Nucleoside Analogs Plus Ritonavir in Stable Antiretroviral Therapy–Experienced HIV-Infected Children:  A Randomized Controlled Trial FREE

In early 1997 many children infected with human immunodeficiency virus (HIV) were receiving single or dual nucleoside analog therapy and were clinically and immunologically stable,^1- 2 despite increasing plasma HIV RNA levels. However, the availability of protease inhibitors and their demonstrated success in adults raised the question of how they should be used in children. Pediatric AIDS Clinical Trials Group (PACTG) Protocol 338 was undertaken to evaluate the safety, tolerance, and virologic efficacy of changing from current antiretroviral treatment in clinically and immunologically stable, protease inhibitor–naive children to either dual nucleoside analog therapy or a 2- or 3-drug regimen containing a protease inhibitor, ritonavir.

Current antiretroviral guidelines for treatment of HIV-infected children and adults recommend initiation of therapy with a combination of antiretroviral drugs, including a protease inhibitor.^3- 4 Preliminary results from PACTG Protocol 338⁵ were instrumental in clarifying the role of protease inhibitors for children in these guidelines. Therapeutic results at 1 year are presented in this article.

The PACTG 338 study was a multicenter, randomized, phase 2 clinical trial that compared change from current therapy to either zidovudine plus lamivudine or 1 of 2 ritonavir-containing regimens (a 3-drug regimen of ritonavir, zidovudine, and lamivudine or a 2-drug regimen of ritonavir and stavudine) in HIV-infected, clinically stable children. All subjects were aged 24 months to 17 years; had stable CD4 cell number or percentage maintained in Centers for Disease Control and Prevention (CDC) immune category 1 or 2 during the 4 months prior to study entry⁶; had experienced no new CDC clinical category C diagnosis in the 12 months prior to study entry; received continuous antiretroviral therapy in the 16 weeks prior to study entry; and were either zidovudine- and lamivudine-naive or had received no more than 6 weeks of zidovudine and lamivudine in the year prior to study entry and none in the 4 months prior to study entry. Exclusion criteria included current grade 3 or 4 adverse effect (as judged by protocol-specified, standard pediatric toxicity criteria); active opportunistic and/or serious bacterial infection; documented hypersensitivity to any of the therapies under study; prior protease inhibitor therapy; or current diagnosis of malignancy or pregnancy.

Children were stratified by CD4 cell percentage (<25% vs ≥25% of the normal range) and randomized in a balanced fashion to 1 of 3 open-label treatment arms. Medication doses dispensed were 160 mg/m² of zidovudine (maximal dose, 200 mg/dose) 3 times per day and 4 mg/kg of lamivudine (maximal dose, 150 mg/dose) twice per day; that same regimen with 350 mg/m² of ritonavir (maximal dose, 600 mg/dose) twice per day; or 350 mg/m² of ritonavir twice per day and 4 mg/kg of stavudine (maximal dose, 40 mg/dose) twice per day.

The primary objective of the study was to evaluate the safety and tolerance of these 3 treatment regimens and compare the change in plasma HIV RNA copy number between entry and study weeks 12 and 48 among the regimens. The duration of study treatment for each patient was initially planned to be 48 weeks but was subsequently extended to 120 weeks. Only 48-week outcomes are presented here. Randomized study treatment was discontinued for children who experienced a virologic failure, disease progression, or persistent grade 3 or higher drug-related adverse effects. Such children were offered the best available therapy at the discretion of their clinician, themselves, or their parents. All subjects were enrolled for the full study period of 48 weeks.

A primary virologic failure at week 12 was defined as a failure to achieve either HIV RNA copy number of at least 2.0 log₁₀ copies/mL below baseline values or to maintain at least 10,000 copies/mL. A subsequent virologic failure was defined as a persistent 0.75 log₁₀ increase above the nadir in HIV RNA copy number for patients who had an initial HIV RNA copy number decrease of more than 2 log₁₀ but whose RNA remained at more than 10,000 copies/mL at study week 12, or as a persistent increase in HIV RNA to more than 10,000 copies/mL for patients who had an initial HIV RNA decrease to less than 10,000 copies/mL.

