Aryl Hydrocarbon Hydroxylase in Asbestos Workers1 SYED M. NASEEM, PETER V. TISHLER, HENRY A. ANDERSON, and IRVING J. SELIKOFF
SUMMARY
Aryl hydrocarbon hydroxylase (AHH) activity and inducibility were studied in the lymphoblasts of 47 asbestos workers and 31 control subjects. Mean inducibility (3-methylcholanthrene [MC]- or dibenz(a,h,)anthracene [DBA]-induced divided by basal enzyme activity) was greater in the occupationally exposed subjects than in the control subjects. The estimated change in mean value (± SE) of MC inducibility was 0.88 ± 0.21 (P < 0.001), and that of DBA inducibility was 1.59 ± 0.56 (P < 0.01). Differences in inducibility values were particularly apparent in 12 pairs of spouses, in all of whom the inducibility values of the occupationally exposed subjects exceeded those of their spouses. The differences in inducibility of AHH resulted from increases in induced activity, not from any appreciable change in basal activities. They were independent of subject age, sex, smoking status, year of first exposure to asbestos, yield of lymphoblasts, or the rate of blastogenesis as measured by the incorporation of tritiated thymidine into nucleic acid. Increased AHH inducibility may bear some relation to the greatly increased risk of bronchogenic carcinoma in asbestos workers who smoke cigarettes.
that asbestos workers who also smoked cigarettes had a 90-fold or greater risk of dying of bronchogenic carcinoma than did asbestos workers who did not smoke. Their risk was augmented 8-fold in comparison to that of smokers with no exposure to asbestos. T h e risk for nonsmokers exposed to asbestos appeared to be only slightly, if any, greater than that for unexposed, nonsmoking populations. Clearly, the combina(Received in original form April 10,1978 and in re- tion of occupational exposure to asbestos and cigarette smoking imposes a risk for the developvised form June 27,1978) ment of lung cancer in humans that is among the 1 From the Channing Laboratory, Department of highest of any of the recognized environmental Medicine, Harvard Medical School, and the Peter carcinogens (3). Bent Brigham Hospital Division of the Affiliated The mechanism by which asbestos and cigaHospitals Center, Inc., Boston, Mass., and the En- rettes interact in this apparently multiplicative vironmental Sciences Laboratory, Mount Sinai manner is unknown. A causal role may be played School of Medicine of the City University of New by the carcinogenic aromatic hydrocarbons of York, New York, N.Y. cigarette smoke. Shabad and associates (4) 2 Supported in part by Grant No. ES00928 from showed in rats that benzo(a)pyrene plus asbesthe National Institute for Environmental Health Scitos dust, when administered either by intratraences to the Environmental Sciences Laboratory, cheal instillation or inhalation, will cause preMount Sinai School of Medicine. malignant histologic change or frank malignan3 Requests for reprints should be addressed to Dr. Peter V. Tishler, Channing Laboratory, 180 Long- cy in the bronchopulmonary tree with a far greater frequency than will either agent alone. wood Ave., Boston, Mass. 02115. Introduction
The association of cigarette smoking with increased risk of lung cancer is well known. More striking, however, is the extraordinary augmentation of this risk by occupational exposure to asbestos. In studies both in the greater New York area and throughout the United States and Canada, Selikoff and associates (1, 2) observed
AMERICAN REVIEW OF RESPIRATORY DISEASE, VOLUME 118, 1978
693
694
NASEEM, TISHLER, ANDERSON, AND SELIKOFF
They also demonstrated that asbestos will both readily absorb benzo(a)pyrene in vitro and decrease the excretion of benzo(a)pyrene in laboratory animals (4, 5). Thus, asbestos may serve as a vehicle for the delivery of large quantities of aromatic hydrocarbons in forms that cannot be readily cleared, and these increased effective doses of carcinogen may be responsible for the increased incidence of neoplasm. However, benzo(a)pyrene and other aromatic hydrocarbons in cigarette smoke undergo a variety of metabolic transformations, yielding oxidized metabolites that are even more carcinogenic in experimental systems than the parent compounds (6-9). T h e microsomal enzyme aryl hydrocarbon hydroxylase ( A H H ) , a ubiquitous enzyme that is inducible both in vivo and in cell culture by a variety of hydrocarbons, seems to play an important role in the early metabolism of these aromatic hydrocarbons ( 1 0 12). T h e degree of inducibility of A H H may thus be another factor influencing chemical carcinogenesis, because this may regulate the rate of formation in vivo of highly carcinogenic