Prevention of Diabetes and Hypertension

1. Background and rationale 1. Evidence available Data obtained from observational studies have shown that subjects with hypertension have an increased prevalence of type II diabetes mellitus compared to normotensives, and that the association between diabetes and hypertension represents a markedly increased risk of cardiovascular disease (1). Also the recent overview of the Blood Pressure Lowering Treatment Trialists' Collaboration indicates that the risks for stroke, coronary events, heart failure, and cardiovascular death are about twice as large in hypertensives with diabetes than in hypertensives without diabetes, independently of absence of treatment or use of different antihypertensive regimens (2). On the other hand, a number of observational and randomised intervention trials has shown that antihypertensive therapy may increase the risk of developing new diabetes mellitus in hypertensives patients. This evidence has been reviewed and meta-analysed in several recent papers (3,4). In brief, of 13 large randomised trials of antihypertensive therapy, 12 have shown a greater incidence of new diabetes mellitus in the treatment arms using thiazide diuretics or beta-blockers (and often the two types of agent in association) than in the treatment arms based on angiotensin converting enzyme inhibitors (ACEIs) or angiotensin receptor blockers (ARBs) or calcium antagonists (CAs) (often with addition of thiazides and sometimes of beta-blockers) (3). Furthermore, a recent review analysed 48 groups of subjects without diabetes mellitus at the time of randomization in 22 clinical trials. 17 of such trials enrolled patients with hypertension, 3 enrolled high risk patients, and only 1 enrolled patients with heart failure. The primary outcome was represented by incident diabetes mellitus. Overall, the antihypertensive classes that were found to be associated with newly developed diabetes mellitus were, in increasing order, ACEIs, ARBs, followed by CAs and placebo (4). It is debated whether thiazide diuretics and beta-blockers are really diabetogenic, by exaggerating or accelerating the time-dependent tendency to develop diabetes, and whether ACEIs and ARBs are antidiabetogenic by retarding this normal trend, or simply lack the diabetogenic effect of thiazides/beta-blockers. The issue, of obvious clinical importance, can only be solved by placebo-controlled studies, but placebo-controlled studies are ethically difficult to be conducted nowadays when evidence favouring active treatment of hypertension and cardiovascular risk is overwhelming. Nonetheless, data from a small number of placebo-controlled studies are available. The diuretic chlorthalidone, often associated with a beta-blocker, has been compared with placebo in elderly patients with isolated systolic hypertension in the Systolic Hypertension in the Elderly Program (SHEP) trial: when the current definition of diabetes was used (fasting glucose, FG, \>=126 mg/dl), a significant higher incidence of diabetes was found in the actively treated group (13 vs 8.7%, p\<0.0001) (5), a finding which substantiates the hypothesis that diuretic and beta-blocker therapy exaggerates the trend towards developing diabetes, i.e. has a diabetogenic effect. Placebo-controlled studies are also available concerning the effects of ACEIs and ARBs on diabetes onset. Unfortunately, in all these trials placebo groups were contaminated by administration of multiple therapies required for the control of the cardiovascular pathologies of the patients under study. In the Study on Cognition and Prognosis in the Elderly (SCOPE), concerning elderly patients with hypertension, the use of hydrochlothiazide and beta-blockers was more extensive in the so called "placebo" arm than in the active (candesartan) treatment arm, and may explain the higher incidence of diabetes in the placebo than the candesartan group (6). Likewise, in four large trials in which an ACEI or ARB were compared to placebo in patients at high cardiovascular risk because of coronary heart disease or heart failure, both the ACEI or the ARB and the placebo were given on top of multiple background therapies required by the severity of the diseases being investigated. Furthermore, little information is available from publications of these trials as to what extent background therapies were modified during the trials and whether these changes were balanced between the placebo and active treatment arms of the trials (see 3). Therefore the issue whether