Two hundred ninety-eight HIV-infected children from 48 sites were enrolled in the study. The institutional review board at each institution approved the study, and informed consent was obtained from all patients or their parents.

Evaluations were performed within 14 days prior to randomization (the preentry visit), at study entry, and every 4 weeks during the study. These visits included a medical history, physical examination, a complete blood cell count with differential and serum chemistries. Lymphocyte surface markers were evaluated at preentry; entry; study weeks 4, 8, 12, and every 12 weeks thereafter and were performed by local laboratories participating in the National Institute of Allergy and Infectious Diseases (NIAID) Flow Cytometry Quality Assurance Program.⁷ Specimens for HIV-1 RNA were obtained at preentry; entry; and at study weeks 4, 12, 24, 36, 44, 48, and every 12 weeks thereafter. The NucliSens Assay (Organon Teknika, Durham, NC)⁸ was used in the assessment of HIV RNA copy number.⁹ Specimens from individuals from preentry through week 12 were assayed for HIV RNA copy number in batched fashion; subsequent specimens were run at the specified individual time points. The lower limit of assay quantification for RNA was 400 copies/mL. Assay results were adjusted using Virology Quality Assurance (VQA) standards.¹⁰

All adverse events were graded using protocol-specified, standard toxicity criteria for pediatric populations. Only those events occurring during therapy or less than 60 days after termination of initial study medication were included in the analysis.

A minimum of 80 children were planned to be enrolled in each of the 3 arms of this study. This sample size would ensure a power of 80% to detect a difference of one third in the average change in log₁₀ RNA copy number between 2 therapy arms using a 2-sided P = .05 level of significance. The primary end point of the study was initially specified to be the change in HIV RNA from baseline. However, since the majority of patients in the ritonavir arms were observed to have HIV RNA copy numbers below the level of assay quantification (<400 copies/mL) at the week 12 interim analysis, the primary end point for the study was changed to the proportion of children with values below the level of assay quantification, consistent with other, similar studies.

Comparisons among treatment groups used the Fisher exact test for categorical variables and Wilcoxon/Kruskal-Wallis test for continuous variables.^11- 12 An RNA assay result was categorized as below the level of quantification (undetectable) if the VQA-adjusted result was less than 400 copies/mL. Comparisons of the rates of undetectable RNA by baseline RNA values used the Fisher exact test for ordered categorical data.¹³ The baseline RNA value was defined to be the geometric mean of the preentry and the entry values. Patients with undetectable baseline RNA values could not contribute to the determination of the change from detectable to undetectable RNA and so were excluded from that analysis.

Some patients experienced virologic failure and discontinued their randomized study treatment. These patients remained on study follow-up but received nonprotocol alternative antiretroviral treatment that could affect their subsequent virologic evaluation during follow-up. To avoid confounding the study-related evaluations by such nonstudy antiretroviral treatment, a child was classified as having an undetectable level of RNA at a specific study week only if the child was still receiving initial randomized study treatment at that time. The number of patients on the treatment arm at baseline serves as the denominator for this conservative definition of response rate.

Periodic interim monitoring of this study by an independent study monitoring committee was planned at weeks 12 and 24 using protocol-specified criteria for the stopping or modification of a treatment arm. The criterion for stopping a specific treatment at the week 12 interim analysis was the detection of a statistically and medically significant difference in virologic efficacy between any 2 treatment arms; a virologically inferior arm was defined as a one in which the average change in HIV RNA copy number was at least 1.5 log₁₀ less than that in another arm or in which the proportion of children who achieved HIV RNA levels below quantification was at least 50% lower than that in another arm. All analyses were based on an intent-to-treat approach.¹⁴ All P values were 2-sided, were not adjusted for multiple comparisons, and were not adjusted for interim analyses.

