| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2006 European Society of Cardiology
Clinical trials update from the European Society of Cardiology Meeting 2005: CIBIS-III. Response to correspondence from R. Willenheimer et al.
Stavanger University Hospital, University of Bergen Norway E-mail address: trout{at}online.no
Received January 11, 2006; As discussant of the CIBIS III presentation at the Hotline session at the ESC Congress in Stockholm 2005, I have been asked to comment on the correspondence from Dr. Willenheimer et al. [1] in response to Professor Cleland's summary of the CIBIS III results [2].
There are several issues that deserve attention, the most central being the interpretation of the trial's main result. The primary objective of CIBIS III was to demonstrate non-inferiority for bisoprolol-first vs. enalapril-first with regard to the primary endpoint, a composite of all-cause mortality and all-cause hospitalisation.
This was not a futility analysis as suggested by Professor Cleland. As stated in the design paper [3], "Non-inferiority will be assessed based on the per-protocol population. Non-inferiority based on the intention-to-treat population may introduce a bias; the fewer patients who actually take study medication the more likely it is that non-inferiority will be shown."
As Dr. Willenheimer explains in his letter, "non-inferiority for bisoprolol-first vs. enalapril-first is considered proven if the entire 95% confidence interval is below a relative risk of 1.125". The result for the primary endpoint was an upper limit for the 95% confidence interval of 1.21 for the per-protocol population and 1.16 for the intention-to-treat population. This means that the per-protocol trial result cannot exclude a 21% increase in risk of the primary endpoint due to bisoprolol-first.
In the main publication [4], the upper limit for the 95% CI is given as 1.17. Dr. Willenheimer writes in his letter that, "In the setting of CIBIS III, a relative risk of 1.125 corresponds to a hazard ratio of 1.17". However, relative risk and the hazard ratio are usually considered to be synonymous in the context of a time-to-event analysis. No matter what terminology distinction is drawn here, the results certainly do not approach any conventional criteria used to establish non-inferiority.
Perhaps the main objection to the CIBIS III design is the concept that patients receive two different treatments for a short period of time, switch to combination therapy in both groups for a long period of time and then the non-inferiority analysis is based on both periods pooled. This is a risky approach that can easily make an ineffective treatment appear comparable to an efficacious treatment.
The design is not compatible with current guidelines or clinical practice, which do not recommend any substantial delay in initiating β-blocker therapy. Dr. Willenheimer dismisses this criticism by explaining that the long period of monotherapy was included "to increase the chance of finding any differences between the two strategies". Although, this might increase the likelihood of the trial detecting a difference, it does not assist the clinician in interpreting the relevance of the results based on an intervention that deviates so substantially from clinical practice.
For these reasons, I agree with Professor Cleland that, "In essence, CIBIS III was a study of bisoprolol vs. enalapril for 6 months". Certainly, all the 6-month endpoints should have been presented and it would have been preferable to have based on the 6-month results. The Kaplan-Meier plot of patients with worsening heart failure requiring hospitalisation or occurring in hospital presented in Fig. 6 of the main publication, demonstrates some cause for concern in this regard. Based on this plot, one would estimate that the hazard ratio for this event occurring during the 6-month bisoprolol monotherapy period to be approximately 1.8 (p=0.01). I was surprised to learn from Dr. Willenheimer's letter that the hospitalisation data due to worsening heart failure is still not analysed. I would recommend that the authors perform this analysis and have a good look at the results after 6 months of monotherapy.
 |
References |
|---|
 Top References |
|---|
- Willenheimer R., Krum H., Veldhuisen D., Funck-Brentano Erdmann E., Meyer W. Comment on "Clinical trials from the European Society of Cardiology Meeting 2005: CIBIS-III. Eur J Heart Fail (2006) 8:219–220.
[Free Full Text] - Cleland J.G., Coletta A.P., Lammiman M., et al. Clinical trial update from the European Society of Cardiology Meeting 2005: CARE-HF extension study, ESSENTIAL, CIBIS III, S-ICD, ISSUE-2, STRIDE-2, SOFA, IMAGINE, PREAMI, SIRIUS-II and ACTIVE. Eur J Heart Fail (2005) 7:1070–1075.
[Abstract/Free Full Text] - Willenheimer R., Erdmann E., Follath F., et alCIBIS III investigators. Comparison of treatment initiation with bisoprolol vs. enalapril in chronic heart failure patients: rationale and design of CIBIS-III. Eur J Heart Fail (2004) 6:493–500.
[Abstract/Free Full Text] - Willenheimer R., van Veldhuisen D.J., Silke B., et alCIBIS III Investigators. Effect on survival and hospitalization of initiating treatment for chronic heart failure with bisoprolol followed by enalapril, as compared with the opposite sequence: results of the randomized Cardiac Insufficiency Bisoprolol Study (CIBIS) III. Circulation (2005) 112:2426–2435.
[Abstract/Free Full Text]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||