INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: allele; genotyping; silicosis; tumor necrosis factor-

Silicosis is an occupational fibrotic lung disease resulting from inhalation of microscopic crystalline silica particles (1). There is a clear relationship between increased cumulative silica exposure and increased risk of silicosis (2, 3). There is also an intensity effect, in that even brief exposure to extremely high concentrations of silica can result in acute silicosis, pathologically distinct from the chronic form of disease that results from exposure to lower concentrations of silica over a period of many years (4). In addition, there appears to be quite pronounced individual variation in the severity of silicosis, suggesting that genetic factors may influence susceptibility to this disease (5).

Silicotic inflammation and fibrosis are initiated when alveolar phagocytes contact or ingest silica particles and become activated, but are unable to further degrade the particles (8). Cytokines and other fibrogenic mediators are released into the local tissues and stimulate an inflammatory response, followed by fibroblast proliferation and collagenization (8). There is a particularly strong body of evidence from experimental animal studies to suggest that tumor necrosis factor (TNF)- plays an important causal role in the inflammatory and fibrotic response to acute challenge with silica (9).

Single nucleotide polymorphisms (SNPs) in the TNF promoter region have been associated with susceptibility to a number of fibrotic, inflammatory, and infectious diseases other than silicosis (13), including pulmonary conditions (22). A small case-control study of 18 Belgian subjects with coal worker's pneumoconiosis (CWP) reported an association of this condition with the 308A TNF- promoter allele (26). Reanalysis of the published data with correction for the small number of subjects shows that this association was of borderline significance (p = 0.08), but the results were notable for the high prevalence (50%) of a relatively uncommon allele, 308A, among cases of CWP, in most of which the disease was in an early stage. This raised the possibility of a large fraction of pneumoconiosis being attributable to TNF- promoter polymorphism.

The main aim of the present study was to investigate the role of SNPs in the TNF- promoter region in determining susceptibility to silicosis in black South African gold miners. The prevalence of chronic silicosis is high among South African miners (27), affecting more than 20% of men aged 40 yr or more at the mines that were the focus of this study (28, 29), and carrying the major health consequences of greatly increased susceptibility to tuberculosis (TB) (28, 30) and increased risk of death from respiratory causes (31).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Case and Control Selection and Specimen Collection

Case patients and control subjects were recruited from a single workforce in the Free State, South Africa. Case patients were identified from company records of compensation claims for silicosis uncomplicated by TB during the period from 1994 to 1997, and their selection was based entirely on annual screening radiographs. Potential control subjects were randomly selected from three separate age groups (30 to 39 yr, 40 to 49 yr, and 50 yr or older) of the remaining employees. Silicosis is strongly associated with age and years worked as a miner (29). Control subjects were randomly selected from each age group on the basis of age-specific silicosis prevalence data (28), to enable an approximate frequency match with the age distribution of cases of the disease.

Miniradiographs (10 × 10 cm), taken annually for screening purposes, were graded for silicosis with the International Labour Office (ILO) system (32). Radiographs were graded independently by two readers without reference to personal details or case status, with final grades reached by consensus in cases of initial disagreement. The ILO system was designed for use with standard-sized films, but the use of miniradiographs for detecting silicosis was validated in a separate study involving the same workforce as in the present study (29). Radiographs from all potential case subjects were graded as showing silicosis of ILO grades 1/ 1 or above by both readers. Cases were graded into the three categories of early, moderate, and severe disease, corresponding to ILO grades 1/1 to 1/2, 2/ 1 to 2/3, and 3/ 2 to 3/3, respectively. Control subjects with radiographs graded by both readers as showing no evidence of silicosis (ILO grade 0/0) were asked to participate. Written consent or witnessed verbal consent was obtained from all participants. Approval for the study was obtained from the Ethics Committees of the London School of Hygiene and Tropical Medicine and the Ernest Oppenheimer Hospital, Welkom.

A sample of blood (8 ml) was collected into heparin from each consenting subject, and DNA was prepared through two rounds of phenol:chloroform:isoamyl alcohol (25:24:1) and one round of chloroform extraction, followed by ethanol precipitation and dissolution of the pellet in 100 µl of sterile water.

Polymerase Chain Reaction of TNF- Fragments

The polymerase chain reaction (PCR) and nested PCR were performed on DNA samples, using 1 µl of DNA (or 1 µl of preamplified DNA); 50 mM KCl; 10 mM Tris-HCl; 0.1% Triton X-100; 2.5 mM MgCl₂; 200 µM each of deoxyadenosine, deoxyguanosine, deoxythymidine, and deoxycytosine triphosphate; 0.5 µM of each primer; and 1 unit of Taq polymerase with a total volume of 20 µl. DNA was heated to 95° C for 3 min; and subjected to 20 cycles of PCR at 95° C for 1 min, at 60° C for 1 min, and at 72° C for 1 min, with primers described by McGuire and colleagues (13). On some samples, which had unexpectedly poor DNA quality, a nested PCR was performed under the same conditions but with the following primers: TNF-423Fwd: 5'-GC CCCTCCCAGTTCTAGTTC-3'; and TNF-225Rev: 5'-GATACCCCT CACACTCCCCA¹-3'.

