Atypical diabetes in children: ketosis-prone type 2 diabetes

Abstract

Ketosis-prone type 2 diabetes mellitus also known as atypical or flatbush diabetes is being increasingly recognised worldwide. These patients are typically obese, middle-aged men with a strong family history of type 2 diabetes. The aetiology and pathophysiological mechanism is still unclear but some initial research suggests that patients with ketosis-prone type 2 diabetes have a unique predisposition to glucose desensitisation. These patients have negative autoantibodies typically associated with type 1 diabetes but have shown to have human leucocyte antigen (HLA) positivity. At initial presentation, there is an impairment of both insulin secretion and action. β Cell function and insulin sensitivity can be markedly improved by initiating aggressive diabetes management to allow for discontinuation of insulin therapy within a few months of treatment. These patients can be maintained on oral hypoglycaemic agents and insulin therapy can be safely discontinued after few months depending on their β cell function.

Background

It is assumed that children presenting with diabetic ketoacidosis (DKA) have type 1 diabetes and will require lifelong insulin treatment, but this is not always the case. Atypical diabetes in children or Flatbush diabetes is an important clinical entity that is being increasingly recognised worldwide. It is important for anyone dealing with childhood diabetes to understand that the natural history of Flatbush or atypical diabetes mellitus (DM) is distinct from either type 1 DM or type 2 DM and awareness of this entity can facilitate early diagnosis and appropriate management.

The following case report will analyse and evaluate the diagnostic dilemma and management of atypical diabetes or ketosis-prone diabetes (KPD) in a child.

Case presentation

An 11-year-old Afro-Caribbean boy presented acutely to the children's assessment unit in January 2011 with a 3-week history of secondary enuresis, abdominal pain, polyuria and polydypsia. His medical history was unremarkable. His initial laboratory blood glucose reading was 16.8 mmol/l and his urine dip showed 3+ glucose and 3+ ketones. The family history of this child is notable. He has a stepbrother on his mother's side who was apparently diagnosed with diabetes in early childhood in Jamaica and was started on oral hypoglycaemic agents for a few months following which he was deemed ‘free of diabetes’. His stepbrother is currently well and not on any treatment. His three maternal aunts and a maternal uncle have type 2 diabetes. His maternal grandfather also had type 2 diabetes and died of renal failure. His father, paternal grandfather, paternal grandmother and paternal uncle also have type 2 diabetes. All the family members apart from his stepbrother were diagnosed with type 2 diabetes in adulthood.

His physical examination showed Acanthosis Nigricans and the rest of his examination was unremarkable.

During his initial presentation, he had a pH of 7.32, pCO₂ of 5.4 kPa, bicarbonate of 22 mEq/l and a base excess of −3.3. His laboratory lactate was 2.1 mmol/l and repeat laboratory blood glucose was 19.2 mmol/l. His glycosylated haemoglobin (HbA1c) at presentation was 84 mmol/mol (9.8%). There was no evidence of acute infection, renal or liver dysfunction. He was diagnosed with probable type 1 diabetes and started on a basal bolus regime that included 12 units of glargine at night and 4 units of novorapid three times a day with meals (total daily dose of insulin 0.5 units/kg/day).

Differential diagnosis

• Type 1 diabetes mellitus

• Ketosis-prone type 2 diabetes

• Maturity onset diabetes of the young

Outcome and follow-up

Over the next few weeks following discharge, he experienced a number of symptomatic hypoglycaemic episodes. His insulin dose was progressively reduced and he eventually came off insulin completely. During this time, his HbA1c dropped to 66 mmol/mol (8.2%), in spite of being on no insulin treatment for nearly 1 month. On several occasions, he became hypoglycaemic while being off insulin, with blood glucose readings as low as 2.2 mmol/l. For the majority of the time, however, particularly in the evening, his blood glucose readings were in the early teens. This did not appear to be related to his activity levels or diet. His continuous glucose monitoring revealed a similar trend of high blood-glucose readings during the evening with varied fluctuations during the day. His autoantibodies to glutamic acid decarboxylase (GAD) and islet cell antibodies (IA-2) were negative.

