Frequency of human immunodeficiency virus (HIV) infection among contemporary anti-HIV- 1 and anti-HIV- 1/2

supplemental test-indeterminate blood donors

M.P. BUSCH, S.H. KLEINMAN, A.E. WILLIAMS, J.W. SMITH, H.E. OWNBY, M.E. LAYCOCK, L.L.L. LEE, c.-P. PAU, G.B. SCHREIBER, AND THE RETROVIRUS EPIDEMIOLOGY DONOR STUDY

Background: Follow-up studies from the mid-I 980s showed that 1 to 5 percent of blood donors testin reactive in anti-human immunodeficiency virus type 1 (HIV-1) enzyme im-

were in the process of seroconverting. The present study was conducted to establish the rate of HIV infection among contemporary anti-HIV-l/HIV type 2 (HIV-2) EIA-reactive, West- ern blot-indeterminate donors. Study Design and Methods: Donations (n = 607) with indeterminate HIV supplemental test results were identified by screening 3,021,342 donations given from November 1990 through August 1993 at five participating blood centers. Consenting donors were enrolled and samples taken 4 to 8 weeks after donation. Follow-up sera were tested by EIA and Western blot for anti-HIV-1 seroconversion and by type-specific peptide assays for anti- bodies to HIV-2 and HIV-1 subtype 0. Peripheral blood mononuclear cells and/or plasma from the follow-up samples were tested for HIV-1 DNA and/or RNA by polymerase chain reaction. The rate of HIV-1 infection among Western blot-indeterminate donors was also estimated by multiplying the incidence rate of HIV-I seroconversion in this donor popula- tion by the estimated duration of the EIA-reactive and Western blot-indeterminate window durin seroconversion (8.5 days). Resuis: Supplemental test-indeterminate donors (n = 355) enrolled a median of 38 days after donation; 265 (75%) of these donors were identified as indeterminate after an anti- HIV-1/2 EIA-reactive donation. Enrolled and non-enrolled donors had similar distributions of demographic characteristics and band patterns. Follow-up samples from all 355 do- nors tested negative for HIV-1 in polymerase chain reaction. Follow-up sera tested West- ern blot-negative in 54 cases (15%) and Western blot-indeterminate in 299 (84%). Two follow-up sera (0.6%) were interpreted, according to manufacturer's package insert crite- ria, as Western blot-positive with p24 and gp41 bands and/or gpl20/160 bands; however, paired testing of index and follow-up sera from these two cases showed identical Western blot and EIA reactivity, and polymerase chain reaction was negative for HIV RNAand DNA, which ruled out HIV infection. The absence of HIV infection in 355 Western blot-indeter- minate donors was consistent with our incidence-based model analysis, which yielded an estimate of one HIV-1 infection for every 21 5 Western blot-indeterminate donations (95%

Conclusion: Contemporary blood donors classified as indeterminate in supplemental HIV testing are infrequently infected with HIV. Donors whose follow-up samples test ne ative in anti-HIV-1/2 ElAs and negative or persistently indeterminate in Western blots saould be considered eligible for reinstatement.

munoassay ( E IA) and testing indeterminate in Western blot were infected with HIV-1 and

CI, 1 /39-1/8333).

Abbrevlatlons: EIA(s) = enzyme lmmunoassay(s); HIV-1 = human immunodeficiency virus type 1 ; HIV-2 = HIV type 2; PBMC(s) = peripheral blood mononuclear cell(s); PCR = polymer- ase chain reaction; REDS = Retrovirus Epidemlology Donor Study; WB(s) = Western blot(s).

From the Invin Memorial Blood Centers and the Department of Labo- ratory Medicine, University of California, San Francisco, San Francisco, California; the Jerome Holland Laboratory, American Red Cross, and Westat, hc., Rcckville, Maryland; the Department of Pathology and Labo- ratory Medicine, University of California. Los Angeles, Los Angeles, Cali- fornia; the Oklahoma Blood Institute, Oklahoma City, Oklahoma; the American Red Cross Blood Services, Southeast Michigan Region, Detroit, Michigan: and the Division of HIVIAIDS, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, Georgia.

Supported by contracts NOI-HB-97077 (superseded by NO1-HB- 471 141, -97078, -97079, -97080, -9708 I , and -97082 from the National Heart, Lung, and Blood Institute.

Received for publication April 24. 1995: revision received August 29, 1995, and accepted August 30, 1995.

TRANSFUSION 1996;36:374.

