Introduction

The introduction of effective antiretroviral therapy (ART) has much improved the outcomes of human immunodeficiency virus (HIV) infection. HIV-infected individuals treated with ART can achieve a near-normal life expectancy and experience improved quality of life because of immune system restoration and viral suppression, which in turn reduces mortality and comorbidity [1, 2]. Therefore, potential related complications arising from long-term treatment have been a topic of interest.

Regarding bone health, osteoporosis is thought of as a silent disease, since bone loss usually arises without any symptoms or signs. It is also characterized by deterioration of bone microarchitecture, increased bone turnover, and low bone mineral density (BMD) [3]. Some studies found that HIV-infected patients had a decreased BMD [4,5,6].

A meta-analysis including 12 cross-sectional studies showed that the prevalence of osteoporosis in HIV-infected patients is 3.7 times higher than that in HIV-uninfected controls [7].

Although multiple analyses have demonstrated decreased BMD in HIV-infected persons and raised concerns for increased fracture risk, the effect of low BMD on fracture risk among patients diagnosed with HIV infection is not well-recognized because of the limited data.

We performed a systematic review and meta-analysis to investigate whether the risk of fracture increased in individuals with HIV infection. The primary goal of the study was to compare the risks of all fractures and fragility fractures between HIV-infected patients and controls. We also estimated the prevalence of fractures in the HIV population.

Materials and methods

We conducted this meta-analysis according to Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [8].

Search method

We looked up PubMed, Medline, Embase, and the Cochrane Library using the keywords “human immunodeficiency virus,” “bone mineral density,” and “fracture” for the period 2000 to 2018.

Google search engine was also used with the following terms to search for studies: “human immunodeficiency virus,” “acquired immunodeficiency syndrome,” “fracture,” “fragility fracture,” “vertebral fracture,” “wrist fracture,” “hip fracture,” “bone mineral density,” “osteopenia,” “osteoporosis,” “prevalence,” and “epidemiology”.

We also reviewed the reference lists of relevant articles for any further relevant studies.

Study selection

The search for relevant studies and exclusion of papers was executed by two authors (CJ Chang, TW Tai) independently. Three authors (CJ Chang, YL Chan, TW Tai) performed data extraction and article appraisal. When discrepancies occurred, we reached consensus through discussion.

Articles designed without a control group, with unclear patient characteristics, or insufficient available data were excluded. We then appraised the studies and identified those eligible for inclusion in our analysis. The authors assessed the quality of each article by using the STrengthening the Reporting of OBservational studies in Epidemiology (STROBE) statement [9].

Data extraction

Two authors (CJ Chang, YL Chan) independently extracted the data from the articles.

We contacted the authors of the included articles for missing data. The following data were extracted from the included articles: (1) characteristics of patients, (2) body mass index, (3) fracture type, (4) fracture incidence and prevalence, (5) duration of follow-up, and (6) lumbar spine and hip bone mineral density.

Statistical analysis

We used the Mantel–Haenszel method and variance-weighted means to analyze the outcomes. The authors used I2 to figure out and quantify the effect of heterogeneity [10]. The range of I2 was from 0 to 100%. An I2 value > 50% was considered as obvious heterogeneity [11]. A random-effects analysis [12] was used to compare studies showing heterogeneity, and a fixed-effects analysis [13] was used to compare studies without obvious heterogeneity. We calculated the heterogeneity, mean difference, and relative risk (RR) for all outcomes in this meta-analysis. By using Egger funnel plots, we also assessed the possibility of publication bias [14]. We conducted this meta-analysis with the Review Manager version 5.3 software (The Nordic Cochrane Centre, The Cochrane Collaboration, Copenhagen, Denmark). The pooled data on fracture incidence were analyzed using the MedCalc Statistical Software version 19.0.5 (MedCalc Software bvba, Ostend, Belgium).

Results

Identified trials

We found 1013 potentially relevant articles by using our search strategy initially. The full texts of 75 articles were reviewed. After applying the inclusion and exclusion criteria, 35 articles were identified and included in our meta-analysis (including 228,901,329 people; 106,994 people living with HIV and 228,794,335 non-infected controls) (Fig. 1). Of these included articles, 18 studies focused on the influence of HIV infection on BMD in HIV patients compared with that in healthy controls (Table 1), and 11 studies discussed differences in the fracture prevalence between these two groups (Table 2). Six articles mentioned both issues [5, 31, 39, 41, 45, 46].

