In this randomized clinical trial among virologically suppressed HIV-infected adults, the introduction of fixed-dose treatment with TDF/FTC lead to decreases in hip and lumbar spine BMD compared with fixed-dose treatment with ABC/3TC. In addition, we detected increases in the majority of measured bone turnover biomarkers in TDF/FTC- compared with ABC/3TC-treated patients. There was no difference in renal function between the two study arms as evaluated by eCrCl or plasma cysC, but a minor increase in urinary NGAL/creatinine ratio was observed in the TDF/FTC arm compared with the ABC/3TC arm.
The strength of the study is the randomized design combined with exploratory analyses of several biomarkers of bone and renal function. Concurrent inflammation was assessed at baseline and did not differ between study groups. Furthermore, the entry criteria required that patients be negative for the HLAB5701 allele, had not previously been exposed to ABC or TDF, and had been virologically suppressed for at least 3 months prior to randomization.
Only one previous randomized clinical trial (STEAL) has evaluated BMD in virologically suppressed HIV-infected patients switching to a TDF- or ABC-based regimen 
. In the STEAL-study TDF/FTC significantly lowered hip and spine T scores at both weeks 48 and 96, whereas both T scores increased in the ABC/3TC group. The changes in T scores in the two study arms in our study were of a similar magnitude as in the STEAL-study (data not shown). However, we used changes in BMD as the principal bone endpoint measure and found reductions in BMD with TDF/FTC-treatment, but stable rather than increased BMD with ABC/3TC-treatment. Our patient material is much smaller and, consequently, our data less robust. Biomarkers of bone turnover increased in patients treated with TDF compared with ABC in a recently presented STEAL-sub study, but early changes in these biomarkers did not predict bone loss at week 96 
. We measured bone turnover biomarker as early as 4 weeks after randomization, which was not done in the STEAL-sub study, but generally the changes seen in biomarkers of bone turnover in the present study are in line with the findings of the STEAL- study. In the SMART-study BMD declined more with continuous than interrupted therapy, but no consistent drug-specific association with BMD decline was found 
. Bone turnover markers were measured in an unpublished sub-study of the SMART study and early changes (4 months after baseline) in some of these markers predicted BMD changes over 12 months 
. However, the SMART-study was not designed to evaluate differences in BMD between patients randomized to ABC- or TDF-based treatment and is not directly comparable to our study. Two HAART-initiation randomized trials of TDF versus ABC have evaluated effects on BMD 
in which BMD reductions with TDF therapy were similar, or slightly larger, than in our material, but comparison with these are not straightforward as decreases of BMD have been observed with HAART-initiation irrespective of regimen used 
Our results support the findings of previous studies 
that TDF treatment compared with ABC is associated with a decrease in BMD, but the clinical implications are not clear. First, this should be interpreted in the context of a normal reduction in BMD after the age of 30 both in premenopausal women (~0.3%/year), perimenopausal women (~1%/year), and men (0–0.5%/year) 
. Second, the loss of BMD is lesser than, but comparable to the 2–10% reduction in lumbar spine BMD that is seen with initiation of corticosteroids 
. Still, initiation of corticosteroid therapy is associated with an increase in incident fractures 
, which has not been observed with initiation or switch to TDF-based HIV-treatment 
, but studies so far have been too small or too short to sufficiently address this issue.
Whether the observed changes in BMD in our study and those of others are clinically relevant is a very interesting but difficult to answer question. The observed loss of BMD is quite small and not likely to confer a consequent increase in fracture risk in the majority of HIV-infected patients. At least there is no data to suggest that patients with higher risk of fractures should not receive TDF, but alternatives may be considered for those with previous fractures or known osteoporosis 
. In addition, no increase in risk of fracture among TDF-treated patients in the STEAL-study was revealed with use of the FRAX® tool 
, but it should be noted that the FRAX® tool is not validated for patients with BMD reductions within the previous year.
It has been suggested that the initial loss of BMD may stabilize after 1 year of treatment and therefore may be caused by a continued effect of viral replication on bone turnover until complete suppression, even though other explanations have also been put forward 
. Our data, and those of the STEAL-study, indicate a direct effect of anti-retroviral therapy (ART) on bone mass density. This is helpful in the efforts to understand how traditional risk factors for low BMD, HIV-infection and ART each contribute to the increased risk of low BMD and fractures among HIV-infected individuals.
