Monitoring Plasma HIV-1 RNA Levels in Addition to CD4+ Lymphocyte Count Improves Assessment of Antiretroviral Therapeutic Response

Methods

Study Design

We prospectively evaluated virological, immunologic, and clinical data from patients who participated in the intensive virology substudy of ACTG (AIDS Clinical Trials Group) Protocol 241; ACTG Protocol 241 was a multicenter, randomized, double-blind, placebo-controlled trial of 398 patients receiving nevirapine, zidovudine, and didanosine compared with zidovudine and didanosine [17]. All patients at 8 of the 16 AIDS Clinical Trials Units who participated in the main study were enrolled in the substudy (n = 198). For 48 weeks, all 198 patients received open-label zidovudine (600 mg/d) and didanosine (400 mg/d for patients weighing ≥ 60 kg and 250 mg/d for patients weighing <60 kg). One hundred of the substudy patients were randomly assigned to receive nevirapine (200 mg/d for the first 2 weeks and 400 mg/d thereafter), and 98 were assigned to receive matching placebo. Participants gave written informed consent, and the protocol was approved by the institutional review board at each participating AIDS Clinical Trials Unit.

The study was funded by the ACTG of the National Institute of Allergy and Infectious Diseases; supplemental funding for virology was provided by Boehringer Ingelheim Pharmaceuticals (Ridgefield, Connecticut). Study drugs were provided by Glaxo Wellcome (Research Triangle Park, North Carolina), Bristol-Myers Squibb (Princeton, New Jersey), and Boehringer Ingelheim Pharmaceuticals. However, all data were gathered by members of the ACTG and were analyzed and interpreted by the authors, who had sole responsibility for the decision to submit the manuscript for publication.

Evaluation of Patients

Stable therapy at baseline was defined as the absence of reported change in antiretroviral therapy from 30 days before the preentry visit until the entry visit. All patients were followed prospectively for progression of HIV-related disease. Progression was defined as the development of a new acquired immunodeficiency syndrome (AIDS)-defining event [18]; a newly diagnosed, deep-seated bacterial infection or bacteremia that was not related to injection drug use or an intravascular catheter; pulmonary or extrapulmonary tuberculosis; recurrent Pneumocystis carinii pneumonia; recurrent toxoplasmosis of the central nervous system; or death. Reports of disease progression were reviewed by the study chair; only events that could be confirmed were used in the analysis.

We measured CD4+ lymphocyte counts, plasma HIV-1 RNA levels, and infectious HIV-1 titers in mononuclear cells of peripheral blood at the preentry visit (within 14 days of starting study treatment), at the entry visit (before starting study treatment and at least 72 hours after the preentry visit), and at the visits 8 and 48 weeks after the start of study treatment. Specimens could be obtained at any time of day. We used the geometric mean of preentry and entry measurements as the baseline value for each variable. The presence of the syncytium-inducing viral phenotype was determined at the entry visit.

Standardized assays were used to determine CD4+ lymphocyte counts [19, 20], infectious HIV-1 titer in mononuclear cells of peripheral blood (in infectious units per million cells) using real-time testing [13, 21], and syncytium-inducing viral phenotype of a virus isolated from mononuclear cells of peripheral blood using MT-2 cells [22]. Plasma samples were frozen at −70°C; HIV-1 RNA levels were measured by quantitative reverse transcription polymerase chain reaction assay (Roche Molecular Systems, Alameda, California, and Branchburg, New Jersey) [23]. The lower limit of detection for this assay was 200 HIV-1 RNA copies/mL. Levels of HIV-1 RNA in plasma samples collected from the same patient at the preentry, entry, week 8, and week 48 visits were determined in a single laboratory assay.

Statistical Analysis

Analysis of plasma HIV-1 RNA levels and infectious HIV-1 titers in mononuclear cells of peripheral blood was done after log₁₀ transformation. Plasma levels of HIV-1 RNA that were below the detectable limit were assigned the value of 200 copies/mL. Infectious HIV-1 titers in mononuclear cells of peripheral blood outside the measurable range (0.22 to 7493 infectious units per million cells) were assigned the value of 0.22 infectious units per million cells if they were below the range and 7493 infectious units per million cells if they were above the range.

