RISK OF DEVELOPING CHRONIC BERYLLIUM DISEASE
AND BERYLLIUM SENSITIZATION ASSOCIATED WITH
HLA-DPB1 AND DRB1 POLYMORPHISMS AND MAGNITUDE
OF BERYLLIUM EXPOSURE
By
Vitri Widyaningsih

A THESIS
Submitted to
Michigan State University
In partial fulfillment of the requirements
For the degree of
Epidemiology - Master of Science
2013

ABSTRACT
RISK OF DEVELOPING CHRONIC BERYLLIUM DISEASE AND BERYLLIUM
SENSITIZATION ASSOCIATED WITH HLA-DPB1 AND DRB1 POLYMORPHISMS
AND MAGNITUDE OF BERYLLIUM EXPOSURE
By
Vitri Widyaningsih
Background: Beryllium exposure is a necessary but not sufficient cause of CBD and BeS. The
presence of HLADPB1 Glutamine 69 and other polymorphisms were shown in previous studies
to influence disease development. Our goal was to examine genetics and exposure effect in the
development of BeS and CBD and progression from BeS to CBD
Methods: DNA-based typing was conducted for all subjects (n=361) consisting of 61 CBD, 41
BeS, and 259 exact matched controls. Exposures were assessed through job history and industrial
hygiene records.
Results: Glutamine 69 increased the risk of Chronic Beryllium Disease and Beryllium
Sensitization (OR respectively 25.7; 95% CI 6.1-108.5 and 6.4; 95% CI 2.4-17.1). Glutamine 71
had an important role in the development of BeS (OR 2.4; 95% CI 1.1-5.6) and was shown to be
protective for CBD among Beryllium sensitized individuals (OR 0.38, 95% CI 0.16-0.87). There
was no clear dose-response or interaction between genetics and exposure, but the matched
controls without susceptible genes, although having the highest exposure, remained healthy.
Conclusion: Glutamine 69 increased the odds of developing CBD and BeS. Glutamine 71
showed an important role in the development of BeS and possibly reducing the risk of
progression to CBD. Further work to explore other polymorphisms is needed to assess exposuregenetic interactions and dose-response associations.
ii

DEDICATION

To my husband Nugroho and our daughter Ardella, without you I’m nothing.

iii

ACKNOWLEDGEMENTS

I would like to acknowledge the following people for helping and supporting me to complete this
work. First, to my thesis committee chair, Dr. Ken Rosenman, I would like to express my
deepest gratitude for encouraging me to pursue this topic, allowing me access to the data, and
mentoring me throughout the process. To Dr. Dorothy Pathak, my academic adviser and thesis
committee, thank you for helping me throughout my graduate years as well as the completion of
this thesis. Your thoughtful advise, time, and support has directed me on the right track for the
completion of my degree. Next, to Dr. Joseph Gardiner, my thesis committee, thank you for the
valuable input and feedback you have given for this thesis and the knowledge you share. Mary Jo
Reilly, thank you for helping me in handling the data and giving great feedback and input
throughout the completion of this thesis. This thesis would not have been completed without all
your support and guidance.
I am also grateful to all faculty and staff in Department of Epidemiology and Biostatistics
Michigan State University, for your contribution on this thesis and the graduate program.
Finally, I would like to thank my family and friends for all the support and encouragement
throughout this journey. Thank you.

iv

TABLE OF CONTENTS

LIST OF TABLES

viii

LIST OF FIGURES

xi

CHAPTER I : BACKGROUND
I.
OVERVIEW
II.
AIMS
III.
HYPOTHESIS

1
1
2
2

CHAPTER II : LITERATURE REVIEW
I.
BERYLLIUM
II.
BERYLLIUM TOXICITY
a. Definitions
b. Epidemiology
c. Factors Affecting Development of Disease
d. Natural History
e. Diagnosis
f. Treatment
g. Prevention
III.
EXPOSURE DISEASE ASSOCIATION
IV.
GENE SUSCEPTIBILITY RELATED TO BERYLLIUM TOXICITY
V.
INTERACTION OF GENETIC SUSCEPTIBILITY AND EXPOSURE
ON BERYLLIUM TOXICITY

3
3
4
4
5
6
8
8
9
10
11
14

CHAPTER III : METHODOLOGY
I.
STUDY DESIGN
II.
SAMPLE CHARACTERISTICS
III.
VARIABLES AND MEASUREMENTS
a. Dependent Variables
1. Beryllium Sensitization
2. Chronic Beryllium Disease
b. Independent Variables
1. HLA-DPB1 and DRB1 polymorphisms
2. Exposure to Beryllium
IV.
DATA COLLECTION
V.
DATA ANALYSIS

20
20
20
23
23
23
23
23
23
24
24
25

CHAPTER IV : RESULTS
I.
DEMOGRAPHICS CHARACTERISTICS
II.
EXPOSURE CHARACTERISTICS

28
28
29

v

17

III.
IV.
V.
VI.
VII.
VIII.
IX.
X.

GENETIC CHARACTERISTICS FOR BOTH PLANTS AND
COMBINED
GENETICS AND EXPOSURE ASSOCIATION WITH CHRONIC
BERYLLIUM DISEASE AND BERYLLIUM SENSITIZATION
GENETIC AND EXPOSURE ASSOCIATION WITH CHRONIC
BERYLLIUM DISEASE
GENETIC AND EXPOSURE INTERACTION WITH BERYLLIUM
EXPOSURE
PROGRESSION OF CBD FROM BERYLLIUM SENSITIZATION
EXPOSURE CHARACTERISTIC BY GENETICS AND DISEASE
STATUS
EXPOSURE CHARACTERISTICS IN INDIVIDUALS WITH
GLUTAMINE 69
ANALYSIS OF THE EFFECT OF TYPE OF EXPOSURE IN THE
DEVELOPMENT OF CBD AND BES USING THE HOCEKY STICK
APPROACH FOR CODING EXPOSURE
a. Hockey Stick Analyses for CBD
b. Hockey Stick Analyses for BeS
c. Hockey Stick Analyses on Progression of BeS to CBD

36
38
38
39
41
42
45

46
47
48
50

CHAPTER V : DISCUSSION

52

CHAPTER VI : CONCLUSION

59

APPENDICES
Appendix 1. Distribution of Cumulative Exposure and Log Cumulative
Exposure
a. Distribution of Cumulative Exposure
b. Distribution of Log Cumulative Exposure
Appendix 2. Wilcoxon Two Sample Test for Difference of Exposure in Plant 1
Appendix 3. Wilcoxon Two Sample Test for Difference of Exposure in All
Subjects
Appendix 4. Proportion of Glutamine 69 and Glutamine 71 Positive
Individuals by Disease State and the Importance of Glutamine 71
in the Absence of Glutamine 69
a. Proportion of Glutamine 69 and Glutamine 71 Positive
Individuals by Disease State
b. The Importance of Glutamine 71 in the Absence of Glutamine
69
Appendix 5. Univariable Conditional Logistic Regression for Development of
CBD
Appendix 6. Univariable Conditional Logistic Regression for Development of
BeS
Appendix 7. Univariable Unconditional Logistic Regression for Progression of
BeS to CBD

62

vi

63
63
64
65
66

67
67
67
68
69
70

Appendix 8. Development of CBD and BeS by Exposures Quartile and
Genetics
a. Development of CBD and BeS by Exposure Quartiles and
Genetics
b. Multivariable Conditional Logistic Regression Using
Quartiles of Log Exposure
Appendix 9. Algorithm of Level of Exposure (Cumulative and Peak) by
Median Value
REFERENCES

71
71
71
72
73

vii

LIST OF TABLES

Table 1.

Studies on Prevalence of Beryllium Sensitization and Chronic Beryllium
Disease from 1990-2012

Table 2.

Current Permissible Exposure Limits and Recommendations of Beryllium
Levels in Different Countries

11

Table 3.

Studies on Genetic and Beryllium Toxicity

15

Table 4.

Studies on Genetic and Exposure Interaction in Development of Beryllium
Toxicity

18

Comparison of Demographic Characteristics among Subjects with Chronic
Beryllium Disease, Beryllium Sensitization, and Controls

28

Table 6.

Exposure Characteristics by Plant

29

Table 7.

Exposure Characteristics by Outcome within Each Plant

30

Table 8.

Comparison of Exposure between Plant 1 and Plant 2 for Chronic Beryllium
Disease, Beryllium Sensitization, and Control Individuals

32

Table 9.

Exposure and Type of Exposure by Outcome

34

Table 10.

Genetics Distribution by Plant

36

Table 11.

Comparison of Gene Distribution between CBD, BeS and Control

37

Table 12.

Factors Significantly Associated with CBD on Univariable Analysis

38

Table 13.

Multivariable Conditional Logistic Regression for the Development of CBD

39

Table 14.

Factors Significantly Associated with BeS on Univariable Analysis

40

Table 15.

Multivariable Conditional Logistic Regression for the Development of BeS

40

Table 16.

Factors that Significantly Differentiate CBD and BeS on Univariable Analysis 41

Table 17.

Unconditional Multivariable Logistic Regression Analysis on Progression
from BeS to CBD

Table 5.

viii

7

42

Table 18.

Table 19.

Table 20.

Table 21.

Table 22.

Table 23.

Comparison of Cumulative, Log Cumulative, Mean, and Peak Exposure
between CBD, BeS, and Control Groups based on HLA-DPB1Glu69 -0201
presence and allele type

43

Comparison of Magnitude and Type of Exposures between CBD, BeS, and
Control Groups based on Individuals with Glutamine 69

45

Conditional Logistic Regression for the Development of CBD by Type of
Exposure with Hockey Stick Analysis

47

Conditional Logistic Regression for the Development of CBD by Type of
Exposure with Hockey Stick Analysis in Glutamine 69 Positive Individuals

48

Conditional Logistic Regression for the Development of BeS by Type of
Exposure with Hockey Stick Analysis

48

Conditional Logistic Regression for the Development of BeS by Type of
Exposure with Hockey Stick Analysis among Glutamine 69 Positive
Individuals

49

Table 24.

Unconditional Logistic Regression Comparing CBD and BeS by Type of
Exposure with Hockey Stick Analysis

50

Table 25.

Unconditional Logistic Regression Comparing CBD and BeS by Type of
Exposure with Hockey Stick Analysis among Glutamine 69 Positive
Individuals

51

Table 26.

Wilcoxon Two Sample Test for Difference of Exposure in Plant 1

65

Table 27.

Wilcoxon Two Sample Test for Difference of Exposure in All Subjects

66

Table 28.

Proportion of Glutamine 69 and Glutamine 71 Positive Individuals by Disease
State
67

Table 29.

Effect of Glutamine 71 in the Absence of Glutamine 69

67

Table 30.

Univariable Conditional Logistic Regression for Development of CBD

68

Table 31.

Univariable Conditional Logistic Regression for Development of BeS

69

Table 32.

Univariable Unconditional Logistic Regression Comparing CBD and BeS

70

Table 33.

Descriptive Analysis of Log Total Exposure

71

ix

Table 34.

Table 35.

Multivariable Conditional Logistic Regression Using Quartiles of Log
Exposure

71

Multivariable Conditional Logistic Regression Using Algorithm of Exposure

72

x

LIST OF FIGURES

Figure 1. Conceptual Framework of Exposure and Genetic Interaction in the Development
of Sensitization and CBD

19

Figure 2. Study Population and Sample Selection

22

Figure 3. Statistical Analysis

27

Figure 4. Distribution of Cumulative Exposure

63

Figure 5. Distribution of Log Cumulative Exposure

64

xi

CHAPTER I
BACKGROUND

I.

OVERVIEW
Beryllium is a naturally occurring metal, which was not reported to cause toxicity
until after its extensive use in industry.(1,2) Beryllium has been commonly used in
aerospace, electronics, and munitions industries.(1) Beryllium toxicity was first reported in
the mid-1950s in the form of acute symptoms and a more chronic progressive form.(3) This
report was soon followed by the implementation of an occupational exposure limit for
beryllium that caused a decrease in the incidence of acute beryllium disease. The chronic
form, in terms of sensitization and chronic beryllium disease, however, is still an
occupational health problem.(2)
The two major types of chronic beryllium toxicity are: a subclinical form of beryllium
disease, beryllium sensitization (BeS), with in vitro proliferation of lymphocytes; and
Chronic Beryllium Disease (CBD), a clinical form characterized by shortness of breath,
cough and granulomas in the lung.(2,4)
Despite reports from multiple studies conducted on beryllium exposure and beryllium
disease, the pathogenesis as well as exposure-disease association is still unclear.(2,4,5)
Several recent studies show that low dose exposure to beryllium well below the OSHA
permissible exposure limit can cause beryllium disease.(6–10) Several other studies have
reported that solubility of beryllium and possible skin exposure influence the development
of the disease.(11,12) A host-disease interaction has also been evaluated in studies of
genetic susceptibility to beryllium sensitization and CBD .(2, 5, 13)
1

Although a lot of focus has been directed to beryllium exposure, how the exposure
interacts with genetic susceptibility in the development of beryllium disease, is still not
well defined.(14–16) This thesis will evaluate the effect of different polymorphisms of
HLA-DPB1 and HLA-DRB-1 and their interaction with detailed exposure level
measurements comprised of mean, cumulative and peak exposure, to assess the geneexposure relationship in the development of beryllium sensitization and beryllium disease.

II.

AIMS
1.

To examine genetics and exposure influence in the development of BeS and CBD

2.

To examine genetics and exposure influence in the progression of BeS to CBD

III. HYPOTHESIS
1.

The risk of developing CBD and BeS is based on both the occurrence of certain
HLA-DPB1 and DRB1 polymorphisms and the magnitude of beryllium exposure

2.

The risk of progressing from BeS to CBD is based on both the occurrence of certain
HLA-DPB1 and DRB1 polymorphisms and the magnitude of beryllium exposure

2

CHAPTER II
LITERATURE REVIEW

I.

BERYLLIUM
Beryllium is a metal commonly found in the environment.(1,2) It can be found in
coal, wood, water, food and stones.(1,2) The general population can be exposed to low
levels of beryllium through air, drinking water, and food.(1) Although people are naturally
exposed, only at a higher level of exposure, mostly though inhalation in industrial
processes, has beryllium been reported to cause disease.(1–3)
Beryllium is commonly used in the manufacturing of aerospace, automotive, energy,
defense, medical, and electronics due to its specific characteristics.(1,2) Beryllium is one of
the lowest density metals, but one of the most rigid, even more rigid than steel.(1)
Exposure in the beryllium processing or manufacturing industries is higher than beryllium
exposure in other industries such as aluminum or nuclear facilities.(1–3)
Since cases of acute beryllium toxicity and chronic lung disease due to beryllium
were first recognized in the 1950s, an occupational exposure limit to beryllium was
implemented by Federal OSHA. (1–3) The OSHA regulations have helped to decrease the
incidence of acute beryllium toxicity, although cases of chronic toxicity to beryllium are
still reported. (1–3,17)

3

II.

BERYLLIUM TOXICITY
a.

Definitions
Beginning in the mid-1950s, it has been known that beryllium can cause
different kinds of diseases; acute beryllium toxicity, beryllium sensitization which
could be assessed after the development of a blood screening test for beryllium in
1989, and chronic beryllium disease.(3,18)
In 1949, the Atomic Energy Commission (AEC) set an occupational
permissible exposure limit (PEL) with a daily 8-hour time-weighted average (TWA)
3

of 2.0 µg/m but it was not until 1971 that OSHA adopted this standard for
industries.(19) After the implementation of the standard, the incidence of acute
beryllium disease was controlled and became very rare.(1,8) Beryllium sensitization
and chronic beryllium disease, however are still prevalent amongst workers who are
exposed, even when exposure is below the permissible limit.(2,5)
Beryllium sensitization is defined as individuals who have positive beryllium
lymphocyte proliferation test results (LPTs) without any positive result on the
following work up for CBD (chest radiograph, lung biopsy).(18) Chronic Beryllium
Disease (CBD) is defined as individuals who had positive beryllium lymphocyte
proliferation test results (LPTs) with non-caseating granuloma on lung biopsy or a
positive bronchial lavage.(4)

4

b.

Epidemiology
Studies conducted in several facilities with beryllium exposure have found a
prevalence of 0.3-16.6% for beryllium sensitization (BeS) and 0-7.6% for Chronic
Beryllium Disease (CBD).(6,7,20–33) The wide range of prevalence is due to
different levels of beryllium exposure, and different ascertainment of cases (previous
worker, current worker, or both), as well as the sensitivity and specificity of
Beryllium Lymphocyte Proliferation Testing (BeLPT) which probably differs across
laboratories.(2,5)
Higher levels of beryllium exposure are often found in beryllium
manufacturing industries, but low level exposure in aerospace industries, nuclear
facilities, weapons or munitions industries, aluminum industries, beryllium
distribution, and mining has also been reported to cause sensitization, and even
chronic disease (Table 1). (6, 7, 17–30) Certain work related processes such as
machining were reported to cause an increase in the prevalence of beryllium toxicity.
(6,34)
In 2004, it was estimated that 134,000 workers in the United States were
exposed to beryllium.(13) Cullen et al (1986) and Henneberger (2004) estimate that
until the 1980s, up to 800,000 workers in government or industries across the United
States were occupationally exposed to beryllium.(13,17)
Schubauer-Berigan and colleagues, in their mortality studies of beryllium
workers found that beryllium exposure was related to lung cancer, Chronic
Obstructive Pulmonary Disease (COPD), and nervous system and urinary tract
cancers independent of cigarette smoking and exposure to other lung carcinogens.(35)
5

The International Agency for Research on Cancer (IARC) has listed beryllium as a
carcinogen, (36) although recent findings for the association between beryllium
exposure and lung cancer were not conclusive.(37) Schubauer-Berigan and
colleagues found a positive association between beryllium exposure and lung cancer,
but Boffetta et al in their review concluded that the causal criteria for an association
was not well established.(37,38) The mortality rate for beryllium toxicity as reported
by Newman et al in 1996, ranged from 5.8 to 38%. The difference in study design,
follow up duration and also type of exposure contributes to this wide range.(4)

c.

