MRI Evaluation of Uterine Masses for Risk of Leiomyosarcoma: A Consensus Statement

Abstract

Laparoscopic myomectomy, a common gynecologic operation in premenopausal women, has become heavily regulated since 2014 following the dissemination of unsuspected uterine leiomyosarcoma (LMS) throughout the pelvis of a physician treated for symptomatic leiomyoma. Research since that time suggests a higher prevalence than previously suspected of uterine LMS in resected masses presumed to represent leiomyoma, as high as one in 770 women (0.13%). Though rare, the dissemination of an aggressive malignant neoplasm due to noncontained electromechanical morcellation in laparoscopic myomectomy is a devastating outcome. Gynecologic surgeons’ desire for an evidence-based, noninvasive evaluation for LMS is driven by a clear need to avoid such harms while maintaining the availability of minimally invasive surgery for symptomatic leiomyoma. Laparoscopic gynecologists could rely upon the distinction of higher-risk uterine masses preoperatively to plan oncologic surgery (ie, potential hysterectomy) for patients with elevated risk for LMS and, conversely, to safely offer women with no or minimal indicators of elevated risk the fertility-preserving laparoscopic myomectomy. MRI evaluation for LMS may potentially serve this purpose in symptomatic women with leiomyomas. This evidence review and consensus statement defines imaging and disease-related terms to allow more uniform and reliable interpretation and identifies the highest priorities for future research on LMS evaluation.

© RSNA, 2022

Summary

Evidence supports the use of T2-weighted, diffusion-weighted (b value of 1000 sec/mm²), and apparent diffusion coefficient MRI for noninvasive preoperative evaluation for the possibility of uterine leiomyosarcoma and extrauterine disease in symptomatic women.

Essentials

• ■ The prevalence of uterine leiomyosarcoma (LMS) in uterine masses is one in 770 patients (0.13%), but if unsuspected, a likely consequence of undergoing the current standard of laparoscopic myomectomy for leiomyoma is disseminated cancer or delayed care.

• ■ Presurgical risk assessment for uterine LMS is of high priority internationally for gynecologists and gynecologic oncologists due to the widespread use of noncontained electromechanical morcellation during laparoscopic myomectomy.

• ■ MRI using T2-weighted imaging, diffusion-weighted imaging (DWI) with a b value of 1000 sec/mm², and apparent diffusion coefficient (ADC) mapping offers accuracy of 88%–95% for detecting uterine LMS, with sensitivity of 83%–100% and specificity of 88%–100%; the variability is attributable in part to nonuniform imaging technique and interpretation criteria.

• ■ Multidisciplinary preoperative planning that incorporates MRI with emphasis on T2, DWI, and ADC values and assessment of extrauterine disease supports safe, effective management pathways for treatment of leiomyoma and LMS.

Introduction

Uterine leiomyomas are benign uterine masses and are the most common pelvic neoplasm in women, with an estimated cumulative incidence by age 50 years of over 80% for Black women and nearly 70% for White women (1,2). Each year, up to one-third of women undergo treatment for symptomatic leiomyomas, and minimally invasive treatment with laparoscopic myomectomy offers pain relief while limiting the recovery period and preserving the uterus in women who wish to preserve fertility (3). Among these women, a subpopulation has unsuspected malignancy at the time of surgery. The reported prevalence of unsuspected sarcoma at surgery for symptomatic leiomyoma ranges widely, from 0.01% (one in 10 000) to 0.28% (one in 352) depending on the inclusion and exclusion criteria for the largest studies, which include up to 136 195 women in a 2017 Agency for Healthcare Research and Quality study (4–18). Therefore, broad use of noncontained, electromechanical morcellation to deliver small portions of masses directly through port sites during laparoscopic myomectomy has a rare but important consequence; for the subpopulation with unsuspected leiomyosarcoma (LMS), delay of care or dissemination of tumor is often oncologically devastating. In late 2013, a young anesthesiologist from a medical center in Boston underwent a laparoscopic hysterectomy with open (noncontained) morcellation that inadvertently seeded unsuspected LMS throughout her abdomen, leading to iatrogenic metastatic disease and her subsequent death 4 years later. Instead of the noncontained morcellation within the peritoneal cavity, an alternative used in other types of laparoscopic surgery is tissue containment, where specimens are placed directly into a plastic pouch within the peritoneal cavity (19,20). The U.S. Food and Drug Administration issued an immediately-in-effect warning for power morcellators in 2014 (4), which recommended against use of these noncontained power morcellators in women with suspected or known uterine cancer. It also advised that women be informed about the risk and spread of unexpected cancer by power morcellation. As of February 25, 2020, there has been restoration of Food and Drug Administration approval for tissue-contained power morcellation given the development of new equipment for gynecologic surgery. The Food and Drug Administration issuances on changing the morcellation technique have made the laparoscopic approach less accessible for gynecologists, leading to higher rates of open hysterectomy or myomectomy and associated longer postoperative recovery. Therefore, gynecologists and gynecologic oncologists seek a better balance of stratifying risk for LMS while allowing minimally invasive surgery for the vast majority without LMS.

