HIV transmission from infected mother to fetus occurs in a significant number of pregnancies worldwide despite advances in techniques for preventing transmission(1). Timely detection of HIV infection is important for appropriate patient management. The current gold standard for detection of infant infection is isolation of HIV by culture of peripheral blood mononuclear cells(2). The polymerase chain reaction (PCR) assay for detection of viral DNA or RNA has been used for early detection of HIV infection(3,4). However, HIV culture is time consuming and expensive, and PCR requires special laboratory skills and equipment. Neither of these techniques is readily adaptable to use in developing countries with limited resources. There is need for a diagnostic tool that is sensitive and specific for detecting infection in infancy and that can be performed in most laboratories.
Although detection of p24 antigen in serum or plasma directly can be done as a simple enzyme-linked immunoassay and results rapidly acquired, its lack of sensitivity has limited its application(5). The sensitivity can be increased by immune complex dissociation (ICD) before assay(6). Studies of ICD p24 antigen in infants demonstrated its utility in early diagnosis(6,7). In this investigation, we describe its use in a large cohort of infants and their mothers enrolled in the Women and Infants Transmission Study (WITS), a multisite prospective natural history cohort study of congenital or perinatal HIV infection.
Informed consent was obtained from the parents or guardians of all subjects. The guidelines of the National Commission for the Protection of Human Subjects of Biomedical and Behavioral Research and requirements of Title 45 Code of Federal Regulations, Part 46, Protection of Human Subjects and the regulations of the authors' institutions were followed.
The infants that made up the study population were born of HIV-1 infected women enrolled in WITS. They were followed at U.S. centers in Chicago, eastern and central Massachusetts, and New York City and in San Juan, Puerto Rico. The women were enrolled while pregnant or within 7 days of delivery. The infants were seen at birth (with a window of 0 to 7 days), 1 month (±2 weeks), 2 months (±2 weeks), 4 and 6 months (±1 month), and periodically thereafter.
As of March 1994, the overall WITS cohort included 790 individual pregnant women, 897 pregnant enrollments (i.e., one to four enrollments per woman), and 657 enrolled infants, including 591 first enrolled singleton infants. Of these, 207 infants had known HIV infection status (47 infected, 160 uninfected) and one or more ICD p24 antigen assay results available through 6 months of age (average 3.1). Most exclusions were caused by indeterminate infection status of the infant or lack of samples for analysis. These 207 infants and their mothers(who had a delivery plasma specimen or a specimen from approximately 34 weeks' gestation) formed the basis for this study. Specimens were available from delivery for 181 mothers, from only 34 weeks of gestation for 16, and from earlier in the gestations for 10. These were combined in this report and referred to as samples taken "at delivery."
Zidovudine was used primarily for therapeutic reasons in this cohort. Of the 197 mothers included, 34 received some zidovudine during pregnancy (3 in ACTG 076), 161 did not, and treatment history could not be ascertained for 2. Of the 47 HIV-1-infected infants, 7 received at least some zidovudine during the first 6 months of life.
Diagnosis of HIV Infection
The diagnosis of HIV infection was based on an algorithm developed from our experience with HIV culture(2). It used the AIDS Clinical Treatment Group (ACTG) consensus quantitative co-culture technique for most and the qualitative co-culture technique for a few(8). HIV concentration in the quantitative technique was defined as infectious units per million mononuclear cells (IUPM)(8). Each site participated in the National Institute of Health Division of AIDs virology quality assurance program.
An infected infant was defined by having two or more positive HIV cultures at any age. The lack of infection was defined by two or more negative cultures, at least one at 1 month or later and at least one at 6 months or later, and no positive cultures at any age. Infants not meeting these criteria were defined as having indeterminate infection status and were excluded from this analysis.
Immune Complex Dissociation p24 Antigen Determination
Blood was drawn from peripheral veins and serum or plasma separated and stored at −70C in aliquots at a central repository for up to 4 years before analysis. Samples were frozen and thawed only once.
