Parallel evolution of drug resistance in HIV: failure of nonsynonymous/synonymous substitution rate ratio to detect selection.

Parallel or convergent evolution at the molecular level has been difficult to demonstrate especially when rigorous statistical criteria are applied. We present sequence data from the protease gene from eight patients infected with the human immunodeficiency virus (HIV-1). These patients have been on multiple drug therapies for at least 2 years. We present sequence data from two timepoints: time zero--the initiation of drug therapy--and a subsequent timepoint between 59 and 104 weeks after the initiation of drug therapy. In addition to the sequence data, we present viral load data from both initial and final timepoints. Our phylogenetic analyses indicate significant evolution of virus from initial to final time points, even in three of eight patients who show low viral loads. Of the five patients who escaped drug therapy, identical amino acid replacements were seen in all five patients at two different codon positions, an indication of parallel evolution. We also measured genetic diversity for these patients and found no correlation between genetic diversity and viral load. Finally, we calculated the nonsynonymous and synonymous substitution rates and showed that the ratio of nonsynonymous to synonymous substitution compared to the value of one may be a poor indicator of natural selection.

Drug resistance during indinavir therapy is caused by mutations in the protease gene and in its Gag substrate cleavage sites.

Two different responses to the therapy were observed in a group of patients receiving the protease inhibitor indinavir. In one, suppression of virus replication occurred and has persisted for 90 weeks (bDNA, < 500 human immunodeficiency virus type 1 [HIV-1] RNA copies/ml). In the second group, a rebound in virus levels in plasma followed the initial sharp decline observed at the start of therapy. This was associated with the emergence of drug-resistant variants. Sequence analysis of the protease gene during the course of therapy revealed that in this second group there was a sequential acquisition of protease mutations at amino acids 46, 82, 54, 71, 89, and 90. In the six patients in this group, there was also an identical mutation in the gag p7/p1 gag protease cleavage site. In three of the patients, this change was seen as early as 6 to 10 weeks after the start of therapy. In one patient, a second mutation occurred at the gag p1/p6 cleavage site, but it appeared 18 weeks after the time of appearance of the p7/p1 mutation. Recombinant HIV-1 variants containing two or three mutations in the protease gene were constructed either with mutations at the p7/p1 cleavage site or with wild-type (WT) gag sequences. When recombinant HIV-1-containing protease mutations at 46 and 82 was grown in MT2 cells, there was a 68% reduction in its rate of replication compared to the WT virus. Introduction of an additional mutation at the gag p7/p1 protease cleavage site compensated for the partially defective protease gene. Similarly, rates of replication of viruses with mutations M46L/I, I54V, and V82A in protease were enhanced both in the presence and in the absence of Indinavir when combined with mutations in the gag p7/p1 and the gag p1/p6 cleavage sites. Optimal rates of virus replication require protease cleavage of precursor polyproteins. A mutation in the cleavage site that enhanced the availability of a protein that was rate limiting for virus maturation would confer on that virus a significant growth advantage and may explain the uniform emergence of viruses with alterations at the p7/p1 cleavage site. This is the first report of the emergence of mutations in the gag p7/p1 protease cleavage sites in patients receiving protease therapy and identifies this change as an important determinant of HIV-1 resistance to protease inhibitors in patient populations.

Emergence of protease inhibitor resistance mutations in human immunodeficiency virus type 1 isolates from patients and rapid screening procedure for their detection.

Patient human immunodeficiency virus type 1 (HIV-1) isolates that are resistant to protease inhibitors may contain amino acid substitutions L10I/V, M46L/I, G-48V, L63P, V82A/F/T, I84V, and L90M in the protease gene. Substitutions at positions 82 and/or 90 occur in variants that display high levels of resistance to certain protease inhibitors. Nucleotide substitutions at these two sites also lead to the loss of two HindII restriction enzyme digestion sites, and these changes make possible a rapid procedure for the detection of drug-resistant variants in patients on protease inhibitor therapy. This procedure was used to detect the emergence of mutated viruses at various times after the initiation of therapy with the HIV-1 protease inhibitor indinavir. The method includes viral RNA isolation from plasma and reverse transcription PCR amplification of the protease gene with fluorescence-tagged primers. The PCR product is digested with HindII, the cleavage products are separated on a urea-acrylamide gel in a DNA sequencer, and the extent of cleavage is automatically analyzed with commercially available software. In viruses from 34 blood samples from four patients, mutations leading to an amino acid change at residue 82 appeared as early as 6 weeks after the start of therapy and persisted throughout the course of the study period (48 weeks). Mutations leading to double substitutions at residues 82 and 90 were seen at a lower frequency and appeared later than the change at position 82. The changes detected by restriction enzyme cleavage were confirmed by DNA sequencing of the cloned protease genes by reverse transcription PCR amplification of viral RNA from isolates in plasma. In addition to the changes at positions 82 and 90, we have identified M46L/I, G48V, and I54V substitutions in isolates derived from indinavir-treated patients. HindII analysis of uncloned, PCR-amplified DNA offers a rapid screening procedure for the detection of virus isolates containing mutations at amino acid residues 82 and 90 in the HIV-1 protease gene. By using other restriction enzymes, the same method can be used to detect additional protease drug-resistant variants and is generally applicable for the detection of mutations.

