| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Albert Einstein College of Medicine and Montefiore
Medical Center, 111 E 210th St., Bronx, NY 10467.
2 Department of Epidemiology and Population Health, Biostatistics Unit, Albert
Einstein College of Medicine and Montefiore Medical Center, Bronx, NY.
Received October 31, 2007;
accepted after revision March 20, 2008.
Address correspondence to M. W. Stein
(mstein17{at}aol.com).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Between March 2005 and January 2007, data from 259 patients (mean age, 47 years; age range, 18–90 years) who underwent transvaginal sonography and contrast-enhanced CT of the pelvis were analyzed retrospectively. The endometrium was quantitatively measured in millimeters on sonography. On CT it was qualitatively categorized as normal, thickened, indeterminate, or not visualized and compared with the sonography findings and original radiology reports. When the endometrium was indeterminate (thickened or triangular in shape on axial images), sagittal reconstructions were performed for final categorization. Two reviewers evaluated the CT scans and sonograms jointly with differences resolved by consensus. Kappa, Wilcoxon's rank sum test, and intraclass correlation statistics were derived.
RESULTS. The overall sensitivity and specificity of CT in detecting the thickened endometrium was 53.1% and 93.5%, respectively, relative to transvaginal sonography. The positive and negative predictive values were 66.7% and 89.1%, respectively. Kappa, the statistical measure of agreement between CT and sonography data, was 0.5049. All cases of a triangular endometrium were normal in size on sagittal reconstruction images.
CONCLUSION. Routine pelvic CT correctly identifies a normal endometrium in most patients. Sagittal reconstruction images are helpful to further evaluate the endometrium on CT in cases with a prominent or triangular endometrium because these are often related to uterine version. CT is relatively insensitive in detecting the thickened endometrium but better able to identify gross rather than subtle thickening, which must be further characterized by transvaginal sonography.
Keywords: CT • endometrium • postmenopausal • premenopausal • transvaginal sonography
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
On CT, criteria for endometrial thickness have not been well described. To our knowledge, there has been only one study evaluating the normal endometrium on CT in postmenopausal women. This study suggested that measuring the endometrium in the short axis of the uterus on axial CT was more consistent than measuring the anteroposterior diameter, but the sample size was too small to conclusively establish normal criteria [2]. Currently, the assessment of the endometrium on CT is qualitative, based on the experience and clinical judgment of the interpreting radiologist rather than quantitative criteria.
CT is a commonly performed imaging study in patients with nongynecologic complaints. Often, CT of the abdomen and pelvis performed for another complaint may incidentally show an enlarged endometrium, prompting referral for transvaginal sonography for definitive diagnosis. In our experience, occasionally an abnormality is found, but frequently it is not. This results in increased patient anxiety and health care costs. Our purpose is to assess the ability of CT in the evaluation of endometrial thickening, using transvaginal sonography as the reference standard.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
We conducted a retrospective review of pelvic CT examinations performed for various clinical indications in patients who also underwent transvaginal sonography at our institution. The indications for CT included abdominal or pelvic pain (n = 204), palpable abdominal mass (n = 16), fever (n = 12), hematuria (n = 7), oncologic follow-up (n = 6), nausea (n = 6), bloating (n = 4), ascites (n = 2), and follow-up for diverticulitis (n = 2). Five postmenopausal patients had a history of abnormal vaginal bleeding. The study was initiated after the institutional review board granted an exempt status. Patient consent for the study was not required. Data were obtained by searching our institutional radiologic database from March 2005 to January 2007 to identify women who underwent both examinations. The interval between CT and sonography in patients who were selected for the study was 48 hours or less for all premenopausal women and those postmenopausal women with vaginal bleeding. The interval between studies for postmenopausal women without vaginal bleeding was 30 days or less. Postmenopausal status was defined as date of last menstrual period (LMP) more than 1 year before the imaging study. Patient medical charts were reviewed to obtain the date of LMP and history of hormone replacement therapy.
Images were analyzed by two of three readers with differences resolved by consensus. Two readers were fellowship-trained, experienced, board-certified radiologists specializing in abdominal imaging. One reader was a board-certified abdominal imaging fellow. Sonograms and CT scans were read on different weeks with readers blinded to identifying information and the results of the other technique to avoid recall bias.
