The combination of the nucleoside analogs zidovudine and zalcitabine has additive to synergistic inhibitory activity against HIV in vitro, and viral isolates resistant to zidovudine remain susceptible to zalcitabine in vitro [12-14]. Combination therapy with zidovudine and zalcitabine causes a greater and more sustained increase in CD4 cell counts and suppression of HIV p24 antigenemia than does either drug used alone in patients with advanced HIV disease who have not previously received therapy with zidovudine or zalcitabine [15]. The clinical correlation of these findings is unknown. These observations prompted us to try to determine whether zalcitabine monotherapy or combination therapy with zidovudine and zalcitabine would improve clinical outcome in patients with advanced HIV disease who had previously received zidovudine.

Methods

Top Methods Results Discussion Author & Article Info References

Study Sample

The study sample consisted of patients with symptomatic HIV disease who had a CD4 cell count of 300 cells/mm³ or less or patients with asymptomatic HIV disease who had a CD4 cell count of 200 cells/mm³ or less. Participants had previously received and tolerated at least 6 months of zidovudine treatment. The eligibility criteria also included a hemoglobin concentration of 92 g/L or more, a neutrophil count of 1.0 x 10⁹/L or more, a platelet count of 75 x 10⁹/L or more, serum alanine aminotransferase and aspartate aminotransferase levels no higher than five times the upper range of normal, Karnofsky performance status of 60 or more, and positive results of serum tests for HIV antibody. We excluded patients with a history of intolerance to zidovudine at a dose of 600 mg/d or less, previous zalcitabine therapy, mild or more severe peripheral neuropathy, or Kaposi sarcoma requiring systemic therapy; women of childbearing age who were pregnant or breast-feeding were also excluded. We accepted patients receiving chemoprophylaxis for Pneumocystis carinii pneumonia, candidiasis and herpes simplex virus infection, and those receiving erythropoietin and granulocyte colony-stimulating factor. We excluded persons taking other antiretroviral drugs, biological response modifiers, cytotoxic chemotherapy, drugs other than isoniazid that cause peripheral neuropathy, or investigational drugs.

We recruited patients from 35 AIDS Clinical Trials Units and 16 National Hemophilia Foundation sites between December 1990 and August 1991. The study was approved by the institutional review board at each institution, and patients gave written informed consent.

Study Design and Treatment Regimen

The study was a multicenter, randomized, double-blind clinical trial that compared the safety and efficacy of zidovudine, zalcitabine, and the combination of zidovudine and zalcitabine. Randomization was weighted at 2:2:3, favoring the combination group. Patients were stratified by HIV disease status (symptomatic or asymptomatic), by duration of previous zidovudine therapy ( 1 year or >1 year), and by P. carinii pneumonia chemoprophylaxis regimen (systemic therapy, nonsystemic therapy, both, or neither). Zalcitabine (Hivid; Hoffmann-La Roche, Inc., Nutley, New Jersey) was given in two 0.375-mg tablets every 8 hours. Zidovudine (Retrovir; Burroughs Wellcome Co., Research Triangle Park, North Carolina) was given in two 100-mg capsules every 8 hours.

The primary end point for the study was time to an acquired immunodeficiency syndrome (AIDS)-defining event or death, whichever occurred first. All end points were reviewed in a blinded manner by the study chairpersons. Patients continued to receive blinded study medication after the development of a primary study end point. The protocol was modified on 20 March 1992 to allow patients who had reached a primary study end point the option to cross over to open-label combination therapy. Secondary end points included survival, CD4 cell count changes, and decreased HIV p24 antigenemia.

Management of Toxic Effects

All study medications were interrupted, regardless of whether the toxicity was believed to be associated with either zidovudine or zalcitabine, in patients in whom a moderate or severe peripheral neuropathy or another severe or worse toxic effect developed. Once the toxicity grade returned to pretreatment values or a lower grade, study medications were restarted at one half the initial dose. Study medications were permanently discontinued in patients who had a recurrent moderate or severe peripheral neuropathy or another severe or worse toxic effect within a 30-day period. Patients who had recurrent toxic effects after more than 30 days, except for those with peripheral neuropathy, continued to receive study medications once the toxicity grade returned to a lower grade or to pretreatment values.

