There is a paucity of data on bone turnover markers in HIV-infected patients after HAART. Some studies demonstrated an increase in OC and CTx [14, 15] and these increases correlated with reduction of bone mineral density (BMD) . By contrast, other authors found significant decreased levels of OC in patients treated with HAART [12, 13]. Our data confirm a parallel increase of soluble markers indicating bone formation (OC) and resorption (CTx) upon initiation of HAART and this effect appeared to be maximal at week 24.
We found a significant increase of OPG after HAART, differently from the study by Brown et al.  who found a decrease of this marker. The authors hypothesized that decreased inflammation after HAART could reduce OPG. In the present study, inflammation was not assessed, but all patients achieved sustained undetectable HIV RNA (the main driver of inflammation), therefore other factors may be implicated.
Higher eGFR at baseline (measured by cystatin C using the CKD-EPI formula) predicted OPG increase from baseline to week 48. Glomerular filtration rate estimated through cystatin C may be a reliable method to assess renal function in HIV infected patients because it is less influenced by muscle mass and liver function than the methods based on creatinine [26, 27]. Therefore, our data suggest that patients with better eGFR are protected since OPG by itself is known to counteract osteoclastic activation by antagonism to RANKL . However, since cystatin C is increased by inflammation [29, 30], it has to be seen whether the pro-inflammatory status at baseline or the actual kidney damage is a better predictor OPG evolution.
Older age and increasing HIV RNA at baseline were independently predictive of CTx increase during the follow-up. Since the increase in CD4+ T-cells is greater in patients with higher HIV RNA at baseline [31, 32] the correlation between higher HIV RNA and CTx increase could be mediated by immune-reconstitution, as recently suggested . However, no significant correlation between Δ[wk 48-bl] of the CD4+ T-cell count and CTx increase was found in our study (data not shown), and this seems to contradict the previous hypothesis . Therefore, greater increase of CTx could be due to the fact that bone health is already impaired in older patients and in those with higher HIV RNA due to a damaging effect of HIV on bone health [34, 35]. However, the underlying mechanisms should be investigated.
Notably, the use of ATV/r was associated with greater CTx increase in our study. Brown et al.  found that OC increase, but not CTx increase, was associated with PI use. Notwithstanding this apparent inconsistency, both studies suggest that PIs are associated with up-regulation of bone remodeling. Although we could not provide reliable explanations, McComsey et al.  already demonstrated a deeper impact of ATV/r on lumbar vertebral BMD than EFV when co-administered with TDF. Our results are in line with--and may explain--this observation.
We found a significant PTH increase after HAART, confirming previous studies [8, 37, 38]. This increase could be due to low 25(OH) vitamin D levels. However, we did not evaluate 25(OH) vitamin D, so this hypothesis could not be proven. Rather, we were interested in 1,25-(OH)2 vitamin D in order to assess the availability of the final product exerting biological activity. We found that 1,25-(OH)2 vitamin D remained stable, while in other studies TDF seemed to increase it [11, 39]. This stability was observed independently from variations due to seasonality, a finding that was not confirmed by others . Therefore, the effect of TDF on 1,25 (OH)2 vitamin D needs further considerations.
This study suffers from several limitations. First of all, BMD was not assessed; therefore the early activation of bone markers could not be coupled with BMD reduction. Second, 1,25-(OH)2 vitamin D levels are unstable and its values are of difficult interpretation . Third, our study is descriptive, so we did not provide any data on possible mechanisms. For instance, it is impossible to infer whether the increase in bone markers was due to PTH increase or to a direct effect of antiretroviral drugs on bone metabolism. Fourth, the number of patients was small, but the study provided statistically significant results. Lastly, we studied a sub-group of patients from the randomized cohort based on the availability of stored samples and maintenance of the initial HAART regimen. Although we could not exclude a selection bias, this is unlikely because no patients dropped out of the study for bone complications and they were not systematically selected for inclusion in the study. Notwithstanding these limitations we believe that our results add interesting information given the paucity of the current literature data.