Bone health and HIV in resource-limited settings: a scoping review

Purpose of review Sub-Saharan Africa and other resource-limited settings (RLS) bear the greatest burden of the HIV epidemic globally. Advantageously, the expanding access to antiretroviral therapy (ART) has resulted in increased survival of HIV individuals in the last 2 decades. Data from resource rich settings provide evidence of increased risk of comorbid conditions such as osteoporosis and fragility fractures among HIV-infected populations. We provide the first review of published and presented data synthesizing the current state of knowledge on bone health and HIV in RLS. Recent findings With few exceptions, we found a high prevalence of low bone mineral density (BMD) and hypovitaminosis D among HIV-infected populations in both RLS and resource rich settings. Although most recognized risk factors for bone loss are similar across settings, in certain RLS there is a high prevalence of both non-HIV-specific risk factors and HIV-specific risk factors, including advanced HIV disease and widespread use of ART, including tenofovir disoproxil fumarate, a non-BMD sparing ART. Of great concern, we neither found published data on the effect of tenofovir disoproxil fumarate initiation on BMD, nor any data on incidence and prevalence of fractures among HIV-infected populations in RLS. Summary To date, the prevalence and squeal of metabolic bone diseases in RLS are poorly described. This review highlights important gaps in our knowledge about HIV-associated bone health comorbidities in RLS. This creates an urgent need for targeted research that can inform HIV care and management guidelines in RLS. Video abstract: http://links.lww.com/COHA/A9.


INTRODUCTION
Resource-limited settings (RLS) that constitute lowand middle-income countries [1] continue to bear dramatically increased survival of HIV-infected individuals in the last 2 decades [3]. With more HIV-infected individuals living longer, it is expected that medical comorbidities such as osteoporosis and fragility fractures will increase. Data from developed countries estimate that up to two-thirds of HIVinfected antiretroviral therapy (ART)-treated and ART-naïve individuals exhibit osteopenia or osteoporosis at the time of low bone mineral density (BMD) diagnosis, with those on ART at increased risk [4]. Importantly, studies in resource rich settings (RRS) are reporting increased evidence of fracture rates in the HIV-infected population, with fracture rates 30-70% higher than those among matched uninfected controls [5,6 & ,7 & , 8,9]. Fragility fractures are associated with significant loss of physical function, independence, and quality of life [10], as well as an increased risk of short-term and long-term mortality [11][12][13]. These data call for more strategic clinical management of HIV individuals that includes prevention or minimization of long-term metabolic complications of HIV infection and its treatment in addition to treating opportunistic infections. In this review, we summarize recently published and presented studies that inform the discussion on bone health among HIV-infected persons in RLS. We highlight the epidemiology of HIV and bone loss in RLS, and among special populations, including HIV-infected young women and perinatally infected adolescents. We focus on three main areas of interest in HIV metabolic bone disease in RLS: effects of HIV and ART, vitamin D insufficiency and other risk factors for bone loss, and fracture risk assessment. We further identify important gaps in research and clinical management as well as make recommendations for future research priorities that would help address these HIV-related, bone health comorbidities in RLS.

Bone mass and HIV in resource-limited settings
The World Health Organization has categorized low BMD into osteopenia and osteoporosis. In postmenopausal women and men, 50 years and above, osteoporosis is defined as a T-score at or below 2.5 SD whereas osteopenia is defined as a T-score between 1 and 2.5 SD below the young adult mean value. Premenopausal women, men below 50 years or children who have a BMD Z-score at or below 2.0 of the age and sex-matched population are classified as having low bone mass. [14]. In the general population, a decline in BMD, assessed by dual-energy X-ray absorptiometry (DXA), is associated with an increased risk of subsequent fractures [15]. Data from RRS consistently show that HIV infection is associated with low BMD and increased fracture risk [5,6 & ,7 & , 8,9]. A meta-analytic review of 11 studies by Brown et al. involving 884 HIV-infected individuals and 654 controls estimated the prevalence of low BMD among HIV-infected individuals to be as high as 67%, 15% of whom had osteoporosis. The magnitude of low BMD was 6.4 times greater and that of osteoporosis 3.7 times greater than in HIV-uninfected controls [4]. Further, in a recent metaanalysis, fracture risk was 1.35-fold higher in HIVpositive compared to HIV-negative controls [7 & ]. Although underlying mechanisms leading to reduced BMD in HIV-infected persons are still unclear, they are believed to be multifactorial and include both traditional and HIV-specific risk factors [4, [16][17][18][19][20][21][22][23][24][25]. Owing to physiological, psychological, and lifestyle factors, HIV-infected persons are likely to have many of the traditional risk factors for low BMD such as physical inactivity, low body weight, nutritional deficiencies (including inadequate calcium and vitamin D intake), depression, smoking, heavy alcohol use, oligo-/amenorrhoea, and hypogonadism [26][27][28][29][30][31][32][33][34][35]. Among the nontraditional causes, a direct effect of HIV and its treatment have been most often quoted; chronic inflammation induced by HIV may impact bone metabolism [36][37][38][39]. In addition, ART significantly contributes

