You are here

  • Home
  • Archive
  • Volume 72, Issue 1
  • Genetic susceptibility to beryllium: a case–referent study of men and women of working age with sarcoidosis or other chronic lung disease

Article Text

Article menu
Workplace
Original article
Genetic susceptibility to beryllium: a case–referent study of men and women of working age with sarcoidosis or other chronic lung disease
Free
  1. Nicola Cherry1,
  2. Jeremy Beach2,
  3. Igor Burstyn3,
  4. Jillian Parboosingh4,5,
  5. Janine Schouchen2,
  6. Ambikaipakan Senthilselvan6,
  7. Larry Svenson7,
  8. Jan Tamminga5,
  9. Niko Yiannakoulias8
  1. 1Department of Medicine, University of Alberta, Edmonton, Canada
  2. 2Department of Occupational Medicine, University of Alberta, Edmonton, Canada
  3. 3Department of Environmental and Occupational Health, Drexel University, Philadelphia, Pennsylvania, USA
  4. 4Department of Medical Genetics, University of Calgary, Calgary, Canada
  5. 5ACH Molecular Diagnostic Laboratory, Alberta Health Services, Calgary, Canada
  6. 6School of Public Health, University of Alberta, Edmonton, Canada
  7. 7Surveillance and Assessment, Alberta Ministry of Health, Edmonton, Canada
  8. 8School of Geography and Earth Sciences, McMaster, Hamilton, Canada
  1. Correspondence to Dr Nicola Cherry, Division of Preventive Medicine, University of Alberta, 5-22 University Terrace, 8303-112St, Edmonton, Alberta, Canada T6G 2T4; Nicola.cherry{at}ualberta.ca

Abstract

Objective The study was designed to investigate whether beryllium exposure was related to illness diagnosed as sarcoidosis. Chronic beryllium disease (CBD) and sarcoidosis are clinically and pathologically indistinguishable, with only the presence of beryllium-specific T-lymphocytes identifying CBD. Testing for such cells is not feasible in community studies of sarcoidosis but a second characteristic of CBD, its much greater incidence in those with a glutamic acid residue at position 69 of the HLA-DPB1 gene (Glu69), provides an alternative approach to answering this question.

Methods Cases of sarcoidosis aged 18–60 years diagnosed in Alberta, Canada, from 1999 to 2005 were approached through their specialist physician, together with age-matched and sex-matched referents with other chronic lung disease. Referents were grouped into chronic obstructive pulmonary disease (COPD), asthma and other lung disease. Participants completed a telephone questionnaire, including industry-specific questionnaires. DNA was extracted from mailed-in mouthwash samples and genotyped for Glu69. Duration of employment in types of work with independently documented beryllium exposure was calculated.

Results DNA was extracted for 655 cases (270 Glu69 positive) and 1382 referents (561 positive). No increase in sarcoidosis was seen with either Glu69 or beryllium exposure (none, <10, ≥10 years) as main effects: longer duration in possible beryllium jobs was related to COPD. In Glu69 positive men with exposure ≥10 years, the trend towards increasing rate of COPD was reversed, and a significant interaction of duration of exposure and Glu69 was detected (OR=4.51 95% CI 1.17 to 17.48).

Conclusions The gene–environment interaction supports the hypothesis that some cases diagnosed as sarcoidosis result from occupational beryllium exposure.

https://doi.org/10.1136/oemed-2014-102359

Statistics from Altmetric.com

    Request Permissions

    If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

    What this paper adds

    • Sarcoidosis is a chronic granulomatous disease with no known cause but environmental and genetic factors are suspected. Chronic beryllium disease (CBD) is clinically and pathologically indistinguishable from sarcoidosis but its diagnosis requires a positive test for sensitisation to beryllium.

    • In those occupationally exposed to beryllium, workers with a glutamic acid in codon 69 (Glu69) in the HLA-DPB1 gene are at much increased risk of developing CBD.

    • If patients (untested for beryllium sensitisation) with possible beryllium exposure at work are at a greater risk of sarcoidosis with the Glu69 variant, this would imply that beryllium contributes to some disease diagnosed as sarcoidosis.

    • The finding in the case–referent study reported here of significant gene–environment interaction (Glu69×possible beryllium exposure for 10 years or more) supports the hypothesis that some cases of sarcoidosis result from beryllium exposure in this population.

    • The result supports the investigation of further gene–environment interactions in sarcoidosis and the use of Mendelian randomisation approaches in occupational epidemiology.