ACEIs and ARBs have an active antidiabetogenic action or simply lack the diabetogenic action of thiazides and beta-blockers remains undecided, and any claim that agents blocking the renin-angiotensin system exert an antidiabetogenic action appears unjustified. Prevention of the onset of diabetes, as well as of hypertension, and even more of the ominous association of diabetes and hypertension, remains, however, of primary public health concern, because of the morbidity burden represented by these ailments and the health care cost involved. Prevention must obviously be exerted before a disease develops, and is most effective in those subjects in whom the risk of developing the disease is high. Conditions predisposing to both hypertension and diabetes are well known. Individuals with so-called high-normal blood pressure (i.e., systolic blood pressure, SBP, 130-139,or diastolic blood pressure, DBP, 85-89 mmHg) have a 40% chance of becoming hypertensives (i.e.,SBP\>=140 or DBP\>=90 mmHg) over 4 years according to the Framingham Heart Study (7)and the initial findings of the Trial Of Preventing Hypertension (TROPHY, 8). High-normal BP together with impaired glucose tolerance (as indicated by fasting glucose, FG, \>=100 mg/dl), abdominal obesity, a low HDL-cholesterol and high triglycerides are often found clustered in a large portion of populations both in USA and Europe, the cluster being commonly referred to as the metabolic syndrome (MS) (9). Independently of the current debate whether all manifestations of MS are due to a common pathophysiological mechanism (i.e., insulin resistance), it is undisputed that coexistence of high-normal BP, a raised FG and an elevated BMI is a very strong predictor of both hypertension and diabetes mellitus. 2. Knowledge effect Lifestyle changes (suitable diet and physical exercise) are known to be effective in preventing both hypertension (10) and diabetes (10-12), but whether addition of an ACEI or ARB may further reduce both conditions is unknown. Individuals with MS are ideal subjects for investigating whether agents blocking the renin-angiotensin system exert a real antidiabetogenic effect in addition to what can be achieved by non-pharmacologic measures, as there are no compelling medical indications for treating these subjects with other drugs that may confound interpretation of the findings. Therefore the PHIDIAS trial plans to study a large cohort of individuals with high-normal BP, high FG (but below cut-off values of diabetes) and abdominal obesity, all instructed to lifestyle changes. They will be randomly and double-blindly assigned to placebo, or an ACEI or an ARB over a period of approximately 3 years, in order to investigate whether blockade of the renin-angiotensin system on top of lifestyle advise can further reduce onset of diabetes, a hypothesis so far based only on animal studies suggesting that some ARBs may be partial agonists of peroxisome proliferator-activated receptor gamma (PPAR-gamma) involved in regulating insulin sensitivity (13). Furthermore, the subject will be followed up for additional 6 months after withdrawal of randomised medications to investigate whether the preventive actions are enduring and survive medication. 3. Implication for AIFA and the NHS Both the conduct and the results of PHIDIAS are expected to have favourable implications both for the Italian Agency for Medicinal Products (AIFA) and the Italian National Health System (NHS). The conduct of a trial such as PHIDIAS that is committed to general practitioners and involve individuals at high risk of cardiovascular disease, will effectively promote education to healthier lifestyle measures and help prevention of cardiovascular disease and diabetes. The results of the trial will also have considerable impact both on AIFA and the NHS: 1) if addition of an ACEI or ARB to lifestyle measures is shown not to have any further preventive effect on diabetes, the claim of an antidiabetogenic action of these agents will be disproved, and their use in diabetes prevention discouraged; 2) if both an ACEI and an ARB are found to prevent hypertension and/or diabetes, but their effects are not significantly different or an ACEI is found more effective than an ARB, the use of a generic ACEI- such as the one to be administered in PHIDIAS- will represent a benefit for the subjects combined with considerable savings for the NHS; 3) if an ARB is found to be significantly more effective than an ACEI in preventing diabetes, this will implicate temporarily increased costs for the NHS, largely compensated on the long term by the decreased or delayed onset of diabetes and hypertension, as well as their complications. 