INTRODUCTION — Since the mid-1990s, increasing attention has been focused on a heterogeneous condition characterized by presentation with diabetic ketoacidosis (DKA) in patients who do not necessarily fit the typical characteristics of autoimmune type 1 diabetes. Earlier reports used the terms "atypical diabetes," "Flatbush diabetes," "diabetes type 1B," and "ketosis-prone type 2 diabetes mellitus" to describe subsets of this condition, and it was noted that in some instances patients presented with DKA as the first manifestation of diabetes and evolved to insulin independence [1]. The prevalence of this condition, now termed ketosis-prone diabetes (KPD), appears to be increasing in a wide range of ethnic groups worldwide [2-6].
The classification, pathophysiology, natural history, and management of KPD will be reviewed here. Patients with islet autoantibodies who do not present with ketosis, including those termed "latent autoimmune diabetes in adults" (LADA), "type 1.5 diabetes" [7,8], and "slowly progressing type 1 diabetes" [9] are discussed elsewhere. (See "Classification of diabetes mellitus and genetic diabetic syndromes".)
CLASSIFICATION OF KPD — The goal of new classification schemes is to enable clinicians to predict which patients with diabetic ketoacidosis (DKA) require temporary insulin treatment versus life-long insulin therapy. They also highlight subgroups for genetic and pathogenetic studies.
Ketosis-prone diabetes (KPD) comprises a group of diabetes syndromes characterized by severe beta cell dysfunction (manifested by presentation with DKA or unprovoked ketosis) and a variable clinical course. These syndromes do not fit the traditional categories of diabetes defined by the American Diabetes Association (ADA). To date, attempts to differentiate patients with KPD into clinically distinct subgroups have resulted in four different classification schemes: the ADA system, a modified ADA system, a body mass index (BMI)-based system, and the Aß system (based on the presence or absence of autoantibodies and the presence or absence of beta cell functional reserve).
In a longitudinal study comparing the four classification schemes for accuracy and predictive value, the Aß system was shown to be the most accurate in predicting long-term insulin dependence 12 months after the index DKA event, with 99 percent sensitivity and 96 percent specificity [10,11].
ADA system — In the American Diabetes Association (ADA) classification, type 1 diabetes is characterized by autoimmune destruction of the pancreatic beta cells, leading to absolute insulin deficiency. Markers of immune-mediated diabetes include antibodies to glutamic acid decarboxylase (GAD), islet tyrosine phosphatase 2 (IA-2), and zinc transporter 8 (ZnT8). Among patients presenting with DKA (absolute insulin deficiency), those who lack autoantibodies are referred to as "idiopathic type 1" or "type 1b"; the latter includes patients with the clinical appearance of type 2 diabetes, with some becoming insulin independent. (See "Classification of diabetes mellitus and genetic diabetic syndromes", section on 'Type 1 diabetes'.)
Modified ADA system — A modification of the American Diabetes Association (ADA) scheme is utilized by investigators in France, which divides KPD patients into three groups [12]. Patients with beta cell autoantibodies are classified as type 1a just as in the ADA scheme, while those who lack autoantibodies are distinguished retrospectively, based on long-term insulin dependence, into "KPD insulin dependent" (KPD-ID) and "KPD non-insulin dependent" (KPD-NID). Both type 1a and KPD-ID patients have clinical characteristics of type 1 diabetes with poor beta cell function, while subjects with KPD-NID have clinical characteristics of type 2 diabetes with preserved beta cell function for a prolonged duration.
BMI system — The body mass index (BMI)-based scheme separates KPD patients into lean (BMI <28 kg/m2, clinically resembling type 1 with low beta cell function) or obese (BMI ≥28 kg/m2, clinically resembling type 2 with preserved beta cell function) [13].
