Rituximab May Not Lead to Increased Infection Rates in Transplant Recipients

Abstract

To the Editor: We were surprised to see that the recent study by Kamar et al. (1Kamar N Milioto O Puissant-Lubrano B Incidence and predictive factors for infectious disease after rituximab therapy in kidney transplant patients..Am J Transplant. 2010; 10: 89-98Abstract Full Text Full Text PDF PubMed Scopus (145) Google Scholar) was published by AJT, because we believe it contains several methodological flaws. A major concern is the inappropriate comparison of patients who received rituximab for various therapeutic reasons with all other kidney transplant recipients. We would argue that these are fundamentally incomparable groups, even with the attempts made to subdivide the control group. For example, the immunological hurdles posed by patients treated with rituximab to allow transplantation across a B-cell positive crossmatch in the presence of DSA are very different than those posed by patients needing treatment for acute rejection. Simply put, neither the experimental nor control groups are sufficiently homogeneous to allow comparison. Of significant concern is the use of a control group that is not contemporary with the experimental group. The experimental group is taken from renal transplant recipients receiving rituximab between April 2004 and August 2008 (n = 77). This is compared with a control group between January 1997 and December 2007 (n = 902). Direct comparisons cannot be made between these two groups as during the time difference of 8 years treatment strategies are likely to have changed. It is unclear why, with a relatively large control group, the authors did not choose to collect contemporary data with the same follow up period as the experimental group. The authors conclude that the use of rituximab in renal transplantation is associated with a high risk of infectious disease, yet this directly contradicts their results which showed a lower (statistically insignificant) rate of infections in the patients who received rituximab (46% vs. 54%). Indeed, the incidence of bacterial infections was the same in rituximab-treated patients and controls, viral infections were lower, and only fungal infections (of which there were 13) were higher. The other significant conclusion was that infection-related deaths were associated with rituximab therapy. However, no detailed data are presented regarding the seven patients who died, and the authors describe an array of additional treatments, including repeated plasma exchange, ATG and high doses of steroids, which are all implicated in infection-related deaths. The criteria for using rituximab are not clearly given, and it is noteworthy that the median number of courses given was 4; there is no justification for repeated doses except in those with lymphoma, since persistent CD20 depletion occurs after a single dose (2Genberg H Hansson A Wernerson A Wennberg L Tyden G Pharmacodynamics of rituximab in kidney allotransplantation..Am J Transplant. 2006; 6: 2418-2428Crossref PubMed Scopus (146) Google Scholar). The authors have failed to acknowledge large studies of rituximab in induction therapy (3Genberg H Kumlien G Wennberg L Berg U Tyden G ABO-incompatible kidney transplantation using antigen specific immunoabsorption and rituximab: A 3 year follow-up..Transplantation. 2008; 85: 1745-1754Crossref PubMed Scopus (152) Google Scholar,4Tyden G Genberg H Tollemar J et al.A randomized, doubleblind, placebo-controlled, study of single dose rituximab as induction in renal transplantation..Transplantation. 2009; 87: 1325-1329Crossref PubMed Scopus (116) Google Scholar) which have not shown any increase in infection rates. We acknowledge that the indication for rituximab by the Toulouse group was mainly as a rescue agent, which may account for stronger cumulative immunosuppression and higher risk of fatal infection. Importantly it is not acceptable to perform a multivariate logistic regression analysis on such few 'events'. Generally, the number of variables that can be entered in such an analysis (including all their subcategories) must be at least one-tenth of the 'events' (5Harrell Jr, FE Lee KL Matchar DB Reichert TA Regression models for prognostic prediction: Advantages, problems, and suggested solutions..Cancer Treat Rep. 1985; 69: 1071-1077PubMed Google Scholar). In this case with only 35 patients getting an infection out of the 77 patients who had RTx they would be allowed to include only four variables in the LR model and not the tens of variables used in their analysis. In summary, the authors have simply shown that infection-related deaths are more common in a selected group of patients, largely comprising those with antibody-mediated rejection, when compared with the general transplant population. These deaths cannot be attributed to rituximab from the data given, and may be equally likely to be due to other immunosuppressive manipulations, including the use of ATG.