High prevalence of severe vitamin D deficiency in combined antiretroviral therapy-naive and successfully treated Swiss HIV patients
"we suggest systematic screening for vitamin D deficiency in all HIV-positive patients. Seasonal changes thereby have to be taken into account.....vitamin D deficiency was three times more frequent during springtime than in fall. Even in fall, after peak sun exposure, only 23-24% of the patients reached the target value of 75 nmol/l or more and only 5-10% did so in spring.....In multivariable analysis, white ethnicity and time since HIV diagnosis were significantly associated with higher 25(OH)D, whereas spring measurements, active IDU and NNRTI-based cART correlated with lower levels (Table 2). With NNRTI use, 25(OH)D levels significantly decreased from 46 [interquartile range (IQR) 30-67] to 39 (IQR 21-60) nmol/l (P = 0.04), whereas a nonsignificant increase from 42 (IQR 25-63) to 43 (IQR 28-66) nmol/l was seen in protease inhibitor-treated patients.|
With 42-52% in spring and 14-18% in fall, vitamin D deficiency was highly prevalent in our HIV-positive patients. This prevalence is higher than the 10.5 or 14% reported from Boston or Sydney [18,20], but compares well with the 29% value described in the Netherlands . In order to reach the target level of 75 nmol/l or higher for maximal vitamin D benefits, as many as 76-95% of the patients would qualify for vitamin D supplementation. Again, such high numbers have been reported earlier [22,23]. From recent assessments of vitamin D levels in western countries, it can be assumed that the prevalence of vitamin D insufficiency and deficiency is high in the general population . A comparison with deficiency rates in the general Swiss population is difficult, as recent data are missing. However, the rate of vitamin D deficiency in our HIV-positive cohort was clearly higher than the 6-15% rate found in the only assessment in the general Swiss population during winter . Consistent with earlier reports , non-white ethnicities with darker skin pigmentation, in particular black patients, were most strongly affected by vitamin D deficiency.
In the general population, supplementation with a minimum of 700-800 IU of 25(OH)D per day has been shown to reduce the risk for hip and nonvertebral fractures [5,8]. Alternatively, 100 000 IU of oral 25(OH)D every 4 months or 300 000 IU every 12 months might be used . The role of vitamin D supplementation in HIV-positive patients, and specific treatment goals in particular, must be defined by prospective studies with HIV-specific endpoints such as disease progression or change in CD4 cell counts. Also, the impact of NNRTIs on 25(OH)D and TDF on 1,25(OH)2D needs further consideration."
15 May 2010
Mueller, Nicolas J; Fux, Christoph A; Ledergerber, Bruno; Elzi, Luigia; Schmid, Patrick; Dang, Thanh; Magenta, Lorenzo; Calmy, Alexandra; Vergopoulos, Athanasios; Bischoff-Ferrari, Heike A; the Swiss HIV Cohort Study
aDivision of Infectious Diseases and Hospital Epidemiology, University Hospital of Zurich, Zurich, Switzerland
bUniversity Clinic for Infectious Diseases, Inselspital and University of Bern, Bern, Switzerland
cDivision of Infectious Diseases and Hospital Epidemiology, University Hospital Basel, Basel, Switzerland
dDivision of Infectious Diseases, Department of Internal Medicine, Cantonal Hospital St. Gallen, St. Gallen, Switzerland
eInfectious Diseases Service, Centre Hospitalier Universitaire Vaudois and University of Lausanne, Lausanne, Switzerland
fDivision of Infectious Diseases, Hospital of Lugano, Lugano, Switzerland
gDivision of Infectious Diseases, Geneva University Hospital, Geneva, Switzerland
hInstitute of Clinical Chemistry, University Hospital Zurich, Switzerland
iCentre on Aging and Mobility, University of Zurich, Switzerland
jDepartment of Rheumatology and Institute of Physical Medicine, University Hospital Zurich, Zurich, Switzerland.
Objectives: To evaluate the prevalence of 25-hydroxyvitamin D [25(OH)D] deficiency in HIV-positive patients, a population at risk for osteoporosis.
Design: Retrospective assessment of vitamin D levels by season and initiation of combined antiretroviral therapy (cART).
Methods: 25(OH)D was measured in 211 HIV-positive patients: samples were taken before initiation of cART from February to April or from August to October as well as 12 (same season) and 18 months (alternate season) after starting cART. 1,25-Dihydroxyvitamin D [1,25(OH)2D] was measured in a subset of 74 patients. Multivariable analyses included season, sex, age, ethnicity, BMI, intravenous drug use (IDU), renal function, time since HIV diagnosis, previous AIDS, CD4 cell count and cART, in particular nonnucleoside reverse transcriptase inhibitor (NNRTI) and tenofovir (TDF) use.
