Secondary Logo

Journal Logo


Chronic Beryllium Disease

The Search for a Dose-Response

Borak, Jonathan MD

Author Information
Journal of Occupational and Environmental Medicine: November 2016 - Volume 58 - Issue 11 - p e355-e361
doi: 10.1097/JOM.0000000000000869
  • Free

The dose-response relationship, “the most fundamental and pervasive concept in toxicology,”1 is central to determining the toxic characteristics of chemicals. That relationship underlies the First Law of Toxicology attributed to the 16th century Swiss physician Paracelsus: “the dose makes the poison.” Likewise, the potency of a toxic chemical is defined by the slope of its dose-response: the steeper the slope, the greater the potency. The presence of clear dose-relatedness (eg, a monotonic “biological gradient”) is also a key component of Hill's Criteria for inferring causality from epidemiologic data: “The clear dose-response curve admits of a simple explanation and obviously puts the case in a clearer light.”2 Thus, the presence of well-defined dose-relatedness provides a solid basis for thinking about the toxicological implications of chemical exposures.

Understanding a chemical's dose-response is also important for developing and implementing appropriate workplace exposure standards and occupational health practices. However, for most occupational exposures in most workplaces, “dose” is not known (The term “dose” has various meanings in occupational epidemiology, ranging from “the amount of material taken into the body” to “the amount of biologically active material at a critical organ or tissue.”)3 Instead, it is generally more correct to speak of “exposure-response” relationships, with the understanding that measures of “exposure” serve as surrogate measures of dose.4 Nevertheless, the terms “dose-response” and “exposure-response” have often been used interchangeably.

Lack of explicit, quantitative dose-response information complicates the establishment of workplace health and safety programs. Even in situations wherein epidemiological data are more than sufficient to conclude that a particular exposure causes a specific disease, uncertainty about the underlying dose- or exposure-relatedness of disease can challenge the setting of appropriate exposure limits. This has been well demonstrated by the sometimes frustrating effort to protect workers from beryllium exposure and chronic beryllium disease (CBD). The absence of clear dose-relatedness has not deterred use of epidemiologic data to conclude that beryllium exposure causes CBD. That association has been recognized for more than 70 years.

But until very recently, the dose-relatedness of beryllium exposure was “obscure”5 and “unclear.”6 In turn, the lack of an objective dose-response curve has complicated development of appropriate health and safety programs for beryllium-exposed workers. This is reflected by long-standing concerns regarding the adequacy OSHA's 2 μg/m3 Permissible Exposure Limit (PEL)7 and historical uncertainty that removal of sensitized workers from further exposure will lower their risks of CBD.8,9 Fortunately, more sophisticated exposure assessments and greater understanding of the genetics of CBD have allowed beryllium's dose-relatedness to be more fully revealed. Much of that increased knowledge has now been incorporated into OSHA's proposed Beryllium Standard.10

The following report describes the evolution of current understanding of beryllium's dose-relatedness, with a particular emphasis on the important role played by the evolving science of exposure assessment. It is based on a review of the author's extensive collection of beryllium-related reports, but it is not intended to represent a systematic review of the entire published literature on beryllium and beryllium disease. Despite efforts to be inclusive, it is likely that some relevant publications have been missed. Readers should recognize that other important scientific advances, including dramatically greater understanding of the genetics and immunotoxicology of beryllium disease and the widespread adoption of lymphocyte proliferation testing and fiberoptic bronchoscopy, have essentially “revolutionized”11 understanding about beryllium-related health effects. All of these advances contribute to the design of optimal worker protection.


CBD is an immunologically mediated granulomatous disease that most importantly affects the lungs. The first clear association between beryllium exposure and CBD was a 1946 report of chronic lung disease in fluorescent light bulb workers exposed to beryllium-containing phosphors.12 Shortly thereafter, CBD was also described among workers at a large Ohio manufacturing facility and (as discussed below) among residents of a contiguous community. The geographic distribution of the cases and the apparent lack of a dose-response relationship led early investigators to speak of the “bizarre” and “puzzling” epidemiology of CBD.13–15

Over the following decades, much scientific interest focused on CBD and considerable information accumulated, but its dose-relatedness remained obscure. The inability to document dose-response was generally attributed to two factors: the immunological basis of CBD and the limitations of the existing industrial hygiene data. This was explicitly discussed in two major reviews of beryllium and lung disease published during the 1980s. Cullen et al5 observed that the “dose-host-response relationship is obscure, especially at low levels of exposure … the dose-response curve … is still largely unknown,” which they attributed to the “incompleteness of the data from the classic studies,” particularly deficiencies of exposure data. Likewise, Kriebel et al,16 in a State of Art review, discussed the important limitations of then existing data for the determination of beryllium's dose-relatedness: “delineating the importance of dose-rate versus total cumulative dose is difficult from epidemiologic data, yet it is crucial to setting of adequate hygienic standards.”

Twenty years later, in the Annual Review of Public Health, Kreiss et al8 described the “absence of effective exposure-response information” for beryllium: “Measuring beryllium mass concentrations in air does not predict risk of sensitization or chronic beryllium disease.” The following year, the National Research Council (NRC) found that the dose-response relationship between beryllium and CBD remained “unclear”: “the available human and animal data are inadequate to support a dose-response analysis.”6 NRC blamed both uncertainty in determining beryllium exposure levels and the immunologic mechanisms of CBD.