Factors Affecting Development of Disease
Exposure to beryllium is a necessary cause for development of BeS and
CBD.(2,5) However, several studies have evaluated potential factors that increase the
risk for beryllium toxicity such as age, gender, race, ethnicity, smoking, respiratory
symptoms, spirometric or radiographic abnormalities, but only a few found positive
associations.(6,20,26,27) Age was reported as a risk factor for development of BeS
and CBD after controlling for duration of exposure.(6) No significant associations
between smoking and the disease were reported from these studies.(20,26,27) Kreiss
et al also showed that allergic history was not a risk factor for sensitization (20)
which was further emphasized by Schuler (2005), who reported that self-reported skin
problems associated with exposure to pickling fluids, coolants, or other work, were
not related to either beryllium sensitization or CBD. (27)

6

Table 1. Studies on Prevalence of Beryllium Sensitization and Chronic Beryllium Disease from 1990-2012
BeS
CBD
Sampl
Studies
Population
Detection of CBD
e
N
%
N
%
Kreiss et al, 1993(20)
Kreiss et al, 1993(21)
Kreiss et al, 1996(6)
Stange et al 1996(7)
Kreiss et al, 1997(22)
Henneberger, 2001(23)
Deubner et al, 2001(24)
Newman et al, 2001(25)
Sackett et al, 2004(26)
Schuler et al, 2005(27)
Rosenman et al, 2005(28)
Stanton et al, 2006(29)
Taiwo et all, 2008(30)
Arjomandi, 2010(31)
Taiwo et al, 2010(32)
Mikulski et al, 2011(33)

Ceramics Company, Colorado
Rocky Flats Nuclear Plant
Beryllia Ceramics Plant, Arizona
Rocky Flats Nuclear Plant
Beryllia Ceramics Plant, Ohio
Beryllia Ceramics Plant, Arizona
Mining Extraction
Machining
Nuclear Weapon
Copper Be Alloy Finishing
Beryllium Plant, Pennsylvania
Beryllium Alloy Distribution
Aluminum Smelter
Nuclear Weapons
Aluminum Smelter
Munitions Plant

505
890
136
4,397
627
151
75
235
2,221
153
577
88
734
1,875
1,932
524

7

9
17
8
107
59
15
3
22
19
10
96
1
2
59
9
8

1.78
1.9
5.9
2.43
9.4
9.9
4.0
9.4
0.9
6.5
16.6
1.1
0.3
3.1
0.47
1.5

9
13
6
29
24
8
1
13
2
6
44
1
2
5
2
0

1.8
1.8
4.4
0.7
3.8
5.3
1.3
5.3
0.09
3.9
7.6
1.1
0.3
0.3
0.1
0

Lung biopsy
Lung biopsy
Lung biopsy
Biopsy + X-ray
Bronchoscopy
Biopsy
Biopsy
Bronchoscopy
Biopsy
Bronchoscopy
Biopsy and X-ray
Biopsy
Not specified
Bronchoscopy
Not specified
Clinical

d.

Natural History
Beryllium sensitization has been reported in workers who were exposed to
Beryllium even for just a few months.(23,25,27,39,40) Of all sensitized individuals,
around 11-31% will develop CBD over the following 4-7 years.(41,42) It is still not
clearly defined why some individuals become sensitized, and what factors play a role
in the progression to CBD.(4,5,43,44) Several factors have been proposed, such as
duration of exposure in which individuals who had longer exposure would be more
likely to develop CBD, but this was still inconclusive since there were other studies
that reported different findings, and it might also be due to the longer latency period
of CBD or a host-related factor.(20,21,23,28)
Several studies recommended a follow up of sensitized individuals to assess
the development of CBD ranging from 2-4 years, including pulmonary function
testing and X-rays to look for clinical symptoms of CBD. (41,42,45) Among the
positive for Beryllium Lymphocyte Proliferation Testing (BeLPT) patients without
CBD symptoms, half become BeLPT negative on their follow up testing and some
workers did not develop CBD even after being followed for 12 years. (41,42) These
results set the foundation for periodic screening for beryllium-exposed individuals to
detect beryllium sensitization and CBD.(41,42,45)

e. Diagnosis
The transformation of lymphocytes due to beryllium exposure in sensitized
individuals enabled screening for the disease in asymptomatic individuals.(4,18)
Beryllium sensitization (BeS) can be measured with beryllium-specific lymphocyte
8

proliferation testing (BeLPT) using white blood cells or broncho-alveolar lavage
cells.(2,18) This test is commonly used to conduct screening and surveillance for
beryllium-exposed workers.(20,26,45) BeLPT can also characterize workplace risk
and evaluate the effectiveness of preventive interventions.(2)
Further examination to identify Chronic Beryllium Disease (CBD) is
conducted for those individuals with positive BeLPT results.(20,46) The examination
includes bronchoscopy with broncho-alveolar lavage and trans-bronchial biopsy, and
a chest radiograph.(44,46) Workers with positive BeLPTs but negative
granulomatous lung disease on further examination are at risk for developing CBD in
the future.(2)
The gold standard for the diagnosis of CBD includes histologic evidence of
granuloma from a lung biopsy and proliferative response of broncho-alveolar cells to
beryllium.(46) CBD can also be diagnosed by radiographic findings consistent with
granulomas and positive blood proliferative response to beryllium and localization of
beryllium inside the granuloma.(46)

f. Treatment
Early detection followed by prompt treatment of CBD can lead to regression
and prevent further progression, hence reducing the morbidity and mortality of the
disease. (46,47) CBD cannot be cured but is treatable; the goal of the treatment is to
reduce morbidity and mortality.(46) Cessation of beryllium exposure and
administration of systemic corticosteroids is the current standard management of
CBD. (46,47) Regression can be obtained by early corticosteroid intervention.(46,47)
9

Avoidance of further exposure to beryllium for sensitized individuals is important to
prevent progression of disease.(45,46)
Patients with BeS are followed up regularly, to detect any signs of early lung
damage.(42,46,47) The examination includes a history and physical examination, a
chest radiograph, and pulmonary function tests.(46,47) Patients with early lung
damage are given 40 mg of prednisone on alternate days for 6 months.(46,47) The
dosage is then tapered by no more than 10 mg every other month unless there is
evidence of renewed disease activity which is evaluated by the same examination as
used for disease progression.(46,47) The lowest dose of prednisone that prevents
disease activity is then maintained.(46,47) It is uncertain whether this treatment has to
be continued for the rest of the individual’s life.(46,47) However, once pulmonary
fibrosis develops, it is not reversible even with corticosteroid treatment.(46,47)
Patients who undergo treatment should be monitored using pulmonary function
testing and high resolution chest computed tomography.(46,47)

g. Prevention
There is no international exposure standard and different countries are
implementing different exposure limits for beryllium (Table 2).(48) Even within the
United States, there are differences in the Permissible Exposure Limit (PEL) for
3

beryllium.(48) The current OSHA PEL in the United States is 2.0 µg/m .(49) This
level prevents acute beryllium disease but recent studies show that it is not protective
enough to prevent BeS and CBD.(7,9,10) Therefore, recommendations have been
10

made for OSHA to lower the current permissible limit for beryllium exposure to 0.2
3

µg/m .(50,51) A lower PEL for beryllium per 8-hour shift was implemented by the
3

Department of Energy (DOE) in 1999 (0.2 µg/m ), (50) by the State of California in
3

2004 (0.2 µg/m ), (52) and also by the National Institute for Occupational Safety and
3

Health (NIOSH) which recommends a limit of 0.5 µg/m for 8 hours of occupational
exposure.(49) Secondary prevention through periodic medical screening is also
recommended for beryllium exposed workers. (6,27)
Table 2. Current Permissible Exposure Limits and Recommendations of Beryllium
Levels in Different Countries (48)
Limit Value
Limit Value
(8 hours)
(Short Term)
Country
3

in µg/m
2
2
2
2
0.15
1
2
2
1
2
0.2
2
2
0.2
2
2
0.5
2
2

Australia**
Austria**
Belgium**
Canada-Ontario
Canada-Quebec**
Denmark
France**
Japan*
Latvia
New Zealand
Poland**
Singapore
South Korea
Spain**
Sweden
Switzerland
USA – NIOSH*
USA – OSHA**
United Kingdom**
*) Recommendation limit
**) Standard limit, with legal implication
11

3

in µg/m
8
10
10
2

10

0.5
5

III. EXPOSURE-DISEASE ASSOCIATION
Exposure to beryllium is a necessary cause for the development of beryllium toxicity,
either in the form of acute disease, sensitization or chronic disease.(3) Since the first few
cases were reported back in the mid-1950s, several measures have been implemented
including the adoption of an occupational exposure limit for beryllium.(34) Several studies,
however, reported that higher exposure to beryllium does not always cause disease, and
that even low exposure to beryllium can cause disease.(7,9,10,29,31) These studies find
that even with exposure lower than the PEL standard by OSHA (7,9,10,27) or minimum
opportunistic contact with beryllium can cause sensitization and even the development of
3

CBD.(7,20,53) Schuler et al (2005) found that exposures higher than 0.2 µg/m were
associated with sensitization and CBD and rarely found cases in areas which maintained air
3

exposure lower than this level.(27) This level of 0.2 µg/m has been adopted as
California’s 8 hour occupational exposure limit since 2004.(54)
Another mechanism that has been proposed to explain this lack of a clear doseresponse relationship is the different solubility of different beryllium compounds and also
the different form and particle size which may influence the entry pathway as well as
pathogenesis of the disease.(11,24,32) A lower rate of beryllium disease was found in the
aluminum smelting environment which might be due to the more soluble form of beryllium
although there was also the consistent use of respiratory protection in the population
studied.(32) Skin contact, which was proposed as an entry for beryllium exposure, has also
been shown to contribute to the development of beryllium disease.(12,55)

12

Studies have also found that beryllium disease can develop years after cessation of
exposure (10,53) Additionally, a longitudinal study that assessed the prevalence of disease
after implementation of several measures to control high beryllium exposure found no
decrease in sensitization or CBD.(13)
The different measurement methods across different studies, in terms of mean,
average, and cumulative exposure, may also influence the variety of results.(51,54)
Henneberger studied peak, cumulative and mean exposure of beryllium and found that the
prevalence of beryllium disease is greater in the long term worker compared to the short
term worker (9.1% vs. 1.4%, p 0.06).(23) However, they were unable to show a distinct
difference in association between disease and peak, cumulative, and mean exposure level
and beryllium toxicity which might be due to limited statistical power.(23)
There is no clear dose-response relationship between beryllium exposure and
beryllium toxicity.(24,41) Several factors that have been proposed to explain this are:
different measures of exposure, solubility of beryllium exposure, and multiple pathway
entries which include skin contact.(12,13,28) The fact that not everybody who had the
same beryllium exposure develops the disease, infers a possible host-agent-environment
dynamic, and studies have been conducted to determine if gene susceptibility of exposed
individuals is related to the pathophysiology of beryllium toxicity.(5,56)

13

IV. GENE SUSCEPTIBILITY RELATED TO BERYLLIUM TOXICITY
The development of CBD and BeS is based on a type IV hypersensitivity mechanism
which involves activation of T cells and MHC Class II antigens.(5) Richeldi et al were the
first to assess several genes related to this mechanism.(56) In their preliminary study, they
found that HLA-DP and not HLA-DR and HLA-DQ genes were related to CBD.(56) In
their 1993 study they found that 97% of CBD patients had residue Glutamate in position 69
of the HLA-DPB1 gene compared to 30% in unaffected subjects, and proposed the use of
HLA-DPB1-Glu69 as a marker for CBD risk.(56,57) Subsequent studies also found that
this allele not only influenced development of CBD but also beryllium sensitization.(58,59)
Some authors reported that the homozygosity of this allele increased the risk of CBD (58–
60), Beryllium sensitization (59,60), as well as CBD severity (61). The odds ratio (OR) for
the association of HLA-DPB1Glu69 with CBD regardless of zygosity ranged from 3.719.14 (60,62,63) and 3.3-6.9 for sensitization.(60,63)
Further studies also showed that the presence of non-0201 alleles of HLA-DPB1
Glu69 was an important marker for beryllium toxicity.(58,59,61,63) Several other alleles
have also been linked to beryllium sensitization (HLA-DRArg74) (62), while DQ-B1-G86
(64), DRB1-S11 (64), DRB1-S13 (61), DQB1-06 (61), were associated with CBD, and
DRB-Glu71 (63) and TNF-α-308 (62) were associated with both BeS and CBD. Rosenman
et al (2011) proposed that the negative charge contributed by specific polymorphisms in
conjunction with DPβ-E69 was associated with CBD and BeS and that this polymorphism
was related to how peptides were presented to T cells involved in the pathophysiology of
CBD and BeS.(63)

14

Author,
Year
Richeldi
et al,
1993
(56)

Study
Design
Case
Control

Wang et
al, 1999
(58)

Case
Control

Wang et
al, 2001
(59)

Case
Control
on BeS

Saltini et Case
al, 2001 Control
(62)
on BeS
and
CBD

Rossma
n et al,
2002(64
)

Case
Control
on BeS
and
CBD

Table 3. Studies on Genetics and Beryllium Toxicity
N
N
Result
Cases Control
33
44
HLA-DPB1-0201 increase the risk of CBD (P <
CBD
0.05)
97% of CBD vs. 30% of controls expressed the
HLA-DPBglu69 (P < 0.001).
Conclusion:
HLA-DP has a role in development of CBD
Residue 69 can be used as a potential marker of
CBD.
20
75
Homozygous DPB1Glu69 in both alleles were found
CBD
more in CBD group (6/20) vs. control group (1/75).
Most Glu69 carriers from the control group had a
DPB1 allele-0201 (68%), while CBD group had a
non-0201 DPB1 Glu69-carrying allele (84%).
Conclusion:
Specific Glu69-containing alleles and their
homozygosity increase the risk of CBD.
25
163
88% of BeS has HLA-DPB1-Glu69, and 24% were
BeS
homozygous.
Conclusion:
HLA-DP related to BeS
23
93
HLA-DPGlu69 associated with CBD (OR 3.7,
BeS
p=0.016, 95% CI 1.4–10.0).
22
High TNF-a-308-2 marker associated with both BeS
CBD
and CBD (OR 7.8, p < 0.0001, 95% CI 3.2–19.1), no
difference between CBD and BeS.
HLA-DRArg74 associated with BeS (OR 3.96,
p=0.005, 95% CI 1.5–10.1).
30
82
HLA-DPB1-E69 was the most important marker for
BeS
sensitization, and did not differentiate BeS and CBD.
25
A significant association with CBD was observed
CBD
with HLA-DQB1-G86 (p=0.04), and HLA-DRB1S11 compared with BeS (p=0.03).
Conclusion:
HLA-DPB1-E69 is a marker for susceptibility to
beryllium sensitization.
HLA amino acid epitopes on HLA-DRB1 and DQB1, in association with or independently of HLADPB1-E69 may be associated with progression to
CBD. Did not find an association with homozygosity.
15

Author,
Year
Maier et
al, 2003
(61)

McCanli
es et al,
2004
(60)

Rosenm
an et al,
2011
(63)

Silveira,
2012
(43)

Table 3 (cont’d). Studies on Genetics and Beryllium Toxicity
Study
N
N
Result
Design Cases Control
Case
50
125
DPB1 Glu69 gene is associated with CBD and BeS
Control BeS
(OR 10.1 for CBD and 9.5 for BeS).
on BeS
104
The majority of BeS and CBD subjects displayed
and
CBD
non-0201 Glu69 alleles.
CBD
Glu69 homozygosity was highest in CBD, and lowest
in control.
DRB1-13 and DQB1-06 were associated with CBD
in the absence of Glu69.
Markers of disease severity were associated with
Glu69 homozygosity.
Conclusion:
DPB1 Glu69 is a marker of sensitization and not
specific for disease.
Case
64
730
HLA-DPB1Glu69 was associated with both CBD
Control BeS
(OR 9.4; 95% CI 5.4, 16.6) and sensitization (OR 3.3,
on BeS
90
95% CI 1.9, 5.9).
and
CBD
CBD and BeS were more likely to be homozygous
CBD
compared to controls (P<0.001).
Conclusion:
Evaluation of HLA-DPB1 haplotypes, gene–
environment and gene–gene interactions will be
important for fully understanding the immunogenic
nature of BeS and CBD.
Matche
44
288
92.3% CBD have HLA-DPbE69 residue (OR 19.14
d case
BeS
(95% CI 7.10 to 55.92) p 1.8310-16) and 79.5% BeS
control
65
(OR 6.20 (95% CI 2.73 to 14.47) p 3.82310-7) and
on BeS CBD
38.5% of control.
and
Conclusion:
CBD
Protective effect of the DPB1-0201 positive
haplotype may involve particular polymorphisms
outside of the DPB1 gene.
Case
502
653
CBD cases were more likely than controls to carry a
Control Bes/C
non-02 E69 allele than 02 Glutamine 60, with odds
BD
ratios ranging from 3.1 (2.1–4.5) to 3.9 (2.6–5.9) (p <
0.0001).
Conclusion:
The less frequent non 02 alleles increase the risk for
CBD more than the 0201 alleles.

16

V.

