The analysed patient group consists of 2 index toxicity and 6 similar non-toxicity control cases. The non-toxicity cases were matched by the proximity of the target volume to optical structures as well as the indication for treatment. The patients were similarly treated on PSI Gantry 3 (ProBeam Connect) in a time period between October 2019 and February 2022.

A dose metric D2 above a threshold dose (above 60 GyRBE) [19] increases the risk for development of late optic pathway toxicities. However, it can also be seen that the two toxicity index cases were associated with considerably lower D2 dose levels: 51.05 GyRBE and 48.95 GyRBE for the cases T4-03 and T3-06, respectively (Fig. 1). Although significant variability in terms of sensitivity to dose exists in the population, our data suggests the existence of other dose promoting adverse effects. LET is widely considered to be an adverse effect promotor, where essentially elevated LET results in an elevated RBE [20]. LET distributions for the given patient group were calculated. However, similarly as with the dose metrics D2, we did not observe a correlation between elevated LET and the location of radiographic injury in the optic apparatus for the index cases, or a variance in LET between index and control cases to suggest an association with optic apparatus injury (Fig. 2). LET on its own is a parameter that is difficult to interpret in terms of association with increased clinical toxicity risk [21,22,23]. Numerous studies have demonstrated a lack of association between LET and radiation-induced toxicity while other series have suggested a causal relation [24].

Dose rate in proton pencil beam scanning is a parameter that can be considered from various perspectives: instantaneous, averaged over the duration of the delivery of a field or averaged over a predefined period of time (for example, 100 ms) [14]. Furthermore, when calculating dose rate, the considered dose level can be factored in (for example, dose rate is considered only for delivered dose in the voxel above 50 cGy level). We derived a global voxel wise maximum dose rate as a parameter (MDR) to gain initial insights into the dose rate related parameters in the context of the development of late toxicities. MDR is related to a maximum instantaneous dose rate that a voxel has been exposed to disregarding the amount of dose delivered at this dose rate. Nevertheless, we observed that for the two toxicity cases T4-03 and T3-06, which were difficult to explain in the context of dose metrics, MDR was shown to be elevated compared to other cases in the group (Fig. 4). In fact, MDR2 as high as 3.9 GyRBE/s and 4.3 GyRBE/s was observed only for the two unexplained toxicity index cases (T4-03 and T3-06). These high MDR values could be traced back to a contribution from a single field in the plan for both cases. Figure 5 has shown that such high dose rates delivered from a field to structures of the optical apparatus are atypical.

In this study we demonstrate the relative differences in MDR between index and control cases. A word of caution should be mentioned with regards to broader adoption of absolute scale. The MDR metric is sensitive to dose threshold value, which in the current study was set to 1 cGy per fraction. By adopting different parametrization, the obtained absolute MDR values may vary considerably. The choice of the dose threshold value in MDR calculations is a subject of further studies.

For the two individual fields contributing to the high MDR values for the cases T4-03 and T3-06 we calculated field-wise MDR distributions and overlayed them with the delineated CT, as shown in Fig. 6. The high MDR regions intersect remarkably well with the damaged left pre-chiasmatic optic nerve structure, which, for both cases, was associated with the late optic toxicity that was identifiable based on follow-up MR images.

To put the observations in the context to the FLASH effect it should be highlighted that, the dose rates for the clinical cases in this study are well below FLASH dose rates, which are currently considered to be above 40 Gy/s. Furthermore, a FLASH effect is achieved at high dose levels. On the contrary to that, we observe low dose levels delivered at high dose rates. For both index cases the dose delivered at 3.5 Gy/s or more was less than 10% of total dose delivered to the toxicity area. Finally, at this point overwhelming majority of FLASH effect demonstrations are related to acute toxicities, while RION, which is considered in the current study, is late toxicity.

When discussing relative biological effectiveness (RBE), there are numerous parameters that define its characteristics [25]. Synergic effects between dose rate and LET could be explored further and might reveal a better predictive model for late toxicity development. In our investigated cases regions of high LET and regions of high dose rate spatially do not correlate. High dose rates can for example also occur on the proximal edge of a field, where the LET is usually low. If a significant dose-enhancing effect linked to dose rate exists, it is probable that predicting the risk of toxicity would be more accurate by considering the cumulative dose delivered at or above a specific dose rate threshold, rather than solely relying on the instantaneous maximum dose rate. We are further exploring the dose above dose rate parameter space in combination with dose delivered at an elevated RBE due to LET.

The strong limitation of this study is the size of the data set. Fortunately, the incidence of optic apparatus toxicities is low in our patients with intracranial disease [11, 12]. Therefore, the possibility to acquire a sufficiently large data set to prove or disprove the hypothesis will require a multi-institutional collaboration, similar as suggested in the European Particle Therapy Toxicity Workshop (Leuven, Belgium, 04.10.2022). Nevertheless, dose rate calculation for conventional proton treatment plans currently is not a feature readily available in commercial treatment planning systems (TPS). Such calculations require in-house developments or dedicated research builds of TPS.

Eventually, other clinical factors may have influenced the development of the toxicity. For instance, one out of two toxicity cases is a patient more than 60 years old.

In conclusion, our observations reveal large variations in instantaneous dose rates experienced by different volumes within our patient cohort, even when considering the same indications and beam arrangement. Furthermore, we observed an overlap between high-dose rate areas and regions of developed optic toxicities as identified on MR images. At this point, it is not feasible to establish causality between exposure to high dose rates and the development of late optic apparatus toxicities due to the low incidence of injury. Since the number of toxicity cases per institute is generally low, a multi-institutional study would be necessary to provide evidence for the correlation between elevated dose rate and the radiation injury. Further radiobiology studies investigating the effects of low/high dose rate proton exposures will hopefully provide additional insights and further elucidate this potential radiobiological impact.