Survey
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project
American Journal of Epidemiology Copyright O 1999 by The Johns Hopkins University School of Hygiene and Public Health All rights reserved Vol. 150, No. 6 Printed In USA. Risk Factors for HIV-1 Shedding in Semen Carl E. Speck, u Robert W. Coombs,34 Laura A. Koutsky,1 Judith Zeh,5 Susan O. Ross,6 Thomas M. Hooton,4 Ann C. Collier,4 Lawrence Corey,34 Anne Cent,3 Joan Dragavon,3 Willa Lee,3 Eric J. Johnson,3 Reigran R. Sampoleo,3 and John N. Krieger* Semen is the body fluid most commonly associated with sexual transmission of human immunodeficiency virus type-1 (HIV-1). Because the male genitourinary tract is distinct immunologically from blood, compartmentdependent factors may determine HIV-1 shedding in semen. To identify these factors, the authors obtained 411 semen and blood specimens from 149 men seen up to three times. Seminal plasma was assayed for HIV-1 RNA and semen was cocultured for HIV-1 and cytomegalovirus (CMV), which may up-regulate HIV-1 replication. The best multivariate model for predicting a positive semen HIV-1 coculture included two local urogenital factors, increased seminal polymorphonuclear cell count (odds ratio (OR) = 12.6 for each log10 increase/mL, 95% confidence interval (Cl) 12.2, 134.5) and a positive CMV coculture (OR = 3.0, 95% Cl 1.2, 7.7). The best multivariate model for predicting semen HIV-1 RNA included two systemic host factors, CD4+ cell counts <200/|xliter (OR = 3.0, 95 percent Cl 1.3, 6.9) and nucleoside antiretroviral therapy (monotherapy: OR = 0.5, 95% Cl 0.3, 1.0; combination therapy: OR = 0.4, 95% Cl 0.2, 0.9), and a positive CMV coculture (OR = 1.7, 95% Cl 1.0, 3.0). Thus, both systemic and local genitourinary tract factors influence the risk of semen HIV-1 shedding. These findings suggest that measures of systemic virus burden alone may not predict semen infectivity reliably. Am J Epidemiol 1999; 150:622-31. acquired immunodeficiency syndrome; coculture; cytomegalovirus; HIV-1; polymerase chain reaction; RNA, viral; semen Compelling epidemiologic data suggest that direct contact with semen is the major route of human immunodeficiency virus type-1 (HTV-1) transmission from seropositive men to their sex partners (1-16). However, little is known about the factors that influ- ence HTV-1 shedding in semen or the biology of HTV1 within the male genital tract. Although HTV-1 nucleic acid (DNA and RNA) is detected in most semen specimens from infected men, infectious virus is cocultured from fewer than half of specimens (17-24). In addition, factors that influence the infectivity of semen may differ from those that influence the infectivity of blood, lymphatic tissue, or the central nervous system because the male reproductive tract is a distinct immunologic compartment (24-27). Most of our work has used semen HTV-1 coculture as the outcome measure. While somewhat insensitive, a positive semen HTV-1 coculture indicates the presence of infectious virus. The results of our earlier studies suggest that shedding of cultivable HIV-1 in semen is intermittent (20, 28); is not statistically associated with clinical stage of HIV-1 disease or CD4+ T-lymphocyte count (CD4+ count) (18, 29); is not reduced by zidovudine monotherapy (18, 20); but is associated with low peripheral blood CD8+ T-lymphocyte counts (CD8+ counts) (20). Taken together, these findings suggest that systemic host factors alone are poor predictors of semen infectivity as assessed by HIV-1 coculture. Reports from other researchers who have used semen HTV-1 coculture as the outcome have varied. In 1992, Anderson et al. (30) reported that seminal HTV-1 cocul- Received for publication May 2, 1997, and accepted for publication February 1, 1999. Abbreviations: AIDS, acquired immune deficiency syndrome; Cl, confidence interval; CMV, cytomegalovirus; GEE, generalized estimating equation; HIV-1, human immunodeficiency virus type-1; OR, odds ratio; PBMC, peripheral blood mononuclear cell; RT-PCR, reverse transcription polymerase chain reaction; PMN, polymorphonuclear; RNA, ribonuclefc acid; SD, standard deviation. 1 Department of Epidemiology, University of Washington, Seattle, WA. 2 Department of Research and Evaluation, Southern California Permanente Medical Group, Pasadena, CA. 3 Department of Laboratory Medicine, University of Washington, Seattle, WA. 4 Department of Medicine, University of Washington, Seattle, WA. 5 Department of Statistics, University of Washington, Seattle, WA. 6 Department of Urology, University of Washington, Seattle, WA. Reprint requests to Dr. John N. Krieger, Department of Urology, University of Washington, Box 356510, Seattle, WA 98195. Written informed consent was obtained from subjects in accordance with guidelines of the University of Washington and the US Department of Health and Human Services. Presented in part in the HIV in Semen: Research Issues workshop, NIH, Bethesda, MD, February 1-2,1996 (Abstract p. 23), and at the Eleventh International Conference on AIDS, Vancouver, British Columbia, July 7-12, 1996 (Abstract We.B.3383). 622 Risk Factors for HIV-1 Shedding in Semen ture positivity was associated with low peripheral blood CD4+ counts and with seminal inflammation. These authors also reported that zidovudine monotherapy significantly reduced HTV-1 shedding in semen. However, in 1993, Hamed et al. (19) reported no association between detection of HTV-1 RNA in semen and clinical stage of HIV-1 infection, CD4+ count, or long-term zidovudine monotherapy. In 1994, Vernazza et al. (23) reported higher semen HTV-1 coculture positivity rates among symptomatic men and among men with CD4+ cell counts <100/nliter of peripheral blood. However, positive peripheral blood plasma HTV-1 coculture did not predict positive semen HTV-1 coculture. Cytomegalovirus (CMV), a sexually transmitted virus, is shed in semen. In 1994, Detels et al. (31) reported that after they controlled for peripheral blood CD4+ cell count, HTV-1 seropositive men who persistently shed CMV in their semen were more likely to develop acquired immune deficiency syndrome (AIDS) than men whose semen was either intermittently positive or persistently negative. In a subsequent analysis, the same group of researchers (32) reported that men who shed multiple strains of CMV in their semen were more likely to progress to AIDS than men who either shed the same strain of CMV or who did not shed CMV. Krieger et