The impact of vitamin D supplementation on musculoskeletal health outcomes in children, adolescents, and young adults living with HIV: A systematic review

Objective HIV-positive children, adolescents, and young adults are at increased risk poor musculoskeletal outcomes. Increased incidence of vitamin D deficiency in youth living with HIV may further adversely affect musculoskeletal health. We investigated the impact of vitamin D supplementation on a range of musculoskeletal outcomes among individuals aged 0–25 years living with HIV. Methods A systematic review was conducted using databases: PubMed/Medline, CINAHL, Web of Knowledge, and EMBASE. Interventional randomised control trials, quasi-experimental trials, and previous systematic reviews/meta-analyses were included. Outcomes included: BMD, BMC, fracture incidence, muscle strength, linear growth (height-for-age Z-score [HAZ]), and biochemical/endocrine biomarkers including bone turnover markers. Results Of 497 records, 20 studies met inclusion criteria. Thirteen studies were conducted in North America, one in Asia, two in Europe, and four in Sub-Saharan Africa. High-dose vitamin D supplementation regimens (1,000–7,000 IU/day) were successful in achieving serum 25-hydroxyvitamin-D (25OHD) concentrations above study-defined thresholds. No improvements were observed in BMD, BMC, or in muscle power, force and strength; however, improvements in neuromuscular motor skills were demonstrated. HAZ was unaffected by low-dose (200–400 IU/day) supplementation. A single study found positive effects on HAZ with high-dose supplementation (7,000 vs 4,000IU/day). Conclusions Measured bone outcomes were unaffected by high-dose vitamin D supplementation, even when target 25OHD measurements were achieved. This may be due to: insufficient sample size, follow-up, intermittent dosing, non-standardised definitions of vitamin D deficiency, or heterogeneity of enrolment criteria pertaining to baseline vitamin D concentration. High-dose vitamin D may improve HAZ and neuromuscular motor skills. Adequately powered trials are needed in settings where HIV burden is greatest. PROSPERO Number: CRD42016042938.


Introduction
The global scale-up of antiretroviral therapy (ART) has dramatically improved survival of those living with HIV and converted what was once a life-threatening infection into a chronic, treatable condition. HIV management now includes treatment of HIV infection, as well as associated chronic comorbidities, for example increased risk of low bone mineral density (BMD) [1][2][3][4]. Low BMD in youth living with HIV has been shown to far exceed those of HIVnegative controls [3,5]. Similarly, multiple observational studies, in both high-and low-middle-income countries (LMIC), have demonstrated vitamin D insufficiency in HIV-positive children, adolescents, and young adults [6-9], with a single study demonstrating increased rates compared to HIV-negative age-matched controls [10].
During childhood and adolescent growth, bones grow in length, width and mineral content until peak bone mass (PBM) is achieved [11]; PBM is a key determinant of future adult osteoporosis and lifetime fracture risk [12][13][14]. 'Low bone mass', defined as a dual energy X-ray absorptiometry (DXA) measured BMD Z-score � -2 has been associated with low 25OHD and altered vitamin D metabolism in HIV-positive youths [15,16]. Furthermore, HIV infection increases bone turnover to reduce BMD even when vitamin D concentration is adequate [16]. HIV-associated alterations in vitamin D and bone metabolism are thought to arise from inflammatory and metabolic properties of the HIV infection itself [17][18][19][20] and/or side effects of ART [21][22][23][24][25][26][27] altering the molecular balance between bone formation and resorption. Intestinal absorption, nutritional intake and/or sun exposure may also be reduced [28][29][30][31]. HIV can cause delayed puberty [32,33] with associated reductions in bone mass [34][35][36] and restricted linear growth [37]. Inadequate dietary vitamin D is associated with growth failure (i. e. height-for-age Z-scores [HAZ] <-2) in HIV-positive children [38]. In sum, direct and indirect effects of HIV on musculoskeletal health are multifactorial and are potentially exacerbated by inadequate vitamin D.
Most current guidelines define vitamin D deficiency and insufficiency as a serum vitamin D value (25OHD) <30 nmol/L (<12 ng/ml) and between 30-50 nmol/L respectively (12-20 ng/ml); however, consensus is lacking and defined thresholds vary between countries and specialty advisory committees [39][40][41]. Some experts advocate higher values (e.g >75 nmol/L [>30 ng/ml]) in order to acieve maximal suppression of parathyroid hormone (PTH) and to optimise bone matrix formation in light of altered vitamin D metabolism with HIV [42]. Vitamin D supplementation has been shown to improve BMD in a range of paediatric chronic diseases such as epilepsy [43], kidney disease [44] and juvenile idiopathic arthritis [45], and to improve muscle function in HIV-negative adolescent girls [46]. Hence, it has been hypothesised that the beneficial effects of vitamin D supplementation can be reproduced in HIV infection.
Globally, there is an absence of evidence-based guidelines for vitamin D supplementation in youth living with HIV. We aimed to systematically review the current evidence examining the relationships between vitamin D supplementation and a range of musculoskeletal outcomes in children, adolescents, and young adults living with HIV, to guide future strategies to optimise musculoskeletal health.

