Rapid Responses to:
|
|
Rapid Responses: Rapid Responses published:
-
![[Read Rapid Response]](/icons/misc/sectionDOWN.gif)
Interleukin-2 receptor monoclonal antibodies and renal graft loss
- Ana Luiza M Gleisner, Daniel Melecchi de Oliveira Freitas, MD (4 May 2003)
-
Re: Interleukin-2 receptor monoclonal antibodies and renal graft loss
- Dwomoa Adu, Natalie J Ives and Keith Wheatley (12 May 2003)
-
Sample power
- Ana Luiza Gleisner, Daniel Melecchi de Oliveira Freitas (25 May 2003)
-
Sample power
- Dwomoa Adu, Natalie J. Ives and Keith Wheatley (18 June 2003)
|
|
|||
|
Ana Luiza M Gleisner, MD Hospital de Clínicas de Porto Alegre, RS Zipcode 90040 143, Daniel Melecchi de Oliveira Freitas, MD
Send response to journal:
|
In the meta-analysis conducted by Adu and Wheatley (1), the addition of interleukin-2 (IL-2) receptor monoclonal antibodies to ciclosporin based immunossupression protocols after renal transplant reduced in 49% the incidence of acute rejection with “no significant reduction in the graft loss rate at one year”. The authors suggest that longer follow-up studies are needed to confirm the hypothesis that the reduction in rejection rates will result in reduced graft lost rate. Acute rejection is believed to cause a 50% decrease in graft survival(2). A 49% reduction in rejection would, therefore, result in an expected overall 25% reduction in graft loss. Adu and Whealey found, in fact, a 22% reduction in graft loss when IL-2 receptor monoclonal antibodies were used (relative risk 0.78; 95% confidence interval 0.58-1.04), or a 25% reduction when only the studies with basiliximab and daclizumab are analyzed (relative risk 0.75; 95% confidence interval 0.56-1.03). This results may have been not statistically significant because both the incidence of the outcome (graft loss) and the effect measured (22% reduction) are not as prominent as are the values for acute rejection. In order to reach statistical significance for the measured effect with 95% confidence and 80% power, we would need to include 5000 patients in a study. The sample included in the meta-analysis would be able to detect a significant 22% reduction only if the incidence of graft loss in controls were twice as much, which may, for instance, happen if the entire sample was followed for a longer period. The authors also found a reduction in cytomegalovirus infections in the patients treated with IL-2 receptor antibodies. This might be explained by the increased incidence in acute rejection in control patients which probably received more immunossupression to treat the rejection episodes. There is no reason to believe that there is a direct effect of IL-2 receptor antibodies on cytomegalovirus infection. Patients in which renal transplant failure end up in dialysis, so differences in mortality are really difficult to obtain in these population. Graft loss is a clinically important surrogate end-point and should be strongly considered. With this estimated effect, 43 patients treated with IL-2 receptor antibodies would prevent one graft loss and about 10% of the losses would be prevented. We thus believe that while we wait for larger studies with longer follow-up as well as studies with tacrolimus, IL-2 receptor antibodies should be used in renal transplant patients on ciclosporin-based immunossupression. References 1. Adu D, Cockwell P,Ives NJ, Shaw J,Wheatley K. Interleukin-2 receptor monoclonal antibodies in renal transplantation: meta-analysis of randomised trials. BMJ 2003;326:789-794. 2.Hariharan S, Johnson C, Bresnahan B, Taranto S, McIntosh M, Stablein D. Improved graft survival after renal transplantation in the United States, 1988 to 1996. N Engl J Med 2000; 342: 605-612. Competing interests: None declared |
|||
|
|
|||
|
Dwomoa Adu, Consultant Nephrologist Department of Nephrology, Queen Elizabeth Hospital. Edgbaston. Birmingham. B15 2TH, Natalie J Ives and Keith Wheatley
Send response to journal:
|
We are grateful for the constructive comments of Gleisner and de Oliveira Freitas on our meta-analysis.1 We agree that it is plausible that a reduction in acute rejection rate will translate into a reduction in graft failure and we implied this, although did not state it explicitly, in the discussion when we said that we also found a non-significant reduction in graft loss. We do, however, believe that cautious interpretation is generally appropriate in meta-analysis and our study does not provide clear evidence that graft loss is reduced since the confidence interval is consistent with no benefit. Our aim was to provide clinicians with information, based on the totality of the currently available evidence, that would allow them to make judgements on the implications of interleukin-2 receptor antibody treatment of patients undergoing renal transplantation with ciclosporin based therapy. Necessarily our conclusions had to be based on the data from the trials. We agree that longer follow up of patients in the current trials may establish more convincingly the effectiveness of this treatment on graft loss, but we are unclear as to how Gleisner and de Oliveira Freitas arrived at