Higher vitamin D levels in HIV-infected out-patients on treatment with boosted protease inhibitor monotherapy

Authors


  • Parts of this work have been previously presented at the XIXth International AIDS Conference, 22–27 July 2012, Washington, DC (Abstract MOPE085) and the 11th International Congress on Drug Therapy in HIV Infection, 11–15 November 2012, Glasgow, UK (Abstract P58).

Abstract

Objectives

We investigated the vitamin D status of patients receiving frequently used types of combination antiretroviral therapy (cART), including boosted protease inhibitor (PI) monotherapy.

Methods

For this cross-sectional study, out of 450 HIV-infected patients followed in the Hospital Severo Ochoa (Madrid, Spain), we selected 352 patients for whom vitamin D levels had been measured (January 2009 to December 2010). We collected the following data: demographics, cART duration, main cART regimen, viral load (VL), CD4 cell count, and concentrations of 25(OH)-vitamin D [25(OH)-D], parathyroid hormone (PTH), albumin and calcium. Vitamin D status cut-off points were: (1) deficiency (vitDd): 25(OH)-D < 20 ng/mL; (2) insufficiency (vitDi): 25(OH)-D from 20 to 29.99 ng/mL; and (3) optimal (vitDo): 25(OH)-D ≥ 30 ng/mL.

Results

The percentages of patients with vitDd, vitDi and vitDo were 44, 27.6 and 28.5%, respectively. Twenty-nine out of 30 (96.7%) Black patients had vitDd or vitDi, vs. 71.6% in the global sample (P < 0.001). Former injecting drug users (IDUs) had a higher prevalence of vitDo (P < 0.001) than patients in other transmission categories. Among patients with vitDd, vitDi and vitDo, the proportions of patients with a VL ≤ 50 HIV-1 RNA copies/mL were 77.4, 68 and 91%, respectively (P < 0.0001). Of the cART regimens, only boosted PI monotherapy was associated with significant differences in vitamin D levels (P = 0.039). Multivariate logistic regression analysis showed an increased risk of vitDi or vitDd associated with the following variables: Black vs. Caucasian ethnicity [odds ratio (OR) 10.6; 95% confidence interval (CI) 1.2–94; P = 0.033]; heterosexual (OR 2.37; 95% CI 1.13–4.93; P = 0.022) or men who have sex with men (MSM) (OR 3.25; 95% CI 1.25–8.50; P = 0.016) transmission category vs. former IDU; and VL > 50 copies/mL (OR 2.56; 95% CI 1.10–7.25; P = 0.040). A lower risk of vitamin D insufficiency or deficiency was found in patients on boosted PI monotherapy vs. no treatment (OR 0.08; 95% CI 0.01–0.6; P = 0.018).

Conclusions

Our data show an increased risk of vitamin D deficiency or insufficiency in patients with detectable VL and a Black ethnic background. Among cART regimens, boosted PI monotherapy was associated with a lower risk of vitamin D deficiency or insufficiency. The more favourable vitamin D status in former IDUs was probably attributable to a higher frequency of outdoor jobs in this group of patients.

Introduction

Concerns about the effects of combination antiretroviral therapy (cART) on bone metabolism have led to many studies being carried out in an attempt to clarify this issue. The use of cART increases by 2.5 times the probability of osteopenia [1], mainly during the first 2 years on treatment. During this period, bone mineral density (BMD) drops by between 2 and 6% [2]. In addition to the associations of cART with osteopenia and osteoporosis, HIV infection in itself has been related to increased rates of bone fractures, including hip and wrist fractures [3, 4].

Recently, many studies have described a high prevalence of vitamin D deficiency in HIV-infected patients [5-8]. However, vitamin D deficiency was more prevalent in noninfected women than in HIV-infected women in the WHIS (Women's health initiative study) [9]. It is possible that the additive effects of HIV infection and use of cART may exacerbate the metabolic complications of vitamin D deficiency.

