250
Letters
to
the Editor
GENETIC ORIGIN OF DIABETES: RE-EVALUATION OF TWIN DATA SIR,—Most investigators have concluded that diabetes mellitus is a genetic disorder’-’. However, Tattersall and Pyke8 have concluded that if one of a pair of monozygotic twins develops diabetes before age 40, the unaffected twin will frequently not become diabetic. This unexpected finding has raised the possibility that a significant number of young diabetics may have an acquired form of the disease. We have re-evaluated Tattersall arfd Pyke’s data in an attempt to determine why their conclusions differ from others. Tattersall and Pyke followed up 96 pairs of monozygotic twins for periods ranging from less than 3 to more than 20 years and reported a 67% frequency of concordance (both twins becoming diabetic). In those twins diagnosed after age 40, however, the concordance-rate was 92%, whereas if a twin was diagnosed as diabetic before 40 years of age, the concordance-rate was only 52%. Tattersall and Pyke concluded that factors other than heredity are the cause of diabetes mellitus in many patients under age 40.
YEARS FOLLOWED
Fig. 1--Concordance for diabetes mellitus
in 96
pairs of identical
twins.
The data of Tattersall and Pyke8 have been used. Concordance is plotted as a function of the years that the twins were followed. Values in parentheses indicate the number of twins at the time of evaluation who had either become concordant or remained discordant during the periods indicated on the abscissa. two important oversights in the analytical used. First, in calculating concordance Tattersall approach and Pyke did not appropriately take into account the varying periods of time for which the pairs were followed up. Secondly, they seem to have assumed that the risk of diabetes developing in a discordant (non-diabetic) twin decreased with time, and that if concordance were to develop it would do so rapidly. The absolute number of twins becoming concordant did decrease each year, but as more twins became concordant so the population at risk decreased. Although the absolute number of twins becoming diabetic may decrease with time, the risk of
There
1. 2.
Fig. 2--Concordance for diabetes mellitus in 21 pairs according to age at onset of diabetes in index twin.
of identical twins
The data of Gottlieb and Root have been used. Concordance is plotted as a function of the years that the twins were followed. Values in parentheses indicate the number of twins at the time of evaluation who had either become concordant or remained discordant during the period on the abscissa.
any
given unaffected unchanged.
twin
becoming
diabetic may
actually
remain
To evaluate the effect of these two variables on the conclusions of Tatters6ll and Pyke, we tabulated the total number of twins who became concordant as a function of the specified period for which they were followed up. The denominator in any calculation of concordance must include all twins who had become concordant as well as those who remained discordant for this time period. The rationale behind this analysis is, first, that once concordance develops it is irreversible and, second, that, as time passes, more of the twins develop concordance and the pool at risk thus decreases. It is apparent, therefore, that in calculating concordance the duration for which the twins were followed up must be considered, and that only those twins studied for a given period of time should be included in the calculation of the percent of twins reaching concordance for that period. As shown in fig. 1 52% (47 of 91) of the twin pairs studied for 3 years by Tattersall and Pyke reached concordance (5 pairs were not followed up for the full 3 years). However, of the twins followed up for 10 years 63 out of 81 (78%) became concordant. When followed up for up to 20 years, 65 of 71 (92%) of the pairs reached concordance. Thus almost all twin pairs studied by Tattersall and Pyke became concordant in time. A like analysis of a smaller series of monozygotic twins compiled by Gottlieb and Root9 produced similar results. The concordance-rate in their series reached 82% in twins followed up for 24 years.
are
Neel, J. V. Am. J. hum. Genet. 1962, 14, 353. Neel, J. V. in Small Blood Vessel Involvement in Diabetes Mellitus (edited by M. D. Siperstein, A. R. Colwell, Sr., and K. Meyer); p. 295. Washington, D C., 1964. 3. Darlow, J. M., Smith, Charles, Duncan, L. J. P. Ann. hum. Genet. 1973, 37, 157 4. Post, R H. Diabetes, 1962, 11, 56. 5. Steinberg, A G. ibid, 1961, 10, 269. 6. Simpson, N. E. Ann. hum. Genet. 1962, 26, 1. 7. Rimoin, D L. Diabetes, 1967, 16, 346. 8. Tattersall, R. B , Pyke, D. A. Lancet, 1972, i, 1120.
