Disease progression and treatment response in data-driven subgroups of type 2 diabetes compared with models based on simple clinical features: an analysis using clinical trial data.
Dennis JM, Shields BM, Henley WE, Jones AG, Hattersley AT. · 2019
Read Disease progression and treatment response in data-driven subgroups of type 2 diabetes compared with models based on simple clinical features: an analysis using clinical trial data. by Dennis JM, Shields BM, Henley WE, Jones AG, Hattersley AT. (2019) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/europmc-pmc6520497.txt.
Introduction
Dennis and co-authors compare diabetes subgroups with simple clinical models. Their analysis appeared in The Lancet Diabetes & Endocrinology. A past cluster study proposed five subgroups of type 2 diabetes. The subgroups seemed to differ in progression and risk.
The team rebuilt the five clusters in the ADOPT trial, with 4,351 patients. Clusters did differ in blood sugar progression. But age at diagnosis alone explained a like amount of variation. Baseline kidney function beat clusters at predicting kidney disease. Therapy checks used the RECORD trial, with 4,447 patients. Simple features again beat clusters for picking treatment. The authors favour outcome-specific measures over fixed subgroups.
Catalog record
| CREATOR | Dennis JM, Shields BM, Henley WE, Jones AG, Hattersley AT. |
|---|---|
| YEAR | 2019 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 51 KB |
| LICENSE | CC BY 4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Disease progression and treatment response in data-driven subgroups of type 2 diabetes compared with models based on simple clinical features: an analysis using clinical trial data By John M Dennis, Beverley M Shields, William E Henley, Angus G Jones, Andrew T Hattersley Published in: The Lancet. Diabetes & Endocrinology DOI: 10.1016/S2213-8587(19)30087-7 ABSTRACT Summary Background Research using data-driven cluster analysis has proposed five subgroups of diabetes with differences in diabetes progression and risk of complications. We aimed to compare the clinical utility of this subgroup-based approach for predicting patient outcomes with an alternative strategy of developing models for each outcome using simple patient characteristics. Methods We identified five clusters in the ADOPT trial (n=4351) using the same data-driven cluster analysis as reported by Ahlqvist and colleagues. Differences between clusters in glycaemic and renal progression were investigated and contrasted with stratification using simple continuous clinical features (age at diagnosis for glycaemic progression and baseline renal function for renal progression). We compared the effectiveness of a strategy of selecting glucose-lowering therapy using clusters with one combining simple clinical features (sex, BMI, age at diagnosis, baseline HbA1c) in an independent trial cohort (RECORD [n=4447]). Findings Clusters identified in trial data were similar to those described in the original study by Ahlqvist and colleagues. Clusters showed differences in glycaemic progression, but a model using age at diagnosis alone explained a similar amount of variation in progression. We found differences in incidence of chronic kidney disease between clusters; however, estimated glomerular filtration rate at baseline was a better predictor of time to chronic kidney disease. Clusters differed in glycaemic response, with a particular benefit for thiazolidinediones in patients in the severe insulin-resistant diabetes cluster and for sulfonylureas in patients in the mild age-related diabetes cluster. However, simple clinical features outperformed clusters to select therapy for individual patients. Interpretation The proposed data-driven clusters differ in diabetes progression and treatment response, but models that are based on simple continuous clinical features are more useful to stratify patients. This finding suggests that precision medicine in type 2 diabetes is likely to have most clinical utility if it is based on an approach of using specific phenotypic measures to predict specific outcomes, rather than assigning patients to subgroups. Funding UK Medical Research Council. INTRODUCTION Type 2 diabetes is a heterogeneous, multifactorial condition, comprising 90–95% of all cases of diabetes and affecting over 400 million people worldwide. There is great interest in better characterising the heterogeneity in type 2 diabetes and in exploiting this heterogeneity to improve care and outcomes for individuals with type 2 diabetes.1, 2, 3 Ahlqvist and colleagues4 identified five replicable clusters of individuals with diabetes in the All New Diabetics in Scania (ANDIS) cohort. The smallest cluster was defined by the presence of glutamic acid decarboxylase autoantibody (GADA), regardless of other characteristics (cluster 1: severe autoimmune diabetes [SAID]). Four type 2-like clusters were then characterised by the absence of GADA positivity and varying