Serum vitamin D levels and risk of prevalent tuberculosis, incident tuberculosis and tuberculin skin test conversion among prisoners

Poor vitamin D status has been associated with tuberculosis (TB); whether poor status is cause or consequence of disease is uncertain. We conducted a case-control study and two nested case-control studies to determine whether vitamin D levels were associated with active TB, tuberculin skin test (TST) conversion, and risk of progression to the active TB in prisoners in Brazil. In multivariable conditional logistic regression, subnormal vitamin D levels (OR, 3.77; 95% CI, 1.04–13.64) were more likely in prisoners with active TB. In contrast, vitamin D was not found to be a risk factor for either TST conversion (OR, 2.49; 95% CI, 0.64–9.66) or progression to active disease (OR, 0.59; 95% CI, 0.13–2.62). Black race (OR, 11.52; 95% CI, 2.01–63.36), less than 4 years of schooling (OR, 2.70; 95% CI, 0.90–8.16), cigarette smoking (OR, 0.23; 95% CI, 0.06–0.79) were identified as risk factors for TST conversion. Risk of progression to active TB was found to be associated with cigarette smoking (OR, 7.42; 95% CI, 1.23–44.70). Our findings in the prison population show that poor vitamin D status is more common in individuals with active TB, but is not a risk factor for acquisition of latent TB or progression to active TB.


Results
Risk factors associated with active TB. The frequency distribution of the variables and risk factors are shown in Table 1. We compared 24 TB cases and 48 matched controls without TB. The mean age of prisoners for the case-control study was 32 years (±7 years) among cases and 33 years (±8 years) among the controls ( Table 2). Mean serum 25(OH)D 3 level was significantly lower among the cases (27.7 ± 7.85 ng/mL) than controls (37.1 ± 8.94 ng/mL, p < 0.01; Table 2). Prisoners with active TB were more likely to have subnormal vitamin D levels (<30 ng/mL) when compared to healthy controls (18/24 [75%] vs 16/48 [33%], p = 0.004; Table 1). As shown in Table 1, there is a clear difference between cases and controls with active TB (p < 0.01). A greater difference existed among TB patients with vitamin D deficiency (3.5×), than those with vitamin D insufficiency (2.0×) when compared with controls. All cases had negative HIV serology and only two controls had positive HIV serology (Table 1).
Bivariable conditional logistic regression analyses were performed to assess the associations between active TB and thirteen potential predictor variables. These variables included drug use over the last year (OR, 4

Risk factors associated with TST conversion.
In the first nested case-control study, we selected individuals who screened negative for TB at baseline by symptoms and sputum culture, and compared those who converted their TST from negative to positive when retested at one year (converters) with those who remained negative upon repeat testing (non-converters). The mean age of the converters was 32 years (±6 years) and 31 years (±7 years) among the non-converters (Table 2). Vitamin D levels did not differ significantly between the TST converters (37.7 ± 11.93 ng/mL) and non-converters (34.5 ± 14.89 ng/mL, p = 0.19). Albumin levels differed significantly between cases (4.63 ± 0.43 g/dL) and controls (4.28 ± 1.00 g/dL, p < 0.01) but were in the normal range (>3.5-5.2 g/dL) ( Risk factors associated with progression to active TB. In the second nested case-control study, we compared baseline characteristics of individuals, who screened negative for active TB at baseline, according to whether they progressed to active TB over the subsequent year ("progressors") or did not ("non-progressors"). Thirty-six percent of those who progressed and 50% of those who did not progress had positive TSTs at baseline. No significant difference in baseline serum 25(OH)D 3 levels was identified between progressors (37.0 ± 14.83 ng/ mL) and non-progressors (37.5 ± 11.04 ng/mL, p = 0.86; Table 2). All cases had negative HIV serology and only one control had positive HIV serology (Table 1).
The average time to active TB diagnosis among progressors was 6.9 months. No statistical difference in serum 25(OH)D 3 level at enrollment was observed among prisoners who progressed to active TB in the first 6 months compared to prisoners who had progressed after 6 months of follow-up (35.5 ng/mL ±12.01 and 38.4 ng/mL ±15.36, respectively; p = 0.46).