The American College of Obstetricians and Gynecologists Committee Opinion (March 2021) recommends counseling patients before morcellation, using the Agency for Healthcare Research and Quality estimate of one of 770 patients as potential risk (21). Several institutions have begun using MRI to identify masses suspicious for LMS preoperatively, justified by the added benefit MRI offers in terms of accurate leiomyoma localization for preoperative planning (22). However, current counseling lacks a standardized preoperative planning approach. Our multi-institutional and international group of clinical experts aimed to review the evidence, clarify terms, suggest imaging technique and interpretation criteria, and identify the highest priorities for future research on LMS evaluation.

Methods

This consensus statement is based on expert opinions from eight gynecologic radiologists and one gynecologist from three countries, including Canada, France, and the United States, with 7–30 years of experience. Invitation criteria included leadership in the field with academic scholarship and a history of interdisciplinary collaboration. Panel members are part of several national and international radiology societies (including the American College of Radiology, Canadian Association of Radiologists, Canadian Society of Abdominal Radiology, European Society of Urogenital Radiology, International Society for Magnetic Resonance in Medicine, Radiological Society of North America, Society of Abdominal Radiology, Society for Advanced Body Imaging, and French Society of Women’s Imaging). A few of these societies are beginning working groups on uterine sarcomas, but these are nascent and in the organizational stages at the time of this writing. The gynecologist participant is a part of several gynecologic societies, including the American College of Obstetricians and Gynecologists and American Association of Gynecologic Laparoscopists, which both have working groups regarding laparoscopic morcellation.

For crafting the decision tree flowchart analysis of atypical uterine masses, the criteria for agreement were determined a priori to be either moderate (five- to six-member consensus of eight total participants) or strong (seven- to eight-member consensus of eight participants). Two anonymous agreement rounds were held to meet these criteria for all features evaluated. The ninth participant in this panel, a gynecologist, abstained from voting on imaging features.

Articles for inclusion in this review were selected by searching for imaging and sarcoma or leiomyosarcoma in PubMed, Embase, Web of Science, and Scopus databases (with the latter two databases sorted in priority by top 50 cited abstracts). The panel reviewed this initial database and culled it by means of group consensus for final inclusion in the review. All chosen seminal articles in MRI that are included in Table 1 and the MRI discussion were either recognized unanimously by the group as seminal articles or evaluated more than six LMS and met group consensus for inclusion.

Table 1:

Consensus Definitions for Terms with a Strong Association in Support of LMS Diagnosis

Background

The Food and Drug Administration 2014 Safety Communication of a systemic review of 23 studies reported the prevalence of unsuspected uterine sarcoma (all types) during myomectomy or hysterectomy to be one in 352 patients (0.28%) and the prevalence of the most common type of uterine sarcoma, uterine LMS, to be one in 498 patients (0.2%) (4). Only studies in which an LMS was identified were included, however, which may have resulted in selection bias by overestimating LMS prevalence. Black race confers an increased risk of uterine sarcoma and is associated with a twofold higher incidence for uterine LMS (but not for other types of uterine sarcoma) compared with the risk for White women (23,24). Postmenopausal status (defined as age above 50 years) is a risk factor for uterine sarcoma, with incidence in women above 50 years of age more than four times greater than the incidence in younger women (6.4 vs 1.5 per 100 000 women) (23).

In some patients, uterine sarcomas (any type) are diagnosed incidentally after a myomectomy or hysterectomy performed for a symptomatic leiomyomatous uterus. The prevalence of malignancy in a fibroid uterus is higher when there are symptoms. Symptomatic uterine sarcoma manifests with the following: abnormal uterine bleeding (postmenopausal bleeding: 31%–46%; premenopausal abnormal uterine bleeding: 27%–34%), abdominal distention (8%–17%), pelvic pain or pressure (4%–13%), or urinary symptoms (1%–2%) (25). These symptoms overlap with those of benign uterine leiomyomas. Foul-smelling vaginal discharge has been described in two studies from China and Israel, occurring in up to 27.4% of patients (26,27); however, this is not specific to uterine sarcomas and can be seen in multiple conditions. Notably, uterine sarcoma is asymptomatic in 1%–2% of patients (25).

Rapid growth (increase of 6 weeks’ gestational size within 1 year, as determined at physical examination by a gynecologist) (28) of a uterine mass in a premenopausal patient was previously considered to be a predictive factor for uterine sarcoma. However, studies have demonstrated that this may no longer be true. One prospective MRI study showed that benign uterine leiomyomas can also undergo rapid growth, as 37 of 101 leiomyomas increased in volume by more than 30% over a 3-month period, with the most rapid growth occurring in leiomyomas under 5 cm in diameter (29). Another study of 1332 women found a similar incidence of uterine sarcoma between the group with a rapidly growing uterus (one of 371 women [0.27%]) and the group without a rapidly growing uterus (two of 961 women [0.15%]) (28). Nevertheless, a new or rapidly enlarging myometrial mass in postmenopausal women (who are not on postmenopausal estrogen therapy) should be viewed as highly suspicious for a uterine sarcoma unless proven otherwise (30).

