Doctors have recognised for some time that dividing diabetes into type 1 and type 2 doesn’t define the patterns that we see in clinical practice.
In March 2018 a study in the Lancet identified five clusters of diabetes that could revolutionize the way we diagnose and manage the disease, and screen for complications.
Current classification of diabetes
Diabetes is a metabolic disease underpinned by a pathological failure of glucose uptake by cells, which if left untreated can affect almost every organ system causing significant morbidity and reducing life expectancy. The disease can be caused either by reduced insulin production or insulin resistance and widespread complications result from chronic hyperglycemia.
In 2006 the WHO published criteria for the diagnosis of diabetes: patients should have symptoms of diabetes coupled with a random venous plasma glucose concentration >11.1 mmol/L, fasting plasma glucose concentration >7.0mmol/L or a two-hour plasma glucose concentration of 11.1mmol/L after a 75g oral anhydrous glucose load. In 2011 they included the use of HbA1c > 48mmol/L (6.5%) for a diagnosis of diabetes. Whilst these criteria establish inadequate glycemic control it does not identify aetiology of the disorder, which may require further antibody and genetic screening, resulting in the risk of inadequate management of the disease.
Traditionally this disorder has been divided into two types. Type 1 affects 10-15% of patients with diabetes. It is characterised by early onset and is caused by autoimmune destruction of insulin-producing cells with patients requiring exogenous insulin administration. Type 2 represents 85% of those with diabetes, however, unlike type 1, this group is largely heterogenous with patients having varying degrees of insulin deficiency and insulin resistance.
The new classification of diabetes
Researchers in Scandinavia followed five cohorts across Sweden and Finland but only the All New Diabetics in Scania (ANDIS) cohort had blood sampling performed at the time of diagnosis. Analysing data from 8980 patients in the ANDIS cohort, who were followed for an average of 3.9 years, the authors identified six key variables: age at diagnosis, body mass index, HbA1c, function of insulin-producing cells of the pancreas, level of insulin resistance, presence of glutamatic acid decarboxylase antibodies (GADA). Using these variables, they identified 5 distinct clusters that separated the cohort out as shown below in Table 1.
Table 1
In addition to these findings, the study authors were able to replicate the findings from the ANDIS cohort and reproduce the cluster types in three of the other four cohorts as well as identifying that there was not one unifying genetic code for “type 2 diabetes” but that each cluster had its own collection of genetic codes that distinguished it from that of another cluster.
Summary
By 2030 diabetes and its complications are forecast to affect 629 million and cost between $2.1-2.5 trillion globally. With this rising health and economic burden, interventions aimed at early diagnosis, rapid and sustained glycemic control, and aggressive screening for complications may have a significant effect both in terms of health outcomes and health economies.
This study reproducibly categories diabetes further and more accurately than its traditional classification. The identification of these clusters paves the way for targeted screening of at-risk individuals. Given that only 15% of the cohort (cluster 3) had the highest risk of developing neuropathy and 17% (cluster 2) had the highest risk of developing retinopathy, aggressive screening of these groups would enable effective allocation of resources with consequent early intervention and reduction in severity of morbidity and a significant increase in economic viability of such programs.
Whilst this study has a number of limitations, particularly that the study group was a relatively homogenous population, it provides an exciting platform for future research to guide the way we diagnose and manage diabetes and reduce the associated global health and economic burden.