(2)
KOEFFLER HP, HASKELL CM: Rhabdomy-
Vol. 84, No. 7, April 1,1992
be identified; e.g., our small laboratory at a city hospital has established more than 800 human cell lines that have been distributed freely to other investigators. GEORGE E. MOORE
Department of Health and Hospitals Division of Surgical Oncology City and County of Denver 777 Bannock St.
Denver, CO 80204-4507
Reference (/)
CAPUTO JL, THOMPSON A, MCCLINTOCK P,
ET AL: An effective method for establishing human B lymphoblastic cell lines using Epstein-Barr virus. J Tiss Cult Meth 13:3944,1991
Asbestos Exposure Epidemiologic study of exposures to known cancer-causing agents provides a continuing challenge to the research community. Evidence is clear that rates of lung cancer and mesothelioma are increased in workers exposed to asbestos. A recent brief communication in the Journal (J) reported an important casecontrol study of colorectal neoplasia and adenomatous polyps associated with asbestos exposure. Comparing exposures of 410 patients with those of 509 control subjects, Neugut et al. found an elevated risk for either disease for those with a history of significant asbestos exposure. In estimating significant asbestos exposure, the authors excluded two men, whom they believed unlikely to have workplace exposure, from their 12 case subjects. One man was a merchant marine for 22~ years; the" other was a teacher. These two men could have incurred significant workplace exposures. National Cancer Institute maps of cancer rates from 1950 to 1979 (2) reveal that coastal counties where shipbuilding constituted a major occupation had elevated rates of lung cancer. While data on exposures to asbestos among teachers and other institutional workers have proven more difficult to assess, undetected, friable asbestos can occur in older institutions. Adding these two additional case subjects to the others
CORRESPONDENCE 537
Downloaded from http://jnci.oxfordjournals.org/ at Florida Atlantic University on September 8, 2015
of cell lines. In matter of fact, most malignant cell lines, despite their grossly abnormal genomes, become extraor(3) MARGOLIS D, ROSS E, MILLER KB: Rhabdomyolysis associated with high-dose cytardinarily stable within a few weeks or abine. Cancer Treat Rep 71:1325-1326, months and maintain the same charac1987 teristics and generate the same cell (4) SINGHAL PC, VENKATESAN J, GIBBONS N, ET AL: Prevalence and predictors of rhabdoproducts decade after decade. myolysis in patients with hypokalemia [pubThe malignant cells in effusions are lished erratum appears in N Engl J Med 324:707, 1991]. N Engl J Med 323:1488, very resilient and will survive several 1990 days at room temperature and more than (5) KNOCHEL JP, BARCENAS C, COTTON JP, ET 1 week at 4 °C. Collection of effusions AL: Hypophosphatemia and rhabdomyolysis. J Clin Invest 62:1240-1246, 1978 into sterile flasks or plastic containers (6) LULIRI P, BOBBIO-PALLAVICINI E, GORINI M: with 10 U of heparin per milliliter is Acute rhabdomyolysis during treatment with customary because it is difficult to e-aminocaproic acid. Description of two cases. Haematologica 68:664-669, 1983 retrieve cells from clotted protein. It is (7) TOMIYAMA J, HASEGAWA Y, NAGASAWA T, not necessary to save or to ship the enET AL: Bacillus cereus septicemia associated tire volume of fluid. Within 2 hours, the with rhabdomyolysis and myoglobinuric renal failure. Jpn J Med 28:247-250, 1989 desirable cells and cell clumps will set(8) HAANEN C, MUUS P, PENNINGS A: The effect tle, and 80% of the volume can be of cytosine arabinoside upon mitochondrial decanted or aspirated and the cell-rich staining kinetics in human hematopoietic cells. Histochemistry 84:609-613, 1986 remaining fluid transferred to a smaller container with or without added culture media. To avoid gas pressure changes, full or nearly full containers are desirUnique Life Down the Drain able for shipping. or Save the Cells A supplemental benefit is available from the voluntary collection of 5-15 mL of heparized peripheral blood from An informal survey indicates that the donor, which permits the parallel esthere is about one malignant effusion for tablishment of permanent B-lymphoid each 1000 hospital admissions. Of these cell lines for comparative genetic effusions from both peritoneal and