Zalcitabine Compared with Zidovudine in Patients with Advanced HIV-1 Infection Who Received Previous Zidovudine Therapy

Methods

Patients

The study sample consisted of patients who had tolerated 48 weeks or more of zidovudine therapy (total daily dose 500 mg) and who had a first episode of Pneumocystis carinii pneumonia or advanced AIDS-related complex at the time of initiation of zidovudine therapy. All patients had to be receiving and tolerating zidovudine at the time of enrollment in the study. Advanced AIDS-related complex was defined by the presence of oral candidiasis, oral hairy leukoplakia, herpes varicella infection, unintentional weight loss exceeding 10% of body weight or 4.5 kg (10 lb), unexplained diarrhea (defined as two or more liquid stools per day persisting for 30 days or more) or unexplained fever (defined as a temperature greater than 38.5 C occurring for more than 14 consecutive days or for more than 15 days in a 30-day period), and a CD4-lymphocyte count of less than 200 cells/mm³. The criteria for eligibility also included a hemoglobin concentration of 95 g/L or more, a total granulocyte count of 0.750 10⁹/L or more, a platelet count of 75 10⁹/L or more, serum transaminase levels no higher than five times the upper limit of normal, a Karnofsky performance score of 60 or more, and seropositivity for HIV antibody as determined by any licensed enzyme-linked immunosorbent assay. At least 20% of the patients were also required to have a serum HIV p24 antigen level of 70 pg/mL or more. Patients were also excluded from the study for the following reasons: interruption of previous zidovudine therapy for more than 30 consecutive days or a total of 90 days, development of a serious toxicity during zidovudine therapy, previous zalcitabine therapy, the presence of visceral or symptomatic Kaposi sarcoma requiring therapy, the presence of peripheral neuropathy, or pregnancy. All patients received inhaled pentamidine isethionate, 300 mg every 4 weeks, for the prevention of P. carinii pneumonia. Maintenance therapy for other AIDS-related opportunistic infections was allowed. The use of other antiretroviral drugs, biologic-response modifiers, systemic corticosteroids, drugs that could cause peripheral neuropathy other than isoniazid, and experimental drugs was not allowed.

The patients were recruited from four AIDS Clinical Trials Units, five university-affiliated medical centers, and private practice groups. The study was approved by institutional review boards at each center, and all patients gave written informed consent.

Study Design and Treatment Regimen

The study was an open-label, randomized trial. Patients were stratified according to a diagnosis of AIDS or advanced AIDS-related complex at the time of initiation of zidovudine therapy and according to the dose of zidovudine that they were receiving at study entry. The intended sample size was 160 patients per treatment arm, the number required to detect a 15% difference in survival outcome with a power of 80% and a two-sided significance level of 0.05. After the enrollment of only 115 patients, the study was closed to enrollment in July 1991 because of slow accrual.

A blocked randomization procedure was used to assign patients in each center to zalcitabine or zidovudine. Zalcitabine (Hivid, Hoffmann-La Roche, Nutley, New Jersey) was given in two 0.375-mg tablets every 8 hours. Zidovudine (Retrovir, Burroughs Wellcome, Inc., Research Triangle Park, North Carolina) was initially given in two 100-mg capsules every 4 to 5 hours and then in one 100-mg capsule every 4 to 5 hours.

Management of Toxicity

If a serious toxicity occurred, therapy with the study medication was interrupted. After abnormal values returned to levels indicating a lower-grade toxicity or to pretreatment values, therapy with the study medication was restarted at half the dose. If severe anemia developed, red-cell transfusions or therapy with recombinant erythropoietin was administered. If patients had a toxicity that was persistent, recurred after reduction of study medication, or was life threatening, their study medication was permanently withdrawn. Zalcitabine therapy was permanently discontinued in patients who developed severe treatment-related peripheral neuropathy. Severe peripheral neuropathy was defined as moderate discomfort associated with the loss of a deep tendon reflex, the presence of paresthesias, or pain that either was refractory to non-narcotic analgesics, amitriptyline, or clonazepam or worsened during a 1-week period.