    Introduction

    Sarcoidosis is a chronic granulomatous disease with no known cause, thought to represent an exaggerated immunological response to as yet unknown antigens.1 Both environmental and genetic factors are associated with sarcoidosis.2 ,3 Suggested occupational risk factors include metal working industries (from an ecological study in Switzerland)4 and exposure to insecticides.2 No clear gene–environmental interaction has been reported for sarcoidosis, contrasting strongly with the highly elevated risk of chronic beryllium disease (CBD) in those with a glutamic acid residue coded for in codon 69 (Glu69) in the HLA-DPB1 gene.5 Given that CBD is clinically and pathologically indistinguishable from sarcoidosis, it is reasonable to hypothesise that genetic variation in the presentation of one or more (as yet unknown) antigens may be causal also in sarcoidosis. The present study was designed to investigate this possibility. This first report addresses a focussed question within this broader thesis: to what extent do cases diagnosed as sarcoidosis result from occupational exposure to beryllium? If CBD is being misdiagnosed as sarcoidosis, a sentinel case reflecting an unrecognised risk in a pool of workers may be missed, together with opportunities for prevention.

    Beryllium is a low-molecular-weight metal with properties (light weight, high tensile strength, corrosion resistance) that make it highly attractive, either as the metal itself or as an alloy or compound, in a wide range of uses including (as a beryllium-copper bronze) non-spark tools and electrical switch parts and as a metal in nuclear reactors. The health effects of exposure have largely been investigated in the USA in the nuclear industry.5 Importantly, no clear dose–response relationship has been found between exposure and CBD, although there is an assumption that sensitisation to beryllium depends on intensity of exposure and perhaps also duration and/or latency.6 Although there have been reports of sensitisation to beryllium after exposures of less than 1 year,7 progression from sensitisation to CBD may take many years.8 In the absence of clear dose–response data (probably reflecting the immunological basis of the disease) and the decision (now disputed) that beryllium is carcinogenic to humans,9 ,10 there is a great deal of uncertainty about a ‘safe’ level of beryllium exposure.11 At present the US Occupational Safety and Health Administration (OSHA) permissible exposure limit is 2 µg/m3, the National Institute for Occupational Safety and Health (NIOSH) recommended exposure limit 0.5 µg/m3 and the American Conference of Governmental Industrial Hygienists (ACGIH) threshold limit value 0.05 µg/m3.

    CBD has two characteristics potentially important for investigating its misdiagnosis as sarcoidosis. First, diagnosis of CBD requires pathological changes and evidence of sensitisation to beryllium on a lymphocyte proliferation test (BeLPT). This test has been used clinically to identify patients with CBD misdiagnosed as sarcoidosis, in individual patients12 and as part of a prospective case series.13 The BeLPT, however, is somewhat exacting and costly, and available only from specialist centres. The second significant feature of CBD is the substantial (perhaps 10-fold) increase in risk in those with the Glu69 susceptibility marker, an important (but rare) example of gene–environment interaction at the workplace.7 ,14 In contrast to the exacting handling requirement of blood (or bronchoalveolar lavage fluid) samples for the BeLPT, identification of the Glu69 genetic marker can be made using mailed-in mouthwash or buccal swab samples.15 This known, and easily identifiable, marker of genetic susceptibility to CBD made possible this study to investigate the misdiagnosis of sarcoidosis: if the genetic susceptibility marker specific to beryllium is more frequent in workers (or subsets of workers) with sarcoidosis, this would support the conclusion (at the population level) that sarcoidosis was, at least in part, attributable to beryllium. Such an approach, sometimes labelled as Mendelian randomisation, has been used to good effect in earlier studies.16 ,17

    The estimation of occupational exposure to beryllium for this study was only possible because of a published analysis of reported beryllium content of dust samples collected during OSHA inspections during 1979–1996 which identified 38 four-digit Standard Industrial Classification (SIC) codes, together with ‘at risk’ tasks in which beryllium had been found in two or more samples at or above the NIOSH recommended standard of 0.5 µg/m3.18 This report of industries, together with typically exposed jobs within each, reduced the seemingly impossible task of assessing occupational beryllium exposure over many years in widely diverse occupations to one of collecting comprehensive data on specific tasks within previously identified industries.

    Methods

    Cases and referents

    A case–referent study was designed of patients living in Alberta, Canada from 1 January 1995 to the time of first diagnosis of sarcoidosis (International Classification of Diseases (ICD)-9 code 135; cases) or other chronic respiratory diseases (ICD-9 codes 466, 490–518; referents). Patients with sarcoidosis diagnosed in Alberta aged 18–60 years from 1 January 1999 to 31 December 2005 were eligible as cases if they had evidence of pulmonary involvement. Only new diagnoses of sarcoidosis were included: a case was considered a new diagnosis if a specialist physician recorded a diagnosis of ICD-9 code 135 in the period 1999–2005 and no diagnosis of ICD-9 135 had been recorded by any Alberta physician for this patient in the period 1995–1998. The diagnostic code was not required to be the main (or first) code recorded for that patient visit, and only the code that made the patient eligible for the study was recorded. Up to five potential referents were chosen for each case, individually matched on sex, age (±2 years) and date of clinic visit (after the case diagnosis but within 12 months). Referents were chosen with ICD-9 codes that indicated lung disease that was chronic or had the potential to become chronic: those with an ICD-9 code of 135 since 1995 were ineligible. Hypersensitivity pneumonitis (ICD-9 code 495) was excluded because of diagnostic overlap with sarcoidosis. A referent was only used once: if there was a match to more than one case, alternative referents were used. Patients with an eligible diagnosis (case or referent) but who were found, on chart review, to have an additional diagnosis of carcinoma or cystic fibrosis were omitted, as were those known to have died.

