| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Genitourinary Imaging |
1 From the Departments of Medical Imaging (K.F., P.N., P.G., D.S.), Obstetrics and Gynecology (R.K.), Pathology (A.L.), and Medical Oncology (M.T.), Sunnybrook and Women’s College Health Sciences Centre, 76 Grenville St, Toronto, Ontario, Canada M5S 1B2; the Departments of Pathology (W.C.) and Radiation Oncology (L.M.), Princess Margaret Hospital-University Health Network, Toronto, Ontario, Canada; and the Centre for Research in Women’s Health, University of Toronto, Ontario, Canada (T.M.). From the 2000 RSNA scientific assembly. Received November 26, 2000; revision requested January 22, 2001; revision received March 9; accepted April 3. Supported by the Canadian Breast Cancer Foundation (Ontario Chapter) and the Helen and Paul Phelan Foundation. Address correspondence to K.F. (e-mail: katherine.fong@swchsc.on.ca).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: The authors prospectively examined 138 women receiving tamoxifen by using transvaginal US, hysterosonography, and office hysteroscopy. The combined hysteroscopic-histopathologic diagnosis was the reference standard. Sensitivity, specificity, positive and negative predictive values, and likelihood ratios of transvaginal US and hysterosonography were calculated.
RESULTS: All 138 women underwent transvaginal US; 104, successful hysterosonography; and 117, successful hysteroscopy. Uterine abnormality was present in 47 (40.2%) of 117 women: 45 with polyps and two with submucosal fibroids. Receiver operating characteristic curve analysis revealed 6 mm to be the optimal endometrial thickness cutoff for diagnosing endometrial abnormalities. When a thickness greater than 6 mm or a focal endometrial finding was considered abnormal, transvaginal US had a sensitivity of 85.1% and a specificity of 55.7%. In 92 women who completed transvaginal US, hysterosonography, and hysteroscopy, hysterosonography was more specific (79.2%; P = .008) but not significantly more sensitive (89.7%; P = .508) than transvaginal US. When women with abnormal transvaginal US findings were further examined with hysterosonography, the sequential combination of transvaginal US and hysterosonography was more specific (77.1%) than transvaginal US alone (P < .001), without a significant decrease in sensitivity (78.7%; P = .25).
CONCLUSION: In asymptomatic postmenopausal women receiving tamoxifen, 6 mm is the optimal endometrial thickness cutoff for diagnosing endometrial abnormalities with transvaginal US. Further examination with hysterosonography can improve specificity by reducing the high false-positive rate of transvaginal US.
Index terms: Uterus, abnormalities, 854.315, 854.317, 854.318, 854.3192, 854.3199 • Uterus, endometrium • Uterus, US, 854.12985, 854.12989
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Transvaginal ultrasonography (US) is often used to evaluate the endometrium. Although endometrial thickening at transvaginal US may indicate abnormality, there is no clear definition of what constitutes abnormal endometrial thickness in women receiving tamoxifen. Various authors (4,5,9–11) have reported a double-layer endometrial thickness of 5, 8, 9, or 10 mm as the cutoff value for identifying endometrial disease in asymptomatic postmenopausal women receiving tamoxifen. However, these studies were limited owing to an inadequate reference standard (4,5), verification bias (9,10), a small sample size (11), the lack of appropriate blinding (4,5,9–11), and the failure to report test reproducibility (4,5,9–11).
Hysterosonography is a technique for evaluation of the endometrial cavity. The introduction of fluid with US guidance improves depiction of the endometrium. Hysterosonography in women treated with tamoxifen has been evaluated in a few studies (12–15). However, according to methodologic standards for diagnostic testing (16), these studies (12–15) have limitations similar to those involving transvaginal US.
The purpose of our study was to determine the test performance characteristics—that is, sensitivity, specificity, positive and negative predictive values, and positive and negative likelihood ratios—of transvaginal US and hysterosonography for diagnosing endometrial abnormality in asymptomatic postmenopausal women with breast cancer who were receiving tamoxifen as adjuvant therapy.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
The exclusion criteria included previous hysterectomy, endometrial ablation, hormone replacement therapy during the past 6 months, known metastatic disease, other known primary malignancy, tamoxifen therapy for more than 5 years, cessation of tamoxifen therapy for more than 4 weeks, vaginal bleeding during the past month, endometrial sampling during the past 6 months, and acute pelvic infection. In addition, at the request of their oncologists, patients at Princess Margaret Hospital were not approached for this study if they were already enrolled in other research trials.
