Vitamin D as Supplementary Treatment for Tuberculosis A Double-blind, Randomized, Placebo-controlled Trial Christian Wejse1,2, Victor F. Gomes1, Paulo Rabna1, Per Gustafson1,3, Peter Aaby1, Ida M. Lisse4, Paul L. Andersen2, Henning Glerup5, and Morten Sodemann1,6 1 Bandim Health Project, INDEPTH Network, Statens Serum Institute, Bissau, Guinea-Bissau; 2Infectious Disease Research Unit, Aarhus University Hospital, Skejby, Denmark; 3Infectious Diseases Research Group, Department of Clinical Sciences, Lund University, Malmo ¨ , Sweden; 4Department of 5 Pathology, Herlev University Hospital, Copenhagen, Denmark; Department of Internal Medicine, Aarhus University Hospital, Silkeborg, Denmark; and 6Department of Infectious Diseases, Odense University Hospital, Odense, Denmark

Rationale: Vitamin D has been shown to be involved in the host immune response toward Mycobacterium tuberculosis. Objectives: To test whether vitamin D supplementation of patients with tuberculosis (TB) improved clinical outcome and reduced mortality. Methods: We conducted a randomized, double-blind, placebocontrolled trial in TB clinics at a demographic surveillance site in Guinea-Bissau. We included 365 adult patients with TB starting antituberculosis treatment; 281 completed the 12-month followup. The intervention was 100,000 IU of cholecalciferol or placebo at inclusion and again 5 and 8 months after the start of treatment. Measurements and Main Results: The primary outcome was reduction in a clinical severity score (TBscore) for all patients with pulmonary TB. The secondary outcome was 12-month mortality. No serious adverse effects were reported; mild hypercalcemia was rare and present in both arms. Reduction in TBscore and sputum smear conversion rates did not differ among patients treated with vitamin D or placebo. Overall mortality was 15% (54 of 365) at 1 year of follow-up and similar in both arms (30 of 187 for vitamin D treated and 24 of 178 for placebo; relative risk, 1.19 [0.58–1.95]). HIV infection was seen in 36% (131 of 359): 21% (76 of 359) HIV-1, 10% (36 of 359) HIV-2, and 5% (19 of 357) HIV-112. Conclusions: Vitamin D does not improve clinical outcome among patients with TB and the trial showed no overall effect on mortality in patients with TB; it is possible that the dose used was insufficient. Clinical trial registered with www.controlled-trials.com/isrctn (ISRCTN35212132). Keywords: vitamin D; tuberculosis; randomized clinical trial; clinical score; HIV

Vitamin D was used for treatment of tuberculosis (TB) in the preantibiotic era (1–3) and before then cod liver oil, rich in vitamin D, was used as well as sun exposure (4). Vitamin D has been attributed an important role in host immune defense against Mycobacterium tuberculosis (5, 6) and observational studies have found evidence of an association with vitamin D deficiency (VDD) and active tuberculosis (7–10). Vitamin D is a low-cost

(Received in original form April 16, 2008; accepted in final form January 23, 2009) Supported by Aarhus University Hospital (Ph.D. scholarship); the Danish Research Council for Developmental Research; the SSAC; and the Skejby Hospital, Segel, and Beckett Foundations. The funding sources and the provider of cholecalciferol had no role in study design, data collection, data analysis, data interpretation, or writing of the report. The principal investigator (corresponding author) had full access to all the data and had final responsibility for the decision to submit for publication. Correspondence and requests for reprints should be addressed to Christian Wejse, M.D., Ph.D., Infectious Disease Research Unit, Aarhus University Hospital, Skejby, Brendstrupgaardsvej, 8200 Aarhus N, Denmark. E-mail: wejse@ dadlnet.dk Am J Respir Crit Care Med Vol 179. pp 843–850, 2009 Originally Published in Press as DOI: 10.1164/rccm.200804-567OC on January 29, 2009 Internet address: www.atsjournals.org

AT A GLANCE COMMENTARY Scientific Knowledge on the Subject

Vitamin D insufficiency is associated with impaired immune function and increased risk of active tuberculosis (TB). Vitamin D has been used in the preantibiotic era in the treatment of TB. It has also been suggested as a supplementary/prophylactic treatment for TB, but only two small trials have assessed this. What This Study Adds to the Field

Vitamin D can be given safely to patients with TB. Our study suggests no overall effect on clinical outcome or mortality with the doses used. The study raises the question of whether vitamin D has a differential effect depending on immune status.

intervention that is easy to administer in resource-poor settings, and has been suggested as prophylaxis in TB household contacts (11) as well as in treatment of bacterial infections (12). A possible mechanism of vitamin D–mediated effect on TB treatment efficacy is that 1,25-dihydroxyvitamin D has been shown to induce antimycobacterial activity in macrophages in vitro, to upregulate protective innate host responses, and to trigger antimicrobial peptides such as cathelicidin (5). No data are available on dosedependent effects of vitamin D on mycobacterial activity, but it has been speculated that pharmacologic doses of vitamin D may elevate serum 25-hydroxyvitamin D [25(OH)D] concentrations to levels saturating the ability of vitamin D–binding protein to bind vitamin D metabolites, leading to an increase in biological availability of 1,25-dihydroxyvitamin D in infected tissues (13). Two small randomized studies (14, 15) have suggested beneficial effects on weight gain and time to sputum conversion without an initial screening for VDD. One review concluded there is evidence that vitamin D modulates antimycobacterial immunity and called for double-blind, randomized, placebo-controlled trials in patients with active TB (13). We aimed to test whether vitamin D supplementation could improve clinical response to treatment or reduce mortality among patients with TB. Vitamin A is known to be a beneficial intervention to children irrespective of vitamin A status (16), and therefore we hypothesized that vitamin D, regardless of initial vitamin D status, would improve immunologic capacities against M. tuberculosis. We report the findings from a community-based trial (the VDTB Trial) testing vitamin D supplementation for patients with TB in Guinea-Bissau. We first reported these results as a poster at the Royal Society of Hygiene and Tropical Medicine Centennial Conference in London, September 2007 (17).