4. Differences from other current trials PHIDIAS substantially differs from a few ongoing trials of diabetes prevention. The Diabetes Reduction Assessment with Ramipril and Rosiglitazone Medication (DREAM) trial (14-15) and the Nateglinide and Valsartan in Impaired Glucose Tolerance Outcomes Research (NAVIGATOR) trial (16) are only directed to diabetes prevention. Both of them intend to test an ACEI (DREAM) or an ARB (NAVIGATOR) versus an antidiabetic agent (rosiglitazone in DREAM, nateglinide in NAVIGATOR ) versus placebo with a factorial design, in individuals selected primarily on the basis of impaired glucose tolerance, so that other cardiovascular conditions are not excluded: in DREAM 43% of recruited subjects are hypertensives and 35% are dyslipidemic, and in NAVIGATOR previous cardiovascular disease or risk factors are criteria for eligibility. Therefore in both studies the investigation of the possible antidiabetogenic effect of the ACEI or ARB is likely to be confounded to some extent by concomitant therapies for hypertension and cardiovascular disease. Finally the TROPHY trial (8) is only planned for prevention of hypertension (by the ARB candesartan) and is not exploring the equally important issue of diabetes prevention by blockade of the renin-angiotensin system. 2. Objectives of the study 1. Primary objectives. In a group of individuals with high-normal blood pressure (SBP 130-139, or DBP 85-89 mmHg), FG in the high but nondiabetic range (100-125 mg/dl), and abdominal obesity (abdominal circumference \>= 102 cm in men and \>= 88 cm in women), i.e. with those characteristics of the so called MS that are major predictors of DM and HT, the trial will explore whether addition of an agent blocking the renin-angiotensin system (either an ACEI or an ARB) over periodically reinforced instructions for suitable lifestyle changes reduces: 1) onset of diabetes (defined as FG \>= 126 mg/dl or need for antidiabetic therapy) or 2) onset of hypertension (defined as SBP \>= 140 or DBP \>= 90 mmHg, or need for antihypertensive therapy) to a significantly greater extent than addition of a placebo. 2. Secondary objectives. In the same group of individuals, the trial will also investigate whether 1) the diabetes preventing effects of an ACEI and ARB significantly differ, because of the different blocking mechanisms on the renin-angiotensin system, and the additional action on bradykinin and, respectively, on PPAR gamma of the two agents, 2) the effects of the ACEI and ARB on the onset of diabetes and hypertension are enduring, persisting 3 to 6 months after substitution of the active drugs with placebo. Obviously, preventive effects substantially surviving medication would strengthen the therapeutic role of these agents, whereas disappearance of the effects after a short period would suggest a masking effect only. 3. Study design 1. Study population 1. Inclusion criteria: . Men or women of any racial background * Age \>= 40 years and \<= 75 years * SBP\>= 130 mmHg and \< 140 mmHg or DBP \>= 85 mmHg and \< 90 mmHg, average of screening and randomisation visits (in absence of any antihypertensive medication) * FG \>=100 mg/dl (5.6 mmol/l) and \< 126 mg/dl (7.0 mmol/l) between screening and randomisation (in absence of any antidiabetic medication) * Waist circumference \>= 102 cm in men and \>= 88 cm in women. 2. Exclusion criteria (see below). 2. Enrolment procedures. The study subjects will be enrolled by general practitioners associated with the Italian Society of General Practitioners (SIMG). Candidates for enrolment will be identified by the participating physicians among subjects in their NHS list attending their clinics. Each physician will be asked to enrol and randomise 21 subjects within 1 year. 3. Trial design. * Type of the study. This is a prospective, Italian, multicentre, double-blind, randomised, placebo-controlled, 3-arm, parallel group comparison with a response dependent on fixed-dose interventions. * Interventions. At the randomisation visit, as well as at follow-up visits, at 6-month intervals, all subjects will receive written detailed instruction for an adequate diet and physical exercise, with the goal of a body weight reduction of at least 5% (see Annex 1). In addition to lifestyle advise, all subjects will be randomised to one of the three following blinded treatments: 1. Placebo : 1 Enalapril placebo tablet and 1 Losartan placebo tablet, once daily. 