    Identification and recruitment of cases and referents

    In Alberta, healthcare is free at the point of delivery and registration with the Alberta Health Care Insurance Plan is mandatory for all residents of the province. Physicians are paid through the scheme and complete administrative billing records that include up to three ICD-9 diagnoses for patient services. Specialist physicians (who have completed a specialist residency recognised by the Royal College of Physicians and Surgeons of Canada) are paid on a different schedule than family physicians and general practitioners. Access to the billing codes on the administrative records was possible only with the agreement of the physician, specifically elicited for this study.

    There was no validated algorithm for the diagnosis of sarcoidosis through administrative records and it was decided to maximise the likelihood that a recorded diagnosis of sarcoidosis was truly sarcoidosis by restricting physician recruitment to internal medicine and pulmonary specialists and to review medical files to identify coding errors. No additional requirement (such as biopsy confirmation) was imposed, although the chart review included confirmation of pulmonary involvement. Specialist physicians were approached by the research team to ask for their agreement that the Alberta Ministry of Health (AMH) might identify all eligible cases for which they had recorded a diagnostic code of 135 and also, for the cases identified, up to five eligible referents. The AMH identified cases and referents for all physicians who gave such written consent. The medical files of patients listed were examined to ensure that the condition was consistent with sarcoidosis or, for referents, other chronic lung disease. The physician then signed an information letter asking patients selected to return a signed consent to the study team. Patients who consented were sent a recruitment package.

    The preinterview package contained three forms to be completed before the interview as an aide memoire for dates of (1) every residence since birth, (2) all occupations since leaving school and (3) employment in any of the 34 industries (online supplementary appendix 1), of which 18 were a somewhat compressed version of the 38 four-digit industry codes previously identified with exposure to beryllium in OSHA samples,18 10 were industries with possible beryllium exposure but not reported to have OSHA beryllium samples and 6 industries employing large numbers within Alberta of interest to the Province. A collection kit for the mouthwash sample was also included in the prequestionnaire package.

    Exposure to beryllium

    When participants reported that they had been employed in one of the industries listed in online supplementary appendix 1 the interviewer completed an industry-specific questionnaire designed for this study. All industries recorded on the job history were coded to US SIC four-digit codes19 and in rare instances where an industry-specific questionnaire was indicated but had not been completed, the job description was scrutinised to determine whether a potentially beryllium-exposed job had been carried out. All periods potentially exposed were classified as (1) ‘listed’: both industry and job/task within that industry were included in the published listing of OSHA samples with beryllium ≥0.5 µg/m3; (2) ‘assumed listed’ periods as a construction labourer with insufficient information about the work site and (3) ‘not-listed’, periods of employment with possible beryllium exposure not included in the list of positive OSHA results. Total duration to diagnosis was calculated for each of these three categories. Date first working in such a job was also determined to allow examination of latency.

    Determination of Glu69 status

    DNA was extracted where possible from each mouthwash sample received using methods adapted from Garcia-Closas et al.15 Samples were thawed at room temperature and then centrifuged at 1500g for 15 min. The supernatant was decanted and the cell pellet was resuspended in Tris-EDTA. The samples were centrifuged a second time at 1500g for 15 min and the supernatant again decanted. The samples were then split into two groups on cell pellet volume, less than or greater than 100 µL. For samples with volumes <100 µL, the QIAamp DNA Investigator kit (Qiagen) was used with the following modifications: cell pellets were lysed overnight at 55°C in 90 µL buffer ATL (Qiagen) and 10 µL of proteinase K (20 mg/mL). The remaining steps were carried out on the QIAcube (Qiagen) according to the manufacturer's instructions. For samples with volumes >100 µL, the QIAamp DNA Mini kit (Qiagen) was used with the following modifications: cell pellets where lysed overnight at 55°C in 180 µL buffer ATL and 20 µL of proteinase K (20 mg/mL). The remaining steps were carried out on the QIAcube (Qiagen) according to the manufacturer's instructions. All DNA samples were eluted in TE and stored at 4°C.

    Genotyping for the Glu69 allele (NM_002121.4 p.Lys98Glu) was performed by sequencing exon 2 of the HLA-DBP1 gene. Genomic DNA was PCR amplified prior to sequencing using the ABI BigDye Terminator Cycle Sequencing kit (V.1.1; Life Technologies) and the ABI 3130 Genetic Analyzer (Life Technologies). Primer sequences and reaction conditions are available on request. Sequence subtraction was performed using Mutation Surveyor software (SoftGenetics).

