British Journal of Radiology (2005) 78, 394-397
© 2005 British Institute of Radiology
doi: 10.1259/bjr/60976544
Accuracy of transvaginal ultrasound in diagnosing endometrial pathology in women with post-menopausal bleeding on tamoxifen
J Weaver, MSc, DCR1,
J M McHugo, FRCR, FRCP, FRCPCH1 and
T J Clark, MD, MRCOG2
1 Department of Radiology and 2 Academic Department of Obstetrics and Gynaecology, Birmingham Women's Hospital, Birmingham B15 2TG, UK
Correspondence: Dr Jo McHugo
 |
Abstract
|
|---|
The incidence of endometrial pathology is increased in women with tamoxifen-induced post-menopausal bleeding (PMB). The aim of this study was to determine the accuracy of transvaginal ultrasound (TVS) in diagnosing endometrial pathology in symptomatic women taking tamoxifen, using endometrial thickness measurements (5 mm and 10 mm cut-offs) and morphological changes within the uterine cavity to define abnormality. The sensitivity of TVS was 97% (95% confidence interval (CI) 83% to 100%) and the specificity 35% (95% CI 17% to 56%) at a 5 mm threshold. The corresponding likelihood ratios for a positive test were 1.47 (95% CI 1.16–2.10) and a negative test were 0.09 (95% CI 0.02–0.53). The use of a higher 10 mm threshold, characterization of endometrial morphology or combining endometrial thickness and morphology did not improve overall accuracy. TVS measurement of endometrial thickness using a 5 mm cut-off is highly accurate in excluding endometrial disease in symptomatic women taking tamoxifen and may negate the need for further diagnostic testing in women unless symptoms recur. A positive TVS result is of little value, so further testing is mandatory in order to detect benign and malignant endometrial disease associated with tamoxifen and PMB.
|
Introduction
|
|---|
Tamoxifen is an effective and widely used adjuvant therapy for women with breast cancer because it reduces recurrence rates and prolongs disease free survival. It is a partial oestrogen agonist in that it exhibits anti-oestrogenic activity in the breast, but has a stimulatory effect on the endometrium. This hormonal activity results in a higher incidence of post-menopausal vaginal bleeding and endometrial pathology such as endometrial polyps, hyperplasia and cancer [1–3].
Transvaginal ultrasound (TVS) measurement of endometrial thickness has been demonstrated to have high accuracy in excluding endometrial polyps, hyperplasia and cancer in women with post-menopausal bleeding [4, 5]. Several studies have examined the usefulness of endometrial thickness measurement using TVS in asymptomatic women on tamoxifen, as they are at increased risk of significant endometrial disease [6, 7]. These studies have consistently shown the accuracy of TVS to be poor, especially in detecting endometrial disease, and thereby of limited value as a screening test. This is partly because tamoxifen can induce specific morphological endometrial changes, particularly subendometrial cysts, and such appearances can make characterization and measurement of the endometrium difficult [8, 9]. Only a few studies have assessed the accuracy of TVS in symptomatic women taking tamoxifen (i.e. those with post-menopausal bleeding) [6, 10] and these women have represented small subgroups of larger asymptomatic populations. The role of TVS in the diagnosis of endometrial pathology in symptomatic women taking tamoxifen is thus less clearly defined. We therefore undertook a study to estimate the accuracy of TVS in post-menopausal women on tamoxifen presenting with abnormal vaginal bleeding.
|
Methods and materials
|
|---|
Ethical approval for the study was obtained from our hospital's Research and Development Committee. A retrospective study was undertaken, at the Birmingham Women's Hospital, using the Radiology Department computer system to identify all women receiving tamoxifen therapy who presented with post-menopausal vaginal bleeding (PMB) and were investigated with a TVS and outpatient endometrial biopsy, between April 1st 1996 and March 31st 2000. Of the 126 women consecutively investigated, data from a random sample of 56 were retrieved for further analysis. The aim was to determine the accuracy of (a) endometrial thickness and (b) morphological changes within the uterine cavity as characterized on TVS in diagnosing serious endometrial disease (endometrial polyps, hyperplasia and cancer). The endometrial thickness was measured at the double layered thickness to include sub endometrial cysts/cystic spaces if visualized. Morphological changes were characterized as normal if the endometrium was regular or subendometrial cystic spaces consistent with tamoxifen change [8, 9] were seen. Structural changes, which could not be attributed to characteristic tamoxifen effects, were considered to represent abnormal morphology (irregular endometrium, heterogeneous echotexture and indistinct endometrial borders). All TVS were performed by Departmental sonographers or radiologists using an Acuson Sequoia with an EV8C4 probes or Acuson 128XP/10 with an EC7 probe (Acuson, Mountain View, CA). Ultrasound findings were reported independently and blinded from the pathologist interpreting the subsequent endometrial biopsy sample, which was the diagnostic reference standard.
