Virus and Antibody Dynamics in Acute West Nile Virus Infection

MAJOR ARTICLE Virus and Antibody Dynamics in Acute West Nile Virus Infection Michael P. Busch, 1,2 Steven H. Kleinman, 1,9 Leslie H. Tobler, 1 Hany T. Kamel, 5 Philip J. Norris, 1,2 Irina Walsh, 1 Jose L. Matud, 3 Harry E. Prince, 3 Robert S. Lanciotti, 7 David J. Wright, 8 Jeffrey M. Linnen, 4 and Sally Caglioti 6 1 Blood Systems Research Institute and 2 University of California, San Francisco, 3 Focus Diagnostics, Cypress, and 4 Gen-Probe, San Diego, California; 5 Blood Systems, Scottsdale, and 6 Blood Systems Laboratories, Tempe, Arizona; 7 Division of Vector-Borne Infectious Diseases, Centers for Disease Control and Prevention, Fort Collins, Colorado; 8 Westat, Rockville, Maryland; and 9 University of British Columbia, Victoria, Canada (See the brief report by Tobler et al., on pages XXX XXX.) Background. The dynamics of the early stages of West Nile virus (WNV) infection can be assessed by follow-up studies of viremic blood donors. Methods. A total of 245 donors with WNV viremia were followed up weekly for 4 weeks and then monthly for up to 6 additional months or until seroconversion. Plasma samples were tested for WNV RNA by transcription-mediated amplification (TMA) and for WNV-specific IgM and IgG antibodies. RNA persistence was investigated by 6 replicate TMA tests; samples that were viremic for 40 days were tested for WNV-neutralizing activity. Follow up of 35 additional viremic donors for up to 404 days was conducted to evaluate persistence of WNV-specific antibody. Results. The median time from RNA detection to IgM seroconversion was 3.9 days; to IgG seroconversion, 7.7 days; to RNA negativity by single-replicate TMA, 13.2 days; and to RNA negativity by 6-replicate TMA, 6.1 additional days after results of single-replicate TMA are negative. For 4 donors in whom RNA persisted for 40 days after the index donation, all samples obtained after this threshold were also positive for WNV IgG and neutralizing activity. The mean times to IgM and IgA negativity were 156 and 220 days, respectively. Conclusions. IgM and IgG develop rapidly after viremia and before RNA levels become undetectable, which occurred a mean of 13.2 days after the index donation among donors in this study. WNV RNA detection by replicate TMA rarely persists for 40 days after the index donation and is accompanied by WNV-specific neutralizing antibody, consistent with an absence of WNV transmission via transfusion of seropositive blood components. West Nile virus (WNV) is an important cause of neuroinvasive disease and febrile illness in the United States [1 4]. In 2002 evidence accumulated that WNV could be transmitted through transfusions [5, 6], and in June Received 13 December 2007; accepted 31 March 2008; electronically published XX August 2008. Potential conflicts of interest: M.P.B. is on the scientific advisory board of Gen-Probe and is under contract with Novartis/Chiron to research issues on blood safety that are unrelated to this study. L.H.T. is under contract with Novartis/ Chiron to research issues on blood safety that are unrelated to this study. J.L.M. and H.E.P. are employed by Focus Diagnostics, which provided the ELISA kits used in this study. J.M.L. is employed by Gen-Probe, which provided the TMA kit used in this study. All other authors: none reported. Presented in part: 2006 Annual Meeting of the American Association of Blood Banks, Miami Beach, FL; 2007 Annual Meeting of the American Association of Blood Banks, Anaheim, CA. Financial support: Centers for Disease Control and Prevention (grant R01-CI- 000214 to M.P.B.). Reprints or correspondence: Dr. Michael P. Busch, Blood Systems Research Institute, 270 Masonic Ave., San Francisco, CA 94118 (MBusch@bloodsystems.org). The Journal of Infectious Diseases 2008; 198:xxx 2008 by the Infectious Diseases Society of America. All rights reserved. 0022-1899/2008/19807-00XX$15.00 DOI: 10.1086/591467 2003 nucleic acid amplification testing (NAT) of blood donors for WNV RNA was begun [7]. Donor samples are tested in minipools (MPs) in a fashion similar to that of NAT for HIV-1 and hepatitis C virus [8]. In addition to MP-NAT, blood centers developed strategies to switch to individual donation (ID) NAT in regions experiencing epidemic WNV activity in order to interdict donations with low-level viremia missed by MP-NAT [9 14]. These approaches have proven highly effective, interdicting 2000 viremic donations and nearly eliminating the risk of transfusiontransmitted WNV [10 16]. A detailed understanding of the dynamics of the early stages of WNV infection is important for developing algorithms for screening and confirming potentially infected donors, deciding when it is safe for such donors to resume blood donations, and evaluating the residual risk of transfusion-transmitted infection. More generally, an understanding of WNV infection dynamics is important for developing diagnostic testing algorithms and design- Virus and Antibody Dynamics in WNV Infection JID 2008:198 (1 October) 1