Search strategy
Search strategy followed PRISMA guidance [47] (Fig 1). Articles were restricted to those published in English and French from the year 2000-2017, reflecting the period of ART availability. Articles were not restricted by geographic location. The following databases were searched: PubMed/Medline, EMBASE, CINAHL, and Web of Knowledge, (S1-S4 Tables). A hand search of the references cited for each retrieved article was performed. Furthermore, available conference abstracts, within the past six years, were reviewed from: (i) the Conference on Retroviruses and Opportunistic Infections, (ii) International AIDS Society, (iii) Infectious Diseases Society of America ID Week, and (iv) the American Society for Bone and Mineral Research.

Study selection
Interventional trials of oral or parenteral vitamin D with or without calcium were assessed, including randomised control trials (RCT) and quasi-experimental studies (QET) (both controlled and uncontrolled). Controlled QET consisted of pre-and post-intervention studies with vitamin D and non-vitamin D arms, in addition to studies comparing high-dose and standard-dosing regimens (standard-dose defined: �800 International Units [IU]/day).
Uncontrolled QETs were considered if they utilised either a one group pre-post-intervention design (vitamin D only) or an interrupted-time-series design [48]. Previous systematic reviews were included. To ensure enough time to achieve outcomes, a minimum of six months followup after enrolment was required for studies reporting radiological or clinical outcomes and three months for those reporting biochemical/endocrine outcomes. Studies that had at least 10 individuals aged 1 month to 25 years, regardless of the mode of HIV acquisition and HIV treatment status (ART naïve or experienced) were considered. Studies assessing bisphosphonates in conjunction with vitamin D supplementation were excluded.
Eligibility for inclusion was determined independently by two reviewers (JP & CR) using an assessment toolkit with a pre-defined inclusion checklist [49,50].

Data extraction and quality assessment
Data were extracted independently using the Cochrane Public Health Group Data Extraction and Assessment Form [50], with risk of bias (ROB) assessed using the Cochrane Handbook for systematic reviews and reported independently as high, low, or unclear by both reviewers (JP & CR) [49]. Data on study latitude/geographic location, participant age, gender, seasonality, adherence, dietary vitamin D intake (IU/day), skin colour/sun exposure and method of 25OHD and 1,25OHD quantification were recorded.
Doses of vitamin D supplementation were standardised into daily doses (IU/day). Definitions of vitamin D deficiency, insufficiency, and sufficiency were extracted for between-study comparison (S5 Table).

Musculoskeletal outcomes (bone, muscle and linear growth)
BMC and BMD measured using the following radiographic techniques were acceptable: DXA, computed tomography (including pQCT and high resolution pQCT), quantitative ultrasound (measured by either speed of sound, broadband ultrasound attenuation, or stiffness index), and/or digital X-ray radiogrammetry. For those not reporting BMD Z-scores, BMD measurements were compared against standardised, age-matched paediatric Z-scores as outlined by the International Society of Clinical Densitometry (ISCD) [51].
Studies assessing muscle function by measuring muscle strength, power, and/or force were considered relevant, as were studies assessing muscle size and mass. Linear growth was reported as HAZ, where SDs and standard error can be computed from Z-score values. HAZ summary statistics by age/sex represent a comparison to WHO reference standards with an expected mean HAZ of zero and SD of one. Z-scores were compared to the WHO Global Database on Child Growth and Malnutrition standard classifications, if not reported by the study authors [52]. Clinical outcomes of incident fractures and prevalent rickets and osteomalacia were extracted.

Adverse events
All adverse effects of vitamin D supplementation were recorded. Hypercalcemia, hypercalciuria, renal calculi, and gastrointestinal upset were specifically screened for as these are recognised complications of cholecalciferol supplementation in healthy populations [53].

