MOSES: Superiority or Noninferiority?
To the Editor:
The MOSES Investigators conclude that for a similar level of blood pressure control eposartan demonstrates a protective effect relative to nitrendipine in hypertensive patients with a history of stroke.1 What is the basis of this claim? No difference in mortality or disability (either cognitive or physical) was demonstrated. The claim appears to be based on a significant difference in the composite end point using the incidence density rather than a traditional survival analysis (Table 3). A Cox proportional hazards analysis is given in Table 4 but only for the components of the primary end point and not the primary end point itself. There is no evidence from this analysis that cerebrovascular events are different between the groups. Yet, the incidence density approach (Table 3), preferred by the authors, suggests that there is. Which are we to believe? We must surely be cautious in accepting a result if it depends on the type of analysis used to obtain it. It would be interesting to see the result of a traditional survival analysis (excluding recurrent events) on the primary end point.
A further point worthy of examination is the inclusion of transient ischemic attacks (TIA) and recurrent events in the composite primary end point. In a discussion of the validity of composite end points2 it has been suggested that the components of a composite end point should be of similar importance. Clearly, a TIA does not have the same importance as death or a disabling stroke. TIA contributed 67% of all cerebrovascular events and thus made a large contribution to the primary end point. The use of recurrent events in the primary end point is unorthodox and requires robust theoretical justification. Even the authors do not seem particularly convinced on this point stating that “The protocol appears to be appropriate … . ”. A few subjects experiencing multiple events could have a disproportionate effect on the outcome and lead to spurious conclusions.
In summary, the claim of the superiority of eposartan over nitrendipine is not supported by the data and analysis presented. Might the data support a conclusion of the noninferiority of eposartan to nitrendipine? The INSIGHT Trial3 sought to test for the superiority of a calcium channel antagonist over a diuretic in hypertensive subjects with one additional risk factor (although not stroke) for cardiovascular events. This study also had a planned secondary noninferiority analysis if superiority was not demonstrated. The margin of inferiority (the maximum difference of no clinical relevance) was set at 2% for the absolute difference in event rates. If we strip out TIA and pulmonary embolism, the primary end point of the MOSES Trial is similar to that of INSIGHT (although ideally any remaining multiple events should also be removed). It would thus be reasonable to use the same margin of noninferiority. If we do this, the absolute event rates are 20.41% and 23.85% for eposartan and nitrendipine, respectively, over the period of the trial. This gives a difference in absolute event rates (eposartan−nitrendipine) of −3.44% with a 95% one-sided upper bound of 0.3%. Because this is <2% we can accept eposartan as noninferior to nitrendipine in this patient population.
Conflicts of interest: The author has received educational support and lecturing honorarium from various pharmaceutical companies, including Solvay pharmaceuticals.
Schrader J, Luders S, Kulschewski A, Hammersen F, Plate K, Berger J, Zidek W, Dominiak P, Diener HC; MOSES Study Group. Morbidity and mortality after stroke, Eposartan compared with Nitrendipine for secondary prevention: principal results of a prospective randomized control study (MOSES). Stroke. 2005; 36: 1218–1226.
Montori VM, Permanyer-Miralda G, Ferreria-Gonzalez I, Busse JW, Pacheco-Huergo V, Bryant D, Alonso J, Akl EA, Domingo-Salvany A, Mills E, Wu P, Schunemann HJ, Jaeschke R, Guyatt GH. Validity of composite end points in clinical trials. BMJ. 2005; 330: 594–596.
Brown MJ, Palmer CR, Castaigne A, de Leeuw PW, Mancia G, Rosenthal T, Ruilope LM. 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; 356: 366–372.
Fournier et al and Boulanger et al have asked some interesting questions that have been discussed by the MOSES authors as well:
Boulanger et al argue that transient ischemic attacks (TIAs) are from a patient’s perspective not as clinically relevant as a stroke because a TIA is not disabling. This is true, but TIAs are strong predictors of stroke, which has been shown again by van Wijk et al recently.1 Therefore, TIAs are of clinical importance, and it had been predefined by MOSES-study protocol to include TIA in end point analysis. TIAs lead to hospitalization and further diagnostic procedures and therefore have economical impact. TIAs have been evaluated—as all end points—by blinded end-point committee in both treatment arms in the same way. The accuracy of a diagnosis of TIA may not be as good as for stroke. Because nearly all patients with TIA in Germany are hospitalized, the diagnosis has been assessed by neurologists as well in the majority of patients. Additionally, members of the end-point committee were neurologists and cardiologists. We agree that one of the key questions is how the eprosartan-based antihypertensive regimen, with or without combinations, compares to the combination of an angiotensin-converting enzyme inhibitor and diuretic, defined by the PROGRESS study, but final conclusions are difficult unless there will be results of a study comparing these alternatives in the future. Till these data are available some convincing results from experimental studies showing cerebroprotective effects of ARB should be considered.2,3,4 Subgroup analysis of clinical large scale trials could add some more information to answer the question how to treat hypertension after stroke.
Fournier et al discussed several interesting aspects of secondary prevention by different antihypertensive agents and compared the MOSES and PROGRESS study. These indirect comparisons between the studies are difficult because of different trial designs, baseline parameters and the comparison of active antihypertensive treatment versus placebo in PROGRESS and 2 active antihypertensive treatments in MOSES. We agree with the points raised by Fournier et al concerning the mechanism of action, but we were not able to discuss them in extension because of limited word count in the article. Clinical studies like MOSES or PROGRESS often lead to speculations about underlying mechanisms of the observed clinical effects, but overall clinical studies usually cannot sufficiently solve these interesting questions on mechanisms.
Ando H, Zhou J, Macova M, Imboden H, Saavedra JM. Angiotensin II AT1 Receptor Blockade Reverses Pathological Hypertrophy and Inflammation in Brain Microvessels of Spontaneously Hypertensive Rats. Stroke. 2004; 35: 1726–1731.
Stenman E, Edvinsson L. Cerebral Ischemia Enhances Vascular Angiotensin AT1Receptor–Mediated Contraction in Rats. Stroke. 2004; 35: 970–974.