|
|
| |||||||||||||||||||||||||||||||||
Background and Methods A placebo-controlled trial has shown that treatment with zidovudine reduces the rate at which human immunodeficiency virus type 1 (HIV-1) is transmitted from mother to infant. We present data from that trial showing the number of infected infants at 18 months of age and the relation between the maternal viral load, the risk of HIV-1 transmission, and the efficacy of zidovudine treatment. Viral cultures were obtained, and HIV-1 RNA was measured by two assays in samples of maternal blood obtained at study entry and at delivery.
Results In 402 mother–infant pairs, the rate of transmission of HIV-1 was 7.6 percent (95 percent confidence interval, 4.3 to 12.3 percent) with zidovudine treatment and 22.6 percent (95 percent confidence interval, 17.0 to 29.0 percent) with placebo (P<0.001). In the placebo group, a large viral burden at entry or delivery or a positive culture was associated with an increased risk of transmission (the transmission rate was greater than 40 percent in the highest quartile of the RNA level). In both groups, transmission occurred at a wide range of maternal plasma HIV-1 RNA levels. Zidovudine reduced plasma RNA levels somewhat (median reduction, 0.24 log). Zidovudine was effective regardless of the HIV-1 RNA level or the CD4+ count at entry. In the zidovudine group, however, after we adjusted for the base-line HIV-1 RNA level and CD4+ count, the reduction in viral RNA from base line to delivery was not significantly associated with the risk of transmission of HIV-1.
Conclusions A high maternal plasma concentration of virus is a risk factor for the transmission of HIV-1 from an untreated mother to her infant. The reduction in such transmission after zidovudine treatment is only partly explained by the reduction in plasma levels of viral RNA. To prevent HIV-1 transmission, initiating maternal treatment with zidovudine is recommended regardless of the plasma level of HIV-1 RNA or the CD4+ count.
Methods
Study Subjects
ACTG 076 was a randomized, double-blind, placebo-controlled study that enrolled HIV-1–infected women between 14 and 34 weeks of pregnancy who had CD4+ T-lymphocyte counts exceeding 200 cells per cubic millimeter and no indications for antiretroviral therapy.1 The women were randomly assigned to receive either zidovudine or placebo. In the zidovudine group, maternal therapy both ante partum and intra partum was combined with six weeks of therapy for the newborn.1
The present study included all mothers eligible for the original study1 who had delivered live infants by January 4, 1994. Mothers who delivered infants thereafter were excluded from this study because their infants could not have completed six weeks of blinded treatment by the time the study was unblinded.
Laboratory Methods
Blood samples were collected from the mothers before they entered ACTG 076 and at the time of delivery in tubes containing heparin (72 percent of samples), acid citrate dextrose (25 percent), or EDTA (3 percent). The specimens were transported at ambient temperature and processed on the day of collection. Peripheral-blood mononuclear cells were isolated by Ficoll–Hypaque density-gradient separation, and qualitative cultures of mixed lymphocytes were performed by previously published methods.3
The cultures were assayed for p24 antigen at 7, 14, 21, and 28 days by a commercially available enzyme-linked immunosorbent assay (Abbott, Chicago). A quality-assurance program evaluated the performance of each participating laboratory.3,4 A culture was scored as positive if there was more than 30 pg of HIV-1 p24 antigen per milliliter of supernatant and if that level doubled within 28 days.3 Some laboratories performed quantitative microcultures of peripheral-blood mononuclear cells (21 percent of all cultures) using established procedures.5
Aliquots of plasma collected in tubes containing heparin were stored at -70°C. All the specimens were shipped on dry ice, then thawed and divided into aliquots for testing by each of two methods: a second-generation quantitative branched-chain DNA (bDNA) signal-amplification assay (Chiron, Emeryville, Calif.)6 and a quantitative reverse-transcription–polymerase-chain-reaction (PCR) amplification assay (Amplicor HIV Monitor Test, Roche Molecular Systems, Branchburg, N.J.).7 In the reverse-transcription–PCR assay, the plasma specimens collected in heparin required pretreatment with heparin lyase I. The bDNA and reverse-transcription–PCR assays were performed by the manufacturers of the assay kits; when the volume of a specimen was small, it was tested only by the latter assay.
