Endosonography for Cancer of the Esophagus and Cardia: Is it Worthwhile?

N. K. Altorki *
H. Snady **
D. B. Skinner *

* The New York Hospital-Cornell Medical Center, 525 East 68th Street, New York, NY 10021
** Mount Sinai Medical Center, One Gustave Levy Plaza, New York, NY 10128, USA

SUMMARY. The introduction of endoscopic ultrasonography has generally resulted in a more accurate assessment of malignant mural penetration and nodal involvement. We evaluated the role of endoscopic ultrasonography in the clinical staging of 55 patients with carcinoma of the esophagus and cardia, who underwent esophagectomy with complete mediastinal and abdominal lymphadenectomy.

The positive predictive value for T1, T2 and T3 tumors were 100%, 31% and 92% respectively. For N0 and N1 disease the positive predictive value was 44% and 79% respectively. Endoscopic ultrasonography correctly predicted tumor-lymph node-metastases stage in 62% of patients.

The data suggest that, although endoscopic ultrasonography is a reasonable modality to evaluate the primary tumor, its ability to predict nodal status is operator-dependent and may result in overstaging of patients without nodal metastases.


The staging of esophageal carcinoma remains imprecise. Various staging modalities, including biplanar esophagogram,1 azygous venography and gallium scanning1 have proved to be of limited value. Although computerized tomography is essential for the evaluation of distant organ metastases, its ability to correctly stage the local extent of the disease is suboptimal.2 Since its introduction, endoscopic ultrasonography (EUS) has become an important tool in the staging of patients with carcinoma of the esophagus and gastroesophageal junction.4 The accuracy of EUS is 85% in assessing depth-of-wall penetration and 80% for assessment of nodal status. 2-4 Despite the wide experience reported by Japanese and some European investigators,5-7 only a few surgical centers in North America have reported their experience with preoperative EUS and its impact on subsequent therapy.8-9 This report summarizes our experience with 55 patients who underwent preoperative EUS followed by esophagectomy and radical lymph-node dissection at The New York -Hospital-Cornell Medical Center.



Between January 1989 and December 1994, 55 patients underwent EUS as part of their preoperative staging workup for carcinoma of the thoracic esophagus. Tumors were located in the lower third of the esophagus or gastroesophageal junction (GEJ) in 39 patients, middle esophagus in 13 and in the upper thoracic esophagus in three patients. Adenocarcinoma of the esophagus or GEJ was present in 36 patients while 19 had squamouscell carcinoma. EUS was performed by a single operator (HS) in 34 patients while the remaining patients were studied by several other physicians. EUS was performed with an Olympus GF-UM system (Olympus Corp., Lake Success, NY, USA). All patients underwent light intravenous sedation. The 13-mm radial scanner ultrasound echoendoscope was passed into the esophagus. Mural penetration was classified as T, if the tumor extended into but not through the submucosa (seen as the second hyperechoic layer), T2 if the muscularis propria was partially invaded (second hypoechoic layer) and T3 , if the last hyperechoic layer, the adventitia, was reached. Tumors were considered T4, if clear invasion of the mediastinal structures was noted. Nodal metastasis, was assessed based on various parameters including nodal size, shape and echogenicity. Exact criteria for assessing nodal status varied among various operators. All patients subsequently underwent esophagectomy (54 transthoracic, one transhiatal) with radical lymphadenectomy excising all nodal stations between the tracheal bifurcation and the celiac axis (Table 1). Clinical and pathologic staging was done using the UICC/AJCC staging system (Table 2).

Table 1 Nodal groups resected at esophagectomy

Mediastinal nodes

Abdominal nodes

Middle paraesophageal

Paracardiac nodes

Lower paraesophageal

Lesser curvature nodes

Subcarinal nodes

Left gastric nodes

Pulmonary hiliar nodes

Celiac nodes

Diaphragmatic nodes

Common hepatic nodes
Splenic nodes


Table 2 Tumour stages

Stage 0




Stage I




Stage IIA




Stage IIB




Stage III






Any N


Stage IV

Any T

Any N




Assessment of T-status

An assessment of mural penetration and nodal status was possible in all patients including 12 (21.8%) patients in whom malignant stenosis precluded passage of the echoendoscope. In the latter patients, the probe was positioned at the superior edge of the tumor and wall penetration, as well as peritumoral node status, was partially assessed.