2. ACEI : 1 Enalapril tablet and 1 Losartan placebo tablet, once daily. 3. ARB : 1 Losartan tablet and 1 Enalapril placebo tablet, once daily. During the first four weeks, 10 mg Enalapril tablets and 50 mg Losartan tablets will be used, and 20 mg Enalapril and 100 mg Losartan tablets will be used for the rest of the treatment phase of the study. At the end of this phase, all subjects will receive the two placebo tablets daily for six months (withdrawal phase). Among the ACEIs enalapril has been chosen as it has been the most widely prescribed ACEI for about 20 years, is presently available as a generic agent, and is one of the ACEIs for which a lower incidence of diabetes has been shown (SOLVD trial \[17\]). Among ARBs losartan may be chosen as animal studies indicates it may be a partial antagonist of PPAR-gamma to a greater extent than other ARBs (18). The detailed flow-chart of the study is in Annex 2. 4. Subject allocation. At the time of random allocation to double blind treatment an individual number will be allocated to each individual subject and noted in the clinical record form (CRF). The individual number determines which of the three treatment schemes the individual will receive throughout the treatment period of the study. By a computer-generated randomisation list prepared by the randomisation centre using appropriate blocks each subject will be assigned to one of the three treatment groups with identical probability. The individual numbers given to each participating physician will be assigned to an individual subject in ascending order. 5. Outcomes - Primary outcomes. Time to first event of: 1\) New diabetes, defined as occurrence of a FG\>= 126 mg/dl at any six-month visit during the treatment phase, to be confirmed at a subsequent visit within one month; or initiation of any antidiabetic treatment during the treatment period of the trial. 2\) New hypertension, defined as occurrence of a SBP\>= 140 mmHg or of a DBP\>= 90 mmHg at any six-month visit during the treatment phase, to be confirmed at a subsequent visit within one month; or initiation of any antihypertensive treatment during the treatment period of the trial. \- Secondary outcomes. Time to first event (during the treatment phase) of: 1. New diabetes (as defined above) or new hypertension (as defined above), which comes first. 2. Major cardiovascular events (myocardial infarction, stroke, cardiovascular death, heart failure, new documented angina, revascularization procedures ) plus death by non-cardiovascular causes. 3. Variations of SBP and DBP during the treatment phase and, separately, during the withdrawal phase. 4. Variations fasting blood glucose during the treatment phase and, separately, during the withdrawal phase. 5. Estimated creatinine clearance (Cockroft and Gault formula) lower than 60 ml/min in subjects with values \>= 60 ml/min at baseline. 6. Serious adverse effects. Incidence at the end of the final 6-month withdrawal phase of: 1\) New diabetes (as defined above). 2) New hypertension (as defined above)\[in order to test the possibility of a "masking" effect of treatment\]. 3\) Variations of SBP and DBP. 4) Variations of fasting blood glucose. 6. Information retrieval. * Forms. All relevant information will be filled up in CRFs for the screening and randomisation visits and all subsequent visits. When required additional documentation will be attached to the related clinical form. * Physical examination. See Annex 3. 7. Blood pressure measurement. See Annex 4. 8. Clinical laboratory measurements. See Annex 5. 9. Clinical Outcomes. See Annex 6. l. Checking of bias and confounding. Compliance to randomised treatment will be checked at each visit by counting return capsules, but even noncompliant individuals, as well as individuals withdrawn from medication, will be kept in the trial. Careful search will be made at each visit of concomitant medication used intermittently or continuously. m. Follow up. As illustrated in Fig. 1 (Annex 2), follow-up visits will occur every 6 months (± 1). The trial (see Sample size calculation) will be event-driven, but it is expected that the treatment phase will be of approximately 3 years. n. Subjects lost to follow up The subjects being all in the NHS list of the participating physicians, it is expected that loss to follow up will be minimal (probably less than 5%), and will consist only of those subjects withdrawing their consent to participate in the study (and not only to take