The ICD p24 antigen determination was performed at a single site(University of Illinois at Chicago, IL, U.S.A.). ICD was performed by acid dissociation using a commercially available kit (Coulter ICD-Prep Kit, Coulter, Miami, FL, U.S.A.) according to the manufacturer's instructions. HIV-1 p24 antigen was then detected by the routine enzyme immunoassay procedure (Coulter HIV-1 p24 Antigen Assay, Coulter, Miami, FL, U.S.A.). Assays were read with an automated microplate enzyme immunoassay reader and quantified using a standard curve established according the kit instructions and reagents and the interpretive software of the reader. Positive cutoffs for each run were determined by adding a predetermined factor of 0.055 to the mean value of three negative controls included on each plate as specified by the manufacturer.
Values corresponding to optical densities below the cutoff on the standard curve could not be accurately quantified because of the nature of enzyme linked immunoabsorbent assays. For some of the statistical analyses, concentrations for optical density readings below the cutoff were assigned a value equal to one half of the smallest concentration corresponding to optical densities above the cutoff.
All samples for a given mother and infant were run on a single plate using the Coulter standards. The patient samples were run singly because of sample limitations, and the standard curves and known positive and negative plasma control samples were run in duplicate with close agreement between replicates. Neutralization was not performed in the case of positive samples that were in agreement with the known infection status. In 8 of 11 instances in which the assay was positive for infants known to be uninfected, specimens were available and were subjected to neutralization according to the manufacturer's instruction.
Lymphocyte phenotyping was done using a standard whole EDTA-anticoagulated blood lysis technique described elsewhere(9,10). All sites participated in the ACTG flow cytometry proficiency testing program and a special WITS quality control program.
Confidence intervals for sensitivity and specificity at each visit were calculated using the exact binomial method(11). The sensitivity and specificity of a particular number of assays (e.g., two) were estimated by calculating for each infant the expected probability of observing one or more assays positive if only two assays were performed. If more than two assays were performed for an individual, all possible pairs of assays were considered and the probabilities averaged to calculate the expected probability. These probabilities were then averaged across infants to obtain the overall estimate. This technique allowed the use of all available pertinent data for calculating test performance characteristics, rather than limiting it to only those patients who had exactly the indicated number of assays. The bootstrap method was used to obtain confidence intervals(12).
Fisher's exact test(13) and logistic regression (14) were used to assess categorical covariates as predictors of transmission. Quantitative levels of ICD p24 were compared between transmitting and nontransmitting women using a two-sample Wilcoxon test(11). Paired comparisons between quantitative ICD p24 and other quantitative variables were assessed with a Spearman rank correlation(11).
Rates of change in ICD p24 antigen levels between adjacent visits were calculated on a logarithmic scale among infected infants having data at both visits. Exact (rather than nominal) ages at each visit were used. These slopes can be interpreted as change in logs per month. Wilcoxon signed rank tests(11) were then used to test for a change in level between each pair of adjacent visits.
To assess the relation between quantitative HIV culture and ICD p24 antigen results, the log of infectious units per million (IUPM) was regressed on log ICD p24 levels at the same visit. The bootstrap method was used to estimate standard errors by resampling on the level of the individual(12). All logarithms are base 10, and p values are two-sided.
Diagnosis of Perinatal HIV Infection
ICD p24 was detected in 27.3% of assays from HIV-infected infants at ≤7 days of age (Fig. 1). At 1 month of age, 76.9% of assays from known infected infants were positive. Thereafter, the proportion of positive assays remained in the range of 71.0% to 80.8% through 6 months of age. Ten percent of the assays from uninfected infants at ≤7 days of age were positive (i.e., false positive) so the specificity at this early age was of 89.8% but thereafter, between 1 and 6 months of age, the specificity was 98.3% to 100%.