Evaluation of Patients

The pretreatment evaluation included a medical history, a physical examination, a signs-and-symptoms questionnaire, an evaluation for peripheral neuropathy, a Karnofsky performance score, and laboratory studies. Lymphocyte phenotyping was done by flow cytometry at the same laboratory at each center on peripheral blood mononuclear cells prepared on a Ficoll-Hypaque gradient or by the whole-blood lysis method with monoclonal reagents. The serum specimens collected for HIV p24 antigen determinations were frozen for batch testing with a commercial test kit (Abbott, North Chicago, Illinois).

All patients were re-evaluated every 2 weeks for the first 12 weeks and every 4 weeks thereafter. Patients who discontinued therapy with the study medication were followed for survival only. Clinical measures of efficacy were survival and time to a first critical event, which was defined as any AIDS-related condition or death. Secondary measures of efficacy were increases in CD4 lymphocyte counts, reduction of serum p24 antigen levels, weight gain, and changes in Karnofsky performance scores.

Statistical Analysis

Differences in proportions were assessed using the Fisher exact test. Time-to-event distributions were estimated using the method of Kaplan and Meier and compared using the Cox proportional-hazards regression model. To assess changes in the CD4 lymphocyte count, the serum p24 antigen level, weight, and Karnofsky performance score, parametric analyses of covariance were used. Subgroup analyses were carried out, with patients stratified according to diagnosis (AIDS or AIDS-related complex) and pretreatment CD4 lymphocyte count ( 100 cells/mm³ or >100 cells/mm³). All analyses were based on an intent-to-treat approach, using clinical data available through 15 June 1991 and survival data collected through 31 December 1991. The hazard ratio expressed as relative risk and 95% two-sided CIs are given when appropriate. All P values were two sided.

Results

Patient Sample

Between January 1990 and June 1991, 111 patients were enrolled in the study. Patients included 110 men and 1 woman (mean age, 36 years). Most patients were white and non-Hispanic. One hundred six patients were homosexual or bisexual, 7 had a history of intravenous drug use, 12 had a history of receipt of blood products, and 8 had heterosexual contact with a person at risk for HIV infection. Twenty-two patients had AIDS and 89 had AIDS-related complex at the time zidovudine therapy was first instituted.

Fifty-nine patients were randomly assigned to receive zalcitabine and 52 to receive zidovudine. No significant differences were noted between the two treatment groups regarding pretreatment characteristics (Table 1). The median duration of study treatment was 279 days (range, 16 to 437 days) for the zalcitabine group and 174.5 days (range, 8 to 400 days) for the zidovudine group. The time to discontinuation of study medication was statistically shorter in the zidovudine group compared with the zalcitabine group (P = 0.001).

Of the 111 patients enrolled, 56 were withdrawn from the study medication, 20 completed the study, and 35 were still receiving study medication when the database was closed for analysis on 15 June 1991. The reasons for discontinuation of study medication included treatment with nonallowed medications (5 patients in the zidovudine group); death (2 patients in the zalcitabine group and 1 patient in the zidovudine group); toxicity (12 patients in the zalcitabine group and 5 in the zidovudine group); self-withdrawal (6 patients in the zalcitabine group and 18 in the zidovudine group); loss to follow-up (2 patients in the zalcitabine group and 3 in the zidovudine group); and administrative reasons (2 patients in the zidovudine group). Significantly more patients were withdrawn from the zidovudine group than from the zalcitabine group (34 patients compared with 22 patients; P = 0.004). For patients who completed the study or discontinued use of the study medication before reaching a study end point, follow-up survival information was not available on 28 patients (54%) in the zalcitabine group and 32 patients (54%) in the zidovudine group.

Clinical End Points

Overall, 19 patients (33%) in the zalcitabine group and 17 patients (33%) in the zidovudine group developed AIDS or died as a first critical event (P = 1.0). The estimated 12-month event-free probabilities were 53% for the zalcitabine group and 57% for the zidovudine group (Figure 1). No significant differences were noted between the two groups (relative risk, 1.02; CI, 0.5 to 2.2; P > 0.2). In addition, no differences were noted between the two treatment groups when patients were grouped by a diagnosis of AIDS or AIDS-related complex or pretreatment CD4 lymphocyte count (Table 2). Fourteen patients (24%) in the zalcitabine group and 13 patients (25%) in the zidovudine group developed an AIDS-defining opportunistic infection or malignancy (P = 1.0). The estimated 12-month AIDS event-free probabilities were 69% for the zalcitabine group and 66% for the zidovudine group (relative risk, 0.95; CI, 0.4 to 2.3; P > 0.2).