ing research protocols to study pathogenesis and immune responses that may influence development or treatment of WNV disease. Use of sensitive WNV RNA detection assays in donor screening has allowed implementation of several approaches to study these issues. The first approach, developed by our group, involved use of the incidence-window period model to comprehensive NAT and IgM incidence data in North Dakota donors to estimate the interval after infection onset that WNV RNA is detectable by MP-NAT; this so-called MP-NAT window is 6.9 days (95% confidence interval [CI], 3.0 10.7 days) [13]. By comparing the yields of donors detected by ID-NAT and MP-NAT in the entire Blood Systems donor population over a 2-year period, we were able to further estimate that the length of an even earlier stage of low-level viremia (defined as a detectable level of WNV RNA by ID-NAT but not MP-NAT) is 1 day [12]. These approaches, however, are limited for studying later stages of infection after the development of IgM and IgG antibody responses. A cross-sectional analysis is likely to be biased because symptoms are more likely during these stages and because symptomatic individuals either do not attempt to donate or are deferred, owing to elevated temperatures or reports of mild symptoms [17]. Despite these limitations, cross-sectional screening data combined with limited follow-up data have allowed for a general picture of the dynamics of the later stages of WNV infection to emerge. After the initial phase of increasing viremia, viral loads begin to decrease before or concurrent with IgM and IgA antibody development, followed shortly thereafter by development of IgG antibody [11, 12, 18 22]. However, the precise duration of viremia and its relationship to antibody development and symptoms have not been well-defined [11 14, 18 22]. Furthermore, available data concerning IgM and IgA persistence are conflicting [18 22]. In this study, we use data derived from serial follow-up of a large number of viremic donors and from replicate testing by use of a very sensitive WNV RNA detection system to further define the natural history of viremia and its relationship to the development and persistence of IgM, IgA, and IgG antibodies. SUBJECTS, MATERIALS, AND METHODS WNV RNA screening, enrollment, and follow-up of viremic donors. Donations from 19 US blood-collection facilities (in the southwest, south, and central plains and in parts of California) owned by Blood Systems were screened for WNV by means of the WNV transcription-mediated amplification (TMA) system (Procleix WNV Assay [Gen-Probe]) in minipools of 16 samples as previously described [7, 12, 23]. The 16 donations that comprised reactive minipools were tested individually to identify the viremic donation. The TMA WNV assay has an analytical sensitivity of 4 RNA copies/ml (i.e., 50% of the lower limit of detection, according to probit analysis) and a sensitivity of 45 copies/ml when used in MP screening [7, 23]. All TMA-reactive donors identified in 2003 were offered enrollment into a follow-up study, with return visits scheduled at approximately weekly intervals for the first 4 weeks, followed by monthly visits for up to 3 additional months or until antibody seroconversion. TMA-reactive donors were classified as having confirmed WNV infection if WNV IgM was detected in either the index or follow-up specimens, if WNV RNA was detected in the index donation by repeat WNV TMA screening and confirmed by an alternate NAT assay, or if WNV RNA was found by TMA in a follow-up sample [7, 12, 23]. The aim of the 2003 study was to investigate viral and antibody dynamics for several weeks after detection of infection (table 1). Follow-up specimens were tested for WNV RNA by TMA and for IgM and IgG WNV-specific antibody (Focus Diagnostics). Persistence of low-level viremia (defined as a WNV RNA level less than the lower limit of ID-TMA detection) was investigated by performing 5 additional replicate ID-TMA assays on all 180 available serial follow-up samples from a subset of 56 subjects in the 2003 donor cohort from whom specimens were collected 30 days following the index donation. A similar protocol was followed for confirmed WNV-positive donors identified in 2004, except that an increased focus was placed on persistence of viremia over a longer interval ( 40 days after the index donation); hence, all donors were recalled weekly for 4 weeks and then monthly for 3 months. Because of limitations in the volume of samples, follow-up samples collected 40 days after the index donation were evaluated by ID-TMA and IgM and IgG enzyme immunoassays (EIAs) but not by replicate TMA. Samples from these earlier visits were used to conduct quantitative viral load and plasma cytokine and chemokine immunoassays; findings are reported elsewhere [24]. All samples collected 40 days after viremic donations were tested by 6-replicate TMA. For 2005, the follow-up algorithm was further modified to perform extensive immunological studies to characterize cellular immune response to WNV [25]. Weekly specimens were collected for 1 month, followed by monthly collection of specimens at 2, 3, 6, and 12 months; these longer-term samples were used in the present study to assess the persistence of WNV RNA and IgM, IgA, and IgG antibodies. For all 3 years, all available serial samples from any donor in whom RNA was detected 40 days after the index donation were tested for neutralizing antibody by WNV plaque-reduction neutralization testing (PRNT). PRNT was also performed on all serial samples from 23 additional representative cases to compare the time of IgM and IgG seroconversion with development of detectable neutralizing antibody activity. WNV serologic assays. Plasma specimens were tested for WNV IgM and IgG by use of US Food and Drug Administration (FDA) licensed enzyme-linked immunosorbent assay (ELISA) 2 JID 2008:198 (1 October) Busch et al.

