Population-based data on cancers associated with acquired immunodeficiency
syndrome (AIDS) in children are lacking.
To determine risk of pediatric AIDS-associated cancers.
Design, Setting, and Participants
Using records from 11 locations in the United States for varying periods
between 1978 and 1996, we linked data for children aged 14 years and younger
at AIDS diagnosis to local cancer registry data.
Main Outcome Measures
Cancer frequency and, in the 2-year post–AIDS onset period, cancer
incidence and relative risk (RR; measured as standardized incidence ratio),
by cancer type.
Among 4954 children with AIDS, 124 (2.5%) were identified as having
cancer before, at, or after AIDS onset, including 100 cases of non-Hodgkin
lymphoma (NHL), 8 of Kaposi sarcoma (KS), 4 of leiomyosarcoma, and 2 of Hodgkin
disease; there were 10 other or unspecified cancers. Expected numbers for
all cancers identified in the study sample, based on population rates (using
area-specific registry data), were less than 1. In the first 2 years after
AIDS diagnosis (5485 person-years), NHL incidence was 510 per 100,000 person-years
(RR, 651; 95% confidence interval [CI], 432-941). Median time for developing
NHL after AIDS diagnosis was 14 months (range, 3-107 months). The most common
type of NHL was Burkitt lymphoma. However, the risk of primary brain lymphoma
(91 per 100,000 person-years) was especially high (RR, 7143; 95% CI, 2321-16,692),
and 4 cases were diagnosed more than 2 years (range, 37-98 months) after AIDS
onset. Leiomyosarcomas also tended to occur several years after AIDS onset,
with 3 of the 4 cases occurring 33 to 76 months after AIDS diagnosis, whereas
KS was reported only at or within 2 years of AIDS diagnosis. Hodgkin disease
risk was also significantly increased (RR, 62; 95% CI, 2-342).
The spectrum of AIDS-associated pediatric cancers resembled that seen
in adults, with the addition of leiomyosarcoma. Both primary brain lymphomas
and leiomyosarcomas tended to occur in children surviving several years after
AIDS onset. Because the expected numbers of these cancers in this population
were less than 1 and because of the small numbers of some types of observed
cancers, the RR estimates are imprecise and caution is warranted in their