Results: At baseline, median 25(OH)D levels were 37 (interquartile range 20-49) nmol/l in spring and 57 (39-74) nmol/l in the fall; 25(OH)D deficiency less than 30 nmol/l was more prevalent in spring (42%) than in fall (14%), but remained unchanged regardless of cART exposure. In multivariable analysis, 25(OH)D levels were higher in white patients and those with a longer time since HIV diagnosis and lower in springtime measurements and in those with active IDU and NNRTI use. 1-Hydroxylation rates were significantly higher in patients with low 25(OH)D. Hepatitis C seropositivity, previous AIDS and higher CD4 cell counts correlated with lower 1,25(OH)2D levels, whereas BMI and TDF use were associated with higher levels. In TDF-treated patients, higher 1,25(OH)2D correlated with increases in serum alkaline phosphatase.
Conclusion: Based on the high rate of vitamin D deficiency in HIV-positive patients, systematic screening with consideration of seasonality is warranted. The impact of NNRTIs on 25(OH)D and TDF on 1,25(OH)2D needs further attention.
Recent studies have increased the awareness of the beneficial effects of vitamin D. In addition to being essential for bone growth  and preservation , vitamin D is antineoplastic by controlling cell proliferation, differentiation and apoptosis  and counteracts arterial hypertension by suppressing renin synthesis . Also, vitamin D expresses immunmodulatory effects with receptors being found on monocytes, macrophages as well as B lymphocytes and T lymphocytes . For example, vitamin D supplementation has been shown to improve CD4 T-cell counts in HIV-positive individuals .
A vitamin D precursor is produced under the influence of Ultraviolet B (UVB) light in the skin, then 25-hydroxylated to stable 25-hydroxyvitamin D [25(OH)D] in the liver. 1-Hydroxylation to an active form, 1,25-dihydroxyvitamin D [1,25(OH)2D], occurs in the kidney and numerous other target organs. For lack of UVB light, little if any vitamin D can be produced above a 35° northern latitude limit between November and February. Several investigators have defined 75 nmol/l as the minimal target serum concentration of 25(OH)D for optimal bone density  as well as fracture and fall prevention in different age groups and across various populations [8-11]. Unfortunately, this target level is only reached in about a third of healthy, non-HIV-infected adult individuals .
Whereas successful combined antiretroviral therapy (cART) has led to an unprecedented decrease in HIV-related mortality, noninfectious diseases associated with older age, such as cardiovascular and neoplastic diseases as well as osteoporosis, gain importance. Because many of these diseases are modulated by vitamin D, we believe that the minimal vitamin D target levels suggested for the general population should also be met by HIV-positive patients, although the effect of vitamin D has been poorly evaluated in HIV-infected individuals and no target level has been validated. Therefore, we defined 75 nmol/l as minimal target level for 25(OH)D while defining values less than 30 nmol/l as vitamin D deficiency.
Among HIV-positive patients, up to 60% of osteopenia and 10-15% of osteoporosis have been documented [13-15]. While chronic inflammation associated with viral replication is believed to promote osteoclast activity in untreated patients, the impact of cART and of specific drugs remains to be clarified . In particular, protease inhibitors [14,15] and tenofovir (TDF)  have been associated with reductions in bone mineral density  and altered vitamin D metabolism .
We determined serum vitamin D levels in HIV-positive patients comparing measurements after peak (fall) and nadir (spring) sun exposure as well as before and after the initiation of cART and compared these levels with the minimal desirable levels recommended . We evaluated the effects of ethnicity as well as the chosen cART regimen [protease inhibitor, nonnucleoside reverse transcriptase inhibitor (NNRTI) and TDF use] and looked for predictors for vitamin D deficiency. Furthermore, we measured serum alkaline phosphatase (sAP) as a surrogate marker of osteoblast activity.
