| Authors |
London, Wendy B ; Villanueva, Gabriela ; Shyr, Derek ; van Heerden, Jaques ; Fuentes-Alabi, Soad ; Saab, Raya ; Harrison, Derek ; Papyan, Ruzanna ; Kim Nguyen, Hoa Thi ; Noun, Dolly ; Kambugu, Joyce ; Rascon, Jelena ; Garami, Miklos ; Guo, Dongjing ; Kao, Paige ; Naranjo, Arlene ; Park, Julie R ; Cohn, Susan L ; Rodriguez-Galindo, Carlos ; Matthay, Katherine K |
| Abstract [eng] |
PURPOSE: Risk/treatment stratification for children with neuroblastoma (NB) relies on clinical, histologic, and genomic factors. However, most children with cancer live in low- and middle-income countries (LMIC), where access to advanced methods for stratification is limited. To address this unmet need, we developed a novel risk/treatment classification, the Adaptive Clinical Neuroblastoma Risk Groups (ACNRG) using clinical prognostic biomarkers. PATIENTS AND METHODS: A survival tree regression analysis of the International Neuroblastoma Risk Group (INRG) Data Commons (N = 14,501, diagnosed 1990-2014) was performed using univariate Cox regression models (age, International Neuroblastoma Staging System, serum lactate dehydrogenase [LDH], and serum ferritin) of event-free survival (EFS), separately for test and validation sets. Within each terminal node of the survival tree, the proportion of patients by initial treatment assignment and outcome achieved on that treatment were used to subjectively assign risk/treatment intensity (low-, intermediate-, and high-risk). For additional validation, the ACNRG was descriptively compared with INRG classification. Guidelines were developed for determining INRGs Staging System (INRGSS) in LMIC, using the minimum essential versus optimal imaging/biopsy procedures. RESULTS: Twelve statistically, clinically significant unique pretreatment risk groups of INRGSS/age/LDH/ferritin were identified (5-year EFS): low-L1/any/any/any (92% ± 0.5%); intermediate-L2/<18 months/<1,400/any (88% ± 1%), MS/any/<1,400/any (86% ± 1.5%), M/<12 months/<1,400/any (76% ± 2.3%); intermediate/high-L2/<18 months/≥1,400/any (73% ± 4.7%), L2/≥18 months/<1,400/<30 (68% ± 3.4%), L2/≥18 months/<1,400/≥30 (59% ± 3.7%), MS/any/≥1,400/any (52% ± 6.3%); high-L2/≥18 months/≥1,400/any (46% ± 4.7%), M/12-18 months/<1,400/any (64% ± 4.1%), M/<18 months/≥1,400/any (60% ± 1.6%), M/≥18 months/any/any (28% ± 0.8%). The concordance and discordance rates of ACNRG versus INRG were 86.6% and 13.4%, respectively (n = 8,152 nonmissing-data intersection). CONCLUSION: The ACNRG classification, using easily obtained clinical markers, is highly prognostic. The ACNRG could transform risk and treatment stratification, improve accuracy of treatment intensity decisions, and potentially improve outcome, for the large number of children with NB in LMIC. Prospective validation of the ACNRG classification is planned in a pilot trial. |
