Fax: (011) +81-6-6646-5800
Transcervical needle biopsy for the differential diagnosis between uterine sarcoma and leiomyoma
Article first published online: 15 MAR 2002
Copyright © 2002 American Cancer Society
Volume 94, Issue 6, pages 1713–1720, 15 March 2002
How to Cite
Kawamura, N., Ichimura, T., Ito, F., Shibata, S., Takahashi, K., Tsujimura, A., Ishiko, O., Haba, T., Wakasa, K. and Ogita, S. (2002), Transcervical needle biopsy for the differential diagnosis between uterine sarcoma and leiomyoma. Cancer, 94: 1713–1720. doi: 10.1002/cncr.10382
- Issue published online: 15 MAR 2002
- Article first published online: 15 MAR 2002
- Manuscript Accepted: 31 OCT 2001
- Manuscript Revised: 22 OCT 2001
- Manuscript Received: 3 SEP 2001
- The Osaka Medical Research Foundation for Incurable Diseases, Osaka, Japan
- Aventis Pharma Ltd., Tokyo, Japan
- uterine sarcoma;
- endometrial stromal sarcoma;
- needle biopsy;
- differential diagnosis
The clinical differential diagnosis between uterine sarcoma and benign leiomyoma is difficult even with magnetic resonance imaging (MRI). Therefore, a considerable number of patients have undergone hysterectomies due to an indication of “suspected malignancy” based on tumor size alone. However, approximately 80% of these hysterectomies have been judged to have been recommended inappropriately. In such situations, reliable preoperative diagnostic tests are required. The authors have evaluated the accuracy of needle biopsy for uterine myoma-like tumors, a procedure that to the authors' knowledge has been performed infrequently.
Transcervical needle biopsy was performed in 435 patients with uterine myoma-like tumors. The biopsy specimens were scored for degree of malignancy according to the histopathologic criteria proposed by Bell et al. Histopathologic evaluation of surgical specimens and clinical outcome after 2 years of follow-up were used as the reference standards.
Of 435 patients, 7 had uterine sarcomas, 4 of which were scored as ≥ 4 points and were diagnosed as “sarcoma” by needle biopsy alone. No sarcoma cases were included in the group of patients with a score of 0. The cutoff score combining the highest sensitivity and specificity with respect to distinguishing uterine leiomyosarcoma from uterine leiomyoma was 2; sensitivity, specificity, and positive and negative predictive values were 100%, 98.6%, 58%, and 100.0%, respectively.
Transcervical needle biopsy using histopathologic scoring is a reliable diagnostic test for the differential diagnosis between uterine sarcoma and leiomyoma. This diagnostic method, combined with MRI screening, could reduce the number of patients currently undergoing unnecessary surgery. Cancer 2002;94:1713–20. © 2002 American Cancer Society.
Uterine leiomyoma is a common disease, occurring in 20–25% of women age > 30 years.1 Uterine leiomyoma is the indication for approximately 30% of all hysterectomies performed, (i.e., an estimated 200,000 per year in the U.S).2 However, approximately 80% of hysterectomies performed because of uterine leiomyoma have been judged to have been recommended inappropriately.3 The most common reason is the lack of adequate diagnostic evaluation;3 despite the presence of no or only mild symptoms, a considerable number of patients have undergone hysterectomies because of a “suspected malignancy” based on tumor size alone.
In contrast, the recent clinical introductions of gonadotropin-releasing hormone (GnRH) agonists4 and uterine arterial embolization5 have meant that patients with uterine leiomyoma, who might previously have undergone surgery, now often are managed conservatively.