al. (33) reported that CMV shedding was associated with HTV-1 shedding in semen, but that this association was attenuated when they controlled for other factors. We used a prospective, repeated measures design, to reassess local urogenital and systemic host factors as determinants of HTV-1 shedding in semen. The local urogenital factors that we examined included seminal leukocyte levels and concurrent seminal CMV shedding, while the systemic host factors included lymphocyte subset counts, HTV-1 level in peripheral blood, clinical status, and antiretroviral therapy use. Because there is no perfect measure of semen HTV-1 infectivity, we used two outcome measures, coculture of semen and detection of HTV-1 RNA by reverse transcription polymerase chain reaction (RT-PCR) amplification of seminal plasma. Coculture techniques are much less sensitive than RT-PCR amplification, but a positive coculture indicates the presence of infectious virus. However, in defining the "infectious transmissible unit," the relation between the detection of semen viral RNA and semen infectivity can only be inferred. MATERIALS AND METHODS Subjects Subjects were recruited using an advertisement in a local gay newspaper (the Seattle Gay News), and through the Harborview Medical Center (Madison) Am J Epidemiol Vol. 150, No. 6, 1999 623 HTV/AIDS Clinic, the University of Washington AIDS Clinical Trials Unit, Center for AIDS Research, and Virology Research Clinic. At the clinic sites, potential subjects were recruited by a study nurse who briefly described the study. Interested men were then introduced to the study manager, who provided details and obtained informed consent. Subjects recruited through the newspaper advertisement called a confidential telephone number to speak with the study manager. Study eligibility. HIV-1 seropositive men, aged >18 years, who had a local address and telephone number and were physically capable of providing three semen specimens at 4-week intervals were eligible for the study. There were no exclusions based on disease stage, antiretroviral therapy status, or CD4+ cell count. Specimen collection. At each visit, blood was drawn, a clinical questionnaire was administered, and a semen specimen was scheduled for the following week. Semen specimens were collected by masturbation into sterile containers after 2 days' abstinence from ejaculation and immediately sent to the University of Washington Male Fertility Laboratory. Laboratory methods Semen processing, cell-staining, and analysis. Semen specimens were generally processed within 2 hours of collection (mean, 72 minutes; standard deviation (SD), 40; range, 20-210). Each specimen was diluted 1:1 with culture medium and the seminal plasma and seminal cell-pellet were separated by centrifugation at 2,940 g for 2-4 minutes (18). An aliquot of seminal plasma was stored at -70°C for HTV-1 RNA analysis. If sufficient specimen was available, each fraction was cocultured separately for HIV-1 and for CMV. For each semen analysis, a thin smear was stained using a modified Bryan-Leishman stain and examined for sperm and leukocyte morphology according to a World Health Organization protocol (34, 35). This method allows discrimination of leukocytes from immature sperm forms. Leukocyte types were enumerated per 100 sperm and the concentrations (millions/mL) of total leukocytes and each leukocyte subset (polymorphonuclear (PMN) cells, lymphocytes, and monocytes) were calculated by multiplying the leukocytes per 100 sperm times the sperm concentration and dividing by 100. We found excellent agreement between this staining and evaluation of the same samples by monoclonal antibody IILe-1 (panleukocyte)-stained slides (36). Weekly quality control was provided by cryopreserved aliquots of semen or by semen samples from normal donors. Detection of cell-free HIV-1 genomic RNA in seminal plasma. A reverse transcriptase-based polymerase chain reaction (RT-PCR) amplification method was 624 Speck et al. used to assay HTV-1 ribonucleic acid (RNA) in the cell-free seminal plasma (Amplicor HTV Monitor, Roche Molecular Systems, Branchburg, New Jersey). Prior to amplification, seminal plasma was pretreated using a silica gel extraction method to remove inhibition of the internal assay standard (24, 37). The manufacturer's internal quantitative standard was used for each assay, and negative controls were routinely run to ensure that results fell within the range of acceptance for truly negative assays. Results were reported as positive or negative. The lower limit of detection for this assay was 400 copies of HTV-1 RNA/mL of diluted seminal plasma (38, 39). We did not use bDNA to assay semen specimens due to insufficient volume. However, we did compare the RT-PCR to bDNA assays using HTV-1 spiked pooled HTV-1 seronegative seminal plasma provided by the National Institute of Allergy and Infectious Disease (NIAID) Division of AEDS-sponsored Virology Quality Assurance Laboratory (VQAL). The relation between the two assays is described by the following equation: RT-PCR log10 copy number/mL = 0.987 (bDNAlog10 copy number/mL) + 0.716, Pearson correlation coefficient, r2 = 0.986 (24). Thus, the semen HTV-1 RNA levels are approximately fivefold greater when measured by the RT-PCR assay than by the bDNA assay. Viral cultures. HTV-1 was cocultured from blood and semen using standard methods (24,40). To coculture for CMV, seminal cell pellets and plasma were diluted with RPMI 1640 medium containing heatinactivated (30 minutes, 56°C) fetal calf serum (16 percent), penicillin G (100 units/mL), streptomycin (100 (Xg/liter), and additional L-glutamine (0.03 mg/mL), then stored at 4°C for <6 hours (41, 42). Human fetal tonsil fibroblast cell monolayers were inoculated with 0.2 mL of the diluted specimen and with 1:10 and 1:100 dilutions, then examined weekly for 5 weeks. CMV isolates were identified by their characteristic cytopathic effects, then confirmed by indirect immunoperoxidase staining and by failure to grow in HeLa-M cells (41^3). Quantitation of HIV-1 RNA in peripheral blood plasma. Using the Quantiplex branched chain DNA assay (Chiron Corporation, Emeryville, California), ACD-anti-coagulated blood plasma was assayed without pretreatment (44). The manufacturer's standards were run with each assay with results reported as HTV-1 RNA copies/mL of blood plasma. The lower limit of HTV-1 RNA detection was 10,000 copies/mL of blood plasma (44). The