INTERACTION

OF

GENETIC

SUSCEPTIBILITY

AND

EXPOSURE

ON

BERYLLIUM TOXICITY
There have been only a few studies that focused on the association of beryllium
exposure, genetic susceptibility, and development of beryllium sensitization or CBD.(14–
16) The first study was conducted in 1997 by Richeldi et al who reported that 32.3% of a
worker population had HLA-DPB1Glu69, and found that the presence of this genetic
biomarker was associated with an increased risk of CBD in highly exposed individuals.(14)
Van Dyke et al in their case control studies on beryllium exposed workers found that
both exposure and genetic susceptibility had an independent effect on the development of
the disease, and found even higher odds ratios in subjects with both genetic susceptibility
and exposure.(15,16) They also found that people with homozygosity were at a greater risk
of developing the disease.(15,16) These studies used the lifetime average level of exposure
of individuals and specifically assessed the HLA-DPB1 Glu69 allele.(15,16) The
interaction between different types of exposure measurements, different genetic
susceptibilities to beryllium with CBD and BeS, and especially how only a percentage of
sensitized individuals go on to develop CBD, however, is still uncertain.(15,16)
Therefore, our study will assess the association of different genetic polymorphisms
associated with CBD and BeS and the interaction of peak, average and cumulative
beryllium exposure of the individual to better understand the association.

17

Table 4. Studies on Genetic and Exposure Interaction in Development of Beryllium Toxicity
Author Study
N
N
Result
, Year Design Cases Control
Richeld Case
6 CBD
119
HLA-DPB1Glu69 present in 30% controls, and
i et al,
Control 2 BeS
83% in CBD (P 0.01), and in none in BeS.
1997
on
The presence of the marker was associated with
(14)
CBD
higher prevalence of CBD (HLA-DPB1Glu69positive machinists 25%; HLA-DPB1Glu69negative machinists 3.2%, P 0.05).
Conclusion: Genetic susceptibility factor adds to
the effect of process-related risk factors.
Van
Case
35
127
Increased odds for BeS and CBD among DPbE69
Dyke et Control BeS
carrier (OR 6.06, 95% CI 1.96 to 18.7).
3
al,
19
Exposure of 0.1 mg/m (lifetime weighted
2011(1
CBD
average) increased the odds of CBD (OR 3.98,
5)
95% CI 1.43 to 11.0).
Those with both risk factors had higher increased
odds (OR 24.1, 95% CI 4.77 to 122).
Conclusion: DPbE69 carriage and high exposure
to beryllium appear to contribute individually to
the development of BeS and CBD.
Van
Case
70
255
HLA-DPB1-E69 carriage increased odds for CBD
Dyke et Control
BeS
(OR, 7.61; 95% CI, 3.66–15.84).
al, 2011
61
Each unit increase in lifetime weighted average
(16)
CBD
exposure increased the odds for CBD (OR, 2.27;
95% CI, 1.26–4.09).
Compared with E69-negative genotypes, a
heterozygote E69-positive 02 allele increased the
odds for BeS (OR, 12.01; 95% CI, 4.28–33.71)
and CBD (OR, 3.46; 95% CI, 1.42–8.43).
A single non-02 E69 allele further increased the
odds for BeS (OR, 29.54; 95% CI, 10.33–84.53)
and CBD (OR, 11.97; 95% CI, 5.12–28.00).
Conclusion: E69 and beryllium exposure both
contribute to the odds of CBD.
Non-02 E69 carriers and E69 homozygote at
higher odds than those with 02 genotypes.

18

Figure 1: Conceptual Framework of Exposure and Genetic Interaction in the Development of
Sensitization and CBD
Genetic Susceptibility

Magnitude

Beryllium
Type IV Delayed
Hypersensitivity

Proliferation of Be
Specific T Lymphocytes

BeS

Local proliferation and
accumulation in the lung

CBD

19

Genetic
Susceptibility

CHAPTER III
METHODOLOGY

I.

STUDY DESIGN
This thesis analyzed the data from a case control study conducted by Rosenman et al
during 1996-2010.(28,63)

II.

SAMPLE CHARACTERISTICS
a.

Population
The study population consisted of workers in two beryllium processing facilities
in eastern Pennsylvania.

b.

Sample selection
Using personnel records, 5490 workers who were on the payroll for 2 or more
days at either of two beryllium facilities were identified. There were 1349 individuals
working between 1958 through 1978 in Plant 1, and 4141 individuals working from
1935 to 2000 in Plant 2.
From various databases, it was determined that as of 12/31/1988, 328 (24.3%)
individuals from Plant 1 and 2293 (55.4%) from Plant 2 had died.
Mailing and follow up phone calls were initiated in 1996 to offer free medical
screening for beryllium-related disease to members of the cohorts not known to have
died as of 12/31/88.

20

One hundred forty eight (11%) workers from Plant 1 and 177 (4.3%) workers
from Plant 2 could not be located. Among the 873 workers who were located from
Plant 1, 160 said that they worked for the company but not in the beryllium
production Plant, 65 declined to participate, 86 completed the questionnaire only, and
562 individuals participated in the medical screening.
From Plant 2, 1671 workers were contacted for the medical screening, 35 said
that they did not work in the beryllium production Plant, 191 declined to participate,
474 completed the questionnaire only, and 971 participated in the medical screening.
Therefore, for both plants a total of 1533 individuals participated in the medical
screening, 256 declined and 560 completed a questionnaire only.
The medical screening occurred from 1996-2001. From the medical screening,
80 people met the case definition for CBD and 55 for BeS. Fifteen of these workers
with CBD and 11 with BeS either did not provide consent or blood for genetic testing,
hence genetic data was available from 65 workers who were diagnosed with CBD and
44 workers who were identified as BeS. A total of 288 individuals who underwent
genetic testing and had completely normal medical testing were chosen as controls.
Complete data on medical testing, genetics, and exposures, were available for 61
CBD, 41 BeS, and 259 controls that were included in the final analysis.

21

Figure 2: Study Population and Sample Selection
PLANT 1

328 died
148 couldn’t be
located

160 didn’t work
at production
65 declined

562 medical
screening

PLANT 2

1349 worker

4141 worker

873 worker

1671 worker

86 questionnaire
only

474 questionnaire
only

2293 died
177 couldn’t be
located

35 didn’t work
at production
191 declined

971 medical
screening

1563 medical screening and
genetics from plant 1 and 2

80 CBD

55 BeS

1299 normal

61 CBD

41 BeS

259 control

22

99 abnormal not
BeS or CBD

III. VARIABLES AND MEASUREMENTS
a.

Dependent Variables
1. Beryllium Sensitization
Beryllium sensitization was defined as individuals who had two positive
beryllium lymphocyte proliferation tests (BeLPTs) without a positive result on the
following work up for CBD (chest radiograph and lung biopsy, if performed).

2. Chronic Beryllium Disease
Chronic Beryllium Disease (CBD) was defined as individuals who had two
positive beryllium lymphocyte proliferation tests (BeLPTs) or positive BeLPT
from a bronchial lavage sample. Probable CBD was characterized as individuals
who had two positive blood BeLPTs and positive lavage BeLPT or a positive
radiograph.

b.

Independent Variables
a. HLA-DPB1 and DRB1 polymorphisms
HLA-DPB1 and HLA-DRB1 polymorphisms were tested by using
Polymerase Chain Reaction (PCR) on Oiagen columns from a venous whole
blood sample that had been frozen the day after the original blood collection.

23

b. Exposure to Beryllium
Beryllium exposure was defined as occupational exposure to beryllium that
was calculated with regards to duration, level of total exposure (cumulative, mean
and peak), and type of exposure.
Exposure data was obtained by calculating the Daily Weighted Average
(DWA), Job Exposure Matrix and Task Exposure Matrix and then calculating the
mean exposure, peak exposure, and cumulative exposure for each worker.
Exposure was also classified to chemical form which consisted of soluble, nonsoluble and mixed as well as physical form which consisted of dust, fume, and
mixed.(28)

IV. DATA COLLECTION
Individuals who had two positive beryllium lymphocyte proliferation tests (BeLPTs)
and/or a chest radiograph reading ≥ 1/0 in parenchymal profusion determined by at least
two of the three physicians certified to interpret chest radiographs for pneumoconiosis were
referred for bronchoscopy, the testing of lavage fluid for beryllium lymphocyte
proliferation and a trans-bronchial biopsy. Individuals with definite or probable CBD and
BeS were classified as cases.
The control group was matched by Plant, gender, and year of birth within 5 years.
Two to three controls were chosen for each case. These controls had completely normal
results on chest radiograph and BeLPT testing. There were initially an additional 35
individuals with suspected CBD or BeS who had had matched controls selected. After
subsequent review it was determined that these 35 individuals did not have CBD or BeS
24

and no genetic analysis was performed on their blood but the now 70 extra controls did
have genetic analyses and were appropriately reassigned to other cases as controls
Data was collected through questionnaire, medical screening, and genetic testing. The
questionnaire included data on demographics, a detailed work history to assess beryllium
exposure, and also history of previous illness. Medical screening was conducted from
1996-2001 for all workers who were eligible and signed informed consent to allow genetic
testing was obtained. Genetic analyses were conducted 4-10 years after collection using
Oiagen columns from a venous whole blood sample that was frozen the day after the
original blood collection. Beryllium exposure was calculated for each participant by
reviewing past sampling data and work processes combined with duration of exposure
obtained from each plant’s employee work history records. Exposure metrics used were
cumulative, peak and average exposure to beryllium. Exposure data also included the type
of beryllium, solubility and physical form.

V.

DATA ANALYSIS
Data analysis was conducted using SAS 9.3 to assess associations between genetics
and exposure with the development of CBD and BeS and progression of BeS to CBD.
Analysis included descriptive statistics of demographic, exposure, and genetic distribution,
as well as multivariable conditional logistic regression to assess the association of genetics
and exposure with CBD and BeS. The steps for data analysis included univariable analysis
on beryllium exposure (mean, peak, and cumulative exposure as continuous variables) and
genetic factors (categorical) as independent variables with the development of CBD or
BeS. Variables that were associated with the development of CBD and BeS on univariable
25

analysis at p ≤ 0.1 were then included in multivariable analyses. Possible interactions were
also ascertained to assess potential effect modifiers.
Separate analysis was conducted using conditional logistics regression for CBD and
their controls (total 216 subjects consisting of 61 CBD cases and 155 controls) and BeS and
their controls (total of 145 subjects consisting of 41 BeS and 104 controls). Further analysis
was also conducted for CBD and BeS cases only (61 CBD and 41 BeS) using unconditional
logistic regression to assess factors influencing the progression from BeS to CBD.
Whenever necessary, detailed analyses and multiple comparisons with Bonferroni
correction were conducted. Additional analysis for coding exposure, the Hockey Stick
method was also used since a proportion of individuals had exposure levels below the limit
of detection (close to zero, or even zero level). For analyses where exposure was coded
using Hockey Stick method, two variables for exposure were created and entered into the
model: 1) individuals with exposure bellow the limits of detection were assigned value of
0, and those that had exposure above the detectable limit were assigned a value of 1, and 2)
those with exposure above zero, were assigned their actual exposure levels. All analyses
evaluating the effect of exposure, genetics and their interaction on risk of CBD and BeS
were repeated using this approach for coding exposure.

26

Figure 3: Statistical Analysis
Outcome

Analysis on
Distribution
Characteristics

Dataset

Analysis on Distribution
Characteristics

61 CBD
155 Controls

61
CBD

Conditional Logistic
Regression
361
Subjects

41
BeS

ANOVA
Chi Square

41 BeS
104 Controls

61 CBD
41 BeS

259
Controls

27

Analysis of Risk Factors
for Progression from BeS
to CBD
Logistic Regression

CHAPTER IV
RESULTS

I.

DEMOGRAPHIC CHARACTERISTICS
We analyzed data from a total of 361 subjects consisting of 61 CBD, 41 BeS, and 259
controls. A comparison of CBD, BeS, and controls regarding their demographic
characteristics is shown in Table 5. There was a significant difference in proportion of race
(p= 0.0122) and gender (p= 0.0267) between CBD, BeS, and controls. Subjects were
mostly male (94.2%) and white (98.1%) and there were more non-white and female
individuals in the sensitized group compared to the control and CBD groups. There were
no significant difference in plant (p=0.5847) or history of smoking (p=0.8732).
Table 5. Comparison of Demographic Characteristics among Subjects with Chronic
Beryllium Disease, Beryllium Sensitization, and Controls
CBD
BeS
Control
Total
Characteristics
P value
N = 61
N = 41
N = 259
N = 361
N (%)
Gender
0.0267*
Male
57 (93.4)
37 (90.2)
246 (95.0) 340 (94.2)
Female
4 (6.6)
4 (9.8)
13 (5.0)
21 (5.8)
Race
0.0122*
61 (100)
39 (95.1)
254 (98.1) 354 (98.1)
White
Other
0 (0)
2 (4.9)
5 (1.9)
7 (1.9)
0.5847
Plant
Plant 1
28 (45.9)
23 (56.1)
133 (51.3) 184 (51.0)
Plant 2
33 (54.1)
18 (43.9)
129 (48.7) 177 (49.0)
0.8732
Smoking
Never
20 (32.8)
16 (39.0)
78 (30.1)
114 (31.6)
Ex- smoker
28 (45.9)
19 (46.3)
126 (48.6) 173 (47.9)
Current smoker
10 (16.4)
5 (12.2)
38 (14.7)
53 (14.7)
Unknown
3 (4.9)
1 (2.4)
17 (6.6)
21 (5.8)
Total
61 (17.0)
41 (11.4)
262 (72.6)
361 (100)
Comparison conducted by Chi Square; * comparison conducted by Fisher’s method
28

II.

EXPOSURE CHARACTERISTICS
Individuals working in Plant 2 (operating from 1935-2000) had significantly higher
exposure than those working in Plant 1 (operating from 1958-1978), in terms of cumulative
exposure, mean exposure and peak exposure (respectively, p = 0.0094; 0.0005; and <.0001)
as seen in Table 6.
There was a significant difference in type of exposure, in which individuals working
in Plant 1 had higher chemical exposure for mixed, non-soluble, and soluble chemical
compared to Plant 2. Although not statistically significant, individuals working in Plant 2
also had higher physical exposures than those in Plant 1.
Table 6. Exposure Characteristics by Plant
Plant
Exposure*
1
2
N=184 N=177
8.63
13.89
Duration (In Years)
Total Exposure
Cumulative
145.15 2555.88
Log Cumulative
4.03
2.47
Mean
1.63
23.20
Peak
3.45
159.49
Type of Exposure
Chemical
Mix
Cumulative
40.66 844.06
Mean
0.44
11.78
Peak
1.52
42.95
Cumulative
84.61 153.95
Non soluble
Mean
0.95
1.47
Peak
3.01
5.49
Cumulative
19.89
59.97
Soluble
Mean
0.24
0.56
Peak
1.09
4.87
Physical
Cumulative
37.14 276.88
Mix
Mean
0.41
3.24
Peak
1.47
13.22
Cumulative
**
309.68
Dust
29

Total
P
Average value
N=361
10.87
0.1233
599.55
4.36
7.47
24.08

0.0094
0.0094
0.0005
<.0001

434.57
6.00
21.83
118.61
1.21
4.23
39.54
0.40
2.94

<.0001
<.0001
<.0001
0.0009
<.0001
0.0008
<.0001
<.0001
<.0001

154.68
1.80
7.23
309.68

0.1358
0.3776
0.1235
n/a

Table 6 (cont’d). Exposure Characteristics by Plant
Plant
Total
P
Exposure*
Average value
1
2
N=361
N=184 N=177
Mean
**
8.29
8.29
n/a
Peak
3.04
29.84
16.18
0.1065
Cumulative
16.03 424.04
216.08 0.3727
Fume
Mean
0.20
2.28
1.22
0.0578
Peak
1.02
19.41
10.04
0.8474
*) Total exposures and different type of exposures were each measured in µg3
3
3
year/m unit for cumulative exposure, in µg/m for mean exposure, and in µg/m
for peak exposure
**) Data on cumulative and mean dust levels not available for Plant 1
P Value was obtained from Wilcoxon two sample test comparing 2 plants

Due to the higher exposure received by workers in Plant 2, we ran separate analyses
of exposure by CBD, BeS, and control status within each Plant and found differences of
duration between the three groups within each plant, but only found significant differences
in Plant 1 for cumulative exposure, and all measures of mixed chemical exposure and
mixed physical exposures among the three groups (Table 7).
Table 7. Exposure Characteristic by Outcome within Each Plant
Plant 1
Plant 2
Exposure*
CBD
BeS Control
P
CBD
BeS
Control
N=28 N=23 N=133 Value N=33 N=18
N=125
8.73
5.22
9.20
5.83
15.34
Duration (Years)
0.0487 8.72
Total Exposure
Cumulative
118.72 77.66 162.39 0.0366 437.9 232.28 1357.90
Log Cumulative
3.86
3.25
4.20
4.06
4.89
0.0366 4.37
Mean
1.42
1.53
1.69
0.2851 12.96 16.65
13.25
Peak
2.63
2.90
3.71
0.0849 24.13 22.25
54.45
Chemical Mix
Cumulative
20.93
9.51
50.20
0.0029 351.5 146.25 1072.75
Mean
0.28
0.37
0.48
11.25
0.0248 12.01 15.03
Peak
1.09
1.09
1.68
52.09
0.0347 20.42 20.33
Non Soluble
Cumulative
84.69 60.86
88.70
0.6954 86.17 86.02
181.40
Mean
0.94
0.83
0.97
0.7539 1.50
1.77
1.43
Peak
2.11
2.60
3.27
0.3259 4.70
3.18
6.02
30

P
Value
0.0090
0.3135
0.3135
0.6440
0.5337
0.4851
0.9544
0.4418
0.1432
0.2013
0.1705

Table 7 (Cont’d). Exposure Characteristic by Outcome within Each Plant
Plant 1
Plant 2
Exposure*
CBD
BeS Control
P
CBD
BeS
Control
N=28 N=23 N=133 Value N=33 N=18
N=125
Soluble
Cumulative
13.10
7.29
23.49
0.2498 0.18
0.00
84.20
Mean
0.21
0.33
0.24
0.5181 0.03
0.00
0.78
Peak
0.65
1.18
1.17
0.2776 2.72
0.00
6.13
Physical Mix
Cumulative
20.59
7.65
45.72
360.62
0.0023 74.61 61.49
Mean
0.25
0.32
0.46
5.95
3.29
0.0127 1.58
Peak
0.99
1.09
1.64
11.48
14.84
0.0260 7.98
Dust
Cumulative
**
**
**
n/a
195.5 123.19 366.21
Mean
**
**
**
n/a
7.45
10.79
8.15
Peak
2.11
2.60
3.32
0.2300 12.38 17.18
36.23
Fume
Cumulative
12.29
6.49
18.47
0.2837 165.5 16.02
550.03
Mean
0.18
0.30
0.19
0.4482 4.51
0.07
2.02
Peak
0.55
1.16
1.10
0.2238 12.74
1.34
23.74

P
Value
0.4226
0.5927
0.5974
0.1031
0.6392
0.1661
0.2759
0.2981
0.2505
0.2242
0.5128
0.2004

3

*) Total exposures and different type of exposures were each measured in µg-year/m unit for
3

3

cumulative exposure, in µg/m for mean exposure, and in µg/m for peak exposure
**) Data on cumulative and mean dust levels not available for Plant 1
P Value was obtained from Kruskal Wallis test comparing CBD, BeS and Controls

For further analysis, exposure was ascertained as cumulative exposure, because it
reflects the total exposure received by each subject by taking into account the duration of
exposure. A skewed distribution of cumulative exposure was detected, hence for further
analysis transformation to log scale for cumulative exposure was used (Appendix 1).
To understand the effect of exposure level on disease outcome, we compared
exposure level between Plant 1 and Plant 2 for each category of outcome (Table 8).