Linear regression analysis [24] was used to compare the mean plasma HIV-1 RNA levels, infectious HIV-1 titers in mononuclear cells of peripheral blood, and CD4+ lymphocyte counts according to patient characteristics at baseline and to assess factors associated with the long-term change (from baseline to week 48) in CD4+ lymphocyte counts. Logistic regression analysis [25] was used to assess the association at baseline of the percentage of patients who had AIDS with virological measures and CD4+ lymphocyte counts. The intrapatient SD of plasma HIV-1 RNA levels was estimated using the method of moments for variance components [26]. Spearman correlation coefficients were used to assess the association between preentry and entry measurements.

Proportional hazards models [27] were used to assess the association between the risk for disease progression or death and baseline levels and early changes (from baseline to week 8) in plasma HIV-1 RNA levels, infectious HIV-1 titers in mononuclear cells of peripheral blood, and log-transformed CD4+ lymphocyte counts as well as baseline syncytium-inducing viral phenotype. These models were stratified by study treatment to control for any differential effects of the two study regimens.

Results

Patient Characteristics at Study Entry

The mean CD4+ lymphocyte count of the 198 patients before treatment was 145 cells/mm³ (range, 1 to 443 cells/mm³). Patients were a median of 39 years of age, predominantly male (81%), predominantly white (76%), and predominantly free of a previous AIDS-defining diagnosis (86%). All but 3 patients had taken zidovudine before study entry, 44% had taken didanosine, and 35% had taken zalcitabine. The median duration of cumulative previous nucleoside therapy was 25 months, and 34% of patients had received therapy for longer than 36 months.

Virological Measures at Baseline by Patient Characteristics

Table 1 shows the mean plasma HIV-1 RNA levels, infectious HIV-1 titers in mononuclear cells of peripheral blood, and CD4+ lymphocyte counts at baseline for patients stratified by characteristics that were significantly associated with viral load. We also assessed the associations with age, sex, racial or ethnic group, self-reported homosexuality, and duration of previous nucleoside therapy, but these associations were not significant.

Patients with a history of AIDS had a significantly higher mean baseline level of HIV-1 RNA in plasma and a significantly lower mean CD4+ lymphocyte count than did those without such a history (Table 1). More patients with a history of AIDS than those without had baseline HIV-1 isolates with the syncytium-inducing viral phenotype (58% compared with 36%; P = 0.015). However, in a multivariate analysis, only the CD4+ lymphocyte count at baseline was significantly associated with a history of AIDS. Thus, disease status at baseline was explained by CD4+ lymphocyte counts and not by any of the virological measures that were considered.

Variability of Virological Measures in Patients Receiving Stable Treatment

Variation in plasma HIV-1 RNA levels was evaluated by comparing the preentry and entry measures from the 167 patients who reported no changes in treatment from 30 days before the preentry visit to the entry visit. The median time between the two visits was 6 days (range, 2 to 15 days). The intrapatient SD of plasma HIV-1 RNA levels was ± 0.19 log₁₀ copies/mL. Figure 1 shows the difference between the preentry and entry HIV-1 RNA plasma levels plotted against the mean of the two measures (the baseline level). The median difference was zero, indicating that plasma HIV-1 RNA levels were stable during this period. Moreover, the magnitude of the difference between the preentry and entry measures was independent of the level at baseline (P = 0.14). The range of differences was between −0.25 and +0.25 log₁₀ copies/mL (1.8-fold) for 80% of patients and between −0.39 and +0.39 log₁₀ copies/mL (2.5-fold) for 90% of patients (the latter range is indicated by the dashed lines in Figure 1).

The intrapatient variability in infectious HIV-1 titers in mononuclear cells of peripheral blood was greater than that for the plasma HIV-1 RNA level. The preentry and entry infectious HIV-1 titers in mononuclear cells of peripheral blood were less well correlated (r = 0.68) than were the two measures of plasma HIV-1 RNA levels (r = 0.94), and the difference between the two determinations of infectious HIV-1 titer was between −1.17 and +1.17 log₁₀ infectious units per million cells (15-fold) for 90% of patients.

