects of a long-term multimodal exercise program (9–106 months) on the BMD of men and women living with HIV. We hypothesized that a multimodal supervised exercise program would be effective to attenuate bone losses in PLHIV vs. controls. Additionally, associations of exercise-related effects vs. determinant factors of BMD were calculated.

Materials and methods

Study design and subjects

This is a retrospective cohort study based on information from patients’ medical records, obtained at the University of Rio de Janeiro State (UERJ) (RJ, Brazil). Between January 2010 and December 2019, 95 PLHIV enrolled in a multimodal supervised exercise program. Eligible participants were adults (>18 y old), diagnosed with HIV infection for at least three years, and treated with cART for at least one year. Exclusion criteria were a) physical training before enrollment in the supervised exercise program; b) less than 9 months of participation in the supervised exercise program, or attendance lower than 75% of training sessions; c) cardiovascular, respiratory, neurological, or musculoskeletal impairments limiting physical exercise; and d) use of hormonal replacement therapy or medications influencing bone metabolism.

Figure 1 summarizes the outflow of group selection and allocation. Of the 95 initial volunteers, 39 were eligible and stratified into groups according to sex and presence of osteopenia or osteoporosis, as follows: 1) men with osteopenia/osteoporosis (MLHIV+; n = 12; 6 with borderline or established osteoporosis); 2) men without osteopenia/osteoporosis (MLHIV−; n = 14); 3) women with osteopenia/osteoporosis (WLHIV+; n = 8; 1 with borderline or established osteoporosis); 4) women without osteopenia/osteoporosis (WLHIV−; n = 5). Diagnosis of osteopenia or osteoporosis was based on World Health Organization (WHO) criteria,²⁸ using T-score data from the dual X-ray absorptiometry (DXA) performed at the beginning of exercise intervention. Body fractioning was assessed every year. Post-training data were extracted from the last DXA performed within 9- to 106 months. This study was approved by the local institutional review board (CAAE 41957620.1.0000.5259), and all patients provided informed written consent regarding the use of medical data in future research.

Figure 1. Flow diagram of the study. HIV; Human Immunodeficiency Virus.

Exercise training program

The multimodal supervised exercise program took place at the University facilities and included aerobic, resistance, and flexibility exercises, as described elsewhere.^29,30 The exercise sessions were performed three times per week and included all exercise modalities, as follows: a) 5-min warm-up including overall calisthenics; b) 20–30 min of treadmill or cycle ergometer exercise with intensity corresponding to 50–80% of heart rate reserve; c) 20–30 min of resistance exercises for the upper, core and lower body (8–11 single and multi-joint exercises) performed using free weights or machines (Selection Line, Technogym^TM, Gambettola, Italy); d) 5–10 min cool down through stretching exercises for the major joints using the static method (2 sets of 6- to 8 exercises during 30 s). Resistance exercises were performed with 2–3 sets of 12–15 repetitions with loads corresponding to 80–90% of 1 repetition maximum (1-RM) tests. Exercises for the upper/core body were chest press, lat machine, shoulder press, biceps curl, triceps curl, and abdominal crunch, and exercises for the lower body were leg press, leg extension, leg curl, hip adduction, and abduction. Participants moved from one modality to another without resting intervals.

The workloads of aerobic and resistance exercises were adjusted every two months. Training sessions were supervised by trained professionals and occurred either in the morning or afternoon (9–11am or 2–4pm), according to the patient’s best convenience. The program took place during all months of the year, except for a vacation interval from mid-December to mid-January, between 2010 and 2019. During this period, patients were instructed to perform light physical activities, such as walking or cycling.

Anthropometry and BMD assessments

Body mass and height were measured using a calibrated electronic scale Filizola^TM (São Paulo, SP, Brazil) and wall stadiometer Sanny^TM (São Paulo, SP, Brazil), respectively. Body mass index (BMI) was calculated (Kg/m²). BMD was measured at the femoral neck, total femur, lumbar spine (L1–L4), and whole-body through DXA (DPX-IQ, Lunar Radiation Corporation^TM, Madison, WI, USA), following recommendations from the International Society for Clinical Densitometry.³¹ Body fat, lean mass, and appendicular skeletal muscle mass (ASM) were also assessed. The ASM index was calculated as the ratio between ASM and height squared.³² All DXA exams were performed by the same experienced technician blinded for the study purposes. Equipment was calibrated according to manufacturer instructions and scans had high resolution. Principles for body composition analyses with DXA have been described elsewhere.³³