In addition to its functions in bone metabolism, vitamin D plays a role in insulin resistance, metabolic syndrome and immune modulation [10]. The metabolic complications of cART (such as reductions in BMD) may be mediated, at least to some extent, by alterations of vitamin D metabolism induced by these drugs through the cytochrome P450 (CYP) pathway [11]. Although many studies have shown that efavirenz is associated with lower vitamin D levels [7, 12, 13], only a few have shown that vitamin D levels decrease over time with its use [14-16]. However, protease inhibitors (PIs) and tenofovir have been found to increase vitamin D levels [1, 7]. As each drug acts on different steps of vitamin D metabolism, we speculated that it may be interesting to investigate the effects of various therapeutic protocols on vitamin D status.

Few studies have assessed the relationships between various cART regimens and vitamin D status, and studies examining the effects of boosted PI monotherapy are lacking. Using a novel approach, Kim et al. [17] investigated the relationships between various cART regimens and vitamin D status. In that study, no statistically significant association was found between cART regimen and vitamin D status, but boosted PI monotherapy was not assessed. In our study, performed in a cohort of HIV-infected patients in an urban Spanish setting, we assessed the associations with vitamin D status of classical factors commonly associated with vitamin D deficiency, factors related to HIV infection, and the most frequently used therapeutic regimens.

Recently, the use of boosted PI monotherapy has been considered an acceptable alternative to cART in specific clinical situations, as a strategy to avoid nucleoside reverse transcriptase inhibitor (NRTI) toxicity [18]. It is therefore possible to include boosted PI monotherapy in antiretroviral regimens.

Methods

For this cross-sectional study, out of 450 HIV-infected out-patients followed in the Hospital Severo Ochoa (Madrid, Spain), we selected 352 patients for whom vitamin D levels had been measured (January 2009 to December 2010). The Hospital Severo Ochoa, located at a latitude of 40° N, has a reference population of 180 000 mostly urban inhabitants. The chosen vitamin D status cut-off points were: (1) deficiency (vitDd): 25(OH)-vitamin D [25(OH)-D] < 20 ng/mL; (2) insufficiency (vitDi): 25(OH)-D from 20 to 29.99 ng/mL, and (3) optimal (vitDo): 25(OH)-D ≥ 30 ng/mL. We adopted this definition of optimal vitamin D level (≥ 30 ng/mL) in the light of increasing evidence that under this cut-off point there can be an increased risk of osteoporosis, fractures, cancer and cardiovascular disease [10, 19, 20]. Data were extracted from the hospital database of adult HIV-infected patients, after approval had been obtained from the local ethical committee. We assessed the following demographic and clinical characteristics: age, gender, race, body mass index (BMI), duration of cART, most frequently used cART regimens, viral load, CD4 cell count, 25(OH)-D and PTH levels (using electrochemiluminescence), and serum albumin and calcium levels.

The electrochemiluminescence vitamin D determination method has been validated with the methods of graphic liquid chromatography–tandem mass spectrometry (r = 0.92) and radio immunoanalysis (r = 0.859).

Patients with renal insufficiency [glomerular filtration rate < 60 mL/min estimated using the modified diet renal disease (MDRD) method], hepatic dysfunction [aspartate aminotransferase (AST) or alanine aminotransferase (ALT) > 5 times the upper limit of normal (ULN)], pregnant patients and patients already receiving vitamin D supplements were excluded from the analysis. We also excluded patients receiving concomitant medications that may interfere with vitamin D status, such as rifampicin, corticosteroids, bile acid sequestrants and CYP3A4 inhibitors. Hepatitis B was defined by the presence of hepatitis B virus surface antigen and hepatitis C by positive serological results or a detectable hepatitis C virus (HCV) plasma viral load.

Statistical analysis was performed with the spss package, version 20.0 for Mac (SPSS, Chicago, IL, USA). To assess the significance of differences among the three vitamin D status groups, continuous variables were compared using the Kruskal–Wallis test and categorical variables using the χ2 test. We used the Kruskal–Wallis test to compare ethnic groups stratified by viral load. Finally, logistic regression was performed to assess factors related to vitamin D insufficiency or deficiency. Those variables having P < 0.20 in univariate analysis were introduced into the model (see Table 3 below). We calculated odds ratios (ORs) and 95% confidence intervals (CIs). P < 0.05 was considered statistically significant.

