Bone Mineral Density Decline According to Renal Tubular Dysfunction and Phosphaturia in Tenofovir-exposed HIV-infected Patients

José L. Casado; Carmen Santiuste; Monica Vazquez; Sara Bañón; Marta Rosillo; Ana Gomez; María J. Perez-Elías; Carmen Caballero; José M. Rey; Santiago Moreno


AIDS. 2016;30(9):1423-1431. 

In This Article


Overall, 90 patients met the inclusion criteria and had detailed serum and urine evaluation (previous and at inclusion), before two successive DXA determinations, without any change in cART, including TDF. Baseline characteristics of the patients at time of the first DXA evaluation are shown in Table 1. Of note, mean age was 47.5 years and only 8% of patients were older than 60 years, 20% were women (18 cases), median time of HIV infection was around 14 years, 63% of patients were pretreated during a median time of 11 years, and HCV coinfection was observed in 28% of cases.

At inclusion, in the first DXA evaluation, after a median of 38 months of TDF [interquartile range (IQR), 19.9–60.7], osteopenia at spine and hip was observed in 49 and 48%, and osteoporosis in 9 and 2%, respectively. The characteristics of the patients with reduced BMD at hip (osteopenia or osteoporosis) are shown in Table 1. Patients were slightly older, with lower BMI, with higher prevalence of HCV coinfection, and with a longer history of HIV infection and of previous cART. BMD in women was similar at spine (0.94 vs 0.95; P = 0.68) and at femoral neck (0.76 vs 0.81; P = 0.15). Although there was no significant difference in time on TDF between both groups, a lineal correlation between time on TDF and a lower BMD at hip was observed (Spearman's rho = -0.27; P = 0.01; Fig. 1), not significant for BMD at spine.

Figure 1.

Bivariate correlation between time on tenofovir disoproxil fumarate and bone mineral density at femoral neck.

In a bivariate correlation (Table 2), a lower BMD was related to older age (r = -0.34–0.35; P < 0.01) and lower BMI (hip, r = 0.2; P = 0.04), but also to longer time of HIV infection, cumulative time on cART, and HCV coinfection (0.74 vs 0.82 g/cm2; P = 0.012). Moreover, there was a trend with lower CD4+ nadir and eGFR. Of note, a higher vitamin D status but not PTH level correlated with better BMD at hip (r = 0.24; P = 0.02), and BMD was significantly lower in case of vitamin D deficiency (0.71 vs 0.79 g/cm2; P = 0.01).

Serum and tubular parameters were evaluated at the same time of DXA scan, at the inclusion visit. Almost 80% of patients had one or more tubular abnormality, and 30% had tubular dysfunction. Among the tubular alterations, proteinuria was observed in 40% (median, 98.1; IQR, 75–164.1), hyperphosphaturia in 50% of cases (fractional excretion of phosphate, median 19.9%, IQR 15.1–25.4), hyperuricosuria in 23% (median 7.2, IQR 5.8–9.3), and glycosuria normoglucemic in 8%. Urinary B2M and RBP were 3.4 log μg/g (IQR, 3.18–3.73) and 2.71 log μg/g (2.58–2.92), respectively. Micro and macroalbuminuria were rare (3 and 1%), and hypophosphataemia was observed in 17% of patients (15 cases). Of note, there was a significant inverse correlation between fractional excretion of phosphate and serum phosphate (Spearman's rho = -0.55; P < 0.01), and hyperphosphaturia was also correlated with tubular proteinuria (RBP, r = 0.27; P < 0.01; B2M, r = 0.28; P < 0.01), and hyperuricosuria (r = 0.22; P = 0.04).

At this time, the association of tubular renal parameters with BMD is shown in the Table 2. In more detail, a lower BMD was associated with number of tubular abnormalities (P = 0.04), and there was a trend with B2M (r = -0.18; P = 0.08), RBP (r = -0.13; P = 0.1), but not with glycosuria, uricosuria, or proteinuria. BALP was increased in 11% of patients (mean, 93, 1 mg/dl; 33–266), and correlated with lower BMD at spine. Notably, a reduced BMD at hip was observed with higher fractional excretion of phosphate (Spearman's rho = -0.25; P = 0.03; Fig. 2a), with significant differences for a fractional excretion more than 10% (0.79 vs 0.92 g/cm2; P = 0.03).

Figure 2.

(a) Correlation between phosphaturia (fractional excretion of phosphate) and bone mineral density at femoral neck at inclusion. (b) Bivariate correlation between bone mineral density at hip and changes in fractional excretion of phosphate (%) before inclusion.

Strikingly, phosphataemia and phosphaturia changed during the follow-up. As the previous serum and urine determination, performed a median of 9 months before (IQR, 4.3–12.9), there was a reduction in hypophosphataemia rate (from 27 to 17%), and an improvement in the fractional excretion of phosphate in 66% of patients. This improvement in phosphaturia was especially observed in patients with hypophosphataemia (-2.9% vs +1.47%; P = 0.03), and with more severe hyperphosphaturia (r = -0.77; P < 0.01), suggesting compensatory factors. Indeed, these improving patients had greater levels of 25OHD (22.1 vs 18.7; P = 0.1), albeit it was not significant. Thus, changes in phosphaturia produce a normalization of serum phosphate in 54% of patients with previous hypophosphataemia (13 out of 24). Of note, patients with this improvement in phosphataemia and phosphaturia showed greater BMD at hip (r = -0.33; P < 0.001; Table 2 and Fig. 2b), and at spine (r = -0.22; P = 0.05).

A second DXA evaluation was performed after a median time of 40.8 months (IQR, 33.8–45.1), and after a median time of 76.3 months on TDF (IQR, 59.8–102; 627.7 patients-year). There was a median decrease of -0.1% (IQR, -2.87 to 2; 24% had ≥3% reduction) at lumbar spine, and of -1.45% (IQR, -4.3 to 1.5%) at hip, with 32 study participants (36%) having BMD loss of at least 3%. Thus, a greater proportion of study participants had osteopenia and osteoporosis (48 and 15% at spine, 57 and 2% at hip).

In univariate linear regression models, hip BMD decline (45 patients, 50%) was correlated with a higher phosphaturia (β, -0.37; P < 0.01), proteinuria (β, 0.32; P = 0.03), and, as expected, with number of tubular abnormalities (β, -0.24, P = 0.03) (Table 3). There was a trend to association with glycosuria, RBP and B2M, and it was greater albeit not significant in patients receiving TDF plus a protease inhibitor (PI, -2.3% vs -0.84%; β, -0.14; P = 0.12). There were no clear relationships between bone loss and age, HCV coinfection, CD4+ cell count nadir, or baseline GFR. Again, BMD decline was not different according to sex (-1.45% in both men and women). A small improvement was observed with higher 25OHD levels (β, 0.17; P = 0.1). Interestingly, BMD decline was associated with better BMD at inclusion (β, -0.28; P = 0.01). In the multivariate lineal regression model, fractional excretion of phosphate (β, -0.31; P = 0.01), use of a boosted protease inhibitor (β, -0.47, P = 0.03), and BMD at inclusion (β, -0.33; P = 0.03) were associated with BMD decline after adjusting by age, sex, BMI, eGFR at inclusion, uPCR, glycosuria, and nadir CD4+. Notably, there was a trend between BMD decline and longer time on TDF (β, -0.29; P = 0.06) and vitamin D deficiency (β, 0.29; P = 0.06). The model was repeated, including number of tubular abnormalities instead of uPCR, glycosuria, or phosphaturia, and this variable was found to be significant (number of tubular abnormalities ≥3, β, -0.41; P = 0.01), without changes in the remaining variables.