EUS correctly staged 7/11 (63%) patients with T1 lesions; three were overstaged to T2 and the tumor was not seen in one patient (TX). Eight patients had lesions partially invading the muscularis propria (T2), of whom five (62%) were correctly staged. Two of the remaining three patients were overstaged to T3 and one understaged as Tis. Transmural tumors (T3) were present in 34 patients of whom 24 (70.5%) were correctly identified by preoperative EUS. Among the remaining ten patients, eight were understaged as T2 and two were overstaged as T4. The two patients with T4 lesions were correctly staged by EUS. Among twelve patients in whom the passage of echoendoscope was not possible, 11 (91%) had T3 tumors. The sensitivity, specificity, positive and negative predictive values for T-stage assessment are shown in Table 3.


Assessment of nodal, status

The average number of resected nodes was 32 nodes per patient. The absence of nodal metastases was confirmed pathologically in 18 patients of whom 12 (66%) were correctly identified by EUS preoperatively. The impact of operator experience was particularly evident in this instance. A correct assessment of negative nodal status was achieved in 91% of patients studied by a single physician. Among 37 patients with N1, disease, 22 patients (59%) were correctly staged preoperatively. The sensitivity, specificity, positive and negative predictive values for N-stage assessment are shown in Table 3.

Correlation of T-stage with N1disease

There was a progressive increase in the frequency of nodal metastases with increasing depth of wall penetration by the tumor. Metastatic nodal disease was present in 3/11 T1 lesions (30%), 5/8 T2 lesions (62%) and 28/34 T3 lesions (82%). Among patients with malignant stenosis precluding passage of the echoendoscope, 11/12 (91.5%) had positive nodes.

Stage assessment

In predicting the overall tumor-lymph node-metastases (TNM) stage, EUS was correct in 34 patients (Table 4). Staging was best for patients with Stage I and Stage III disease but was poor for patients with Stage II tumors. Importantly, 11/17 patients (64%) with the Stage I or IIA tumors were correctly staged while 36% were overstaged. Overstaging was again reduced to 11% in the subset of patients studied by a single operator.


Endoscopic ultrasonography is probably the most significant advance in endoscopy over the past decade. Housed in the tip of a regular endoscope, the high-frequency ultrasound transducer produces detailed images of the various anatomic layers of the esophagus which appear as alternating hyperechoic and hypoechoic regions. The technique allows the identification of various degrees of involvement of these layers by hypoechoic lesions, benign or malignant, and the presence or absence of nodal metastases. However, the criteria for predicting nodal involvement are not strictly defined, making that assessment operator-dependent to some extent. Several investigators worldwide have substantiated the efficacy of EUS in accurately predicting the T and N stages for esophageal carcinoma. 3-9

Mural invasion is generally predicted with 85% accuracy while the N stage is diagnosed with 80% accuracy.

Table 3 Sensitivity, specificity, positive and negative predicative values for T and N status



(+) P value

(-) P value

































Table 4 Stage analysis


EUS Stage

Path Stage























The data reported herein confirm those reported by others insofar as T1 and T3 lesions are concerned, with a positive predictive value of 100% and 92% respectively. However, ability to positively predict T2 lesions was quite poor: 31 %. The small number of T2 lesions may be a possible, though unlikely, explanation of this interpretive error. Others have observed a similar inadequacy of EUS to predict T2 lesions.10 Catalano reported a significant discrepancy between experienced endosonographers in diagnosing T2 tumors.10 Whether the problem relates to the experience of the observer or to machine-dependent factors is unknown at this time. Tumors invading other mediastinal structures, i.e. T4 lesions, were also associated, in our experience, with a low positive predictive value of 50%. However, the sensitivity and specificity of EUS for T4lesions was 100% and 96% respectively. Among four patients in whom EUS predicted T4 status, only two were pathologically confirmed. We therefore continue to require corroboration of the EUS finding of a T4 tumor by other staging methods (usually computerized tomography) before deeming the patient inoperable. Occasionally, EUS may delineate the intact fat plane between the aorta and the esophagus not otherwise discernable by computerized tomography.