Uterine LMS is characterized by aggressive behavior, with a 5-year survival rate ranging from 19% to 68%, which varies widely according to different stages (31). Up to 33% of patients with newly diagnosed uterine LMS present with distant metastatic disease (stage IVB), typically involving the lungs, liver, or upper abdomen (30,32).

Morphologic Information on LMS Relevant to Imaging Evaluation

LMS is the most common subtype of uterine sarcoma, followed by endometrial stromal sarcoma (ESS), undifferentiated endometrial sarcoma, and adenosarcoma (33,34). Of note, carcinosarcoma (also termed mixed malignant mullerian tumor) is no longer considered a sarcoma subtype. In 2009, the International Federation of Gynecology and Obstetrics reclassified uterine carcinosarcoma as a variant of endometrial carcinoma; therefore, moving forward, it should not be included as a sarcoma subtype in sarcoma studies (31). LMS is the subtype of uterine sarcoma that manifests most commonly as a myometrial-based mass and is therefore the subtype most likely to mimic a benign leiomyoma (31). ESS arises from the stroma of the endometrium; 90% of ESS manifest with abnormal uterine bleeding and nearly all involve the endometrium, only rarely manifesting without endometrial thickening with an isolated myometrial mass (35). Deep curettage of the endometrium can be helpful in preoperative diagnosis of ESS; in contradistinction, LMS does not typically involve the endometrium and will therefore not typically be detected at endometrial sampling.

Imaging Evaluation of Uterine LMS Favors MRI

US Evaluation

Leiomyomas typically manifest as a well-circumscribed, homogeneous, hypoechoic mass with variable vascularity and overlap in sonographic appearance with LMS. In prior studies, LMS was associated with sonographic features of large size (>8 cm), irregular borders, areas of cystic change or necrosis, increase in central and peripheral vascularity, higher peak systolic velocity and lower resistive index (36–38) than leiomyomas, and rapid growth. Sensitivity and specificity in a small study (eight LMS, 21 cellular leiomyomas, three smooth muscle tumors of unknown malignant potential [STUMPs], and 225 benign leiomyomas) were as high as 100% and 86%, respectively, but with a positive predictive value of 19% with use of increased central and peripheral vascularity for diagnosis of LMS (39). However, performance of US remains highly uncertain, as another small study of 111 patients (six with LMS, seven with carcinosarcoma, and 98 with leiomyoma) reported no evidence of a difference in US appearance, including in the mean resistive index, between LMS and leiomyomas (38).

CT and PET/CT Evaluation

CT is not typically used for the diagnosis and local staging of uterine LMS.

PET/CT with fluorine 18 (¹⁸F) fluorodeoxyglucose (FDG) has limited use for detection of LMS. While uterine leiomyomas typically show mild ¹⁸F-FDG uptake, they can occasionally show intense uptake, particularly in premenopausal women (40,41), possibly due to hormonal dependency, cellularity, vascularity (microvessel density), tumor cell proliferation (growth factor expression including basic fibroblast growth factor, Ki-67, and so on), glucose transporter 1 and hexokinase expression, endometrial tissue, or inflammatory cells (41,42). LMS usually shows moderate to intense ¹⁸F-FDG uptake, with a mean standardized uptake value of 6.4 ± 4.3 (SD) (range, 2.4–10.2) (43) compared with average uptake of 1.74 ± 0.5 (range, 0.66–3.95) in leiomyomas (41). A 2017 study of 34 patients (15 with LMS and 19 with leiomyoma) showed that a maximum standardized uptake value threshold of 7.5 excluded leiomyoma with 80.8% sensitivity and 100% specificity (44). Thus, non–¹⁸F-FDG avidity makes LMS highly unlikely. Newer tracers, such as fluoro-17β-estradiol and 3′-deoxy-3′-¹⁸F-fluorothymidine, are under investigation, but with very small study numbers to date (45,46).

MRI Evaluation

Among imaging modalities, MRI offers the highest accuracy for characterization of uterine masses before intervention (47) due to improved soft-tissue contrast, larger field of view, diffusion sequences, and multiplanar sequences. For procedural planning, MRI offers better localization of lesion position in the uterus (including intracavitary, intramural, subserosal, cervical, or broad-ligament location) relative to sonography and differentiation from ovarian solid mass lesions and can be used to assess viability and arterial supply of lesions (infarcted leiomyomas, ovarian vs uterine arterial supply) (47–52). Use of MRI for mapping and characterization of symptomatic leiomyomas before intervention is associated with changes in the treatment approach in up to 20% of patients (49,50,52). In the context of preprocedural planning, MRI features have been evaluated for performance in separating LMS from leiomyomas or atypical leiomyomas.