TNF- Allele-Specific Oligonucleotide Probing

PCR products were denatured for 2 min at 94° C, and genotyping was performed with sequence-specific oligonucleotide (SSO) probing. Oligonucleotide probes of 18-mer were designed for each allele at each polymorphic site. These were as follows: TNF--376G: 5'-TGTCTG GAAGTTAGAAGG-3'; TNF--376A: 5'-TGTCTGGAAATTAGAA GG-3'; TNF--308G: 5'-AGGGGCATGGGGACGGGG-3'; TNF--308A: 5'-AGGGGCATGAGGACGGGG-3'; TNF--238G: 5'-GAA TCGGAGCAGGGAGGA-3'; TNF--238A: 5'-GAATCAGAGCAG GGAGGA-3'.

Probes were 3' end-labeled with digoxigenin (Boehringer Mannheim, Mannheim, Germany) and used at 2 pmol/ml. Hybridization was performed in 5-ml volumes of 3M tetramethylammonium chloride (TMAC), 50 mM Tris, 0.1% sodium dodecyl sulfate (SDS), and 2 mM ethylene diamine tetraacetic acid (EDTA) solution at 53° C for 1.5 h, after which excess probe was removed by washing twice for 10 min each in 2× saline-sodium phosphate-EDTA/0.1% SDS at room temperature and twice for 10 min each in 3M TMAC buffer at 56° C. Detection of hybridization was done with antidigoxigenin Fab antibody (alkaline phosphatase conjugated) and chemoluminescent detection was done with cold shock protein D (CSPD) according to the manufacturer's recommendations (Boehringer Mannheim). Allele scoring was checked by two independent investigators.

Data Analysis

Data were analyzed with STATA 5.0 software (Stata Corporation, College Station, TX). The chi-squared and Fisher exact tests were used to test the significance of associations between categorical variables. The significance of differences in employment duration and age was determined with t tests.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Baseline characteristics of the 121 case and 120 control participants are shown in Table 1. All were black African men. Participation rates were 79% and 74% for approached case and control subjects, respectively. Participation rates were significantly lower in the Basotho (70%) than in other tribes (88%, p < 0.001), making Basothos relatively underrepresented in both the case and control populations. Case and control subjects were well matched for age, but there were significant differences in employment duration and job type, consistent with the known association between duration and intensity of dust exposure and silicosis. There was also a significant association between tribe and case status, with Basothos being overrepresented in the case populations. This remained significant when nonparticipants were included in the analysis (p = 0.015), and so reflects a genuine association of tribe with disease in the initially selected subjects, rather than an artefact introduced by the difference in refusal rates of Basothos.

TNF polymorphisms were not significantly associated with age, type of job, or mean years spent as a miner, but were significantly associated with tribal group and grade of silicosis. Basotho tribe members had a higher frequency of 308A alleles than did members of other tribes, with the combined AA/AG frequencies being 46% for Basothos, 33% for Xhosas, and 25% for members of other tribes (Fisher's exact p = 0.033), but there were no significant tribal variations at the 238 or 376 loci. There was strong linkage disequilibrium between SNPs at positions 238 and 376, as previously found for West and East Africans (14), so that the 376A allele occurred only in association with 238A (p < 0.001).

Case status was unrelated to polymorphism at any locus. The respective frequencies of AA/AG alleles in cases and controls were 40% and 39% at 308 (Fisher's exact p = 1.0), 8.3% and 5.8% at 238 (Fisher's exact p = 0.62), and 7.4% and 5.0% at 376 (Fisher's exact p = 0.60). However, there was a significant association between AG/AA alleles at 238 and 376 and silicosis grade, suggesting a possible association between TNF promoter polymorphisms and silicosis severity. This possibility was explored further, as shown in Table 2. The main finding was of a high prevalence of rare alleles among the subjects with severe (ILO grade 3) silicosis. At each of the three loci there were significant differences between the subjects with severe silicosis and controls, but not between subjects with early/moderate silicosis and controls. The frequency of A alleles (AA or AG) at each locus was also significantly higher among subjects with severe silicosis than among those with less severe silicosis (Fisher's exact p values = 0.028, 0.027, and 0.019 for the 308, 238, and 376 loci, respectively).

