Vitamin E Status Is Inversely Associated with Risk of Incident Tuberculosis Disease among Household Contacts

Abstract

Background

Few studies have previously assessed how pre-existing vitamin E status is associated with risk of tuberculosis (TB) disease progression.

Objective

We evaluated the association between baseline plasma concentrations of 3 vitamin E isomers (α-tocopherol, γ-tocopherol, and δ-tocopherol) and TB disease risk.

Methods

We conducted a case-control study nested within a longitudinal cohort of household contacts (HHCs) of pulmonary TB cases in Lima, Peru. We defined cases as HHCs who developed active TB disease ≥15 d after the diagnosis of the index patient, and we matched each case to 4 control cases who did not develop active TB based on age by year and gender. We used univariate and multivariate conditional logistic regression to calculate ORs for incident TB disease by plasma concentrations of α-tocopherol, γ-tocopherol, and δ-tocopherol.

Results

Among 6751 HIV-negative HHCs who provided baseline blood samples, 180 developed secondary TB during follow-up. After controlling for possible confounders, we found that baseline α-tocopherol deficiency conferred increased risk of incident TB disease (adjusted OR: 1.59; 95% CI: 1.02, 2.50; P = 0.04). Household contacts in the lowest tertile of δ-tocopherol were also at increased risk of progression to TB disease compared to those in the highest tertile (tertile 1 compared with tertile 3, adjusted OR: 2.29; 95% CI: 1.29, 4.09; P-trend = 0.005). We found no association between baseline concentration of γ-tocopherol and incident TB disease.

Conclusions

Vitamin E deficiency was associated with an increased risk of progression to TB disease among HHCs of index TB cases. Assessment of vitamin E status among individuals at high risk for TB disease may play a role in TB control efforts.

Introduction

Tuberculosis (TB) remains a global public health challenge with an estimated worldwide incidence of 10.4 million cases in 2015 and 1.8 million TB deaths (1). Studies in humans and animal models have shown that undernutrition is a risk factor for incident TB disease (2–4). There is also increasing evidence of an association between various micronutrient deficiencies and active TB disease (2, 5, 6). However, most prior investigations of the association between specific micronutrients and TB disease have assessed micronutrient status after TB diagnosis. Hence, the causal direction of the association between incident TB disease and micronutrient concentration is unclear. TB disease causes diverse metabolic disturbances, but there is limited evidence on whether underlying micronutrient deficiencies are the result of, or contribute to, TB disease progression.

Although in vitro studies have elucidated some of the mechanisms by which specific micronutrients, such as vitamins A and D, influence host defense against Mycobacterium tuberculosis (Mtb) infection (7, 8), there is limited evidence on the role of other micronutrients in TB pathogenesis. For instance, multiple studies have found lower concentrations of vitamin E among TB patients compared to healthy controls (9–13). Mtb infection induces an oxidative burst (14, 15), but excess oxidative stress leads to tissue injury and impairs the immune system (16, 17). Thus, antioxidant micronutrients, such as vitamin E, may also have an important role in the defense against Mtb. However, to date few prospective studies, to our knowledge, have evaluated the association between pre-existing vitamin E concentration and the risk of developing TB disease.

We evaluated the association between baseline concentrations of 3 isomers of vitamin E (α-tocopherol, γ-tocopherol, and δ-tocopherol) and the risk of progression to TB disease among adults and children with household exposure to a confirmed TB case.

Methods

Ethics statement

The study was approved by the Institutional Review Board of Harvard School of Public Health and the Research Ethics Committee of the National Institute of Health of Peru. All participants or guardians provided written informed consent.

Setting and study design

We conducted a case-control study nested within a prospective longitudinal cohort of household contacts (HHCs) of TB cases in an area of Lima, Peru that included 20 districts and ∼3.3 million residents. Between September 2009 and August 2012, we identified patients aged ≥15 y who were diagnosed with pulmonary TB at any of the 106 participating health centers; these were defined as “index” patients. Within 2 wk of enrolling an index patient, we enrolled all HHCs and screened for TB disease. HHCs with symptoms were referred to a health center for evaluation by sputum smear microscopy and culture. Those who tested positive were treated according to Peru's national guidelines (18). HHCs aged ≤19 y and those with specified comorbidities were offered isoniazid preventive therapy (IPT).

