EDITORIALS
Using Nature to Understand Nurture The full implications of investigating the contribution of environmental influences to certain health risk factors by controlling for genetic effects through studies of identical twins have not been sufficiently appreciated. In the current issue of this journal, Newman and colleagues' report the results of their investigations in 250 pairs of White, male monozygotic (MZ) twins from the National Heart, Lung and Blood Institute (NHLBI) Twin Study.2,3 But, instead of focusing on resemblances within twin pairs (as might be expected from the usual twin study), they have, metaphorically speaking, crossed "through the looking glass"4 and used the MZ twin cohort to examine nongenetic influences on diabetes and cardiovascular disease (CVD) risk factors. Specifically, they have calculated the proportion of the variance in such risk factors as diastolic blood pressure, systolic blood pressure, and serum total cholesterol that can be explained by differences in body mass index (BMI) and weight gain between members of the MZ twin pairs. In other words, by comparing within genetically identical MZ twin pairs, they have eliminated genetic variability, thereby creating a pure culture of nongenetic determinants-determinants they define as being "environmental exposure" and "personal behaviors." Apart from the imaginative approach used by the authors to discover whether associations exist between obesity and other risk factors for diabetes and CVD independently of genetic influences, their report is noteworthy for several specific features. First, the size of the cohort of MZ twins is gratifyingly large; second, excellent height and weight data are available on the subjects at military induction (ages 17 to 28) and during middle age; and third, measures of an array of CVD risk factors were obtained on the subjects when they were 42-55 years old (at the time of the study examination). The resulting data have made it possible for the authors to assess the contribution of the following independent variables to risk factor status in middle age: BMI at the time of military induction (average age 21.9 years); BMI at the time of the study examination (average age 47.8 years); and amount of weight gained between military induction and the study examination. Comparison of the heavier and lighter brothers at the time of the study examination disclosed a mean intrapair difference in BMI of 1.8 kg/M2 (approximately 5.5 kg); on average, the heavier co-twin weighed about 81.4 kg while the lighter co-twin weighed about 75.9 kg. It was noteworthy that, within the MZ twin pairs at attained weight, the levels of all the measured indicators of CVD and diabetes risk (blood pressure, serum lipids, and post-load blood glucose) were significantly higher in the heavier than the lighter co-twins. The mean intrapair differences in CVD and diabetes risk factors were relatively small at the time of the study examination (total cholesterol differed by 0.15 mmolIL and onehour, post-load glucose by 0.3 mmol/L. The average man in the cohort gained 12.7 kg between age 20 and age 48. In comparison with the co-twins who gained less weight (or did not gain any weight) between military induction and middle age, the co-twins who gained more weight exhibited substantially higher levels of risk factors for diabetes and CVD. These excesses were independent of initial BMI and attributable solely to nongenetic influences. Although the induction BMI was independently related AJPH June 1990, Vol. 80, No. 6
to blood pressure, blood sugar, and serum lipids in the matched analyses of the twin pairs, the weight change during adult life was found to be more important than BMI at military induction in predicting diabetes and CVD risk factor status in middle age. Moreover, when the military induction and study examination data were made part of an inclusive analysis, level of obesity attained in middle age emerged as being most important in elevating the individual's risk from CVD and diabetes. Thus, Newman, et al, have been able to demonstrate convincingly that fatness arising from environmental/ behavioral influences can be responsible for increasing an individual's risk of developing diabetes and CVD, entirely apart from any genetic effects. This is not to say that obesity, hypertension, diabetes, and hyperlipidemia do not have genetic determinants-of course they do. What Newman, et al, have established is that behaviors and environmental exposures during adult life are at least partly responsible for the development of human obesity and, via the obesity, for the diabetes and CVD risk factors that are generated or exacerbated by increasing corpulence. The corollary to this observation is that, by decreasing a person's obesity, appropriate changes in the environment (food) and in behavior (eating and exercise) can be expected to reduce the associated risk of diabetes and CVD. But, when Newman, et al, state at one point that ... the influence of obesity on other CVD risk factors appears to be limited to less than 15 percent of the nongenetic variability of each risk factor . . .," it has to be kept in mind that they are referring to variances that are determined in part by the genetic make-up of the individuals under study and in part by the nature of the nongenetic influences acting upon them over time. Thus, in theory, if