Combination therapy with didanosine and interferon-alpha in human immunodeficiency virus-infected patients: results of a phase I/II trial.

A nonrandomized trial was undertaken to evaluate the combination of didanosine and interferon-alpha (INF-alpha) in human immunodeficiency virus (HIV)-infected patients. Thirty-six volunteers with >200 x 10(6) CD4 cells/L received didanosine (one 100-, 250-, or 375-mg sachet twice daily) for at least 6 weeks, following which IFN-alpha (1, 5, 10, or 15 MU/day) was begun. Didanosine (one 375-mg sachet twice daily) was substituted for zidovudine in 14 additional patients who had received IFN-alpha and zidovudine for 7-45 months. Thirty-five patients completed the 34-week study. Clinical or chemical pancreatitis was the most common (6 patients) dose-limiting toxicity. CD4 cell counts increased with didanosine but declined following the addition of IFN-alpha; CD4 cell percents tended to increase and remain elevated. Thus, combination therapy with didanosine and IFN-alpha can be safely administered to patients with HIV infection. The clinical benefit of this combination therapy will require further evaluation.

Increases in CD4 T lymphocytes with intermittent courses of interleukin-2 in patients with human immunodeficiency virus infection. A preliminary study.

BACKGROUND: Interleukin-2 is an important regulatory cytokine of the immune system, with potent effects on T cells, B cells, and natural killer cells. In vitro, interleukin-2 can induce the proliferation and differentiation of peripheral-blood mononuclear cells from patients infected with the human immunodeficiency virus (HIV). METHODS: We treated 25 HIV-infected patients with interleukin-2 administered as a continuous infusion at a dosage of 6 to 18 million IU per day for 5 days every 8 weeks during a period of 7 to 25 months. All patients also received at least one approved antiviral agent. Immunologic and virologic variables were monitored monthly. RESULTS: In 6 of 10 patients with base-line CD4 counts higher than 200 per cubic millimeter, interleukin-2 therapy was associated with at least a 50 percent increase in the number of CD4 cells. Changes ranged from -81 to +2211 cells per cubic millimeter. Interleukin-2 therapy resulted in a decline in the percentage of CD8 lymphocytes expressing HLA-DR and an increase in the percentage of CD4 lymphocytes that were positive for the p55 chain of the interleukin-2 receptor. Four patients had a transient but consistent increase in the plasma HIV RNA level at the end of each infusion. In the remaining 15 patients, who had CD4 counts of 200 or fewer cells per cubic millimeter, interleukin-2 therapy was associated with increased viral activation, few immunologic improvements, and substantial toxic effects. CONCLUSIONS: Intermittent courses of interleukin-2 can improve some of the immunologic abnormalities associated with HIV infection in patients with more than 200 CD4 cells per cubic millimeter.

Application of branched DNA signal amplification to monitor human immunodeficiency virus type 1 burden in human plasma.

A branched DNA (bDNA)-based quantitation of plasma human immunodeficiency virus type 1 (HIV-1) RNA was used to monitor the virologic status of 102 patients (29-906 CD4 cells/mm3) enrolled in clinical trials of antiretroviral and immune-based therapies. Virion-associated RNA was measurable in plasma of 74% of patients tested (10,000-10,000,000 RNA equivalents/mL). Virus levels measured by the bDNA assay exceeded titers obtained by quantitative plasma culture and were inversely correlated (r = -.378; P < .05) with total CD4 cell counts. The assay was used to demonstrate a significant decline (mean, 5-fold; range, 0- to 30-fold), relative to pretreatment, in virus load after beginning antiviral therapy and a transient increase (mean, 15-fold; range, 2- to 50-fold) after treatment with interleukin-2. The decrease in RNA was more dramatic than changes in serum p24 antigen. The bDNA assay yields reproducible results, is relatively easy, and should be useful in measuring HIV-1 RNA in patients in clinical trials.

