INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Chronic beryllium disease (CBD) is an occupational, granulomatous lung disease that continues to occur despite efforts to reduce beryllium exposure in industry (1). A purely linear disease-exposure relationship has not been found for beryllium and CBD (3). Individuals with casual exposure, such as secretaries, security guards, and spouses of exposed workers, have been diagnosed with CBD (4, 5). In addition to exposure to beryllium, a beryllium-specific, cell-mediated immune response is required for the development of CBD (6). The immune response to beryllium appears to depend on an individual's genetic susceptibility. Studies exposing guinea pigs and mice to beryllium have shown that strains that vary in their major histocompatibility complex (MHC) loci produce different immune responses to beryllium, suggesting a genetic predisposition to beryllium disease (5). In humans, familial occurrences of CBD have been noted in identical twins and in parents and children (5). Richeldi and colleagues (9, 10) and Stubbs and coworkers (11) have shown allelic differences in the MHC class II molecules human leukocyte antigen (HLA)- DP (9, 10) and HLA-DR (11) in patients with CBD as compared with nondiseased, beryllium-exposed individuals. To date, no other genetic markers for CBD have been found, although it is likely that the genetic susceptibility to CBD is multifactorial and that other genes are involved in regulating the immune and inflammatory response in pathogenesis of the disease (5).

We have previously shown that serum levels of angiotensin converting enzyme (ACE) activity are associated with CBD disease severity measured according to symptoms of shortness of breath, bronchoalveolar lavage (BAL) cellularity (12), abnormalities found through chest computed tomography (CT) (13), and concentrations of an inflammatory marker, the  subunit of the IL-2 receptor (14). Differences in serum ACE activity have been shown to be genetically regulated (15), and are associated with an insertion (I) and a deletion (D) in intron 16 of the ACE gene (16). Specifically, three genotypes occur, DD, ID, and II, depending on the absence or presence of a 287-bp sequence in the intron, accounting for 50% of the variance in serum ACE activity (16). Normal individuals and sarcoidosis patients with the DD genotype tend to express higher serum levels of ACE activity, indicating a genotypic effect even in sarcoidosis patients with increased serum ACE activity (16, 17). Associations between the DD polymorphism and disease have been observed in coronary artery disease (18), cardiomyopathy (19, 20), asthma (21), and the progression of nephropathy (22, 23) and cardiovascular disease (19). In addition, there is an indication that ACE and its product, angiotensin II, may be important in fibrogenesis in cardiovascular models (24) and in the immune response in granulomatous disease (25). These findings, especially the association between the ACE polymorphisms and disease progression in cardiovascular and renal disease, led us to hypothesize that the ACE genotype may be important in the progression of disease in cases of CBD. The purpose of the study reported here was: (1) to investigate whether or not individuals with CBD express an increased frequency of the DD ACE genotype as compared with a control population; and (2) to determine whether the ACE genotype is associated with serum ACE activity or markers of disease severity in CBD.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

We designed a case-control study of subjects with CBD and two control groups, one consisting of beryllium-exposed workers with neither CBD nor beryllium sensitization, and another consisting of office workers from a non-beryllium-associated industry. Fifty cases and 50 controls from each control group were chosen for the study, providing a power of 80% to detect a difference between cases and control groups if the prevalence of the DD genotype was 60% in the cases versus 30% in the control group, corresponding to an odds ratio (OR) of 3.5. The controls were stratified as subsequently described. Cases and controls who were African-American, Hispanic whites, or members of another ethnic group than non-Hispanic whites were excluded from the study to avoid any potential bias related to race or ethnicity. The protocol was approved by the Human Subjects Institutional Review Board of the National Jewish Medical and Research Center, and informed consent was obtained from the study participants.

Case and Control Definitions

Cases of CBD were defined by the presence of: (1) a beryllium-specific immune response as demonstrated by an abnormal blood beryllium lymphocyte proliferation test (BeLPT) and/or BeLPT done on bronchoalveolar lavage fluid (BALF); and (2) histologic verification of granulomas or mononuclear infiltrates on lung biopsy, usually via transbronchial biopsy (1, 6). We defined beryllium sensitization as the presence of a beryllium-specific immune response noted on blood and/or BALF BeLPT without the demonstration of granulomas on biopsy (1, 6). A confirmed abnormal blood BeLPT result was defined as two positive tests obtained on two occasions.