degrees of differences in age at diagnosis, and baseline measures of BMI, HbA1c, and homoeostatic model assessment (HOMA) 2 measured insulin resistance and β-cell function. The four type 2 diabetes clusters were cluster 2, severe insulin-deficient diabetes (SIDD); cluster 3, severe insulin-resistant diabetes (SIRD); cluster 4, mild obesity-related diabetes (MOD); and cluster 5, mild age-related diabetes (MARD). Ahlqvist and colleagues showed potentially clinically important differences in disease progression and risk of complications between the clusters in observational follow-up, most notably a striking increase in the risk of diabetic kidney disease in cluster 3 (SIRD). The key question for any subgroup analysis is the clinical utility of the subgroups, and in particular whether the proposed subgroups differ in response to therapy, which could help to inform treatment strategies.2 Ahlqvist and colleagues suggested but did not show that the clusters could be useful to guide choice of therapy.5 The only stratified approaches in type 2 diabetes showing large differences in response between treatments have used subgroups defined by routine clinical measures such as sex and BMI.6 A further key question, raised by van Smeden and colleagues7 in response to the original study, is whether assigning individuals to clusters has greater clinical utility for predicting outcomes than an approach that combines continuous clinical features to predict outcomes for individual patients. Research in contextEvidence before this studyA study by Ahlqvist and colleagues proposed a novel stratification method for patients with diabetes, using a data-driven cluster analysis in Scandinavian registry data to identify five reproducible subgroups of adult-onset diabetes. The authors showed differences between the clusters in disease progression and risk of complications in observational follow-up. The authors suggested the clusters might help with therapy selection in the future but did not test whether the clusters could inform therapy choice. We searched Scopus, Web of Science, and Google Scholar for citations of the original study, searching for follow-up studies assessing the reproducibility, clinical utility, and role in treatment selection of the proposed data-driven clusters up to Jan 1, 2019. We identified a study that identified similar clusters in a Chinese population and a small mixed American population but that did not examine any aspect of clinical utility because clinical follow-up was not available. A second study of Danish patients applied a similar cluster analysis and, with duration of diabetes as an additional input variable, identified five subgroups of type 2 diabetes that differed to those in the original study, and differed in the prevalence of diabetes complications. No studies were found that tested the clinical utility and the role in treatment of the proposed cluster-based approach.Added value of this studyThis study advances the concept of heterogeneity in type 2 diabetes by testing the clinical utility of the data-driven cluster approach proposed by Ahlqvist and colleagues. The cluster analysis was repeated, and differences by cluster in disease progression and treatment response were assessed in newly diagnosed participants in the ADOPT trial with randomised, protocol-driven, follow-up data available. We found that the clusters were reproducible and differed in progression and treatment response. However, simple clinical measures were as or more useful than were the clusters for stratifying each outcome assessed.Implications of all the available evidencePatients with type 2 diabetes differ in treatment response and risk of disease progression, raising the possibility of a practical, stratified approach that is clinically orientated. Our study suggests a prediction model approach, combining phenotypic measures to predict specific outcomes for individual patients, is likely to have greater clinical utility than approaches that use clinical features to assign individuals to subgroups. We aimed to establish the clinical utility of the clusters by analysing two large existing trial datasets of individuals randomised to metformin, sulfonylurea, and thiazolidinedione therapy, ADOPT and RECORD.8, 9 By contrast with the observational follow-up by Ahlqvist and colleagues,4 these trial datasets provided protocol-driven, randomised follow-up to assess clinical outcomes and differences in response to therapy. We compared the utility of the data-driven clusters with simpler approaches based on routine clinical measures available in any diabetes clinic. METHODS Study population The primary study population comprised newly diagnosed, drug-naive, individuals with type 2 diabetes, recruited between April, 2000, and June, 2002, followed up until June, 2006, who had participated in the ADOPT trial of glycaemic durability, randomly assigned to metformin, sulfonylurea (glibenclamide), or thiazolidinedione (rosiglitazone) monotherapy