To generate measures of diagnostic accuracy, 2 x 2 contingency tables were constructed. Endometrial disease was defined as endometrial polyps, hyperplasia or cancer. Atrophic endometrium, functional changes and non-diagnostic endometrial samples were considered to be negative for disease. Where more than one endometrial condition was identified histologically, the more serious condition was taken as the final diagnosis (cancer>hyperplasia>polyp). Normal and abnormal test results were defined firstly on the basis of endometrial thickness measurements using two different thresholds to define abnormality (abnormal test >5 mm and >10 mm). These thresholds were chosen as the former represents a recommended cut-off for investigating PMB [4, 5] and the latter cut-off has been recommended by some [11, 12] for investigating post-menopausal women taking tamoxifen (the higher threshold accounting for expected tamoxifen induced endometrial/subendometrial changes). Abnormality was then defined on the basis of abnormal endometrial morphology (present or absent) and finally on taking into account both thickness and morphology (abnormal test >5 mm or 10 mm, respectively, and/or abnormal morphology vs normal test 
5 mm or 10 mm, respectively and normal morphology).
Sensitivity, specificity, predictive values and likelihood ratios (LRs) were derived from the 2 x 2 tables to enable estimation of diagnostic accuracy. The LRs indicate by how much a given test would raise or lower the probability of having pathology. An LR of >1 increases the probability of having pathology and an LR of <1 decreases the probability of having pathology with a positive and negative test result, respectively. Moderate accuracy can be achieved with LR values of 5–10 and 0.1–0.2, whereas LRs of 1–5 and 0.2–1 would be poor [13, 14].
|
Results
|
|---|
Of the 56 women undergoing TVS and endometrial biopsy, 30 of 56 (54%) had endometrial disease. There was only one case of endometrial cancer, 10 cases of endometrial hyperplasia (one with cytological atypia) and 19 endometrial polyps.
Endometrial thickness
The 2 x 2 contingency table using a 5 mm cut-off for abnormality on TVS is shown in Table 1
. Sensitivity was 97% (95% confidence interval (CI) 83–100%) and the specificity 35% (95% CI 17–56%). Positive and negative predictive values were 63% (95% CI 48–77%) and 90% (95% CI 56–100%), respectively. The corresponding likelihood ratios for a positive test were 1.47 (95% CI 1.16–2.10) and a negative test were 0.09 (95% CI 0.02–0.53).
View this table:
[in this window]
[in a new window]
|
Table 1. Accuracy of transvaginal ultrasound (TVS) in diagnosing endometrial disease (polyps, hyperplasia or cancer) using a >5 mm cut-off to define abnormality
|
|
Increasing the cut-off for abnormality to 10 mm reduced the sensitivity to 77% (95% CI 57–90%), but increased the specificity to 54% (95% CI 33–73%). Positive and negative predictive values were 66% (95% CI 48–81%) and 67% (95% CI 43–85%) respectively (Table 2). The corresponding likelihood ratios for a positive test were 1.67 (95% CI 1.09–2.75) and a negative test were 0.43 (95% CI 0.20–0.88).
View this table:
[in this window]
[in a new window]
|
Table 2. Accuracy of transvaginal ultrasound (TVS) in diagnosing endometrial disease (polyps, hyperplasia or cancer) using a >10 mm cut-off to define abnormality
|
|
Endometrial morphology
The 2 x 2 contingency table for endometrial morphology on TVS is shown in Table 3. Sensitivity was 17% (95% CI 6–35%) and the specificity 81% (95% CI 61–93%). Positive and negative predictive values were 50% (95% CI 19–81%) and 46% (95% CI 31–61%), respectively. The corresponding likelihood ratios for a positive test were 0.87 (95% CI 1.30–2.55) and a negative test were 1.03 (95% CI 0.79–1.38).