Eighty-one percent (38 of 47) of the infected infants and 86% (136 of 158) of the uninfected infants had more than one assay performed at 1 to 6 months. The sensitivity for correctly identifying an infected infant who had one, two, or three assays was 81.0% (95% CI = 70.7% to 89.8%), 86.8% (76.7% to 97.0%), and 87.5% (70.8% to 100.0%), respectively. The specificity for a single assay was 99.2% (97.7% to 100.0%), two assays was 99.5% (98.5% to 100.0%), and three assays was 98.9%(96.6% to 100.0%).
The origin of the false-positive results at ≤7 days of age was explored. Five had sufficient specimen volume from cord blood or peripheral blood from ≤7 days to perform neutralization studies. All had insignificant reduction in ICD p24 antigen as would be expected for a false-positive result. Only 32.5% of the mothers had a positive assay result at delivery.
Immune Complex-Dissociated p24 Antigen Concentration as a Measure of Circulating Viral Load
The median concentration among infected infants at<7 days of age was below the cutoff (Table 1). The median concentration increased and ranged from 107.1 to 317.5 pg/ml between 1 and 6 months of age. The median maternal delivery specimen ICD p24 antigen concentration was only 9.9 pg/ml for transmitting mothers and below the cutoff for nontransmitting mothers.
The change in concentration in samples for a given patient between adjacent visits was examined. From birth to 1 month of age, 80% (12 of 15) of infants had an increase in concentration. The slope or rate of increase was substantial. The mean log change in pg/ml(log10[ICD p24]) was 1.90 ± 1.86 (mean ± 1 SD) from birth to 1 month (p < 0.001) or an increase of 79 times. The mean changes in concentration between 1 and 2 months and between 2 and 4 months of age were not significant. Between 4 and 6 months of age, however, the change increased significantly (mean change of 0.115 logs per month, p = 0.02). When the analysis was repeated ignoring any ICD p24 antigen data obtained after zidovudine treatment, the slopes remained similar to those reported earlier, and the statistically significant intervals(birth to 1 month and 4 to 6 months) remained the same.
The relation between the ICD p24 antigen concentration and the level of circulating infectious virus measured by quantitative culture was also examined. There was a significant relation between the ICD p24 antigen concentration and the concurrent IUPM. For the mothers. IUPM increased 1.18 logs (15.1-fold) for every log (10-fold) increase in ICD p24 antigen concentration (β ± SE= 1.18 ± 0.56, p = 0.04). For the 16 infants who had an average of 2.13 paired results for analysis, IUPM increased 0.82 logs (6.6-fold) for each log increase in ICD p24 antigen (0.82 ± 0.41, p = 0.05).
Maternal Immune Complex-Dissociated p24 and Perinatal HIV Transmission
Detection of ICD p24 antigen in the mother at delivery was associated with an increased risk of transmission to her infant (OR = 3.17, 95% CI = 1.58 to 6.37, p = 0.002, Fisher's exact test) (Table 2). The transmission rates in women with and without detectable ICD p24 were 35.9% and 15.0%, respectively. Therefore, the risk of transmission was higher if the mother had detectable ICD p24 antigen at delivery, but its presence was not accurately predictive of transmission (i.e., positive predictive value of 35.9%, negative predictive value of 85.0%).
The association between qualitative ICD p24 (above versus below cutoff) and transmission remained significant even after adjusting for covariates found to be predictors of transmission in previous analyses: duration of ruptured membranes (≤4 hours versus >4 hours), mean maternal CD4% during pregnancy (<29% versus≥29%), and maternal hard drug use during pregnancy(yes versus no)(15). The adjusted odds ratio (OR) was 2.25 (p = 0.04). In a separate model, the qualitative ICD p24 covariate and the maternal HIV culture result from samples drawn at delivery were included simultaneously as covariates. The ICD p24 covariate was significant(OR = 2.52, p = 0.01), and the viral culture covariate was marginally significant (OR = 2.28, p = 0.12), suggesting that detectable ICD p24 may carry more predictive information than culture positivity alone.