In both of the above situations, there is an important common clinical problem: distinguishing uterine sarcomas from benign leiomyomas. In the majority of cases, the clinical differential diagnosis of the tumors is difficult before a histopathologic evaluation of surgical specimens is performed. Magnetic resonance imaging (MRI) has proven to be one of the most useful imaging modalities in making the preoperative differentiation between these tumors, but even with MRI it is difficult to distinguish between uterine sarcomas and uterine leiomyomas with degeneration.6, 7
The most reliable preoperative diagnostic method is a biopsy of the tumor. However, routine endometrial biopsy will not provide a correct diagnosis unless the tumor has reached the surface of the endometrial cavity, with a low sensitivity of approximately 30%.8, 9 Therefore, direct biopsy of tumors, such as a needle biopsy, is indicated. Although needle biopsies often are performed for solid tumors in other organs such as the prostate,10 breast,11 and liver,12 to our knowledge, the use of needle biopsies for distinguishing uterine leiomyosarcoma from leiomyoma has not gained widespread acceptance, most likely due to limited experience with its use and uncertainty with regard to the reliability and risk of the biopsy.13
Since May 1994, we have performed transcervical needle biopsies for uterine myoma-like tumors.14 Factors related to the usefulness of this biopsy, such as sampling certainty, safety, and simplicity, have improved with the development of medical imaging technologies and biopsy devices.14 However, to our knowledge, accuracy, the most significant factor in diagnostic tests, remains unknown.
The purpose of the current study was to evaluate the accuracy of transcervical needle biopsy in the differential diagnosis between uterine sarcoma and leiomyoma.
MATERIALS AND METHODS
Between May 1994 and April 2001, 1253 patients who had been recommended to undergo surgery for presumed uterine leiomyoma by outside physicians were seen at our outpatient clinic at Osaka City University Hospital and were selected as potential study subjects. All patients had at least one of following: uterine corpus exceeding a size consistent with 12 weeks' gestation, myoma-related symptoms including hypermenorrhea or pelvic pain, or infertility with no other explanation apart from uterine myoma.
All 1253 patients underwent MRI at least once during the study period and of these 473 patients were enrolled who met ≥ 1 of the following criteria: myoma with unusual MRI findings (high intensity on T1-weighted and/or T2-weighted images; 452 patients), rapidly enlarging tumor (enlargement of > 3 times original volume within 1 year; 27 patients), high serum lactate dehydrogenase concentrations (46 patients), or a desire to undergo biopsy to rule out malignancy (3 patients). Patients were excluded if they had a tendency to bleed, pelvic infection within the past month, suspected subserosal pedunculate tumor mimicing an ovarian tumor, or pregnancy.
Of the 473 patients, 453 patients (age [mean ± the standard deviation] 44.5 ± 7.3 years) provided written informed consent and underwent transcervical needle biopsies (Fig. 1). The ethics committee of Osaka City University Hospital approved the study protocol.
Transcervical Needle Biopsy of Uterine Myoma-Like Tumor
The procedure used in transcervical needle biopsy of uterine myoma-like tumors was previously described.14 Briefly, biopsies were performed using a Pro-Mag 2.2 biopsy system (Manan Medical Products, Northbrook, IL) with an automatic cutting needle (25-cm long, 16-gauge, 17-mm notch) and a 20-cm long straight stainless-steel guide pipe, with an external greatest dimension of 4 mm and an internal greatest dimension of 3 mm. In the first 159 cases, a manually manipulated biopsy needle (30-cm long, 17-gauge, 17-mm notch [Type M for liver biopsy; Hakko, Tokyo, Japan]) was used instead of the automatic biopsy system.
Before biopsy, the targeted myoma-like nodule was located and its direction and distance from the uterine cavity were confirmed on T2-weighted MRI images. If two or more tumors were observed in the uterus, the largest tumor was biopsied.9 However, when apparent differences were observed on the MRI between or among these tumors, each tumor was biopsied.
Patients were placed in the lithotomy position and the guide pipe was inserted transcervically into the uterine cavity. Next, the operator manipulated the uterine corpus and the guide pipe to bring them into line with the lesion to be biopsied under transabdominal ultrasonic guidance. Next, the biopsy needle was inserted through the guide pipe into the uterine muscle. After confirmation that the tip of the needle was within the lesion, the biopsy gun was fired and a core of tissue was obtained. To prevent sampling error, three or more biopsy cores per patient were obtained.
