Ketosis-Prone Diabetes—A New Subgroup of Patients with Atypical Type 1 and Type 2 Diabetes?

One objective for classification of a disease is the opportunity to study its epidemiology, etiology, and pathogenesis to provide various effective interventions for its prevention and treatment. The paper of Maldonado et al. (1) in this issue of JCEM reports on the classification of four groups of diabetic patients who presented with diabetic ketoacidosis (DKA). Of the two most common types of hyperglycemic crises, DKA most often occurs in type 1 diabetes (DM-1), and hyperglycemic hyperosmolar state most frequently arises in type 2 diabetes (DM-2). However, the occurrence of these acute metabolic emergencies is not specific to one type of diabetes or the other (2). Maldonado et al. (1) have carefully and meticulously studied a well-defined multiethnic group of patients with diabetes who presented with DKA. The study cohort was divided into four groups based on positive or negative β-cell insulin function (B+ or B−, respectively), as well as positive or negative autoimmune markers for DM-1 (A+ or A−, respectively). They also studied human leukocyte antigen class II genotype in these patients, profiled their demographic characteristics, and followed their response to insulin therapy on a longitudinal basis. Based on these criteria, they identified four groups: A+ B−, A− B−, A+ B+, and A− B+. This study demonstrates the heterogeneity associated with DKA and stimulates further discussion regarding the development of standardized nomenclature and a global database to establish a more accurate classification of ketosis-prone diabetes (KPD). According to the consensus report of the Expert Committee on Diagnosis and Classification of Diabetes Mellitus (3) and the World Health Organization (4), group A+ B− in the work of Maldonado et al. (1) would consist of type 1A diabetes mellitus, whereas A− B− would be considered as type 1 idiopathetic, because this group is nonobese with negative autoimmune markers and requires insulin for prevention of recurrence of DKA. Patients in this group, similar to the type 1A, also have low residual C-peptide but have a strong family history of DM-2, are frequently nonobese, and often are nonwhite with variable insulin requirement. Such patients have been reported in African countries as early as 1985 (5). This group is sometimes referred to as type 1.5 diabetes (6). The two groups with B+, however, are more heterogenous and share certain characteristics such as positive family history of DM-2, overweight or obese, mostly nonwhite, and often present in DKA as newly diagnosed DM. However, in this group, recurrence of DKA is rare after the initial event and often can initially be treated with diet or oral antidiabetic agents. These two groups have been described globally by various names, including “Flatbush diabetes,” “obese DKA,” “type 2 DM with DKA,” “newly onset with DKA,” “idiopathetic type 1 DM,” “atypical DM,” “type 3 KPD,” as well as “KPD B+.” Table 1 (courtesy of Dr. Mario Maldonado) summarizes the clinical characteristics and demographics of these patients as reported by various investigators (7–14). Of the two groups’ B+ cohorts, A− B+ most likely would consist of a heterogenous group of DM-2 patients with various degrees of insulin resistance and varying degrees of insulin secretory capacity. However, at the time of appearance of DKA, they have relatively lower C-peptide and increased counterregulatory hormones compared with hyperglycemic nonketosis patients (15). The last group, A+ B+, on the other hand poses the most challenging task as to the exact nomenclature for its classification or assignment of its etiopathological mechanism. Included in this category is a subclassification of individuals who had previously been labeled as latent autoimmune diabetes of adult (16–19). Maldonado et al. (1) designated 11 patients under the A+ B+ category and further subdivided them into two groups. Although the numbers in each subgroup are not large enough to make a definitive conclusion, the results are provocative enough that one may be tempted to propose certain hypotheses. Of the 11 patients in this category, six are positive for susceptibility gene DQB1*02, five of whom require insulin to prevent recurrence of DKA and have average glycosylated hemoglobin (HbA1c) of about 8.6%. All patients in the second subgroup of five lack DQB*02 genes; four require no insulin for future therapy, and their HbA1c average is about 6.5%. Most interesting in this A+ B+ group is the autoimmune profile of these patients, compared with the A+ B− (type 1A) patients (Table 2). Whereas in A+ B− group, whose HbA1c is greater than 10%, the major autoantibodies are IA-2 (61%) with 50% glutamic acid decarboxylase (GAD)65 and 17% combined GAD65 + IA-2, the A+ B+ patients exhibit 82% GAD65, only 18% IA-2, and