The Association Between Red Blood Cell Transfusions and Development of Non-Hodgkin Lymphoma.

A relationship between red blood cell (RBC) transfusions and the subsequent development of NHL has been suggested from previous observational reports. The mechanism is unclear but could be related to the known immunomodulatory effect of blood transfusion. This report is a meta-analysis of such observational studies and helps clarifying the strength of such an association, if any.

We searched MEDLINE from January 1966 through June 2009 for observational studies on the association between RBC transfusions and NHL in adults using the keywords “transfusion” and “lymphoma”. Prospective studies and case control studies that reported relative risks (RR), hazard ratios or odds ratios with 95% confidence intervals (CI) were included. One author (SD) gathered the data and another (JC) reviewed the data. A fixed-effect model (FEM) was used to assess the combined outcome of individual studies while a random-effects model (REM) was used, if needed, to account for heterogeneity between studies. Heterogeneity was evaluated using the Cochrane Q and I² statistics. Publication bias was assessed by direct observation of a funnel plot as well as trim-and-fill statistics. Quality of the studies was assessed independently by another author (SP) using the Newcastle-Ottawa scale.

Our initial search rendered 1830 articles. After reviewing the abstracts, 21 papers were selected, from which 5 prospective and 9 case-control studies were included in the final analysis. Based on case-control studies, cases receiving RBC transfusions were associated with an RR of 1.37 (95% CI 0.94-1.87) of developing NHL (p=0.11); a REM was used given the heterogeneity found between studies (I²=90%; Q=90.9, p<0.0001). No publication bias was found. Based on prospective studies, RBC transfusions had a RR of 1.57 (95% CI 1.23-1.99) of developing NHL (p=0.0003); a REM was used despite finding minimal heterogeneity (I²=34%; Q=7.7, p=0.17). Publication bias analysis found 2 imputed studies, which would have not altered our results. When pooling data from retrospective and prospective studies, RBC transfusions had a RR of 1.43 (95% CI 1.12-1.84; p=0.005); a REM was used given a high degree of heterogeneity (I²=85%; Q=102.0, p<0.0001). No evidence of publication bias was found when pooling all the studies.

In both retrospective and prospective studies, there is an increased risk of developing NHL after getting a RBC transfusion. The possibility that non-diagnosed NHL was the initial indication for transfusions is not addressed by our study, limiting the generalization of our conclusions. After pooling the available data, cases were 43% more likely to develop NHL if they had a RBC transfusion. This risk should be considered non-negligible. Our findings confirm the previously reported association between RBC transfusions and the development of NHL, and further emphasize a conservative approach to RBC transfusions.

No relevant conflicts of interest to declare.