For staging of the N status, our data suggest that despite the significant advantage of EUS over computerized tomography, several problematic areas remain unresolved. The positive predictive value of N0 stage was only 44% with a sensitivity and specificity of 66% and 50% respectively. The presence of metastatic disease in lymph nodes less than 1 cm in diameter is a frequent even in patients with esophageal cancer. To date, no precise correlation has been established between the presence or absence of nodal metastases and the size, shape or echogenicity of lymph nodes. Although the sizes of all positive lymph nodes were not documented in this retrospective analysis, it is probable that metastatic disease was present in nodes less than 1 cm. An alternative explanation is our routine use of extended lymphadenectomy, thus excising nodal stations throughout the mediastinum and upper abdomen that are beyond the resolution of current EUS probes. When nodal metastases were present (N1) the positive predictive value EUS was 79% overall, and up to 93%. This is a definite improvement over our reported 50% accuracy in assessing metastatic nodal disease by computerized tomography.11 A 20% false positive rate is disturbing, since patients may be denied the benefit of a possible curative resection. In our own experience and that of other investigators, the five-year survival for node-negative patients is about 60%.12,13 The recent introduction of linear-array echoendoscopes equipped with a biopsy channel may allow fine needle aspiration under EUS guidance and decrease the frequency of false positive results.14 Furthermore, as noted in our data, the increase in incidence of nodal metastasis with increasing mural penetration by the primary tumor may permit an indirect evaluation of the significance of identified nodes. The ability of EUS to correctly predict TNM stage is 62% primarily due to inadequate staging of patients with IIA or IIB. The reasons for this have already been described. A troublesome observation is the overstaging (11-35%) of patients without nodal metastases; with a possible adverse impact on treatment strategy which should be primary curative surgery for such patients.

The role of EUS in treatment planning for patients with carcinoma of the esophagus and cardia remains unclear. The use of preoperative chemotherapy or chemoradiotherapy for patients with N1disease remains largely investigational. Despite the reported survival advantage of preoperative therapy in most phase II trials,15-17no such benefit has been borne out in random assignment studies.18-20 In fact, Roth reported on a randomized trial of 36 patients randomized to receive preoperative cisplatin, bleomycin and vindisine versus surgery alone.18 No difference was noted in the outcome of both arms of the trial except in a small subgroup of patients who were partial responders. Similar results were reported by Nygaard and Schlag. 19,20 A national intergroup trial currently in progress continues to evaluate the merit of preoperative chemotherapy. The use of preoperative chemoradiotherapy has also not been shown to confer a survival advantage compared to surgery alone.21 However, when chemoradiotherapy was evaluated against primary radiation therapy, a significant survival benefit was noted for the chemoradiation group. 22 Unfortunately, the local failure rate was at least 35%. Our local recurrence rate (defined as mediastinal or anastomotic) following esophagectomy is 12% and only 6% after en bloc resection. We therefore continue to believe that primary radical surgical resection is the standard form of therapy. Although the presence of positive nodes by EUS portends a poor prognosis, it is unclear what impact it might have on treatment plans but EUS is invaluable for benefit/risk and prognosis discussions with patients without distant metastasis. With the introduction of more effective chemotherapeutic agents, EUS may assume a more critical role in determining the preoperative management of patients with carcinoma of the esophagus and cardia.


1. Skinner D B. En bloc resection for neoplasms of the esophagus and cardia. J Thorac Cardiovasc Surg 1983; 85: 59-60.

2. Tio T L, Cohen P, Coene P P et al. Endosonography and computed tomography of esophageal carcinoma: preoperative classification compared to the new (1987) TNM system. Gastroenterology 1989; 96: 1478.

3. Snady H. Role of endoscopic ultrasonography in diagnosis, staging, and outcome of gastrointestinal diseases. The Gastroenterologist 1994; 2: 91-110.