View this table:
[in this window]
[in a new window]
|
Table 3. Accuracy of transvaginal ultrasound (TVS) in diagnosing endometrial disease (polyps, hyperplasia or cancer) using abnormal endometrial morphology to define abnormality
|
|
Combination of endometrial morphology and thickness
The 2 x 2 contingency table for endometrial thickness (5 mm cut-off) and morphology combined on TVS is shown in Table 4. Sensitivity was 97% (95% CI 83–100%) and the specificity 11% (95% CI 2–30%). Positive and negative predictive values were 56% (95% CI 41–70%) and 75% (95% CI 19–99%), respectively. The corresponding likelihood ratios for a positive test were 1.09 (95% CI 0.92–1.37) and a negative test were 0.29 (95% CI 0.04–1.90).
View this table:
[in this window]
[in a new window]
|
Table 4. Accuracy of transvaginal ultrasound (TVS) in diagnosing endometrial disease (polyps, hyperplasia or cancer) using a >5 mm cut-off and/or abnormal endometrial morphology to define abnormality
|
|
The 2 x 2 contingency table for endometrial morphology on TVS is shown in Table 5. Sensitivity was 83% (95% CI 65–94%) and the specificity 50% (95% CI 30–70%). Positive and negative predictive values were 66% (95% CI 49–80%) and 72% (95% CI 47–90%), respectively. The corresponding likelihood ratios for a positive test were 1.67 (95% CI 1.14–2.65) and a negative test were 0.34 (95% CI 0.14–0.77).
View this table:
[in this window]
[in a new window]
|
Table 5. Accuracy of transvaginal ultrasound (TVS) in diagnosing endometrial disease (polyps, hyperplasia or cancer) using a >10 mm cut-off and/or abnormal endometrial morphology to define abnormality
|
|
|
Discussion
|
|---|
This study shows that endometrial thickness measurement using a 5 mm cut-off to define an abnormal ultrasound is highly accurate in excluding endometrial disease in symptomatic women taking tamoxifen and is thus useful in clinical decision making. A positive test result is of little use in detecting disease. These findings are in keeping with accuracy estimated in the general PMB populations using TVS and a 5 mm cut-off [4, 5]. A 10 mm cut-off has been advocated in women on tamoxifen so that characteristic "benign" tamoxifen induced endometrial changes (e.g. subendometrial cysts) are accounted for, in an attempt to reduce the false positive rate and subsequent unnecessary medical intervention [11, 12]. However, by using such a cut-off in this series, the accuracy of a negative test in excluding endometrial disease was reduced with no substantial improvement in the predictive ability of a positive ultrasound. This cut-off cannot therefore be recommended in women on tamoxifen presenting with PMB. The use of morphological endometrial changes to define abnormality was not discriminative and thereby of no use to the clinician. Accuracy was not improved when morphology was combined with endometrial thickness measurements.
It is well recognized that tamoxifen-induced changes to the uterine cavity can make characterization and measurement of the endometrium difficult [6, 7, 12]. Accuracy data for TVS in symptomatic women with PMB on tamoxifen is sparse. Most available data represent small subgroups from larger asymptomatic populations [6, 10]. Our study therefore provides important information as it evaluated symptomatic women from the outset using endometrial morphology in addition to endometrial thickness measurements. TVS was conducted by several radiologists and radiographers and this may raise concerns regarding the reproducibility and accuracy of the TVS findings. However, a prospectively agreed protocol for performing and reporting endometrial findings on TVS was followed in all cases, thereby limiting any such biases.
In addition to its clinical importance, strengths of the study include complete verification of test diagnoses and blinding of reference standard interpretation. Inferences from this study are limited by the small sample size and the retrospective, non-consecutive recruitment of patients. Although a consecutive series of eligible women may be considered ideal, our findings should still be considered generalizable because a random sub-selection of all eligible women was studied [15]. Our study may also be criticised for the sole use of endometrial histology obtained using outpatient miniature endometrial biopsy as the diagnostic reference standard. However, these devices have been shown to display good accuracy in diagnosing endometrial cancer and hyperplasia [16, 17] and its use as a diagnostic reference standard is thus reasonable and indeed well-established when assessing the accuracy of TVS [4, 5] and as part of standard clinical practice. The accuracy of endometrial biopsy in diagnosing the presence or absence of endometrial polyps is however, less well established and the addition of hysteroscopy may have been useful to improve accuracy of the reference test. Despite these potential limitations, our findings are in keeping with precise estimates of accuracy derived from systematic reviews of the general PMB population [4, 5].