Upon enrollment of HHCs, we obtained the following information: age, gender, height, weight, Bacillus Calmette-Guérin (BCG) vaccination scars, initiation of IPT, alcohol and tobacco use, TB disease history, self-reported diabetes mellitus (DM), other comorbid diseases, and housing information (housing type, number of rooms, water supply, sanitation facilities, lighting, composition of exterior walls and floor and roof materials). In HHCs with no prior history of TB infection or disease, we obtained a tuberculin skin test (TST). We offered HIV testing to HHCs and invited all HHCs to provide a venous blood sample; 60% of those ≥10 y old provided this sample.

HHCs were evaluated for pulmonary and extra-pulmonary TB disease at 2, 6, and 12 mo after enrollment. We considered HHCs to have developed incident secondary TB disease if the diagnosis was confirmed by a physician >14 d after index case enrollment; we considered HHCs to have coprevalent TB disease if they were diagnosed before that time. We defined secondary TB disease in HHCs aged <18 y according to consensus guidelines for diagnosing TB disease in children (19).

For this case-control study, we defined cases to be HIV-negative HHCs who developed incident secondary TB disease during 1 y of follow-up. From among HHCs who did not develop TB disease within 1 y, we randomly selected 4 controls for each case matching on gender and age by year. We considered HHCs to be infected with TB at baseline if they reported previous TB disease or a positive TST or had a TST ≥10 mm at enrollment.

Laboratory methods

Samples were protected from light and stored at –80°C after collection until analysis after study follow-up ended. Laboratory staff were not aware of the specimens’ case-control status; samples were handled identically and assayed randomly. We measured concentration of vitamin E isomers (α-tocopherol, γ-tocopherol, and δ-tocopherol) and vitamin A using HPLC (20). The interassay coefficients of variation for all measured micronutrients were <5%.

Statistical analysis

We defined α-tocopherol deficiency as serum α-tocopherol <5.00 mg/L (21, 22). We also categorized all vitamin E isomers into tertiles. We defined vitamin A deficiency (VAD) as serum retinol <200 µg/L (23). We calculated BMI as weight in kilograms divided by square of height in meters (kg/m²). Among children and adolescent HHCs <20 y, we classified those with BMI z score <–2 as underweight and those with z score >2 as overweight according to WHO age and gender-specific BMI z score tables (24). Among adults ≥20 y, we categorized nutritional status as: underweight (BMI <18.5), normal (BMI 18.5 to <25) and overweight (BMI ≥25). We classified HHCs as heavy drinkers if they reported drinking ≥40 g or ≥3 alcoholic drinks/d based on past-year recall. We derived a socioeconomic status (SES) score using principal components analysis of housing asset weighted by household size (25).

We used a joint model multiple imputation analysis to determine missing baseline covariate data, assuming our data were missing at random. We considered a flat prior for β and for the covariance matrix. We imputed 10 datasets, accounting for the clustered data structure due to our nested case-control matching design. For each dataset, we used conditional logistic regression to evaluate the association between tertiles of vitamin E isomers and risk of incident TB disease. We performed tests for linear trend across tertiles. We also assessed the association between α-tocopherol deficiency and TB disease risk using conditional logistic regression. We included in the multivariate models baseline covariates identified a priori as potential confounders. We previously reported that VAD confers increased risk of TB disease in this cohort (26), and therefore we also adjusted for VAD in our multivariate analysis. We report pooled estimates from imputed datasets according to Rubin's rule (27) using the “jomo” package in R statistical analysis software (28).

We evaluated the interaction between age and vitamin E on the risk of TB disease using the likelihood ratio test. In sensitivity analyses, we repeated the primary analysis among cases (and their matched controls) who were not infected with TB at enrollment, and among cases (and matched controls) diagnosed with TB disease ≥60 d after index case enrollment. Under our method of control selection, ORs approximate risk ratios because secondary TB disease was relatively rare in our study cohort (29).