Beryllium-exposed, nondiseased controls were defined as those individuals who had worked in the beryllium industry and who had negative blood BeLPT results on two separate occasions 2 yr apart. Nonexposed controls were individuals who had been employed in a non-beryllium-associated industry.

Study Populations

Case subjects were enrolled from the outpatient Occupational and Environmental Medicine Clinic at the National Jewish Medical and Research Center. CBD was diagnosed in these subjects after they were referred because of respiratory symptoms or an abnormality noted during participation in a beryllium disease surveillance program. These subjects were employed in ceramics manufacture, nuclear weapons production, beryllium processing and manufacturing, and other beryllium-associated industries. Demographic characteristics of CBD cases and controls are shown in Table 1.

A beryllium-exposed, nondiseased control group was drawn from a cohort of workers at a precision metal machining factory where beryllium oxides, alloys, composites, and pure beryllium metal are machined. This facility experienced an index case of CBD in 1995, triggering a plantwide beryllium surveillance program from May 1995 until December 1996. All 213 employees participated in an initial blood BeLPT-based medical surveillance program. After excluding 14 individuals with confirmed abnormal blood BeLPT results, we stratified the beryllium-exposed control group by years of beryllium exposure at the time of the first blood BeLPT (tenure), using 10-yr intervals. One beryllium-exposed control subject was randomly sampled for each case, again with stratification by years of beryllium exposure. This stratified sampling helped mitigate potential differences in duration of exposure between the cases and the beryllium-exposed controls.

The non-beryllium-exposed control population was drawn from a group of 310 office workers in a non-beryllium-associated industry. This group was studied originally to examine the relation between ACE genotype and cardiovascular disease. The nonexposed control population was stratified by age, after which one control was randomly sampled by decade of age (e.g., 20 to 29, 30 to 39, 40 to 49, 50 to 59, and 60 to 69 yr) for each case. Because the ACE genotype has been associated with cardiovascular disease and death from cardiovascular disease (19, 20), the stratification scheme described here should control for any possible change in the prevalence of ACE genotypes that might occur with aging.

Clinical Evaluation

The following clinical evaluation was performed on the cases of CBD after they gave informed consent. Details of the testing methods have been previously outlined (6, 7). All test results used in our study were obtained within 12 mo of the date of diagnosis for each case.

Questionnaire. A self-administered questionnaire based on the American Thoracic Society (ATS) standard respiratory questionnaire was completed by each of the CBD case patients (7). Demographic information was also obtained, and included age, sex, time of first exposure to beryllium, smoking status, and corticosteroid use. We defined a never-smoker as one who had smoked less than 20 packs of cigarettes in his or her lifetime.

Serum ACE. Serum ACE activity was measured with standard commercial laboratory methods (12). Serum ACE activity is expressed in units per liter (U/L), with a normal range for this test of 8 to 52 U/L.

Radiography. Standard posterioanterior chest radiographs were scored by a certified B-reader using the International Labour Organization classification system for pneumoconiosis radiographs (26). For purposes of data analysis, the profusion scores for interstitial opacities were converted to a 10-point scale, with profusion categories of 0/- and 0/0 being scored as zero, and category 3/3 being scored as 10.

Pulmonary physiology. FVC and FEV₁ were measured with a recording spirometer or a pneumotachograph. The maximum value obtained from three satisfactory attempts was recorded. Normal predicted values were obtained from Morris and colleagues (27). Total lung capacity (TLC) and thoracic gas volume (Vtg) were measured in a constant-pressure body plethysmograph. The predicted normal values were derived from those of Goldman and Becklake (28). Results of all of these measurements were expressed as percentages of predicted values.

The single-breath method of Ogilvie and coworkers was used to evaluate the diffusion capacity for carbon monoxide (DL_CO) and the ratio of DL_CO to alveolar volume (DL_CO/VA) (29), both of which were expressed as percentages of the predicted normal values reported by Crapo and Morris (30). Gas exchange, maximum exercise capacity, and maximum oxygen consumption (VO₂max) were determined with a 380 B cycle ergometer (Siemens-Elma, Solna, Sweden) with continuous cardiac rhythm and oxygen saturation monitoring. Inspired and expired oxygen and carbon dioxide were measured through mass spectrometry. By using an indwelling arterial catheter, we measured arterial blood gas at rest and after each minute of exercise. The results are reported as the partial pressure of oxygen (PO₂) and alveolar-arterial oxygen difference ([A-a]PO₂) at rest and at maximum exercise (PO₂max and [A-a]PO₂max).