Aß system — The collaborative group at Baylor College of Medicine and the University of Washington developed a classification system (Aß classification) that distinguishes four KPD subgroups based on the presence or absence of autoantibodies and the presence or absence of beta cell functional reserve, as measured by a fasting or glucagon-stimulated C-peptide level [14] (see 'Beta cell secretory reserve' below). The four subgroups are defined as follows:
●A+ß- autoantibodies present, beta cell function absent
●A+ß+ autoantibodies present, beta cell function present
●A-ß- autoantibodies absent, beta cell function absent
●A-ß+ autoantibodies absent, beta cell function present
A+ß- and A-ß- patients are immunologically and genetically distinct from each other but share clinical characteristics of type 1 diabetes with decreased beta cell function, and both subgroups would be termed type 1 diabetes (type 1 and 1b) in the current ADA classification system. A+ß+ and A-ß+ patients are immunologically and genetically distinct from each other but share clinical characteristics of type 2 diabetes with preserved beta cell functional reserve and would be termed type 2 diabetes in the ADA scheme.
A-ß+ patients comprise the largest KPD subgroup (approximately 50 percent) in multi-ethnic cohorts of KPD patients in the United States (table 1). They are also the patients who most commonly come to the notice of clinicians because they present with DKA yet have the clinical features and subsequent behavior of type 2 diabetes [1,14,15]. In the interest of defining and investigating novel syndromes of beta cell dysfunction, the broader terminology of "ketosis-prone diabetes" with its four subgroups subsumed under the Aß classification, rather than ketosis-prone type 2 diabetes, is more useful and does not presume to define a syndrome a priori.
PATHOPHYSIOLOGY OF KPD SYNDROMES
Autoantibodies present, beta cell function absent or present — There is a spectrum of beta cell destruction in patients with antibody-positive diabetes. Distinguishing A+ß- from A+ß+ ketosis-prone diabetes (KPD) permits investigators to explore different autoimmune pathways leading to clinically distinct patterns of beta cell loss, such as different latencies and variable degrees of beta cell destruction.
The later onset and more moderate clinical course (ability to discontinue insulin for over two years following the index episode of diabetic ketoacidosis [DKA] in 50 percent of the patients) of A+ß+ KPD compared with A+ß- KPD appears to be related in part to epitope-specific antibodies to the 65-kDa isoform of glutamic acid decarboxylase (GAD65). A specific amino-terminal epitope defined by monoclonal antibody DPD correlated with higher beta cell functional reserve and was associated with the milder A+ß+ phenotype [16]. The mechanisms that result in this autoantibody specificity and give rise to variable beta cell functional reserve remain to be elucidated.
In a majority of healthy individuals, GAD65 antibodies (GAD65Ab) are present in the sera but are masked by anti-idiotypic antibodies; in contrast, overtly GAD65Ab-positive patients with autoimmune type 1 diabetes lack these anti-idiotypic antibodies [17]. Masked GAD65Ab specific for the epitope DPD are strongly associated with preserved beta cell functional reserve among patients with KPD. Absence of GAD65Ab(DPD) reactivity is associated with two human leukocyte antigen (HLA) class II susceptibility haplotypes for autoimmune type 1 diabetes.
The presence of circulating insulin DNA serves as a biomarker for A+ß+ KPD patients, indicating a chronic, ongoing inflammatory process in the islets associated (at least for many months to years after the index DKA) with a significant degree of preserved beta cell function. Circulating insulin DNA is absent in A+ß- KPD patients [18].
Autoantibodies and beta cell function absent — A-ß- KPD is characterized by beta cell failure without evidence of detectable autoimmunity. Some A-ß- KPD patients may possess recently validated or newly described islet autoantibodies [19,20] or untested autoantibodies, such as SOX13 (SRY-related HMG box antigen 13) [21]. Alternatively, some A-ß- KPD patients may be misclassified as "A-" because of a decline in autoantibody titers over time. In one cohort, only 10 percent of patients who were A-ß- had new-onset diabetes when identified at presentation with DKA; the majority had insulin-dependent diabetes for many years previously [14].