    Measurement of potential confounders

    Information on confounders for sarcoidosis or other chronic lung disease was collected using the questionnaire. This included ‘cigarette years’ (usual number of cigarettes/day while smoking×total years smoked to date of diagnosis) for those ever smoking as much as 1 cigarette/day for 12 months, ethnicity self-reported as black, education (completed high school or equivalent), together with sex and age at diagnosis. Sarcoidosis in a first-degree blood relative (parent, sibling, child), as participant by the patient, was also recorded.

    Statistical methods

    Distribution of participants was examined by diagnosis (sarcoidosis) and three subgroups of referents (asthma (ICD-9 code 493), chronic obstructive pulmonary disease (COPD; ICD-9 codes 490–492, 496) and other (ICD-9 codes 466, 494, 501–518)). Duration of time spent in each of the 34 industries of interest was examined and those with 50 or more participants ever employed (within sex) identified for more detailed analysis of diagnosis (sarcoidosis, asthma, COPD, other) by duration of employment in the industry. Total duration of employment in jobs potentially exposed to beryllium was calculated and examined by diagnosis, adjusting for potential confounders. Sarcoidosis was then examined in those with and without a Glu69 marker, stratifying for exposure, and a logistic regression analysis carried out (with sarcoidosis against all referents as the dependent variable), including a gene–exposure interaction, and adjusting for potential confounders. Because a relatively large proportion of participants recruited did not form part of a cluster (ie, a case was recruited but not any of the referents or referents were recruited but the case was not) an unconditional regression was used. Sensitivity analyses were carried out to examine the dependence of the interaction effect on the definition of exposure. These included varying the duration of exposure required (with 10 years or more as the initial breakpoint), stratifying for latency and, in separate analyses, excluding periods in ‘assumed listed’ employment and adding periods in ‘not listed’ work.

    Power

    Power calculations are given in online supplementary appendix 2. Briefly, if the prevalence of biologically significant exposure to beryllium was found to be 5% in referents, a gene–environment effect producing a threefold increase in risk could be found with 80% power using 617 cases and 1234 referents assuming (1) Glu69 marker present in 30% of the population, (2) the exposure in the absence of Glu69 confers only a 10% excess odds of the disease and (3) Glu69 in the absence of exposure does not confer risk.

    Results

    Participation

    Six hundred and fifty-two respiratory and internal medicine specialists were identified of whom 467 were approached. Those omitted were either retired or had a practice unlikely to see adult patients with sarcoidosis. Of the 467 physicians approached 254 agreed. The most frequent reasons given by those who refused were that they had not seen any eligible patient with sarcoidosis. At least one eligible case was found for 110 specialists, with a total of 1826 cases and 5935 referents identified in the administrative records for these physicians. An attempt was made to screen the clinical records of all these patients but this was achieved for only 6480 (1629 cases and 5211 referents). Review of the charts eliminated 238 cases and their matched referents (N=1129). Of 1392 cases and 4081 referents approached, questionnaires were completed by 684 cases and 1454 referents. In both cases and referents those who participated were older than those who did not, and among the referents, men were less likely to take part.20 Among the 684 cases, 30% (204) had had a confirmatory lung biopsy and a further 20% (140) biopsy of a lymph node or some other organ.20 Genotyping for Glu69 status was obtained for 2037 (655 cases (378 men) and 1382 referents (692 men); figure 1).

    Figure 1
    Figure 1

    Recruitment flow diagram.

    Description of the sample

    Asthma was by far the most frequent referent diagnosis with the only other diagnosis importantly represented being COPD (online supplementary appendix 3). Confounders differed with diagnostic group (table 1). Those with sarcoidosis were least likely to smoke (but with similar smoking patterns in asthmatics) and had a higher proportion with black ethnicity and self-reported first-degree relatives diagnosed with sarcoidosis. Those with COPD were less likely to have completed high school, more likely to be heavy smokers and had the highest median age. All factors shown in table 1 differed significantly (p<0.001) among the four diagnostic groups.

    View this table:
    Table 1

    Description of the sample

    Relation of diagnosis to exposures at work

    The distribution of participants who had worked in each of the 34 industries of interest was examined and the relation to diagnosis explored for those industries with 50 or more participants (within sex) having any period of employment (other than exclusively office work). For men there were eight such industries (online supplementary appendix 4): for women there were only three (oil and gas, agriculture and healthcare). For men in some industries (eg, construction) there was a clear increase in COPD with increasing duration of employment (online supplementary appendix 4) but the risk of sarcoidosis increased with duration of employment only in healthcare, where the unadjusted OR for 10 years or greater employment was 2.60 (95% CI 1.04 to 6.53). This reduced to 1.87 (95% CI 0.71 to 1.95) when adjusted for confounders. For women, asthma, but not sarcoidosis, tended to increase with years of work in healthcare: unadjusted OR for employment of 10 years or greater was 1.42 (95% CI 0.98 to 2.06), only slightly reduced to 1.39 (95% CI 0.94 to 2.04) with allowance for confounding.

