256
Proc. roy, Soc. Med. Volume 68 April 1975
16
Table I Number of islets per cm' pancreas in post-mortem material from untreated caaes of juvenile diabetes mellitus (I)M), insulin-treated cases and matched nondiabetic controls No.
Untreated juvenile DM: Patients
Age at death (years) Mean Range
9
15
6-21
Controls
13
15
7-21
Insulin-treated juvenile DM: Patients
12
32
19-52
Controls
13
31
20-52
No. of isletsper Duration of DM cm' pancreas Mean Range Average Range
2.5 1-4 weeks weeks
85 214
18.5 years
6-38 years
48
196
7-3000 84-399
0-204102-410
0 < 84 in 7 patients * < 102 in 11 patients
history, and for scattered insulitis in a proportion ofpatients.
that juvenile diabetes is an organ-specific autoimmune disease.
The loss-of islets might hypothetically result from (1) congenital deficiency, (2) 'overwork atrophy' of normal islets due to excessive stimulation as a result of extrapancreatic diabetogenic factors, (3) viral infection, (4) autoimmune destruction.
REFERENCES Doniach I & Morgan A G (1973) Clinical Endocrinology 2, 233 Gamble D R, Taylor K W & Cumming H (1973) British MedicalJournal ii, 260 Gepts W (1965) Diabetes 14,619 Goldstein D C, Drash A, Fibbs J & Blizard R M (1970) Journal ofPediatrics 77, 304 Irvine W J, Scarth L, Clarke B F, Cullen D R & Duncan L J P (1970) Lancet ii, 163 LeCompte P M & Legg M A (1972) Diabetes 21,762 Maclean N & Ogilvie R F (1955) Diabetes 4,367-376 Nerup J, Andersen 0 0, Bendixen G, Egeberg J & Poulsen J E (1971) Diabetes 20,424 Schmidt M B (1902) Malnchener Medizinische Wochenschrift 49, 51 Warren S, LeCompte P M & Legg M A (1966) The Pathology of Diabetes Mellitus. 4th edn. Henry Kimpton, London
Hypothesis (1) is difficult to refute, but the histological findings suggest it is unlikely, i.e. in some fatal cases the number is not reduced, in others many of the surviving islets are atrophic rather than missing. Hypothesis (2) is popular; to me it seems unlikely, since most endocrine glands respond to excessive maintained stimulation by marked hyperplasia rather than by atrophy. Hypothesis (3), supported by epidemiological studies (Gamble etal. 1973), might underlie Dr D R Gamble (The Public Health Laboratory, the insulitis found in a proportion of patients. West Park Hospital, Epsom, Surrey) Hypothesis (4), suggested by immunological studies (Goldstein et al. 1970, Irvine et al. 1970 Viral and Epidemiological Studies and Nerup et al. 1971), is not fully supported by the histological findings, i.e. the absence of A number of viruses can produce pancreatitis in plasma cells in the infiltrate and the total absence man and animals and in some cases this may be followed by diabetes. We have recently reported of insulitis in chronic cases. diabetes in CD-1 mice following Coxsackie B4 The main functional lesion is the failure of virus infection and these animals showed lymphonormal beta-cell regeneration in the face of the cytic infiltration of the islets of Langerhans, and hyperglycxcmic stimulus. The direct basis of this is B cell degranulation and degeneration (Coleman unknown but can be regarded as the essential et al. 1973). Similar changes have been described islet cell abnormality, possibly inherited, in in mice with diabetes induced by encephalojuvenile diabetes. The end result is atrophy and carditis virus (Craighead & McLane 1968), and loss of islets which might be accelerated by the similarity of these changes to those seen in excessive functional demands and in some cases fatal cases of insulin-dependent diabetes in precipitated by viral infection. The finding of children (Gepts 1972) suggests the possibility of a excessive prevalence of thyroid and gastric auto- viral etiology in human diabetes. antibodies in juvenile diabetes is intriguing and, Direct experimental verification of this sugthough suggestive of a common factor in genetic make-up, does not thereby necessarily indicate gestion is difficult but our recent epidemiological
17
257
Section ofEndocrinology
The age incidence (Fig 2) increases erratically from birth to a peak at 11-12 years and there may be secondary peaks at 5, and 8-9 years. These peaks may be related to school entry at 4-5 years, transfer to junior school at 7-8 years, and to secondary school at 11-12 years (Fig 2). At these ages, susceptible and infected children are mixed and waves of infection might be expected. The high incidence of Coxsackie B and mumps virus infection in 5-6-year-old children is probably explained in this way. 1973
Four-week period
Fig 1 Seasonal incidence of1100 new cases of diabetes in childhood
studies on cases of childhood diabetes notified to the British Diabetic Association in 1973 provide some relevant information.
