ESTRO 2024 - Abstract Book

S2057

Clinical - Paediatric

ESTRO 2024

Material/Methods:

The study devised the Alopecia_Hope (A_H) contouring protocol on CT scans using RayStation 12A-SP1 software, identifying two scalp volumes: the whole scalp (WS) and posterior cranial fossa scalp (PCFs). After selecting patients for creating WS and PCF atlases, and tester patients, excluding those with notable post-neurosurgery bone changes, contouring time was recorded and analyzed, alongside performing statistical analysis using metrics like 3D DICE similarity coefficient (DSC), mean distance to agreement (MDA), and time saved percentage. Post validation, patients who underwent craniospinal irradiation from 2006 to 2021 at the specified hospital were selected, focusing on pediatric patients with a minimum of 12 months follow-up and known alopecia outcomes. The PAST Protocol was utilized to identify WS and PCFscalp volumes for each patient. Various data including age at diagnosis, disease specifics, surgical procedures, risk levels, high-dose Thiotepa usage, and radiotherapy timing were analyzed, alongside examining endocrinological effects and alopecia onset timing post-radiotherapy. Dosimetric variables were analyzed and recalculated using automated scripts, focusing on permanent alopecia lasting at least a year. Statistical analysis was conducted on collected demographic, clinical, and molecular data using Chi-square and Mann-Whitney U tests for nominal and ordinal variables, and Cox regression for univariate analysis. In the WS and PCF groups, ROC and AUC were used to identify dosimetric cut-offs. Only for the WS group, the analysis was repeated, stratifying it based on Thiotepa administration and daily radiotherapy fractions. Additionally, we created a modelling of WS permanent alopecia on the dose-volume histogram (DVH). PAST study: 52 CT scans were collected, median contouring time with the guidelines, the A_H protocol and automatic scripting has been 19 minutes. 8 testers were selected and mean contouring time has been 17.5 min . The contouring time by the automatic atlas was 11 min and the mean time of the expert correction was 4.5 min. Considering the work done by the atlas is in background, this resulted in a time savings of 80% for the operator. The WS DICE has resulted 0,76 +/- 0,05 and the PCF DICE has resulted 0,77 +/- 0,04. The cohort of HOPE project consisted of 48 patients, of which 18 were females (37.5%) and 30 were males (62.5%). The average age at diagnosis was 8.6 years, with a standard deviation of 4.68 years. 26 (54.2%) patients did not present with alopecia in WS (Whole Spine) or PCF (Posterior Cranial Fossa), 11 (22.9%) showed G1 alopecia in WS, and another 11 exhibited G2 alopecia. In PCF, 7 (14.6%) patients exhibited G1 alopecia, and 15 (31.3%) displayed G2 alopecia. For the study of the WS cohort and the PFC cohort, in the univariate analysis, Thiotepa p0.010, the number of radiotherapy fractions per day 0.04, as well as many dosimetric variables, turned out to be statistically significant. The results of the ROC (Receiver Operating Characteristic) analysis and the AUC (Area Under the Curve) for the entire scalp skin and PCF are reported in Table 1. Various dosimetric and volumetric variables have been identified with their cutoffs. Results:

In the ROC curves stratified by Thiotepa and the number of daily fractions, we have not identified statistically reliable cutoffs.

Conclusion:

The PAST project offers a more efficient and precise method for identifying the scalp and reducing the workload of the radiation therapist. The identification of dosimetric cutoffs can influence the prevention of permanent alopecia. Whole brain irradiation of the scalp at lower parameters than the PCF carries a higher risk. These results of HOPE to be PAST protocol could be the pilot for a larger retrospective and prospective study that allows for dose stratification also based on the presence or absence of other risk factors such as high-dose chemotherapy.