Patients
Fifty-four consecutive patients re-irradiated for HNC between 2011-2017 in our institution were retrospectively analysed. The inclusion criteria were initial radiotherapy with curative intent (≥60 Gy) for HNC, a second course of radiotherapy for recurrence of HNC or second primary HNC where the intent was to achieve cure or local control (≥40 Gy), and an overlapping volume of these two treatments (Fig. 1). Consistent with this definition of re-irradiation, the cumulative dose in what was considered the overlapping volume was ≥100 Gy in EQD2, i.e. V100. To account for different fractionation schedules and heterogeneous dose distributions, all radiotherapy doses are reported in an equivalent dose of 2 Gy fractions (EQD2) based on the linear-quadratic model (15), using α/β=3 Gy.
A re-treatment dose ≥40 Gy was considered to reflect the clinician’s intent to achieve local control, in contrast to palliation; the same dose has been used in other studies of re-irradiation as a cut-off for significant re-irradiation dose (6, 13). Thus, patients re-irradiated with palliative doses were excluded, as were patients re-irradiated for other tumours than HNC. One patient was also excluded because of missing data from the initial radiotherapy treatment. Three patients went on to have a second course of re-irradiation. Two of these patients had a second recurrence and one patient had a new second primary HNC. For these three patients the overlapping volumes and near-maximum doses were extracted from the total cumulative radiotherapy given. Patient and treatment characteristics at the initial treatment are reported in Table 1.
In our institution approximately 230 patients with primary HNC are treated with curative intent every year. At first presentation patients are typically treated with chemoradiotherapy. Oropharyngeal cancer will typically receive definitive chemoradiotherapy without prior surgery and tumours of the oral cavity will typically undergo surgery before radiotherapy. During the studied period, patients with a tumour in the base of tongue often received a brachytherapy boost to the primary tumour after completing external beam radiotherapy.
The study was approved by the National Ethical Review Authority.
Treatment
At re-irradiation, 93% of the patients were treated with highly conformal radiotherapy: either with IMRT or volumetric modulated arc therapy (VMAT), brachytherapy or a combination of external-beam radiotherapy and brachytherapy (Table 2). In the original treatments multiple techniques and modalities were used, including brachytherapy and external-beam therapy with 6 and 18 MV photon fields, and sometimes electrons. All treatment plans were computed tomography (CT) based and for all except four patients, both the original treatment plan and the re-treatment plan were available electronically. For the patients where the original treatment plan was not available electronically, plans were reconstructed on the re-treatment planning-CT from original treatment data available as print-outs (including field parameters, beam apertures, a selection of CT slices with isodoses and dose-volume histograms), to get complete dose data for all the included patients and all treatment courses. The reconstructions were performed by an experienced physicist; field shapes and weights were adapted to the anatomy in the available CT images, resulting in a plan reflecting the clinical practice at the time of treatment. Treatments were originally planned in Eclipse (Varian, USA), TMS (Helax, Sweden) or Pinnacle (Philips Radiation Oncology Systems, USA) and any reconstructions were made in Eclipse. All external-beam dose distributions were calculated in Eclipse using the AAA algorithm, also for plans imported from a different system. Brachytherapy treatments were planned in Oncentra (Elekta, Sweden).
The 3D-dose distributions of all included external-beam plans and brachytherapy plans were exported to an in-house application converting the dose in each voxel to EQD2. The CT-, structure- and dose data were imported into a research version of Raystation (Raysearch Laboratories, Sweden) where, for each patient, the planning-CT images were registered to the most recent CT used for external-beam planning, using a grey-level based non-rigid registration. The deformed dose-distribution from each treatment was calculated on the reference CT, finally giving the cumulative dose distribution in EQD2 including all the treatment plans. The dose to the hottest 1 cm3 (D1cc) of the patient volume, i.e. the near-maximum dose, as well as V100, were extracted from the cumulative dose distribution.
In this study, the accumulated dose from the original treatment and the re-treatment was derived by registering the planning-CT images non-rigidly and summing the deformed 3D dose. Non-rigid image registration has been shown to significantly improve the estimation of accumulated dose for re-irradiation (16). While image registration in the head-and-neck region is challenging due to anatomical differences naturally appearing over time and due to different tilts of the head and different mouth fixation used for different treatment courses and modalities, non-rigid registration techniques appear to perform well for this anatomical site (17). However, the CT images used for brachytherapy planning had extensive artefacts and a very limited field-of-view. For this reason, the brachytherapy planning-CT was never used as a reference image. Despite the challenges, visual evaluation of each non-rigid registration showed a good result.
Oncologic and toxicity outcomes
Data of patient outcome were collected from a local quality registry, with prospectively gathered data, and supplemented with a review of medical records. Acute and late toxicities were graded according to Radiation Oncology Group (RTOG) and the European Organization for Research and Treatment of Cancer (EORTC) Radiation Morbidity Schema. Patients are invited for routine follow-up visits every 3 months the first 2 years after treatment, and then every 6 months for another 3 years. Patient outcome data were collected and recorded in the local quality registry every 6 months during this time. Toxicities were considered acute if presented within 90 days of the last day of re-irradiation. Any toxicities presenting later were considered late toxicities. Toxicities specifically investigated were osteoradionecrosis, soft tissue necrosis, trismus, dysphagia and carotid blowout. The Eastern Cooperative Oncology Group (ECOG) Scale of Performance status (PS) was used to quantify the functional status of the patients.
Oncologic endpoints included OS and PFS. OS was defined as the time between the last day of radiotherapy to the time of death or the last date of clinical follow-up. PFS was defined as the time from the last day of re-irradiation to the time of progression, death or the last date of clinical follow-up. Progression was defined as either progression on diagnostic imaging, a positive biopsy or a clinical progression assessed by a clinician. Carotid blowout syndrome was defined as massive pharyngeal bleeding in the absence of local recurrence.
Statistics
The Kaplan-Meier method was used to estimate OS and PFS from the last day of re-irradiation. Re-irradiation dose, overlapping re-treated volume, site of recurrence, PS at re-irradiation and age were used as predictive variables. As next step, in order to find out independent predictors, multivariate analysis was carried out using Cox regression models with hazard ratios and 95% confidence intervals. The chi-square test was used to test differences in toxicity. A test result below 5% was considered as statistically significant, and R version 3.6.1 was used for the data management and the analysis.