Between June 1996 and March 2000, we recruited 138 women: 57 from Princess Margaret Hospital and 81 from Sunnybrook and Women’s College Health Sciences Centre. After written informed consent was obtained, we collected information on the patients’ age, time since menopause, time since diagnosis of breast cancer, and duration of tamoxifen therapy. Transvaginal US and hysterosonography were performed during the same visit. Office hysteroscopy was scheduled for either the same day or a subsequent visit.
Transvaginal US
Transvaginal US (HDI 5000 or HDI-Ultramark 9; Advanced Technology Laboratories, Bothell, Wash) was performed by one of two designated sonographers (one of whom was D.S.). The frequency of the vaginal probe was either 4–8 or 5–9 MHz. The uterus was scanned in the sagittal and coronal planes. The double-layer endometrial thickness was measured at the widest point between the endometrial-myometrial interface in the sagittal plane by using electronic calipers. If there was fluid in the endometrial cavity, the amount (<2 or 
2 mL) was recorded, the anterior and posterior walls of the endometrial cavity were measured separately, and the measurements were added to obtain the endometrial thickness. The presence of focal endometrial thickening or a focal mass within or deforming the endometrial central echo was noted. The transvaginal US examination was supervised by a radiologist (K.F., P.N., or P.G.), who reviewed the images and the endometrial thickness measurements.
Interobserver reliability for measuring endometrial thickness was assessed prior to the study in a separate group of 36 postmenopausal women with breast cancer who had been receiving tamoxifen for at least 6 months. Two observers—a radiologist (K.F.) and a sonographer (D.S.)—performed transvaginal US within 10 minutes of each other in a blinded randomized fashion. The intraclass correlation coefficient was 0.95 (lower 95% CI, 0.92). We were therefore confident that reliable measurements of endometrial thickness were possible.
Hysterosonography
Hysterosonography was performed by a radiologist (K.F., P.N., or P.G.) who was blinded to the results of transvaginal US. A vaginal speculum was inserted, and the cervix was cleansed with povidone-iodine solution. A 5-F Tampa catheter (Ackrad Laboratories, Cranford, NJ) or a catheter with a small balloon near the tip (H/S catheter; Ackrad Laboratories) was introduced through the cervical os into the uterus. When it was necessary to use a tenaculum and/or a cervical dilator, benzocaine 20% spray (Hurricaine Spray; Beutlich Pharmaceuticals, Waukegan, Ill) was used as a local anesthetic. The vaginal probe was inserted following withdrawal of the speculum. With US guidance, approximately 5 mL of sterile normal saline solution was injected into the uterine cavity. Images of the endometrial cavity were obtained on films, and the examination was videotaped. By using electronic calipers, the anterior and posterior walls were measured separately where there was no focal abnormality. The presence of any intracavitary mass or focal area of endometrial thickening was noted.
Women did not undergo hysterosonography when transvaginal US depicted at least 2 mL of fluid in the endometrial cavity. In such women, the US appearance essentially mimics that on a hysterosonogram. Therefore, hysterosonography would not have provided any additional information.
Hysteroscopy
The mean interval from transvaginal US and hysterosonography to office hysteroscopy was 8.8 days (range, 0–23 days). Women were advised to take oral analgesics, such as acetaminophen, and/or a nonsteroidal antiinflammatory agent 1 hour prior to the procedure. The gynecologists (one of whom was R.K.) who performed the hysteroscopic procedures were blinded to the transvaginal US and hysterosonographic findings. The patient was prepared, and a paracervical block was performed. The cervix was grasped with a tenaculum and gently dilated with a size 5 Hegar dilator, if necessary, and a 5-mm 25° panoramic cystoscope (Richard Wolf, Knittlinger, Germany) was inserted. Normal saline solution was used to distend the uterine cavity for optimal visualization. The appearance of the endometrium (atrophic, proliferative, or hyperplastic) and any abnormality (polyps, fibroids, or carcinoma) were recorded. Directed biopsy of any abnormal area was performed through the 3-mm operating channel. This was followed by Pipelle biopsy in all patients.
Histopathologic Analysis
All specimens were immediately placed into a 10% neutral buffered formalin solution. Hematoxylin and eosin–stained sections prepared from the paraffin-embedded material were reviewed by two reference pathologists (A.L., W.C.) (
for interpathologist agreement, 0.87; 95% CI: 0.77, 0.96). Any discrepant diagnoses were resolved at consensus review. The pathologists were blinded to the transvaginal US and hysterosonographic findings.