Beginning with the seminal paper by Goto et al in 2002 (53), analysis of MRI features associated with LMS began gaining traction in the gynecologic community. Over the next 2 decades, MRI features that were continually noted in multiple studies as associated with LMS began to emerge, including the following features: intermediate to high signal intensity of the mass at T2-weighted imaging, irregular margins of the uterine mass with the adjacent myometrium, and high signal intensity at high–b value DWI and corresponding low signal intensity on ADC maps (22,53–65). While seemingly obvious as a sign of malignancy, peritoneal implants and enlarged lymph nodes were also noted as predictive MRI features in the diagnosis of LMS. These warrant specific inclusion in the MRI evaluation for LMS, as detection of both peritoneal implants and borderline enlarged nodes can be challenging diagnostically at MRI (54,57,60); both are infrequently observed (seen in approximately one of 20 patients with LMS) (57) and are easier to inadvertently overlook (Fig 1). The details of the articles demonstrating use of these features are briefly summarized and referenced in Tables 1–4.

Figure 1:

(A–E) MRI scans depict features with a strong association with leiomyosarcoma (LMS). All images are in the axial plane, except for the second column of row C, which is a sagittal T2-weighted image. The short arrow in the left panel of D indicates signal intensity in the LMS, and the long arrow shows signal intensity in the endometrium. In the right panel of D, the arrow shows signal intensity in a lymph node. Apparent diffusion coefficients (ADCs) are expressed in square millimeters per second × 10⁻³. b1000 = b value of 1000 sec/mm², Circumf = circumference, Dia = diameter, DWI = diffusion-weighted imaging, Max = maximum, Min = minimum, ROI = region of interest, SI = signal intensity, T2WI = T2-weighted imaging.

Table 4:

Seminal Articles on LMS to Date

Table 2:

Consensus Definitions for Terms with an Indeterminate Association in Support of LMS Diagnosis

Problems with the sensitivity and specificity of various MRI features observed in association with LMS have been attributed not only to the overlapping appearance of subtypes of benign uterine leiomyomas (such as cellular or myxoid subtypes or red/hemorrhagic degeneration of a leiomyoma), but also due to the lack of a consensus and understanding of the terms and features used in each study. For example, review of the seminal literature demonstrates a complete lack of an actual definition of the term “irregular” for the margin of a leiomyoma, with different terms used for “irregular,” including “nodular” or “lobulated” (55,57,60–62,64–66). Thus, we use the precedent set in the American College of Radiology Thyroid Imaging Reporting and Data System lexicon for defining “irregular,” “lobulated,” and “smooth” margins (67) (Fig 2). Additionally, there are varying definitions for the concept of a solid mass, with some groups considering any region of the mass as solid, and others considering solid masses as only those with internal vascularity (ie, enhancement) (54,57,60). This confusion led certain groups to include benign leiomyomas undergoing red degeneration in a malignant category due to restricted diffusion in the hemorrhagic regions of the avascular solid component (which is avoidable if only regions demonstrating enhancement are evaluated). Therefore, we clearly define enhancement in our lexicon and note throughout the flowchart (Fig 3) that only solid enhancing regions of the mass can be evaluated for restricted diffusion.

Figure 2:

Imaging depiction of consensus definitions of margins of uterine masses. All images are axial T2-weighted images (T2WI).

Figure 3:

Flowchart for atypical uterine mass evaluation at MRI. * = Likely benign has a low likelihood of malignancy; however, given the conflicting data in the literature for the consensus on which absolute threshold of the apparent diffusion coefficient (ADC) to use for capturing all leiomyosarcoma, it is still recommended that masses that fall under this category be evaluated with a multidisciplinary team for management, with consideration for open myomectomy. ** = Suspicious for malignancy indicates that many of the mass lesions in this category are malignant; however, cellular leiomyomas and smooth muscle tumors of unknown malignant potential occasionally fall into this category. These masses should be managed with a multidisciplinary team approach, as described in Figure 6. DWI = diffusion-weighted imaging.

Other confusion in terms used in many of these articles exists, some of which were ultimately not included in the lexicon table (Tables 1–3) due to the added complexity. For example, we omitted the term “focal low T2 signal,” as some groups apply this definition to nonenhancing, centrally located hemorrhagic material (where the low T2 signal intensity may be a malignant feature in an LMS or a benign feature in an infarcted LMS) (57), with other groups applying this definition to enhancing solid viable tissue (where the low T2 signal intensity was a reported benign feature) (54).