31

Table 8. Comparison of Exposure between Plant 1 and Plant 2 for Chronic Beryllium Disease, Beryllium Sensitization, and
Control Individuals
CBD
BeS
Control
Exposure*
Plant 1 Plant 2
P
Plant 1 Plant 2
P
Plant 1
Plant 2
P value
N=28
N=33
value
N=23
N=18
value N=133
N=125
Duration
Duration in years
8.73
8.72
0.9711
5.22
5.83
0.2263
9.20
15.34
<.0001
Total Exposure
Cumulative
118.72 437.91 0.4053 77.66
232.28 0.2026 162.39
1357.9
0.0258
Log Cumulative
3.86
4.37
0.4053
3.25
4.06
0.2026
4.20
4.89
0.0263
Mean
1.42
12.96 0.3657
1.53
16.65
0.0640
1.69
13.25
0.0024
Peak
2.63
24.13 0.1097
2.90
22.25
0.0568
3.71
54.45
<.0001
Type of Exposure
Chemical
Mix
Cumulative
20.93
351.49 0.0009
9.51
146.25 0.0037 50.20
1072.7
<.0001
Mean
0.28
12.01 0.0002
0.37
15.03
0.48
11.25
0.0079
<.0001
Peak
1.09
20.42 0.0026
1.09
20.33
1.68
52.09
0.0224
<.0001
Non Soluble
Cumulative
84.69
86.17 0.0379 60.86
86.02
0.0854 88.70
181.40
0.0246
Mean
0.94
1.50
0.83
1.77
0.1456
0.97
1.43
0.0054
0.0013
Peak
2.11
4.70
2.60
3.18
0.1607
3.27
6.02
0.0141
0.0359
Soluble
Cumulative
13.10
0.18
7.29
0.00
84.20
0.0054
0.0035 23.49
<.0001
Mean
0.21
0.03
0.33
0.00
0.24
0.78
0.0072
0.0035
<.0001
Peak
0.65
2.72
1.18
0.00
1.17
6.13
0.0039
0.0018
<.0001
Physical Mix
Cumulative
20.59
74.61 0.3270
7.65
61.49
0.5838 45.72
360.62
0.2396
Mean
0.25
1.58
0.6778
0.32
5.95
0.9628
0.46
3.29
0.2203
Peak
0.99
7.98
0.3230
1.09
11.48
0.8931
1.64
14.84
0.1640
32

Table 8 (Cont’d). Comparison of Exposure between Plant 1 and Plant 2 for Chronic Beryllium Disease, Beryllium
Sensitization, and Control Individuals
CBD
BeS
Control
Exposure*
Plant 1 Plant 2
P
Plant 1 Plant 2
P
Plant 1
Plant 2
P value
N=28
N=33
value
N=23
N=18
value N=133
N=125
Dust
Cumulative
Mean
Peak
Fume
Cumulative
Mean
Peak

**
**
2.11

195.54
7.45
12.38

n/a
n/a
0.6460

**
**
2.60

123.19
10.79
17.18

n/a
n/a
0.3852

**
**
3.32

366.21
8.15
36.23

n/a
n/a
0.0495

12.29
0.18
0.55

165.51
4.51
12.74

0.4419
0.8519
0.4866

6.49
0.30
1.16

16.02
0.07
1.34

0.4172
0.1130
0.0964

18.47
0.19
1.10

550.03
2.02
23.74

0.2837
0.1443
0.5413

3

*) Total exposures and different type of exposures were each measured in µg-year/m unit for cumulative exposure, in
3

3

µg/m for mean exposure, and in µg/m for peak exposure
**) Data was not obtained from Plant 1 for cumulative and mean physical dust exposure
Comparison was conducted using non parametric Wilcoxon two sample test comparing the level of exposures for each disease
category (Chronic Beryllium Disease, Beryllium Sensitization, and controls) between the two plants

33

Significant differences were found between Plant 1 and Plant 2 for duration,
cumulative exposure, peak exposure, and mean exposure in the control group, but not in
BeS or CBD (Table 8). Plant 2 had higher cumulative, mean, and peak exposure compared
to Plant 1 in control individuals. Controls in Plant 2 also had longer duration of exposure
compared to Plant 1. We also found a significant difference of exposure level between
Plant 1 and Plant 2 when analyzed within each disease status (Table 8).
Mixed chemical exposure was higher in Plant 2 for all disease categories (CBD, BeS
and controls). Non soluble chemical were only significantly different between plant 1 and
plant 2 for CBD and control, in which workers in Plant 2 had higher exposure than those
working in Plant 1. However, CBD and BeS individuals in Plant 1 had a higher exposure of
soluble chemical than those in Plant 2 (Table 8).
When data from the two plants was combined, there was no significant difference of
exposure between CBD, BeS, and control individuals, except for duration (p = 0.0010),
cumulative exposure (p = 0.0236), as well as in type of exposures for mixed chemical
exposure (p = 0.0230) and all measures of mixed physical exposure (p value respectively =
0.0007; 0.0247 and 0.0052 for cumulative, mean, and peak physical exposure) as shown in
Table 9.
Table 9. Exposure and Type of Exposure by Outcome
Outcome
Total
Exposure
CBD
BeS
Control
Average
N=61
N=41
N=259
Duration
Duration in years
Total Exposure
Cumulative
Log cumulative exposure
Mean
Peak

P
Value

8.89

5.49

12.19

10.87

0.0010

291.40
4.13
7.66
14.26
34

145.54
3.60
8.17
11.39

743.99
4.53
7.32
28.40

599.55
4.36
7.47
24.08

0.0236
0.0236
0.4931
0.1480

Table 9 (Cont’d). Exposure and Type of Exposure by Outcome
Outcome
Total
Exposure*
CBD
BeS
Control
Average
N=61
N=41
N=259
Type of Exposures
Chemical
Mix
Cumulative
199.76
69.54
547.65
434.57
Mean
6.62
6.81
5.72
6.00
Peak
11.54
9.54
26.20
21.83
Non Soluble
Cumulative
85.49
71.91
133.80
118.61
Mean
1.24
1.25
1.19
1.21
Peak
3.51
2.86
4.61
4.23
Soluble
Cumulative
6.11
4.09
53.03
39.54
Mean
0.11
0.19
0.50
0.40
Peak
1.77
0.66
3.58
2.94
Physical
Mix
Cumulative
49.81
31.29
198.92
154.68
Mean
0.97
2.79
1.84
1.80
Peak
4.77
5.65
8.06
7.23
Dust
Cumulative
195.54
123.19
366.21
309.68
Mean
7.45
10.79
8.15
8.29
Peak
7.66
9.00
19.33
16.18
Fume
Cumulative
95.18
10.67
277.07
216.08
Mean
2.52
0.20
1.08
1.22
Peak
7.14
1.24
12.11
10.04

P
Value

0.0230
0.2151
0.0599
0.1540
0.4830
0.0594
0.1461
0.3209
0.2371

0.0007
0.0247
0.0052
0.2759
0.2981
0.0627
0.1464
0.3790
0.1398
3

*) Total exposures and different type of exposures were each measured in µg-year/m unit
3

3

for cumulative exposure, in µg/m for mean exposure, and in µg/m for peak exposure
Comparison was conducted by Kruskal Wallis non parametric test

35

III. GENETIC CHARACTERISTICS FOR BOTH PLANTS AND COMBINED
Table 10 shows that there was no significant difference of genetics distribution
between Plant 1 and Plant 2 except for Serine 13 and Serine 11 (Table 10). However, when
compared by disease states (Table 11) these two genes showed no significant difference in
proportion among CBD, BeS, and controls.
Table 10. Genetics Distribution by Plant
Gene
Plant 1
Plant 2
(N, %)
N = 184
N = 177
Glutamine 69
Positive
95 (51.6)
94 (53.1)
Negative
89 (48.4)
83 (46.9)
Glutamine 71
Positive
48 (26.1)
48 (27.1)
Negative
136 (73.9)
129 (72.9)
Serine 11
Positive
141 (76.7)
109 (61.6)
Negative
43 (23.3)
68 (38.4)
Serine 13
Positive
128 (69.6)
102 (57.6)
Negative
56 (30.4)
75 (42.4)
Arginine 74
Positive
37 (20.1)
25 (14.1)
Negative
147 (79.9)
152 (85.9)
Asparagine 37
Positive
73 (39.7)
59 (33.3)
Negative
111 (60.3)
118 (66.7)
Histidine 32
Positive
89 (48.4)
73 (41.2)
Negative
95 (51.6)
104 (58.8)
Phenyl alanine 47
Positive
142 (77.2)
130 (73.4)
Negative
42 (22.8)
47 (26.6)
Tyrosine 26
Positive
38 (20.7)
27 (15.2)
Negative
146 (79.3)
150 (84.8)
Comparison was conducted using Chi Square

36

P value
0.7788

0.8245

0.0020

0.0184

0.1318

0.2111

0.1735

0.4114

0.1821

Glutamine 69 and Glutamine 71 were significantly different between the three groups
(respectively, p =<.0001 and 0.0026) as shown in Table 11. We also found a significant
difference in the distribution of homozygosity (p value < 0.0001) and non-0201 alleles
among glutamine 69 positive individuals (p=0.0167) in which BeS and CBD had a higher
proportion of non-0201 alleles compared to controls.
Table 11. Comparison of Gene Distribution between CBD, BeS and Control
Gene
CBD
BeS
Control
Total
P value
(N, %)
N = 61
N = 41
N=259
N = 361
Glutamine 69
Positive
56 (91.8) 32 (78.1) 101 (39.0) 189 (52.4) <.0001
Negative
5 (8.20)
9 (21.9) 158 (61.0) 172 (47.6)
Glu69 homozygosity
Homozygous
10 (16.4)
8 (19.5)
17 (6.6)
35 (9.7)
<.0001
Heterozygous
46 (75.4) 24 (58.5) 84 (32.4) 154 (42.7)
Negative
5 (8.2)
9 (22.0) 158 (61.0) 172 (47.6)
Glu69-0201 allele
Positive
28 (45.9) 17 (41.5) 72 (27.8) 117 (32.4)
0.0167
Negative
28 (45.9) 15 (36.6) 29 (11.2) 72 (20.0)
Glu69 negative
5 (8.20)
9 (21.9) 158 (61.0) 172 (47.6)
Glutamine 71
Positive
16 (26.2) 20 (48.8) 60 (23.2) 96 (26.6)
0.0026
Negative
45 (73.8) 21 (51.2) 199 (76.8) 265 (73.4)
Serine 11
Positive
44 (72.1) 32 (78.1) 174 (67.2) 250 (69.3) 0.3248
Negative
17 (27.9)
9 (21.9)
85 (32.8) 111 (30.7)
Serine 13
Positive
38 (62.3) 30 (73.2) 162 (62.6) 230 (63.7) 0.4083
Negative
23 (37.70) 11 (26.8) 97 (37.4) 131 (36.3)
Arginine 74
Positive
10 (16.4)
5 (12.2)
47 (18.1) 62 (17.2)
0.6335
Negative
51 (83.6) 36 (87.8) 212 (81.9) 299 (62.8)
Asparagine 37
Positive
18 (29.5) 20 (48.8) 94 (36.3) 132 (36.6) 0.1384
Negative
45 (70.5) 21 (51.2) 165 (63.7) 229 (63.4)
Histidine 32
Positive
27 (44.3) 21 (51.2) 114 (44.0) 162 (44.9) 0.6860
Negative
34 (55.7) 20 (48.8) 145 (56.0) 199 (55.1)

37

Table 11 (Cont’d). Comparison of Gene Distribution between CBD, BeS and Control
Gene
CBD
BeS
Control
Total
P value
(N, %)
N = 61
N = 41
N=259
(N = 361)
Phenyl alanine 47
Positive
46 (75.4) 30 (73.2) 196 (75.7) 272 (75.3) 0.9419
Negative
15 (24.6) 11 (26.8) 63 (24.3) 89 (24.7)
Tyrosine 26
Positive
10 (16.4)
5 (12.2)
50 (19.3) 65 (18.0)
0.5114
Negative
51 (83.6) 36 (87.8) 209 (80.7) 296 (82.0)
Comparison was conducted using Chi Square

IV. GENETICS AND EXPOSURE ASSOCIATION WITH CHRONIC BERYLLIUM
DISEASE AND BERYLLIUM SENSITIZATION
a.

Genetic and Exposure Association with Chronic Beryllium Disease
On univariable conditional logistic regression HLA-DPB1glu69 and allele type were

found to have a significant association with the development of CBD relative to glutamine
69 negative individuals (Table 12, complete univariable analysis is shown in Appendix 5).
The 0201 negative allele had a greater association with development of CBD compared to
0201 positive.
Table 12. Factors Significantly Associated with CBD on Univariable Analysis
Variable
Coefficie Standard
OR
95% Confidence
P value
nt
Error
Interval
3.2970
0.7278
27.03
6.49
112.56
Glutamine 69
<.0001
Glutamine 69 allele
0201 negative
1.3668
0.3170
35.02
7.96
154.01
<.0001
0201 positive
0.8224
0.3164
20.32
4.63
89.24
0.0094

Multivariable conditional logistic regression was then conducted, by including several
variables that had a marginally significant association with CBD and also checked for
biologically plausible interactions. From multivariable logistic regression, HLADPB1glu69 was found to be the only significant factor related to the development of CBD after
38

adjusting for log cumulative exposure (Table 13) with an Odds Ratio of 27.52 (95% CI
6.56-115.35). There was no significant interaction found in the analysis.
Table 13. Multivariable Conditional Logistic Regression for the Development of CBD
Coeffici Standard
95% Confidence
Variable
OR
P Value
ent
Error
Interval
Glutamine 69
Glutamine 69
3.3147
0.7312
27.52
6.56 115.35
<.0001
Log cumulative exposure
0.0231
0.0872
1.02
0.86
1.21
0.7915
By allele type
Glutamine 69 (0201 -)
3.5861
0.7596
36.09
8.15 159.95
<.0001
Glutamine 69 (0201+)
3.0234
0.7571
20.56
4.66
90.69
<.0001
Log cumulative exposure
0.0356
0.0887
1.04
0.87
1.23
0.6879
Glutamine 69 negative
Ref
Comparison
Glutamine 69 (0201- vs.
0.5697
0.4109
1.75
0.78
3.92
0.1714
0201+)

When analyzed based on allele type, we found that subjects with non-0201 alleles had
higher OR compared to those with 0201 alleles (respectively, OR 36.09 95% CI 8.15159.95; OR 20.56 95% CI 4.66-90.69), although when contrasted, the difference between
non-0201 alleles and 0201 alleles was not significant (p=0.1714).

b. Genetic and Exposure Interaction with Beryllium Sensitization
The same procedure was conducted for BeS and controls (total of 145 subjects, 41
BeS and 104 controls). HLA-DPB1glu69 and allele type had a significant association with
the development of BeS as shown in table 14 (complete univariable analysis is shown in
Appendix 7).

39

Table 14. Factors Significantly Associated with BeS on Univariable Analysis
95%
Standard
Variable
Coefficient
OR
Confidence
P value
Error
Interval
Glutamine 69
1.8430
0.4650
6.32
2.54 15.71
<.0001
Glutamine 71
0.9992
0.3842
2.72
1.28
5.77
0.0093
Glutamine 69 allele
< .0001
0201 positive
0.0787
0.2944
4.49
1.70 11.83
0.7893
0201 negative
1.3433
0.3946
15.88
4.25 59.35
0.0007

Multivariable conditional logistic regression was then conducted by including several
variables that had a marginally significant association with BeS (Appendix 7) and also
biologically plausible interaction in the model. From multivariable logistic regression,
glutamine 69 and glutamine 71 were found as significant factors related to the development
of BeS after adjusting for other variables in the model with OR respectively 7.08 95% CI
2.59-19.35 and 0R 2.54 95% CI 1.06-6.12 (Table 15). There was no significant interaction
found on the analysis.
Table 15. Multivariable Conditional Logistic Regression for the Development of BeS
95%
Coeffici Standard
P
Variable
OR
Confidence
ent
Error
Value
Interval
Glutamine 69
Glutamine 69
1.9565
0.5134
7.08
2.59
19.35 0.0001
Glutamine 71
0.9337
0.4479
2.54
1.06
6.12
0.0371
Log cumulative exposure
-0.2286
0.1230
0.80
0.63
1.01
0.0632
By allele
Glutamine 69 (0201 -)
3.1966
0.7875
24.45
5.22
114.45 <.0001
Glutamine 69 (0201+)
1.5676
0.5696
4.80
1.57
14.64 0.0059
Glutamine 71
1.0005
0.4657
2.72
1.09
6.78
0.0317
Log cumulative exposure
-0.2949
0.1378
0.75
0.57
0.98
0.0324
Comparison
Glutamine 69 (0201 – vs.
1.5486
0.6485
4.71
1.32
16.77 0.0170
0201 +)

40

When analyzed based on allele type, we found that subjects with non-0201 alleles had
higher OR compared to those with 0201 alleles (respectively, OR 24.45 95% CI 5.22114.45; OR 4.80 95% CI 1.57-14.64). This difference was significant when contrasted
between non-0201 alleles and 0201 alleles (p=0.0170).