Types of opportunistic infections among patients in the zalcitabine group included candidiasis (two patients), cytomegalovirus infection (two patients), cryptococcosis (three patients), Mycobacterium avium complex infection (seven patients), P. carinii pneumonia (three patients), and others (three patients). Types of opportunistic infections among patients in the zidovudine group included candidiasis (one patient), cytomegalovirus infection (five patients), M. avium complex infection (three patients), P. carinii pneumonia (four patients), and others (three patients).

Survival

Ten patients in the zalcitabine group and 13 in the zidovudine group died (P > 0.2). The estimated 12-month survival rate was 81% for the zalcitabine group and 75% for the zidovudine group (Figure 2). No significant differences between the two treatment groups were noted (relative risk, 1.39; CI, 0.5 to 3.8; P > 0.2). No differences in survival between the two treatment groups were noted when patients were grouped by a diagnosis of AIDS or AIDS-related complex or by pretreatment CD4 lymphocyte count (Table 3). The estimated 12-month survival for patients in the zalcitabine group and for patients in the zidovudine group was 73% and 54%, respectively, for patients with AIDS and 82% and 83%, respectively, for patients with advanced AIDS-related complex. For patients with a pretreatment CD4 lymphocyte count of 100 cells/mm³ or less, the estimated 12-month survival rate was 64% for zalcitabine recipients and 54% for zidovudine recipients. For patients with a pretreatment CD4 lymphocyte count greater than 100 cells/mm³, the estimated 12-month survival rate was 100% for both groups.

Causes of death among zalcitabine recipients included general debilitation (five patients), cytomegalovirus infection (one patient), P. carinii pneumonia (one patient), M. avium-complex infection (one patient), lymphoma (one patient), and unspecified pneumonia (one patient). Causes of death among zidovudine recipients included general debilitation (four patients), M. avium complex infection (three patients), sepsis (two patients), P. carinii pneumonia and cytomegalovirus infection (one patient), unspecified pneumonia (one patient), Kaposi sarcoma (one patient), and unknown causes (one patient).

Immunologic Studies

The mean CD4 lymphocyte count decreased over time for both treatment groups (Figure 3). Patients in the zalcitabine group showed significant decreases in the CD4 lymphocyte count from pretreatment values ( 20 cells/mm³ at week 28 and 60 cells/mm³ at week 56; P < 0.01 for both comparisons). Patients in the zidovudine group also showed significant decreases in the CD4 lymphocyte count from pretreatment values ( 41 cells/mm³ at week 28 and 55 cells/mm³ at week 44; P < 0.01 for both comparisons). Patients in the zidovudine group showed a significantly greater decrease in the CD4 lymphocyte count at weeks 28 ( 41 cells/mm³) and 36 ( 53 cells/mm³) compared with patients in the zalcitabine group ( 20 cells/mm³ at week 28 and 22 cells/mm³ at week 36; P = 0.03 and P = 0.05, respectively).

Overall, the mean CD4 lymphocyte counts for patients with AIDS and those with a pretreatment CD4 lymphocyte count of 100 cells/mm³ or less remained near pretreatment values through week 24 of treatment, and no differences between the two treatment groups were noted. Patients with advanced AIDS-related complex and those with a pretreatment CD4 lymphocyte count of more than 100 cells/mm³ had greater decreases in CD4 cell counts. Among patients with a pretreatment CD4 lymphocyte count of more than 100 cells/mm³, patients in the zidovudine group had greater decreases in CD4 cell counts during weeks 28 to 40 compared with those in the zalcitabine group (P < 0.05).

To better assess changes in CD4 lymphocyte counts over time, a linear relation between CD4 lymphocyte count and time since initiation of study medication was fit for each patient using simple linear regression. The estimated rate of decline in CD4 lymphocyte count was statistically greater in the zidovudine group (slope = 0.17 cells/day) compared with the zalcitabine group (slope = 0.08 cells/day; P = 0.05).