Since earliest reports, CBD has been viewed as a “puzzling disease,”12 in part due to the apparent lack of a classical dose-response between exposure and disease. During the 1940s, for example, CBD was described in 10 adults who lived in the vicinity of a large beryllium manufacturing facility. Those “neighborhood cases” had never worked with beryllium, and did not reside in the same household as a beryllium worker.13,14 Further investigation of the plant and its neighbors revealed a striking inconsistency, which the investigators described as “bizarre” and “puzzling.”15

On the one hand, the geographic distribution of “neighborhood cases” did seem to follow a classical dose-response relationship. Individuals living closer to the plant, who were exposed to higher levels of beryllium in plant effluent, had higher incidence rates than those living further away. Within one-fourth mile of the plant, the incidence rate was approximately 1%; the rate declined as distance from the plant increased, and no cases were found in residents who lived more than three-fourth mile from the plant.15 On the other hand, the reported incidence among plant workers, where exposures were routinely “orders of magnitude higher” than those in the neighborhood (>1 mg/m3 vs <1 μg/m3), was about 0.35%, significantly less than that among the neighbors.15,17

To explain this “bizarre epidemiology,” Sterner and Eisenbud correctly proposed that CBD was an acquired immunological disorder and that susceptibility (ie, the capacity of an individual to become sensitized) was largely determined by inherited capacity, not dose. However, to explain the distribution of neighborhood cases, they also proposed that sensitization, and therefore chronic disease, followed a classical dose-response among those with appropriate susceptibility. They suggested that the distribution of neighborhood cases indirectly defined “the lowest effective sensitizing stimulus.”15

Lacking the toxicological and epidemiological dose-response data necessary to calculate a worker-protective occupational exposure limit (OEL), in 1949, the Atomic Energy Commission (AEC) recommended an OEL of 2 μg/m3 based on analogy to other toxic metals,18 a standard that was widely adopted.7 For nearly 35 years thereafter, evidence seemed to indicate that 2 μg/m3 was an exposure-response threshold below which CBD did not occur.

In the 1980s, however, the adequacy of the 2 μg/m3 standard came to be questioned because of several reports of CBD in workers who seemingly had not been exposed above the standard. Along with those questions came renewed interest in the details of beryllium's dose-response. In 1983, Cotes et al19 described four cases of CBD in workers whose “estimated average daily exposure” based on annual averages of area sampling data did not exceed 2 μg/m3. They concluded that it was not possible to know whether the standard was adequate because so “little information is available … on the dose-response relationship.” Cullen et al,20 in 1987, described CBD in five metal refinery workers whose beryllium exposures were said to have averaged less than 2 μg/m3. They noted that “the dose-response to beryllium … has puzzled investigators for 4 decades.”


An important aspect of the renewed concerns about the 2 μg/m3 standard and beryllium's dose-response was the adequacy of the data used to characterize beryllium exposures. As in other scientific fields, exposure assessment methods have evolved significantly over the past decades, both sampling methods and analytical technologies. Early efforts depended on fixed monitors to obtain area samples or cumbersome hand-held monitoring devices for short-term samples, with results assigned to all workers in a job or area.4,21 Later, personal monitoring methods were developed that allowed full-shift samples to be obtained from individual workers, but reliance on fixed samplers persisted in many work sites. Over the years, a number of such approaches were employed to characterize workplace beryllium exposure. Results of exposure assessments differed depending on the assessment methods used.

The original AEC method involved collection of total beryllium particulate using high-flow air samplers for both general area and breathing zone exposures, and also time-motion studies of each worker's shift routine. Air sampling results collected over a quarter year were combined with findings of the time-motion study to calculate a worker's “daily weighted average” (DWA) exposure.17,22,23 In most cases, DWAs reflected averaged worker exposures by job title, using a formula that incorporated “average general area, full-shift area and breathing zone measurements based on time studies for most jobs,”24 thus measuring “exposures for jobs rather than for individual workers.”25

That approach was also adopted by NIOSH in 1972.26 The NIOSH method included “general air samples” for total beryllium (the average of at least three area samples of 30 to 120 minutes duration per week for each location) and total beryllium breathing zone samples (the average of at least three samples of 3 minutes or longer for “each operation”). Sampling results were combined with time-motion studies to calculate “time-weighted average” (TWA) exposures “for each work area,” but not necessarily for each worker. (Although referred to by NIOSH as a TWA, exposures were calculated in the same manner as the AEC-DWA).

