COEXISTENCIA DE FOLIOS DE MATRÍCULA INMOBILIARIA SOBRE UN MISMO BIEN.

This study on adult patients with intracranial tumours receiving PBT investigated the relation of acute side effects and dose to associated OARs. Overall, PBT was well tolerated with very low incidences of side effects grade 3. Dose-volume parameters in the high-dose region were prognostic for erythema and alopecia. Fatigue was found to be associated with gender and high doses to the brain-CTV. While the NTCP models for erythema as well as alopecia showed similar or even improved performance for the validation cohorts compared to the exploration cohort, the model for fatigue could not be validated.

Several relatively small studies documented acute side effects in the brain tumour patients receiving PBT according to CTCAE v3 or v4 (Hauswald et al., 2012; Weber et al., 2012; Brown et al., 2013; Grosshans et al., 2014; Shih et al., 2015). The incidence rate of alopecia grade 2 in the exploration cohort (63 %) is comparable to the results of Hauswald et al. (2012) (68 %), even though in this study, the 19 glioma grade I-II patients were treated with an active scanning technique and a median dose of 54 Gy(RBE). Wilkinson et al. (2016) assessed 58 patients with grade II glioma treated with PBT doses between 50.4 and 54 Gy(RBE). During treatment, only grade 1 or 2 toxicities were observed with alopecia (81 %), dermatitis (78 %), fatigue (47 %), and headache (40 %) being the most common. These incidence rates of side effects are of the same range as those reported in this thesis. Weber et al. (2012) investigated 39 skull base meningioma patients treated with spot-scanning PBT with a total dose of 56 Gy(RBE) and Shih

4 Modelling of side effects following cranial proton beam therapy

(A) Alopecia grade≥2.

(B) Erythema grade≥1.

Figure 4.2: NTCP models for acute alopecia grade ≥ 2 and erythema grade ≥ 1. Regression curve (left),

calibration plot (centre) and ROC curves (right) are displayed. Each data point and error bar represents the mean value and standard deviation of each patient set, see section 4.1. Adapted from Dutz et al. (2019).

et al. (2015) conducted a study on 20 glioma patients treated with a passive scattering technique with a total dose of 54 Gy(RBE). Both studies did not report any cases of alopecia grade 2, but an incidence rate for alopecia grade 1 of 60 % and 85 %, respectively. In two studies by Grosshans et al. (2014) and Brown et al. (2013) investigating 15 skull base chordoma and chondrosarcoma patients treated with spot-scanning PBT with doses of 69.8 Gy(RBE) and 68.4 Gy(RBE), and 19 medulloblastoma patients receiving a total dose of 54 Gy(RBE), respectively, no erythema grade 2 was documented. This is comparable to validation cohort 1 in this thesis. The incidence rate of fatigue grade ≥1 both for the exploration (69 %) and validation cohort (75 %) are in line with the findings of Grosshans et al. (2014) (66 %). For nausea grade≥1, Shih et al. (2015) reported similar results with 20 %, compared to the exploration cohort with 16 %.

The findings of this thesis regarding the association of skin dose-volume parameters with alopecia and erythema are in line with other publications, in which photon doses to the skin

4.2 Modelling of early side effects

were associated with these side effects. Reduction of the irradiated volume V 24Gy and V 30Gy of follicle hair-bearing scalp prevented alopecia in a study using 11-field IMRT (Mahadevan et al., 2015). In this thesis, the skin dose-volume parameter V 25Gy (RBE ) showed also a significant association with alopecia. Nevertheless, the prognostic performance was slightly superior for D2% (alopecia grade≥1) as well as for D5% (alopecia grade≥2) in internal cross-validation.

Another study investigated side effects after XRT in 61 cranially irradiated patients (Lawenda et al., 2004). More than 50 % of the patients receiving a follicle dose>43 Gy developed permanent moderate to severe alopecia. The higher dose value of this prognostic parameter compared to the findings in this thesis may be due to the endpoint of permanent manifestation of alopecia which was investigated in this study. Sung et al. (2016) found the dose-volume parameter V 35Gy as the most significant predictor for radiation dermatitis grade ≥ 2 after hybrid IMRT for 101 breast cancer patients. As the same dose-volume parameter was found as the most predictive parameter for erythema in this thesis, skin reaction in thoracic regions may be similar to those after cranial irradiation. Mendelsohn et al. (2002) reported early dry desquamation (erythema grade ≥ 1) if the total skin dose during conventional radiotherapy does not exceed 30 Gy and acute moist desquamation (erythema grade≥2 and 3) as an effect of total skin dose>40 Gy. These findings are in line with the results of this thesis as the reduction of V 35Gy (RBE ) is predicted to lower the risk of erythema. However, in this study, it was observed that patients who had only a low absolute skin volume exposed to 35 Gy(RBE) already had a high probability of erythema grade ≥ 1, see figure 4.2B. This indicates that many patients who had developed erythema were exposed to lower skin doses than 35 Gy(RBE). Thus, a high dose parameter (e.g. D2% or D5%) that covers a wide range of doses continuously may be more appropriate to describe this dose-effect relationship than a volume parameter which may introduce a threshold effect.

