Interobserver agreement among endosonographers for staging of pancreatic cancer by endoscopic ultrasound

INTEROBSERVER AGREEMENT AMONG ENDOSONOGRA]PEgERS FOR STAGING OF PANCREATIC CANCER BY ENDOSCOPIC ULTRASOUND. F. Gross, D. Ciaeeia, C. Schmitt, M. Catalano, J. Affrenfi, K. Binmceller, P. Stevens, T. Savides, M. Bhutani, L. RJcobein, N. Niekl, D. Faigel, J, Birk, C. Lightdale. Indiana Univemity Medical Center, University of TennesseeChattanooga Unit, St. Luke's Medical Center, University of Florida, University Hospital Hamburg, Columbia Presbyterian Medical Center, University of California at San Diego, Wright State University, University of Kentucky, Portland VAMC, Winthrop University HoepitaYSUNYStony Brook. Endoscopic Ultrasound (EUS) has been reported to be an accurate modality for staging panerentic cancer (CA). However, there is limited data regarding the effects of intembserver variation on the overall accuracy and clinical utility of EUS for staging pancreatic CA. AIM OF STUDY: To determine interobserver reliability for EUS staging of pancreatic tumors. METHODS: Twelve patients with previously diagnosed pancreatic duetal adenecarcinoma undenvent staging with EUS. Surgical correlation was obtained in aJl pafica'Rs. A blinded observer developed a study videotape of real-time EUS for each case. The videotape was then distributed to ten endosonograpbers with at least one year of previous experience who independently reviewed the tape and recorded their interpretation of the EUS tumor stage for eada case based upon previously reported EUS criteria for staging pancreatic CA. A multiple observer kappa (k) statistic was calculated for eneh staging category (T, N, and T+N), vascalar invasion and overall k. Kappa accounts for agreement due to chance alone and standardized ranges are available. The Light (G) statistie was utilized to allow for the presence of the gold standard. Accuracy was assessed using a simple matrix format. RESULTS: Overall agreement was highly statistically significant for T stage: k=0.22 ( G=5.52, Z=2.35, p<0.0l).Observers were more likely to agree with T3 stage tumors (94%) than "1"2 (40%) or T1 (33%) lesions. The Overall agreement for N stage was highly statistically significant: k=0.23 (G----4.44, Z=4.90, p<0.01). Agreement for absence of nodal involvement (63%) was better than that for the presence of nodal involvement (56%). For overall staging if+N): k=0.18 (G-=9.19, Z=4.98, p<.01). For vascular invasion: k------~.26 (G=-3.90, z=3.33, p<.01). The agreement for T and N stage improved as stage increased (p<0.01) and in the presence of vascular invasion (p<0.01). CONCLUSION: Interobserver agreement is good for staging pancreatic tumors by EUS. However, it appears to be better for some tumor stages than others. Experience may play a role in the successful use of this medality to evaluate pancreatic rumors.