Changing picture of diabetes classification: a case report

The use of immunological and genetic investigations can help aid diagnosis (Ramachandra et al, 2009). The presence of beta-cell specific auto-antibodies indicates that an autoimmune process behind the beta-cell depletion is likely and usually suggests T1D. However this is not absolute. T1D may present with negative auto-antibodies at diagnosis. The absence of antibodies therefore does not rule out T1D if the clinical picture is classical. Maahs et al (2010) recommended subclassifying T1D into type 1A (typical, autoimmune-mediated) and type 1B (idiopathic, more commonly seen in specific ethnic groups and less frequently in Caucasian patients). The absence of auto-antibodies suggests an alternative form of diabetes, such as T2D or maturity-onset diabetes of the young (MODY), should be considered (Craig et al, 2014), see Table 1. It should be noted, however, that patients with features typical of T2D may be positive for diabetes auto-antibodies: 15–40% of paediatric patients diagnosed with T2D are autoantibody-positive, including those not requiring insulin for over 1 year (Reinehr, 2013). Non-insulin dependent adults positive for auto-antibodies have been referred to as having latent autoimmune diabetes of adults (LADA). For a diagnosis of LADA, patients must be positive for one or more of the four antibodies commonly found in T1D, be aged at least 30 years and not be treated with insulin within the first 6 months after diagnosis (Ramachandra et al, 2009). The presence of auto-antibodies may be an indication of risk that the patient will more rapidly develop a need for insulin (Zeitler et al, 2014; International Society for Pediatric and Adolescent Diabetes, 2018). 

We discuss the case of a paediatric patient whose diabetes classification changed over time. We discuss the clinical management, possible explanations for changing classification, and mention parallels with latent autoimmune diabetes in adults (LADA). We conclude with clinical learning points to take from this case.

Case study
A 12-year-old girl was referred for review due to a random blood glucose finding of 10.9 mmol/L  during primary care investigations for fatigue, menorrhagia and abdominal pains. She had a high BMI of 30 and a past medical history of congenital left-sided sensorineural hearing loss. Her family history included two great uncles with T1D and a grandparent with T2D. 

At initial review, the patient was asymptomatic and a urine dip was negative for ketones and glucose. On examination she had acanthosis nigricans in her groin and axilla. Investigations found:

C-peptide was not measured at this point. Her HOMA-IR result indicated increased insulin resistance, as the upper limit of the normal range is 4.5 in pubertal children. The patient’s auto-antibodies were weakly positive for GAD 32 but negative for IA2 and ZnT8 antibodies. It was therefore concluded that she had T2D. The patient’s diabetes was managed with diet and exercise to achieve weight loss, and metformin 250 mg once daily commenced.

Four months on from diagnosis, the patient had lost 10 kg following her diet and exercise plan, which equated to 13% weight loss, and her BMI had dropped to 24.9. Her HbA_1C was 31 mmol/mol (5%). At subsequent reviews she continued to lose weight, and her HbA_1C remained within non-diabetic range, therefore her metformin was stopped and she remained in apparent remission. The patient’s classification was changed to ‘transient hyperglycaemia of unknown cause’ and she was given advice to remain alert to symptoms of diabetes and counselled that she would remain at a higher risk of diabetes in the future. 

Eighteen months after her first presentation, she reported fatigue, mild polyuria, polydipsia and recent weight loss. Her HbA_1C had increased to 72 mmol/mol (8.7%) and she was negative for ketones. The results of oral glucose tolerance tests were higher than on first presentation (12.7 mmol/L at 0 hours and 23.4 mmol/L at 2 hours). Her C-peptide level was 483 pmol/L, which is at the lower end of the normal range for healthy normoglycaemic individuals (350–1800 pmol/L). At this time, T2D recurrence was presumed in view of the patient’s previous insulin resistance and lack of ketones, and treatment with metformin was recommenced. The patient’s autoantibody results were subsequently found to be positive for GAD 32, IA2 and ZnT8, in keeping with T1D and insulin was promptly started and metformin subsequently stopped. A year after diagnosis with T1D, her HbA_1C is 59 mmol/mol (7.5%). 

Discussion
Our case included a number of features initially consistent with T2D: 

The weak presence of GAD antibody suggested the potential for autoimmune destruction, but was thought to be of doubtful significance. GAD antibody is present in around 20% of young people with T2D, and GAD-positive individuals seem to have no difference in C-peptide levels or predicted future need for insulin treatment compared to GAD-negative individuals (Dabelea et al, 2007; Bingley, 2010). In keeping with a diagnosis of insulin resistance, which is known to improve with weight loss, our patient went on to complete remission of biochemical measures of diabetes following weight reduction. 

Our paediatric patient fulfilled the criteria for LADA apart from the age limit. Reinehr (2013) proposed that this type of presentation could be termed latent autoimmune diabetes mellitus in youth (LADY), and currently all patients positive for auto-antibodies are considered at risk for future T1D. Bingley (2010) discussed whether this specific patient group might benefit from a different management approach, but this has yet to be determined. 

When our patient later presented with polyuria/polydipsia and a normal BMI, it could be argued that insulin be commenced pending her autoantibody results; however on the basis of her past history and known insulin resistance, metformin was restarted. Holding off insulin initiation increases the risk of developing ketosis, however this risk must be balanced against the potentially negative impact of a more prolonged admission to start insulin, if this is unnecessary. Individual circumstance will determine the choice made, but clinicians should sensitively share these uncertainties with the family and ensure they have an awareness of the symptoms of ketosis and remain in close contact until the antibody results are known.

There are two possible pathways for changes in the clinical phenotype of diabetes over time: 

In our case, while the patient was clearly insulin resistant at initial presentation, autoimmune or hyperglycaemia-mediated beta-cell destruction may have already been under way. Weight loss likely contributed to the short-term resolution of insulin resistance, with the subsequent start of an autoimmune process and deterioration of insulin secretion. 

Conclusion
Patients do not always present with a clear type of diabetes. It is our opinion that for patients who have clinical features suggestive of T2D and who are positive for auto-antibodies, clinicians be alert to a greater risk of beta-cell destruction, T1D, and early requirement for insulin. Patients should be counselled about this risk and the associated symptoms. The monitoring frequency for these patients should be planned accordingly. The presence of auto-antibodies also indicates the need to monitor for associated autoimmune conditions; such monitoring is not indicated in antibody-negative T2D annual screening.