Reply to El-Mallawany et al

We agree with El-Mallawany and colleagues [1] that Kaposi sarcoma (KS) causes a high burden of disease in human immunodeficiency virus (HIV)–infected children in Eastern and Southern Africa, the regions most heavily affected by the HIV/AIDS epidemic. Worldwide, 1.8 million children aged 0–14 years are HIV-infected and most of these (1.04 million; 58%) live in Eastern and Southern Africa [2]. HIV infection increases the risk of developing cancer by compromising the immune system and increasing susceptibility to oncogenic viruses [3, 4]. We recently showed that KS and non-Hodgkin lymphoma (NHL) are the cancers with the highest incidence in HIV-infected children in South Africa [5]. In a clinical trial in Malawi, median survival in HIV-infected children and adolescents with KS was <6 months [6]. Another study from South Africa showed that 10% of HIV-infected children with cancer died of treatment-related complications and severe infections [7].

Data on cancer burden and risk factors in HIV-infected children in African settings are scarce [8]. Even in well-funded Antiretroviral therapy (ART) programs in South Africa, cancer cases are frequently underreported. We previously linked data of 5 pediatric HIV programs participating in the International Epidemiology Databases to Evaluate AIDS (IeDEA) Southern Africa [9] with data of 4 referral pediatric oncology departments in South Africa using probabilistic record linkage methods [5]. We included data of 12448 HIV-infected children and identified a total of 47 prevalent and 24 incident cancer cases. Of all cancer cases identified, 15.5% were only identified in the HIV cohort, 15.5% were identified both in the HIV cohort and in the oncology departments, and 69% were identified in the oncology departments only. Including cases identified through record linkage increased the rate per 100000 person-years from 24 to 34 for KS and from 0 to 31 for NHL [5]. The risk of developing cancer was lower in children on ART (hazard ratio [HR], 0.29; 95% confidence interval [CI], .09–.86) than in children not on ART. Other risk factors were higher age (>10 vs <3 years: HR, 7.3; 95% CI, 2.2–24.6) and immunodeficiency at enrollment (advanced/severe vs no/mild: HR, 3.5; 95% CI, 1.1–12.0). In regions where pediatric oncology departments or functioning cancer registries exist, probabilistic record linkages with these data sources might be a valuable approach to improve cancer ascertainment. Alternatively, dedicated studies will be needed to better estimate cancer burden and monitor trends over time in HIV-infected children in African settings.

Notes

Financial support. Research reported in this publication was supported by the National Institute of Allergy and Infectious Diseases of the National Institutes of Health (award number U01AI069924); the National Cancer Institute (supplement to 5U01AI069924-07); and the Swiss National Science Foundation (Ambizione-PROSPER PZ00P3_160407 to J. B.).

Potential conflicts of interest. All authors: No potential conflicts. All authors have submitted the ICMJE Form for Disclosure of Potential Conflicts of Interest. Conflicts that the editors consider relevant to the content of the manuscript have been disclosed.