Study characteristics
A total of 607 articles were retrieved from database searches, a further 27 from hand searches of references, and 7 conference abstracts. After removal of duplicates, 497 articles were screened by title and abstract. The resulting 190 articles were then reviewed in full, of which 37 studies met inclusion criteria (Fig 1) [65][66][67] and a third study assessing BMD in conjunction with longitudinal biochemical data [68].

Study participants
Overall, the 20 trials included individuals aged six months to 25 years old. With the exception of one Hepatitis C co-infected individual in the Foissac et al. [69]

Study outcomes
The outcomes assessed can be broadly classified into four categories, with several studies reporting primary outcomes in more than one of the following categories: 14 studies assessed biochemical/endocrine parameters, most notably serum 25OHD concentration (Table 1), five No baseline 25OHD exclusion criteria 1 . Mean age based on group allocation and/or overall age when reported. 2 . All oral cholecalciferol. 3 . Calculated based on 30 days/month. 4 . Means (standard deviation) unless otherwise specified. 5 . ng/ml transformed to nmol/L 6 . Gender reported as percentage male 7 . Bimonthly defined: once every 2 months. 8 . Median (interquartile range). 9 . 60,000 and 120,000 IU/month groups considered together except in bone turnover marker analysis  studies reported bone outcomes (Table 2), two studies described muscle function and structure findings (Table 3), and four studies analysed HAZ ( [54] reported measures of potential harm and/or side-effects of the intervention although none were powered for safety. No significant adverse events directly related to the intervention were observed apart from two cases of renal calculi, although one was remote from the intervention and the other part of the placebo group receiving only 400 IU/day of vitamin D. Minor adverse events, when reported, consisted of transient hypercalcemia and hypercalciuria. Biochemical/endocrine outcomes. The majority of studies aimed to raise concentrations of 25OHD >75 nmol/L with the goal of maximal PTH suppression (S5 Table). Kakalia et al. demonstrated persistent hyperparathyroidism with serum 25OHD values between 50-75 nmol/L [73]. To achieve a measurement >75 nmol/L, higher supplementation doses (>1,000    [71]. Although less frequently a reported outcome, changes in serum calcium, phosphate and BAP concentrations were found to be largely unaffected after   67,73]. Interestingly, in one study, changes in PTH and BAP were only observed in the cholecalciferol supplemented group treated with TDF [67]. Changes in markers of bone remodelling (CTX, P1NP, OC) were not seen in two studies where the treatment arm used doses of 2,000 IU/day or less [67,77] but when the treatment dose was increased to 4,000 IU/day a decrease in CTX and P1NP was observed [77]. These findings are similar to those found in the preliminary data by Sudjaritruk et al., however, changes in bone turnover markers were demonstrated in both high-and standard-dose groups without intergroup differences [54]. Increased FGF23 was reported in response to cholecalciferol supplementation, although only in two studies, and increases were only observed in the group receiving TDF-containing ART [57, 78].
Musculoskeletal outcomes. In the five studies examining bone outcomes, no significant differences in BMD nor BMC were noted between various high-and standard-doses of cholecalciferol versus placebo, despite significant increases in 25OHD. Notably, as demonstrated by Arpadi et al., 75% who received supplementation did not consistently maintain serum 25OHD >75 nmol/L; 30% had at least one monthly serum 25OHD concentration <50 nmol/L [56]. Preliminary results of a 48-week randomised open-labelled trial assessing BMD after 'highdose' cholecalciferol (3,200 IU/day) versus 'normal dose' (400 IU/day), both groups also supplemented with oral calcium, has recently been presented at the 9th International Workshop on HIV paediatrics [54]. Analyses were stratified by the presence of low lumbar spine BMD Zscore at baseline. BMD gains were seen in both 'high-dose' and 'normal-dose' groups who had low-BMD at baseline. A weak trend was suggested towards greater BMD gains amongst the 'high-dose' group; however, full trial data are awaited. Body size adjustment was not consistently considered, despite recommendations for size adjustment by the ISCD [79], which limits the validity of bone outcomes.
In the one study focused on muscle outcomes, cholecalciferol supplementation did not improve muscle power, force, or strength amongst youth undergoing jumping mechanography, ankle and knee isometric/isokinetic testing and grip strength dynamometry, despite substantial cholecalciferol dosing (7,000 IU/day) which achieved significant increases in 25OHD [62]. However, post-hoc multivariate analysis suggested participants with the greatest increases in 25OHD tended to have greater jump power and energy, although it is unclear if this effect was dependent on baseline 25OHD concentration. Interestingly, in one study, cholecalciferol supplementation did have a beneficial effect on neuromuscular motor skills measured using the Bruininks-Oseretsky Test of Motor Proficiency [80]. This included tests of fine motor precision and integration, dexterity, coordination, balance, and agility. Muscle cross-sectional area was unaffected by cholecalciferol supplementation at the same daily dose in a separate study [57]. Three of four studies examining the effect of cholecalciferol on HAZ provided cholecalciferol as part of a multi-micronutrient supplement and in much lower doses (200-400 IU/day) than the aforementioned studies examining bone and muscle outcomes. No differences in HAZ were observed after 6-18 months on these low-dose regimens [59][60][61]. Similarly, no differences in HAZ were found in a separate study supplementing 4,000 IU/day to 30 individuals. However, at a much higher dose (7,000 IU/day) this same study found significant differences in HAZ after 12 weeks when compared with 4,000 IU/day [71]. There was again no placebo group for comparison. Furthermore, 32% and 28% of participants, all HIV-positive, in studies by Mda [59] and Steenhoff et al. [71] were reported as 'already stunted' (HAZ < -2) at baseline. Those supplemented with cholecalciferol in studies by Ndeezi [60] and Chhagan et al.