At 4 months from diagnosis, his HbA1c had increased slightly from 66 mmol/mol (8.2%) to 69 mmol/mol (8.5%) and he continued to be off-insulin therapy for nearly 8 weeks. His blood glucose meter confirmed a number of high blood glucose tests that were spread through out the day, but most strikingly, after school and in the evening. He continued to experience episodes of hypoglycaemia late at night and in the early hours of morning with no correlation with his diet and activity levels. At 4 months from initial presentation, his C-peptide level was 994 pmol/l or 0.994 nmol/l (normal 364–1655 pmol/l) and insulin level was 231 pmol/l (normal 14–145 pmol/l) that suggested he had adequate β-cell function. The high insulin level would suggest hyperinsulinaemia similar to type 2 diabetes in childhood. His repeat GAD and IA-2 at 4 months from initial diagnosis was again negative.

At 6 months, his HbA1c had increased to 95 mmol/mol (10.8%) and his blood glucose readings were persistently high between 13–25 mmol/l. He had been restarted on a variable amount of glargine at home but was not on any regular novorapid injections. It was decided to start on oral metformin 500 mg once a day and this was subsequently increased to twice a day over the next few weeks.

At 9 months, his blood glucose trends had improved but he still continued to have low blood glucose levels intermittently. On further review at 1 year from initial diagnosis, his HbA1c had dramatically improved to 61 mmol/mol (7.7%) and his day-to-day blood glucose readings were predominantly within the normal range.

He re-presented in March 2012 (14 months from initial diagnosis) with lethargy, tiredness, polyuria and polydypsia. He had returned from a trip to America and had not been feeling well there. He had persistently high blood glucose levels and admitted to missing his metformin in the preceding few weeks. His blood gas revealed a pH of 7.22, CO₂ 4.2 pKa, base excess of −15 and bicarbonate of 11 mEq/l. His laboratory blood glucose was 42 mmol/l and his urine had 4+ ketones and 4+ glucose. He appeared moderately dehydrated and was tachycardic. A diagnosis of DKA was made and he was treated according to the British Society of Paediatric Endocrinology and Diabetes (BSPED) recommended DKA protocol (2009) with intravenous insulin and fluids. His metabolic derangements normalised in 2 days and he was started on a low-dose basal bolus regime and continued on metformin.

Conclusion

This report summarises the progress of an 11-year old Afro-Caribbean child who presented with hyperglycaemia and ketoacidosis. He had a strong family history of type 2 diabetes and no evidence of β-cell autoimmunity on serological testing. He had clinical signs of insulin resistance and was started on a basal bolus regime, but this resulted in unpredictable hypoglycaemia and his insulin doses were gradually reduced and eventually discontinued 6 weeks after diagnosis. There was a transient improvement in his HbA1c when he was not on any insulin therapy. He was commenced on metformin as his HbA1c deteoriated. Repeat investigations showed good β-cell reserve and persistent negative islet cell antibodies. His HbA1c improved dramatically following metformin monotherapy. He represented with DKA at 14 months from initial diagnosis, probably secondary to poor compliance with metformin therapy, and was restarted on low-dose basal bolus therapy along with metformin.

Discussion

Introduction

KPD is a widespread, emerging, heterogeneous syndrome characterised by patients who present with DKA or unprovoked ketosis with hyperglycaemia but do not necessarily have the typical phenotype of autoimmune type 1 diabetes.^{1 2} Atypical diabetes was originally described by Banerji et al³ as a unique form of diabetes among African-American patients who presented with DKA as their initial manifestation of diabetes. However, in contrast to type 1 diabetes, patients with atypical diabetes undergo spontaneous remission and maintain long-term insulin independence. At presentation, they have impairment of both insulin secretion and insulin action, but intensified diabetes management results in significant improvement in β-cell function and insulin sensitivity to allow discontinuation of insulin therapy within few months of treatment.^{4 5} Because of mixed features of type 1 diabetes and type 2 diabetes, this variant has been given several names including diabetes mellitus type 1b, idiopathic type1 diabetes, Flatbush diabetes mellitus, type 1.5 diabetes mellitus and ketosis-prone type 2 diabetes mellitus.