Table 3:

Consensus Definitions for Terms Associated with a “Very Low Risk” Leiomyosarcoma Diagnosis

Finally, the technique for ADC assessment, in terms of the size of the region of interest (ROI) required to obtain a reading of 0.905 × 10⁻³ mm²/sec or less, requires that the ADC measurement be placed in a region of viable enhancing tissue, since hemorrhage in a nonviable region can have low ADC. Additionally, the ADC ROI must be placed in the mass region with the lowest signal intensity, not a whole-lesion ADC value, and using a small ROI rather than a whole-lesion ROI (Abdel Wahab et al [54] used the smallest ROI of all the articles evaluated, with an ROI size of 24 mm² and ADC of 0.905 × 10⁻³ mm²/sec; Thomassin-Naggara et al [65] used an ADC of 0.978 × 10⁻³ mm²/sec with a mean ROI size of 46.1 mm²; and Tasaki et al [64] used an ADC of 0.91 × 10⁻³ mm²/sec with a minimum ROI size ≥50 mm²). Other studies that used ADC values higher than 0.905 × 10⁻³ mm²/sec as the discriminatory cutoff for diagnosing LMS had large ROIs that likely incorporated higher signal regions, raising the discriminatory ADC value (including Tamai et al [62]: 1.29 × 10⁻³ mm²; Namimoto et al [59]: 1.05 × 10⁻³ mm²/sec; Sato et al [61]: 1.1 × 10⁻³ mm²/sec; and Lin et al [58]: 1.08 × 10⁻³ mm²/sec; all with an ROI size of "as large as possible").

Given the conflicting data in the literature for the consensus on which absolute threshold of ADC to use for capturing all LMS, a mass with an ADC greater than 0.901 × 10⁻³ mm²/sec is placed into the “likely benign” (or “unlikely LMS”) arm of the flowchart (Figs 3, 4), indicating that such a mass has a low likelihood of malignancy. However, this is a nonzero likelihood of malignancy (acknowledging the rare likelihood that a STUMP or LMS may potentially fall into the higher ADC threshold ranges suggested by Tamai et al [62], Namimoto et al [59], Sato et al [61], and Lin et al [58]); therefore, until further data validating the ADC threshold results of Abdel Wahab et al (54) are obtained, a mass in the “likely benign” arm would still meet requirement for multidisciplinary evaluation of the uterine mass, with this group’s recommendation being to perform an open myomectomy (per Fig 2). In our experience, the majority of these masses (with high DWI b value of 1000 sec/mm² and mildly low ADC) will be cellular fibroids.

Figure 4:

Atypical uterine mass evaluation for leiomyosarcoma (LMS) and its mimickers: application of the atypical uterine mass flowchart. This figure demonstrates the real-time application of the Figure 3 flowchart, beginning with assessment for uterine mass enhancement in column 2. As the columns progress to the right, the various features in the top row are applied. If the mass meets criteria for further analysis, an arrow is seen in the box, indicating that the next feature be applied. Once the mass is determined to fit benign criteria, a stop symbol is shown and subsequent sequences to the right are grayed out, indicating that these features should not be evaluated. Masses characterized as benign according to this imaging flowchart analysis do not warrant additional work-up from an imaging perspective. Clinical treatment of such masses can proceed along standard gynecologic management of benign leiomyomas. All images were acquired in the axial plane, except for the T2-weighted image in the second row, which was acquired in the coronal plane. Apparent diffusion coefficient (ADC) is expressed in square millimeters per second × 10⁻³. b1000 = b value of 1000 sec/mm², CE = contrast-enhanced, DWI = diffusion-weighted imaging, endo = endometrium, int. = intermediate, STUMP = smooth muscle tumor of unknown malignant potential, T2WI = T2-weighed imaging.

These various nuanced proposed terms and definitions are outlined in Tables 1–3, with pictorial depiction of various features in Figures 1, 2, 4, and 5.

Figure 5:

Difficult atypical uterine mass lesion evaluation using the atypical uterine mass flowchart. This figure demonstrates the real-time application of the Figure 3 flowchart, beginning with assessment for uterine mass enhancement in column 2. As the columns progress to the right, the various features in the top row are applied. If the mass meets criteria for further analysis, an arrow is seen in the box, indicating that the next feature be applied. Once the mass is determined to fit benign criteria, a stop symbol is shown and subsequent sequences to the right are grayed out, indicating that these features should not be evaluated. Masses characterized as benign according to this imaging flowchart analysis do not warrant additional work-up from an imaging perspective. Clinical treatment of such masses can proceed along standard gynecologic management of benign leiomyomas. Note the important role of the diffusion-weighted imaging (DWI) sequence with a b value of 1000 sec/mm² (b1000) for distinguishing between leiomyosarcoma (LMS) and benign leiomyoma subtypes, a key feature initially described by Abdel Wahab et al (54). All images are in the axial plane. Apparent diffusion coefficient (ADC) is expressed in square millimeters per second × 10⁻³. CE = contrast-enhanced, endo = endometrium, int. = intermediate, LN = lymph node, STUMP = smooth muscle tumor of unknown malignant potential, T2WI = T2 weighed imaging.

Based on the summative evaluation of MRI features at T2-weighted imaging, DWI with a b value of 1000 sec/mm², and ADC mapping as outlined in Tables 1–3, imaging features with at least moderate consensus were included in the atypical uterine mass flowchart (Fig 3). As shown in Table 4, using combined T2-weighted imaging, DWI with a b value of 1000 sec/mm², and ADC mapping, combination MRI analysis offers accuracy of 88%–94.6% for detecting uterine LMS, with sensitivity of 83%–100% and specificity of 88%–100%; the variability is attributable in part to nonuniform imaging techniques and interpretation criteria (54,58,61).