the strength of the nongenetic influences acting on the cohort had been either greater or less great, the proportion of the variance attributable to the environment would have differed correspondingly. Two studies that utilized the same population of male MZ twins from which Newman's sample was drawn are relevant. First, in a bivariate analysis of obesity in identical twins,5 it was concluded that genetic background largely determines the propensity to become obese; but whether a predisposed person actually becomes obese and the extent of the obesity depends on the nature of the obesity-promoting factors in the environment and the duration of exposure to those factors. Moreover, some genotypes may be much more sensitive than others to the environment. Second, the high concordance in BMI of members of MZ twin pairs decreases as the twin pairs become increasingly overweight;6 the concordance at followup for MZ twins who were 20 percent or more overweight was 60 percent; for those more than 40 percent overweight, it was only 36 percent. Thus, when co-twins are both obese, the correlation in the extent of their overweight is low. It was suggested that this reduced correlation in obese MZ co-twins might be explained as the consequence of "varying exposure to environmental conditions in genetically vulnerable individuals."5 If this hypothesis is correct, one could then postulate that when co-twins are increasingly sensitive to obesity-promoting forces in the environment, variations in that environment or in personal behavior (for example, in food selection) will assume a correspondingly greater importance, thereby accounting for the lower concordance in BMI that obtains in obese MZ co-twins. 657
EDITORIALS
If further research bears out this interpretation, the public health problem is not merely one of undertaking community-wide measures designed to prevent excessive weight gain and promote weight loss. We must also find practical ways of identifying individuals within the population who are especially vulnerable to obesity-promoting behaviors and environmental factors before they become severely overweight. Further studies focused on obese MZ twin pairs may help in this endeavor. Recent research7'8 has suggested that the pattern of regional fat distribution and (perhaps more specifically) the quantity of visceral fat drained by the portal venous system are better predictors than the BMI of diabetes and CVD risk factor status. For this reason, it would be helpful to conduct further investigations of MZ twin pairs in order to determine, first, the proportion ofthe variance in regional fat distribution (i.e., waist-to-hips circumference ratio [WHR]) that can be explained by nongenetic influences and, second, to compare the effects of nongenetic variations in BMI and nongenetic variations in WHR on diabetes and CVD risk factor status. Although Newman, et al, have used the term "obesity" in referring to the overweight members of their cohort, even at ages 42-55 years, the mean BMI of the heavier co-twins (26.6 kg/M2) is considerably less than the BMI of >27.8 kg/M2 proposed by the 1985 NIH Consensus Development Conference on Obesity9 as being the cutoff point for "overweight" in men. Given the small standard deviation of 3.2 kg/M2 for the overall group at study examination, it would appear that only a minority of the cohort were frankly obese. It would be interesting to know what the relative contribution of the most overweight members was to the reported risk factor status of the heavier co-twins. Finally, it is impressive that even the moderate elevation in BMI exhibited by the heavier co-twins in middle age was associated with significant increases in diabetes and CVD risk factors. We share the authors' optimism about the implications for intervention arising from their data. In terms of
I
long-term results, treatment of lesser degrees of overweight appears to be more successful than treatment of severe overweight. Clearly, the new ground broken by Newman, et al, deserves to be carefully cultivated by further research. REFERENCES 1. Newman B, Selby JV, Quesenberry CP Jr, et al: Nongenetic influences of obesity on other cardiovascular disease risk factors: An analysis of identical twins. Am J Public Health 1990; 80:675-678. 2. Feinleib M, Garrison RJ, Fabitz R, et al: The NHLBI Twin Study of Cardiovascular Disease Risk Factors: Methodology and summary of results. Am J Epidemiol 1977; 106:284-295. 3. Jablon S, Neel JV, Gershowitz H, et al: The NAS-NRC Twin Panel: Methods of construction of the panel, zygosity diagnosis, and proposed use. Am J Hum Genet 1967; 19:133-161. 4. Lewis Carroll: Through the Looking Glass and What Alice Found There. New York: Knopf, 1986. 5. Price RA, Stunkard AJ: Commingling analysis of obesity in twins. Hum Hered 1989; 39:121-135. 6. Stunkard AJ, Foch TT, Hrubec Z: A twin study of human obesity. JAMA 1986; 256:51-54. 7. Larsson B, Svardsudd K, Welin L, et al: Abdominal adipose tissue distribution, obesity, and risk of cardiovascular disease and death: 13-year follow-up of participants in the study of men born in 1913. Br Med J 1984; 288:1401-1404. 8. Bjorntorp P: Obesity and the risk of cardiovascular disease. Ann Clin Res 1985; 17:3-9. 9. Burton BT, Foster WR, Hirsch J, VanItallie TB: Health implications of obesity: An NIH Consensus Development Conference. Int J Obes 1985; 9:155-170.