Fig. 1
figure 1

Flow chart of article selection

Full size image
Table 1 Included studies that provided data of bone mineral density
Full size table
Table 2 Included studies that provided data of fractures
Full size table

Bone mineral density analysis

Twenty-two articles provided lumbar spine BMD data, and 19 articles provided hip BMD data for both HIV-infected patients and non-HIV-infected controls.

The means of lumbar spine BMD ranged from 0.85 to 1.25 g/cm2 in the HIV-infected group and 0.92 to 1.31 g/cm2 in the non-HIV-infected group. The pooled data showed that lumbar spine BMD was significantly lower in HIV-infected patients than that in non-HIV-infected controls (−0.04, 95% confidence interval (CI): −0.05~−0.03, p < 0.00001, random effect model) (Fig. 2a).

Fig. 2
figure 2

Comparison of (a) spine and (b) hip bone mineral density (BMD) between people with and without HIV infection

Full size image

The means of hip BMD ranged from 0.8 to 1.05 g/cm2 in HIV-infected patients and 0.81 to 1.12 g/cm2 in non-HIV-infected controls. The pooled result revealed that HIV-infected patients also had significantly lower hip BMD compared with controls (−0.04, 95% CI: −0.05~−0.04, p < 0.00001, random effect model) (Fig. 2b).

Among the 22 articles that provided lumbar spine BMD data, there were 9 studies in which age and body mass index (BMI) were well-matched between the HIV-infected and HIV-uninfected subjects. The pooled data from these 9 studies showed that lumbar spine BMD was significantly lower in HIV-infected patients than in controls (p < 0.05) (Fig. 3a). In addition, seven age- and BMI-matched studies reported that hip BMD was also significantly lower in HIV-infected patients than that in controls (p < 0.05) (Fig. 3b).

Fig. 3
figure 3

Comparison of (a) spine and (b) hip BMD between people with and without HIV infection in age-matched and BMI-matched studies

Full size image

Fracture risk analysis

The all fracture prevalence in HIV-infected people and controls was reported in 17 studies (Table 3). Among HIV-infected people, the prevalence was 4.08% for all fractures, 2.66% for fragility fractures, 1.26% for vertebral fractures, 1.38% for hip fracture, and 1.38% for wrist fractures.

Table 3 Fracture prevalence among HIV-infected patients and non-HIV-infected controls
Full size table

The pooled data revealed that HIV-infected patients had a higher prevalence of all fractures than controls (4.08% versus 0.44%; relative risk (RR): 1.91, 95% CI: 1.46–2.49, p < 0.00001) (Fig. 4a). The prevalence of fragility fractures (including vertebral, hip, and wrist fractures) was provided in 13 studies. The pooled results found a significantly higher prevalence of fragility fractures in the HIV-infected group than in controls (2.66% versus 2.19%, RR: 1.68, 95% CI: 1.40–2.01, p < 0.00001) (Fig. 4b).

Fig. 4
figure 4

Comparison of the risk of (a) all fractures and (b) fragility fractures between people with and without HIV infection

Full size image

In addition to the analysis of fragility fractures, we also performed subgroup analyses of vertebral, hip, and wrist fractures. Six studies mentioned vertebral fracture rates, and the pooled results showed that the prevalence of vertebral fractures was significantly higher in HIV-infected patients than in controls (1.26% versus 0.37%; RR: 1.97, 95% CI: 1.22–3.2, p = 0.006) (Fig. 5a). Six articles provided data on the prevalence of hip fractures. The pooled data indicated a trend in that HIV-infected people were more likely to have hip fractures than healthy controls (1.38% versus 0.81%, RR: 1.88, 95% CI: 0.99–3.57, p = 0.05) (Fig. 5b). The pooled results for wrist fractures from four studies found that HIV-infected people were more likely to have wrist fractures than controls (1.38% versus 1.29%, RR: 1.67; 95% CI: 1.13–2.45; p = 0.009) (Fig. 5c).

Fig. 5
figure 5

Comparison of the risk of (a) vertebral fractures, (b) hip fractures, and (c) wrist fractures between people with and without HIV infection

Full size image

Six articles provided information about fracture incidence. The pooled data showed that the incident fracture rate per 100 person-years was higher in HIV-infected patients than that in healthy controls (1.72 versus 1.29 per 100 person-year) (Fig. 6).

Fig. 6
figure 6

Pooled data on the incidence of fractures in people with HIV infection (a) and non-HIV-infected controls (b)

Full size image

Discussion

The most important findings of this meta-analysis were that people living with HIV had lower lumbar spine and hip BMD and a higher prevalence of all fractures as well as fragility fractures, vertebral fractures, and wrist fractures than controls. People with HIV also had a higher incidence of fractures than controls.