844

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 179

METHODS Study Site The Bandim Health Project, a disease surveillance site located in Guinea-Bissau, West Africa. A poor, urban population of 92,000 is under continued epidemiologic surveillance, and we have previously reported a high TB incidence of 470 per 100,000 in this area (18).

Study Design We conducted a randomized, double-blind, placebo-controlled trial. Trial inclusion criteria included either a diagnosis of TB by sputum examination (smear microscopy; no culture was available) or by World Health Organization (WHO, Geneva, Switzerland) clinical criteria (19), age 15 years or more, and residence in the study area. There were no exclusion criteria.

Patients Field assistants daily identified new patients with TB initiating tuberculosis chemotherapy at the three health centers and at the national TB hospital situated in the study area, inviting patients to be included in the trial the next day. All received antituberculosis treatment consisting of 2 months of daily observed treatment with ethambutol (E), isoniazid (H), rifampicin (R) and pyrazinamide (Z) followed by 6 months of H1E collected by the patient twice per month. Adherence to daily observed treatment was noted daily by the nurse at the health center; patients were referred to sputum examinations after 2, 4, and 6 weeks of treatment in addition to the regular sputum examinations provided by the national tuberculosis program at 2, 5, and 8 months of treatment.

Enrollment Procedures Patients were randomized to either 100,000 IU of cholecalciferol or identical placebo ampoules at inclusion. Cholecalciferol and placebo (vegetable oil without cholecalciferol) were given in ampoules with drinkable content. At inclusion patients received 100,000 IU of cholecalciferol or placebo, and this was repeated 5 and 8 months after inclusion. Hence patients completing treatment received in total 300,000 IU of cholecalciferol or three placebo doses. The rationale behind the choice of dosage and time points was a concern that vitamin D might induce hypercalcemia, and therefore no intervention was given when patients were seen at 2 months; instead, samples were collected for calcium measurement and hypercalcemia was assessed at the first interim analysis. We chose administration of a few large doses so as not to further increase the substantial pill burden on the patients and because it was a simple intervention applicable to low-resource settings. A single large dose has been shown as effective as daily administration, and the half-life is 2 months (20). The dosage was known to be efficient for treatment of vitamin D deficiency and also safe to give with no deficiency present and even in pregnancy (21).

Follow-up Patients were invited to clinical examinations after 2, 5, and 8 months of treatment, followed by a household visit 12 months after initiation of treatment, or until death or moving out of the study area.

Randomization Procedures The random allocation sequence was computer generated; a list of continuous study numbers was generated with a random allocation to treatment 1 or 2. Study numbers were consecutive and given to patients by the field assistant at inclusion, and patients were recorded in a book with prewritten study numbers and allocation sequence numbers 1 or 2. Study medicine was provided in identical containers labeled lot 204 (allocation sequence number 1) or lot 205 (allocation sequence number 2). A physician gave the trial information, obtained patient’s consent, and conducted the clinical examination; a trial nurse administered study medicine according to sequence number.

Blinding Patients, staff, and researchers assessing outcome were blinded. Trial medicine was available in two lots and for logistical reasons it was not concealed whether a patient was on lot 1 or lot 2 during the trial.

2009

Blinding was not assessed among patients, but we assessed blinding among 10 field assistants and investigators through tasting both lots. Five voted in favor of lot 204 being vitamin D, five voted in favor of lot 205. The randomization code was broken by the primary investigator in December 2006 on completion of data analysis.

Adverse Effects We questioned the patients for the following adverse effects related to hypercalcemia: nausea, vomiting, excessive thirst, anorexia, symptoms of kidney stones, and confusion. We measured calcium concentrations in 120 patients seen at 2 months, and in 120 patients completing 8 months of treatment. After the first interim analysis of calcium concentrations in control samples taken at 2 months, we proceeded to give vitamin D at 5 and 8 months.

Measurements To determine the body mass index, we used the following formula: body mass index 5 weight/(height)2. Height was measured with a meter scale; weight was measured in kilograms, using the same weight scale at each patient visit. Mid–upper arm circumference was measured at the midpoint between the acromion and olecranon over the biceps of the nondominant arm, using a nonstretchable measuring tape (TALC, Guilford, UK) to the nearest 0.2 cm (22). Severity of TB disease was assessed by the TBscore, which counts signs, symptoms, and anthropometry as described in OUTCOME. The TBscore has been validated in another cohort and has been grouped in severity classes as follows: I (0–5 points), II (6 or 7 points), or III (8 points or more) (23). We allocated patients into HIV-infected and HIV-uninfected groups, but there was a high frequency of HIV-2-infected patients as shown in Table 1. Despite milder clinical progression and lower mortality in HIV2–infected than HIV-1–infected individuals (24, 25), HIV-2 patients were more comparable to HIV-1 patients than to uninfected patients: 79% (22 of 28) of HIV-2 patients had a CD41 T-lymphocyte count less than 500 cells/ml and 12-month mortality was 17% (6 of 36). We measured serum 25-hydroxy metabolites of vitamins D2 and D3 [25(OH)D213] simultaneously by isotope-dilution liquid chromatography– tandem mass spectrometry on an API 3000 mass spectrometer (Applied Biosystems, Foster City, CA), according to a method adapted from Maunsell and colleagues (26). The inter- and intraassay coefficients of variation were 9.4 and 9.7%. We defined VDD as serum 25(OH)D3 not exceeding 50 nmol/L and vitamin D insufficiency (VDI) as 25(OH)D3 not exceeding 75 nmol/L, according to Vieth (27). Serum calcium and albumin were measured by absorbance (COBAS Integra; Roche Diagnostics, Mannheim, Germany). We corrected total serum calcium for individual variations in albumin by the following equation: adjusted serum calcium (mmol/L) 5 serum calcium total (mmol/L) 3 0.00086 3 (650 2 serum albumin [mmol/L]). T-lymphocyte subsets were determined at the National Public Health Laboratory, Bissau, by flow cytometry (FACStrak; Becton Dickinson, San Jose, CA) with the use of three two-color immunofluorescence reagents, CD45/CD14, CD3/CD4, and CD3/CD8 (Simultest; Becton Dickinson). Leukocyte and differential counts were performed manually. Because of technical problems on several occasions CD41 T-lymphocyte counts were available only from 276 patients at inclusion and from 187 patients at 8 months.