Although in the more recent cases the majority of patients underwent the biopsy without anesthesia, a paracervical block with 1% lidocaine (6–18 mL) and/or venous injection of both pentazocine (15–30 mg) and diazepam (10 mg) were given if needed.
Degrees of malignancy were established based on the histopathologic criteria for uterine smooth muscle tumors proposed by Bell et al.15 Three parameters (mitotic index, cytologic atypia, and coagulative tumor cell necrosis) were evaluated on a 3-point scale (0, 1, or 2 points) (Table 1) according to our previous report on simulation-biopsy.16
|Mitotic index (mf/10 hpf)||0||1-9||≥ 10||—-||—-|
|Cytologic atypia||Absent to mild||—-||Moderate to severe||—-||—-|
|Coagulative tumor cell necrosis||Absent||Suspected||Present||—-||—-|
|Variant (myxoid, epithelioid)||Absent||Present||—-||—-||—-|
|Endometrial stromal origin||Absent||Present||—-||—-||—-|
Mitotic figures were counted in the most mitotically active areas of the specimen. At least 4 sets of 10 high-power fields were counted using a ×40 objective and ×15 ocular lens. The highest counts were recorded as the mitotic indices of the tumors.
The degrees of atypia were recorded as “insignificant atypia” (absent to mild) or “significant atypia” (moderate to severe). To be considered “significant,” the atypia nearly invariably had to be detectable at low magnification (×10 objective and ×15 ocular) by virtue of the fact that significant atypia usually is manifested by nuclear pleomorphism.
Coagulative tumor cell necrosis
Coagulative tumor cell necrosis was characterized by an abrupt transition from viable cells to necrotic cells without an interposed zone of granulation tissue or hyalinized tissue between viable and necrotic cells (Fig. 2). Coagulative “ghost” cells without abrupt transitions were recorded as “suspected coagulative tumor cell necrosis.”
Specimens with a score of ≥ 4 were considered to be leiomyosarcoma according to the criteria established for the common type (spindle cell type) of uterine smooth muscle tumors.15 When the specimens demonstrated uncommon types such as epithelioid or myxoid variants or endometrial stromal origin, evaluation was performed in the same manner as for the common type and an extra point (+1) was added to the score (Table 1).
Histopathologic scoring was determined based on the most advanced lesions in the specimens by two independent observers who were unaware of the clinical data. Interobserver differences in the histopathologic scoring were resolved by a consensus of the two observers after reevaluation.
The reference (gold) standard for distinguishing between benign and malignant tumors is the histopathologic evaluation of surgical specimens. Histopathologic definitions of uterine sarcoma were performed independently, according to the criteria proposed by Bell et al.,15 by two observers who were unaware of the clinical data. Interobserver differences were resolved by consensus after reevaluation.
We applied a second reference standard to patients who did not undergo surgery; uterine sarcoma status was determined to be negative when no outcome event suggested malignancy for ≥ 2 years after the biopsy was performed.
A conventional receiver operating characteristic (ROC) curve was used to determine the cutoff points that yielded the highest combined sensitivity and specificity with respect to distinguishing uterine leiomyosarcoma from uterine leiomyoma. The 95% confidence intervals (95% CI) for sensitivity, specificity, and positive and negative predictive values were calculated using standard z tests.
From a total of 480 biopsies taken from 453 patients, we obtained 462 satisfactory biopsy specimens (96.3%) from 435 patients. Of these 435 patients, 141 underwent a hysterectomy or myomectomy. Sixteen patients underwent surgery because of positive or suspicious biopsy results despite their symptoms, and 3 patients underwent surgery because of positive results from endometrial curettage (endometrial carcinoma). The remaining 122 patients (benign diagnoses) underwent surgery because of myoma-related symptoms, infertility, or the patient's desire to undergo the surgery (some patients were overlapped). Histopathologic evaluation of the surgical specimens from these 141 patients revealed 5 leiomyosarcomas and 1 endometrial stromal sarcoma.