metabolites from environmental sources including cigarette smoke. Several correlative studies of A H H inducibility and chemical carcinogenesis in animals or spontaneous neoplasia in humans have provided data consistent with such a relationship (13-20). However, definitive evidence for any causal role for A H H in oncogenesis, particularly in humans, is lacking (21-25). W e postulated that significant asbestos exposure may (via inhalation of airborne asbestos particles) lead not only to long-term pulmonary accumulation of aromatic hydrocarbons, but also to increased inducibility of A H H . T o test this hypothesis, we investigated lymphoblast A H H inducibility, an index of A H H inducibility in at least certain cells in the bronchopulmonary tree of subjects without lung cancer (26-29), in asbestos workers. A preliminary report of this study has been presented (30). Materials and Methods Patient population. Forty-seven men exposed to asbestos were selected from the current rosters of both the International Association of Heat and Frost Insulators and Asbestos Workers in New York City and Newark, N. J., and former employees of an asbestos factory in Paterson, N. J. Previous studies have documented the high lung cancer mortality associated with asbestos exposure and cigarette smoking in these groups (1, 2). The subjects' ages ranged from 33 to 78 years. All 47 subjects began their work exposure at least 14 years, and 45 more than 20 years, be-
fore their participation in this study. Eleven subjects were still employed in an asbestos industry at the time of this study. Four subjects had not been employed in this industry for 3 to 8 years before the study, whereas the remainder had left the industry at least 20 years previously. No subject had lung cancer at the time of study, although one subject had had a partial pneumonectomy for this disease 4 years previously. Forty-one subjects had radiographic evidence of asbestosis. Twenty-three subjects were current cigarette smokers, and 2 were former cigarette smokers who smoked a pipe, cigars, or both at the time of study. Sixteen subjects had not smoked for at least 6 months before this study, and 6 had never smoked. The 31 control subjects, 11 men and 20 women, included 20 relatives of the exposed subjects, 12 of whom were spouses, and 11 employees of the health clinic attended by the exposed subjects. Their ages ranged from 20 to 72 years. Eleven control subjects were current smokers, 9 had not smoked for at least 6 months, and 11 had never smoked. Blood samples were obtained from different occupationally exposed and control subjects on each of 6 separate examination dates (table 1), and were processed without knowledge of the asbestos exposure experience of the subject. Relatives serving as control subjects were always sampled on the same date as their occupationally exposed relation. Each subject was studied only once. In addition, multiple blood samples, collected on different dates, were obtained from 6 normal laboratory volunteers (5 men and 1 woman) for the purpose of comparing intra- versus intersubject variance. Lymphocyte culture and enzyme assay. Samples of venous blood (20 ml) were diluted 1:1 with Joklik modified Eagle's medium (F-13 minimal essential medium) containing 20 units of heparin per ml. The diluted blood samples were transported immediately from New York or New Jersey to Boston by air, arriving within 4 hours in nearly all cases. The lymphocytes were isolated immediately by the FicollHypaque gradient technique of Boyum (31). The cells were suspended at 2 to 3 X 105 cells per ml in F-13 minimal essential medium supplemented with 15 per cent fetal calf serum, 100 units of penicillin per ml, 100 ^g of heparin per ml, 1 per cent phytohemagglutinin, and 1 per cent pokeweed mitogen. Four-ml aliquots of cell suspension were incubated at 37° C in 20- by 125-mm round glass culture tubes under a mixture of 5 per cent C 0 2 and 95 per cent air. For the induction of enzyme activity, 3-methylcholanthrene (MC) and dibenz (a,h)anthracene (DBA) were added separately to portions of each culture after 48 hours of incubation to achieve a final concentration of 1.5 pM (32). The remainder of the culture received vehicle (10 yX of acetone) only. Incubation was continued for an additional 24 hours. Cells were counted in a hemocytometer before AHH assay. At least 90 per cent of all cells from all cultures were mononuclear, and 99 per cent ex-
695
ARYL HYDROCARBON HYDROXYLASE
TABLE 1
DESIGN OF STUDY: NUMBER OF CONTROL AND ASBESTOS-EXPOSED SUBJECTS STUDIED AT EACH EXAMINATION Control Subjects Exam
Date
1 2 3 4 5
1/1976 4/1976 7/1976 10/1976 3/1977
6
7/1977
Totals
{no.)
