Screening for the mitochondrial DNA A3243G mutation in children with insulin-dependent diabetes mellitus.

Since recent studies demonstrated the occurrence of the mitochondrial DNA (mtDNA) mutation A3243G in patients with adult-onset diabetes, an investigation was undertaken to determine the frequency of this mutation in a pediatric population with insulin-dependent diabetes mellitus (IDDM). DNA was extracted from peripheral blood of 270 pediatric patients with IDDM. The presence of the mtDNA A3243G mutation was screened for by minisequencing and mutation-specific ApaI endonuclease restriction after polymerase chain reaction (PCR) amplification of mtDNA. The A3243G mtDNA mutation was not found in any IDDM patients examined. This mutation is uncommon in children with IDDM from various ethnic and racial groups. Therefore, the contribution of the mutation to the pathogenesis of IDDM, if any, is minimal.

Anti-islet autoantibodies detected by monoclonal antibody 1A2: further studies suggesting a role in the pathogenesis of IDDM.

Insulin-dependent diabetes mellitus (IDDM) is associated with autoantibodies to several pancreatic islet antigens. We have described an assay in which autoantibodies displace a radiolabelled monoclonal anti-islet antibody. Sera from 87% of 429 children at time of diagnosis of IDDM were positive, while sera from control groups had much lower prevalences (1.3-19%). Sera from 41.9% of diabetic subjects remained positive after 20 years duration of IDDM. Sera from 23.6% of parents and 37.9% of non-diabetic siblings were positive. Twenty relatives who subsequently developed IDDM had the same prevalence of the antibodies (85%) as did the patients at time of diagnosis. These findings confirm that the autoantibodies detected by monoclonal antibody (mAb) 1A2 are common at the onset of IDDM and their presence prior to the onset of hyperglycaemia suggests that this method may be useful in screening non-diabetic populations. The high prevalence of antibodies in relatives reduces the efficacy for diabetes prediction, but suggests either that generation of these antibodies is an autosomal dominant trait, or that the antigen detected by these antibodies is cross-reactive with a common environmental antigen. Differentiation between these hypotheses will await the identification of the specific islet-cell antigen detected by mAb 1A2.

Postprandial thermogenesis at rest and postexercise before and after physical training in lean, obese, and mildly diabetic men.

To determine the independent impact of physical training on postprandial thermogenesis at rest and after 1 hour of cycling at 100 W, 10 lean (15% +/- 1% body fat), 10 obese (33% +/- 2% fat), and six obese diet-controlled, type II diabetic men (34% +/- 4% fat) underwent 12 weeks of vigorous cycle ergometer training (4 h/wk at approximately 70% of maximum oxygen consumption [VO2max]) while maintaining body weight and composition. Body weight was held constant by refeeding the energy expended in each training session. Cardiorespiratory fitness increased by approximately 27%, but body weight and fat did not change. Before and at least 4 days after the last exercise session, energy expenditure was measured for 3 hours under four conditions: (1) rest, no meal; (2) rest, after a 720-kcal mixed meal; (3) postexercise after 1 hour cycling, no meal; and (4) postexercise, meal after exercise. The thermic effect of food was calculated as postprandial minus postabsorptive energy expenditure at rest and postexercise (kcal/3 h). Before and after training, the thermic effect of food during rest was lower in obese than in lean men, and lower in diabetic than in obese men (P less than .05). Thermogenesis was improved after short-term exercise in obese and diabetic men compared with that at rest, but was not normalized (P less than .05 for lean v obese, diabetic men). A significant effect of training on thermogenesis was due to a small but significant increase after training for diabetic men under the postexercise condition. Thus, while short-term exercise enhances but does not normalize thermogenesis in obese and diabetic men, long-term exercise training leading to increased cardiorespiratory fitness, in the absence of changes in body composition, leads to a small increase in thermogenesis in diabetic men, which manifests only after a short period of exercise.

Gestational diabetes mellitus: evidence for autoimmunity against the pancreatic beta cells.