Patient age was known at time of interpretation of all studies. Readers usually knew LMP or menopausal status or both when analyzing both techniques because the sonography technologist routinely documents this information on the first sonogram in the PACS. Chart review was performed for patients whose LMP was not available in the PACs to determine the LMP. When the LMP or menopausal status could not be obtained from chart review or the PACS, it was assumed that those women less than 50 years old were premenopausal and those 50 years and over were postmenopausal. LMP was unknown in a total of 55 patients—10 patients who were less than 40 years, 10 patients 40–49 years, three patients 50–59 years, and 32 patients more than 60 years.
Sonography was performed with an Acuson XP or Sequoia system (Siemens Medical Solutions) using narrow-band variable-frequency transducers (5-8–MHz simple end-fire transducer for transvaginal examinations) or with an ATL HDI 5000 system (Philips Healthcare) using wideband transducers (4-8–MHz transvaginally). The highest frequency transducer possible was used for each patient. CT was performed with either the 4-MDCT LightSpeed Plus or 64-MDCT HiSpeed Advantage scanner (GE Healthcare). Scans were obtained after IV administration of iohexol 300 (Omnipaque 300, GE Healthcare) or iopromide 300 (Ultravist 300, Bayer HealthCare) administered with a single uniphasic bolus of 100–140 mL via a power injector at 2–3 mL/s. Routine CT of the abdomen and pelvis was performed in the portal venous phase, with a 70–80 second delay and a pitch between 1.2 and 1.5 before contrast medium was excreted into the bladder. Images were acquired at 1.5-mm slice thickness and reviewed in the axial plane at 5-mm slice thickness and in the coronal plane at 3-mm slice thickness.
The endometrium was assessed quantitatively on sonography by measuring it
in the sagittal plane on its thickest portion, excluding the hypoechoic inner
myometrium, from echogenic border to echogenic border using electronic
calipers, and its thickness was recorded in millimeters
[3]. Because the endometrium
and intracavitary fluid usually cannot be distinguished on CT, any intervening
fluid was included in the sonographic measurements. The endometrium was
considered normal by sonographic criteria if it was 
16 mm in
premenopausal women and 5 mm in postmenopausal patients
[1]. Because the phase of
menses is generally unknown at the time of CT interpretation, the upper limits
of normal for the secretory phase ( 16 mm) was used to differentiate a
thickened from nonthickened endometrium for the premenopausal cases.
|
|
|
|
|
|
On sonography, cases with a thickened endometrium were measured, and an increment between top normal endometrial thickness for age and measured thickness on sonography was calculated. Both discordant and concordant CT results of these patients were compared with sonography. Statistical comparison between the incremental values was performed for both the concordant and discordant cases.
Interobserver reliability studies using an intraclass correlation coefficient for sonography, given that the data were continuous, and a kappa test for CT, given that the data were dichotomous, were performed to compare the results from this study with the official report previously generated [4]. If the endometrium was not specifically mentioned in the official CT report, it was presumed to be normal. A focused interreader variability analysis was performed for the cases in which the endometrium appeared enlarged or triangular, comparing our interpretation using sagittal reconstructions with that of the original reviewers.
In addition, we performed an interreader variability study using an intraclass correlation coefficient for sonography and kappa test for CT to compare the results of cases from this study, which were derived from consensus, with those results from an additional independent reviewer, experienced in abdominal imaging. This reviewer used similar criteria for endometrial assessment as the consensus readers in our study.
The sensitivity, specificity, and positive and negative predictive values of the visual assess ment of the endometrium on CT compared with the conventional quantitative assessment on transvaginal sonography were calculated for the overall study population and stratified by meno pausal status.
Kappa statistics were derived to analyze the agreement between CT and sonography for the entire study population and for the pre- and postmenopausal subgroups. Kappa statistics were interpreted according to Landis and Koch [5]: < 0, less than chance agreement; 0–0.20, low agreement; 0.21–0.40, moderate; 0.41–0.60, good; 0.61–0.80, substantial; and > 0.81, perfect agreement. Results were considered statistically significant for p values less than 0.05.
