INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The pathology of fibrosing alveolitis (FA) is characterized by persistent alveolar inflammation and interstitial pulmonary fibrosis, processes mediated by proinflammatory and profibrotic cytokines (1). Although the basis of an individual's susceptibility to fibrosing alveolitis is unknown, a genetic component is suggested by approximately 3% of cases of cryptogenic fibrosing alveolitis (CFA) being familial, with pathology that is indistinguishable from nonfamilial forms (2).

A number of cytokine gene polymorphisms have been described within the interleukin-1 (IL-1) gene cluster on chromosome 2 (3) and in the TNF- gene (TNF-A) on chromosome 6 (4), which are associated with susceptibility to inflammatory, autoimmune, and infectious diseases. The three known IL-1 genes are arranged in close proximity on chromosome 2q13 (3). The IL-1A and IL-1B genes encode agonist proteins, the proinflammatory cytokines IL-1 and IL-1, with well-known roles in inflammation and innate immunity. The third known gene in this cluster (IL-1RN) produces a related protein, the IL-1 receptor antagonist (IL-1ra), which binds the IL-1 signaling receptor (IL-1R Type 1) but does not elicit a response (5). A single base variation occurs at position +2018 in IL-1RN (6), with the rarer allele, [IL-1RN (+2018) allele 2], being implicated in inflammatory diseases; for example, coronary artery disease (7). A polymorphism in the TNF locus has been identified in the promoter region of TNF-A (8) and the rarer allele [TNF-A (-308) allele 2] implicated in inflammatory diseases, including asthma (9) and chronic bronchitis (10).

We hypothesized that susceptibility to FA may be determined by the occurrence of these disease-associated alleles, IL-1RN (+2018) allele 2 and TNF-A (-308) allele 2, and have tested this hypothesis in two independent FA case-control cohorts, one English and one Italian.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patient Populations

The English case-control sample comprised 88 consecutively recruited cases and age- and sex-matched controls recruited from the same primary care physician lists as the index cases, and who were part of a larger case-control study previously reported from Nottingham, England (11). Although the original study included more than 200 cases, blood for DNA extraction was stored only from later stages of the study. Of the 88 available cases, 51 were prevalent at the beginning of recruitment into the study, and 37 were incident during an 18 mo recruitment period. The Italian cohort comprised 61 cases collected from the Istituto di Patologia Medica of the University of Bologna, Italy, from a study described previously (12), and 103 controls obtained by taking 414 consecutive ethnically matched, healthy blood donors from the Bologna area and, blinded to genotype, taking the upper quartile of age distribution (> 54 yr of age) to obtain an appropriate age distribution for the controls.

FA was diagnosed either from an open lung biopsy or by the presence of basal inspiratory pulmonary crackles, bilateral interstitial lung shadowing on chest radiograph, and restrictive lung function (%FEV₁/ FVC > 70% together with FVC or DL_CO < 80% of predicted). In the absence of restrictive lung function patients were included only if there were pathognomonic changes of FA on a high-resolution computed tomography (HRCT) scan and if there was no evidence of another parenchymal process caused by occupational or domestic exposures, sarcoid, bronchiectasis, or connective tissue disease (11). Of the English cases, 25 of 88 had undergone lung biopsy and a further 42 of 88 had had a HRCT scan. FA was diagnosed by open lung biopsy in 11 of 61 Italian cases, 56 of 61 cases had had HRCT scans and all patients fulfilled the exclusion criteria as previously detailed (11, 12).

Ethics approval for the English study was granted by the Nottingham City Hospital Medical Ethics Committee, for the Italian study by the Institutional Review Board of the University Hospital, Bologna, and for the genetic analyses by the South Sheffield Research Ethics Committee.

Genotyping

The IL-1RN (+2018) polymorphism is a single base variation (C/T) at +2018 in exon 2 of the IL-1RN gene (6). The TNF-A (-308) polymorphism is a single base variation (G/A) at -308 in the promoter region (8). Genotyping was performed blind to case status either by restriction-enzyme digest of PCR products, as previously described (6, 8), or by TaqMan allelic discrimination (13). TaqMan probes were purchased from ABI-PE (Warrington, UK) and were as follows: for IL-1RN (+2018) Probe 1: 5'-C(·FAM)AA CCA ACT AGT TGC TGG ATA CTT GCA AG(·TAMRA)-3' Probe 2: 5'-C(·TET)AA CCA ACT AGT TGC CGG ATA CTT GCA AG(·TAMRA)-3' Forward Primer: 5'-AAG TTC TGG GGG ACA CAG GAA G-3' Reverse Primer: 5'-ACG GGC AAA GTG ACG TGA TG-3'. For TNF-A (308) Probe 1: 5'-A(·TET) CC CCG TCC CCA TGC CC (·TAMRA)-3' Probe 2: 5'-A(·FAM) AC CCC GTC CTC ATG CCC C (·TAMRA)-3' Forward Primer: 5'-GGC CAC TGA CTG ATT TGT GTG T-3' Reverse Primer: 5'-CAA AAG AAA TGG AGG CAA TAG GTT-3'. The English samples were genotyped by the PCR-RFLP method and the Italian samples by the TaqMan method; 10% of samples were selected at random and tested by both techniques as quality-control. Genotype frequencies in patients and control subjects in both populations were not significantly different from those predicted under Hardy Weinberg equilibrium.