Risk of bias
The majority of studies were deemed of fair to good quality (Fig 2).

Discussion
In response to cholecalciferol supplementation, our review found no clear improvements in BMD, BMC, or in muscle power, force or strength. High-dose cholecalciferol appeared to show some effect on HAZ in a single study. The lack of effect on a broad range of musculoskeletal health outcomes may be due to differences in the definition of vitamin D deficiency used as compared to those in place for the general population (S5 Table) and/or the application of these definitions to the variable baseline 25OHD concentrations (Tables 1-4), which allowed enrollment of individuals with 25OHD concentrations which could in some cases already be deemed sufficient, prior to cholecalciferol supplementation.

The effect of cholecalciferol supplementation on biochemical/endocrine outcomes
The majority of studies aimed to raise 25OHD concentrations >75 nmol/L, intending to supress PTH, although more recent guidelines recommend lower 25OHD concentrations to prevent poor musculoskeletal health in the general population [40,41]. The need for cholecalciferol dosing above that recommended for HIV-negative individuals may be due to reduced intestinal absorption and/or altered vitamin D metabolism, particularly in children and adolescents [31,81]. The omission of those with very low baseline 25OHD biases results against maximal effect; these children and adolescents would largely meet the criteria for rickets/osteomalacia and as such would be expected to benefit most from long-term supplementation. Inclusion of individuals with vitamin D concentrations deemed adequate may have similarly biased results. Alternatively, potentially even higher cholecalciferol doses and thresholds >75 nmol/L for 25OHD may be required. We identified submaximal PTH suppression in 3/7 studies, particularly when lower 25OHD thresholds were used or when 25OHD targets were inconsistently maintained, highlighting the importance of adherence and quality control monitoring in studies. Overall, when assessing for normalisation of biochemical parameters, the studies evaluated seem to favour higher 25OHD targets, e.g. �75 nmol/L.
The Endocrine Society Task Force clinical practice guideline remains one of few to contain recommendations addressing vitamin D deficiency in children and young adults living with HIV, although the evidence-base for these recommendations is limited [42]. Our findings do not dispute their recommendation of a standardised maintenance daily dose of cholecalciferol 2-3 times that recommended for age, even in the absence of universal 25OHD screening. This dose may be simplified to 1,500-2,000 IU/day for all paediatric patients, a recommendation which aligns with optimal dose modelling data by Foissac et al. [82]. Higher doses are likely to be required for treatment of documented vitamin D deficiency, and if used, the full safety profile should be substantiated under medical supervision. The Endocrine Society guideline suggests treatment doses of 2,000 IU/day for infants and 4,000 IU/day for children 1-18 years old, which should be accompanied by routine follow-up 25OHD measurements to ensure repletion of biochemical stores. Expert consultation is advocated with higher dosing.
Unfortunately, PTH changes in response to cholecalciferol supplementation were variable. In the five studies addressing bone outcomes, PTH was not measured, not reported, or when reported failed to consistently normalise. Persistently elevated PTH may explain the lack of improvement in BMD/BMC despite increases in 25OHD. Conflicting results in the context of TDF treatment are reported where TDF, by lowering serum 25OHD concentrations, is associated with secondary hyperparathyroidism [83][84][85]. Havens et al. (2012b) observed a reduction in PTH in those receiving TDF and cholecalciferol. Whilst, in those not receiving TDF, PTH values were unchanged despite increased 25OHD, indicating a possible vitamin D independent effect (67]. Dougherty et al. showed superior 25OHD responses whilst receiving efavirenz (EFV), but without effects on PTH [72]. This finding contrasts with Eckard et al. who demonstrated no differences in 25OHD in those receiving EFV [77]. Likewise, Havens et al. (2012a) showed no PTH response from treatment with either TDF or EFV [66]. Potentially, different ART regimens may influence musculoskeletal outcomes, independent of ARTs' known impacts on vitamin D metabolism [22,23,86].
Low 25OHD concentrations, with secondary hyperparathyroidism, increase bone turnover (i.e. CTX, P1NP, OC); this is seen in young people with HIV just as it is the wider population [7,87]. Surprisingly, Havens et al. (2012b) failed to corroborate this finding, even with normalisation of PTH in the TDF group [67]. Early treatment with cholecalciferol to prevent subclinical hyperparathyroidism may moderate effects on bone turnover. However, this may require high cholecalciferol dosing. Eckard et al. only demonstrated changes in CTX and P1NP with 4,000 IU/day (although no PTH response was seen) [77].
FGF23, an important regulator of phosphate homeostasis, is secreted by osteocytes in response to 1,25OHD. FGF23 mediated phosphate regulation in HIV-associated vitamin D deficiency remains poorly understood. TDF is associated with phosphaturia which may perpetuate a hypothesised 'functional vitamin D deficiency', explained by higher concentrations of vitamin D binding protein (VDBP) reducing free 1,25OHD [21]. Eckard et al. postulated a compensatory decrease in VDBP with EFV after cholecalciferol supplementation [88]. Their findings suggest that TDF and EFV may modify FGF23 response to vitamin D supplementation in adolescents living with HIV by altering vitamin D metabolism, at least in the short term [65]. Further study is needed.
FGF23 increases in renal impairment [89][90][91], yet only three studies considered renal function in their analysis [67,77,78]. Others excluded patients with renal impairment altogether, or inconsistently measured this at baseline. We further identified heterogeneity in 25OHD assays with five different methods used.

