Diabetogenic effect of antihypertensive treatment: primum nil nocere

Hypertension and diabetes mellitus have been shown to exhibit a complex and multifactorial interrelationship. As part of this, the incidence of diabetes is enhanced in hypertensive patients and this finding is only in part explained by the higher percentage of overweight and obese patients in both populations [1–3]. Moreover, evidence suggests that the rate of new-onset diabetes mellitus in hypertensive patients may also depend on the choice of antihypertensive treatment [4–15]. Among the more modern trials comparing antihypertensive treatment strategies, the Captopril Prevention Project (CAPPP) observed a statistically higher rate of new-onset diabetes in patients randomized to treatment with conventional therapy (diuretics and β-blockers) as compared with those receiving the angiotensin-converting enzyme (ACE) inhibitor captopril [16]. The present review analyses more studies that reported differences in the incidence of new-onset diabetes mellitus when comparing antihypertensive treatment strategies, and attempts to define the magnitude and possible clinical significance of the observed differences. Since the CAPPP trial, several studies have described differences in the rate of new-onset diabetes mellitus in hypertensive patients treated with various antihypertensive strategies. Table 1 summarizes results from five prospective, randomized hypertension trials (CAPPP [16], INSIGHT [17], LIFE [18], ALLHAT [19] and INVEST [20]) in which different treatment strategies were compared on various morbidity and mortality end points and from one study investigating an ACE inhibitor vs placebo in patients with high cardiovascular risk (HOPE) [21]. In all hypertension studies analysed, there was a significant difference in the incidence of new-onset diabetes between patients treated with β-blockers or diuretics as compared with more modern antihypertensive agents. Interestingly, in the HOPE trial, ACE inhibition—when compared with placebo—was associated with significantly less new-onset diabetes in a population at high cardiovascular risk, in which β-blockers and diuretics were used at a rate of ∼40 and 15%, respectively [21]. Table 2 gives the calculated ‘numbers needed to treat’ (NNT) [22] for new-onset diabetes mellitus in the cited trials. Data are also corrected for the respective observation periods and expressed as NNT/year. Numbers of patients in the different studies and of patients developing new-onset diabetes mellitus mellitus in five hypertension trials comparing different antihypertensive treatment strategies and in one placebo-controlled study performed in patients with high cardiovascular risk (HOPE) aHOPE compared the ACE inhibitor ramipril with placebo. Except for INSIGHT, the numbers given reflect patients without diabetes at randomization. Data for the ALLHAT trial are given for two time points (2 and 4 years). CAPPP, Captopril Primary Prevention Project; INSIGHT, International Nifedipine GITS Study: Intervention as a Goal in Hypertension Treatment; LIFE, Losartan Intervention for Endpoint Reduction in Hypertension Study; ALLHAT: Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial; INVEST, International Verapamil Trandolapril Study; HOPE, Heart Outcomes Prevention Evaluation Study. Numbers of patients in the different studies and of patients developing new-onset diabetes mellitus mellitus in five hypertension trials comparing different antihypertensive treatment strategies and in one placebo-controlled study performed in patients with high cardiovascular risk (HOPE) aHOPE compared the ACE inhibitor ramipril with placebo. Except for INSIGHT, the numbers given reflect patients without diabetes at randomization. Data for the ALLHAT trial are given for two time points (2 and 4 years). CAPPP, Captopril Primary Prevention Project; INSIGHT, International Nifedipine GITS Study: Intervention as a Goal in Hypertension Treatment; LIFE, Losartan Intervention for Endpoint Reduction in Hypertension Study; ALLHAT: Antihypertensive and Lipid-Lowering Treatment to Prevent Heart Attack Trial; INVEST, International Verapamil Trandolapril Study; HOPE, Heart Outcomes Prevention Evaluation Study. Numbers needed to treat (NNT) for new-onset diabetes mellitus in five hypertension trials comparing different antihypertensive treatment strategies and in one placebo-controlled study performed in patients with high cardiovascular risk (HOPE) Data are given as NNT over the trial period or over the time periods given, and as NNT/year. In addition, values for NNT/30 years are also given assuming linear relationships for