However, a decline in antibody titer is less likely as GAD autoantibodies are reported to be quite durable [22-24]. In addition, extensive HLA typing reveals that the frequencies of major class II alleles associated with susceptibility to autoimmune type 1 diabetes are not significantly higher in A-ß- KPD patients than in ethnic-matched population controls, whereas they are significantly higher in A+ß- KPD patients, suggesting significant differences in the causes of absent beta cell function between the two populations [25].
The strong family history of diabetes among relatives of most A-ß- KPD patients suggests a familial trait and the possibility that genes required for beta cell development, regeneration, or function may be defective. Potentially significant variants in the genes TCF1, PAX-4, and PDX-1, encoding the key beta cell transcription factors hepatocyte nuclear factor-1-alpha (HNF1a), PAX-4, and pancreas-duodenum homeobox-1 (PDX-1), are enriched in A-ß- KPD patients compared with ethnicity-specific population controls; these may contribute to a monogenic etiology for some patients with the A-ß- phenotype [26].
While the "A-" subgroup is characterized by absence of humoral islet autoimmune markers, there could be a role for "occult" cellular islet autoimmunity. A significant proportion of A-β- KPD patients manifest strong T-cell responses to islet antigens, and higher percentages of circulating proinflammatory CD14+CD16+ monocytes [27].
Autoantibodies absent, beta cell function present — The A-ß+ phenotype is characterized by partially reversible beta cell dysfunction, which may be due to metabolic, genetic, or viral etiologies [28-30]. Increased oxidant stress in the islets may also contribute to A-ß+ KPD. One study of West African patients suggested that X-linked glucose-6-phosphate dehydrogenase (G6PD) deficiency contributes to depressed beta cell defense against oxidant stress in the face of acute hyperglycemia in KPD patients, but its cause does not appear to be a genetic mutation [31].
Short-term glucotoxicity and lipotoxicity do not appear to be critical factors in the development of beta cell decompensation in KPD patients of this subgroup [32,33].
Complex dysregulation of intermediary metabolism likely underlies the proclivity to develop DKA in A-ß+ KPD patients who do not have a clinically identifiable precipitating or provoking factor ("unprovoked" A-ß+ KPD). A plasma metabolomics survey indicated aberrant pathways of branch chain amino acid (BCAA) and arginine/citrulline metabolism in these patients [34]. Targeted kinetic measurements to investigate both pathways at the whole-body level have demonstrated distinctive pathogenic sequences:
●When clinically stable, unprovoked A-ß+ KPD patients with aberrant leucine metabolism have impaired ketone oxidation and fatty acid utilization for energy, leading to accelerated leucine catabolism and transamination of alpha-ketoglutarate to glutamate, with impaired tricarboxylic acid anaplerosis of glutamate carbon [34]. These findings highlight a novel process of defective energy production and ketosis in this form of KPD.
●When clinically stable, unprovoked A-ß+ KPD patients with aberrant arginine metabolism have increased intracellular arginine availability in the euglycemic state, indicating a higher requirement. This is compromised during hyperglycemia, with an inadequate supply of arginine to sustain insulin secretion during a hyperglycemic crisis. In a pilot study, exogenous arginine administration restored normal insulin secretory response in the face of sustained hyperglycemia [35].
●During an acute episode of DKA, A-ß+ KPD patients resemble patients with type 1 diabetes in having impaired BCAA catabolism and accelerated fatty acid flux to ketones, a reversal of their distinctive BCAA metabolic defect when stable. Thus, the natural history of A-β+ KPD appears to be marked by chronic but varying dysregulation of BCAA metabolism [36].
NATURAL HISTORY OF KPD SYNDROMES — The natural history of ketosis-prone diabetes (KPD) after the initial episode of diabetic ketoacidosis (DKA) depends upon the presence of autoantibodies and long-term beta cell reserve. Long-term beta cell reserve is the key determinant of long-term glycemic management and insulin dependence [37,38].
The natural history of KPD is best detailed in large cohorts with longitudinal follow-up [2,12,13,25,39]. One of the largest of these, the Houston cohort, includes 185 multi-ethnic adult patients admitted with DKA between 1999 and 2001 and followed for a mean of 5.5 years. The most frequent KPD subgroup was A-ß+ (54 percent), followed by A-ß-, A+ß-, and A+ß+ accounting for 20, 18, and 8 percent of patients, respectively [25].