STUDIES FROM THE LATE 1980s designed to determine the frequency of human immunodeficiency virus (HIV) infection among Western blot (WB)-indeterminate blood donors showed that 1 to 5 percent of allogeneic donors with reactive enzyme immunoassay (EIA) and indeter- minate WBs were in early seroconversion.l-1° Concern over this low rate of infection has prevented the Food and Drug Administration from allowing re-entry of WB-in- determinate donors. It has also led to recommendations for additional testing of WB-indeterminate donors, either through the performance of supplemental tests on the index donation (e.g., p24 antigen, immunofluorescence

37

38 BUSCH ET AL. TRANSFUSION Vol. 36. No. 1-1996

assay, and polymerase chain reaction [PCR])12113 or by the repetition of EIA and WB on follow-up specimens ob- tained 1 to 6 months after the donation.'**

Because of recent changes in the demographics of HIV infection in the donor setting and in the performance of screening and supplemental HIV antibody assays, w e decided to reinvestigate the rate of HIV infection among contemporary WB-indeterminate donors. Changes likely to have led to a decrease in the rate of early HIV infec- tion among WB-indeterminate donors include 1) decreas- ing prevalence of HIV infection among donors, due to improved donor selection and interview-based screen- ingI4*I5; 2) increased sensitivity and reproducibility of Food and Drug Administration-licensed WBs (relative to earlier unlicensed reagents), such that truly infected per- sons are classified as WB-positive earlier in the process of s e r o c o n ~ e r s i o n ~ ~ - ~ ~ ; and 3) revision of WB interpretive criteria, such that the presence of two bands rather than three is adequate for defining a recently infected person as WB-p~s i t ive ." . '~ In addition, with the implementation of anti-HIV type 1 and 2 (HIV-1/2) combination EIAs in early 1992, the rate of EIA reactivity among noninfected donors increased significantly (from <O. 1 to >0.2%)20; this declining specificity led to an increase in the rate of WB-indeterminate donations, thereby reducing the pro- portion of WB-indeterminate donors likely to be in the HIV seroconversion phase. On the other hand, these new- format EIAs are more sensitive to early HIV-1 serocon- version, and therefore they may detect seroconverting donors earlier in the WB-indeterminate phase of sero- conversion.16,21-22 Finally, variant subtypes of HIV, includ- ing HIV-2 and HIV-1 subtype 0, are of increasing con- cern, and sera from persons infected with these variants frequently show WB-indeterminate pattern^.^^-^^

In this study, we estimate the rate of HIV infection among contemporary EIA-reactive and WB-indetermi- nate donors by two approaches, one a direct measure based on the follow-up of a cohort of WB-indeterminate donors and the second based on an incidence window-period

Materials and Methods Identification and enrollment of WB-indeterminate donors

The Retrovirus Epidemiology Donor Study (REDS) is a study of viral infection in blood donors, conducted by five par- ticipating blood centers (Irwin Memorial Blood Centers, San Francisco, CA; Oklahoma Blood Institute, Oklahoma City, OK; and American Red Cross Regional Blood Centers based in Baltimore, MD, Detroit, MI, and Los Angeles, CA), a coordi- nating center (Westat, Inc., Rockville, MD), and a central labo- ratory (SRA Technologies, Rockville, MD). As one element of REDS, demographic information and viral marker EIA reactivity and supplemental test findings for all donations at the partici- pating blood centers are tracked by the coordinating center. Donations are routinely screened for HIV antibodies by EIA (Abbott Laboratories, Abbott Park, IL) with confirmation of

EIA-repeatably reactive units by WB (Cambridge BioTech, Worcester, MA); they have also been screened by HIV-2 EIA (Genetic Systems Corp., Redmond, WA) since the implemen- tation of HIV-1/2 screening EIAs. WBs are graded according to Centers for Disease Control and Prevention and Food and Drug Administration criteria,".19 which specify that an inde- terminate interpretation be made when one or more visible bands are observed but criteria for positivity are lacking (i.e., 1+ reactivity for at least two of the following bands: p24, gp41, gp 120/160).

For this study, donors 18 years of age or older with reactive HIV-1 or HIV-112 EIAs and indeterminate HIV-1 WBs and/or reactive HIV-2 EIAs were identified as eligible for enrollment. Donors were recalled and notified of their HIV test results by the participating blood centers, according to each center's stan- dard operating procedures. At the time of notification and coun- seling, donors were asked to consent to the collection of a fol- low-up blood sample. For already enrolled donors, blood centers were asked to retrieve an aliquot of serum from the index do- nation. All testing was conducted under code in batches after the completion of subject enrollment in late 1993.

Laboratory studies

Follow-up sera from the enrolled indeterminate donors were tested for anti-HIV by EIA (HIV-112 EIA, Abbott) and WB (Cambridge BioTech). In selected cases, we performed HIV-1 EIA and WB testing in parallel on the index donation and fol- low-up sera; we used single reagent kits to distinguish changes due to reagent lot variability from changes in seroreactivity of individual donors over time. All follow-up sera were also tested with an unlicensed, supplemental HIV-1/2 assay employing a series of recombinant HIV-1 and HIV-2 antigens in a strip immunoblot format (HIV- 1/2 RIBA, Chiron Corp, Emeryville, CA); we used the results of this test to assess the validity of HIV-2 EIA reactivity. All sera were also tested for antibody reactivity to HIV-1 subtype 0-specific antigens by a series of peptide EIAs, as detailed elsewhere.*'

DNA was extracted from cryopreserved peripheral blood mononuclear cells (PBMCs, 5 x lo6 PBMCs/subject) and as- sayed for HIV-1 DNA by using a commercial PCR kit (Roche Molecular Systems, Branchburg, NJ) that includes the uracil- N-glycosylase carryover-prevention procedure.*' PBMCs from 30 HIV-1-positive samples were tested for HIV DNA under code in parallel; all 30 samples tested positive. We performed RNA PCR by using a commercially available kit (HIV-1 RNA Moni- tor Assay, Roche)28 on the index donation and follow-up sera for the four WB-indeterminate donors from whom PBMCs were not available. RNA PCR was also performed on index dona- tions and follow-up sera from the two donors in whom WB re- activity on the follow-up sera suggested possible HIV sero- conversion. Ten anti-HIV- I-positive donor sera were tested in parallel as controls; all 10 were positive for HIV-1 RNA.

