Comparative dosimetric analysis of hippocampal-sparing vs conventional craniospinal irradiation in central nervous system leukemia: A single-center study

To evaluate dosimetric differences between hippocampal-sparing (HS) and conventional craniospinal irradiation (CSI) in central nervous system leukemia (CNSL) patients. Methods This analysis included 40 CNSL patients (25 males, 15 females; 10–45 years with a mean age of 30.35 ± 8.02 years) who underwent CSI at Chengdu BOE Hospital between September 2021 and December 2024. Patients were stratified into HS (observation group) and non-HS (control group) cohorts. Comparative dosimetric parameters included target coverage homogeneity index (HI), hippocampal dose constraints (Dmax, Dmean, Dmin, D40 %), and critical organ doses. Results Both groups demonstrated comparable mean radiation doses (control: 23.47 ± 3.23 Gy vs observation: 23.38 ± 3.21 Gy, P > 0.05). The HS technique exhibited differences in target homogeneity: maximum dose (24.23 ± 3.37 Gy vs 25.21 ± 4.17 Gy), D2 % (23.96 ± 3.29 Gy vs 24.12 ± 3.24 Gy), D98 % (23.02 ± 3.07 Gy vs 19.03 ± 2.42 Gy), and HI (0.04 ± 0.01 vs 0.22 ± 0.06) (all P < 0.05). Most notably, the observation group demonstrated significant dose reductions: - Dmax: 14.24 ± 2.00 Gy vs 23.95 ± 3.33 Gy (Δ9.71 Gy). - Dmean: 9.20 ± 0.97 Gy vs 23.54 ± 3.29 Gy (Δ14.34 Gy). - Dmin: 7.30 ± 0.82 Gy vs 23.03 ± 3.19 Gy (Δ15.73 Gy). - D40 %: 9.29 ± 0.96 Gy vs 23.86 ± 3.31 Gy (Δ14.57 Gy). (all P < 0.001).No significant differences were observed in the brainstem doses between the groups. Conclusion This study demonstrates the dosimetric feasibility of HS-CSI for patients with CNSL, achieving significant hippocampal dose reduction while ensuring adequate target coverage. These findings provide a crucial technical foundation and justify the initiation of prospective clinical trials. The critical next step is to validate whether these dosimetric advantages translate into tangible benefits in neurocognitive preservation and quality of life for this vulnerable patient population.