●(A+ß-) and (A-ß-) – The A+ß- and A-ß- KPD patients displayed a typical course of complete insulin dependence and difficulty in attaining and achieving excellent long-term glycemic management. Although there was no difference in the mean age of the patients admitted with DKA during study recruitment, there were significant group differences in mean age at diabetes diagnosis and duration of diabetes (table 1). Compared with patients with the A-ß+ and A+ß+ phenotypes, patients with the A+ß- and A-ß- phenotypes were diagnosed at an earlier age (approximately 25 versus 40 years) and had a longer duration of diabetes (approximately nine versus two years).
●(A+ß+) – The majority of A+ß+ KPD patients had new-onset diabetes. Shortly after resolution of DKA, approximately 50 percent of patients had adequate beta cell functional reserve and could discontinue insulin; the others remain insulin dependent (table 1).
●(A-ß+) – Approximately 50 percent of A-ß+ KPD patients had new-onset diabetes and developed DKA without a clinically evident precipitating factor ("unprovoked" A-ß+ KPD), while the remainder had had longstanding diabetes prior to presentation with DKA and developed ketoacidosis in association with acute illness or noncompliance with antidiabetic treatment ("provoked" A-ß+ KPD). Unprovoked A-ß+ KPD patients displayed a striking male predominance (2.6:1, male-to-female ratio) that is quite distinct from provoked A-ß+ KPD patients (0.7:1); this sex imbalance has been noted also in patients with the unprovoked A-ß+ KPD phenotype in other cohorts [12,39,40].
Longitudinal data suggest other phenotypic differences between the unprovoked and provoked subgroups of A-ß+ KPD patients. Prospective assessment of 83 unprovoked and 64 provoked A-ß+ KPD patients revealed that despite equivalent degrees of hyperglycemia and beta cell functional reserve at initial testing following the index DKA episode, the two subgroups had different genetic characteristics, natural histories of beta cell function, and insulin requirements [41]. Unprovoked A-ß+ KPD was characterized by reversible beta cell dysfunction with male predominance and increased frequency of DQB1*0602 (resistance allele for autoimmune diabetes), whereas provoked A-ß+ KPD was characterized by progressive loss of beta cell reserve and increased frequency of the human leukocyte antigen (HLA) susceptibility alleles for autoimmune type 1 diabetes, DQB1*0302 and DRB1*04. In this prospective assessment, unprovoked DKA predicted long-term beta cell functional reserve, insulin independence, and glycemic management in KPD patients.
Longitudinal characterization of KPD patients presenting with concomitant acute pancreatitis and DKA reveals that despite greater clinical severity at presentation, KPD patients with acute pancreatitis have better preserved beta cell function than those without acute pancreatitis. β+ KPD patients presenting with acute pancreatitis have worse long-term glycemic management than those with other causes of provoked DKA. Factors other than beta cell function negatively impact glycemic management in KPD patients presenting with acute pancreatitis [42].
In a smaller cohort study of patients with new-onset, unprovoked DKA, beta cell functional reserve was preserved in a greater proportion of individuals with obesity than in those who were lean [13,39].
In another cohort study with 10-year follow-up, KPD patients with a probable unprovoked A-ß+ phenotype initially achieved insulin independence and good glycemic management with oral agents [12]. At the end of the follow-up period, 40 percent were still insulin independent. In those that required insulin, the mean duration until relapse to insulin dependence was 40 months. Some of these patients experienced relapsing and remitting ketosis.
Longer-term follow-up of a cohort of unprovoked A-β+ KPD patients indicates that abrupt relapse to ketosis can occur after varying periods of remission. Early relapse is associated with younger age, lack of robust increase (usually severalfold) of beta cell functional reserve, and suboptimal glycemic management after one year. Among unprovoked A-β+ KPD patients, measurements of these objective parameters one year after the index DKA episode can be used reliably to justify clinical decisions and provide reassurance, if beta cell function has at least doubled over baseline, for continued insulin independence [43]. A report of patients in Thailand with a phenotype corresponding to unprovoked A-β+ KPD suggests that those who maintain insulin independence may also be quite insulin sensitive, with low circulating leptin and high circulating adiponectin levels [44].