The remaining 294 patients were managed conservatively and followed every 1–6 months according to their symptoms. Tumor and uterine sizes were measured by ultrasonography at every visit. If the tumor volume increased by > 3 times within 1 year, MRI and chest X-rays were taken and another biopsy was performed (27 patients). Among the 294 patients who were managed conservatively, 1 patient died of leiomyosarcoma and the remaining 293 patients were found to have no evidence of uterine malignancy during follow-up. Of a total of 312 biopsies from the 294 patients, 90 were excluded from the statistical analysis because patient follow-up was < 2 years after biopsy. Accordingly, 372 biopsies from 351 patients were analyzed to establish the accuracy of needle biopsy (Fig. 1).
Table 2 demonstrates the relation between the histopathologic scores and the sarcoma status based on the histopathology of the surgical specimens or clinical outcomes after 2 years of follow-up. No sarcomas were found in those patients whose needle biopsy scores were 0. The details of the 18 patients with scores of ≥ 1 point are listed in Table 3. Of the seven sarcomas found, four could not be detected by endometrial biopsy. According to the analysis using a ROC curve, the cutoff score (which combined the highest sensitivity and specificity with respect to distinguishing uterine leiomyosarcoma from leiomyoma) was 2. At the cutoff score of 2, the sensitivity, specificity, and positive and negative predictive values were 100% (95% CI, 59–100%), 98.6% (95% CI, 97.4–99.8%), 58% (95% CI, 30–86%), and 100.0% (95% CI, 99.0–100.0%), respectively.
|Total histopathologic score||Sarcoma (n = 7)||No sarcoma (n = 365)|
|Patient no.||Age (yrs)||Endometrial Biopsy||Needle biopsy||Surgical specimen||Stage||Survival|
|Total score||Mitotic index||Cytologic atypia||CTCN||Others||Diagnosis||Mitotic index||Cytologic atypia||CTCN|
|7||48||No malig||2||0||Moderate||−||Atypical leiomyoma||1||Moderate||−||NED||61|
|8||38||No malig||2||0||Moderate||−||Atypical leiomyoma||1||Moderate||−||NED||47|
|9||44||No malig||2||0||Moderate||−||Atypical leiomyoma||1||Moderate||−||NED||49|
|15||53||No malig||4||15||Moderate||+||LMS||> 30||Moderate||+||III||DOD||5|
|16||67a||LMS||5||12||Moderate||?||ND (skin meta)||IV||DOD||3|
|17||53||ESS||5||15||Moderate||−||ESS||ESS (high grade)||> 30||Moderate||+||III||DOD||1|
The mean blood loss was 11 g (range, 1–115 g). There were 3 cases (0.6%) in which the blood loss during the procedure was > 50 g. The level of patient pain and discomfort during the needle biopsy was lower than that reported during endometrial curettage in > 80% of patients. No major complications, such as infection, intraperitoneal hemorrhage, or an injury of adjacent structures that required surgery was reported to have occurred.14
The results of the current study demonstrate that transcervical needle biopsy using histopathologic scoring is highly precise, with an especially high negative predictive value, in distinguishing uterine sarcoma from uterine leiomyoma.
A needle biopsy of a solid tumor is one of the most useful diagnostic tests for distinguishing malignant from benign disease. However, to our knowledge this type of biopsy has been performed only infrequently in patients with uterine myoma-like tumors. One possible reason could be explained by the “threshold model,” a rationale for clinical applications of diagnostic tests proposed by Pauker et al.13 The threshold of whether to perform a diagnostic test is determined by four factors: the reliability and potential risks of a diagnostic test, and the benefits and risks of a specific treatment. Hysterectomy is an effective treatment for uterine leiomyoma and the surgical risk is not significantly high. However, a hysterectomy is more invasive than a biopsy.17, 18 According to the model of Pauker et al., a diagnostic test with low reliability and high potential risks will be performed less frequently. However, because these factors can improve with improvements in medical technology, some diagnostic tests that traditionally have been performed less frequently due to their low reliability and high potential risks will be reevaluated.