Sample Size To show the safety of vitamin D supplementation in patients with TB, we investigated whether vitamin D increased serum calcium concentrations significantly. A significant change in serum calcium was defined as albumin-corrected serum calcium increasing to a concentration greater than 2.75 mmol/L in a patient with no hypercalcemia before vitamin D supplementation. Estimating no hypercalcemia before vitamin D supplementation and hypercalcemia in 15% afterward, 60 patients in each group would be needed to reject the null hypothesis with a power of 80% and a 5% significance level. For vitamin D supplementation we used the clinical end point of weight increase and mortality for sample size assessment; the primary end point TBscore was still under development at the time of sample size calculation, but weight was known to be a major component. We estimated that a weight increase of more than 10% would be seen for 70% of the patients, and with intervention 80–85% (relative risk,

Wejse, Gomes, Rabna, et al.: Vitamin D TB Trial

0.82–0.87), and for this we would need 134–286 patients in each arm with a power of 80% and a 5% significance level. For the mortality outcome we stipulated case fatality to be 27% (24), and relative risk after intervention to be 0.6, demanding a sample size of 286 patients in each arm. We estimated enrollment of 300 patients per year, and the planned and funded inclusion period was 2 years, but the incidence estimates we had based this on proved to be too high. After 2 years we had enrolled only 365 patients and the interim analysis showed no difference in mortality or any trends for any of the arms being superior, and hence we decided not to prolong the trial.

Outcome The primary outcome was clinical improvement as assessed by TBscore (23). The TBscore is a newly developed tool aimed at assessment of change in clinical state in patients with TB. It is based on points assigned to signs and symptoms, including cough, hemoptysis, dyspnea, chest pain, night sweating, anemia, tachycardia, lung auscultation finding, fever, low body mass index, and low mid–upper arm circumference, giving patients a TBscore from 0 to 13. Change in TBscore has been shown to detect clinical change well; a high TBscore correlates well with mortality and low TBscores correlate with favorable outcomes, cure, and completed treatment. The secondary outcome was all-cause mortality at 12 months of follow-up. A verbal autopsy was conducted on all deaths, with a physician using a standardized questionnaire to obtain information from the nearest relative. No traumatic deaths were recorded; all died of causes that may be related to TB or HIV. We further assessed sputum conversion in smear-positive patients, weight gain and changes in immunologic response by changes in CD41 T-lymphocyte count. The primary end point was available only for the patients completing 8 months of treatment. Mortality was analyzed by ‘‘intention to treat,’’ that is, for all included patients regardless of number of follow-ups and study drug treatments.

Statistical Analysis We expressed variables by their means or medians and standard deviations or range. The Pearson chi-square (x2) was used to assess statistical differences in proportions between groups (P , 0.05); the Student t test to assess differences in means between two groups when a normal distribution was present; and the Wilcoxon rank-sum test when nonparametric analysis was needed. Linear and logistic regression analyses were used as multivariate models to adjust clinical outcomes for effects of other factors. Cox regression and the Wilcoxon-Breslow-Gehan log-rank test for equality of survivor functions were used to analyze mortality, and Kaplan-Meier survival graphs were used to estimate the survival function. A two-sided P , 0.05 was considered significant for the primary outcome, for mortality P , 0.03 was considered significant because of interim analyses. A false discovery rate was used to correct for multiple testing: test P value 5 [(no. of tests 1 1)/(2 3 no. of tests)] 3 crude P value, hence for exploratory subgroup analyses a P value of 0.03 were considered significant. Therefore 97% confidence intervals were used in estimates in all subgroup analysis. Statistical analyses were performed with STATA version 9 software (StataCorp, College Station, TX).

Interim Analysis A data and safety monitoring board was established to monitor the study. Interim safety analyses were done at 5 and 16 months for calcium concentrations and mortality, and were blinded to treatment group. No clinical outcomes were analyzed, only safety; the predefined stopping rule was a difference in mortality between the two treatment arms at the 5% significance level.

Ethics Pretest and posttest counseling was provided for HIV testing; HIVinfected patients with TB were referred for cotrimoxazole treatment, social support and retesting at a nongovernmental organization voluntary counseling and testing center in Bissau. No antiretroviral treatment was available in Bissau during the study period. Written informed consent was obtained before enrollment. The study was permitted by the Health Ministry of Guinea-Bissau and approved by the National Science and Ethics Committee as well as the Danish National Committee on Biomedical Research Ethics.

