*4533THE DETECTION OF CELIAC LYMPHADENOPATHY INESOPAHGEAL CANCER BY ENDOSONOGRAPHY ISSYNONYMOUS WITH MALIGNANT INVOLVEMENT.Mohamad A. Eloubeidi, Michael B. Wallace, Carolyn E. Reed, NevenHadzijahic, Annette van Velse, Robert Etemad, Koji Matsuda, Rig S. Patel,Robert H. Hawes, Brenda J. Hoffman, Med Univ of South Carolina,Charleston, SC.Background: EUS-guided FNA is the most accurate method of confirmingloco-regional malignant lymphadenopathy. The additional yield of FNAbeyond endosonographic characteristic is unknown. Aims: 1) To determineif the detection of a CLN by EUS, independent of FNA, indicates malignantinvolvement. 2) To evaluate the accuracy of EUS in detecting CLN metas-tasis. Methods: We reviewed all cases of esophageal cancer that underwentEUS at our institution from 1/26/94 to 11/1/99. All staging was performedwith a radial scanning echoendoscope (UM-20 or UM-130). FNA was per-formed of all accessible CLN with a linear scanning echoendoscope (UC-30P, UCT-30, UM-30P). Patients were included in this study if they under-went surgery (n= 59), or if they had FNA of a celiac LN (n=44). The accu-racy of EUS compared to cytology or histology was subsequently deter-mined. Results: 103 patients with esophageal cancer met inclusion criteria.Seventy eight percent were male and 76% were Caucasian. Fifty five per-cent had adenocarcinoma of the esophagus and 79% of the tumors wereconfined to the distal esophagus or GE junction. Twenty five percentunderwent dilation to 45 Fr to complete the examination. No complicationswere encountered. EUS imaging identified 48 true positive patients withCLN, 6 false positive, 14 false negative and 35 true negative. Therefore, thesensitivity of EUS in detecting CLN was 77% (95% CI, 67-88), the speci-ficity 85% (95% CI, 75-96), the negative predictive value 71%, and the pos-itive predictive value 89%. The overall accuracy of EUS was 81%. EUSFNA confirmed the nature of a CLN in 88% of the cases. Seventy eight per-cent (21/27) of EUS-detected CLN ≤ 1cm were malignant while 100%(25/25) of EUS-detected CLN >1 cm were malignant (p=0.02). Conclusions:Approximately ninety percent of CLN detected by EUS in patients withesophageal cancer are ultimately proven to be malignant. Since cytologicalproof of malignant involvement is critical in clinical decision making, allvisible CLN should undergo FNA. IF a CLN (> 1 cm) is imaged by EUS andFNA is not technically feasible, this study suggests that the patient shouldbe considered to have CLN malignant involvement and should be managedaccordingly.

*4534EUS STAGING PRE CHEMO-RADIATION IN ESOPHAGEALCANCER PREDICTS SURVIVAL.Kairasp C. Noshirwani, Walter E. Smalley, Panos Sechopoulos, Howard R.Mertz, Vanderbilt Univ Med Ctr, Nashville, TN; Vamc, Nashville, TN.BACKGROUND: Endoscopic ultrasound (EUS) is superior to CT in stag-ing esophageal cancer (EC). Its prognostic value has been shown in a pre-vious study of EC patients treated primarily by surgery. We specificallyevaluated the prognostic value of EUS staging for EC patients treated ini-tially by chemo-radiation prior to possible surgery.METHODS:Recordsfrom both Vanderbilt University Medical Center and the Veteran’sHospital Nashville were reviewed in 31 patients with esophageal cancer.All had EUS staging prior to recieving chemo-radiation. The EUS was per-formed by a single examiner using a Pentax 32 UA (n=21 ) and anOlympus UM 20 (n=10). Pathological staging on post surgical specimanswas available in 21 cases. Survival data was subsequently analyzed.Ninepatients had squamous cell cancer and 22 had adenocarcinoma. Twentyseven patients subsequently underwent esophagectomy. Results: Medianfollow-up period from date of EUS was 723 days. Kaplan-Meier survivalestimates showed that survival was significantly related to overall EUSstaging (log rank test: p=0.0055) and independently to EUS T staging(p=0.0001), EUS N staging (p=0.0025)and EUS M (celiac node involove-ment)staging (p=0.0435). No patients with EUS T stage 1 or 2 died of dis-ease (mean follow-up 1040 days). Of EUS T3 cases 8/14 (57%) died of dis-ease (median survival 448 days)and all stage T4 cases died of disease(median survival 486 days). Only 1/11 (9%)patients staged as N0 by EUSdied of disease (mean follow-up 998 days) whereas 14/20 (70%) N1 patientswere dead from disease (median survival time 510 days). Conversely,pathological N,T and overall staging did not significantly correlate withsurvival. Conclusion: EUS staging pre chemo-radiation in EC patientsaccurately predicts survival and is superior to pathological staging.Patients staged T3,T4 and N1 have a poor prognosis inspite of adjuventchemo-radiation and better treatment options are needed.