Incidence window-period model The incidence of seroconversion from HIV- 1 EIA-negative

to HIV-1 EIA-reactive and WB-positive was determined for all donors who gave two donations between January 1, 1990, and December 31, 1993.26 We determined the denominator for the incidence rate by summing the individual observation periods for each donor, defined as days between their first and last donation to a REDS center during the observation period. Sero- converting donors were assumed to have seroconverted at the midpoint of the interdonation period preceding their seropositive donations.

TRANSFUSION 1 9 9 6 V o I . 36. No. I RATE OF INFECTION IN HIV-INDETERMINATE DONORS 39

1/91 6/91 1/92 6/92 1/93 6/93 Monthiyear

FIG. 1. Rate of EM-reactive (A), WB-indeterminate (*), and WB-posi- tive (+) donations during study period. Arrow indicates date of implemen- tation of anti-HIV-1R EIAs.

An estimate of the duration of the EIA-reactive, WB-inde- terminate window (8 .5 days) was derived from the study by Henrard et aLZ9 by testing serial samples from plasmapheresis donors who seroconverted to HIV seropositivity in anti-HIV- 1/2 EIA and WB. We then calculated the frequency of dona- tions from seroconverting donors given during the WB-indeter- minate window by multiplying the incidence of HIV-1 infection in repeat donors at REDS centers by the fraction of a year that a seroconverting subject tested EIA-reactive and WB-indeter- minate (83365.25). We then combined this number with the

80 "I 70 --

00 --

a 0, 50- - a K a 2 40 -- a

+

a 30 --

20 --

l o -- (1

overall rate of WB-indeterminate donations during the study period to estimate the proportion of donors with WB-indeter- minate donations who are HIV-seroconverting donors.

Results During the approximately 3-year study period ( 1 1/90 through

8/93), a total of 3,021,342 donations were given at the five participating blood centers. Figure 1 shows the rate of EIA re- activity over time according to overall WB interpretation. In contrast to the stable, low rate of donations testing EIA-reac- tive and WB-positive, the rate of EIA-reactive donations test- ing WB-negative and -indeterminate fluctuated markedly and tended to increase over time. Noteworthy is the marked increase in nonspecific EIA reactivity beginning in the fall of 1991, a finding attributed in part to influenza vaccination^.^^ There was also an increase in WB-negative and -indeterminate results af- ter the implementation of the anti-HIV-112 EIA into routine screening in 1992.

A total of 593 donations tested EIA-reactive and WB-inde- terminate during the study period (0.01 96% WB-indeterminate rate). Of these, 152 were identified by reactivity in anti-HN-l EIA and 441 by reactivity in anti-HIV-l/2 EIA. An additional 14 donations that tested reactive by anti-HIV- 112 EIA, but were negative in HIV- 1 WB, tested reactive by anti-HIV-2 EIA; these are included as eligible cases for the present study, because a reactive anti-HIV-2 EIA is given an overall indeterminate sta- tus on HIV-112 supplemental testing." Thus, a total of 607 do- nations were considered eligible for the follow-up study.

Figure 2 compares the donor demographics for the 607 do- nations that tested WB-indeterminate in supplemental testing with the demographics of donors of EIA-reactive and WB-posi- tive, EIA-reactive and WB-negative, and EIA-nonreactive do-

Sex Age Race or ethnicity FIG. 2. Comparison of demographic characteristics of HIV supplemental test-indeterminate donors (m) with those of WB-positive (H), WB-negative

(0). and HIV EIA-aonreactive (m) donors.

TRANSFUSION Vol. 36. No. 1-1996 BUSCH ET AL. 40

Table 1. Comparison of enrolledandnonenrolleddonors by HlV-7 and HlV-2 supplemental testpatterns Anti-HIV-1 EIA Anti-HIV-lI2 EIA

24 87 31 0 51 9 55 7 66 1 120 2 160 2 17and24 2 17/24 and 51 155 7 17/24 and 66 0 17/24 and 51 155 and 66 2 24 and 12011 605 0 51 and 55/66 1 55 and 66 2 55/66 and 12011 60 0 120/160 8 Nonviral onlyll 15 HIV-1 WB-negative,