V.

PROGRESSION OF CBD FROM BERYLLIUM SENSITIZATION
It is still uncertain what factors influence the progression of BeS to CBD.
Unconditional logistic regression was conducted to assess this association only in
individuals with CBD (61 subjects) and BeS (41 subjects). From univariable logistic
regression we found Glutamine71 as a significant predictor blocking the progression to
CBD (Table 16).
Table 16. Factors Which Significantly Differentiate CBD and BeS on Univariable
Analysis
Standard
95% Confidence
P
Variable
Coefficient
OR
Error
Interval
Value
Glu69
1.1474
0.6002
3.15
0.97
10.21
0.0559
Glutamine71
-0.9852
0.4270
0.37
0.16
0.86
0.0210
Allele
0201 1.2119
0.6430
3.36
0.95
11.85
0.0595
0201 +
1.0867
0.6369
2.97
0.85
10.33
0.0880

From multivariable logistic regression, only Glutamine 71 continued to be a factor
related to reducing the risk of progression to CBD after adjusting for log cumulative
exposure as seen in Model 2 of Table 17 (OR = 0.38 95% CI 0.16-0.87). There was no
significant interaction found in the analysis.

41

Table 17. Unconditional Multivariable Logistic Regression Analysis on Progression from
BeS to CBD
Standard
95% Confidence
P
Variable
Coefficient
OR
Error
Interval
Value
Model 1
Intercepts
-0.2625
0.8506
0.7577
0201 +
0.4931
0.7232
1.64
0.40
6.76
0.4953
0201 0.4912
0.7857
1.63
0.35
7.62
0.5319
Glutamine 71
-0.7948
0.5244
0.45
0.16
1.26
0.1296
Log Cum Exposure
0.1372
0.1119
1.15
0.92
1.43
0.2204
Model 2
Intercepts
0.1979
0.4967
0.6902
Glutamine 71
-0.9812
0.4309
0.38
0.16
0.87
0.0228
Log Cum Exposure
0.1455
0.1106
1.16
0.93
1.44
0.1884

VI. EXPOSURE CHRACTERISTICS BY GENETICS AND DISEASE STATUS
From univariable and multivariable conditional logistic regression analysis, the
amount of exposure was shown to have no significant association with the development of
CBD and BeS although exposure is a necessary cause for beryllium-related toxicity.
Glutamine 69 was consistently shown to be the significant factor related to the
development of CBD and BeS. Approximately 39% of control individuals who tested
positive for HLA-DPB1Glu69 gene, however, were not sensitized and remained disease
free. Hence, another analysis was conducted to assess the exposure by HLA-DPB1glu69
status, to see whether control individuals who were HLA-DPB1 positive had lower
exposure, and therefore did not develop beryllium toxicity.
From Table 18 we can see that there was a difference although not statistically
significant in cumulative exposure among control subjects, who were glu69 positive and
glu69 negative (p= 0.4106). The result also shows that control individuals with the non0201 allele have higher exposure than those with the 0201 allele, which infers that this
allele cannot explain why these individuals remain healthy.
42

In CBD and BeS individuals however, the cumulative exposure is higher in those
positive for Glutamine 69 compared to individuals with glutamine 69 negative, but these
individuals were all positive for glutamine 71 which from our previous reports was shown
to be a risk factor in development of CBD and BeS in the absence of glu69.(63) (Table
18). When comparing based on alleles type, individuals carrying 0201 allele had higher
exposure compared to those with non-0201 alleles, inferring that individuals who are 0201
negative are more likely to get CBD or BeS, even with less exposure, although the
difference is not significant (Table 18).
Table 18. Comparison of Cumulative, Log Cumulative, Mean, and Peak Exposure between
CBD, BeS, and Control Groups based on HLA-DPB1Glu69 presence and allele type
Outcome
Exposure
N
Glu69
Mean
P value
Control
101
Positive
312.17
Cumulative
0.4061
158
Negative
1020.03
101
Positive
4.27
Log cumexp
0.4061
158
Negative
4.70
By gene
101
Positive
6.83
Mean
0.7197
158
Negative
7.62
101
Positive
11.84
Peak
0.4500
158
Negative
38.98
29
0201 negative
407.15
Cumulative
72
0201 positive
273.91
0.4106
158
Negative
1020.03
29
0201 negative
4.60
Log cumexp
72
0201 positive
4.14
0.4106
158
Negative
4.70
By allele
29
0201 negative
8.15
Mean
72
0201 positive
6.30
0.6000
158
Negative
7.62
29
0201 negative
13.27
Peak
72
0201 positive
11.26
0.3339
158
Negative
38.98
CBD
56
Positive
307.65
By gene
Cumulative
0.9685
5
Negative
109.36
Log cumexp
56
Positive
4.15
0.9685
43

Table 18 (Cont’d). Comparison of Cumulative, Log Cumulative, Mean, and Peak Exposure
between CBD, BeS, and Control based on HLA-DPB1Glu69 presence and allele type
Outcome
Exposure
N
Glu69
Mean
P value
5
Negative
3.94
56
Positive
8.16
Mean
0.7524
5
Negative
2.10
56
Positive
15.28
Peak
0.7030
5
Negative
2.92
28
0201 negative
176.31
Cumulative
28
0201 positive
438.99
0.1849
5
Negative
109.36
28
0201 negative
3.69
Log cumexp
28
0201 positive
4.61
0.1849
5
Negative
3.94
By allele
28
0201 negative
10.27
Mean
28
0201 positive
6.05
0.9469
5
Negative
2.10
28
0201 negative
18.43
Peak
28
0201 positive
12.12
0.5815
5
Negative
2.92
BeS
56
Positive
166.95
Cumulative
0.3060
5
Negative
69.40
56
Positive
3.76
Log cumexp
0.3060
5
Negative
3.03
By gene
56
Positive
9.43
Mean
0.1807
5
Negative
3.67
56
Positive
13.42
Peak
0.1705
5
Negative
4.20
15
0201 negative
132.27
Cumulative
17
0201 positive
197.55
0.3147
9
Negative
69.40
15
0201 negative
4.03
Log Cumexp
17
0201 positive
3.52
0.3147
9
Negative
3.03
By allele
15
0201 negative
12.91
Mean
17
0201 positive
6.36
0.3274
9
Negative
3.67
15
0201 negative
14.68
Peak
17
0201 positive
12.31
0.3736
9
Negative
4.20
Comparison was conducted by Wilcoxon two sample test and Kruskal Wallis non parametric
test
44

VII. EXPOSURE CHARACTERISTICS IN INDIVIDUALS WITH GLUTAMINE 69
To see whether there was a different level of exposure among those who were
susceptible to developing CBD or BeS, we compared exposure levels among the 189
subjects who were positive for glutamine 69 (Table 19). We found no significant difference
in cumulative, mean, and peak exposure as well as type of exposure between CBD, BeS,
and controls in glutamine 69 positive individuals. The only significant difference we found
was in duration of exposure, in which the controls had a longer exposure followed by CBD
and BeS (p=0.0217). However, we observed a trend of dose-response in peak exposure in
which subjects with CBD had higher exposure than BeS and controls (respectively, peak
exposure 15.28, 13.42, 11.84) although the difference was not significant (p=0.6162).
Table 19. Comparison of Magnitude and Type of Exposures between CBD, BeS, and
Control Groups based on Individuals with Glutamine 69
Outcome
Total
Exposure
CBD
BeS
Control Average P Value
N=189
N=56
N=32
N=101
Total Exposure
Duration
9.18
5.08
11.51
9.73
0.0217
Cumulative
307.65
166.95
312.17
286.24
0.3329
Log Cumulative exposure
4.15
3.76
4.27
4.15
0.3329
Mean
8.16
9.43
6.83
7.67
0.4739
Peak
15.28
13.42
11.84
13.12
0.6162
Chemical
Cumulative
215.31
78.12
175.27
170.69
0.4746
Mean
7.10
7.84
5.37
6.30
0.7408
Mix
Peak
12.45
11.25
10.05
10.96
0.6204
Cumulative
91.20
84.56
114.29
102.41
0.5331
Mean
1.32
1.53
1.36
1.38
0.2453
Non Soluble
Peak
3.69
3.49
3.44
3.52
0.0662
Cumulative
1.10
4.28
18.53
10.95
0.3369
Mean
0.07
0.16
0.12
0.11
0.3962
Soluble
Peak
1.87
0.65
0.58
0.98
0.3896
Physical
Cumulative
53.06
31.86
116.65
83.45
0.0947
Mix
Mean
0.96
2.73
2.42
2.04
0.1743
45

Table 19 (Cont’d). Comparison of Magnitude and Type of Exposures between CBD, BeS,
and Control Groups based on Individuals with Glutamine 69
Outcome
Total
Exposure
CBD
BeS
Control Average P Value
N=189
N=56
N=32
N=101
Peak
5.10
6.33
5.25
5.39
0.2497
Cumulative
208.16
184.61
198.71
200.03
0.5315
Mean
7.93
16.17
6.62
8.27
Dust
0.0117
Peak
8.21
11.36
8.16
8.72
0.0497
Cumulative
97.05
9.96
34.71
48.99
0.8933
Mean
2.69
0.15
0.57
1.13
0.7378
Fume
Peak
7.70
1.33
3.22
4.23
0.8763
3

*) Total exposures and different type of exposures were each measured in µg-year/m unit
3

3

for cumulative exposure, in µg/m for mean exposure, and in µg/m for peak exposure
Comparison was obtained using Kruskal Wallis non parametric test

VIII. ANALYSIS OF THE EFFECT OF TYPE OF EXPOSURE IN THE
DEVELOPMENT OF CBD AND BES USING THE HOCEKY STICK APPROACH
FOR CODING EXPOSURE
Based on previous reports, the type of beryllium exposure has been associated with
the development of CBD and BeS.(5,11) To test this hypothesis, we ran analyses on
exposures that showed marginally significant differences with the Kruskal Wallis test in
previous section based on Table 19. However, as noted previously, proportion of
individuals had exposure levels below the limit of detection (close to zero, or even zero
level).
To assess the effect of exposures that contained zero or minimal exposure, we used
the hockey stick method, which assigns a categorical variable for individuals exposed
higher than the detectable level and lower than the detectable level (zero values), and a
continuous variable for individuals that had higher than the detectable level. The
categorical variable compared individuals with lower than detectable levels to individuals
46

with detectable levels of exposure, while log scale showed the effect of detectable levels of
exposure on the development of CBD. From the analyses, type of exposure shows no
significant association with development of CBD (Table 20). These types of analyses were
repeated for assessing the effect of exposure on the development of CBD, BeS, and
progression from BeS and CBD, as well as repeating the analyses for glu69 positive only.
a. Hockey Stick Analyses for CBD
Table 20. Conditional Logistic Regression for the Development of CBD by Type of
Exposure with Hockey Stick Analysis
Coeffici Standard
95% Confidence
Variable
OR
P Value
ent
Error
Interval
Mean Mixed Chemical
Glutamine 69
3.2945
0.7306
26.96
6.44
112.90 <.0001
Mean mixed chemical
-0.1658
0.4171
0.85
0.37
1.92
0.691
Log mixed chemical
0.0251
0.1124
1.03
0.82
1.28
0.8236
Peak Mixed Chemical
Glutamine 69
3.2744
0.7288
26.43
6.33
110.26 <.0001
Peak Mixed Chemical
-0.0779
0.4812
0.93
0.36
2.38
0.8713
Log peak mixed chemical
-0.0558
0.1544
0.95
0.70
1.28
0.718
Cumulative Soluble
Glutamine 69
3.3435
0.746
28.32
6.56
122.20 <.0001
Chemical Soluble
-0.72
1.0561
0.49
0.06
3.86
0.4954
Log cum soluble
-0.0274
0.2896
0.97
0.55
1.72
0.9247
Peak Soluble
Glutamine 69
3.3953
0.7573
29.82
6.76
131.58 <.0001
Peak Soluble
-0.5268
0.7586
0.59
0.13
2.61
0.4874
Log Peak Soluble
-0.3155
0.5137
0.73
0.27
2.00
0.5391

To have better understanding on the effect of exposure on the susceptible individuals,
we run for only those that were glu69 positive. Similar results were also found as seen on
table 21. There was no significant effect of different types of exposures on the
development of Chronic Beryllium Disease.

47

Table 21. Conditional Logistic Regression for the Development of CBD by Type of
Exposure with Hockey Stick Analysis in Glutamine 69 Positive Individuals
Coeffici Standard
95% Confidence
Variable
OR
P Value
ent
Error
Interval
Mean Mixed Chemical
Mean mixed chemical
0.1521
0.4562
1.16
0.48
2.85
0.7387
Log mixed chemical
-0.0303
0.1180
0.97
0.77
1.22
0.7972
Peak Mixed Chemical
Peak mixed chemical
0.3304
0.5298
1.39
0.49
3.93
0.5330
Log Peak mixed chemical
-0.1079
0.1669
0.90
0.65
1.25
0.5180
Chemical Soluble
Chemical Soluble
1.5400
1.2311
4.66
0.42
52.09
0.2110
Log cum soluble
-0.7568
0.4558
0.47
0.19
1.15
0.0968
Peak Soluble
Peak Soluble
0.0902
0.8185
1.09
0.22
5.44
0.9123
Log Peak Soluble
-0.8516
0.7934
0.43
0.09
2.02
0.2831

b. Hockey Stick Analysis for BeS
Similar analyses were conducted for factors associated with the development of BeS
(Table 22). Only log mixed chemical exposure showed a significant exposure influence on
the development of BeS (OR 1.5 with 95% CI 1.1-2.2, p =0.0171), which means that
increases in mean mixed chemical exposure among people exposed at higher than the
detectable limit, also increased the likelihood for the development of sensitization.
Table 22. Conditional Logistic Regression for the Development of BeS by Type of Exposure
with Hockey Stick Analysis
Coeffici Standard
95% Confidence
Variable
OR
P Value
ent
Error
Interval
Mean Mixed Chemical
Glu71
1.2711
0.4968
3.56
1.35
9.44
0.0105
Glu69
2.1322
0.5587
8.43
2.82
25.21
0.0001
Mean mixed chemical
-0.6448
0.4730
0.52
0.21
1.33
0.1728
Log mixed chemical
0.4255
0.1785
1.53
1.08
2.17
0.0171
Peak Mixed Chemical
Glu71
1.1331
0.4702
3.11
1.24
7.80
0.0160
Glu69
2.0218
0.5225
7.55
2.71
21.03
0.0001
Peak Mixed Chemical
-0.6992
0.5458
0.50
0.17
1.45
0.2001
Log peak mixed chemical
0.2285
0.2009
1.26
0.85
1.86
0.2554
48

Table 22 (Cont’d). Conditional Logistic Regression for the Development of BeS by Type of
Exposure with Hockey Stick Analysis
Coeffici Standard
95% Confidence
Variable
OR
P Value
ent
Error
Interval
Cumulative Soluble
Glu71
1.1615
0.4682
3.19
1.28
8.00
0.0131
Glu69
1.9452
0.5073
7.00
2.59
18.91
0.0001
Chemical Soluble
1.4402
1.0612
4.22
0.53
33.79
0.1748
Log cum soluble
-0.5472
0.3559
0.58
0.29
1.16
0.1241
Peak Soluble
Glu71
1.1166
0.4528
3.05
1.26
7.42
0.0137
Glu69
2.0222
0.5156
7.55
2.75
20.75
<.0001
Peak Soluble
0.9557
1.3997
2.60
0.17
40.41
0.4948
Log Peak Soluble
-0.5571
1.3520
0.57
0.04
8.11
0.6803