0
5 8 7 6
-
3/2 3/6 4/3 1/5
5
0/5
31
11/20
eluded trypan blue. Preliminary studies demonstrated that lymphocyte culture under these conditions yielded maximal basal and induced activities of AHH. AHH activity in both the uninduced (basal) and induced samples of lymphoblasts was assayed immediately thereafter by the fluorometrk method of Nebert and Gelboin (33), as modified by Gurtoo and co-workers (34). This assay measures the fluorescence of hydroxylated metabolites of the substrate benzo (a)pyrene. Both basal and induced AHH activities were linear with respect to time of incubation (0 to 60 min; the assay was routinely incubated for 30 min) and the number of lymphoblasts (1 to 8 million cells). AHH activity is expressed as pmoles of hydroxylated product generated per min per 106 cells. The inducibility of AHH in any sample was derived by dividing the MC- or DBA-induced enzyme activity by the basal enzyme activity. Incorporation of [3H]thymidine into nucleic acid. This was studied in the blood samples obtained in examination 6 only. [3H] thymidine was added to all tubes 24 hours before termination of the culture to achieve a final concentration of 0.25 juCi per ml. AHH activity was determined in these cultures in the standard manner. To the aqueous phase remaining after the hexane: acetone extraction of the products of the AHH reaction were added 8 ml of cold 2 per cent potassium acetate in 95 per cent ethanol (35). Samples were stored overnight at 4° C and were sonicated briefly to disperse the cell pellet. After centrifugation, the clear supernatant was removed. The pellet was washed twice with 2 per cent potassium acetate in 95 per cent ethanol and once with absolute methanol. The pellet was solubilized by the addition of 0.05 ml of water, 0.45 ml of NCS®, and incubation at room temperature for 24 hours. Aliquots of the solution were counted in 15-ml toluene-based scintillation fluid (36) in a liquid scintillation spectrometer. Because the counting efficiency was essentially invariant (38 per cent), results were expressed as counts per min (cpm) per 106 cells. Materials. Culture medium and all supplements
Exposed Subjects
(male/female)
{no.) 6 9 6 9 9
8 47 (all male)
were purchased from Grand Island Biological Co., Grand Island, N. Y. Two lots of fetal calf serum (no. R745720 and no. A362218), selected from numerous samples because they elicited maximal and similar AHH inducibility ratios in lymphoblasts from laboratory volunteers, were used throughout this study. Only one lot was used in cultures from any individual examination. Ficoll (molecular weight 400,000 daltons), preservative-free heparin, and cofactors were obtained from Sigma Chemical Co., St. Louis, Mo.; 90 per cent Hypaque-M® from Winthrop Laboratory, New York, N Y.; MC and benzo (a)pyrene from Calbiochem, La Jolla, Calif.; DBA from Aldrich Chemical Co., Milwaukee, Wis.; methyl-[3H]thymidine, specific activity 6.7 Ci per mmole, from New England Nuclear Co., Boston, Mass.; and NCS Tissue Solubilizer from Amersham, Arlington Heights, 111. The 3-hydroxybenzo (a)pyrene was a gift from Dr. H. V. Gelboin and the National Cancer Institute Carcinogenesis Chemical Repository. All other reagents were of analytic grade. An Aminco-Bowman 4-8202 spectrophotofluorometer and a Beckman LS-230 liquid scintillation spectrometer were used in the enzyme and radioisotope studies, respectively. Statistical procedures. "Student's" t tests, linear regression analyses, and one-way analyses of variance were carried out according to standard methods (37). For comparison of data derived from the multiple examinations for both control and exposed subjects, the data were first analyzed by one-way analyses of variance to determine their comparability among all examinations. In most instances, the data were found not to be comparable, which precluded the use of analyses of pooled data. In these situations, an estimated difference in mean change (D, the over-all average difference between exposed and control subjects, derived from the weighted mean differences for individual examinations) and its standard error (SE) were derived from data of all examinations by means of the special case of a 2-way analysis of variance applied to an R X 2 table (38). The test statistic, \ , is given by
696
NASEEM, TISHLER, ANDERSON, AND SELIKOFF
D
A.
= SE
>ICO'l;f'MM(,,)~
rv
~~~~~~ :li.~ ,.-,.l!),.-,.- N
tN-K,
'0
where N = total number of subjects, and K = total number of subject groups (exposed and control groups counted separately) in all examinations.