Patient Evaluation

Patients were evaluated at weeks 0, 4, 8, and 12, and every 8 weeks thereafter. The CD4 cell counts were measured within 60 days before study enrollment, at weeks 0, 2, and 4, every 4 weeks until week 52, and every 8 weeks thereafter. Blood for HIV p24 antigen determinations was collected at weeks 0, 2, 4, 8, 12, 24, 36, 48, and 60 and every 16 weeks thereafter. Determinations of HIV p24 antigen were done simultaneously on stored serum aliquots by an enzyme-linked immunosorbent assay (Abbott Laboratories, North Chicago, Illinois). A positive assay result was one that detected an antigen level of 25 pg/mL or more. Laboratories measuring CD4 cell counts and serum p24 antigen levels had to meet the performance-monitored standards of the AIDS Clinical Trials Group.

Statistical Analysis

We assessed differences in proportions using the Fisher exact test. Time-to-event distributions were estimated using the Kaplan-Meier method and compared using the log-rank test and Cox proportional-hazards model [16], with stratification for HIV disease status, duration of previous zidovudine therapy, and P. carinii pneumonia prophylaxis [17]. Analyses were based on an intention-to-treat approach [18]. We censored data on toxic effects 30 days after crossover or discontinuation of study medications. We used the hazard ratios, expressed as relative risk, and 95% two-sided confidence intervals. All P values were two-sided and were not adjusted for multiple comparisons.

A test of trend evaluating the interaction between pretreatment CD4 cell counts and treatment effect on the primary end point was based on Cox models and included the study stratification factors as covariates and pretreatment CD4 cell counts as a continuous variable. For completeness, two additional tests of trend used stratified Cox models in which pretreatment CD4 cell counts were modeled as a categorical variable. In one analysis, pretreatment CD4 cell count was modeled as a binary variable, above and below the median CD4 cell count (119 cells/mm³). In a second analysis, pretreatment CD4 cell count was modeled as a discrete variable with three levels ( –1,0, 1) corresponding to the three planned CD4 cell subgroups. Changes in CD4 cell counts and serum HIV p24 antigen levels were expressed as the percentage change from pretreatment values for each patient. Trends in CD4 cell counts over time were evaluated by estimating the slopes from pretreatment to the time of aggregate peak (week 2) and separately estimating the slopes for subsequent time points. We compared the slopes using nonparametric tests [19].

We also analyzed treatment differences for patients with pretreatment CD4 cell counts of less than 50 cells/mm³, 50 to 150 cells/mm³, and 150 cells/mm³ or greater. The three CD4 cell count subgroups were specified by the study chairpersons in June 1992, which was before any interim review of the primary end-point data and unblinding to the study results by the study chairpersons (March 1993). The three CD4 cell count subgroups were chosen based on documented associations between CD4 cell counts and survival and development of zidovudine resistance.

Results

Top Methods Results Discussion Author & Article Info References

Study Sample

We enrolled 1001 patients in the study between December 1990 and August 1991. No patients were ineligible for the study, except for those granted an exemption by the study chairpersons. We excluded 10 patients from the analyses: 7 who never received study medications and were not followed and 3 who were lost to follow-up within the first 2 weeks of treatment. The patients included 888 (90%) men and 103 (10%) women; their median age was 37 years. Overall, 809 patients (82%) were white, 152 (15%) were black, and 3% were neither; 117 (12%) were Hispanic and 884 (88%) were non-Hispanic. Eight hundred nineteen (83%) patients had symptomatic HIV disease. The median duration of previous zidovudine therapy was 18 months. The median pretreatment CD4 cell count was 119 cells/mm³, and 254 of 925 patients had detectable levels of serum HIV p24 antigen ( 25 pg/mL) before treatment. The treatment groups were well balanced with regard to pretreatment characteristics (Table 1).

We randomly assigned 283 patients to the zidovudine group, 285 to the zalcitabine group, and 423 to the combination group. Three hundred eighty-six patients had 150 or more CD4 cells/mm³, 336 had 50 to 150 CD4 cells/mm³, and 269 had fewer than 50 CD4 cells/mm³. Overall, the treatment groups within each CD4 cell subgroup were well balanced with regard to pretreatment characteristics (data not shown). The median duration of follow-up was 17.7 months: 17.9 months for the zidovudine group, 17.8 months for the zalcitabine group, and 17.7 months for the combination group (P = 0.97). The median duration of blinded study treatment was 11.7 months: 10.5 months for the zidovudine group, 12.2 months for the zalcitabine group, and 11.9 months for the combination group (P = 0.22). We evaluated compliance with zidovudine therapy by comparing mean corpuscular volume during treatment. We found no differences in mean corpuscular volume values compared with pretreatment values for the zidovudine or combination groups. We noted an initial progressive decrease in mean corpuscular volume values for the zalcitabine group.