The data of Gottlieb and Root9 also demonstrate that the which concordance develops depends upon the age at which the disease appears in the index twin (fig. 2). Equally important, the data of Gottlieb and Root reveal that in the discordant twin of a diabetic of onset before age 40, the risk of developing diabetes remained constant. These data indicate that concordance for diabetes in identical twins requires a longer period of time to develop when the disease afflicts the index twin before the age of 40 than when it occurs at an older age. Tattersall and Pyke8 did acknowledge that concordance occurred later in younger than in older onset diabetics, and data from a later publication 10 indicate that, in their concordant twins, 100% concordance was achieved in about 3 years in the older age-group but took 9 years if diabetes developed in the index twin before age 40. These data indicate that a genetic predisposition for diabetes rate at
9. Gottlieb, M. S., Root, H. F. Diabetes, 1968, 17, 693. 10. Pyke, D. A., rattersall, R. B. ibid 1973, 22, 613.
251 mellitus may well have existed in those monozygotic twins of Tattersall and Pyke who were discordant for diabetes mellitus at the time of study. The findings of Gottlieb and Root demonstrate that discordant monozygotic twins must be followed up for two to three decades before it is assumed that diabetes mellitus in younger patients does not have a genetic basis. In Tattersall and Pyke’s series approximately half the 28 discordant pairs of younger twins were followed up for less than 10 years. Since the data of Gottlieb and Root indicate that the rate of development of concordance is unchanged in younger diabetic twins for at least 24 years, had Tattersall and Pyke followed their twin pairs for an adequate time, concordance in the early and older onset diabetics may well have been comparable. We conclude that the evidence of Tattersall and Pyke and of Gottlieb and Root is consistent with the conclusion that diabetes mellitus in man, regardless of age at onset, is primarily genetic in origin, with environmental factors apparently influencing only the time of appearance of this disease. This study was performed under Veterans Administration projects MRIS 7927-01 and MRIS 3194-01 and was supported by program project grant HL-06285, projects A-9 and A-10, from the National Heart and Lung Institute. We thank Peter Martin and Vera Vaughn for their assistance. Metabolism Section, Medical Service, Veterans Administration Hospital, San Francisco, California 94121, U.S.A. and Department of Medicine and the Cardiovascular Research Institute, University of California San Francisco
MONROE B. ROSENTHAL IRA D. GOLDFINE MARVIN D. SIPERSTEIN
glucose. 2 parents were found to have maturity-onset diabetes. No additional J .D.M. patients were found. Except for B7, there was no segregation distortion of the HLA antigens, irrespective of whether the sibs were diabetic or not. Of the 35 families 10 had one B7-positive parent. 8 of these were heterozygous; the other 2 were B7 homozygous. Data from several other informative families have been published (see table). Depending on whether 1 or both parents are B7 heterozygous, one expects, respectively, 50% or 75% of the children also to be B7 positive. This distribution is confirmed in the healthy sibs, but only 1 of the 16 diabetics is B7 positive (P
Letters
to
the Editor
GENETIC ORIGIN OF DIABETES: RE-EVALUATION OF TWIN DATA SIR,—Most investigators have concluded that diabetes mellitus is a genetic disorder’-’. However, Tattersall and Pyke8 have concluded that if one of a pair of monozygotic twins develops diabetes before age 40, the unaffected twin will frequently not become diabetic. This unexpected finding has raised the possibility that a significant number of young diabetics may have an acquired form of the disease. We have re-evaluated Tattersall arfd Pyke’s data in an attempt to determine why their conclusions differ from others. Tattersall and Pyke followed up 96 pairs of monozygotic twins for periods ranging from less than 3 to more than 20 years and reported a 67% frequency of concordance (both twins becoming diabetic). In those twins diagnosed after age 40, however, the concordance-rate was 92%, whereas if a twin was diagnosed as diabetic before 40 years of age, the concordance-rate was only 52%. Tattersall and Pyke concluded that factors other than heredity are the cause of diabetes mellitus in many patients under age 40.
YEARS FOLLOWED
Fig. 1--Concordance for diabetes mellitus
in 96
pairs of identical
twins.
The data of Tattersall and Pyke8 have been used. Concordance is plotted as a function of the years that the twins were followed. Values in parentheses indicate the number of twins at the time of evaluation who had either become concordant or remained discordant during the periods indicated on the abscissa. two important oversights in the analytical used. First, in calculating concordance Tattersall approach and Pyke did not appropriately take into account the varying periods of time for which the pairs were followed up. Secondly, they seem to have assumed that the risk of diabetes developing in a discordant (non-diabetic) twin decreased with time, and that if concordance were to develop it would do so rapidly. The absolute number of twins becoming concordant did decrease each year, but as more twins became concordant so the population at risk decreased. Although the absolute number of twins becoming diabetic may decrease with time, the risk of
There
1. 2.