new-onset diabetes over time. Numbers needed to treat (NNT) for new-onset diabetes mellitus in five hypertension trials comparing different antihypertensive treatment strategies and in one placebo-controlled study performed in patients with high cardiovascular risk (HOPE) Data are given as NNT over the trial period or over the time periods given, and as NNT/year. In addition, values for NNT/30 years are also given assuming linear relationships for new-onset diabetes over time. Two studies have provided information about the time course of the manifestation of new-onset diabetes mellitus in treated hypertensive patients. In the LIFE study, an apparently linear increase in the incidence of diabetes over the 5 year period of the study could be observed in the patients randomized to either atenolol or losartan-based antihypertensive treatment. Accordingly, the difference in the incidence of diabetes in the two groups of patients increased in a linear fashion over time. Since antihypertensive therapy in most patients is a life-long process, it therefore appears legitimate also to estimate the respective NNT for extended treatment periods. Therefore, in Table 2, NNT were also estimated for a treatment period of 30 years. Notably, the number needed to harm with respect to new-onset diabetes in the five cited hypertension studies ranged between 4 and 25. In contrast, in the ALLHAT trial, patients manifesting fasting glucose values of ≥125 mg/dl increased markedly over the first 2 years of chlortalidone without a further increase at 4 years. However, no information is available on the use of anti-diabetic agents from that ‘large and simple trial’, and it therefore seems likely that the time dependence of new-onset diabetes in ALLHAT is obscured by concomitant anti-diabetic therapy. Therefore, in Tables 1 and 2, data are given taking into account the incidence values at both 2 and at 4 years. Looking at the differences in the NNT values between the studies analysed and excluding the 4 year data of the ALLHAT study, CAPPP stands out as the study with the highest NNT. In this context, it is interesting to note that in this study captopril, 50 mg/day, could be given o.d or b.i.d., and that 48% of the patients randomized to captopril took the medication o.d. Also, calcium channel blockers were an optional add-on therapy in CAPPP. Thus insufficient dosing of captopril and also add-on therapy may have diluted the difference in new-onset diabetes in CAPPP. Interestingly, the low NNT calculated for ALLHAT on the basis of the (probably more reliable) 2 year data should be seen against the background that the average body mass index (BMI) in that study (29.8 kg/m2) was higher than in CAPPP (27.9 kg/m2) or LIFE (28.0 kg/m2). It is, therefore, tempting to speculate that the risk of new-onset diabetes during antihypertensive treatment may also be a function of the underlying risk. The clinical significance of the adverse effects of a diuretic- and β-blocker-based antihypertensive therapy on metabolic parameters remains a matter of controversy. In ALLHAT, the cited figures have been played down, arguing that “these metabolic differences did not translate into more cardiovascular events or into higher all-cause mortality in the chlortalidone group compared with the other 2 groups” [19]. This argument applies to a number of studies with respectable effects of so-called ‘conventional therapy’ on cardiovascular morbidity and mortality end points. However, studies of up to 5 years duration may be insufficient to detect these end points since translation of new-onset diabetes into morbidity and mortality may simply take more time. It is interesting, therefore, to note the results from a prospective population-based cohort study from Uppsala, Sweden. The results from this long-term study observing patients over almost two decades suggest that the increase in blood glucose observed during antihypertensive treatment with diuretics and β-blockers represents a significant, independent risk factor for myocardial infarction [23]. In summary, antihypertensive treatment has a major influence on the rate of new-onset diabetes, with a markedly higher incidence in patients on β-blockers and diuretics. This finding represents a clinically important limitation for the use of these agents, at least when used without concomitant administration of one of the more modern antihypertensive agents. Primum nil nocere. Conflict of interest statement. None declared. 1Universitätsklinikum Bonn, Medizinische Universitäts-Poliklinik and 2Zentrum für Innere Medizin, Klinik für Endokrinologie und Stoffwechselkrankheiten, Magdeburg, Germany