Three hundred seventy-two patients (38%) withdrew from the study medication arm before reaching a primary study end point. The median time to withdrawal for these 372 patients was 7.7 months: 6.7 months for the zidovudine group, 8.5 months for the zalcitabine group, and 8.0 months for the combination group (P = 0.25). The most common reasons for discontinuation of study medication among patients who withdrew were the desire to receive other antiretroviral therapy (34%), patient request (28%), and milder grades of toxicity (23%). In addition, eight patients withdrew because of site closure. Sixty-three patients (6.4%) were lost to follow-up. Twenty-eight percent of the participants were still receiving blinded study medication when the database was closed for analysis on 15 January 1993. Ninety-five patients who reached a primary study end point elected to cross over to open-label combination therapy: 27 in the zidovudine group, 32 in the zalcitabine group, and 36 in the combination group (P = 0.49).

Clinical Disease Progression

Progression of HIV infection or death occurred in 405 patients (41%): 118 (42%) in the zidovudine group, 123 (43%) in the zalcitabine group, and 164 (39%) in the combination group. We found no significant difference among the treatment groups in the time to disease progression or death (P = 0.26; see Figure 1). The most common causes of disease progression were P. carinii pneumonia (22%), cytomegalovirus infection (13%), disseminated Mycobacterium avium complex infection (11%), Kaposi sarcoma (8%), and esophageal candidiasis (8%); 58 patients (14%) died without a preceding AIDS-defining event. The distribution of causes of disease progression was similar among the CD4 cell subgroups. The 12-month, event-free probabilities were 70% for the zidovudine group, 67% for the zalcitabine group, and 73% for the combination group. We found no differences among the three treatment groups when the patients were grouped by HIV disease status (symptomatic, P = 0.30; asymptomatic, P = 0.75), duration of previous zidovudine therapy ( 1 year, P = 0.38; >1 year, P = 0.16), and type of P. carinii pneumonia chemoprophylaxis (systemic therapy, P = 0.25; nonsystemic therapy, P = 0.44; neither or both, P = 0.43). Overall, 94% of the patients received P. carinii pneumonia chemoprophylaxis during the study, and we noted no differences among the three treatment groups.

View larger version (24K): [in this window] [in a new window]   Figure 1. Estimated distributions of time until disease progression. Proportion of patients with disease progression or death, whichever occurred first, by treatment group overall (A) and for patients with 150 or more CD4 cells/mm3 (B), 50 to 150 CD4 cells/mm3 (C), and fewer than 50 cells/mm3 (D).

View larger version (20K): [in this window] [in a new window]   Figure 2. Hazard ratio comparisons and 95% confidence intervals for time to clinical progression and survival. Hazard ratio comparisons and 95% confidence intervals of combination therapy compared with zidovudine (A), zalcitabine compared with zidovudine (B), and combination therapy compared with zalcitabine (C).

A trend analysis, which divided patients into two groups (above and below the median pretreatment CD4 cell count of 119 cells/mm³), showed a significant difference for the combination group (P =0.013) and for the zalcitabine group (P = 0.009) compared with the zidovudine group. A trend analysis, which modeled pretreatment CD4 cell counts as three discrete levels corresponding to the three CD4 cell subgroups, yielded nonsignificant results for the combination (P = 0.09) and zalcitabine groups (P = 0.19) compared with the zidovudine group.

Survival

One hundred seventy-two patients (17%) died: 43 (15%) in the zidovudine group, 51 (18%) in the zalcitabine group, and 78 (18%) in the combination group. Thirty-five patients (20%) died while receiving study medication or within 1 month of discontinuing therapy with study medication. The rest of the patients died an average of 7.1 months after discontinuing study medication. We noted no difference in survival among the treatment groups (P = 0.54).