:::l
"0
c
:::l
CD
:>
U CD
~ +'" t'O
E +'" > (") 0:;; I
U
g
~J: 0
..
~
:I:
SUMMARY
Aryl hydrocarbon hydroxylase (AHH) activity and inducibility were studied in the lymphoblasts of 47 asbestos workers and 31 control subjects. Mean inducibility (3-methylcholanthrene [MC]- or dibenz(a,h,)anthracene [DBA]-induced divided by basal enzyme activity) was greater in the occupationally exposed subjects than in the control subjects. The estimated change in mean value (± SE) of MC inducibility was 0.88 ± 0.21 (P < 0.001), and that of DBA inducibility was 1.59 ± 0.56 (P < 0.01). Differences in inducibility values were particularly apparent in 12 pairs of spouses, in all of whom the inducibility values of the occupationally exposed subjects exceeded those of their spouses. The differences in inducibility of AHH resulted from increases in induced activity, not from any appreciable change in basal activities. They were independent of subject age, sex, smoking status, year of first exposure to asbestos, yield of lymphoblasts, or the rate of blastogenesis as measured by the incorporation of tritiated thymidine into nucleic acid. Increased AHH inducibility may bear some relation to the greatly increased risk of bronchogenic carcinoma in asbestos workers who smoke cigarettes.
that asbestos workers who also smoked cigarettes had a 90-fold or greater risk of dying of bronchogenic carcinoma than did asbestos workers who did not smoke. Their risk was augmented 8-fold in comparison to that of smokers with no exposure to asbestos. T h e risk for nonsmokers exposed to asbestos appeared to be only slightly, if any, greater than that for unexposed, nonsmoking populations. Clearly, the combina(Received in original form April 10,1978 and in re- tion of occupational exposure to asbestos and cigarette smoking imposes a risk for the developvised form June 27,1978) ment of lung cancer in humans that is among the 1 From the Channing Laboratory, Department of highest of any of the recognized environmental Medicine, Harvard Medical School, and the Peter carcinogens (3). Bent Brigham Hospital Division of the Affiliated The mechanism by which asbestos and cigaHospitals Center, Inc., Boston, Mass., and the En- rettes interact in this apparently multiplicative vironmental Sciences Laboratory, Mount Sinai manner is unknown. A causal role may be played School of Medicine of the City University of New by the carcinogenic aromatic hydrocarbons of York, New York, N.Y. cigarette smoke. Shabad and associates (4) 2 Supported in part by Grant No. ES00928 from showed in rats that benzo(a)pyrene plus asbesthe National Institute for Environmental Health Scitos dust, when administered either by intratraences to the Environmental Sciences Laboratory, cheal instillation or inhalation, will cause preMount Sinai School of Medicine. malignant histologic change or frank malignan3 Requests for reprints should be addressed to Dr. Peter V. Tishler, Channing Laboratory, 180 Long- cy in the bronchopulmonary tree with a far greater frequency than will either agent alone. wood Ave., Boston, Mass. 02115. Introduction
The association of cigarette smoking with increased risk of lung cancer is well known. More striking, however, is the extraordinary augmentation of this risk by occupational exposure to asbestos. In studies both in the greater New York area and throughout the United States and Canada, Selikoff and associates (1, 2) observed
AMERICAN REVIEW OF RESPIRATORY DISEASE, VOLUME 118, 1978
693
694
NASEEM, TISHLER, ANDERSON, AND SELIKOFF
They also demonstrated that asbestos will both readily absorb benzo(a)pyrene in vitro and decrease the excretion of benzo(a)pyrene in laboratory animals (4, 5). Thus, asbestos may serve as a vehicle for the delivery of large quantities of aromatic hydrocarbons in forms that cannot be readily cleared, and these increased effective doses of carcinogen may be responsible for the increased incidence of neoplasm. However, benzo(a)pyrene and other aromatic hydrocarbons in cigarette smoke undergo a variety of metabolic transformations, yielding oxidized metabolites that are even more carcinogenic in experimental systems than the parent compounds (6-9). T h e microsomal enzyme aryl hydrocarbon hydroxylase ( A H H ) , a ubiquitous enzyme that is inducible both in vivo and in cell culture by a variety of hydrocarbons, seems to play an important role in the early metabolism of these aromatic hydrocarbons ( 1 0 12). T h e degree of inducibility of A H H may thus be another factor influencing chemical carcinogenesis, because this may regulate the rate of formation in vivo of highly carcinogenic metabolites from environmental