Triarilcarbinoles como sensores de armas químicas

Early patient identification based on simple and robust parameters bears the potential to reduce resources which are needed for detailed individual treatment plan comparisons. For HNSCC pa- tients the tumour location is a simple feature assessed routinely before treatment. The presented analysis revealed that for some of the considered NTCP models and DVH parameters a clear distinction of the benefit of PT between the subgroups exists, in particular for mucositis and dys- phagia related toxicities like larynx oedema. These differences can be explained by differently delineated elective target volumes depending on the extent of the primary tumour volume. Parts of the OAR relevant for a specific toxicity may thus have a different overlap with the targeted area depending on the subgroup. In regions with little overlap, the steeper dose gradient of IMPT leads to a better conformity, inducing less dose to OAR, which reduces the toxicity risk.

Another effect was seen in the risk assessment of xerostomia. The toxicity model is based on mean doses to the contralateral parotid gland and was significantly less reduced for patients of Subgroup B. The reason for this originates in the definition of the sequential boost volume, con- sisting of the primary tumour and involved lymph nodes, being distant in cranio-caudal direction to the glands for patients of Subgroup B. As a consequence, sparing of the contralateral parotid gland was well possible in Series III with IMXT and IMPT. Mean doses and the resulting NTCP values were smaller for Subgroup B compared to the other two subgroups. Additionally, the dif- ferences in mean doses between the two modalities were smaller for patients of Subgroup B. This also translated into smaller benefit estimation with NTCP reduction.

Small initial NTCP values were additionally impacted by another factor: the shape of the NTCP function representing the dose-effect relationship. Due to the sigmoidal shape of the NTCP curve, cf. Figure 2.8, the same dose differences lead to larger differences in the estimated NTCP values if the initial NTCP value is around 50 %. This illustrates that a simple dose reduction evaluation will not give the same results as the biological evaluation with NTCP models. Patients with small NTCP values, e.g. patients of Subgroup B for xerostomia risk, benefit less from a similar dose reduction than patients of the other subgroups. Figure 3.17 shows this exemplarily for two patients: patient 1 from Subgroup A had a decrease in NTCP of about 17 % from 26 % to 9 % xerostomia risk with 8 Gy(RBE) reduction in mean dose, while for patient 2 from Subgroup B with

Figure 3.17: The difference in NTCP, ∆NTCP, for identical dose differences ∆D depends on the initial

absolute NTCP value. Values for patient 1 are depicted in green, patient 2 in pink, IMXT treatment in solid lines, IMPT in dotted lines.

Aiming on a patient selection via subgrouping, two different parameters are relevant: first, the

∆NTCP values in a subgroup must be significantly different from the other subgroups. Second, the estimated benefit from IMPT must be of a relevant magnitude, e.g. ∆NTCP_≥10 %. With these requirements, Subgroup A would be most preferred for referral to PT. A significantly de- creased risk compared to the other two subgroups were seen for all dysphagia related toxicities at a relevant magnitude of∆NTCPMod. Furthermore, xerostomia and trismus were significantly less reduced for Subgroup B. The only significant subgroup difference favouring Subgroup B occurred for oral mucositis. Subgroup C was intermediate between the other two subgroups in most cases, as could be expected from the tumour location extending into areas of both, Subgroup A and B, and showed no advantage compared to the other two subgroups in any toxicity. Subgroup A has also been found to be preferable when using a PT selection criterion of∆NTCPMod ≥10 % and counting patients who fulfil this constraint for both, xerostomia 12 months after treatment and physician-rated swallowing dysfunction: 12 patients of Subgroup A were eligible for PT (67 % of Subgroup A), while only 1 patient of Subgroup B and 1 patient of Subgroup C fulfilled both crite- ria. Furthermore, a higher number of toxicities with∆NTCPMod ≥10 % per patient was seen for patients of Subgroup A compared to those of Subgroup B and C: 14 with at least two toxicities for Subgroup A but only 3 and 2 in Subgroup B and C. However, a general bias in the analysis of the selection concept is the different number of patients in each subgroup, with the largest number of patients in Subgroup A (18), followed by Subgroup C (14) and B (13). Additionally, the exclusion of patients from the NTCP calculations due to the infiltration of the OAR relevant for modelling further increased the discrepancy of patients numbers.

Although differences in median benefit of IMPT were seen between the subgroups, a large spread of benefit still occurred within a subgroup for some toxicities. This may have several rea- sons. First, depending on the toxicity to evaluate, subgrouping with respect to tumour location may not adequately represent patient similarities. Second, defining the correct subgroup of a patient according to the tumour location, although being conceptual simple to assess, may be difficult in some cases, especially in advanced tumour stages. Thus, the assignment of a patient to a subgroup may be controversial in some cases. These uncertainties emphasise that even though patient selection according to subgroups may be reasonable, an individual treatment plan comparison between IMXT and IMPT may be required to identify the expected individual benefit.

Furthermore, advanced tumours may grow into other anatomical areas. Involved lymph nodes may support such a trespassing tumour growth. In the presented treatment planning study, Sub- group C was created to contain such cases. With about a third of the examined patient collective, this subgroup is an essential part of the patient population with advanced HNSCC. Due to the size and extension of the tumour, patients of this subgroup often have high toxicity risks leading to similar or higher NTCP values compared to the other two subgroups, without higher NTCP reductions. Thus, patient selection should not be exclusively restricted to a subgroup, but also allow for the referral of other patients in special cases, e.g. if no applicable IMXT plan can be created.

Independent of the question concerning the treatment modality, patient subgrouping was eval- uated for the purpose of excluding patients with large disadvantage of the higher dose escalation level of 2.6 Gy(RBE). Based on the evaluated NTCP values and DVH parameters, subgrouping with this goal does not seem required, since almost no differences between the two dose esca- lation levels occurred. For this reason, the choice of the dose escalation level used for treatment intensification can be a decision purely based on medical requirements, e.g. by assessing the risk of relapse based on the magnitude of FMISO uptake after 20 Gy(RBE) of treatment.

Altogether, subgrouping with respect to tumour location allows for a patient stratification ac- cording to a probable benefit of PT in HNSCC RT favouring HNSCC patients with tumours in the upper pharyngeal region, e.g. oropharynx and oral cavity carcinoma. Due to the partially large variation of benefit within this subgroup, the patient selection should preferentially be followed by an individual treatment plan comparison to assess the actual benefit for a specific patient. The subgrouping can thus be used as a pre-selection of patients to keep the workload of individual treatment plan comparisons within feasible limits (a reduction to 40 % of the patients in a patient population similar to the presented one), while still recognising the majority of patients with signif- icant benefit. However, this should not be a strict a-priori limitation, since the individual benefits differ within the subgroups and individual patients with other primary tumour locations can also have a considerable benefit from IMPT. Those patients could be referred to PT when a generated IMXT plan violates specified constraints.