PARTVIII. METHODOLOGIES FOR RISK ASSESSMENT
METHODOLOGIES FOR RISK ASSESSMENT IN OCCUPATIONAL CARCINOGENESIS Umberto Saffiotti Division of Cancer Cause & Prevention National Cancer Institute BetFesda, Maryland 20014
The review of the present state of our knowledge of the occupational causation of cancers-accomplished by the conference on which this volume is based -brings us to the final crucial question: How can we assess cancer risks and what can we do about them? There are two aspects to this problem. One concerns those risks that have already been contracted in the past by a population, and that can be directly measured by the institution of appropriate epidemiologic studies, provided that the exposure conditions, the size of the population, and the time of observation are sufficient for meaningful statistical comparisons. The other aspect concerns the assessment of those risks that are still in the process of being acquired or that are potential, such as those risks which accrue to a population of industrial workers when a new chemical process is introduced. For the latter category of risks, methods of epidemiologic observation cannot be used, short of setting up “human experiments” which would give their results in decades and, if positive, would have already inflicted irreversible damage on the exposed people by the time it is detected. For the assessment of these potential risks, animal experimentation provides an extremely valuable means of detecting the carcinogenic activity of many chemicals. The development of a battery of rapid prescreening methods, using criteria of chemical structure, interaction with cellular macromolecules, mutagenesis, and neoplastic transformation of cells in culture, will provide-after their validation in the near future-a much-needed additional short-term methodology for an early warning against potential risks of new products, applicable before the exposure of workers in production or utilization plants. It might be worth emphasizing that the detection of carcinogenicity by animal bioassays is severely limited by the sensitivity of such bioassay methods, which seldom can detect the induction of tumor incidences below 10%. Species variations in the susceptibility to carcinogens are well known and imply that a chemical may be carcinogenic for man and yet not be detected by conventional experimental tests. Conversely, there may be chemicals that are carcinogenic in a rodent species and not in humans; but in this case, we have no way of establishing this exclusion with any certainty, especially in view of the extreme genetic and environmental variability of the individuals in a human population. Therefore, at the present time, qualitative evidence that a chemical is carcinogenic in animals is sufficient to assume the likelihood that it be carcinogenic also in humans. The next logical step is that exposures to all carcinogens should be avoided or, if avoidance is not possible, minimized. Quantitative extrapolations of risk from animals to man have been proposed using the results of animal experiments together with mathematical models of dose-response extrapolation and the estimation of species-conversion factors.l*
393
394
Annals New York Academy of Sciences
One must, however, be extremely cautious of the many pitfalls of such complex exercises, which are based on several arbitrary selections of models and on a biological variability that is hard to predict. Risk assessment is often part of a necessary societal evaluation of the “function” in our society of an industrial process involving exposure to carcinogens. When the use of unproven methods of extrapolation to alleged “safe” or “acceptable” levels based on limited animal tests is proposed for regulatory action, we must first ask these questions: Have all other approaches to the reduction of risk been analyzed and, if possible, implemented? What evidence is there of the need to expose people to such carcinogens? What are the technological alternatives? In many cases the problem is discussed in terms of theoretical extrapolation to parts per trillion levels for a hypothetical population when the problem is the continued exposure of real people to massive and uncontrolled amounts in the workplace. A good example is provided by the current situation concerning the exposure to benzidine in the United States where, following the promulgation of a temporary standard by the Occupational Safety and Health Administration (Department of Labor) on May 3, 1973, it appears that the few major manufacturers have stopped production of this well-recognized occupational carcinogen, and a large number of plants that use benzidine in the production of dyes are now making their own benzidine and then reacting it to make the dyes. No official information exists on the exact number and location of these plants (but they were estimated to be over 1,000), since no registration is required. The Environmental Protection Agency hastened to publish a proposed standard to control the emission of benzidine in effluent waters from these plants, based on a hypothetical risk-extrapolation formula. The crucial issue is whether benzidine has any real social utility in our society. Industries of several countries have stopped its manufacture altogether, as in