Association between Virological Measures and CD4+ Lymphocyte Count at the Entry Visit

The level of HIV-1 RNA in plasma was significantly associated with the CD4+ lymphocyte count at study entry (Figure 2). The mean plasma HIV-1 RNA level (Figure 2) increased by 0.39 log₁₀ copies/mL (2.5-fold) for each reduction of 100 cells/mm³ in CD4+ lymphocyte count at entry (95% CI, 0.29 to 0.49 log₁₀ copies/mL [P < 0.001]). For any particular CD4+ lymphocyte count, however, plasma HIV-1 RNA levels varied considerably among individual patients, as indicated by the wide vertical scatter of points around the line in Figure 2. For example, among patients with CD4+ lymphocyte counts of 200 to 300 cells/mm³ at the entry visit, plasma HIV-1 RNA levels ranged from below the limit of detection of the assay (2.30 log₁₀ copies/mL; 200 copies/mL) to 5.37 log₁₀ copies/mL (234 000 copies/mL). The association between infectious HIV-1 titer in mononuclear cells of peripheral blood and CD4+ lymphocyte count at the entry visit was similar in magnitude to that for plasma HIV-1 RNA levels: a mean increase in titer of 0.35 log₁₀ infectious units per million cells (2.2-fold) for each reduction of 100 cells/mm³ in entry CD4+ lymphocyte count (CI, 0.21 to 0.49 log₁₀ infectious units per million cells).

A significant, although moderate, correlation of 0.66 was seen between infectious HIV-1 titers in mononuclear cells of peripheral blood and plasma HIV-1 RNA levels obtained at the entry visit.

Plasma HIV-1 RNA Levels Predict a Decline in CD4+ Lymphocyte Count

Among the 141 patients who had CD4+ lymphocyte counts measured at week 48, the mean change ±SD from baseline was a loss of 4 ± 64 cells/mm³. All virological measures and their changes from baseline to week 8 were significant univariate predictors of the long-term change in CD4+ lymphocyte count at week 48. In particular, the average long-term change in CD4+ lymphocyte count at week 48 was significantly associated with the plasma HIV-1 RNA level at baseline (Figure 3, top) (P = 0.021) and the early treatment-mediated change in plasma HIV-1 RNA levels at week 8 (Figure 3, bottom) (P < 0.001). However, the wide vertical scatter of points in these figures shows the large variability of change in CD4+ lymphocyte count during therapy in patients regardless of the individual plasma HIV-1 RNA levels at baseline or any treatment-mediated change. In a multivariate analysis adjusted for CD4+ lymphocyte counts at baseline and the early change in CD4+ lymphocyte counts to week 8, each 1 log₁₀ copies/mL lower plasma HIV-1 RNA level was associated with an increase of 15 cells/mm³ in mean CD4+ lymphocyte count to week 48 (CI, 3 to 27 cells/mm³ [P = 0.015]). Moreover, each additional decrease of 1 log₁₀ copies/mL in plasma HIV-1 RNA levels from baseline to week 8 was associated with an additional increase of 30 cells/mm³ in mean CD4+ lymphocyte count to week 48 (CI, 20 to 44 cells/mm³ [P < 0.001]). These associations did not differ significantly by treatment assignment.

When they were added to the multivariate model, infectious HIV-1 titers in mononuclear cells of peripheral blood at baseline and the early change in infectious titers from baseline to week 8 were not significantly associated with long-term change in CD4+ lymphocyte count at week 48. Patients with the syncytium-inducing phenotype of an HIV-1 isolate at baseline had a mean long-term decrease in CD4+ lymphocyte count that was increased by 23 cells/mm³ when added to the model (CI, 0 to 46 cells/µL [P = 0.050]).

Plasma HIV-1 RNA Levels Predict Clinical Progression

Progression of HIV-related disease or death was confirmed in 34 patients (17%) during the 48 weeks of follow-up. For 30 patients, the first event was disease progression; for 4 patients, it was death (all deaths were HIV-related). We did not see a significant difference between the treatment groups in the number of patients who had progression or died (22 for the three-drug combination compared with 14 for the two-drug combination [P > 0.2]). All the virological measures and the CD4+ lymphocyte counts at baseline were significant univariate predictors of the risk for disease progression or death (Table 2).

