2432 - Reirradiation for Pediatric Central Nervous System Tumors: Validation of an LQ Model-Based Approach for Critical Structure Dose Constraints and Clinical Outcomes
Presenter(s)
N. Martynova1, V. Nikolay1,2, M. Linnik1, G. Andreev1, A. Lyubinskiy1, N. Kataev1, and K. Suprun1,2; 1Dr. Berezin Medical Institute, Saint-Petersburg, Russian Federation, 2Saint-Petersburg State University, Saint-Petersburg, Russian Federation
Purpose/Objective(s): The management of recurrent brain tumors in children poses significant challenges due to the balance between therapeutic efficacy and treatment-related toxicities. We represent the management of CNS reirradiation in single institution through comparison of LQ-based model approach and real clinical practice
Materials/Methods: Forty pediatric patients received CNS reirradiation at a single institution (2018–2023), with a median interval of 21.9 months between courses. A novel LQ model incorporating time-dependent neural tissue recovery was developed to calculate cumulative biologically effective dose (BED) limits for critical structures. Retrospective analysis included 29 patients with complete dosimetric and outcome data. Prescribed doses were compared to LQ-derived constraints, and clinical outcomes (local control, radiation necrosis, neurological deficits) were correlated with dose deviations.
Results: Median follow-up was 13.7 months. Twenty-six patients had dose exceeding to one or more critical structures compared to LQ-model calculations. Meanwhile 15 patients had doses to brainstem and spinal cord lower by more than 15%, accompanying target undercoverage in 6 patients. Eight patients had local recurrence, 5 of them had target undercoverage. Nine patients (31%) developed radiation necrosis in reirradiated area. Frequency of neurodeficiency correlated significantly with necrosis (r=0.489; p=0.033). Radiation necrosis arose in patients with doses higher than retrospectively calculating levels (OR=12.8 [95% CI: 1.2–130.5]; p=0.03). ROC-analysis showed the increasing risk of necrosis in periventricular and brainstem area, when prescribed dose exceeded calculated dose by =35.2% (AUC=0.919±0.061 [0.799–1.000]; p=0.002). In three patients prescribed doses did not exceed calculated levels for any structure. No neurodeficiency observed in this group of patients.
Conclusion: LQ model-derived dose constraints appear safe, with adherence mitigating neurological toxicity. Exceeding constraints by =35.2% significantly elevates radiation necrosis risk, particularly in periventricular/brainstem regions. These findings support integrating LQ-based models into reirradiation planning to optimize therapeutic ratios. Prospective validation is warranted to refine thresholds and assess long-term functional outcomes.