Efficacy and Toxicity of Antiretroviral Therapy Using 4 or More Agents: Application of a Strategy for Antiretroviral Management in Human Immunodeficiency Virus–Infected Children

Objective  To characterize the long-term tolerance and virologic efficacy of combination antiretroviral therapy consisting of 4 or more agents in a clinical setting.

Methods  An observational review of 36 children infected with human immunodeficiency virus 1 (HIV-1) treated with 4 or 5 antiretroviral agents in 2 university hospital clinics between April 1, 1996, and October 31, 2000. Highly active antiretroviral therapy regimens were chosen with regard to the child's past antiretroviral exposure or results of genotypic resistance data. Plasma HIV-1 RNA levels were monitored weekly to monthly after initiation of highly active antiretroviral therapy, and adherence efforts were actively supported and monitored.

Main Outcome Measure  Number of children with undetectable plasma HIV-1 RNA levels at longest follow-up.

Results  Four- or 5-drug highly active antiretroviral therapy reduced plasma HIV-1 RNA levels to less than 50 copies/mL in 32 (89%) of 36 children. After a median of 28.7 months of observation, 28 children (78%) remained at this level of suppression. Adverse reactions were limited to mild neutropenia and mild transient or persistent elevations in alanine aminotransferase levels in 11% of children.

Conclusions  Treatment with 4 or 5 antiretroviral agents was well tolerated in HIV-1–infected children and resulted in a high degree of viral suppression, even in children with previous antiretroviral drug experience.

MOST STUDIES^1-3 of combination antiretroviral therapy in human immunodeficiency virus 1 (HIV-1)–infected children published to date suggest that long-term control of viral replication is difficult to achieve in children, as only 27% to 48% experience suppression of plasma HIV-1 RNA levels to below the limits of detection when treated with "standard" highly active antiretroviral therapy (HAART) regimens that include 2 nucleoside reverse transcriptase inhibitors (NRTIs) and a protease inhibitor (PI). Recently, results of 2 clinical trials^4,5 using up to 4 antiretroviral agents demonstrated improved virologic outcome (60%-63% viral suppression at 6-12 months), suggesting that sustained viral suppression may be better achieved with multidrug regimens.

Concern has been raised, however, about potential increased toxic effects and difficulty in adherence to regimens that include more than 3 antiretroviral agents.⁶ Our clinical experience treating HIV-1–infected children with 4 or more antiretroviral agents suggests that these regimens are well tolerated and result in sustained viral suppression in most treated children.

All children with HIV-1 followed at the pediatric HIV clinics at Children's Hospital and Regional Medical Center (CHRMC), Seattle, Wash, and Oregon Health Sciences University (OHSU), Portland, treated with 4 or more antiretroviral agents between April 1, 1996, and October 31, 2000, were included in this review. This included children switching from 3-drug regimens owing to concerns about incomplete viral suppression and all children newly initiating HAART after mid-1997, except for 2 children who were randomized to a 3-drug arm of the Pediatric AIDS Clinical Trials Group (PACTG) 377.⁵ The multidrug HAART regimen for each child was chosen for ease of administration and for potency, with avoidance of agents to which the virus was unlikely to be susceptible based on previous use, known patterns of drug cross-resistance, and/or "resistance" mutations found on genotypic testing.^7,8 Six children received treatment as part of the PACTG protocol: 3 participated in the PACTG 382,⁴ 2 in the PACTG 377,⁵ and 1 in the PACTG 397.⁹ This retrospective review was approved by the institutional review boards of CHRMC and OHSU.

The level of HIV-1 RNA in plasma was determined for most children every 3 months before initiating combination antiretroviral therapy; at the start of therapy; weeks 1, 2, and 4 after initiating therapy; then monthly until undetectable for at least 2 consecutive months; and then every 2 to 3 months thereafter. Plasma RNA concentration was measured using a branched chain DNA assay (Chiron Corp, Emory, Calif) sensitive to 500 copies/mL before June 1997, and by reverse transcription polymerase chain reaction (Amplicor 1.0; Roche Molecular Systems, Somerville, NJ) sensitive to 400 copies/mL after that date. Beginning December 1997, plasma samples with RNA levels less than 400 copies/mL were quantified using the ultrasensitive reverse transcription polymerase chain reaction (Amplicor 1.0) sensitive to 50 copies/mL.