Results

Vitamin D levels were assessed in 352 adult patients. One-hundred and sixteen (33%) were female, and their median age was 28 years. A Caucasian ethnic background prevailed in the sample (84.9%). The median CD4 cell count was 501 cells/μL; the median HIV viral load was 40 HIV-1 RNA copies/mL, and 277 patients (78.7%) had a viral load ≤ 50 copies/mL. A total of 310 patients (88.1%) were on cART. Vitamin D levels were measured in summer or autumn in 60.5% of the sample.

The proportions of vitDd, vitDi and vitDo were 44, 27.6 and 28.4%, respectively (Table 1).

Table 1. Baseline characteristics of HIV-infected patients according to vitamin D status
 vitDd (n = 155; 44%)VitDi (n = 97; 27.6%)VitDo (n = 100; 28.4%)P-value
  1. ART, antiretroviral therapy; HBV, hepatitis B virus; HCV, hepatitis C virus; IDU, injecting drug use; PTH, parathyroid hormone; IQR, interquartile range; MSM, men who have sex with men; vitDd, vitamin D deficiency; vitDi, vitamin D insufficiency; vitDo, vitamin D optimal level.
  2. *Statistically significant difference at P < 0.05.
Age (years) [median (IQR)]28 (23–34)28 (25–35)27 (23–35)0.77
Sex [n (%)]   0.06
Female59 (38.1)23 (23.7)34 (34)
Male96 (61.6)74 (76.3)66 (66)
Ethnic background [n (%)]    
Caucasian125 (80.6)83 (85.6)91 (91)0.07
Black22 (14.2)7 (7.2)1 (1)0.001*
Hispanic (South American)8 (5.2)7 (7.2)8 (8)0.64
Risk group [n (%)]    
Heterosexual76 (49.1)34 (35.1)27 (27)0.001*
MSM18 (11.6)27 (27.8)9 (9)0.0001*
IDU59 (38.1)33 (34)63 (63)0.0001*
Other2 (1.3)3 (3.1)1 (1)0.45
Calcium (mg/dL) [median (IQR)]9.3 (9–9.6)9.2 (8.5–9.5)9.2 (8.9–9.5)0.08
Albumin (g/dL) [median (IQR)]4.4 (4.2–4.6)4.4 (4.1–4.6)4.4 (4.2–4.6)0.59
PTH (pg/mL) [median (IQR)]48.2 (34.2–67.8)46 (32.8–59.4)45.2 (34.5–59.9)0.51
HCV infection [n (%)]63 (40.6)39 (40.2)67 (67)0.0001*
HBV infection [n (%)]3 (1.9)6 (6.2)4 (4)0.22
CD4 count (cells/μL) [median (IQR)]519 (334–700)449 (307–637)505 (360–660)0.36
Viral load [n (%)]   0.0001*
> 50 copies/mL35 (22.6)31 (32.4)9 (9)
≤ 50 copies/mL120 (77.4)66 (68)91 (91)
ART [n (%)]73 (47.1)45 (46.4)67 (67)0.003*
Season [n (%)]    
Spring52 (33.5)15 (15)1 (1)0.0001*
Summer20 (12.9)25 (25.8)37 (37)0.0001*
Autumn35 (22.6)38 (39.2)58 (58)0.0001*
Winter48 (31)19 (19.6)4 (4)0.0001*

There were significant differences according to ethnic background (Black, Caucasian and South American Hispanic). We found that 96.7% of Black patients had vitDd or vitDi, vs. 71.6% in the global sample (P < 0.001).

There was an increased percentage of patients with vitDo in those with HCV coinfection, for whom the proportions of vitDd, vitDi and vitDo were 40.6, 40.2 and 67%, respectively (P < 0.0001).

