Diabetes Care Publish Ahead of Print, published online January 8, 2009 Bezafibrate for Prevention of Diabetes Antidiabetic Action of Bezafibrate in a Large Observational Database James H Flory BA*, Susan Ellenberg PhD*, Philippe O Szapary MD, MSCE*†¥, Brian L Strom MD* MPH, Sean Hennessy PharmD PhD* * Center for Clinical Epidemiology and Biostatistics, University of Pennsylvania School of Medicine, Blockley Hall, 423 Guardian Drive, Philadelphia, PA, 19103 (all authors) † Centocor Research and Development, Malvern, PA 19355 ¥ Preventive Cardiovascular Medicine and Lipid Clinic, Department of Medicine, University of Pennsylvania Health System, Philadelphia, PA 19104. Correspondence to: James Flory, [email protected] Submitted 2 October and accepted 4 January 2009. This is an uncopyedited electronic version of an article accepted for publication in Diabetes Care. The American Diabetes Association, publisher of Diabetes Care, is not responsible for any errors or omissions in this version of the manuscript or any version derived from it by third parties. The definitive publisher- authenticated version will be available in a future issue of Diabetes Care in print and online at http://care.diabetesjournals.org. Copyright American Diabetes Association, Inc., 2009 Bezafibrate for Prevention of Diabetes Objective: To test the hypothesis that bezafibrate, an approved fibrate, can prevent or delay type 2 diabetes. Research Design and Methods: Retrospective cohort study using data from routine medical practice in the United Kingdom, as captured by the General Practitioner’s Research Database (GPRD). Individuals chronically exposed to bezafibrate were compared to individuals chronically exposed to other fibrates. Hazard ratios for incident type 2 diabetes mellitus were calculated using a Cox proportional hazards model. A post-hoc analysis examined bezafibrate’s effect on progression to use of oral antidiabetic medications or insulin in individuals with diabetes at baseline. Results: Bezafibrate users had a lower hazard for incident diabetes than users of other fibrates (HR 0.66, 95% CI 0.53, 0.81). This effect became stronger with increasing duration of therapy. Post-hoc analysis of the effect of bezafibrate on progression of pre-existing diabetes also showed lower hazard for progression to use of antidiabetic medication (HR 0.54, 95% CI 0.38, 0.76) or progression to use of insulin (HR 0.78, 95% CI 0.55, 1.10). Conclusions: Bezafibrate appears to have clinically important antidiabetic properties. Randomized controlled trials should be considered to assess bezafibrate’s utility in treating patients with diabetes or in preventing diabetes in high risk patients. 2 Bezafibrate for Prevention of Diabetes ype 2 diabetes is a major public status as a pan-PPAR agonist may give it health threat, expected to affect antidiabetic properties unique among fibrates. T over 221 million people While not approved in the United worldwide by 2010 (1). One key target for States, bezafibrate has been widely prescribed diabetes drug development is the peroxisome for dyslipidemia in the United Kingdom. We proliferator activated receptor (PPAR) (2, 3). used observational data to examine the a There are three isotypes that are of specific priori hypothesis that bezafibrate is unique interest in metabolic diseases: PPAR-gamma, among fibrates in reducing diabetes risk. PAR-alpha, and PPAR-delta. The thiazolidinediones (e.g., pioglitazone) are RESEARCH DESIGN AND METHODS PPAR-gamma agonists used to treat diabetes We conducted a cohort study using the through improvement of insulin response. General Practice Research Database (GPRD). The fibrates are PPAR-alpha agonists used to Personal information was removed prior to treat dyslipidemia by raising high-density inclusion in the database. The requirement lipoproteins and lowering triglycerides. for informed consent was waived by the PPAR-delta remains an investigational drug University of Pennsylvania institutional target with the potential uses in diabetes, review board and the GPRD Independent dyslipidemia, and obesity (4). Because Scientific Advisory Committee. dyslipidemia and diabetes are commonly Data Source: The GPRD contains comorbid, attempts have been made to create data abstracted from a computerized medical dual PPAR alpha/gamma agonists or pan record system used by a subset of general PPAR agonists, although none of these has practices in the United Kingdom. Ninety- reached the market (3). eight percent of the UK population receives In response to efforts to develop these agents, all forms of health care through their general at least one observer has pointed out that the practitioners. The database is broadly fibrate bezafibrate actually is a pan-PPAR representative of the UK population