br The Charlson score was used
The Charlson score was used to quantify preexisting comorbidities. It is a tool for classifying clinical comorbidities and risk adjustment in analysis of administrative data . Propensity scores were estimated using a logistic model that Thonzonium Bromide included the following variables: age, sex, Charlson score, clinical T, clinical N, clinical stage, histologic grading, tumor location, and tumor length. One-to-one matching analysis between the definitive CRT and esophagectomy-alone groups was performed on the basis of propensity score to minimize bias caused by the nonrandom allocation of treat-ments. Each patient treated by esophagectomy alone was matched with a patient treated by definitive CRT.
To investigate the effect on overall survival, all of the following clinical variables were included into univariate and multivariate analyses before and after propensity score matching: age, sex, Charlson score, clinical T, clin-ical N, clinical stage, grade, tumor location, tumor length, and treatment. Statistical analysis with a p value of less than 0.05 was considered to be significant.
During the study period, 5,487 esophageal SCC clinical stage I to III patients treated by definitive CRT (4,251 [77.5%]) or esophagectomy alone (1,236 [22.5%]) were analyzed. The clinical demographics are summarized in Table 1. The mean age was 57.7 11.3 years in the definitive CRT group and 56.7 10.7 years in the esophagectomy-alone group (p ¼ 0.0097). Before pro-pensity score matching, a higher percentage of patients in the definitive CRT group were male with a higher clinical T, clinical N, and clinical stage, and a lower grade tumor and upper third location compared with the
GENERAL THORACIC 1062 WANG ET AL Ann Thorac Surg
Table 1. Clinical Demographics of Esophageal Squamous Cell Carcinoma Patients Who Underwent Definitive Chemoradiotherapy or Esophagectomy
Definitive CRT Esophagectomy
a Continuous variables are shown as the mean SD and categorical variables as number (%). CRT ¼ chemoradiotherapy.
esophagectomy-alone group. In addition, the definitive matching, the two groups were well-matched with
CRT group had a lower average Charlson score and a respect to clinical and pathologic variables (Table 1).
longer average tumor length compared with the
The 3-year overall survival rates for matched patients
esophagectomy-alone group. After propensity score treated by definitive CRT and esophagectomy alone
Table 2. Overall Survival Rate
Definitive CRT Esophagectomy
CRT ¼ chemoradiotherapy.
WANG ET AL 1063 ESOPHAGECTOMY IN ESOPHAGEAL CANCER
Fig 1. Kaplan-Meier (K-M) survival curves for (A) all esophageal squa-mous cell carcinoma (SCC) patients,
(B) clinical stage I patients, (C) clin-ical stage II patients, and (D) clinical stage III patients stratified based on treatment with definitive chemo-radiotherapy (CRT) versus esoph-agectomy (p < 0.0001 for all).
Fig 2. Kaplan-Meier (K-M) survival
curves for (A) all esophageal SCC
and (D) clinical stage III patients (p ¼
definitive chemoradiotherapy (CRT)
versus esophagectomy after propensity
THORACIC 1064 WANG ET AL Ann Thorac Surg
GENERAL were 26.14% and 46.17%, respectively (p < 0.001, Table 2). The overall survival curve of the unmatched patients according to clinical stage was stratified by treatment strategy (Fig 1). The survival curve of the unmatched patients is shown in Figure 1A. Patients treated with esophagectomy alone had a significantly superior 3-year overall survival rate (p < 0.0001). The survival curve was assessed according to clinical stage, and for clinical stages I, II, and III, respectively, the overall survival rate of patients receiving esoph-agectomy alone was better than cholecystokinin of patients treated by definitive CRT (p < 0.001, Figs 1B–D).
The overall survival rates of matched patients are shown in Figure 2. When patients were divided into different clinical stages, the overall survival rate in pa-tients undergoing esophagectomy alone was higher in clinical stages I and II. Analysis of the propensity-matched clinical stage III patients demonstrated that there was no significant difference between the two groups (p ¼ 0.2455, Fig 2D). Univariate survival analysis for unmatched patients identified sex, clinical T, clinical N, clinical stage, grade, location, length, and treatment modality as prognostic factors. Regarding univariate survival analysis for matched patients, Charlson score, clinical T, clinical N, clinical stage, tumor length, and treatment modality were identified as the prognostic factors (Table 3).