Table 3 shows the number of patients who had progression or died, stratified by plasma HIV-1 RNA levels and CD4+ lymphocyte counts at baseline. The risk for disease progression or death was greatest for patients who, at baseline, had the lowest CD4+ lymphocyte counts and the highest plasma HIV-1 RNA levels. The predictive value of plasma HIV-1 RNA levels and CD4+ lymphocyte counts was improved when both values were taken into account. For example, 26 of 66 patients (39%) who had a CD4+ lymphocyte count in the lowest third (≤ 63 cells/mm³) developed AIDS or died. Adding plasma HIV-1 RNA levels at baseline to the CD4+ lymphocyte counts in these patients improved the determination of risk: Seventeen of 35 patients (49%) who also had a plasma HIV-1 RNA level in the highest third (≥ 113 000 copies/mL), compared with 9 of 31 patients (29%) who had lower levels (<113 000 copies/mL), had progression to AIDS or died (Table 3).

In a multivariate proportional hazards model that included only baseline predictors, plasma HIV-1 RNA levels and CD4+ lymphocyte counts were both significantly associated with risk for disease progression (Table 2). Infectious HIV-1 titers in mononuclear cells of peripheral blood and syncytium-inducing viral phenotype at baseline were not significantly associated with more rapid progression when added to this model.

Progression of HIV-related disease was confirmed in 28 patients after week 8. In addition to the values at baseline, changes from baseline to week 8 in plasma HIV-1 RNA levels and infectious HIV-1 titers in mononuclear cells of peripheral blood (but not changes in CD4+ lymphocyte counts) were significantly associated with risk for progression in univariate analyses (Table 2). Table 4 shows the importance of plasma HIV-1 RNA levels at baseline and the effect of early suppression at week 8 on prognosis: Five of 65 patients (8%) who had decreases in excess of 0.39 log₁₀ copies/mL (2.5-fold) had progression or died, 9 of 48 patients (19%) who had no change or decreases of 0.39 log₁₀ copies/mL or less (no change to 2.5-fold decrease) had progression or died, and 14 of 57 (25%) with any increase had progression or died (Table 4). This trend was evident among patients in the upper two thirds of the distribution of HIV-1 RNA levels at baseline (Table 4).

In a multivariate proportional hazards model with adjustment for plasma HIV-1 RNA levels and CD4+ lymphocyte counts at baseline, early suppression of plasma HIV-1 RNA levels of 1 log₁₀ copies/mL (10-fold) at week 8 was associated with a 52% reduction in the risk for progression or death after week 8 (CI, 6% increase to 79% reduction [P = 0.071]) (Table 2). The extent of this association between treatment-mediated suppression of HIV-1 RNA plasma levels and disease progression was similar to an association seen for each 10-fold lower HIV-1 RNA level at baseline in this model (risk was reduced by 56% for each 10-fold lower HIV-1 RNA level at baseline; CI, 8% to 79% reduction [P = 0.028]) (Table 2). In addition, each 2-fold higher CD4+ cell count at baseline was associated with a 67% reduction in risk for disease progression (CI, 42% to 81% reduction [P < 0.001]) (Table 2). These associations did not differ significantly by treatment assignment.

Discussion

These data show how the prediction of a decline in CD4+ lymphocyte counts and disease progression during 1 year of antiretroviral therapy can be optimized with currently available assays. This optimization can be achieved if CD4+ lymphocyte counts and plasma HIV-1 RNA levels are measured before an antiretroviral regimen is started or changed and if plasma HIV-1 RNA levels are measured shortly thereafter.