a figure of 5000 patients being needed to demonstrate a reduction in graft failure. Were a 22% reduction still to be observed with a larger number of events (it is the number of events, not the number of patients, that is important), the result would become conventionally significant with just 25% more events (i.e. about 250 compared to the current 200). The apparent reduction in cytomegalovirus infections could still be simply due to the play of chance. We agree that this non- significant reduction in cytomegalovirus infections could be due to fewer rejection episodes and treatment for this. Reference 1. Adu D, Cockwell P, Ives NJ, Shaw J, Wheatley K. Interleukin-2 receptor monoclonal antibodies in renal transplantation: meta-analysis of randomised trials. BMJ 2003;326:789-791. Competing interests: DA is an investigator in a trial of fluvastatin in renal transplant recipients that is funded by Novartis; DA has a grant from Roche to study chronic allograft nephropathy |
|||
|
|
|||
|
Ana Luiza Gleisner, MD Hospital de Clínicas de Porto Alegre 90040 193, Daniel Melecchi de Oliveira Freitas
Send response to journal:
|
We would like to clarify our previous comments on the interesting meta-analysis conducted by Adu et al (1). Not only the number of events but also the sample size are important to characterize the study power. Either increasing the sample size or the follow-up- and ultimately the frequency of the outcome- can be regard as in order to estimate if statistical significance would be attained in case the results found were reproduced. In that way, using a computer program(2), we found that we would need 5000 patients - about half of all renal patients transplanted yearly in United States in the past decade(3)- with a 11.9% incidence of graft loss in the control group (as found on Adu’s meta-analysis) to obtain a statistically significant 22% reduction in graft loss with intervention with 95% confidence and 80% power. Alternatively, all 1858 patients from the meta-analysis could be followed for about 5 years- the anticipated time to achieve a 25% graft loss (3)-, if the rate reduction remained 22% in the entire period. That would represent an increase from 200 events in the study conducted by Adu et al. to about 350 events (not 250). What we are trying to point out is that a p value of 0.05 instead of the 0.09 found whilst could be very hard to achieve, does not guarantee the evidence is irrefutable. We believe that background knowledge- the fact that acute rejection episodes will decrease graft survival- should be consider when conclusions are drawn from trials’ data. Conclusions based merely on the p value do not take into account the strength of the association neither the possibility of a beta-error (4). Medical inference cannot be based solely on statistical hypothesis, even when conclusions are being made from a single study. 1. Adu D, Cockwell P, Ives NJ, Shaw J, Wheatley K. Interleukin-2 receptor monoclonal antibodies in renal transplantation: meta-analysis of randomized trials. BMJ 2003; 326:789. 2.Computer programs for epidemiologists (PEPI), version 3.0 by Abramson JH and Gahlinger PM 3. Hariharan S et al. Improved graft survival after renal transplantation in the United States, 1988 to 1996. NEJM 2000; 342:605. 4.Goodman SN. Toward Evidence-Based Medical Statistics. 1: The p Value Fallacy. Ann Intern Med 1999; 130: 995. Competing interests: None declared |
|||
|
|
|||
|
Dwomoa Adu, Consultant Nephrologist Queen Elizabeth Hospital, Edgbaston, Birmingham. B15 2TH, Natalie J. Ives and Keith Wheatley
Send response to journal:
|
We believe that our previous response addressed the points raised again by the latest letter (25th May) from Gleisner and de Oliveira Freitas. To briefly reiterate, with the incidence rate reported in our meta-analysis (graft loss: 9.6% and 11.9%), a single confirmatory trial would need 5700 patients (based on comparison of proportions, it is 3800 patients if time to event analysis with the log-rank test is performed) to have 80% power to detect a 22% odds reduction in graft loss at p=0.05. It is however highly debatable as to whether a trial of this size would actually be feasible (the largest previous trial recruited just 380 patients in one year(1). In any case, this rather misses the point of a meta-analysis, which aims to pool all the available randomised data and for which standard sample size calculations are not appropriate. On the issue of study power, we would direct the interested reader to the work of Machin et al.(2) (page 176), which clearly states that the power of a log- rank test is determined solely by the number of critical events, thereby demonstrating that longer follow-up will on its own improve the estimates in our meta-analysis. 1 Nashan B, Moore R, Amlot P, Schmidt A-G, Abeywickrama K, Soulillou J-P. Randomised trial of basiliximab versus placebo for control of acute cellular rejection in renal allograft recipients. Lancet 1997;350:1193-8. 2 Machin D, Campbell M, Fayers P, Pinol A. Sample size tables for clinical studies. Blackwell Science Oxford. Competing interests: None declared |
|||