Additional features defined in Tables 1–3 have been suggested as imaging features that allow differentiation between leiomyoma and LMS (such as hemorrhage or central necrosis); however, for our flowchart, we did not incorporate these features, as they did not have enough support in the literature to validate inclusion. Future studies are needed to evaluate the reproducibility between radiologists of the various features proposed in Tables 1–3 and the associated accuracy, sensitivity, and specificity for differentiation between leiomyoma, atypical leiomyoma, and LMS.

Management of Potential LMS after Concern Raised at MRI

Management algorithms for the diagnosis of uterine LMS have been proposed by several groups, based on use of various combinations of MRI feature analysis and laboratory values (22,47,54,68,69). Every institution represented by this consensus group has a different individualized approach to the management of potential LMS as suggested by an abnormal MRI interpretation. At minimum, all these cases should undergo multidisciplinary discussion between the gynecologist and radiologist, weighing clinical factors such as patient age and desire for fertility, presence of abnormal uterine bleeding, and the level of suspicion at MRI, combined with results of any prior imaging studies, correlation with laboratory values, and exclusion of other cancer types before the operation (with use of endometrial biopsy and Papanicolaou test). Ultimately, the patient typically will undergo a surgical procedure, balancing the patient’s desire for fertility preservation with the need for an effective oncologic procedure. In this group’s opinion, in cases where suspicion for LMS is raised at MRI evaluation according to the atypical uterine mass flowchart (Fig 3), there is no evidence for endometrial or cervical cancer at preoperative endometrial biopsy and cervical Papanicolaou test, and the patient wishes to preserve fertility, then the gynecologic oncologists (or gynecologists, depending on expertise and level of comfort) should consider obtaining the patient’s consent for an open myomectomy; they must also acknowledge that the patient may need a definitive oncologic operation (ie, second operation) depending on the final pathologic findings from the open myomectomy (Fig 6). This area requires further research to balance the need for uterine preservation and fertility in women with a benign leiomyoma (which has overlapping suspicious features at MRI with LMS) with the need for optimal treatment of a uterine LMS.

Figure 6:

Consensus flowchart of the generalized management algorithm for the atypical uterine mass visualized with MRI. This management algorithm provides general guidance for multidisciplinary teams using MRI for leiomyosarcoma (LMS) screening for patients in whom any level of suspicion of LMS is raised at MRI. After cervical cancer is excluded by means of Papanicolaou (Pap) smear and endometrial cancer by means of endometrial biopsy, the patient who wishes to preserve her uterus is offered a two-step operative procedure. Two-step means that the patient will potentially have two operations. The first operation is an open myomectomy, with resection of the uterine mass. Because frozen-section analysis is only accurate in up to 88% of patients with uterine masses, a definitive pathologic result is obtained, which can take up to a week. After the final pathologic result is available, if the patient has a malignant neoplasm like LMS, then a definitive hysterectomy and/or lymphadenectomy is performed. If the final pathologic result is benign, then the patient is reassured, her fertility is preserved, and no further operation is required.

Optimal MRI Protocol

Table 5 lists the various sequences used in MRI evaluation of the female pelvis (53–65). These include axial T1-weighted in- and out-of-phase images and axial T1-weighted frequency-selective fat-saturated images, sagittal and axial thin-section (high-resolution) T2-weighted turbo spin-echo images, axial DWI with a high b value (1000 sec/mm² recommended), and pre– and post–contrast-enhanced T1-weighted fat-saturated imaging with recommended generation of subtraction images from the postcontrast data set (57).

Table 5:

Imaging Protocol for Atypical Uterine Mass Characterization at 1.5-T or 3.0-T MRI

Ideally, patients should fast 3–6 hours before MRI for less peristalsis of the bowel (and associated motion blurring and artifact) (70,71).

Data on the benefit of antiperistaltic agents are limited; however, the European Society of Urogenital Radiology recommends their use to decrease artifact generated by bowel motion (71,72). Additionally, patients should be coached on how to breathe gently through their nose to limit movement of the anterior abdominal wall, and scanning technologists should use carefully placed saturation bands over the anterior abdominal wall to decrease wrap and ghosting artifact.

Most MRI studies in the literature have been performed at 1.5 T rather than 3.0 T, with similar documented rates of LMS detection at 1.5 T and 3.0 T for studies that included both scanner types (22,57). Therefore, scanners of either magnet strength can be used interchangeably.

Variants of Leiomyoma That Overlap in MRI Appearance with Uterine LMS

Leiomyoma variants represent approximately one in 100 leiomyomas (73). There are many mimickers of uterine LMS at MRI, of which five are discussed herein: cellular leiomyoma, hemorrhagic/degenerating (red degeneration) leiomyoma, lipoleiomyoma, myxoid leiomyoma, and STUMP (Figs 4, 5). Other rarer subtypes and several variant leiomyoma appearances are not specifically reviewed herein (69,74).