CD4 and CD8 lymphocyte counts were determined from blood samples collected in potassium-EDTA by standard 2-color fluorescent-activated cell sorter analysis using the Coulter EPICS XL Flow Cytometry System (Coulter Electronics, Miami, Fla).

Before initiating a combination regimen, the goal of complete viral suppression was discussed with families, including the need for close to 100% adherence to prevent the emergence of HIV-1 that is resistant to therapy. The families were actively involved in choosing the regimens with consideration for the lifestyle of the family and the medication-taking skills of the child. The families were provided with written instructions regarding the medications, including dosages, frequency, and dates of dosage changes. In most cases, the first dose of HAART was administered in the clinic. In addition, if the child had had difficulty tolerating liquid medications in the past, therapy was delayed until either the child was taught to swallow pills or a gastrostomy tube was placed. Personnel from the HIV team contacted the family by telephone frequently after the regimen was started. Most families were seen at 1- or 2-week intervals several times after initiation of therapy and then at least monthly until the child's RNA level was less than 50 copies/mL. Adherence to the regimen was discussed with the family by the authors (A.J.M., P.F.L., K.M.M., or L.M.F.) at each of these visits. The child's caretaker was asked about difficulties with medications during the past month and specifically about missed doses in the past week. Attempts were made to address any difficulties identified. Families were also provided with a data table showing the child's sequential RNA levels and CD4 cell counts, growth variables, and results of all tests performed to monitor drug toxic effects. Medications for the CHRMC cohort were filled on a monthly basis at the hospital pharmacy, and lapses in refills were recorded. At OHSU, medications were filled at outside pharmacies, with the clinic physician (P.F.L.) called for refills.

Blood specimens for a complete blood cell count, serum alanine aminotransferase measurement, and glucose and cholesterol (nonfasting) measurements were obtained to monitor for hematologic, hepatic, and metabolic toxic effects, respectively, every 4 to 16 weeks. Toxic effects were graded according to the Division of AIDS Toxicity Table for Grading of Pediatric Adverse Experiences. Clinical and laboratory adverse events are graded on a scale of increasing toxicity (grades 1-4).

When an increase in HIV-1 plasma RNA concentration was detected at greater than 50 copies/mL, the pol genotype of the rebounding virus was determined by direct consensus or end point sequencing of plasma viral RNA.⁷ Two milliliters of plasma was used when the viral copy number was less than 2000 copies/mL. The viral particles were extracted using silica.¹⁰ The entire protease gene and amino acids 1 through 375 of the reverse transcriptase gene were amplified through nested polymerase chain reactions after reverse transcription polymerase chain reaction using random hexamers. The amplified DNA was directly sequenced using the Biosystems Genetic Analyzer 310 (Perkin Elmer, Foster City, Calif) and was compared with wild-type HIV-1 (Consensus B from the Los Alamos database⁸).

All 36 children treated with 4- or 5-drug HAART were included in this analysis (27 from CHRMC and 9 from OHSU). Their median age at the start of therapy was 6 years (range, 0.2-16.0 years). Eight children were naïve to antiretroviral therapy before starting 4- or 5-drug HAART; the remaining children had previous experience with a variety of NRTIs (n = 14), NRTIs and PIs (n = 9), or NRTIs, nonnucleoside reverse transcriptase inhibitors (NNRTIs), and PIs (n = 4). Past treatment records were not available for 1 child. Twelve children had been treated initially with a 3-drug HAART regimen and were changed to a 4- or 5-drug regimen owing to either viral rebound (9 children) or concerns regarding the potency of their initial regimen (3 children). Six children (17%) were treated with a 5-drug regimen. The median time being treated with a 4- or 5-drug regimen was 28.7 months (range, 3.5-43.0 months). Twenty-eight (78%) of the 36 children were receiving 4- or 5-drug HAART for at least 18 months (median, 31 months).