In each assay, the HIV-1 RNA copy number was assessed by using the manufacturer's reference standards. The lower limit of sensitivity of the bDNA assay varied with the volume of the specimen; the limit was 500, 1000, and 10,000 copies of RNA per milliliter for volumes of 1, 0.5, and 0.05 ml, respectively. The lower limit of sensitivity of the reverse-transcription–PCR assay was approximately 400 copies per milliliter, but the limit varied among samples, since the test result for each specimen was based on the reverse-transcription–PCR amplification of an internal RNA standard that was added to each specimen. The assays differed in the range of the RNA concentration over which the detection signal was linear.6,7 Recent data suggest that of the two, the bDNA assay has more variability in the low range (below 3000 copies per milliliter).8 The upper limit of this linear range was 750,000 RNA copies per milliliter for the reverse-transcription–PCR assay and 1.6 million RNA copies per milliliter for the bDNA assay.6,7
Statistical Analysis
The updated comparison of treatment groups with respect to the efficacy of zidovudine treatment was based on the difference between groups in the percentage of eligible live-born infants who had HIV-1 infection at 18 months of age, expressed as a simple ratio with a binomial 95 percent confidence interval. Twins were counted as a single infant; discordant twins (one of whom was infected but the other not) were counted as a single infected infant. Infants were defined as infected if assays for HIV-1 were positive on two separate occasions or if a condition considered to define the presence of the acquired immunodeficiency syndrome (AIDS) was noted.9 An uninfected infant was defined as one who did not meet this definition and who met at least one of the following two criteria: (1) two or more negative serologic tests between the ages of 6 and 18 months or a single negative serologic test after the age of 18 months,9 or (2) two negative HIV cultures after the age of 4 weeks, one of which was performed at the age of 24 weeks or thereafter. Infants with a single positive culture but for whom there were not enough data for classification as uninfected were classified as infected; all other infants were considered to have indeterminate data and were excluded from all the analyses.
Specimens of maternal blood collected more than four days after the start of the study treatment or more than four days after delivery were excluded from study. RNA measurements were analyzed on a logarithmic scale (base 10) to reduce the tendency of the data to become more variable with increasing levels of RNA. We excluded bDNA measurements if the coefficient of variation between duplicate measurements exceeded 30 percent. Specimens in which HIV-1 RNA was undetectable were assigned values equal to half the reported detection limit of the assays of those specimens.
HIV-1 RNA levels were compared between groups by the two-sample Wilcoxon test. Nonparametric confidence intervals with at least 95 percent confidence were calculated around the median values.10 Changes in RNA levels from study entry to delivery were calculated as log(delivery RNA) - log(entry RNA). Tests for trend were used to compare transmission rates according to the quartile of the RNA measurements in each study group.11 Associations between prognostic factors were assessed with Spearman rank correlations. Logistic-regression analyses were used to assess whether virologic and immunologic measures predicted the risk of HIV-1 transmission and to estimate the proportion of the treatment effect that the measures explained.12
Results
Four hundred twenty-five mothers (211 in the zidovudine group and 214 in the placebo group) who were randomized by December 20, 1993, gave birth to 433 infants (417 singleton infants and 16 twins) by January 4, 1994. Among these 425 mother–infant pairs, there were 402 (198 in the zidovudine group and 204 in the placebo group) for which there were sufficient data for the infant's infection status to be classified.
Fifteen infants in the zidovudine group and 46 infants in the placebo group were infected with HIV-1. An estimated 7.6 percent (95 percent confidence interval, 4.3 to 12.3 percent) of the infants in the zidovudine group were infected, as compared with 22.6 percent (95 percent confidence interval, 17.0 to 29.0 percent) of those in the placebo group (P<0.001).
The characteristics of the study cohort have been reported previously.1 Not all the mothers had blood specimens obtained at entry and at delivery for virologic testing. Of the 402 mother–infant pairs, 333 (163 in the zidovudine group and 170 in the placebo group) had at least one plasma HIV-1 RNA measurement obtained at entry by either assay that could be evaluated. Of these mother–infant pairs, 286 (139 in the zidovudine group and 147 in the placebo group) had paired plasma RNA measurements obtained at entry and at delivery that could be evaluated. There were no significant differences between either of these subgroups and the entire cohort with respect to the characteristics of the mother, the delivery, or the newborn, nor were there differences between the two subgroups of mothers and their counterparts who did not have measurements that could be evaluated.
