Factors Predictive of 90-day Mortality After Surgical Resection for Oral Cavity Cancer: Development of a Recursive Partitioning Analysis for Risk Stratification

Presentation: AHNS-QS-100
Topic: Mucosal - HPV Negative
Type: Quickshot
Date: Thursday, May 2, 2019
Session: 5:30 PM - 6:00 PM
Authors: Ashwin Shinde, MD1, Bernard L Jones, PhD2, Richard Li, MD1, Scott Glaser, MD1, Sana Karam, MD, PhD2, Erminia Massarelli, MD, PhD1, Morganna L Freeman, DO1, Thomas J Gernon1, Ellie Maghami, MD1, Robert Kang, MD1, Zachary S Zumsteg, MD3, Arya Amini, MD1
Institution(s): 1City of Hope National Medical Center, 2University of Colorado School of Medicine, 3Cedars-Sinai Medical Center

Introduction: Treatment of oral cavity cancer (OCC) is primarily surgical, followed by adjuvant therapy based on pathologic risk factors. While post-operative mortality (POM) has been associated with individual patient, disease, and treatment related factors, methods to create an algorithm to enable clinicians to better identify patient groups at various risks of POM have not been created. This study sought to evaluate predictors of 90-day POM in oral cavity patients and create a tool for clinicians to utilize to identify subgroups at highest risk for POM.

Materials and Methods: Patients with non-metastatic OCC diagnosed from 2004 to 2015 were identified from the National Cancer Database. Patients who underwent upfront surgical resection were evaluated. Patients were only included if they had known pathologic tumor (pT) and pathologic nodal (pN) staging. Baseline demographics, pathological, and treatment related factors were collected for use as covariates. Surgery was defined as wide local excision (including partial and hemiglossectomy) and radical (including total glossectomy and composite mandibular and/or maxillary resection). 90-day POM was evaluated using chi-square, multivariate logistic regression (MLR), and recursive partitioning analysis (RPA).

Results: We identified 33,845 patients. Ninety-day POM for the entire cohort was 3.2%. Most (95.3%) patients had squamous histology. Adjuvant radiation (RT) was delivered to 43.3% of patients, and 18.9% received chemotherapy (CT). Most (87.9%) patients receiving CT received it concurrently with RT.

On MLR, factors that predicted for a higher likelihood of 90-day mortality included older age, higher Charlson Deyo (CD) co-morbidity score, higher pT and pN stage, positive margins, pathologic extracapsular extension (ECE), and undergoing radical surgery compared to wide local excision. Factors that were protective against 90-day mortality included private insurance, those from higher income counties, receipt of RT and CT, and gum or hard palate primary subsite.

RPA was created incorporating all factors from the MLR analysis, focusing on four variables most predictive of higher rates of 90-day POM: pT, pN, patient age, and CD score. According to RPA, 90-day POM was highest in patients ≥ 60 years old, with a CD score ≥2 and T3-4, N2-3 disease (POM: 24.4%). Ninety-day POM was lowest in any age patient with T1-2 N0 disease (POM 1.4%) or those <70 years, with a CD score ≤1 and T1-2, N1-N3 disease and (90-day POM: 1.2%).

Conclusions: Patients undergoing curative intent resection for OCC appear to have a range of 90-day POM based primarily on age, comorbidity score, and pathologic tumor and nodal stage. Older patients with multiple comorbidities who present with higher-stage disease are at the greatest risk for 90-day POM, with rates exceeding 20%. Socioeconomic and insurance statuses also affect 90-day POM on MLR. For these high-risk individuals, multidisciplinary care with close monitoring following hospital discharge and early incorporation of additional supportive care services may be needed.