no GAD65 + IA-2 antibodies. It therefore appears that type 1A patients who have higher frequency of IA-2 antibodies and more severe DKA than the group A+ B+ support the recent emerging concept that IA-2 positivity is associated with more aggressive β-cell destruction in autoimmune diabetes (20). Based on the above information, it is tempting to hypothesize that of the A+ B+ group the subgroup of six patients, who are more similar to DM-1 than the other subgroup of five, can justifiably be labeled latent autoimmune diabetes of adult. Furthermore, this group could have shared gene background with DM-1similar to the patients reported in the earlier study in which possible genetic interaction has occurred between DM-1 and DM-2 (21). This subgroup may also be considered to have a cluster of DM-1 and DM-2 and thus may benefit from preventive measures, such as immunosuppression of antibodies, including the use of GAD65 antibody vaccination that is now in phase II clinical trial (22). The latter approach, however, requires careful assessment of side effects, including possible theoretical neurological complications that may occur as a consequence of antibody development in the brain tissue. The second less aggressive subgroup of A+ B+ may have less autoimmune antibody if a more quantitative measurement of GAD65 antibody were used. The etiology of DKA in this group may be secondary to sudden exposure to hyperglycemia causing desensitization of the islets, in contrast to the glucose toxicity that leads to the destruction of pancreatic β-cells in DM-1 (23). Whereas the former condition may be reversible with establishment of euglycemia, the latter condition may be irreversible (24). In conclusion, the presentation of Maldonado et al. (1) provides a challenging thesis regarding future directions for studies delineating and classifying common traits of patients who experience DKA and hyperglycemic hyperosmolar state (25). These studies will be instrumental in building accurate multinational database registries with extensive and detailed family histories of DM-1 and DM-2, as well as assessment of prenatal and immediate postnatal environmental factors and nutritional history. Further advancement in this area necessitates agreement in four related areas of clinical investigation: 1) a uniform nomenclature, 2) a standardized assessment for GAD65 antibody and other autoimmune markers (26), 3) a standardized test for assessment of α- and β-cell capabilities (27), and 4) the utilization of more advanced technical procedures with the use of proteomics and genomics for genetic mapping (28). Comparative table of non-DM-1 presenting with KPD β+, Preserved β-cell functional reserve; BMI, body mass index. The ethnicity of the subjects was either sub-Saharan African or Afro-Caribbean. Mean ± sem. Calculated average of the two groups of reported KPD subjects (for the Sobngwi et al. report, 11 subjects with long-term insulin treatment and 10 subjects with insulin discontinuation; for the Maldonado et al. report, 11 subjects with positive antibodies and preserved β-cell functional reserve and 51 subjects with negative antibodies and preserved β-cell functional reserve). Percentage of ideal body weight (mean ± sem). Median BMI range was 14.2 to 41.8 kg/m2. Percentage of subjects who were overweight. Area under the curve for the glucagon stimulation test. The mean and sd for the two groups with preserved β-cell functional reserve were calculated. Comparative table of non-DM-1 presenting with KPD β+, Preserved β-cell functional reserve; BMI, body mass index. The ethnicity of the subjects was either sub-Saharan African or Afro-Caribbean. Mean ± sem. Calculated average of the two groups of reported KPD subjects (for the Sobngwi et al. report, 11 subjects with long-term insulin treatment and 10 subjects with insulin discontinuation; for the Maldonado et al. report, 11 subjects with positive antibodies and preserved β-cell functional reserve and 51 subjects with negative antibodies and preserved β-cell functional reserve). Percentage of ideal body weight (mean ± sem). Median BMI range was 14.2 to 41.8 kg/m2. Percentage of subjects who were overweight. Area under the curve for the glucagon stimulation test. The mean and sd for the two groups with preserved β-cell functional reserve were calculated. Frequency of autoantibody types in the two A+ groupsa Maldonado, M., personal communication. Frequency of autoantibody types in the two A+ groupsa Maldonado, M., personal communication. The work of the authors and colleagues mentioned were supported in part by General Clinical Research Center Grant RR00211, Division of Research Resources, National Institutes of Health (Bethesda, MD). Diabetic ketoacidosis; type 1 diabetes mellitus; type 2 diabetes mellitus; glutamic acid decarboxylase; glycosylated hemoglobin; KPD, ketosis-prone diabetes.