Combination Therapy with Zidovudine and Didanosine Compared with Zidovudine Alone in HIV-1 Infection

Zidovudine (AZT, Retrovir) delays progression of human immunodeficiency virus type 1 (HIV-1) infection and prolongs survival in persons with advanced HIV-1 disease [1-5]. However, with prolonged therapy, clinical disease progresses, CD4+ cell counts decrease, and variants of HIV-1 occur that have decreased in vitro susceptibility to zidovudine [6-12]. Didanosine (Videx), another nucleoside analogue that inhibits the reverse transcriptase of HIV-1, increases CD4+ cell counts and delays progression of HIV-1 disease among patients who previously received zidovudine therapy [13-16]. The combination of zidovudine and didanosine has additive to synergistic inhibitory activity against HIV-1 in vitro [17, 18] and is being used clinically, because both agents are licensed and available. We did a clinical trial to characterize the safety and efficacy of a range of doses using combination zidovudine and didanosine therapy compared with zidovudine therapy alone. We found that this combination therapy is well tolerated and is associated with enhanced in vivo activity as shown by higher and more prolonged increases of CD4+ cell counts, with more frequent decreases in plasma HIV-1 RNA titers, and with more stable hematologic measurements than zidovudine therapy alone. Table 1. Pretreatment Characteristics of Patients in the Six Treatment Groups Methods Patients Between February 1990 and August 1991, 69 patients were enrolled in the study: 30 at University of Washington, 14 at University of Miami, 14 at New England Medical Center and Tufts University, and 11 at University of California, San Francisco. Patients had HIV-1 infection, CD4+ cell counts fewer than 400/mm3, fewer than 121 days of previous zidovudine therapy (Retrovir; Burroughs Wellcome Company, Research Triangle Park, North Carolina), and no previous didanosine therapy (Videx; Bristol Laboratories, Princeton, New Jersey). Patients also had granulocyte counts of 1200 cells/mm3 or more, hemoglobin levels greater than 90 g/L, platelet counts greater than 90 000/mm3, creatinine levels less than 1.5 times the upper limit of normal, and aspartate aminotransferase levels less than 5 times the upper limit of normal. Patients were excluded if they had visceral or progressive Kaposi sarcoma; more than four stools per day for 4 weeks; opportunistic infections requiring maintenance therapy; a history of pancreatitis, seizures, peripheral neuropathy, or zidovudine or didanosine intolerance; were pregnant or nursing; were taking experimental medications; or required chronic acyclovir therapy. Study Design The study was an open-label, partially randomized study of five different combination regimens of zidovudine and didanosine and one dosage regimen of zidovudine alone (Figure 1). Zidovudine was given as capsules three times per day. Didanosine was given twice daily as sachets containing a citrate/phosphate buffer with a neutralizing capacity of 30 to 40 mEq/dose. Because of initial concerns about the safety of this combination, enrollment was required in the lower dose groups (groups 1, 2, and 3) before enrollment in groups 4 and 5. In March 1991, after enrollment was completed in groups 1 to 5, the protocol was modified to enroll consecutively identified eligible patients in the zidovudine-alone regimen (group 6), to permit a comparison with standard therapy. Because standard zidovudine therapy in the United States at the time this trial was done was 500 to 600 mg daily, the groups that included zidovudine at 600 mg daily (groups 3, 5, and 6) were planned to have larger enrollment than the other groups. Treatment duration was 24 weeks. Patients were replaced if they discontinued treatment permanently before week six for reasons other than toxicity. Figure 1. Outline of the study design. Patient Evaluation The study protocol was approved by institutional review boards of participating institutions, and patients gave written informed consent. Patients had a standardized clinical and laboratory evaluation at enrollment and had weekly follow-up visits for 4 weeks and then every 2 weeks. Symptoms and signs were assessed at each visit and were graded as absent or within normal limits, or as mild, moderate, severe, or life-threatening severity. Patients at the University of Washington in groups 3 (n = 9) and 5 (n = 6) had blood samples collected at 0, 0.5, 1, 1.5, 2, 2.5, 3, 4, 6, and 8 hours for pharmacokinetic assays, after a single 200-mg zidovudine dose, 2 days before starting study drugs, and after a single 167- or 250-mg didanosine dose on the day before starting combination therapy. In addition, 26 patients in groups 3 and 5 had plasma collected on a similar schedule during weeks 2 (n = 26) and 12 (n = 23). Doses of study drugs were modified if patients had moderate peripheral neuropathy or certain severe toxicities. Study drugs were permanently discontinued for severe or life-threatening toxicities. Laboratory Measurements