Bronchoalveolar lavage. Bronchoalveolar lavage was performed by instilling four 60-ml aliquots of sterile saline at room temperature through a wedged bronchoscope into the subsegmental bronchus of the right middle lobe or lingula as described previously (6, 7). The lavage fluid was retrieved by gentle hand suctioning. Total BAL white blood cell (WBC) counts were made on an aliquot of uncentrifuged, pooled lavage fluid according to the methods of Willcox and coworkers (31). Cytocentrifuge preparations were stained with Wright-Giemsa stain and differential cell counts were performed for lymphocytes, macrophages, eosinophils, and neutrophils. Results are expressed as number of cells per milliliter of BALF and as percentages of the total cell count.

BeLPT. The blood and BALF BeLPTs were done according to the methods of Mroz and colleagues (8). Briefly, either BAL cells or Ficoll-Hypaque-separated peripheral blood cells were cultured in quadruplicate in the presence or absence of beryllium sulfate for 5 and 7 d for BAL cells and for 3, 5, and 7 d for blood cells. During the last 4 h of culture, cells were pulsed with 0.5 µCi [³H]thymidine deoxyriboside (Amersham, Arlington Heights, IL), harvested onto glass filters, and counted in a liquid scintillation counter. The counts per minute (cpm) from each set of quadruplicate cultures were averaged and expressed as a ratio of the cpm of the beryllium-treated cells to that of the untreated cells (stimulation index, SI). The peak SI is reported for blood and BAL cells.

Exposure Assessment

Because the CBD cases were drawn from different industries with varying levels of beryllium exposure and varying amounts of sampling data, we could not use industrial hygiene data to evaluate exposure. Instead, the job title of machinist was used as a surrogate marker of exposure, since it has been associated with higher levels of exposure and increased risk of developing disease (2, 3, 10). We classified a worker as a machinist if he or she had ever had the job title of a machinist or had performed machining tasks.

Evaluation of ACE Genotype

Genomic DNA was extracted from peripheral blood leukocytes, BAL cells, or buccal epithelial cells as previously described (20). Briefly, the cells were pelleted and resuspended in 425 µl of buffer consisting of sterile sodium chloride, Tris, and ethylenediamine tetraacetic acid (EDTA) (STE). Fifty microliters of 20 mg/ml proteinase K (Sigma, St. Louis, MO) and 25 µl 20% sodium dodecyl sulfate (SDS) were added to the cell suspension and the mixture was heated at 56° C overnight. The DNA was extracted with 500 µl of Tris-buffered phenol:chloroform (1:1 [vol/vol]), separating the phases by centrifugation. The aqueous phase was mixed with 500 µl of chloroform:isoamyl alcohol (24:1 [vol/vol]) and centrifuged. To precipitate the DNA, 8 M ammonium acetate (pH 4) and isopropanol were added to the aqueous phase from the previous reaction. To increase the yield, this reaction mixture was stored at 20° C overnight. The extracted genomic DNA pellet was resuspended in sterile Tris-EDTA buffer to yield a final concentration of approximately 1 mg/ml.

After extraction of the genomic DNA, a polymerase chain reaction (PCR) was used to evaluate the insertion/deletion polymorphisms in intron 16 of the ACE gene. In the single-step method, primers and PCR conditions were based on those described by Tiret and colleagues (15). Briefly, the PCR reaction was run in a 100-µl reaction volume including 10x PCR buffer, 1.5 mM MgCl₂, 0.125 mM deoxyribonucleoside triphosphates (dNTPs), and 2 U of Taq polymerase (Perkin Elmer, Branchburg, NJ) for 35 cycles with a 7-min extension cycle. The PCR products were electrophoresed on a 2.0% (wt/vol) agarose gel with an ethidium bromide stain. The D polymorphism was 190 bp in length, on an agarose gel, whereas the I gene resolved at 490 bp. An individual was classified as DD only if the 190-bp band was apparent, and as II only if the 490-bp band was seen, or as ID if both bands were visualized (Figures 1A and 1B).