    The distribution of exposure to beryllium (derived from ‘listed’ or ‘assumed listed’ employment) is shown by sex and diagnosis in table 2. Only 1.2% of women (12/1023) had ever been in an occupation with possible exposure to beryllium, but in these there was perhaps some trend, with sarcoidosis being seen more in those with some exposure to beryllium dust. Among men nearly 20% (218/1115) had been employed at some point in a job potentially exposed, though only 61 (5.5%) for 10 years or more. Without consideration of genotype, years of exposure to possible beryllium in dust was associated with COPD rather than sarcoidosis (only 14.5% having COPD in those with no beryllium exposure but 29.5% in those exposed for 10 years or more). In a logistic regression analysis, with a diagnosis of COPD as the outcome, the OR for employment in (dusty) jobs with possible beryllium exposure >0 to <10 years was 1.99 (95% CI 1.26 to 3.14) and for 10 years or greater 2.02 (95% CI 1.04 to 3.92), compared with no exposure, having adjusted for cigarette-years of smoking, black ethnicity, first degree relative with sarcoidosis, completing high school and age at diagnosis.

    View this table:
    Table 2

    Estimated exposure to beryllium dust by diagnosis and sex

    The proportion of cases and referents with one or more Glu69 positive markers is shown by sex in online supplementary appendix 5: no relation was seen overall between the genetic susceptibility marker and diagnosis (χ2=2.95; df=3: p=0.4). Table 3 gives the relation between genotype, beryllium exposure and diagnosis in the 1070 men genotyped: very little influence of Glu69 is seen in those estimated not to be exposed to beryllium (χ2=0.78; df=3: p=0.9) or those exposed for <10 years (χ2=4.41; df=3: p=0.2) but the distribution of cases with exposure ≥10 years is unlikely to be due to chance (χ2=9.08; df=3: p=0.03). Among the 58 men exposed for at least 10 years, half were Glu69 positive and these had a higher rate of sarcoidosis (44.8%) than those with similar estimated exposure but Glu69 negative (13.8%), whose diagnosis was most likely to be COPD. In a logistic regression analysis with a diagnosis of sarcoidosis as the outcome and main effects for genotype, exposure to beryllium (10 years or more) and with allowance for smoking (cigarette-years), completing high school, first degree relative with sarcoidosis, age at diagnosis and black ethnicity, the interaction term for exposure 10 years or longer together with a positive Glu69 marker was significantly elevated (OR=4.51 95% CI 1.16 to 17.48; table 4). Within the Glu69 negative group the proportion of cases of COPD increased steadily with years of exposure to dust possibly containing beryllium (from 14% to 38%) but this effect was importantly modified towards sarcoidosis in the presence of the Glu69 marker (table 3).

    View this table:
    Table 3

    Diagnosis by genotype stratified by beryllium exposure (men)

    View this table:
    Table 4

    Logistic regression analysis of sarcoidosis in men typed on Glu69 (N=1066)*

    Among the 211 men genotyped and exposed to beryllium dust, a majority (126:59.7%) was first exposed more than 20 years before diagnosis with only 18 first exposed <10 years before. When those with latency from first exposure of <10 years were excluded the size of the OR of interaction changed little (increasing to 4.54). Omitting those with <15 years latency increased the OR of interaction to 5.95 (95% CI 1.31 to 27.05). In those with latency ≥20 years, the point estimate of OR increased further, to 11.23 (95% CI 1.17 to 108.18), although with a very wide CI. Other sensitivity analyses, (1) excluding ‘assumed listed’ exposures, (2) including non-listed exposures or (3) changing the cut off from 10 to 5 years or 15 years, resulted in somewhat smaller (but always positive) estimates of the interaction term (data not shown). No similar interaction with genotype was found for any of the individual industries included in online supplementary appendix 4.

    Discussion

    The analysis reported here describes the relation between duration of employment in certain industries and diagnosis of chronic lung disease. Specifically it considers the interaction between a genetic marker of susceptibility to beryllium and estimates of potential beryllium exposure at work. It demonstrates such an interaction in men in potentially exposed jobs held over at least 10 years and in doing so is consistent with beryllium being a cause of the disease recorded as sarcoidosis. Clinical confirmation of beryllium in an individual patient would best be carried out by arranging a BeLPT, not easily available in Alberta, but at the population level, the relationship is supported.

    This observation, while of considerable interest in demonstrating that such an approach can be informative, does not help with identifying the cause of sarcoidosis in the large majority of cases: the nature of other such gene–environment interactions (with perhaps many alleles specific to many different exposures) remains to be identified. The clear excess of men, who constituted 56.7% of the 1826 sarcoidosis cases of working age in records of participating physicians, is consistent with at least some of the relevant exposures being occupational, but a genome-wide (or at least complete HLA) analysis of alleles in the harvested DNA samples would be needed to make full use of the detailed exposure data to identify further antigens implicated in sarcoidosis.