If the onset of diabetes is related to school entry, children who start school younger than usual should develop diabetes early. Fig 3 shows that the maximum incidence in children who start school at 5 years occurs at 11-12 years, but at 5 years in those who start school at age 3 or less. Children who start school at age 4 have, an
A seasonal variation in incidence of juvenile diabetes has been reported (Adams 1926, Gamble & Taylor 1969, Gamble et al. 1973) and our recent studies have confirmed this seasonal pattern (Fig 1). Incidence varies with an autumn peak about October and a winter peak from December to March. This implies the presence of seasonal, environmental etiological factors and virus infection seems a likely candidate.
Children
starting
10
school
aged 4 years
(239 cases)
A g-e
i n
y e a r s
Fig 2 Age distribution of 900 new cases ofchildhood diabetes notified to the British Diabetic Association in 1973. Histograms show number ofnewfull-time pupils, by age group, at maintained and independent schools in England and Wales (Department ofEducation & Science 1973)
1
2
3 A G E
4 5
6
I N
7
8
9
10 11 12 13
14 15 16
Y E A R S
Fig 3 Age distribution (°) ofnew cases ofdiabetes, by age ofstarting school
258
Proc. roy. Soc. Med. Volume 68 April 1975
intermediate pattern with similar incidences at 5 and 11 years. These incidence data also suggest that children who start school at age 3 or 4 have a relatively high incidence in their first school year, whereas those who start at 5 do not. This phenomenon may also be related to virus infection since we know that certain viruses, particularly enteroviruses, spread with extraordinary facility among infants in nurseries and institutions.
The second wave of diabetes seems to occur in children aged 5 who started school at less than 5, when they are joined by those who start at 5; but the latter children, who are exposed to the same environment, have a relatively low incidence in their first school year. Presumably, an environmental factor is present in school life before the age of 5. Virus infections in this period may produce cumulative pancreatic damage leading eventually to diabetes after further infections at the age of 5 or 6. Similarly, in children who start school at 5 cumulative infections may lead to diabetes in later years.
Although these epidemiological features could result from other influences, virus infection may provide the most plausible explanation. Moreover virus induced diabetes in animals, and evidence of an excess of Coxsackie B4 virus antibodies in juvenile diabetics (Gamble et al. 1969, Gamble et al. 1973) give circumstantial support to this idea.
If diabetes is precipitated by virus infection, what is the significance of the lymphocytic infiltration of the islets of Langerhans in human and animal diabetes? Is it a harmless coincidental response to virus infection, or is it an autoaggressive phenomenon playing an essential part in the pathogenesis of diabetes ? T cell activity is usually directed against viruses which bud from the host cell surface and such viruses might well provoke an autoaggressive immune response since they possess antigens derived from the host cell. The available evidence however suggests that, if viruses induce diabetes, enteroviruses may be the most likely candidates. These viruses are liberated by destruction of the host cell; they are not budded from the cell surface and there is no evidence that they possess host-derived antigen, or antigen which crossreacts with host cells. Since their method of replication is to divert host protein and nucleic acid synthesis to the production of viral components - leading eventually to destruction of the cell - there would seem to be little need to invoke
the agency of an autoaggressive immunological process to account for reduced insulin synthesis. In any case I suspect that it all happens too quickly. The seasonal pattern of juvenile diabetes suggests that, if it has an autoimmune basis, the process is triggered by a seasonal factor - and a virus infection would probably be the most likely seasonal trigger. Since I have argued that diabetes may sometimes result from a number of infections producing cumulative pancreatic damage, at what stage would autoimmunity develop ? One would expect the autoimmune process to be a self-feeding mechanism, and once triggered, the resulting tissue damage would sustain the process. One might therefore expect that the first infection to damage the pancreas would start the process. If diabetes then followed rapidly it would be inconsistent with my suggestion of multiple infections to explain the epidemiological features. If, on the other hand, diabetes developed slowly over a long period, the seasonal incidence of the initial virus infection would probably be obscured and would not be reflected in the incidence of diabetes.