*4535IS ENDOSCOPIC ULTRASOUND ACCURATE IN THE RESTAGINGOF ESOPHAGEAL CANCER FOLLOWING CHEMORADIATION? John P. Czarnecki, Stuart R. Gordon, Dartmouth Hitchcock Med Ctr,Lebanon, NH.OBJECTIVE: Endoscopic ultrasound (EUS) is an important tool in stagingof esophageal tumors, however its utility following chemoradiation forrestaging or predicting complete or partial response remains unclear.These qualities would be highly desirable, as clinical course may bechanged and surgery could be avoided entirely in some cases, reducingmorbidity and mortality. In addition, few studies have looked at surrogateEUS markers that may help improve predictive value, beyond the TNMsystem. This study seeks to review the experience with EUS followingchemoradiation for esophageal cancer at our medical center, as well as toevaluate its ability to predict response based on change in tumor thicknessas a surrogate marker. METHODS: Between 3/98 and 8/99, 13 consecutivepatients with esophageal cancer underwent EUS staging pre and postchemoradiation regimens, prior to esophagectomy and pathologic staging.6 additional patients had EUS following chemoradiation administeredelsewhere, prior to surgery. 18 patients had distal esophageal adenocarci-noma; 1 patient had proximal squamous cell carcinoma. Oncologists deter-mined chemotherapy and radiation regimens individually. TNM stagingand maximal tumor thickness were noted in all cases and compared withpathologic findings. RESULTS: Post chemoradiation EUS accurately pre-dicted pathologic stage in only 4/19 patients (21%). Post chemoradiationEUS correctly demonstrated T stage in 12/19 patients (63%) and N stagein 9/19 (47%). In those undergoing pre and post chemoradiation EUS, casesdemonstrating reduction in T stage when comparing initial EUS T stage tofinal pathologic T stage showed an average change in maximal tumorthickness of 6.15mm (N=6). Those without change in T stage showed anaverage of 1.2mm tumor thickness reduction (N=7). In those who haddowngrading of EUS T stage from pre to post chemoradiation EUS, aver-age change in maximal tumor thickness was 5.8mm, and all matched finalpathologic T stage (N=5). If T stage remained the same on both EUS, aver-age change in maximal tumor diameter was 2.9mm (N=8). CONCLU-SIONS: The predictive value of post chemoradiation EUS was low and notuseful to guide clinical decision making. Accuracy for T staging was betterwhen compared to N staging, but still remained suboptimal to makeimportant clinical decisions. However, it may be possible to define a thresh-old of change in tumor thickness that would be predictive of clinicalresponse. More studies are needed to appropriately define this threshold.

*4536DO COLLATERALS AROUND ESOPHAGEAL WALL ANDPERFORATING VEINS INFLUENCE VARICEAL RECURRENCE? -ANALYSIS OF THIRTY-THREE PATIENTS BY ENDOSCOPICULTRASONOGRAPHY-.Hitoshi Oyama, Atsushi Irisawa, Katsutoshi Obara, Ayako Saito, TakutoHikichi, Tsuyoshi Rai, Tadayuki Takagi, Goro Shibukawa, Yukio Sato, ReijiKasukawa, fukushima Med Univ Sch of medicine, Fukushima, Japan.Background: The development of the esophageal varices has been demon-strated to be related to the severity of the esophageal collateral veins(ECVs) adjacent to the muscularis externa (peri-ECVs) and external to theesophageal wall but with out contact with the muscularis externa (para-ECVs) as well as the veins perforating the esophageal wall(Gastrointestinal Endosc 1999;50). In the present study, we investigatedwhether the recurrence of esophageal varices after initial endoscopic injec-tion sclerotherapy (EIS) is correlated with the severity of peri- and para-ECVs and the perforating veins. Patients and Methods: Thirty-threepatients who had been treated once with EIS were enrolled in this study.During a follow-up of more than 2 years after the initial EIS, 10 patientswere found to have recurrent esophageal varices and 23 patients showedno signs of recurrence. The collateral and perforating veins were analyzedwith a 20-MHz ultrasound catheter probe. Recurrence of esophagealvarices was determined on the basis of the appearance of the esophagealvarices, with a grade of F0 or higher together with red color sign positive,or varices with F1 or higher forms. Results: In the recurrence group, ahigher incidence of severe-type peri-ECVs (8/10, 80% vs. 2/23, 8.7%;p<0.001) and more and more dilated perforating veins (1.30±0.66 vs.0.22±0.41; p<0.001, and 2.00±1.15 vs. 0.35±0.71; p<0.001, respectively)were recognized in a comparison with these in the non-recurrence group.The severity of para-ECVs did not correlate with the recurrence. The veinsin the esophago-gastric junction (EGJ) were demonstrated in 13 of 33(39.4%) of the patients. However, the presence of veins in the EGJ did notcorrelate with the recurrence. Conclusion: The severity of the esophagealcollateral veins adjacent to esophageal wall and also of the perforatingveins detected by endoscopic ultrasonography after the initial EIS is sig-nificantly indicative of the recurrence of the esophageal varices in patientswith portal hypertension.

VOLUME 51, NO. 4, PART 2, 2000 GASTROINTESTINAL ENDOSCOPY AB161