An internally controlled virion PCR for the measurement of HIV-1 RNA in plasma.

We have developed an assay to measure the HIV-1 RNA in patients' plasma or sera using an infectious mutant virus as an internal control. The mutant virus VX-46 has a 25-bp insert in a conserved region between the primer-binding and major splice donor sites. To utilize this virus as an internal control, different dilutions of this virus were added to aliquots of plasma sample to be measured, RNA was isolated and reverse-transcribed to cDNA. PCR was performed with primers selected to include the sequences on either side of the insert contained in the externally added virus. The DNA product from the control virus is 25 bp longer than that from the virus present in plasma. The amount of viral RNA present in a plasma sample is calculated after the PCR-amplified products are separated by gel electrophoresis. Unlike other quantitative PCR assays, this internally controlled virion PCR (ICVPCR) assay eliminates errors introduced by variable recovery during the RNA purification step, therefore, enhancing the accuracy of the assay.

Persistence of four related human immunodeficiency virus subtypes during the course of zidovudine therapy: relationship between virion RNA and proviral DNA.

Human immunodeficiency virus (HIV) virion RNA and proviral DNA sequences have been examined over a 1-year period in an HIV-seropositive patient, commencing with the start of zidovudine treatment. By characterizing the variable V3 and V4 env domains, four related but structurally discrete genotypes could be identified prior to the start of therapy and during the subsequent 60-week period of therapy. Each of the four subtypes showed a unique pattern in the preservation of glycosylation sites. A comparison of the V3 amino acid sequences in peripheral blood mononuclear cell proviral DNA and plasma virion RNA at 0, 24, 36, and 60 weeks demonstrated that proviral DNA did not serve as a predictor of the structure of virion RNA. HIV virion RNA subtype 3 was the most prevalent virion RNA subtype at three of the four periods studied, yet no corresponding proviral DNA was detected. Other virion subtypes have been observed, but only on a transient basis. The present data are consistent with a model of HIV infection in which related but different HIV substrains coexist and evolve independently within an individual. Characterization of virion RNA may be required to identify the unique properties of the virus involved in disease progression; characterization of proviral DNA will not yield this information.

Regulation of HIV-1 envelope protein synthesis by Tat and Rev in 293 cells.

A plasmid expression vector (B2) with the HIV-1 envelope sequence downstream of the adenovirus type 5 early region 3 promoter could direct the synthesis of envelope protein in the absence of Rev when transfected into 293 cells. We investigated this further using pNL4.3 delta TR, and HIV-1 mutant which lacks the first exon of Tat and Rev and pNL4.3 delta R, an HIV-1 mutant with a premature termination codon in the second coding exon of Rev. In cells transfected with pNL4.3 delta TR and a Tat-expressing vector or with pNL4.3 delta R alone, analysis of RNA revealed the accumulation of cytoplasmic Env mRNA in the absence of Rev. However, envelope protein synthesis was observed in the absence of Rev only in cells transfected with pNL4.3 delta TR and a Tat-expressing vector, not in cells transfected with pNL4.3 delta R. The Env mRNAs synthesized from pNL4.3 delta R can have 536 to 548 nucleotides of 5' non coding sequence, whereas the Env mRNA from pNL4.3 delta TR will have a shortened noncoding sequence of 321 nucleotides. These results indicate that the mRNA sequences 5' to the Env protein initiation codon have a role in Env expression.

Detection of HIV in bone allografts prepared from AIDS autopsy tissue.

Processed bone from an AIDS patient was tested for the presence of the human immunodeficiency virus type 1 (HIV-1). The preliminary procedures used to process bone allografts included removal of adventitious material and two cycles of freeze thawing. Although infectious virus was readily observed in plasma and bone marrow cells taken at autopsy, no infectious virus was detected in processed bone fragments. However, by using the polymerase chain reaction procedure, the presence of proviral HIV DNA could be demonstrated in processed bone allografts from this donor. Whereas the best safeguard against transmission of HIV by allografts is rigorous criteria for the exclusion of seropositive individuals as donors, proper procedures for processing bone allografts can further reduce the possibility of HIV transmission by bone allografts in cases where tissue from an infected donor is collected and processed.

Clinical utility of an enhanced human immunodeficiency virus type 1 p24 antigen capture assay.

