Antiretroviral Therapy Containing HIV Protease Inhibitors Enhances Fracture Risk by Impairing Osteoblast Differentiation and Bone Quality

Abstract

Long-term antiretroviral therapy is associated with increased fracture risk, but the mechanism remains elusive. We measured serum undercarboxylated osteocalcin and pentosidine (markers of poor bone quality) in human immunodeficiency virus–infected patients treated with protease inhibitors (PIs) or an integrase strand transfer inhibitor–containing regimen. The results demonstrated significantly higher undercarboxylated osteocalcin and pentosidine in PI-treated patients. Switching to integrase strand transfer inhibitor significant decreased these markers. We also showed impaired bone mechanical properties with higher undercarboxylated osteocalcin level in PI-treated mice and inhibited osteoblast differentiation in PI-treated osteogenic cells. The results confirmed the adverse effects of PIs on bone quality and osteoblast differentiation.

Antiretroviral therapy (ART) has beneficial and preventive effects for both AIDS-related and non–AIDS-related events in individuals infected with human immunodeficiency virus (HIV) [1]. Based on these properties, it is recommended that ART be introduced at earlier stages of HIV disease. However, because life-long treatment is necessary once treatment is initiated, some authors have raised concern about the long-term adverse effects of ART on the cardiovascular system, kidney, and bone.

HIV-infected individuals are at higher risk of fragility fracture [2]. Bone mineral density (BMD) is widely used to assess fracture risk, and several studies have demonstrated low BMD in HIV-infected individuals [3]. Although HIV infection itself affects BMD, the use of ART is also associated with a reduction in BMD, which varies among the available ART regimens [4], suggesting that selection of appropriate antiretroviral agents is critical for bone health.

Interestingly, a meta-analysis has shown that although various ART regimens accelerated BMD loss during the first couple of years of treatment, BMD remained stable or slightly increased in the long term [5]. Another study showed that patients who received ART exhibited significant increase in the risk of fragility fracture compared with population controls, and that this increase was observed over a period of 10 years, rather than being limited to the first couple of years after ART initiation [6]. Such higher fracture risk cannot be rationally explained only by reduced BMD. Other factors, such as impaired bone quality, are more likely to be associated with the high fracture risk.

Protease inhibitors (PIs) have been reported to facilitate osteoclast differentiation, resulting in decreased BMD [7]. However, there is little or no information on the effect of antiretroviral agents on bone quality or osteoblast differentiation. In the present study, the physiological effects of PIs or integrase strand transfer inhibitor (INSTI) on bone quality and osteoblast differentiation were determined in HIV-infected patients, PI-treated mice, and cultured osteogenic cells.

METHODS

Clinical Studies of HIV-Infected Patients

The AIDS Clinical Center at the National Center for Global Health and Medicine stocks serum samples of HIV-infected patients under signed informed consent for future use in medical research. The original consent was obtained under an institutional review board–approved protocol. The ethics committee approved the use of such samples in this study. Serum samples were collected from 30 ART-naive patients before and 1 year after ART introduction. The 3 ART regimens were as follows, in 10 patients each: darunavir (DRV; 800 mg/d), ritonavir (RTV; 100 mg/d), abacavir (ABC; 600 mg/d), and lamivudine (3TC; 300 mg/d); lopinavir (LPV; 800 mg/d), RTV (200 mg/d), stavudine (60 or 80 mg/d), and 3TC (300 mg/d); and raltegravir (RAL; 800 mg/d), tenofovir (300 mg/d) and emtricitabine (200 mg/d). To analyze the effect of regimen switch from PIs to INSTI, we also collected serum samples from another 10 patients before and 6 months after the switch of their regimen from LPV (800 mg/d), RTV (200 mg/d), ABC (600 mg/d), and 3TC (300 mg/d) for 2–5 years, to RAL (800 mg/d), ABC (600 mg/d), and 3TC (300 mg/d).

Animal Studies

Eight-week-old male ICR mice were treated with either LPV/RTV (20 mg/kg/5 mg/kg; Kaletra) or vehicle orally once a day for 4 weeks (n = 10 in each group). The femurs, tibias, and serum samples were collected at 12 weeks old. All animal experiments were performed after obtaining approval from the Animal Study Committee of Tokyo Medical and Dental University.

Micro–Computed Tomographic Analysis

The left femurs of LPV/RTV-treated mice were used for micro–computed tomographic analysis (Comscan Techno). Two-dimensional images of the distal femur were obtained and reconstructed using TRI/3D-BON software (RATOC).

Histomorphometric Analysis

To evaluate bone remodeling histologically, we measured bone formation or resorption parameters with an osteometric analysis system (OsteoMetrics), using nondecalcified sections of the left tibias from LPV/RTV-treated mice.

Mechanical Property Test

To evaluate the fracture risk in LPV/RTV-treated mice, the right femurs were subjected to 3-point bending tests, in which a load is placed on the bone and increased gradually until the bone is broken. At the crosshead speed of 2 mm/min, we obtained the load displacement curve and calculated the maximum bending load, stiffness, and energy absorption.