KEY POINTS
Low BMD is generally higher among HIV-infected individuals in RLS compared with RRS, regardless of ART use.
There are very limited published data in any RLS regarding longitudinal changes in BMD in HIV-infected persons, very minimal data on the effect of TDF-based ART on BMD regardless of study design, no longitudinal data on the effect of ART initiation on BMD among ART-naïve cohorts, and no data at all on fractures among HIV-infected individuals in RLS.
There are high but similar prevalence of low vitamin D among HIV-infected individuals regardless of ART use in both RLS and RRS.
There are no data on BMD and fracture risk in key populations of HIV-infected women on DMPA for contraception both in RLS and RRS, and no data in HIVinfected adolescents in RLS.
Unlike RRS where medications such as EFV are no longer preferred, and alternatives to TDF with less bone toxicity are likely to be more frequently used, there are currently no strategies in RLS for minimizing bone loss among HIV-infected individuals which poses an urgent need for research that can inform management guidelines in metabolic bone complications in RLS.
to bone loss among HIV-infected persons [40]. Among individuals on ART, studies in RRS consistently report a 2-6% decline in BMD over the first few years after treatment initiation [25,41], regardless of ART choice [26].
In RLS with a disproportionately high burden of HIV and background nutritional deficiencies [42], known risk factors for low BMD remain similar to those in RRS [25,43,44]. However, some of these risk factors such as low BMI, malnutrition, advanced disease, longer duration since HIV diagnosis and higher HIV viral load are more common in HIV- 48]. These risk factors coupled with more widespread use of non-BMD sparing ART-like tenofovir disoproxil fumarate (TDF) and efavirenz (EFV) make the extremely high prevalence of low BMD in some RLS almost inevitable. Unfortunately, data on BMD among HIV-infected individuals are currently scanty and subject to methodological concerns such as cross-sectional design, lack of appropriate control groups, and local BMD reference data. The majority of the studies did not use local noninfected controls for comparison; the United States National Health and Nutrition Examination Survey reference data being used instead and comparisons were not adjusted for differences in body composition and size. Our review revealed overlapping prevalence of low BMD in RLS and RRS, with a generally higher prevalence of low BMD in RLS overall compared to RRS (Table 1 and [48] and middle-income countries (South Africa [49], Brazil [27], Turkey [50], China [51,52], Israel [53], and Thailand [54]) as well as mixed settings (South Africa, India, Thailand, Malaysia, and Argentina [55]) show varying levels of low BMD with some studies reporting a high prevalence of low BMD in HIV-positive individuals of up to 85% [53]. However, a few authors such as Hamill et al. [49] from South Africa have reported comparable BMD levels between HIV-infected women and appropriate uninfected controls regardless of disease severity. The high BMI of participants in this study may have had a sparing effect on bone loss. In contrast, a study comparing ART-naïve to ART-experienced patients on long-term suppressive ART in western India found extremely high prevalence of low BMD, 80.4% among ART-experienced, and 67% among ART-naïve patients, but no local uninfected controls were used [46]. Another crosssectional study among young HIV-infected Israeli women of Ethiopian and Caucasian origin found a higher prevalence of low BMD, 85% among Ethiopians compared to 40% seen in the Caucasians [53] which the authors attributed to poorer vitamin D status among Ethiopian women [53]. Similar proportions of low BMD have been reported by recently published data from RRS [56 & , [57][58][59][60] with the exception of a few studies [61][62][63].
There are very limited data in any RLS regarding BMD longitudinal changes among HIV-infected persons. In a 48-week, multisite, second-line trial in South Africa, India, Thailand, Malaysia, and Argentina [55], HIV-infected patients who initiated a second-line regimen experienced additional bone loss. We did not find any longitudinal data on the effect of ART initiation on BMD among ART-naïve cohorts, or any data on fractures among HIVinfected individuals in RLS.