However, among the susceptible individuals (those with Glu69 positive), the
association is not observed (Table 24). There was no significant effect of different types of
exposures in the development of BeS among individuals with the glutamine 69
polymorphism.
Table 23. Conditional Logistic Regression for the Development of BeS by Type of Exposure
with Hockey Stick Analysis among Glutamine 69 Positive Individuals
Coeffici Standar
95% Confidence
Variable
OR
P Value
ent
d Error
Interval
Mean Mixed Chemical
Glutamine 71
-0.2126
0.6417
0.81
0.23
2.84
0.7404
Mean mixed chemical
-0.9344
0.6337
0.39
0.11
1.36
0.1404
Log mixed chemical
0.2803
0.2161
1.32
0.87
2.02
0.1946
Peak Mixed Chemical
Glu71
-0.2548
0.6173
0.78
0.23
2.60
0.6798
Peak Mixed Chemical
-1.1842
0.7817
0.31
0.07
1.42
0.1298
Log peak mixed chemical
0.2415
0.2819
1.27
0.73
2.21
0.3917
Chemical Soluble
Glu71
-0.4032
0.6359
0.67
0.19
2.32
0.5260
Chemical Soluble
0.4066
1.3436
1.50
0.11
20.91
0.7622
Log cum soluble
-0.6738
0.5344
0.51
0.18
1.45
0.2074
Peak Soluble
Glu71
-0.4053
0.6130
0.67
0.20
2.22
0.5085
Peak Soluble
0.8233
3.1929
2.28
0.00
1189.61 0.7965
Log Peak Soluble
-1.5991
3.4192
0.20
0.00
164.44
0.6400
49

c. Hockey Stick Analyses on Progression of BeS to CBD
Analyses were also conducted to evaluate the influence of these exposures on the
development of CBD from BeS. No significant association was found between exposures
and CBD vs. BeS (Table 24).
Table 24. Unconditional Logistic Regression Comparing CBD and BeS by Type of
Exposure with Hockey Stick Analysis
Coeffici Standar
95% Confidence
Variable
OR
P Value
ent
d Error
Interval
Mean Mixed Chemical
Intercepts
0.5548
0.3702
0.1340
Glutamine 71
-0.9265
0.4341
0.40
0.17
0.93
0.0328
Mean mixed chemical
0.3958
0.4313
1.49
0.64
3.46
0.3587
Log mixed chemical
-0.1452
0.1297
0.86
0.67
1.12
0.2628
Peak Mixed Chemical
Intercepts
0.6212
0.3730
0.0959
Glu71
-0.9832
0.4346
0.37
0.16
0.88
0.0237
Peak Mixed Chemical
0.3830
0.4941
1.47
0.56
3.86
0.4383
Log peak mixed chemical
-0.0925
0.1678
0.91
0.66
1.27
0.5814
Chemical Soluble
Intercepts
0.9835
0.3085
0.0014
Glu71
-1.1802
0.4498
0.31
0.13
0.74
0.0087
Chemical Soluble
-1.3302
0.8847
0.26
0.05
1.50
0.1327
Log cum soluble
0.2456
0.3013
1.28
0.71
2.31
0.4150
Peak Soluble
Intercepts
0.9813
0.3055
0.0013
Glu71
-1.1184
0.4426
0.33
0.14
0.78
0.0115
Peak Soluble
-1.0305
0.7326
0.36
0.08
1.50
0.1596
Log Peak Soluble
0.1629
0.4965
1.18
0.44
3.11
0.7428
We conducted similar analysis in individuals with glutamine 69 positive and found
similar result as seen in Tale 25. There were no significant association found between the
type of exposure and development of CBD compared to BeS on subjects positive for
Glutamine 69 (Table 25).
50

Table 25. Unconditional Logistic Regression Comparing CBD and BeS by Type of
Exposure with Hockey Stick Analysis among Glutamine 69 Positive Individuals
Coeffici Standar
95% Confidence
Variable
OR
P Value
ent
d Error
Interval
Mean Mixed Chemical
Intercepts
0.4445
0.3822
0.2448
Glutamine 71
-0.6531
0.5139
0.52
0.19
1.42
0.2037
Mean mixed chemical
0.6029
0.4732
1.83
0.72
4.62
0.2027
Log mixed chemical
-0.1800
0.1385
0.84
0.64
1.10
0.1939
Peak Mixed Chemical
Intercepts
0.4600
0.3827
0.2293
Glu71
-0.6985
0.5136
0.50
0.18
1.36
0.1738
Peak Mixed Chemical
0.8010
0.5628
2.23
0.74
6.71
0.1547
Log peak mixed chemical
-0.1767
0.1816
0.84
0.59
1.20
0.3305
Chemical Soluble
Intercepts
1.0015
0.3130
0.0014
Glu71
-0.9636
0.5243
0.38
0.14
1.07
0.0661
Chemical Soluble
-1.0539
0.8634
0.35
0.06
1.89
0.2222
Log cum soluble
0.0581
0.3350
1.06
0.55
2.04
0.8622
Peak Soluble
Intercepts
1.0008
0.3129
0.0014
Glu71
-0.9617
0.5241
0.38
0.14
1.07
0.0665
Peak Soluble
-1.0844
0.7570
0.34
0.08
1.49
0.1520
Log Peak Soluble
0.1427
0.4913
1.15
0.44
3.02
0.7715

51

CHAPTER V
DISCUSSION

Our results show the importance of glutamine 69 and glutamine 71 in the development of
beryllium toxicity. Although there was no clear dose-response association of beryllium exposure
for the development of CBD or BeS, we observed a trend of increasing peak levels and the
prevalence of CBD and BeS in individuals with glutamine 69, although this difference was not
statistically significant. There was no interaction between genetics and exposure levels observed
in our analyses, but we did find that despite having the highest exposure, individuals without
either the glutamine 69 or glutamine 71 polymorphisms remained healthy.
Our analysis was conducted on one of the largest cohorts available for studying the effect
of beryllium exposure and genetics, and consisted of 361 subjects. We found a significant
difference in the proportion of individuals with glutamine 69 among CBD, BeS and controls (p
value <0.0001) in which this gene were present in 91.8% of CBD cases and 78.1% of Bes
compared to 39.0% in controls (Table 11). This result is consistent with previous studies that
also showed a higher proportion of glutamine 69 positive individuals in CBD cases (56) followed
by BeS, and the lowest proportion of glutamine 69 positive was found in controls.(58)
We also found significant differences in homozygosity (p <0.0001) between CBD, BeS,
and controls. Approximately 17.8% (10/56) of CBD and 28.1% (9/32) of BeS cases were
homozygous compared to 16.8% (17/101) of controls. This higher proportion of homozygosity in
BeS cases compared to controls is consistent with previous results.(58, 59)
The importance of non-0201 alleles in the development of CBD and BeS had also been
previously reported.(58,59) In our study, a higher proportion of 0201 negative alleles were found
52

in CBD and BeS cases compared to controls. Among those with HLA-DPB1Glu69, the highest
proportion of non-0201 carrying individuals were found in CBD (50%) compared to BeS
(46.8%) and controls (28.1%). This is also consistent with previous studies conducted by Wang
et al who reported a higher proportion of non-0201 allele carriage among CBD, and control
individuals (proportion respectively 84% and 32%).(58,59)
The importance of glutamine 71 in beryllium toxicity, in accordance with a previous
report on this cohort, was reconfirmed in this analysis.(63) The significantly higher proportion
of glutamine 71 carrying individuals was found in BeS cases (48.8%) compared to CBD and
controls (proportions respectively 26.2% and 23.2%). In further analyses we also found that all
diseased individuals (CBD and BeS) had either or both glutamine 69 or glutamine 71
polymorphisms, compared to only 32.4% of controls. However, when we assessed the
importance of glutamine 71 in the development of CBD and BeS among the HLA-DPB1Glu69
negative individuals we found no significant association (p value 0.9984) (Appendix 4).
Consistent with a previous report on this cohort, the highest level of exposure was found
in control individuals, followed by CBD and BeS.(28) On pairwise comparison, we found a
significant difference between BeS and controls for cumulative mixed chemical exposure, peak
mixed chemical exposure, and cumulative mixed physical exposure (Appendix 2). There were no
significant differences in total exposure between CBD, BeS and controls, except for duration of
work and cumulative exposure, (p value 0.0010; 0.0236; respectively). For different type of
exposures we found significant difference for mixed chemical exposure (p = 0.0230) and all
measures of mixed physical exposure (Table 9). The highest levels were found in controls,
followed by CBD and BeS (Table 9). On pairwise comparison, the significant differences were

53

only found among BeS and controls for duration, cumulative mixed physical exposure, and peak
mixed physical exposure (Appendix 3).
Observing higher exposure in controls, it is unlikely that a policy to move diseased
workers from exposed areas is responsible for the higher level of exposures that were found in
controls, because the diagnosis of CBD and BeS was generally made many years or even
decades after the worker had left the Plant and the exposure had ceased. Following the findings
from the previous report on this cohort, we also hypothesized that host factors, i.e. genetic
susceptibility might explain these findings.(28)
To assess the importance of genetic susceptibility in the development of CBD and BeS
we ran separate conditional logistic regression analyses on CBD and their controls and on BeS
and their controls. Univariable conditional logistic regression showed a significant association of
glutamine 69 with the development of CBD (Table 12 and Appendix 5) with a crude odds ratio
of 27.0 (95% CI 6.5-112.6).
When considering allele type we found that, consistent with the previous report, non0201 carriage has a higher risk for developing CBD (OR for non-0201 carrier was 35.02 and
95% CI 7.96-154.01, OR for 0201 carrier was 20.32 and 95% CI 4.63 – 89.24) as shown on table
12. This result is consistent with previous findings that showed the importance of glutamine 69
in the development of CBD especially the non-0201 carriage.(60,62,63) We did not find a
significant effect of other demographic, genetics, or exposure variables with the development of
CBD. When adjusted for log cumulative exposure in the model, the OR for glutamine 69 was
27.52 (95% CI 6.56-115.35).
Univariable analysis of BeS and their controls (41 BeS and 104 controls) found that in
addition to HLA-DPB1glu69, glutamine 71 also showed a significant association (respectively
54

OR for glutamine 69 is 6.32 with 95% CI 2.54-15.71; OR for glutamine 71 2.72 with 95% CI
1.28-5.77). In multivariable regression analyses adjusted for log cumulative exposure, glutamine
69 and glutamine 71 remain the two significant predictors with Odds Ratios of 7.08 (95% CI
2.59-19.35) and 2.54 (95% CI 1.06-6.12), respectively. A previous study has also linked
glutamine 69 as a significant predictor for BeS (15,16,59) but to the best of our knowledge
glutamine 71 significance was only recently reported from the previous study of this same study
population from two facilities in eastern Pennsylvania.(63)
When considering allele type, we found that individuals with non-0201 alleles had higher
OR compared to those with 0201 alleles (OR 24.45; 95% CI 5.22-114.45 and OR 4.79; 95% CI
1.57-14.64 respectively). Contrasting non-0201 carriage with 0201 carriage, we found a
significant difference on OR (p=0.0170) where individuals who carry non-0201 alleles had
higher risk compared to those with 0201 alleles (OR4.71; 95% CI = 1.32-16.77). This result
corroborates Van Dyke et al who also reported a higher odds ratio for non-0201 carriers
compared to 0201-carriers.(15,16)
We did not observe a dose-response association and genetic-exposure interaction in the
development of CBD and BeS such as that which was reported by Van Dyke et al.(15) Our
analyses showed that exposure did not have a significant association with development of
beryllium toxicity on univariable and multivariable models, and did not show a significant
interaction with genetic characteristics on our multivariable model. When we categorized
exposure into quartiles, similar to how analyses were conducted by Van Dyke et al, we still did
not find a significant association (Appendix 8). Further analyses using algorithms to define
individuals who possibly had high cumulative exposure but never had high peak exposure, or

55

between individuals with high peak exposure but who maintained a lower cumulative exposure
found no significant associations (Appendix 9).
We explored whether different exposure measurement categorization methods between
our study and the Van Dyke et al study might explain the differences in results between our
study and Van Dyke et al. Our exposure metrics were based on job personnel records and
collected well before medical examinations were conducted to determine disease status, which
would have minimized recall bias. For this cohort, we calculated and assigned cumulative, mean,
and peak exposure for each individual based on their actual job history from company records
and actual exposure data from workplace industrial hygiene reports where beryllium exposures
in the plant were measured over different time periods. In contrast, Van Dyke assigned exposures
based on personal interview of job history after completion of the medical examination which
might introduce bias and potential exposure misclassification.(15,28) Although it is unlikely that
study methodology greatly influenced these differences, it is important to note that we also used
a different control selection method compared to Van Dyke. We assigned controls through exact
matching based on gender, Plant, and year of birth, while Van Dyke and colleagues assigned
controls through frequency matching based on gender, race, work status, and decade of
hire.(15,16)
Van Dyke et al also reported finding a genetic-exposure interaction in the development of
CBD and BeS.(15) Although we found no significant association of exposure with disease
development and no significant genetic and exposure interactions on our multivariable model,
we did find HLA-DPB1Glu69 has a role in explaining why exposure is higher in controls
compared to cases. Our analysis showed that individuals who were negative for HLADPB1Glu69 had a significantly higher cumulative exposure than controls that carried the HLA56

DPB1Glu69 gene, although this finding was not statistically significant. This result showed that
the absence of this susceptibility gene is protective for beryllium toxicity despite the significantly
higher cumulative exposure.
We further investigated whether non-0201 alleles of HLA-DPB1glu69 influence this
association. Our analysis showed that control individuals with non-0201 alleles had higher
exposures than those with 0201 alleles. This result shows that non-0201 alleles, which are
associated with increased susceptibility for beryllium toxicity, cannot explain why these control
individuals would remain healthy. If non-0201 alleles had an important role, the average
exposure would have been expected to be lower in non-0201 carriers than those with 0201
carriers. We did not find any significant difference of exposure between non-0201 and 0201
carrying individuals among CBD and BeS (Table 18).
Although we did not observe a clear dose-response association in our multivariable
model, our analyses showed that among those positive for glutamine 69 there was a trend
suggesting a dose-response in regards to peak exposure although the trend was not statistically
significant.
Previous reports have suggested the importance of the type of beryllium exposure in the
development of CBD and BeS, as well as progression to CBD in sensitized individuals.(5,11)
We found significant differences in cumulative mixed chemical exposure and cumulative,
mean, as well as peak mixed physical exposure, but the highest exposures were found in
controls, followed by CBD and BeS (Table 8). These analyses however, were conducted among
all individuals regardless of beryllium susceptibility. To have better comparable groups by taking
genetic susceptibility into account, we analyzed the difference of exposure level and exposure
type between CBD, BeS and control groups only in individuals with Glutamine 69 (Table 19).
57

We found no significant difference; however, for some exposures i.e. peak exposure, peak
chemical mix, and peak soluble, we did find that CBD individuals had the highest level of
exposure compared to BeS and controls, which would suggest a dose-response association
(Table 19).
With further analyses using the hockey stick method for coding exposure, we only found
a significant effect of log mixed chemical exposure in the development of BeS but did not find
any significant association for other types of exposure (Table 22). This suggests that for people
who had a detectable level of exposure, an increase in mean mixed chemical exposure will
increase the likelihood of the development of BeS (OR 1.5 95% CI 1.1-2.2; p value = 0.0171).
We did not observe this effect when analyses were run only for susceptible individuals (Table
23).
Our study also aimed to evaluate factors that influenced the development of CBD from
BeS. Using unconditional logistic regression we found that the glutamine 71 gene has a
protective effect on the development of CBD (OR 0.38, 95% CI 0.16-0.087), adjusted for log
cumulative exposure. This result should be interpreted cautiously and further research is still
needed to determine whether the presence of glutamine 71 is truly protective against CBD in
sensitized individuals, and what mechanism is involved.
Previous studies have proposed that duration of exposure or genetic factors, including
homozygosity of HLA-DPB1Glu69 (61), are linked to this progression.(20,21,23,28) In our
study, we found that homozygosity or exposure was not a predictor for disease severity or the
development of CBD (Appendix 7).
There have only been a few studies on the interaction of genetics and exposure on the
development of beryllium toxicity. The strength of our study is the availability of specific
58

measurements of beryllium exposure, which included cumulative, mean, and peak exposure as
well as different types (i.e., chemical, physical) of beryllium exposures. We also assessed several
genes potentially involved in the mechanism of beryllium toxicity, using a relatively large study
population with an exact matched case-control design.
A possible limitation of our study includes potential measurement error of exposure
which could influence our results, as either over- or under-estimating exposures for various jobs
or departments over time. It is unlikely that recall bias of job history would influence our
imputation and assigned exposure measurements, because we used personnel records to assess
job history and these records were accessed independently of case determination (medical
examination). One further potential exposure bias was that our study did not assess the
possibility of exposure pathways beyond airway exposure. In particular, skin exposure was not
able to be assessed for this study.
Another limitation of our study was the potential differences between the two facilities.
Our analyses showed that Plant 2 had significantly higher cumulative, mean, and peak exposure
(Table 6) compared to Plant 1 (p value 0.0094, 0.0005, and <0.0001 respectively). The mean
3

3

exposure in Plant 2 (23.20 µg/m ) was also higher than the permissible limit of 2 µg/m (49). A
significant difference was also seen in the type of exposure, in which Plant 2 had significantly
higher chemical exposure compared to Plant 1 but there was no significant difference in physical
exposure (Table 6). This difference in type of exposure suggests that each Plant had a different
industrial environment. Different lengths of time during which each Plant was in operation as
well as decade of operation between the two facilities might explain this difference. Plant 2 was
open longer and earlier, from 1935-2000, while Plant 1 was open from 1958 to 1978.(63) The
59

longer duration and earlier starting of operations at Plant 2 might contribute to the different
exposure characteristics and exposure levels found between the two facilities. Further, the
recommended level of beryllium exposure for workplaces was not implemented until after cases
of acute beryllium toxicity were reported in late 1950s.(3) Despite these differences, our study
design did match cases with controls within each plant, so any results would presumably control
for confounding due to which plant an individual worked.
Our findings on the importance of genetic susceptibility of HLA-DPB1Glu69 in the
development of CBD and BeS corroborate previous reports. In addition, we also found that
glutamine 71 is important in the development of BeS and decreasing the risk of progressing from
BeS to CBD. Our results also showed that either one or both genes were present in all cases of
BeS and CBD. The significance of glutamine 71 in the development of beryllium toxicity
especially in the absence of glutamine 69, as well as the role of this gene in decreasing the risk of
progressing from CBD to BeS, warrants further research.
Our results also imply that there are other factors contributing to the development of
CBD and BeS as well as progression of CBD from BeS other than the magnitude and type of
exposure, as well as glutamine 69 and glutamine 71 polymorphisms. These factors might
influence the interaction of exposure and genetics, which might explain the lack of a doseresponse effect of beryllium even in susceptible individuals in our cohort. It has been proposed
that it is not only the genetic susceptibility that plays an important role in the development of
beryllium toxicity, but also the local environment of the epitopes.(63) Studying the local
environment of additional polymorphisms might be important to understand exposure-genetic
interaction in the development of beryllium toxicity.