HIV-2 El A-repeatably reactive, HIV-2 WB-negative

HIV-2 EIA-repeatably reactive, HIV-1 WB-negative,

HIV-2 WB 16/26 HIV-1 WB-negative,

HIV-2 El A-repeatably reactive, HIV-2 WB nonviral only

Total 152

57.2 0 5.9 4.6 0.7 1.3 1.3 1.3 4.6 0 1.3 0 0.7 1.3 0 5.2 9.9

4 56 0 6 5 0 1 1 1 3 0 0 0 1 1 0 5 8

92

HIV-1 WB Total Enrolled Not enrolled Total Enrolled Not enrolled band pattern’ Number Percent Number Percent Number Percent Number Percent Number Percent Number Percent 17 7 4.6 4.3 3 5.0 22 4.8 15 5.7 7 3.7

60.9 31 0 0 6.5 3 5.4 2 0 1 1.1 1 1.1 1 1.1 1 3.3 4 0 0 0 2 0 0 1.1 0 1.1 1 0 0 5.4 3 8.4 7

60

51.6 246 54.1 0 1 0.2 5.0 11 2.4 3.3 13 2.9 1.7 9 2.0 1.7 0 0 1.7 14 3.1 1.7 12t 2.6 6.7 17$ 3.7 0 11 2.4 3.3 6 1.3 0 3 0.7 0 11 2.4 1.7 2 0.4 0 3 0.7 5.0 1 0.2

11.7 59 13.0

5 1.1

7 1.5

2 0.4

136 1 9 9 5 0 7 7 7

10 3 3 3 2 3 1

34

4

5

1 455 265

51.3 0.4 3.4 3.4 1.9 0 2.6 2.6 2.6 3.8 1.1 1.1 1.1 0.7 1.1 0.4

12.7

1.5

1.9

0.4

110 0 2 4 4 0 7 5

10 1 3 0 8 0 0 0

25

1

2

1 190

57.9 0 1 .o 2.1 2.1 0 3.7 1.4 5.2 0.5 1.6 0 4.2 0 0 0

13.2

0.5

1 .o

0.5

Slash (4 in WB band pattern column indicates presence of either one or both of the antigen-specific bands listed. t One donation HIV-2 repeatably reactive (not enrolled). $Two donations HIV-2 repeatably reactive (1 enrolled, 1 not enrolled). 5 Intensity of p24 and/or gpl20/160 bands was 4 +. II If nonviral bands were present in addition to virus-specific bands, classification was based on virus-specific bands.

nations. Generally, the age, gender, and racial or ethnic distri- butions of the supplemental test-indeterminate donors are simi- lar to those of EIA-nonreactive and WB-negative donors. In particular, the overrepresentation in the WB-positive group of men, persons of non-Hispanic black race or ethnicity, and per- sons aged 25 to 34 years is not observed in the supplemental test-indeterminate donor group, which suggests that the inde- terminate group does not contain a large number of infected donors.

Of donors with indeterminate supplemental test results, 355 enrolled a median of 38 days after donation. These 355 donors gave 357 indeterminate donations, or 58.5 percent of the 607 eligible donations. Table 1 summarizes the distributions of donations from eligible, enrolled, and nonenrolled donors ac- cording to the EIA used to screen the index donation (HIV-I vs. HIV-1/2), the HIV-I WB band pattern, and HIV-2 supple- mental test results. Seventy-one percent of the index donations (431/607) were classified as WB-indeterminate on the basis of the presence of a single virus-specific band (with or without nonviral bands). This reactivity was directed against a gag antigen (p17, p24, p55) in 382 donations, a pol antigen (p31, p51, p66) in 3 1 donations, and an env antigen (gp120, gp160) in 18 donations. Reactivity to multiple viral bands was seen in 88 samples (14.5%). Reactivity to nonviral antigens only was the basis for an indeterminate classification in 74 donations (12.2%). Of the 14 samples with a negative HIV-1 WB and a

reactive HIV-2 EIA, 9 were HIV-2 WB indeterminate, while 5 tested negative in HIV-2 WB.

For the 355 enrolled donors, 92 donations (26%) were origi- nally identified as a result of anti-HIV-1 EIA screening, while 265 (74%) were detected by anti-HIV-112 screening. The rate of enrollment was similar for both screening periods (60% and 58%. respectively). Enrollment was also similar for donors with different WB band patterns. Finally, enrolled and nonenrolled donors were similar with respect to demographic characteris- tics, especially those characteristics that were seen more fre- quently among WB-positive donors (Fig. 2)-that is, male gen- der (47.9% vs. 48.8% for enrolled and nonenrolled donors, respectively), non-Hispanic black race or ethnicity (8.5% vs. 11.5%), and age 25 to 34 years (29.6% vs. 24.2%).