in terms agonist and affects insulin resistance (5). of sex, age, and geography (16). We used data Post-hoc analyses of a placebo-controlled from 1988 through 2002. randomized trial showed that bezafibrate may The information prospectively postpone or prevent Type 2 diabetes (5-6). collected in the database includes During a mean six years of follow-up, hazard demographic information, all prescriptions ratios vs. placebo for incident diabetes were written by the general practitioner, clinical 0.59 (95% CI 0.39-0.91) in obese patients (5) diagnoses, specialty consultation notes, and and 0.70 (95% CI 0.49 to 0.99) in pre-diabetic hospital discharge diagnoses. Medical patients (6). These clinical endpoint data diagnoses are classified using Read Clinical were supported by biochemical evidence Classification and the Oxford Medical showing that bezafibrate slowed progression Information System codes. of insulin resistance (7). Participating general practices follow Studies of the other fibrates prospectively designed protocols for (gemfibrozil, fenofibrate, ciprofibrate, and recording computerized clinical information clofibrate) have not found such effects (8-14), and uploading it to the research database. and these drugs are far more selective for Data reaching predefined quality standards PPAR-alpha than bezafibrate (15). Hence, it are so designated. More than 400 published is reasonable to hypothesize that bezafibrate’s epidemiological studies have been performed using the GPRD (16, 17). 3 Bezafibrate for Prevention of Diabetes Study Cohort: From all patients being defined by exposure to other fibrates without followed in the GPRD, we included only any history of bezafibrate use. The pre- person-time from individuals who were specified plan for this study was to maximize exposed to a fibrate. Individuals were only power by considering all non-bezafibrate included who had been registered and up-to- fibrates as a single exposure and only standard with the GPRD for at least twelve distinguishing between individual fibrates in a months prior to initiation of the exposure secondary analysis. drug, making this an inception cohort. Any patient who switched from one study Because the primary outcome of group to another was censored at the time of interest was incident diabetes, any diagnostic the switch. In a secondary analysis, each code for diabetes mellitus, any use of home specific fibrate constituted its own exposure glucose-monitoring equipment or of drugs group, compared to bezafibrate. that are only used to treat diabetes (insulin, Outcome Definition: The outcome of biguanides, sulfonylureas, thiazolidinediones, interest was clinical diagnosis of or treatment or acarbose) prior to the first fibrate for diabetes, defined by at least two codes prescription, or within the first 90 days of indicative of diabetes. Such codes included fibrate therapy, excluded that individual from any diagnostic code for diabetes mellitus, any participation in the primary analysis. The prescription for home glucose monitoring rationale for this exclusion was to avoid equipment, or any prescription for insulin or including prevalent diabetics in the study an oral antidiabetic drug. cohort. Post-hoc analysis: In a post-hoc Exposure Definition: The study group analysis, two additional cohorts were created. included those with more than one These consisted of individuals who would prescription for bezafibrate, as a way to have been eligible for the primary study but identify those on chronic treatment. Because were excluded because of diabetes occurring we excluded persons developing diabetes prior to the 91st day of fibrate treatment. within the first 90 days of therapy, we began These individuals with baseline diabetes were follow-up with the 91st day of fibrate therapy. divided into two groups. The new cohort The duration of each prescription was either consisted of individuals who had untreated provided in the database or, when this diabetes at baseline (as identified by medical information was missing, estimated from the codes for diabetes or use of home glucose number of pills dispensed. Exposure was monitoring equipment but no use of any assumed to continue 30 days after the end of antidiabetic medication). For this cohort the the expected duration of the last prescription. outcome of interest was progression to use of Gaps over 60 days longer than expected any antidiabetic medication. between prescriptions were considered to In addition, individuals who were using oral mark a last prescription, although a patient antidiabetic therapy (a biguanide, could re-enter the cohort with the next sulfonylurea, thiazolidinedione, or acarbose) prescription. Clustering methods were used to at baseline were treated as a separate cohort, account