Safety of filtered leukocyte-reduced blood products for prevention of transfusion-associated cytomegalovirus infection.

In a recent Blood report, Bowden et al’ concluded that filtration is an effective alternative to the use of seronegative blood products for prevention of transfusion-associated cytomegalovirus (CMV) infection in CMV seronegative marrow transplant patients. Furthermore, they suggest their results justify abandoning the maintenance of dual inventories of seronegative and seropositive/unscreened blood products and that serologic screening of blood products for CMV could be eliminated altogether. We believe their data do not support such strongly stated conclusions and, in fact, point to the possibility of important problems with filtered products. We address the most striking finding first: an overall increased incidence of fatal CMV disease in the group receiving filtered products. In the group of 250 patients receiving filtered products, all 6 patients who showed evidence of CMV infection developed CMV disease, including 5 cases of fatal CMV pneumonia. In contrast, of 252 patients receiving seronegative blood products, only 4 showed evidence of CMV infection, and none had CMV disease. The investigators’ actuarial analysis confirms an increased probability of developing CMV disease by day 100 in the filtered arm (2.4% v O%, P = .03). Also, it is unlikely that chance alone could account for six Occurrences of disease among six infections in the filtered arm versus no disease among four infections in the seronegative arm (P = .005 by Fisher’s exact test). We agree with the investigators that these findings are surprising and unexplained, but they cannot be ignored. The investigators partly discount the difference in disease rates on the basis of their primary analysis, which included only those infections occuning between days 21 and 100 after transplant. However, even in this subgroup analysis there were 2 cases of fatal CMV pneumonia out of 3 infections in the filtered arm compared with no cases of CMV disease out of 2 CMV infections in the seronegative arm. Although the difference in actuarial estimates of CMV disease rates now do not achieve statistical significance (1.2% for the filtered arm v 0% for the seronegative arm, P = .25), it is a fundamental error (a so-called type II error): to interpret this to mean the two arms are clinically equivalent. The investigators designed their study to have sufficient power to detect, at best, a difference of 5% in the incidence of CMV infection/ disease rates. However, the use of seronegative products results in a CMV transmission rate of just a few percent, and their study was not designed to rule out a doubling or even tripling of the risk for developing fatal CMV disease in the filtered arm. To put this in perspective, the investigators’ data are consistent with the possibility of 2 to 5 extra CMV pneumonia deaths per 250 patients when filtered products are used in place of seronegative products. In our opinion this would be a clinically important difference, and their study cannot rule this out. Thus, Bowden et a1 have demonstrated a real possibility that using filtered blood rather than CMV seronegative blood puts bone marrow transplant patients at an unacceptably higher risk of fatal CMV disease. To conclude equivalency of seronegative products and filtered products for the prevention of CMV disease is not warranted by their data.