Enzyme-Linked Immunosorbent Assay

The concentrations of undercarboxylated osteocalcin and carboxylated osteocalcin were measured using Glu-osteocalcin and Gla-osteocalcin enzyme-linked immunosorbent assay kits (Takara), respectively.

Cell Culture

Calvarial preosteoblasts were isolated from 4-day-old mouse pups by enzymatic digestion. To induce osteogenic differentiation, we incubated preosteoblasts in alpha-modified Eagle’s medium, supplemented with 10% fetal bovine serum, 0.1 mg/mL ascorbic acid (Sigma Aldrich), and 10 mmol/L β-glycerophosphate (Sigma Aldrich). RAL, DRV, LPV, and RTV were added during osteogenic induction. After 7-day treatment, alkaline phosphatase activity was measured with an alkaline phosphatase assay kit (Wako). Alizarin red staining was also performed after 14-day treatment to evaluate the amount of mineralization. For this purpose, cultured osteoblasts were fixed with 10% formalin for 20 minutes and then incubated with 10 mg/mL alizarin red S (Sigma Aldrich) for 30 minutes.

Statistical Analysis

All data are represented as means with standard deviations. Statistical analyses were performed using Student or Welch t tests or 1-way analysis of variance followed by Bonferroni multiple-comparison test, as appropriate. Differences were deemed statistically significant at P < .05.

RESULTS

Poor Bone Quality in PI-Treated HIV-Infected Patients and Its Prevention by RAL

We collected serum samples of 30 HIV-infected patients before and 1 year after treatment with PIs (DRV/RTV or LPV/RTV) or an INSTI (RAL)-containing regimen. To investigate the effects of PIs or INSTI on bone quality, we measured the concentration of serum undercarboxylated osteocalcin. Undercarboxylated osteocalcin is secreted from osteoblasts, and high concentration reflects poor bone quality and predicts an increased risk of bone fracture [8, 9]. Serum undercarboxylated osteocalcin levels were higher under both treatment regimens. However, the rate of increase in undercarboxylated osteocalcin was significantly higher in the PI-treated group (Figure 1A), suggesting more extensive reduction of bone quality by PIs than by RAL.

We also examined the concentration of serum pentosidine, a nonenzymatic collagen cross-link produced by advanced glycation end products and widely used as a marker of poor bone quality [10]. High urinary or serum pentosidine levels are associated with vertebral fracture risk, independent of BMD [10]. In our study, 1-year treatment with PIs significantly increased serum pentosidine levels (Figure 1B). These findings indicate that PIs, but not RAL, adversely affect bone quality.

To compare the effects of PIs and RAL in the same patient, we analyzed serum samples from another group of 10 patients, whose treatment regimens were switched from PIs to RAL. These patients were initially treated with a PI-containing regimen for 2–5 years. This is a time point at which their BMDs have been reported to be stable after ART initiation [5]. Interestingly, 6 months after the switch to RAL, serum undercarboxylated osteocalcin and pentosidine levels were significantly lower than before switching (Figure 1C and 1D), suggesting that RAL rescued impaired bone quality caused by PIs.

Impairment of Bone Quality and Increased Fragility Fracture Risk in PI-Treated Mice

To elucidate the pathophysiological effects of PIs on bone metabolism, 8-week-old ICR mice were treated orally with LPV/RTV (Kaletra) for 4 weeks. Micro–computed tomographic and histomorphometric analyses showed no significant differences in bone volume and bone histomorphometric parameters between the LPV/RTV-treated and control groups (Supplementary Figure S1A and S1B). On the other hand, 3-point bending test showed a significantly lower energy absorption in the LPV/RTV-treated group (Figure 2A). We also measured the serum undercarboxylated osteocalcin and pentosidine levels in these mice. Although serum pentosidine was not up-regulated in LPV/RTV-treated mice (data not shown), serum undercarboxylated osteocalcin levels were significantly higher in the treated mice (Figure 2B). These findings suggest that the LPV/RTV regimen significantly impairs bone mechanical properties and reduces bone quality in vivo, resulting in increased fracture risk, despite minimal bone loss.

Inhibition of Osteoblast Differentiation by PIs but Not RAL

We also investigated the effects of antiretroviral drugs on osteoblast differentiation. Mouse primary osteoblasts or human mesenchymal stem cells were cultured in the presence of PIs (DRV/RTV or LPV/RTV) or INSTI (RAL) during osteogenic induction. Alkaline phosphatase assay showed that PIs significantly decreased alkaline phosphatase activity (Figure 2C and Supplementary Figure S2). Alizarin red staining also confirmed the inhibitory effect of PIs on osteoblast mineralization (Figure 2D), suggesting that PIs, but not RAL, inhibit osteoblast differentiation.