View larger version (7K): [in this window] [in a new window]   View larger version (22K): [in this window] [in a new window]   Figure 1.   Determination of ACE genotypes with the single-step PCR. (A) In the single-step PCR, the PCR primers span the insertion in intron 16 of the ACE gene, yielding two bands of different lengths, depending on the presence or absence of the insertion. The insertion is shown as the hatched segment. (B) Results of the single-step PCR reaction are shown for the ACE genotype. The I allele resolves at 490 bp, whereas the D allele is apparent at 190 bp. The presence or absence of each of these bands determines the genotype.

Since this PCR reaction may fail to amplify the I polymorphism across the insertion, and may thus potentially lead to misclassification of an I allele as a D allele (32), we used a nested PCR reaction that extends the insertion segment (33). The first reaction in this PCR amplifies intron 16 across the insertion, with use of the two primers GIIS (5'-CTCAAGCACGCCCCTCACAGGACTG-3') and GAS (5'-GATGTGGCCATCACATTCGTCATCAGAT-3'). The reaction mixture was heated to 95° C for 1.5 min and then amplified for an additional 20 cycles of 95° C for 30 s, 62° C for 30 s, and 72° C for 30 s. A 50-µl reaction volume was used, containing 400 pmol of primers, 2 mM MgCl₂, 0.25% dimethyl sulfoxide (DMSO), and 2.5 U Taq polymerase. The tubes were cooled to 4° C after 20 cycles, and 400 pmol of GIIS and FYM primers (5'-ATCACGAGGTCAGGAGATCGGGAGAC-3') were added to the reaction mixture. The PCR was then continued for an additional 20 cycles. The FYM primer is internal to the insertion in intron 16. As a result, the second PCR reaction amplifies only the I allele (Figure 2A). A 561-bp PCR product results from the extension of the I allele between the GIIS and GAS primers, and a 274-bp product results for the D allele. The extension between the FYM and GIIS primers results in a 376-bp product for the I allele only (Figures 2A and 2B) (33). The samples were run and interpreted by a technician who was blinded to the subject's status or diagnosis for both PCR reactions.

View larger version (6K): [in this window] [in a new window]   View larger version (38K): [in this window] [in a new window]   Figure 2.   Determination of ACE genotypes with the nested PCR technique. (A) The diagram represents the location of the three PCR primers on the I and D alleles. The first PCR reaction, involving GIIS and GAS primers, results in two products of different lengths, depending on the presence or absence of the insertion in intron 16. The insertion is indicated by the hatched segment. The second PCR reaction produces a 376-bp product for the I allele only, since the FYM primer is internal to the insertion in intron 16. (B) The nested PCR results for the ACE genotype are shown. The first reaction with the GIIS and GAS primers results in a 561-bp fragment for the I allele and a 274-bp fragment for the D allele. The second PCR reaction, using the GIIS and FYM primers, amplifies the 376-bp fragment within the insertion. The II genotype is indicated by presence of the 376-bp and 561-bp bands, the DD genotype is represented only by the 274-bp band, and the ID genotype is represented by the presence of all three bands.

We found 15% misclassification between the single-step and nested PCR reactions (n = 109). The majority of the discrepancies resulted in misclassification of an ID as a DD genotype with the single-step method (12% disagreement, or 13 of 109 samples). Three subjects of genotype II were misclassified as having an ID genotype (3%). It has been our experience that the yield of DNA after extraction is inversely proportional to the WBC count and the age of the sample. Upon review of our samples, we found that older DNA samples were the ones most likely to be misclassified with the single-step PCR. Furthermore, we found that the nested PCR reaction could reveal the ACE genotype in a few individuals in which the single-step method could not. As a result, the nested PCR was used for the remainder of the study samples and for purposes of analysis.

Statistical Analysis

To examine possible interactions between demographic variables and genotypes in the cases versus the controls, we used stepwise logistic regression analysis. The following variables were entered into the logistic regression model: age at the time of diagnosis, sex, job tenure, and an interaction term for the combination of each of these variables. Because no significant associations were found between any of these variables and the DD genotype for cases as compared with the controls in either control group, we calculated an unadjusted OR with 95% confidence intervals (CIs) to evaluate the degree of association between the DD genotype versus non-DD genotype (ID and II) in cases of CBD as compared with each of the control groups. Continuous variables were compared in cases and controls, and between DD cases and non-DD cases, using Student's t test or Wilcoxon's rank sum test, as appropriate. For the three-way analysis of continuous variables, the Kruskal-Wallis test was employed. Chi-square or Fisher's exact tests were used to compare categorical variables and cases versus controls or genotypes. Dunn's method was used to control for each set of three pairwise comparisons. Spearman's rank correlation coefficient () was used to evaluate the relationship between continuous variables. All statistical analyses were performed using JMP-SAS or SAS (SAS Institute, Cary, NC). Comparisons were considered significantly different when p < 0.05. All tests were two-sided.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Demographics of Cases and Controls