Patient Preference Assessment
The patients were asked to rate their experience with transvaginal US, hysterosonography, and hysteroscopy by using a five-point scale: 1 indicated no discomfort; 2, minimal discomfort; 3, mild discomfort; 4, moderate discomfort; and 5, severe discomfort. In addition, they were asked to indicate which of the three procedures would be acceptable to them in the future.
Outcome Criteria, Sample Size, and Statistical Analyses
The primary outcome was the presence or absence of uterine (endometrial) abnormality. Uterine abnormality was defined as the presence of one or more of the following: (a) hysteroscopic or histopathologic evidence of submucosal fibroid and (b) histopathologic confirmation of endometrial polyp, endometrial hyperplasia, or malignancy. The uterus was considered normal if no finding other than an atrophic or proliferative endometrium was documented at combined hysteroscopic-histopathologic analysis.
Secondary outcomes were complications of hysterosonography and office hysteroscopy, as well as patient acceptability of transvaginal US, hysterosonography, and office hysteroscopy. Complications evaluated included (a) infection, hemorrhage, and uterine perforation resulting from hysterosonography and (b) infection, hemorrhage, cervical laceration, uterine perforation, and creation of a false passage in the cervix secondary to office hysteroscopy.
We chose a sample size that would yield a 95% CI of plus or minus 0.1 for a sensitivity of 0.90 for transvaginal US, assuming the prevalence of endometrial abnormality was 30%. When a 5% dropout rate and a 5% failure rate for hysteroscopy were assumed, the required sample was determined to be 126 women. Because the sensitivity of hysterosonography was expected to be the same as or higher than that of transvaginal US, this sample size was deemed to be sufficient.
Analyses were based on women who successfully completed hysteroscopy. The primary analysis was performed to determine the cutoff level of endometrial thickness that maximized the accuracy of transvaginal US in identifying endometrial abnormality. Statistical analyses were performed by using a computer software program (Statistical Analysis System, version 6.12; SAS Institute, Cary, NC). The receiver operating characteristic curve for transvaginal US was generated by using multiple cutoff points of 1-mm endometrial thickness. The area under the curve and its standard error were calculated by using the equivalent of the Mann-Whitney test (17). The cutoff point of endometrial thickness that maximized the accuracy of diagnosis (sensitivity and specificity) was determined. Test performance characteristics—that is, sensitivity, specificity, positive and negative predictive values, and likelihood ratios—were calculated for this cutoff point, as well as for this cutoff point combined with the finding of any focal endometrial abnormality at transvaginal US.
The test performance characteristics of hysterosonography were determined by using the presence of either an intracavitary mass or focal endometrial thickening as an abnormal finding. Ninety-five percent CIs for sensitivity, specificity, and positive and negative predictive values were calculated by using the Wilson method (18). Differences in test characteristics among the tests were examined by using the exact form of the McNemar test (19). The complication rates of hysterosonography and hysteroscopy were calculated. Differences in patient preferences for the three tests also were investigated by using the McNemar test (19).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
|
|
|
|
|
|
|
|
Office hysteroscopy was not successful in 13 (10%) of 130 women because of severe pain, cervical stenosis, or equipment failure. The complication rate was 3.1%: One patient had a hemorrhage, and three had false passages in the cervix.
Patient preference assessment forms were completed by 118 of 130 patients who underwent hysteroscopy. As shown in Figure 6, transvaginal US was well tolerated by the majority (102 of 118 [86.4%]) of patients, with no, minimal, or mild discomfort. Severe discomfort was experienced by three (2.5%) of 118 patients. However, 17 (15.2%) of 112 patients reported having severe discomfort with hysterosonography; and 16 (13.6%) of 118, with office hysteroscopy. Of the 102 patients who answered all three questions regarding their preference for the different tests, 96 (94.1%) considered transvaginal US an acceptable procedure, as compared with 85 (83.3%) patients who considered hysterosonography an acceptable procedure (P = .001) and 86 (84.3%) who considered office hysteroscopy acceptable (P = .041).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Results of several studies (3–5,11) show that asymptomatic postmenopausal women receiving tamoxifen have a thicker endometrium than do control subjects (range of reported mean endometrial thicknesses, 9.1–13.1 mm vs 3.7–4.8 mm). Using hysteroscopy with or without histopathologic analysis as the reference standard, various authors (9–11) have reported a transvaginal US endometrial thickness of 5, 9, or 10 mm as the cutoff value for identifying endometrial disease. In the study by Love et al (9), abnormalities (polyps and submucosal fibroids) were identified in only 20% of postmenopausal women with an endometrial thickness greater than 5 mm, that is, this cutoff was associated with a high false-positive scan rate. Franchi et al (10) reported that an endometrial thickness greater than 9 mm had a sensitivity of 84% and a specificity of 80% for predicting endometrial disease (polyps and hyperplasia). According to Exacoustos et al (11), an endometrial thickness of at least 10 mm was always associated with an endometrial lesion (polyp or hyperplasia) and had a sensitivity of 83% and a specificity of 100%. The results of these studies are limited by verification bias (9,10) and small sample size (11).