Cellular leiomyomas often will have a benign appearance at T2-weighted imaging, with homogeneous low-to-intermediate signal intensity and well-defined margins; on postcontrast images, there is homogeneous avid contrast enhancement (75). However, cellular leiomyoma can show restricted diffusion (high signal intensity on high–b value images and low signal intensity [sometimes lower than 0.905 ×10⁻³ mm²/sec] on ADC maps), mimicking an LMS (59,64).

Red (hemorrhagic) degeneration of a leiomyoma typically demonstrates complete lack of enhancement after contrast agent administration, which allows the diagnosis to be made with certainty, but occasionally it may only show central necrosis, which mimics that in an LMS. These lesions, however, are typically well-defined and demonstrate internal hemorrhage on T1 fat-suppressed images (diffusely high signal intensity or peripheral high signal intensity). A low–signal intensity rim related to methemoglobin deposits is a typical feature. In contradistinction, an LMS typically has a thickened enhancing rim of soft tissue with intermediate T2 signal intensity and associated restricted diffusion (high signal intensity on high–b value images with associated low signal intensity on ADC maps) (76,77).

Lipoleiomyomas are relatively rare (0.8% of all leiomyomas), with debated histogenesis, but current thought is that most represent fatty metaplasia of a leiomyoma. These are characterized by the presence of internal bulk fat (on frequency-selective fat-saturated images) or intravoxel fat (on chemical-shift opposed-phase T1-weighted images) (69).

Myxoid leiomyoma will have high signal intensity on T2-weighted images that can mimic that of an LMS, although usually the signal intensity of myxoid leiomyoma is closer to that of water and may appear cyst-like. After contrast agent administration, myxoid leiomyomas have delayed gradual homogeneous enhancement and lack restricted diffusion (74,78).

STUMPs may also have dense cellularity that causes restricted diffusion, mimicking an LMS. Descriptions of STUMPs are variable and are primarily based on case series, with general consensus that STUMPs can appear as large, heterogeneous masses at T1- and T2-weighted imaging, with early avid nonuniform enhancement after contrast agent administration (55,58,79).

The overlap in imaging appearance of these mimickers with LMS stems from the overlap in their inherent pathologic characteristics, described in the next section.

Pathologic Limitations in Diagnosis of LMS

The pathologic diagnosis of uterine sarcoma requires evaluation of the mitotic index, cellular atypia, cellularity, and an abrupt transitional zone from viable cells to necrotic cells lacking hyalinized tissue (termed coagulative necrosis) (80). The lesion is more likely to behave in a clinically aggressive manner when there is a higher number of these pathologic findings. These findings have imaging correlates that reflect the increased cellularity (ie, restricted diffusion, high signal intensity at DWI with a b value of 1000 sec/mm², and low ADC) and necrosis (central vascular necrosis) often observed in LMS.

A conventional leiomyoma (also termed spindled, ordinary, classic, or typical) has three features: mitotic index of fewer than five mitotic figures per 10 high-power fields, mild cytologic atypia, and no tumor cell necrosis (81). At gross pathologic examination, conventional leiomyomas have a usual whorled pattern. In contrast, an LMS diagnosis requires two of the following three features: a mitotic index of greater than or equal to 10 mitotic figures per 10 high-power fields, moderate to severe cytologic atypia, and the presence of tumor cell necrosis (80). Tumor cell necrosis has an abrupt transition between viable and dead tumor without an interface of granulation tissue–like response. This can be difficult for the pathologist to distinguish from ischemic or hyaline-type necrosis, as seen in red degeneration of a leiomyoma. In cases where necrosis type is ambiguous, correlation with the degree of cytologic atypia and the mitotic index may help with classification. Dedicated gynecologic pathologists have only moderate agreement (κ = 0.44) in their interpretation of tumor cell necrosis (82). Gross pathologic features of uterine sarcoma include a loss of the typical whorl pattern, homogeneous texture, yellow color, soft consistency, absence of a bulging surface under the capsule, and ill-defined margins causing difficulty in excising from the myometrium (80,83). Frozen-section analysis is not reliable for the exclusion of uterine sarcoma, due in part to inherent sampling bias within the tumor (84). A recent study noted that of nine LMS and 103 leiomyomas, frozen-section diagnosis concurred with the permanent-section diagnosis in 88% of all lesions (99 of 112) and 88.9% of the LMS (eight of nine) (85).

Cellular leiomyoma is a benign histologic variant of leiomyoma and is defined as having increased cellular density relative to background myometrium. There is no specific quantitative minimum degree of cellularity for a tumor to be considered cellular, and application of the term varies among pathologists.

Uterine STUMPs are a histologically and biologically heterogeneous group of smooth muscle tumors that do not meet criteria of a benign leiomyoma or those of an LMS. Smooth muscle tumors with high mitotic counts (>15 per 10 high-power fields) are uncommon, and some have demonstrated aggressive behavior (including recurrence, dissemination, or transformation into LMS). These typically have findings that do not fit neatly into leiomyoma or LMS categories, such as a mass that has more than 15 mitotic figures per 10 high-power fields but is cytologically bland and lacking tumor cell necrosis (86,87). Alternatively, a STUMP could have no cytologic atypia and low mitotic index, but tumor cell necrosis. The management of uterine STUMP is controversial, with a minimum recommendation for surveillance imaging every 6 months for the first 5 years and annual surveillance for the next 5 years (88).