A total of 150 non-Hispanic white individuals were enrolled in this study, consisting of 50 cases and 50 subjects from each of the two control groups (beryllium-exposed and non-beryllium-exposed). The characteristics of these 150 participants are shown in Table 1. No significant difference between the CBD cases and the beryllium-exposed controls was found in the number of years worked in a beryllium-associated industry (job tenure) (p = 0.46). The beryllium-exposed controls were younger than the CBD cases (median age: 42 yr versus 48 yr, respectively, p = 0.002). The non-beryllium-exposed control group contained more women than did the CBD group (p < 0.0001). The cases came from beryllium-associated industries, which have a predominantly male workforce, whereas the controls were taken from an office-based workplace with a mixed male and female workforce. We observed no difference in the age of the CBD cases and non-beryllium-exposed controls (p = 0.189).

Relationship between ACE Genotype and CBD

The ACE genotype frequencies of the I and D allele did not differ in the CBD cases versus either control group (p = 0.12 for the CBD cases and the beryllium-exposed controls and p = 0.56 for the CBD cases versus the non-beryllium-exposed controls) (Table 2). The genotype distribution for the cases and each of the control groups is shown in Table 2. We observed no statistically significant difference between the ACE genotypes when the CBD cases (DD versus non-DD) were compared with each of the control groups (p = 0.29 for the CBD cases versus the beryllium-exposed controls, and p = 0.84 for the CBD cases versus the non-beryllium-exposed controls). However, the frequency of the DD genotype was noted to be slightly higher among the CBD cases than among the beryllium-exposed controls (38% versus 28%), whereas the frequency of the II genotype was lower in the cases (12% versus 24%). The OR for the DD versus non-DD cases was 1.58 (95% CI: 0.68 to 3.66), as compared with the beryllium-exposed control group, and 1.090 (95% CI: 0.48 to 2.46) as compared with the non-beryllium-exposed controls. The gene frequencies were in Hardy-Weinberg equilibrium for each of the three groups (p = 0.83 for CBD cases, p = 0.99 for the beryllium- exposed controls, and p = 0.96 for the non-beryllium-exposed controls).

If only the single-step PCR had been used to classify the ACE genotypes of the cases and controls, instead of the nested PCR, markedly different results would have been achieved. With the single-step PCR method, 58% of the CBD cases (n = 48) would have been classified as DD, whereas 27% of the beryllium-exposed controls (n = 26) and 31% of the non-beryllium-exposed controls (n = 35) would have been classified as DD. Because misclassification of the ID genotype was most prominent among the cases, an association between the DD genotype and CBD would have been found with the single-step PCR. That is, a statistically significant difference would have been found between the DD cases and beryllium-exposed controls, with an OR of 3.80 (95% CI: 1.37 to 10.54; p = 0.01), and between the CBD cases and the non-beryllium-exposed controls (OR = 3.06; 95% CI: 1.23 to 7.57; p = 0.01).

Characteristics of CBD Cases by ACE Genotype

As shown in Table 3, we observed no difference in the sex, use of corticosteroids among, or smoking status of CBD cases with the DD and non-DD genotypes (ID or II). Individuals with the DD genotype were as likely to have been diagnosed through a medical surveillance program rather than from clinical symptoms as were the non-DD cases (n = 12 DD cases, or 63%, of all DD cases, that were diagnosed through medical surveillance, versus n = 20 non-DD cases, or 65%, that were so diagnosed, p = 0.92). Although there was no statistically significant difference in the age at diagnosis among the DD subjects, they tended to be younger than non-DD cases at the time of diagnosis (median age: 44 yr versus 50 yr, respectively, p = 0.29). The DD cases also tended to have a shorter latency or time from first exposure to beryllium to disease diagnosis than did the non-DD cases (median: 11.5 yr versus 18.1 yr, respectively, p = 0.26). The ID cases had a median latency of 18 yr, whereas the II cases had a latency of 20 yr from first beryllium exposure to diagnosis of CBD, which in the latter instance was almost twice that of the DD cases. A correlation was found between the age at diagnosis and time since first exposure to beryllium, such that younger cases of DD genotype were more likely to have had a shorter time since first beryllium exposure ( = 0.64, p < 0.0001). No association was found between work as a machinist and the DD genotype (p > 0.05, n = 7 of 19, or 37%, versus n = 8 of 31, or 26%, for the non-DD cases).

