Angiotensin II antagonists for hypertension: There are differences in efficacy

A recent paper by Conlin et al¹ describes a meta-analysis of certain published studies of the AT₁-receptor blockers losartan, valsartan, irbesartan, and candesartan. The authors conclude that these compounds have comparable antihypertensive efficacy, and a nearly flat dose-response curve. We are concerned that the approach adopted is not truly “meta-analytical” and is rather more a pooling of data sets that simply serves to obfuscate the issue. Rather than enhancing the precision of estimated treatment effects, the analysis as performed, lacks the power to detect the large clinically important differences in efficacy, which are present and are more apparent by a thorough and systematic review of the available evidence.

Our main concerns are as follows: first, no placebo correction was employed, despite the known large variability of placebo antihypertensive effects between different trial populations. Second, a wide variety of trial comparisons and methodologies were included, from AT₁-receptor blocker comparisons with placebo, to comparisons with other classes of antihypertensive drug. Third, only 4 of the 43 trials analysed were direct, head-to-head comparisons between AT₁-receptor blockers. Fourth, methods of measuring blood pressure varied widely between trials. Fifth, only secondary data were available to the authors, rendering rigorous examination of primary data and trial methodology impossible. Sixth, important data were lacking from the analysis, as recent studies, unpublished studies (but with abstracts presented at major meetings), and studies using ambulatory blood pressure monitoring, were not included.

Such indirect comparisons between drugs introduce so much ‘noise’ into pooled data that any possible differences in efficacy are completely obscured. As a result, the Conlin meta-analysis has very low statistical power, and it is therefore not surprising that the method was unable to detect any differences.

The most appropriate way to compare the efficacy of two or more drugs is in a single, prospective, randomized, head-to-head, double-blind study, avoiding the inherent problems of meta-analysis. Such studies may have the greater statistical power than the analysis described by Conlin and colleagues. A number of such studies indicate that several AT₁-receptor blockers are more effective than losartan. In the case of irbesartan, two definitive studies have demonstrated its superior antihypertensive efficacy when compared to losartan.^2,3 Direct, head-to-head comparisons between losartan and candesartan show significantly greater antihypertensive efficacy of candesartan compared to losartan,^4–6 whether compared at the starting dose or at the maximum recommended dose. Similar findings have been apparent when valsartan^7,8 and telmisartan⁹ have been compared with losartan, although in one of the studies⁸ the differences between valsartan and losartan did not achieve statistical significance.

A number of these direct, comparative studies have been performed using ambulatory blood pressure monitoring, enabling important information to be gathered on the variability of blood pressure control between doses. Such studies confirm another important difference between these agents, namely that whereas the antihypertensive effects of candesartan are maintained for 24 hours and, indeed, up to 48 hours after dosing, those of the losartan combination are not, and by 48 hours after dosing have virtually disappeared.⁶ This is likely to be clinically important in everyday practice, as most patients miss at least one dose every week, and many more fail to take their medication at the recommended intervals.

We would also take issue with Conlin et al's analysis of dose-response relationships, which is not supported by the available data. It is true that the recommended doses of losartan are at the top of the dose-response curve, with little difference in efficacy between 50 and 100 mg daily.¹⁰ On the other hand, there is no disputing the clear increase in efficacy from 4 to 16 mg daily of candesartan,¹¹ or from 75 to 300 mg daily of irbesartan.¹²

We conclude, therefore that the Conlin meta-analysis is potentially misleading and does not correspond either with clinical experience or with the data derived from definitive studies.