Full-shift personal lapel breathing zone (PBZ) air sampling was only slowly adopted. During the 1980s, most beryllium facilities relied on calculated DWAs to characterize worker exposures. In addition, some facilities used neither DWA nor lapel sampling methods, instead relying almost entirely on general area sampling. For example, at Rocky Flats Environmental Technology Site, a former nuclear weapons facility, daily beryllium samples were collected beginning in 1958, yielding a total of more than 500,000 individual samples by 1988.27 These were almost entirely area samples; only about 20 “useful” PBZ samples were obtained before 1984.28 Moreover, area samples were obtained using fixed airhead (FAH) samplers affixed to various machines, some of which were installed in locations such that they were not representative of worker exposure.27

Reliance on area samples poses problems for historical exposure reconstruction, as well as efforts to characterize dose-relatedness, because area samples and calculated DWAs correlate poorly with PBZ samples. In a 1980 NIOSH study, three different beryllium sampling methods were conducted simultaneously and compared22: (1) the AEC-DWA method; (2) TWAs based on PBZ total beryllium samples; and (3) TWAs based on PBZ respirable beryllium samples. Exposure concentrations measured by the three methods differed significantly. Correlations between AEC-DWA results and corresponding PBZ TWA results ranged from 0.33 [R2 = 0.11] (respirable beryllium) to 0.49 [R2 = 0.24] (total beryllium). In addition, there were wide confidence intervals (CIs) for estimated AEC-DWAs; AEC-DWAs were significantly lower than corresponding PBZ total beryllium TWAs, and no consistent conversion factor could be determined. The NIOSH authors concluded that it was “not possible” to accurately estimate personal respirable or total time-weighted average exposures from DWA values.22

Similar difficulties were documented in other studies. An evaluation of sampling at Rocky Flats found virtually no correlation between matched pairs of FAH and PBZ samplers (R2 = 0.014).27 The geometric mean of PBZ samples was more than 10-fold greater than that of FAH samples (0.64 vs 0.06 μg/m3) and 95% CIs around FAH and PBZ samples did not come close to overlapping (0.10 to 0.22 vs 0.79 to 1.29 μg/m3). In another study of nuclear workers, there was “no correlation” between 118 pairs of stationary and personal air samples (“correlation coefficient = 0.038”) [R2 = 0.001], and historical exposure could not be reconstructed.29 Likewise, a study of a beryllium manufacturing facility found that the correlation between 103 pairs of AEC-DWA and PBZ TWA samples was only 0.26 [R2 = 0.07] and PBZ samples were significantly higher than matched DWA samples: “DWAs seem to be a poor estimate of personal exposure.”30 Nevertheless, reflecting efforts to make the most of the limited data, some studies relied on exposure assessments in which DWA and PBZ samples were interconverted and pooled using a fixed conversion factor.25

Thus, older beryllium workplace exposure assessments could provide useful qualitative estimates, for example, such data might serve to compare the relative exposures of different jobs or they could be used to evaluate the relative effectiveness of different exposure controls. But they rarely provided accurate exposure measures suitable for quantitative risk assessment, instead providing mainly crude approximations of workers’ exposure levels.

Even among contemporary studies that included comprehensive full-shift PBZ sampling, estimation of cumulative and life-time average exposures has often depended on exposure reconstructions utilizing older exposure data, almost always calculated as DWAs. For example, in a study of beryllium ceramics workers employed between 1981 and 1992, there were 4890 general area samples, 4133 breathing zone samples of 1 to 15 minutes each, but only 75 PBZ-TWA samples, all obtained after 1991.24 In the study by Couch et al,31 a retrospective NIOSH study of 9199 workers at seven beryllium facilities employed between 1940 and 2005, all pre-1995 exposure estimates were based on calculated DWAs; personal lapel sampling data were available only after 1995. But, because lapel samples “cannot be correlated to the method used prior to 1995” (ie, DWAs), the authors excluded the lapel samples. Instead, post-1995 exposures were “extrapolated from the previous exposure estimates.”

A different problem has been how best to reconstruct historical exposures when some of the historical data are missing. For example, in a study of beryllium machining workers, Kelleher et al32 performed rigorous exposure measurements during 1996 to 1999, but historical data were available for only 1981 to 1984. The authors could not find data for 1985 to 1995, and they could not determine the original purpose or “specific methods of analysis” of the 1981 to 1984 samples. In lieu of empirical data, the authors performed exposure reconstruction by assuming that the missing exposure data were the same as those measured during 1996 to 1999.

In a recent risk assessment of that facility's workers, OSHA incorporated the Kelleher data along with additional historical facility exposure data located by Madl et al,33 data from two surveys by an OSHA contractor, an earlier OSHA site inspection, comments submitted to OSHA in 1976, and discussions with the plant's industrial hygienist. Nevertheless, OSHA warned that its findings should be interpreted “with caution” because of data limitations that included “the size of the dataset, relatively limited exposure data from the plant's early years … and limited follow-up time on many workers.”10


Rather than relying on potentially inaccurate exposure reconstruction data, some investigators turned to surrogate measures to help explain beryllium's exposure-response characteristics. Exposure duration has been often used as such a surrogate, even in studies that also included quantitative industrial hygiene data. Results have been inconsistent. Some studies found that duration of exposure positively predicted CBD incidence rates. In the report of an early CBD screening program in the ceramics industry, Kreiss et al34 relied on self-reported work histories, not industrial hygiene data, to characterize exposures in 505 workers. Disease rates were positively associated with “degree of beryllium exposure,” particularly duration of exposure (“cumulative months”). In another ceramics facility, Henneberger et al35 compared CBD rates in 74 workers with short-term (less than 6 years) and 77 with long-term (8 to 40 years) exposure. CBD was found “almost exclusively” among the long-term workers (7/77 vs 1/74).

