On the basis of screening studies on healthy volunteers, coeliac disease is now recognised to affect 1% of the population.¹ The current optimal serological testing for coeliac disease may be immunoglobin (Ig)A endomysial, IgA tissue transglutaminase or a combination of the two, with a positive predictive value in excess of 95% provided IgA deficiency is excluded. Patients with positive serology then require an endoscopic duodenal biopsy to confirm the presence of villous atrophy and thus a firm diagnosis of coeliac disease. A link between coeliac disease and tuberculosis was first suggested as far back as 1952, but there has been a paucity of data exploring this relationship other than case reports and case series.² The initial focus was on gastrointestinal tuberculosis being one of the differential diagnoses for villous atrophy of the small bowel. However, in 1988 a report from Birmingham, UK, suggested that patients with coeliac disease seemed to have an increased prevalence of tuberculosis (6/76 patients with coeliac disease reporting a history of tuberculosis).³

The Swedish study reported in this edition of Thorax explores this overlap/relationship in greater detail. Ludvigsson et al⁴(see p 23) investigated the risk of tuberculosis in 14 335 patients with coeliac disease and made comparisons against 69 888 matched individuals in a general population-based cohort study. Patients with coeliac disease were reported to have an increased risk of subsequent tuberculosis (HR 3.74, 95% CI 2.14 to 6.53, p<0.001). This epidemiological study is fairly robust and reflects the strengths of medical data collection in Sweden as a result of the Swedish National In-Patient Register. The findings of this study are, however, limited by the fact that only 24 actual patients with coeliac disease (24/14 335) developed tuberculosis. In the context of a large database study, this observation is a positive statistical finding but identifies a small increase in absolute risk. A caveat, however, is that patients with tuberculosis were ascertained by hospitalisation for this condition, so that some clinically relevant cases might be missed. Importantly, Ludvigsson et al⁴ have looked for an association of coeliac disease with only one respiratory condition rather than many. In the past others have explored the positive associations between coeliac disease and a variety of respiratory conditions such as sarcoidosis and pulmonary fibrosis.^{5,6,7,8,9,10,11,12} This study would thus have been strengthened by study of other infectious and inflammatory lung diseases, such as pneumonia for example, that may not be associated. This would support the view of a mechanistic basis for an association between coeliac disease and tuberculosis, and also persuade readers that the observed association is not simply due to ascertainment bias, that is, that individuals admitted to hospital for one condition, coeliac disease, may then acquire additional diagnoses more frequently than their control population. The authors make rather a leap of faith from finding an association to assuming it is causal and then speculating on a mechanism.

If their finding is indeed a causal relationship, why should patients with coeliac disease be more susceptible to tuberculosis? The authors speculate that this may be related to malabsorption of vitamin D. Vitamin D was recently shown to upregulate antimicrobial peptides such as LL-37, which are implicated in intracellular killing of Mycobacterium tuberculosis.¹³ Vitamin D can also induce nitric oxide synthase in macrophages, which again may suppress growth of Mtuberculosis.¹⁴ An alternative possibility is that the HLA-DQ2 haplotype, which is strongly linked to coeliac disease, is associated with specific alleles of HLA class I and II molecules as well as with genes for tumour necrosis factor (TNF) α and complement factors C2 and C4. TNF α polymorphisms may regulate host responses in tuberculosis, although coeliac disease is associated with the TNF-308A promoter polymorphism that may protect against tuberculosis.¹⁵ The C2 and C4 molecules are also polymorphic and, as the authors suggest, perhaps a variant C2 allele may promote tuberculosis infection in a subgroup of patients.¹⁶ Further investigation of these complex haplotypes could provide biological evidence of the relationship between coeliac disease and tuberculosis, or further insight into the pathogenesis of either condition. At a clinical level, a serological study on the prevalence of coeliac disease in patients with tuberculosis might also serve to establish whether there is a relationship between these two disease processes.

How does this study change clinical practice? The findings would be unlikely to lower the threshold for investigating patients with coeliac disease for tuberculosis, given the relatively small number of actual cases reported. What can perhaps be inferred from this study is that, if a relationship between coeliac disease and tuberculosis is confirmed in further studies, then clinicians should consider serological testing for coeliac disease in those patients with tuberculosis who have gastrointestinal symptoms or apparent drug resistance.¹⁷