Table 1. Characteristics of studies involving blood donors infected with West Nile virus, by year of cohort enrollment. Year Enrolled viremic donors, no. No. of index and follow-up specimens, no. Sampling frequency Study emphasis 2003 182 632 Weekly for 4 weeks, then monthly for 2 months or until seroconversion 2004 63 341 Weekly for 4 weeks, then monthly for 3 months 2005 35 509 Intensive sampling for first 4 weeks then monthly for 2 months, then sampling at 6 months and 1 year Viral and antibody dynamics early after infection; substudy of longer-term lowlevel viral persistence by replicate TMA that used samples collected 30 days after the index donation Viral and antibody dynamics early after infection; substudy of longer-term lowlevel viral persistence by replicate TMA that used samples collected 40 days after the index donation Dynamics of antibody persistence or seroreversion; immunologic response early after infection (samples were collected 28 days after the index donation) Total 280 1482...... NOTE. TMA, transcription-mediated amplification. kits (Focus Diagnostics). The assays were performed according to instructions in the package inserts [26, 27]. WNV IgA was measured using an in-house alpha-capture ELISA (Focus Diagnostics) modeled after the WNV IgM ELISA. PRNT assays for WNV and St. Louis encephalitis virus (performed to rule out cross-reactivity) were conducted using Vero cell monolayers in accordance with a standardized protocol developed by the Centers for Disease Control and Prevention [28]. Titers were expressed as the reciprocal of the plasma dilution that reduced or neutralized the challenge inoculum by 90%. Statistical methods. Actual donor return times varied, resulting in intermittent collection of blood specimens. Because changes in antibody or RNA status would have occurred during the interval between 2 particular donations, interval-censored data methods were used [29]. The median window periods (defined as the intervals between the selected viral RNA thresholds and antibody seroconversion, as well as the times to achievement of undetectable WNV RNA, IgM, and IgG levels) were derived from an accelerated failure time model [30]. The window period distributions were then estimated [31]. The 95th and 99th percentiles (interpreted as inclusion bounds) were derived from the estimated window period distributions. For purposes of illustration, the index donation (i.e., the donation in which WNV RNA was detected) was assumed to have occurred at the midpoint of the window of RNA detectability. The duration of this window has previously been estimated to be 6.9 days and begins a day or two following incident WNV infection [13]. We performed regression analysis using several linear mixedeffects models to determine the time at which IgM and IgA EIA seroreversion was detected by EIA. The mixed-effect models account for the multiple measures per donor by estimating random (i.e., separate) slope and intercept effects for each patient and fixed slope and intercept effects for the population represented by the sample of donors. We also chose the logtransformation of both dependent and independent variables (log-log method), as it provided the best data fit and improved conformance to the models assumptions [32]. The independent variable was the log-transformed EIA sample-to-calibrator ratio (S/C), and the dependent variable was the log-transformed duration after the index visit, restricted to 21 days in order to model only the descending section of the antibody-response curve. The models were fit to the data by use of the R statistical package [33]. Human subjects research approval. Donor screening and follow-up testing to establish donors true infectious status were conducted with the approval of the Western Investigational Review Board. Additional follow-up testing in 2004 and 2005 was approved by the University of California, San Francisco, Committee for Human Research. All donors gave written informed consent for WNV screening and follow-up testing. RESULTS Of 292 viremic donors identified by MP-NAT in 2003 2004, a total of 245 (84%) were enrolled in our follow-up studies. These donors contributed a total of 973 specimens that were examined for WNV RNA and antibody (table 1). The durations of various intervals (designated as window periods) were estimated as illustrated in figure 1 and detailed in table 2. The median window periods from RNA detection to IgM and IgG seroconversion were 3.9 days (95% CI, 3.4 4.4 days) and 7.7 days (95% CI, 6.9 8.5 days), respectively. The median time between IgM and IgG seroconversion was 3.4 days (95% CI, Virus and Antibody Dynamics in WNV Infection JID 2008:198 (1 October) 3