History of KPD

Reports of African or African-American patients whose clinical features were intermediate between those of type 1 and type 2 diabetes have appeared since the late 1960s. Adadevoh⁶ and Dodu⁷ reported that some adult patients with DKA displayed only transient insulin dependence and required revision of their type of diabetes over time. This atypical presentation has also been reported in other ethnicities including the Japanese, Chinese, Native Americans, Hispanic and white populations.⁸ This unique, transient, insulin-requiring profile was mainly recognised in newly diagnosed diabetics and was reported as ‘temporary DM in adult Nigerians’ and it was difficult to classify such patients as having type 1 or 2 diabetes. Winter et al¹ in 1987 described a cohort of African-American children with an atypical form of diabetes. These patients were noted to have an acute presentation, an autosomal dominant pattern of inheritance and negative islet cell antibodies. In 1994, Banerji et al³ described a case series of adult overweight Afro-Caribbean patients who presented with DKA. The term ‘Flatbush diabetes’ entered the literature at this point in recognition of the region in New York, where most of these patients resided.

Umpiereez et al⁴ reported obese African-American patients in Atlanta, who had late-onset diabetes presenting with DKA. The presence of measurable pancreatic insulin reserve, absence of autoimmune indicators of β-cell destruction and increased frequency of HLA-DR3 and HLA-DR4 was also noted. The concept of the body mass index (BMI) as a means to distinguish two phenotypes (obese and lean) of patients presenting with DKA, based on their immunological and β-cell functional differences was introduced.⁹

Prevalence

The exact incidence and prevalence of ketosis-prone type 2 diabetes (KPD) mellitus is not well known. Based on reports, it is clearly evident that the incidence of this type of diabetes is underestimated and under-reported. In the USA, the prevalence of KPD has been estimated to be between 20% and 50% in African-American and Hispanic patients presenting with DKA¹⁰ and half of these patients had Flatbush diabetes mellitus. African studies have also reported a similar incidence.^{11 12} Asian and white populations show a lower prevalence and may represent fewer than 10% of cases presenting with DKA.^{13 14}

Classification of KPD

Attempts to differentiate patients with KPD into clinically distinct and relevant subgroups have resulted in four different classification schemes: the American Diabetes Association (ADA) classification, a BMI-based system, a modified ADA classification, and the Aβ system. The ADA classification scheme lacks flexibility to accommodate heterogeneity of KPD, because it applies only to patients with type 1 diabetes. To address this discrepancy, Balasubramanyam et al¹⁵ suggested a classification system based on autoimmunity (A) and β-cell function (β) for KPD patients. This is discussed below:

KPD type 1A (type 1A diabetes mellitus)

A+ β−: those with autoantibodies and absent β-cell function, they require lifelong exogenous insulin therapy.

KPD type 1B (type 1B diabetes mellitus)

A− β−: those without antibodies and permanent and complete β-cell failure; they require lifelong exogenous insulin therapy.

KPD type 2A (latent autoimmune diabetes in adults)

A+ β+: those with autoantibodies but preserved β-cell functional reserve; in long run, these patients lose their β-cell preserve and require lifelong exogenous insulin therapy.

KPD type 2B (flatbush DM)

A− β+: those without antibodies but preserved β-cell functional reserve. The majority (especially new onset) can discontinue exogenous insulin therapy and can be managed with oral hypoglycaemic agents long term.