MANAGEMENT OF KPD — Clinical management of ketosis-prone diabetes (KPD) includes:
●Acute management of diabetic ketoacidosis (DKA)
●Outpatient management shortly after resolution of DKA, including classification of the patient according to KPD subgroup and evaluation of predictive factors
●Long-term management
Initial treatment is the same as for patients with type 1 diabetes and ketoacidosis. However, the long-term outcomes and insulin requirements of these patients are variable. In the Houston cohort described above, insulin was successfully discontinued after 12 months of treatment in 50 and 44 percent of patients in the A+ß+ and A-ß+ subgroups, respectively, whereas all patients in the ß- subgroups remained insulin dependent (table 1). Thus, all ß- patients and some ß+ patients will require long-term insulin therapy to prevent recurrent ketoacidosis.
As these patients are heterogeneous and the "type" of diabetes is unclear at presentation, they should all be maintained on insulin initially. In addition, in some patients, ketoacidosis may result from transient suppression of beta cell function due to an increased sensitivity to glucose toxicity [1]. In these patients, insulin therapy is required until recovery from the effects of glucose toxicity.
Acute management of DKA — All patients who present with diabetic ketoacidosis (DKA) should be treated according to established principles of acute management of the metabolic decompensation. Standard inpatient hospital protocols requiring aggressive fluid replacement, continuous insulin therapy, assessment for and treatment of precipitating factors, monitoring for resolution of hyperglycemia, ketoacidosis and electrolyte disorders, and transition from intravenous insulin to subcutaneous insulin regimens have been well-described [1]. (See "Diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults: Treatment".)
Inpatient treatment during the episode of DKA should be the same regardless of the apparent phenotype of the KPD patient, and all KPD patients should be discharged from the hospital on a regimen that provides 24-hour insulin coverage. Any attempt to withdraw insulin treatment should be based upon precise classification of the KPD subgroup and assessment of predictive factors, which should be performed at the first outpatient visit one to three weeks following hospital discharge.
Evaluation in the first 2 to 10 weeks following resolution of DKA — Assessment of beta cell secretory reserve and beta cell autoimmunity can be performed one to three weeks after resolution of ketoacidosis to minimize the acute effects of glucose toxicity or desensitization on beta cell function. Beta cell secretory reserve (as measured by fasting plasma C-peptide, C-peptide response to glucagon stimulation, and C-peptide-to-glucose ratio) following DKA resolution is the strongest predictor of long-term glycemic management and insulin dependence [37,38,45].
Beta cell secretory reserve — Beta cell function can be assessed by measuring fasting or glucagon-stimulated C-peptide concentrations. These tests are performed after an overnight fast and at least 10 hours after the last dose of short-acting or intermediate-acting insulin, metformin, or thiazolidinedione, and at least 24 hours after a dose of sulfonylurea, glucagon-like peptide 1 (GLP-1) receptor agonist, or long-acting insulin (glargine, detemir) [14]. The fasting C-peptide measurement is less expensive and easier to perform.
Patients are classified as "ß-" if the fasting serum C-peptide concentration is less than 1 ng/mL (0.33 nmol/L) and the peak serum C-peptide response to glucagon (measured at 5 and 10 minutes after intravenous injection of 1 mg glucagon) is less than 1.5 ng/mL (0.5 nmol/L) [14]. They are classified as "ß+" if the fasting serum C-peptide concentration is at least 1 ng/mL (0.33 nmol/L) or the peak serum C-peptide response to glucagon is at least 1.5 ng/mL (0.5 nmol/L). Although the cutpoints noted above do not independently predict the potential for successful and safe withdrawal of insulin, a high ratio (>11) of fasting C-peptide (in nmol/L) to glucose (in mmol/L) at six months predicts such a course among ß+ patients [38].
