General Comments and LimitationsTable 2 summarizes the accuracy of EUS compared with CT scans for the TNM staging of various major GI neoplastic diseases. CT values are for standard dynamic CT technique with intravenous contrast agent. Data for special CT scanners and techniques such as arterial postography are not included. Accuracy is the best overall measure by which tests can be compared because it is calculated from true-positive, true-negative, falsepositive, and false-negative results. In general, gains in accuracy with EUS over other imaging methods have resulted from a significant improvement in sensitivity and, to a lesser degree, in negative and positive predictive value. There is usually only a modest improvement in specificity. A test's level of sensitivity and specificity is directly related to what it measures. CT scans rely on measurements of size and alterations of tissue planes and organ boundaries to evaluate T stage. Thus, CT cannot differentiate T1, mucosal or submucosal, disease from T2 disease, in which the deeper muscularis propria wall layer is invaded but not breached. Accurate documentation of invasion into contiguous organs (stage T4), especially when invasion is superficial, has been equally disappointing with CT. The major contribution of EUS to staging tumors has been in the evaluation of T stage. No other imaging technique can differentiate TI and T2 lesions and document superficial invasion of contiguous vessels and organs. EUS errors in T stage do occur because EUS cannot always make the distinction between benign inflammation or fibrosis and neoplastic tissue. Table 2
a Values are median estimates in percent from references in text. |