The concentration of maternal ICD p24 antigen was higher in transmitting than nontransmitting women (p = 0.001 by Wilcoxon rank sum; see Table 1). However, the concentration had no effect on the risk for transmission. The concentrations in transmitting women with measurable ICD p24 antigen above the positive or negative cutoff were not different from those who did not transmit(p = 0.62), even if mothers who took zidovudine were excluded from analysis (p = 0.48). Similarly, among women above the cutoff, the transmission rate for those in the highest quartile ICD p24 antigen concentration was nearly identical to those below the 75% percentile(37.5% and 35.4%, respectively).
The maternal ICD p24 concentration had an association with the infected infant concentration only at 1 month of age (Spearman rank correlation = 0.49, p = 0.01). It had no effect at birth or 2 through 6 months of age (all p > 0.15).
Immune Complex Dissociated p24 Antigen Concentration and the Immune System in Infants
Higher ICD p24 antigen levels were significantly associated with lower CD4+ concentrations at 1, 2, and 6 months (Spearman correlation, p ≤ 0.05, n = 14 to 25 at different ages). The ICD p24 antigen concentrations in the infants at 2 and 4 months of age were positively associated with the concentration of CD8+HLA-DR+-bearing lymphocytes.
We applied the ICD p24 antigen assay to infants and women enrolled in WITS, a large multifaceted cohort study of perinatal HIV infection. The ICD p24 serum or plasma antigen assay was a useful diagnostic tool in perinatal HIV infection at 1 to 6 months of age, but not in the first week of life. Although the sensitivity of a single ICD p24 antigen test was moderately high (71% to 81%) for ages 1 to 6 months, this assay was not as sensitive for correctly identifying infection as other techniques such as HIV-1 culture (82% to 91%) and DNA PCR (90% to 100%)(2,5,16). The specificity was, however, very high (98% to 100%). The low cost and ease of performing the assay suggested that the assay has value as a diagnostic test, particularly when more sophisticated tests are not available or as an independent supplemental test in conjunction with HIV-1 DNA PCR or HIV-1 culture.
Only a limited number of infected infants are identified as positive by HIV-1 culture (24%) or DNA PCR(29%) in the first few days of life(2,5,16). Similarly, ICD p24 antigen was detected at ≤7 days of age in only 27%. It is possible that the lower proportion of infected infants who had positive assays in the first week of life reflected the proportion of infants who were infected in the immediate intrapartum period(17). The apparent lack of sensitivity may be a biologic effect and not a deficiency in the assay.
The lower specificity of ICD p24 antigen at ≤7 days of age (89.8%) compared with later is not as readily explained. It is possible that factors such as the more frequent occurrence of hemolysis in newborn samples or excess bilirubin may explain some of them. It is also possible that p24 protein or protein fragments complexed with antibody from the HIV-infected mother are actively transported across the placenta and are detected by the assay. We suggest that the assay should not be used as an isolated diagnostic test in the first week of life.
The ICD step increases the sensitivity of p24 antigen determination. Previous studies without ICD reported lower sensitivity, possibly because of inability to detect immune-complexed p24 antigen(5,6). The acid dissociation step increases the detectable p24 antigen and increases the sensitivity of the assay.
Miles et al. evaluated ICD p24 antigen determination as a diagnostic tool in a group of 40 infants with cord blood specimens and found that 5 of 8 who were known to be infected had positive ICD p24 antigen assays at birth and 20 of 22 uninfected had negative assays(7). Others have found similar or higher levels of sensitivity of ICD p24 antigen detection as a diagnostic tool in infants beyond the newborn period(6).
Some investigators suggest that there is a need for neutralization to confirm a positive result(18). The very low occurrence of false-positive results in this study outside the immediate newborn period (2 of 158 infants) argues against the need to use it routinely, particularly when it is used in conjunction with other techniques.
