HIV was successfully genotyped in 124 of 127 children with sufficient pretreatment plasma at viral loads ≥100,000 copies/mL (63 children in the nevirapine arm and 61 children in the lopinavir/ritonavir arm; ). Similar to the parent trial [17
], these 124 children started therapy at median 9 months of age (interquartile range [IQR], 5–14 months; ) and received induction HAART with lopinavir/ritonavir for median 9 months (IQR, 7–12 months) before randomization. The presence of archived nevirapine-resistant HIV in cells while viremia was suppressed with induction HAART was assessed in 96 children (53 in the nevirapine arm and 43 in the lopinavir/ritonavir arm; ) who also had paired plasma genotypes. Pretreatment viral loads, CD4+
T cell percentages, World Health Organization (WHO) clinical staging, and months of receipt of induction HAART were similar for the 2 treatment groups ().
Nevirapine Resistance in Plasma Prior to HAART
For analysis of plasma genotypes, a median of 10,868 sequence reads per patient (IQR, 7715–14,551 sequence reads per patient) were obtained by means of ultradeep pyrosequencing. The prevalence of nevirapine resistance prior to HAART was 33% (41 of 124 children; ). The median frequency of nevirapine-resistant virus in plasma was 25.6% (IQR, 14.4%–73.7%). Most children (32 [78%] of 41) had only 1 resistance mutation; however, 8 (20%) had 2 mutations and 1 child had 3 mutations.
Figure 2A. Overall prevalence and median frequency of nevirapine-resistant human immunodeficiency virus (HIV) in the pretreatment plasma virus population in children initiating highly active antiretroviral therapy (HAART) between 2 and 24 months of age. Also shown (more ...)
As expected, the persistence of nevirapine-resistant HIV in plasma differed significantly by age. Younger children starting HAART had a significantly higher prevalence of resistance: 16 (48%) of 33 infants aged <6 months, compared with 19 (37%) of 51 and 6 (15%) of 40 of children aged 6–12 and >12 months, respectively (P = 0.005; ). However, when resistance was detectable in plasma, there was a trend toward higher frequencies in older children. The median frequencies of nevirapine-resistant virus were 18.9% (IQR, 10.5%–52.8%) for children aged <6 months, 25.6% (IQR, 11.2%–92.3%) for children aged 6–12 months, and 62.5% (IQR, 31.4%–98.6%) for children aged >12 months (P = .23; ).
In plasma, Y181C was the most common mutation detected and was present in samples from 31 (76%) of 41 children with resistant virus (median frequency, 25.6% [IQR, 13.4%–71.9%]). The K103N mutation was detected in samples from 27% of children (median frequency, 14.6% [IQR, 1.3%–82.7%]). Three other nevirapine resistance mutations were also detected, although at relatively low levels: G190A/E (15% prevalence; median frequency, 4.4% [IQR, 2.1%–12.4%]), V106A/M (7% prevalence; median frequency, 1.5% [IQR, 1.2%–7.4%]), and K101E (in 1 child in 3.0% of plasma virus).
Archived Nevirapine Resistance during Suppressive Induction HAART
From a median of 12,696 sequence reads per patient (IQR, 10,630–17,728 sequence reads per patient), archived nevirapine resistance mutations were detectable in 59 (61%) of 96 children at a median of 18 months of age (IQR, 15–25 months; ). The median frequency of cells with nevirapine-resistant HIV during suppressive induction HAART was 13.6% (IQR, 3.3%–52.2%). Moreover, the frequencies of infected cells with nevirapine resistance directly correlated with frequencies of resistant virus in plasma prior to therapy (r = 0.580; P < .001). As observed in plasma, most children (39 [66%] of 59) had 1 nevirapine resistance mutation; 13 (22%) had 2 mutations, 6 (10%) had 3 mutations, and 1 child had 4 mutations.
We did not detect a difference in prevalence of archived nevirapine resistance according to age. Fourteen (54%) of 26 children aged <6 months, 27 (73%) of 37 children aged 6–12 months, and 18 (55%) of 33 children aged >12 months at the start of HAART had archived nevirapine-resistant HIV (P = .19; ). However, younger children had significantly higher frequencies of infected cells with resistant virus: a median of 52.3% (IQR, 7.6%–67.9%) in children initiating therapy at age <6 months, 18.3% (IQR, 4.5%–43.2%) in children initiating therapy at age 6–12 months, and 3.8% (IQR, 2.4%–10.5%) in children initiating therapy at age >12 months (P = .03; ).
All 8 nevirapine resistance mutations [22
] were detected in cells. As observed in plasma, Y181C was the most frequent mutation detected and was present in specimens from 34 (58%) of 59 children with resistance. The median frequency of HIV-infected cells with Y181C was 22.4% (IQR, 6.5%–52.5%). G190A/E mutations were also commonly detected in cells (44%) and were present in a median of 2.6% of infected cells (IQR, 1.3%–3.5%). The K103N mutation was present in 14% of children in a median of 37.0% of infected cells (IQR, 4.0%–76.4%), and V106A/M mutations were present in 17% of children in a median of 4.8% of infected cells (IQR, 1.5%–28.5%). The mutations L100I, K101E, V108I, and Y188C/H were detected in 2%–7% of children at median frequencies ranging from 1.2% to 7.3% of cells.