60

CHAPTER VI
CONCLUSION

From our study we found that HLA-DPB1Glu69 increased the risk of Chronic Beryllium
Disease (adjusted OR 27.52; 95% CI 6.56-115.35). Individuals with HLA-DPB1Glu69 non-0201
alleles, had a higher OR compared to those with 0201 alleles (adjusted OR 36.09 95% CI 8.15159.95 and 20.56 95% CI 4.66-90.69 respectively). HLA-DPB1Glu69 was also significantly
associated with development of Beryllium Sensitization (adjusted OR 7.08; 95% CI 2.59-19.35),
and among these beryllium sensitized workers, the individuals with non-0201 alleles also had
higher ORs compared to 0201 alleles (adjusted OR 24.45; 95% CI 5.22-114.45 and 4.80; 95% CI
1.57-14.64 respectively).
In addition to Glutamine 69, Glutamine 71 also significantly increased the risk of
development of Beryllium Sensitization (adjusted OR 2.54; 95% CI 1.06-6.12). However,
Glutamine 71 was protective for CBD among BeS and CBD subjects (adjusted OR 0.38, 95% CI
0.16-0.87). Further study is needed to examine the role of Glutamine 71 in the development of
CBD and BeS and in the progression to CBD from BeS.
Although our results show no clear dose-response association between the magnitude and
type of exposure and beryllium toxicity, we found that the control individuals with the highest
exposure are those who do not have the HLA-DPB1Glu69 polymorphism and presumably
remain healthy because they are not genetically susceptible. Further work to explore other
polymorphisms for an exposure genetic interaction is needed to determine if controlling for these
additional polymorphisms will elucidate a dose-response.

61

APPENDICES

62

Appendix 1. Distribution of Cumulative Exposure and Log Cumulative Exposure
Distribution of Cumulative Exposure
Figure 4. Distribution of Cumulative Exposure
Distribution of Cumulative Exposure
500

400

300

Percent

a.

200

100

0
0

4000

8000

12000

16000

Cumulative Exposure
Curves – Normal (Mu=599.55 Sigma=1850) – Kernel (c=0.79)
63

20000

Distribution of Log Cumulative Exposure
Figure 5. Distribution of Log Cumulative Exposure
Distribution of Log Cumulative Exposure
25

20

Percent

b.

15

10

5

0
-4.8

-2.4 -2.4

0.0

2.4

4.8

7.2

9.6

Log Cumulative Exposure
Curves – Normal (Mu=4.3599 Sigma=2.1547 – Kernel (c=0.79)

64

12.0

Appendix 2. Wilcoxon Two Sample Test for Difference of Exposure in Plant 1
Table 26. Wilcoxon Two Sample Test for Difference of Exposure in Plant 1
Average Level
P Value
Exposure*
CBD
BeS
Control CBD vs. CBD vs.
BeS vs.
N=28
N=23
N=133
BeS
Control Control
Duration
In Years
8.73
5.22
9.20
0.3203
0.4886
0.0202
Total Exposure
Cumulative
118.72
77.66
162.39
0.2764
0.3066
0.0256
Log Cumulative
3.86
3.25
4.20
0.2764
0.3066
0.0256
Mean
1.42
1.53
1.69
0.7548
0.3872
0.1910
Peak
2.63
2.90
3.71
0.9773
0.2826
0.0407
Type of Exposures
Chemical Mix
Cumulative
20.93
9.51
50.20
0.5689
0.0663
0.0030**
Mean
0.28
0.37
0.48
0.9059
0.0698
0.0567
Peak
1.09
1.09
1.68
0.8987
0.2270
0.0127**
Physical Mix
Cumulative
20.59
7.65
45.72
0.6313
0.0531
0.0024**
Mean
0.25
0.32
0.46
0.9652
0.0451
0.0327
Peak
0.99
1.09
1.64
0.9315
0.1305
0.0181
3

*) Total exposures and different type of exposures were each measured in µg-year/m unit
3

3

for cumulative exposure, in µg/m for mean exposure, and in µg/m for peak exposure
Comparison was obtained using Kruskal Wallis non parametric test
**) with Bonferroni correction for multiple comparison in Wilcoxon two sample test,
difference of mean is significant at < 0.0167 level

65

Appendix 3. Wilcoxon Two Sample Test for Difference of Exposure in All Subjects
Table 27. Wilcoxon Two Sample Test for Difference of Exposure in All Subjects
Exposure Level
P Value
CBD
BeS
Control
CBD
CBD
BeS vs.
Exposure*
N=61
N=41
N=259 vs. BeS
vs.
Control
Control
Duration (Years)
8.89
5.49
12.19
0.1017 0.0922 0.0009**
Cumulative
291.40
145.54
743.99 0.2932 0.1636
0.0187
Log cumexp
4.13
3.60
4.53
0.2932 0.1636
0.0187
Chemical Mix
Cumulative
199.76
69.54
547.65 0.2904 0.1111
0.0375
Physical mix
Cumulative
49.81
31.29
198.92 0.3678 0.0193 0.0020**
Mean
0.97
2.79
1.84
0.7452 0.1188
0.0209
Peak
4.77
5.65
8.06
0.6463 0.0479 0.0071**
3

*) Total exposures and different type of exposures were each measured in µg-year/m unit for
3

3

cumulative exposure, in µg/m for mean exposure, and in µg/m for peak exposure
Comparison was obtained using Kruskal Wallis non parametric test
**) with Bonferroni correction for multiple comparison in Wilcoxon two sample test,
difference of mean is significant at < 0.0167 level

66

Appendix 4. Proportion of Glutamine 69 and Glutamine 71 Positive Individuals by Disease
State and the Importance of Glutamine 71 in the Absence of Glutamine 69
a.

Proportion of Glutamine 69 and Glutamine 71 Positive Individuals by Disease State
Table 28. Proportion of Glutamine 69 and Glutamine 71 Positive Individuals by Disease
State
Outcome (N, %)
Glutamine 69 and Glutamine 71
Negative
Either
Both
Total
CBD
0 (0)
50 (81.97) 11 (18.03)
61
BeS
0 (0)
30 (73.17) 11 (26.83)
41
Control
128 (49.4) 101 (39.0)
30 (11.6)
259
Total
128 (35.5_ 181 (50.2)
52 (14.3)
361

b.

The Importance of Glutamine 71 in the Absence of Glutamine 69

Testing the hypothesis that Glutamine 71 influences the development of CBD and BeS in the
absence of Glutamine 69, the convergence is not reached for CBD and control and CBD and
BeS.
For BeS and Control
Table 29. Effect of Glutamine 71 in the Absence of Glutamine 69
Parameter
DF Estimate Standard
Wald Pr > ChiSq
Error Chi-Square
Glutamine 71
1 20.5517
10261.2
0.0000
0.9984

67

Appendix 5. Univariable Conditional Logistic Regression for Development of CBD
Table 30. Univariable Conditional Logistic Regression for Development of CBD
95%CI
Wald
Variable
LR test Score T
OR
test
L
U
Smoking (Current vs. never)
1.022
0.395
2.645
Ex vs. never
0.9794
0.9792
0.9792
0.966
0.484
1.927
Unknown vs. never
0.834
0.198
3.525
Race (Black vs. white)
<0.001
<0.001
>999.9
0.2813
0.4733
0.9999
No answer vs. white
<0.001
<0.001
>999.9
Age
0.4609
0.4933
0.4974
1.086
0.884
1.334
Glu69
<0.001 <0.0001 <0.0001 27.030
6.491 112.561
Ser13
0.4203
0.4224
0.4233
1.180
0.647
2.151
Tyr26
0.4424
0.4462
0.4494
0.705
0.320
1.550
His32
0.4781
0.4774
0.4780
1.171
0.652
2.105
Arg74
0.6760
0.6782
0.6764
0.803
0.362
1.783
Ser11
0.1628
0.1689
0.1718
1.433
0.750
2.738
Phe47
0.9857
0.9857
0.9857
0.915
0.461
1.816
Asp37
0.4938
0.4965
0.4971
0.760
0.404
1.427
Glu71
0.5515
0.5477
0.5483
1.281
0.663
2.476
Homozygosity (heterozygous)
26.519
6.325 111.193
<0.0001 <0.0001 <0.0001
Homozygous
30.267
5.809 157.687
Peak exposure
0.1655
0.3604
0.3118
0.994
0.983
1.005
Cum exposure
0.0377
0.0947
0.1083
1.000
0.999
1.000
Log CEMEX
0.3299
0.3265
0.3286
0.930
0.804
1.076
Mean Exposure
0.9514
0.9512
0.9513
1.000
0.983
1.018
Cum Chemical mix
0.0581
0.1455
0.1385
1.000
0.999
1.000
Mean chemical mix
0.8789
0.8796
0.8796
0.999
0.982
1.016
Peak Chemical mix
0.1609
0.3650
0.2983
0.994
0.982
1.005
Cum Chemical N
0.4356
0.4739
0.4944
1.000
0.998
1.001
Mean Chemical NS
0.6883
0.6814
0.6851
1.019
0.930
1.118
Peak Chemical NS
0.4544
0.4617
0.4738
0.987
0.954
1.021
Cum Chemical sol
0.0603
0.2023
0.3101
0.996
0.988
1.004
Mean Chemical sol
0.2428
0.3049
0.3481
0.766
0.440
1.336
Peak Chemical sol
0.2443
0.2873
0.3260
0.989
0.968
1.011
Cum physical mix
0.0074
0.1070
0.0562
0.998
0.996
1.000
Mean physical mix
0.2550
0.3256
0.3815
0.973
0.916
1.034
Peak physical mix
0.1285
0.1493
0.1641
0.987
0.969
1.005
Cum physical dust
0.3337
0.3660
0.3822
1.000
0.999
1.000
Mean physical dust
0.8061
0.8078
0.8081
0.997
0.973
1.022
Peak physical dust
0.0505
0.3196
0.1421
0.987
0.969
1.004
Cum physical fume
0.2948
0.4224
0.4581
1.000
0.999
1.000
Mean physical fume
0.3687
0.3338
0.3742
1.012
0.985
1.040
Peak physical fume
0.6675
0.6712
0.6721
0.997
0.984
1.011
68

Appendix 6. Univariable Conditional Logistic Regression for Development of BeS
Table 31. Univariable Conditional Logistic Regression for Development of BeS
95%CI
Score
Wald
Variable
LR test
OR
test
test
L
U
Smoking (Current vs. never)
0.698
0.213 2.282
Ex vs. never
0.3305
0.3806
0.4237
0.626
0.270 1.450
Unknown vs. never
0.204
0.024 1.711
Race (Black vs. white)
2.000
0.125 31.97
0.2222
0.1969
0.8868
No answer vs. white
1.060
0.926 1.213
Age
0.3161
0.3678
0.3994
Glu69
<0.0001
<0.0001
<0.0001
6.315
2.539 15.71
Ser13
0.5260
0.5290
0.5297
1.292
0.581 2.872
Tyr26
0.5764
0.5831
0.5842
0.746
0.260 2.134
His32
0.8622
0.8623
0.8623
0.942
0.477 1.860
Arg74
0.5764
0.5831
0.5842
0.746
0.260 2.134
Ser11
0.3651
0.3704
0.3722
1.490
0.620 3.579
Phe47
0.9443
0.9443
0.9443
1.030
0.452 2.346
Asp37
0.3652
0.3631
0.3645
1.387
0.684 2.811
Glu71
0.0316
0.0297
0.0331
2.231
1.067 4.666
Homozygous (heterozygous)
5.614
2.153 14.63
<0.0001
<0.0001
<0.0001
Homozygous
8.987
2.600 31.06
Peak exposure
0.2033
0.3175
0.3810
0.993
0.978 1.008
Cum exposure
0.0070
0.0766
0.1068
0.999
0.997 1.000
Log Cum exposure
0.0081
0.0086
0.0118
0.756
0.608 0.940
Mean Exposure
0.6608
0.6560
0.6577
1.006
0.980 1.033
Cum Chemical mix
0.0251
0.1198
0.2658
0.999
0.997 1.001
Mean chemical mix
0.4825
0.4782
0.4847
1.009
0.983 1.036
Peak Chemical mix
0.9969
0.9969
0.9699
0.994
0.979 1.008
Cum Chemical NS
0.1577
0.2529
0.2493
0.999
0.996 1.001
Mean Chemical NS
0.8734
0.8759
0.8762
0.992
0.891 1.104
Peak Chemical NS
0.6766
0.6818
0.6835
0.951
0.871 1.037
Cum Chemical sol
0.0510
0.2907
0.2621
0.985
0.960 1.011
Mean Chemical sol
0.5014
0.5862
0.6954
0.961
0.787 1.173
Peak Chemical sol
0.0145
0.2530
0.1379
0.949
0.730 1.233
Cum physical mix
0.0623
0.2267
0.2750
0.998
0.993 1.002
Mean physical mix
0.2184
0.2094
0.3098
1.034
0.969 1.103
Peak physical mix
0.2140
0.3235
0.3854
1.000
0.979 1.022
Cum physical dust
0.1513
0.2147
0.2570
0.999
0.998 1.001
Mean physical dust
0.8720
0.8720
0.8721
1.002
0.977 1.028
Peak physical dust
0.1749
0.2792
0.2551
0.996
0.978 1.015
Cum physical fume
0.0139
0.1502
0.2490
0.994
0.984 1.004
Mean physical fume
0.0633
0.1473
0.2446
0.801
0.552 1.164
Peak physical fume
0.4268
0.5432
0.6931
0.945
0.877 1.018
69

Appendix 7. Univariable Unconditional Logistic Regression for Progression of BeS to CBD
Table 32. Univariable Unconditional Logistic Regression Comparing CBD and BeS
Variable
LR test
Score test
Wald test
Smoking status
0.7535
0.7606
0.7662
Race
0.1615
0.2274
0.9997
Age
0.7632
0.7632
0.7632
Glu69
0.0562
0.0542
0.0625
Ser13
0.2185
0.2220
0.2242
Tyr26
0.5216
0.5257
0.5272
His32
0.5663
0.5662
0.5664
Arg74
0.5216
0.5257
0.5272
Ser11
0.4568
0.4597
0.4607
Phe47
0.8544
0.8542
0.8542
Asp37
0.0618
0.0612
0.0632
Glu71yn
0.0257
0.0253
0.0270
Homozygous
0.0944
0.0922
0.1043
Peak exposure
0.5646
0.5696
0.5720
Cumulative exposure
0.1573
0.2038
0.2524
Mean exposure
0.8658
0.8664
0.8665
Cumulative chemical mix
0.1263
0.1987
0.2558
Mean Chemical mix
0.9942
0.9942
0.9942
Peak chemical mix
0.8231
0.8222
0.8225
Cumulative chemical non soluble
0.7046
0.7091
0.7114
Mean Chemical non soluble
0.9734
0.9735
0.9736
Peak chemical non soluble
0.7707
0.7696
0.7700
Cumulative chemical soluble
0.7261
0.7339
0.7396
Mean Chemical soluble
0.4355
0.4311
0.4410
Peak chemical soluble
0.0761
0.1351
0.2550
Cumulative physical mix
0.3953
0.4138
0.4281
Mean physical mix
0.2846
0.2929
0.3664
Peak physical mix
0.6679
0.6709
0.6721
Cumulative physical dust
0.4720
0.4902
0.5010
Mean physical dust
0.5921
0.5874
0.5900
Peak physical dust
0.7404
0.7468
0.7510
Cumulative physical fume
0.0716
0.1558
0.3349
Mean physical fume
0.2128
0.3680
0.5437
Peak physical fume
0.0761
0.1351
0.2550

70

Appendix 8. Development of CBD and BeS by Exposure Quartiles and Genetics
a.

Descriptive Analysis of Log Total Exposure
Exposure
Logcumexp
Logmeanexp
Logpeakexp

b.