    It is also important to consider the interpretation of negative results from this analysis. First, there was no general increase in risk of sarcoidosis in those with a positive Glu69: the increased risk was found only in those (potentially) exposed to beryllium in dusty jobs. Environmental exposure to beryllium, although detected in air quality monitoring data in Alberta21 and in snowpack near oil sands development,22 appears to be too low to affect risk at the population level: historically, environmental cases of CBD have been reported in the vicinity of large industrial users and in the family of workers in the USA.12 ,23 Second, even with occupational exposure, the gene–environment interaction was seen only in those with at least 10 years employment in work with the potential for beryllium exposure, and particularly in those with longer latency. While this is a credible pattern, the failure to find an excess of granulomatous pulmonary disease in those with shorter exposure or in Glu69 negative workers should be noted, together with the relatively small numbers in the 10-year plus duration group. If there had not been a clear and biologically plausible prior hypothesis, the significance of the interaction term would have been easily overlooked.

    Although the study was designed with referents as ‘all other chronic lung disease’ the selection of cases and referents from the same physician, within the year of first presentation, made it reasonable (although unorthodox) to examine other diagnostic groups within the overall category of chronic lung disease, while recognising that because of the initial design sarcoidosis would appear more frequently than by chance. In this way we considered, for example, the relation of occupation to asthma (in online supplementary appendix 4) or, as reported in more detail here, to COPD. The strong relationship, particularly in those Glu69 negative, between years of exposure to (beryllium containing) dust and COPD was not anticipated. Given that sarcoidosis is consistently found more frequently in non-smokers,1 and COPD in heavy smokers, this analysis would be particularly susceptible to confounding between smoking and long exposure in dusty (beryllium) jobs. Those with longer exposure in such jobs were indeed more likely to be heavy smokers and residual confounding may in part explain the unexpectedly strong risk of COPD. It is not clear, however, that some such effect could explain the pattern of diagnosis in those with long duration of employment who were Glu69 positive, where sarcoidosis was found more frequently than COPD.

    This study has important strengths. First, it is sufficiently large to address the question of gene–environment interaction (at least for relatively common genetic markers and exposures). Second, it has detailed occupational histories, allowing us to compute duration of employment by industry and task. Third, in the present analysis, the estimate of beryllium exposure was performed using an independent exposure assessment based on quantitative measures and so avoiding ‘professional judgment’ as a criterion of likely exposure. While there might be differences in occupational practices between Alberta and the US workplaces sampled by OSHA, these are neither likely to be substantial, nor to result in a falsely positive gene–environment interaction. It is of interest that potential beryllium exposures not reported by Henneberger et al18 did not add to the gene–environment interaction. This might be because beryllium was not, in fact, present (OSHA samples existed but were not positive) or that there were no OSHA samples because the jobs were not dusty. Newman and Newman24 recently suggested that occupational sarcoidosis may result from the presence of an antigen (which here would be beryllium) and contemporaneous dust exposures causing inflammation. The results reported here would be consistent with that hypothesis, with the antigen being presented to the immune system more efficiently by those who were Glu69 positive and inflammation being a consequence of the dusty work environment. If that were the case, beryllium sensitisation might not be a pre-requisite and, if negative, the disease may be correctly diagnosed as sarcoidosis.

    Use of a job-exposure-like approach was the only feasible way to assess possible beryllium exposure in this population, spread over a Canadian province that is geographically vast and with many years of potential exposure. Its use, without direct measurements of exposure, precludes investigating misclassification. The estimate that only 5% of men in this study were exposed for 10 years or more was consistent with our initial power calculations for ‘biologically significant’ exposure but we cannot be sure this is accurate. If we assume that sensitisation to beryllium may have occurred in some of these men, it may be that longer exposure increases the likelihood of a beryllium exposure sufficiently high to cause sensitisation, with (given the presumed low exposure overall) a relatively indolent progression to CBD/sarcoidosis.

    As in the present analysis, previous studies have not found a consistent relation between the Glu69 marker and sarcoidosis. An early study in the UK found a strong relation with 26 of 41 cases (63.4%) positive but only 26/76 controls (38.2%) but a further and larger study from the same group found only a small increase in cases.25 ,26 Similarly in the ACCESS study of biopsy confirmed sarcoidosis in the USA, in which 474 case–referent pairs were assessed, the Glu69 variant was found somewhat less frequently in sarcoidosis cases.27 The hypothesis under test here, that exposure was important particularly in the presence of the susceptibility marker, and conversely that the genetic susceptibility marker was only important in the presence of exposure, could not be examined in these earlier studies that did not estimate beryllium exposure. Where such exposure is somewhat rare, the effect of such an interaction may be too small to be noticeably reflected in the main effect, as was recently demonstrated in a theoretical calculation, under these same assumptions.28 The relationship of CBD to the Glu69 allele is one of the very few gene–environment effects in the occupational literature in which the size of the interaction is potentially of importance in advising workers.29 The present study suggests that preventive intervention targeting beryllium exposures should not ignore industries where exposure is presumed to be low or infrequent, and that a high index of suspicion be maintained where sarcoidosis is diagnosed in dusty workplaces where beryllium may be present.

