Results and discussion

Pathology in GI organs can be imaged with EUS using anatomical landmarks for orientation. In patients without gastric surgery, all landmarks can be located in at least 90% of cases. However, scanning all organs completely would take more than an hour. Thus, in any given examination, certain unrelated structures or areas need not be recorded, if the focus is a specific anatomical site.

Table 13.2 - Accuracy (%) of GI tumour staging and resectability with EUS and CT.

		EUS	CT
Oesophagus	T stage	85	60
	N stage	80	55
	M stage	70	70
	Resectability	80	55
Gastric	T stage	80	40
	N stage	75	50
	M stage	75	75
	Resectability	80	70
Pancreas	T stage	90	50
	N stage	75	50
	M stage	65	65
	Resectability	80	40
Biliary System	T stage	85	45
	N stage	60	50
	M stage	85	85
	Resectability	80	50
Rectum	T stage	85	70
	N stage	80	55
	M stage	-	75

Values are median estimates in % from references in text.

Table 13.2 shows the median accuracy of TNM staging and assessment of resectability for EUS compared to CT scan for various major GI neoplasms. 4 Errors in T stage occur because EUS cannot always distinguish between neoplastic tissue and benign inflammation or fibrosis. Because high frequency Sound waves have a penetration depth of only 2-4 cm from the probe, optimal focus of vessels or structures around a tumour larger than 5 cm may not always be achieved, and vessel or organ involvement may be missed.

Errors in N stage can occur for similar reasons. 4,29,52,55 A malignant node can appear to have benign characteristics because micrometastases have not yet caused parenchymal changes that can be seen sonographically. Criteria for lymph node boundaries and echogenicity appear to overlap less than those for size and shape (Table 13.3). Most metastatic lymph nodes are <10mm. By relying almost exclusively on size to evaluate N stage, CT has been insensitive for neoplastic regional lymph nodes. In contrast, because most lymph nodes >10 mm are usually neoplastic, specificity is good when CT is positive. Unlike CT scan, size resolution is not a limitation for EUS. Lymph nodes >3 mm can be found easily. However, even though EUS is superior to other imaging methods in differentiating malignant from inflammatory benign lyph nodes, criteria (Table 13.3) overlap and still require improvement.

In a prospective series of 1000 patients where EUS was performed in an office setting, the major complication rate was 0.2% (Snady H, unpublished observations, 1996). Two perforations occurred in patients with oesophageal tumours, one related to preEUS dilation of the tumour. Transient laryngospasm occurred in a third patient. In a worldwide retrospective survey of 42,105 patients,67 the "major complication rate was also reported to be low (0.05%). Two-thirds of upper GI EUS complications occurred in patients with oesophageal strictures. In 10 of 13 perforations, oesophageal dilation had been performed immediately prior to EUS. Mortality within 30 days of EUS occurred in only I of 42,105 patients surveyed, and was related to one such perforation. Therefore, aggressive dilation of all oesophageal stricture at the time of EUS is not recommended.

Improvements in correlations of sonographic findings to histology will continue to improve with further studies. Development of various agents to enhance or change sound wave characteristics of tissues and differentiate normal and pathological states will become valuable. Further human and animal studies will continue to establish and clarify parameters that will decrease interobserver variability.36-38,47 The role of EUS in selection of appropriate treatment will depend upon alternatives available for amelioration of symptoms and improvement of quality of life, survival and outcome.4,14 Utility of EUS will continue to increase with application of major advances in ultrasound technology.68