studies. The efficiency of such transforpleural cavities, about 30% have sigmations with Epstein-Barr virus is about nificant populations of cancer cells for 95% (/). culture in vitro. In most medical centers, there are cell There are approximately 34 000 000 culture laboratories which would welhospital admissions yearly in the United come malignant effusions to establish States; therefore, perhaps 11 000 biounique cell lines. In my opinion, a minilogically significant malignant effusions mum of 100 cell lines should be estabshould be available for research. Usuallished, characterized, and banked from ly, only a small aliquot of large effueach kind of cancer with representative sions is used for a cytological diagnosis, lines having the most frequent heteroand the remainder are discarded. logous characteristics. Many of these significant effusions In summary, many unique malignant have cancer cells which have adapted to "effusionsrparticularly those "from rare" self-sustaining"growth.'Thus, they~have cancers and those which are very difgreat biological advantages, and 30%ficult to culture from solid tumors, are 50% of them can be established as perbeing wasted. Such effusions would manent cell lines. In contrast, the provide exceptional opportunities for successful establishment of cell lines the establishment of diverse malignant from biopsy specimens ranges from 1% human cell lines. Concurrent normal Bto 10%, except for the exceptionally lymphoid cell lines are useful for comhigh rates of 10%-20% for small-cell parative genetic studies. I suggest that carcinomas of the lung, melanomas the collection and culture of malignant (usually from metastases), and carcinoeffusions and blood from the donor mas of the ovary and colon. patients should be formalized, and Critics, with some justification, note regional and local donor hospitals and the selectivity of the in vitro culture recipient cell culture laboratories should technique and emphasize the instability olysis as a complication of 5-azacytidine. Cancer Treat Rep 62:573-574, 1978
538
Expanded use of techniques that appropriately combine similar studies through meta-analysis will enhance the ability to identify cancer-causing agents and chemo-preventive materials. DEVRA LEE DAVIS
National Academy of Sciences National Research Council 2101 Constitution Ave.,N.W. Washington, DC 20418
References (/)
NEUGUT AI, MURRAY TI, GARBOWSKI GC, ET
AL: Association of asbestos exposure with colorecta! adenomatous polyps and cancer. J Natl Cancer Inst 83:1827-1828, 1991 (2)
PICKLE LW, MASON TJ, HOWARD N, ET AL:
Atlas of U.S. Cancer Mortality Among Whites 1950-80. DDHS Publ No. (NIH)872900. Washington, DC: US Govt Print Off, 1987 (3)
DAVIS DL, MANDULA B, VAN RYZIN J: As-
sessing the power and quality of epidemiologic studies of asbestos-exposed populations. Toxicol Ind Health 1:93-110, 1985 (4)
(5)
(6)
FRUMKIN H, BERLIN J: Asbestos exposure
and gastrointestinal malignancy review and meta-analysis [published erratum appears in Am J Ind Med 14:493, 1988]. Am J Ind Med 14:79-95, 1988 National Toxicology Program, Technical Report 295, November, 1985, Toxicology and Carcinogenesis Studies of Chrysotile Asbestos (CAS NO. 12001-29-5) in F344/ Nrats (feed studies), Research Triangle Park, NC: National Toxicology Program, National Institute of Environmental Health Sciences, 1985 CHALMERS TC, LEVIN H, SACKS HS, ET AL:
Meta-analysis of clinical trials as a scientific discipline. I. Control of bias and comparison with large co-operative trials. Stat Med 6:315-328, 1987 (7)
MORRIS MD, KIMBALL KT, ALDRICH TE, ET
AL: Statistical approach to combining the results of similar experiments, with application to the hematologic effects of extremely-low-frequency electric field exposures. Bioelectromagnetics 10:23-34, 1989 (8)
