Margin Assessment Methods in Oral Cavity Squamous Cell Carcinoma and Recurrence: Tumor Bed vs Resection Specimen Sampling.

Importance: Positive margins and margin clearance are risk factors for recurrence in oral cavity squamous cell carcinoma (OCSCC), and these features are used to guide decisions regarding adjuvant radiation treatment. However, the prognostic value of intraoperative tumor bed vs resection specimen sampling is not well defined. Objective: To determine the prognostic implications of intraoperative margin assessment methods (tumor bed vs resection specimen sampling) with recurrence among patients who undergo surgical resection for OCSCC. Design, Setting, and Participants: This was a retrospective study of patients who had undergone surgical resection of OCSCC between January 1, 2000, and December 31, 2021, at a tertiary-level academic institution. Patients were grouped by margin assessment method (tumor bed [defect] or resection specimen sampling). Of 223 patients with OCSCC, 109 patients had localized tumors (pT1-T2, cN0), 154 had advanced tumors, and 40 were included in both cohorts. Disease recurrence after surgery was estimated by the cumulative incidence method and compared between cohorts using hazard ratios (HRs). Data analyses were performed from January 5, 2023, to April 30, 2023. Main Outcome and Measures: Recurrence-free survival (RFS). Results: The study population comprised 223 patients (mean [SD] age, 62.7 [12.0] years; 88 (39.5%) female and 200 [90.0%] White individuals) of whom 158 (70.9%) had defect-driven and 65 (29.1%) had specimen-driven margin sampling. Among the 109 patients with localized cancer, intraoperative positive margins were found in 5 of 67 (7.5%) vs 8 of 42 (19.0%) for defect- vs specimen-driven sampling, respectively. Final positive margins were 3.0% for defect- (2 of 67) and 2.4% for specimen-driven (1 of 42) margin assessment. Among the 154 patients with advanced cancer, intraoperative positive margins were found in 29 of 114 (25.4%) vs 13 of 40 (32.5%) for defect- and specimen-driven margins, respectively. Final positive margins were higher in the defect-driven group (9 of 114 [7.9%] vs 1 of 40 [2.5%]). When stratified by margin assessment method, the 3-year rates of local recurrence (9.7% vs 5.1%; HR, 1.37; 95% CI, 0.51-3.66), regional recurrence (11.0% vs 10.4%; HR, 0.85; 95% CI, 0.37-1.94), and distant recurrence (6.4% vs 5.0%; HR, 1.10; 95% CI, 0.36-3.35) were not different for defect- vs specimen-driven sampling cohorts, respectively. The 3-year rate of any recurrence was 18.9% in the defect- and 15.2% in the specimen-driven cohort (HR, 0.93; 95% CI, 0.48-1.81). There were no differences in cumulative incidence of disease recurrence when comparing defect- vs specimen-driven cases. Conclusions and Relevance: The findings of this retrospective cohort study indicate that margin assessment methods using either defect- or specimen-driven sampling did not demonstrate a clear association with the risk of recurrence after OCSCC resection. Specimen-driven sampling may be associated with reduced surgical margin positivity rates, which often necessitate concurrent chemotherapy with adjuvant radiation therapy.

Concurrent carboplatin and paclitaxel definitive radiation therapy for locally advanced head and neck cancer.

BACKGROUND: We report the outcomes of cisplatin-ineligible HNSCC patients treated with definitive chemoradiation and concurrent carboplatin and paclitaxel. MATERIALS AND METHODS: We included consecutive HNSCC patients treated from 2013 to 2021 that received definitive chemoradiation with carboplatin and paclitaxel. Locoregional recurrences (LRR) and distant metastases (DM) were estimated using cumulative incidence functions. Progression free survival (PFS) and overall survival (OS) were estimated using Kaplan-Meier methods. RESULTS: Sixty-five patients were identified with median age of 71 years (range 44-85). Median radiation dose was 70 Gy and the median doses of carboplatin and paclitaxel were AUC 1 and 40 mg/m2 , respectively. At a median follow-up of 29 (range 5-91) months, the 2-year rates of LRR, DM, PFS, and OS were 8.8%, 9.4%, 72.2%, and 88.7%, respectively. In total, there were 5 LRR, 7 DM, and 12 deaths. CONCLUSIONS: Chemoradiation with carboplatin and paclitaxel is an excellent option for cisplatin-ineligible HNSCC patients.