Initial multidrug HAART regimens are detailed in Table 1. Children were treated with full recommended dosages of all agents. Nelfinavir was administered at 55 mg/kg every 12 hours and ritonavir at 400 mg/m² every 12 hours, except for the child younger than 2 years, who was given 450 mg/m². Saquinavir was initially administered at 350 mg/m² every 12 hours when given in conjunction with ritonavir until February 2000,¹⁰ when it was increased to 50 mg/kg every 12 hours.

The median RNA level before HAART was 4.9 log₁₀(range, 3.3-6.5 log₁₀) (Table 2). After initiation of HAART, all children experienced a decrease in plasma HIV-1 RNA concentration, with the plasma RNA nadir being below 50 copies/mL (1.7 log₁₀) in 32 children (89%). The median RNA level at the last follow-up during 4- or 5-drug HAART was 1.7 log₁₀(range, 1.7-5.8 log₁₀), with 28 children (78%) having RNA levels suppressed to below 50 copies/mL. Virologic rebound, defined as an increase in RNA levels to greater than 50 copies/mL on at least 2 time points 4 weeks apart or 1 value greater than 1000 copies/mL, occurred in 13 children during 4- or 5-drug therapy. Failure to achieve RNA suppression or viral rebound was associated with documented nonadherence, either chronic or episodic, in 10 of the 13 children.

In 5 of the 13 children, viral suppression was again achieved after adjusting their regimens. New resistance mutations did not develop at the time of the viral rebound (RNA increases to 3.1 and 3.4 log₁₀) in 2 children, and viral levels decreased to below 50 copies/mL with only minor modifications in their regimens (changing from nelfinavir to ritonavir in one child and increasing the dosage of saquinavir in the other). New resistance mutations were present in 3 of the 5 children at the time of viral rebound, which occurred in 2 of the 5 children after decreased adherence associated with emesis caused by an intercurrent illness. Suppression of plasma RNA levels to less than 50 copies/mL was again achieved in each of these 3 children after changing 2 to 4 drugs in their regimens. The HAART was suspended in the remaining 8 children at the time of rebound while issues relating to nonadherence were addressed.

Median CD4 cell count and percentage at baseline were 406 cells/µL (range, 7-3249 cells/µL) and 18.5% (range, 1%-41%), respectively (Table 2). After a median of 28.7 months of 4- or 5-drug HAART, the median CD4 cell count was 986 cells/µL (range, 349-2895 cells/µL) and the percentage was 34% (range, 9%-50%), representing a median increase of 339 cells/µL (range, –358 to 1845 cells/µL) and 12% (range, –7% to 37%), respectively. CD4 cell counts improved to or remained at age-appropriate levels for all children maintaining HIV plasma RNA levels less than 50 copies/mL during HAART.

Adherence, as assessed through parental report, was greater than 95% (missing only 1 or 2 doses a month) in 24 (67%) of 36 children. These families also were timely in obtaining medication refills. For the remaining families, either parental report or pharmacy refill data suggested chronic nonadherence (8 families), or the clinicians felt unable to judge the level of adherence (4 families). All of the children whose families reported chronic nonadherence experienced virologic rebound. As detailed in the "Virologic Responses After Initiation of HAART" subsection, viral rebound occurred in 2 children during adherence difficulties associated with an intercurrent illness. Both of these children again achieved viral suppression after instituting a new regimen. A variety of factors were associated with chronic poor adherence in the remaining families, including family disorganization, parental ambivalence regarding the value of HAART, and the child's dislike of taking medicines (liquid or pills). The number of antiretroviral agents in the regimen was specifically noted as a barrier to adherence for 5 families.