Clinical Disease Progression Categorized by CD4 Cell Count Subgroup

As suggested by the trend analysis, we noted a difference in time to disease progression or death among the treatment groups when we grouped patients by pretreatment CD4 cell counts. Disease progression or death was significantly less likely to occur in patients receiving combination therapy who had 150 or more CD4 cells/mm³ than in patients receiving zidovudine therapy who had 150 or more CD4 cells/mm³ (relative risk, 0.51; CI, 0.28 to 0.93; P = 0.029; Figure 2). We found no differences between the zalcitabine and zidovudine groups (relative risk, 0.74; CI, 0.40 to 1.36; P = 0.33) and between the combination and zalcitabine groups (relative risk, 0.70; CI, 0.36 to 1.35; P = 0.29) for patients who had 150 or more CD4 cells/mm³. We noted no differences among the treatment groups for patients with pretreatment CD4 cell counts of 50 to 150 cells/mm³ (P = 0.69) and for patients with pretreatment CD4 cell counts of less than 50 cells/mm³ (P = 0.57, Figure 2).

We observed no differences in survival among the three treatment groups for patients in the three CD4 cell count subgroups. Fatality rates per 100 person-years of follow-up were 4.1, 1.9, and 1.7, respectively, for the zidovudine, zalcitabine, and combination groups among patients who had pretreatment CD4 cell counts of 150 or more cells/mm³. For patients with pretreatment CD4 cell counts of 50 to 150 cells/mm³, fatality rates were 9.0, 12.3, and 11.2; for patients with pretreatment CD4 cell counts less than 50 cells/mm³, fatality rates were 27.4, 30.8, and 36.5, respectively, for the zidovudine, zalcitabine, and combination groups.

Changes in CD4 Cell Counts

Combination therapy resulted in a greater rate of initial increase in CD4 cells compared with zidovudine therapy (P = 0.0004) but not compared with zalcitabine therapy (P = 0.06) and a slower rate of decline in CD4 cells compared with zidovudine therapy (P = 0.007) but not compared with zalcitabine therapy (P = 0.06; Figure 3).

View larger version (26K): [in this window] [in a new window]   Figure 3. Median CD4 cell count trends. Median CD4 cell count changes from pretreatment CD4 cell counts categorized by treatment group.

Changes in HIV p24 Antigen Levels

We had 925 pretreatment samples to determine serum HIV p24 antigen levels. Sixty-nine patients (26%) in the zidovudine group, 72 (27%) in the zalcitabine group, and 113 (29%) in the combination group had detectable levels of p24 antigen ( 25 pg/mL). The median percentage decrease in serum p24 antigen at week 2 was 37% for the combination group, 11% for the zidovudine group, and 13% for the zalcitabine group. The combination group had a greater initial decrease in serum p24 antigen levels at week 2 than did either the zidovudine or zalcitabine groups (P = 0.0001 and P = 0.0009, respectively). This decrease in serum p24 antigen levels appeared to persist over time. Similar findings were noted when we grouped patients by pretreatment CD4 cell counts, although significance was not achieved for patients with fewer than 150 CD4 cells/mm³.

Dose Modifications and Discontinuation of Therapy with Study Medication

The time to a first study medication modification (dose interruption, reduction, or discontinuation) was not significantly different among the treatment groups (P = 0.57). The percentage of visits at which the patients received full doses of study medications was slightly less for the combination group (73%) than for the zidovudine or zalcitabine groups (78% in both groups). Similarly, the percentage of visits at which study medication was interrupted was slightly greater for the combination group (12%) than for the zidovudine or zalcitabine groups (9% and 10%, respectively).

The proportion of patients with an interruption or reduction in study medication was greater for patients with fewer than 50 CD4 cells/mm³ (90%) than for those with 50 to 150 CD4 cells/mm³ (82%) or for those with 150 or more CD4 cells/mm³ (66%; P < 0.001). Within each subgroup, we noted no significant differences among the treatment groups related to the proportion of patients with an interruption or reduction in study medication.

The percentage of visits at which the patients received full doses of study medication was the lowest for patients receiving combination therapy and greater for patients with 150 or more CD4 cells/mm³ (83%) than for patients with 50 to 150 CD4 cells/mm³ (74%) or patients with fewer than 50 CD4 cells/mm³ (58%). The percentage of visits at which study medication was interrupted was greater for patients with fewer than 50 CD4 cells/mm³ and greater for the combination group (20%) than for the zidovudine (14%) or zalcitabine (11%) groups among patients with fewer than 50 CD4 cells/mm³.

Overall Toxic Effects

The 12-month probabilities for severe or worse toxic effects were 45%, 40%, and 49%, respectively, for the zidovudine, zalcitabine, and combination groups (P = 0.06). Table 2 lists common toxic effects. Neutropenia was the most common toxic side effect associated with zidovudine therapy, and peripheral neuropathy (any grade) was the most common toxic effect associated with zalcitabine therapy.