Peer Review of Head and Neck Cancer Planning Target Volumes in Radiation Oncology.

Purpose: Radiation treatment plans undergo peer review during chart rounds, but changes to treatment volumes would require replanning. Our group implemented weekly head and neck cancer "volume rounds" to peer review all target volumes for head and neck cancer before radiation therapy (RT) planning and chart rounds. Methods and Materials: We analyzed modifications made to planning target volumes (PTVs) at volume rounds for consecutive nonproton head and neck cancer cases from May 2020 to May 2021. Nine head and neck radiation oncologists participated in weekly volume rounds during this time. Recommendations were categorized as no changes, minor changes, major changes, additional workup (eg, biopsy or imaging), and consultation or tumor board discussion needed before the start of RT. Minor changes to PTVs generally did not require a second review before treatment planning while major changes did. Results: PTVs for 511 cases involving 432 patients underwent peer review and 298 (58.3%) of these cases did not require any modifications before treatment planning. Minor and major changes were recommended in 75 (14.7%) and 86 (16.8%) cases, respectively. Forty-five (8.8%) cases were recommended to have additional workup and 23 (4.5%) required additional consultation with nonradiation surgeons or medical oncologists. Of the 45 cases that were recommended for additional workup, 40 underwent biopsy or imaging. Positive findings on imaging or biopsy occurred in 13 patients, leading to a significant change in management, including 4 patients who underwent additional surgery after positive findings before the start of RT. Conclusions: Prospective peer review during head and neck cancer volume rounds led to frequent minor and major alterations to PTVs. Significant changes in the overall treatment plan, such as additional surgery before start of RT, occurred in a minority of patients.

Prospectively-validated deep learning model for segmenting swallowing and chewing structures in CT.

Objective.Delineating swallowing and chewing structures aids in radiotherapy (RT) treatment planning to limit dysphagia, trismus, and speech dysfunction. We aim to develop an accurate and efficient method to automate this process.Approach.CT scans of 242 head and neck (H&N) cancer patients acquired from 2004 to 2009 at our institution were used to develop auto-segmentation models for the masseters, medial pterygoids, larynx, and pharyngeal constrictor muscle using DeepLabV3+. A cascaded framework was used, wherein models were trained sequentially to spatially constrain each structure group based on prior segmentations. Additionally, an ensemble of models, combining contextual information from axial, coronal, and sagittal views was used to improve segmentation accuracy. Prospective evaluation was conducted by measuring the amount of manual editing required in 91 H&N CT scans acquired February-May 2021.Main results. Medians and inter-quartile ranges of Dice similarity coefficients (DSC) computed on the retrospective testing set (N = 24) were 0.87 (0.85-0.89) for the masseters, 0.80 (0.79-0.81) for the medial pterygoids, 0.81 (0.79-0.84) for the larynx, and 0.69 (0.67-0.71) for the constrictor. Auto-segmentations, when compared to two sets of manual segmentations in 10 randomly selected scans, showed better agreement (DSC) with each observer than inter-observer DSC. Prospective analysis showed most manual modifications needed for clinical use were minor, suggesting auto-contouring could increase clinical efficiency. Trained segmentation models are available for research use upon request viahttps://github.com/cerr/CERR/wiki/Auto-Segmentation-models.Significance.We developed deep learning-based auto-segmentation models for swallowing and chewing structures in CT and demonstrated its potential for use in treatment planning to limit complications post-RT. To the best of our knowledge, this is the only prospectively-validated deep learning-based model for segmenting chewing and swallowing structures in CT. Segmentation models have been made open-source to facilitate reproducibility and multi-institutional research.