Gastrostomy tubes were placed for the purpose of medication administration in 7 children, and they were successful aids to adherence for 5 children, as assessed by the families and the clinicians. In these cases, the main obstacle to adherence was convincing the children to take unpalatable medicines. In the other 2 children, poor adherence was in large part due to ambivalence on the part of the caretaker, and sustained virologic suppression was not achieved in these children despite use of the gastrostomy tube.

Both the 4- and 5-drug regimens were well tolerated. Several children initiating therapy that included an NNRTI developed a mild to moderate rash. However, in all children the rash resolved without modification of therapy, and no child discontinued therapy because of rash. Biochemical and hematologic toxic effects were minimal, with no toxicities greater than grade 1 present at the latest point during 4- or 5-drug HAART (Table 3). Serum alanine aminotransferase level was elevated at the longest follow-up in 3 children (8%), although none above grade 1 toxicity (range, 33-147 U/L). Serum alanine aminotransferase level was moderately elevated (range, 165-331 U/L) for 16 weeks in an additional child treated with ritonavir; the elevation resolved spontaneously without any change in the medication regimen. Grade 1 neutropenia (absolute neutrophil count, 750-1200 cells/µL³) was present in 4 children (11%) at longest follow-up. These children developed persistent neutropenia during therapy that fluctuated between grades 1 and 2 (absolute neutrophil count, 400-1200 cells/µL³) without any clinical symptoms. No child developed significant anemia or thrombocytopenia during therapy. Grade 1 elevation in cholesterol level (measured on nonfasting blood specimens) (range, 171-499 mg/dL [4.42-12.90 mmol/L]) developed in 27 (79%) of 34 children.

We began using 4- and 5-drug HAART regimens after 4 of the first 10 children treated at CHRMC and OHSU with 3-drug HAART either did not achieve viral suppression or rebounded within the first 6 months of treatment.¹¹ In addition to increasing the number of antiretroviral agents included in a regimen, we modified our approach to initiating HAART in several other ways. Regimens were selected to include antiretroviral agents not previously used or unlikely to have high levels of cross-resistance based on genotypic analysis, and the agents were prescribed at the highest recommended dosages. Frequent monitoring of plasma HIV-1 RNA levels was instituted, and antiretroviral drug regimens were intensified soon after any detectable viral rebound to minimize the selection of mutations associated with high-level resistance to PIs. Adherence was facilitated by prescribing regimens that were given twice daily, maintaining frequent contact with the families, monitoring pharmacy refills, and evaluating and discussing adherence at every clinic visit.

Using this approach, 28 children (78%) treated with 4- or 5-drug regimens at CHRMC and OHSU achieved and maintained plasma HIV-1 RNA levels of less than 50 copies/mL during a median of more than 28 months of HAART. This occurred despite most children being highly antiretroviral drug experienced (28 children [78%] with previous exposure to multiple NRTIs and 9 [25%] with previous PI exposure), although data¹² suggest that the effectiveness of therapy is adversely affected by previous treatment with antiretroviral agents.

It is not clear which component of this strategy is the most important. In most previous studies,^1-3 the rate of suppression of viral replication with HAART in children has been low (<50% in most studies), and only short-lived suppression of viral replication has been documented in children. However, recently the PACTG 377 demonstrated an improved virologic outcome in children treated with 4 agents. Sixty percent of children treated with the combination stavudine, lamivudine, nevirapine, and nelfinavir had plasma RNA levels less than 400 copies/mL at 6 months.⁵ Similarly, improved outcome was seen in PACTG 382,⁴ an investigation of efavirenz and nelfinavir with 1 or 2 NRTIs. In this study, which in addition used therapeutic drug monitoring, 63% of children had RNA levels less than 50 copies/mL through 48 weeks.

In adults, the number of new drugs included in a regimen^13,14 and the number of agents to which the virus seems susceptible by genotypic testing^15,16 have been associated with the duration of viral suppression. More potent HAART regimens may be needed for children because of their higher viral burden,^17,18 relatively immature immune system,¹⁹ or unfavorable antiretroviral pharmacokinetics.^20,21 Alternatively, a 4- or 5-drug regimen may provide greater efficacy owing to the low probability of selection of virus resistant to all agents^22,23 or because it may allow for a greater degree of nonadherence.