Our multivariate models show the importance of CD4+ T-lymphocyte counts and plasma HIV-1 RNA levels at baseline as independent predictors of a decline in CD4+ lymphocyte counts and risk for progression of HIV-related disease (Table 2). We also found that determining CD4+ lymphocyte count and plasma HIV-1 RNA levels once predicted the risk for disease progression better than either determining CD4+ lymphocyte count twice without determining plasma HIV-1 RNA levels or determining plasma HIV-1 RNA levels twice without determining CD4+ lymphocyte counts (data not shown). This finding again emphasizes the benefit of measuring both variables at baseline. Our findings complement those of studies in which plasma HIV-1 RNA levels were measured retrospectively. Retrospective measurement of HIV-1 RNA levels was done by O'Brien and colleagues [7] in patients who had symptomatic HIV-1 infection and in two studies [9, 10] of patients who had CD4+ lymphocyte counts less than 300 cells/mm³. In contrast, results from the studies of Mellors and coworkers [8] and Katzenstein and associates [11] suggested that the plasma HIV-1 RNA level at baseline was a more important prognostic indicator than CD4+ lymphocyte count at baseline. However, patients in these studies had higher mean CD4+ lymphocyte counts at baseline and were followed for longer periods of time than were patients in our study, which suggests that CD4+ lymphocyte counts may be less important than plasma HIV-1 RNA levels for monitoring patients at earlier stages of HIV-1 infection. More recently, further analyses by Mellors [28] of a larger cohort of patients with higher CD4+ lymphocyte counts and a study of patients given antiretroviral therapy who had higher CD4+ lymphocyte counts than those that we studied [29] indicated that prediction of progression is improved by measuring both plasma HIV-1 RNA levels and CD4+ lymphocyte counts at baseline.

Of note, the higher average plasma HIV-1 RNA levels among patients with AIDS in our study was explained by differences in CD4+ lymphocyte counts at baseline. This implies that a patient's current clinical status is primarily associated with the current extent of CD4+ lymphocyte depletion and is not independently associated with a greater current viral load.

Our findings are also consistent with data that have documented the importance of treatment-mediated suppression of plasma HIV-1 RNA levels as a determinant of clinical disease progression and decline in CD4+ lymphocyte count during antiretroviral treatment. By multivariate analysis, each 1 log₁₀ copies/mL (10-fold) reduction in plasma HIV-1 RNA levels within 8 weeks of treatment initiation in our study was associated with a 52% reduction in risk (CI, 6% increase to 79% reduction [P = 0.071]) for disease progression or death for the next 40 weeks (Table 2). Although this result did not reach statistical significance, it complements similar findings for patient groups that start treatment with higher mean CD4+ lymphocyte counts [7, 11]. Our results also extend those of two retrospective analyses of patients at a similar stage of HIV-related immunodeficiency [9, 10]; the results of these studies implied 52% (P = 0.04) and 65% (P = 0.07) reductions in risk for clinical progression for each 1 log₁₀ copies/mL (10-fold) decrease from baseline in plasma HIV-1 RNA levels at 4 to 8 weeks after treatment initiation. These earlier reports involved patients who had received antiretroviral therapy [9] and those who had received only minimal treatment before entering the study [10]. Our findings also agree with those of a recent prospective study for which preliminary results are available [30].

The consistent magnitude of this effect across studies and the statistical significance of the result in patients with less advanced disease [7, 11] and in patients with minimal previous antiretroviral therapy [10] suggest that an early treatment-mediated decrease in plasma HIV-1 RNA levels is associated with an improved prognosis during therapy. However, proof that a treatment-mediated decrease in plasma HIV-1 RNA levels is a surrogate end point for clinical outcome in treatment trials requires that the difference in the effects on clinical outcome of two treatments be explained by differences in effects on plasma HIV-1 RNA levels. This possibility has been investigated by O'Brien and colleagues [7], with promising results, but further studies must be done to confirm this in other clinical trials that show a significant difference in clinical outcome between study treatments.

The reduction in risk that was associated with each 1 log₁₀ copies/mL treatment-mediated suppression of plasma HIV-1 RNA levels at week 8 was similar to the difference in risk that was associated with each difference of 1 log₁₀ copies/mL in the plasma HIV-1 RNA level at baseline. This finding is added to other data [9-11] that suggest 1) that the clinical benefit of antiretroviral treatment is obtained rapidly and 2) that antiretroviral suppression of plasma HIV-1 RNA levels confers the full reduction in risk for clinical progression that is associated with differences in plasma HIV-1 RNA levels at baseline among different patients. Thus, maximizing and sustaining early reductions in plasma HIV-1 RNA levels seem to be critical if the clinical benefit of therapy is to be improved.