Figure 1. Dynamics of West Nile virus (WNV) RNA and WNV-specific antibody positivity and negativity. The top 3 intervals represent window periods that start with the index donation (i.e., the donation that tested positive for WNV in minipools of 16 donor specimens [MP] by transcription-mediated amplification [MP-TMA]). Although positive results of MP-TMA can occur anytime during the 6.9-day window period when results of MP nucleic acid amplification testing (NAT) are positive, for illustrative purposes the first 3 window periods are depicted as beginning at the midpoint of this window period. Values are not additive (i.e., the time from RNA detection to IgM detection [3.9 days] plus the time from IgM antibody to IgG antibody detection [3.4 days] does not equal the time from RNA detection to IgG antibody detection [7.7 days]), as noted in table 2. ID, individual donation; 1 ID, single replicate; 6 ID, 6 replicates. 2.6 4.3 days). The median time to RNA-negative results of a single-replicate TMA was 13.2 days (95% CI, 11.8 14.6 days). The 95% inclusion bounds for all of these intervals are given in table 2. The 95% and 99% inclusion bounds for RNA-negative results of ID-TMA were 41 days (95% CI, 21 50 days) and 50 days (95% CI, 22 85 days), respectively. Two subset analyses were performed for donors enrolled in 2003. The first evaluated 56 donors for WNV RNA persistence by performing TMA on 6 replicate aliquots from a given study visit. RNA was detectable for an additional 6.1 days (95% CI, 4.2 8.0 days) after results of single-replicate TMA were negative. In the second analysis, PRNT was performed to detect WNV- Table 2. Window periods of West Nile virus (WNV) detection, by results of tests for detection of antibodies and WNV RNA. Window period Duration, median 95% inclusion (95% CI), days a bound, days b From RNA positivity by MP-TMA to IgM positivity c 3.9 (3.4 4.4) 10 From RNA positivity by MP-TMA to IgG positivity c 7.7 (6.9 8.5) 16 From RNA positivity by MP-TMA to RNA negativity by single-replicate ID-TMA d 13.2 (11.8 14.6) 41 From IgM positivity to IgG positivity 3.4 (2.6 4.3) 8 From IgM positivity to RNA negativity by single-replicate ID-TMA 9.5 (7.7 11.3) 28 From IgG positivity to RNA negativity by single-replicate ID-TMA 5.8 (4.0 7.7) 17 From RNA negativity by single-replicate ID-TMA to RNA negativity by 6-replicate ID-TMA e 6.1 (4.2 8.0) 15 NOTE. See Subjects, Materials, and Methods for a description of window periods. ID, individual donation; MP, minipool of 16 donor specimens; TMA, transcription-mediated amplification. a Values are not additive (i.e., the time from RNA detection to IgM detection [3.9 days] plus the time from IgM antibody to IgG antibody detection [3.4 days] does not equal the time from RNA detection to IgG antibody detection [7.7 days]) because the window period components are not necessarily independent and the interval-censored methods adjust for the uncertainty of transition between successive window period components. b A 95% inclusion bound is the 95th percentile of the distribution window for an individual with incident WNV infection. c IgM was detected in the index donation for 53 of 280 donors, and IgG was detected in the index donation for 48 of 280 donors. For such donors, the time from the index donation to IgM positivity or IgG positivity was 0 days (not interval censored) and used in the analysis. d For policy purposes, we also calculated the 99% inclusion limit for this interval (see Results). e Determined from the 2003 donor cohort, whereas other parameters were determined from the 2003 and 2004 donor cohorts combined. 4 JID 2008:198 (1 October) Busch et al.

specific neutralizing antibody in serial samples from 23 donors; this activity was detected concurrently with IgM and/or IgG seroconversion in all cases. On the basis of results from the 2003 cohort, we performed a more extensive evaluation of cases to determine whether RNA could persist for 40 days. Four (3.2%) of 175 donors in the 2003 2005 cohorts showed this extended persistence of RNA when evaluated by 6-replicate TMA (table 3). In each case, a low percentage (e.g., 20% in 2003, 16.7% in 2004, and 33.3% in 2005) of the replicate TMA assays were reactive during a given study visit, indicating very low RNA concentrations. In 2 cases, RNA detection was intermittent, with partially reactive replicates separated by WNV RNA negative specimens. The donor with the longest duration of RNA detection had reactive results on days 57, 83, and 104 after the index donation. PRNT documented WNV-specific neutralizing activity (defined as a titer of 160) in all RNA-positive and RNA-negative samples collected 8 days after the index donation in all 4 persistent viremia cases, 2 of which showed IgM seroreversion (table 3, donors 3 and 4). In 2005, a total of 22 and 13 of 35 enrolled donors were followed up long enough to document complete loss of IgM and IgA, respectively; IgG persisted with no decrease in reactivity in all cases (figures 2 and 3). The mean times to IgM and IgA negativity were 156 days (95% CI, 70 423 days) and 220 days (95% CI, 48 2100 days), respectively. A minimum follow-up period of 3 months was achieved for 31 donors (89%), and 23 donors were followed up for the full 12 months. At 12 months, all 23 donors (100%) were IgG positive, 4 (17%) were IgM positive, and 13 (57%) were IgA positive (P.013, IgA positivity vs. IgM positivity). DISCUSSION Our data, combined with previously reported observations, allow us to formulate a comprehensive picture of the dynamics of viremia and antibody production in persons with WNV infection (figure 1). The dynamics of the very earliest phase of infection, defined as the time from a bite by an infected mosquito to the first detectable level of WNV RNA in plasma, are not known. Although viremia is first detectable 1 2 days after inoculation in experimental WNV infection involving immunosuppressed humans and mammalian model systems, it is unclear whether these same time frames are applicable to natural infection in healthy humans [34 37]. Once viremia becomes detectable by sensitive ID-NAT, there is a very brief period ( 1 day) when viremia is detectable only by ID-NAT but is below the limit of