Trends in Treatment of T1N0 Esophageal Cancer

Tara R. Semenkovich, MD, MPHS; Jessica L. Hudson, MD, MPHS; Melanie Subramanian, MD; Daniel K. Mullady, MD; Bryan F. Meyers, MD, MPH; Varun Puri, MD, MSCI; Benjamin D. Kozower, MD, MPHY


Annals of Surgery. 2019;270(3):434-443. 

In This Article

Patients and Methods

Data Source

Patients were identified from the National Cancer Database (NCDB) Participant Use File for esophageal cancer from 2004 to 2014. The NCDB is a retrospective dataset capturing >70% of all new cancer diagnoses nationwide from >1500 facilities, sponsored by the American Cancer Society and the Commission on Cancer of the American College of Surgeons. This study was exempt from Washington University's Institutional Review Board approval because the dataset is deidentified.

Patient Population

Patients with clinically staged T1N0 esophageal cancer were included in this study. Patients were excluded if they had tumors located in the cervical esophagus or upper third, had evidence of metastatic disease, or if it was unknown whether surgery was received. Patients were categorized based on receipt of: endoscopic therapy, esophagectomy, chemotherapy and/or radiation, and no treatment. A subgroup analysis was performed for T1a (tumor invading the lamina propria or muscularis mucosa) and T1b (tumor invading the submucosa) patients that had detailed staging information from 2010 to 2014, based on the effective date of the American Joint Committee on Cancer (AJCC) 7th edition cancer staging manual.


We abstracted and categorized the following covariates: age, sex, race (white vs non-white), insurance status (private vs non-private), education by zip code (lowest quartile of >21% without a high school diploma vs <21%), median income by zip code (lowest quartile of <$38,000 vs >$38,000), population by zip code (>250,000 vs <250,000), treatment center type (academic vs nonacademic), Charlson Deyo Score (0, 1, ≥2), tumor size, histology (squamous cell vs adenocarcinoma), grade, and year of diagnosis.

Statistical Analyses

All statistical analyses were performed using SAS version 9.4 (SAS Institute Inc., Cary, NC). Descriptive statistics to compare baseline characteristics of patients receiving various treatment modalities were performed using Wilcoxon rank sum tests or chi-square tests, as appropriate. The chi-squared test for trend was used to assess changes in use of treatment modalities over time. Univariable and multivariable analyses were performed using stepwise binary logistic regression to identify variables associated with receipt of endoscopic treatment versus esophagectomy. Variables that had a P < 0.1 on univariable analyses were considered for inclusion into a multivariable model, and variables that were significant with P < 0.05 were kept in the final model. Kaplan-Meier analysis and Cox proportional hazard modeling were performed to compare overall survival by treatment modality and assess for association with increased mortality hazard. Statistical significance was defined as P < 0.05.