Malignancy in perinatally human immunodeficiency virus-infected children in the United States

Objectives: To determine the incidence of and factors associated with malignancy in perinatally human immunodeficiency virus (HIV)-infected children in the United States. Methods: Included were 2969 children followed in the Pediatric AIDS Clinical Trials Group (PACTG) 219/219C cohort from 1993 through 2003. Cancer incidence by sex, race, age, histology and highly active antiretroviral therapy (HAART) era (pre-HAART, 1993-1997; HAART, 1998-2003) was estimated, and the standardized incidence ratio contrasting infected and uninfected children was determined. Poisson regression was used to further investigate the relation between HAART use (≥3 drugs of ≥2 classes, 1 of which was a protease inhibitor), CD4⁺% and cancer. Results: There were 37 cancers (17 prevalent and 20 incident) diagnosed in 2969 children for a prevalence of 0.6% [95% confidence interval (CI), 0.3, 0.9] and an incidence of 1.56/1000 person-years (95% CI 0.95, 2.41). Compared with uninfected children, the standardized incidence ratio was 10.08 (95% CI 5.87, 16.14). Incidence did not significantly differ by sex, race, age or HAART era. Of the cases, 35% were immunocompetent (CD4⁺ ≥25%), 25% were moderately immunosuppressed (15%≤ CD4⁺ ≤24%) and 40% were severely immunosuppressed (CD4⁺ <15%) at diagnosis. In multivariate regression, the cancer rate was 3.09 (95% CI 1.22, 7.85) times higher in children with ≤2 years of HAART use than in children with >2 years of HAART and 3.20 (95% CI 1.32, 7.76) times higher in children with CD4⁺ < 15% at cohort enrollment than in children with CD4⁺ ≥15%. Conclusion: Cancer incidence in this U.S. pediatric cohort was lower than that of European cohorts but was markedly higher than that of HIV-uninfected children. Cancer incidence was highest in children who were severely immunosuppressed and in children who received HAART for ≤2 years.