Role of tenofovir
A strong body of evidence from longitudinal data in RRS shows that among the different antiretroviral drugs, the potential effect of TDF on bone health is particularly concerning [64][65][66][67][68][69][70]. In ART-naïve HIVpositive individuals, initiating TDF-containing ART was associated with greater bone loss over the first few years compared to TDF-sparing regimens [67,[69][70][71]. With ART-initiation, there is a rapid acceleration of bone turnover; bone resorption outstrips bone formation, likely accounting for the decrease in BMD [72,73]. Consistent with these findings, Brown and others [66,67,74] have shown that ART initiation is associated with a 2-6% loss of BMD over the first 48-96 weeks of therapy that does not return to baseline after prolonged HIV RNA suppression and also reoccurs after reinitiation of ART after treatment failure. In another adult study comparing TDF-containing and noncontaining regimens, Gallant et al. [64] observed increased bone resorption and loss in the TDF-containing arm compared to patients receiving an alternate NRTI (stavudine), at both the LS (À3.3 vs. À2.0%) and hip (À3.2 vs. À1.8%). Importantly, the majority of BMD loss was observed within the first 24-48 weeks of treatment, and thereafter, BMD loss slowed, but BMD did not recover over the 144 weeks of the study. Similarly, a study comparing TDF to abacavir an NRTI revealed a greater loss of BMD at total hip (À3.6 vs. À1.9%) and LS (À2.4 vs. À1.6%) in the TDF group. Again, BMD loss occurred closer to initiation of therapy and was maximal in the spine at 24 weeks and in the hip at 48 weeks [66]. More interestingly, switching from a TDF-containing regimen to an alternative NRTI leads to an increase in BMD [71]. Though the mechanism through which TDF reduces bone mass is not clear, there is more evidence suggesting that TDF induces renal dysfunction [75][76][77][78][79][80][81][82][83][84][85][86][87]. TDF has been shown to induce proximal renal tubular In this study, osteopenia and osteoporosis were defined as T-or Z-score <

À1
and T-or Z-score <
c In this study, low BMD was defined as T-score <1 in those older than 40 years or Z <2.0 in those younger than 40 years, respectively.
In RLS, the two WHO recommended first line ART treatment regimens for adults and children above 15 years contain TDF; TDF, lamivudine (3TC) and EFV or TDF, emtricitabine (FTC), and EFV, which exposes many HIV-infected individuals to the negative impact of TDF on bone health [3,95]. Conversely, there are scarce data on the effect of TDF-based ART on BMD in these settings. Martin et al. [55] reported that HIV-infected patients who initiated a second-line regimen had a greater bone loss if they were on TDF for longer duration during the 48 weeks of the study. For every 1 year of TDF use, the femur BMD reduced by 1.58% and spine BMD by 1.65% (P < 0.001).

Vitamin D and bone health in HIV
Worldwide, it's estimated that more than one billion people are characterized as having vitamin D deficiency (<20 ng/ml), or insufficiency (<30 ng/ ml) regardless of the economic setting. According to a recent review by Mansueto (Table 2 and Fig. 2). The authors ascribed the high prevalence of vitamin D deficiency seen among Turkish [50], and Israeli [53] to skin coverage with resultant reduced sunlight exposure. Among individuals on ART, several cross-sectional studies from both RRS and RLS have shown an association between EFV use and low 25-hydroxyvitamin D (25(OH)D) [30,97,104,[110][111][112][113]. NNRTIs, especially EFV which are widely used to treat HIV infection in RLS are hypothesized to enhance 25(OH)D catabolism through the induction of cytochrome P450 enzymes (CYP24A) [112] which reduce 25(OH)D concentrations. Among HIV-infected individuals, vitamin D insufficiency has been associated with a higher risk of HIV disease progression, death and virologic failure after ART [96,114]. In addition, vitamin D deficiency has been reported to independently increase the risk of low BMD [115]. In view of this, supplementation with vitamin D has been reported to mitigate bone loss [61,105]. In a recent randomized trial Overton et al. [61] found that BMD loss in the first year after ART initiation may be minimized by calcium and vitamin D supplementation D. By way of contrast, none of the studies we reviewed supported an association between vitamin D insufficiency and low BMD [61]. Though a cross-sectional study by Shahar et al.  Overlapping prevalence of osteopenia and osteoporosis in low-/low-middle-income countries and high-income countries was found, with a generally higher prevalence of low bone mineral density in low-to middle-income countries overall compared to high-income countries.   , and the provision of reproductive health services is a crucial part of their HIV care. However, certain types of hormonal contraception have been associated with long-term metabolic dysregulation, particularly low BMD. In RLS with the highest unmet need for contraception, depot medroxyprogesterone acetate (DMPA) is the preferred contraceptive option across the different age groups [116] with approximately 15 million current users in the sub-Saharan African region alone [117]. Among HIVinfected women in particular, DMPA remains effective [118] because of its lack of interactions with antiretroviral drugs [119][120][121]. However, owing to its hypoestrogenemic effects [122], DMPA has also been associated with reduced BMD [123][124][125][126][127][128][129][130]. The few published observational studies on the association between DMPA and fracture risk in RRS suggest increased risk of fractures among DMPA users [131,132]. For example, a large populationbased control study by Meier et al. [131] showed a 50% increased risk of incident fractures among 20 to 44-year-old European DMPA users receiving 10 or more injections compared to nonhormonal users, among those who had received 10 or more injections. It must also be noted that all the above studies were conducted among HIV-negative individuals. Our review did not yield any published data on the effect of DMPA on BMD or fracture risk among HIV-infected women either in RRS or RLS. This presents a critical gap in policy and clinical management guidelines for HIV infected women.  [134][135][136][137]. Several studies from RRS support an independent, dose-response relationship between BMD and T h a i l a n d 1 * * T h a i l a n d 2 T h a i l a n d 3 T h a i l a n d 4 * * B r a z i l 1 B r a z i l 2 risk of osteoporotic fractures [135,[138][139][140][141][142][143][144][145][146][147][148]. For example, a 10% increase in PBM in young women is associated with an estimated 50% reduction in fracture risk after menopause [135]. Although there have been a few controversies among HIV individuals on ART [133,149], the general conclusion from a number of studies in RRS is that TDF treatment decreases BMD with stronger associations being seen in children and adolescents than in adults [64,[150][151][152]. Thus, BMD may be more affected during the active period of bone growth and development. Among HIV-infected adolescents living in RLS, additional highly prevalent factors, including protein and energy malnutrition, micronutrient deficiencies, and childhood infections that are known to adversely affect bone mass accrual may pose additional threats to bone acquisition. To date, there are currently no published data in RLS where over 90% of infected adolescents live. This has inadvertently lead to lack of prevention and clinical management guidelines for this unique age group who may be at considerable risk of bone complications during a critical period of PBM attainment and subsequent lifelong ART exposure.