CD4+ and CD8+ cells from peripheral blood were enumerated using monoclonal antibodies and flow cytometry [19]. Sera from each patient, frozen at 30C, were assayed simultaneously for HIV-1 p24 antigen by an enzyme-linked immunosorbent assay (Abbott Laboratories, North Chicago, Illinois). A positive result for HIV-1 was defined as 12 pg/mL using an HIV-1 p24 antigen reference standard supplied by the AIDS Clinical Trials Group Virology Reference Laboratory. Sera from patients enrolled at the University of Washington were also assayed for immune-complex dissociated-HIV p24 antigen (Abbott Laboratories). Plasma from patients enrolled at the University of Washington, stored at 70C, was assayed for HIV-1 RNA by a semiquantitative polymerase chain reaction technique, as previously described [20]. All assays were done in duplicate on coded samples, and the semiquantitative assessment of the polymerase chain reaction-signal strength was graded using scanning densitometry without knowledge of any clinical data or treatment group. Zidovudine concentrations in plasma were determined by radioimmunoassay [21]. Plasma didanosine concentrations were analyzed by high-performance liquid chromatography [22]. Data Analysis Comparisons among groups were made using chi-square analysis or the Fisher exact test for discrete variables and using the Wilcoxon Mann-Whitney test and Kruskal-Wallis test among three or more groups for continuous variables. All P values were two-tailed. To evaluate the effect of therapy for sequentially monitored variables, area-under-the-curve (AUC) analyses were done based on absolute change from pretreatment or time-averaged absolute change from pretreatment values. The baseline CD4+ cell count was an average of two pretreatment counts for each patient. The AUC for CD4+ cell counts was calculated by multiplying the average CD4+ cell count between two successive time points by the time elapsed between those time points, summing these areas for the study period, and dividing the result by the duration of the study period. This AUC was strongly related to the pretreatment CD4+ cell count. Analysis of covariance, adjusting for the initial CD4+ cell count, was used to examine the relation between the treatment group and the AUC. Residuals from the model were analyzed, and no pattern was discernible, suggesting the model was appropriate. Area-under-the-curve analysis, based on absolute CD4+ cell changes from pretreatment values, was used to compare these areas across groups using nonparametric statistics. The proportion of patients with an increase in absolute CD4+ cells of 50 and 100 cells/mm3, on two consecutive measurements above the pretreatment value, was evaluated using contingency tables. The duration of response was defined as the time to two consecutive counts at or below baseline and was analyzed using Kaplan-Meier survival techniques. Pharmacokinetic parameter estimates were calculated by standard methods and included the peak concentration (Cmax), the time to peak concentration (Tmax), the elimination half-life (t1/2), the AUC, and the apparent oral clearance. The evaluation of a pharmacokinetic drug interaction between zidovudine and didanosine was done by paired t-test analysis of parameter estimates obtained for those patients in groups 3 and 5 who received single doses of zidovudine alone (day 2) and didanosine alone (day 1) followed by coadministration of both compounds with pharmacokinetic evaluations in weeks 2 and 12. Parameter estimates were considered statistically different for P values less than 0.05. Results Study Patients Sixty-seven (97%) patients were men; 61 (88%) were white, 2 (3%) were black, and 6 (9%) were of other races. The mean age (SD) was 34 6 years. Fifty-four (78%) were homosexual or bisexual men, 8 (12%) were homosexual or bisexual men who also used injection drugs, 2 (3%) were injection drug users, 3 (4%) had heterosexually acquired HIV-1, 1 (1%) had hemophilia, and 1 (1%) had unknown risk behavior for HIV. Thirty-eight (55%) patients were asymptomatic, 22 (32%) had constitutional symptoms, and 9 (13%) had the acquired immunodeficiency syndrome (AIDS). The median pretreatment CD4+ cell count was 259 cells/mm3, and the median Karnofsky performance score was 100. No statistical differences were seen in demographic, clinical, or laboratory characteristics among the six dosing groups, except for the expected difference in frequency of detectable HIV-1 p24 antigen (Table 1). The median HIV-1 p24 antigen level for patients in group 1 was 140 pg/mL (range, 17 to 282 pg/mL). Thirty-one patients had taken previous zidovudine therapy; the mean duration of therapy (SD) was 65 days ( 36 days). Fifty-eight (84%) patients completed 24 weeks of study treatment and 11 (16%) discontinued study treatment, 5 before week 6. Pretreatment characteristics of patients who discontinued treatment were similar to those who completed therapy. The reasons for study medication discontinuation were toxicity (3 patients), administrative reasons (4