Patients were at least 18 years old, had essential arterial hypertension with blood pressure values beyond 140/90 mmHg (> 130/85 mmHg in diabetic patients) and previous antihypertensive therapy of at least 8 weeks duration had proven unsuccessful. In addition to this one of the following risk factors had to be present: Diabetes mellitus, dyslipidemia, abdominal obesity, hs-CRP ≥ 2 mg/l or confirmation of microalbuminaria (MAU). Patients who fulfilled these criteria and who's treating physician deemed a treatment with 16 mg Candesartan/12.5 mg HCTZ to be necessary were documented.

The AHA/NHLBI 2004 definition was used to determine the presence of the metabolic syndrome 17. It requests that 3 out of the following 5 criteria are met: 1) waist circumference > 102 cm in men and > 88 cm in women; 2) blood pressure readings ≥ 130 mmHg systolic or ≥ 85 mmHg diastolic; 3) fasting glucose ≥ 5.6 mmol/L (≥ 100 mg/dL) or known diabetes mellitus; 4) triglycerides ≥ 1.7 mmol/l (≥ 150 mg/dL); 5) HDL-Cholesterol < 1.0 mmol/L (40 mg/dL) in men, and < 1.3 mmol/L (50 mg/dL) in women.

The following parameters were used for the calculation of the score 18: sex, age, HDL-C, total cholesterol (either indicated or calculated by the Friedewald formula), systolic BP and smoking status. In the presence of diabetes, the score was increased by a factor of 2 in men and by a factor of 4 in women. Calculations were done without replacement of missing values. To calculate the blood pressure risk reduction attributable to diabetes, the baseline SCORE risk was modified by replacing the blood pressure baseline value by the 8 week follow-up value but keeping the baseline values for metabolic risk factors. To calculate the risk attributable to metabolic changes, baseline SCORE risk was modified by replacing the baseline values of metabolic parameters by the 8 week values but keeping the baseline blood pressure within the model.

Physicians were asked to document at least 3 patients with arterial hypertension and concomitant cardiovascular risk factors. The observational period was 8 weeks and patients had a voluntary additional visit at 4 weeks and the case report forms were retrieved by the CRO Christine Franzen Consulting, (Stolberg, Germany) and screened routinely for plausibility and completeness. Source data verification was however not performed. Confidentiality: Patient data were recorded anonymously (age and gender only). Electronic data processing was conducted in accordance with local laws and regulations and the participating GPs received remuneration for the documentation of each patient which was in accordance with the "Gebührenordnung für Ärzte" (GOÄ).

Regarding safety, the trial was adequately sized (n = 3,787) to identify rare AEs, i.e. those that may not have been detected in previous clinical studies, (incidence 1: 1,000) with a probability of > 99%. These data have been published previously 16. The statistical analysis was performed descriptively and was interpreted in an explorative way. Comparisons were made for a number of variables and analyzed using descriptive statistics. The number of patients is given for each value and differences were calculated in patients with values at baseline and follow-up (per protocol). The analysis of data was performed with ACCESS 2003 and Winstat for Microsoft Excel. Tests applied are indicated in the legends of tables and figures.

The risk factors making up the metabolic syndrome were elevated throughout all patient subgroups (metabolic syndrome, with diabetes mellitus and neither condition) with hypertension (inclusion criterion), dyslipidemia (78.2%) and abdominal obesity (65.4%) being the most frequent. Tables 1, 2 and 3 display the course of these risk factors throughout the study in patients with the metabolic syndrome, with diabetes mellitus and neither condition. Blood pressure was significantly reduced in all patients (-27.2/-13.4 p < 0.0001) and there was no significant difference between the extent of blood pressure lowering (p = n.s.) between groups. Fasting blood glucose and triglycerides were unchanged in patients without the metabolic syndrome or diabetes only (p = n.s.); every other parameter was significantly changed versus baseline in a positive direction.

Figure 1 displays the relative change in risk factors in comparison between the three groups of patients. While fasting plasma glucose was unchanged (p = n.s.) in patients with neither the metabolic syndrome nor diabetes it was gradually reduced in patients with the metabolic syndrome (-1.6 ± 12.1%; p < 0.0001 vs. baseline) and strongly reduced in patients with diabetes (-5.6 ± 21.6%; p < 0.0001 vs. baseline). This corresponded to a mean absolute reduction of -2.5 ± 13.1 and -10.8 ± 30.1 mg/dL respectively. Triglycerides were similarly unchanged in the control group while they were reduced in MetSyn (-4.1 ± 43.9%; p < 0.0001 vs. baseline) and Diabetes (-4.9 ± 29.4%; p < 0.0001). Mean absolute values were -19.1 ± 61.0 and -23.3 ± 76.2 mg/dl respectively.

Finally absolute cardiovascular risk was calculated according to the SCORE method. The baseline risk differed substantially in the three patient subgroups (Diabetes 19.8 > MetSyn 7.4 > Neither 6.4%) as did the absolute risk reduction (Diabetes -8.7 > MetSyn -3.2 > Neither -2.7%). For details see table 4. The SCORE value was also calculated assuming that a) blood pressure was unchanged (risk reduction of -2.6, -2.9 and -7.6% in patients with neither, MetSyn or diabetes respectively) and b) that all other risk factors were unchanged (-0.4, -0.7 and -1.4) to estimate the proportion of risk reduction from the antihypertensive effect of candesartan (Table 4).