There were differences according to the presumed transmission category (P < 0.001), with an increased proportion of former injecting drug users (IDUs) having vitDo (Table 1).

Seasonality was also observed in our study. In spring and winter, only 1 and 4% of the blood samples showed vitDo, whereas in summer and autumn these percentages were 37 and 58%, respectively (P < 0.0001).

There were no differences in CD4 cell count among the vitamin D status groups (P = 0.349). However, there was a significant difference among these groups in the proportion of patients with an HIV viral load ≤ 50 copies/mL, with 77.4, 68 and 91% respectively, of patients in the vitDd, vitDi and vitDo groups having a viral load ≤ 50 copies/mL (P < 0.0001).

We compared vitamin D status between patients with detectable (> 50 copies/mL) and undetectable (≤ 50 copies/mL) viral loads, and between patients with Black and non-Black ethnic backgrounds. Patients with an undetectable viral load had significantly higher vitamin D levels than those with a detectable load (P = 0.045). There were significant differences among ethnic groups. In Black patients, the median vitamin D level was 14.05 ng/mL [interquartile range (IQR) 9.2–20.975 ng/mL], in Caucasian patients it was 23.3 ng/mL (IQR 15.8–32.5 ng/mL) and in Hispanic patients it was 24.2 ng/mL (IQR 13–31.8 ng/mL) (P < 0.0001).

Within ethnic groups (Black, Caucasian and Hispanic), no significant differences in vitamin D level were found according to viral load (> 50 vs. 50 copies/mL) (P = 0.539, 0.184 and 0.624, respectively).

We assessed the relationship between the most frequently used cART regimens and vitamin D level (Table 2). As we found an association between a viral load > 50 copies/mL and vitDd, we performed an analysis restricted to patients with a viral load ≤ 50 copies/mL (Fig. 1). cART was classified into nine regimes. Of the cART regimens, only boosted PI monotherapy was associated with significant differences in vitamin D level. The proportion of patients treated with boosted PI was different in the three groups: 7.6, 3.1 and 14.3% in the vitDd, vitDi and vitDo groups, respectively (P = 0.039).

Table 2. Stratification of vitamin D status by treatment regimen, for patients with viral load ≤ 50 copies/mL
 vitDd (n = 118)VitDi (n = 64)VitDo (n = 91)P-value
  1. NNRTI, nonnucleoside reverse transcriptase inhibitor; PI, protease inhibitor; EFV, efavirenz; NVP, nevirapine; ETV, etravirine; TDF, tenofovir disoproxil fumarate; LPV, lopinavir; DRV, darunavir; ATV, atazanavir; FAMP, fosamprenavir; SQV, saquinavir; TPV, tipranavir; r, ritonavir booster; vitDd, vitamin D deficiency; vitDi, vitamin D insufficiency; vitDo, vitamin D optimal level.
  2. aStatistically significant difference at P < 0.05.
  3. bEFV, 58 patients in total (72.5%); NVP, 22 (27.5%).
  4. cEFV, 14 (42.4%); NVP, 18 (54.5%); ETV, 1 (3.1%).
  5. dLPV/r, 9 (13.2%); DRV/rt, 15 (22%); ATV/r, 25 (36.8%); FAMP/r, 11 (16.2%); SQV/r, 7 (10.3%); TPV/r, 1 (1.5%).
  6. eLPV/r, 8 (21%); DRV/r, 5 (13.2%); ATV/r, 13 (34.2%); FAMP/r, 5 (13.1%); SQV/r, 4 (10.5%).
  7. fLPV/r, 17 (70.8%); DRV/r, 6 (25%); ATV/r, 1 (4.2%).
Duration on ARVs (months) [median (IQR)]85 (40–143)64 (28–138)86 (43–140)0.344
NNRTIb + TDF [n (%)]36 (30.5)23 (35.9)21 (23.1)0.258
NNRTIc + no TDF [n (%)]16 (13.6)8 (12.5)9 (9.9)0.745
PId + TDF [n (%)]28 (23.7)15 (23.4)25 (27.5)0.728
PIe + no TDF [n (%)]14 (11.3)9 (14.1)15 (16.5)0.602
NNRTI + PI + TDF [n (%)]1 (0.8)0 (0)0 (0)0.519
NNRTI + PI + no TDF [n (%)]3 (3)1 (1.6)3 (3.3)0.781
Other + TDF [n (%)]2 (2.5)1 (1.6)0 (0)0.475
Other + no TDF [n (%)]9 (7.6)5 (7.8)5 (5.5)0.821
Boosted PI monotherapyf [n (%)]9 (7.6)2 (3.1)13 (14.3)0.039a
Figure 1.

