Combining Clinical and Dosimetric Features in a PBS Proton Therapy Cohort to Develop a NTCP Model for Radiation-Induced Optic Neuropathy

Assessment of Radiation-Induced Optic Neuropathy in a Multi-Institutional Cohort of Chordoma and Chondrosarcoma Patients Treated with Proton Therapy

“…Older age, hypertension, and tumor involvement with the optic apparatus have been significantly associated with the onset of toxicity, confirming previous reports for both photon and proton-based radiotherapy [ 5 , 6 , 9 , 19 , 20 ]. In contrast to previous reporting [ 5 , 10 ], and although both sexes were equally represented (152F /137M) in our cohort, sex was not significant in our dataset. In addition, we found that the number of surgeries (>3) prior to therapy was associated with RION in the univariable correlation analysis.…”

“…A patient cohort with a higher number of surgeries would be needed for further confirmation of the significance of this as a risk factor for RION. Interestingly, and in agreement with previous studies from our group [ 10 ], no dose metric was found to be significantly associated with RION other than the inclusion criterion of at least 45 Gy RBE being planned to the optic apparatus. While this cut-off has been set arbitrarily, there is general agreement that high doses are a prerequisite for the onset of RION [ 4 , 9 , 21 ], and toxicity has rarely been reported in clinical studies for doses below 45Gy [ 6 ].…”

“…In a previous publication, we introduced two RION NTCP models [ 10 ] from a cohort of skull-base and head and neck cancer patients treated with PBS proton therapy based on age, tumor involvement, and either sex or hypertension, respectively. Although the present cohort is limited to chordoma and chondrosarcoma, and hence the inhomogeneity of populations, prevents its use for the external validation of the models mentioned above, it is nevertheless possible to confirm some of the hypothesis and risk factors discussed in our previous study.…”

“…In particular, since no other dose metric was found to influence the risk for RION other than the inclusion criterion, our speculation that above a certain level, clinical characteristics such as age, hypertension, or tumor proximity determine the actual risk for this side effect is supported by this multi-institutional analysis. Moreover, the reassessment of statistical power for sex, or the lack of thereof, indicates that its significance as a risk factor as reported in [ 10 ] might have been a statistical anomaly with limited generality.…”

“…RION typically appears several months or a few years after treatment [ 9 ] and therefore requires sufficient follow-up to be detected. In addition, clinical risk factors such as age or hypertension have been demonstrated to also play a role in RION onset and have recently been integrated into NTCP models [ 5 , 10 ]. This however requires datasets with a great level of detail, including dosimetry, clinical patient characteristics, and sufficient follow-up, which, in conjunction with the low incidence rate of RION, is challenging to acquire.…”