RESEARCH DESIGN AND METHODS

Study design
As reported previously (8), the DIABHYCAR Study aims to show that ACE inhibition by itself protects normotensive and hypertensive type 2 diabetic patients with micro- or macroalbuminuria from cardiovascular and renal events. The primary outcome is the combined incidence of the following events: cardiovascular death (including sudden death), acute myocardial infarction, stroke, end-stage renal failure, and acute congestive heart failure requiring urgent medical assistance (this last item was added to the initial protocol [8] by an amendment of the Scientific Committee as a primary outcome in 1997 because death from heart failure is predicted by microalbuminuria in type 2 diabetic patients [1] and because this condition has emerged as an important cause for mortality in Westernized populations during recent years [9]). Secondary outcomes were as follows: all-cause mortality, coronary bypass surgery or angioplasty, transient ischemic attack, nonurgent congestive heart failure, peripheral artery bypass surgery or angioplasty, lower-limb amputation above the metatarso-phalangial joint, loss of vision in one eye, and doubling of serum creatinine on two consecutive visits. The DIABHYCAR Study is a double-blind multicenter multinational parallel trial comparing 1.25 mg/day ramipril with placebo for a minimum of 3 years. A low dose of ramipril was chosen because this drug can affect renal and myocardial functions independently of its blood pressure–lowering effects (10,11). A feasibility study was performed in France. The vast majority of the French type 2 diabetic patients are cared for by their general practitioners. For this feasibility study, the general practitioners were selected randomly from lists organized according to their geographical origins. Type 2 diabetic patients were screened for microalbuminuria during their routine visits on a consecutive basis for a 6-week period starting 1 November 1993. On this occasion, the patients potentially eligible for the DIABHYCAR trial were asked if they were willing to participate, and most of them answered positively (12). These data suggest that the patients enrolled in the DIABHYCAR trial are representative of the French type 2 patients with microalbuminuria. However, no feasibility study was performed in the other participating countries. Patients were recruited there by specialty clinics.

Patient selection. The main inclusion criteria were as follows: men or women aged >50 years with type 2 diabetes treated with oral antidiabetic drugs at inclusion, with or without hypertension, and with persistent micro- or macroalbuminuria, identified by two successive random urine samples containing >20 mg/l urinary albumin (assayed centrally) within 6 months. This selection procedure was chosen because it predicts a urinary albumin excretion (UAE) level >30 mg/24 h two to three times out of three consecutive 24-h urine samples with a positive predictive value >98% (13). Main exclusion criteria were a plasma creatinine level >150 µmol/l, treatment with ACE inhibitors or angiotensin subtype 1 receptor antagonists, or myocardial infarction <3 months before inclusion. The study started in France in February 1995 with general practitioners as local investigators and started in other European and North African countries during 1997 (Austria, Belgium, Croatia, Czech Republic, Germany, Greece, Hungary, Morocco, the Netherlands, Slovenia, Spain, Switzerland, Tunisia, and the U.K.). The study protocol was first approved by the Ethics Committee of Angers University Hospital (Angers, France) and then by all local ethics committees elsewhere as requested. Prestudy assumptions indicated that 4,000 patients must be followed-up for a minimum of 3 years, taking into account a 5% annual incidence of the primary outcome, a 20% risk reduction, and risks of 5% of and 10% of error types (8). However, the measured incidence of the primary outcome was 50% less than expected 2 years after study onset. To compensate for this possible lack of study power, the recruitment phase was prolonged, as recommended by the DIABHYCAR Scientific Committee, in all countries until 31 March 1998. Follow-ups were therefore scheduled for all patients through 31 March 2001.