Data were analyzed using SAS version 9.4 (SAS Institute) and R version 3.1.0 (R Foundation).

Results

Among 6751 HIV-negative HHCs who provided blood samples, we identified 258 cases of active TB disease, of which 66 were coprevalent and 192 were secondary cases. Of the 180 secondary TB cases with viable blood samples, 147 (81.7%) were microbiologically confirmed (Figure 1).

FIGURE 1.

Flow diagram for the case-control study. HHC, household contact; TB, tuberculosis.

Table 1 lists baseline characteristics for cases and controls. Table 2 shows that cases had lower baseline median concentrations of α-tocopherol compared to controls (P = 0.02). In the univariate analysis, we found that HHCs with baseline α-tocopherol deficiency had a 2-fold increased risk of incident TB disease (OR: 2.14; 95% CI: 1.46, 3.13; P < 0.001) (Table 3). Contacts in the lowest tertile of α-tocopherol similarly had nearly 2-fold increased risk of TB disease compared to those in the highest tertile (tertile 1 compared with tertile 3, OR: 1.86; 95% CI: 1.18, 2.92; P = 0.01). We also found an inverse univariate association between incident TB disease and baseline tertiles of δ-tocopherol (tertile 1 compared with tertile 3, OR: 2.04; 95% CI: 1.23, 3.36; P = 0.01). Baseline concentrations of γ-tocopherol were not associated with risk of TB disease (Table 3).

TABLE 1.

Baseline characteristics of incident secondary TB cases and matched controls¹

	Cases (N = 180)	Cases,2	Control (N = 709)	Controls,2
	n (%)	N	n (%)	N
Age categories, y
<10	4 (2.2)	—	16 (2.3)	—
10–19	50 (27.8)	—	200 (28.2)	—
≥20	126 (70.0)	—	493 (69.5)	—
Male	94 (52.2)	—	366 (51.6)	—
BMI categories3		179		706
Underweight	8 (4.5)		6 (0.9)
Overweight	45 (25.1)		299 (42.4)
Normal	126 (70.4)		401 (56.8)
Socioeconomic status		171		698
Lowest tertile	77 (45.0)		228 (32.7)
Middle tertile	66 (38.6)		326 (46.7)
Highest tertile	28 (16.4)		144 (20.6)
Heavy alcohol use4	14 (8.1)	174	64 (9.3)	692
Current smoking	13 (7.4)	176	78 (11.2)	699
Self-reported diabetes	6 (3.4)	179	11 (1.6)	701
Comorbid disease	37 (20.6)	—	175 (24.7)	—
Isoniazid preventive therapy	7 (3.9)	—	108 (15.2)	—
BCG scar	159 (88.3)	—	628 (88.6)	—
History of TB	34 (18.9)	—	55 (7.8)	708
TB infection at baseline	145 (82.4)	176	281 (41.1)	683
Index patient characteristics
Smear positive	156 (86.7)	—	486 (68.7)	707
Cavitary disease	54 (30.2)	179	175 (25.1)	697

¹Values are n (%). BCG, Bacillus Calmette-Guérin; TB, tuberculosis.

²Total number of subjects with data for corresponding variable.

³BMI categories (in kg/m²): Underweight, <18.5; normal, 18.5 to <25; overweight, ≥25.

⁴Self-report of drinking ≥40 g or ≥3 alcoholic drinks/d based on past-year recall.

TABLE 2.

Baseline plasma concentrations of vitamin E isomers among household contacts of index TB patients¹

	Cases (N = 180)	Control (N = 709)	P 2
α-Tocopherol, mg/L	5.79 (4.56, 7.24)	6.06 (5.08, 7.39)	0.02
α-Tocopherol deficient (<5.00 mg/L)	67 (37.2)3	169 (23.8)3	<0.001
γ-Tocopherol, mg/L	0.96 (0.69, 1.28)	0.98 (0.74, 1.28)	0.49
δ-Tocopherol, mg/L	0.23 (0.14, 0.36)	0.26 (0.16, 0.41)	0.13

¹Values are n (%) or median (IQR) unless otherwise indicated. TB, tuberculosis.