The presence of p24 core antigen in the serum of individuals with human acquired immunodeficiency syndrome has been used as one of the important prognostic markers of HIV-1 infection and also as an end point in evaluating antiviral drugs and vaccines. Unfortunately the majority of p24 antigen present in serum exists as an antigen-antibody complex and is not detected with the commercial kits currently available to measure p24 antigen. In this study, we report a simple procedure utilizing treatment of serum samples with glycine buffer (pH 1.85) to dissociate antigen-antibody complexes prior to assaying for p24 antigen. A 300% increase in the number of p24-reactive samples and a 3- to 12-fold increase in the quantity of antigen detected were observed when samples were pretreated with 1.5 M glycine buffer (pH 1.85) for 1 hr. Glycine treatment of samples did not result in non-specific positive tests and samples previously shown to be reactive remained positive. In reconstruction experiments the release of antigen was found to be inversely proportional to the amount of p24 antibody present in the serum. The percentage of HIV-1-infected patients positive for p24 antigen was clearly a function of CD4 count. Forty-nine percent of patients with more than 500 CD4 cells and 100% of patients with less than 200 CD4 were p24 positive.(ABSTRACT TRUNCATED AT 250 WORDS)

Prevention of the spread of HIV-1 infection with nonnucleoside reverse transcriptase inhibitors.

Certain bisheteroarylpiperazines (BHAPs) directly inhibit the replication of human immunodeficiency virus type 1 (HIV-1) and block the spread of infection to susceptible populations of cells. At a 1 microM concentration three analogs, U-87201, U-88204, and U-89674, inhibited the replication of HIV-1 in MT-2 cells by 83, 100, and 93%, respectively. At the same concentration, U-88204 completely inhibited replication of primary HIV-1 isolates in peripheral blood mononuclear cells. Replication of 3'-azido-2',3'-dideoxythymidine (AZT)-resistant strains of HIV-1 was also inhibited by U-88204. When MT-2 cells that were lytically infected with HIV-1 were mixed with uninfected MT-2 cells, U-88204 provided complete protection to the uninfected cells. Integrated proviral DNA sequences were not detected by the polymerase chain reaction technique in this culture after 15 days in the presence of drug. The resultant healthy cell culture was subsequently maintained without drug with no evidence of latent proviral DNA. Serial passage of a laboratory strain and a primary isolate of HIV-1 in cell culture in the presence of increasing concentrations of U-88204 yielded virus populations which were at least 100-fold resistant to the drug. These resistant viruses also showed cross-resistance to the pyridinone class of nonnucleoside inhibitors but were sensitive to AZT. Analysis of the nucleotide sequence of resistant viruses revealed mutations at conserved regions of the reverse transcriptase (RT) gene. The results presented here suggest the therapeutic potential of U-88204 in the combination therapy for HIV-1 infection.

Envelope-specific antibodies in the saliva of individuals vaccinated with recombinant HIV-1 gp160.

HIV-1-specific antibodies have been detected in the saliva of seropositive individuals and may play a role in preventing oral transmission of the virus. We have analyzed saliva samples obtained from HIV-1-seronegative individuals who were immunized with various dosages of a recombinant HIV-1 envelope glycoprotein (gp160) vaccine for the presence of antibodies to HIV-1. Antibodies specific for envelope glycoproteins were detected in saliva from all of the volunteers, with those vaccinated with the higher doses of 640 and 1,280 micrograms showing the strongest responses. Peak salivary antibody titers were obtained 4-14 weeks after vaccination; they then gradually dropped in parallel with serum antibody titers. These envelope-specific antibodies were detected in whole saliva and in submandibular saliva but not in parotid saliva, suggesting that the source of antibodies in saliva is from serum transudation. The class of reactive antibodies was found to be IgG. The HIV-1-specific antibodies in the saliva of vaccinated individuals may offer local protection against HIV-1 infection.

Isolation of HIV-1 from plasma of infected individuals: an analysis of experimental conditions affecting successful virus propagation.

Experimental conditions affecting the successful propagation of HIV-1 from the plasma of seropositive individuals were examined. It was determined that whole blood samples collected with lithium heparin as the anticoagulant, immediate plasma separation, and immediate culturing were best suited for obtaining viable virus from plasma. Virus was isolated by infecting fresh phytohemagglutinin-stimulated normal donor peripheral blood mononuclear cells (PBMCs) with plasma followed by weekly cocultivation with new target cells. The plasma virus isolation rate was the greatest and HIV-1 titers were the highest for those individuals with less than 200 CD4+ cells/mm3 and decreased as the level of CD4+ cells approached normal values. We were able to obtain positive cultures from 29.5% of those patients with CD4+ counts greater than 500/mm3. HIV-1 titers in plasma also correlated with high serum p24 antigen levels when serum was treated with glycine to dissociate antigen-antibody complexes.