Discussion
Despite the low power, this is the largest study ever published that evaluate the association of vitamin D levels with progression to TST conversion and progression of active disease in the prison population. We found that prisoners with active TB had significantly lower vitamin D levels at the time of diagnosis compared with prisons without active TB. However, baseline vitamin D levels were not predictive of TST conversion or progression to active disease. Notably, while 75% of prisoners with active TB had subnormal vitamin D levels at TB diagnosis, less than 33% of all other groups of participants (cases and controls) presented subnormal levels. These findings provide support for a pattern of disrupted vitamin D metabolism that results from the presence of active TB disease, and do not provide evidence for subnormal vitamin increasing susceptibility to infection or disease.
Our primary findings of poor vitamin D status associated with active TB among prisoners were consistent with the results of case-control studies conducted in the general population 8,[17][18][19][20][21][22][23][24][25][26] . Furthermore, previous cohort studies conducted among household contacts of TB cases have suggested that low levels of vitamin D may be associated with risk of subsequent M. tuberculosis infection 27 and development of active TB 28,29 ; we did not find these associations in this study when vitamin D levels were examined months prior to diagnosis and found to be largely normal. The number of subjects experiencing the outcome of interest in previous studies was limited, however, ranging from 3 to 18 over follow-up periods 2 to 4 years. These studies have been found to vary with study design, geographical location 8 , as well as ethnicities 30 and gender 24 of subjects. Thus, studies with larger sample sizes in different populations are needed to help clarify whether the association between vitamin D and susceptibility to TB infection or disease is causal in nature.
Notably, we previously identified high rates of conversion (28%) and incident active disease (1.39%) over one year in this cohort study 10 . These rates provided more robust sample sizes of 30 and 24 cases, respectively, for the nested case-control studies to examine the impact of vitamin D status at baseline. Surprisingly, the majority of healthy prisoners show normal vitamin D. In the individuals without active disease at enrollment, both diet and sun exposures in this incarceration setting appear to be adequate to sustain precursor vitamin D levels in healthy prisoners. Vitamin D status under these conditions appeared not to be associated with either the acquisition of latent infection, or the development of active disease in this prison population. It is possible that there was insufficient heterogeneity in vitamin D in this population and setting to detect such effects. Here, factors other than vitamin D levels (e.g. smoking status) appear to contribute risk along the TB disease pathway.
Previous studies have recognized that inflammatory process may lower 25(OH)D 3 levels as the precursor supply for activated vitamin D 31,32 . The normal immune response against M. tuberculosis produces inflammatory cytokines that activate CYP27B1 (1α-hydroxylase), converting available 25(OH)D 3 to its activated form 1,25(OH) 2 D 3 31,33 and contributing to the measurement of low 25(OH)D 3 levels 34 . Thus, the low levels of 25(OH) D 3 (precursor) identified among prisoners with active TB in our case-control and among in several previous studies may have been associated with high conversion rates to 1,25(OH)2D 3 caused by persistent disease. Notably, elevated plasma 1,25(OH)2D3 levels have been found in TB patients compared to healthy individuals 35,36 , suggesting hyperresponsive metabolism to form activated vitamin D may be associated with the disease in humans. However, vitamin D supplementation as part of randomized clinical trials during TB treatment has not been shown to reduce time to sputum culture conversion after 8 weeks [37][38][39][40][41][42][43][44] .
One of the major defense mechanisms involving vitamin D and M. tuberculosis is associated with Toll-like receptor (TLR) 45 . Mycobacterial antigens, especially lipoproteins, are recognized by TLR2 or associated with TLR1 or TLR6 which when activated participate in the expression of vitamin D receptor (VDR) and activation of 25(OH)D, assisting in the release of cathelicidin, a peptide capable of inhibiting the intracellular growth of M. tuberculosis, in vivo 46 and in vitro 45 .
Vitamin D levels may be influenced by anti-TB drugs, especially rifampicin and isoniazid 47 , and if vitamin D measurement is performed after the initiation of treatment, these levels may be considerably lower among patients with TB 17,48-51 . One strength of our study was that all serum samples used to measure vitamin D levels were collected before treatment was initiated removing this potential factor impacting vitamin D.