detection by MP-NAT. This level of WNV has been demonstrated to transmit infection by the transfusion route, which has led to the implementation of targeted ID-NAT donor screening during epidemics in specific locales [9, 11, 12, 15, 16]. Viremia is then detectable by MP-NAT for a mean of 6.9 days [13]; during this period, viral concentrations increase and then decrease below the threshold of MP-NAT detectability, as indicated by 2 independent reports that quantitated viral load in 579 NAT-reactive blood donations [11, 12]. The median viral load was 5 10 3 copies/ml for 90% of the MP-NAT positive donations detected before the development of IgM antibody and was below the limit of quantitation (defined as 50 100 copies/ ml) for IgM-positive donations [11, 12]. Our current data establish that IgM antibody becomes detectable a median of 3.9 days after the midpoint of the MP-NAT detection interval (which roughly coincides with the time of peak viremia), with IgM appearance occurring 1 day before viremia decreased to a level below the limit of MP-NAT detection [7, 11, 12, 21, 22, 24]. IgG develops within 3.4 days of IgM detection. Although IgA levels were not directly measured in our 2003 and 2004 cohorts, results from a previous study reported by our group and data from the current 2005 donor cohort (data not shown) indicate that IgA antibody development follows a very similar time course to that of IgM and is consistently detected several days before IgG [22]. Detection of WNV-specific neutralizing antibody activity by PRNT is also approximately coincident with detection of seroconversion by commercially available IgM and IgG EIAs. In the early stages of antibody development, plasma viremia remains detectable if sensitive NAT assays are used; we detected RNA for an additional 13.2 days by single-replicate ID-TMA and for a further 6.1 days by 6-replicate ID-TMA. In all cases, this tail-end viremia was found in association with IgM, IgA, and IgG, including WNV-specific neutralizing antibodies in the 4 cases in which RNA levels persisted for 40 days. Therefore, if RNA assays with sufficient sensitivity are used, it appears that antibody development generally precedes loss of detectable RNA by several weeks. The viral and antibody dynamics that we report are influenced by the assays we used. Our antibody assays could not detect antibody in immune complexes, and our RNA assay did not detect cell-associated WNV, which has been demonstrated to occur [38]. To explore these relationships, further studies would be needed. We previously estimated that IgM and IgA antibodies persist for 218 and 232 days, respectively, by extrapolating the rate of antibody EIA signal loss over time for a set of donors whose follow-up samples were generally obtained 180 days after viremic donations [22]. In the current study, we analyzed a data set that includes follow-up antibody results for 51 specimens from 35 donors that were collected 180 days after the index donation. We altered the log-linear model we used in the previous study in order to better fit these expanded data. IgM disappeared before IgA (mean times to seroreversion, 156 days for IgM and 220 days for IgA). IgM persisted for 1 year in 17% of cases; IgA persisted for 1 year in 57% of cases. This latter finding has important diagnostic implications, because the presence of IgM or IgA is not necessarily indicative of infection during a current- Virus and Antibody Dynamics in WNV Infection JID 2008:198 (1 October) 5

Table 3. Serial test results for 4 blood donors with West Nile virus (WNV) RNA detected >40 days after the index donation. Donor, by year Days after index donation TMA positivity a WNV antibody titer b Immunoglobulin detection, S/C ratio c IgA IgM IgG 2003 Donor 1 Index donation 0 2/2 10 NA NA NA Follow-up specimen First 4 2/2 10 1.0 0.90 1.30 Second 7 2/2 1:20 1.0 1.23 1.30 Third 11 2/2 1:640 20.5 5.96 1.30 Fourth 14 2/2 1:5120 24.38 2.25 2.25 Fifth 23 2/2 1:2560 21.32 5.65 3.02 Sixth 49 1/5 NA 9.28 5.21 3.80 Seventh 85 0/5 1:640 4.93 3.52 3.63 Eighth 144 0/5 NA 1.84 3.74 5.25 2004 Donor 2 Index donation 0 2/2 NA NA 0.90 1.30 Follow-up specimen First 33 NT 1:5120 NA 5.87 2.66 Second 40 2/5 1:2560 NA 6.18 2.46 Third 47 0/5 1:5120 NA 5.25 2.58 Fourth 56 NA NA NA NA NA Fifth 62 1/5 1:5120 NA 3.95 4.12 Donor 3 Index donation 0 2/2 10 NA 0.90 1.30 Follow-up specimen First 8 NT 1:40 NA 3.99 1.30 Second 21 NT 1:640 NA 4.94 2.62 Third 28 NT 1:320 NA 4.25 3.45 Fourth 35 NT 1:640 NA 3.75 3.51 Fifth 57 2/6 1:640 NA 2.09 2.88 Sixth 83 1/6 1:640 NA 1.03 3.82 Seventh 104 1/6 1:1280 NA 0.90 3.58 2005 Donor 4 Index donation 0 2/2 NA......... Follow-up specimen First 43 0/3 1:160 1.0 2.14 3.87 Second 48 1/4 1:160 1.23 1.78 4.07 Third 50 3/5 1:320 2.7 1.56 3.93 Fourth 55 2/5 1:160 1.31 1.39 4.0 Fifth 57 0/2 1:320 1.21 1.37 4.2 Sixth 62 0/4 1:320 1.08 1.25 4.63 Seventh 69 0/4 1:320 0.74 1.19 5.45 Eighth 76 1/4 1:320 1.02 0.96 5.38 Ninth 83 1/5 1:160 0.27 0.94 4.33 Tenth 90 0/5 1:320 0.82 0.91 4.94 Eleventh 181 0/3 1:160 0.84 0.55 4.19 NOTE. NA, not available; NT, not tested. a Data are no. of samples with a positive TMA result/no. of samples tested. b Determined by a plaque-reduction neutralization assay. A positive test result was defined as a titer 1:20. c Data are sample-to-calibrator (S/C) ratios determined by an enzyme immunoassay. A positive test result for IgA was defined as an S/C ratio 1.0; for IgM, an S/C ratio 1.1; and for IgG, an S/C ratio 1.5.