University of Washington Retro virology Laboratory participates in the VQAL proficiency testing program for HTV-1 RNA quantitation (39, 45). T-lymphocyte subset analyses. Counts of total lymphocytes and of lymphocyte subsets were done in the University of Washington's Hematopathology Laboratory using a standard, quality assured flow cytometric technique (46). Blood cells were fractionated by Ficoll-Hypaque density gradient centrifugation (40). The mononuclear cell-enriched layer was then incubated with OKT-3, OKT-4, and OKT-8 fluorescein-conjugated monoclonal antibodies. Reactions were read on an EPICS flow cytometer (Beckman Coulter, Inc., Fullerton, California) and the proportions of each subset were calculated. The absolute number of lymphocytes was determined on a simultaneous complete blood count and the absolute numbers of each T-cell subset were calculated. This procedure was run daily. Statistical analyses Continuous variables were transformed on a logarithmic base-10 scale. The SAS package (SAS Institute, Inc., Cary, North Carolina) was then used to perform generalized estimating equation (GEE) analyses to determine the best univariate and multivariate predictors for each outcome measure (47). The GEE approach was chosen because ordinary logistic regression techniques assume that all data points are independent, and thus fail to account for the intra-subject correlation inherent in longitudinal data sets such as that produced in this study. When intra-subject correlation is not accounted for, incorrect indications of statistically significant effects may result. Usually, GEE approaches attenuate this problem by calculating wider confidence intervals. Although risk estimates generated by ordinary and GEE logistic regression analyses are nearly identical, the GEE approach is almost always more conservative and confidence intervals are wider. In the univariate and multivariate models, we performed both "unrestricted" and "restricted" analyses. The restricted analyses included only observations for which results were available for HTV-1 and CMV coculture and seminal plasma HIV-1 RNA PCR assays. We conducted these restricted analyses to determine whether odds ratios differed between data sets with complete information for all variables and data sets with either some outcome or CMV coculture data missing or both missing. This facilitated assessment of potential bias caused by differential testing rates of semen specimens by HTV-1 coculture, CMV coculture, and HIV-1 RNA assays. In all, four analyses were run for each potential predictor variable for each of the two study outcomes (univariate unrestricted; univariate restricted; multivariate unrestricted; and multivariate restricted). When we constructed the multivariate models, we evaluated all factors associated with HTV-1 shedding in Am J Epidemiol Vol. 150, No. 6, 1999 Risk Factors for HIV-1 Shedding in Semen the univariate models, as well as factors that have been consistently identified in the literature to influence detection of HTV-1 in semen, such as antiretroviral therapy. For each model, we selected the factor most strongly associated with HTV-1 shedding in semen, and then placed other variables into the model one at a time. If the variable was statistically significant in the multivariate model, it was retained. In all multivariate modeling, we used categorical (as opposed to continuous) CD4+ cell count because the relation between CD4+ cell count and HIV-1 shedding in semen was non-linear. RESULTS Overview A total of 161 men were enrolled between April 4, 1994 and August 18, 1995. Of these 161 men, 12 were lost to follow-up prior to providing their first semen specimen. Of the remaining 149 men, 13 (8.7 percent) completed only one visit; 8 (5.4 percent) completed exactly two visits; and 128 (85.9 percent) completed all three visits. The 149 study participants provided 411 semen specimens overall. The mean CD4+ count for subjects who completed <3 visits was 277 (SD, 281), while that for subjects who completed all 3 visits was 375 (SD, 241) (t = -1.66, p = 0.10). The mean log10 copy number of HTV-1 RNA/mL of blood plasma for subjects who completed <3 visits was 4.68 (SD, 0.68); the mean for subjects who completed 3 visits was 4.38 (SD, 0.56) (t = 1.43, p = 0.15). At study entry, about 25 percent of the subjects had clinical AIDS (Centers for Disease Control Stage C); about 30 percent had CD4+ counts <200/(xliter of peripheral blood and just under 40 percent of subjects took one or more nucleoside antiretroviral medications. Average semen volume was 1.84 mL (range, 0.05-6.5 mL). These and other baseline subject characteristics are summarized in table 1. HIV-1 isolation from semen by coculture. HTV-1 was isolated from 34 (12.4 percent) of 274 specimens cocultured for HTV-1. Of the 34 culture positives, 28 (82.3 percent) were positive only in the cellular fraction of the specimen; 4 (11.8 percent) were positive in both fractions of the specimen, and 2 (5.9 percent) were positive in the plasma fraction only. Of 53 men cocultured for HTV-1 at all 3 visits, 8 (15.1 percent) were positive at visit 1, 9 (17.0 percent) had >1 positive coculture at visit 2, and 12 (22.6 percent) had >1 positive coculture after visit 3. CMV isolation from semen by coculture. Of 320 specimens evaluated for CMV, 126 (39.4 percent) were positive. Of these 126 positive specimens, 94 (74.6 percent) were positive in both the cellular and plasma fractions of the specimen. Of the 69 men Am J Epidemiol Vol. 150, No. 6, 1999 625 TABLE 1. Sociodemographlc and Immunologic characteristics of 149 participants In study of risk factors for HIV-1* shedding In semen, Seattle, Washington State, 1994-1995 Characteristic Age (years) 20-29 30-39 40-49 50-59 Race Caucaslon Hispanic African American Other Current employment status Full time Part time Not working Recruitment procedure Harborview Madison Clink; Seattle Gay News advertisement Other AIDS»-related programs Physician referral Another enrolled patient HIV-1 risk factors Male-male sex Injection drug use Health care/occupational exposure Female-male sex Blood transfusion Multiple risk factors Unknown Baseline CD4 count): £500 (Class 1) 200-499 (Class 2) <2OO(C4ass3) Baseline clinical stage of HIV-1 disease Asymptomatic (Class A) Mild constitutional symptoms (Class B) Clinical AIDS Baseline CD4 count and clinical stage of HIV-1 disease A-1 A-2 A-3 B-1 B-2 B-3 C-1 C-2 C-3 Baseline