Effect of Nevirapine Resistance on Virologic Control with Nevirapine-based HAART
As in the parent trial [17
], confirmed viremia of more than 1000 copies/mL by 52 weeks was more common in children who switched to nevirapine-based HAART than in children who continued to receive lopinavir/ritonavir–based HAART (14 [22%] of 63 vs 2 [3%] of 61; P
= .002). We examined whether frequencies of nevirapine-resistant HIV in plasma prior to therapy would adversely affect virologic suppression after switch to nevirapine-based HAART. We did not detect a difference in the risk of virologic failure for children with nevirapine resistance at ≥1% frequency compared with those without detectable resistance (HR, 2.0 [95% CI, 0.7–5.9]; P
= .19). ROC analysis was used to identify a threshold frequency of nevirapine-resistant virus in plasma of 25% to maximize the sensitivity and specificity of predicting failure in nevirapine-treated children despite initial treatment with the lopinavir/ritonavir regimen. Indeed, children who had pretreatment nevirapine resistance at or above this threshold frequency of 25% had a 3.5-fold higher risk of virologic failure than children without resistance (95% CI for HR, 1.1–10.8; P
= .03). There was no suggestion of a difference in the risk of failure between children with resistance in plasma below 25% frequency and those without resistance (HR, 0.9 [95% CI, 0.2–4.5]; P
= .91). Of the 13 children with nevirapine resistance at ≥25% frequency (median frequency, 83.9% [IQR, 38.9%–95.8%]), 6 (46%) experienced failure, whereas 2 (15%) of 13 children with <25% frequency of resistance (median frequency, 14.4% [IQR, 3.0%–16.8%]) and 6 (16%) of 37 children with no resistance experienced failure on nevirapine-based therapy. The observed differences in the time to virologic failure for these 3 groups are displayed in .
Figure 3. Kaplan-Meier estimates of virologic suppression (<1000 copies/mL at ≥2 visits) by 52 weeks after switch to nevirapine-based highly active antiretroviral therapy (HAART) for children in the study with resistance at ≥25% frequency (more ...)
Archived nevirapine resistance detected in cells just prior to randomization was not predictive of virologic failure for patients receiving nevirapine-based HAART (HR, 1.2 [95% CI, 0.3–4.5]; P = .77). Unlike for resistance detected in plasma, we were unable to identify a threshold frequency of nevirapine-resistant cells associated with failure. The frequencies of nevirapine-resistant cells were, however, linked with the frequencies of resistant virus maintained in pretreatment plasma for children treated with nevirapine-based HAART. All 10 of the children who had a frequency of resistance in plasma of more than the threshold of 25% had detectable archived resistance in a median of 75.3% of infected cells (IQR, 49.6%–87.7%). This archived resistance was higher in prevalence and frequencies than in children who started HAART with <25% of plasma virus with nevirapine resistance (10 [77%] of 13 children; a median of 23.1% of infected cells [IQR, 7.6%–52.2%]) and without plasma resistance (17 [57%] of 30 children; a median of 4.1% of infected cells [IQR, 2.4%–7.6%]; P = .02).
We also examined whether other factors were predictive of virologic control with the switch to nevirapine-based HAART. Pretreatment viral loads and CD4+ T cell percentages were not significantly associated with virologic failure. Children who started induction HAART by 6 or 6–12 months of age were also not at an increased risk of virologic failure compared with children who started at age >12 months (<6 months: HR, 0.8 [95% CI, 0.3–2.5]; P = .73; 6–12 months: HR, 0.2 [95% CI, 0.02–1.6]; P = .12). Even when age at start of HAART was included in multivariate analysis, the association between plasma resistance at ≥25% threshold frequency and virologic failure persisted (HR, 6.4 [95% CI, 1.5–27.4]; P = .01).
Antiretroviral Drug Resistance Mutations among Children with Virologic Failure
HIV genotypes were amplified from specimens obtained from 12 of 14 children who experienced failure on nevirapine-based therapy (). Overall, adherence was considered suboptimal (<95% adherence) for only 2 of 12 patients who experienced failure (patients 2 and 11). Seven (58%) of the 12 children experienced failure with mutations that were either detected in pretreatment plasma or archived in cells before reexposure to nevirapine. Three children experienced failure with nonnucleoside reverse transcriptase inhibitor (NNRTI) mutations not detected in pretreatment plasma or cells (patients 1, 4, and 5), and 2 with wild-type virus (patients 7 and 12). Ten children (83%) experienced failure with Y181C, and 5 children with mutations not typically seen during failure involving nevirapine (K101E, V106A/M, and G190A; ). The lamivudine resistance mutation, M184V, was detected in 8 children (75%), but none experienced failure with thymidine analog mutations ().
Nevirapine Resistance in Pretreatment Plasma, Long-lived Cells, and at Rebound Viremia for Children Experiencing Virologic Failure after Switch to Nevirapine-Based Highly Active Antiretroviral Therapy