Predictive Variables at Study Entry
The zidovudine group was compared with the placebo group with respect to plasma HIV-1 RNA levels before the study treatment began (Figure 1). The median RNA levels at entry were similar. In both groups there was transmission from mothers with entry RNA levels spanning virtually the entire range of values.
|
RNA levels at entry were compared between mothers who transmitted HIV-1 to their infants and those who did not (Table 1). By the reverse-transcription–PCR assay, in both study groups the mothers who transmitted the virus were found to have higher median RNA levels at entry. By the bDNA assay, those in the placebo group who transmitted the virus were found to have higher median RNA levels than those who did not, but in the zidovudine group no significant difference was detected.
|
|
|
Immunologic data obtained at entry into the study were available for 400 mothers. Transmission rates in the placebo group increased as CD4+ cell counts decreased (Figure 3); no significant trend was detected in the zidovudine group. In each subgroup determined on the basis of the CD4+ count, the transmission rate was lower in the zidovudine group than in the placebo group. The presence of a lower proportion of CD4+ cells and a higher proportion of CD8+ cells was associated with a higher risk of transmission in the placebo group, but not in the zidovudine group (Table 2). CD4+ counts at entry were negatively correlated with plasma RNA levels at entry (reverse-transcription–PCR assay, r = -0.38; bDNA assay, r = -0.31; P<0.001 for both).
|
Viral Load and Effect of Zidovudine Treatment
In assessing whether RNA levels explained the effect of zidovudine treatment, we focused on the larger set of data obtained by the reverse-transcription–PCR assay. Among mothers for whom there were plasma RNA measurements obtained at both study entry and delivery, those in the zidovudine group had median RNA values 1.7 times (0.24 log) lower at delivery than the values in the placebo group (median reduction, 0.28 vs. 0.04 log copies per milliliter; P<0.001). There was a larger median reduction (0.34 log) among zidovudine-treated mothers whose RNA levels at entry exceeded 1000 copies per milliliter than among their counterparts in the placebo group (0.8 log), but the net reduction was similar (0.26 log).
Most specimens obtained at delivery were collected within one (92 percent) or two (97 percent) days after delivery; the median changes in RNA levels were similar when the analysis was restricted to these narrower intervals. The median time between the sample collections (at entry and at delivery) was 13 weeks; there was no relation in either group between the change in RNA and the time between the collections. The proportion of positive cultures at delivery as compared with entry did not differ between treatment groups; in most mothers (75.5 percent), the results of the cultures were concordant, and those who had discordant cultures were divided equally in both groups between those with negative cultures at entry but positive cultures at delivery and those with the reverse.
We used logistic-regression models to assess whether measures of viral load at delivery or changes in RNA levels from the base-line values predicted the risk of transmission (Table 2), explained the effect of zidovudine treatment, or both. In the zidovudine group, a higher absolute level of RNA at delivery as measured by the reverse-transcription–PCR assay was the only significant predictor of an increased risk of transmission (odds ratio, 1.95; P<0.05). In the placebo group, a higher absolute level of RNA at delivery as measured by either assay was significantly associated with a higher risk of transmission (odds ratio with reverse-transcription–PCR assay, 2.21; with bDNA assay, 1.80; P<0.01 for both), as was an increase in RNA from entry to delivery as measured by the reverse-transcription–PCR assay and adjusted for the base-line RNA value and the percentage of CD4+ cells (odds ratio, 0.52; P<0.05). A negative culture at delivery was associated with a lower risk of transmission. A significant effect of zidovudine treatment remained after adjustment for each predictive variable. The estimated proportion of the treatment effect that was explained by each measure of viral load at delivery was 10.8 percent (95 percent confidence interval, -0.5 to 31.5 percent) for RNA as measured by the reverse-transcription–PCR assay, 6.7 percent (no stable 95 percent confidence interval could be determined) for RNA as measured by the bDNA assay, and 6.3 percent (95 percent confidence interval, 0.5 to 18.2 percent) for a positive HIV-1 culture of peripheral-blood mononuclear cells. The change in the RNA level (after adjustment for the base-line RNA level and proportion of CD4+ cells) accounted for approximately 16.6 percent (95 percent confidence interval, 3.9 to 41.2 percent) of the effect of zidovudine treatment when the reverse-transcription–PCR assay was used and 9.2 percent (95 percent confidence interval, -3.3 to 30.4 percent) of the effect when the bDNA assay was used.