The effect of cholecalciferol supplementation on musculoskeletal outcomes
The absence of any significant effect of cholecalciferol on BMD/BMC in the five retrieved studies may be a real finding or explained by a number of factors. (i) Intermittent dosing may prove insufficient to sustain steady state 25OHD above essential thresholds. Daily, as opposed to monthly/bi-monthly dosing, may be superior but must be balanced against adherence. However, evidence pertaining to optimal dosing schedules remains contentious, particularly as it relates to effects on bone mass, fractures, and falls [92,93] (ii) Arpadi et al.'s exclusion of individuals with the lowest 25OHD measurements may have biased results towards the null, omitting participants who may benefit most from supplementation [56]. (iii) inclusion of individuals with adequate 25OHD concentrations across studies may have equally biased results towards the null, similar to biochemical/endocrine outcomes. (iv) Reported intra-group changes in BMD were not consistently analysed relative to age, changes in body composition (Tanner Staging), and skeletal maturation (particularly height adjustments) [94,95]. Rovner et al. [63] was the only study to explicitly state that such height adjustments were performed whereas Arpadi et al. was the only study to report supplementary analyses of Tanner stage advancement [56]. (v) DXA measurements are size-dependent, hence size correction is crucial, otherwise low BMD/BMC may be explained by reduced height compared to a control population, rather than an actual deficit on bone mass [79]. BMD must be interpreted relative to stature [96,97].
No paediatric studies from LMIC were available. Promisingly, preliminary results by Sudjaritruk et al. represents the first supplementation trial assessing BMD in youth living with HIV outside of the USA; it will be important to see final BMD outcomes correctly size-adjusted and so taking into account growth differences between groups [54]. Furthermore, published data of concurrent changes in biochemical/endocrine markers will be valuable.
Our review highlights a lack of data on muscle size or functional outcomes following cholecalciferol supplementation in HIV-positive youth. Although one study identified increases in neuromuscular motor skills, no effects on muscle power, force or strength were found; perhaps because only 33% achieved 25OH concentrations �80 nmol/l. Exploratory post-hoc analysis suggested a responder effect such that participants with increased 25OHD after supplementation did show a positive response in jump power and energy [62]. These neuromuscular improvements are important as poor motor function, evaluated by assessing muscular tone, strength, and muscle volume, have been associated with HIV disease progression [98].
The only study to find an effect of cholecalciferol supplementation on HAZ used high-dose supplementation (7,000 IU/day) and examined linear growth as a secondary outcome [71]. HAZ was the primary outcome in just one, much larger study, but of a much younger, heterogenous population with high attrition rates secondary to death [61]. With the exception of the Steenhoff et al. trial [71], concurrent 25OHD measurements were not evaluated making it difficult to confirm the extent to which 25OHD concentrations improved after supplementation. In addition, high rates of baseline stunting may represent a missed opportunity, as 'catchup' growth may be unattainable even with adequate micronutrient supplementation. It remains unclear at what age intervention may be beneficial. Follow-up time in all studies was �18 months, likely insufficient to detect an effect on HAZ, particularly outside of the peripubertal growth period. Lastly, studying cholecalciferol as part of a multi-micronutrient supplement may mean effects are confounded by other micronutrients, supporting the need for placebo controlled studies of high-dose vitamin D supplementation alone.