Diabetes mellitus is a frequent transient or rare permanent complication of pregnancy. The role of autoimmune phenomena in this gestational form of diabetes is incompletely understood. We have examined sera from 312 pregnant women who had abnormal glucose tolerance (based on a screening examination during the second trimester) for the presence of islet cell surface antibodies or insulin autoantibodies. Fifty-eight of these women were lost to follow-up. Of the remaining subjects, 144 (57.1%) had gestational diabetes diagnosed by formal glucose tolerance testing and the others (42.9%) were normal. Sixty percent of the women with gestational diabetes eventually required insulin to control their blood glucose during pregnancy. One serum from the non-diabetic women was positive for insulin antibodies (0.9%); 8 of the sera from the patients with gestational diabetes were positive (5.6%). Subsequent analysis revealed that all nine of the women whose sera were positive for insulin autoantibodies had been treated with insulin previously. Islet cell surface antibodies were strongly correlated with gestational diabetes. Forty-five of 144 gestational diabetic sera were positive (31.3%) whereas only 9 of 108 suspect control sera (8.3%) and 7 of 60 unknown sera (11.7%) were positive. These data suggest that a high percentage of pregnant women who screen positive for glucose intolerance have serological evidence of an autoimmune response against the pancreatic islets, in spite of the state of relative immune tolerance during pregnancy. These data suggest that autoimmune phenomena may play a role in gestational diabetes and that the presence of islet cell antibodies can predict insulin-requiring gestational diabetes.

Strong association between diabetes and displacement of mouse anti-rat insulinoma cell monoclonal antibody by human serum in vitro.

In an attempt to identify novel pancreatic beta-cell surface antigens, mouse monoclonal antibodies (MoAbs) were raised against rat insulinoma (RIN5F) cells with standard techniques. Several clones were identified whose antibodies bound specifically to RIN5F cells but not to other rat, mouse, and human target cells. Each of these MoAbs was radiolabeled, and the specificity of binding of each MoAb was determined by the ability of excess cold homologous MoAb to displace the labeled MoAb. Six RIN5F cell-specific MoAbs of different epitopic specificities were identified. The relevance of these beta-cell epitopes to human insulin-dependent diabetes (IDDM) was demonstrated by the differential ability of human serums from control and diabetic children to displace the radiolabeled MoAbs from the RIN5F cells. Serums from 333 children without diabetes or a family history of diabetes and from 156 newly diagnosed IDDM patients were tested. Only one IgM MoAb was specifically displaced by the IDDM serums, i.e., 146 of 156, compared to serums from control children, i.e., 10 of 333. With immunofluorescence, the serum component responsible for the displacement of the mouse MoAb was identified as IgG. Most of the positive control serums were from children with active autoimmune thyroiditis. Serums from children with other forms of glucose intolerance did not displace MoAb 1A2. There was no correlation between age and the degree of displacement of 1A2. Thus, the displacement of 1A2 is a specific and sensitive marker of diabetes susceptibility easily applicable to mass screening.(ABSTRACT TRUNCATED AT 250 WORDS)

No excess of DR*3/4 in Ashkenazi Jewish or Hispanic IDDM patients.

The gene frequencies, haplotype relative risks, and zygotic assortments of HLA-DR in three ethnically defined samples of insulin-dependent diabetes mellitus (IDDM) patients were determined in a prospective family study. Although DR3 and DR4 were positively associated with IDDM in the probands of 123 northern European, 94 Ashkenazi Jewish, and 49 New York Hispanic families, significant excess of DR*3/4 heterozygotes was observed only among the probands from families of northern European ancestry. There was also a significant decrease in the frequency of Bw62,DR4 haplotypes derived by northern European patients from their mothers compared with their fathers. This difference, together with data reported in the literature, suggests that the expressivity of the susceptible genotype(s) in IDDM patients may be modified by protective maternal effects associated with Bw62,DR4 and probably other DR4 haplotypes. Samples of IDDM patients from populations with high frequencies of these modifiers should have different DR-gene frequencies contributed by fathers and mothers, capable of accounting for the observed Hardy-Weinberg disequilibrium. We postulate that, because the mechanism of action of these modifiers is distinct from that of the susceptibility gene, the difference must be considered in devising strategies for elucidation of the mode of inheritance of the disease and for understanding the molecular nature of the susceptibility.

Functional properties of a naturally occurring Trp1200----Ser1200 mutation of the insulin receptor.