These cutpoints (established using the Linco radioimmunoassay [RIA] for human C-peptide) have been shown to accurately predict beta cell function and glycemic management after one year [10,37]; however, they have not as yet been standardized across other C-peptide assays.
Beta cell autoimmunity — Although beta cell functional reserve provides sufficient information to predict the patient's clinical course, quantitative assessment of beta cell autoantibodies is also clinically useful, specifically for patients with the A+ß+ KPD phenotype. Patients of this subgroup follow one of two divergent clinical courses within the first two years of diagnosis; approximately 50 percent maintain stable beta cell function and remain insulin independent, while the others convert to ß- status and become insulin dependent. Hence, identification of patients as A+ß+ rather than simply ß+ alerts the clinician to follow them closely. Furthermore, these are the patients in whom human leukocyte antigen (HLA) genotyping provides additional prognostic markers of clinical behavior as the presence of specific autoimmune type 1 diabetes susceptibility alleles such as HLA DQB1*02 are associated with higher risk of progressive beta cell functional loss. (See "Pathogenesis of type 1 diabetes mellitus", section on 'Genetic susceptibility'.)
Glutamic acid decarboxylase (GAD65) and islet tyrosine phosphatase 2 (IA-2) autoantibody titers can be measured in the patients' sera by highly sensitive and specific assays, with the proviso that upper limits of the normal range for the autoantibody levels may differ between regions and ethnic groups [14]. Measuring serum titers of autoantibodies to the zinc transporter 8 (ZnT8) antigen increases the accuracy of classifying patients as "A-" or "A+". Patients may be classified as "A+" if the autoantibody index for any one of these autoantibodies exceeds the ethnic-specific 99th percentile or "A-" if the indices for all are below the 99th percentile.
Clinical practice — Long-term management of KPD can be guided rationally by accurate classification based upon assessment of beta cell functional reserve, beta cell autoantibodies, and in some instances, HLA typing. Although assessment of these parameters in all patients presenting with DKA is ideal, cost constraints and assay availability may be prohibitive in some regions.
For patients presenting with DKA and a typical type 1 diabetes clinical phenotype (young onset, lean), it is probably unnecessary to check C-peptide or antibodies. These patients should be continued on insulin indefinitely.
However, for anyone presenting in an atypical fashion (older onset, overweight), C-peptide and antibodies (GAD65, IA-2) should be measured, with referral to a specialist for detailed assessment (genetic or immunologic), classification of KPD type, and for greater confidence in selecting patients to attempt insulin withdrawal. In regions where such testing is unavailable, future likelihood of insulin independence in atypically presenting patients can be assessed by reducing the insulin dose and monitoring serum glucose and serum or urine ketones. The safety of the latter approach depends upon a patient's ability to self-monitor and the attentiveness of the provider. (See 'Assessing for insulin independence' below.)
Glycemic management in the first 2 to 10 weeks following discharge
Continue insulin — Because the "type" of diabetes is unclear at presentation, all patients should be maintained on insulin initially. Insulin should never be discontinued in patients who are classified as "ß-," regardless of whether they are A+ or A-, as these patients invariably require insulin to avoid ketosis. Evidence from the Houston cohort shows that no patient initially classified as "ß-" by the above protocol has recovered beta cell function sufficiently to warrant a trial of insulin withdrawal. Some ß+ patients will also require long-term insulin therapy to prevent recurrent ketoacidosis.
Assessing for insulin independence — Future likelihood of insulin independence in patients who are initially categorized as "ß+" may be assessed as follows:
●Patients are placed on long- or intermediate-acting insulin (with or without short acting pre-meal insulin) at the time of hospital discharge, with the dose determined by the mean daily insulin requirement during the previous two hospital days.
•Patients with known diabetes may be given insulin at the dose they were receiving before the onset of DKA.