randomised medication). o. Monitoring of the study Monitoring of the entire study, committees, centres, and participating practitioners will be the responsibility of the coordinating institution, under the supervision of the Principal investigator. Monitoring of participating units will be done electronically on a continuous way, with the support of a professional agency. p. Sample size estimates The primary hypothesis is that incidence of diabetes and, separately, hypertension will be reduced by active treatment (cumulative data of enalapril and losartan-treated groups) by 25% with respect to placebo treatment. For evaluating this difference between the placebo group and the two actively treated groups together, a two-tailed type I error of 5% and a power of 90% will be accepted. The main secondary hypothesis is that either of the two active treatments (enalapril versus losartan) will reduce incidence of diabetes by 25-30% more than the other, with a two-tailed type I error of 5% and a power of 70-90%. On the basis of the Framingham Heart Study data in high-normal BP subjects (7) and on initial findings of TROPHY (8), incidence of hypertension in our study population can be expected to be close to 10% per year. Assessment of the expected incidence of diabetes is more difficult because of variable data provided by previous studies. In the hypertension trials that have suggested a possible antidiabetogenic action of ACEIs and ARBs, annual incidence of diabetes was between 1.2 and 4% (3), but those trials that have also explored predictors of new diabetes suggest higher annual incidences in subjects at higher risk of diabetes, as follows: 2.6% in ELSA, patients with metabolic syndrome (19), 2.7% in CAPPP, upper risk tertile (20), 3.4% in LIFE, upper risk quartile (21), and 2.7% and 7.2% in VALUE, third and fourth risk quartile (22) 6.2% in DREAM (15). Three recent trials investigating interventions to prevent diabetes in subjects with impaired glucose tolerance (presumably rather similar to those in PHIDIAS, as a multivariable definition of risk such has ours has been shown to outperform a glucose tolerance test in predicting onset of diabetes report yearly incidences of diabetes of 7.5% (23), 11% (23) and 12.7% (24), reduced to 3.7-4.8% by intense lifestyle counselling. Finally, the DREAM study (15) is based on the hypothesis that patients with impaired glucose tolerance have a diabetes annual accrue rate of 4.5%. On the basis of this evidence, we have chosen a conservative hypothesis of an yearly accrue rate of diabetes of 3.5%, i.e. an incidence of 10.5% in 3 years. This being a lower incidence rate than that expected for hypertension, the sample size has been calculated on the diabetes rate. It has been assumed that, supposing a 4.8% loss to follow-up, a placebo-treated group of 2000 subjects will develop 200 cases of diabetes in three years, and the two actively treated groups, of 2000 subjects each, will develop altogether 300 cases of diabetes. These numbers will provide the required alpha of 5% and power of 90% to detect 25% reduction of diabetes incidence in the two active treatment groups together (7.5% in 3 years) compared to the placebo group (10% in 3 years). As to the secondary hypothesis, 2000 subjects in each of the enalapril or losartan groups will be sufficient to detect a lower incidence of 25% in one group (6.4% incidence in 3 years) versus the other (8.5%) (alpha 5%, power 72%) or a lower incidence of 30% (6.2 versus 8.8 in 3 years) (alpha 5%, power 88%), again supposing a 4.8% loss to follow-up. In any case the study will be event driven, and continued until 500 cases of diabetes are observed. q. Organisation PHIDIAS will be governed by a Steering Committee, chaired by the Principal Investigator (A.Zanchetti) and formed by 7 other members (G. Mancia, B. Trimarco, M. Volpe, E. Agabiti Rosei, C.Cricelli, A. Filippi, G. Corrao). A Data Monitoring and Safety Committee, chaired by L. Berrino and formed by G. Corrao, and E. Ambrosioni, will monitor the study and review two interim analyses (180 and 360 accumulated cases of diabetes) using previously agreed conservative warning rules for efficacy and harm. An Operating Committee (A. ZAnchetti, M. Valentini, F. Gregorini, A. Filippi, E. Romagnoli)will supervise the daily management of the trial. An Endpoint Committee will provide an independent and blinded assessment of efficacy endpoints (G. Mancia, M. Volpe). Participating Units will be general medicine physicians associated with SIMG. The Italian Society