A Wilcoxon's rank sum test was used to compare the difference in the mean incremental endometrial thickness above the normal among those detected and not detected on CT. Age and the presence of submucosal fibroids on CT or sonography were analyzed as possible factors influencing interpretation of the endometrium on CT and sonography using multiple logistic regression analysis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
The mean endometrial thickness of the nine patients whose endometrium was nonvisualized on CT was 4 mm (range, 1–7 mm) on sonography. For detection of endometrial thickening in the entire patient population, sensitivity, specificity, and positive and negative predictive values of CT were 53.1%, 93.5%, 66.7%, and 89.1%, respectively. The overall kappa, the statistical measure of agreement between the two techniques, was 0.5049 (95% CI, 0.3653–0.6445) (Table 1).
|
In the premenopausal groups, 162 patients had a visualized endometrium on CT. Of these, 155 patients had normal and seven had a thickened endometrium on sonography. Of the 155 patients with normal endometrial thickness on sonography, the endometrium was normal on CT in 146 (94.2%). Of the seven patients who had a thickened endometrium on sonography, the endometrium was thickened on CT in three (42.9%) patients. For detection of endometrial thickening in the premenopausal group, sensitivity, specificity, and positive and negative predictive values of CT were 42.9%, 94.2%, 25.0%, and 97.3%, respectively. Kappa was 0.2763 (95% CI, 0.0022–0.5548) (Table 2).
|
Of 88 postmenopausal patients whose endometrium was seen on CT, 46 had normal and 42 had a thickened endometrium on sonography. Of the 46 patients with a normal endo metrium on sonography, the endometrium was normal on CT in 91.3% (42/46). Of the 42 patients who had a thickened endometrium on sonography, the endometrium was thickened on CT in 54.8% (23/42). For detection of endometrial thickening in the postmenopausal group, sensitivity, specificity, and positive and negative predictive values of CT were 54.8%, 91.3%, 85.2%, and 68.9%, respectively. Kappa was 0.4679 (95% CI, 0.2930–0.6428) (Table 2).
Interrater variability was calculated for CT and sonography of the entire study population by comparing the results in the original radiology report with the results obtained by our reviewers. The intraclass correlation coefficient (ICC) calculated for transvaginal sonography measurement of the endometrium was found to have strong agreement, ICC = 0.8886 (95% CI, 0.848–0.906). However, for CT, the agreement was found to be 0.3769 (95% CI, 0.2153–0.5385).
The kappa value for CT measurement for the independent reviewer using the same criteria for endometrial evaluation compared with the consensus readers was 0.8757 (95% CI, 0.7962–0.9552). The ICC calculated for transvaginal sonography measurement for the independent reviewer of the endometrium was 0.8886 (95% CI, 0.848–0.906).
A special interrater variability study was performed on the triangular endometrium because our colleagues not involved in this study who generated the CT reports did not perform sagittal reconstruction imaging. The kappa was –0.0909 (95% CI, –0.1908 to 0.0089), indicating less than chance agreement. An additional interrater variability study was performed on the 14 cases that lacked consensus. The kappa was 0.1429 (95% CI, –0.1242 to 0.4099).
Analysis of factors such as age (p = 0.1510) and submucosal fibroids on sonography or CT (p = 0.9714, p = 0.9629, respectively) did not reveal any factor predictive of concordance or discordance between the two techniques. Twenty-one (8.1%) of the patients had submucosal fibroids seen on sonography, and 25 (9.7%) had submucosal fibroids seen on CT.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Our study shows that when CT depicts an apparently normal endometrial thickness, there is a high likelihood that the endometrium is indeed normal on sonography, as reflected by the high negative predictive value (89.1%). However, our study found an overall low sensitivity (53.1%) of CT in detecting a thickened endometrium regardless of age or menopausal status. CT detected only half of the cases of thickened endometrium diagnosed on sonography. Yet, when the endometrium is identified as thickened on CT, it is confirmed to be truly thickened on sonography in two-thirds of cases. CT overcalls endometrial thickening in one-third of cases. Therefore, transvaginal sonography remains the reference standard and should be used to confirm the CT findings.
In the analysis of cases with discordant results between CT and sonography, we found that when the endometrium was interpreted as normal on CT but thickened on sonography, the mean difference between the measured endometrial thickness and accepted top normal sonographic value was found to be small (3 mm). In contrast, among cases with concordant results, when the endometrium was interpreted as thickened on both techniques, the mean difference between the measured endometrial thickness and accepted top normal sonographic value was larger (7 mm) (p < 0.002). Thus, those cases misclassified on CT had a significantly thinner endometrium on sonography compared with those classified correctly. This indicates that CT better detects gross rather than subtle endometrial thickening.
It was observed in some cases that the uterus is imaged in its true coronal plane on axial CT because of ante- or retroversion. In these cases, on axial CT, the endometrium has a distinctive triangular shape and may be misinterpreted as thickened. We performed sagittal reconstruction imaging of these cases, which showed a normal endometrium in all. Similarly, sagittal reconstructions were also useful in the cases where consensus of endometrial status could not be reached showing a normal-appearing endometrium in half of these cases. Therefore, sagittal reformation images are helpful in the evaluation of endometrial thickness in indeterminate cases.