Constraints to diagnosis and management of bone loss in resource-limited settings
In 2015, 11 out of the 16 million people receiving ART globally were in the WHO Africa region alone [153]. However, unlike RRS where medications such as EFV are no longer preferred, and alternatives to TDF with less bone toxicity are likely to be more frequently used, there are currently no strategies in RLS for minimizing bone loss among HIV-infected individuals. The already limited funding, poor healthcare infrastructures, and sparse personnel pose tremendous challenges toward prevention and management of metabolic bone complications in RLS. As the standard assessment tool for BMD, DXA has only limited value as a single assessment. Serial assessments during HIV patient monitoring while on ART provide more information on the pattern of BMD changes [154]. In RLS, use of DXA scans in assessing BMD is limited by availability, cost, and training. In addition, once the diagnosis is obtained, the current cost of treatment medications for osteoporosis, for example, bisphosphonates is prohibitive. Furthermore, most healthcare personnel in most RLS lack the expertise to make appropriate diagnoses and provide relevant care.

Research needs
With more people starting ART [153] and living longer with HIV than ever before, more individuals will continue to experience osteoporosis and its sequelae, including fragility fractures [155]. Given low clinical and research capacity for metabolic bone disease in RLS, there is urgent special need for building capacity in bone healthcare and research. Expanding knowledge about bone health in RLS will not only provide significant insights into the burden of HIV-related bone loss in RLS but also predictors, and evolution of bone metabolic comorbidities in the time course of HIV infection and its lifelong treatment. An initial focus is needed to establish the epidemiology of metabolic bone diseases in both the general and HIV populations. We recommend prioritization of the following research agenda in RLS: (1) Cost-effective and feasible strategies to prevent osteoporosis for both HIV-infected and noninfected populations.
(2) Identification of simple-low cost tools to detect early osteopenia. (3) Strategies to minimize or avoid ARV-associated bone loss such as ART choice, dose optimization, and ARV switching. (4) Research among HIV-infected populations focusing on women of reproductive age and special populations such as perinatally infected children and adolescents.
To successfully conduct research addressing the above mentioned gaps in bone health comorbidities in RLS, there is need to work through several existing research networks either regionally or globally. This will ensure effective design and quality implementation approaches are employed. Importantly, involving key policy makers both domestically and regionally upfront will make the future policy implementation more successful.

CONCLUSION
The review reveals overlapping prevalence of low BMD in RLS and RRS, with a generally higher prevalence of low BMD in RLS overall compared to RRS. We highlight important gaps in our knowledge about HIV-associated bone health comorbidities in RLS. In particular, there are scarce data on bone health mainly from cross-sectional studies that call for urgent need for research that can inform management guidelines in metabolic bone complications in RLS.
Research and Training in Africa, Makerere University School of Public Health and University of the Witwatersrand.
Financial support and sponsorship F.K.M. has received an R01 grant from the National Institute of Allergy and Infectious Diseases of the National Institutes of Health (NIH) under Award Number R01AI118332NIH for bone health-related work as the Principal Investigator, and support as a site investigator on NIH funded microbicide trials network protocols. K.R. has received support from Senior Research Scholar, Thailand Research Fund (TRF) for his work.

Conflicts of interest
There are no conflicts of interest.