The change in CD4+ lymphocyte count at week 8 was not significantly associated with subsequent disease progression or death. This finding concurs with those of other studies that examined treatment-mediated changes within 4 to 8 weeks [9-11]. However, it contrasts with results of an evaluation of change in CD4+ lymphocyte count at 6 months that significantly predicted a more rapid disease progression [7]. This might be explained by the fact that treatment effects on CD4+ lymphocyte count are established more slowly than are effects on plasma HIV-1 RNA levels. If this is the case, then plasma HIV-1 RNA levels provide a more rapid measurement of prognosis after treatment is initiated or changed.

The infectious HIV-1 titer in mononuclear cells of peripheral blood was not a significant predictor in models that also included plasma HIV-1 RNA levels, which suggests that little additional information is obtained by measuring both factors. Measurement of the syncytium-inducing viral phenotype provided information in addition to plasma HIV-1 RNA level and CD4+ lymphocyte count in predicting change in CD4+ lymphocyte counts for 48 weeks (P = 0.050). However, it did not provide any significant additional information in predicting disease progression (P > 0.1).

The proportional hazard analyses of risk for disease progression indicate that monitoring CD4+ lymphocyte counts and plasma HIV-1 RNA levels improves the assessment of drug effectiveness in clinical trials. Our results also indicate that measuring both variables is useful in determining prognosis for 1 year for an individual patient. As shown in Table 3, it seemed that predictions of risk for disease progression were more accurate when both variables were measured rather than only one; this finding is consistent with the results of the multivariate model (Table 2). Even if both variables are measured, however, a patient cannot be perfectly categorized as a progressor or a nonprogressor during therapy. We need to further evaluate the possibility that additional improvements in risk prediction may be achieved by measuring other factors, such as viral resistance, host cell entry coreceptor genetic polymorphisms, or pharmacologic factors.

Despite the dynamic nature of HIV-1 replication [31-37], it has been hypothesized that the plasma HIV-1 RNA level is maintained at a relatively stable set point, or quasi-steady-state level, over brief periods [30]. Such stability was noted between the baseline measures (taken at the preentry visit and the entry visit), which were obtained a median of 6 days apart from patients who reported receiving stable therapy (Figure 1). The patients with higher levels of HIV-1 RNA in plasma had a set point for this period that was as stable as the set point for patients with lower levels. Our estimate of the intrapatient SD of plasma HIV-1 RNA levels was ± 0.19 log₁₀ copies/mL; this reflects both errors in laboratory measurement and biological variation between specimens from the same patient, including the possibility of diurnal variation. This value is not much larger than the SD that can be attributed to laboratory error for the assay that we used (± 0.16 log₁₀ copies/mL, estimated by repeated testing of several aliquots from the same specimen [McCreedy B. Personal communication]). Thus, errors in laboratory measurement seem to contribute more than biological variation to intrapatient variability of repeated measurements taken several days apart during stable therapy.

For the monitoring of individual patients in clinical practice, our study shows that changes in plasma HIV-1 RNA levels of more than 0.39 log₁₀ copies/mL (2.5-fold) occur in 10% of patients receiving stable therapy. This result is similar to that reported for patients with CD4+ lymphocyte counts between 200 and 500 cells/mm³[11]. Thus, a 2.5-fold or greater decrease in plasma HIV-1 RNA levels between measurements obtained before and after a change of treatment probably indicates a true change in the plasma HIV-1 RNA level that is not caused by biological fluctuations or errors in laboratory measurement. A 2.5-fold or greater reduction is therefore required to categorize an individual patient as responding to a new treatment. Conversely, a 2.5-fold increase in plasma HIV-1 RNA levels during stable therapy should be the minimum criterion for defining failure of virological treatment. These data support and refine a recent recommendation to use changes of at least 0.5 log₁₀ copies/mL (3.2-fold) for these criteria [38].

Many patients in our study who had less than a 2.5-fold reduction in plasma HIV-1 RNA levels from baseline to week 8 after starting study treatment experienced disease progression within 1 year (Table 4). Given our data (Table 2) and results from other studies [7-11, 29], which suggest that an early treatment-mediated decrease in plasma HIV-1 RNA levels is associated with an improved prognosis, clinicians might consider alternate treatments if a patient's plasma HIV-1 RNA levels decrease by less than 0.39 log₁₀ copies/mL (2.5-fold) within several weeks after a new antiretroviral regimen is initiated.

