Interpreting the patterns of local failure following postoperative volumetric-modulated arctherapy in oral cavity and oropharynx cancers: Impact of the different methods of analysis.

Purpose: Intensity-modulated radiation therapy or volumetric-modulated arctherapy is nowadays the recommended radiation technique for the treatment of head and neck cancers. However, by providing a significant dose gradient between target volumes and organs at risk, there is a risk of target missing and thus recurrence in case of inadequate delineation. It is therefore necessary to determine the origin of these recurrences to improve clinical practice. Over the past years, different methods have been described for the analysis of recurrences. Using the patterns of failure of patients with oral cavity and oropharynx carcinoma, treated with postoperative volumetric-modulated arctherapy in our institution, the purpose of this work was to analyse the sites of local recurrences and to evaluate the disparity in the classification of recurrences when different methods were used.

Material and methods: Between 2011 and 2019, 167 patients who underwent postoperative volumetric-modulated arctherapy for oral cavity or oropharyngeal cancers were included (60 and 40 % respectively). Two or three dose levels were prescribed (54Gy, 59.4/60Gy±66Gy). Local recurrence occurred in 17 patients (10.2 %). We assessed the patterns of local recurrences according to four methods: 1/ volume-based method using the volume overlap between the recurrence volume and initial target volumes; 2/ volume-based method of overlap between the recurrence volume and the 95 % treatment isodose; 3/ point-based method using the position of the barycentre of the recurrence volume; 4/ combined centroid method classifying recurrences according to both the initial target volumes and dose distribution. Each case was reviewed to make a clinical judgment on these classifications and assessed them as "appropriate", "possible", or "inappropriate".

Results: For the volume-based method using overlap between the recurrence volume and the initial clinical target volume, this classification was clinically judged as inappropriate in 11 out of 17 cases (65 %). For the volume-based method using overlap between the recurrence volume and the 95 % prescribed isodose, this classification was clinically judged as appropriate in 15 out of 17 cases (88 %). For the point-based method, this classification was clinically judged as appropriate in 14 out of 17 cases (82 %). Thirteen out of 17 local recurrences had the same classification between this point-based method and the volume-based method of overlap between the recurrence volume and the 95 % prescribed isodose. For the combined centroid method, among 17 local recurrences nine were classified as type A, two as type B, two as type C, three as type D and one as type E. This classification was clinically judged as appropriate in 15 out of 17 cases (88 %). Only five out of 17 of the local recurrences were classified the same way according to the four different methods (29 %).

Conclusion: Recurrences that are "marginal" or "outfield" represent a major challenge for intensity-modulated radiation therapy/volumetric-modulated arctherapy quality assurance and improvement of delineation recommendations. To date, there are no published methods that give complete satisfaction.