845

RESULTS Recruitment began in November 2003 and ended in December 2005. Follow-up was completed in December 2006. Three hundred and sixty-seven patients were enrolled and received study drug at inclusion. Two patients were excluded as a revision of chest X-rays showed no infiltrations, they were smear negative, and therefore did not meet the WHO criteria for tuberculosis. Hence 365 patients were analyzed by intention to treat. One patient was mistaken for another patient and wrongly received placebo instead of the study drug at 2 months of follow-up. He was censored at that date. Two hundred and thirty-three patients came for 8-month follow-up; 9 of these had not received the second dose at 5 months but all were analyzed for the primary outcome. Fifty-seven patients were monitored (found alive at home visit) but were not seen at the final follow-up, 47 patients died during the first 8 months, and 24 patients abandoned treatment or transferred out (they were censored). The flow of participants is outlined in Figure 1. Baseline characteristics, including median age, body mass index, type of TB, prevalence of HIV, baseline CD41 T-lymphocyte (CD4) counts, baseline vitamin D concentrations, and TB severity score were similar between the vitamin D supplementation and placebo groups (Table 1). Baseline characteristics for dropouts were analyzed separately and did not differ significantly from those with complete follow-up (data not shown). Adverse Events

The symptoms asked for as adverse effects were reported most frequently at inclusion, at which time 103 of 365 reported any of the symptoms before receiving the study drug. At 2 months only 24 reported any symptom, most frequently excessive thirst: 10 of 157 (6%) reporting adverse effects in the vitamin D group and 14 of 147 (9%) in the placebo group (P 5 0.31). At 2 months of follow-up mean albumin-corrected serum calcium in the first 120 patients was 2.05 mmol/L (95% confidence interval [CI], 1.98–2.12) for the vitamin D–treated group and 2.09 (95% CI, 2.01–2.17) for the placebo group. One patient exceeded the reference range in the vitamin D group with an albumincorrected serum calcium of 2.93 mmol/L and two patients in the placebo group with serum calcium at 2.92–3.02 mmol/L. At 8 months the mean serum calcium values were slightly higher (2.17 vs. 2.19 mmol/L), but no cases of hypercalcemia were found. Primary Outcome

Changes in TBscore and time to clinical improvement (progression to low-severity class) were similar in the two groups (Figure 2). Sputum samples were not obtained from all patients every second week, but sputum conversion rates were not different in the two groups among the 247 initially smear-positive patients (data not shown); smear positivity over time is presented in Figure 3. Additional Outcomes

We further analyzed weight gain and change in CD4 counts as additional outcomes. Median weight gain at 8 months was 5.9 kg or 11.1% of body weight in the vitamin D–treated group and 5.7 kg or 11.8% in the placebo group (P 5 0.9). In the vitamin D group 56% (70 of 124) gained more than 10% in weight, and in the placebo group 52% (56 of 108) gained more than 10% in weight (P 5 0.48). Weight gain did not differ by HIV status (data not shown). The mean CD4 count at 8 months was higher than at inclusion, but did not differ in the two treatment arms. In the vitamin D–treated group 97 patients had a mean CD4 count of 592 cells/ml (95% CI, 513–671) and in the placebo group 90 patients had a mean CD4 count of 635 cells/ml (95% CI, 556–714). Paired CD4 counts at both inclusion and 8 months

846

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 179

2009

Figure 1. Flowchart of participants.

were available from 133 patients. Among the 70 vitamin D– treated patients there was a mean rise in CD4 count of 50 cells/ml (95% CI, 257 to 158) and in the placebo group a mean fall of 7 cells/ml (95% CI, 2109 to 95). Results differed when stratified by HIV status: Among 41 HIV-infected individuals the vitamin D–treated patients had a fall in CD4 count of 3 cells/ml (95% CI, 282 to 76) versus a mean rise of 19 cells/ml (95% CI, 278 to 117) in placebo-treated patients (P 5 0.71). In contrast, 92 HIVuninfected individuals had a mean rise in CD4 count of 78 cells/ml (95% CI, 282 to 238) in the vitamin D–treated group and a mean fall of 17 cells/ml (95% CI, 2153 to 120) in the placebo group (P 5 0.37). However, this shift in change in CD4 count was not significant (test of homogeneity, P 5 0.47). The Secondary Outcome: Mortality

Fifty-four all-cause deaths were recorded during the 12-month follow-up period, 30 of 187 (16%) in the vitamin D group and 24

of 178 (13%) in the placebo group (Table 2) (Kaplan-Meier estimate shown in Figure 4, log-rank test P 5 0.45). In univariate analyses, the only variables affecting the estimate were the presence of VDI (increasing the hazard ratio [HZ] by 14%) and HIV-1 infection (decreasing the HZ by 8%), and for HIV-1 we noted a tendency for interaction (test of interaction, P 5 0.08). When adjusting for both HIV-1 and VDI in a logistic regression model, the estimate was only moderately increased, HZ 5 1.3 (95% CI, 0.7–2.3). Stratified Analysis 1: Vitamin D Insufficiency

Because the effect of vitamin D supplementation may be seen only among patients with vitamin D deficiency or insufficiency, we assessed the outcomes in these two groups separately in an exploratory, not prespecified, analysis. At treatment initiation, mean TBscores differed according to vitamin D status, although not significantly. The 181 patients with pulmonary TB with

Wejse, Gomes, Rabna, et al.: Vitamin D TB Trial

847

TABLE 1. BASELINE CHARACTERISTICS AT ENROLLMENT Parameter Age, years, mean (SD) Men Ethnicity Balanta Fula Mandinga Manjaco Pepel Others Smear positive Extrapulmonary TB, no. HIV distribution HIV-1 HIV-2 HIV-112 HIV negative Missing Mean CD41 T-lymphocyte count, cells/ml (range) CD41 cell count , 200/ml, no. (%) Weight, kg, mean (SD) Mean BMI (range) MUAC, cm, mean (range) TBscore* (95% CI) Distribution of severity class I II III Mean s-25(OH)D3 (SD) No. with s-25(OH)D3 , 50 nmol/L (%) No. with s-25(OH)D3 , 75 nmol/L (%) Albumin-corrected serum calcium, mmol/L, mean (SD)