Based on the sequence of cDNA encoding the intracellular domain of the insulin receptor beta-subunit, we recently defined a heterozygous point mutation causing a Ser for Trp substitution at position 1200 in the tyrosine kinase domain of a patient (BI-2) with the type A syndrome of insulin resistance. We have now sequenced the remainder of BI-2's insulin receptor cDNA-coding region and find no additional alterations in the encoded proreceptor protein. The nucleotide sequence of cDNA encoding the portion of the beta-subunit which includes Trp1200 was normal in BI-2's unaffected mother. Hybridization of a mutant allele-specific oligonucleotide to polymerase chain reaction-amplified cDNA confirmed the presence of the mutant allele in the proband and excluded it in her unaffected sister and mother, 18 normal control subjects, and six other subjects with insulin resistance. To determine whether this mutation had functional consequences for receptor signalling, we reconstructed it into a full-length insulin receptor cDNA expression vector. Chinese hamster ovary cells were transfected with mutant cDNA, and the expressed insulin receptors were compared to receptors expressed by cells transfected with wild-type receptor cDNA. Both mutant and wild-type receptors were properly processed into receptor alpha- and beta-subunits, were expressed on the cell surface, and displayed similar insulin-binding affinity. In contrast, insulin-stimulated autophosphorylation of the mutant receptors was severely impaired, whether assessed in intact cells or with a partially purified receptor preparation.(ABSTRACT TRUNCATED AT 250 WORDS)

Frequency of B cells committed to the production of antibodies to insulin in newly diagnosed patients with insulin-dependent diabetes mellitus and generation of high affinity human monoclonal IgG to insulin.

Circulating autoantibodies to insulin can be detected in patients with insulin-dependent (type I) diabetes mellitus (IDDM) at the onset of the clinical disease. To characterize the autoantibody response in IDDM patients, we determined the frequency of circulating B cells committed to the production of IgM, IgG, and IgA to insulin in 12 newly diagnosed IDDM patients and, for comparison, in 9 healthy subjects and 17 insulin-treated IDDM patients. We found that B cells committed to the production of anti-insulin IgG, but not IgM, autoantibodies are present at much higher frequency in the circulation of newly diagnosed IDDM patients before insulin treatment (0.209 +/- 0.142%, mean value +/- SD of total IgG-producing cell precursors) as compared with age-matched healthy controls (0.032 +/- 0.030% of total IgG-producing cell precursors). In IDDM patients who had been treated with insulin, cells producing IgG antibody to insulin were 0.177 +/- 0.139% of total IgG-producing cell precursors. Generation of IgG mAb from B cells of IDDM patients revealed that they were monoreactive, i.e., they bound to insulin, but to none of the other Ag tested, and displayed a high affinity for insulin (Kd approximately 10(-7) moles/liter). In contrast, the IgG mAb derived from healthy subjects were polyreactive, i.e., they bound to all Ag tested, and displayed a low to moderate affinity for insulin (Kd approximately 10(-5) to 10(-6) moles/liter). These findings show that lymphocytes committed to the production of high affinity IgG autoantibodies to insulin are common in the B cell repertoire at the onset of IDDM.

Insulin-dependent diabetes mellitus.

Much progress has been made in the treatment of IDDM in children and adolescents during the past 15 years. The perfect tools required to ensure relative normoglycemia at all times are still not available. Until the time that a perfect and permanent indwelling glucose sensor can be developed and connected to an insulin infusing device, or much less toxic immunosuppressive drugs enable pancreatic or beta-cell transplantation on a wide basis, we should endeavor to provide the best possible care for our patients with this most difficult disease. Education, psychosocial support, and physician availability and knowledge are all important to this effort. Early results from groups treating large numbers of insulin-dependent diabetic youngsters and adolescents suggest that it is possible to keep glycohemoglobin levels in the normal or good-control range in upwards of 50% of patients, although the early Diabetes Control and Complications Trial data has found this to be particularly difficult in adolescents. In our own patients, we have found major hypoglycemic attacks requiring glucagon or intravenous glucose to be rare (much less than one per patient per year). Minor episodes marked by the feeling that one's blood glucose concentration is decreasing rapidly, however, are common. Therefore, education and appropriate snacking are crucial. In our population, we have noted improvement in school performance with improved blood glucose control, but these early impressions have yet to be documented through time. It is the pediatrician's role to start these youngsters on the right track early to avoid complications that usually do not occur until the patient reaches an age when he or she is treated by an internist.(ABSTRACT TRUNCATED AT 250 WORDS)

DQA1 restriction fragment length polymorphisms and insulin-dependent diabetes mellitus: a BglII fragment labels a subset of B8,DR3 haplotypes uniquely associated with insulin-dependent diabetes mellitus.