The dose-volume parameters of the skin were highly correlated so that other parameters may be relevant on other cohorts. A principal component analysis was conducted to create inde- pendent parameters (principal components) representing different dose regions, see appendix B tables B.9 and B.10. NTCP models based on these components showed similar AUC values in internal and external validation, but they are more difficult to interpret. Therefore, they are less suitable for use in clinical routine.

A shift in the calibration curves for alopecia and erythema was observed in both validation co- horts for the presented model as well as for the models based on principal components. Different rates of erythema and alopecia could be caused by differences in several patient and tumour characteristics, such as tumour type and volume, see table 4.1. Nevertheless, other consider- able factors such as malnutrition, age, and tobacco abuse history may also be important (Ginot et al., 2010). Different treatment techniques in exploration (DS) and validation cohort 1 (PBS) may also lead to differences in acute side effects. A study investigating skin dose differences between PBS and DS PBT in prostate cancer patients found a lower skin dose in patients re- ceiving actively formed PBT compared to patients treated with passive PBT (Arjomandy et al.,

4 Modelling of side effects following cranial proton beam therapy

2009). With cranial irradiation, however, the target is located less deep in the body than with pelvic irradiation. Moreover, in many brain tumour patients, the target volume is located directly under the scalp, so that skin doses differ less between both techniques. A comparison of dose- volume parameters of the skin between the exploration and validation cohort 1 showed significant differences only in the low dose range, see appendix B table B.7. These findings and the shift in calibration for validation cohort 2 (PBS) support the hypothesis, that the treatment techniques are probably not the reason for different toxicity incidences in this thesis. Hence, the differences are most likely due to different toxicity scoring by the physicians, even though the same grading system was used. Therefore, a study investigating the interobserver variability of toxicity assess- ment was conducted that is described in section 4.4. Di Maio et al. (2015) detected frequent under-reporting of subjective side effects, such as alopecia and fatigue, by physicians at three different centres to a variable extent compared to patients’ self-assessment. The incorporation of patient-reported outcome versions of the CTCAE scale (PRO-CTCAE™) in clinical trials may reduce such centre-specific differences in toxicity assessment (Basch et al., 2014).

For alopecia and erythema, univariable NTCP models have been developed. The clinical co- factors surgery, prescribed dose, and CTV were highly correlated to both acute side effects and dose-volume parameters of the skin. As diagnosis determines the dose prescription, tumour volume, and margin size, patients with large CTV had significantly higher prescribed total doses (p < 0.001), leading to larger dose-volume parameter values. Due to these high correlations, the clinical cofactors were not included in potential multivariable NTCP models, since including redundant information would not improve their prognostic ability.

For acute fatigue, a bivariable model including the dose-volume parameter D2% of brain-CTV and the clinical cofactor gender was developed in this thesis. This model could not be validated. Gulliford et al. (2012a) investigated the association of dose to cranial structures and acute fatigue in 67 patients treated with 3D-CRT or IMRT and found significantly higher mean and maximum dose values for posterior fossa, brain stem, and cerebellum for patients suffering from acute fatigue grade≥2 (p<0.01). Ferris et al. (2018) found a significant correlation between patient- reported fatigue based on Multidimensional Fatigue Inventory scores and maximum dose to the brain stem as well as medulla in 124 patients treated with IMRT or VMAT (p < 0.05). In this thesis, no statistically significant dose relationship between dose to the brain stem and acute fatigue for patients treated with passively scattered proton beams could be observed; instead, a significant correlation between fatigue grade≥1 and female gender was found. One of the few studies investigating cancer-related fatigue according to the EORTC QLQ-C30 scale also found female gender associated with greater fatigue severity (Pater et al., 1997). However, as there are only a few studies on gender differences in cancer-related fatigue, there is still concern about the prevalence and severity of this side effect (Miaskowski, 2004).

Finally, no association between dose to the brain and acute nausea or headache was found in this study. These endpoints are difficult to measure and may also be alternating. Radiation dose