Discussion
The initial report of the controlled trial of the efficacy of zidovudine to prevent maternal–infant transmission of HIV-1 included 363 mother–infant pairs for whom at least one viral culture from the infant was available.1 We now present analyses based on clinical, virologic, and serologic data obtained after an 18-month follow-up in 402 mother–infant pairs. The revised estimates of the proportions of infected infants in the zidovudine and placebo groups strongly support our earlier report1 and are consistent with published reports of uncontrolled studies.13,14,15
In the placebo group, an increased maternal load of HIV-1 at entry or delivery, as measured by either the bDNA assay or the reverse-transcription–PCR assay in plasma or by culture of peripheral-blood mononuclear cells, was associated with an increased risk of maternal–infant transmission of HIV-1. In the zidovudine group, only an increased viral load at entry or delivery as measured by the reverse-transcription–PCR assay was significantly associated with the risk of transmission. This finding is in accordance with data obtained in small U.S. cohorts.16,17,18 In contrast to these reports,16,17 however, we found that transmission occurred at lower absolute levels of plasma RNA and across the entire range of measurable RNA levels, as well as from some mothers with undetectable plasma viremia. Although the rates of transmission decreased as RNA levels decreased, our data do not suggest that there is an absolute plasma RNA level below which transmission does not occur.
The effect of zidovudine treatment was seen with all plasma HIV-1 RNA levels in samples obtained at study entry. Overall, zidovudine treatment was associated with only a small reduction in circulating levels of plasma RNA. Neither the change in the plasma RNA level from entry to delivery nor a critical level of plasma RNA at delivery accounted for the substantially reduced transmission rate in the zidovudine group. Only a small part of the treatment effect could be explained by the observed RNA measurements, which is further evidence that the protective effect of zidovudine results at least in part from a mechanism other than the reduction of the maternal plasma viral burden.
Explanations need to be considered for the apparent lack of association between the observed RNA levels and the effect of zidovudine treatment. The possibility of a prophylactic effect both in utero and intra partum, during and after exposure, is raised. A recent case–control study of health care workers with percutaneous exposures suggests that zidovudine use may be protective after exposure to HIV-1.19 Moreover, the plasma RNA level is only one measure of viral activity. The timing of transmission (either in utero or intra partum), maternal infection with a zidovudine-resistant strain of virus, or other maternal factors ante partum or intra partum may all have influenced the success of zidovudine therapy and the risk of transmission.
We used both the bDNA and the reverse-transcription–PCR assays to assess the influence of the viral load on the risk of vertical transmission. Confirming our results by two independent quantitative methods was considered important because no other placebo-controlled trials of antiretroviral agents in the United States are likely, given that zidovudine has been accepted for use in preventing perinatal transmission.2 Since the two assays had not been developed when the ACTG 076 trial began, specimens were not always collected by methods optimal for the performance of these assays.
In general, the results obtained by both assays were in agreement. The higher absolute values obtained from the reverse-transcription–PCR assay, as well as the stronger observed association with the risk of transmission, underscore the problems with interpreting these comparisons. First, since the reverse-transcription–PCR assay requires a smaller volume of plasma than the bDNA assay, specimens whose volume was limited were tested only by the reverse-transcription–PCR assay. The statistical power of the study to allow conclusions based on the results of the bDNA assay was therefore more limited. Second, there are no standard references for use in calibrating the results of the two assays. The copy number obtained with the bDNA assay tended to be lower than that obtained by the reverse-transcription–PCR assay. Among specimens found to have detectable RNA by both techniques, the RNA copy numbers measured differed by a median factor of two. Third, the assays have different performance characteristics. many specimens tested had relatively low RNA copy numbers. The nonlinear performance of the bDNA assay in specimens with low copy numbers may in part explain the weaker association observed with the risk of transmission. Finally, the performance of an assay can be affected by the handling and processing of specimens. The stored specimens were collected in heparin, which had to be removed with heparin lyase I before the reverse-transcription–PCR assay could be performed. The choice of anticoagulant has been shown to interfere with quantification by both the reverse-transcription–PCR and the bDNA assays; plasma samples collected in heparin may yield values up to 38 percent lower than those of specimens collected in EDTA.6 The relatively low median RNA levels reported may partly be due to delays in separating plasma from whole blood, which result in increasing degradation of RNA with time.20 Despite the variations that may have occurred in the collection and handling of specimens, there was no evidence that the median RNA copy number differed significantly according to study site or over time.
