Insights on donor screening for West Nile virus

Abstract

Four articles in this issue of TRANSFUSION document the outcome and limitations of interventions implemented in 2002 and 2003 to prevent transmission of West Nile virus (WNV) by transfusion.1-4 Deferral of symptomatic donors and the interdiction of approximately 1000 confirmed viremic donations by viral testing in 2003 undoubtedly prevented many infections in transfusion recipients and significant morbidity and mortality. Evidence of low-level viremia below the sensitivity of screening tests on minipool samples and documentation of six cases of transfusion-transmitted WNV despite screening,5 however, showed that these measures did not eliminate all risk. The results of these and other studies formed a basis for modified approaches taken in 2004. WNV, a mosquito-borne agent known to cause potentially fatal meningoencephalitis, was introduced into the United States in 1999 in Queens, New York, most likely from importation of exotic birds.6 Although recognized since 1937 as a cause of human outbreaks in Africa, Asia, and the Middle East, WNV was first shown to be transmissible by transfusion and organ transplantation during investigations of an unprecedented massive human epidemic in the United States in 2002.7 After these discoveries, first reported in August 2002, efforts to prevent transfusion transmission of WNV in the United States focused initially on vigilance in deferring donors with symptoms of illness and withdrawal of products at increased risk based on postdonation information.8 This strategy was extended in December 2002 by a voluntary withdrawal of approximately 60,000 frozen blood components that were collected in areas of high clinical WNV disease incidence during the outbreak period. The scope of the epidemic, the serious nature of the transfusion-transmitted infections observed, and the fact that 80 percent of infected persons are asymptomatic led to the conclusion that a laboratory test would be needed to screen blood donations. Because viremia precedes seroconversion in WNV infection, and in the absence of any known chronic viremic state, priority was given to development of nucleic acid tests (NAT), although the potential for additive benefit of serologic testing was not dismissed. Technological progress in this area was extremely rapid owing to antecedent scientific knowledge about WNV genetic structure,9 the availability of NAT platforms for human immunodeficiency virus and hepatitis C virus,10 and a high level of cooperation among blood organizations, diagnostic product manufacturers, state public health laboratories, and the federal public health agencies. A task force convened by the AABB played a special role in fostering open communication and coordination of efforts. As a result, investigational donor testing for WNV by NAT became widely available in the United States and Canada in July 2003. Various studies from the 2002 and 2003 WNV epidemics have provided additional understanding of the threat to blood safety and the performance characteristics of available assay systems. Using independent assays to detect WNV RNA (Gen-Probe Inc. and Roche Molecular Systems) and antibodies (Focus Technologies and Abbott Laboratories), Tobler and coauthors1 investigated the prevalence of WNV markers in 1468 withdrawn frozen plasma units originally collected in 2002 in areas of high WNV activity. Their finding of 15 units positive for the presence of WNV immunoglobulin M (IgM), immunoglobulin G (IgG), or both by various assays demonstrated the high incidence of WNV infections during the clinical epidemic in affected regions. The finding of one RNA-positive and antibody-negative unit at a titer consistent with detection by NAT in minipools confirmed 1) that WNV exposures of transfusion recipients had taken place (i.e., from the transfused components of the corresponding donations), 2) the value of the voluntary plasma withdrawals to prevent some further exposures, and 3) the need for donor screening by NAT. The additional finding of 2 units with virus titers below the threshold of detection by NAT on minipool samples and positive for IgM and IgG, however, raised a concern that is still unresolved, namely, whether transfusion of low-titer units that contain neutralizing antibodies can cause infection and illness in recipients. With the widespread introduction of donor screening by NAT in the United States and Canada in July 2003, follow-up testing of donors with reactive screening tests has allowed the performance of WNV screening to be evaluated. Additionally, aggregation of blood center data has permitted temporal, geographic, and demographic characterization of the human WNV epidemic. Kleinman and coauthors2 reported the finding of 877 NAT-reactive donations, among 2,512,218 donations screened between July 1 and October 31, 2003, at 90 participating facilities of America's Blood Centers, representing approximately 50 percent of US collections.2 The reactive donations included 498 viremic cases: 430 confirmed positive by donor follow-up and 68 “presumptive positives” found reactive by additional testing of the index (i.e., donation) sample, but without donor follow-up. The sensitivity and positive predictive value of a presumptive-positive NAT determination were estimated at 92 and 99 percent, respectively, supporting the value of this surveillance criterion for policy making. The highest WNV activity was found in the central plains states of Colorado, South Dakota, Wyoming, and North Dakota with rates between 67.7 and 102.0 per 10,000 donations, although at different times. In any given affected region, the epidemic curve rose and fell over a period of 6 to 8 weeks, and WNV activity was highly geographically focal. Cameron and coauthors3 performed a follow-up study on 14 donors found reactive by NAT in Canada in 2003 either by minipool or by individual sample testing. Although the screening test used (Roche TaqScreen, Canada) was broadly reactive for multiple flaviviruses, further testing with a WNV-specific NAT at Canadian Blood Services confirmed the virus to be WNV in all cases. Virus titers ranged from below a limit of detection (in the