Two hundred ninety-eight children entered the study between February 6 and April 30, 1997. One child was not included in the analysis because the child never started study therapy due to an illness (Figure 1). Baseline patient characteristics (Table 1) were well balanced among the treatment groups. The 11 children with prior stavudine experience who entered the study were equally distributed among the treatment arms. The median duration of follow-up was 12.7 months, varying between 12.4 and 12.8 months for the 3 treatment groups.

The week 12 interim analysis of the first 136 children enrolled was conducted on August 11, 1997. This analysis showed that the proportion of children whose HIV RNA copy number reached an undetectable level (<400 copies/mL) in the zidovudine and lamivudine group (14% [6/43]) was significantly less than in the 2 ritonavir-containing treatment groups (61% [28/46] in the triple-therapy group and 57% [27/47] in the dual-therapy ritonavir group; P<.001 for both comparisons). Based on this interim analysis, children in the zidovudine and lamivudine group with HIV RNA levels greater than 10,000 copies/mL were offered combination ritonavir, nevirapine, and stavudine treatment identified as the step 2 phase of the study; children who had HIV RNA levels of up to 10,000 copies/mL continued taking their original, randomized zidovudine and lamivudine treatment. Forty-eight children enrolled in step 2; 25% (12/48) enrolled in step 2 between study weeks 30 and 36 of their initial randomization, and the remainder enrolled after study week 36. Therefore, analyses of the zidovudine and lamivudine group patients in this article are only included through study week 24 (when all patients were still receiving their original, randomized treatment).

The overall week 12 analysis results are presented in Table 2. At study week 12, 12% of the zidovudine plus lamivudine group had undetectable plasma HIV RNA compared with 52% and 54% of those in the 2- and 3-drug ritonavir-containing arms, respectively (P<.001). The proportion of children (taking initial treatment) at study week 24 who had HIV RNA below the level of assay quantification was 8% (8 of 95) for the zidovudine and lamivudine group, 34% (31 of 92) for the stavudine and ritonavir group, and 47% (44 of 93) for the triple-therapy group. The pairwise differences between the zidovudine and lamivudine group and the other 2 treatment groups at study weeks 4, 12, and 24 were statistically significant (P<.001). For children in the 2 ritonavir-containing study arms, the differences between these 2 groups at study weeks 4, 12, 24, and 36 were not statistically significant. However, a significant difference was observed at study week 48. The proportion of children receiving initial treatment at study week 48 with HIV RNA below the level of quantification was 27% (25 of 92) for the stavudine and ritonavir group and 42% (39 of 93) for the triple-therapy group (P = .04). At the same time, the proportion of children with HIV RNA less than 10,000 copies/mL was similar in both ritonavir-containing treatment groups (P = .19).

We analyzed whether baseline HIV RNA copy number, CD4 cell count, or age were predictors of long-term virologic success in the ritonavir-containing treatment groups. In those children continuing their initial treatment at study week 48, only lower baseline HIV RNA copy number was associated with a greater proportion of children achieving HIV RNA levels below quantification (Table 3). Sixty-nine percent of children with baseline HIV RNA between 2.6 and 3 log₁₀ copies/mL achieved RNA levels below quantification at study week 48 compared with only 19% of children with baseline HIV RNA between 5 and 6 log₁₀ copies/mL. There was no statistically significant association of baseline CD4 cell count or age at study entry with virologic response (data not shown).

There was no difference in the median CD4 cell count for all 3 treatment groups combined between study entry and week 24 (671 vs 744 × 10⁶/L) (Figure 2). In the ritonavir-containing study arms, median CD4 cell counts at week 48 for the children receiving stavudine and ritonavir and those receiving triple therapy were 767 and 818 × 10⁶/L, respectively (P = .47). However, there was a significant difference in CD4 percentage between the 2 ritonavir-containing treatment groups at study week 48, with medians of 29% vs 33% for the 2-drug vs 3-drug combinations (P<.01).