and colorectal adenomatous polyps. While a large number of individuals, probably the majority of the population, have had some contact with asbestos at some point in their lives, a qualitatively and quantitatively significant exposure is much less common. Since our study used self-reports of asbestos exposure, it was inevitable that nonsignificant exposures would be reported to the interviewers. As a result, we were compelled to review the individual assessment of the exposure in a blinded fashion and to exclude those who our trained reviewers believed were not significantly exposed. These procedures are carefully described in our brief communication. Even with these exclusions, we did find a borderline statistically significant relationship between asbestos exposure and colorectal cancer and, even more interestingly, between asbestos exposure and colorectal adenomatous polyps. As noted by Dr. Davis, a case-control study is a relatively insensitive method to detect these associations because of the low level of exposure in the general population. Other types of studies would have more ability to detect significant relationships. Meta-analyses would be appropriate and have been done for colon cancer. Our study, however, is the first to explore the relationship between asbestos exposure and colorectal adenomatous polyps; thus, meta-analysis would not be useful at this time. We believe that the next step in studying this relationship would be a screening study of asbestos-exposed workers. ALFRED I. NEUGUT* GAIL C. GARBOWSKI
JOHNSON CC, FEINGOLD M, TILLEY B: A
Division of Epidemiology Columbia University School of Public Health New York, N.Y.
meta-analysis of exposure to phenoxy acid herbicides and chlorophenols in relation to risk of soft tissue sarcoma. Int Arch Occup Environ Health 62:513-520, 1990 (9)
TROCK B, LANZA E, GREENWALD P: Dietary
fiber, vegetables, and colon cancer: Critical review and meta-analyses of the epidemiologic evidence. J Natl Cancer Inst 82:650-661.1990
Reference (/)
Response We agree with Dr. Davis that our recent report in the Journal (/) had relatively low statistical power to detect an association between asbestos exposure
NEUGUT AI, MURRAY TI, GARBOWSKI GC, ET
AL: Association of asbestos exposure with colorectal adenomatous polyps and cancer. J Natl Cancer Inst 83:1827-1828, 1991
*Correspondence to: Alfred I. Neugut, M.D., Ph.D.. Division of Epidemiology, Columbia University School of Public Health, 600 W. 168th St.. New York, NY 10032.
Journal of the National Cancer Institute
Downloaded from http://jnci.oxfordjournals.org/ at Florida Atlantic University on September 8, 2015
would further strengthen the association that was detected in this study. Two major types of errors can occur with epidemiologic studies of carcinogens: 1) A study can falsely link exposure to a substance with a form of cancer (commonly known as a type I error, or a false positive); or 2) a study can falsely exonerate a toxicant that, in fact, causes a form of cancer (commonly known as a type II error, or a false negative). Misclassification either of exposure or of disease will tend to bias results toward a type II error. Thus, in this study, excluding two persons who were probably exposed from the total of 12 cases produced an underestimate of the true risk of colorectal neoplasia due to asbestos exposure. The power, or probability, that this study could detect an excess risk is thereby reduced. Properly conducted case-control studies usually involve small numbers of people, which generally limits their power to detect an effect. Further, the expected increase in relative risks for many environmental or occupational hazards is often twofold or less. One remedy to this problem is to pool results of similar studies, which increases the power of detecting an effect. An early attempt at this pooling reviewed all relevant occupational studies of asbestos and gastrointestinal cancer. Of the negative studies, fewer than one half had a power greater than 0.8 to detect a relative risk of 2.5 (3). In the positive studies, there was less than a .01 probability that chance alone would have produced the results. Thus, the recent report in the Journal is consistent with a number of epidemiologic studies that have reported an excess of colorectal neoplasia linked to asbestos exposure (4). These findings are also congruent with animal studies, indicating that lymphatic drift can occur after asbestos exposure (5). Both consistency of the effect across studies and biologic plausibility are thus demonstrated for the association of colorectal neoplasia with asbestos exposure. While originally developed to allow the combination of data from clinical trials (6), meta-analysis has considerable promise for studies of environmental and occupational exposures (7-9).