Class II restriction fragment length polymorphism studies of 38 pedigrees with multiple cases of insulin-dependent diabetes mellitus revealed the existence of a DQA1-related polymorphism that distinguishes two kinds of HLA-B8,DR3 haplotypes. One of these, characterized by the presence of DQA1-BglII 7.20 kb, was present in all 14 examples inherited by patients and in 6 of the 12 B8,DR3 haplotypes not so inherited. The apparently complete association between the presence of this fragment and the "affected" status of B8,DR3 haplotypes (p = 0.004) was confirmed in a separate group of 26 simplex pedigrees selected for the presence of this haplotype in the respective probands (combined p less than 0.0001).

Definition of IDDM-associated HLA DQ and DX RFLPs by segregation analysis of multiplex sibships.

Genetic variation of the DQ alpha and beta and of the DX alpha genes, detectable as RFLP in genomic DNA digests, has been suggested to improve the identification of individuals at high risk for insulin-dependent diabetes mellitus (IDDM). DNA from all members of 32 IDDM multiplex families was digested with six restriction endonucleases and the resulting fragments analyzed in Southern blots for hybridization with labeled cDNA probes for those genes. A computerized segregation analysis procedure was then used to assign fragments to haplotypes. Associations among fragments and between fragments and haplotypes characterized serologically and biochemically for their class II genes and IDDM-carrier status were calculated. The results indicate that the alleles of the DX alpha polymorphism maintain linkage disequilibrium with those of the DQ beta genes responsible for the well-known DQ beta 3.2-IDDM association, so that IDDM-carrier haplotypes carry disproportionally often both DQ beta 3.2 and DX alpha-TaqI-2.2kb. Thus, these RFLPs identify a DR-DQ-DX haplotype in high linkage disequilibrium, rather than the locus or loci that account for their high relative risk. However, four DR4-DQ beta 3.2 haplotypes that lack DX alpha-TaqI-2.2kb were encountered, two of which are "affected." These haplotypes suggest that the identification of the "disease locus" can be facilitated by the study of unusual haplotypes in which distinct IDDM-associated alleles occur separated from their neighbors of the standard genetic configurations.

Relationship of stressful housing conditions to the onset of diabetes mellitus induced by multiple, sub-diabetogenic doses of streptozotocin in mice.

The effect of the stress of crowded housing conditions (10 mice/cage) on the onset of diabetes after multiple, sub-diabetogenic doses of streptozotocin (MSZ) in male C57BL/KsJ mice was investigated. Prior to MSZ treatment, the group-housed and individually-housed animals had similar plasma glucose levels, while the former group's plasma corticosterone (CS) levels were elevated (54 +/- 8 ng/ml, p less than 0.03; 166% of the latter group). The group-housed animals became hyperglycemic (253 +/- 23 mg/dl) 2 days after the MSZ (day 7), with maximum hyperglycemia (506 +/- 23 mg/dl) developing by day 10. The individually housed animals did not become hyperglycemic until day 10 (303 +/- 24 mg/dl, p less than 0.001), and did not reach maximal hyperglycemia until between days 31 and 46, when plasma glucose levels were no longer different from the group-housed mice (507 +/- 37 mg/dl). There was a significant and progressive rise in CS levels of the stressed animals, reaching 218 +/- 25 ng/ml at day 46. The rise in CS of the unstressed animals was not significant until day 46, when the mean value reached 96 +/- 19 ng/ml (p less than 0.001 vs. basal). However, even at the conclusion of the experiment, the mean CS in the stressed animals was still 227% of that in the unstressed group (p less than 0.001). These studies demonstrate that the effects of stress (biochemically documented as an increase in CS levels) act synergistically with streptozotocin to promote an earlier onset of diabetes mellitus in males of this murine strain.

Molecular characterization by high-resolution isoelectric focusing of the products encoded by the class II region loci of the major histocompatibility complex in humans. I. DR and DQ gene variants.

