Sex and BMI Alter the Benefits and Risks of Sulfonylureas and Thiazolidinediones in Type 2 Diabetes: A Framework for Evaluating Stratification Using Routine Clinical and Individual Trial Data

Abstract

Objective: The choice of therapy for type 2 diabetes after metformin is guided by overall estimates of glycemic response and side effects seen in large cohorts. A stratified approach to therapy would aim to improve on this by identifying subgroups of patients whose glycemic response or risk of side effects differs markedly. We assessed whether simple clinical characteristics could identify patients with differing glycemic response and side effects with sulfonylureas and thiazolidinediones.

Research design and methods: We studied 22,379 patients starting sulfonylurea or thiazolidinedione therapy in the U.K. Clinical Practice Research Datalink (CPRD) to identify features associated with increased 1-year HbA1c fall with one therapy class and reduced fall with the second. We then assessed whether prespecified patient subgroups defined by the differential clinical factors showed differing 5-year glycemic response and side effects with sulfonylureas and thiazolidinediones using individual randomized trial data from ADOPT (A Diabetes Outcome Progression Trial) (first-line therapy, n = 2,725) and RECORD (Rosiglitazone Evaluated for Cardiovascular Outcomes and Regulation of Glycemia in Diabetes) (second-line therapy, n = 2,222). Further replication was conducted using routine clinical data from GoDARTS (Genetics of Diabetes Audit and Research in Tayside Scotland) (n = 1,977).

Results: In CPRD, male sex and lower BMI were associated with greater glycemic response with sulfonylureas and a lesser response with thiazolidinediones (both P < 0.001). In ADOPT and RECORD, nonobese males had a greater overall HbA1c reduction with sulfonylureas than with thiazolidinediones (P < 0.001); in contrast, obese females had a greater HbA1c reduction with thiazolidinediones than with sulfonylureas (P < 0.001). Weight gain and edema risk with thiazolidinediones were greatest in obese females; however, hypoglycemia risk with sulfonylureas was similar across all subgroups.

Conclusions: Patient subgroups defined by sex and BMI have different patterns of benefits and risks on thiazolidinedione and sulfonylurea therapy. Subgroup-specific estimates can inform discussion about the choice of therapy after metformin for an individual patient. Our approach using routine and shared trial data provides a framework for future stratification research in type 2 diabetes.

Conflict of interest statement

Conflict of Interest statement

WEH declares a grant from Quintiles, ERP declares personal fees from Lily, Novo Nordisk, and Astra Zeneca. NS declares personal fees from Boehringer Ingelheim, Eli Lilly, Novo Nordisk, Janssen and a grant from Astra Zeneca. RHH declares research funding from Bayer, Astra Zeneca, MSD and honoraria from Amgen, Bayer, Elcelyx, Jannsen, Intarcia, MSD, Novartis, Novo Nordisk and Servier. SJ is an employee and stockholder of GSK. Representatives from GSK, Takeda, Janssen, Quintiles, AstraZeneca and Sanofi attend meetings as part of the industry group involved with the MASTERMIND consortium. No industry representatives were involved in the writing of the manuscript or analysis of data. For all authors these are outside the submitted work; no other relationships or activities that could appear to have influenced the submitted work.

© 2018 by the American Diabetes Association.

Figures

Figure 1. CPRD: One year glycemic response (baseline adjusted change in HbA1c) with thiazolidinediones (red dots) and sulfonylureas (blue triangles), by sex and obesity defined subgroup.

Data are presented as least square means adjusted for baseline HbA1c ± 95% CI. A reduction (improvement) in HbA1c is represented as a negative value.

Figure 2. 5 year glycemic response (change from baseline in HbA1c) and weight change (percentage change from baseline) with thiazolidinediones (TZD, red dots) and sulfonylureas (SU, blue triangles), by sex and obesity defined subgroup.

Data are presented as means at each study visit ± standard error from mixed effects models. A reduction (improvement) in HbA1c is represented as a negative value. For AUC and treatment difference estimates positive values favour SU, negative values favour TZD. For RECORD weight change data see Supplementary Figure 7. a) ADOPT trial: absolute glycemic response (mmol/mol) b) RECORD trial: absolute glycemic response (mmol/mol) c) ADOPT trial: weight change from baseline (%)

Figure 2. 5 year glycemic response (change from baseline in HbA1c) and weight change (percentage change from baseline) with thiazolidinediones (TZD, red dots) and sulfonylureas (SU, blue triangles), by sex and obesity defined subgroup.

Data are presented as means at each study visit ± standard error from mixed effects models. A reduction (improvement) in HbA1c is represented as a negative value. For AUC and treatment difference estimates positive values favour SU, negative values favour TZD. For RECORD weight change data see Supplementary Figure 7. a) ADOPT trial: absolute glycemic response (mmol/mol) b) RECORD trial: absolute glycemic response (mmol/mol) c) ADOPT trial: weight change from baseline (%)

Figure 2. 5 year glycemic response (change from baseline in HbA1c) and weight change (percentage change from baseline) with thiazolidinediones (TZD, red dots) and sulfonylureas (SU, blue triangles), by sex and obesity defined subgroup.

Data are presented as means at each study visit ± standard error from mixed effects models. A reduction (improvement) in HbA1c is represented as a negative value. For AUC and treatment difference estimates positive values favour SU, negative values favour TZD. For RECORD weight change data see Supplementary Figure 7. a) ADOPT trial: absolute glycemic response (mmol/mol) b) RECORD trial: absolute glycemic response (mmol/mol) c) ADOPT trial: weight change from baseline (%)