The reliability of a biopsy depends on the accuracy of both the sampling and the interpretation of the specimens. Sampling accuracy is determined by identification of the appropriate lesion, monitoring of sampling of the appropriate lesion, and the efficiency of the biopsy devices. For uterine myoma-like tumors, the development of MRI, ultrasonography, and the automatic needle biopsy device are technological developments that have improved these three factors.
MRI is one of the most useful clinical modalities for distinguishing between uterine leiomyomas and leiomyosarcomas. Uterine leiomyomas often demonstrate homogeneous low intensity in both T1-weighted and T2-weighted images.19 In contrast, leiomyosarcomas demonstrate heterogeneous high intensities in T2-weighted or both T1-weighted and T2-weighted images, reflecting secondary pathologic changes such as intranodular hemorrhage or tumor necrosis, appearances that are believed to be characteristics of leiomyosarcoma.19 However, because leiomyomas with degeneration can have an appearance that mimics leiomyosarcomas, the differential diagnosis between these tumors is difficult.7, 19
In the current study the positive predictive value of MRI evaluation for distinguishing uterine sarcomas from leiomyomas based on such characteristics was low (approximately 10%). In contrast, the negative predictive value was high (100% in those patients who underwent surgery or who were followed for ≥ 2 years). These results indicate that MRI could be used to screen for uterine sarcomas and to provide useful information regarding the location of suspicious lesions.19 It was the clinical introduction of MRI for such characteristics that instigated the current study in which the use of needle biopsy for uterine myoma-like tumors was reevaluated.
In the current study, three patients did not undergo biopsy. We could not insert the guide pipe in two of these patients because the external os of the uterus was invisible due to vaginal tumor or marked deviation of the uterine cervix as a result of the myoma. In another case, calcifications of the nodule prevented insertion of the needle. In 15 cases, we obtained insufficient materials, such as endometrium or myometrium. The factors that were predictive of the degree of ease and/or difficulty with this biopsy were the location and the size of nodule and the thickness of the abdominal fat. The most difficult case was a small nodule that was located at the subserosal anterior/posterior wall in an anteflexed/retroflexed uterine corpus (the difficult location required a guide pipe at a nearly right angle to the uterine cavity) in an overweight woman because of poor transabdominal ultrasound images.
The other factor that determines the reliability of a biopsy is the accuracy of the interpretation of specimens. The gold standard is the histopathologic evaluation of surgical specimens. However, differential diagnoses between benign and malignant disease in nonepithelial tumors (including uterine smooth muscle tumors) generally are more difficult than those in epithelial tumors. Multisection evaluation of a surgical specimen often is required, especially in borderline cases. The most widely accepted histopathologic criteria for uterine smooth muscle tumors are those proposed by Bell et al. in 1994.15 The degree of malignancy of common-type smooth muscle tumors are determined by three factors: mitotic index, the degree of cytologic atypia, and the presence or absence of coagulative tumor cell necrosis. Evaluation using these factors should be performed in the most advanced areas. However, such areas are not always included in the small specimens obtained by a needle biopsy, resulting in an underevaluation. Therefore, a descriptive diagnosis is not preferred for reporting on this type of biopsy because clinicians may misunderstand the descriptive diagnosis with a potential underevaluation given as the final diagnosis. To avoid such misunderstandings and to apply the results of needle biopsies for practical clinical use, we attempted to use an original scoring system as the clinician reporting system.16 This system was scored based on the degree of malignancy as well as on the biopsy result for gastric and colorectal tumors.