    Acknowledgments

    The authors gratefully acknowledge many Alberta pulmonologists and other specialists in internal medicine who gave consent to the study and contacted potential participants. Dr James Archibald and Dr Neil Brown piloted questionnaires and study procedures. Dr Robert Cowie acted as a co-investigator to ensure the study process worked effectively in Calgary and surrounding areas.

    References

    1. ↵
      1. Valeyre D,
      2. Prasse A,
      3. Nunes H, et al
      . Sarcoiosis. Lancet 2014;383:1155–67.
      OpenUrlCrossRefPubMedWeb of Science
    2. ↵
      1. Newman LS,
      2. Rose CS,
      3. Bresnitz EA, et al
      .; ACCESS Research Group. A case control etiologic study of sarcoidosis: environmental and occupational risk factors. Am J Respir Crit Care Med 2004;170:1324–30.
      OpenUrlCrossRefPubMedWeb of Science
    3. ↵
      1. Müller-Quernheim J,
      2. Schürmann M,
      3. Hofmann S, et al
      . Genetics of sarcoidosis. Clin Chest Med 2008;29:391–414.
      OpenUrlCrossRefPubMedWeb of Science
    4. ↵
      1. Deubelbeiss U,
      2. Gemperli A,
      3. Schindler C, et al
      . Prevalence of sarcoidosis in Switzerland is associated with environmental factors. Eur Respir J 2010;35:1088–97.
      OpenUrlAbstract/FREE Full Text
    5. ↵
      1. Van Dyke MV,
      2. Martyny JW,
      3. Mroz MM, et al
      . Exposure and genetics increase risk of beryllium sensitisation and chronic beryllium disease in the nuclear weapons industry. Occup Environ Med 2011;68:842–8.
      OpenUrlAbstract/FREE Full Text
    6. ↵
      1. Henneberger PK,
      2. Cumro D,
      3. Deubner DD, et al
      . Beryllium sensitization and disease among long-term and short-term workers in a beryllium ceramics plant. Int Arch Occup Environ Health 2001;74:167–76.
      OpenUrlCrossRefPubMedWeb of Science
    7. ↵
      1. Van Dyke MV,
      2. Martyny JW,
      3. Mroz MM, et al
      . Risk of chronic beryllium disease by HLA-DPB1 E69 genotype and beryllium exposure in nuclear workers. Am J Respir Crit Care Med 2011;183:1680–8.
      OpenUrlCrossRefPubMedWeb of Science
    8. ↵
      1. Newman LS,
      2. Mroz MM,
      3. Balkissoon R, et al
      . Beryllium sensitization progresses to chronic beryllium disease: a longitudinal study of disease risk. Am J Respir Crit Care Med 2005;171:54–60.
      OpenUrlCrossRefPubMedWeb of Science
    9. ↵
      International Agency for Research on Cancer. World Health Organization. IARC Monographs on the Evaluation of Carcinogenic Risks of Humans. Volume 58. Beryllium, Cadmium, Mercury, and Exposures in the Glass Manufacturing Industry. 1993.
    10. ↵
      1. Boffetta P,
      2. Fryzek JP,
      3. Mandel JS
      . Occupational exposure to beryllium and cancer risk: a review of the epidemiologic evidence. Crit Rev Toxicol 2012;42:107–1810.
      OpenUrlCrossRefPubMed
    11. ↵
      1. Borak J
      . The beryllium occupational exposure limit: historical origin and current inadequacy. J Occup Environ Med 2006;48:109–16.
      OpenUrlCrossRefPubMedWeb of Science
    12. ↵
      1. Newman LS,
      2. Kreiss K
      . Nonoccupational beryllium disease masquerading as sarcoidosis: identification by blood lymphocyte proliferative response to beryllium. Am Rev Respir Dis 1992;145:1212–14.
      OpenUrlCrossRefPubMedWeb of Science
    13. ↵
      1. Müller-Quernheim J,
      2. Gaede KI,
      3. Fireman E, et al
      . Diagnoses of chronic beryllium disease within cohorts of sarcoidosis patients. Eur Respir J 2006;27:1190–5.
      OpenUrlAbstract/FREE Full Text
    14. ↵
      1. Richeldi L,
      2. Kreiss K,
      3. Mroz MM, et al
      . Interaction of genetic and exposure factors in the prevalence of berylliosis. Am J Ind Med 1997;32:337–40.
      OpenUrlCrossRefPubMedWeb of Science
    15. ↵
      1. García-Closas M,
      2. Egan KM,
      3. Abruzzo J, et al
      . Collection of genomic DNA from adults in epidemiological studies by buccal cytobrush and mouthwash. Cancer Epidemiol Biomarkers Prev 2001;10:687–96.
      OpenUrlAbstract/FREE Full Text
    16. ↵
      1. Cherry N,
      2. Mackness M,
      3. Durrington P, et al
      . Paraoxonase (PON1) polymorphisms in farmers attributing ill health to sheep dip. Lancet 2002;359:763–4.