Statistical analysis

An achieved statistical power of 0.64 was obtained by performing post hoc analysis (GPower^TM 3.0.10; Kiel, University of Kiel, Germany) based on the sample size, predefined p-value, and effect size of 0.75 for a BMD difference of 0.03 g.cm⁻² (standard deviation of 0.04 g.cm⁻²).²¹ Data normality was confirmed by the Shapiro Wilk test and therefore results are expressed as mean ± standard deviation (SD) for continuous variables, and frequencies for categorical variables. To check whether pre-training data had been influenced by the lack of adherence of 51 of the initial 95 participants, we compared all subjects including dropouts vs. those included in the study at baseline employing t-tests (basically, an intention-to-treat approach without the second evaluation).

Baseline characteristics of patients were compared using either unpaired Student-t or Fisher exact tests. Longitudinal changes during follow-up were compared between groups using unpaired t-tests. One-tailed Pearson correlation coefficients were calculated to verify the association between sample characteristics (femoral neck T-score, time of follow-up, ASM index, and age) vs. changes from baseline in femoral neck BMD. The BMD at this particular site is recommended by the WHO for identifying osteopenia or osteoporosis.²⁸ All calculations were performed using the GraphPad^TM software (Version 6.0, La Jolla, CA, USA), and statistical significance was set at p ≤ 0.05.

Results

In all cases, the results of the treatment intention were not different than the analysis for adherence to training. Baseline clinical characteristics and body composition of men and women are presented in Table 1. As expected, BMD for the whole body and at all regional sites was higher in MLHIV− vs. MLHIV+ (p ≤ 0.05, for all outcomes). On the other hand, no difference between groups was found for age, years diagnosed with HIV or cART treatment, T-CD4 count, cART regimen, body mass, body fat, lean body mass, ASM index, or BMI (p ≥ 0.10, for all outcomes). As for women, WLHIV+ and WLHIV − were similar in regards to all characteristics, except for age (WLHIV+ older, p = 0.01), femoral neck BMD, and total femur BMD (WLHIV+ lower, p < 0.01).

Table 1. Baseline clinical characteristics and body composition in PLHIV with osteopenia/osteoporosis (MLHIV + and WLHIV+) and without osteopenia/osteoporosis (MLHIV − and WLHIV−) (n = 39).

	MLHIV+ (n = 12)	MLHIV- (n = 14)	p-values	WLHIV+ (n = 8)	WLHIV- (n = 5)	p-values
Clinical characteristics
Age (y)	47.4 ± 6.3	48.2 ± 8.7	0.79	54.6 ± 3.7	42.6 ± 8.7	0.01
Years diagnosed with HIV	14.2 ± 5.4	18.6 ± 8.5	0.17	13.1 ± 3.9	13.7 ± 2.8	0.79
Years taking cART	11.4 ± 5.4	14.5 ± 9.0	0.34	11.7 ± 4.9	7.3 ± 5.7	0.19
T CD4 (cell.mm^−3)	647.0 ± 182.7	605.8 ± 414.0	0.82	719.4 ± 127.6	601.9 ± 522.0	0.60
cART regimen
PI (n, %)	9 (75)	9 (64)	0.68	4 (50)	3 (60)	1.00
NRTI (n, %)	12 (100)	13 (92)	1.00	8 (100)	5 (100)	1.00
NNRTI (n, %)	3 (25)	6 (42)	0.42	4 (50)	2 (40)	1.00
II (n, %)	1 (8)	1 (7)	1.00	0 (0)	0 (0)	1.00
Body composition
Body mass (kg)	71.3 ± 15.9	78.8 ± 11.0	0.17	54.8 ± 7.0	59.8 ± 9.6	0.30
Body fat (kg)	18.0 ± 8.7	19.1 ± 6.5	0.73	16.1 ± 3.7	18.6 ± 6.4	0.38
Lean body mass (kg)	51.5 ± 9.2	56.9 ± 6.1	0.10	36.4 ± 4.7	38.9 ± 5.4	0.39
ASM index (kg.m^−2)	12.1 ± 2.1	13.4 ± 1.5	0.10	8.6 ± 1.3	9.5 ± 0.8	0.22
BMI (kg.m^−2)	23.0 ± 3.7	25.0 ± 3.0	0.16	21.9 ± 2.7	23.7 ± 2.5	0.27
Whole body BMD (g.cm^−2)	1.02 ± 0.07	1.19 ± 0.07	<0.001	1.00 ± 0.12	1.07 ± 0.12	0.37
Femoral neck BMD (g.cm^−2)	0.82 ± 0.07	1.09 ± 0.08	<0.001	0.82 ± 0.05	1.01 ± 0.11	0.001
Total femur BMD (g.cm^−2)	0.83 ± 0.08	1.11 ± 0.09	<0.001	0.88 ± 0.08	1.09 ± 0.16	0.01
Lumbar spine BMD (g.cm^−2)	1.02 ± 0.15	1.21 ± 0.16	<0.01	0.99 ± 0.16	1.12 ± 0.27	0.29