Figure 2. Sample-to-calibrator (S/C) index of West Nile virus (WNV) specific IgM (A), IgA (B), and IgG (C) for each donor, by time after the viremic donation. Each chart represents index ratios measured in untransformed linear scales. Positive values are 1.1 for IgM, 1.0 for IgA, and 1.5 for IgG.

Figure 3. Fitting of linear mixed models to the log-transformed sample-to-calibrator (S/C) ratio of West Nile virus (WNV) specific IgM antibody (A) and WNV-specific IgA antibody (B), by log-transformed time after the index donation. Diagonal lines, mean antibody S/C (95% confidence interval). year epidemic; in some cases, these antibodies could represent an immunologic response to WNV infection acquired during the previous mosquito season. We have further established that IgG persists at essentially the same titer (as indicated by stable S/C ratios) for at least 1 year. Because of the logistics of blood donor testing, it is very difficult to notify and enroll donors within the first several days after a viremic donation. Given rapid viremia development and seroconversion, all studies of these dynamic phenomena in blood donors have several limitations, including the lack of specimens collected at closely spaced intervals immediately after infection, the inability to determine the point in the MP-NAT detection interval during which each donor presented for donation, and intradonor variation in the time of sampling. In this study we used interval-censored data methods [29], which allowed us to make reasonably precise estimates of the median duration of window periods, associated confidence intervals, and inclusion bounds, akin to other studies [39 41]. Our finding that viremia is present for at least 7 days before development of IgM and persists for 9.5 days (95% CI, 7.7 11.3 days) after IgM seroconversion predicts that plasma RNA testing by use of sensitive assays might be useful in the early diagnosis of symptomatic WNV infection. This has recently been documented in patients who presented for diagnosis 8 days after developing WNV symptoms: approximately one-third of patients with acute WNV infection diagnosed were RNA positive and IgM negative [42]. In this early postsymptomatic period, 8% of patients tested positive for both RNA and IgM, similar to the rate of IgM positivity detected by MP-NAT among asymptomatic donors in our studies. Furthermore, the mean viral load among WNV RNA positive patients in the diagnostic setting was 7.5 10 3 copies/ml, which is very similar to that detected for asymptomatic donors by use of MP-NAT [7, 11]. This suggests that the sensitivity of the NAT used in the diagnostic laboratory (expressed by Tilley et al. [42] in plaque-forming units rather than copies/ml) was comparable to that of MP-TMA and that the dynamics of RNA loss and IgM development may be similar in asymptomatic and symptomatic cases. Further study will be needed to corroborate this hypothesis. Eight transfusion-transmitted cases of WNV infection have been documented since the implementation of MP-NAT in 2003, and in all cases that could be evaluated, transmission resulted from units that were retrospectively found to be RNA positive by ID-NAT and IgM and IgG negative [15, 16, 43]. These clinical surveillance data have generated the current working hypothesis that viremic units that are also positive for WNVspecific antibody are not infectious when transfused. This hypothesis is supported by in vitro and epidemiologic evidence. First, as demonstrated by our evaluation of 27 donors in this study, WNV-specific neutralizing antibody is detectable at approximately the same time as IgM and IgG are first detected and, in the 4 cases described in table 3, persisted throughout the longest intervals of viremia that we observed. Second, no documented transmissions from antibody-positive units occurred during 2003 before the implementation of MP-NAT, despite the likelihood that a large number of RNA-positive, IgM-positive, and IgG-positive units were transfused. We used our current data on WNV dynamics to estimate the number of antibody-positive blood components with low-level viremia (i.e., blood components in which WNV was detectable only by multiple replicate ID-NAT) that were transfused during 2003 2005. Our calculation is based on the relative durations of the window periods of MP-NAT positivity (6.9 days) and 6-replicate TMA positivity (6.1 days) and on nationally compiled data indicating that at least 1038 WNV-positive donations confirmed by MP-NAT were detected during 2003 2005 [42]. We estimate that at least 917 (1038 [6.1/6.9]) seropositive units with low-level viremia (i.e., WNV RNA levels below the 8 JID 2008:198 (1 October) Busch et al.