Increasing viral burden in CD4+ T cells from patients with human immunodeficiency virus (HIV) infection reflects rapidly progressive immunosuppression and clinical disease.

OBJECTIVE: To determine over time the relation between viral burden and immunologic decline in patients with asymptomatic human immunodeficiency virus (HIV) infection. DESIGN: Blind analysis of cell samples from matched cohorts for HIV proviral DNA by polymerase chain reaction, retrospective analysis of clinical data on patients, and prospective follow-up of patients seropositive for the human immunodeficiency virus type 1 (HIV-1). SETTING: National research clinic and academic medical centers. PATIENTS: Cohort 1 included 12 healthy HIV-1-seropositive patients (average follow-up, 14 months): Six patients had stable disease and 6 developed rapidly progressive disease. Cohort 2 included 15 healthy HIV-1-seropositive patients from the Multi-center AIDS Cohort Study (average follow-up, 32 months): Eight patients had stable disease and 7 developed rapidly progressive disease. LABORATORY STUDIES: Quantitative polymerase chain reaction was done to determine the HIV-1 viral burden in sort-purified CD4+ T cells obtained from patients at various timepoints. MEASUREMENTS AND MAIN RESULTS: In patients who remained asymptomatic, frequencies of HIV-infected CD4+ T cells were low (less than 1/10,000 to 1/1000) at study entry and increased only minimally (none higher than 1/1000). In contrast, among patients who developed HIV-related symptoms including the acquired immunodeficiency syndrome (AIDS) despite having similar CD4 counts, frequencies of HIV-infected CD4+ T cells were higher at entry (greater than 1/1000) and increased substantially (greater than 1/100) in most within 3 months of developing progressive disease. This increase in HIV burden coincided with a significant decline over time in the percent of T4 cells (31% to 16%), whereas the percent of T4 cells was unchanged in persons who remained asymptomatic (33% to 34%). CONCLUSIONS: Increasing viral burden in peripheral blood CD4+ T-cells is directly associated with a progressive decline in CD4+ T cells and deteriorating clinical course in HIV-infected patients.

Suppression of HIV-1 reverse transcriptase activity by mycoplasma contamination of cell cultures.

The detection of HIV-1 in human peripheral blood lymphocytes is routinely carried out by cocultivation of test cells with normal peripheral blood mononuclear cells (PBMC). The presence of virus is evidenced by cytologic observation of syncytia or by detecting viral reverse transcriptase (RT) and/or p24 antigen in the culture supernatant fluid. Syncytia formation is almost always associated with the presence of virus as measured by RT, although many RT-positive cultures do not form syncytia. As part of a large screening program, we identified three cultures that showed syncytia but were RT negative. The basis for these discrepant observations was contamination of cultures with mycoplasma that interfered with the RT assay and thereby obscured virus detection. Treatment of cultures with BM-cycline removed mycoplasma contamination and restored RT activity. The present findings indicate the need for caution in the interpretation of negative RT results during HIV-1 isolation and especially in cultures that show evidence of syncytia formation.

Integrated proviral human immunodeficiency virus type 1 is present in CD4+ peripheral blood lymphocytes in healthy seropositive individuals.

Evidence of a latent human immunodeficiency virus type 1 (HIV-1) infection in healthy, seropositive individuals who do not have viral antigens in their sera and from whom virions cannot be rescued in cocultivation experiments was examined. Proviral DNA was detected by amplification by the polymerase chain reaction procedure. In each of 10 seropositive individuals, the presence of HIV-1 proviral sequences was demonstrated in their peripheral blood mononuclear cells. By using fluorescence-activated cell sorting, we obtained highly enriched subpopulations of peripheral blood mononuclear cells and found that the CD4+ T-cell subset is the cell subset that consistently harbors the HIV-1 proviral sequences. The number of HIV-1-infected CD4+ T cells was variable among the 10 healthy individuals, ranging from 1 in 100 to 1 in 40,000. While in vitro infection of CD4+ T cells causes down regulation and eventual loss of CD4 surface molecules, this is not true in vivo where it is only the CD4+ population that harbors the virus. This disparity may reflect differences between a latent infection in vivo with the lytic response of cells infected in vitro.