Results are expressed as mean ± SD or %. p values in bold denote statistically significant differences within a given sex. HIV; Human Immunodeficiency Virus; cART, combined antiretroviral therapy; PI, protease inhibitors; NRTI, nucleoside reverse transcriptase inhibitor; NNRTI, non-nucleoside reverse transcriptase inhibitor; II, integrase inhibitors; ASM, appendicular skeletal muscle mass; BMI, body mass index; BMD, bone mineral density. Significant differences in bold.

Overall, the time of follow-up (or time under training) ranged from 9 to 106 months or 0.8 to 8 y (average 3.4 ± 2.2 y). No death was reported in this period, nor discomforts or injuries related to the exercise intervention. The mean duration of exercise training was similar between MLHIV groups (MLHIV+: 34.1 ± 23.8 vs. MLHIV−: 33.0 ± 25.9 months; p = 0.91), and WLHIV groups (WLHIV+: 42.5 ± 23.2 vs. WLHIV−: 43.0 ± 29.4 months; p = 0.97).

Figures 2 and 3 exhibit longitudinal changes from baseline for BMD in MLHIV and WLHIV groups, respectively. MLHIV+ exhibited greater increases than MLHIV − in BMD at femoral neck (0.01 ± 0.01 vs. −0.02 ± 0.02 g.cm⁻²; p < 0.001), total femur (0.01 ± 0.02 vs. −0.01 ± 0.02 g.cm⁻²; p < 0.01), and lumbar spine (0.03 ± 0.04 vs. −0.00 ± 0.02 g.cm⁻²; p = 0.03), while no difference was detected between groups for the whole-body assessment (0.00 ± 0.01 vs. 0.01 ± 0.03 g.cm⁻²; p = 0.55). In women, reductions in BMD at femoral neck were slightly greater in WLHIV+ vs. WLHIV− (−0.04 ± 0.03 vs. 0.00 ± 0.04 g.cm⁻²; p = 0.04), whereas no difference was observed for BMD at total femur (−0.03 ± 0.03 vs. −0.02 ± 0.04 g.cm⁻²; p = 0.59), lumbar spine (−0.05 ± 0.04 vs. −0.02 ± 0.02 g.cm⁻²; p = 0.25), or whole body (−0.03 ± 0.02 vs. −0.01 ± 0.04 g.cm⁻²; p = 0.35). The visual inspection of individual data in Figures 2 and 3 suggested that non-responders to exercise training were more prevalent in WLHIV+ than MLHIV+. Moreover, most of non-responders in WLHIV+ had follow-ups longer than 36 months (5 out of 8 patients).

Figure 2. Individual and mean data depicting the change from baseline (Δ) for bone mineral density for the whole body (A), femoral neck (B), total femur (C), and lumbar spine (D) of men with (MLHIV+) and without osteopenia/osteoporosis (MLHIV−). *p < 0.05.

Figure 3. Individual and mean data depicting the changes from baseline (Δ) for bone mineral density for the whole body (A), femoral neck (B), total femur (C), and lumbar spine (D) of women with (WLHIV+) and without osteopenia/osteoporosis (WLHIV−). *p < 0.05.

Figures 4 and 5 exhibit Pearson correlations between changes in BMD at femoral neck and selected baseline characteristics in MLHIV and WLHIV, respectively. In men, changes from baseline were inversely correlated to femoral neck T-score (r = −0.62; p < 0.001), but not to time of follow-up (r = −0.26; p = 0.09), ASM index (r = −0.28; p = 0.08), or age (r = 0.12; p = 0.26). In women, changes from baseline were inversely correlated to time of follow-up (r = −0.58; p = 0.01) and age (r = −0.70; p < 0.01), and positively correlated to femoral neck T-score (r = 0.46; p = 0.05) and ASM index (r = 0.47; p = 0.05).

Figure 4. Correlations between changes from baseline (Δ) for bone mineral density in femoral neck and baseline femoral neck T-score (A), time of follow-up (B), appendicular skeletal muscle mass index (C), and age (D) in men living with HIV.