KOEFFLER HP, HASKELL CM: Rhabdomy-
Vol. 84, No. 7, April 1,1992
be identified; e.g., our small laboratory at a city hospital has established more than 800 human cell lines that have been distributed freely to other investigators. GEORGE E. MOORE
Department of Health and Hospitals Division of Surgical Oncology City and County of Denver 777 Bannock St.
Denver, CO 80204-4507
Reference (/)
CAPUTO JL, THOMPSON A, MCCLINTOCK P,
ET AL: An effective method for establishing human B lymphoblastic cell lines using Epstein-Barr virus. J Tiss Cult Meth 13:3944,1991
Asbestos Exposure Epidemiologic study of exposures to known cancer-causing agents provides a continuing challenge to the research community. Evidence is clear that rates of lung cancer and mesothelioma are increased in workers exposed to asbestos. A recent brief communication in the Journal (J) reported an important casecontrol study of colorectal neoplasia and adenomatous polyps associated with asbestos exposure. Comparing exposures of 410 patients with those of 509 control subjects, Neugut et al. found an elevated risk for either disease for those with a history of significant asbestos exposure. In estimating significant asbestos exposure, the authors excluded two men, whom they believed unlikely to have workplace exposure, from their 12 case subjects. One man was a merchant marine for 22~ years; the" other was a teacher. These two men could have incurred significant workplace exposures. National Cancer Institute maps of cancer rates from 1950 to 1979 (2) reveal that coastal counties where shipbuilding constituted a major occupation had elevated rates of lung cancer. While data on exposures to asbestos among teachers and other institutional workers have proven more difficult to assess, undetected, friable asbestos can occur in older institutions. Adding these two additional case subjects to the others
CORRESPONDENCE 537
Downloaded from http://jnci.oxfordjournals.org/ at Florida Atlantic University on September 8, 2015
of cell lines. In matter of fact, most malignant cell lines, despite their grossly abnormal genomes, become extraor(3) MARGOLIS D, ROSS E, MILLER KB: Rhabdomyolysis associated with high-dose cytardinarily stable within a few weeks or abine. Cancer Treat Rep 71:1325-1326, months and maintain the same charac1987 teristics and generate the same cell (4) SINGHAL PC, VENKATESAN J, GIBBONS N, ET AL: Prevalence and predictors of rhabdoproducts decade after decade. myolysis in patients with hypokalemia [pubThe malignant cells in effusions are lished erratum appears in N Engl J Med 324:707, 1991]. N Engl J Med 323:1488, very resilient and will survive several 1990 days at room temperature and more than (5) KNOCHEL JP, BARCENAS C, COTTON JP, ET 1 week at 4 °C. Collection of effusions AL: Hypophosphatemia and rhabdomyolysis. J Clin Invest 62:1240-1246, 1978 into sterile flasks or plastic containers (6) LULIRI P, BOBBIO-PALLAVICINI E, GORINI M: with 10 U of heparin per milliliter is Acute rhabdomyolysis during treatment with customary because it is difficult to e-aminocaproic acid. Description of two cases. Haematologica 68:664-669, 1983 retrieve cells from clotted protein. It is (7) TOMIYAMA J, HASEGAWA Y, NAGASAWA T, not necessary to save or to ship the enET AL: Bacillus cereus septicemia associated tire volume of fluid. Within 2 hours, the with rhabdomyolysis and myoglobinuric renal failure. Jpn J Med 28:247-250, 1989 desirable cells and cell clumps will set(8) HAANEN C, MUUS P, PENNINGS A: The effect tle, and 80% of the volume can be of cytosine arabinoside upon mitochondrial decanted or aspirated and the cell-rich staining kinetics in human hematopoietic cells. Histochemistry 84:609-613, 1986 remaining fluid transferred to a smaller container with or without added culture media. To avoid gas pressure changes, full or nearly full containers are desirUnique Life Down the Drain able for shipping. or Save the Cells A supplemental benefit is available from the voluntary collection of 5-15 mL of heparized peripheral blood from An informal survey indicates that the donor, which permits the parallel esthere is about one malignant effusion for tablishment of permanent B-lymphoid each 1000 hospital admissions. Of these cell lines for comparative genetic effusions from both peritoneal and