Table 2 summarizes the results of testing follow-up samples from the 355 enrolled donors. Anti-HIV- 1/2 was detected in 162 (45.6%) follow-up sera in EIA, with a significantly higher rate of reactivity in sera from donors whose index donation had screened reactive in anti-HIV-1/2 EIA (56.4%) than in the sera of donors originally screened by anti-HIV-1 EIA (14.3%). Over- all, 299 (84.2%) follow-up sera tested indeterminate and 54 (15.2%) tested negative in WB, with similar rates for anti-HIV-1 and anti-HIV- 1/2-screened index donations. Two follow-up sera were interpreted as WB-positive according to the manufac- turer’s package insert. In one case, the follow-up sample WB had p24, gp41, and gp120/160 bands (all I + intensity), whereas

TRANSFUSION 1 9 9 6 V o l . 36. No. 1 RATE OF INFECTION IN HIV-INDETERMINATE DONORS 41

Table 2. Results of HIV testing of follow-up samples from 355 blood donors with WB-indeterminate supplemental test results Donation HIV-112 EIA HIV-1 WB Number positive

EIA tested reactive (“A) Negative Indeterminate Positive’ PCR peptide subtype 0 EIA screening Number Number repeatably Number (%) HIV-1 HIV-2 HIV-1

HIV-1 91 13 (14.3) 7 (7.7) 83(91.2) l ( l . 1 ) 0 0 0 H IV- 1 /2 264 149 (56.4) 47(17.8) 216(81.8) l(0.4) 0 0 0

Total 355 162 (45.6) 54 (1 5.2) 299 (84.2) 2 (0.6) 0 0 0

* False-positive interpretations of WBs (see text).

the index donation had been reported as gp120/160 only. In the second case, the follow-up WB was considered positive on the basis of p24 (2+), gp41 (I+), p66 (+/-), and gp120 (+/-) reac- tivity, whereas the index-donation WB had only p24 and p66 bands.

HIV-I DNA PCR was negative in follow-up PBMCs from all 351 subjects for whom PBMCs were available, including the two with WB-positive results on follow-up samples. For the four donors lacking PBMCs, the index donation and follow-up sera were tested for HIV- 1 RNA by PCR; they were negative. HIV-I RNA PCR was also performed on the index donations and follow-up sera from the two donors classified as WB-posi- tive; all four samples tested negative. All 360 follow-up sera also tested negative for HIV-2 antibodies on the basis of spe- cific HIV-2 peptide reactivity in the HIV-1/2 RIBA and nega- tive for HIV-I subtype 0-specific antibody reactivity in pep- tide EIAs.

Although a high proportion of donors (299/355,84%) tested indeterminate in WB on both the index and follow-up samples, comparison of the results of WBs performed on the index do- nations and those obtained in the present study on the follow- up samples showed differences in band patterns. In 107 cases, there were significant changes in band patterns (defined as a gain or loss of at least one viral antigen band on the follow-up sample relative to the index donation: 5 1 gains, 3 1 losses, 25

both gains and losses). Of the remaining I92 donors in whom testing showed no major viral band gain or loss, 38 had changes in viral band intensity (+/- to > I + , or vice versa), and 78 had appearance or disappearance of nonviral bands.

To determine whether these changes in band patterns were the result of changes in the seroreactivity of donors over time or of differences in WB reagent lots, we performed side-by-side WBs on the index and follow-up sera from 12 donors showing significant band pattern changes. The change in viral band pat- terns was reproducible on side-by-side WB analysis in only 3 of the 12. In particular, the two donors with apparent sero- conversion (Subjects K and L in Fig. 3) had identical band patterns when index and follow-up sera were studied with WB from a single reagent lot. Thus, in the majority of cases, changes in WB band patterns were due to lot-to-lot variations in reagents, rather than to changes in donor seroreactivity over time. We also performed side-by-side anti-HIV-1R EIAs on five donors who appeared to have reverted from EIA-reactive to -nonreactive. Both samples tested nonreactive in all five cases, which indi- cates that changes in EIA reactivity were also reagent lot-re- lated, rather than due to changes in donor seroreactivity.

The follow-up study therefore showed that 0 of 355 inde- terminate donors were infected with HIV (95%, upper CI 1/97 [ 10.3/1000]). As an alternative approach for estimating the rate of HIV infection among HIV-indeterminate donors, we used

FIG. 3. Reproducibility of WB results on serial samples from WB-indeterminate donors. Side-by-side WBs were performed on the index donation and follow-up samples from five indeterminate donors whose follow-up sample WBs showed a major viral band not detected on the index donation WB (A-E). five donors with apparent loss in amajor viral band on the follow-up specimens (F-J). and the two donors whose follow-up specimens were classified as WB- positive (K. L). The change in viral band patterns was reproducible on side-by-side WB analysis in only 3 of the 12 donors: the appearance of a p24 band in two donors (A and E) and the loss ofa gp160 band in one ( G ) .

42 BUSCH ET AL. TRANSFUSION Vol.36.No. 1-1996

data on HIV seroconversion rates in blood donors to project the rate at which infected donors could be identified as EIA-reac- tive and WB-indeterminate. The rate of HIV seroconversion among repeat blood donors in REDS centers was determined to be 40 seroconversions per million person-years.26 By multi- plying the annual incidence rate by the fraction of a year dur- ing which an incident case would be expected to test WB-in- determinate (8.5 days divided by 365.25 dayslyear = 0.0233 years), we calculated that 0.93 WB-indeterminate donations per million donations may be from HIV-infected, seroconverting donors (40 seroconverting donations/106 person-years x 8.5 days divided by 365 daydyear). Given the overall rate of detecting EIA-reactive and WB-indeterminate donations in this study (607/3,02 1,342; 0.0200%), approximately 200 indeterminate donations are observed per million donations. Combining these data, we estimated that 1 in 215 WB-indeterminate donations (4.6511000) is derived from an HIV-infected seroconverting donor (95% CI = 1/39 to 1/8333 [0.12-25.65/1OOO]).