Vitamin D status stratified by combination antiretroviral therapy (cART) regimen in patients with HIV viral load ≤ 50 copies/mL. NNRTI, nonnucleoside reverse transcriptase inhibitor; PI, protease inhibitor; TDF, tenofovir.

In multivariate analysis, the following variables were related to an increased risk of vitDi or vitDd: Black vs. Caucasian ethnic group (OR 10.6; 95% CI 1.2–94; P = 0.033); heterosexual (OR 2.37; 95% CI 1.13–4.93; P = 0.022) or men who have sex with men (MSM) (OR 3.25; 95% CI 1.25–8.50; P = 0.016) transmission category vs. IDU; and VL > 50 copies/mL (OR 2.56; 95% CI 1.10–7.25; P = 0.04) (Table 4).

Table 3. Univariate analysis of factors related to vitamin D insufficiency or deficiency (< 30 ng/mL)
 OR (95% CI)P
  1. ART, antiretroviral therapy; BMI, body mass index; CI, confidence interval; HBV, hepatitis B virus; HCV, hepatitis C virus; IDU, injecting drug use; iPTH, intact parathyroid hormone; MSM, men who have sex with men; OR, odds ratio.
  2. *Statistically significant difference at P < 0.05.
Age (reference < 50 years)  
≥ 50 years1.16 (0.66–2.01)0.606
Sex (reference male)  
Female0.94 (0.57–1.53)0.793
Ethnic background (reference Caucasian)  
Black12.69 (1.7–94.56)0.013*
Hispanic (South American)0.82 (0.34–2)0.664
HIV risk group (reference IDU)  
Heterosexual2.79 (1.64–4.73)0.0001*
MSM3.42 (1.56–7.5)0.002*
Other3.42 (0.39–30)0.266
Calcium (reference ≥ 8.5 mg/dL)  
< 8.5 mg/dL0.79 (0.23–2.67)0.701
Albumin (serum) (reference ≥ 3.4 g/dL)  
< 3.4 g/dL2.41 (0.29–20.31)0.420
iPTH (reference ≤ 66 pg/mL)  
> 66 pg/mL1.26 (0.73–2.18)0.413
BMI (reference ≥ 18.5)  
< 18.50.69 (0.19–2.39)0.554
HBV positive0.89 (0.27–2.95)0.848
HCV positive0.33 (0.21–0.54)0.0001*
CD4 count (reference ≥ 500 cells/μL)  
< 200 cells/μL2.53 (0.93–6.87)0.069
200–350 cells/μL0.99 (0.53–1.89)0.99
350–500 cells/μL0.82 (0.46–1.45)0.49
Viral load (reference ≤ 50 copies/mL)  
> 50 copies/mL3.59 (1.71–7.52)0.001*
ART (reference not receiving ART)  
Receiving ART0.43 (0.27–0.70)0.001*
Season (reference spring)  
Summer0.018 (0.002–0.137)0.0001*
Autumn0.019 (0.002–0.139)0.0001*
Winter0.250 (0.027–2.296)0.220
Table 4. Multivariate analysis of factors related to vitamin D insufficiency (< 30 ng/mL)
 OR (95% CI)P
  1. ART, antiretroviral therapy; CI, confidence interval; IDU, injecting drug use; MSM, men who have sex with men; NNRTI, nonnucleoside reverse transcriptase inhibitor; OR, odds ratio; PI, protease inhibitor; TDF, tenofovir disoproxil fumarate.
  2. *Statistically significant difference at P < 0.05.
Ethnic background (reference Caucasian)  
Black10.64 (1.20–94)0.013*
Hispanic (South America)0.18 (0.05–0.68)0.012*
HIV risk group (reference IDU)  
Heterosexual2.37 (1.13–4.93)0.022*
MSM3.25 (1.25–8.51)0.016*
Other9.37 (0.54–162)0.124
Viral load (reference ≤ 50 copies/mL)  
> 50 copies/mL2.56 (1.1–7.25)0.040*
Season (reference spring)  
Summer0.015 (0.002–0.116)0.0001*
Autumn0.013 (0.002–0.099)0.0001*
Winter0.257 (0.027–2.443)0.237
ART (reference ART naïve)  
NNRTI + TDF0.35 (0.06–2.18)0.262
NNRTI + no TDF0.19 (0.03–1.31)0.092
PI + TDF0.25 (0.04–1.5)0.129
PI + no TDF0.12 (0.02–0.76)0.025*
NNRTI + PI + TDF21885246 (0.000–)0.999
NNRTI + PI + no TDF0.29 (0.03–2.89)0.291
Other + TDF265624819 (0.000–)0.999
Other + no TDF0.31 (0.04–2.39)0.258
Boosted PI monotherapy0.08 (0.01–0.65)0.018*