Great Britain and Italy. Technological alternatives for making dyes with similar color shades have been developed, but they are estimated to be somewhat more expensive. How does our society assess the value of certain shades of colors in textiles versus the proven risk of cancer for benzidine workers and the possible risk for the general population? What, then, are the methodologies for risk assessment? I propose the following steps: (a) A registration of all uses of products known to be carcinogenic by conclusive evidence obtained either from human observations or from animal bioassays. (b) An estimate of exposure levels in the workplace and in the environment, supported by a definition of the adequate analytical methods used to detect such exposures. (c) An analysis of technologies available for reducing or eliminating exposure, including the use of alternate technologies. (d) An analysis of the total impact of the use of the chemical on health as well as on other societal needs, for an evaluation of the balance of benefits and risks critically documented on both sides. If society will then conclude that it needs to maintain a given hazardous operation, the estimate of presumptive risk should be made by the best inductive guessing methods that scientific expertise can provide, but not as a replacement for the commitment to reduce or eliminate exposure to carcinogens whenever possible. Methods for risk assessment from occupational carcinogens are primarily methods for identifying the extent of exposure and for assessing technological alternatives. The right of the workers to a safe workplace, as recognized by the Occupa-
Saffiotti: Methodologies for Risk Assessment
395
tional Safety and Health Act of 1970, implies their right to know if and when they are exposed to a carcinogen and to know what is being done to avoid such exposure. The worker has a right to know also that the present state of our scientific knowledge of the chemical causation of cancer is such that we cannot recognize any level of exposure to a carcinogen as absolutely safe. REFERENCES 1. AD Hoc COMMITTEE ON THE EVALUATION OF Low LEVELSOF ENVIRONMENTAL
CARCINOGENS. 1971. Evaluation of environmental carcinogens-Report to the Surgeon General. I n Chemicals and the Future of Man. Hearings before the Subcommittee on Executive Reorganization and Government Research of the Committee on Government Operations of the US. Senate, 92nd Congress, 1st Session. U.S. Govt. Print. Off. Wash., D.C. 2. HOEL,D. G., D. W. GAYLOR, R. L. KIRSCHSTEIN, U. SAFFIOTTI & M. A. SCHNEIDERMAN. 1975. Estimation of risks of irreversible, delayed toxicity. J. Toxicol. Environ. Health 1: 133-1 5 1.
METHODOLOGIES FOR RISK ASSESSMENT IN OCCUPATIONAL CARCINOGENESIS Umberto Saffiotti Division of Cancer Cause & Prevention National Cancer Institute BetFesda, Maryland 20014
The review of the present state of our knowledge of the occupational causation of cancers-accomplished by the conference on which this volume is based -brings us to the final crucial question: How can we assess cancer risks and what can we do about them? There are two aspects to this problem. One concerns those risks that have already been contracted in the past by a population, and that can be directly measured by the institution of appropriate epidemiologic studies, provided that the exposure conditions, the size of the population, and the time of observation are sufficient for meaningful statistical comparisons. The other aspect concerns the assessment of those risks that are still in the process of being acquired or that are potential, such as those risks which accrue to a population of industrial workers when a new chemical process is introduced. For the latter category of risks, methods of epidemiologic observation cannot be used, short of setting up “human experiments” which would give their results in decades and, if positive, would have already inflicted irreversible damage on the exposed people by the time it is detected. For the assessment of these potential risks, animal experimentation provides an extremely valuable means of detecting the carcinogenic activity of many chemicals. The development of a battery of rapid prescreening methods, using criteria of chemical structure, interaction with cellular macromolecules, mutagenesis, and neoplastic transformation of cells in culture, will provide-after their validation in the near future-a much-needed additional short-term methodology for an early warning against potential risks of new products, applicable before the exposure of workers in production or utilization plants. It might be worth emphasizing that the detection of carcinogenicity by animal bioassays is severely limited by the sensitivity of such bioassay methods, which seldom can detect the induction of tumor incidences below 10%. Species variations in the susceptibility to carcinogens are well known and imply that a chemical may be carcinogenic for man and yet not be detected by conventional experimental tests. Conversely, there may be chemicals that are carcinogenic in a rodent species and not in humans; but in this case, we have no way of establishing this exclusion with any certainty, especially in view of the extreme genetic and environmental variability of the individuals in a human population. Therefore, at the present time, qualitative evidence that a chemical is carcinogenic in animals is sufficient to assume the likelihood that it be carcinogenic also in humans. The next logical step is that exposures to all carcinogens should be avoided or, if avoidance is not possible, minimized. Quantitative extrapolations of risk from animals to man have been proposed using the results of animal experiments together with mathematical models of dose-response extrapolation and the estimation of species-conversion factors.l*