All women in our study were referred for outpatient hysteroscopy, with or without directed biopsy, as well as for Pipelle endometrial biopsy, irrespective of the endometrial thickness measurement. The results of our study demonstrate that an endometrial thickness greater than 6 mm is the optimal cutoff for identifying endometrial abnormalities with maximal test accuracy (sensitivity and specificity). However, if increased test sensitivity is desired, a lower cutoff, such as 5 mm, can be used but at the cost of decreased specificity.
In women treated with tamoxifen, the presence of endometrial or subendometrial cysts or adenomyosis makes the endometrial-myometrial interface indistinct to the extent that measurement of the endometrial thickness may be difficult. Prior to this study, we established that endometrial thickness could be reliably measured by two experienced observers at our institution (Sunnybrook and Women’s College Health Sciences Centre). The intraclass correlation coefficient of 0.95 (lower 95% CI, 0.92) indicated almost perfect agreement between the observers. However, transvaginal US does not consistently depict the true endometrial thickness in patients who are receiving tamoxifen. The false-positive appearance of a thick endometrium at transvaginal US has been reported by several investigators (22–24). This may account for the much lower specificity of transvaginal US in diagnosing endometrial abnormality in women receiving tamoxifen, as compared with that of transvaginal US performed in women with postmenopausal bleeding who are not receiving tamoxifen (21).
Hysterosonography is a useful adjunct to transvaginal US. It enables identification of intracavitary or endoluminal lesions and can be used to determine whether an abnormality is endometrial or subendometrial. At hysterosonography, changes of adenomyosis are seen in the subendometrial inner myometrium (23). Transvaginal US, hysterosonography, and hysteroscopy performed in women treated with tamoxifen have been evaluated in a few studies (12–15). However, in these studies, women underwent hysterosonography only when the endometrial thickness at transvaginal US was greater than 4, 5, or 8 mm (12–15). Other methodological problems, including the small number of hysterosonograms obtained (12,14,15), verification bias, and the lack of appropriate blinding (12–15), limited the usefulness of these studies for evaluation of the test characteristics of hysterosonography.
To our knowledge, our study, which included 92 successful hysterosonograms with hysteroscopic-histopathologic correlation, is the largest study of hysterosonography performed in asymptomatic postmenopausal women with breast cancer who were receiving tamoxifen. In our study, hysterosonography had a higher specificity than did transvaginal US. However, the specificity of hysterosonography in our population was lower than that reported for women with either abnormal uterine bleeding or postmenopausal bleeding who were not receiving tamoxifen (specificity, 91% and 96%, respectively) (25,26). In our study, the appearance of a cystic or edematous endometrium due to endometrial cystic atrophy was sometimes mistaken for focal areas of endometrial thickening, and this accounted for some of the false-positive diagnoses (Fig 5).
The current reference standard for endometrial assessment is operative hysteroscopy combined with histopathologic analysis. Because surgical hysteroscopy would have been too invasive for the asymptomatic women in our study, we used office hysteroscopy as the reference standard. In women with abnormal uterine bleeding, the reported sensitivity of office hysteroscopy for the detection of endometrial abnormalities is 90%, as compared with that of surgical hysteroscopy; the reported specificity is 91% (25). In asymptomatic postmenopausal women receiving tamoxifen, office hysteroscopy has a reported sensitivity of 77% and a reported specificity of 92% for identifying polyps (14). In our study, to optimize the detection of endometrial abnormalities, office hysteroscopy was performed by experienced gynecologists.