ICD p24 antigen concentration and quantitative HIV-1 culture results were found to be significantly associated with a ratio of roughly 1:1. However, there is considerable scatter in the results, which might be explained by the fact that the two methods probably assess different aspects of infection. Quantitative culture is a function of the number of cells with replicating virus, and the ICD p24 antigen assay is a function of circulating plasma HIV-1 virus. The results are likely to be the consequence of slightly different virologic and immunologic mechanisms.
We found, as did St. Louis et al. that the qualitative presence of detectable ICD p24 antigen was an important risk factor in determining HIV-1 transmission(19). The association was not substantially affected by adjusting for other covariates such as duration of rupture of membranes, maternal CD4+ lymphocyte percentages, and maternal drug use during pregnancy. However, the proportion of women who were ICD p24 antigen positive in St. Louis' study was much lower, only 6% compared with 32% in the current study, and therefore the direct comparison of the African cohort to the U.S. cohort must be questioned. The concentration of ICD p24 antigen, compared with its mere presence, did not correlate with risk of transmission in our study or in the study by Jackson et al.(20). Others, in contrast to the results of this study, have not found a significant relation between the presence of maternal ICD p24 antigen and HIV transmission(5).
In this study, the concentration of the infant ICD p24 antigen was inversely correlated with the concentration of CD4+ T cells at various time points and positively correlated with the concentration of activated CD8+ T cells (CD8+HLA-DR+). In advancing HiV disease, the levels of CD4+ T cells fall, and activated levels of CD8+ T cells usually rise(21) as the circulating viral load increases. We questioned why the correlation was not sustained throughout the 6-month observation period. The immunologic consequence of HIV infection is a complex, dynamic process that is further complicated in infants by the ongoing maturation of the immune system. We suggest the possibility that that an initial burst of viral production destroys many of the CD4+ T cells before the mechanisms of immune activation come into play. Activated CD8+ cytotoxic lymphocytes dampen the exuberant burst of viral production that occurs shortly after infection. However, activated CD8+ T lymphocytes are more readily destroyed by apoptotic and other mechanisms, resulting in a fall in the concentration(22). The sum of the consequences of the viral load factors and the immune system may vary with time.
When used with an awareness of its performance characteristics and limitations, the ICD p24 antigen assay provides a potentially useful diagnostic adjunct for assessment of an infant's infection status. It may have particular value when other, more sensitive techniques are not available and as a technique for understanding the changes that occur in congenital or perinatal HIV infection. Further study is suggested to examine whether it is possible to improve its diagnostic efficiency by combining it with other assay techniques.
Acknowledgments: Financial support was provided by National Institutes of Health grants (AI-82507, AI-82506, AI-85005, AI-82505, and AI-05072; HD-8-2913). The material was presented in part at the Third Conference on Retroviruses and Opportunistic Infections, January 28 through February 1, 1996, Washington, D.C. Principal investigators, study coordinatiors, and program officers include Ruth Tuomala, Ellen Cooper, Donna Mesthene (Brigham and Women's Hospital and Boston City Hospital, Boston, MA), Harold Fox, Jane Pitt, Alice Higgins (Columbia Presbyterian Hospital, New York, NY), Sheldon Landesman, Hermann Mendez, Gail Moroso(State University of New York, Brooklyn, NY), Clemente Diaz, Edna Pacheco-Acosta (University of Puerto Rico, San Juan), Kenneth Rich, Geraldine Alexander (University of Illinois at Chicago, IL), Mary Glenn Fowler. Judy Lew (National Institute of Allergy and Infectious Diseases. Bethesda, MD), Lynne Mofenson, Jack Moye (National Institute of Child Health and Human Development, Bethesda, MD), and Sonja McKinlay, Kathy Sherrieb (New England Research Institute, Watertown, MA).