Discussion Our results indicate a very low rate of HIV infection

among donors with indeterminate HIV supplemental test results. We failed to detect any confirmed HIV- 1 or HIV- 2 infections during follow-up investigations of 355 con- temporary supplemental test-indeterminate donors. This follow-up study included PCR for HIV-1 DNA and/or RNA, as well as type-specific serologic studies to rule out HIV-2 and HIV-1 subtype 0. Using an alternative ap- proach based on modeling the rate of HIV seroconversion among donors and the duration of the indeterminate win- dow during seroconversion, we estimated that one true HIV-1 infection would be detected for every 215 WB- indeterminate donations. Thus, the empirical and mod- eled estimates are statistically consistent and reinforce the conclusion that HIV infection is rare among supplemen- tal test-indeterminate blood donors.

Although it is possible that failure to detect any sero- converting donors in our follow-up study may have been due to participation bias, we believe that this is unlikely, for several reasons. First, WB band patterns were very similar for enrolled and nonenrolled donors (Table 1). This was particularly true for those band patterns most frequently observed among seroconverting subjects (i.e., patterns with p24 or gpl2O/gp 160 rea~tivity).'~~'*~'~~*'~** Second, the demographic characteristics (age, sex, race or ethnicity) of enrolled and nonenrolled donors were similar, particularly with respect to the characteristics more commonly observed among known HIV-infected (i.e., WB-positive) donors. Third, no indeterminate do- nors who refused enrollment in the follow-up study were determined to be HIV-infected on the basis of additional testing performed by the participating blood centers.

The two cases in which donors appeared to have sero- converted to WB positivity, but who were subsequently established not to be infected, serve to emphasize the need for caution in interpreting WB results in a low-prevalence and low-incidence setting. False-positive WB results among donors with patterns similar to those observed in

our two cases have recently been reported in Australia3' and the United States.32 It has been estimated that up to 6 percent of all WB-positive results reported among blood donors may in fact be false-positive results.33 The present study shows that caution against overinterpretation of WBs should be exercised in the examination of follow- up samples from indeterminate donors as well as during initial supplemental testing of index donations. We rec- ommend the two approaches employed in the present study (side-by-side WB testing of index and follow-up sera and/or PCR analyses of follow-up samples) to resolve these cases. In addition to the two donors who were mis- classified as positive on the follow-up specimen, there were frequent changes in band patterns and band inten- sities when we compared WB results from the index and follow-up samples. These changes were largely attribut- able to lot-to-lot variability in viral lysate-based WB re- agents, rather than to biologic changes in donors.

Our results reinforce and update previous reports from the 1 9 8 0 ~ ~ - ~ - ~ ' and a recent report from a long-term, fol- low-up study of WB-indeterminate donors.34 These stud- ies all indicate a very low rate of HIV infection among donors with indeterminate supplemental test results. On the basis of these findings, we support the modification of donor re-entry algorithms to include donors with WB- indeterminate results who subsequently test negative in anti-HIV-1/2 EIA. If the follow-up WB is negative or, more typically, indeterminate, HIV infection is excluded and donor re-entry should be allowed. Unfortunately, our results showing persistent EIA reactivity of donors who were initially screened with recombinant antigen-based anti-HIV-1/2 EIAs indicate that only a minority of such donors will be eligible for re-entry until modifications to improve the specificity of the screening EIA are imple- mented.20

Irrespective of eligibility for reinstatement as blood donors, persons with indeterminate HIV supplemental test results should be reassured regarding their very low probability of HIV infection. Given this very low prob- ability, we do not recommend that donations with inde- terminate supplemental test results be routinely evaluated by DNA PCR, RNA PCR, or p24 antigen assays, because of the high cost and projected low yield of such test^.'^.'^ In the absence of HIV risk factors, supplemental test-in- determinate donors should be informed that supplemen- tal testing did not confirm HIV infection. If a donor ac- knowledges a risk factor or is particularly anxious, unequivocal evidence of the presence or absence of HIV infection can be obtained by the performance of a PCR on a properly saved, donation-derived sample or of a WB on a follow-up sample.

Acknowledgments The authors thank Roger Dodd, PhD (Transmissible Diseases

Laboratory, American Red Cross, Rockville. MD). for assistance in accessing index donation sera from ARC donors, Leslie Tobler, DrPH

RATE OF INFECTION IN HIV-INDETERMINATE DONORS 43 TRANSFUSION 1996-Vol. 36. No. I

(Irwin Memorial Blood Centers), for technical assistance; Kevin Watanabe and Robin McEntire, PhD (Westat), for data analysis; and Barbara Johnson (Irwin Memorial Blood Centers) for manuscript preparation. They acknowledge Ortho Diagnostic Systems, Chiron Corporation, and Roche Molecular Systems for supplying the WB, HIV- 112 RIBA, and HIV DNA and RNA PCR reagents, respectively, at no cost to the study. They also acknowledge the efforts of the nurses and donors who made this study possible.