A+ β− and A− β− patients are immunologically and genetically distinct from each other but share clinical characteristics of type 1 diabetes with very low β-cell function, whereas A+ β+ and A− β+ patients are immunologically and genetically distinct from each other but share clinical characteristics of type 2 diabetes with preserved β-cell functional reserve.

Pathophysiology of flatbush diabetes (metabolic and genetic basis)

Umpierrez et al¹⁶ studied obese African-American patients with the phenotype of A− β+ KPD after resolution of the initial episode of DKA and examined the roles of glucotoxicity and lipotoxicity in causing a severe but partially reversible β-cell functional defect. It was demonstrated that acute hyperglycaemia but not acute hyperlipidaemia caused severe blunting of the C-peptide response to glucose stimulation, and chronic hyperglycaemia was associated with reduced expression and insulin-stimulated threonine-308 phosphorylation of Akt2 in skeletal muscle. This study revealed that severe glucotoxic blunting of an intracellular pathway which leads to insulin secretion may contribute to the reversible β-cell dysfunction, that is, characteristic of A− β+ KPD patients, and that hyperglycaemia may be exacerbated by defects in skeletal muscle glucose uptake as a result of glucotoxic downregulation of skeletal muscle insulin signalling. One possible mechanism of glucotoxic β-cell dysfunction is increased oxidant stress in the islet cells of the pancreas.

There is clearly a genetic susceptibility to KPD, but it is unclear whether the model is polygenic or has a major gene influence. Some researchers have found an increased frequency of HLA-DR3 and HLA-DR4 when compared with non-diabetic population.³ Other researchers have failed to find an association with HLA susceptibility alleles.¹⁷ Sobngwi et al,¹¹ in 2002, investigated the possibility of x-linked glucose-6-phosphate-dehydrogenase (G6PD) deficiency as a genetic basis for the male predominant Flatbush diabetes phenotype in West-African patients. The prevalence of functional G6PD deficiency was found to be higher in the KPD patients as compared with type 2 diabeties along with a relationship between β-cell functional reserve and erythrocyte G6PD activity.

Mauvais-Jarvis et al⁵ found a high frequency of polymorphism leading to an amino acid substitution (R133W) in PAX4, a transcription factor essential for islet β-cell morphogenesis, among patients with phenotypes A− β+KPD. The pathophysiological significance of this variant is unclear in KPD because it is found in higher percentage of West Africans and African-Americans with and without type 2 diabetes, but not in Caucasians.

Clinical presentation and natural course of disease

The phenotypic profile of patients with Flatbush diabetes is markedly different from typical DKA patients with type 1 DM. Most patients with KPD are obese, middle-aged with newly diagnosed diabetes and present with unprovoked DKA. Their initial presentation is usually acute and they have a history of polyuria, polydypsia and weight loss for less than 4–6 weeks.¹³

The mean age at diagnosis is 40 years (range between 33 years and 53 years). Several series of patients with ketosis-prone type 2 diabetes show a twofold-to-threefold higher prevalence in men.¹⁸ This is in contrast to a series of white patients with type 1 diabetes, which report that women are more likely than men to develop DKA. The male predominance in ketosis-prone type 2 diabetes seems to be independent of the degree of obesity and age at presentation. The reason for the gender difference is unclear; nevertheless, it has been ascribed to hormonal factors, body-fat distribution and changes in insulin sensitivity. Blood glucose level and acid–base parameters at presentation are similar to those reported in patients with type 1 DM and DKA.

After the initial episode of DKA, the disease course of ketosis-prone type 2 diabetes depends on the presence of autoantibodies and long-term β-cell reserve. The key determinant of long-term glycaemic control and insulin dependence is the absence of autoimmunity and presence of β-cell reserve. McFarlane et al¹⁹ followed the clinical course of African-American patients admitted to hospital with newly diagnosed ketoacidosis for at least 1 year. Remission was defined as an HbA1c of 45 mmol/mol or less (6.3% or less) and fasting plasma glucose of less than 6.6 mmol/l 3 months after therapy when all pharmacological agents were discontinued. He showed that 42% of patients achieved remission after a mean of 83 days and remained in remission during 20 months of follow-up. It was demonstrated that there was no statistical differences in age, gender, plasma-glucose level at presentation, changes in BMI, magnitude of weight change or pharmacological agents used between patients who achieved remission and those who did not.