      OpenUrlCrossRefPubMedWeb of Science
    17. ↵
      1. Smith GD,
      2. Ebrahim S
      . Mendelian randomization: can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol 2003;32:1–22.
      OpenUrlAbstract/FREE Full Text
    18. ↵
      1. Henneberger PK,
      2. Goe SK,
      3. Miller WE, et al
      . Industries in the United States with airborne beryllium exposure and estimates of the number of current workers potentially exposed. J Occup Environ Hyg 2004;1:648–59.
      OpenUrlCrossRefPubMedWeb of Science
    19. ↵
      Office of Management and Budget. Standard Industrial Classification Manual. Washington DC: U.S. Government Printing Office, 1987.
    20. ↵
      1. Schouten J
      . Biological factors in the etiology of pulmonary sarcoidosis. A thesis submitted in partial fulfillment of the requirements for the degree of Master of Science in Epidemiology at the University of Alberta. 2011.
    21. ↵
      Alberta Environment. Air Quality Monitoring, Northwest Edmonton. Interim report. 2002. http://environment.gov.ab.ca/info/library/6389.pdf
    22. ↵
      1. Kelly EN,
      2. Schindler DW,
      3. Hodson PV, et al
      . Oil sands development contributes elements toxic at low concentrations to the Athabasca River and its tributaries. Proc Natl Acad Sci USA 2010;107:16178–83.
      OpenUrlAbstract/FREE Full Text
    23. ↵
      1. Eisenbud M,
      2. Lisson J
      . Epidemiological aspects of beryllium-induced nonmalignant lung disease: a 30-year update. J Occup Med 1983;25:196–202.
      OpenUrlCrossRefPubMedWeb of Science
    24. ↵
      1. Newman KL,
      2. Newman LS
      . Occupational causes of sarcoidosis. Curr Opin Allergy Clin Immunol 2012;12:145–50.
      OpenUrlCrossRefPubMed
    25. ↵
      1. Lympany PA,
      2. Petrek M,
      3. Southcott AM, et al
      . HLA-DPB polymorphisms: Glu 69 association with sarcoidosis. Eur J Immunogenet 1996;23:353–9.
      OpenUrlCrossRefPubMedWeb of Science
    26. ↵
      1. Foley PJ,
      2. Lympany PA,
      3. Puscinska E, et al
      . Analysis of MHC encoded antigen-processing genes TAP1 and TAP2 polymorphisms in sarcoidosis. Am J Respir Crit Care Med 1999;160:1009–14.
      OpenUrlCrossRefPubMedWeb of Science
    27. ↵
      1. Rossman MD,
      2. Thompson B,
      3. Frederick M, et al.
      ; CCESS Group. HLA-DRB1*1101: a significant risk factor for sarcoidosis in blacks and whites. Am J Hum Genet 2003;73:720–35.
      OpenUrlCrossRefPubMedWeb of Science
    28. ↵
      1. Luo H,
      2. Burstyn I,
      3. Gustafson P
      . Investigations of gene-disease associations: costs and benefits of environmental data. Epidemiology 2013;24:562–8.
      OpenUrlCrossRefPubMedWeb of Science
    29. ↵
      1. Christiani DC,
      2. Mehta AJ,
      3. Yu CL
      . Genetic susceptibility to occupational exposures. Occup Environ Med 2008;65:430–6.
      OpenUrlAbstract/FREE Full Text

    Supplementary materials

    • Supplementary Data

      This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

      Files in this Data Supplement:

    Footnotes

    • Contributors NC, JB, IB and AS designed the project with LS and NY developing and executing the algorithms for identifying the patients. JS visited the offices of all collaborating physicians and confirmed patient eligibility in discussion with JB. NC, JB and IG devised and executed the exposure assessment. JP carried out the genotyping with the assistance of JT who extracted DNA from mouthwash samples. NC and JS carried out the data analysis. All authors commented on the draft report and have seen and agreed the final version.

    • Funding The work was funded in part by an establishment grant to NC from the Alberta Health Foundation for Medical Research with additional funding as a gift from the Canadian National Sarcoidosis Organisation.

    • Competing interests None.

    • Patient consent Obtained.

    • Ethics approval Health ethics boards of the University of Alberta and the University of Calgary.

    • Provenance and peer review Not commissioned; externally peer reviewed.

    • Data sharing statement Additional data would be made available to researchers interested in collaboration after discussion of an agreed protocol.