reference assay) of 2.8 log10 NAT units per mL to 4.7 log10 NAT units per mL. The virus titer in two cases detected by individual sample NAT was below detection by the reference assay and presumed to be very low based on detection in only one of 38 or 39 replicates in this assay. Thirteen evaluable donors were either seropositive at the index donation or else seroconverted, confirming true positivity of the screening NAT. Busch and coauthors4 performed a study to assess and compare the sensitivities of the NAT screening and additional (supplemental) assays that were available in 2003.4 Coded serum panels were provided to manufacturers of WNV NAT assays including WNV standards of known viral titers, low-level viremic samples obtained through donor screening in 2003, and negative controls. For the screening assays, the limits of detection (LoD) were determined both on neat samples and at dilutions corresponding to minipools. For the screening assay systems of Gen-Probe Inc. and Roche Molecular Systems, the 50 percent LoD values were 3.4 and 29 copies per mL for neat samples and 43 and 309 copies per mL for samples at 1:16 and 1:6 dilutions, respectively. The 50 percent LoD sensitivities of the supplemental assays of National Genetics Institute, Bayer Diagnostics, and Gen-Probe Inc. were, respectively, 6.1, 7.7, and 1.5 copies per mL for neat samples, clearly adequate for confirmation of samples reactive in minipools, but possibly inadequate for resolution of some samples found reactive on an individual sample NAT. By comparing the LoD of the screening tests to the viral titers of 142 donations found reactive in minipool based screening using the Gen-Probe system, it was determined that 13 and 24 percent of donor samples had titers below the 50 percent LoD of the Gen-Probe and Roche assays, respectively. Analysis of diluted samples suggested that reducing the dilution of minipools would not significantly improve assay sensitivity compared with individual sample testing. As the United States contemplated the resurgence of a WNV epidemic in 2004, the results of these and other studies provided a scientific foundation for public health planning. The demonstration of limited sensitivity of minipool NAT relative to the very low viral loads seen in some asymptomatic donors, the retrospective and prospective demonstration of the added “yield” of individual sample NAT testing, and the documentation of six cases of transfusion-transmitted WNV despite donor screening in 2003 convinced policy makers in government and industry that NAT on individual rather than minipool samples could provide additional benefit and should be evaluated in high-incidence areas. This realization led to the development of expanded WNV NAT capability at central testing locations in advance of the 2004 outbreak and the devising of criteria to trigger initiation and cessation of individual sample testing. Additionally, as noted by Busch and colleagues,4 modifications were made by Roche Molecular Systems to increase the analytical sensitivity of its NAT system for WNV. Despite the demonstration of low-level viremia in a significant proportion of IgM- and IgG-positive donations in outbreak regions, all cases of transfusion-transmitted WNV to date have been associated with antibody-negative donations. Considering this fact, and the expected high rate of unneeded deferrals if donors were screened for antibodies, few experts advocate antibody-based screening at this time. Studies in animals of the infectivity of viremic and antibody-positive units, however, are still needed to address this further. Additionally, as NAT screening tests evolve, seroconversion to WNV remains a primary tool for confirmation of positive results. Antibody testing is also of major value for diagnosis of patients with illnesses suggestive of WNV and for surveillance of the clinical epidemic. A resurgent, seasonal WNV epidemic already has run its course in the United States and Canada in 2004. As predicted, the epidemic reemerged in all areas of previous activity, and the highest incidence was seen in the new areas of its progressive westward expansion. For as yet undetermined reasons, the overall size of the human epidemic was lower in 2004 than it had been in 2002 and 2003. Rapid implementation of NAT on individual donor samples was accomplished successfully in many regions, although not soon enough to prevent all transmissions by detectably viremic units. Efforts are ongoing to improve automation and throughput for individual sample testing, but this capacity remains limited. Predictors of the human outbreak have proven elusive, as have efforts to explain or predict its remarkable geographic focality. The possibility of unexpected outbreaks and the challenges of implementing seasonal testing regimens have, to date, caused the blood system to remain “on alert” and test all donations for WNV throughout the year, even in regions with colder climates where mosquito activity ceases periodically. Despite implementation of WNV intervention programs, including mosquito control as well as donor screening, we still have many unanswered questions. Will large outbreaks of WNV continue to occur in the United States? How long does viremia in an asymptomatic donor persist, and how long should donors continue to be deferred after an initial finding of viremia? Are the clinical consequences of infection the same for exposure to lower versus higher levels of virus? Are low-level viremic donations infectious when antibodies are present? How do the public health benefits of testing for WNV on individual donor samples compare with the benefits of testing only minipool samples? Will it be possible to turn donor screening on and off based on predictors of WNV activity? Whereas we cannot predict the ultimate course of seasonal human outbreaks of WNV, it is clear that an effective response to this and other emerging infectious diseases will continue to require close cooperation of all stakeholders to bring forward the necessary policies and technologies. Outcome studies on WNV testing such as those published in this issue of TRANSFUSION provide important information for addressing this ongoing challenge.