Umpierrez et al⁴ showed that two-third of patients with KPD type 2 diabetes relapsed into hyperglycaemia within 2 years if treated with diet alone. It was further shown that treatment with sulphonylurea or metformin in this group was effective in prolonging the duration of normoglycaemic remission and in preventing ketoacidosis.^{20 21}

Management of ketosis-prone type 2 DM

Clinical management of ketosis-prone type 2 diabetes can be broadly divided into

• Acute management of hyperglycaemia and ketosis/DKA

• Long-term management after resolution of DKA

Acute management of DKA

All patients who present with DKA should be treated according to established principles of acute management of metabolic decompensation irrespective of their diabetes phenotype.

Standard inpatient hospital protocol requires fluid replacement, correction of hyperglycaemia and acidosis, and management of electrolyte imbalance. All patients should be discharged on a regimen that provides 24 h insulin cover.

Long-term management after resolution of DKA

Generally 4–8 weeks after resolution of ketoacidosis assessment should be performed to measure β-cell functional reserve and autoimmunity. This time interval is needed to minimise any acute effects of glucose toxicity or desensitisation of β-cell function. Assessment of pancreatic β-cell function involves fasting blood glucose, C-peptide levels and C-peptide response to glucagon. Patients are classified as β− (absent β-cell function), if the fasting serum C-peptide concentration is less than 0.33 nmol/l (1 ng/ml) and the peak serum C-peptide response to glucagon (measured at 5 and 10 min after intravenous injection of 1 mg of glucagon) is less than 0.5 nmol/l (1.5 ng/ml). They are classified as β+ (adequate β cell function), if the fasting serum C-peptide concentration is at least 0.33 nmol/l (1 ng/ml), or the peak serum C-peptide response to glucagon is at least 0.5 nmol/l (1.5 ng/ml). These cut-offs accurately predict β cell function after 6 months and 1 year.¹⁵ The long-term dependence of patients on insulin is predicted by their autoimmune status, hence autoimmunity against β cells should also be assessed. Positive autoimmunity indicates long-term dependence on insulin.

Categorisation of patients should be carried out based on the above results (A β classification) and only patients who are A− β+ (KPD type 2B, Flatbush diabetes DM) can be cautiously tried on oral hypoglycaemic agents under careful supervision. It is not advisable to stop insulin with other KPD categories. After discontinuation of insulin therapy in A− β+, it is recommended that oral therapy with metformin or low-dose sulphonylurea is started. The duration of insulin withdrawal and initiation of oral hypoglycaemics is variable and may range from 10 weeks to 14 weeks or longer. Patients with positive autoimmunity or with inadequate insulin secretion are more likely to relapse and should be continued on insulin therapy with careful monitoring for recurrence of hyperglycaemia or ketosis.

Learning points.

• It is often assumed that children presenting with diabetic ketoacidosis (DKA) have type 1 diabetes and will require lifelong insulin treatment. However, this is not always the case.

• The early management of atypical diabetes is similar to that of type 1 diabetes.

• Rapidly falling requirements of insulin over the first few weeks in patients with diagnosis of type 1 diabetes should alert the clinician to the possibility of atypical or Flatbush diabetes, and their pancreatic autoantibodies and function should be assessed to predict insulin dependence.

• The natural history of Flatbush or atypical diabetes mellitus (DM) is distinct from either type 1 or type 2 DM and awareness of this entity can facilitate early diagnosis and appropriate management.

• Patients presenting in DKA with atypical diabetes will have spontaneous resolution of their diabetes within a few months; most will relapse within 2 years of diagnosis and will require insulin and or an oral hypoglycaemic agent.