THEODORE B. VANITALLIE, MD ALBERT J. STUNKARD, MD Address reprint requests to Theodore B. VanItallie, MD, St. Luke's/ Roosevelt Hospital Center, Amsterdam Avenue at 114th Street, New York, NY 10025. Dr. Van Itallie is Professor Emeritus of Medicine, Columbia University College of Physicians & Surgeons, New York City; Dr. Stunkard is Professor of Psychiatry, University of Pennsylvania School of Medicine. © 1990 American Journal of Public Health 0090-0036/90$1.50
ERRATUM
I
In: Evans MR, Henderson DK, Benett JE: Potential for laboratory exposures to biohazardous agents found in blood. Am J Public Health 1990; 80:423-427. On page 424, the last sentence in the section "Hepatitis B Surface Antigen Detection" should read as follows: "Within run variation for the confirmation runs was expressed as the CV calculated for the negative control (values ranged from 3 percent to 5 percent with a mean of 4 percent)." Due to a percent sign "%" being mis-read by the typesetter as a number "9", the percentages in the published version are incorrect. The Journal staff and the proofreader regret the error, and apologize to the authors and readers.
658
AJPH June 1990, Vol. 80, No. 6
Using Nature to Understand Nurture The full implications of investigating the contribution of environmental influences to certain health risk factors by controlling for genetic effects through studies of identical twins have not been sufficiently appreciated. In the current issue of this journal, Newman and colleagues' report the results of their investigations in 250 pairs of White, male monozygotic (MZ) twins from the National Heart, Lung and Blood Institute (NHLBI) Twin Study.2,3 But, instead of focusing on resemblances within twin pairs (as might be expected from the usual twin study), they have, metaphorically speaking, crossed "through the looking glass"4 and used the MZ twin cohort to examine nongenetic influences on diabetes and cardiovascular disease (CVD) risk factors. Specifically, they have calculated the proportion of the variance in such risk factors as diastolic blood pressure, systolic blood pressure, and serum total cholesterol that can be explained by differences in body mass index (BMI) and weight gain between members of the MZ twin pairs. In other words, by comparing within genetically identical MZ twin pairs, they have eliminated genetic variability, thereby creating a pure culture of nongenetic determinants-determinants they define as being "environmental exposure" and "personal behaviors." Apart from the imaginative approach used by the authors to discover whether associations exist between obesity and other risk factors for diabetes and CVD independently of genetic influences, their report is noteworthy for several specific features. First, the size of the cohort of MZ twins is gratifyingly large; second, excellent height and weight data are available on the subjects at military induction (ages 17 to 28) and during middle age; and third, measures of an array of CVD risk factors were obtained on the subjects when they were 42-55 years old (at the time of the study examination). The resulting data have made it possible for the authors to assess the contribution of the following independent variables to risk factor status in middle age: BMI at the time of military induction (average age 21.9 years); BMI at the time of the study examination (average age 47.8 years); and amount of weight gained between military induction and the study examination. Comparison of the heavier and lighter brothers at the time of the study examination disclosed a mean intrapair difference in BMI of 1.8 kg/M2 (approximately 5.5 kg); on average, the heavier co-twin weighed about 81.4 kg while the lighter co-twin weighed about 75.9 kg. It was noteworthy that, within the MZ twin pairs at attained weight, the levels of all the measured indicators of CVD and diabetes risk (blood pressure, serum lipids, and post-load blood glucose) were significantly higher in the heavier than the lighter co-twins. The mean intrapair differences in CVD and diabetes risk factors were relatively small at the time of the study examination (total cholesterol differed by 0.15 mmolIL and onehour, post-load glucose by 0.3 mmol/L. The average man in the cohort gained 12.7 kg between age 20 and age 48. In comparison with the co-twins who gained less weight (or did not gain any weight) between military induction and middle age, the co-twins who gained more weight exhibited substantially higher levels of risk factors for diabetes and CVD. These excesses were independent of initial BMI and attributable solely to nongenetic influences. Although the induction BMI was independently related AJPH June 1990, Vol. 80, No. 6
