Analysis of clinical factors associating with improvement of TNM staging accuracy in pancreatic head adenocarcinoma        XU Wei, LI Jing-dong, ZHAO Guo-gang, et al. Department of General Surgery, Institute of Hepatobiliary-pancreatic-intestinal Diseases, Affiliated Hospital of North Sichuan Medical College, Nanchong 637000, China
Corresponding author：LI Jing-dong, E-mail: lijingdong358@126.com
Abstract    Objective    To explore possible clinical factors contributing to improvement of TNM staging accuracy in pancreatic head adenocarcinoma (PHA). Methods    The retrospective study was conducted on 121 consecutive cases of PHA surgically treated from January 2004 to December 2010 in Affiliated Hospital of North Sichuan Medical College. Univariate and multivariate analysis using ordinal logistic regression explored clinical factors relating to postoperative tumor staging (pTNM staging). Cross-validation was done to test discriminant accuracy of multivariate analysis results when compared with preoperatively radiologically TNM staging evaluation. Results    Altogether 121 cases of potentially-resectable PHA entered the study. Preoperatively radiologically TNM staging evaluation revealed as followed: unidentified: 28 cases (23.1%), IA 9 cases (7.4%)，IB 45 cases (37.2%), IIA 18 cases (14.9%)，IIB 15 cases (12.4%) and III 6 cases (5.0%), respectively. Simple radiological TNM staging accuracy was frustrating, since pTNM staging results as followed: IA 2 cases (1.7%) , IB 7 cases (5.8%)，IIA 14 cases (11.6%)，IIB 6 cases (5.0%), III 63 cases (52.1%) and IV 29 cases (24.0%) , respectively. Multivariate analysis demonstrated that radiologically TNM staging，tumor diameters, ALT≥84.6U/L, neutrophil-lymphocyte ratio and serum CA19-9 values contributed to improvement of accurate TNM staging. Cross-validation based on multivariate results exhibited better discriminant performance, demonstrating the number of observations correctly classified as 71.7% and 66.7% of cross-validated grouped cases correctly classified. When pTNM staging reclassified three categories as IA+IB+IIA+IIB,III and IV，78.6% of original grouped cases correctly classified and 74.6% of cross-validated grouped cases correctly classified. Conclusion    Clinical factors associating with TNM staging could improve accuracy of PHA TNM staging and may be incorporated into current management strategy for PHA.