Vitamin D (n 5 187)

Placebo (n 5 178)

37 (13) 116 (62%)

38 (14) 106 (60%)

28 29 20 36 41 33 123

29 25 8 32 37 47 124

42/187 18/187 12/187 111/187 4/187 505

(15%) (16%) (10%) (19%) (22%) (17%) (66%) 7 (23%) (10%) (6%) (59%) (2%) (16–1733)

30/138 (22%) 51.9 18.8 237.1 6.7

(9.4) (12–33) (154–358) (6.4–7.0)

60 (32%) 69 (37%) 58 (31%) 77.5 (23.8) 17 (10%)

34/178 18/178 7/178 117/178 2/178 561

(16%) (14%) (5%) (18%) (21%) (26%) (70%) 4 (19%) (10%) (4%) (66%) (1%) (6–2328)

23/138 (17%) 51.1 18.5 236.7 6.8 44 78 56 79.1 13

(8.7) (12–27) (144–328) (6.5–7.1) (25%) (44%) (32%) (21.8) (8%)

86 (48%)

77 (45%)

2.03 (0.26)

2.03 (0.24)

Definition of abbreviations: BMI 5 body mass index; CI 5 confidence interval; MUAC 5 mid–upper arm circumference; s-25(OH)D3 5 serum 25-hydroxyvitamin D3; TB 5 tuberculosis; TBscore 5 clinical scoring system for tuberculosis. Data are presented as number (%) unless otherwise stated. * Only patients with pulmonary TB.

normal vitamin D status had a mean TBscore of 6.5 (95% CI, 6.2– 6.8) compared with 157 patients with pulmonary TB with VDI having a mean TBscore of 6.8 (95% CI, 6.5–7.1), and the 29 patients with pulmonary TB with VDD had a mean TBscore of 7.1 (95% CI, 6.3–7.9). The TBscore at end of treatment differed

Figure 2. TBscore (clinical scoring system for tuberculosis) at various time points for the two treatment arms.

Figure 3. Proportion of patients with positive acid-fast bacteria sputum examination.

significantly by vitamin D status at treatment start. Among the 220 patients with pulmonary TB seen at 8 months with an initial vitamin D measurement, the 117 with normal vitamin D status had a mean TBscore of 0.7 (95% CI, 0.5–0.9) compared with 1.3 (95% CI, 1.0–1.6) among 103 patients with initial VDI. Among these VDI patients, the TBscore was insignificantly lower at 8 months: among 55 vitamin D–treated the mean score was 1.1 (95% CI, 0.8–1.4) compared with 1.4 (95% CI, 1.0–1.9) among 48 placebo-treated (P 5 0.21). For the secondary outcome, mortality, the crude mortality ratio of 1.2 changed moderately when looking at the 163 patients with initial VDI, HZ 5 1.4 (95% CI, 0.5–3.7). But when looking at the 30 patients with VDD, HZ for death among vitamin D–treated patients was 0.7 (95% CI, 0.1–6.4). Adjusting for HIV and background factors did not change these estimates. Stratified Analysis 2: The Role of HIV

Because HIV infection may modify the effect of health interventions, for example, vitamin A, we examined the impact of vitamin D among both HIV-infected and HIV-uninfected individuals in an exploratory, not prespecified, analysis. As demonstrated in Table 3, the data suggest a tendency after the first dose of vitamin D toward a more rapid decrease in the TBscore among HIV-uninfected patients treated with vitamin D and a more slow decrease in HIV-infected patients treated with vitamin D, TBscores being significantly different by HIV status in the vitamin D–treated group (P 5 0.008) after 2 months of followup but not in the placebo group (P 5 0.45). This pattern was similar at 5 and 8 months with significant differences by HIV status in the vitamin D–treated group, although TBscores at inclusion did not differ significantly. Table 3 also shows a lower TBscore at the end of treatment in the HIV-uninfected group; however, this difference was not significant. In addition, we analyzed whether mortality differed when stratified for HIV status. Among the 131 HIV-infected patients, 12-month mortality was high, as shown in Table 2. In a Cox regression model we examined mortality separately for both HIV-1–infected patients and for all HIV-infected patients. HZ estimates were not changed when looking at all HIV-infected patients, and among the 95 HIV-1–infected patients the mortality risk was not significantly higher in the vitamin D–treated arm, HZ 5 1.8 (95% CI, 0.8–4.1) or significantly lower among the 228 HIV-uninfected patients, HZ 5 0.9 (95% CI, 0.3–2.8). This

848

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 179

2009

TABLE 3. TBscore IN TREATMENT GROUPS BY TIME AND HIV STATUS Mean TBscore (NVitamin D/NPlacebo)

Vitamin D (SD)

Placebo (SD)

P Value

Inclusion HIV1* (72/58) HIV2 (104/114)

6.9 (2.1) 6.5 (1.9)

6.8 (2.1) 6.7 (1.9)

0.79 0.54

Two months HIV1 (57/49) HIV2 (94/97)

2.8 (2.3) 1.9 (1.7)

2.4 (2.0) 2.1 (2.0)

0.34 0.35

Five months HIV1 (49/43) HIV2 (88/91)

1.6 (1.6) 1.0 (1.4)

1.6 (2.2) 1.2 (1.3)

0.89 0.35

Eight months HIV1 (39/32) HIV2 (81/74)

1.3 (1.4) 0.7 (1.0)

1.2 (1.3) 1.0 (1.4)

0.91 0.09

†

Figure 4. Kaplan-Meier plot of cumulative incidence proportion (1 – survival) for the two groups.