Serum ACE Activity and Genotype

Serum ACE activity was associated with the ACE genotype in CBD cases (p = 0.005), as shown in Figure 3. Individuals with the DD genotype had the highest serum ACE activity (median: 47.5 U/L), whereas cases with the II genotype had the lowest activity, at a median of 13 U/L, and cases with the ID genotype had intermediate activity (median: 43 U/L). Using Dunn's adjustment for multiple comparisons to compare the cases of DD genotype versus those of the ID and II genotypes, and cases of the ID versus the II genotype, we found a significant difference in serum ACE activity for the DD versus II and the ID versus II genotypes (p < 0.05).

View larger version (14K): [in this window] [in a new window]   Figure 3.   Serum ACE activity is associated with the ACE genotype in CBD (p = 0.005). Using Dunn's method for multiple comparisons, we found a significant difference between serum ACE activity for the DD and II and for the ID and II genotypes (*p  0.05).

Relationship between ACE Genotype and Markers of Disease Severity in CBD

We evaluated blood and BAL BeLPT results, BAL cellularity (Table 4), chest radiographs, and pulmonary physiology (Table 5). Although peak blood and BALF BeLPT results were not statistically associated with ACE genotype (p = 0.20 for blood BeLPT and p = 0.26 for BALF BeLPT), the cases of DD genotype tended to have a lower median BeLPT response in both the blood and BALF. This was consistent across all three genotypes, such that cases of CBD with the II genotype showed the highest response (median SI: 31.15 in blood and 120.25 in BALF), and cases with the DD genotype showed the lowest response (median SI: 5.95 in blood and 16.7 in BALF), and the cases with the ID genotype were intermediate (median SI: 12.91 in blood and 23.25 in BALF). These differences were not statistically significant for comparison of blood (p = 0.44) or lavage (p = 0.10) values for the three genotypes, although a trend was apparent in both fluids. Because the blood and BALF BeLPT results were not correlated ( = 0.031, p = 0.84), it is unlikely that the association that was observed was solely the result of a few individuals with low or high blood and BALF BeLPT responses. No association was found between other BALF markers and ACE genotype in CBD, including total WBC/ml or percent or absolute number of lymphocytes or macrophages (Table 4).

Although chest radiographic abnormalities have previously been associated with serum ACE activity in CBD (13), we found no association between profusion of small opacities on chest radiography and ACE genotype in our cases (p = 0.34) (Table 5). Neither pulmonary function indices nor exercise physiology variables were associated with ACE genotype in CBD cases (Table 5).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

CBD is likely to be a multigenetic disease, with a number of genetic factors contributing to the development of an immune response to beryllium and the progression to disease (5). We hypothesized that polymorphisms in the ACE genotype would be important in the development of CBD and in its severity. However, we did not observe an association between the ACE genotype and CBD in comparisons with two separate control populations. We did confirm previous findings that the ACE genotype accounts for variability in serum ACE activity, and achieved results suggesting that the DD polymorphism may play a role in the progression to disease in CBD and the development of an immune response to beryllium.

Angiotensin-1 converting enzyme (EC 3.4.15.1) converts angiotensin I to angiotensin II. It inactivates bradykinin, a vasodilatory peptide implicated in inflammatory reactions. Serum ACE levels have been known to be elevated in sarcoidosis, a granulomatous disease of unknown etiology, as well as in CBD and some other diseases, since the 1970s (34). The significance of a biologic role for serum ACE activity in sarcoidosis and CBD remains unclear. It has been postulated that serum ACE activity reflects the body's granulomatous burden, and has in fact been shown to reflect the amount of ACE mRNA in an animal model of granulomatous disease (35). Earlier findings of higher serum ACE activity in some ethnic groups (34), and similar ACE levels in families (15), led to the hypothesis that serum ACE activity may be genetically determined. Subsequently, a polymorphism in the ACE gene, which varied with the insertion or deletion of a 287-bp sequence in intron 16, was shown to account for 47% of the variability in serum ACE activity (16), with individuals of the DD genotype exhibiting the highest activity, those of the II genotype the lowest activity, and heterozygotes exhibiting intermediate serum ACE activity. It is unclear whether the observed effects on serum ACE activity are related to the ACE gene locus or whether they are due to another gene in linkage disequilibrium with, or located closely to, the ACE gene; Tiret and coworkers have provided evidence that the ACE locus controls serum ACE activity, although the I/D polymorphism is most likely a marker of a nearby gene that regulates ACE activity (15). The ACE genotype has been associated with various other diseases, through unclear mechanisms (18).

Despite the well-established increase in serum ACE activity in granulomatous disorders, three previous studies, two in Japan (17, 36) and one in Europe (37), failed to find an association between sarcoidosis and the ACE genotype in comparisons with ethnically matched controls. We had questioned whether the lack of an association between ACE genotype and sarcoidosis was a result of that disease's heterogeneity. It is possible that the association between ACE and sarcoidosis is limited to a subset of individuals with sarcoidosis, or to individuals with a specific clinical pattern of that disorder. Previous studies have combined individuals with different forms and severities of sarcoidosis, potentially masking significant findings (17, 36, 37). The ACE genotype had not been evaluated previously in CBD. Even though CBD is a more homogeneous granulomatous disease than is sarcoidosis, we found no association between the ACE genotype and CBD in comparisons with two control populations. The frequency of the DD genotype observed in our cases was higher, but similar, to that reported for other white control populations (24 to 30%) (18, 32). Although the difference was not statistically significant, our beryllium-exposed control group was found to have a slightly lower frequency of the DD genotype (28% versus 38%) and a lower frequency of the D allele (52% versus 63%) than did our CBD cases. It is possible that a beryllium-exposed work population could have a lower frequency of the DD genotype than that of CBD cases if the cases are more likely to be of the DD genotype and are thus removed from the beryllium-exposed workforce because of their disease. However, our power to detect this difference between the beryllium-exposed controls and CBD cases in our study was only 16%, given our limited sample size. A much larger sample size would have been necessary to detect a difference of 38% versus 28% in the DD genotype in our cases and controls.

The use of two control groups in our study offers an advantage over a design using only a non-beryllium-exposed workforce or hospital-based control group, since it avoids any bias in genotypes that might be related to a healthy workforce effect with or without beryllium exposure. Stratifying the beryllium-exposed controls in our study by years of exposure should have insured that the duration of exposure to beryllium was similar to that for the cases of CBD, and should have decreased lead-time bias or the chance of misclassification related to length of follow-up. The cases of CBD and the beryllium-exposed controls originated from different manufacturing industries, although many of both the cases and controls were employed in machining beryllium. Beryllium machining is associated with higher rates of sensitization in both the ceramics and nuclear weapons industries (2, 3). Using this crude surrogate marker of exposure, we found no association with our DD cases. This is not surprising, considering the different industries and potentially different exposures that the CBD cases may have had. To accurately evaluate the impact of beryllium exposure and genetics, a case-control format with individuals from the same facility should be used in future studies, with industrial hygiene measures of exposure.

The use of a nested PCR reaction with primers internal to the insertion site in the ACE gene (33) reduced potential misclassification of the genotypes in our study. The nested PCR method has previously been shown to produce results identical to those of hybridization techniques employing oligonucleotides specific to the I and D alleles (33). Previous studies have found a misclassification rate of approximately 4 to 5% or less with the single-step PCR reaction (32, 33, 36). We observed a higher rate of misclassification in our cases and controls, with an overall misclassification rate of 15%. In the past, we had observed a lower rate of DNA recovery in DNA extraction from older samples and those with a lower WBC count, as well as misclassification with use of the single-step PCR reaction on such samples. The DNA samples from our CBD cases that were misclassified tended to be older samples than those from the controls, and tended to yield less DNA after DNA extraction. If we had not run a confirmatory second reaction, our results would have been skewed, since we primarily found that cases of ID genotype were misclassified as DD to a greater extent than in the control groups. Specifically, had we not used the nested PCR method, we would have concluded that the DD ACE genotype was more frequent in our CBD cases than in either control group, and was thus associated with CBD. This raises a note of caution for investigators who may have based their conclusions on the single-step PCR method in studying other disorders.

Although we did not find any association between ACE genotype and CBD, we did confirm previous findings that serum ACE levels are associated with ACE genotype (16, 17). Specifically, we found that cases of the DD genotype produced the highest levels of serum ACE, whereas CBD cases with the II genotype produced the lowest levels, and heterozygotes produced intermediate levels. Similar associations have been found between ACE genotype and T-cell ACE activity (38) and serum ACE levels in patients with sarcoidosis (17, 36). As a result, Furuya and colleagues have suggested that normal serum ACE values should be adjusted for ACE genotype, in an effort to improve the use of the test as a marker of disease activity in sarcoidosis (17).

Newman and coworkers had previously shown an association between serum ACE levels and the inflammatory response in the lung as measured through BALF cellularity and lymphocytosis and radiographic abnormalities in CBD (12, 13). Thus, we hypothesized that the DD genotype would be associated with markers of disease severity in CBD. Previous studies have found an association between the DD genotype and the severity of other diseases, including proteinuria (22) and decline in the glomerular filtration rate (39) in diabetic nephropathy, requirement for renal replacement therapy in IgA nephropathy (23), and death in hypertrophic cardiomyopathy (19). In our CBD cases we found no association between ACE genotype and markers of disease severity including radiographic abnormalities, pulmonary function test results, exercise tolerance, or gas-exchange abnormalities. This lack of an association could have resulted from the small number of cases in each genotype category in our study. Because the abnormalities just described tend to occur late in CBD (6), and many of the patients in our study were diagnosed as having CBD through medical screening programs at a time when they were relatively asymptomatic, it is not surprising that we found no association between disease severity and ACE genotype. Our study suggests that an individual's ACE genotype may be associated with disease latency, given that CBD subjects with the DD genotype were younger at the time of diagnosis and developed disease after a shorter exposure to beryllium than did those of the II or ID genotype. Because these two variables were correlated, it is not surprising that we observed that cases of the DD genotype both tended to be younger and to have had a shorter period to diagnosis after their first beryllium exposure. Thus, an individual's ACE genotype may enhance the progression to disease in CBD even if it is not associated with a more severe disease outcome. Analogously, in IgA nephropathy, the DD genotype has been associated both with onset of disease at a younger age and with disease progression (23).

A further finding was that the BeLPT response in both blood and BALF appeared to be weaker in the DD cases than in the non-DD cases. Interestingly, prior studies have shown a correlation between BALF but not blood BeLPT results and markers of disease severity, in that individuals with a greater BALF BeLPT result had a more cellular BALF, more restrictive pulmonary function test findings, and decreased exercise tolerance (7). This contrasts with what we would have expected, given the observed associations between disease severity and the DD genotype in previous studies (19, 22, 23, 39). Although it did not reach statistical significance, this finding still raises the question of whether ACE may have an effect on the immune response in CBD. Treatment with an ACE inhibitor of mice infected with Schistosoma mansoni led to a decreased granulomatous response in the liver and the lung, indicating that ACE may be important in the inflammatory response in granulomas (25). ACE expression has been found for alveolar macrophages (40) and epithelioid cells within granulomas (41). Although these studies suggest that ACE may play a role in the immune response in granulomatous disease, the precise mechanism of its action and function in granulomatous diseases remains elusive and was not the purpose of the present study.

The findings of our study raise a number of issues about the role of genetics in the development of CBD. Although we did not find a direct association between ACE genotype and CBD, our study fuels speculation about the potential role of the ACE genotype in the cellular immune response to beryllium and in progression to disease. Because CBD is most likely a multigenetic disease process (5), the ACE locus and other genetic loci may interact with one another in the development of disease. For example, Tiret and coworkers found that the association between ACE genotype and myocardial infarction was limited to a subset of individuals with a specific point mutation in the gene for the angiotensin II type I receptor, showing that these two gene loci interacted in conferring disease risk (18). Furthermore, the effects of the interaction were most pronounced in individuals with an otherwise low risk of coronary artery disease (18). This emphasizes the need to evaluate the contribution both of exposure and of multiple genetic influences, especially in an occupationally mediated disease such as CBD, which is most likely the result of both susceptibility and exposure. Similarly, it will be important for future studies to evaluate the potential interactions of known CBD genetic risk factors in the HLA domain (9, 11) with angiotensin-system genes in order to better define the multigenic contributors to CBD susceptibility. Currently, the interaction between HLA and the angiotensin-system genes is unknown. By studying the interaction between these genetic markers and facets of exposure, we hope to gain greater insight into the pathogenesis of CBD and other granulomatous diseases, and ultimately to provide the ability to predict and prevent these diseases.