ArticlesIncident diabetes in clinical trials of antihypertensive drugs: a network meta-analysis
Introduction
The propensity for some antihypertensive drugs to lower glucose tolerance and precipitate diabetes has been known since at least 1958.1, 2, 3, 4, 5, 6 Because hypertension is often associated in large populations with impaired glucose tolerance, insulin resistance, and obesity, many patients with hypertension develop diabetes,7, 8, 9 even when treated with placebo.10, 11 Some long-term clinical trials of antihypertensive agents have shown significant differences in the rates of new cases of diabetes between treatment groups.6, 12, 13, 14 Four independent meta-analyses have shown that direct inhibitors of the renin-angiotensin system reduce the risk of incident diabetes, but no comparison was attempted in these studies between angiotensin-converting-enzyme (ACE) inhibitors and angiotensin-receptor blockers (ARB).15, 16, 17, 18 Traditional meta-analyses that attempt to summarise all the existing data from clinical trials comparing one drug class with any other treatment show significant heterogeneity for ACE inhibitors, calcium-channel blockers (CCB),19 diuretics, and β blockers. When the nine trials of an initial CCB versus any other therapy (including placebo in the Felodipine Events Reduction trial [FEVER]20) are separated by comparator, the significant p value for homogeneity (<0·0001) for the overall meta-analysis becomes less significant. In four trials comparing an initial CCB with an initial ACE inhibitor or ARB, the CCB was associated with a significantly increased risk (by 26%, 95% CI 15–39, homogeneity p=0·77). In six trials that compared an initial CCB with either a diuretic or a β blocker or both, the proportion of patients developing diabetes was significantly smaller in the CCB group than in the other groups (by 21%, 16–26, homogeneity p=0·04).
Network meta-analysis is a fairly new statistical technique that allows both direct and indirect comparisons to be undertaken, even when two of the strategies have not been directly compared (eg, ACE inhibitor vs ARB).21 This type of analysis can summarise randomised clinical trials of several different treatment strategies, and provide point estimates (and 95% CI) for their association with a given endpoint, as well as an estimate of incoherence (a measure of how well the entire network fits together, with smaller values suggesting better internal agreement of the model). We therefore used the network meta-analysis technique to estimate the relative odds of developing diabetes during long-term treatment with an initial class of antihypertensive drug, on the basis of the reported numbers of participants with, or at risk of, incident diabetes in randomised clinical trials.
Section snippets
Identification of trials
We undertook a systematic review to identify long-term randomised clinical trials of antihypertensive drugs that reported the number of new cases of diabetes from 1966 to Sept 15, 2006. This search was repeated four times by one reviewer independently of the other. Data abstraction was done by one reviewer and verified independently by the other. We searched MEDLINE, the Cochrane Collaboration's Database of Systematic Reviews, PubMed, and OvidWeb using the MeSH terms: “diabetes mellitus-Type 2,
Results
Table 1 shows a summary of the trials included in the base-case network meta-analysis. Figure 2 shows the network of clinical trials according to the comparison of specific classes of initial antihypertensive drugs, as well as point estimates for the results of traditional meta-analyses. In this figure the arrows point towards the class of drug with the higher risk of incident diabetes according to the summary odds ratio for a traditional meta-analysis of studies comparing the two drugs
Discussion
Our findings show that ARB and ACE inhibitors are the antihypertensive agents least associated with incident diabetes followed by CCB and placebo, β blockers, and diuretics.
These results summarise international experience of incident diabetes in long-term clinical trials of antihypertensive agents, incorporating both direct and indirect comparisons of agents, including those that have never been compared directly (ie, ACE inhibitor vs ARB). The network meta-analysis technique overcomes the
References (60)
Renin-angiotensin system inhibition prevents type 2 diabetes mellitus: part 1: a meta-analysis of randomised clinical trials
Diabetes Metab
(2004)- et al.
Angiotensin-converting enyzme inhibitors or angiotensin receptor blockers for prevention of type 2 diabetes: a meta-analysis of randomized clinical trials
J Am Coll Cardiol
(2005) - et al.
Prevention of cardiovascular events with an antihypertensive regimen of amlodipine adding perindopril as required versus atenolol adding bendroflumethiazide as required, in the Anglo-Scandinavian Cardiac Outcomes Trial-Blood Pressure Lowering Arm (ASCOT-BPLA): a multicentre randomised controlled trial
Lancet
(2005) - et al.
Effect of angiotensin-converting-enzyme inhibition compared with conventional therapy on cardiovascular morbidity and mortality in hypertension: the Captopril Prevention Project (CAPPP) randomised trial
Lancet
(1999) - et al.
Morbidity and mortality in patients randomised to double-blind treatment with a long-acting calcium-channel blocker or diuretic in the International Nifedipine GITS study: Intervention as a Goal in Hypertension Treatment (INSIGHT)
Lancet
(2000) - et al.
Cardiovascular morbidity and mortality in the Losartan Intervention For Endpoint reduction in hypertension study (LIFE): a randomised trial against atenolol
Lancet
(2002) - et al.
Randomised trial of effects of calcium antagonists compared with diuretics and β-blockers on cardiovascular morbidity and mortality in hypertension: The Nordic Diltiazem (NORDIL) study
Lancet
(2000) - et al.
Randomised trial of old and new antihypertensive drugs in elderly patients: cardiovascular mortality and morbidity the Swedish Trial in Old Patients with Hypertension-2 study
Lancet
(1999) - et al.
Outcomes in hypertensive patients at high cardiovascular risk treated with regimens based on valsartan or amlodipine: the VALUE randomised trial
Lancet
(2004) - et al.
Effects of candesartan on mortality and morbidity in patients with chronic heart failure: the CHARM-Overall programme
Lancet
(2003)