4. Van Dam J. Endosonographic evaluation of the patient with esophageal carcinoma. Chest Surg Clinic North Am 1994; 4: 269-284.

5. Yoshikane H, Tsukomato Y, Niwa Y et al. Superficial esophageal carcinoma: evaluation by endoscopic ultrasonography. Am J Gastroenterol 1994; 89: 702-707.

6. Toh Y, Baba K, Ikebe M et al. Endoscopic ultrasonography in the diagnosis of an early esophageal carcinoma. Hepatogastroenterology 1993; 40: 212-216.

7. Dittler H J, Siewert J R. Role of endoscopic ultrasonography in esophageal carcinoma. Endoscopy 1993; 25: 156-161.

8. Peters J H, Hoeft S F, Heimbucher J M et al. Selection of patients for curative or palliative resection of esophageal cancer based on preoperative endoscopic ultrasonography. Arch Surg 1994; 129: 534-539.

9. Rich T W, Boyce G A, Sivak M V et al. Esophageal ultrasound and the preoperative staging of carcinoma of the esophagus. J Thorac Cardiovasc Surg 1991; 101: 536-544.

10. Catalano M F, Sivak M V, Bedford R A et al. Observer variation and reproducibility of endoscopic ultrasonography. Gastrointest Endosc 1995; 41: 115-120.

11. Skinner D B, Little A G, Ferguson M K et al. Selection of operation for esophageal cancer based on staging. Ann Surg 1986; 204: 391.

12. Matsubara T, Mamoru U, Yanagida O et al. How extensive should lymph node dissection be for cancer of the thoracic esophagus? J Thorac Cardiovasc Surg 1994; 107: 1073-1078.

13. Lerut T, De Leyn P, Coosemans W et al. Surgical strategies in esophageal carcinoma with emphasis on radical lymphadenectomy. Ann Surg 1992; 216: 583-590.

14. Wiersema M J, Wiersema L M, Khusro Q et al. Combined endosonography and fine-needle aspiration cytology in the evaluation of gastrointestinal lesions. Gastrointest Endosc 1994; 40: 199-206.

15. Forastiere A A, Orringer M B, Perez-Tamayo C et al. Preoperative chemoradiation followed by transhiatal esophagectomy for carcinoma of the esophagus: final report. J Clinic Oncol 1993; 11: 1118-1123.

16. Schlag P, Herrmann R, Raeth V et al. Preoperative chemotherapy in esophageal cancer: a phase II study. Acta Oncol 1988; 27: 811.

17. Kelsen D P, Minsky B, Smith M et al. Preoperative therapy for esophageal cancer: a randomized comparison of chemotherapy versus radiation therapy. J Clin Oncol 1990; 8: 1352-1361.

18. Roth J A, Pass H I, Flanagan M M et al. Randomized clinical trial of preoperative and postoperative adjuvant chemotherapy with cisplatin, vindesine, and bleomycin for carcinoma of the esophagus. J Thorac Cardiovasc Surg 1988; 96: 242-248.

19. Nygaard K, Hagen S, Hansen H S et al. Pre-operative radiotherapy prolongs survival in operable esophageal carcinoma: a randomized, multicenter study of pre-operative radiotherapy and chemotherapy. The second Scandinavian trial in esophageal cancer. World J Surg 1992; 16: 1104-1110.

20. Schlag P M. Randomized trial of preoperative chemotherapy for squamous cell cancer of the esophagus. The chirurgische Arbeitsgemeinschaft für Onkologie der Deutschen Gesellschaft f für Chirurgie Study Group. Arch Surg 1992; 127: 1446-1450.

21. Urba S, Orringer M, Turrisi A et al. A randomized trial comparing transhiatal esophagectomy (THE) to preoperative concurrent chemoradiation followed by esophagectomy in locoregional esophageal carcinoma. Proc Am Soc Clin Oncol 1995; 14: (Abstract).

22. Herskovic A, Martz K, Al-Sarraf M et al. Combined chemotherapy and radiotherapy compared with radiotherapy alone in patients with cancer of the esophagus. N Engl J Med 1992; 326:1593-1598.