We describe a new approach to the analysis of the structural polymorphism of the DR beta, DQ alpha, and DQ beta polypeptide chains of human histocompatibility class II antigens. In comparison to conventional two-dimensional gel studies, this method provides sharper definition of the protein bands and side-by-side comparisons within the same gel, thereby permitting the detection of minor differences in the isoelectric points of the protein chains. Using this methodology we have analyzed the IEF polymorphism and the variability in the number of the DR beta chains encoded by different DR haplotypes. Twenty DR beta chain variants, which include the products of no less than two separate DR beta loci, have been thus far identified. Alleles at one of these loci are assumed to code for DR beta chains carrying the DR alloespecificities DR1, DR2, DR3, DR4, DR5, DRw6, DR7, and DR8. Alleles at a second DR beta locus encode DR beta chains that may be shared by serologically DR-different haplotypes and carry supertypic serologic specificities (i.e., DRw52 and DRw53). We also demonstrate here that the structural polymorphisms of the DQ alpha and DQ beta chains are more extensive than previously thought, report the characterization of 14 DQ beta variants, and define their relationship to the previously described DQw serologic specificities. In addition, we describe the class II haplotype associations observed for the different DR and DQ variants characterized.

Human monoclonal multiple-organ-reactive autoantibodies distinguished by mouse monoclonal anti-idiotypic antibodies: expression of idiotopes in humans with and without autoimmune diseases.

Previously we reported on the production and characteristics of a number of human monoclonal autoantibodies. All of these autoantibodies were of the IgM class and reacted with antigens in multiple organs. In this study we generated IgG murine monoclonal anti-idiotypic antibodies against five human monoclonal autoantibodies, (i.e., MOR-h2, MOR-h3, MOR-h4, CG1, and CG2). These anti-idiotypic antibodies reacted strongly with the corresponding human monoclonal autoantibody, but minimally or not at all with other human monoclonal autoantibodies. By using these anti-idiotypic antibodies as probes, we screened sera obtained from normal individuals and patients with insulin-dependent diabetes mellitus, Hashimoto's thyroiditis, and systemic lupus erythematosus for the expression of idiotopes. Our study showed that the idiotopes recognized by three of the anti-idiotypic antibodies, i.e., anti-CG1, anti-CG2, and anti-MOR-h2, were not expressed, but the idiotopes recognized by two of the anti-idiotypic antibodies, i.e., anti-MOR-h3 and anti-MOR-h4, were expressed in normal individuals. In patients with autoimmune disorders, there was no increase in the expression of the CG1, CG2, and MOR-h2 idiotopes, but 45 and 23% of the patients with systemic lupus erythematosus showed a significant increase in the expression of the MOR-h3 and MOR-h4 idiotopes respectively. These findings show that there is widespread expression in the B cell repertoire of certain autoantibody-associated idiotopes.

Randomized, controlled trial of diabetic patient education: improved knowledge without improved metabolic status.

We randomized 749 insulin-treated patients on the rolls of the Mount Sinai Medical Center Diabetes Clinic in a controlled trial of diabetic patient education; 345 agreed to participate, of whom 165 were assigned to the education group and 180 to the control group. Cognitive scores increased from 5.3 +/- 1.6 to 5.8 +/- 1.6 in the education group, but there was no change in the control group, whose score was 5.3 +/- 1.7 before and after the intervention (P = .0073). HbA1c fell from 6.8 +/- 2.1 to 6.1 +/- 2.0% in the education group and from 6.6 +/- 2.0 to 6.3 +/- 2.0% in the control group, an insignificant difference (P = .1995). The fasting blood glucose decreased from 223 +/- 94 to 179 +/- 73 mg/dl in the education group and from 199 +/- 81 to 185 +/- 76 mg/dl in the controls (P = .1983). Triglycerides, high- and low-density lipoprotein cholesterol, and insulin dosage also failed to show significant variation among groups. The foot-lesion score showed similar progression in the education and control groups. Neither diastolic nor systolic blood pressure showed significantly greater change in the education or the control group, with falls noted, particularly in diastolic pressures, in both patient groups. Differences between the groups were not significant for sick days, hospitalizations, emergency room visits, or outpatient visits. The sample sizes of the study and control populations were sufficiently large to detect a difference in means between the education and control groups in the HbA1c, the primary outcome variable, of greater than 1.0%, with alpha = .05 and a power of .95. Thus, our study suggests that patient education may not be an efficacious therapeutic intervention in most adults with insulin-treated diabetes mellitus.

Multiple low-dose streptozotocin-induced diabetes in the mouse: further evidence for involvement of an anti-B cell cytotoxic cellular auto-immune response.