Pancreatitis was uncommon, and we noted no differences among the treatment groups. One death secondary to pancreatitis, which we believed to be associated with study medication (zalcitabine), occurred in the combination group [20].

Toxic Effects Categorized by CD4 Cell Count Subgroup

Severe or worse toxic effects were uncommon among patients with 150 or more CD4 cells/mm³, and we found no differences among the treatment groups (Table 3). However, for patients with 150 or more CD4 cells/mm³, moderate or worse peripheral neuropathy was more common among patients in the zalcitabine (16%) and combination (21%) groups than in the zidovudine group (9%; P = 0.03). Although toxic effects were slightly more common among patients with 50 to 150 CD4 cells/mm³, only the incidence of stomatitis was different among the treatment groups, which was more common in the zalcitabine group (8%) than in the zidovudine (1%) and combination (2%) groups (P = 0.03). Severe or worse toxic effects were common in patients with fewer than 50 CD4 cells/mm³, and we noted differences among the treatment groups, with a higher incidence of severe or worse neutropenia in the zidovudine (41%) and combination (35%) groups than in the zalcitabine group (18%, P = 0.003) and a higher incidence of severe or worse hepatic toxicity in the zidovudine group (14%) than in the zalcitabine (5%) or combination (4%) groups (P = 0.02).

Discussion

Top Methods Results Discussion Author & Article Info References

Patients with advanced HIV disease who receive long-term zidovudine therapy are at risk for disease progression, the emergence of a more virulent viral strain, and the development of viral isolates with decreased susceptibility to zidovudine [14, 21, 22]. The combination of zidovudine and zalcitabine has additive to synergistic inhibitory activity against HIV in vitro and yields enhanced and more persistent CD4 cell count responses in patients who receive no previous antiretroviral therapy [13, 15]. Based on these findings, combination therapy with zidovudine and zalcitabine has become widely used to treat patients with advanced HIV disease. No data, however, are available on the clinical benefits of this combination regimen and the potential indications for its use. Because zidovudine therapy is associated with waning effectiveness over time, we designed our study to assess several treatment strategies for patients who previously received zidovudine therapy. We designed the study specifically to determine whether switching therapy to zalcitabine or adding zalcitabine to zidovudine therapy rather than continuing zidovudine would improve the outcome of patients at greatest risk for disease progression.

We observed no differences in clinical outcome among the three treatment groups, which suggests that changing therapy to zalcitabine or adding zalcitabine to zidovudine provided no additional clinical benefits compared with continuing zidovudine in these patients. In contrast, CD4 cell and serum p24 antigen responses favored combination therapy but did not achieve the magnitude or sustained effects noted when treating patients with no previous nucleoside therapy [15]. These findings raise questions about the optimal treatment strategy for patients with advanced HIV disease and the best time to use combination therapy or to change therapy to treat HIV disease. The disparity between these findings and those noted when evaluating patients without previous therapy emphasizes the fact that previous antiretroviral therapy may affect subsequent responses to other therapies. These data also suggest that CD4 cell count and serum p24 antigen responses may not accurately reflect clinical response in patients who previously received nucleoside therapy, or that the study population had varying treatment outcomes, with only certain subgroups benefiting.

Several potential explanations exist for the overall findings in our study. The median duration of previous zidovudine therapy was considerably longer (18 to 19 months) than anticipated when this study was planned. Because long-term zidovudine therapy is associated with zidovudine resistance and zidovudine resistance is associated with a greater risk for disease progression [23], adding a single additional nucleoside may be unlikely to produce synergistic anti-HIV activity. Further, although laboratory isolates resistant to zidovudine in vitro remain susceptible to other nucleosides, patients with viral isolates highly resistant to zidovudine, compared with those with susceptible isolates, are at a greater risk for disease progression whether they continue zidovudine or receive another nucleoside, such as didanosine [23]. However, switching therapy from zidovudine to didanosine is associated with an improved outcome in patients with previous zidovudine therapy [10]. Thus, it is difficult to know if our findings reflect the relative lack of antiretroviral activity of zalcitabine in these patients or the late treatment intervention with zalcitabine. Earlier intervention with combination or zalcitabine monotherapy, before viral resistance, high viral burden, or phenotypic changes occur, may have produced a better outcome.