Summary of recommendations for future trials of vitamin D supplementation in young people
Moving forward, studies in LMIC are of particular importance. Trials are needed to establish the effect of vitamin D supplementation on musculoskeletal outcomes before PBM is achieved, targeting the key pubertal stages of maximal growth velocity, when impact may be greatest. Supplementation is needed to avoid secondary hyperparathyroidism which is the primary stimulus for bone turnover [7,99]. Hence, concurrent PTH and 25OHD measurements (at appropriate intervals relative to supplementation doses) are needed in studies measuring BMD/BMC and linear growth. Future studies may also wish to investigate practical adjuncts, such as muscle strength training and weight bearing activity. Fracture incidence should be reported in longitudinal studies, at least as a secondary outcome.
Optimal dosing regimens need to be established. Safety profiles need continued evaluation especially at higher doses and for rare drug-related adverse events, missed by smaller studies. Vitamin D should be supplemented at doses and in regimens that aim to provide sustained 25OHD above pre-defined thresholds. Consensus on threshold 25OHD concentrations defining vitamin D deficiency, insufficiency, and sufficiency would be welcome to standardise studies and permit future meta-analyses. The majority of studies employ a 25OHD target of �75 nmol/ L as sufficient, with values 50-75 nmol/L considered insufficient and <50 nmol/L deficient. We suggest future studies try not to exclude those with the lowest vitamin D concentrations. Whilst, clinical trials where equipoise is lacking are unethical, for example in cases of symptomatic vitamin D deficiency i.e. rickets, trials in asymptomatic youth who have incidental findings of low 25OHD concentrations which may simply reflect seasonality, can be justified. Where equipoise is lacking, alternative study designs are more appropriate, such as longitudinal study designs evaluating musculoskeletal outcomes before and after cholecalciferol replacement. Studies should also address the long-term effects of supplementation in relation to baseline concentrations, stratifying analysing according to adequacy of baseline 25OHD concentrations.
The combined effect of calcium and vitamin D supplementation on musculoskeletal outcomes in youth living with HIV remains to be established. Future studies need to consistently report and consider the effects of renal function, latitude, season, ethnicity, and local policies on dietary fortification. Lastly, we recommend standardisation of both DXA measurements to take account of size-adjustment as per the revised 2013 ISCD Pediatric Official Position Guideline and the type of 25OHD assay used [79,100]. Use of alternative modalities in the measurement of bone quality such as pQCT, high resolution pQCT and DXA measured trabecular bone score may prove beneficial and should be investigated in this population [101].

Limitations
Our analysis was limited by the four databases searched and to studies published in English and French. Unfortunately, we were unable to perform a meta-analysis on the available data given the heterogeneity in study designs and populations investigated. This heterogeneity extended to a wide age and geographical range of study particpants, variablity in modes of HIV infection and treatment, and a variety of cholecalciferol supplementation regimes which limits identification of clear patterns in outcomes.

Conclusion
Our systematic review identified few, small studies, with heterogeneous study designs from which we were unable to draw firm conclusions to guide future evidence-based vitamin D supplementation strategies to optimise musculoskeletal health in youth living with HIV. However, we were able to make a series of recommendations which we feel should be considered by all researchers performing much needed further work in this field. Given the successful role out of universal ART and the transition to HIV-associated chronic disease management, there is an urgent need to identify any interventions that may attenuate the musculoskeletal consequences of a lifetime of HIV infection and treatment.
Supporting information S1