sources including cigarette smoke. Several correlative studies of A H H inducibility and chemical carcinogenesis in animals or spontaneous neoplasia in humans have provided data consistent with such a relationship (13-20). However, definitive evidence for any causal role for A H H in oncogenesis, particularly in humans, is lacking (21-25). W e postulated that significant asbestos exposure may (via inhalation of airborne asbestos particles) lead not only to long-term pulmonary accumulation of aromatic hydrocarbons, but also to increased inducibility of A H H . T o test this hypothesis, we investigated lymphoblast A H H inducibility, an index of A H H inducibility in at least certain cells in the bronchopulmonary tree of subjects without lung cancer (26-29), in asbestos workers. A preliminary report of this study has been presented (30). Materials and Methods Patient population. Forty-seven men exposed to asbestos were selected from the current rosters of both the International Association of Heat and Frost Insulators and Asbestos Workers in New York City and Newark, N. J., and former employees of an asbestos factory in Paterson, N. J. Previous studies have documented the high lung cancer mortality associated with asbestos exposure and cigarette smoking in these groups (1, 2). The subjects' ages ranged from 33 to 78 years. All 47 subjects began their work exposure at least 14 years, and 45 more than 20 years, be-
fore their participation in this study. Eleven subjects were still employed in an asbestos industry at the time of this study. Four subjects had not been employed in this industry for 3 to 8 years before the study, whereas the remainder had left the industry at least 20 years previously. No subject had lung cancer at the time of study, although one subject had had a partial pneumonectomy for this disease 4 years previously. Forty-one subjects had radiographic evidence of asbestosis. Twenty-three subjects were current cigarette smokers, and 2 were former cigarette smokers who smoked a pipe, cigars, or both at the time of study. Sixteen subjects had not smoked for at least 6 months before this study, and 6 had never smoked. The 31 control subjects, 11 men and 20 women, included 20 relatives of the exposed subjects, 12 of whom were spouses, and 11 employees of the health clinic attended by the exposed subjects. Their ages ranged from 20 to 72 years. Eleven control subjects were current smokers, 9 had not smoked for at least 6 months, and 11 had never smoked. Blood samples were obtained from different occupationally exposed and control subjects on each of 6 separate examination dates (table 1), and were processed without knowledge of the asbestos exposure experience of the subject. Relatives serving as control subjects were always sampled on the same date as their occupationally exposed relation. Each subject was studied only once. In addition, multiple blood samples, collected on different dates, were obtained from 6 normal laboratory volunteers (5 men and 1 woman) for the purpose of comparing intra- versus intersubject variance. Lymphocyte culture and enzyme assay. Samples of venous blood (20 ml) were diluted 1:1 with Joklik modified Eagle's medium (F-13 minimal essential medium) containing 20 units of heparin per ml. The diluted blood samples were transported immediately from New York or New Jersey to Boston by air, arriving within 4 hours in nearly all cases. The lymphocytes were isolated immediately by the FicollHypaque gradient technique of Boyum (31). The cells were suspended at 2 to 3 X 105 cells per ml in F-13 minimal essential medium supplemented with 15 per cent fetal calf serum, 100 units of penicillin per ml, 100 ^g of heparin per ml, 1 per cent phytohemagglutinin, and 1 per cent pokeweed mitogen. Four-ml aliquots of cell suspension were incubated at 37° C in 20- by 125-mm round glass culture tubes under a mixture of 5 per cent C 0 2 and 95 per cent air. For the induction of enzyme activity, 3-methylcholanthrene (MC) and dibenz (a,h)anthracene (DBA) were added separately to portions of each culture after 48 hours of incubation to achieve a final concentration of 1.5 pM (32). The remainder of the culture received vehicle (10 yX of acetone) only. Incubation was continued for an additional 24 hours. Cells were counted in a hemocytometer before AHH assay. At least 90 per cent of all cells from all cultures were mononuclear, and 99 per cent ex-
695
ARYL HYDROCARBON HYDROXYLASE
TABLE 1
DESIGN OF STUDY: NUMBER OF CONTROL AND ASBESTOS-EXPOSED SUBJECTS STUDIED AT EACH EXAMINATION Control Subjects Exam
Date
1 2 3 4 5
1/1976 4/1976 7/1976 10/1976 3/1977
6
7/1977
Totals
{no.)