Virologic Studies

Ten patients (19%) in the zalcitabine group and seven patients (15%) in the zidovudine group had detectable levels of serum p24 antigen at enrollment in the study. Two patients (20%) in the zalcitabine group and one patient (14%) in the zidovudine group who had detectable pretreatment p24 antigen levels had decreases in serum p24 antigen to below detectable levels. Five patients (12%) in the zalcitabine group and three patients (8%) in the zidovudine group who did not have detectable pretreatment p24 antigen levels developed detectable levels (>31 pg/mL). No differences between the two treatment groups were noted.

Karnofsky Performance Status and Weight

Overall, the Karnofsky performance status decreased in all patients. However, patients in the zalcitabine group had a significantly slower decrease from pretreatment values (slope = 0.02 per day) compared with patients in the zidovudine group (slope = 0.07 per day; P = 0.02). Significantly less pronounced decreases from pretreatment values were noted at weeks 20 and 24 among patients with AIDS in the zalcitabine group compared with patients with AIDS in the zidovudine group (P < 0.05, respectively).

The mean body weight for patients in the zalcitabine group remained at or near pretreatment values through the first 40 weeks of the study, whereas body weight slightly decreased among patients in the zidovudine group. Patients in the zalcitabine group showed average weight gains of 0.5 kg and 0.4 kg at weeks 20 and 24, respectively, whereas patients in the zidovudine group showed weight losses of 1.8 kg and 2.4 kg at these time points (P = 0.04 and P = 0.05, respectively). The estimated rate of decline of body weight was 0.003 kg/d for the zalcitabine group compared with 0.016 kg/d for the zidovudine group (P = 0.03). More prominent differences in weight between the two treatment groups were noted for patients with advanced AIDS-related complex and those with pretreatment CD4 lymphocyte counts of more than 100 cells/mm³.

Adverse Experiences

Twelve patients (20%) in the zalcitabine group and 5 (10%) in the zidovudine group were withdrawn from study medication because of a toxicity. Reasons for discontinuation of therapy in the zalcitabine group included peripheral neuropathy (six patients [10%]), pancreatitis (one patient [2%]), esophageal ulcers (one patient [2%]), mouth ulcers (one patient [2%]), granulocytopenia (two patients [3%]), and hepatotoxicity (one patient [2%]). Among the patients in the zidovudine group, the reasons for discontinuation of study medication included anemia (one patient [2%]), granulocytopenia (two patients [4%]), thrombocytopenia (one patient [2%]), and hepatotoxicity (one patient [2%]).

Overall, 26 patients (44%) in the zalcitabine group and 3 patients (6%) in the zidovudine group had a treatment-related toxicity. However, only 19 patients (32%) in the zalcitabine group and 3 patients (6%) in the zidovudine group had a moderate or severe treatment-related toxicity that required interruption or withdrawal of study medication. The incidence of treatment-related toxicities is shown in Table 4.

The most common toxicity associated with zalcitabine therapy was ulcerative stomatitis (10 patients [17%]) and peripheral neuropathy (15 patients [25%]). Moderate or severe treatment-related ulcerative stomatitis or esophagitis occurred in 10 patients (17%). Only two of these patients required withdrawal of study medication. Moderate or severe treatment-related peripheral neuropathy occurred in 10 patients (17%). Six patients (10%) discontinued zalcitabine therapy because of peripheral neuropathy. Two additional patients had unresolved moderate peripheral neuropathy shortly after or at the time study medication was withdrawn for other reasons. The most common manifestations of moderate or severe treatment-related peripheral neuropathy were hypoesthesia in the lower extremities (six patients [10%]), paresthesia in the lower extremities (four patients [7%]), and pain in the lower extremities (three patients [5%]). Weakness in the lower extremities was uncommon (one patient). In a Cox regression model, patients who had a pretreatment CD4 lymphocyte count of less than 50 cells/mm³ were more likely to develop peripheral neuropathy than were patients with a count of 50 cells/mm³ or more (5 of 19 patients compared with 6 of 40 patients); also, patients who had a creatinine clearance rate of greater than 1.83 mL/s (110 mL/min) were less likely to develop peripheral neuropathy than were patients with a creatinine clearance rate of 1.83 mL/s (110 mL/min) or less (1 of 11 patients compared with 10 of 43 patients).

Few hematologic toxicities or hepatotoxicities were seen with zalcitabine therapy. Three patients (5%) developed serious neutropenia (<750 cells/mm³); three (5%), serious anemia (hemoglobin <75 g/L); three (5%), serious eosinophilia (>1000 cells/mm³); and four (7%), serious elevations in the serum glutamic-oxaloacetic transaminase level (>250 U/L).