twofold difference in HZ ratio estimates, although not significant, showed indications of effect modification of HIV on vitamin D treatment (test of interaction, P 5 0.08). By adjusting for the presence of VDI the mortality risk for HIV-1–infected individuals was 2.4 (95% CI, 0.95–6.5). Vitamin D Status at 2 and 8 Months

Vitamin D status was remeasured at 2 and 8 months, which were 2 and 3 months after the first and second doses were given at inclusion and at 5 months. At 2 months 270 samples were available for analysis, and at 8 months 226 samples were available. A general rise was seen in serum 25(OH)D3 concentrations, in 203 patients with paired samples at inclusion, 2 months, and 8 months (Figure 5). In this group mean serum 25(OH)D3 increased from 78 nmol/L (95% CI, 75–81) to 103 nmol/L (95% CI, 99–108) at 2 months of follow-up and 98 nmol/L (95% CI, 94–102) at 8 months. The increase was also present in the placebo group: Mean 25(OH)D3 increased to 105 nmol/L (95% CI, 99–110) in the treatment arm and 103 nmol/L (95% CI, 96–110) in the placebo arm at 2 months and to 102 nmol/L (95% CI, 96–107) in the treatment arm and 95 nmol/L (95% CI, 89–102) in the placebo arm at 8 months. Fractions of patients with VDD/VDI were not lower to at inclusion but were lower in the treatment arm at 2 and 8 months, although not significantly different (data not shown); crude data are displayed in Figure 5. A nonsignificant effect on serum vitamin D levels was noted among vitamin D–insufficient patients at baseline: Patients with 25(OH)D3 concentrations less than 75 nmol/L increased from 60 nmol/L (95% CI, 57–63) to 102 nmol/L (95% CI, 93–110) whereas placebo-treated patients increased from 60 nmol/L (95% CI, 57–63) to only 95 nmol/L (95% CI, 85–106).

DISCUSSION Vitamin D supplementation of patients with TB in GuineaBissau did not result in serious adverse events, and we showed no effect on the chosen outcomes in the entire group of patients. The data show no significant differential effects of vitamin D TABLE 2. MORTALITY BY HIV STATUS Infection HIV-1 and HIV-112 infected HIV-2 only infected HIV-1 and HIV-2 uninfected HIV status unknown Total

Vitamin D

Placebo

20/54 1/18 7/111 2/4 30/187

10/41 5/18 8/117 1/2 24/178

Mortality Ratio 1.52 0.2 0.92 1 1.19

(0.80–2.88) (0.03–1.55) (0.35–2.46) (0.18–5.5) (0.72–1.95)

* HIV infected (HIV-1 and HIV-2 and HIV-112 combined). † HIV uninfected.

supplementation in either HIV-infected or vitamin D–insufficient subjects. We may, however, speculate that there is a trend toward positive effects for HIV-uninfected and vitamin D– insufficient subjects and a trend toward higher mortality among HIV-infected patients treated with vitamin D. None of the HIV-positive patients were receiving antiretroviral treatment although 53 of 273 (19%) had CD41 T-lymphocyte counts indicating a need for such treatment, which unfortunately was not available in Bissau at the time of the study. Therefore the findings cannot be generalized to HIV-positive patients treated with antiretroviral therapy. As reported elsewhere (28), VDI was common among patients with TB in the study area, but severe vitamin D deficiency was rare. Our findings indicate a slightly beneficial effect among patients with TB with low vitamin D, but the study was not sufficiently powered or designed to determine this. Furthermore, because measuring vitamin D is costly and requires a certain level of laboratory facility, it is unlikely that measuring vitamin D in all patients with TB will be routine in resource-poor settings. Our aim was to test vitamin D treatment in such a setting to determine whether vitamin D is of value for patients with TB in resourcepoor areas. Our study addresses the controversy of hypercalcemia in patients with TB (29). We found no evidence of TB-associated risk of hypercalcemia in this population, but we report a slight rise in calcium concentrations during antituberculosis treatment irrespective of vitamin D treatment arm. This observation is in line with Fuss and colleagues, who did not find that vitamin D supplementation of patients with TB resulted in hypercalcemia (30). In 1969 Brincourt reported the result of repeated oral supplementation with 15 mg (600,000 IU) of vitamin D2 (3) as supplementary treatment to antibiotics in 150 patients in a study with no control group and noted an effect on dissolving cavities; Brincourt observed no hypercalcemia. Our findings of lack of clinical effect are contradictory to what Brincourt, as well as numerous studies from the preantibiotic era, have reported. However, they all used much higher dosages than in this trial, and all of these studies were uncontrolled (31–33). In modern times we know of only two small randomized trials enrolling 24 children with TB in Egypt (15) and 67 adults in Indonesia (14). They used oral vitamin D at 1,000 and 10,000 IU/ day, respectively, or placebo. They reported a more evident clinical and radiographic improvement in the treated group in Egypt (although not statistically significant), and improved sputum conversion rates in the Indonesian study. These studies were small and HIV status was not reported; however, it is unlikely that these studies included many HIV-infected individ-

Wejse, Gomes, Rabna, et al.: Vitamin D TB Trial

Figure 5. Vitamin D serum levels before and during study.