Appendix

Additional members of the ACTG 241 Protocol Virology Substudy Team are John Kappes, PhD, University of Alabama at Birmingham School of Medicine and Birmingham Veterans Affairs Medical Center, Birmingham, Alabama; Scott Hammer, MD, New England Deaconess Hospital and Harvard Medical School, Boston, Massachusetts; and Song-heng Liou, Harvard School of Public Health, Boston, Massachusetts.

The following persons participated in the ACTG 241 Protocol Virology Substudy: Roy Byington, Jacqueline Gillis, MS, Elizabeth Ladd, MA, and Rhonda Vincent (Massachusetts General Hospital, New England Deaconess Hospital, and Harvard Medical School, Boston, Massachusetts); Ronald Lollar, BS, MT(ASCP) (Rush Medical College, Chicago, Illinois); Joan Conway, BA, Julie Decker, BS, Byron Lambert, BS, and C. Brian Overbay, BA, BS (University of Alabama at Birmingham School of Medicine, Birmingham, Alabama); Diane Havlir, MD, Chris Fegan, RN, Janet Lathey, PhD, and Douglas Richman, MD (University of California, San Diego, San Diego, California); Christopher Horton, BA, and Marlene Saxer, MSc (University of California, San Francisco, San Francisco, California); Bev Putnam, RN, Robert Schooley, MD, David Shugarts, MS and Russ Young, MS (University of Colorado Health Sciences Center, Denver, Colorado); Fred Breakenridge, BS, Maria Fitterman, BS, Jorge Lopez, BS, and Cristina Woodson, BS (University of Miami School of Medicine, Miami, Florida); Henry Balfour, MD, and Kathy Brandt, BS (University of Minnesota, Minneapolis, Minnesota); Maureen Myers, PhD (Boehringer Ingelheim Pharmaceuticals, Inc., Ridgefield, Connecticut); Jeanette Harris, BS, and Bruce McCreedy, PhD (Laboratory Corporation of America, Inc., Research Triangle Park, North Carolina); Elaine Gebhardt, MA (Statistical and Data Management Center, Harvard School of Public Health, Boston, Massachusetts); Karen Kazial, RN (Statistical and Data Management Center, Frontier Science and Technology Research Foundation, Amherst, New York).

From Harvard School of Public Health, Massachusetts General Hospital, and Harvard Medical School, Boston, Massachusetts; London School of Hygiene and Tropical Medicine, London, England; University of Alabama at Birmingham School of Medicine and Birmingham Veterans Affairs Medical Center, Birmingham, Alabama; Rush Medical College, Chicago, Illinois; University of California, San Francisco, San Francisco, California; University of Minnesota Medical School, Minneapolis, Minnesota; University of Colorado Health Sciences Center and the Veterans Affairs Medical Center, Denver, Colorado; University of Miami School of Medicine, Miami, Florida; and University of California, San Diego, San Diego, California.

Dr. Johnson: University of Alabama at Birmingham School of Medicine and Birmingham Veterans Affairs Medical Center, 229 Tinsley Harrison Tower, 1900 University Boulevard, Birmingham, AL 35294-0006.

Drs. Hirsch and Basgoz: Infectious Disease Unit and AIDS Research Center, Massachusetts General Hospital, 55 Fruit Street, Boston, MA 02114-2698.

Dr. Bremer: Rush Medical College, 1653 West Congress Parkway, Chicago, IL 60612.

Dr. Elbeik: San Francisco General Hospital, Building 80, Ward 84, 995 Potrero Avenue, San Francisco, CA 94110.

Dr. Erice: University of Minnesota Health Center, Box 609, Harvard Street at East River Road, Minneapolis, MN 55455.

Dr. Kuritzkes: University of Colorado Health Sciences Center, 4200 East 9th Avenue, B-168, Denver, CO 80262.

Drs. Scott and Fischl: Department of Biochemistry and Molecular Biology, University of Miami School of Medicine, PO Box 016129, Miami, FL 33101.

Dr. Spector: University of California, San Diego, Clinical Sciences Building, 9500 Gilman Drive, La Jolla, CA 92093-0672.

Dr. D'Aquila: Infectious Disease Unit and AIDS Research Center, Massachusetts General Hospital, 149 13th Street, Charlestown, MA 02129.