Background. Higher levels of parathyroid hormone have been associated with the use of tenofovir disoproxil fumarate (TDF) in people with and without HIV infection. Yet, alterations in calcium levels have never been elucidated in detail. Objective. To compare the association of parathyroid hormone with serum calcium levels and other markers of calcium and bone metabolism in people living with HIV on TDF- and non-TDF-containing antiretroviral therapy. Patients and Methods. A retrospective single center cohort study in Munich, Germany. Median and interquartile ranges and absolute and relative frequencies were used to describe continuous and categorical variables, respectively. The Mann–Whitney U test and chi2-test were used for comparisons. Multivariate median regression was performed in a stepwise backward approach. Results. 1,002 patients were included (786 (78.4%) male; median age 48 (40–55) years). 564 patients (56.3%) had a TDF-containing ART regimen. PTH concentrations were 46.9 (33.0–64.7) pg/mL and 35.2 (26.4–55.4) pg/mL (P=0.001), 43.3 (30.8–59.8) pg/mL and 31.8 (22.3–49.6) pg/mL (P<0.001), 46.1 (29.5–65.4) pg/mL and 33.4 (22.6–50.1) pg/mL (P<0.001), and 37.8 (25.3–57.9) pg/mL and 33.8 (20.1–45.3) pg/mL (P=0.012) within the first, second, third, and fourth quartile of corrected calcium levels for patients with and without TDF-containing ART, respectively. In multivariate median regression, PTH concentration was significantly associated with Cacorr. (−32.2 (−49.8 to −14.8); P<0.001), female sex (5.2 (1.2–9.2); P=0.010), 25(OH)D (−0.4 (−0.5 to −0.3); P<0.001), and TDF-use (9.2 (6.0–12.5); P<0.001). Discussion. Higher levels of PTH seem to be needed to maintain normal calcium levels in PLWH on TDF-containing ART compared to non-TDF-containing ART. Optimal concentrations for 25-hydroxy vitamin D and calcium might therefore be different in people using TDF than expected from general populations but also people living with HIV with non-TDF-containing antiretroviral therapy. This might require different supplementation strategies but warrants further investigation.
On Multivariate Median Regression