uals. It is worth noting that the Indonesian study showed benefit with a combined dose 1 log higher than the present study; the reason for this trial not showing effect may therefore be that the dose was insufficient. We based our choice of dose on what has been shown effective toward osteomalacia, but the optimal dose for an antituberculous effect is not known and it may well be higher. As shown in Figure 5, the doses of vitamin D used did not have a lasting effect on vitamin D levels in the patients who received vitamin D and the lack of effect may therefore be due to a suboptimal dose concentration or a suboptimal dosage interval. The finding that vitamin D levels were higher at 2 months of follow-up in both treatment arm and placebo arm is curious and unexpected. The half-life for a single large dose is 2 months and no sample was available earlier, and therefore a difference may have been present right after vitamin D supplementation. A large proportion of the patients had normal vitamin D levels before the intervention, and if the dose given was suboptimal even for nondeficient patients no difference between groups would be expected 2 months later. There are no previous data on repeated vitamin D measurements during a course of standard tuberculosis treatment, and therefore a rise in vitamin D levels at 2 months may be a normal phenomenon. It is likely that patients with a TB diagnosis in the intensive phase of treatment are allowed a better share of food by the family to compensate for their illness; in our experience the patients with TB are focused on obtaining a rich diet with meat and eggs if possible and may therefore have increased their dietary vitamin D intake. A change in sun exposure may also explain the increased vitamin D levels, but we have no knowledge of this being the case. Activated macrophages have been shown to hydroxylate 25(OH)D (34) to 1,25dihydroxyvitamin D [1,25(OH)2D]; active synthesis of 25(OH)D from nonhydroxylated vitamin D by macrophages no longer occupied by mycobacteria may therefore be a possible additional mechanism for increased vitamin D in the placebo group. Our findings also contrast with those of Range and colleagues (35), who reported a 50% reduction in mortality among HIVinfected patients with TB treated with multivitamin supplementation including vitamin D in a randomized clinical trial in Tanzania. This was a subgroup analysis result, currently being tested in a new trial, and they used much lower vitamin D doses as

849

well as a variety of other micronutrients such as zinc, which was shown to have a separate beneficial effect. We observed trends in CD4 counts indicating an increase in CD4 count in vitamin D–treated subjects who were not HIV infected and a decrease in HIV-infected patients. We therefore speculate that vitamin D has differential effects depending on whether the immune system is intact. We may hypothesize that a small beneficial effect is present in HIV-uninfected patients with TB; vitamin D has been shown to exert an effect in vitro on host immune response to Mycobacterium tuberculosis (5). HIV-infected patients, on the other hand, are known to suffer from further immune activation (36), which may cause the slower clinical improvement and the nonsignificantly higher mortality present among vitamin D–treated HIV-1– infected patients. We may later analyze additional nonplanned outcomes such as HIV viral load in this cohort to furnish more data on this matter. Indeed, vitamin D supplementation may have only a minor effect on the outcome of TB treatment. Future studies may therefore instead aim to evaluate vitamin D as an agent to improve immune function in order to prevent latent TB infection. In conclusion, we found satisfactory safety when supplementing patients with TB with vitamin D at high doses. In this trial vitamin D supplementation showed no beneficial effect for patients with TB in general, but this may have been due to suboptimal dosage. Further trials investigating larger, perhaps daily, vitamin D doses for prevention of latent TB infection appear warranted. Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Acknowledgment: The authors thank the TB study team at the Bandim Health Project; the staff at the Bandim, Belem, and Cuntum Health Centers and at Sant D’Egidio TB Hospital; Henrik Friis and Søren Johnsen for useful suggestions early in the study; and Lars Østergaard, Lars Pedersen, and Henrik Toft Sørensen for serving on the Data and Safety Monitoring Board and Jesper Eugen-Olsen and Jens Nielsen for advice on interpreting the data. The authors thank Crinex Pharmaceuticals for providing the cholecalciferol and placebo ampules.

References 1. Dowling GB, Prosser Thomas EW. Treatment of lupus vulgaris with calciferol. Lancet 1946;1:919–922. 2. Ellman P, Anderson K. Calciferol in tuberculosis peritonitis with disseminated tuberculosis. BMJ 1948;1:394. 3. Brincourt J. Liquefying effect on suppurations of an oral dose of calciferol. Presse Med 1969;77:467–470. 4. Bennett JH. The pathology and treatment of pulmonary tuberculosis, 1st ed. Edinburgh: Sutherland and Knox; 1853. p. 138. 5. Liu PT, Stenger S, Li H, Wenzel L, Tan BH, Krutzik SR, Ochoa MT, Schauber J, Wu K, Meinken C, et al. Toll-like receptor triggering of a vitamin D–mediated human antimicrobial response. Science 2006; 311:1770–1773. 6. Martineau AR, Wilkinson RJ, Wilkinson KA, Newton SM, Kampmann B, Hall BM, Packe GE, Davidson RN, Eldridge SM, Maunsell ZJ, et al. A single dose of vitamin D enhances immunity to mycobacteria. Am J Respir Crit Care Med 2007;176:208–213. 7. Wilkinson RJ, Llewelyn M, Toossi Z, Patel P, Pasvol G, Lalvani A, Wright D, Latif M, Davidson RN. Influence of vitamin D deficiency and vitamin D receptor polymorphisms on tuberculosis among Gujarati Asians in West London: a case–control study. Lancet 2000; 355:618–621. 8. Ustianowski A, Shaffer R, Collin S, Wilkinson RJ, Davidson RN. Prevalence and associations of vitamin D deficiency in foreign-born persons with tuberculosis in London. J Infect 2005;50:432–437. 9. Nnoaham KE, Clarke A. Low serum vitamin D levels and tuberculosis: a systematic review and meta-analysis. Int J Epidemiol 2008;37:113–119. 10. Gibney KB, MacGregor L, Leder K, Torresi J, Marshall C, Ebeling PR, Biggs BA. Vitamin D deficiency is associated with tuberculosis and latent tuberculosis infection in immigrants from sub-Saharan Africa. Clin Infect Dis 2008;46:443–446. 11. Bellamy R. Evidence of gene–environment interaction in development of tuberculosis. Lancet 2000;355:588–589. 12. Zasloff M. Fighting infections with vitamin D. Nat Med 2006;12:388–390.