The most common toxicities associated with zidovudine therapy were hematologic. Four patients (8%) developed serious anemia (hemoglobin <75 g/L); six (12%), serious neutropenia (<750 cells/mm³); and one (2%), serious thrombocytopenia (<50 000 platelets/mm³). Four patients (8%) required discontinuation of therapy. Three patients (6%) had serious elevations in serum glutamic-oxaloacetic transaminase levels (<250 U/L).

Discussion

Zidovudine therapy in patients with advanced HIV disease prolongs survival, decreases the severity and frequency of opportunistic infections, and results in transient increases in CD4 lymphocyte counts and decreases in HIV replication [1-3]. Although continued benefits have been described with prolonged therapy [5], long-term zidovudine treatment in patients with advanced HIV disease is associated with progressive HIV disease, decreases in CD4 lymphocyte counts, and emergence of viral strains with decreased in vitro susceptibility to zidovudine [1-7]. The reasons for the waning effectiveness of zidovudine during long-term treatment in patients with advanced HIV disease may be related to the limitations of nucleoside analogs in the presence of high viral titers and severe immunosuppression [12, 13]. The role of resistant viral strains is unclear. Based on previous experience with acyclovir for the treatment of herpesvirus infection in the immunocompromised host [14-16], it is likely that the presence of resistant viral strains is at least partly responsible for the waning effectiveness of zidovudine. Because cross-resistance with other nucleoside analogs does not necessarily occur [11], the administration of other potentially effective nucleosides or the combination of nucleosides may result in improved clinical outcome.

Our study was designed to evaluate the effectiveness and safety of zalcitabine in patients who had previously received 48 weeks or more of zidovudine therapy and who were, therefore, at an increased risk for disease progression and the development of viral strains with decreased susceptibility to zidovudine. Thus, the study was designed to determine specifically if changing therapy to another nucleoside analog would improve the subsequent outcome in patients who had received long-term zidovudine treatment. Because similar findings have not been noted in patients with early manifestations of HIV disease [17], the study sample was limited to patients with advanced HIV disease.

Because of the slow accrual of patients, the study was closed to accrual after enrollment of only 35% of the intended number of patients. Thus, the ability to detect a 15% difference in clinical outcome measures was limited by the smaller-than-expected sample size. Because patients were randomly assigned to the treatment groups, the smaller-than-expected sample size should not favor one study medication over the other, but it does limit the precision with which one can interpret the results of the study. One can only speculate as to the reasons for slow accrual. However, one plausible explanation was the broad availability of didanosine (ddI) through an expanded access program for patients with progressive HIV disease who had received previous zidovudine therapy. Availability of didanosine may have also resulted in the higher-than-expected rate of premature discontinuation of study medication, which would tend to decrease the sensitivity of detecting differences between study treatments. Because significantly more patients discontinued zidovudine therapy, it is possible that the study results reflect this limited zidovudine therapy.

Despite these limitations, several interesting results were noted for both clinical and surrogate measures of outcome. The estimated 12-month survival rate was 75% for patients assigned to receive zidovudine, which is consistent with other findings [1, 6], and 81% for those assigned to receive zalcitabine. Similarly, the relative risk for developing an AIDS-defining event or dying was greater for the zidovudine group than for the zalcitabine group. Even though the estimated 12-month event and survival rates favored zalcitabine, the imprecision of these measurements, as reflected by the wide confidence intervals, makes it difficult to ascribe a greater clinical benefit to zalcitabine. Using other potential markers of clinical benefit, such as Karnofsky performance status and weight, we found significant differences between the two treatment groups, favoring zalcitabine therapy. Because of our high dropout rate, caution should be used in ascribing a greater clinical benefit to zalcitabine.

Increases in CD4 lymphocyte count have been seen in patients treated with both zidovudine and zalcitabine [1-3, 9, 10]. Increases in CD4 lymphocyte counts were not noted in either treatment group in our study. This lack of increase in CD4 lymphocyte counts may reflect both the advanced stage of HIV disease in these patients and previous treatment with zidovudine. However, the rate of decline in CD4 lymphocyte counts was significantly slower among zalcitabine recipients than among zidovudine recipients. Cellular viral burden is closely related to CD4 lymphocyte counts, and increasing viral burden in peripheral blood CD4 lymphocytes is directly associated with progressive decreases in CD4 lymphocyte counts and clinical deterioration [18, 19]. Although the clinical significance of a slower rate of decline in CD4 lymphocyte counts with zalcitabine therapy could not be determined in our study, this finding is encouraging and may imply a potential CD4-lymphocyte benefit with switching to zalcitabine in this group of patients.