Overall, 72% of the children experienced a moderate (grade 2) or worse toxic event while receiving initial therapy, and 21% experienced a severe (grade 3) or worse toxic event. However, there were no significant differences between the treatment groups with respect to the overall rates of toxicity. The overall rate of severe or worse toxic effects was 22% for those receiving zidovudine and lamivudine; 23% for the stavudine and ritonavir group; and 17% for the triple-therapy group. The most commonly observed adverse events were nausea or vomiting (24%) (grade 2 or higher: >3 episodes of vomiting per day, duration of >3 days, nausea resulting in decreased oral intake); rash (19%) (grade 2 or higher: diffuse maculopapular rash, dry desquamation, or worse); fever of 38.5°C or more (18%); and neutropenia (absolute neutrophil count <750 × 10⁶/L) (18%). Moderate or worse nausea or vomiting was more frequent in children receiving triple therapy, compared with those receiving zidovudine and lamivudine alone (Table 4). Neutropenia of a moderate or worse degree was less frequent in children receiving stavudine and ritonavir than either of the study arms that contained zidovudine.

Three reasons that children needed permanent discontinuation of their initial randomized study treatments were: (1) HIV RNA greater than 10,000 copies/mL, (2) adverse events or intolerance, or (3) other nontoxicity or virologic failure reasons. By study week 48, initial therapy had been discontinued prematurely for 62% of patients in the zidovudine and lamivudine group, 35% in the stavudine and ritonavir group, and 28% in the triple-therapy group (Figure 3). While intolerance was the most frequent reason for discontinuation of initial ritonavir therapy by study week 4, by study week 48 the rates of discontinuation due to intolerance were similar for the 3 treatment groups. Thirty-six percent of the 197 children in the ritonavir-containing arms were receiving full-dose therapy, and 31% were no longer receiving ritonavir at study week 48. Discontinuation of initial treatment due to an increase in HIV RNA copy number after study week 12 was not common in the zidovudine and lamivudine group until study week 36, at which time it became the predominant reason.

This study was the first large, randomized clinical trial to evaluate the use of ritonavir in HIV-infected children. Of the 297 evaluable children who entered the study, 197 received ritonavir-containing antiretroviral treatment regimens. After 12 weeks of therapy, a significant and sustained decrease in HIV RNA copy number was observed in children changed from their previous nucleoside analog antiretroviral therapy to a ritonavir-containing regimen. Fifty-three percent of children who received a ritonavir-containing regimen had HIV RNA levels below the limit of assay quantification at study week 12 compared with only 12% of children who had been randomized from their previous regimen to the zidovudine and lamivudine study arm (P<.001). Thus, changing from a single or dual nucleoside regimen to another dual nucleoside regimen had little virologic benefit for clinically stable pediatric patients. As a consequence, children randomized to the zidovudine and lamivudine study arm who had HIV RNA levels of more than 10,000 copies/mL after study week 12 were changed to a protease-containing regimen (stavudine, nevirapine, and ritonavir) in the step 2 phase of this study and are still undergoing follow-up.

After 48 weeks of therapy, 42% of the children receiving triple therapy maintained HIV RNA levels below the limit of assay quantification, compared with only 27% of children receiving dual therapy with ritonavir and stavudine. This difference was not noted at study weeks 24 and 36. In addition, there was no significant difference in the proportion of children with HIV RNA viral number below 10,000 copies/mL between the 2 ritonavir-containing treatment groups at study week 48. These data suggest that dual therapy including ritonavir has comparable activity to triple therapy including ritonavir in terms of moderate reduction in HIV RNA copy numbers (to levels <10,000 copies/mL) but that triple therapy with ritonavir was superior to dual therapy in maintaining HIV RNA copy numbers below the level of assay quantitation. Although initial evaluation at week 24 suggested that the 2-drug combination of stavudine and ritonavir was as effective as zidovudine, lamivudine, and ritonavir in effecting an undetectable viral load, this outcome did not appear to be fully sustained when a longer interval (48 weeks) was examined. It is important to note that even triple therapy was less than 50% effective in maximally suppressing the viral load. It is difficult to evaluate the effect of therapy on clinical outcomes, as few outcome events had occurred by 48 weeks. The effect of treatment must therefore be evaluated over extended periods before definitive conclusions can be drawn regarding any comparisons of highly active antiretroviral therapies. There were no cases of Pneumocystis carinii pneumonia reported during the study period. One case of cytomegalovirus disease was reported in the stavudine plus ritonavir treatment group and none in any of the other treatment groups.