nucleoside antiretroviral therapy status Combination tnerapy§ Monotherapyfl No antiretroviral therapy No. %t 27 73 44 5 18.1 49.0 29.5 3.4 121 12 12 4 81.2 8.1 8.1 2.7 44 16 89 29.5 10.7 59.7 54 75 8 6 6 362 50.3 5.4 4.0 4.0 121 4 3 2 1 17 1 812 2.7 2.0 1.3 0.1 11.4 0.1 47 58 44 31.6 38.9 29.5 41 71 37 27.5 47.7 24.8 20 18 3 15 34 22 0 12 25 13.4 12.1 2.0 10.1 22.8 14.8 0.0 8.1 16.8 17 38 94 11.4 25.5 63.1 • HIV-1, human Immunodeficiency virus type-1; AIDS, acquired Immunodeficiency syndrome. t Percent was calculated based on the 149 men who completed at least one visit t Per mm 1 of peripheral blood. § Use of 22 licensed nucleoside analog drugs. H Use of exactly one licensed nucleoside analog drug. cocultured three times for CMV, 25 (36.2 percent) were positive at visit 1, 30 (43.5 percent) had been positive at least once at visit 2, and 33 (47.8 percent) had been positive at least once after visit 3. HIV-1 RNA detection in seminal plasma by RTPCR. Of 350 specimens evaluated for HTV-1 RNA, 210 (60.0 percent) were positive. Among 91 men assayed at all three time points, 56 (61.5 percent) were positive at visit 1, 65 (71.4 percent) were positive after 626 Speck et al. visit 2, and 71 (78.0 percent) were positive at least once after visit 3. HIV-1 and CMV isolation from semen by coculture. A total of 236 specimens were of sufficient volume to be cocultured for both HTV-1 and CMV. Of those 236 specimens, 18 (7.6 percent) were positive for both HTV-1 and CMV; 69 (29.2 percent) were CMV positive but HTV-1 negative; 15 (6.4 percent) were HTV-1 positive but CMV negative; and 134 (56.8 percent) were negative for both viruses. HIV-1 in semen by RNA RT-PCR and coculture. Of 274 specimens cocultured for HTV-1, 244 (89.1 percent) were also assayed for HTV-1 RNA. Twenty-five (89.3 percent) of 28 HTV-1 culture positive samples were positive for HTV-1 RNA. Of 216 HTV-1 culture negative samples, 121 (56.0 percent) were positive for HTV-1 RNA. HIV-1 in semen by RNA RT-PCR, HIV-1 coculture, and CMV coculture (table 2). Of 210 semen specimens with results for all three assays, 127 (60.5 percent) were HTV-1 RNA positive, 27 (12.9 percent) were HTV1 coculture positive, and 75 (35.7 percent) were CMV coculture positive. Eighteen (24.0 percent) of 75 CMV coculture positive specimens were HTV-1 negative by both coculture and RT-PCR and 16 (21.3 percent) were positive for HTV-1 by both coculture and RT-PCR. In contrast, only 8 (5.9 percent) of the 135 CMV coculture negative specimens were positive for HTV-1 by both coculture and RT-PCR. Detection of HIV-1 in peripheral blood plasma by RNA RT-PCR and coculture. Among 398 peripheral blood specimens assayed for HIV-1 RNA, 300 (75.4 percent) were positive, with a median of 39,400 HIV1 RNA copies/mL (range, 10,400-1,600,000). Of 400 peripheral blood specimens cocultured for HFV-1, 362 (90.5 percent) were positive, with a median of 28.3 infectious units/10* cells (range, 0.13-5,608). Seminal leukocyte counts. Seminal lymphocyte, PMN cell, and monocyte cell levels were assessed per TABLE 2. HIV-1 and CMV Isolation in semen by coculture and HIV-1 RNA detection In semen by RNA RT-PCR amplification among 149 men, Seattle, Washington State, 1994-1995* Results ot HIV-1 Isolation RNA RT-PCR, cocutture No. of specimens CMV coculture-posltive No. Negative, negative Negative, positive Positive, negative Positive, positive 80 3 103 24 18 0 41 22.5 0.0 39.8 16 66.7 Total CMV-posWve 210 75 35.7 * Abbreviations: HIV-1, human immunodeficiency virus type-1; CMV, cytomegaiovirus; RNA, ribonudeic add; RT-PCR, reverse transcription polymerase chain reaction. 100 sperm in 285 (69.3 percent) of the 411 specimens. Of those 285 specimens, 44 (15.4 percent) had one or more seminal lymphocytes (range, 0-12); 77 (27.0 percent) had one or more PMN cells (range, 0-22); and 3 (1.1 percent) had one or more monocytes (range, 0-2). Because few specimens contained monocytes, we did not evaluate this variable as a potential predictor of HTV-1 in semen. HIV-1 coculture-positivity in relation to systemic and local host factors (tables 3 and 4). We analyzed the relation between detection of HTV-1 in semen by coculture and a variety of systemic host factors and local urogenital tract factors. Systemic host factors included: peripheral blood CD4+ and CD8+ counts, clinical stage of HTV-1 infection, antiretroviral therapy use, the presence and levels of HTV-1 RNA in peripheral blood plasma, and quantitative and qualitative HTV-1 peripheral blood mononuclear cell culture result. Local urogenital tract factors included: concurrent seminal CMV shedding, seminal PMN and lymphocyte cell counts per mL of semen, and semen volume. Predictors of positive semen HIV-1 cocultures: unlvariate analyses (table 3) Systemic host factors. None of the systemic variables evaluated was associated with semen HTV-1 coculture positivity in the univariate analyses. Odds ratios were elevated for specimens from men with <200 CD4+ cells/(i.liter of peripheral blood, but not significantly so. Local urogenital tract factors. Urogenital tract factors associated with a positive HIV-1 coculture included log|0 seminal PMN cells/mL of semen (odds ratio (OR) = 3.6, 95 percent confidence interval (CI) 1.1, 12.3) and semen CMV coculture positivity (OR = 2.3, 95 percent CI 1.0, 5.4). Thus, for each one-log10 increase in the number of PMN cells/mL of semen, the odds of a positive semen HTV-1 coculture increased by a factor of 3.6. The odds ratios for a one-log|0 increase in seminal PMN cells per 100 sperm and per ejaculate were significantly increased (data not shown). HTV-1 shedding rates were increased in specimens that had higher seminal lymphocyte counts/mL of semen, but this association did not reach statistical significance. The univariate predictors of semen HTV1 coculture outcome did not differ in the restricted analyses (data not shown). Predictors of positive semen HIV-1 cocultures: multlvariate analyses (table 4) Systemic host factors. Although not associated with HTV-1 coculture positivity in the univariate analyAm J Epidemiol Vol. 150, No. 6, 1999 Risk Factors for HIV-1 Shedding in Semen 627 TABLE 3. Unlvarlate predictors of semen HIV-1 Isolation by coculture and detection of HIV-1 RNA In seminal plasma by RT-PCR amplification assay estimated using GEE analysis, Seattle, Washington State, 1994-1995* Semen HIV-1 coculture Odds ratio A Systemic host factors Peripheral Wood CD4+ count/jiLt £500 200-499 <200 Peripheral blood CD8+ count/|iLt 51,200 800-1,199 <800 CD4:CD8 ratio (for each unit Increase) Clinical stage of disease Asymptomatic (Class A) Symptoms (Class B) AIDS (Class C) HIV-1 PBMC coculture titer (for each log10 increase) Blood plasma RNA Negative Positive Blood plasma RNA level (for each log10 increase) Antiretroviral therapy No therapy Monotherapyf Combination therapy§ B. Local urogenital factors Semen CMV coculture Negative Positive Seminal PMN count (for each log10 increased Seminal lymphocyte count (for each log10 increased Seminal volume (for each 1-mL Increase in ejaculate volume) 95% Cl No. of specimens/ men Semen HIV-1 RNA RT-PCR Odds ratio 95% Cl 274/127 1.0 0.8 2.2 348/132 1.0 1.6 0.3, 2.4 0.8, 6.2 2.4 0.8, 2.9 1.2,4.8 274/127 1.0 1.2 2.0 0.5 348/132 1.0 0.3, 4.0 0.6, 6.7 0.1,2.2 0.7 274/127 274/127 1.1 2.4 1.1 0.3, 3.6 0.6, 9.0 0.8, 1.6 1.0 1.0 1.4 0.9, 1.1 0.8, 2.5 1.4 0.3 1.0 1.7 1.0 1.0 0.9 0.8 No. of specimens/ men 3.3 268/127 268/127 1.5 1.0 2.3 268/127 274/127 2.0 0.9, 3.3 1.4,7.9 1.0,2.0 1.0, 5.4 1.1, 12.3 0.1, 348.7 0.7 0.1,3.1 340/132 345/131 1.5,3.6 1.3,3.3 0.5, 1.6 0.2, 1.2 276/121 2367119 1.0 2.3 3.6 5.8 348/132 350/132 350/132 1.0 0.9 0.5 0.3, 2.8 0.3, 2.1 0.4, 1.4 0.7, 2.7 0.1,0.8 237/114 237/114 1.0 1.8 1.6 5.6 1.1,3.1 0.6, 4.6 0.3, 96.7 307/124 307/124 270/126 0.9 0.4, 2.0 344/131 * Abbreviations: Cl, confidence interval; CMV, cytomegalovirus; GEE, generalized estimating equation; HIV-1, human immunodeficiency virus type-1; PBMC, peripheral blood mononuclear cells; PMN polymorphonuclear; RNA, ribonucleic acid; RT-PCR, reverse transcription polymerase chain reaction. t Per mm3 of peripheral blood. t Use of exactly one licensed nucleoside analog drug. § Use of £2 licensed nucleoside analog drugs. H Per mL of semen. sis, we assessed CD4+ count and antiretroviral therapy as potential confounders by including them (alone and together) in the model containing CMV shedding and log|0 number of PMN cells/mL of semen. Neither of the systemic host factors was statistically associated with HTV-1 coculture positivity in these analyses, nor did they substantially alter risk estimates for local urogenital factors. Therefore, neither of these systemic host factor variables was included in the final model. Local urogenital tract factors. PMN cells per mL of semen and semen CMV coculture outcome were tested in multivariate models because they were statistically associated with shedding of HTV-1 in the univariate analyses. When placed together in a model, PMN cell Am J Epidemiol Vol. 150, No. 6, 1999 count (irrespective of the manner in which it was assessed) and concurrent CMV shedding each independently predicted positive semen HTV-1 coculture (ORs (95 percent CIs) 12.6 (1.2, 134.5) and 3.0 (1.2, 7.7), respectively). Results were similar in the restricted analyses (data not shown). Predictors of HIV-1 RNA in seminal plasma: univariate analyses (table 3) Systemic host factors. Systemic factors associated with HTV-1 RNA detection in semen included: CD4+ count <200/(iliter (compared with CD4+ count = 500; OR = 2.4, 95 percent Cl 1.2, 4.8), CD4+:CD8+ cell ratio (for each one unit increase, OR = 0.3, 95 percent 628 Speck et al. TABLE 4. Multivariate models for predicting semen HIV-1 Isolation (n = 207 specimens from 108 men) and detection of HIV-1 RNA In seminal plasma by RT-PCR (n n 275 specimens from 121 men) estimated using GEE analysis, Seattle, Washington State, 1994-1995* Semen HIV-1 cocutture Odds ratio if in model 95% Cl A. Systemic (actors Peripheral Wood CD4+ countf 2500 200-t99 <200 Antiretroviral therapy No therapy MonotherapyJ Combination therapy§ B. Local urogenltal factors Semen CMV coculture Negative Positive Each log t0 increase in seminal PMN cell courrtH 1.0 3.0 12.6 1.2,7.7 Semen HIV-1 RNART-PCR Odds ratio [fin model 95% a 1.0 1.6 3.0 0.8, 3.3 1.3,6.9 1.0 0.5 0.4 0.3, 1.0 0.2, 0.9 1.0 1.7 1.0,3.0 1.2, 134.5 • Abbreviations: Cl, confidence interval; CMV, cytomegalovirus; GEE, generalized estimating equation; HIV-1, human Immunodeficiency virus type-1; PMN polymorphonuclear; RNA, ribonucleic acid, t Per mm3 of peripheral Wood. $ Use of exactly one licensed nucleoside analog drug. § Use of two or more licensed nucleoside analog drugs. H Per mL of semen. Cl 0.1, 0.8), class "C" clinical disease (AIDS) (OR = 3.3, 95 percent Cl, 1.4, 7.9), log10 copy number of infectious units/10* peripheral blood mononuclear cells (OR = 1.5, 95 percent Cl 1.0, 2.0), and the presence (OR = 2.3, 95 percent Cl 1.5, 3.6) and levels (OR = 2.0, 95 percent Cl 1.3, 3.3) of HTV-1 RNA in peripheral blood plasma. The odds ratio for clinical stage B (mild constitutional symptoms) was elevated but not statistically significant. Combination nucleoside antiretroviral therapy was associated with a decrease in the odds of detecting HTV-1 RNA in seminal plasma, but this decrease was not significant. CD8+ count was not associated with semen HTV-1 RNA assay outcome. Local urogenital tract factors. Of the local urogenital risk factors assessed, only semen CMV coculture positivity was associated with detection of HTV-1 RNA in seminal plasma (OR = 1.8, 95 percent Cl 1.1, 3.1). Seminal PMN and lymphocyte cell counts/mL of semen, per 100 sperm, and per ejaculate, and semen volume, were not associated with detection of HTV-1 RNA in seminal plasma. Results were similar in the restricted analyses (data not shown). Predictors of HIV-1 RNA in seminal plasma: multivarlate analyses (table 4) Systemic host factors. Systemic factors statistically associated with seminal plasma HTV-1 RNA copy number in univariate analyses were considered for inclusion in the multivariate model. Antiretroviral therapy was also considered to assess its effect after adjustment for CD4+ count. Of these systemic host factors, low CD4+ count <200/jxliter of peripheral blood (compared with £500) increased the odds of detection of HTV-1 RNA in seminal plasma (OR = 3.0, 95 percent Cl 1.3, 6.9). In contrast, monotherapy (OR = 0.5, 95 percent Cl 0.3, 1.0) or combination nucleoside antiretroviral therapy (OR = 0.4, 95 percent Cl 0.2, 0.9) decreased the odds of detecting HTV-1 RNA in seminal plasma. Local urogenital tract factors. Semen CMV coculture positivity, the only local urogenital host factor associated with detection of HIV-1 RNA in semen in the univariate analyses, was also evaluated for inclusion in the multivariate model. When peripheral blood CD4+ count and antiretroviral therapy status