With 42-52% in spring and 14-18% in fall, vitamin D deficiency was highly prevalent in our HIV-positive patients. This prevalence is higher than the 10.5 or 14% reported from Boston or Sydney [18,20], but compares well with the 29% value described in the Netherlands . In order to reach the target level of 75 nmol/l or higher for maximal vitamin D benefits, as many as 76-95% of the patients would qualify for vitamin D supplementation. Again, such high numbers have been reported earlier [22,23]. From recent assessments of vitamin D levels in western countries, it can be assumed that the prevalence of vitamin D insufficiency and deficiency is high in the general population . A comparison with deficiency rates in the general Swiss population is difficult, as recent data are missing. However, the rate of vitamin D deficiency in our HIV-positive cohort was clearly higher than the 6-15% rate found in the only assessment in the general Swiss population during winter . Consistent with earlier reports , non-white ethnicities with darker skin pigmentation, in particular black patients, were most strongly affected by vitamin D deficiency. The reasons for the negative correlation with active IDU are unknown, both malnutrition and limited sunlight exposure may play a role. The positive correlation with the time since HIV diagnosis may be related to a generally poorer health condition of patients who have been recently diagnosed. The observed trend of lower 25(OH)D levels in individuals with higher BMI is probably caused by vitamin D sequestration in body fat . However, this association must be interpreted with caution, as body weight was normal in almost all patients.
Although cART exposure had no significant impact on vitamin D levels, NNRTI use was associated with significantly lower 25(OH)D levels. This correlation, with a consecutively increased risk of developing hyperparathyroidism, has been documented earlier  and may be caused by an increased catabolism of 25(OH)D through induction of CYP450 by NNRTIs. Consistently, a case report has described severe 25(OH)D deficiency after starting efavirenz .
Neither NNRTI nor protease inhibitor medication showed a significant effect on 1,25(OH)2D levels in our analyses, although protease inhibitor-treated patients had shown lower 1,25(OH)2D values in an earlier study . Many enzymes of the cytochrome P450 cytochrome family, which are known to interact with NNRTIs and/or protease inhibitors, are involved in vitamin D metabolism; for example, 25-hydroxylation in the liver (CYP27A1, CYP2R1 or CYP3A4), 1-hydroxylation in the kidney (CYP27B1) or the catabolism of activated vitamin D (CYP3A4 rather than CYP24) . In vitro, inhibitory effects of ritonavir, indinavir and nelfinavir on CYP450 enzymes resulted in a net reduction of activated vitamin D, suggesting that protease inhibitor use could contribute to bone demineralization .
Further clinical data are needed to evaluate the clinical net effect of NNRTIs and protease inhibitors on bone metabolism in general and on vitamin D in particular.
In clinical routine, vitamin D assessment is based on 25(OH)D measurements, because of its stability with a 2-week half-life in blood circulation . At low 25(OH)D levels, parathyroid hormone (PTH) stimulates the tubular reabsorption of calcium, phosphaturia and the 1-hydroxylation of 25(OH)D. This is consistent with our finding of a logarithmic increase in the 1-hydroxylation ratio with decreasing 25(OH)D levels, particularly with levels below 30 nmol/l . In our patients, this compensatory increase in 1-hydroxylation partially compensated the seasonal differences in 25(OH)D. Our data underline that 1,25(OH)2D should not be used to diagnose vitamin D deficiency, as normal values may hide both 25(OH)D deficiency and secondary hyperparathyroidism.
Surprisingly, a history of AIDS-defining events was associated with lower 1,25(OH)2D, but not with lower 25(OH)D levels, which would have suggested impaired nutritional supply and/or lower sun exposure. AIDS may result in an inflammation-related impairment of 1-hydroxylation. As a matter of fact, earlier studies have reported low 1,25(OH)2D levels in patients with advanced HIV infection and immunologic hyperactivation [31,32]. The significant negative correlation of hepatitis C seropositivity with 1,25(OH)2D, but not with 25(OH)D, found in our analysis might be related to an inflammation-related impairment of 1-hydroxylation due to replicating hepatitis. However, the hypothesis of an inflammation-related impairment of 1-hydroxylation is contradicted by the unexplained negative correlation of 1,25(OH)2D with CD4 cell counts.
TDF-use correlated with both higher serum 1,25(OH)2D and serum alkaline phosphatase levels indicating osteoblast activation . This pattern was not seen in protease inhibitor-treated or zidovudine-treated patients (data not shown). Importantly, higher 1,25(OH)2D levels were not due to higher 25(OH)D, but to an increased rate of 1-hydroxylation. 1-Hydroxylation is not only stimulated by PTH in response to low calcium and 25(OH)D levels but also, possibly mediated through phosphatonins [33,34], by hypophosphatemia . Several publications have documented TDF-related proximal renal tubulopathy with excessive phosphate losses [36,37], but also decreasing bone mineral density . The stimulated 1-hydroxylation observed in our TDF-treated patients may represent an autoregulatory process to compensate for drug-related phosphate losses by the stimulation of intestinal phosphate resorption and/or mobilization of bone phosphate through higher 1,25(OH)2D levels. Stimulating intestinal phosphate resorption may thereby be protective, whereas mobilizing bone phosphate may promote osteopenia and osteoporosis in these individuals.