Several reasons may have contributed to the relatively low success rate. First, children tend to have higher plasma HIV RNA levels than adults,¹⁵ and the higher the viral load, the lower the success rate appears. Second, although more than two thirds of the children were still taking their initially assigned ritonavir-containing treatment at study week 48, only half of them were still receiving the full dose of ritonavir, and resistant virus may have been present. Third, all these children were antiretroviral-treatment experienced and often only changed 2 drugs, not 3. Fourth, adherence to multiple-drug regimens is difficult in children. These factors may have contributed to the overall lower success rates when compared with adult antiretroviral trials.

There were no significant differences in absolute CD4 cell counts after 48 weeks of therapy between the 2 ritonavir-containing regimens; however, a greater increase in CD4 percentage was seen in children receiving triple therapy. This discrepancy between CD4 cell count and percentage suggests that CD4 percentage may be a more sensitive or accurate measure of immune response to antiretroviral therapy than absolute CD4 cell count when nucleoside analog therapy suppresses white blood cell counts over time. Another potential reason for the lack of a significant change in CD4 cell number between treatment regimens is that many children entered the study with absolute CD4 cell counts that were already normal for age (the median age at entry was approximately 7 years, and the median entry CD4 cell count in the study arms was 644 to 693 × 10⁶/L; normal for this age group is >500 × 10⁶/L). Additional studies are needed to investigate the importance of more specialized phenotypes within the larger CD4 cell group.

Ritonavir was relatively well tolerated by the children in this study. Seventy percent continued taking ritonavir for the 48 weeks of the study. While nausea and vomiting appeared to be the most common adverse effect in the ritonavir arms, taste was the biggest obstacle for the health care providers. To increase the palatability of the ritonavir, coating the mouth with peanut butter or with chocolate or vanilla pudding or applying ice chips were recommended. In many cases, these methods or others devised at individual sites were effective in permitting successful drug administration. When possible, capsules were substituted for the liquid formulation. A few children able to tolerate ritonavir early in the study were unable to tolerate it weeks or months later. Other toxic events noted in the study included fever and skin rash, which appeared at approximately the same rates in all 3 treatment groups, and neutropenia, which appeared more commonly in the zidovudine and lamivudine treatment arms. There were no differences overall in the rate of grade 3 or 4 toxic events across all treatment groups.

Baseline HIV RNA copy number was an important prognostic factor in virologic response to therapy. Sixty-nine percent of children who entered the study with HIV RNA levels under 1000 copies/mL achieved HIV RNA levels below the limit of assay quantitation by study week 48, compared with only 19% of children who entered with baseline HIV RNA levels greater than 100,000 copies/mL. A similar finding has been reported in HIV-infected adults¹⁶ but was not observed in a smaller study in children.¹⁷ These data suggest that change in antiretroviral therapy in children should be considered when HIV RNA levels are only moderately elevated, rather than waiting until HIV RNA levels become excessively high.

This study demonstrated that ritonavir-containing treatment regimens have potent antiviral effects, and, therefore, children who are nucleoside-experienced should be switched to a protease inhibitor–containing treatment regimen to successfully decrease their viral load. To extend the durability of the viral load suppression, 2 nucleosides instead of 1 should be part of the combination. Ritonavir was generally well tolerated and associated with a toxicity profile commonly seen with protease inhibitor therapy. While there appeared to be some late-onset problems with intolerance, 70% of children continued their ritonavir-containing treatment through study week 48. Protease inhibitor–containing combination therapy should be viewed as part of the standard therapy for children with HIV disease.