Risk of TST conversion was not influenced by baseline vitamin D status in the prisoners, but other risk factors showed associations consistent with the previous reports. For example, previous studies reported an association between being a current smoker and TB 52,53 . Evidence suggests that smoking decreases TST reactivity 54,55 but increases the risk of progression to active TB 56 by directly impairing host immunity, especially the responses of T-cells and macrophages to M. tuberculosis 57 52,57 . While the variable cigarette smoking was found here to exert a protective effect on risk of TST conversion to latent infection, it dramatically increased risk of progression to active TB. TST conversion was also associated with black race, which increased the odds of infection by M. tuberculosis. Evidence suggests that black race may increase risk of M. tuberculosis infection greater than other racial backgrounds 59 . Although the mechanism underlying this association has not been fully elucidated, it may be attributed to environmental factors, unfavorable socioeconomic conditions 60,61 , and genetic factors 62,63 . Black skin can impair vitamin D synthesis from UVB exposure, but notably, vitamin D status was found to be an independent risk factor for active TB when controlling for race in the case-control study. Although it is not yet clear why race is a risk factor for TST conversion, we believe that in addition to the polymorphisms vitamin D receptors, other genetic factors may be associated. In our study, in the final model, the level of Vitamin D was adjusted by race and even then, the serum level of vitamin D was not associated with TST conversion. Other polymorphisms associated with black race should be investigated.
Low levels of schooling also showed a tendency toward increased risk and may be directly related with incarceration mixing patterns of high risk individuals during incarceration 64 . The prisoners are combined in cell blocks of mixed socioeconomic conditions, thus increasing the probability of becoming infected with M. tuberculosis 65,66 . However, the fact that all prisoners were exposed to the same conditions during incarceration, reinforces the need for further studies to be conducted to explain the reasons behind the racial and education inequalities among prisoners with active TB.
We examined risk of incident TB among all prisoners rather than those who were TST positive at baseline. The majority of TB disease in this high transmission setting occurs due to recent infection rather that late reactivation. Indeed, the majority of incident TB cases over one year in this cohort occurred among individuals who were TST negative at baseline.
While studies remain challenged by power, this work represents the largest study ever conducted to evaluate TST conversion and progression of active disease. We used the TST conversion to assess TB infections and progression to active TB, however, some studies indicate that TST can cause false-positive results by BCG vaccination and exposure to non-tuberculosis mycobacteria (NTM). In Brazil, BCG vaccination usually performed in the first days of life that does not interfere with the TST result if carried out 10 years or more after vaccination 67,68 . In this study the average age was 32.1 (±6.87, p = 0.50) in TST conversion and 30.8 (±8.90, p = 0.63) in progression to active TB, it seems unlikely that BCG vaccination interfered with the TST results. As well as, exposure to non-tuberculosis mycobacteria (NTM) may interferes with TST results only in populations with a high prevalence of NTM sensitization and low TB infection, that does not happen in our study population, where the incidence of active TB is at least 30 times higher than that observed in the general population 10 . Interferon-ɤ release assays (IGRA) to determinate TST conversion may be more useful in populations where the BCG vaccine has not been administered or if the test was performed within 10 years of vaccination, thus, in our study, the TST and IGRAs likely would generate similar findings 67 .
Data suggest that seasonality 69 and the incarceration time 70 may favor a decrease in vitamin D levels. A strength of our design however, was that all prisoners were screened in the same months, and cases and controls were paired according to the prison cell block and incarceration time, having the same time of potential sun exposure and similar diet; these features help minimize confounding in our analysis, however is very difficult to perfectly match due to the prolonged and variable infectious period, variable social mixing, and frequent movement of inmates between cells. Overmatching among cases and controls, may have decreased the variability in vitamin D levels limiting the ability to identify important associations over the spectrum of vitamin D with the disease pathway. As a caveat, as many participants have sufficient vitamin D this may play a minor role in relation to other risk factors.