1 .

2.

3.

4.

5 .

6.

7.

8.

9.

10.

11.

12.

13.

14.

15.

16.

References Kleinman S. The significance of HIV- 1 -indeterminate western blot results in blood donor populations. Arch Pathol Lab Med

Dodd R, Fang C. The western immunoblot procedure for HIV antibodies and its interpretation. Arch Pathol Lab Med

Celum CL, Coombs RW, Lafferty W, et al. Indeterminate human immunodeficiency virus type 1 western blots: seroconversion risk, specificity of supplemental tests, and an algorithm for evalu- ation. J Infect Dis 1991;164:656-64. Lefrere JJ, Courouce AM, Lucotte G, et al. Follow-up of sub- jects with isolated and persistent anti-core (anti-p24 or anti-pl7) antibodies to HIV. AIDS 1988;2:287-90. Van der Poel C, Lelie P, Reesink H, et al. Blood donors with indeterminate anti-p24 gag reactivity in HIV- 1 western blot: absence of infectivity to transfused patients and in virus culture.

Jackson JB, MacDonald KL, Cadwell J, et al. Absence of HIV infection in blood donors with indeterminate western blot tests for antibody to HIV-I. N Engl J Med 1990;322:217-22. Mariotti M, Rouger P, Thauvin M, et al. Failure to detect evi- dence of human immunodeficiency virus type l (HIV- l ) infec- tion by polymerase chain reaction assay in blood donors with isolated core antibodies (anti-p24 or -PI 7) to HIV- 1. Transfu- sion 1990;30:704-6. Dock NL, Kleinman SH, Rayfield MA, et al. Human immuno- deficiency virus infection and indeterminate western blot pat- terns. Prospective studies in a low prevalence population. Arch Intern Med 1991;151:525-30. Eble BE, Busch MP, Khayam-Bashi H, et al. Resolution of in- fection status of human immunodeficiency virus (H1V)-sero- indeterminate donors and high-risk seronegative individuals with polymerase chain reaction and virus culture: absence of persis- tent silent HIV type I infection in a high-prevalence area. Trans- fusion 1992;32:503-8. Coombs RW, Henrard DR, Mehaffey WF, et al. Cell-free plasma human immunodeficiency virus type 1 titer assessed by culture and immunocapture-reverse transcription-polymerase chain re- action. J Clin Microbiol 1993;3 1: 1980-6. Revised recommendations for the prevention of human immu- nodeficiency virus (HIV) transmission by blood and blood prod- ucts. Bethesda: Center for Biologics Evaluation and Research. Food and Drug Administration, April 23, 1993. Henrard DR, Phillips J, Windsor I, et al. Detection of human immunodeficiency virus type 1 p24 antigen and plasma RNA: relevance to indeterminate serologic tests. Transfusion

Courouce AM, Barin F, Maniez M, et al. Effectiveness of assays for antibodies to HIV and p24 antigen to detect very recent HIV infections in blood donors. The Retrovirus Study Group of the French Society of Blocd Transfusion (letter). AIDS 1992;6: 1548- 50. Ness PM, Douglas D, Koziol D, et al. Decreasing seroprevalence of human immunodeficiency virus type 1 in a regional blood donor population. Transfusion 1990;30:201-6. Petersen LR, Doll LS, White CR, et al. Heterosexually acquired human immunodeficiency virus infection and the United States blood supply: considerations for screening of po- tential blood donors. HIV Blood Donor Study Group. Transfu- sion 1993;33:552-7. Busch MP, Lee LLL, Satten GA, et al. Time course of detection of viral and serologic markers preceding human immunodefi-

1990; I14:298-303.

1990;114:240-5.

VOX Sang 1989;56: 162-7.

1994;34:376-80.

17.

18.

19.

20.