Table 33. Descriptive Analysis of Log Total Exposure
Mean 25th Pctl Median 75th Pctl Minimum Maximum
4.36
2.78
4.44
5.76
-2.63
9.92
0.62
-0.23
0.46
1.34
-3.82
4.88
1.39
0.30
1.20
2.08
-3.61
7.57

Multivariable Conditional Logistic Regression Using Quartiles of Log Exposure
Table 34. Multivariable Conditional Logistic Regression Using Quartiles of Log Exposure
Coeffici Standar
95% Confidence
P
Variable
OR
ent
d Error
Interval
Value
Development of CBD
Glutamine 69
3.2800
0.7297
26.58
6.36
111.06
<.0001
logcumexpcat 1 vs. 0
-0.0817
0.5467
0.92
0.32
2.69
0.8813
logcumexpcat 2 vs. 0
0.1355
0.5218
1.15
0.41
3.18
0.7951
logcumexpcat 3 vs. 0
-0.2983
0.5434
0.74
0.26
2.15
0.5830
Development of BeS
Glutamine 69
2.0587
0.5369
7.84
2.74
22.44
0.0001
Glutamine 71
0.9504
0.4500
2.59
1.07
6.25
0.0347
logcumexpcat 1 vs. 0
-0.4438
0.7317
0.64
0.15
2.69
0.5442
logcumexpcat 2 vs. 0
-0.1960
0.6002
0.82
0.25
2.67
0.7440
logcumexpcat 3 vs. 0
-1.2120
0.7370
0.30
0.07
1.26
0.1001
Progression of CBD from BeS
Intercept
0.4927
0.4079
0.2270
Glutamine 71
-0.9724
0.4299
0.38
0.16
0.88
0.0237
logcumexpcat 1 vs. 0
0.4325
0.5523
1.54
0.52
4.55
0.4336
logcumexpcat 2 vs. 0
0.1895
0.5441
1.21
0.42
3.51
0.7276
logcumexpcat 3 vs. 0
0.6339
0.6684
1.89
0.51
6.99
0.3429

71

Appendix 9. Algorithm of Level of Exposure (Cumulative and Peak) by Median Value
Table 35. Multivariable Conditional Logistic Regression Using Algorithm of Exposure
Variable
Coeffi Standard
OR
95% Confidence
P
cient
Error
Interval
Value
CBD and Control
Glutamine 69
logcumpeakcat3 1 vs. 0
logcumpeakcat3 2 vs. 0
logcumpeakcat3 3 vs. 0
logcumpeakcat3 4 vs. 0
logcumpeakcat3 5 vs. 0
logcumpeakcat3 6 vs. 0
logcumpeakcat3 7 vs. 0
BeS and Control
Glutamine 69
Glutamine 71
logcumpeakcat3 1 vs. 0
logcumpeakcat3 2 vs. 0
logcumpeakcat3 3 vs. 0
logcumpeakcat3 4 vs. 0
logcumpeakcat3 5 vs. 0
logcumpeakcat3 6 vs. 0
logcumpeakcat3 7 vs. 0
CBD and BeS
Intercept
Glutamine 71
logcumpeakcat3 1 vs. 0
logcumpeakcat3 2 vs. 0
logcumpeakcat3 3 vs. 0
logcumpeakcat3 4 vs. 0
logcumpeakcat3 5 vs. 0
logcumpeakcat3 6 vs. 0
logcumpeakcat3 7 vs. 0

3.5721
2.6282
-0.034
0.0200
0.9753
-0.756
0.4616
-0.509

0.7786
1.7921
0.6435
0.8083
0.6680
0.7425
0.9357
0.6491

35.59
13.85
0.97
1.02
2.65
0.47
1.59
0.60

7.74
0.41
0.27
0.21
0.72
0.11
0.25
0.17

163.71
464.40
3.41
4.97
9.82
2.01
9.93
2.15

<.0001
0.1425
0.9576
0.9802
0.1443
0.3086
0.6218
0.4332

2.2418
1.0497
0.4400
0.3501
-0.843
-0.844
0.3887
-0.686
-1.115

0.5910
0.4754
1.2889
0.9119
0.9878
0.9571
0.7444
1.5123
0.8952

9.41
2.86
1.55
1.42
0.43
0.43
1.48
0.50
0.33

2.96
1.13
0.12
0.24
0.06
0.07
0.34
0.03
0.06

29.97
7.25
19.42
8.48
2.98
2.81
6.35
9.76
1.89

0.0001
0.0272
0.7328
0.7011
0.3934
0.3776
0.6016
0.6501
0.2126

0.5943
-1.131
-0.323
0.4588
0.2377
1.1500
-0.518
0.8427
0.5171

0.4454
0.4572
1.1093
0.6421
0.8355
0.8065
0.6636
1.2555
0.7539

0.32
0.72
1.58
1.27
3.16
0.60
2.32
1.68

0.13
0.08
0.45
0.25
0.65
0.16
0.20
0.38

0.79
6.37
5.57
6.52
15.34
2.19
27.21
7.35

0.1821
0.0134
0.7711
0.4749
0.7761
0.1539
0.4346
0.5021
0.4927

72

REFERENCES

73

REFERENCES

1.

Kolanz ME. Introduction to Beryllium: Uses, Regulatory History, and Disease. Applied
Occupational & Environmental Hygiene. 2001 May;16(5):559–67.

2.

Kreiss K, Day GA, Schuler CR. Beryllium: A Modern Industrial Hazard*. Annual Review
of Public Health. 2007;28(1):259–77.

3.

Hardy HL, Stoeckle JD. Beryllium disease. Journal of Chronic Diseases. 1959 Feb;9(2–
3):152–60.

4.

Newman LS, Lloyd J, Daniloff E. The natural history of beryllium sensitization and chronic
beryllium disease. Environmental Health Perspectives. 1996 Oct;104(Suppl 5):937.

5.

McCanlies EC, Kreiss K, Andrew M, Weston A. HLA-DPB1 and Chronic Beryllium
Disease: A HuGE Review. Am. J. Epidemiol. 2003 Mar 1;157(5):388–98.

6.

Kreiss K, Mroz MM, Newman LS, Martyny J, Zhen B. Machining risk of beryllium disease
and sensitization with median exposures below 2 μg/m3. American Journal of Industrial
Medicine. 1996 Jul 1;30(1):16–25.

7.

Stange AW, Hilmas DE, Furman FJ. Possible health risks from low level exposure to
beryllium. Toxicology. 1996 Jul 17;111(1–3):213–24.

8.

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. Journal of Occupational and Environmental Hygiene.
2007;4(6):448–66.

9.

Mikulski MA, Leonard SA, Sanderson WT, Hartley PG, Sprince NL, Fuortes LJ. Risk of
beryllium sensitization in a low‐exposed former nuclear weapons cohort from the cold war
era. American Journal of Industrial Medicine. 2011 Mar 1;54(3):194–204.

10. Redlich CA, Welch LS. Chronic Beryllium Disease Risk from Low-Level Exposure. Am. J.
Respir. Crit. Care Med. 2008 May 1;177(9):936–7.
11. Deubner DC, Sabey P, Huang W, Fernandez D, Rudd A, Johnson WP, et al. Solubility and
Chemistry of Materials Encountered by Beryllium Mine and Ore Extraction Workers.
Journal of Occupational and Environmental Medicine. 2011 Oct;53(10):1187–93.
12. Tinkle SS, Antonini JM, Rich BA, Roberts JR, Salmen R, DePree K, et al. Skin as a Route
of Exposure and Sensitization in Chronic Beryllium Disease. Environmental Health
Perspectives. 2003 Jul;111(9):1202.

74

13. Henneberger PK, Goe SK, Miller WE, Doney B, Groce DW. Industries in the United States
with Airborne Beryllium Exposure and Estimates of the Number of Current Workers
Potentially Exposed. Journal of Occupational and Environmental Hygiene. 2004;1(10):648–
59.
14. Richeldi L, Kreiss K, Mroz MM, Zhen B, Tartoni P, Saltini C. Interaction of genetic and
exposure factors in the prevalence of berylliosis. American Journal of Industrial Medicine.
1997 Oct 1;32(4):337–40.
15. Van Dyke MV, Martyny JW, Mroz MM, Silveira LJ, Strand M, Cragle DL, et al. Exposure
and genetics increase risk of beryllium sensitisation and chronic beryllium disease in the
nuclear weapons industry. Occup Environ Med. 2011 Nov 1;68(11):842–8.
16. Van Dyke MV, Martyny JW, Mroz MM, Silveira LJ, Strand M, Fingerlin TE, 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 Jun 15;183(12):1680–8.
17. Cullen M, Cherniack M, Kominsky J. Chronic Beryllium Disease in the United States.
Seminars in Respiratory and Critical Care Medicine. 2008 Mar 20;7(03):203–9.
18. Kreiss K, Newman LS, Mroz MM, Campbell PA. Screening blood test identifies subclinical
beryllium disease. J Occup Med. 1989 Jul;31(7):603–8.
19. Occupational Exposure to Beryllium; Request for Information - 67:70707-70712 [Internet].
[cited 2013 Apr 12]. Available from:
http://www.osha.gov/pls/oshaweb/owadisp.show_document?p_table=FEDERAL_REGIST
ER&p_id=17469
20. Kreiss K, Wasserman S, Mroz MM, Newman LS. Beryllium disease screening in the
ceramics industry. Blood lymphocyte test performance and exposure-disease relations. J
Occup Med. 1993 Mar;35(3):267–74.
21. Kreiss K, Mroz MM, Zhen B, Martyny JW, Newman LS. Epidemiology of beryllium
sensitization and disease in nuclear workers. Am. Rev. Respir. Dis. 1993 Oct;148(4 Pt
1):985–91.
22. Kreiss K, Mroz MM, Zhen B, Wiedemann H, Barna B. Risks of Beryllium Disease Related
to Work Processes at a Metal, Alloy, and Oxide Production Plant. Occup Environ Med.
1997 Aug 1;54(8):605–12.
23. 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. International Archives of Occupational and Environmental Health. 2001 Apr
7;74(3):167–76.

75

24. Deubner D, Kelsh M, Shum M, Maier L, Kent M, Lau E. Beryllium Sensitization, Chronic
Beryllium Disease, and Exposures at a Beryllium Mining and Extraction Plant. Applied
Occupational & Environmental Hygiene. 2001 May;16(5):579–92.
25. 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 Mar;43(3):231–7.
26. Sackett HM, Maier LA, Silveira LJ, Mroz MM, Ogden LG, Murphy JR, et al. Beryllium
medical surveillance at a former nuclear weapons Plant during cleanup operations. J.
Occup. Environ. Med. 2004 Sep;46(9):953–61.
27. Schuler CR, Kent MS, Deubner DC, Berakis MT, McCawley M, Henneberger PK, et al.
Process‐related risk of beryllium sensitization and disease in a copper–beryllium alloy
Plant. American Journal of Industrial Medicine. 2005 Mar 1;47(3):195–205.
28. Rosenman K, Hertzberg V, Rice C, Reilly MJ, Aronchick J, Parker JE, et al. Chronic
Beryllium Disease and Sensitization at a Beryllium Processing Plant. Environmental Health
Perspectives. 2005 May 26;113(10):1366–72.
29. Stanton ML, Henneberger PK, Kent MS, Deubner DC, Kreiss K, Schuler CR. Sensitization
and Chronic Beryllium Disease Among Workers in Copper Beryllium Distribution Centers.
Journal of Occupational and Environmental Medicine. 2006 Feb;48(2):204–11.
30. Taiwo OA, Slade MD, Cantley LF, Fiellin MG, Wesdock JC, Bayer FJ, et al. Beryllium
sensitization in aluminum smelter workers. J. Occup. Environ. Med. 2008 Feb;50(2):157–
62.
31. Arjomandi M, Seward J, Gotway MB, Nishimura S, Fulton GP, Thundiyil J, et al. Low
Prevalence of Chronic Beryllium Disease Among Workers at a Nuclear Weapons Research
and Development Plant. Journal of Occupational and Environmental Medicine. 2010
Jun;52(6):647–52.
32. Taiwo OA, Slade MD, Cantley LF, Kirsche SR, Wesdock JC, Cullen MR. Prevalence of
beryllium sensitization among aluminium smelter workers. Occupational Medicine. 2010
Oct;60(7):569–71.
33. Mikulski MA, Sanderson WT, Leonard SA, Lourens S, Field RW, Sprince NL, et al.
Prevalence of Beryllium Sensitization Among Department of Defense Conventional
Munitions Workers at Low Risk for Exposure. Journal of Occupational and Environmental
Medicine. 2011 Mar;53(3):258–65.
34. Safety and Health Information Bulletin - Preventing Adverse Health Effects From Exposure
to Beryllium on the Job [Internet]. [cited 2012 Oct 19]. Available from:
http://www.osha.gov/dts/hib/hib_data/hib19990902.html
76

35. Schubauer-Berigan MK, Couch JR, Petersen MR, Carreón 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 May 1;68(5):345–53.
36. Beryllium and Beryllium Compounds (IARC Summary & Evaluation, Volume 58, 1993)
[Internet]. [cited 2012 Nov 16]. Available from:
http://www.inchem.org/documents/iarc/vol58/mono58-1.html
37. Boffetta P, Fryzek JP, Mandel JS. Occupational exposure to beryllium and cancer risk: A
review of the epidemiologic evidence. Critical Reviews in Toxicology. 2012
Feb;42(2):107–18.
38. Schubauer-Berigan MK, Deddens JA, Couch JR, Petersen MR. Risk of Lung Cancer
Associated with Quantitative Beryllium Exposure Metrics Within an Occupational Cohort.
Occup Environ Med. 2011 May 1;68(5):354–60.
39. Cummings KJ, Deubner DC, Day GA, Henneberger PK, Kitt MM, Kent MS, et al.
Enhanced Preventive Programme at a Beryllium Oxide Ceramics Plant Reduces Beryllium
Sensitisation Among New Workers. Occup Environ Med. 2007 Feb 1;64(2):134–40.
40. Thomas CA, Bailey RL, Kent MS, Deubner DC, Kreiss K, Schuler CR. Efficacy of a
Program to Prevent Beryllium Sensitization Among New Employees at a Copper-Beryllium
Alloy Processing Plant. Public Health Reports. 2009;124(Suppl 1):112.
41. Newman LS, Mroz MM, Balkissoon R, Maier LA. Beryllium Sensitization Progresses to
Chronic Beryllium Disease A Longitudinal Study of Disease Risk. Am. J. Respir. Crit. Care
Med. 2005 Jan 1;171(1):54–60.
42. Duggal M, Deubner D, Curtis A, Cullen M. Long-term follow-up of beryllium sensitized
workers from a single employer. BMC Public Health. 2010 Dec 1;10(1):1–10.
43. Silveira LJ, McCanlies EC, Fingerlin TE, Dyke MVV, Mroz MM, Strand M, et al. Chronic
Beryllium Disease, HLA-DPB1, and the DP Peptide Binding Groove. J Immunol. 2012 Oct
15;189(8):4014–23.
44. Harber P, Bansal S, Balmes J. Progression from Beryllium Exposure to Chronic Beryllium
Disease: An Analytic Model. Environmental Health Perspectives. 2009 Jun;117(6):970–4.
45. Stange AW, Joseph Furman F, Hilmas DE. The beryllium lymphocyte proliferation test:
Relevant issues in beryllium health surveillance. American Journal of Industrial Medicine.
2004 Nov 1;46(5):453–62.
46. Rossman MD. Chronic beryllium disease: diagnosis and management. Environmental
Health Perspectives. 1996 Oct;104(Suppl 5):945.
47. Sood A. Current Treatment of Chronic Beryllium Disease. Journal of Occupational and
Environmental Hygiene. 2009;6(12):762–5.
77

48. GESTIS International Limit Values [Internet]. [cited 2012 Nov 16]. Available from:
http://limitvalue.ifa.dguv.de/WebForm_ueliste.aspx
49. Beryllium (Be) Toxicity: What Are the U.S. Standards for Beryllium Exposure? | ATSDR Environmental Medicine & Environmental Health Education - CSEM [Internet]. [cited
2012 Nov 16]. Available from: http://www.atsdr.cdc.gov/csem/csem.asp?csem=5&po=5
50. OSHA Standard for Beryllium Exposure Inadequate, Groups Charge [Internet]. [cited 2012
Nov 16]. Available from: http://www.chronicberylliumdisease.com/news/nw_091401oshapel.htm
51. Borak J. The Beryllium Occupational Exposure Limit: Historical Origin and Current
Inadequacy. Journal of Occupational and Environmental Medicine. 2006 Feb;48(2):109–16.
52. Environmental Resource Center - Hazardous Waste Training, DOT Hazardous Materials
Training, OSHA Training, Environmental Consulting [Internet]. [cited 2012 Nov 16].
Available from: http://www.ercweb.com/resources/viewtip.aspx?id=6603
53. Maier LA, Martyny JW, Liang J, Rossman MD. Recent Chronic Beryllium Disease in
Residents Surrounding a Beryllium Plant. Am. J. Respir. Crit. Care Med. 2008 May
1;177(9):1012–7.
54. Wambach PF, Laul JC. Beryllium health effects, exposure limits and regulatory
requirements. Journal of Chemical Health and Safety. 2008 Jul;15(4):5–12.
55. Day GA, Dufresne A, Stefaniak AB, Schuler CR, Stanton ML, Miller WE, et al. Exposure
Pathway Assessment at a Copper–Beryllium Alloy Plant. Ann Occup Hyg. 2007 Jan
1;51(1):67–80.
56. Richeldi L, Sorrentino R, Saltini C. HLA-DPB1 Glutamate 69: A Genetic Marker of
Beryllium Disease. Science. 1993 Oct 8;262(5131):242–4.
57. Newman LS. To Be2+ or not to Be2+: immunogenetics and occupational exposure.
Science. 1993 Oct 8;262(5131):197–8.
58. Wang Z, White PS, Petrovic M, Tatum OL, Newman LS, Maier LA, et al. Differential
Susceptibilities to Chronic Beryllium Disease Contributed by Different Glu69 HLA-DPB1
and -DPA1 Alleles. J Immunol. 1999 Aug 1;163(3):1647–53.
59. Wang Z, Farris GM, Newman LS, Shou Y, Maier LA, Smith HN, et al. Beryllium
sensitivity is linked to HLA-DP genotype. Toxicology. 2001 Aug 13;165(1):27–38.
60. McCanlies EC, Ensey JS, Schuler CR, Kreiss K, Weston A. The association between HLA‐
DPB1Glu69 and chronic beryllium disease and beryllium sensitization. American Journal
of Industrial Medicine. 2004 Aug 1;46(2):95–103.

78

61. Maier LA, McGrath DS, Sato H, Lympany P, Welsh K, Du Bois R, et al. Influence of MHC
CLASS II in Susceptibility to Beryllium Sensitization and Chronic Beryllium Disease. J
Immunol. 2003 Dec 15;171(12):6910–8.
62. Saltini C, Richeldi L, Losi M, Amicosante M, Voorter C, Van Den Berg-Loonen E, et al.
Major Histocompatibility Locus Genetic Markers of Beryllium Sensitization and Disease.
Eur Respir J. 2001 Oct 1;18(4):677–84.
63. Rosenman KD, Rossman M, Hertzberg V, Reilly MJ, Rice C, Kanterakis E, et al. HLA
class II DPB1 and DRB1 polymorphisms associated with genetic susceptibility to beryllium
toxicity. Occup Environ Med. 2011 Jul 1;68(7):487–93.
64. 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 Mar 15;165(6):788–94.

79