ID-NAT detection limit) were donated in these 3 years [44]. This likely would have resulted in the transfusion of 1330 WNVpositive components (917 donations 1.45 components per donation), because such components would not have been detected by MP-NAT and were also unlikely to have been detected by targeted ID-NAT. Because all 7 cases of posttransfusion WNV infection during 2003 2005 were traced to units that were negative for WNV-specific antibody but positive for WNV RNA by single-replicate ID-NAT, this strongly suggests that very lowlevel WNV viremia in the presence of WNV antibody is either not infectious or, if rarely infectious, does not produce clinical symptoms. Because the potential for transmissions from low-level viremic, antibody-positive units has not been ruled out by statistically powered human transfusion based look-back studies, the FDA has recommended that donors be deferred from donation for 120 days after detection of WNV RNA [45]. This recommendation was based on the infrequent persistence of RNA beyond day 40 after the index donation and an additional precautionary adjustment to allow for even longer persistence. In this report, we have estimated that 99% of donors with WNV viremia will clear WNV from the blood 50 days after the index donation, with an upper 95% confidence bound of 85 days. We found only 1 donor (0.6%) who failed to clear viremia before the 85-day upper confidence bound (RNA was detectable in this donor for 104 days after the index donation). Our estimates therefore suggest that the 120-day deferral period is a conservative precautionary policy, particularly because all samples with persistent viremia had high titers of WNV-specific neutralizing antibodies present. Combined with the clinical surveillance data discussed above, it is highly unlikely that a blood donation obtained after low-level WNV viremia becomes persistent would be infectious. We conclude that the FDA donor-deferral recommendations probably err on the side of caution and that there is no risk of WNV transmission through transfusion of blood from previously viremic donors who return to donate after their 120-day deferral interval and have blood samples obtained at the second donation that test negative for WNV by MP-NAT. References 1. Petersen LR, Marfin AA. West Nile virus: a primer for the clinician. Ann Intern Med 2002; 137:173 9. 2. Petersen LR, Hayes EB. Westward ho? the spread of West Nile virus. N Engl J Med 2004; 351:2257 9. 3. Watson JT, Pertel PE, Jones RC, et al. Clinical characteristics and functional outcomes of West Nile fever. Ann Intern Med 2004; 141:360 5. 4. Gea-Banacloche J, Johnson RT, Bagic A, Butman JA, Murray PR, Agrawal AG. West Nile virus: pathogenesis and therapeutic options. Ann Intern Med 2004; 140:545 54. 5. Pealer LN, Marfin AA, Petersen LR, et al. Transmission of West Nile Virus through blood transfusion in the United States in 2002. West Nile Virus Transmission Investigation Team. N Engl J Med 2003; 349:1236 45. 6. Dodd RY. Emerging infections, transfusion safety, and epidemiology. New Engl J Med 2003; 349:1205 6. 7. Busch MP, Tobler LH, Saldanha J, et al. Analytical and clinical sensitivity of West Nile virus RNA screening and supplemental assays available in 2003. Transfusion 2005; 45:492 9. 8. Stramer SL, Glynn SA, Kleinman SH, et al. Detection of HIV-1 and HCV infections among antibody-negative U.S. blood donors by nucleic acid amplification testing. NHLBI-REDS NAT Study Group. New Engl J Med 2004; 351:760 8. 9. Custer B, Tomasulo P, Murphy EL, et al. Triggers for switching from minipool testing by nucleic acid testing for West Nile virus: analysis of 2003 data to inform 2004 decision making. Transfusion 2004; 44:1547 54. 10. Kleinman S, Glynn SA, Busch M, et al. The 2003 West Nile virus United States epidemic: the America s Blood Centers experience. NHLBI Retrovirus Epidemiology Study (REDS). Transfusion 2005; 45:469 79. 11. Stramer SL, Fang CT, Foster, GA, et al. West Nile virus among blood donors in the United States, 2003 and 2004. N Engl J Med 2005; 353: 451 9. 12. Busch MP, Caglioti S, Robertson EF, et al. Screening the blood supply for West Nile virus RNA by nucleic acid amplification testing. N Engl J Med 2005; 353:460 7. 13. Busch MP, Wright DJ, Custer B, et al. West Nile virus infections projected from blood donor screening data, United States, 2003. Emerg Infect Dis 2006; 12:395 402. 14. Custer B, Busch MP, Marfin AA, Petersen LR. The cost-effectiveness of screening the US blood supply for West Nile virus. Ann Intern Med 2005; 143:486 92. 15. Montgomery SP, Brown JA, Kuehnert M, et al. Transfusion-associated transmission of West Nile virus, United States 2003 2005. Transfusion 2006; 46:2038 46. 16. Centers for Disease Control and Prevention. West Nile virus transmission South Dakota, 2006. MMWR Morb Mortal Wkly Rep 2007; 56: 76 9. 17. Orton SL, Stramer SL, Dodd RY. Self-reported symptoms associated with West Nile virus infection in RNA-positive blood donors. Transfusion 2006; 46:272 7. 18. Martin DA, Biggerstaff BJ, Allen B, Johnson AJ, Lanciotti RS, Roehrig JT. Use of immunoglobulin M cross-reactions in differential diagnosis of human flaviviral encephalitis infections in the United States. Clin Diagn Lab Immunol 2002; 9:544 9. 