were taken into account, this variable remained associated with detection of HTV-1 RNA in seminal plasma (OR = 1.7, 95 percent Cl 1.0, 3.0). The factors that comprised the best multivariate model were therefore CD4+ count, antiretroviral therapy use, and semen CMV coculture result. The same factors formed the multivariate model in the restricted analyses (data not shown). Am J Epidemiol Vol. 150, No. 6, 1999 Risk Factors for HIV-1 Shedding in Semen DISCUSSION We believe this study to be one of the largest investigations of risk factors for detection of HTV-1 in semen, as well as one of the few studies in which rigorous epidemiologic and biostatistical methods were incorporated into the design. Subjects were seen at 4week intervals, and semen and blood were collected at each study visit. Subject compliance was excellent; most men (85.9 percent) completed all three study visits. Nevertheless, we did three things to assess the potential for bias due to differential dropout rates among study subjects. First, we conducted restricted analyses in which only observations with complete data for all three semen assays (HTV-1 coculture, CMV coculture, and HTV-1 RNA) were analyzed. Neither risk estimates nor the risk factors that comprised the multivariate models differed in these analyses. Second, we compared baseline CD4+ counts and systemic viral load values between men who completed less than three visits and those who completed all three visits. These baseline values did not differ significantly. Third, we conducted univariate analyses using a GEE approach to assess the association between number of visits completed (<3 visits vs. 3 visits), semen HTV-1 coculture results, and semen HTV-1 RNA assay results. Number of visits completed was not associated with either of these outcomes (data not shown). CMV shedding in semen, a local urogenital tract factor, was the only variable assessed that predicted both positive semen HTV-1 coculture and positive HTV-1 RNA assay of seminal plasma. Increased seminal PMN cell count, another local urogenital tract factor, predicted positive semen HTV-1 coculture but did not predict detection of HTV-1 RNA in seminal plasma. In contrast, low CD4+ count and lack of antiretroviral therapy, both systemic host factors, predicted detection of HTV-1 RNA in seminal plasma, but not by semen cocultivation. We are not aware of prior reports that concurrent shedding of CMV in semen predicts detection of HTV1 in semen by coculture or RT-PCR after controlling for CD4+ counts and antiretroviral therapy. However, the relation between CMV shedding and HTV-1 shedding in semen is biologically plausible. In vitro data suggest that bi-directional interactions occur between HTV-1 and CMV (48); these interactions could provide mechanisms for the activation of T-lymphocytes, which has been proposed as a prerequisite for efficient HTV-1 replication (49), and is analogous to immunologic activation of retroviruses in murine models (50). It has also been suggested that herpes group viruses, such as CMV, may enhance HTV-1 production by infected immune cells (51), forming a positive feedback loop among herpes viruses and HTV-1 -induced Am J Epidemiol Vol. 150, No. 6, 1999 629 growth factors, cell proliferation, and HTV-1 replication (52). A chemokine receptor homologue (US28) encoded by CMV could be an additional cofactor for HTV-1 entry (53). Together, these observations suggest that interactions between HTV-1 and CMV may play a role in the sexual transmission of HTV-1. Our report that seminal PMN levels predict detection of HTV-1 by coculture but not by viral RNA is consistent with the observations of other investigators (23, 30, 54, 55) on the importance of sexually transmitted diseases and genital tract inflammation as potential risk factors for increased shedding of HTV-1. These observations suggest that there may be a compartmentalization of infectious virus to the cell-associated semen pellet, while cell-free viral RNA in the seminal plasma may arise from other cellular compartments within the male genital tract (24). In addition, these observations suggest that efforts to treat urethritis, prostatitis, and other inflammatory conditions that affect the male urogenital tract may limit the potential for sexual transmission of HTV-1 (55). Our findings are to some degree consistent with other reports that low CD4+ count, a systemic host factor, predicts shedding of HTV-1 in semen, in that detection of HTV-1 RNA in seminal plasma was more likely if specimens were obtained from men with <200 CD4+ cells/nliter of peripheral blood (19, 30). However, the reports by others were based on coculture. This study is consistent with our previous work, in that we did not find an association between CD4+ counts and HTV-1 shedding in semen as assessed by coculture (29, 33). Our finding that both mono- and combination nucleoside antiretroviral reduced shedding of HTV-1 in semen confirms the report by Anderson et al. (30) that zidovudine use was associated with decreased rates of detection of HTV-1 in semen. In contrast to the local urogenital tract factors, systemic host factors were inconsistent predictors of HTV1 shedding in semen. In univariate analyses, clinical stage of HTV-1 infection, counts of CD4+ and CD8+ cells in peripheral blood, CD4+:CD8+ cell ratio, nucleoside antiretroviral therapy, and systemic viral load (assessed by either peripheral blood mononuclear cell culture or plasma HTV-1 RNA) were associated with detection of HTV-1 RNA in the seminal plasma, but not with cocultivation of HTV-1 from semen. In the multivariate analyses, none of the systemic host factors approached significance as potential predictors of HTV-1 shedding in semen as assessed by cocultivation. CD4+ cell counts <200/^liter of peripheral blood, mono-antiretroviral therapy, and combination antiretroviral therapy predicted detection of cell-free HTV1 RNA in the seminal plasma. These findings suggest 630 Speck et al. that our study may have lacked sufficient power to detect an effect of CD4+ cell count or combination nucleoside antiretroviral therapy on isolation of cultivable HTV-1 in semen. Quantitative determinations of HTV-1 RNA levels in semen or studies that employ even larger sample sizes may be necessary to establish the effect of these factors on HTV-1 shedding in semen (24). These findings do suggest that highly active antiretroviral agents, such as protease inhibitors, that