850

AMERICAN JOURNAL OF RESPIRATORY AND CRITICAL CARE MEDICINE VOL 179

13. Martineau AR, Honecker FU, Wilkinson RJ, Griffiths CJ. Vitamin D in the treatment of pulmonary tuberculosis. J Steroid Biochem Mol Biol 2007;103:793–798. 14. Nursyam EW, Amin Z, Rumende CM. The effect of vitamin D as supplementary treatment in patients with moderately advanced pulmonary tuberculous lesion. Acta Med Indones 2006;38:3–5. 15. Morcos MM, Gabr AA, Samuel S, Kamel M, el Baz M, el Beshry M, Michail RR. Vitamin D administration to tuberculous children and its value. Boll Chim Farm 1998;137:157–164. 16. Benn CS, Bale C, Sommerfelt H, Friis H, Aaby P. Hypothesis: vitamin A supplementation and childhood mortality: amplification of the nonspecific effects of vaccines? Int J Epidemiol 2003;32:822–828. 17. Wejse C, Gomes VF, Rabna P, Aaby P, Lisse I, Andersen PL, Glerup H, Sodemann M. Vitamin D as supplementary treatment for tuberculosis: a randomised double-blind placebo-controlled trial. Poster P2.106 presented at the Royal Society of Tropical Medicine and Hygiene Centenary Conference. September 13–15, 2007, London. 18. Gustafson P, Gomes VF, Vieira CS, Rabna P, Seng R, Johansson P, Sandstrom A, Norberg R, Lisse I, Samb B, et al. Tuberculosis in Bissau: incidence and risk factors in an urban community in subSaharan Africa. Int J Epidemiol 2004;33:163–158. 19. Harries AD, Maher D, Graham S. TB/HIV: a clinical manual, 2nd ed. Geneva: WHO; 2004. WHO/HTM/TB/2004.329. 20. Vieth R. Vitamin D supplementation, 25-hydroxyvitamin D concentrations, and safety. Am J Clin Nutr 1999;69:842–856. 21. Trivedi DP, Doll R, Khaw KT. Effect of four monthly oral vitamin D3 (cholecalciferol) supplementation on fractures and mortality in men and women living in the community: randomised double blind controlled trial. BMJ 2003;326:469. 22. Powell-Tuck J, Hennessy EM. A comparison of mid upper arm circumference, body mass index and weight loss as indices of undernutrition in acutely hospitalized patients. Clin Nutr 2003;22:307–312. 23. Wejse C, Gustafson P, Nielsen J, Gomes VF, Aaby P, Andersen PL, Sodemann M. A clinical score system for monitoring tuberculosis in a low-resource setting. Scand J Infect Dis 2008;40:111–120. 24. Seng R, Gustafson P, Gomes VF, Vieira CS, Rabna P, Larsen O, Larouze B, Norberg R, Lisse IM, Samb B. Community study of the relative impact of HIV-1 and HIV-2 on intrathoracic tuberculosis. AIDS 2002;16:1059–1066.

2009

25. Gustafson P, Gomes VF, Naucler A, Vieira CS, Samb B, Aaby P, Lisse I. Clinical predictors for death in HIV-positive and HIV-negative tuberculosis patients in Guinea-Bissau. Infection 2007;35:69–80. 26. Maunsell Z, Wright DJ, Rainbow SJ. Routine isotope-dilution liquid chromatography–tandem mass spectrometry assay for simultaneous measurement of the 25-hydroxy metabolites of vitamins D2 and D3. Clin Chem 2005;51:1683–1690. 27. Vieth R. What is the optimal vitamin D status for health? Prog Biophys Mol Biol 2006;92:26–32. 28. Wejse C, Olesen R, Rabna P, Kaestel P, Gustafson P, Aaby P, Andersen PL, Glerup H, Sodemann M. Serum 25-hydroxy-vitamin D in a West African population of tuberculosis patients and un-matched healthy controls. Am J Clin Nutr 2007;86:1376–1383. 29. Fuss M, Pepersack T, Gillet C, Karmali R, Corvilain J. Calcium and vitamin D metabolism in granulomatous diseases. Clin Rheumatol 1992;11:28–36. 30. Fuss M, Karmali R, Pepersack T, Bergans A, Dierckx P, Prigogine T, Bergmann P, Corvilain J. Are tuberculous patients at a great risk from hypercalcemia? Q J Med 1988;69:869–878. 31. Winterberg H. Vigantol therapy in early tuberculous infections of the lungs, and in exsudative pulmonary tuberculosis. Z Tuberk 1950;95: 146–150. 32. Ghosh PK. Vitamin D2 in tuberculosis. J Indian Med Assoc 1957;29: 90–93. 33. Trautwein H, Stein E. Vitamin D2 and pulmonary tuberculosis; effects and therapeutic results. Beitr Klin Tuberk Spezif Tuberkuloseforsch 1952;107:295–313. 34. Adams JS, Ren SY. Autoregulation of 1,25-dihydroxyvitamin D synthesis in macrophage mitochondria by nitric oxide. Endocrinology 1996; 137:4514–4517. 35. Range N, Changalucha J, Krarup H, Magnussen P, Andersen AB, Friis H. The effect of multi-vitamin/mineral supplementation on mortality during treatment of pulmonary tuberculosis: a randomised twoby-two factorial trial in Mwanza, Tanzania. Br J Nutr 2006;95: 762–770. 36. Brenchley JM, Price DA, Schacker TW, Asher TE, Silvestri G, Rao S, Kazzaz Z, Bornstein E, Lambotte O, Altmann D, et al. Microbial translocation is a cause of systemic immune activation in chronic HIV infection. Nat Med 2006;12:1365–1371.