Statistical Analysis

Demographic details were compared between and within case-control populations by Z test and chi-square tests as appropriate and odds ratios (ORs) estimated by logistic regression using STATA (version 5) software. ORs for the matched case-control study were estimated by conditional logistic regression, comparing carriers of allele 2 for IL-1RN (+2018) or TNF-A (308), in total or as 1,2 heterozygotes or 2,2 homozygotes, with 1,1 homozygotes. A similar analysis was performed for the unmatched case-control study (Italian) using logistic regression. Multiplicative terms were used to look for evidence of interaction between genotype effects within populations, comparing carriage of allele 2 for either gene with 1,1 homozygotes, and to look for effect modification by age or sex. To estimate a summary OR for IL-1RN and TNF-A effects, across both study populations and combining matched and unmatched data, we used a fixed effects meta-analysis method to derive a weighted average of the log odds, the weight being the inverse of the variance of the log OD. The significance test for the summary OD was a Z test calculated as the log of the summary OD divided by its standard error (14).

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Demographic information for the two study populations and pulmonary function parameters for the FA cases are shown in Table 1. There were no significant demographic differences between or within the English and Italian cases and controls, but the Italian cases had significantly lower values for FVC and DL_CO than the English cases (p < 0.0001, Z test). The distribution and frequency of IL-1RN (+2018) and TNF-A (308) genotypes in FA cases and controls are shown in Table 2. The frequency of IL-1RN allele 2 was increased in the English FA cases, of whom 44.3% carried allele 2 compared with 31.8% of controls (Table 2). Most of these subjects were heterozygous for the mutation, but of the 12 who were homozygous for allele 2, 10 were cases. The OD for FA in those homozygous for IL-1RN (+2018) allele 2 was 10.2 (95% CI, 1.26 to 81.4; p = 0.03), for heterozygotes it was 1.43 (95% CI, 0.70 to 2.92; p = 0.3), and for carriage of allele 2 compared with 1,1 homozygotes it was 1.85 (95% CI, 1.15 to 3.06; p = 0.075) (Table 3). In the Italian cohort the frequency of the IL-1RN (+2018) allele 2 in the patients with FA was 57.4% compared with 36.0% of control subjects (Table 2) and the OD for FA for homozygotes was 2.54 (95% CI, 0.68 to 9.5; p = 0.2), for heterozygotes it was 2.38 (95% CI, 1.21 to 4.67; p = 0.012) and for carriage of allele 2 versus 1,1 homozygotes it was 2.40 (95% CI, 1.26 to 4.60; p = 0.008) (Table 3). Estimated ODs obtained by combining these estimates across the two studies were 3.77 (95% CI, 1.24 to 11.5; p = 0.01) for allele 2 homozygotes, 1.87 (95% CI, 1.15 to 3.06; p = 0.006) for heterozygotes, and 2.12 (95% CI, 1.33 to 3.38; p = 0.001) for carriage of allele 2 compared with 1,1 homozygotes. In view of previously described linkage disequilibrium within the IL-1 gene cluster and the association of IL-1B gene polymorphisms with other inflammatory diseases, we subsequently examined two informative polymorphisms at positions 511 and +3954 in the IL-1B gene (3). There was no significant difference in the genotype distributions at these loci between patients and control subjects in either population (data not shown).

Carriage of TNF-A (308) allele 2 was also increased in both case populations, occurring in 39.8% of the English cases compared with 27.3% of controls and 34.4% of Italian cases compared with 15.6% of controls (Table 2). Few homozygotes were identified for TNF-A (-308), none in the Italian cohort and only three patients and two control subjects in the English cohort. For carriage of allele 2 versus 1,1 homozygotes, the OD for FA was 1.85 (95% CI, 0.94 to 3.63; p = 0.075) in the English cohort and 2.50 (95% CI, 1.14 to 5.47; p = 0.022) in the Italian cohort (Table 3). The combined estimated OD for carriage of allele 2 was 2.10 (95% CI, 1.26 to 3.50; p = 0.002).