produce larger reductions in the systemic viral burden, may be more effective in reducing seminal shedding of HTV-1 (56). From a clinical standpoint, the important implication is that healthy, HTV-1 seropositive patients on antiretroviral therapy should still regard themselves as infectious for their sexual partners. This report has several strengths and limitations. Its strengths include the large number of men enrolled and specimens collected; its high patient compliance; the use of random effects techniques in statistical analyses to recognize and account for intra-subject correlations; and the use of sensitive molecular techniques for detection of HTV-1 RNA in seminal plasma. Thus, because many semen specimens were of insufficient volume to be evaluated by HTV-1 coculture, CMV coculture, and HTV-1 RNA RT-PCR, study power was reduced in models containing more than two variables. Neither the univariate nor the multivariate models presented here differed in the restricted analyses, which lessens the likelihood that systematic bias substantially influenced our results. However, the low study power may have prevented us from detecting significant relations between semen HTV-1 coculture positivity and some of the other risk factors investigated. In addition, other than direct measurement of cultivable virus from semen, we can only infer the impact of presence of viral RNA on semen infectivity. Furthermore, the insensitivity of coculture may also have limited study power to detect relations between infectivity, presence of viral RNA, and our inferences about sexual transmissibility. In summary, our findings confirm and extend the results of our prior reports and those other researchers that local urogenital tract factors, such as seminal PMN cell count and concurrent shedding of CMV in semen, were the best predictors of HTV-l shedding in semen, hi contrast, systemic viral load was a poor predictor of HIV-1 shedding. CD4+ cells counts <200/|iliter of peripheral blood and the use of nucleoside antiretroviral therapy were associated with shedding of HTV-1 as assessed by cell-free HTV-1 RNA assay, but not by cocultivation. In addition to direct measurements of HTV-1 in semen, an assessment of local urogenital tract factors may be necessary to determine the potential for sexual transmission of HTV-1. ACKNOWLEDGMENTS Supported by NIH grant nos. DK-49477, AI-27664, AI30731, and AI-27757 and the Paul G. Allen Foundation for Medical Research. Dr. Speck was a University of Washington NIH/NIAID STD/HIV Predoctoral trainee, supported by grant no. 05-T32-AI-07140. The authors thank Dr. Diana Petitti for her help in assembling the final manuscript and Pam Easterling for technical assistance in preparing the manuscript. REFERENCES 1. Blattner WA. HTV epidemiology: past, present, and future. FasebJ 1991;5:2340-8. 2. Calabrese LH, Gopalakrishna KV. Transmission of HTLV-DI infection from man to woman to man. (Letter). N Engl J Med 1986;314:987. 3. Friedland GH, Klein RS. Transmission of the human immunodeficiency virus. N Engl J Med 1987;317:1125-35. 4. Allain JP. Prevalence of HTLV-IH/LAV antibodies in patients with hemophilia and in their sexual partners in France. (Letter). N Engl J Med 1986;315:517-18. 5. Fischl MA, Dickinson GM, Scott GB, et al. Evaluation of heterosexual partners, children, and household contacts of adults with AIDS. JAMA 1987;257:64(M. 6. Kreiss JK, Kitchen LW, Prince HE, et al. Antibody to human T-lymphotropic virus type i n in wives of hemophiliacs. Evidence for heterosexual transmission. Ann Intern Med 1985;102:623-6. 7. Mann JM, Quinn TC, Francis H, et al. Prevalence of HTLVin/LAV in household contacts of patients with confirmed AIDS and controls in Kinshasa, Zaire. JAMA 1986;256: 721-4. 8. Peterman TA, Stoneburner RL, Allen JR, et al. Risk of human immunodeficiency virus transmission from heterosexual adults with transfusion-associated infections. JAMA 1988;259:55-8. 9. Redfield RR, Markham PD, Salahuddin SZ, et al. Frequent transmission of HTLV-in among spouses of patients with AIDS-related complex and AIDS. JAMA 1985;253:1571-3. 10. Padian NS, Shiboski SC, Jewell NP. Female-to-male transmission of human immunodeficiency virus. JAMA 1991 ;266: 1664-7. 11. De Gruttola V, Mayer KH. Assessing and modeling heterosexual spread of the human immunodeficiency virus in the United States. Rev Infect Dis 1988; 10:138-50. 12. Johnson AM, Petherick A, Davidson SJ, et al. Transmission of HTV to heterosexual partners of infected men and women. AIDS 1989;3:367-72. 13. European Study Group. Risk factors for male to female transmission of HTV. European Study Group. BMJ 1989;298: 411-15. 14. Laurian Y, Peynet J, Verroust F. HTV infection in sexual partners of HTV-seropositive patients with hemophilia. (Letter). N Engl J Med 1989;320:183. [Published erratum appears in N Engl J Med 1989;321:268]. 15. Zagury D, Bernard J, Leibowitch J, et al. HTLV-UI in cells cultured from semen of two patients with AIDS. Science 1984; 226:449-51. 16. Ho DD, Schooley RT, Rota TR, et al. HTLV-UI in the semen and blood of a healthy homosexual man. Science 1984;226: 451-3. 17. van Voorhis BJ, Martinez A, et al. Detection of human immunodeficiency virus type 1 in semen from seropositive men using culture and polymerase chain reaction deoxyribonucleic acid amplification techniques. Fertil Steril 1991;55:588-94. 18. Krieger JN, Coombs RW, Collier AC, et al. Fertility parameAm J Epidemiol Vol. 150, No. 6, 1999 Risk Factors for HIV-1 Shedding in Semen ters in men infected with human immunodeficiency virus. J Infect Dis 1991;164:464-9. 19. Hamed KA, Winters MA, Holodniy M, et al. Detection of human immunodeficiency virus type 1 in semen: effects of disease stage and nucleoside therapy. J Infect Dis 1993; 167:798-802. 20. Krieger JN, Coombs RW, Collier AC, et al. Intermittent shedding of human immunodeficiency virus in semen: implications for sexual transmission. J Urology 1995;154:1035-40. 21. Rasheed S, I i Z, Xu D. Human immunodeficiency virus load. Quantitative assessment in semen from seropositive individuals and in spiked seminal plasma. J Reprod Med 1995;40: 747-57. 22. Mermin JH, Holodniy M, Katzenstein DA, et al. Detection of human immunodeficiency virus DNA and RNA in semen by the polymerase chain reaction. J Infect Dis 1991;164:769-72. 23. Vernazza PL, Eron JJ, Cohen MS, et al. Detection and biologic characterization of infectious HTV-1 in semen of seropositive men. AIDS 1994;8:1325-9. 