Finally, there is not much evidence of any gross abnormality in the immunological response of diabetics to infection. Their response to most viruses - including most of the coxsackieviruses appears to follow the usual pattern. Is it likely that an inherited immunological defect would manifest itself in response to a particular virus? or is it more likely that there is an inherited susceptibility of the islet cells to particular viruses ? In conclusion, although I have suggested in discussion that the immune response may not be of pathogenic significance in diabetes, we must keep an open mind. While I think that this may be true in virus-induced experimental diabetes in animals, the animal model may not be relevant to the human disease.
Acknowledgment: This investigation was supported by a grant from British Diabetic Association. REFERENCES
Adams S F (1926) Archives of Internal Medicine 37,861 Coleman T J, Gamble D R & Taylor K W (1973) British MedicalJournal iii, 25 Craighead E & McLane M F (1968) Science 162,913 Department of Education and Science (1973) Statistics of Education 1972, volume l: Schools. HMSO, London Gamble D R, Kinsley M L, Fitzgerald M G, Bolton R & Taylor K W (1969) British Medical Journaliii, 627 Gamble D R & Taylor K W (1969) British MedicalJournaliii, 631 Gamble D R, Taylor K W & Cummings H (1973) British MedicalJournal iv, 260 Gepts W (1972) In: Handbook of Physiology. Ed. D F Steiner & N Freinkel. American Physiological Society, Washington DC;
1,289
Proc. roy, Soc. Med. Volume 68 April 1975
16
Table I Number of islets per cm' pancreas in post-mortem material from untreated caaes of juvenile diabetes mellitus (I)M), insulin-treated cases and matched nondiabetic controls No.
Untreated juvenile DM: Patients
Age at death (years) Mean Range
9
15
6-21
Controls
13
15
7-21
Insulin-treated juvenile DM: Patients
12
32
19-52
Controls
13
31
20-52
No. of isletsper Duration of DM cm' pancreas Mean Range Average Range
2.5 1-4 weeks weeks
85 214
18.5 years
6-38 years
48
196
7-3000 84-399
0-204102-410
0 < 84 in 7 patients * < 102 in 11 patients
history, and for scattered insulitis in a proportion ofpatients.
that juvenile diabetes is an organ-specific autoimmune disease.
The loss-of islets might hypothetically result from (1) congenital deficiency, (2) 'overwork atrophy' of normal islets due to excessive stimulation as a result of extrapancreatic diabetogenic factors, (3) viral infection, (4) autoimmune destruction.