Figure 5. Correlations between changes from baseline (Δ) for bone mineral density in femoral neck and baseline femoral neck T-score (A), time of follow-up (B), appendicular skeletal muscle mass index (C), and age (D) in women living with HIV.

Discussion

The present study investigated the effects of a prolonged multimodal exercise intervention on BMD in PLHIV with and without advanced osteopenia – seven patients in trained groups had borderline or established osteoporosis, mainly among men. Exercise-related effects on bone health seemed to rely on patients’ sex and BMD status at baseline, given that increases in BMD occurred only in men with advanced osteopenia. Correlations between changes in BMD vs. baseline characteristics were also different in men and women. In men, increases in BMD were greater when the baseline level was poorer, irrespective of other factors. In women, BMD outcomes always tended to decrease. Those reductions appeared to be more pronounced in older women with longer follow-up, and among those with lower BMD at baseline.

There is little research on the effects of exercise training on the bone health of PLHIV, and their data are mixed. Favorable outcomes were reported by three trials,^19–21 while in the other three improvements in BMD or bone turnover did not occur.^16–18 Interestingly, experiments that failed to detect increases in BMD included young eugonadal patients (not exhibiting advanced osteopenia)¹⁶ and applied short-term exclusive resistance training. Of the studies reporting increased BMD due to exercise intervention, one applied multimodal training²⁰ and two applied exclusive resistance training for 12 weeks to non-osteoporotic patients aged 18–50 y – increases of 3% in the BMD of lumbar spine,^19,21 8% in the femoral neck, and 5% in the proximal radius²¹ were observed.

Multimodal exercise training has been shown to preserve or improve BMD in older adults, postmenopausal women, or cancer survivors.^24,25,34,35 This modality of intervention has been also indicated for PLHIV instead of exclusive exercise training.^22,23 Despite this, we could find a single randomized trial²⁰ investigating the effects of combined resistance and aerobic training on the BMD of patients showing poor BMD. Improvements vs. baseline in the lumbar spine, femoral neck, and 1/3 radius were detected after 6 months of intervention, concomitant with increased IGF-1 and lowered TNF-α, IL-6, and myostatin. Changes in BMD were similar between men and women. However, it worth noticing that differences vs. controls have not been found for any outcome. In the present study, increases of approximately 3- to 4% were observed in the BMD of MLHIV+ vs. MLHIV−, which is similar to prior research.^19–21 In women, the BMD remained mostly stable with reductions of about 6% in the femoral neck among WLHIV+ compared to WLHIV−.

Our findings contribute to the current knowledge by demonstrating that improved bone health in PLHIV following chronic multimodal exercise depends on the baseline BMD, at least in men. This is relevant for exercise prescription, considering that PLHIV with normal BMD may require higher exercise intensity or volume to benefit from significant osteogenic gains. Another important aspect of our study refers to the characteristics of exercise training – we could not locate prior interventions combining aerobic, resistance, and stretching exercises that improved BMD. In addition, in comparison with the current literature, the time of follow-up was markedly longer, and outcomes were compared between individuals with borderline/established osteoporosis vs. normal BMD. Previous trials included patients with low BMD, sometimes with comorbidities as lipodystrophy,^17,21 or wasting syndrome,¹⁶ but the short interventions along with non-osteoporotic samples limited their findings. Anyway, evidence about the effects of multimodal training on the BMD of PLHIV is still insufficient, and further studies are warranted.

As abovementioned, increases in BMD were observed only in men with advanced osteopenia at baseline. Given the link between bone loss, immune system activation, and inflammation,^9,36 MLHIV with poorer levels of BMD would be more likely to present higher levels of pro-inflammatory cytokines triggered by chronic HIV infection, and thereby experience greater anti-inflammatory and osteogenic effects due to exercise training. Along with this possibility, there is evidence that exercise-related increases in BMD might be concomitant with decreased inflammation levels in either general population,^37,38 or PLHIV.²⁰ Unfortunately, measurements of blood inflammatory markers, endocrine response, and bone formation/resorption could not be presently performed, which would help to explain our findings.