Methods
At the first screening visit, the following variables were recorded by the local investigators on a standardized screening form: age, sex, weight (kg) and height (cm) in light clothes without shoes, sitting blood pressure (measured two times with a mercury sphygmomanometer), and, from medical records, duration of type 2 diabetes and hypertension (blood pressure >140/90 mmHg and/or antihypertensive treatment), tobacco (cigarettes/day) and alcohol (drinks/day) consumption, antecedent myocardial infarction, angina pectoris, stroke (as evidenced by hospital records), lower-limb arteritis (as defined by lack of foot pulses at palpation or amputation), diabetic retinopathy (five or more microaneurysms at fundoscopic examination), and fasting plasma glucose and creatinine (measured locally during the previous 3 months). These variables (together with current treatment) were reported on a screening form and were input into a central database by dedicated staff in the central laboratory (Biochemistry Laboratory, Angers, France). Completion of the database was obtained because expedition to the investigator of the urinalysis (as described below) was blocked automatically if response to one item was lacking. In such a case, the investigators were interviewed by telephone or fax until clarification was achieved. Then, this clarification (signed by the investigator) was confirmed by regular courier to the central laboratory. A random urine sample was then collected and sent with the screening form through private express mail to the central laboratory. UAE was measured immediately by nephelometry (14) (sensitivity 1 mg/l, inter- and intra-assay coefficient of variation 4 and 2%, respectively), and leucocyturia and hematuria were screened with dipsticks (Multistix, Bayer Diagnostics, Ruteaux, France). Results of the urinalysis were sent to investigators the same day. If urinary albumin was <20 mg/l, the screened patient was excluded; if it was >20 mg/l with leucocyturia and/or hematuria, a repeat urine sample could be sent again by the investigator for rescreening. If urinary albumin was >20 mg/l without any likely cause for error, then a second urine sample was collected by the investigator at the next outpatient visit. The second urinalysis was handled the same way as the urinalysis for the first sample. If the second urinary albumin was >20 mg/l without cause for error, then the patient was eligible for the study. The eligible patients could be included by the investigator through an automated telephone connection with the coordinating center (Department of Clinical Pharmacology, Lyon, France). For all included patients, a blood sample was then sent to the central laboratory for HbA_1c measurement (15) (range of normal values 4.0–5.6%). In addition, serum, plasma, and DNA samples were collected from the French participants on this occasion for ongoing substudies.

Statistical analysis
Results are given as means ± SD or median (range). For variables with a skew distribution, a log-transformation was performed, as requested. The aim of this work was to examine the contributions of selected variables on the risk of selection for the DIABHYCAR trial because of high urinary albumin by comparing patients selected for the trial to those not selected and by taking into account the patients' geographical origin. Because the majority of patients were French, we grouped patients of several countries of the same geographical area to cut down the contribution of the French participants in the data analysis. Country groups were constituted so that patient numbers could be rounded up to 1,000: Germany (sufficient by itself); group 1 (Northwest Europe: Belgium, the Netherlands, Switzerland, and the U.K.); group 2 (Central Europe: Austria, Czech Republic, Hungary, and Slovenia); group 3 (Mediterranean Europe: Croatia, Greece, Spain, and Turkey); and group 4 (North Africa: Morocco and Tunisia). A two-factor analysis of variance was used to examine the independent effects of inclusion into the study (yes/no), of the geographical origin, and of their interaction on selected quantitative variables. For qualitative variables, ² tests were used. To examine the independent effects of selected variables on risk for selection into the trial because of micro- and macroalbuminuria in the screened type 2 diabetic patients, a logistic regression analysis was performed using the following variables (the quantitative variables were dichotomized because their distributions were not normal): age (median <65/>65 years), sex (female/male), BMI (mean <28.8/>28.8 kg/m²), systolic blood pressure (sBP) (<140/>140 mmHg), diastolic blood pressure (dBP) (<85/>85 mmHg), tobacco consumption (no cigarettes/>1 cigarette per day), alcohol consumption (no drinks/>1 drink per day), previous myocardial infarction (no/yes), retinopathy (no/yes), fasting plasma glucose (mean <9.3/>9.3 mmol/l), and geographical origin. Because the respective contributions of several potential determinants for micro- and macroalbuminuria (16) may vary according to geographical origin, this analysis was repeated independently for each of the country groups.

RESULTS

Recruitment procedure
The investigators from the various countries participated as follows (number of centers per number of screened patients per number of included patients): Austria 79/330/86, Belgium 45/387/43, Croatia 4/167/82, Czech Republic 31/343/150, France 6,429/20,296/3,438, Germany 104/918/259, Greece 5/516/152, Hungary 29/167/64, Morocco 30/863/204, the Netherlands 3/209/24, Slovenia 1/129/62, Spain 6/59/34, Switzerland 79/196/63, Tunisia 41/431/95, Turkey 8/217/63, and the U.K. 22/227/118. A series of patients were eliminated from the screening procedure: 374 because their age was <50 years, 152 because they did not take oral antidiabetic drugs, 53 because they were taking insulin, and 134 because their plasma creatinine was >150 µmol/l. These patients were not taken for analysis in this study. As a whole, 25,455 type 2 diabetic patients were screened. After urine specimens with a possible cause for error (leucocyturia and/or hematuria) had been double-checked, 14,987 urine samples were negative (UAE <20 mg/l) at first specimen examination, 9,753 were positive (UAE >20 mg/l), and 715 were positive but with concomitant leucocyturia and/or hematuria. As a result, a urine specimen for second screening was obtained from 7,922 subjects. Of them, 1,747 specimens were negative (UAE <20 mg/l), 5,972 were positive (>20 mg/l), and 203 were positive but with concomitant leucocyturia and/or hematuria. Then, 5,972 patients were eligible; of them, 4,937 completed the randomization procedure on 31 March 1998 (Fig. 1).

Patient characteristics
The main characteristics of the screened patients are given in Tables 1–4, according to whether or not they were randomized into the DIABHYCAR trial and according to their geographical origins. Compared with the patients not selected, patients randomized into the trial were older, more often men than women, and more obese, and smoked tobacco and drank alcohol more often (Table 1). Their diabetes duration was longer and they had poorer glycemic control and displayed retinopathy more often (Table 2). They displayed higher sBP and dBP, and the percentage of hypertensive subjects was higher among the randomized patients (Table 3). Lastly, previous angina pectoris, myocardial infarction, stroke, lower-limb arteritis, and diabetic retinopathy were more frequent among the randomized patients than among the other patients (Table 4).

Incidence of primary outcome at completion of the recruitment phase of the DIABHYCAR Study
On 31 March 1998, the following primary outcomes had been observed after 6,545 patient-years of follow-up: 59 myocardial infarctions, 31 sudden deaths, 68 cardiovascular deaths, 61 strokes, 99 cases of congestive heart failure, and 1 end-stage renal failure.

CONCLUSIONS — The DIABHYCAR Study was designed to ascertain if the peculiar renal properties of ACE inhibitors, which convey a clinical benefit for renal prognosis in type 1 diabetic patients, may be of benefit to the type 2 diabetic patients, whose high urinary albumin predominantly predicts cardiovascular events, as illustrated by the incidence of primary outcome to date. The rationale of the DIABHYCAR Study is supported by several recent trials in type 2 diabetic subjects. In the Appropriate Blood Pressure Control in Diabetes Trial (17) and Fosinopril Versus Amlodipine Cardiovascular Events Randomized Trial (18), which both compare an ACE inhibitor to a long-acting dihydropyridine, the relative risk for cardiovascular events (especially myocardial infarction) was two to five times higher with treatment with dihydropyridine than with treatment with an ACE inhibitor. But the risk (or the benefit attributable to dihydropyridine or ACE inhibitor) was not ascertained specifically in the absence of a placebo group or in the absence of a group on conventional treatment. Conversely, two studies testing long-acting dihydropyridine as first-line treatment to aggressively reduce sBP (19) or dBP (20) for prevention of cardiovascular events concluded this strategy brings a peculiar benefit to the diabetic patients, as compared with the nondiabetic patients. In the type 2 diabetic patients participating in the Captopril Prevention Project Study (21), risk for myocardial infarction was reduced, but risk for stroke was equivalent to that in patients using captopril, by comparison to diuretics and/or -blockers, while the event rates of these two types of complications were similar. Regarding kidney function, a monocenter study comparing lisinopril to atenolol in type 2 diabetic patients with proteinuria did not favor lisinopril in terms of glomerular filtration rate evolution (6). Finally, the hypertensive type 2 diabetic patients of the UKPDS who received intensive antihypertensive treatment did not benefit from captopril compared with atenolol, and the incidence of microalbuminuria or proteinuria was similar (7). Thus, a strategy for type 2 diabetic patients based on ACE inhibitors administered because of high urinary albumin must be validated independently of the ACE inhibitors' hypotensive effect. This is why a low dose of ramipril, able to act on surrogate end points like microalbuminuria (10) or left ventricular hypertrophy (11) without discernable effects on blood pressure, was chosen for comparison versus placebo for the DIABHYCAR Study (8). This treatment is tested in addition to the currently recommended treatments for type 2 diabetic subjects. Blood pressure targets recommended by the DIABHYCAR Scientific Committee are <140/90 mmHg initially. Because of the results of the UKPDS (22) and the Hypertension Optimal Treatment Study (20), theses targets were recently adjusted to a dBP of <85 mmHg.

APPENDIX

Members of the DIABHYCAR Study Group
Ph. Passa (President), Fr. Alhenc-Gelas, J.P. Boissel, F. Cambien, G. Chatellier, S. Etienne, A. Girault-Louvel, P. Gueret, M. Lievre, J. Mann, M. Marre, J. Menard, P.F. Plouin, D. Vasmant, L. Vaur, G.C. Viberti, and C. Weisselberg.

---