1. Mofenson LM. Epidemiology and determinants of vertical HIV transmission. Semin Pediatr Infect Dis
2. McIntosh K, Pitt J, Brambilla D, Carroll S, Diaz C, Handelsman E, Moye J, Rich K. Blood culture in the first 6 months of life for the diagnosis of vertically transmitted human immunodeficiency virus infection. J Infect Dis
3. Kline MW, Lewis DE, Hollinger FB, et al. A comparative study of human immunodeficiency virus culture, polymerase chain reaction and anti-human immunodeficiency virus immunoglobulin A antibody detection in the diagnosis during early infancy of vertically acquired human immunodeficiency virus infection. Pediatr Infect Dis J
4. Palumbo P, Kwok S, Waters S, et al. Viral measurement by polymerase chain reaction-based assays in human immunodeficiency virus-infected infants. J Pediatr
5. Borkowsky W, Krasinski K, Pollack H. Hoover W, Kaul a, Ilmet-Moore T. Early diagnosis of human immunodeficiency virus infection in children <6 months of age: comparison of polymerase chain reaction, culture, and plasma antigen capture techniques. J Infect Dis
6. Quinn TC, Kline R, Moss MW, Livingston RA, Hutton N. Acid dissociation of immune complexes improves diagnostic utility of p24 antigen detection in perinatally acquired human immunodeficiency virus infection. J Infect Dis
7. Miles SA, Balden E, Magpantay L, et al. Rapid serologic testing with immune-complex-dissociated HIV p24 antigen for early detection of HIV infection in neonates. N Engl J Med
8. Hollinger FB, ed. ACTG virology manual for HIV laboratories, version 2.1
. Bethesda. MD: Division of AIDS, National Institute of Allergy and Infectious Diseases, National Institute of Health, September 1993.
9. Clavelli T, Denny TN, Paxton H, Gelman R, Kagan J. Guideline for flow cytometric immunophenotyping: a report from the National Institute of Allergy and Infectious Diseases, Division of AIDS. Cytometry
10. Landay AL, Brambilla D, Pitt J, et al. Interlaboratory variability of CD8 subset measurements by flow cytometry and its application to multicenter clincal trials. Clin Diagn Lab Immunol
11. Rosner B. Fundamentals of biostatistics
. 3rd ed. Boston: PWS-Kent, 1990.
12. Efron B, Tibshirani RJ. An introduction to the bootstrap
. New York: Chapman & Hall, 1993.
13. Mehta CR, Patel NR. A network algorithm for performing Fisher's exact test in rxc contingency tables. J Am Stat Assoc
14. Hosmer DW, Lemeshow S. Applied logistic regression. New York: John Wiley & Sons, 1989.
15. Landesman SH, Kalish LA, Burns DN, et al. Obstetrical factors and the transmission of human immunodeficiency virus type 1 from mother to child. N Engl J Med
16. Bremer JW, Lew J, Cooper E, et al. Diagnosis of infection with human immunodeficiency virus type 1 by a DNA polymerase chain reaction assay among infants enrolled in the Women and Infants Transmission Study. J Pediatr
17. Bryson YJ, Luzuriaga K, Sullivan JL, Wara DW. Proposed definitions for in utero versus intrapartum transmission of HIV-1. N Engl J Med
18. Fauvet M, Henrard D, Delage G, et al. Correspondence: Early detection of HIV in neonates. N Engl J Med
19. St. Louis ME, Kamenga M, Brown C, et al. Risk for perinatal HIV-1 transmission according to maternal immunologic, virologic, and placental factors. JAMA
20. Jackson BJ, Kataaha P, Hom DL, et al. Beta2
-microglobulin. HIV-1 p24 antibody and acid-dissociated HIV-1 p24 antigen levels: predictive markers for vertical transmission of HIV-1 in pregnant Ugandan women. AIDS
21. Plaeger-Marshall S, Hultin P, Bertolli J, et al. Activation and differentiation antigens on T cells of healthy, at-risk, and HIV infected children. J Acquir Immune Defic Syndr
22. Akbar AN, Borthwick N, Salmon M, et al. The significance of low bcl-2 expression by CD45RO T cells in normal individuals and patients with acute viral infections: the role of apoptosis in T cell memory. J Exp Med