²Univariate P values adjusted for matching factors (age and sex).

³ n (%).

TABLE 3.

Plasma vitamin E concentrations and the risk of TB disease among household contacts of index TB¹ patients

	Median	Cases/	Univariate		Multivariate
	(range)	controls, n	OR (95% CI)	P	OR2 (95% CI)	P
α-Tocopherol deficient (<5.00 mg/L)		67/169	2.14 (1.46, 3.13)	<0.001	1.59 (1.02, 2.50)	0.04
α-Tocopherol, mg/L
Tertile 1 (n = 296)	4.55 (0.18, 5.34)	76/220	1.86 (1.18, 2.92)	0.01	1.29 (0.76, 2.18)	0.35
Tertile 2 (n = 297)	6.02 (5.34, 6.86)	51/246	1.01 (0.65, 1.56)	0.97	0.83 (0.51, 1.36)	0.46
Tertile 3 (n = 296)	8.33 (6.87, 35.2)	53/243	1.00		1.00
				P-trend = 0.02		P-trend = 0.36
γ-Tocopherol, mg/L
Tertile 1 (n = 296)	0.64 (0.10, 0.81)	68/228	1.20 (0.80, 1.80)	0.37	0.86 (0.54, 1.38)	0.54
Tertile 2 (n = 297)	0.97 (0.81, 1.16)	52/245	0.85 (0.56, 1.28)	0.43	0.71 (0.44, 1.14)	0.15
Tertile 3 (n = 296)	1.47 (1.17, 5.53)	60/236	1.00		1.00
				P-trend = 0.45		P-trend = 0.55
δ-Tocopherol, mg/L
Tertile 1 (n = 296)	0.13 (0.04, 0.19)	72/224	2.04 (1.23, 3.36)	0.01	2.29 (1.29, 4.09)	0.005
Tertile 2 (n = 297)	0.25 (0.19, 0.34)	59/238	1.35 (0.88, 2.08)	0.17	1.66 (1.01, 2.72)	0.04
Tertile 3 (n = 296)	0.49 (0.34, 3.50)	49/247	1.00		1.00
				P-trend = 0.01		P-trend = 0.005

¹TB, tuberculosis.

²Adjusted for matching factors (age and sex), vitamin A deficiency, BMI categories, socioeconomic status, heavy alcohol consumption (≥40 g or ≥3 alcoholic drinks/d), tobacco use, isoniazid preventive therapy, TB history, comorbid disease, self-reported diabetes, and index patient smear status; 67 household contacts had missing information on ≥1 covariate.

When we adjusted for matching factors, VAD, BMI, SES, heavy alcohol use, tobacco use, IPT, TB history, comorbid disease, self-reported DM, and index patient smear status, we found that baseline α-tocopherol deficiency remained associated with increased risk of TB disease [adjusted (a)OR: 1.59; 95% CI: 1.02, 2.50; P = 0.04] (Table 3). A low concentration of δ-tocopherol also remained associated with increased TB disease risk (tertile 1 compared with tertile 3, aOR: 2.29; 95% CI: 1.29, 4.09; P = 0.005). We found no association between baseline tertiles of γ-tocopherol and incident TB disease in the adjusted analysis (Table 3).

Given the altered pathophysiology of TB disease among adults and children, we considered whether the association between vitamin E and incident TB was similar among children and adolescents and adults. We found no significant difference in the association between α-tocopherol deficiency on TB disease by age group (P-interaction = 0.34). Similarly, we found no evidence that age modified the association between γ-tocopherol and δ-tocopherol concentrations and TB disease risk (P-interaction = 0.40 and 0.39, respectively).