Statistical considerations limit the conclusions that can be drawn from the study. The small number of women in the zidovudine group who transmitted HIV-1 to their infants limited the statistical power of the study to detect an association between changes in the RNA level and the risk of transmission. The estimated proportions of the treatment effect that are explained by this measure may have been attenuated by errors of measurement in reporting the RNA results.21
In other studies, advanced maternal immunosuppression, as evidenced by a low CD4+ cell count,22,23 has been associated with an increased risk of perinatal HIV-1 transmission. In the placebo group, lower base-line CD4+ counts, lower percentages of CD4+ cells, and higher percentages of CD8+ cells were all associated with an increased risk of transmission. In the zidovudine group, no immunologic predictors were significantly associated with this risk. The effect of zidovudine treatment was seen in all groups regardless of the CD4+ count at study entry.
In our study there was HIV-1 transmission with a wide range of plasma RNA levels and CD4+ counts at entry, the treatment effect was seen in all RNA subgroups, and the reduction in plasma RNA from entry to delivery did not account for the success of zidovudine therapy. Therefore, our data strongly support the current practice of offering zidovudine to all HIV-1–infected pregnant women and their newborns, regardless of the maternal plasma viral burden and CD4+ count. Even when alternative antiretroviral drugs are indicated to treat the mother's disease, the regimen of zidovudine that has been proved to prevent perinatal transmission of HIV-1 must be considered.
Supported in part by the Pediatric AIDS Clinical Trials Group under a contract (N01 AI95030) with the National Institute of Allergy and Infectious Diseases; by the General Clinical Research Center units funded by the National Center for Research Resources; by the Pediatric–Perinatal HIV Clinical Trials Network, National Institute of Child Health and Human Development; by Glaxo Wellcome; by the Chiron Corporation; by Roche Molecular Systems; and by the Agence Nationale de Recherche sur le SIDA (France).
We are indebted to Bethann Cunningham (clinical data manager); Elaine Gebhardt (virology data manager); Shirley Traite (statistical programmer); Paula Britto (biostatistician); Mary Culnane (program officer); Robert Kuzma, Helen Payne, and Kristina Hasselbring (scientists); and Siuhing Sheila Leung (statistician) for their contributions.
Source Information
From the Department of Obstetrics, Gynecology, and Reproductive Science, Mount Sinai School of Medicine, New York (R.S.S.); the Statistical and Data Analysis Center, Pediatric AIDS Clinical Trials Group, Harvard School of Public Health, Boston (D.E.S.); University of Washington School of Medicine, Seattle (R.W.C.); Chiron Corporation, Emeryville, Calif. (J.A.T.); Roche Molecular Systems, Branchburg, N.J. (S.A.H.); Department of Pediatrics, University of Medicine and Dentistry of New Jersey–New Jersey Medical School, Newark (G.D.M.); University of Miami School of Medicine, Miami (M.J.O.); Tulane University School of Medicine, New Orleans (R.B.V.); San Juan City Hospital, San Juan, P.R. (E.J.); Agence Nationale de Recherche sur le SIDA–Hôpital Necker, Paris (C.R.); St. Jude Children's Research Hospital, Memphis, Tenn. (P.M.F.); and the University of Massachusetts, Worcester (J.L.S.). Other authors were Stephen A. Spector, M.D. (University of California at San Diego Medical Center, San Diego), Clemente Diaz, M.D. (University of Puerto Rico, San Juan), James Rooney, M.D. (Glaxo Wellcome, Research Triangle Park, N.C.), James Balsley, M.D., Ph.D. (National Institute of Allergy and Infectious Diseases, Bethesda, Md.), Richard D. Gelber, Ph.D. (Harvard School of Public Health, Boston), and Edward M. Connor, M.D. (University of Medicine and Dentistry of New Jersey–New Jersey Medical School, Newark).
Address reprint requests to Dr. Sperling at Mount Sinai Medical Center, Box 1173, 1 Gustave Levy Pl., New York, NY 10029.