24. Coombs RW, Speck CE, Hughes JP, et al. Association between culturable human immunodeficiency virus type 1 (HIV-1) in semen and HTV-1 RNA levels in semen and blood: evidence for compartmentalization of HTV-1 between semen and blood. J Infect Dis 1998;177:320-30. 25. Krieger JN, Coombs RW, Collier AC, et al. Recovery of human immunodeficiency virus type 1 from semen: minimal impact of stage of infection and current antiviral chemotherapy. J Infect Dis 1991;163:386-88. 26. B'Elec L, T'Evi BeC, Lu XS, et al. Local synthesis of IgG antibodies to HTV within the female and male genital tracts during asymptomatic and pre-AIDS stages of HTV infection. AIDS Res Hum Retroviruses 1995;ll:719-29. 27. Zhu T, Wang N, Carr A, et al. Genetic characterization of human immunodeficiency virus type 1 in blood and genital secretions: evidence for viral compartmentalization and selection during sexual transmission. J Virol 1996;70:3098-107. 28. Coombs RW, Krieger JN, Collier AC, et al. Plasma viremia and recovery of HTV from semen: implications for transmission and therapy. In: Me'lica F, ed. AIDS and human reproduction. Basel: Karger, 1992:146-9. 29. Krieger JN, Coombs RW, Collier AC, et al. Recovery of human immunodeficiency virus type 1 from semen: minimal impact of stage of infection and current antiviral chemotherapy. J Infect Dis 1991;163:386-8. 30. Anderson DJ, O'Brien TR, Politch JA, et al. Effects of disease stage and zidovudine therapy on the detection of human immunodeficiency virus type 1 in semen. JAMA 1992;267: 2769-74. 31. Detels R, Leach CT, Hennessey K, et al. Persistent cytomegaloviras infection of semen increases risk of AIDS. J Infect Dis 1994;169:766-8. 32. Leach CT, Detels R, Hennessey K, et al. A longitudinal study of cytomegalovirus infection in human immunodeficiency vims type 1-seropositive homosexual men: molecular epidemiology and association with disease progression. J Infect Dis 1994;170:293-8. 33. Krieger JN, Coombs RW, Collier AC, et al. Seminal shedding of human immunodeficiency virus type 1 and human cytomegalovirus: evidence for different immunologic controls. J Infect Dis 1995;171:1018-22. 34. Coutere M, Ulstein M, Leonard J, et al. Improved staining method for differentiating immature sperm from white blood cells in human seminal fluid. Andrologia 1976;8:61—6. 35. World Health Organization Task Force on Sperm Antigens. Light and electron microscopic immunolocalization of sperm proteins identified by monoclonal antibodies from the World Health Organization Task Force on Sperm Antigens. J Reprod Immunol 1992;3:237-56. 36. Branigan EF, Muller CH. Efficacy of treatment and recurrence rate of leukocytospermia in infertile men with prostatitis. Fertil Steril 1994;62:580-4. Am J Epidemiol Vol. 150, No. 6, 1999 631 37. Boom R, Sol CJA, Salimans MMM, et al. Rapid and simple method for purification of nucleic acids. J Clin Microbiol 1990;28:495-503. 38. 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 infection. J Clin Microbiol 1994,32:292-300. 39. Lin HJ, Myers LE, Yen-Lieberman B, et al. Multicenter evaluation of methods for the quantitation of plasma HTV-1 RNA. J Infect Dis 1994; 170:553-62. 40. ACTG virology manual for HTV laboratories. Bethesda, MD: Division of AIDS, National Institutes of Allergy and Infectious Diseases, NIH, 1994. (Publication no. NIH-94-3828). 41. Goldstein LC, Corey L, McDougall JK, et al. Monoclonal antibodies to herpes simplex viruses: use in antigenic typing and rapid diagnosis. J Infect Dis 1983;147:829-37. 42. Corey L, Dragavon J, Benjamin D. Immunoperoxidase staining in the clinical virology laboratory. In: Tilton R, ed. Proceedings of the 3rd International Symposium on Rapid Methods and Automation in Microbiology. Washington, DC: American Society of Microbiology, 1982:246-53. 43. Wentworth BB, Bonin P, Holmes KK, et al. Isolation of viruses, bacterial and other organisms from venereal disease clinic patients: methodology and problems associated with multiple isolations. Health Lab Sci 1973;10:75-81. 44. Todd J, Yeghiazarian T, Hoo B, et al. Quantitation of human immunodeficiency virus plasma RNA by branched DNA and reverse transcription coupled polymerase chain reaction assay methods: a critical evaluation of accuracy and reproducibility. Serodiagn Immunother Infect Dis 1994;6:233-9. 45. Brambilla D, Leung S, Lew J, et al. Absolute copy number and relative change in determinations of human immunodeficiency virus type 1 RNA in plasma: effect of an external standard on kit comparisons. J Clin Microbiol 1998;36:311-14. 46. Kagan J, Gelman R, Waxdal M, et al. NIAID Division of AIDS flow cytometry quality assessment program. Ann NY Acad Sci 1993;677:50-2. 47. Liang K-Y, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika 1986;73:13-22. 48. Skolnik PR, Kosloff BR, Hirsch MS. Bidirectional interactions between human immunodeficiency virus type 1 and cytomegalovirus. J Infect Dis 1988; 157:508-14. 49. Zagury D, Bernard J, Leonard R, et al. Long-term cultures of HTLV-m-infected T cells: a model of cytopathology of T-cell depletion in AIDS. Science 1986;231:85O-3. 50. Hirsch MS, Phillips SM, Solnik C, et al. Activation of leukemia viruses by graft-versus-host and mixed lymphocyte reactions in vitro. Proc Natl Acad Sci USA 1972;69:1069-72. 51. Osborn L, Kunkel S, Nabel GJ. Tumor necrosis factor alpha and interleukin 1 stimulate the human immunodeficiency virus enhancer by activation of the nuclear factor kappa B. Proc Natl Acad Sci USA 1989;86:2336-4O. 52. Laurence J. Molecular interactions among herpesviruses and human immunodeficiency viruses. J Infect Dis 1990; 162: 338-46. 53. Pleskoff O, Treboute C, Brelot A, et al. Identification of a chemokine receptor encoded by human cytomegalovirus as a cofactor for HTV-1 entry. Science 1997;276:1874-8. 54. Moss GB, Overbaugh J, Welch M, et al. Human immunodeficiency virus DNA in urethral secretions in men: association with gonococcal urethritis and CD4 cell depletion. J Infect Dis 1995; 172:1469-74. 55. Cohen MS, Hoffman IR, Royce RA, et al. Reduction of concentration of HTV-1 in semen after treatment of urethritis: implications for prevention of sexual transmission of HTV-1. Lancet 1997;349:1868-73. 56. Gupta P, Mellors J, Kingsley L, et al. High viral load in semen of human immunodeficiency virus type 1-infected men at all stages of disease and its reduction by therapy with protease and nonnucleoside reverse transcriptase inhibitors. J Virology 1997;71:6271-5.