There was no significant association between TNF alleles and age, employment duration, or job type, so that the associations between TNF alleles and severe silicosis are unlikely to have been due to confounding by any of these variables. There was potential confounding with tribe, however, since A alleles at the 308 locus were significantly more common in members of the Basotho than in those of other tribes (Figure 1), and all subjects with severe silicosis were Basotho (Fisher's exact p value = 0.012 for the association between severe silicosis and Basotho tribe membership). Furthermore, the TNF gene lies within the highly polymorphic human leukocyte antigen (HLA) region, which is known to vary considerably with ethnicity (14).

View larger version (41K): [in this window] [in a new window]   Figure 1.   TNF- promoter region genotypes and allele frequencies in subjects belonging to the Basotho (n = 140) and other tribes (n = 92), not including subjects with severe silicosis. Subjects with severe silicosis are not included because of their nonrepresentative frequency of rare alleles. **p = 0.003 with Fisher exact test for the 3 by 2 comparison of variation in genotype (GG, AG, AA) and tribe. The difference in 308 allele frequency remained significant when the analysis was limited to control subjects only (p = 0.02). There were no other significant tribal differences (p = 0.5 and 0.3 for the 238 and 376 loci, respectively). , GG; , AG; , AA.

Because of the small number of cases of severe silicosis, all of which involved members of a single tribe, direct adjustment for potential confounding by tribe and/or HLA haplotype was not possible. Instead, we repeated the analysis, limiting all subjects to members of the Basotho tribe, as shown in the lower half of Table 2. There was confounding at the 308 locus, in that A allele frequency did not vary significantly between controls and subjects with severe silicosis among Basotho tribe members (p = 0.15). However, the association with AA/AG at the 238 and 376 sites remained significant when the analysis was limited to Basotho tribe members only (Fisher's exact p = 0.011). Therefore, confounding does not explain this association: a causal relationship remains possible.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Silicosis, one of the most extensively studied fibrotic lung diseases, remains a common occupational disease in industrializing countries, despite being preventable by dust control measures (1, 27). The risk of disease is related to total dose and intensity of exposure to silica dust (1), but there is also strong evidence for individual variation in susceptibility to silicosis both in humans (5) and in experimental animals (6, 7, 12, 33). Although environmental factors may also contribute (33), pronounced variability according to strain has been demonstrated for inbred mice (6, 7), implying that susceptibility to silicosis is at least partly determined by genetic factors.

The present study demonstrates significant associations between the 238 and 376 A alleles of the TNF- promoter and severe silicosis among black South African mineworkers. The 238A and 376A alleles are in linkage disequilibrium, as in other African populations (14), so that the associations at these loci are not independent. The association with severe silicosis and the 308A allele was more tentative, and, unlike the association with 238A and 376A, was not significant when single-tribe subanalysis was performed to investigate potential confounding by tribal background. This polymorphism, however, is of particular interest in having been previously associated with CWP in Europeans (26). In the present study, significant differences were found only when subjects with severe silicosis were compared either with unaffected control subjects or with subjects with less severe silicosis, but there were no differences between control subjects and those with less severe disease. This is in contrast to CWP in Europeans, in whom a borderline significant association between the 308A allele and early to moderate pneumoconiosis was reported for 18 case and 60 control participants in Belgium (26).

Our results have to be interpreted with caution because of the small numbers of subjects and because associations were found only on subgroup analysis, but it is striking that TNF- promoter polymorphisms have been associated only with the most severe manifestations of disease in a number of other studies (13, 14, 17, 21). For example, the 308A and 376A alleles are associated with susceptibility to cerebral malaria, but not to less severe episodes of the disease (13, 14), and the 308A allele has been associated with a poor prognosis rather than an increased risk of meningococcal disease (17), septic shock (18), and inflammatory bowel disease (21).

Polymorphisms in the TNF- promoter region are unusually strong candidates as putative silicosis-susceptibility loci, since TNF- has a central causal role in the pathophysiology of acute silicosis (9), and significant associations between silicosis and certain HLAs have been reported (34, 35), but with inconsistency of the implicated antigens in different populations, suggesting linkage disequilibrium with a nearby susceptibility locus, rather than a genuine HLA effect. The TNF- locus lies within the HLA complex, so that linkage disequilibrium with HLA antigens would be expected among diseases associated with polymorphisms in the TNF- promoter region (14). The basis for associations between TNF- promoter polymorphisms and disease susceptibilities is becoming clearer. In some cases SNPs have been associated with functional effects such as binding of nuclear transcription factors, increased basal or inducible transcriptional activity, or increased TNF- production in vivo (14, 16, 24, 36). The genetics and function of this region, however, remain incompletely determined (38).

The main finding of the present study is that individual variation in the severity of chronic silicosis, but not overall susceptibility to the disease, may be partly explained by polymorphisms in the TNF- promoter region. If this is confirmed in other populations, then silicosis joins a number of other diseases in which TNF- plays a central pathophysiologic role, and for which promoter region polymorphisms affect disease severity but not frequency.