Clinical Acceptability of Automatically Generated Elective Lymph Node Volumes for Head and Neck Cancer Patients

PURPOSE/OBJECTIVE(S)

Manual contouring of head and neck lymph node levels is a time-intensive process prone to provider-specific variation. The purpose of this work is to generate a clinical segmentation tool while minimizing the amount of manual effort required by physicians to develop training datasets and review contours. Here we investigate an approach to curate, develop, and clinically validate an auto-contouring model for standard cervical lymph node volumes in the head and neck using a publicly available deep learning architecture. This model updates our previously validated tool to reflect modern practices in lymph node segmentation.

MATERIALS/METHODS

With the assistance of a resident physician, five radiation oncologists manually contoured individual lymph node levels on CT scans for three separate patients treated definitively with radiation or chemoradiation for oropharynx cancer, resulting in 15 unique ground truth cases. These cases were then used to train an nnUnet deep-learning model to generate automated contours for 32 additional cases. These 32 cases were reviewed, manually edited, and used to create the final model. Finally, the model was used to generate contours on the original 15 CT scans (testing cohort), and providers compared these automated contours with the ground-truth (manual) contours. Two blinded studies were performed. In a double-blinded fashion, providers were first asked to select which set of contours they would prefer to use in clinical practice as a starting point for actual cases. Second, they scored each contour on a Likert scale (1-5) to indicate clinical acceptability, ranging from completely unusable to usable without modification.

RESULTS

Across all lymph node levels (IA, IB, II, III, IV, V, RP), average Dice Similarity Coefficient ranged from 0.77 to 0.89 for AI vs manual contours in the testing cohort. These AI and manual lymph node contours were reviewed by 5 physicians each, resulting in 525 preference scores. Across all lymph nodes, the AI contour was superior to or equally preferred to the manual contours at rates ranging from 75% to 91% in the first blinded study. In the second blinded study, physician preference for the manual vs AI contour was statistically different for only the RP contours (p < 0.01). Thus, there was not a significant difference in clinical acceptability for nodal levels I-V for manual versus AI contours. Across all physician-generated contours, 82% were rated as usable with stylistic to no edits, and across all AI-generated contours, 92% were rated as usable with stylistic to no edits.

CONCLUSION

An approach to generate clinically acceptable automated contours for cervical lymph node levels in the head and neck was demonstrated. Furthermore, for nodal levels I-V, there was no significant difference in clinical acceptability in manual vs AI contours. Because we were able to generate and validate a model for each lymph node level individually, the output is applicable to a complete range of disease in which cervical lymph nodes are treated.

Proton Based Stereotactic Radiotherapy for Skull Base Patients: Dosimetric Comparison to 4 Modern Radiation Treatment Modalities

PURPOSE/OBJECTIVE(S)

Re-irradiation with ablative doses to a smaller target volume and strict critical structure constraint is a challenge for modern radiation planning and delivery systems. Several advanced radiation treatment techniques can be used for fractionated stereotactic ablative radiosurgery (FSRS) in select patients with unresectable recurrent head and neck tumors. In this study, in order to better understand the dosimetry advantage of each technique, we compare the stereotactic treatment plans of our new small spot size Hitachi proton treatment unit to those of CyberKnife stereotactic radiosurgery (CK), Gamma Knife radiosurgery (GK), volumetric modulated arc therapy (VMAT), and MR Linac radiotherapy (MRL).

MATERIALS/METHODS

Ten FSRS skull base patients treated at our institution using VMAT (n = 5) or GK (n = 5) techniques. Intensity-modulated proton therapy (IMPT) plans were created in Raystation using Monte Carlo dose calculation algorithm. VMAT, CK, GK and MRL plans were generated in RayStation, Accuray Precision, Leksell Gamma Plan, and Monaco treatment planning systems, separately. Planning goals were to achieve the best target coverage of prescribed dose without compromising the critical organs at risk dose volume constraints of the clinical treatment plans. Plans were compared based on percent CTV coverage, Paddick conformity index (PCI), gradient index (GI, V50/V100), dose homogeneity index (HI, (D2-D98)/D50), low dose bath volume (LDBV, ratio of total volume irradiated between 20% and 50% prescription dose and the target volume), beam-on-time (BOT), and mean/maximum doses to brainstems.

RESULTS

The median target volume was 15.5 cm3 (range 1.0 - 36.23 cm3). The prescription was 45 Gy in 5 fractions for VMAT patients, and 21 - 27 Gy in 3 fractions for GK patients. The comparison of the treatment plans of these 5 delivery modalities was shown in table. All techniques achieved comparable CTV coverage. GI was superior for GK plans and outstanding in CK and IMPT plans. IMPT plans were also outstanding in regard to BOT and PCI. Significantly improved HI, LDBV and brainstem mean doses were achieved in IMPT plans. For adjacent brainstem and other OARs, maximum doses were comparable among all techniques.

CONCLUSION

In these five advanced radiation therapy modalities, proton therapy SBRT showed dosimetric advantage over other modalities to spare nearby OARs without sacrifice of target coverage. Further studies are needed to utilize this clinical benefit and investigate plan robustness.

Definitive Radiotherapy for Oligometastatic and Oligoprogressive Thyroid Cancer

PURPOSE/OBJECTIVE(S)

Local consolidative radiotherapy (LCT) for oligometastatic disease is a promising paradigm improving outcomes for various malignancies but has been underexplored for metastatic thyroid cancer. We hypothesize that LCT to distant sites with definitive RT doses can yield favorable outcomes and defer systemic therapy escalation for these patients.