Because of the increased life expectancy of people living with HIV, osteoporosis and fragility fractures may become emerging causes of morbidity. Low BMD and increased fracture risk in HIV-infected patients have been reported in a number of studies (studies included in Tables 1 and 2) [4,5,6, 15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46]. Previous meta-analysis conducted by Schiau et al. revealed that people with HIV infection have increased risk for all fractures and fragility fractures compared with non-HIV-infected population [47]. However, the relationship between low BMD and increased fracture risks in patients with HIV is not well-established. Whether low bone density leads to more fractures in people living with HIV is inconclusive [5, 48].

HIV-infected people have three-fold increase in osteoporosis with abnormal cortical and trabecular microarchitecture, which diminishes the bone stiffness. The risk of fracture, especially vertebral fracture, is higher in HIV-infected person, which is consistent with our results [7, 41, 49]. A recent systematic review performed by Ilha et al. showed that the risk of vertebral fracture in HIV-infected individuals is approximately double when compared with HIV-uninfected patients [50], which resembled the finding in our study (RR, 1.97). The proposed mechanisms include immune system reaction, vitamin D metabolism, insulin resistance, and lipodystrophy. However, a complex of involved mechanisms transforms the mechanical properties and affects bone quality and turnover. [51,52,53]. The complex regulation involved may transform bone mechanical properties and thus affect quality and turnover rate. HIV was found to lower BMD and cause bone fragility [54]. Several studies have assessed serum markers of bone formation and resorption in patients with HIV infection [55,56,57]. The serum bone formation marker, osteocalcin, was decreased in advanced stages of the disease and was correlated with CD4 cell counts [55, 58]. Increased IL-6 plasma levels were associated with greater BMD loss among HIV-infected individuals but not uninfected controls [59]. Increased levels of soluble CD14, which is responsible for macrophage activation, was inversely correlated with bone mineral density and content in HIV-infected males [60].

Highly active antiretroviral therapy has been proved to be related to reduced BMD in most studies. Most antiretroviral drugs produce complications that interfere with bone metabolism [61]. Brown TT et al. reported a 1.6 times greater loss in BMD among those on a protease inhibitor (PI) regimen compared to those on highly active antiretroviral therapy without a protease inhibitor [7].

Nucleoside reverse transcriptase inhibitors, such as ritonavir, have been associated with low bone mass when compared with non-nucleoside reverse transcriptase inhibitors [62]. Tenofovir, also a nucleoside reverse transcriptase inhibitor, could possibly lead to proximal tubule toxicity and was associated with reduced BMD in people with HIV [63]. Tenofovir may also cause an acute decrease in BMD [64]. A recent longitudinal study including 3251 people living with HIV in Japan concluded that the risk of fractures must be considered regardless of patient age and gender if they had tenofovir disoproxil fumarate (TDF) administration more than 5 years [65].

Low BMI was reported as the strongest predictor of low BMD, regardless of HIV status, in some populations of men who have sex with men [29]. In this meta-analysis, HIV-infected patients had significantly lower lumbar spine BMD than controls in age-matched and BMI-matched studies. These findings suggested that both HIV infection and low BMI are independent risk factors for low bone mass. The Fracture Risk Assessment Tool (FRAX) has been used to estimate fracture risk. However, a cross-sectional study revealed that FRAX had poor sensitivity and specificity for low bone mass among HIV-infected men with a median age of 45 years [66]. The ideal screening protocols of bone health among HIV-infected individuals are not clearly defined and are still controversial. The European AIDS Clinical Society recommends screening fracture risk in HIV-infected adults for individuals older than 40 years, while other authors have suggested screening people older than 50 years [46, 61, 67]. In our opinion, we suggest that the age of fracture risk screening should be lower in HIV-infected groups than in non-HIV-infected people.

This study had some limitations, despite being the largest scale systemic review of the impact of HIV infection on bone health and fracture prevalence. Some included articles were not specifically designed to evaluate fracture prevalence or BMD as their primary purpose. The information we extracted from those studies may have been influenced by confounding factors. This meta-analysis included different types of studies, including case-control, cross-sectional, and cohort studies, which could also lead to clinical heterogeneity.

People with HIV infection had lower lumbar spine and hip BMD and a higher prevalence of all fractures as well as fragility fractures, vertebral fractures, and wrist fractures than controls. People with HIV also had a higher incidence of fractures than controls. Targeted screening with management to maintain skeletal strength and reduce fragility is urgent for people with HIV.