393
394
Annals New York Academy of Sciences
One must, however, be extremely cautious of the many pitfalls of such complex exercises, which are based on several arbitrary selections of models and on a biological variability that is hard to predict. Risk assessment is often part of a necessary societal evaluation of the “function” in our society of an industrial process involving exposure to carcinogens. When the use of unproven methods of extrapolation to alleged “safe” or “acceptable” levels based on limited animal tests is proposed for regulatory action, we must first ask these questions: Have all other approaches to the reduction of risk been analyzed and, if possible, implemented? What evidence is there of the need to expose people to such carcinogens? What are the technological alternatives? In many cases the problem is discussed in terms of theoretical extrapolation to parts per trillion levels for a hypothetical population when the problem is the continued exposure of real people to massive and uncontrolled amounts in the workplace. A good example is provided by the current situation concerning the exposure to benzidine in the United States where, following the promulgation of a temporary standard by the Occupational Safety and Health Administration (Department of Labor) on May 3, 1973, it appears that the few major manufacturers have stopped production of this well-recognized occupational carcinogen, and a large number of plants that use benzidine in the production of dyes are now making their own benzidine and then reacting it to make the dyes. No official information exists on the exact number and location of these plants (but they were estimated to be over 1,000), since no registration is required. The Environmental Protection Agency hastened to publish a proposed standard to control the emission of benzidine in effluent waters from these plants, based on a hypothetical risk-extrapolation formula. The crucial issue is whether benzidine has any real social utility in our society. Industries of several countries have stopped its manufacture altogether, as in Great Britain and Italy. Technological alternatives for making dyes with similar color shades have been developed, but they are estimated to be somewhat more expensive. How does our society assess the value of certain shades of colors in textiles versus the proven risk of cancer for benzidine workers and the possible risk for the general population? What, then, are the methodologies for risk assessment? I propose the following steps: (a) A registration of all uses of products known to be carcinogenic by conclusive evidence obtained either from human observations or from animal bioassays. (b) An estimate of exposure levels in the workplace and in the environment, supported by a definition of the adequate analytical methods used to detect such exposures. (c) An analysis of technologies available for reducing or eliminating exposure, including the use of alternate technologies. (d) An analysis of the total impact of the use of the chemical on health as well as on other societal needs, for an evaluation of the balance of benefits and risks critically documented on both sides. If society will then conclude that it needs to maintain a given hazardous operation, the estimate of presumptive risk should be made by the best inductive guessing methods that scientific expertise can provide, but not as a replacement for the commitment to reduce or eliminate exposure to carcinogens whenever possible. Methods for risk assessment from occupational carcinogens are primarily methods for identifying the extent of exposure and for assessing technological alternatives. The right of the workers to a safe workplace, as recognized by the Occupa-
Saffiotti: Methodologies for Risk Assessment
395
tional Safety and Health Act of 1970, implies their right to know if and when they are exposed to a carcinogen and to know what is being done to avoid such exposure. The worker has a right to know also that the present state of our scientific knowledge of the chemical causation of cancer is such that we cannot recognize any level of exposure to a carcinogen as absolutely safe. REFERENCES 1. AD Hoc COMMITTEE ON THE EVALUATION OF Low LEVELSOF ENVIRONMENTAL
CARCINOGENS. 1971. Evaluation of environmental carcinogens-Report to the Surgeon General. I n Chemicals and the Future of Man. Hearings before the Subcommittee on Executive Reorganization and Government Research of the Committee on Government Operations of the US. Senate, 92nd Congress, 1st Session. U.S. Govt. Print. Off. Wash., D.C. 2. HOEL,D. G., D. W. GAYLOR, R. L. KIRSCHSTEIN, U. SAFFIOTTI & M. A. SCHNEIDERMAN. 1975. Estimation of risks of irreversible, delayed toxicity. J. Toxicol. Environ. Health 1: 133-1 5 1.