We also examined the messenger RNA (mRNA) level of osteocalcin (Ocn) in osteoblasts cultured with DRV, RTV, or RAL. Osteocalcin is a protein secreted from osteoblasts and widely used as a marker for mature osteoblasts [11]. Quantitative real-time PCR showed that RTV significantly suppressed Ocn mRNA expression compared with the control group (Supplementary Figure S3A). Osteocalcin secreted from osteoblasts incubated with RTV was also significantly reduced (Supplementary Figure S3B). To further investigate this mechanism, we focused on Runx2, because Runx2 binds to the osteocalcin promoter sequence elements, OSE2, and stimulates the transcription of Ocn mRNA [12]. Western blot analysis demonstrated significant down-regulation of Runx2 in the DRV- or RTV-treated group. The expression of osterix, a downstream target of Runx2, was also down-regulated in the PI-treated group (Supplementary Figure S3C). These results suggest that PIs significantly inhibit osteoblast differentiation by down-regulation of Runx2 and suppression of osteocalcin promoter activity, whereas RAL did not have such an effect.

Discussion

The long-term adverse effects of HIV antiretroviral agents could be serious, at least in some patients. ART is known to reduce BMD and increases fracture risk in HIV-infected patients [6]. Previous clinical studies indicated that treatment with PIs accelerates BMD loss during the first couple of years [13, 14]. However, BMD loss caused by PIs stabilizes in the long term [5, 13], although the fracture risk remains high during treatment [6]. These findings suggest that the high fracture risk is due not only to the reduced BMD but also to other factors, such as impaired bone quality. However, the effects of antiretroviral agents on bone quality and osteoblast differentiation have not been addressed.

In the present study, serum undercarboxylated osteocalcin and pentosidine levels were significantly higher in HIV-infected patients treated with DRV/RTV or LPV/RTV, than in those treated with RAL (Figure 1A and 1B). High undercarboxylated osteocalcin or pentosidine levels are correlated with poor bone quality and high fracture risk [8]. Our findings suggest that PIs adversely affect bone quality in the clinical setting. Interestingly, switching to RAL from PIs was associated with a significant decrease in the concentrations of these markers (Figure 1C and 1D), indicating that RAL protects against impairment of bone quality.

To determine the pathophysiological effects of PIs on bone quality in vivo, we treated wild-type mice with LPV/RTV. The results showed no significant differences in bone volume and histomorphometric parameters (Supplementary Figure S1A and S1B), indicating that the inhibitory effect of PIs on bone mass is limited in the noninfectious condition. It is noteworthy that LPV/RTV-treated mice showed decreased bone mechanical properties despite minimal bone loss (Figure 2A). Bone strength is attributed to BMD and bone quality, suggesting that PIs seem to affect bone quality rather than BMD. In fact, high serum undercarboxylated osteocalcin level was noted in LPV/RTV-treated mice (Figure 2B). In our study, serum pentosidine concentration was not up-regulated in the treated mice. Pentosidine is a well-known marker of poor bone quality in human. However, the significance of pentosidine in rodents is controversial. Several previous studies have reported that serum pentosidine level was not correlated with impaired bone quality in mice or rats [15].

In the in vitro arm of the present study, we showed the inhibitory effect of antiretroviral agents on osteoblast differentiation. DRV/RTV or LPV/RTV significantly suppressed osteoblast differentiation (Figure 2C and 2D), whereas RAL had no such effect. We also demonstrated that RTV suppressed Ocn mRNA and Runx2 protein levels (Supplementary Figure S3A and S3C). Ocn expression is regulated by Runx2 binding to the OSE2 element in the osteocalcin promoter, suggesting that RTV-induced down-regulation of Runx2 caused the suppression of the promoter activity of osteocalcin. Our present findings suggest that the effect on bone quality should be considered when selecting the most appropriate ART regimen. INSTI-containing regimens are strongly recommended for the treatment of HIV-infected patients at higher risk of fragility fracture.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online. Consisting of data provided by the authors to benefit the reader, the posted materials are not copyedited and are the sole responsibility of the authors, so questions or comments should be addressed to the corresponding author.

Notes

Acknowledgments.  We gratefully acknowledge Kureha Special Laboratory, Tokyo, for conducting bone mechanical property tests and bone histomorphometric analysis.

Financial support.  This work was supported by the National Center for Global Health and Medicine (grant 28-SI-1102) and the Japan Agency for Medical Research and Development (grants AIDS-H26 and AIDS-H28).

Potential conflicts of interest.  H. G. has received honoraria from ViiV Healthcare, Merck Sharp & Dohme, Janssen Pharmaceutical, and Torii Pharmaceutical. S. O. has received honoraria and research grants from Merck Sharp & Dohme, AbbVie GK, Janssen Pharmaceutical, Pfizer, ViiV Healthcare, and Roche Diagnostics and honoraria from Astellas Pharmaceutical, Bristol-Myers Squibb, Daiichisankyo, Dainippon Sumitomo Pharma, GlaxoSmithKline, Taisho Toyama Pharmaceutical, and Torii Pharmaceutical. All other authors report no potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.

References

Author notes