We measured several bone turnover markers to provide additional information on the effects of these treatments on bone remodeling and potentially the mechanisms underlying the changes seen in BMD. Bone turnover markers increased in the TDF/FTC arm within the first 12 weeks after treatment switch and remained stably elevated hereafter. This indicate that loss of BMD is associated with an increased rate of bone resorption and formation that can be detected as early as 12 weeks after initiation of TDF, and possibly already after 4 weeks. The early time point evaluated in this study is particularly interesting, as comparable studies have not provided data on bone turnover biomarkers so shortly after treatment shift to a TDF-based regimen. Both OC and OPG showed initial decreases in the TDF/FTC arm followed by increase or stabilization, which raises the question if certain signaling events relevant for the bone remodeling effects of TDF are taking place within the first few weeks of treatment switch. However, this remains speculative and the meaning of these finding are uncertain. Whether the changes in markers of bone turnover are reflecting direct effects of the treatment on the osteoclasts or osteoblasts or both 
or an indirect effect via effects on the calcium homeostasis, including reduced activation of vitamin D and secondary increases in parathyroid hormone, remains to be shown 
. It is also unclear whether these bone turnover markers can be used for monitoring patients during treatment as early changes in these markers did not predict BMD loss similarly to what was observed in the STEAL sub-study 
We did not detect any differences between study arms in the changes in eCrCl or plasma cysC, but still observed a minor increase in 1 of 3 measured biomarkers of renal dysfunction among patients in the TDF/FTC arm. This is comparable to the findings in the HAART-initiation ASSERT-study 
where retinol-binding protein and β-2 microglobulin increased significantly more in the TDF/FTC group. In other HAART-initiation studies biochemical parameters of renal function were similar between TDF/FTC and ABC/3TC treated patients 
, but more patients in the TDF groups developed clinical renal impairment 
. No difference in eGFR, creatinine clearance or other renal endpoints was found among virologically suppressed HIV-patients in the STEAL-study 
The primary limitation of our study is the failure to reach the intended number of enrolled patients, which reduced the statistical power to detect differences between study arms. In addition, DXA was not performed in all patients and we only have 48 weeks of follow-up, which is too short to acknowledge the long-term effect on BMD reductions that appear to occur primarily within the first 24–48 weeks with stabilization hereafter. Interestingly, while eCrCl decreased over time in both study arms, no change in plasma cysC was observed indicating a stable GFR. We did not include GFR measurement by an exogenous marker and thus cannot compare these two estimates with more precise GFR-measurements. However, the eCrCl estimate based on plasma creatinine is dependent on factors other than GFR, including changes in muscle mass and tubular secretion of creatinine, and is possible that either the HIV infection or the anti-viral treatment may influence such factors. Previous studies comparing eCrCl and cysC based GFR estimates on HIV patients with normal renal function and on anti-viral treatment have generated mixed results 
. The 24-hour urine collections were not complete precluding us from determining 24-hour excretion of albumin, tubular markers, and the fractional phosphate excretion. Furthermore, due to incomplete sampling, urinary biomarkers were not measured in all patients at all time points. We did not, however, detect differences in plasma phosphate levels between study arms.
Long-term effects of HAART, including effects on bone and renal function, will continue to be outcomes of interest in future clinical trials. Beside this, more prospective studies will be needed to evaluate potential interventions to prevent or reverse loss of BMD. While several clinical trials have shown that bisphosphonate, vitamin D and calcium, either separate or in combination, are safe and effective in the treatment of decreased BMD among HIV-infected patients on HAART 
, there are still no published studies on interventions to prevent bone loss associated with TDF or HAART initiation.
In conclusion, we observed decreases in hip and lumbar spine BMD, in conjunction with early increases in bone turnover markers, in patients randomized to TDF/FTC-based treatment compared with ABC/3TC. Changes in renal function were not different between study arms although a very small increase in NGAL was detected among TDF-treated patients. The clinical significance of the latter is unclear.