of Hypertension (SIIA) in accordance with SIMG will identify a number of hospital hypertension centres located in areas close to the participating units, able to provide consulting assistance to the participating physicians, and yearly sessions of lifestyle counselling for the enrolled subjects. A Monitoring Unit will be established at the coordinating institution, liaising with a professional agency (Yghea, Bologna, Italy). A Statistical Unit will also be located at the coordinating institution under the supervision of G. Corrao, with the responsibility of randomising treatments, collecting data and analysing them at trial end. No centralized laboratory is considered nor structured lifestyle support, nor oral glucose tolerance tests for diagnosis of diabetes, as the trial intends to reproduce, as far as possible, the conditions under which the NHS can operate. r. Feasibility The Principal Investigator, Alberto Zanchetti, has a long and extensive experience in running trials of antihypertensive treatment and treatment of atherosclerosis (see curriculum vitae). As Scientific Director of the Coordinating Institution, he confirms that Istituto Auxologico Italiano, IRCCS, accepts to conduct the study and agrees for the use of the human and technological resources described in the study protocol. s. Timing The trial is expected to start about 6 months after eventual AIFA approval. Recruitment of subjects is expected to be completed in one year. The treatment phase, although its duration is event-driven, is expected to last an average of 3 years, followed by a controlled withdrawal phase of 6 months. Predefined check points for recruitment are at 6, 9, 12 months from study start. Interim analyses from the DMSC are also considered (see above). A final report is expected within 1 year from study end. t. Statistical analyses Data analysis will proceed according to CONSORT guidelines for randomised controlled trials and to the intention-to-treat principle. Analysis are detailed in Annex 7. In brief, there will be summary statistics on baseline variables, while primary and secondary efficacy variables will be analysed as time to onset of diabetes or hypertension, using a Cox regression model adjusted for selected number of variables. u. Ethical aspects. The trial is not expected to have potential risks for study subjects. They have clinical characteristics for which no drug treatment is currently indicated, and all of them will benefit by receiving detailed healthy lifestyle counselling. The subjects will receive full written information on the study, and sign an informed consent document. Approval has been obtained from the Ethical Review Committee of the Coordinating Institution and will be obtained from those of the NHS units from which the participating physicians depend. 4. Annex 1. Annex 1. Lifestyle Interventions. At the randomising visit, all subjects will receive written detailed instructions for adequate diet and physical exercise, with the goal of a body weight reduction of at least 5%, as follows: diet: total fat intake of less than 5% of energy consumed, saturated fat to less than 10% of energy consumed, frequent intake of vegetables, fruit, olive oil (see ref. 8); exercise: moderate exercise, such as walking, jogging, swimming for 30 min at least 5 days a week Instructions will be reiterated at each 6-month visit, when adherence to lifestyle measures and their results will be checked and annotated in the clinical form. Recruited subjects will also be offered the possibility of an annual teaching session on healthy lifestyle provided by a hospital hypertension centre located in the vicinity 2. Annex 2. Study Flow-Chart Notes: 1.The randomised treatment phase may be prolonged or shortened (and the withdrawal phase delayed or anticipated) according to the accrual time for the primary outcome diabetes (500 cases) 2.Each time a BP or diabetes outcome is measured, an additional confirmatory visit should be scheduled within one month 3.In individual cases, additional visits may be scheduled at the physician's discretion 4.The randomised treatment doses are fixed for all subjects. However in individual cases, the physician may decide to withdraw medication (possibly temporarily, followed by rechallenge) but the subject should be maintained in the trial to preserve intention to treatment. c. Annex 3. Physical examination. For all physical examinations attention should be focused on cardiovascular signs and symptoms. The initial physical examination (screening visit) and