0
5 8 7 6
-
3/2 3/6 4/3 1/5
5
0/5
31
11/20
eluded trypan blue. Preliminary studies demonstrated that lymphocyte culture under these conditions yielded maximal basal and induced activities of AHH. AHH activity in both the uninduced (basal) and induced samples of lymphoblasts was assayed immediately thereafter by the fluorometrk method of Nebert and Gelboin (33), as modified by Gurtoo and co-workers (34). This assay measures the fluorescence of hydroxylated metabolites of the substrate benzo (a)pyrene. Both basal and induced AHH activities were linear with respect to time of incubation (0 to 60 min; the assay was routinely incubated for 30 min) and the number of lymphoblasts (1 to 8 million cells). AHH activity is expressed as pmoles of hydroxylated product generated per min per 106 cells. The inducibility of AHH in any sample was derived by dividing the MC- or DBA-induced enzyme activity by the basal enzyme activity. Incorporation of [3H]thymidine into nucleic acid. This was studied in the blood samples obtained in examination 6 only. [3H] thymidine was added to all tubes 24 hours before termination of the culture to achieve a final concentration of 0.25 juCi per ml. AHH activity was determined in these cultures in the standard manner. To the aqueous phase remaining after the hexane: acetone extraction of the products of the AHH reaction were added 8 ml of cold 2 per cent potassium acetate in 95 per cent ethanol (35). Samples were stored overnight at 4° C and were sonicated briefly to disperse the cell pellet. After centrifugation, the clear supernatant was removed. The pellet was washed twice with 2 per cent potassium acetate in 95 per cent ethanol and once with absolute methanol. The pellet was solubilized by the addition of 0.05 ml of water, 0.45 ml of NCS®, and incubation at room temperature for 24 hours. Aliquots of the solution were counted in 15-ml toluene-based scintillation fluid (36) in a liquid scintillation spectrometer. Because the counting efficiency was essentially invariant (38 per cent), results were expressed as counts per min (cpm) per 106 cells. Materials. Culture medium and all supplements
Exposed Subjects
(male/female)
{no.) 6 9 6 9 9
8 47 (all male)
were purchased from Grand Island Biological Co., Grand Island, N. Y. Two lots of fetal calf serum (no. R745720 and no. A362218), selected from numerous samples because they elicited maximal and similar AHH inducibility ratios in lymphoblasts from laboratory volunteers, were used throughout this study. Only one lot was used in cultures from any individual examination. Ficoll (molecular weight 400,000 daltons), preservative-free heparin, and cofactors were obtained from Sigma Chemical Co., St. Louis, Mo.; 90 per cent Hypaque-M® from Winthrop Laboratory, New York, N Y.; MC and benzo (a)pyrene from Calbiochem, La Jolla, Calif.; DBA from Aldrich Chemical Co., Milwaukee, Wis.; methyl-[3H]thymidine, specific activity 6.7 Ci per mmole, from New England Nuclear Co., Boston, Mass.; and NCS Tissue Solubilizer from Amersham, Arlington Heights, 111. The 3-hydroxybenzo (a)pyrene was a gift from Dr. H. V. Gelboin and the National Cancer Institute Carcinogenesis Chemical Repository. All other reagents were of analytic grade. An Aminco-Bowman 4-8202 spectrophotofluorometer and a Beckman LS-230 liquid scintillation spectrometer were used in the enzyme and radioisotope studies, respectively. Statistical procedures. "Student's" t tests, linear regression analyses, and one-way analyses of variance were carried out according to standard methods (37). For comparison of data derived from the multiple examinations for both control and exposed subjects, the data were first analyzed by one-way analyses of variance to determine their comparability among all examinations. In most instances, the data were found not to be comparable, which precluded the use of analyses of pooled data. In these situations, an estimated difference in mean change (D, the over-all average difference between exposed and control subjects, derived from the weighted mean differences for individual examinations) and its standard error (SE) were derived from data of all examinations by means of the special case of a 2-way analysis of variance applied to an R X 2 table (38). The test statistic, \ , is given by
696
NASEEM, TISHLER, ANDERSON, AND SELIKOFF
D
A.
= SE
>ICO'l;f'MM(,,)~
rv
~~~~~~ :li.~ ,.-,.l!),.-,.- N
tN-K,
'0
where N = total number of subjects, and K = total number of subject groups (exposed and control groups counted separately) in all examinations.
:::l
"0
c
:::l
CD
:>
U CD
~ +'" t'O
E +'" > (") 0:;; I
U
g
~J: 0
..
~
:I:
