Of the 3553 HIV-infected children enrolled in PACTG 219 and 219C, 3193 were perinatally infected and 2398 of these (75%) had at least one neurological examination. The perinatally HIV-infected children with a neurological examination were generally similar to the 795 without, although those excluded were more likely to be female children, of older age, and to have higher CD4 cell percentage levels. At baseline, 126 cases of HIV encephalopathy were identified among 2398 children with a neurological examination, resulting in a prevalence of 5.3% [95% confidence interval (CI) 4.4–6.2%]. The median age at diagnosis of these prevalent cases was 1.7 years (Q1, Q3: 0.9, 3.9). Twenty of the 126 prevalent cases (16%) had initiated HAART before diagnosis, 60 (48%) had initiated non-HAART regimens, and 46 (36%) had not initiated any antiretroviral therapy before diagnosis of HIV encephalopathy. There were 42 deaths among the 126 prevalent cases with a median survival after diagnosis of 1.4 years (Q1, Q3: 0.7, 2.9). The majority of the deaths (62%) occurred among children who never initiated HAART during their lifetime.
The baseline characteristics of the 2272 children followed for incident HIV encephalopathy and overall survival analyses are provided in . Half of the children were female, 48% were less than or equal to 5 years of age at the time of their first neurological examination, 82% were born prior to 1995, 55% were black, 24% were classified as having low birth weight (<2500 g), and 19% had severe immunosuppression (CD4 cell <15%). Of the 1044 children with HIV-RNA information at baseline, 16% had viral load of at least 100 000 copies/ml. At the time of their first neurological examination, 35% of the children were on a HAART regimen and 27% were on a high CNS-penetrating regimen.
Characteristics of study population followed for incident HIV encephalopathy and overall survival at time of first neurologic examination (N = 2272).
Over a median duration of 6.4 years of follow-up, (Q1, Q3: 3.6, 9.9), 77 incident cases of HIV encephalopathy occurred, yielding an incidence rate of 5.1 cases per 1000 person-years (95% CI 4.0–6.3, person years: 15 178). The median age at diagnosis of the 77 incident cases was 6.3 years (Q1, Q3: 3.3, 11.4). In , the incidence rates and the percentage of children on HAART in the study population are summarized from 1994–2006. Beginning in 1996, there was a 10-fold decrease in the incidence of HIVencephalopathy followed by a relatively stable incidence rate after 2002. Conversely, there was a significant increase in the percentage of children on HAART regimens after 1996, suggesting an impact of HAART on the decreased incidence of HIV encephalopathy over time. By the end follow-up, 1806 children (79%) had initiated HAART and 31 of the 77 HIV encephalopathy cases (40%) were observed among them. Four hundred and sixty-six children never initiated HAART with 46 of the 77 HIV encephalopathy cases (60%) observed among them. Of the 1741 children who had initiated a high CNS-penetrating antiretroviral regimen by end of follow-up, 34 (2%) had a diagnosis of HIV encephalopathy. Twenty-four (9%) of the 267 children who ended follow-up on a medium CNS-penetrating regimen had a diagnosis of HIV encephalopathy, and 19 (7%) of the 264 children on a low CNS-penetrating regimen had a diagnosis of HIV encephalopathy by end of follow-up.
Incidence of HIV encephalopathy and percentage of children on HAART from 1994 to 2006.
Over a median follow-up of 6.5 years (Q1, Q3: 3.9, 10) from the first neurologic examination until death or censoring, 207 deaths were reported, resulting in a mortality rate of 13.5 per 1000 person-years (95% CI 11.7–15.4, person-years = 15 389). By the end of follow-up, 1826 children had initiated HAART and 101 deaths (6%) were observed among them. The number initiating HAART is higher than that reported above for the HIV encephalopathy incidence analysis as some children initiated HAART after their diagnosis. Four hundred and forty-six children never initiated HAART and 106 deaths (24%) were observed among them. Of the 1756 children who had initiated a high CNS-penetrating regimen, 111 (6%) had died by the end of follow-up. Twenty-six deaths (10%) were observed among 255 children on a medium CNS-penetrating regimen and 70 deaths (27%) were observed among 261 children on a low CNS-penetrating regimen.
The total person-time accrued for the 77 incident cases followed for survival was 219 person-years. Forty-three deaths were observed during follow-up resulting in a mortality rate of 196.3 per 1000 person-years (95% CI 142.1–264.5). Median survival after diagnosis was 2.0 years (Q1, Q3: 0.1, –). By the end of follow-up, 51 of the 77 children with a diagnosis of HIV encephalopathy had initiated HAART and 18 deaths were observed among them, as compared with 25 deaths among the 26 children who never initiated HAART. The numbers of deaths occurring by level of CNS-penetrating regimen were 19, 11, and 13 among 50, 12, and 15 children ending follow-up on a high, medium, or low-penetrating regimen.
Children who initiated HAART had a 50% lower risk of developing HIV encephalopathy compared with those who were not on HAART (hazard ratio 0.50, 95% CI 0.29–0.86) (). Baseline CD4 cell values less than 15% were associated with a greater than eight-fold increased risk of HIV encephalopathy (hazard ratio 8.41, 95% CI 4.79–14.76). Age less than or equal to 1 year at first neurologic examination was also independently associated with a greater than three-fold increase in HIV encephalopathy with a hazard ratio of 3.38 (95% CI 1.36–8.44). In the secondary analysis restricted to the subset of the population with HIV-RNA measurements, the HIV encephalopathy hazard ratio comparing HAART to no HAART was even stronger, 0.32 (95% CI 0.12–0.85) after adjustment for viral load. We also considered adjustment for calendar year, but treatment effects were attenuated and became nonsignificant due to the high correlation between calendar year and introduction of HAART (analyses not presented).
Estimated effects of HAART and central nervous system-penetrating antiretroviral regimens on incident HIV encephalopathy.
High CNS-penetrating regimens, as defined by CNS-penetrating scores, were associated with a 41% reduced incidence of HIV encephalopathy compared with low-penetrating regimens, although this association was not statistically significant (hazard ratio 0.59, 95% CI 0.31–1.10; ). After adjustment for viral load, the effect of high CNS-penetrating regimens on incidence of HIV encephalopathy was even stronger compared with low CNS-penetrating regimens (hazard ratio 0.41, 95% CI 0.13–1.29). Given the restricted sample size in this subanalysis, this large protective association was not statistically significant. Similar to the HAART analyses, these treatment effects were attenuated with adjustment for calendar year due to the correlation between calendar year and introduction of effective CNS-penetrating regimens (analyses not presented).
In the overall population, HAART and effective CNS-penetrating regimens were associated with increased survival compared with no HAART and low CNS-penetrating regimens, respectively (). Children with a diagnosis of HIVencephalopathy, however, had 12 times the risk of death compared with children without a diagnosis of encephalopathy (hazard ratio 12.42, 95% CI 8.46–18.24). Among the 77 children with an incident diagnosis of HIVencephalopathy, HAART use halved the risk of death after diagnosis compared with non-HAART regimens (hazard ratio 0.51, 95% CI 0.25–1.05), but use of high CNS-penetrating regimens conferred a larger survival benefit (74% reduction in risk of death) after HIV encephalopathy diagnosis compared with low CNS-penetrating regimens (hazard ratio 0.26, 95% CI 0.11–0.61; ).
Estimated effects of HAART and central nervous system-penetrating antiretroviral regimens on overall survival and survival after diagnosis of incident HIV encephalopathy.