the final one comprise the examination of the head, thorax, abdomen, spinal column, auscultation of the heart and lungs, inspection of the skin, and measurement of bodyweight and body height, and measurement of waist circumference. To measure waist circumference, locate top of right iliac crest and place a measuring tape in a horizontal plane around the abdomen at the level of the iliac crest. Measure at the end of a normal expiration. At follow-up visits, physical examination comprises a short check of all organ systems including auscultation of the heart and lungs, and a check for signs and symptoms of cardiovascular diseases. Waist circumference and body weight must be measured at all follow-up visits. Data will be recorded in specific boxes on Clinical Report Forms d. Annex 4. Blood pressure measurement. Using a calibrated standard sphygmomanometer or a validated digital device and appropriate size cuff, blood pressure will be measured with the subjects' arm supported at heart level. Systolic and diastolic blood pressure will be recorded by the investigator at phase I and V of Korotkoff sounds, or as indicated by the digital device. The cuff should be deflated at a rate not greater than 2 mmHg/sec with the patient sitting quietly for several minutes. Blood pressure will be measured 3 times at 1 min intervals. Blood pressure will be measured in all subjects at the screening and randomization visits, and during all follow-up visits at 6-month intervals. If a value of SBP\>= 140 or DBP\>=90 mmHg (cut-offs for hypertension) is measured at a follow-up visit, this should be confirmed during a second visit within one month. If the participating physician decides for antihypertensive treatment, the use of a calcium antagonist is less likely to interfere in either way with development of diabetes. e. Annex 5. Clinical laboratory. Haematology: haemoglobin, haematocrit, red blood cell count, white blood cell count, platelet count will be measured before randomization and study end. Biochemistry: fasting glucose will be measured before randomization and at all follow-up visits at 6-month intervals Fasting triglycerides, total cholesterol, HDL- cholesterol, creatinine, potassium, AST (SGOT), ALT (SGPT), as well as proteinuria (dip stick) will be measured before randomization and subsequently at yearly intervals. f. Annex 6. Clinical outcomes. Death and the cause of death will be documented in the clinical report form of the first visit missed by a deceased individual. Non fatal cardiovascular and renal events will be documented by careful inquiry with the subject and family members, and every effort will be made to obtain hospital or similar documentation when available. g. Annex 7.Statistical analysis. Data analysis will be performed according to CONSORT guidelines for randomised controlled trials and to the intention-to-treat principle. 1. Comparability between groups Summary statistics on demographic and clinical baseline variables will be calculated by treatment group, and comparisons between treatment groups will be performed. Analysis of variance models or nonparametric tests will be utilized for continuous variables, whereas Mantel-Haenszel, Cochran-Mantel-Haenszel, chi-square or Fisher's exact tests will be performed for between-group comparison of categorical variables, as indicated. 2. Analysis of primary efficacy variables The time-to-event primary efficacy variables (time to onset of diabetes and time to onset of hypertension, during the treatment phase of the trial) will be analysed separately for diabetes and hypertension, using a Cox regression model including treatment (placebo vs treatment), and the following baseline covariates: age, gender, race, BMI, FG, SBP, DBP. Subgroup analyses will be performed based at least on the following criteria: age groups, gender, BMI and FG groups (above and below medians), SBP and DBP (above and below medians). The assumption of proportional hazards will be checked for using graphic methods, and a secondary analysis of sensitivity will be performed utilizing the log-rank test. 3. Analyses of secondary efficacy variables Secondary variables for which time-to-event can be defined will be analysed using the same models as for the primary variable. For the secondary efficacy variables for which time-to-event cannot properly be calculated (incidence of diabetes and hypertension at the end of the final 6-month withdrawal phase), logistic regression models will be utilized adjusting for factors as for primary variable analysis.