In the present study a blood pressure reduction of -27.2/-13.4 mmHg was documented. Baseline blood pressure values were comparable throughout the subgroups as was the reduction in blood pressure induced by 16 mg candesartan/12.5 mg HCT. In a previous study patients with previously untreated severe hypertension (mean blood pressure 178/117 mmHg) who received 16 mg candesartan/12.5 mg HCTZ experienced a rapid blood pressure reduction that peaked at -38/-29 mmHg 21. The blood pressure reduction in the present study matches well with previous data and extends these by the finding that patients previously uncontrolled will have a substantial reduction in blood pressure when prescribed the fixed dose combination. The absolute amount of blood pressure lowering (in mmHg) appeared to be dependent on baseline blood pressure but did not differ between patient types (diabetes, metabolic syndrome and neither condition). Therefore patients with a higher blood pressure value can expect a higher blood pressure lowering effect of candesartan/HCTZ.

Throughout the study a reduction of cardiovascular risk factors was observed. However one should consider the short time frame (observational period 8 weeks) in which the changes were observed and its relative magnitude (Figure 1). While a slight reduction of waist circumference could be observed in all three patient groups, changes were non-significant with respect to fasting blood glucose, triglycerides and HDL cholesterol in patients that neither had the metabolic syndrome nor diabetes (eumetabolic). Patients with metabolic risk factors however showed a pronounced response in these metabolic risk factors when treated with candesartan/HCTZ. Previous studies including the ALPINE (OR 0.12; 95%CI 0.01–0.97 vs. Thiazides) and the combined SCOPE and CHARM studies (OR 0.80; 95%CI 0.67–0.95 vs. placebo) have already demonstrated the reduced diabetes incidence in patients taking candesartan 121422. In the randomized double blind ALPINE study newly diagnosed hypertensive patients on a low dose diuretic alone or in combination with a betablocker were compared in their use of candesartan without a calcium channel blocker and followed for a year 14. At a similar level of blood pressure reduction, patients on the candesartan fixed dose combination had lower levels of plasma glucose, triglycerides and higher HDL-cholesterol. At 12 months, 18 patients in the hydrochlorothiazide group versus five in the candesartan group had a 'metabolic syndrome', as defined by the World Health Organization (P 0.007) despite 1 year of active blood pressure-lowering therapy.

The underlying mechanism for the eumetabolic effects of Candesartan are largely speculative. While there currently is great attention to the PPARγ activating properties of some ARBs 23 candesartan is less lipophilic 24 and probably does not directly activate PPARγ. It has however recently been reported that candesartan enhanced the gene expression of PPARγ and induced the increased expression of adipokines and a decrease in the pro-inflammatory cytokine TNF-alpha 25. On the leptin contrary was reduced because of the repression of the Leptin gene by activated PPARγ. Whether this is an adequate explanation for the changes observed can only be speculated at present.

Previous studies in primary care, similar in design as the present study, have shown comparable results to the ones reported here 2627. These studies were substantially longer (6 and 9 month observations respectively) and point towards the conclusion that the observed effects of candesartan/HCTZ may well be even more pronounced longer term. Differential effects of ARB treatment between patients with and without the metabolic syndrome have also been observed in the study by Kintscher et al. 26, confirming the present result that severely metabolic patients (diabetes > metabolic syndrome) will particularly benefit from an ARB based combination treatment.

Cardiovascular risk scores have been developed to estimate the cardiovascular risk of a given patient. They reflect findings from epidemiological studies like Framingham and PROCAM that a particular set of risk factors makes up the largest proportion of risk. The SCORE method has been developed under the auspices of the European Society of Cardiology (ESC) and has been adapted to fit particular countries in Europe 18 and specific risk tables for Germany have been published 28. 5% has been suggested to be the intervention threshold for primary prevention by a working group of the German Society of Cardiology 29.

The documented baseline risk of in the present was well beyond the intervention threshold of 5% disregarding that most of these patients were eligible for intervention based on a diagnosis of diabetes mellitus or concomitant cardiovascular disease. Treating patients with the fixed dose combination of 16 mg candesartan/12.5 mg HCTZ resulted in a substantial risk reduction. Similar analyses have been conducted in the primary care setting in a previous study. Bramlage et al. showed that Irbesartan treatment (with or without HCTZ) led to a reduction of cardiovascular risk in hypertensive type-2 diabetic patients 30. The effect was higher in men than in women but no specific data were reported for different patient subgroups. Also no effort was made to attribute SCORE reductions to different risk factors making up the score. Data from the present study point at an only partial responsibility of the blood pressure lowering effect (-27.2/-13.4 mmHg) but to the favourable metabolic changes induced by candesartan. This was unexpected given the dominant blood pressure lowering effect of the drug and it is also welcome based on earlier data from the Hypertension Optimal Treatment Study, that despite good blood pressure control, the residing metabolic risk factors were found to be an important cause of residual risk 9. Whether or not a candesartan based treatment may narrow the gap between the cardiovascular risk of naturally normal and well controlled patients is however beyond the scope of the present study.

The present results have to be considered against the background of potential limitations. The study was not controlled and therefore the role of a placebo effect or the withdrawal of antihypertensive agents is unknown. Second, in the absence of a randomization procedure and not consecutive inclusion the influence of unknown biases, e.g. through patient selection, cannot be ruled out. Third, changes in concomitant medication influencing the metabolic profile (lipid lowering agents, oral antidiabetic agents or insulin) have not been documented in the present study. Fourth, randomly high values for blood pressure readings or metabolic changes may have resulted in artificially large reductions consistent with the phenomenon "regression to the mean", the extent of which is difficult to determine. Among the strengths of the study was the choice of the setting. Observational studies in primary care, which include typical patient groups and reflect current treatment approaches, are useful for complementing the findings of randomized controlled trials 31.