•In previously insulin-naive patients, another approach is to administer multidose insulin at a dose of 0.6 to 0.8 units/kg per day, with 50 percent as regular or rapid-acting insulin (in divided doses before meals) and 50 percent as basal insulin [46,47].
The dose can be titrated with frequent glucose monitoring until an optimal dose is established. The first clinic visit is within two to four weeks of discharge from the hospital. (See "Management of blood glucose in adults with type 1 diabetes mellitus", section on 'Designing an MDI insulin regimen'.)
●If capillary blood glucose values before each meal and at bedtime during a two-week period attain American Diabetes Association (ADA) goals for fasting/preprandial (80 to 130 mg/dL [4.4 to 7.2 mmol/L]) and bedtime/peak postprandial plasma glucose levels (180 mg/dL [10.0 mmol/L]), the insulin dose is reduced by 50 percent and the patient reassessed in the clinic one week later. Patients are advised to use urine ketone strips or a blood ketone testing meter to check for significant ketosis if the blood glucose level rises above 200 mg/dL.
●If, upon decreasing the insulin dose, blood glucose values increase with development of ketosis, the insulin regimen is intensified and no further attempts are made to discontinue insulin. (See "Management of blood glucose in adults with type 1 diabetes mellitus".)
●If mean blood glucose values remain at goal for two consecutive clinic visits, insulin is discontinued and the patient is monitored closely. The duration of this process of insulin withdrawal is variable and may range from 10 to 14 weeks to longer.
●If blood glucose values remain at goal, the patient is instructed to continue lifestyle modification and monitored without pharmacologic therapy.
●If blood glucose values increase without development of ketosis, the patient is placed on oral hypoglycemic agents.
•In A-ß+ patients, insulin-sensitizing agents such as metformin are recommended as these patients have the highest frequency of the metabolic syndrome among KPD groups [48]. (See "Initial management of hyperglycemia in adults with type 2 diabetes mellitus".)
In a small trial of metformin, sitagliptin, or placebo in 48 African American patients with obesity and new-onset DKA (majority A-ß+ KPD) who had achieved near-normoglycemia remission, metformin and sitagliptin prolonged remission compared with placebo [49]. The prolongation of remission was due to improvement in beta cell function. The trial was limited by a significant drop-out rate (11 patients lost to follow-up, an additional 2 withdrew from the study).
•If blood glucose levels do not achieve therapeutic targets within eight weeks, addition of a sulfonylurea, thiazolidinedione, meglitinide, dipeptidyl peptidase 4 (DPP-4) inhibitor, GLP-1 receptor agonist, or alpha-glucosidase inhibitor should be considered. Sodium-glucose co-transporter 2 (SGLT2) inhibitor drugs are not recommended for KPD patients, due to their propensity for provoking ketoacidosis. (See "Management of persistent hyperglycemia in type 2 diabetes mellitus".)
Long-term management
●General management – Once the patient has been classified for KPD type, assessed for predictive factors, and begun on an appropriate treatment course, standards of diabetes management should be followed. Aggressive management of the metabolic syndrome (including weight loss and exercise) and cardiovascular risks are important in all subgroups of KPD patients. (See "Metabolic syndrome (insulin resistance syndrome or syndrome X)", section on 'Lifestyle modification'.)
Patients should be counseled by a registered dietitian and diabetic educator periodically. Physical activity for at least 150 minutes per week should be recommended, and weight loss is advised for all patients with obesity. Smoking cessation should be reinforced. Screening and treatment for microvascular and macrovascular complications of diabetes are advised. (See "Overview of general medical care in nonpregnant adults with diabetes mellitus".)
●Predictors of future beta cell function – As described earlier, A-ß+ KPD patients may be divided into a new-onset, "unprovoked" group (presenting with DKA in the absence of a significant stressful precipitant) and a previously diagnosed, "provoked" group (DKA associated with a significant stressful event) (see 'Autoantibodies absent, beta cell function present' above). Patients in the former subgroup have a significantly greater rate of insulin discontinuation and better long-term glycemic management than the latter.