19. Tardei G, Ruta S, Chitu V, Rossi C, Tsai TF, Cernescu C. Evaluation of immunoglobulin M (IgM) and IgG enzyme immunoassays in serologic diagnosis of West Nile virus infection. J Clin Microbiol 2000; 38:2232 9. 20. Roehrig JT, Nash D, Maldin B, et al. Persistence of virus-reactive serum immunoglobulin M antibody in confirmed West Nile virus encephalitis cases. Emerg Infect Dis 2003; 9:376 9. 21. Prince HE, Hogrefe WR. Detection of West Nile virus (WNV)-specific immunoglobulin M in a reference laboratory setting during the 2002 WNV season in the United States. Clin Diagn Lab Immunol 2003; 10: 764 8. 22. Prince HE, Tobler LH, Lape-Nixon M, Foster GA, Stramer SL, Busch MP. Development and persistence of West Nile virus immunoglobulin M (IgM), IgA, and IgG during follow-up of viremic blood donors. J Clin Microbiol 2005; 43:4316 20. 23. Linnen JM, Deras ML, Cline J, et al. Performance evaluation of the Procleix West Nile virus assay on semi-automated and automated systems. J Med Virol 2007; 79:1422 30. 24. Tobler LH, Cameron MJ, Norris PJ, et al. Interferon and interferoninduced chemokine expression is associated with control of acute viremia in West Nile virus infected blood donors. J Infect Dis 2008; 198: XXX XX [in this issue]. 25. Lanteri M, Heitman J, Owen R, et al. Comprehensive analysis of West Nile virus specific T cell responses in human infection. J Infect Dis 2008; 197:1296 306. 26. Flavivirus (West Nile) IgM Capture ELISA [package insert]. Cypress, CA: Focus Technologies. Available at: http://www.focustechnologies.com. Accessed 27 September 2004. Virus and Antibody Dynamics in WNV Infection JID 2008:198 (1 October) 9

27. Flavivirus (West Nile) ELISA IgG [package insert]. Cypress, CA: Focus Technologies. Available at: http://www.focustechnologies.com. Accessed 27 September 2004. 28. Specter S, Hodinka RL, Wiedbrauk DL, Young S. Diagnosis of viral infections. In: Richman DD, Whitley RJ, Hayden FG, eds. Clinical virology. Washington, DC: ASM Press, 2002:263 4. 29. Lindsey JC, Ryan LM. Tutorial in biostatistics methods for intervalcensored data. Stat Med 1998; 17:219 38. 30. Cantor A. SAS survival analysis techniques for medical research. 2nd ed. Cary, NC: SAS Publishing, 2003. 31. Turnbull BW. The empirical distribution function from arbitrarily grouped, censored and truncated data. J Royal Statistical Soc Series B 1976; 38:290 5. 32. Inheiro JC, Bates DM. Fitting linear mixed-effects models. In: Pinheiro JC, Bates DM, eds. Mixed-effects models in S and S-plus. New York, NY: Springer-Verlag, 2002:133 97. 33. R Development Core Team. R: a language and environment for statistical computing. Vienna, Austria: R Foundation for Statistical Computing, 2006. Available at: http://www.r-project.org. Accessed 14 July 2008. 34. Southam CM, Moore AE. Induced virus infections in man by the Egypt isolates of West Nile virus. Am J Trop Med Hyg 1954; 3:19 50. 35. Daffis S, Samuel MA, Keller BC, Gale M Jr, Diamond MS. Cellspecific IRF-3 responses protect against West Nile virus infection by interferon-dependent and -independent mechanisms. PLoS Pathogens 2007; 3:e106. 36. Bourne N, Scholle F, Silva MC, et al. Early production of type I interferon during West Nile virus infection: role for lymphoid tissues in IRF3-independent interferon production. J Virol 2007; 81: 9100 8. 37. Shrestha B, Wang T, Samuel MA, et al. Gamma interferon plays a crucial early antiviral role in protection against West Nile virus infection. J Virol 2006; 80:5338 48. 38. Rios M, Daniel S, Chancey C, Hewlett IK, Stramer SL. West Nile virus adheres to human red blood cells in whole blood. Clin Infect Dis 2007; 45:181 6. 39. Hannigan A, O Mullane DM, Barry D, Schafer F, Roberts AJ. A reanalysis of a cries clinical trial by survival analysis. J Dent Res 2001; 80: 427 31. 40. Flandre P, Alcais A, Descamps D, Morand-Joubert L, Joly V. Estimating and comparing reduction in HIV-1 RNA in clinical trials using methods for interval censored data. J Acquir Immune Defic Syndr 2004; 35:286 92. 41. Breastfeeding and HIV International Transmission Study Group. Late postnatal transmission of HIV-1 in breast-fed children: an individual patent data meta-analysis. J Infect Dis 2004; 189:2154 66. 42. Tilley PAG, Fox JD, Jayarman GC, Preiksaitas JK. Nucleic acid testing for West Nile virus RNA in plasma enhances rapid diagnosis of acute infection in symptomatic patients. J Infect Dis 2006; 193:1361 4. 43. Centers for Disease Control and Prevention. Transfusion-associated transmission of West Nile virus Arizona, 2004. MMWR Morb Mortal Wkly Rep 2004; 53:842 4. 44. Stramer SL, Brodsky JP, Caglioti S, Strong DM. Confirmation of infection among donors identified by minipool and individual donation nucleic acid testing for West Nile virus in the United States from 2003 2005 [abstract]. Transfusion 2006; 46(Suppl):2A. 45. Center for Biologics Evaluation and Research, Food and Drug Administration. Guidance for industry: assessing donor suitability and blood product safety in cases of known or suspected West Nile virus Infection. June 2005. Available at: http://www.fda.gov/cber/gdlns/wnvguid.htm. Accessed 14 July 2008. 10 JID 2008:198 (1 October) Busch et al.