In previously published reports (14,15, 27), the hysterosonography failure rate in postmenopausal women receiving tamoxifen ranges from 12.5% to 37.0%. In our study, hysterosonography was not successful in 21.8% of cases because of severe pain and/or cervical stenosis. Because hysterosonography is technically difficult to perform in some women receiving tamoxifen and often requires cervical dilation, we now strongly advise patients to take oral analgesics, such as acetaminophen, and/or nonsteroidal antiinflammatory agents prior to the procedure.
Office hysteroscopy could not be completed in 10% of patients, which is within the range of the 8%–19% reported in previous studies (9,10,14). More than 80% of our study patients considered hysterosonography and office hysteroscopy to be acceptable procedures, although a significantly higher proportion of patients (94.1%) considered transvaginal US acceptable.
Hysteroscopic-histopathologic abnormalities (ie, polyps and submucosal fibroids) were present in 40.2% of our study population. There were no cases of endometrial hyperplasia or carcinoma. Our findings are similar to those of Love et al (9) but different from those of Franchi et al (10) and Exacoustos et al (11), who reported endometrial hyperplasia prevalences of 4% and 11%, respectively. A few cases of endometrial carcinoma were also identified in other studies (3–5). The median duration of tamoxifen therapy in our study was comparable to that in previous studies with endometrial hyperplasia and carcinoma (3–5,10). However, the maximum duration was longer in previous studies, 75–96 months (3–5,10), and the dosages of tamoxifen exceeded 20 mg/day (3,4,11)—up to 40 mg/day in one study (3). These factors may partly account for the occurrence of endometrial hyperplasia and/or carcinoma in those studies, because it has been suggested that the risk of endometrial cancer increases with the duration of tamoxifen use and with the cumulative dose (28–30). Tamoxifen therapy (20 mg/day) is now recommended for a maximum of 5 years, so our study population and results are probably more representative of and more relevant to current clinical situations.
On the other hand, our sample size may not have been large enough to enable the detection of endometrial cancer in asymptomatic tamoxifen users. According to results of the National Surgical Adjuvant Breast and Bowel Project B-14 trial and the P-1 study (6,7), the average annual rate of endometrial cancer is approximately two per 1,000 women—symptomatic and asymptomatic—who are receiving tamoxifen. The low frequency of significant findings in our study and in several recent reports (9,31,32) brings into question the value of screening asymptomatic women receiving tamoxifen (33–35). However, our study was not designed to address screening issues.
In conclusion, we report the results of a study designed to determine the test performance characteristics of transvaginal US and hysterosonography for diagnosing endometrial abnormality in patients treated with tamoxifen. We agree with recent study findings that transvaginal US is a poor screening tool because of its high false-positive rate (9,31). However, our study results demonstrate that when women with abnormal transvaginal US findings—that is, endometrial thickness greater than 6 mm or focal endometrial abnormality—are further examined with hysterosonography, the sequential combination of transvaginal US and hysterosonography has a significantly higher specificity, as compared with that of transvaginal US alone, without a significant decrease in sensitivity.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  D. L. Berridge and T. C. Winter Saline Infusion Sonohysterography: Technique, Indications, and Imaging Findings J. Ultrasound Med., January 1, 2004; 23(1): 97 - 112. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. E. Hann, C. M. Kim, M. Gonen, R. Barakat, P. H. Choi, and A. M. Bach Sonohysterography Compared With Endometrial Biopsy for Evaluation of the Endometrium in Tamoxifen-Treated Women J. Ultrasound Med., November 1, 2003; 22(11): 1173 - 1179. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 K. Fong, P. Causer, M. Atri, A. Lytwyn, and R. Kung Transvaginal US and Hysterosonography in Postmenopausal Women with Breast Cancer Receiving Tamoxifen: Correlation with Hysteroscopy and Pathologic Study RadioGraphics, January 1, 2003; 23(1): 137 - 150. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
K. M. Dudiak Invited Commentary RadioGraphics, January 1, 2003; 23(1): 151 - 155. [Full Text] [PDF]
|
|
|
|
|
|
P. C. Davis, M. J. O'Neill, I. C. Yoder, S. I. Lee, and P. R. Mueller Sonohysterographic Findings of Endometrial and Subendometrial Conditions RadioGraphics, July 1, 2002; 22(4): 803 - 816. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
) did not undergo hysterosonography (HSG) because transvaginal US (TVUS) depicted at least 2 mL of fluid in the endometrial cavity. One hundred seventeen women (*) completed hysteroscopy, 92 of whom underwent successful hysterosonography.