Intensive virologic monitoring after initiation of a new antiretroviral regimen allows for rapid determination of response to therapy, documentation of complete suppression to below 50 copies/mL, and timely determination of viral rebound, thus allowing initiation of a salvage regimen before development of high-level PI resistance. Failure to achieve complete viral suppression has been associated with increased risk of virologic failure and development of viral resistance mutations.²⁴ The plasma RNA level and the presence of high-level PI resistance mutations have been associated with the failure of salvage therapy.^25,26 Documenting the decline of plasma RNA levels may also provide encouragement to children and their families because they can see the effects of their efforts to adhere to the medications.

Lack of adherence to the medication regimen has been cited as the primary reason for failure of HAART.^3,27,28 The level of adherence necessary to optimize virologic outcome has been shown to be 95% or greater in adults undergoing HAART.²⁹ Adherence to the prescribed therapy is complicated for children because of many factors: the caretaker's schedule must be meshed with that of the child, young children lack understanding and motivation to take the medications and may resist for behavioral reasons, and young children may have difficulty in ingesting large quantities of unpalatable liquids or large numbers of capsules or tablets. Watson and Farley³ found that only 58% of children in their clinic were adherent to their medications at a level of 75% or greater. We attempted to support and monitor adherence through frequent contact with the families, monitoring pharmacy refills, placement of gastrostomy tubes (particularly in younger children), and careful assessment of parental ability and commitment before initiation of HAART. Despite all of these efforts, nonadherence seemed to be the primary cause of viral rebound in 10 (77%) of the 13 children who experienced virologic rebound while taking 4 or 5 antiretroviral agents. Although most treated children were believed to be adherent to the level of greater than 95%, the accuracy of the assessment of level of adherence is limited by the use of self-report measures, which, in general, overestimate adherence.³⁰

Many health care workers and HIV-1–infected individuals are concerned that the increased potency potentially provided by a multidrug regimen might be offset by increased toxic effects and difficulty in adherence, and thus could diminish the ultimate efficacy of multidrug regimens. Although the children in our cohort experienced few adverse effects from the medications, as discussed previously, the absolute level of adherence could not be determined with confidence. It is possible that increased toxic effects from the 4- and 5-drug regimens would have been demonstrated if all patients achieved 100% adherence. Nevertheless, our clinic experience reported herein, and the higher continuation of the 4- vs 3-drug regimens in the PACTG 377,⁵ indicate that multidrug HAART regimens are well tolerated in children.

We believe that in addition to the use of multidrug HAART at the highest recommended dosages, the intensive monitoring of plasma HIV-1 RNA levels and active support of adherence was critical to the success of therapy in these children. We hypothesize that intensive monitoring allows timely intensification of regimens before the development of high-level resistance to PIs and assists families in affirming the benefit of their high-level adherence to therapy. The success of HAART in our clinic population suggests that HAART therapy can result in long-term virologic suppression in most children and that strategies critical to achieving a high level of treatment success should be investigated.

Accepted for publication February 7, 2002.

This study was supported in part by grants AI32910 and R01 HD36184 from the National Institutes of Health, Bethesda, Md.

Antiretroviral treatment of HIV-1–infected children is difficult, and most published studies using HAART have demonstrated suppression of HIV-1 RNA to undetectable levels in only 40% to 60% of treated children. In addition, previous studies have been short term, reporting on 24 to 48 weeks of therapy. We developed a treatment strategy using multidrug antiretroviral regimens, frequent monitoring of HIV-1 RNA levels, and aggressive adherence support. Although this is an observational review of our clinical experience, the results suggest that use of 4- or 5-drug regimens to treat HIV-1 in children is well tolerated and results in long-term viral suppression in most children.

Corresponding author and reprints: Ann J. Melvin, MD, MPH, Children's Hospital and Regional Medical Center, 4800 Sandpoint Way NE, CH-32, Seattle, WA 98105 (e-mail: amelvi@chmc.org).