In both men and women, the BMD status at baseline significantly correlated with changes resulting from exercise intervention. Men with lower BMD showed the greatest increases, while women with lower BMD showed the greatest reductions. The poorer BMD at baseline among men might have contributed to a larger ‘training window’ in comparison with those with greater BMD. In this sense, it is worth noticing that half of the trained men had borderline or established osteoporosis, while this proportion was of only 1/8 among women. On the other hand, the training progression might not have been adequate to elicit skeletal gains and overcome BMD losses in women, particularly in ages close to menopause.^39,40 Menopause dramatically reduces estrogen, which is largely responsible for regulating and maintaining the bone turnover in women.²⁶ Postmenopausal women have been shown to exhibit an impaired osteogenic response to mechanical strain, associated with lowered estrogen levels and receptors sensitivity.^27,41 In the present study, women assigned to WLHV+ were older than WLHIV− (please, refer to Table 1), and were at an age compatible with menopause at baseline (of 8 women, 7 were > 50 y, and 4 > 55 y). At the end of the intervention, many of them were over 60 y old, which might have hampered the effects of exercise training. This possibility is reinforced by the fact that most meta-analyses report low overall effect size of exercise-related BMD improvement in postmenopausal women,^13,24,25 and some research has even observed negative effects.⁴²

A recent meta-analysis suggested that the effects of exercise on the BMD of postmenopausal women might be diluted by inadequate exercise protocols,¹³ and this cannot be discarded in the case of our study. The load progression may have been insufficient to stimulate bone synthesis to a greater extent than the losses – increases in BMD of estrogen-deficient women seem to occur in response to high-intensity, but not moderate-intensity progressive resistance loads.^13,24 Moreover, improvements in bone turnover are related to the overall volume (total amount of weight lifted) within chronic resistance training,^12,41 which has not been individually quantified in our study. Dose-response relationships of exercise training to stimulate osteogenesis in young and older men and women with PLHIV, as well as the optimal loading characteristics to preserve their skeletal integrity, must be determined by future research.

Those features may have also influenced the association between BMD and time of follow-up (e.g. time under training) among women. The negative correlation found in WLHIV+ could be disputable, given the accumulated evidence suggesting that physically active patients would be more likely to exhibit preserved muscle mass and bone health.^22,43,44 However, the time of HIV infection and cART treatment are acknowledged to accelerate muscle losses in PLHIV, predisposing to overall frailty and compromising bone health.^6,9,45 Bias due to collinearity between age (at the end of the follow-up, some women were > 65 y), time of HIV infection under cART, and prolonged follow-up (8 y in some cases) might have concurred to produce an inverse relationship between the duration of training intervention and BMD among women. This premise is reinforced by the finding that the majority of unresponsive women were followed for more than 36 months.

The small sample size, especially of women, along with the lack of a non-exercise control group, are the major limitations of the present study. However, as the prevalence of HIV infection is lower among women than men,^46,47 trials including large sample sizes of WLHIV are limited.⁴⁸ The interval for workload adjustment (every two months) along with the wide range of intervention durations (<10 to > 100 months) may have influenced the BMD responses, particularly among women – WLHIV had more non-responders than MLHIV, especially when the time of follow-up was over 36 months. Bias related to this feature may have affected the dose-response relationship, with inadequate stimuli for osteogenic improvement in women. Moreover, although participants have been encouraged to maintain healthy nutritional habits, their diet, and nutritional status could not be retrospectively assessed. Finally, hormones, inflammatory, and blood bone markers were not measured, which would allow insights into sex differences and physiological mechanisms associated with bone formation and resorption following prolonged exercise intervention.

In conclusion, a prolonged multimodal exercise program (2010 to 2019, from 9 to 106 months) was capable of improving the bone health of HIV-infected men with advanced osteopenia and osteoporosis. However, it was not effective to mitigate bone losses in men with normal BMD and women regardless of their initial BMD levels. Factors as BMD, age, and muscle mass at baseline seem to be moderators of BMD changes in response to prolonged multimodal training. Our findings suggest that this modality of exercise should be considered as a strategy to preserve bone health in PLHIV, especially those with advanced osteopenia. Further research is warranted to investigate potential moderators of BMD improvement and to determine optimal dose-response relationships and loading characteristics of exercise routines aiming to preserve the skeletal integrity of men and women living with HIV.

Disclosure statement

The authors report no conflict of interest.

Additional information

Funding

This work was partially supported by the National Council for Technological and Scientific Development (CNPq) under Grant (303629/2019-3, recipient PF), Carlos Chagas Filho Foundation for Research Support in the State of Rio de Janeiro (FAPERJ) under Grants (E-26/110.184/2013, recipient PF; E-26/202.720/2019 and E-26/010.100935/2018, recipient JB), and Coordination for the Improvement of Higher Education Personnel (CAPES, process number 88882.463218/2019-01, recipient KGL).