Existing Non–Imaging-based Preoperative Evaluation for LMS

Postmenopausal patients with new symptoms or with a new or growing uterine mass are evaluated as having a potential malignant neoplasm. Biopsy of the endometrium (not a targeted biopsy of the suspicious leiomyoma) can be used to detect some uterine sarcomas if the tumor extends into the endometrial cavity; this is more frequent with ESS than with LMS. This can be performed either by means of an office endometrial biopsy or dilation and curettage, with similar sensitivity for either technique (89). One of the largest studies evaluating endometrial biopsy from Canada studied 68 women with LMS who underwent endometrial biopsy before surgery and found that the sensitivity for diagnosis of malignancy was 52% (sensitivity for LMS in particular was 35%) (89). Serum lactate dehydrogenase and lactate dehydrogenase isozyme 3, as well as cancer antigen 125, have not been compelling in the literature for detection of LMS (90,91). Thus, the impetus mainly falls on surgeons to operate while considering the low risk of LMS by applying careful dissection, controlled morcellation, or open myomectomy resection and requesting frozen sections for unusual-appearing leiomyomas (despite controversy regarding the accuracy of frozen sections for uterine LMS, they are still used by gynecologists confronted intraoperatively with an unusual mass) (84,85). Hysterectomy is not typically advised solely on the basis of avoiding the possibility of LMS.

Areas of Priority for Improved Management

There are several unanswered questions around LMS and proposed screening protocols. One unanswered question relates to the true prevalence of LMS, whether the prevalence is increasing, and whether the average age of diagnosis is decreasing (as seen in other cancers, notably colon cancer). Another important question is whether MRI of LMS is cost-effective. A study by Tong et al (22) in 2019 evaluated the consideration of cost-effectiveness by means of a preliminary cost analysis and demonstrated that the MRI screening protocol increased life expectancy by 0.04 year at a cost of $12 937 per life-year gained, arguing that the modality is cost-effective when coordinated with referring clinicians. Another query is whether advances in US, a cheaper modality, could lead to similar effectiveness as that of MRI for LMS screening. From a surgical perspective, questions arise as to whether there is a role for more aggressive partial uterine resection (extended myomectomy) for indeterminate lesions and whether there is a role for targeted uterine biopsy of the suspicious mass. One problem with extended myomectomy is the difficulty to reconstruct the uterus to preserve fertility once an extensive portion of the uterus is removed. Additionally, there are multiple questions around management of patients with suspected LMS at MRI, where the imaging findings do not fall neatly into a “highly suspicious” category for LMS, nor into a “benign” category. Tong et al proposed applying a seven-point confidence level scale (from <1% suspicion for LMS up to >98% suspicion) for the diagnosis of LMS at MRI, working with gynecologists to manage atypical uterine masses by the degree of suspicion raised.

The first author’s institution used a suspicion level of “less likely” (defined arbitrarily as ≤25% suspicion for LMS) to refer the patient to gynecology-oncology, counseling, and consent for a two-step operative procedure aimed at preserving fertility for benign mimickers (22). However, this management algorithm reflects a single institution’s model and has not been validated by application at other institutions. Similarly, in our flowchart analysis (Fig 3), there was varying confidence among our group regarding using the threshold of 0.905 × 10⁻³ mm²/sec as a definitive means of distinguishing a benign leiomyoma (usually cellular subtype) from a STUMP or LMS. As a cautious approach, we agreed to place masses with a high value at DWI with a b value of 1000 sec/mm² but with a higher ADC than 0.9 × 10⁻³ mm²/sec into a “likely benign” heading, with the recommended option to manage these masses with open myomectomy. (Some centers with greater experience with these masses may choose to manage these differently, such as with laparoscopic myomectomy or close interval follow-up, per patient wishes and their institution’s comfort level.) More data from multiple centers are needed to further refine this ADC threshold (and incorporate additional features for analysis not included in this version of the flowchart) and to further refine a management approach to these masses. Other centers use age thresholds (typically around 45 years, where likelihood of LMS increases) to guide management (22,65). Multi-institutional studies, including prospective assessment of diagnostic performance, would guide further development of optimal test-and-treat management strategies for uterine masses.

Conclusion

For the possibility of uterine leiomyosarcoma (LMS) in symptomatic women, evidence supports the use of MRI for noninvasive preoperative evaluation, particularly T2- and diffusion-weighted imaging appearance, apparent diffusion coefficient values, and assessment of extrauterine disease. This review of the clinical need, proposed lexicon, decision flowchart analysis, and management suggestions may enhance the accurate diagnosis of these LMS preoperatively while preserving fertility in women without LMS. Future studies are needed to further evaluate diagnostic accuracy of MRI for LMS and refine the proposed ideas and evaluations.