Letter to the Editor
Response to Questions Regarding the Design and Results of the IMPACT Trial
Version of Record online: 30 DEC 2010
©2010 The Authors Journal compilation©2010 The American Society of Transplantation and the American Society of Transplant Surgeons
American Journal of Transplantation
Volume 11, Issue 1, pages 177–178, January 2011
How to Cite
Humar, A. (2011), Response to Questions Regarding the Design and Results of the IMPACT Trial. American Journal of Transplantation, 11: 177–178. doi: 10.1111/j.1600-6143.2010.03370.x
- Issue online: 30 DEC 2010
- Version of Record online: 30 DEC 2010
To the Editor:
As investigators of the IMPACT study, we think that the viewpoint of Kalil et al. is unfortunately based on a number of erroneous conclusions and a misinterpretation of the IMPACT study (1). We clarify these points below.
The IMPACT study was a meticulously designed double-blinded study, which has been scrutinized and considered robust enough in design, endpoints and outcome, by US and EU health authorities, so as to lead to global label changes in prescribing valganciclovir for kidney transplant patients. Kalil et al. state a major weakness of the IMPACT study was inadequate blinding. Active drug and placebo were indistinguishable and fully matched in terms of shape, size, color and labeling. All study investigators, site staff and sponsor, were fully blinded to the treatment allocation until after the analysis of the primary endpoint. Therefore, any conclusions regarding the lack of valid blinding are not scientifically sound. Withdrawal from treatment between groups was indeed different, however the reason for withdrawal was not due to compliance or unblinding as suggested, but due to the development of cytomegalovirus (CMV) disease after the first 100 days. Obviously patients who are being treated for CMV disease can't continue to get blinded study drug. The suggestion to separately analyze CMV disease in patients who completed 200 days of drug simply does not make sense. Furthermore, a compliance issue did not occur. Compliance was closely monitored, with drug dispense and return logs being performed at each study visit and was comparable between arms. Similarly, the author's conclusion that patients in the 100-day group had more frequent dose adjustments is incorrect; the results show that each group was similar for renal function over time. Dose adjustments were carried out alike for all patients regardless of group, according to renal function.
The suggestion that randomization was not adequate is also incorrect. Patients were randomized sequentially in 1:1 ratio at each study center in the order in which they were enrolled. The number of patients receiving antilymphocyte induction therapy was almost identical in the two arms and maintenance immunosuppression was very similar. While much is made of the analysis of opportunistic infections in the first 50 days (of which most were minor infections i.e. orolabial HSV, or thrush), the real comparison of opportunistic infections should be at the time when CMV developed (between months 3 and 9). During this time period opportunistic infections were very similar between the two groups providing strong evidence that both groups were equally immunosuppressed.
Kalil et al. also suggest that follow-up was different between the two groups. Our data clearly show that this assumption is wrong, as the benefit from 200 days of prophylaxis remains, regardless of whether the data are analyzed at a set point in time (i.e. 6, 12 or 24 months posttransplant) or at a given duration post active prophylaxis (i.e. 6, 9, 12 or 18 months following the end of active prophylaxis). The 2-year analysis of IMPACT demonstrates that very few cases of CMV disease developed between 12 and 24 months and confirms the initial results of a reduced CMV disease incidence (21.3%—200 days, vs. 38.7%—100 days, p < 0.001) (2). Furthermore, the suggestion that the primary outcome should be analyzed as a time-dependent variable is invalid, since the clinically important question is whether prolonging CMV prophylaxis will prevent CMV disease in high-risk kidney transplant patients. The IMPACT trial clearly demonstrates this, and furthermore shows sustainability of that benefit. Manipulation of statistics does not change that simple fact. CMV disease and acute rejection were not interlinked and sensitivity analysis to account for this did not change the result. Any suggestion of competitive risk is therefore purely theoretical.
Other arguments put forth by Kalil et al. are similarly flawed. Hospitalizations due to CMV may be uncommon in the United States but the trial was carried out internationally where hospitalization practices may differ. Similarly, while heavily criticizing the statistical methodology used in the IMPACT trial, the authors at the same time seem to have no hesitation in drawing broad and definitive conclusions regarding degree of immunosuppression from what are actually nonsignificant differences in acute rejection rates. Similarly, Kalil et al. have inappropriately compared data from two different arms in two trials conducted 5 years apart (3). PV16000 used an outdated definition of CMV disease whereas IMPACT used the definition currently recommended by the American Society of Transplantation (AST) for use in clinical trials and approved by the Food and Drug Administration (FDA). In fact, the incidence of investigator treated CMV disease in kidney transplant patients receiving valganciclovir in PV16000 was 32.1% (26/81; 95% CI of 22–42%) compared to 36.8% (60/163; 95% CI of 29–45%) in IMPACT (p = 0.56; corrected chi-square). We do agree with Kalil et al. on their final point. There is no need for a confirmatory trial. The IMPACT study has demonstrated through a robust study design, that extending valganciclovir prophylaxis to 200 days in high-risk patients significantly reduces the incidence of CMV disease and viremia. While Kalil et al. quote a number of older studies and meta-analysis (the latter which analyze these older studies) to justify their viewpoint, more recent studies are in fact very consistent with the results demonstrated in the IMPACT study (4,5). Therefore, future trials could instead focus on refinement of strategies to predict late-onset disease and optimization of prophylaxis duration for other high-risk groups.
Belgium: P. Peeters and D. Abramowicz; Canada: A. Humar; France: Y. Lebranchu; Germany: I. Hauser; United Kingdom: A. Jardine; United States: A.P. Limaye, F. Vincenti, J.D. Punch and E. Blumberg.
The authors of this manuscript have conflicts of interest to disclose as described by the American Journal of Transplantation.
The IMPACT study was sponsored by F. Hoffman-La Roche AG.
A.H. has received grant support and speaker honorariums from F. Hoffman-La Roche AG, and grant support and consulting fees from Astellas.
A.P.L. has undertaken contract research, and has received consulting and lecture fees from F. Hoffman-La Roche Ltd, Viropharma Inc., Vical Inc., and Novartis.
E.A.B., I.A.H., J.D.P., P.P., A.G.J., F.V., Y. L. and D.A. have no conflict of interest.
- 2Extended valganciclovir prophylaxis in D+/R- kidney transplant recipients is associated with long-term reduction in cytomegalovirus disease: IMPACT study 2-year results. Transplant 2010; Dec Suppl (), , et al.