In addition, we found poor overall agreement between the original CT reports and our study findings on CT, whereas the interrater agreement with respect to the original sonography interpretation was excellent. This poor agreement between the CT report and study results may be attributed to the absence of universal standards for endometrial size on CT in contradistinction to sonography. Also, it was suspected that those cases with a triangular appearance of the endometrium on CT were overcalled as thickened on the original reports because the interpreting radiologists were not aware of this pitfall and sagittal reconstruction images were not obtained.
To our knowledge, there has been only one other publication in the literature addressing the ability of CT to assess endometrial thickness. This study was composed of postmenopausal women and was limited by a small sample size [2]. It concluded that the short-axis measurement of the endometrium was the most accurate and reproducible method of assessing endometrial thickness. Although we did not measure the endometrium, we evaluated its thickness along its transverse axis. In addition, several articles have studied the ability of CT to stage myometrial invasion in endometrial carcinoma [6, 7]. The goal of our study was not to identify or stage endometrial pathology but to assess the ability of CT to identify endometrial thickening against sonography as the reference standard. Endometrial thickening by itself has numerous causes, and it may or may not be associated with an actual morphologic or pathologic lesion but if identified must be further investigated.
One limitation of this study is its retrospective design, which resulted in limited patient clinical information, such as LMP, clinical symptoms, and hormonal status in some of our patients. However, a prospective study may have unnecessarily subjected some patients to radiation, iodinated contrast media, or transvaginal sonography and would not be ethical. Another limitation is our assumption in those patients with unknown menstrual status that an age of less than or more than 50 years can define pre-versus postmenopausal status. Time of meno pause is influenced by a multitude of factors other than age. However, among patients in the perimenopausal age group (50–59-years) when onset of menopause is particularly variable, only three patients in this age group (n = 43) had unknown menopausal status, of whom only one had a question of an enlarged endometrium on sonography. The other two patients had a normal endometrium even using the strict postmenopausal criteria.
Also, our study group has an inherent selection bias compared with the general popula tion because these women were referred for pelvic sonography and CT on the basis of symptoms and physical findings. Some were premenopausal and some postmenopausal and thus had different risk profiles for endometrial pathology. It must be noted that sonography, not histologic correlation, was considered the reference standard in our study against which we qualitatively assessed the endometrium on CT.
Another limitation of this study is our visual qualitative evaluation of endometrial thickness instead of using quantitative measurements because measuring small structures on CT is known to be inaccurate. In the case of the endometrium, it is particularly difficult because the outer borders of the endometrial lining are not sharply delineated on CT. Accuracy of CT measurement depends on many variables including slice thickness, scanning technique, and bolus timing. Also, it probably depends on patient factors that affect uterine enhancement including individual uterine vasculature, patient age, and menstrual status. It is possible that our routine CT protocol is less than optimal for endometrial imaging because endometrial-to-inner myometrial contrast varies with different phases of contrast enhancement [8]. We used a 70–90 second delay for abdominal and pelvic imaging, and we did not perform delayed phase imaging. In the majority of our patients, 250 of 259 (96.5%), the endometrium was identified using this routine protocol. Also, more independent reviewers would have been beneficial to achieve more reliable estimates of interreader reliability.
For the purposes of this study, we did not classify premenopausal patients by phase of menstrual cycle. This was done to mirror clinical practice because during CT interpretation the LMP and menstrual phase are gener ally unknown. Assuming all premenopausal women are in the secretory phase is a limitation of this study because such is not the case in day-to-day sonographic practice in which LMP and menstrual phase are known.
In conclusion, CT is an insensitive technique for detecting mild endometrial thickening in both pre- and postmenopausal women. However, the endometrium is often sufficiently visualized on contrast-enhanced CT to detect gross thickening and should not be overlooked at routine interpretation. Patients with a thickened endometrium should be evaluated further by sonography, which remains the technique of choice for assessment of the endometrium. Having a good negative predictive value, CT correctly identifies a normal endometrium in the majority of cases. In patients whose endometrium appears triangular on axial images, sagittal reformation images can be easily obtained on a PACS workstation to confirm the normal appearance and obviate further evaluation with sonography. In all patients in our series, the triangular-appearing endometrium on axial CT was associated with normal endometrial thickness and reflected the degree of uterine version.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