The toxicity management algorithm that we used required that both study medications be interrupted if any serious toxic effects developed. The time off study medications for patients with lower CD4 cell counts tended to be greater for the combination group than for either monotherapy group. One of the potential advantages of combination therapy is the greater likelihood of administering continuous antiretroviral therapy. Because of the toxicity management, this study may have underestimated the effectiveness of combination therapy among patients with lower CD4 cell counts, and different results may be seen with less rigid management of toxic effects. Because the duration of study follow-up was also longer than the duration of study treatment, the results may again underestimate potential outcome differences among the treatment groups. Because of this concern, we did an exploratory analysis to gain insight into the possible association between early treatment cessation and treatment outcome. When we censored data within 2 months of stopping study treatment, combination therapy was associated with a significantly lower rate of disease progression or death than were either zidovudine or zalcitabine monotherapy. This type of analysis is known to be biased. Patients receiving more toxic agents, for example, may terminate study treatment earlier and be excluded from the study results. Therefore, caution should be used when interpreting this exploratory analysis. Finally, the inclusion of a large portion of patients with low CD4 cell counts who deteriorated rapidly and insufficient follow-up of patients with higher CD4 cell counts who may have benefited from zalcitabine or combination therapy may also explain the overall negative results in this study.

Although we noted no overall differences among the three treatment groups, we found an interaction between pretreatment CD4 cell count and treatment outcome. A trend analysis showed that the relative efficacy of combination therapy was significantly associated with the pretreatment CD4 cell count. As the CD4 cell count increased, combination therapy, but not zalcitabine monotherapy, produced a significantly lower rate of disease progression compared with zidovudine. These data underscore the potential importance of CD4 cell counts in subsequent responses to antiretroviral therapy and suggest that this combination regimen may be beneficial for a subgroup of patients with advanced HIV disease who previously received long-term zidovudine therapy.

A planned subgroup analysis based on pretreatment CD4 cell counts also showed that combination therapy relative to zidovudine therapy was associated with a decrease in disease progression for patients with a pretreatment CD4 cell count of 150 or more cells/mm³ but not for those with a pretreatment CD4 cell count less than 150 cells/mm³. Caution should be used when interpreting the subgroup analysis because subgroup analyses may produce spurious results, and we did not make adjustments for multiple comparisons. The trend analysis does not suffer from the same limitations. Consistency between the results of the subgroup analysis and more definitive trend analysis strengthens the conclusions that combination therapy may be beneficial in patients with higher CD4 cell counts. The interaction between pretreatment CD4 cell counts and treatment outcome is important when interpreting the results of this study and suggests the need to evaluate treatment alternatives earlier in the course of HIV disease and after a shorter duration of zidovudine therapy.

The toxicity profile for either drug or the combination was not different than anticipated from other studies [1, 15, 24]. Although pancreatitis has occurred during zalcitabine therapy, the data from our study suggest that the rate of pancreatitis is low. The incidence of severe or worse toxic effects was inversely correlated with the pretreatment CD4 cell count, and patients with 150 or more CD4 cells/mm³ tolerated therapy with no apparent difference in the rates of severe or worse toxic effects among the treatment groups. In contrast, patients with lower pretreatment CD4 cell counts had a higher rate of toxicity, especially in the groups treated with regimens containing zidovudine.

The incidence of severe or worse peripheral neuropathy was low and did not differ among the treatment groups. However, the incidence of moderate peripheral neuropathy was higher with zalcitabine therapy than with zidovudine monotherapy and was similar to that in previous reports [24, 25]. In contrast to overall toxic effects, differences in the rates of peripheral neuropathy among the three treatment groups were most notable among patients with higher CD4 cell counts. We noted no differences for patients with lower CD4 cell counts, suggesting that the risk for peripheral neuropathy, whether drug or HIV related, was similar. Careful monitoring and interruption of therapy may decrease the incidence of severe peripheral neuropathy but will not necessarily prevent less severe peripheral neuropathy in this setting.

The lack of effectiveness of zalcitabine used alone or combined with zidovudine and the higher rates of toxicity in patients with low CD4 cell counts prompt questions about continued nucleoside therapy in these patients. However, although we noted no overall benefit from zalcitabine monotherapy or combination therapy, our study suggests that combination therapy may be beneficial in patients with higher CD4 cell counts. In addition, the data from this study show the importance of evaluating alternative treatment strategies earlier in the course of HIV disease or after a shorter interval of initial antiretroviral therapy, particularly for patients with advanced HIV disease. Finally, this study evaluated treatment strategies for patients who previously received zidovudine therapy, and these results do not address the use of combination therapy as an initial treatment for HIV disease.