WEDNESDAY, SEPTEMBER 6 57

FC3.22 HYSTEROSCOPY: DIAGNOSIS FC3.22.03

FC3.22.01 INTEROBSERVER VARIATION IN THE ASSESSMENT OF HYSTEROSCOPIC IMAGES

OUTPATIENT DIAGNOSTIC HYSTEROSCOPY WITH HYDRODILATATION M. A. Siddig Dept. OB/GYN Royal Cornwall Hospital, Truro, Cornwall, UK

J. M. Martinho, P. Xavier, .I. Monsanto. Departamento de Ginecologia e Obstetricia, Hospital de S. JoPo, Faculdade de Medicina do Porte, Portugal

Objectives: To evaluate the success, indications, complications and results 286 diagnostic hysteroscopies performed with hydrodilatation (without anaesthesia) in a district general hospital. Design: Retrospective analysis of hysteroscopy charts between 1996 and 1998. Setting: Royal Cornwall District General Hospital, Patients: 286 patients referred for outpatient diagnostic hysteroscopy for different indications. Intervention: Diagnostic hysteroscopy was performed in 86% of patients without anaesthesia or cervical dilatation. Hydrodilatation technique was used by increasing the pressure of the infused saline to 200cm H,O. The hysteroscope was introduced under direct vision using a camera. Results: The most common indication for hysteroscopy was PMB (26.5%). Hysteroscopy was successfully performed in 90.9%. In 86% of women we accessed the uterine cavity with hydrodilatation technique under direct vision without premeditation, local anaesthesia or cervical dilatation. In 39.8% intrauterine pathology was diagnosed and in 10% further surgical treatment in the form of hysterectomy was carried out. The most common cause of failure was cervical stenosis (61.4%). Conclusion: Diagnostic outpatient hysteroscopy with hydrodilatation and introduction of the hysteroscope under direct vision is feasible in overwhelming majority of cases without local anaesthesia. The low complication rate and the high detection rate for intrauterine pathology justifies performing this procedure in the office as routine in the new century.

Objective: Evaluation of interobserver agreement in the assessment of hysteroscopic images. Methods: Sixty consecutively recorded hysteroscopic video-tapes with good quality image were selected, obtained with a 5 mm Hamou II hysteroscope and saline as distension media. The tapes were reviewed by three experienced specialists in hysteroscopy, who were asked to detect and identify intra-cavitary lesions as polyps, myomas, neoplasia, adhesions, malformations and IUD and to classify the endometrial pattern as atrophic, proliferative, secretory or hyperplastic. For statistical analysis, the proportions of agreement (PA) and the k statistic, with 95% confidence intervals (95%CI) were used. Results: Interobserver agreement was excellent regarding the detection of intra-cavitary lesions (PA=O.86, 95%CI 0.84-0.89; K=0,79,95%CI 0.70.0.89), but only fair to good in the classification of the endometrial pattern (PA=O.65, 95%CI 0.61-0.69; K=O,49,95%CI 0.40-0.59). Conclusions: A good agreement between specialists is found in hysteroscopic detection of intra-cavitary lesions, but less so in the classification of endometrial patterns. Care should be taken in the interpretation of endometrial patterns as obtained through the hysteroscope, as this may not be very reproducible.

FC3.22.04 CLASSIFICATION OF MICROHYSTEROSCOPIC IMAGES AND THEIR CORRELATION WITH HISTOLOGICAL DIAGNOSES J.E. Dotto, B. Lema, Instituto Argentino de Diagnbstico, Buenos Aires, Argentina.

FC3.22.02 THE ACCURACY OF HYSTEROSCOPIC VISUAL IMPRESSION FOR DIAGNOSING ENDOMETRIAL COMPLEX ATYPICAL HYPERPLASIA OR CANCER G. Del Priore, S. Feinstein, F.S.Williams, A. Lui, NYU School of Medicine, 530 First avenue, suite 9R, New York, NY, United States, 10016.

Objective: The aim of the study was to determine the accuracy of hysteroscopic visual impression (HSC) for diagnosing endometrial complex atypical hyperplasia (CAH) or endometrial cancer (CA). Methods: Using the medical record and pathology department databases from 1994 to 1998, we identified all patients diagnosed with CAH or CA. Those patients who had also had hysteroscopy with dilation and curettage (HSC D&C) as part of their work-up are the subjects of this study. Results: There were approximately 700 patients with either CAH or CA, 44 of which had a HSC D&C. Medical records were available for all of them. Patient ages were between 23 and 87 years with a mean of 58.1. Uterine size was between 4 and 13 weeks with a mean of 6.8. The indication for HSC D&C was abnormal uterine bleeding in 54.5%, a suspicious prior office biopsy in 20.5%, abnormal sonogram in 13.6%, other in 4.5% and not specified in 9.1%. The pre-operative diagnosis was rule out cancer or CAH in 26.8%. This number did not change after HSC. However, in two patients, the pre-operative diagnosis of CAH or cancer was changed to an incorrect benign diagnosis. In two other patients, who had not been diagnosed pre-operatively with CA or CAH, visual impression under hysteroscopy led to these correct diagnoses. The operative findings were reported as completely normal in 7.3% and various other benign findings were reported in 82.9%. The preoperative diagnosis was changed by hysteroscopic appearance in 9.8% of cases. Compared to final pathologic diagnosis, the intra-operative HSC detection of CA or CAH (sensitivity 26.8%) added nothing to the pre- operative clinical impression (sensitivity 26.8%). Conclusions: The operative impression based on HSC appearance rarely matches the post-op pathologic diagnosis of CAH or CA. Cancer is missed by HSC even when the clinical scenario suggests malignancy. We found that physicians are not more likely to correctly predict CAH or CA with HSC information. Further study is needed. However, the data suggests that hysteroscopy is of little benefit in diagnosing CAH or CA.

We have suggested a method for the systematic classification of hysteroscopic images based on our experience with 1436 hysteroscopies and their corresponding biopsies. Images are classified as follows: normal hysteroscopy, benign disease, low risk hyperplasia, high-risk hyperplasia and carcinoma. There is a strong correlation between the classification of images and the histological diagnosis; in the first two groups the correlation was present in every case; there were some false positives and false negatives in the low and high risk hyperplasias and carcinomas, the causes of which have been analyzed. If this classification system is used for the diagnostic exploration and submitted to the final histological diagnosis it can be useful for a systematic approach to hysteroscopic findings and to improve communication between specialists involved in a case. It is based on the degree of hysteroscopic suspicion aimed at the early detection of endometrial cancer and its precursor lesions. Correlation between microhisteroscopy images and histopatological confirmation

C.H. CARC. TOTAL 1MAGES

N.H.I.:

L.R.H.:

CARC:

S.H.:

NORMAL HYSTEROSCOPlC

1MAGE

LOW RlSK

HYPERPlASlA

CARClNOMA

SlMPLE HYPERPlASlA

B.D.: BENlGN DlSEASE

H.R.H.: HlGH RlSK

HYPERPlASlA

N.H.; NORMAL HlSTOLOGY

C.H.: COMPLEX

HYPERPlASlA

The correlation was significant @<O.OOl) Cramer V correlation coefficient = 0.925

The Lambda coefficient = 0.96