studies. The efficiency of such transforpleural cavities, about 30% have sigmations with Epstein-Barr virus is about nificant populations of cancer cells for 95% (/). culture in vitro. In most medical centers, there are cell There are approximately 34 000 000 culture laboratories which would welhospital admissions yearly in the United come malignant effusions to establish States; therefore, perhaps 11 000 biounique cell lines. In my opinion, a minilogically significant malignant effusions mum of 100 cell lines should be estabshould be available for research. Usuallished, characterized, and banked from ly, only a small aliquot of large effueach kind of cancer with representative sions is used for a cytological diagnosis, lines having the most frequent heteroand the remainder are discarded. logous characteristics. Many of these significant effusions In summary, many unique malignant have cancer cells which have adapted to "effusionsrparticularly those "from rare" self-sustaining"growth.'Thus, they~have cancers and those which are very difgreat biological advantages, and 30%ficult to culture from solid tumors, are 50% of them can be established as perbeing wasted. Such effusions would manent cell lines. In contrast, the provide exceptional opportunities for successful establishment of cell lines the establishment of diverse malignant from biopsy specimens ranges from 1% human cell lines. Concurrent normal Bto 10%, except for the exceptionally lymphoid cell lines are useful for comhigh rates of 10%-20% for small-cell parative genetic studies. I suggest that carcinomas of the lung, melanomas the collection and culture of malignant (usually from metastases), and carcinoeffusions and blood from the donor mas of the ovary and colon. patients should be formalized, and Critics, with some justification, note regional and local donor hospitals and the selectivity of the in vitro culture recipient cell culture laboratories should technique and emphasize the instability olysis as a complication of 5-azacytidine. Cancer Treat Rep 62:573-574, 1978
538
Expanded use of techniques that appropriately combine similar studies through meta-analysis will enhance the ability to identify cancer-causing agents and chemo-preventive materials. DEVRA LEE DAVIS
National Academy of Sciences National Research Council 2101 Constitution Ave.,N.W. Washington, DC 20418
References (/)
NEUGUT AI, MURRAY TI, GARBOWSKI GC, ET
AL: Association of asbestos exposure with colorecta! adenomatous polyps and cancer. J Natl Cancer Inst 83:1827-1828, 1991 (2)
PICKLE LW, MASON TJ, HOWARD N, ET AL:
Atlas of U.S. Cancer Mortality Among Whites 1950-80. DDHS Publ No. (NIH)872900. Washington, DC: US Govt Print Off, 1987 (3)
DAVIS DL, MANDULA B, VAN RYZIN J: As-
sessing the power and quality of epidemiologic studies of asbestos-exposed populations. Toxicol Ind Health 1:93-110, 1985 (4)
(5)
(6)
FRUMKIN H, BERLIN J: Asbestos exposure
and gastrointestinal malignancy review and meta-analysis [published erratum appears in Am J Ind Med 14:493, 1988]. Am J Ind Med 14:79-95, 1988 National Toxicology Program, Technical Report 295, November, 1985, Toxicology and Carcinogenesis Studies of Chrysotile Asbestos (CAS NO. 12001-29-5) in F344/ Nrats (feed studies), Research Triangle Park, NC: National Toxicology Program, National Institute of Environmental Health Sciences, 1985 CHALMERS TC, LEVIN H, SACKS HS, ET AL:
Meta-analysis of clinical trials as a scientific discipline. I. Control of bias and comparison with large co-operative trials. Stat Med 6:315-328, 1987 (7)
MORRIS MD, KIMBALL KT, ALDRICH TE, ET
AL: Statistical approach to combining the results of similar experiments, with application to the hematologic effects of extremely-low-frequency electric field exposures. Bioelectromagnetics 10:23-34, 1989 (8)