The patients with the lowest risk of vitDi or vitDd were those treated with PIs without tenofovir (vs. those not on cART, OR 0.12; 95% CI 0.018–0.76; P = 0.025) or boosted PI monotherapy (vs. those without treatment, OR 0.08; 95% CI 0.01–0.6; P = 0.018), Hispanic South American patients (vs. Caucasians, OR 0.18; 95% CI 0.05–0.68; P = 0.012) and those for whom blood had been drawn in summer (vs. spring, OR 0.015; 95% CI 0.002–0.116; P < 0.0001) or autumn (vs. spring, OR 0.013; 95% CI 0.02–0.099; P < 0.0001).

Discussion

The prevalence of low vitamin D levels was very high in our cohort of HIV-infected patients recruited in Hospital Severo Ochoa in Madrid, Spain. Our ethnically diverse cohort, in which the majority of patients are receiving cART, is probably representative of the present HIV epidemic in Spain. Overall, 44% had vitDd, and vitDi was found in 27.6%. Only 28.4% of patients had vitDo. Our results confirm those of other studies revealing a high prevalence of vitDd or vitDi, ranging from 74 to 91% [5, 6, 12, 13]. In a published study by our group [8], in which we investigated the influence of dietary intake and lifestyle, only hours of sunlight exposure was related to the presence or absence of vitDo (P = 0.045).

Increased amounts of skin pigment reduce the capacity of the skin to synthesize vitamin D3 [21]. We showed that the association of dark skin (Black ethnic background) with suboptimal vitamin D levels was predictive of vitamin vitDd or vitDi in multivariate analysis, in accordance with previous research [5, 12, 13, 22]. The analysis of the effect of ethnicity adjusted for viral load (> or ≤ 50 copies/mL) showed an effect of skin pigmentation on vitamin D levels.

As in previous studies [13, 23], we were unable to show an association between CD4 cell count and vitamin D level. However, there was a relationship between HIV viraemia (> 50 copies/mL) and vitDd. In a recent study, Kim et al. [17] demonstrated the existence of an interaction between HIV viraemia and lipopolysaccharides (LPSs), proinflammatory cytokines and Toll-like receptor (TLR) signalling pathways in macrophages, which induce CYP27B1 activation [1α-hydroxylase of 25(OH)-D].

Early HIV infection of the intestinal wall is accompanied by bacterial translocation and intestinal lymphoid tissue destruction. With HIV viraemia, circulating LPS and cytokine levels are increased. TLR signalling pathways in macrophages are activated by LPS and circulating HIV. Activated macrophages increase the concentrations of proinflammatory cytokines, including tumour necrosis factor and interferon-γ. The latter, inducing CYP27B1 in macrophages and acting synergistically with LPS, reduces 25(OH)-D levels through hydroxylation of 1,25(OH)-D. This induction takes place even in the absence of 25(OH)-D.

