Modelling and external validation of late side effects of brain tumours following proton therapy

Purpose:

To investigate late physician-rated side effects and their association with dosimetric parameters of various organs at risk (OARs) as well as clinical cofactors in adult brain tumour patients following proton beam therapy (PBT).

Material and methods: Adult patients with brain tumours who underwent PBT at three different institutes were included in this study (N1=57, N2=47, N3=63). The radiation-induced side effects alopecia, fatigue, headache, memory impairment, hearing impairment, optic nerve disorder, dry eye and seizure (CTCAE v4.0) at 12 and 24 months after PBT were investigated. Side effects with sufficiently high incidence were dichotomised and correlated to different dose-volume histogram (DVH) parameters of associated OARs, such as skin, remaining brain, brainstem, cerebellum, hippocampi and cochlea. Clinical parameters comprised age, gender, tumour volume, prescribed dose, concomitant chemotherapy, resection of the tumour, diagnosis and WHO grading. Normal tissue complication probability (NTCP) models were developed on a combined cohort from two institutes (N=104) using logistic regression. The area under the receiver operating characteristic curve (AUC) was used to assess the prognostic ability in external validation on the remaining cohort.

Results: In all cohorts, low toxicity rates were observed at 12 and 24 months after PBT. Most common side effects were fatigue (grade≥1: 32%/27%, grade≥2: 14%/6% at 12/24 months), alopecia (grade≥1: 30%/22%) and mild memory impairment (grade≥1: 23% at 24 months). Mild headache and hearing impairment (grade≥1) occurred in 20%/19% and 8%/9% of all patients at 12/24 months, respectively. Logistic regression revealed significant correlations between the incidence of alopecia grade≥1 at both times and high dose regions of the skin (D2%, p<0.001, figure A). Hearing impairment grade≥1 at 24 months after PBT was associated with the median dose to the ipsilateral cochlea. For both endpoints, the developed NTCP models were successfully validated (AUC≥0.78, figure B).

Conclusion: Significant correlations between the occurrence of late side effects and DVH parameters of associated OARs were observed and externally validated for patients with brain tumours receiving PBT. Similar DVH parameters were associated to late alopecia as described for early alopecia following PBT [1]. The relation between median cochlear dose and persistent hearing loss is in agreement with several studies on photon therapy patients. In the future, these NTCP models in combination with models on neurocognition may be used to identify patients who are likely to benefit most from PBT.

[1] Dutz A et al. (2019) Radiother Oncol 130, 164-171.