REFERENCES Doniach I & Morgan A G (1973) Clinical Endocrinology 2, 233 Gamble D R, Taylor K W & Cumming H (1973) British MedicalJournal ii, 260 Gepts W (1965) Diabetes 14,619 Goldstein D C, Drash A, Fibbs J & Blizard R M (1970) Journal ofPediatrics 77, 304 Irvine W J, Scarth L, Clarke B F, Cullen D R & Duncan L J P (1970) Lancet ii, 163 LeCompte P M & Legg M A (1972) Diabetes 21,762 Maclean N & Ogilvie R F (1955) Diabetes 4,367-376 Nerup J, Andersen 0 0, Bendixen G, Egeberg J & Poulsen J E (1971) Diabetes 20,424 Schmidt M B (1902) Malnchener Medizinische Wochenschrift 49, 51 Warren S, LeCompte P M & Legg M A (1966) The Pathology of Diabetes Mellitus. 4th edn. Henry Kimpton, London
Hypothesis (1) is difficult to refute, but the histological findings suggest it is unlikely, i.e. in some fatal cases the number is not reduced, in others many of the surviving islets are atrophic rather than missing. Hypothesis (2) is popular; to me it seems unlikely, since most endocrine glands respond to excessive maintained stimulation by marked hyperplasia rather than by atrophy. Hypothesis (3), supported by epidemiological studies (Gamble etal. 1973), might underlie Dr D R Gamble (The Public Health Laboratory, the insulitis found in a proportion of patients. West Park Hospital, Epsom, Surrey) Hypothesis (4), suggested by immunological studies (Goldstein et al. 1970, Irvine et al. 1970 Viral and Epidemiological Studies and Nerup et al. 1971), is not fully supported by the histological findings, i.e. the absence of A number of viruses can produce pancreatitis in plasma cells in the infiltrate and the total absence man and animals and in some cases this may be followed by diabetes. We have recently reported of insulitis in chronic cases. diabetes in CD-1 mice following Coxsackie B4 The main functional lesion is the failure of virus infection and these animals showed lymphonormal beta-cell regeneration in the face of the cytic infiltration of the islets of Langerhans, and hyperglycxcmic stimulus. The direct basis of this is B cell degranulation and degeneration (Coleman unknown but can be regarded as the essential et al. 1973). Similar changes have been described islet cell abnormality, possibly inherited, in in mice with diabetes induced by encephalojuvenile diabetes. The end result is atrophy and carditis virus (Craighead & McLane 1968), and loss of islets which might be accelerated by the similarity of these changes to those seen in excessive functional demands and in some cases fatal cases of insulin-dependent diabetes in precipitated by viral infection. The finding of children (Gepts 1972) suggests the possibility of a excessive prevalence of thyroid and gastric auto- viral etiology in human diabetes. antibodies in juvenile diabetes is intriguing and, Direct experimental verification of this sugthough suggestive of a common factor in genetic make-up, does not thereby necessarily indicate gestion is difficult but our recent epidemiological
17
257
Section ofEndocrinology
The age incidence (Fig 2) increases erratically from birth to a peak at 11-12 years and there may be secondary peaks at 5, and 8-9 years. These peaks may be related to school entry at 4-5 years, transfer to junior school at 7-8 years, and to secondary school at 11-12 years (Fig 2). At these ages, susceptible and infected children are mixed and waves of infection might be expected. The high incidence of Coxsackie B and mumps virus infection in 5-6-year-old children is probably explained in this way. 1973
Four-week period
Fig 1 Seasonal incidence of1100 new cases of diabetes in childhood
studies on cases of childhood diabetes notified to the British Diabetic Association in 1973 provide some relevant information.
If the onset of diabetes is related to school entry, children who start school younger than usual should develop diabetes early. Fig 3 shows that the maximum incidence in children who start school at 5 years occurs at 11-12 years, but at 5 years in those who start school at age 3 or less. Children who start school at age 4 have, an
A seasonal variation in incidence of juvenile diabetes has been reported (Adams 1926, Gamble & Taylor 1969, Gamble et al. 1973) and our recent studies have confirmed this seasonal pattern (Fig 1). Incidence varies with an autumn peak about October and a winter peak from December to March. This implies the presence of seasonal, environmental etiological factors and virus infection seems a likely candidate.
Children
starting
10
school
aged 4 years
(239 cases)
A g-e
i n
y e a r s
Fig 2 Age distribution of 900 new cases ofchildhood diabetes notified to the British Diabetic Association in 1973. Histograms show number ofnewfull-time pupils, by age group, at maintained and independent schools in England and Wales (Department ofEducation & Science 1973)
1
2
3 A G E
4 5
6
I N
7
8
9
10 11 12 13
14 15 16
Y E A R S
Fig 3 Age distribution (°) ofnew cases ofdiabetes, by age ofstarting school
258
Proc. roy. Soc. Med. Volume 68 April 1975
intermediate pattern with similar incidences at 5 and 11 years. These incidence data also suggest that children who start school at age 3 or 4 have a relatively high incidence in their first school year, whereas those who start at 5 do not. This phenomenon may also be related to virus infection since we know that certain viruses, particularly enteroviruses, spread with extraordinary facility among infants in nurseries and institutions.