Using this scoring system in combination with MRI screening and multiple sampling, four of seven cases of uterine sarcoma were scored as ≥ 4 points and were diagnosed as “sarcoma” based on needle biopsy alone, with the remaining three cases scoring ≥ 2 points. When the cutoff score was set at 2, all seven cases of sarcoma were diagnosed positively. At that cutoff score, the false-positive rate (1 − specificity) was 42%. However, all 5 false-positive cases were atypical leiomyomas or smooth muscle tumors of low malignant potential, and were assigned to the negative category in the current study because they were regarded as being clinically benign tumors despite their morphologically potential malignancy.15 Their benign status was determined based on the premise that the tumors would be removed surgically. When these tumors are managed conservatively, their behavior is uncertain. The value of this biopsy is not in providing a final histopathologic diagnosis, but in identifying unequivocal cases of benign leiomyoma for which surgery is unnecessary. Therefore, our management program recommended definitive surgery for patients with tumors scored as ≥ 1 point considered to be high risk. The patients who were so regarded but in whom the final diagnosis was not sarcoma comprised only 2.4% of the patients who underwent needle biopsy and 0.9% of the patients who were recommended to undergo surgery by the outside physicians.
Through MRI screening, the relative incidence (prior probability) of leiomyosarcoma versus leiomyoma in the current study was 1.9% (7 of 372 cases). Although the relative incidence was higher than that in previous reports (range, 0.12–0.7%),8 it also may be too low to justify performing a needle biopsy even with its minimal invasiveness. However, although approximately 70% of patients in the current study reported no or only mild myoma-related symptoms, they had been recommended for surgery because of the possibility of malignancy based on the tumor size alone. Of these patients, > 90% could have avoided surgery based on the results of their needle biopsies. Therefore, a biopsy with minimal invasiveness is justifiable compared with surgery. Furthermore, using noninvasive MRI screening, we could reduce the number of patients who required a biopsy. In this study, using MRI screening, the number of patients required to undergo biopsy was reduced by approximately 33%. Using our current MRI screening criteria, the screening positive rate, including borderline cases, was < 20%.
Because the gold standard (i.e., the histopathologic diagnosis of surgical specimens) is not available for those patients who do not undergo surgery, the results in the current study included some “verification bias.”20 To reduce this influence, we applied a second reference standard to patients who did not undergo surgery. Because the median survival rate of patients with residual uterine sarcoma is < 12 months,21 2 years of follow-up appeared long enough to determine the screening negative for patients who received no treatment. Accordingly, we used the data that included the follow-up patients in our subsequent calculations.
In addition to the six leiomyosarcoma cases, one endometrial stromal sarcoma case and eight carcinosarcoma cases were found during our study period. The preoperative differential diagnosis between leiomyoma and endometrial stromal sarcoma is as difficult as that for leiomyosarcoma. Furthermore, the histopathologic differential diagnosis between an endometrial stromal nodule (benign) and low-grade endometrial stromal sarcoma also is difficult using the analysis of small specimens obtained by needle biopsy. However, the incidence of endometrial stromal tumors is much lower than that of uterine smooth muscle tumors (2 surgical cases vs. 377 surgical cases during our study period), and the rate of malignancy for endometrial stromal tumors is much higher than that of uterine smooth muscle tumors (1 of 2 surgical cases vs. 5 of 377 surgical cases). Therefore, when a needle biopsy specimen indicates an endometrial stromal origin, we recommend surgery.
In this study, we excluded carcinosarcoma cases from analysis because those patients usually report abnormal genital bleeding, and the carcinomatous elements of such tumors usually are detected by routine endometrial biopsy before MRI screening.
Transcervical needle biopsy using histopathologic scoring was found to be a reliable diagnostic test for the differential diagnosis between uterine sarcoma and uterine leiomyoma. The needle biopsy may be of clinical use because it also is less invasive and is simple to perform.22 This diagnostic method, combined with MRI screening, may provide a rationale for the conservative management of patients with uterine leiomyoma and may reduce the number of patients undergoing unnecessary surgery.
- 19FujiiS, editor. An atlas of MRI with histopathology on smooth muscle tumors of the uterus. Tokyo: Medical View, 1999.