to blood pressure, blood sugar, and serum lipids in the matched analyses of the twin pairs, the weight change during adult life was found to be more important than BMI at military induction in predicting diabetes and CVD risk factor status in middle age. Moreover, when the military induction and study examination data were made part of an inclusive analysis, level of obesity attained in middle age emerged as being most important in elevating the individual's risk from CVD and diabetes. Thus, Newman, et al, have been able to demonstrate convincingly that fatness arising from environmental/ behavioral influences can be responsible for increasing an individual's risk of developing diabetes and CVD, entirely apart from any genetic effects. This is not to say that obesity, hypertension, diabetes, and hyperlipidemia do not have genetic determinants-of course they do. What Newman, et al, have established is that behaviors and environmental exposures during adult life are at least partly responsible for the development of human obesity and, via the obesity, for the diabetes and CVD risk factors that are generated or exacerbated by increasing corpulence. The corollary to this observation is that, by decreasing a person's obesity, appropriate changes in the environment (food) and in behavior (eating and exercise) can be expected to reduce the associated risk of diabetes and CVD. But, when Newman, et al, state at one point that ... the influence of obesity on other CVD risk factors appears to be limited to less than 15 percent of the nongenetic variability of each risk factor . . .," it has to be kept in mind that they are referring to variances that are determined in part by the genetic make-up of the individuals under study and in part by the nature of the nongenetic influences acting upon them over time. Thus, in theory, if the strength of the nongenetic influences acting on the cohort had been either greater or less great, the proportion of the variance attributable to the environment would have differed correspondingly. Two studies that utilized the same population of male MZ twins from which Newman's sample was drawn are relevant. First, in a bivariate analysis of obesity in identical twins,5 it was concluded that genetic background largely determines the propensity to become obese; but whether a predisposed person actually becomes obese and the extent of the obesity depends on the nature of the obesity-promoting factors in the environment and the duration of exposure to those factors. Moreover, some genotypes may be much more sensitive than others to the environment. Second, the high concordance in BMI of members of MZ twin pairs decreases as the twin pairs become increasingly overweight;6 the concordance at followup for MZ twins who were 20 percent or more overweight was 60 percent; for those more than 40 percent overweight, it was only 36 percent. Thus, when co-twins are both obese, the correlation in the extent of their overweight is low. It was suggested that this reduced correlation in obese MZ co-twins might be explained as the consequence of "varying exposure to environmental conditions in genetically vulnerable individuals."5 If this hypothesis is correct, one could then postulate that when co-twins are increasingly sensitive to obesity-promoting forces in the environment, variations in that environment or in personal behavior (for example, in food selection) will assume a correspondingly greater importance, thereby accounting for the lower concordance in BMI that obtains in obese MZ co-twins. 657
EDITORIALS