and colorectal adenomatous polyps. While a large number of individuals, probably the majority of the population, have had some contact with asbestos at some point in their lives, a qualitatively and quantitatively significant exposure is much less common. Since our study used self-reports of asbestos exposure, it was inevitable that nonsignificant exposures would be reported to the interviewers. As a result, we were compelled to review the individual assessment of the exposure in a blinded fashion and to exclude those who our trained reviewers believed were not significantly exposed. These procedures are carefully described in our brief communication. Even with these exclusions, we did find a borderline statistically significant relationship between asbestos exposure and colorectal cancer and, even more interestingly, between asbestos exposure and colorectal adenomatous polyps. As noted by Dr. Davis, a case-control study is a relatively insensitive method to detect these associations because of the low level of exposure in the general population. Other types of studies would have more ability to detect significant relationships. Meta-analyses would be appropriate and have been done for colon cancer. Our study, however, is the first to explore the relationship between asbestos exposure and colorectal adenomatous polyps; thus, meta-analysis would not be useful at this time. We believe that the next step in studying this relationship would be a screening study of asbestos-exposed workers. ALFRED I. NEUGUT* GAIL C. GARBOWSKI
JOHNSON CC, FEINGOLD M, TILLEY B: A
Division of Epidemiology Columbia University School of Public Health New York, N.Y.
meta-analysis of exposure to phenoxy acid herbicides and chlorophenols in relation to risk of soft tissue sarcoma. Int Arch Occup Environ Health 62:513-520, 1990 (9)
TROCK B, LANZA E, GREENWALD P: Dietary
fiber, vegetables, and colon cancer: Critical review and meta-analyses of the epidemiologic evidence. J Natl Cancer Inst 82:650-661.1990
Reference (/)
Response We agree with Dr. Davis that our recent report in the Journal (/) had relatively low statistical power to detect an association between asbestos exposure
NEUGUT AI, MURRAY TI, GARBOWSKI GC, ET
AL: Association of asbestos exposure with colorectal adenomatous polyps and cancer. J Natl Cancer Inst 83:1827-1828, 1991
*Correspondence to: Alfred I. Neugut, M.D., Ph.D.. Division of Epidemiology, Columbia University School of Public Health, 600 W. 168th St.. New York, NY 10032.
Journal of the National Cancer Institute
Downloaded from http://jnci.oxfordjournals.org/ at Florida Atlantic University on September 8, 2015
would further strengthen the association that was detected in this study. Two major types of errors can occur with epidemiologic studies of carcinogens: 1) A study can falsely link exposure to a substance with a form of cancer (commonly known as a type I error, or a false positive); or 2) a study can falsely exonerate a toxicant that, in fact, causes a form of cancer (commonly known as a type II error, or a false negative). Misclassification either of exposure or of disease will tend to bias results toward a type II error. Thus, in this study, excluding two persons who were probably exposed from the total of 12 cases produced an underestimate of the true risk of colorectal neoplasia due to asbestos exposure. The power, or probability, that this study could detect an excess risk is thereby reduced. Properly conducted case-control studies usually involve small numbers of people, which generally limits their power to detect an effect. Further, the expected increase in relative risks for many environmental or occupational hazards is often twofold or less. One remedy to this problem is to pool results of similar studies, which increases the power of detecting an effect. An early attempt at this pooling reviewed all relevant occupational studies of asbestos and gastrointestinal cancer. Of the negative studies, fewer than one half had a power greater than 0.8 to detect a relative risk of 2.5 (3). In the positive studies, there was less than a .01 probability that chance alone would have produced the results. Thus, the recent report in the Journal is consistent with a number of epidemiologic studies that have reported an excess of colorectal neoplasia linked to asbestos exposure (4). These findings are also congruent with animal studies, indicating that lymphatic drift can occur after asbestos exposure (5). Both consistency of the effect across studies and biologic plausibility are thus demonstrated for the association of colorectal neoplasia with asbestos exposure. While originally developed to allow the combination of data from clinical trials (6), meta-analysis has considerable promise for studies of environmental and occupational exposures (7-9).