The use of both exposure duration and historical reconstruction was juxtaposed in the study by Schubauer-Berigan and colleagues,31,36 a mortality study that included workers employed at seven beryllium facilities between 1940 and 1970. At three facilities, workers’ exposure were estimated using “qualitative” job exposure matrices based on historical DWA calculations, although one of those plants had essentially no pre-1970 data37; all facilities had duration of employment data. The analysis did not directly consider CBD, instead analyzing “categories containing CBD” (eg, “pneumoconiosis other than asbestosis or silicosis”). The study found “sharply increasing SMRs with increased employment duration,” a highly significant trend (P < .0001), but no significant associations with estimated cumulative or maximum exposure.

Other studies found no association between exposure duration and disease. Rosenman et al38 conducted a screening program of 577 former workers at a large beryllium processing facility in order to investigate “possible exposure-response relationships.” There was no difference in exposure duration for individuals with CBD compared with those who had “no evidence of beryllium disease.” Stange et al39 reviewed CBD screening histories of 5173 current and former workers in the Rocky Flats Beryllium Health Surveillance Program (BHSP). The authors noted “no strong association” between the occurrence of CBD and the duration of exposure to beryllium, measured as the length of time spent in beryllium production.

Another approach considered CBD dose-response rates using specific job tasks as exposure surrogates. Stange et al39 reported a dose-response “for the majority” of 22 job groupings based on their “relative potential for exposure to beryllium,” but there were “no quantifiable exposure data” for the majority of job groups. That study and others have repeatedly found highest rates in beryllium machinists.24,30,34,39,40 Such an association might reflect average or peak levels of exposure, which are often higher in workers who perform machining and related beryllium work than workers performing nonmachining jobs in the same facilities. But the association may also be due to specific properties of the particulate generated during those processes, such as particle size, particle surface area, and particle number.24,41–44 Thus, higher rates of CBD might be attributable to characteristics of the work process, rather than dose. Nevertheless, knowledge of process-related risks provides useful guidance for planning workplace surveillance programs.


The existence of a positive dose-response between exposure and CBD has been best demonstrated in three more recent studies that employed particularly robust data samples to quantify worker exposures at the individual level. Two studies used exposure reconstructions based on historical exposure measurements from similar facilities, while a third used contemporaneous data, thereby avoiding the need to reconstruct historical exposures.

Viet et al28 published the results of a nested case-control study of 50 CBD cases, 74 beryllium sensitization (BeS) cases, and comparable numbers of matched controls from the Rocky Flats BHSP. Individual exposures were estimated by means of a job-exposure matrix (JEM) developed from a database of more than 250,000 daily FAH air samples obtained between 1960 and 1989 in one machine shop (“Building 444”). There were essentially no breathing zone data, and the FAH data were not adjusted to reflect personal exposures. Because air sampling data were not sufficient to characterize exposures in other buildings wherein some of the cases worked, Building 444 data were “generalized to the rest of the facility buildings” by assigning qualitative “relative exposure estimates” for the jobs and tasks in those buildings. Each worker's lifetime cumulative exposure estimate (CEE) and long-term time-weighted mean exposure estimate (MEE) were calculated on the basis of that worker's tenure, job and job factors, and work locations.

Compared with controls, CBD cases had significantly longer duration of employment (P < 0.01) and significantly greater CEE (P < 0.001) and MEE (P < 0.001). There was also a highly significant positive linear dose-response (P = 0.0006), with risk increasing 6.9-fold for every 10-fold increase in CEE. Similar results were found for MEE. Thus, this study demonstrated a “significant association between the probability of CBD with increasing average and cumulative exposure” as well as with exposure duration, thus confirming that CBD is positively dose-related. However, because worker exposures were based exclusively on FAH samples and were therefore only “relative” values, and because some exposures were extrapolated across worksites, the results do not serve to characterize response thresholds or establish OELs.

Van Dyke et al45 performed a second nested case-control study of the Rocky Flats BHSP participants, evaluating both beryllium exposure and genetics as risk factors for CBD and BeS. The study included 386 workers: 61 with CBD; 70 with BeS; and 255 matched controls. Individual worker exposures were reconstructed on the basis of worker interviews and historical measurements. For each job task, exposure levels were calculated as the “arithmetic mean of the available task and time period-specific exposure measurements” using exposure data obtained at Rocky Flats and also data from two “similar facilities.” Then, on the basis of each worker's job history, CEE and estimated lifetime weighted average exposures were estimated along with maximum task-based exposure levels. The study also determined workers’ HLADPβ1 genotype.

As noted above, the facility's historical exposure database was very large, but comprised almost exclusively of FAH general area samples. By contrast, the first of the “similar facilities” had an equally large database of personal lapel samples.46 The second “similar facility” had been the subject of two surveys, which described substantially different quantities and types of exposure data.32,33 Van Dyke et al45 did not explain how data obtained in such differing ways from three different facilities were combined and averaged to estimate exposures for “each of the unique beryllium exposure tasks” at Rocky Flats.

The study found that CBD cases had significantly higher estimated cumulative and average exposures than control subjects (P = 0.008). Risks of CBD were significantly increased in workers with estimated maximum task-based exposure >2 μg/m3 (P = 0.032), but not lesser exposures. Multiple logistic regression analyses confirmed a significant positive dose-response. CBD cases also tended to have more work time directly exposed to beryllium than did controls (P = 0.068). Thus, this study demonstrated significant associations between risk of CBD and both cumulative and mean exposure, and a borderline association with exposure duration.

When genetic factors were considered, the risk of a worker developing CBD more than doubled for every unit increase in lifetime-weighted average exposure (P = 0.010). Figure 1 demonstrates the predicted probability of CBD by estimated lifetime-weighted average exposure and genotype.

Predicted probability of chronic beryllium disease (CBD) by HLA-DPB1 E69 genotype and lifetime-weighted average exposure based on weighted logistic regression (reprinted with permission of the American Thoracic Society. Copyright © 2016 American Thoracic Society).45

These results clearly document a positive dose-response between cumulative and life-time weighted average beryllium exposure and CBD, and further document that dose-relatedness is significantly modified by genotype. (A just published NIOSH report indicates that more precise estimates of risk can be achieved by grouping workers according to allele-specific DNA sequences,47 an approach that would probably also yield significantly more accurate estimates of dose-response).

This study provides very strong evidence of beryllium's dose-relatedness, but the quantitative accuracy of the underlying data is uncertain. As noted above, the study analyses were not based on the workers’ actual exposures. Instead, they relied on the arithmetic means of data derived from three different facilities, using differing sampling methods, and aggregated in a manner that was not described. Because of such limitations, it is not clear how these data can be used to set specific exposure limits.

In a third study that relied exclusively on recent exposure data, Schuler et al48 evaluated 264 beryllium workers who started work after 1993 and were employed in 1999, thus avoiding the need to estimate long-term (ie, pre-1993) historical exposures. The study considered only a relatively short time period: work tenure ranged from 0.2 to 72.7 months and there was apparently no post-work follow-up. Six workers had CBD. Individual exposures were calculated using a JEM. Exposure levels for “most” jobs at the plant were determined in 1999 using 4022 personal closed-face lapel samples for total beryllium particulate and 198 personal impactor lapel samples for size-separated particulate.49 In addition, 76,349 area samples obtained during 1994 to 1999 were used to evaluate time trends in beryllium air levels. For each worker, a beryllium exposure profile was generated that included nine separate metrics: average exposure, cumulative exposure, and highest job exposure levels were each determined for total beryllium mass, respirable beryllium mass, and submicron beryllium mass.

Risks of CBD increased significantly with increasing cumulative total exposure (P < 0.05). Risks were marginally increased with increasing cumulative respirable exposure and highest job respirable exposure (0.05 ≤ P < 0.10). Using logistic regression, odds ratios (ORs) for CBD were significantly increased for cumulative total mass (OR = 1.66, 95% CI: 1.02 to 2.99) and cumulative respirable mass (OR = 1.68, 95% CI 1.02 to 3.28). The OR for cumulative submicron exposure was borderline significant (OR = 1.58, 95% CI: 1.00 to 2.83). For the other exposure metrics, ORs were increased, but not significantly.

This study provides very strong evidence of significant dose-relatedness between quantitative beryllium exposure and CBD. However, study limitations should be noted. Because individual exposures were assigned using a JEM (albeit one based on full-shift personal samples), individual exposures may be inaccurate. And because of the relatively short observation period, there was likely under ascertainment of cases. Despite such limitations, this study probably provides the best available data for setting an OEL.


The discussion above focused on CBD and did not consider the possible dose-relatedness of BeS. There are special challenges to characterizing BeS dose-response. Because individuals identified as BeS are asymptomatic, with normal lung function, normal exercise tolerance, and normal chest X-rays,50–52 there are no clinical signs to indicate when sensitization developed. In many studies, it has been possible to determine the elapsed time from first exposure to first positive beryllium lymphocyte proliferation test (BeLPT), but not the elapsed time to onset of BeS.

This is a particular concern because findings of studies that specifically considered workers with negative BeLPT at time of hire, and with no previous beryllium exposure, document onset of BeS after relatively short durations of exposure. Henneberger et al,35 a study of ceramics workers, found essentially no difference in the prevalence of BeLPT between long-term and short-term workers (10.4% vs 9.5%), but short-term workers with less than 1 year of exposure had five-fold greater BeS prevalence (16% vs 3%) than short-term workers with exposures greater than 1 year. Newman et al53 evaluated workers in a beryllium machining plant. Four of 60 newly hired workers without prior exposure tested BeLPT positive within 3 months of first exposure. Bailey et al54 reported nine beryllium production workers with negative BeLPT at the time of hire who subsequently developed BeS. Three of the nine had positive tests within 3 months, eight were positive within 6 months, and all were positive during their first year of work. Schuler et al,40 a study of copper-beryllium alloy workers, reported an “elevated prevalence of sensitization” (13%) among workers within 1 year of first exposure. Donovan et al55 reviewed more than 10,000 BeLPTs obtained over 12 years from almost 2400 workers. The peak prevalence of positive tests was observed within 4 to 8 months of employment, three-fold higher than in workers employed more than 1 year.

These findings suggest that BeS does not follow traditional dose-relatedness, at least with respect to cumulative and average exposure levels. Perhaps for such reasons, it has been more difficult to document significant dose-relatedness for BeS than CBD. For example, a 2010 report on CBD screening of DOE workers found that “Clear exposure-response relationships among DOE workers with BeS are not apparent from the existing literature.”56 OSHA,10 NRC,6 and NIOSH researchers48 have suggested that BeS is more associated with short-term peak exposures and highest-job exposures: “one might expect short-term ‘peak’ exposure to be more important than the same cumulative exposure delivered over a longer period.”6 However, peak exposures cannot be identified or measured by means of full-shift PBZ sampling, and real-time monitoring of workers’ PBZ exposures is not currently available. Some studies used highest job-worked33,48 or highest year exposed33 as proxies for peak exposures. However, there currently exist insufficient empirical data to evaluate relationships between BeS and peak exposures.


Nearly 70 years since Sterner and Eisenbud first struggled to understand its “puzzling” and “bizarre” epidemiology, it is now clear that the incidence of CBD is positively related to both cumulative and long-term average exposure, in conjunction with inherited susceptibility factors. Such susceptibility factors may involve antigen recognition and inflammatory response,57–64 but questions remain. Less is known about the specific dose-relatedness of BeS, which seems unrelated to cumulative and average exposure levels and may prove to be idiosyncratic.

Such understanding of dose-response has important implications for beryllium-related occupational health and safety programs. It is now apparent that reducing average and cumulative exposure will reduce long-term risks of CBD. It is also generally agreed that the current OSHA PEL is not adequately protective. In its proposed Beryllium Standard, OSHA recommended lowering the PEL from 2 to 0.2 μg/m3,10 but numerous commenters including ACOEM called for an even lower level. While waiting for the final version of the OSHA Standard, employers should aim to lower exposure levels at least to the proposed 0.2 μg/m3 PEL.

Whether risks of BeS are equally decreased by controlling airborne levels may depend on how well peak exposures are controlled and also on how well workers are protected from dermal exposure. Skin exposure can lead to sensitization, and the efficacy and limitations of personal skin protection have been documented in beryllium workers.65–67 The American Thoracic Society68 and NRC6 have both expressed concern that skin exposure leads to BeS, and OSHA includes dermal protection in its proposed Standard.10

The most recent quantitative evidence of dose-relatedness28,45,48 provides a compelling argument that sensitized workers should be discouraged from on-going beryllium exposure in order to limit their cumulative exposure and thereby limit their long-term risks of CBD. Although no empirical data directly support medical removal, it is widely recommended as “prudent”6,9,68 and a medical removal provision is included in OSHA's proposed Standard.10

There is still much to understand about beryllium exposure and CBD, but lack of knowledge about its dose-response characteristics is no longer an important impediment to more optimal workplace health and safety. While waiting for OSHA to issue its final Beryllium Standard, employers and workers should begin to implement more stringent controls on airborne exposures, greater use of skin protection, and medical removal for workers with BeS.


1. Eaton DL, Gilbert SG. Klaassen CD. Principles of toxicology. Casarett and Doull's Toxicology. The Basic Science of Poisons 8th ed.New York: McGraw-Hill Education, LLC; 2013. 13–48.
2. Hill AB. The environment and disease: association or causation? Proc R Soc Med 1965; 58:295–300.
3. Checkoway H, Pearce NE, Kriebel D. Research Methods in Occupational Epidemiology. 2nd ed.New York: Oxford University Press; 2004.
4. Smith TJ, Kriebel D. A Biologic Approach to Environmental Assessment and Epidemiology. New York: Oxford University Press; 2010.
5. Cullen MR, Cherniack MG, Kominsky JR. Chronic beryllium disease in the United States. Semin Respir Med 1986; 7:203–209.
6. National Academy of Sciences. Managing Health Effects of Beryllium Exposure. Washington, DC: National Academies Press; 2008.
7. Borak J. The beryllium occupational exposure limit: historical origin and current inadequacy. J Occup Environ Med 2006; 48:109–116.
8. Kreiss K, Day GA, Schuler CR. Beryllium: a modern industrial hazard. Ann Rev Pub Health 2007; 28:259–277.
9. Kreiss K. Beryllium: a paradigm for occupational lung disease and its prevention. Occup Environ Med 2011; 68:787–788.
10. OSHA. Occupational exposure to beryllium and beryllium compounds; proposed rule. Fed Reg 2015; 80:47566–47827.
11. Maier LA. Beryllium health effects in the era of the beryllium lymphocyte proliferation test. Appl Occup Environ Hyg 2001; 16:514–520.
12. Hardy HL, Tabershaw IR. Delayed chemical pneumonitis occurring in workers exposed to beryllium compounds. J Ind Hyg Toxicol 1946; 28:197–211.
13. DeNardi JM, Van Ordstrand HS, Carmody MG. Chronic pulmonary granulomatosis: report of ten cases. Am J Med 1949; 7:345–355.
14. Eisenbud M, Wanta RC, Dustan C, Steadman LT, Harris WB, Wolf BS. Non-occupational berylliosis. J Ind Hyg Toxicol 1949; 31:282–294.
15. Sterner JH, Eisenbud M. Epidemiology of beryllium intoxication. Arch Ind Hyg Occup Med 1951; 4:123–151.
16. Kriebel D, Brain JD, Sprince NL, Kazemi H. The pulmonary toxicity of beryllium. Am Rev Respir Dis 1988; 137:464–473.
17. Eisenbud M. The standard for control of chronic beryllium disease. Appl Occup Environ Hyg 1998; 13:25–31.
18. Eisenbud M. Origins of the standards for control of beryllium disease (1947–1949). Environ Res 1982; 27:79–88.
19. Cotes JE, Gilson JC, McKerrow CB, Oldham PD. A long-term follow-up of workers exposed to beryllium. Br J Ind Med 1983; 40:13–21.
20. Cullen MR, Kominsky JR, Rossman MD, et al. Chronic beryllium disease in a precious metal refinery. Clinical epidemiologic and immunologic evidence for continuing risk from exposure to low level beryllium fume. Am Rev Respir Dis 1987; 135:201–208.
21. Dinardi SR. The Occupational Environment: Its Evaluation and Control. Fairfax, VA: American Industrial Hygiene Association; 1997.
22. Donaldson HM, Stringer WT. Beryllium sampling methods. Am Ind Hyg Assoc J 1980; 41:85–90.
23. Kolanz ME, Madl AK, Kelsh MA, Kent MS, Kalmes RM, Paustenbach DJ. A comparison and critique of historical and current exposure assessment methods for beryllium: implications for evaluating risk of chronic beryllium disease. Appl Occup Environ Hyg 2001; 16:593–614.
24. Kreiss K, Mroz MM, Newman LS, Martyny J, Zhen B. Machining risk of beryllium disease and sensitization with median exposures below 2 mg/m3. Am J Ind Med 1996; 30:16–25.
25. Kriebel D, Sprince NL, Eisen EA, Greaves IA. Pulmonary function in beryllium workers: assessment of exposure. Br J Ind Med 1988; 45:83–92.
26. National Institute for Occupational Safety Health. Criteria for a Recommended Standard ... Occupational Exposure to Beryllium. Washington, DC: U.S. Department of Health, Education and Welfare; 1972.
27. Barnard AE, Torma-Krajewski J, Viet SM. Retrospective beryllium exposure assessment at the Rocky Flats Technology Site. Am Ind Hyg Assoc J 1996; 57:804–808.
28. Viet SM, Torma-Krajewski J, Rogers J. Chronic beryllium disease and beryllium sensitization at Rocky Flats: a case-control study. Am Ind Hyg Assoc J 2000; 61:244–254.
29. Kreiss K, Mroz MM, Zhen B, Martyny JW, Newman LS. Epidemiology of beryllium sensitization and disease in nuclear workers. Am Rev Respir Dis 1993; 148:985–991.
30. Kreiss K, Mroz MM, Zhen B, Wiederman H, Barna B. Risks of beryllium disease related to work processes at a metal, alloy and oxide production plant. Occup Environ Med 1997; 54:605–612.
31. Couch JR, Petersen M, Rice C, Schubauer-Berigan MK. Development of retrospective quantitative and qualitative job-exposure matrices for exposures at a beryllium processing facility. Occup Environ Med 2011; 68:361–365.
32. Kelleher PC, Martyny JW, Mroz MM, et al. Beryllium particulate exposure and disease relations in a beryllium machining plant. J Occup Environ Med 2001; 43:238–249.
33. Madl AK, Unice K, Brown JL, Kolanz ME, Kent MS. Exposure-response analysis for beryllium sensitization and chronic beryllium disease among workers in a beryllium metal machining plant. J Occup Environ Hyg 2007; 4:448–466.
34. Kreiss K, Wasserman S, Mroz MM, Newman LS. Beryllium disease screening in the ceramics industry: blood lymphocyte test performance and exposure-disease relations. JOM 1993; 35:267–274.
35. Henneberger PK, Cumro D, Deubner DD, Kent MS, McCawley M, Kreiss K. Beryllium sensitization and disease among long-term and short-term workers in a beryllium ceramics plant. Int Arch Occup Environ Health 2001; 74:167–176.
36. Schubauer-Berigan MK, Couch JR, Petersen MR, Carreon T, Jin Y, Deddens JA. Cohort mortality study of workers at seven beryllium processing plants: update and associations with cumulative and maximum exposure. Occup Environ Med 2011; 68:345–353.
37. Sanderson WT, Petersen MR, Ward EM. Estimating historical exposures of workers in a beryllium manufacturing plant. Am J Ind Med 2001; 39:145–157.
38. Rosenman K, Hertzberg V, Rice C, et al. Chronic beryllium disease and sensitization at a beryllium processing facility. EHP 2005; 113:1366–1372.
39. Stange AW, Hilmas DE, Furman FJ, Gatliffe TR. Beryllium sensitization and chronic beryllium disease at a former nuclear weapons facility. Appl Occup Environ Hyg 2001; 16:405–417.
40. Schuler CR, Kent MS, Deubner D, et al. Process-related risk of beryllium sensitization and disease in a copper-beryllium alloy facility. Am J Ind Med 2005; 47:195–205.
41. Martyny JW, Hoover MD, Mroz MM, et al. Aerosols generated during beryllium machining. J Occup Environ Med 2000; 42:8–18.
42. Kent MS, Robins TG, Madl AK. Is total mass or mass of alveolar-deposited airborne particles of beryllium a better predictor of the prevalence of disease? A preliminary study of a beryllium processing facility. Appl Occup Environ Hyg 2001; 16:539–558.
43. McCawley MA, Kent MS, Berakis MT. Ultrafine beryllium number concentration as a possible metric for chronic beryllium disease risk. Appl Occup Environ Hyg 2001; 16:631–638.
44. Stefaniak AB, Hoover MD, Day GA, et al. Characterization of physicochemical properties of beryllium aerosols associated with prevalence of chronic beryllium disease. J Environ Monit 2004; 6:523–532.
45. Van Dyke MV, Martyny JW, 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; 183:1680–1688.
46. Johnson JS, Foote K, McClean M, Cogbill G. Beryllium exposure control program at the Cardiff Atomic Weapons Establishment in the United Kingdom. Appl Occup Environ Hyg 2001; 16:619–630.
47. Kreiss K, Fechter-Leggett ED, McCanlies EC, Schuler CR, Weston A. Research to practice implications of high-risk genotypes for beryllium sensitization and disease. J Occup Environ Med 2016; 58:855–860.
48. Schuler CR, Virji MA, Deubner DC, et al. Sensitization and chronic beryllium disease at a primary manufacturing facility, part 3: exposure-response among short-term workers. Scand J Work Environ Health 2012; 38:270–281.
49. Virji MA, Park JY, Stefaniak AB, et al. Sensitization and chronic beryllium disease at a primary manufacturing facility, part 1: historical exposure reconstruction. Scand J Work Environ Health 2012; 38:247–258.
50. US. Department of Energy. Chronic Beryllium Disease Prevention Program; Final Rule. Fed Reg 1999; 64:68854–68914.
51. Fontenot AP, Maier LA. Genetic susceptibility and immune-mediated destruction in beryllium-induced disease. Trends Immunol 2005; 26:543–549.
52. Newman LS, Maier LA. Sullivan JB, Krieger GR. Beryllium. Clinical Environmental Health and Toxic Exposures 2nd ed.Philadelphia, PA: Lippincott Williams & Wilkins; 2001. 919–926.
53. Newman LS, Mroz MM, Maier LA, Daniloff EM, Balkissoon R. Efficacy of serial medical surveillance for chronic beryllium disease in a beryllium machining plant. J Occup Environ Med 2001; 43:231–237.
54. Bailey RL, Thomas CA, Deubner DC, Kent MS, Kreiss K, Schuler CR. Evaluation of a preventive program to reduce sensitization at a beryllium metal, oxide, and alloy production plant. J Occup Environ Med 2010; 52:505–512.
55. Donovan EP, Kolanz ME, Galbraith DA, Chapman PS, Paustenbach DJ. Performance of the beryllium blood lymphocyte proliferation test based on a long-term occupational surveillance program. Int Arch Occup Environ Health 2007; 81:165–178.
56. Arjomandi M, Seward J, Gotway MB, et al. Low prevalence of chronic beryllium disease among workers at a nuclear weapons research and development facility. J Occup Environ Med 2010; 52:647–652.
57. Mayer AS, Hamzeh N, Maier LA. Sarcoidosis and chronic beryllium disease: similarities and differences. Semin Respir Crit Care Med 2014; 35:316–329.
58. Snyder JA, Demchuk E, McCanlies EC, et al. Impact of negatively charged patches on the surface of MHC class II antigen-presenting proteins on risk of chronic beryllium disease. R Soc Interface 2008; 5:749–758.
59. Rosenman KD, Rossman M, Hertzberg V, et al. HLA class II DPB1 and DRB1 polymorphisms associated with genetic susceptibility to beryllium toxicity. Occup Environ Med 2011; 68:487–493.
60. Rossman MD, Stubbs J, Lee CW, Argyris E, Magira E, Monos D. Human leukocyte antigen Class II amino acid epitopes: susceptibility and progression markers for beryllium hypersensitivity. Am J Respir Crit Care Med 2002; 165:788–794.
61. Saltini C, Richeldi L, Losi M, et al. Major histocompatibility locus genetic markers of beryllium sensitization and disease. Eur Respir J 2001; 18:677–684.
62. McCanlies EC, Yucesoy B, Mnatsakanova A, et al. Association between IL-1A single nucleotide polymorphisms and chronic beryllium disease and beryllium sensitization. J Occup Environ Med 2010; 52:680–684.
63. Maier LA, Sawyer RT, Bauer RA, et al. High beryllium-stimulated TNF-a is associated with the -308 TNF-a promotor polymorphism and with clinical severity in chronic beryllium disease. Am J Respir Crit Care Med 2001; 164:1192–1199.
64. Tooker BC, Bowler RP, Orcutt JM, Maier LA, Christensen HM, Newman LS. SELDI-TOF derived serum biomarkers failed to differentiate between patients with beryllium sensitisation and patients with chronic beryllium disease. Occup Environ Med 2011; 68:759–764.
65. Cummings KJ, Deubner DC, Day GA, et al. Enhanced preventive programme at a beryllium oxide ceramics facility reduces beryllium sensitisation among new workers. Occup Environ Med 2007; 64:134–140.
66. Day GA, Dufresne A, Stefaniak AB, et al. Exposure pathway assessment at a copper-beryllium alloy facility. Ann Occup Hyg 2007; 51:67–80.
67. Armstrong JL, Day GA, Park JY, et al. Migration of beryllium via multiple exposure pathways among work processes in four different facilities. J Occup Environ Hyg 2014; 11:781–792.
68. Balmes JR, Abraham JL, Dweik RA, et al. An official American Thoracic Society Statement: diagnosis and management of beryllium sensitivity and chronic beryllium disease. Am J Respir Crit Care Med 2014; 190:e34–e59.
Copyright © 2016 by the American College of Occupational and Environmental Medicine