Given that TB disease is often accompanied by a range of metabolic disturbances, we considered the possibility that some of the HHCs with low vitamin E concentration may have had early undetected TB disease at the time of their baseline evaluation. To assess this, we evaluated the association between vitamin E concentrations and TB disease among HHCs who were TB uninfected at baseline. After adjusting for possible confounders, we found that contacts in the lowest tertile of α-tocopherol had a 3-fold increased risk of incident TB disease (tertile 1 compared with tertile, 3 aOR: 3.18; 95% CI: 0.89, 11.38), although this was not significant at the 0.05 threshold (P = 0.07) (Table 4). The association between δ-tocopherol concentration and incident TB disease was no longer statistically significant (Table 4). In a second sensitivity analysis, when we excluded cases (and matched controls) diagnosed <60 d after index case enrollment, we found that δ-tocopherol remained inversely associated with TB disease (tertile 1 compared with tertile 3, aOR: 2.72; 95% CI: 1.19; 6.22; P-trend = 0.02). However, α-tocopherol concentrations were no longer associated with incident TB (Table 5). Our study was not powered to detect an interaction between vitamins A and E. We did not stratify by IPT use because few HHCs received IPT.

TABLE 4.

Plasma vitamin E concentrations and risk of TB disease among subset of initially TB uninfected household contacts¹

	Multivariate
	OR (95% CI) N = 433	P
α-Tocopherol, mg/L
Tertile 1 (n = 144)	3.18 (0.89, 11.38)	0.07
Tertile 2 (n = 145)	3.16 (0.92, 10.85)	0.07
Tertile 3 (n = 144)	1.00
		P-trend = 0.12
α-Tocopherol deficient (<5.00 mg/L)	1.69 (0.72, 3.97)	0.23
γ-Tocopherol, mg/L
Tertile 1 (n = 152)	1.83 (0.64, 5.25)	0.26
Tertile 2 (n = 145)	1.18 (0.39, 3.63)	0.77
Tertile 3 (n = 136)	1.00
		P-trend = 0.23
δ-Tocopherol, mg/L
Tertile 1 (n = 128)	1.91 (0.65, 5.57)	0.24
Tertile 2 (n = 142)	2.70 (1.01, 7.20)	0.05
Tertile 3 (n = 163)	1.00
		P-trend = 0.19

¹Adjusted for matching factors (age and sex), vitamin A deficiency, BMI categories, socioeconomic status, heavy alcohol consumption (≥40 g or ≥3 alcoholic drinks/d), tobacco use, isoniazid preventive therapy, TB history, comorbid disease, self-reported diabetes, and index patient smear status. TB, tuberculosis.

TABLE 5.

Plasma vitamin E concentrations and risk of TB disease diagnosed ≥60 d after index case enrollment¹

	Cases/	Multivariate
	controls, n	OR (95% CI) N = 454	P value
α-Tocopherol, mg/L
Tertile 1 (n = 151)	33/118	0.81 (0.36, 1.82)	0.61
Tertile 2 (n = 152)	29/123	0.81 (0.41, 1.60)	0.54
Tertile 3 (n = 151)	29/122	1.00
			P-trend = 0.60
α-Tocopherol deficient (<5.00 mg/L)	30/98	1.00 (0.52, 1.94)	1.00
γ-Tocopherol, mg/L
Tertile 1 (n = 151)	31/120	0.73 (0.37, 1.44)	0.37
Tertile 2 (n = 152)	28/124	0.67 (0.33, 1.37)	0.27
Tertile 3 (n = 151)	32/119	1.00
			P-trend = 0.39
δ-Tocopherol, mg/L
Tertile 1 (n = 151)	38/113	2.72 (1.19, 6.22)	0.02
Tertile 2 (n = 152)	27/125	1.58 (0.85, 2.93)	0.15
Tertile 3 (n = 151)	26/125	1.00
			P-trend = 0.02

¹Adjusted for matching factors (age and sex), vitamin A deficiency, BMI categories, socioeconomic status, heavy alcohol consumption (≥40 g or ≥3 alcoholic drinks/d), tobacco use, isoniazid preventive therapy, TB history, comorbid disease, self-reported diabetes, and index patient smear status. TB, tuberculosis.

Discussion

We found that deficiency of the primary bioactive form of vitamin E, α-tocopherol, conferred increased risk of progression to TB disease among HHCs of index TB cases. Concentrations of δ-tocopherol were also inversely associated with risk of incident TB disease. Conversely, we found no association between γ-tocopherol concentration and risk of progression to TB disease.