Conclusion
We identified that prisoners with active TB had poor vitamin D status; however, vitamin D levels were not identified to be a risk factor for TST conversion or the development of active TB over a follow-up period of one year in the evaluated prison population. Other risk factors such as black race and low education were found to be risk factors for TST conversion, while smoking was the only variable associated with disease progression. Our findings suggest active TB may alter vitamin D metabolism, leading to compromised precursor supply, but that vitamin D status is not necessarily a risk factor for susceptibility to infection or disease.

Methods
Population. In 2013, the state of Mato Grosso do Sul (MS), located in the west-central Brazil, had the country's seventh largest prison population, including 14,904 prisoners, and the higher rate of national incarceration (568.9 prisoners per 100,000 inhabitants) 71 . A one-year prospective cohort study was conducted in 2013 in 12 prisons (8 male prisons, with 6,552 prisoners and 4 female prisons, with 669 prisoners) to identify the incidence of latent and active TB 10,16 . The case-study and two nested case-studies here are part of this larger cohort study.
Prisoners who were over 18 years old and consented to participate in the study were recruited from a randomly stratified sample at the 12 prisons. A structured questionnaire was administered and a (TST) (PPD RT23, Staten Serum Institute, Copenhagen) was performed. Two sputum samples were obtained from all participants reporting cough, and smear microscopy and culture, using the Ogawa-Kudoh technique, was performed. Although X-ray screening shows good sensitivity, and is an important method for screening suspected cases of TB, the specificity of X-ray screening is low. Additionally, chest radiography is not available in the study prisons. For these reasons, we used sputum culture to define TB cases. Individuals with at least one smear or culture result positive for M. tuberculosis were considered TB cases. A venous blood sample was collected, and plasma was frozen for assessment of vitamin D levels. TST positivity was defined as an induration of ≥10 mm measured by a trained TST reader, 48 hours after injected. After one year, in 2014, a second TST was performed, and two sputum samples were collected for smear microscopy and culture from all participants reporting cough at the second time point. TST conversion was defined as TST ≥10 mm and an induration increase of at least 6 mm in an individual with a baseline TST <10 mm 10 .
Incident TB cases were defined as those occurring after the first screening time point, including those diagnosed between the two screening dates (day 0 and day 365) and those diagnosed on the second screening date. Those diagnosed between the screening dates were diagnosed through routine medical services provided by the prison, in which symptomatic individuals are clinically evaluated and sputum is sent for smear and/or culture as determined the prison clinician. These TB cases occurring between the screening dates were identified through prison medical records and the National Notifiable Disease database (Sistema de Informação de Agravos de Notificação National, SINAN).
A plasma sample for vitamin D levels was collected at the time of enrollment, which was prior to TB diagnosis, and therefore, prior to initiation of treatment. Due to the low rates of TB in the female prisoners, the vitamin D evaluation was conducted in only male prisoners 10 .
All eligible participants provided written informed consent prior to study participation. All methods described in this study were approved by Research Ethics Committee of the Federal University of Grande Dourados (Number 191,877 and 793,740) and were performed in accordance with the relevant guidelines and regulations. The study was authorized by the State Agency of Administration of the Penitentiary System (AGEPEN) who gave full support during the study period.
Risk factors associated with active TB. During enrollment in the cohort study, a total of 29 male prisoners from 2,861 were culture positive for M. tuberculosis and, therefore, had active TB 16 (Fig. 1). Based on this finding, a case-control study was performed in which 25(OH)D levels were measured in serum samples obtained from 24 prisoners with active TB and compared with those from samples obtained from 48 controls, 25 of whom were TST-negative and 23 were TST positive, who were matched on the basis of age (±5 years), prison cell block and incarceration time (±6 months) (Fig. 1). Other risk factors of interest were age, marital status, education, resides in MS, smoking, drug use over the last year, alcohol, contact with a TB-positive individual and previous incarceration. The participant's race/skin color (i.e., white, black, indigenous, asian or mixed) was self-reported. The presence of a Bacillus Calmette-Guérin (BCG) vaccine scar was determined by inspecting the participant's arm. The body mass index (BMI) wasn't included in evaluation of active TB because we do not know if cause or effect.
Risk factors associated with TST conversion. A nested case-control study was performed to identify risk factors associated with TST conversion. Thirty prisoners randomly who converted their TST from negative to positive (converters) over the course of 12 months were compared with sixty prisoners who had negative TST tests at baseline and repeat testing at 12 months (non-converters) (Fig. 1). Controls were matched based on age (±5 years), prison cell block and incarceration time (±6 months). Other risk factors assessed for TST conversion were age, marital status, education, resides in MS, smoking, drug use over the last year, alcohol, serum albumin, contact with a TB-positive individual, BMI and previous incarceration. The presence of a BCG vaccine scar was determined by inspecting the participant's arm.

Risk factors associated with progression to active TB.
A nested case-control study was conducted to investigate the association between vitamin D levels and risk of subsequent TB disease. Of the 1,235 prisoners who remained incarcerated after 1 year in the male prison system (Fig. 1), we selected 24 individuals without TB at baseline who were diagnosed with active TB over the course of 12 months (progressors) and compared them with 48 individuals who were TB negative throughout 12 months (non-progressors), as documented by active TB screening at baseline and 12 months. Controls were matched to cases on the basis of age (±5 years), prison cell block and incarceration time (±6 months) (Fig. 1). The risk factors assessed for progression to active TB were age, marital status, education, resides in MS, smoking, drug use over the last year, alcohol, serum albumin, contact with a TB-positive individual, BMI and previous incarceration. The presence of a BCG vaccine scar was determined by inspecting the participant's arm.
Biochemical analyses. Vitamin D levels were measured in serum samples, which were stored away from light and at −20 °C until the time of analysis. Levels of 25(OH)D (ng/ml) were identified using the automated electrochemiluminescence immunoassay method and Cobas ® e411 analyzer (Roche Diagnostics, Mannheim, Germany). Serum 25(OH)D levels were considered normal when values were ≥30 ng/mL [72][73][74] . Subnormal vitamin D was used to indicate combined groups for vitamin D insufficiency (29-20 ng/ml) and deficiency (<20 ng/ ml). Serum albumin levels were measured in samples obtained from prisoners included the nested case-control studies using the Cobas Integra ® 400 plus analyzer (Roche Diagnostics, Mannheim, Germany). All procedures were performed in the Laboratory of the University Hospital of the Federal University of Grande Dourados. Data analysis. All questionnaires were double entered into the Research Electronic Data Capture database (REDCap). SAS version 9.2 (SAS Institute, Cary, NC, USA) was used to perform the analyses. We used the t-test for continuous variables, Chi-square test for categorical variables and bivariable conditional logistic regression analyses to examine the associations between dependent and independent variables. Vitamin D insufficient and all variables with p ≤ 0.10 in the bivariable analyses were included in multivariable conditional logistic regression analysis. P values ≤0.05 were considered statistically significant.