Detection of antibodies to human immunodeficiency virus type 1 in whole blood and saliva by using a passive hemagglutination test.

A passive hemagglutination test (PHA) for detecting human immunodeficiency virus type 1 antibodies in serum samples by using envelope glycoprotein (gp160)-coupled sheep erythrocytes was described earlier (M.B. Vasudevachari, K. U. Uffelman, T.C. Mast, R.L. Dewar, V. Natarajan, H.C. Lane, and N.P. Salzman, J. Clin. Microbiol. 27:179-181, 1989). In the study reported here, the applicability of the PHA test to the detection of antibodies in whole-blood and saliva samples has been investigated. We observed a 100% correlation between PHA and commercial enzyme-linked immunosorbent assay in 101 whole-blood samples and 98% correlation between PHA and reactivity to envelope proteins in Western blots (immunoblots) of 53 saliva samples. Furthermore, salivary antibodies could be detected in 19 of the 22 seropositive individuals. As in serum, antibodies to envelope proteins were widely prevalent in all the Western blot-reactive saliva samples.

The reservoir for HIV-1 in human peripheral blood is a T cell that maintains expression of CD4.

Human immunodeficiency virus type 1 (HIV-1) selectively infects cells expressing the CD4 molecule, resulting in substantial quantitative and qualitative defects in CD4+ T lymphocyte function in patients with acquired immunodeficiency syndrome (AIDS). However, only a very small number of cells in the peripheral blood of HIV-1-infected individuals are expressing virus at any given time. Previous studies have demonstrated that in vitro infection of CD4+ T cells with HIV-1 results in downregulation of CD4 expression such that CD4 protein is no longer detectable on the surface of the infected cells. In the present study, highly purified subpopulations of peripheral blood mononuclear cells (PBMCs) from AIDS patients were obtained and purified by fluorescence-automated cell sorting. They were examined with the methodologies of virus isolation by limiting dilution analysis, in situ hybridization, immunofluorescence, and gene amplification. Within PBMCs, HIV-1 was expressed in vivo predominantly in the T cell subpopulation which, in contrast to the in vitro observations, continued to express CD4. The precursor frequency of these HIV-1-expressing cells was about 1/1000 CD4+ T cells. The CD4+ T cell population contained HIV-1 DNA in all HIV-1-infected individuals studied and the frequency in AIDS patients was at least 1/100 cells. This high level of infection may be the primary cause for the progressive decline in number and function of CD4+ T cells in patients with AIDS.

Zidovudine in patients with human immunodeficiency virus (HIV) infection and Kaposi sarcoma. A phase II randomized, placebo-controlled trial.

STUDY OBJECTIVE: To evaluate the toxicity, effects on immune function, antitumor effects, antiretroviral effects, and pharmacokinetics of zidovudine therapy in patients with early human immunodeficiency virus (HIV) infection and Kaposi sarcoma. DESIGN: Randomized, double-blind, placebo-controlled trial. SETTING: National Institutes of Health, a referral-based research institution (single site). PATIENTS: Physician-referred volunteer patients with HIV infection, Kaposi sarcoma, CD4+ lymphocyte counts greater than 0.2 x 10(9)/L, and no systemic symptoms or history of opportunistic infection. Of 41 patients enrolled, 4 had not met all entry criteria and were therefore not evaluable. INTERVENTIONS: Patients were randomized to one of four treatment groups for an initial 12-week treatment period: oral placebo (9 patients); zidovudine, 250 mg orally every 4 hours (9 patients); zidovudine, 0.5 mg/kg body weight intravenously every 4 hours (9 patients); and zidovudine, 2.5 mg/kg intravenously every 4 hours (10 patients). After at least 12 weeks of therapy at their assigned dose, patients were treated with oral zidovudine, generally 250 mg every 4 hours, with a mean 42-week follow-up. MEASUREMENTS AND MAIN RESULTS: Anemia and granulocytopenia were the major toxicities. Significant increases in platelet counts and declines in serum HIV antigen and IgG and IgM levels occurred in treated patients. Treated patients were more likely than those on placebo to clear HIV from the cerebrospinal fluid. There were no differences in tumor progression or CD4+ or CD8+ lymphocyte counts among the groups. CONCLUSIONS: Zidovudine was well tolerated and had antiretroviral activity in patients with early HIV infection and Kaposi sarcoma but it had no significant effect on the extent of Kaposi sarcoma or on immune function.