Numerous studies have documented low vitamin E concentration among patients diagnosed with TB disease compared to healthy controls (9–13). Although Plit et al. (13) found that vitamin E concentration remained low after TB treatment, few other studies have reported on vitamin E concentration among patients treated for TB disease. To our knowledge, only one published study to date, conducted among HIV-infected patients, has prospectively examined the effect of underlying vitamin E concentration on the risk of developing TB disease. In a multinational cohort of HIV patients, Tenforde et al. (30) found that low vitamin E levels prior to initiating ART did not increase the risk of TB disease during follow-up. One prior study found that vitamin E supplementation had no effect on the overall incidence of TB disease during follow-up (31). Other studies on the role of vitamin E supplementation in TB treatment have only assessed vitamin E with other vitamins; these studies found no effect of supplementation on TB treatment outcomes (5).

Vitamin E is a potent antioxidant that scavenges free radicals and protects FAs, a major structural component of cell membranes, from peroxidation (32). Host innate immune response to infection includes stimulation of neutrophils to produce reactive oxygen species (ROS), which leads to pathogen destruction but also contributes to immune cell death, tissue injury, and deleterious chronic inflammation (16, 17). Antioxidants such as vitamin E thus play a role in protecting immune cells from oxidative damage (17, 32, 33). Vitamin E is also recognized to influence immune function in other ways. Animal models indicate that vitamin E deficiency impairs cell-mediated immune responses (34), whereas animal and human studies have demonstrated that vitamin E supplementation stimulates T-cell differentiation and proliferation (34–36), particularly enhancing Th1-mediated immune responses (36, 37). Experimental studies in animals have further shown that vitamin E supplementation protects against bacterial and viral infections (38–41). However, several studies of the effect of vitamin E supplementation on infectious diseases, particularly respiratory tract infections, in humans have not consistently demonstrated a benefit (42–49). Some studies have reported an adverse effect of vitamin E supplementation on the incidence and severity of respiratory infections (50, 51).

Few studies to date have specifically evaluated the role of vitamin E in immune responses to Mtb infection. In vitro studies have shown that part of the host immune defense against Mtb involves ROS production (14, 15), and epidemiologic studies have demonstrated elevated markers of oxidative stress in TB patients compared to healthy controls (9, 10). Given its role as an antioxidant, vitamin E may thus attenuate the adverse effects of increased oxidative stress on cells and the immune system after TB infection. Furthermore, vitamin E stimulates Th1 immune response (36, 37), which is recognized as a critical part of an effective immune response to tuberculosis (52); hence, vitamin E may also directly modulate the immune response to TB infection.

We also found that 3 isomers of vitamin E are differentially associated with risk of progression to TB disease; low baseline concentrations of α- and δ-tocopherol increased TB disease risk, whereas γ-tocopherol concentration was not associated with the risk of incident TB. α-Tocopherol and γ-tocopherol are the most abundant isomers of vitamin E, although tissue concentrations of α-tocopherol are 10-fold higher than those of γ-tocopherol; thus α-tocopherol has greater antioxidant capacity than γ-tocopherol (53). If the observed association between low vitamin E and increased TB disease risk is mediated by its antioxidant properties, this may partially explain our findings that α-tocopherol deficiency influences TB disease risk whereas γ-tocopherol status does not. In vitro and mice studies have also shown that α-tocopherol has anti-inflammatory action in the lungs, reducing leukocyte migration and recruitment, whereas γ-tocopherol mediates the opposing, proinflammatory immune response in the lungs (53, 54). Considering that classic pulmonary TB results from an exaggerated immune response to Mtb infection (55), low concentrations of α-tocopherol may result in a net proinflammatory state in the lungs that contributes to development of TB disease after exposure. Although δ-tocopherol exists in even lower concentration than α- and γ-tocopherol, we found that concentrations of δ-tocopherol were also inversely associated with TB disease risk. However, the clinical relevance of this finding remains unclear. Given the potential heterogenous effects of vitamin E, further research is needed to elucidate the exact mechanisms by which vitamin E influences TB pathogenesis.