There was no evidence of interaction between genotypes or effect modification because of age or sex for either IL-1RN (+2018) or TNF-A (308) allele 2 in either population.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of this study suggest carriage of allele 2 for IL-1RN (+2018) or TNF-A (308) is significantly associated with increased risk of development of FA. This effect was observed in two independent populations, one English and one Italian. As far as we are aware, this is the first demonstration of genetic factors that may contribute to individual susceptibility to FA. Our sample sizes were determined by the numbers of samples available from cases of FA in existing archives rather than by estimates of numbers required to achieve statistical power, and samples were drawn from matched and unmatched study designs. We were therefore unable to pool data from these studies directly, but we were able to produce a single best estimate of combined OD by a simplified meta-analysis technique. Cases included in the study were pulmonary physician-diagnosed and fulfilled normal clinical criteria for the diagnosis of FA, with the great majority having had a HRCT scan and biopsy information available in a quarter of cases. Misclassification of case and control status cannot be completely excluded but would, if anything, dilute the observed genotypic associations.

The essential role of cytokines in mediating both inflammatory and fibrotic processes in the lung is well established (1). IL-1 and TNF-, key proximal cytokines, are produced predominantly by alveolar macrophages and can, in turn, stimulate the production of other proinflammatory and profibrotic cytokines. IL-1ra, the naturally occurring antagonist of IL-1, counteracts the proinflammatory functions of IL-1 and the balance between IL-1 and IL-1ra is seen as a crucial ratio in destructive inflammatory disease (5). IL-1ra production in lung tissue of patients with FA has been localized to alveolar macrophages (15), hyperplastic type II pulmonary epithelial cells, and fibroblasts (16). Alveolar macrophages from patients with FA produce higher levels of IL-1ra in vitro than do those from healthy control subjects (15) and increases in IL-1ra levels have been documented in bronchoalveolar lavage fluid from patients with FA (16). There is evidence for a protective role of exogenous IL-1ra in a number of animal models of acute lung injury. Overexpression of IL-1ra, targeted to distal airway epithelium under the transcriptional control of the surfactant protein-C gene promoter, has recently been shown to give partial protection from IL-1 induced airway inflammation and injury (17). Moreover, Piguet and colleagues (18) have shown that exogenous administration of IL-1ra via an intraperitoneal pump can partially reverse changes of pulmonary fibrosis and to a lesser extent BAL cellularity in bleomycin-induced fibrosis in mice.

The functional effects of IL-1RN (+2018) allele 2 are uncertain. There is evidence for the association of allele 2 both with reduced IL-1ra protein production in inflammatory bowel disease samples ex vivo (19) and with increased IL-1ra production upon stimulation of peripheral blood monocytes in vitro (20). Such effects may well be stimulus- and cell-type specific, and further study is clearly required. However, the association of IL-1RN allele 2 with a number of other inflammatory diseases (3, 7), in addition to FA, implies a proinflammatory effect of IL-1RN allele 2.

TNF- has also been implicated in the pathogenesis of FA, with increased expression of TNF- protein in lung tissue of patients with FA (21). The murine TNF- gene has been overexpressed under the control of the human surfactant protein SP-C promoter in transgenic mice, and the pulmonary pathology showed a striking resemblance to human FA, with a leukocytic alveolitis, type II cell hyperplasia and extensive fibrosis (22). The TNF-A (308) allele 2 promoter polymorphism has previously been associated with a number of inflammatory lung diseases, including asthma (9) and chronic bronchitis (10), and it was recently shown to be a much more powerful transcriptional activator of the TNF-A gene than the wild-type allele 1 (23).

Identification and understanding of the role of genetic risk factors for FA is currently at a very early stage of development, but it may lead to significant therapeutic opportunities in the management of patients with FA, and to the identification of those who are at increased risk of developing pulmonary fibrosis in the future. The latter category may include those in high-risk occupations, those with other diseases associated with an increased risk of development of pulmonary fibrosis such as rheumatoid arthritis or systemic sclerosis, and those exposed to other recognized risk factors for pulmonary fibrosis such as amiodarone or bleomycin therapy. Insights into the pathophysiology of FA gained from genetic studies are likely to be particularly important, by identifying key molecular regulators of the proinflammatory and profibrotic processes in the lung and thus potential therapeutic targets at a molecular level. Both the genetic associations identified in this study are thus of potential pathophysiologic and therapeutic relevance to the understanding of a progressive and currently untreatable disease.