1379 Frequency of Indeterminate Western Blot Tests in Healthy Adults at Low Risk for Human Immunodeficiency Virus Infection Karen Midthun, Leslie Garrison,· Mary Lou Clements, Homayoon Farzadegan, Bruce Fernie, Thomas Quinn, and the NIAID AIDS Vaccine Clinical Trials Network From the Center for Immunization Research and Departments of International Health and Epidemiology, School of Hygiene and Public Health, and Department of Medicine, School of Medicine, Johns Hopkins University, Baltimore, and Division of Molecular Virology and Immunology, Georgetown University, Rockville, Maryland As part of a phase 1 trial of a candidate AIDS vaccine, blood specimens were collected from 168 healthy adult volunteers at minimal or no risk for becoming infected with human im- munodeficiency virus type 1 (HIV-l). These specimens were screened for evidence ofHIV-l infec- tion by enzyme immunoassay (EIA) and the Biotech/Du Pont Western blot (n = 168), culture (n = 122), and polymerase chain reaction assay (n = 20). None of the subjects had a positive test result by any of these assays, but 32% had indeterminate Western blot tests, most of which demonstrated a single band of low intensity. The most common bands were p24 (47%), p55 (34%), and p66 (36%); envelope bands were unusual (gp41, 2%; gp120, 2%). No serum specimen col- lected after 2-11 months from individuals with indeterminate Western blot results was positive by EIA or Western blot. There was 91% agreement in the test results ofthe first and second serum samples when the same lot of Western blot kit was used but only 36% agreement when different lots were used. The Biotech/Du Pont Western blot kit thus frequently yields indeterminate test results in the absence of HIV-Iinfection, the reproducibility of which is subject to lot-to-Iot vari- ability. Testing for human immunodeficiency virus type 1 (HN-l) infection is encouraged for those at high risk and is manda- tory for blood and organ donors, for individuals applying for military service, for immigration, and, in many cases, for life and health insurance policies [1]. As routine testing increases, concern has been expressed that the false-positive rate ofHN-l serologic tests might be unacceptably high among populations with a low prevalence of HIV-l infection [2]. Two recent studies conducted in low-risk populations concluded that the false-positive rate for detection of HN-l infection in the set- ting of a positive enzyme immunoassay (EIA) and Western blot (WB) was extremely low [3, 4]. However, many studies have noted the occurrence of indeterminate WB tests that have one or more bands but do not meet the criteria necessary for a positive result [4-13]. Indeterminate WB results lead to con- siderable confusion and anxiety because their meaning is un- clear. Received 18 June 1990. Presented in part: 28th Interscience Conference on Antimicrobial Agents and Chemotherapy, Los Angeles, October 1988 (abstract 835); V Interna- tional Conference on AIDS, Montreal, June 1989 (abstract T.B.P.l23). Informed consent wasobtained from all subjects, and experimentation guide- lines of the authors' institutions were followed. Financial support: AI-62515 (National Institutes of Health). Reprints or correspondence: Dr. Karen Midthun, Johns Hopkins Univer- sity Center for Immunization Research, 624 N. Broadway, Rm. 125, Balti- more, MD 21205. '" Present address: Immunex, Seattle. The Journal of Infectious Diseases 1990;162:1379-1382 © 1990 by The University of Chicago. All rights reserved. 0022-1899/90/6206-0025$01.00 As part of a multicentered phase 1 safety and immunoge- nicity trial of a candidate AIDS vaccine, we screened healthy adults at minimal or no risk for acquiring HIV-l infection. A surprisingly high percentage, although negative for HIV-l antibody by commercial EIA, had an indeterminate WB re- sult. Additional WB testing in individuals followed prospec- tively for 2-11 months allowed us to characterize the pattern of their test results over time and to determine the degree of lot-to-lot variability among different WB kits. Materials and Methods Participating sites. Six institutions, designated as AIDS vaccine evaluation units (AVEUs), participated in this study as part of an AIDS vaccine trial. They were located at the Johns Hopkins Univer- sity School of Hygiene and Public Health (Baltimore) and the Schools of Medicine of Baylor (Houston), Marshall (Huntington, WV), Maryland (Baltimore), Rochester (NY), and Vanderbilt (Nashville, TN) universities. The central laboratory was located at Georgetown University (Washington, DC). Subjects. Healthy volunteers 18-55 years old were screened be- tween December 1987 and April 1988 to determine their eligibility to participate in an AIDS vaccine trial [14]. Volunteers were excluded if they had a history of behavior that placed them at risk for acquir- ing HIV-l infection, of hepatitis B virus (HBV) infection or vaccina- tion, or of blood product transfusion within the preceding 6 months. Eligible volunteers were tested by EIA and WB for anti-HIV-l anti- bodies (n = 168), by HIV-l P 24 antigen test (n = 168), and by HIV-l culture (n = 122). They were also screened for serologic mark- ers of HBV. Volunteers who had an indeterminate WB result were ineligible to participate in the vaccine trial, regardless of negative tests for by Clark Baker on August 12, 2011 jid.oxfordjournals.org Downloaded from