MATERIALS/METHODS

We reviewed 96 thyroid cancer patients who received 175 LCT courses from 2010-2022 to 228 metastatic sites, including: thorax (45%), bone (40%), brain (6%), head/neck (5%), and abdomen (3%). Common prescriptions were 50-55Gy/4-5fxs or 56-70Gy/8-10fxs for lung; 52.5-60Gy/15fxs for mediastinum; and 18-24Gy/1fx or 27-30Gy/3fxs for bone. RECIST v1.1 and CTCAE v5.0 were used to define progression and toxicities, respectively. Outcomes were evaluated via Kaplan-Meier and associations examined via Cox proportional hazards modeling.

RESULTS

Median age was 63 years (range: 26-92), with 62 oligometastatic cases (total 1-5 sites) and 34 oligoprogressive (with 1-5 growing sites). Primary disease was controlled in all patients, with 39% receiving post-op RT and 66% prior RAI. Histologies included papillary (40%), anaplastic (25%), follicular (12%), medullary (9%), Hurthle (7%), and poorly-differentiated (7%). Median time from initial diagnosis to LCT was 3 yrs (IQR 1-8), and median follow-up from 1st LCT was 21 mos (IQR 9-51). Patients received an average 2 LCT courses (range 1-8) treating 1-4 sites. Median survival (OS) from 1st LCT was 9 yrs (95% CI = 5-14). On multivariable analysis (MVA), worse OS was associated with anaplastic histology (HR 4.6, p<.01), but longer OS was associated with prior RAI (HR 0.33, p = .02) and oligometastatic disease (HR 0.3, p = .01). For anaplastic histology, median OS was 1.2 years vs. 9.3 years for non-anaplastic; 3-yr OS was 36% vs. 88% (log-rank, p<.01). Five-year OS for oligometastatic cases was 75% vs 53% for oligoprogressive (log-rank, p = .04). Median progression free survival (PFS) from 1st LCT was 15.5 mos (95% C I = 11-20). On MVA for all LCT courses, time to any progression (TTP) was negatively associated with anaplastic histology (HR 1.7, p = .02) and 2nd or higher LCT course (HR 1.45, p = .05), but favorably associated with thoracic site (HR 0.49, p<.01). Following later LCT courses, median TTP was 11 mos vs 17 mos for initial LCT course (log-rank, p = .03). After LCT to lung/chest, TTP was 18.6 mos vs 9.5 mos for non-thoracic sites (log-rank, p<.01). Only 6% of failures occurred at previously treated lesions. Most LCT courses (67%) were without ongoing chemotherapy, while 25% entailed continuing the same regimen and 9% had planned treatment post-RT. There were 2 Grade 3 toxicities (pneumonitis and esophagitis) and no Grade 4-5 events.

CONCLUSION

With high local control rates and minimal toxicity, LCT can be a feasible strategy to defer systemic therapy escalation for oligometastatic and oligoprogressive thyroid cancer.

Disease Control and Toxicity Outcomes after Intraoperative HDR Brachytherapy for Neck Recurrences in Patients Previously Irradiated Head and Neck Cancer

PURPOSE/OBJECTIVE(S)

Despite recent advances in local and systemic therapies for head and neck squamous cell carcinomas (HNSCC), the most common types of treatment failure remain local and regional. Patients with HNSCC also have a higher risk of developing secondary head and neck tumors due to a field cancerization phenomenon. Treatment of these recurrences poses a challenge, particularly in patients having previously received full dose external beam radiotherapy (EBRT) to the head and neck region. This study evaluates the clinical outcomes of intraoperative Harrison-Anderson-Mick applicator (HAM) HDR brachytherapy (BT) after salvage neck dissection for neck recurrences in patients previously irradiated for HNSCC.

MATERIALS/METHODS

All patients having received HAM BT for HNSCC neck recurrences at a single institution were identified. Demographic, treatment, disease control and toxicity data were collected retrospectively. Toxicity was graded as per Common Terminology Criteria for Adverse events version 5. The Kaplan Meier method was used to assess overall survival (OS) and ipsilateral regional control (IRC) from the time of HAM BT.

RESULTS

Nineteen patients with regional recurrence of HNSCC were treated with HAM BT Ir-192 between July 2006 and June 2022, with a median follow-up for survival of 51.4 months (range: 2.5-186.7). The median BT dose was 12.5 Gy delivered intraoperatively at 5 mm tissue depth/1 cm from source. Fifteen patients (78.9%) also received postoperative EBRT to a median dose of 46 Gy in 23 fractions and 13 (68.4%) received systemic therapy. Six patients (31.6%) developed ipsilateral neck recurrences (4 who had adjuvant EBRT and 2 who did not), 2 patients had primary site recurrences and 5 patients developed distant metastases. In terms of acute toxicity, 4 patients had grade 3 dysphagia requiring feeding tube (FT) placement and one patient required surgical intervention for wound necrosis. In terms of late toxicity, 2 additional patients required FT placement and 1 was able to have it removed. No grade 4+ late toxicities were noted. The IRC at the site of HAM BT at 1 year was 71.5% and 63.6% at 5 years. The 1-year OS was 81.3% and 5-year OS was 40.9%.

CONCLUSION

Intraoperative HAM BT resulted in favorable rates of IRC with acceptable toxicity in the reirradiation setting for HNSCC.