References
The following institutions and persons participated in the Pediatric AIDS Clinical Trials Group Protocol 076 Study Group. University of Miami: I. Madera, W.A. Scott, and F. Breckenridge; Tulane–Louisiana State University: S.R.S. Rangan, M. Perrin, and J.G. Price; San Juan City Hospital: E. O'Neill, E. Abreu, and L. Perez; France: J. Puel, H. Fleury, and J. Cottalorda; St. Jude Children's Research Hospital: W.T. Hughes, J. Shenep, and R.V. Srinivas; University of Massachusetts and Baystate Medical Center: K. Luzuriaga, B. Stechenberg, and J.E. Robinson; University of California, San Diego: M. Besser, M. Caffery, and P. Franklin; University of Puerto Rico: G. Hillyer, C. Zorrilla, and C. Rivera; Mount Sinai Medical Center: D. Hodes, L. Rhone, and E. Chusid; Children's Hospital of Michigan: T. Jones, E.C. Moore, and D.D. Harrison; University of North Carolina at Chapel Hill: W. Lim, V. Katz, V. Dudek, and S. Fiscus; Children's Hospital at Albany Medical Center: R. Samelson, N. Wade, M. Lepow, and M. Horgan; University of Washington: S. Burchett, K. Mohan, and E. Johnson; Texas Children's Hospital: H. Hammill, R. Jackson, B. Hollinger, and W. Shearer; University of Illinois: M. Vajaranant, K. Rich, D. Koukul, and K. Hayani; Johns Hopkins University Hospital: J. Lambert, J. Anderson, P. Vink, and L. Alger; Columbia University: A. Gershon, J. Pitt, H. Fox, and A. Higgins; New York University–Bellevue Hospital: W. Borkowsky, P.T. Tao, T. Moore, and M.L. Liu; Los Angeles County–University of Southern California Medical Center: S. Rasheed, A. Kovacs, A. Stek, and M. Khoury; University of Colorado–Denver Children's Hospital: M.J. Levin, E.J. McFarland, C. Salbenblatt, and J. McGregor; Howard University: D. Smith, S. Rana, T. Dawkins, and M. Akinsiku; Albert Einstein College of Medicine: A. Rubinstein, L. Bernstein, J. Youcha, and Y. Mizrachi; University of California, San Francisco: D.W. Wara, M. Benson, T. Elbeik, and M. Shannon; University of California, Los Angeles: Y.J. Bryson, P. Boyer, M. Dillon, and D. Wafer; Boston Children's Hospital: K. McIntosh, R. Tuomala, R. Bishop, and R. Galvin; Northwestern University–Children's Hospital of Chicago: R. Yogev, E. Chadwick, P. Garcia, and D. Stanislawski; Duke University: C.M. Wilfert, R.E. McKinney, M. Donnelly, and E. Livingston; University of Rochester: J. Lambert, L. Frenkel, B. Murante, and C. Emsthausen; Ramon Ruiz Arnau University Hospital: D.E. Garcia Trias, R. Aguayo, E. Reyes, and L. Diaz; State University of New York (SUNY), Brooklyn: S. Fikrig, E. Handelsman, and P. Driscoll; SUNY, Syracuse: C.K. Cunningham, L.B. Weiner, K.A. Contello, and K.P. Kirkwood; University of Alabama: M. Crain, L. Jones, J. Kappes, and J.A. Conway; University of Medicine and Dentistry of New Jersey: A.D. Bardeguez and J.M. Oleske; University of Cincinnati: K. Skahan, R. Baker, N. McOwen, and S. Woods; Boston City Hospital: S.I. Pelton, A.M. Regan, F.P. Donegan, and M. Mirochnick; Medical University of South Carolina: G. Johnson, R.B. Turner, R.B. Newman, and M.G. Connelly; SUNY, Stony Brook: S. Nachman, D. Baker, A. Vomero, and D. Hickey; Children's Hospital of Philadelphia: H. Friedman, R. Hodinka, E. Anday, and K. Mooney; New York Medical College: A. Gupta, N. Kirschenbaum, E. Ahern, and M. Galvin; Bronx Lebanon Hospital Center: A. Wiznia, G. Lambert, W. Pabon, and P. Cihak; Rhode Island Hospital–Women and Infants Hospital: P.S. Smith, P. Sweeney, C. Kneut, and B.L. Means-Tavares; Cornell Medical Center: J.S. Cervia, K.D. LaGuardia, G. Hinds, and K. McGuinness; University of Texas Southwestern Medical Center: G. Wendell, J. Squires, J. Dax, and A. Finegan; National Institute of Child Health and Human Development: J. Moye; ACTG Operations Office, Social and Scientific Systems, Inc.: E. Hawkins; Pediatric ACTG Data Management Center: Frontier Science and Technology Research Foundation, Inc., Amherst, N.Y.; Statistical and Data Analysis Center: Harvard School of Public Health, Boston.
| |||||||||||||||||||||||||||||||||
This article has been cited by other articles:
HOME | SUBSCRIBE | SEARCH | CURRENT ISSUE | PAST ISSUES | COLLECTIONS | PRIVACY | TERMS OF USE | HELP | beta.nejm.org Comments and questions? Please contact us. The New England Journal of Medicine is owned, published, and copyrighted © 2009 Massachusetts Medical Society. All rights reserved. |