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

ciency virus type 1 seroconversion: implications for screening of blood and tissue donors. Transfusion 1995;35:9 1-7. Simon F, Pepin JM, Brun-Vezinet F, et al. Reliability of western blotting for the confirmation of HIV- 1 seroconversion (letter). Lancet 1992;340: 1541-2. Soriano V, GutiCrrez M, Heredia A, et al. Evaluation of different supplementary assays for the confirmation of HIV- 1 and HIV-2 infections (letter). Vox Sang 1994;66:82-3. Interpretive criteria used to report western blot results for HIV- I-antibody testing-United States. MMWR Morb Mortal Wkly Rep 199 1 ;40:692-5. Jurgensen P, Evans J, Skora C, et al. HIV-2: effect on specificity (abstract). Transfusion 1994;34(Suppl):44S. Gallarda JL, Henrard DR, Liu D, et al. Early detection of anti- body to human immunodeficiency virus type 1 by using an an- tigen conjugate immunoassay correlates with the presence of IgM antibody. J Clin Microbiol 1992;302379-84. Zaaijer HL, van Exel-Oehlers P, Kraaijeveld T, et al. Early de- tection of antibodies to HIV-I by third-generation assays. Lan- cet 1992;340:770-2. O’Brien TR, George JR, Holmberg SD. Human immunodeficiency virus type 2 infection in the United States. Epidemiology, diagno- sis, and public health implications. JAMA 1992;267:2775-9. Loussert-Ajaka I, Ly TD, Chaix ML, et al. HIV-IIHIV-2 seronegativity in HIV- I subtype 0 infected patients. Lancet 1994;343: 1393-4. Schable C, Zekeng L, Pau CP, et al. Sensitivity of United States HIV antibody tests for detection of HIV-I group 0 infections. Lancet 1994;344: 1333-4. Schreiber GB, Busch MP, Gilcher RO, et al. Incidence rates of infectious disease markers in repeat blood donors (abstract). Transfusion 1994;34(Suppl):7 1 S. Jackson JB. Ndugwa C, Mmiro F, et al. Non-isotopic polymerase chain reaction methods for the detection of HIV-I in Ugandan mothers and infants. AIDS 1991 ;5: 1463-7. Mulder J, McKinney N, Christopherson C, et al. Rapid and simple PCR assay for quantitation of human immunodeficiency virus type 1 RNA in plasma: application to acute retroviral in- fection. J Clin Microbiol 1994;32:292-3OO. Henrard DR, Phillips J, Phelps BH. et a]. Detection of HIV p24 antigen and plasma RNA can identify true HIV- I infection among EIA-reactiveheroindeterminate specimens (abstract). Vox Sang 1994;67(Suppl2):30. MacKenzie WR, Davis JP, Peterson DE, et al. Multiple false- positive serologic tests for HIV, HTLV-I, and hepatitis C follow- ing influenza vaccination, 1991. JAMA 1992;268:1015-7. Healey DS, Bolton WV. Apparent HIV-I glycoprotein reactivity on western blot in uninfected blood donors. AIDS 1993;7:655-8. Sayre KR, Dodd RY, Tegtmeier G. et al. False-positive human immunodeficiency virus type 1 Western blot tests in noninfected blood donors. Transfusion 1996;36:45-52. Genesca J, Shih JW, Jett BW, et al. What do western blot inde- terminate patterns for human immunodeficiency virus mean in EIA-negative blood donors? Lancet 1989;2: 1023-5. Jackson JB, Hanson MR, Johnson GM, et al. Long-term follow- up of blood donors with indeterminate human immunodeficiency virus type 1 results on Western blot. Transfusion 1995;35:98- 102.

Michael P. Busch, MD, PhD, Associate Professor in Residence, Department of Laboratory Medicine, University of California, San Francisco; and Vice-President, Research and Scientific Services, Irwin Memorial Blood Centers, 270 Masonic Avenue, San Francisco, CA 941 18-4496. [Reprint requests]

Steven H. Kleinman, MD, Professor, Department of Pathology and Laboratory Medicine, and Co-Director, Transfusion Medicine, University of California, Los Angeles, Los Angeles, CA.

Alan E. Williams, PhD, Research Scientist, Transmissible Dis- eases Laboratory, American Red Cross, Jerome Holland Laboratory, Rockville, MD.

James W. Smith, MD, PhD, Associate Medical Director, Okla- homa Blood Institute, Oklahoma City, OK.

44 BUSCH ET AL. TRANSFUSION Vol. 36. No. 1-1996

Helen E. Ownby, PhD, REDS Coordinator, American Red Cross Blood Services, SE Michigan Region, Detroit, MI.

Megan E. Laycock, BA, Research Assistant, Westat, Inc., Rock- ville, MD; current address: Research Project Coordinator, Scientific Services, Irwin Memorial Blood Centers.

Lillian L. L. Lee, MS. MT(ASCP), Assistant Director, Scientific Services, Irwin Memorial Blood Centers.

Chou-Pong Pau, PhD, Research Chemist, Division of HIV/AIDS, National Center for Infectious Diseases, Centers for Disease Control and Prevention, Atlanta, GA.

George B. Schreiber, DSc. REDS Principal Investigator, Westat, Inc.

The REDS Group is presently the responsibility of the follow- ing persons:

Blood Centers:

Potomac Region: A.E. Williams (Holland Laboratory), C.C. Nass American Red Cross Blood Services, Greater Chesapeake and

American Red Cross Blood Services, Southeastern Michigan Region: H.E. Ownby, D.A. Waxman

American Red Cross Blood Services, Southern California Region: S.H. Kleinman (University of California, Los Angeles, Medical Cen- ter), S. Hutching

Irwin Memorial Blood Centers: E.L. Murphy (University of Cali- fornia, San Francisco), M.P. Busch

Oklahoma Blood Institute: R.O. Gilcher, J.W. Smith Medical Coordinating Center:

National Heart, Lung, and Blood Institute: G.J. Nemo Steering Committee Chair: T.F. Zuck (Hoxworth Blood Center)

Westat, Inc.: G.B. Schreiber, R.A. Thomson