The findings of our study have implications for diagnostic work-up of women presenting with PMB on tamoxifen. Additional testing is required following a positive TVS (5 mm cut-off) in the general PMB population this usually constitutes an outpatient endometrial biopsy in order to exclude endometrial cancer or hyperplasia [4, 5]. However, disease prevalence is increased in women with tamoxifen induced PMB [1–3]. Our study suggests that over 80% of women will test positive if a 5 mm TVS cut-off is employed and of these, two thirds will have endometrial pathology and approximately two-thirds of this pathology will be endometrial polyps. Additional testing with blind endometrial biopsy alone is therefore inadequate as focal pathology may be missed [18]. Saline infusion ultrasound or outpatient hysteroscopy [19, 20] should therefore be employed followed by polypectomy if appropriate, in addition to endometrial biopsy, which provides tissue for histology.
In conclusion, a normal TVS (endometrial thickness 5 mm) may negate the need for further diagnostic testing in women with tamoxifen-induced PMB unless symptoms recur. Characterization of endometrial morphology does not appear to usefully improve diagnostic performance of TVS. A positive TVS result is of little value, so further testing is mandatory in order to detect benign and malignant endometrial disease associated with tamoxifen and PMB. Future research should look at the use of Doppler blood flow changes [21], saline infusion ultrasound [20] and three-dimensional ultrasound in influencing the diagnostic accuracy of TVS in women with tamoxifen-induced PMB. One randomized controlled trial has compared ultrasound and outpatient hysteroscopy for screening asymptomatic women on tamoxifen [22]. Such trial should also be conducted in symptomatic women to compare the diagnostic effectiveness of different diagnostic work up strategies.
Received for publication July 27, 2004.
Revision received October 25, 2004.
Accepted for publication November 23, 2004.
|
References
|
|---|
- Neven P, Vernaeve H. Guidelines for monitoring patients taking tamoxifen treatment. Drug Saf 2000;22:1–11.[CrossRef][Medline]
- Barakat RR, Gilewski TA, Almadrones L, Saigo PE, Venkatraman E, Hudis C, et al. Effect of adjuvant tamoxifen on the endometrium in women with breast cancer: a prospective study using office endometrial biopsy. J Clin Oncol 2000;18:3459–63.[Abstract/Free Full Text]
- Bergman L, Beelen ML, Gallee MP, Hollema H, Benraadt J, van Leeuwen FE. Risk and prognosis of endometrial cancer after tamoxifen for breast cancer. Comprehensive Cancer Centres' ALERT Group. Assessment of liver and endometrial cancer risk following tamoxifen. Lancet 2000;356:881–7.[CrossRef][Medline]
- Gupta JK, Chien PFW, Voit D, Clark TJ, Khan KS. Ultrasonographic endometrial thickness for diagnosing endometrial pathology in women with postmenopausal bleeding: a meta-analysis. Acta Obstet Gynecol Scand 2002;81:799–816.[CrossRef][Medline]
- Smith-Bindman R, Kerlikowske K, Feldstein VA, Subak L, Scheidler J, Segal M, et al. Endovaginal ultrasound to exclude endometrial cancer and other endometrial abnormalities. JAMA 1998;280:1510–7.[Abstract/Free Full Text]
- Fung MF, Reid A, Faught W, Le T, Chenier C, Verma S, et al. Prospective longitudinal study of ultrasound screening for endometrial abnormalities in women with breast cancer receiving tamoxifen. Gynecol Oncol 2003;91:154–9.[CrossRef][Medline]