This study has some limitations. The population studied was homogenous (mostly men and white) and relatively healthy. Its retrospective nature prevented broader analyses, such as the evaluation of calcium and phosphate metabolism, and no conclusions on the potential benefit of supplementation can be drawn. The stringent inclusion criteria including the requirement of availability of samples, CD4 cell measurements at all three time points as well as documentation of successful cART by a HIV viral load less than 50 copies/ml resulted in a relatively small overall sample size. Therefore, documented associations will need confirmation by independent studies. Given the numerous HIV-independent risk factors for vitamin D deficiency in our patients, it is almost impossible to establish a HIV-negative control group to assess the influence of HIV on vitamin D levels. On the contrary, we believe that our study design comparing patients before and after cART initiation is most suitable to elucidate the impact of viral control and antiretroviral drug classes.
On the basis of our prevalence data and the growing recognition of vitamin D-related health benefits, we suggest systematic screening for vitamin D deficiency in all HIV-positive patients. Seasonal changes thereby have to be taken into account. If many benefits of vitamin D supplementation have already been documented in the general population, an adequate supply may be even more important in HIV-infected individuals with their additional risk factors for osteopenia, neoplasias or cardiovascular disease. In the general population, supplementation with a minimum of 700-800 IU of 25(OH)D per day has been shown to reduce the risk for hip and nonvertebral fractures [5,8]. Alternatively, 100 000 IU of oral 25(OH)D every 4 months or 300 000 IU every 12 months might be used . The role of vitamin D supplementation in HIV-positive patients, and specific treatment goals in particular, must be defined by prospective studies with HIV-specific endpoints such as disease progression or change in CD4 cell counts. Also, the impact of NNRTIs on 25(OH)D and TDF on 1,25(OH)2D needs further consideration.
Baseline measurements were performed between February and April (spring) for 103 and between August and October (fall) for 108 patients. Participants were 158 (75%) men, 185 (88%) whites, with a median age of 37 (range 32-45) years, BMI of 23 (21-25) kg/m2 and cGFR of 107 (94-122) ml/min. The presumed mode of HIV transmission was homosexual in 85 (40%), heterosexual in 96 (45%) and IDU in 30 (14%) cases. The median CD4 cell count was 226 (135-333) cells/µ l; 30 (14%) had a previous diagnosis of AIDS. Baseline characteristics of the 103 cases with first measurement in spring and the 108 cases with first measurement in fall did not differ by season. The initiated cART included TDF in 35 (17%), NNRTI in 91 (43%) and protease inhibitors in 118 (56%).
25-Hydroxyvitamin D status
Figure 1a illustrates median 25(OH)D levels at baseline before cART initiation as well as after 12 months (same season as baseline but with cART) and 18 months (opposite extreme season with cART). At each time point, 25(OH)D levels were significantly higher in fall than in spring (P < 0.001), without significant change with respect to cART. Table 1 shows the relative distribution of vitamin D deficiency and insufficiency for each time point. With 42-52%, vitamin D deficiency was three times more frequent during springtime than in fall. Even in fall, after peak sun exposure, only 23-24% of the patients reached the target value of 75 nmol/l or more and only 5-10% did so in spring.
Ethnicity was a significant predictor for serum 25(OH)D values (P < 0.001). In both seasons, whites had a lower prevalence of vitamin D deficiency and insufficiency (Fig. 1b). All ethnicities but black, showed significantly higher 25(OH)D levels in fall compared with spring measurements.
In multivariable analysis, white ethnicity and time since HIV diagnosis were significantly associated with higher 25(OH)D, whereas spring measurements, active IDU and NNRTI-based cART correlated with lower levels (Table 2). With NNRTI use, 25(OH)D levels significantly decreased from 46 [interquartile range (IQR) 30-67] to 39 (IQR 21-60) nmol/l (P = 0.04), whereas a nonsignificant increase from 42 (IQR 25-63) to 43 (IQR 28-66) nmol/l was seen in protease inhibitor-treated patients.
1,25-Dihydroxyvitamin D status
1,25(OH)2D was measured in a subgroup of 74 patients. Figure 2 correlates the 1,25(OH)2D-to-25(OH)D ratio with 25(OH)D levels. Black circles indicate measurements in fall, white circles in spring. The exponential increase in renal 1-alpha-hydroxylation with decreasing 25(OH)D levels resulted in a less pronounced seasonal swing for 1,25(OH)2D (minus 3-13%) than for 25(OH)D (minus 36-43%) (Table 3).