Anti-B cell auto-immunity may play a role in the pathogenesis of diabetes in mice resulting from multiple subdiabetogenic doses of the pancreatic B cell toxin, streptozotocin. In the present study we have investigated the cytotoxic anti-B cell response in these mice. A major role for B lymphocytes, macrophages, or their products in the cytotoxic response originally detected in vitro was eliminated by passing splenocytes from the mice treated with multiple subdiabetogenic doses of streptozotocin over a nylon wool column. The removal of the adherent cells enhanced the cytotoxicity against a rat insulinoma cell line in vitro by that expected due to enrichment of T-lymphocytes by approximately two-fold. The induction of diabetes after multiple subdiabetogenic doses of streptozotocin is strain dependent. Mice of five strains were immunized with rat insulinoma cells, but only splenocytes from the two strains susceptible to multiple subdiabetogenic doses of streptozotocin demonstrated a significant cytotoxic response against the rat insulinoma cells in vitro. Mice pre-immunized with either the rat insulinoma cells or with syngeneic islets labelled in vitro with the hapten trinitrophenol developed hyperglycaemia more rapidly than control mice after multiple subdiabetogenic doses of streptozotocin. In the latter experiment the control mice immunized with complete Freund's adjuvant alone also became hyperglycaemic after a modified multiple subdiabetogenic dose of streptozotocin that did not cause diabetes in non-immunized mice. In mice pre-treated with either adjuvant or cyclophosphamide and then given a modified multiple subdiabetogenic dose of streptozotocin (35 mg/kg X 5 rather than 40 mg/kg) the degree of hyperglycaemia was reduced and there was no protective effect of cyclophosphamide.(ABSTRACT TRUNCATED AT 250 WORDS)

Erythrocyte Na+-K+ ATPase activity in childhood: regulation by genetic factors independent of body weight.

A number of previous investigations in obese and normal-weight adults suggested a link between (usually) lower than normal erythrocyte Na+-K+ ATPase activity and obesity, although more recent studies have disputed this link. Developmental patterns of ATPase activity in children have not been studied. Infants of diabetic mothers show an increased incidence of macrosomia and such infants are at risk for later development of obesity. These children provide an ideal group to study the link (if any) between this enzyme and obesity, as well as changes in the level of the enzyme which may occur with age and/or the onset of obesity. The Na+-K+ ATPase levels in erythrocytes from 87 normal children and 27 infants of diabetic mothers were therefore studied by an ouabain-binding method. The study populations contained representatives of three ethnic groups (black, Caucasian and Hispanic) and individuals of normal body weight, as well as overweight and obese individuals. The ATPase activities were not correlated with body weight, sex, or age, nor was umbilical cord red cell ATPase correlated with birthweight. Several children who are being followed have shown considerable variation in percentage ideal body weight (IBW); however their ATPase levels have remained unchanged. There were, however, significant differences among the ethnic groups, with Caucasians having the highest values (0.443 +/- 0.012 pmol ouabain bound/10(9) cells). These data demonstrate the importance of matching study groups by race when investigating factors affecting this enzyme. The differences among the ethnic groups suggest a genetic component in the determination of the activity of this enzyme. This link was further investigated in 14 families and revealed considerable heterogeneity within a given ethnic group, with family members generally having similar enzyme activities, thus supporting a genetic basis for the different enzyme levels. Thus erythrocyte Na+-K+ ATPase in childhood is primarily regulated by genetic factors which are generally independent of body weight.

Excess of DR3/4 in type I diabetes. What does it portend?

The HLA-DR genotypes of 158 new type I diabetic probands from simplex families are compared with those of 43 multiply affected sibships. There were no significant differences in the gene frequencies of the insulin-dependent diabetes mellitus (IDDM)-associated alleles, DR3 and DR4, and whereas the DR3/4 heterozygotes were as frequent among simplex probands as among the first affected of multiplex sibships, subsequently affected sibs displayed lower frequencies of this genotype in this as well as in previously reported samples, indicating that the excessive risk associated with DR3/4 heterozygosity depends on the order of affection and thus on environmental factors. It is proposed that the penetrance of the susceptibility gene is enhanced by epistatic effects of this genotype and that this enhancement is strongest under conditions of low environmental liability. Thus, the excessive risk for DR3/4 individuals appears to depend on secondary interactions between DR and the environmental factors that trigger the onset of this disease and does not in itself indicate the existence of distinct susceptibility alleles in coupling with these genes, i.e., of genetic heterogeneity or overdominance.