New-onset diabetes, older age at onset of diabetes, and high levels of beta cell functional reserve (fasting C-peptide-to-glucose ratio >11) may be used as reliable predictors of insulin discontinuation in ß+ patients [38,45,50,51]. In a multivariate model, new-onset diabetes and beta cell functional reserve remained predictive.
The presence of beta cell autoantibodies is a determinant of future beta cell function. In analyses that do not differentiate the four Aß subgroups, KPD patients with autoantibodies tend to have lower beta cell function both shortly after the correction of the acidosis and on long-term follow-up [12,52]. However, this is not an infallible criterion, as approximately 50 percent of A+ß+ KPD patients maintain long-term beta cell functional reserve. Most A+ß+ KPD patients are able to come off insulin therapy initially, but they require close monitoring for at least two years since the evolution of their beta cell function is the least predictable of the KPD groups. HLA typing may play a useful role in the management of this group of KPD patients as it may help to identify those who are likely to experience a more aggressive course or may be candidates for future immunomodulatory therapy.
SUMMARY AND RECOMMENDATIONS
●General principles – Ketosis-prone diabetes (KPD) is a heterogeneous syndrome characterized by the presence of diabetic ketoacidosis (DKA) in patients who may lack the typical clinical phenotype of autoimmune type 1 diabetes. Recognition of KPD coincides with the emergence of the concept that early beta cell dysfunction is likely to be a primary defect in the pathophysiology of diabetes, regardless of "type." Syndromes of KPD are increasingly recognized worldwide, especially among urban, multi-ethnic populations. They present challenges to both clinicians and researchers but also offer the prospect of revealing novel mechanisms of beta cell dysfunction relevant to common forms of diabetes. (See 'Introduction' above.)
●Classification – There are four different classification schemes for KPD. The Aß classification distinguishes four KPD subtypes based upon the presence or absence of autoantibodies and beta cell functional reserve. This classification most accurately predicts long-term insulin dependence 12 months after DKA presentation. (See 'Classification of KPD' above.)
●Evaluation and management
•DKA should be treated according to established principles. (See "Diabetic ketoacidosis and hyperosmolar hyperglycemic state in adults: Treatment".)
•When patients are discharged from the hospital after resolution of DKA, we recommend initial treatment with insulin, rather than oral agents, regardless of the apparent phenotype of the KPD patient (Grade 1B). (See 'Management of KPD' above.)
•Assessment of beta cell reserve and beta cell autoimmunity after resolution of DKA helps predict clinical course and long-term treatment. This assessment is typically performed one to three weeks after resolution of ketoacidosis. (See 'Evaluation in the first 2 to 10 weeks following resolution of DKA' above.)
●Natural history – The natural history of KPD after the initial episode of DKA depends upon the presence of autoantibodies and long-term beta cell reserve (table 1).
•(ß-) – Patients with poor beta cell function (ß-) after resolution of the index DKA event typically require long-term exogenous insulin therapy, regardless of autoantibody status. (See 'Glycemic management in the first 2 to 10 weeks following discharge' above.)
•(A-ß+) – Patients with beta cell secretory reserve who are antibody negative (A-ß+) are often able to discontinue insulin, especially if they had unprovoked DKA as the initial manifestation of diabetes. The duration of the process of insulin withdrawal is variable and may range from 10 to 14 weeks to longer. (See 'Assessing for insulin independence' above.)
If after discontinuation of insulin blood glucose values increase without development of ketosis, treatment with oral or injectable agents to lower blood glucose is required.
If the patient develops ketosis upon decreasing the insulin dose, insulin should be intensified. In this setting, we suggest not attempting to withdraw insulin a second time (Grade 2C). (See 'Assessing for insulin independence' above.)
•(A+ß+) – Patients with preserved beta cell function who have autoantibodies (A+ß+) have a variable course with some demonstrating progressive beta cell deterioration and others long-term preservation. This group of individuals requires more careful monitoring, and these patients may benefit from human leukocyte antigen (HLA) genotyping to provide additional prognostic markers of clinical behavior. (See 'Beta cell autoimmunity' above.)