Still, median 1,25(OH)2D values remained lower in patients with 25(OH)D deficiency with 28 (IQR 21-36) ng/l and 25(OH)D insufficiency with 37 (25-48) ng/l as compared with 54 (37-72) ng/l in individuals with 25(OH)D values 75 nmol/l or more (P < 0.001).
Univariable and multivariable analyses of factors influencing 1,25(OH)2D levels revealed a distinct pattern from 25(OH)D (Table 2). While season and ethnicity were not associated with 1,25(OH)2D levels, we observed a negative correlation with hepatitis C seropositivity. On the contrary, previous AIDS and higher CD4 cell counts were associated with low levels and BMI with high 1,25(OH)2D levels. Although use of NNRTI showed no significant correlation in multivariable analysis, use of TDF correlated with higher 1,25(OH)2D. Consistently, the 1-hydroxylation rates in TDF-treated patients were 25% higher (P = 0.002) in the absence of 25(OH)D deficiency, that is at values 30 nmol/l or higher. In vitamin D-deficient patients, however, use of TDF did not stimulate the 1,25(OH)2D-to-25(OH)D ratio any further (P = 0.6).
Calculated glomerular filtration rate and serum alkaline phosphatase
With a median baseline calculated glomerular filtration rate (cGFR) of 109 ml/min (IQR 93-124), all but one patient had cGFR values higher than 30 ml/min and, therefore, no structural reason for a 1-hydroxylation deficiency in the kidney. Neither TDF-containing nor TDF-sparing cART significantly changed cGFR compared with baseline (data not shown).
Use of TDF was associated with a significant increase in sAP from a median of 69 (IQR 55-84) to 92 (70-116) U/l (P = 0.01) as compared with 65 (55-86) to 79 (65-98) U/l (P = 0.5) in patients on a TDF-sparing regimen and 70 (58-93) to 80 (67-98) U/l in patients on protease inhibitors (P = 0.8). Regression analysis provided a correlation coefficient of 6.6 (1.0-12.2; P = 0.02) between higher 1-hydroxylation and increases in sAP between baseline and the 1-year follow-up in patients exposed to TDF.
The Swiss HIV cohort study (SHCS, www.shcs.ch) includes data on more than 15 600 patients. At 6-monthly intervals, clinical data, plasma and cell samples, laboratory values and information on cART are collected with written informed consent by all participants and ethical committee approval.
For this study, 211 patients were identified with available stored plasma samples (-80° C) for three predefined time points: baseline, just before cART initiation during spring or fall; 12 months (± 1 month) after cART initiation; 18 months (± 1 month) after cART initiation, that is in the opposite extreme season compared with the prior time points. For inclusion, plasma viral load had to be less than 50 copies/ml 6 months before the second visit and in subsequent visits. One hundred and three patients had their baseline measurement between February and April (spring) and 108 patients between August and October (fall).
Patient demographics, HIV-related parameters including time since HIV diagnosis, CDC stage, CD4 cell counts, the cART regimen, calculated glomerular filtration rates (cGFRs) by Cockcroft-Gault formula and sAP values were extracted from the electronic database. No data on diet or vitamin D supplementation are routinely collected.
Measurement of 25-hydroxyvitamin D and 1,25-dihydroxyvitamin D
25(OH)D and 1,25(OH)2D were measured in EDTA (39.5%) or sodium citrate plasma (60.5%) by radioimmunoassay (RIA), DiaSorin and ImmunoDiagnostic Systems, respectively. For sodium citrate plasma, a dilution factor was determined in a separate validation study and used to adjust those samples. 1,25(OH)2D was measured in a subset of 74 patients, including 35 individuals with TDF as part of cART and 39 randomly selected patients without TDF.
The following cut-offs were defined for 25(OH)D: deficiency, less than 30 nmol/l; insufficiency, 30-74 nmol/l; target level, 75 nmol/l or more. For crude comparisons, we used t-tests for continuous and χ 2-tests for discrete variables. We used univariable and multivariable regression models for analyzing predictors of 25(OH)D and 1,25(OH)2D status, including the following parameters: age, sex, ethnicity, BMI, season of measurements (fall versus spring), cGFR, time since HIV diagnosis, active intravenous drug use (IDU), hepatitis C seropositivity, previous AIDS, current CD4 cell count and TDF and NNRTI comedication. There were a few HIV patients with missing BMI. In order to include all observations in the multivariable analysis, these were replaced by the mean BMI of all available measurements. Excluding HIV patients with missing BMI measurements did not change our findings.
All analyses were conducted with STATA 10 software (Stata Corp., College Station, Texas, USA). All P are two-sided.