- Love CD, Muir BB, Scrimgeour JB, Leonard RC, Dillon P, Dixon JM. Investigation of endometrial abnormalities in asymptomatic women treated with tamoxifen and an evaluation of the role of endometrial screening. J Clin Oncol 1999;17:2050–4.[Abstract/Free Full Text]
- McGonigle KF, Shaw SL, Vasilev SA, Odom-Maryon T, Roy S, Simpson JF. Abnormalities detected on transvaginal ultrasonography in tamoxifen-treated postmenopausal breast cancer patients may represent endometrial cystic atrophy. Am J Obstet Gynecol 1998;178:1145–50.[CrossRef][Medline]
- Goldstein SR. Unusual ultrasonographic appearance of the uterus in patients receiving tamoxifen. Am J Obstet Gynecol 1994;170:447–51.[Medline]
- Giorda G, Crivellari D, Veronesi A, Perin T, Campagnutta E, Carbone A, et al. Comparison of ultrasonography, hysteroscopy, and biopsy in the diagnosis of endometrial lesions in postmenopausal tamoxifen-treated patients. Acta Obstet Gynecol Scand 2002;81:975–80.[CrossRef][Medline]
- Strauss HG, Wolters M, Methfessel G, Buchmann J, Koelbl H. Significance of endovaginal ultrasonography in assessing tamoxifen-associated changes of the endometrium. A prospective study. Acta Obstet Gynecol Scand 2000;79:697–701.[CrossRef][Medline]
- Gerber B, Krause A, Muller H, Reimer T, Kulz T, Makovitzky J, et al. Effects of adjuvant tamoxifen on the endometrium in postmenopausal women with breast cancer: a prospective long-term study using transvaginal ultrasound. J Clin Oncol 2000;18:3464–70.[Abstract/Free Full Text]
- Jaeschke R, Guyatt GH, Sackett DL. User's guides to the medical literature: III. How to use an article about a diagnostic test: B. What are the results and will they help me in caring for my patients? JAMA 1994;271:703–7.[CrossRef][Medline]
- Bakour SH, Khan KS, Gupta JK. Transvaginal ultrasonography and endometrial histology in peri- and postmenopausal women on hormone replacement therapy. Br J Obstet Gynaecol 2000;107:295.
- Bossuyt PM, Reitsma JB, Bruns DE, Gatsonis CA, Glasziou PP, Irwig LM, et al. The STARD statement for reporting studies of diagnostic accuracy: explanation and elaboration. Ann Intern Med 2003;138:W1–12.[Abstract/Free Full Text]
- Clark TJ, Mann CH, Shah N, Song F, Khan KS, Gupta JK. Accuracy of outpatient endometrial biopsy in the diagnosis of endometrial hyperplasia: a systematic review. Acta Obstet Gynecol Scand 2001;80:784–93.[CrossRef][Medline]
- Clark TJ, Mann CH, Shah N, Song F, Khan KS, Gupta JK. Accuracy of outpatient endometrial biopsy in the diagnosis of endometrial cancer: a systematic review. Br J Obstet Gynecol 2002;109:313–21.
- Guido RS, Kanbour-Shakir A, Rulin MC, Christopherson WA. Pipelle endometrial sampling: sensitivity in the detection of endometrial cancer. J Reprod Med 1995;40:553–5.[Medline]
- Neven P, De Muylder X, Van Belle Y, Van Hooff I, Vanderick G. Longitudinal hysteroscopic follow-up during tamoxifen treatment. Lancet 1998;351:36.[Medline]
- Garuti G, Cellani F, Grossi F, Colonnelli M, Centinaio G, Luerti M. Saline infusion sonography and office hysteroscopy to assess endometrial morbidity associated with tamoxifen intake. Gynecol Oncol 2002;86:323–9.[CrossRef][Medline]
- Develioglu OH, Omak M, Bilgin T, Esmer A, Tufekci M. The endometrium in asymptomatic breast cancer patients on tamoxifen: value of transvaginal ultrasonography with saline infusion and Doppler flow. Gynecol Oncol 2004;93:328–35.[CrossRef][Medline]
- Timmerman D, Deprest J, Bourne T, Van den Berghe I, Collins WP, Vergote I. A randomized trial on the use of ultrasonography or office hysteroscopy for endometrial assessment in postmenopausal patients with breast cancer who were treated with tamoxifen. Am J Obstet Gynecol 1998;79:62–70.
Top
Abstract
Introduction