Most of the drug-related toxicities occurred among patients in the zalcitabine group. It should be noted that enrollment was restricted to patients who tolerated zidovudine for 48 weeks or more. Thus, the lower incidence of toxicities and the low rate of discontinuation of zidovudine therapy because of toxicity is not surprising because most zidovudine-related toxicities occur early in the course of treatment [2-6]. In addition, because this was an open-label study, it is possible that physicians ascribed less toxicity to a drug with which they were more familiar.

Nineteen patients in the zalcitabine group experienced a moderate or severe treatment-related toxicity. The most frequently reported treatment-related toxicity was peripheral neuropathy. Overall, 17% of the patients treated with zalcitabine developed a moderate or severe treatment-related peripheral neuropathy. However, zalcitabine therapy was discontinued in only 10% of the patients for this reason. Careful monitoring for signs and symptoms of peripheral neuropathy, including hypoesthesias, paresthesias or pain, and loss of a deep tendon reflex is important in the prevention of serious, debilitating peripheral neuropathy. For patients who develop moderate or severe symptoms, zalcitabine therapy should be interrupted until all manifestations of peripheral neuropathy resolve. For patients with less severe symptoms or those whose symptoms rapidly resolve, lower doses of the drug may be tried [20]. It should also be noted that milder forms of zalcitabine-related peripheral neuropathy can occur that may not necessarily require immediate interruption or discontinuation of therapy.

Ulcerative stomatitis was also associated with zalcitabine therapy, occurring early in the course of treatment and appearing to resolve rapidly within 1 to 2 weeks in most patients despite continued therapy. Symptomatic topical treatment should be tried while continuing zalcitabine therapy. Although uncommon, esophageal ulcers may also occur. Patients with esophageal ulcers should stop zalcitabine therapy until other causes of esophageal ulcers are evaluated.

Pancreatitis has been described with another nucleoside analog, didanosine [21, 22]. The pathogenesis of pancreatitis is unclear, and pancreatitis can be seen with HIV disease and other drugs used to treat complications of HIV disease, such as P. carinii pneumonia with pentamidine isethionate. One case of severe pancreatitis was noted in our study. The total incidence of pancreatitis reported among all patients receiving zalcitabine therapy is less than 1% (Personal communication. Hoffmann-La Roche). Based on this experience, it does not appear that pancreatitis is a frequent complication of zalcitabine therapy.

Data from the current study suggest that the administration of zalcitabine may be of some benefit to patients who have received previous zidovudine therapy for 48 weeks or more. However, because of the smaller-than-anticipated sample size and a shorter duration of therapy among zidovudine recipients, the results of our study must be interpreted with caution. Further studies are needed to better define the efficacy of zalcitabine in this group of patients.

Appendix

The following persons were also involved in the study: OakLawn Physicians Group, Dallas, Texas: Robert C. Andruczk, MD; Davies Medical Center, San Francisco: Stephanie La Carruba, MSN; Mount Zion Medical Center, University of California, San Francisco, California: W.L. Drew, MD, PhD, Eileen Glutzer, RN, Ranaldo Loya, PA-C; Graduate Hospital, Philadelphia, Pennsylvania: Bernard A. Mason, MD, and Arthur P. Staddon, MD; University of Cincinnati, Cincinnati, Ohio: Pamela A. Daniel, RN, and Lisa Veach, MD; Albany Medical College, Albany, New York: David L. Herman, MD, and Danielle M. Powell, MS, RN; Robert Wood Johnson Medical School, New Brunswick, New Jersey: David J. Gocke, MD; Johns Hopkins Hospital, Baltimore, Maryland: Saralyn Elkin, RN; and Hoffmann-La Roche, Nutley, New Jersey: Donna Sutherland, Lucille Donatacci, MS, Aileen T. Ward, MS, Darien Wilson, and Cheryl Ward.