The second wave of diabetes seems to occur in children aged 5 who started school at less than 5, when they are joined by those who start at 5; but the latter children, who are exposed to the same environment, have a relatively low incidence in their first school year. Presumably, an environmental factor is present in school life before the age of 5. Virus infections in this period may produce cumulative pancreatic damage leading eventually to diabetes after further infections at the age of 5 or 6. Similarly, in children who start school at 5 cumulative infections may lead to diabetes in later years.
Although these epidemiological features could result from other influences, virus infection may provide the most plausible explanation. Moreover virus induced diabetes in animals, and evidence of an excess of Coxsackie B4 virus antibodies in juvenile diabetics (Gamble et al. 1969, Gamble et al. 1973) give circumstantial support to this idea.
If diabetes is precipitated by virus infection, what is the significance of the lymphocytic infiltration of the islets of Langerhans in human and animal diabetes? Is it a harmless coincidental response to virus infection, or is it an autoaggressive phenomenon playing an essential part in the pathogenesis of diabetes ? T cell activity is usually directed against viruses which bud from the host cell surface and such viruses might well provoke an autoaggressive immune response since they possess antigens derived from the host cell. The available evidence however suggests that, if viruses induce diabetes, enteroviruses may be the most likely candidates. These viruses are liberated by destruction of the host cell; they are not budded from the cell surface and there is no evidence that they possess host-derived antigen, or antigen which crossreacts with host cells. Since their method of replication is to divert host protein and nucleic acid synthesis to the production of viral components - leading eventually to destruction of the cell - there would seem to be little need to invoke
the agency of an autoaggressive immunological process to account for reduced insulin synthesis. In any case I suspect that it all happens too quickly. The seasonal pattern of juvenile diabetes suggests that, if it has an autoimmune basis, the process is triggered by a seasonal factor - and a virus infection would probably be the most likely seasonal trigger. Since I have argued that diabetes may sometimes result from a number of infections producing cumulative pancreatic damage, at what stage would autoimmunity develop ? One would expect the autoimmune process to be a self-feeding mechanism, and once triggered, the resulting tissue damage would sustain the process. One might therefore expect that the first infection to damage the pancreas would start the process. If diabetes then followed rapidly it would be inconsistent with my suggestion of multiple infections to explain the epidemiological features. If, on the other hand, diabetes developed slowly over a long period, the seasonal incidence of the initial virus infection would probably be obscured and would not be reflected in the incidence of diabetes.
Finally, there is not much evidence of any gross abnormality in the immunological response of diabetics to infection. Their response to most viruses - including most of the coxsackieviruses appears to follow the usual pattern. Is it likely that an inherited immunological defect would manifest itself in response to a particular virus? or is it more likely that there is an inherited susceptibility of the islet cells to particular viruses ? In conclusion, although I have suggested in discussion that the immune response may not be of pathogenic significance in diabetes, we must keep an open mind. While I think that this may be true in virus-induced experimental diabetes in animals, the animal model may not be relevant to the human disease.
Acknowledgment: This investigation was supported by a grant from British Diabetic Association. REFERENCES
Adams S F (1926) Archives of Internal Medicine 37,861 Coleman T J, Gamble D R & Taylor K W (1973) British MedicalJournal iii, 25 Craighead E & McLane M F (1968) Science 162,913 Department of Education and Science (1973) Statistics of Education 1972, volume l: Schools. HMSO, London Gamble D R, Kinsley M L, Fitzgerald M G, Bolton R & Taylor K W (1969) British Medical Journaliii, 627 Gamble D R & Taylor K W (1969) British MedicalJournaliii, 631 Gamble D R, Taylor K W & Cummings H (1973) British MedicalJournal iv, 260 Gepts W (1972) In: Handbook of Physiology. Ed. D F Steiner & N Freinkel. American Physiological Society, Washington DC;
1,289