If further research bears out this interpretation, the public health problem is not merely one of undertaking community-wide measures designed to prevent excessive weight gain and promote weight loss. We must also find practical ways of identifying individuals within the population who are especially vulnerable to obesity-promoting behaviors and environmental factors before they become severely overweight. Further studies focused on obese MZ twin pairs may help in this endeavor. Recent research7'8 has suggested that the pattern of regional fat distribution and (perhaps more specifically) the quantity of visceral fat drained by the portal venous system are better predictors than the BMI of diabetes and CVD risk factor status. For this reason, it would be helpful to conduct further investigations of MZ twin pairs in order to determine, first, the proportion ofthe variance in regional fat distribution (i.e., waist-to-hips circumference ratio [WHR]) that can be explained by nongenetic influences and, second, to compare the effects of nongenetic variations in BMI and nongenetic variations in WHR on diabetes and CVD risk factor status. Although Newman, et al, have used the term "obesity" in referring to the overweight members of their cohort, even at ages 42-55 years, the mean BMI of the heavier co-twins (26.6 kg/M2) is considerably less than the BMI of >27.8 kg/M2 proposed by the 1985 NIH Consensus Development Conference on Obesity9 as being the cutoff point for "overweight" in men. Given the small standard deviation of 3.2 kg/M2 for the overall group at study examination, it would appear that only a minority of the cohort were frankly obese. It would be interesting to know what the relative contribution of the most overweight members was to the reported risk factor status of the heavier co-twins. Finally, it is impressive that even the moderate elevation in BMI exhibited by the heavier co-twins in middle age was associated with significant increases in diabetes and CVD risk factors. We share the authors' optimism about the implications for intervention arising from their data. In terms of
I
long-term results, treatment of lesser degrees of overweight appears to be more successful than treatment of severe overweight. Clearly, the new ground broken by Newman, et al, deserves to be carefully cultivated by further research. REFERENCES 1. Newman B, Selby JV, Quesenberry CP Jr, et al: Nongenetic influences of obesity on other cardiovascular disease risk factors: An analysis of identical twins. Am J Public Health 1990; 80:675-678. 2. Feinleib M, Garrison RJ, Fabitz R, et al: The NHLBI Twin Study of Cardiovascular Disease Risk Factors: Methodology and summary of results. Am J Epidemiol 1977; 106:284-295. 3. Jablon S, Neel JV, Gershowitz H, et al: The NAS-NRC Twin Panel: Methods of construction of the panel, zygosity diagnosis, and proposed use. Am J Hum Genet 1967; 19:133-161. 4. Lewis Carroll: Through the Looking Glass and What Alice Found There. New York: Knopf, 1986. 5. Price RA, Stunkard AJ: Commingling analysis of obesity in twins. Hum Hered 1989; 39:121-135. 6. Stunkard AJ, Foch TT, Hrubec Z: A twin study of human obesity. JAMA 1986; 256:51-54. 7. Larsson B, Svardsudd K, Welin L, et al: Abdominal adipose tissue distribution, obesity, and risk of cardiovascular disease and death: 13-year follow-up of participants in the study of men born in 1913. Br Med J 1984; 288:1401-1404. 8. Bjorntorp P: Obesity and the risk of cardiovascular disease. Ann Clin Res 1985; 17:3-9. 9. Burton BT, Foster WR, Hirsch J, VanItallie TB: Health implications of obesity: An NIH Consensus Development Conference. Int J Obes 1985; 9:155-170.
THEODORE B. VANITALLIE, MD ALBERT J. STUNKARD, MD Address reprint requests to Theodore B. VanItallie, MD, St. Luke's/ Roosevelt Hospital Center, Amsterdam Avenue at 114th Street, New York, NY 10025. Dr. Van Itallie is Professor Emeritus of Medicine, Columbia University College of Physicians & Surgeons, New York City; Dr. Stunkard is Professor of Psychiatry, University of Pennsylvania School of Medicine. © 1990 American Journal of Public Health 0090-0036/90$1.50
ERRATUM
I
In: Evans MR, Henderson DK, Benett JE: Potential for laboratory exposures to biohazardous agents found in blood. Am J Public Health 1990; 80:423-427. On page 424, the last sentence in the section "Hepatitis B Surface Antigen Detection" should read as follows: "Within run variation for the confirmation runs was expressed as the CV calculated for the negative control (values ranged from 3 percent to 5 percent with a mean of 4 percent)." Due to a percent sign "%" being mis-read by the typesetter as a number "9", the percentages in the published version are incorrect. The Journal staff and the proofreader regret the error, and apologize to the authors and readers.
658
AJPH June 1990, Vol. 80, No. 6