Fig. 2--Concordance for diabetes mellitus in 21 pairs according to age at onset of diabetes in index twin.
of identical twins
The data of Gottlieb and Root have been used. Concordance is plotted as a function of the years that the twins were followed. Values in parentheses indicate the number of twins at the time of evaluation who had either become concordant or remained discordant during the period on the abscissa.
any
given unaffected unchanged.
twin
becoming
diabetic may
actually
remain
To evaluate the effect of these two variables on the conclusions of Tatters6ll and Pyke, we tabulated the total number of twins who became concordant as a function of the specified period for which they were followed up. The denominator in any calculation of concordance must include all twins who had become concordant as well as those who remained discordant for this time period. The rationale behind this analysis is, first, that once concordance develops it is irreversible and, second, that, as time passes, more of the twins develop concordance and the pool at risk thus decreases. It is apparent, therefore, that in calculating concordance the duration for which the twins were followed up must be considered, and that only those twins studied for a given period of time should be included in the calculation of the percent of twins reaching concordance for that period. As shown in fig. 1 52% (47 of 91) of the twin pairs studied for 3 years by Tattersall and Pyke reached concordance (5 pairs were not followed up for the full 3 years). However, of the twins followed up for 10 years 63 out of 81 (78%) became concordant. When followed up for up to 20 years, 65 of 71 (92%) of the pairs reached concordance. Thus almost all twin pairs studied by Tattersall and Pyke became concordant in time. A like analysis of a smaller series of monozygotic twins compiled by Gottlieb and Root9 produced similar results. The concordance-rate in their series reached 82% in twins followed up for 24 years.
are
Neel, J. V. Am. J. hum. Genet. 1962, 14, 353. Neel, J. V. in Small Blood Vessel Involvement in Diabetes Mellitus (edited by M. D. Siperstein, A. R. Colwell, Sr., and K. Meyer); p. 295. Washington, D C., 1964. 3. Darlow, J. M., Smith, Charles, Duncan, L. J. P. Ann. hum. Genet. 1973, 37, 157 4. Post, R H. Diabetes, 1962, 11, 56. 5. Steinberg, A G. ibid, 1961, 10, 269. 6. Simpson, N. E. Ann. hum. Genet. 1962, 26, 1. 7. Rimoin, D L. Diabetes, 1967, 16, 346. 8. Tattersall, R. B , Pyke, D. A. Lancet, 1972, i, 1120.
The data of Gottlieb and Root9 also demonstrate that the which concordance develops depends upon the age at which the disease appears in the index twin (fig. 2). Equally important, the data of Gottlieb and Root reveal that in the discordant twin of a diabetic of onset before age 40, the risk of developing diabetes remained constant. These data indicate that concordance for diabetes in identical twins requires a longer period of time to develop when the disease afflicts the index twin before the age of 40 than when it occurs at an older age. Tattersall and Pyke8 did acknowledge that concordance occurred later in younger than in older onset diabetics, and data from a later publication 10 indicate that, in their concordant twins, 100% concordance was achieved in about 3 years in the older age-group but took 9 years if diabetes developed in the index twin before age 40. These data indicate that a genetic predisposition for diabetes rate at
9. Gottlieb, M. S., Root, H. F. Diabetes, 1968, 17, 693. 10. Pyke, D. A., rattersall, R. B. ibid 1973, 22, 613.
251 mellitus may well have existed in those monozygotic twins of Tattersall and Pyke who were discordant for diabetes mellitus at the time of study. The findings of Gottlieb and Root demonstrate that discordant monozygotic twins must be followed up for two to three decades before it is assumed that diabetes mellitus in younger patients does not have a genetic basis. In Tattersall and Pyke’s series approximately half the 28 discordant pairs of younger twins were followed up for less than 10 years. Since the data of Gottlieb and Root indicate that the rate of development of concordance is unchanged in younger diabetic twins for at least 24 years, had Tattersall and Pyke followed their twin pairs for an adequate time, concordance in the early and older onset diabetics may well have been comparable. We conclude that the evidence of Tattersall and Pyke and of Gottlieb and Root is consistent with the conclusion that diabetes mellitus in man, regardless of age at onset, is primarily genetic in origin, with environmental factors apparently influencing only the time of appearance of this disease. This study was performed under Veterans Administration projects MRIS 7927-01 and MRIS 3194-01 and was supported by program project grant HL-06285, projects A-9 and A-10, from the National Heart and Lung Institute. We thank Peter Martin and Vera Vaughn for their assistance. Metabolism Section, Medical Service, Veterans Administration Hospital, San Francisco, California 94121, U.S.A. and Department of Medicine and the Cardiovascular Research Institute, University of California San Francisco
MONROE B. ROSENTHAL IRA D. GOLDFINE MARVIN D. SIPERSTEIN
glucose. 2 parents were found to have maturity-onset diabetes. No additional J .D.M. patients were found. Except for B7, there was no segregation distortion of the HLA antigens, irrespective of whether the sibs were diabetic or not. Of the 35 families 10 had one B7-positive parent. 8 of these were heterozygous; the other 2 were B7 homozygous. Data from several other informative families have been published (see table). Depending on whether 1 or both parents are B7 heterozygous, one expects, respectively, 50% or 75% of the children also to be B7 positive. This distribution is confirmed in the healthy sibs, but only 1 of the 16 diabetics is B7 positive (P