Vitamin D supplementation may be considered as adjunctive treatment for hepatitis C, especially in patients who have low vitamin D levels [24]. The paradox of more favourable vitamin D levels in former IDUs and in patients with hepatitis C (the main risk factor for which is drug use) could be explained by more intense sunlight exposure as a result of having outdoor jobs or more time spent in the open. We did not determine whether the patients who had hepatitis C with vitDo had better control of their disease (lower HCV viral load), this not being the objective of this study.

The low levels of vitamin D in samples drawn in spring and winter in our cohort confirm the low efficiency of UV light in the generation of vitamin D above latitude 35° N from December to March. Studies in other cohorts, such the Swiss Cohort, have obtained similar findings [5].

Both in multivariate analysis and in the analysis restricted to those with viral loads ≤ 50 copies/mL we found that boosted PI monotherapy was significantly associated with an increased percentage of patients with vitDo. We did not find an association with NRTIs, probably because the combination with other antiretroviral drugs partially offsets the well-known effect of efavirenz on vitamin D catabolism, mediated by the induction of CYP3A4 and CYP24 [12]. In the Swiss Cohort, Mueller et al. [5] showed that patients treated with tenofovir had increased levels of 1,25(OH)-D, as a consequence of the activation of 1α-hydroxylase by a renal autoregulation process linked to phosphate loss through the kidney. Tenofovir causes secondary hyperparathyroidism, especially with vitDi [22]. The effect of the absence of this tenofovir-associated hyperparathyroidism could explain why patients on treatment with boosted PI monotherapy or with regimens in which PIs were combined with NRTIs other than tenofovir had higher levels of vitamin D. In addition, through CYP3A4 or CYP24 inhibition, PIs may inhibit hydroxylation of 25(OH)-D to 1,25(OH)-D and increase vitamin D levels [25].

Our study has several limitations. The most important limitation is a consequence of its design: it was a cross-sectional, retrospective study, so we could not determine the causality of the results. We could not investigate whether PIs increase vitamin D levels with time, as our vitamin D measurements were obtained at a single time-point for each patient. Although seasonality may interfere with vitamin D levels, the results were unlikely to have been biased by seasonal variations, as 60.5% of the samples were drawn in summer or autumn during the study period. Use of cART was not informed by knowledge of the patient's vitamin D level, and so no bias can have arisen from the selection of cART. Higher levels of vitamin D in patients under boosted PI monotherapy were not due to better virological control; we found no significant differences in viral load when we compared boosted PI monotherapy with other regimens (P = 0.078).

In conclusion, our data show an increased risk of vitamin D deficiency or insufficiency in HIV-infected out-patients with detectable viral load and a Black ethnic background. Among cART regimens, boosted PI monotherapy was associated with a lower risk of vitamin D deficiency or insufficiency. The more favourable vitamin D status in former IDUs was probably attributable to a higher frequency of outdoor jobs in this group of patients.

In a previous study [8], we investigated the influence of dietary intake and lifestyle on vitamin D status, and found that only hours of sunlight exposure was related to the presence or absence of an optimal vitamin D level. Smoking status was not associated with vitamin D level (P = 0.094).

We could not determine whether patients who had hepatitis C and optimal vitamin D levels had better control of their disease (lower HCV viral load), nor we could establish the relationship between vitamin D status and BMD, as these were not objectives of this study. In future investigations, we will investigate the effect of supplemental vitamin D on HCV viral load and BMD.

More studies with appropriate designs are required to show that antiretroviral therapy based on boosted PIs is associated with higher vitamin D levels. There is only a single study of the effect of switching regimens, the MONET trial (Monotherapy in Europe with Tmc114) [16], in which switching from efavirenz and/or zidovudine to darunavir/ritonavir during the trial led to an increase in vitamin D levels.

Funding

The authors received no funding for this work.

Ancillary