We observed a wide range of α-tocopherol concentrations with deficient concentrations among 37% of cases and 24% of controls. Although overt clinical manifestations of vitamin E deficiency are rare in humans, reported estimates of the prevalence of α-tocopherol deficiency vary in different populations, ranging from 1% in the US-based National Health and Nutrition Examination Survey (NHANES) (56) to 9% in a cohort of Brazilian children (57), 20% among adolescent Hispanics (58) and >60% in one small sample of elderly Greeks (59). There is conflicting evidence on whether dietary intake correlates with plasma vitamin E concentrations (22), and there are limited data on the expected distribution of α-tocopherol concentrations in vulnerable populations at high risk for TB disease such as our study population. It should be noted that the cutoff for α-tocopherol deficiency is based on concentrations associated with normal hydrogen-peroxide-induced hemolysis of red blood cells (22), and thus it is unknown whether α-tocopherol concentrations that influence TB pathogenesis differ from this established cutoff.

Our study is limited by the relatively short period of follow-up considering the slow pathogenesis of tuberculosis. Although 40% of HHCs did not provide a blood sample, most of these were children and adolescents; differences in baseline characteristics were therefore due to age-related risk factors. It is unlikely that the differences among individuals who gave blood biased our observed association between vitamin E and TB disease risk. We also did not assess serum lipids, which may have influenced tocopherol concentrations since vitamin E is transported in lipoproteins (22). However, serum α-tocopherol measurements provide an adequate estimate of vitamin E in individuals with normal serum lipids (22). While low serum lipids may lead to an underestimation of α-tocopherol concentrations, we found low prevalence of undernutrition based on BMI in our study population, and we further adjusted for BMI in our analysis. We examined several correlated exposures and categorized α-tocopherol in different ways, which may have increased the risk of type 1 error. However, the effect estimates in our main analysis were consistent across tests, which lends support to our conclusion about the association between α-tocopherol and TB disease. Our study may also not have been powered to detect an association between γ-tocopherol and incident TB.

Although we attempted to ascertain vitamin E concentration prior to development of TB disease, we cannot rule out that some HHCs had undiagnosed TB at enrollment, which may have led to relatively lower vitamin E concentration compared to controls. To address this, we conducted 2 sensitivity analyses, one of which stratified by baseline infection status and one of which assessed the effect of vitamin E concentrations in people diagnosed after 2 mo. In the first analysis, we found that among HHCs who were TB uninfected at baseline, those in the lowest tertile of α-tocopherol concentration had a 3-fold increased risk of TB disease, although this did not reach statistical significance. In the final analysis, δ-tocopherol remained inversely associated with TB disease whereas α-tocopherol deficiency did not predict an increased risk of TB disease diagnosed later during follow-up. However, the smaller number of HHCs with α-tocopherol deficiency diagnosed later may have limited our power and biased the results towards the null. Our findings raise the possibility that low vitamin E, particularly α-tocopherol, concentration among people at high risk for TB may serve as a marker for those at increased risk of developing TB disease in the near future who could be targets for early intervention.

In conclusion, we found that low concentrations of α-tocopherol and δ-tocopherol isomers of vitamin E were associated with increased risk of TB disease among HHCs of TB cases. These findings raise the possibility that assessing vitamin E concentration among individuals at high risk for TB disease may play a role in TB control efforts.

Support for this project was provided by the National Institute of Allergy and Infectious Diseases grants U19 AI076217 to MBM and U01 AI057786 to MCB. Funding support was provided by National Institute on Drug Abuse training grant T32 DA013911 and NIH grant: Brown Initiative in HIV and AIDS Clinical Research for Minority Communities, #5R25MH083620 to OA.

Author disclosures: OA, MFF, C-CH, JTG, RC, ZZ, MCB, ERS, CC, RY, LL, and MBM, no conflicts of interest.

Abbreviations used

BCG

Bacillus Calmette-Guérin

DM

diabetes mellitus

HHC

household contact

IPT

isoniazid preventive therapy

Mtb

Mycobacterium tuberculosis

ROS

reactive oxygen species

SES

socioeconomic status

TB

tuberculosis

TST

tuberculin skin test