Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Uterine Sarcomas: Are There MRI Signs Predictive of Histopathological Diagnosis? A 50-Patient Case Series with Pathological Correlation

Uterine Sarcomas: Are There MRI Signs Predictive of Histopathological Diagnosis? A 50-Patient... Hindawi Sarcoma Volume 2021, Article ID 8880080, 10 pages https://doi.org/10.1155/2021/8880080 Research Article Uterine Sarcomas: Are There MRI Signs Predictive of Histopathological Diagnosis? A 50-Patient Case Series with Pathological Correlation 1 1 1 1 Siegfried He ´lage , Ste ´phanie Vandeventer, Jean-Noe ¨l Buy, Corinne Bordonne ´, 2 3 4 5 Pierre-Alexandre Just, Denis Jacob, Michel Ghossain, Pascal Rousset, and Elisabeth Dion Department of Radiology, Hoˆtel-Dieu de Paris (AP-HP), 1 Place Du Parvis Notre-Dame, Paris 75004, France Department of Pathology, Hoˆpital Cochin (AP-HP), 27 Rue Du Faubourg Saint-Jacques, Paris 75014, France Department of Gynecological Surgery, Clinique Bizet, 21 Rue Georges Bizet, Paris 75016, France Department of Radiology, CHU Hoˆtel-Dieu de France, Beirut, Lebanon Department of Radiology, Centre Hospitalier Lyon Sud, Hospices Civils de Lyon, Pierre Be´nite, France Correspondence should be addressed to Siegfried He´lage; siegfriedhelage@yahoo.fr Received 6 September 2020; Revised 18 April 2021; Accepted 25 May 2021; Published 1 July 2021 Academic Editor: Eugenie S. Kleinerman Copyright © 2021 Siegfried Helage et al. )is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Purpose. To make clear distinction between two radiological types of uterine sarcomas. Methods. 50 preoperative MRI were analyzed retrospectively, blinded to histopathology: 11 endometrial stromal sarcomas (ESS), 19 leiomyosarcomas (LMS), 18 carcinosarcomas/malignant mixed Mullerian tumors (MMMT), and 2 smooth muscle tumors of uncertain malignant potential (STUMP). Results. According to their locations, two radiological types of sarcomas were identified: type 1: intracavitary (ESS, MMMT) and type 2: intramyometrial (LMS, STUMP). In both types, all tumors displayed intermediate T2-weighted signal (p< 0.001) and high diffusion-weighted imaging (DWI) b1000 signal (p< 0.001). Dynamic contrast-enhanced (DCE) MRI showed intratumoral pathologic vessels (98%) and heterogeneity at venous phase (p< 0.001). In the type 1 subgroup, all tumors displayed local spread: invasion of junctional zone on T2-weighted imaging (T2WI), irregular margins on DWI, and disruption of arcuate arteries subendometrial ring on DCE-MRI. In the type 2 subgroup, all tumors displayed irregular margins on T2WI, DWI, and DCE-MRI. Tumor heterogeneity was due to necrosis (p< 0.001). Most commonly the tumor was single (61%). In both types, −3 2 apparent diffusion coefficient (ADC) lesser than or equal to 0.86 ×10 mm /s (sensitivity � 73%, specificity � 92%) was suggestive of malignancy. Conclusion. It may be feasible to get close to histological type of a uterine sarcoma based on our topographic classification into two radiological subgroups, corresponding to two kinds of diagnostic difficulties. Advances in knowledge. MRI signs suggestive of histopathological malignancy are identifiable, considering the triad T2WI/DWI/DCE-MRI, easily for type 1 but −3 2 less easily for type 2; the threshold value for ADC is 0.86 ×10 mm /s. sarcomas and myometrial sarcomas [3–5]. MR features 1. Introduction predictive of histological malignancy have not been much Uterine sarcomas are very rare neoplasms with poor evaluated, especially a combination of several signs. prognosis. )ey account for only 1 to 3% of gynecological Considering myometrial tumors, pathological distinc- tumors and 3 to 7% of uterine tumors. )e 5-year survival tion between benign and malignant types is often difficult to rate in advanced stages is less than 10% [1, 2]. In the lit- such an extent that there are borderline types termed erature, radiological series include very few patients. Besides, “STUMP” for which only tumor progression during follow- authors never make a distinction between endometrial up will allow physicians to conclude [6]. 2 Sarcoma )e aim of the present observational study was to define intrauterine and extrauterine extension (adenopathy, peri- two types of uterine sarcomas based on tumor macroscopic toneal effusion, peritoneal nodules of carcinomatosis). location within the uterus, found to be closely related to An intermediate T2 signal was defined as a signal be- pathological subtypes of uterine sarcomas. )us, we assessed tween pelvic muscle and adipose tissue, and a T2 signal was MR features predictive of histological malignancy for each of considered high if similar to that of bladder urine. these two location types. Lastly, we pointed out the value of A high b1000 signal intensity was defined as a higher DCE-MRI, especially concerning angiograms analysis, to signal than the myometrium. For diffusion analysis, we determine the malignant character of the tumor and his- recorded the ADC values of each tumor by placing a circular tological differential diagnoses. region of interest (ROI) on the ADC map in the area with the lowest ADC, avoiding hemorrhagic, necrotic, or cystic portions within the lesion by referring to T1-weighted, T2- 2. Materials and Methods weighted, and contrast-enhanced images. Several ROI were placed on the maps, and the lowest value of mean ADC was 2.1. Study Population. After institutional ethical board ap- recorded. proval, our study population included retrospectively 50 DCE-MRI evaluated the presence of tumoral vessels at women, aged 20–95 years (mean age 60) who underwent arterial phase and percentage of necrosis at venous phase MRI for various uterine masses before surgery, in 6 medical and later phases. centers in France, from October 2009 to July 2016. We defined two radiological types of uterine sarcomas: All patients underwent surgery (colpohysterectomy with or without bilateral adnexectomy). )e diagnosis of all tu- (i) Type 1: endometrial epicenter or protruding into the mors was proved at pathology. endometrial cavity Pathological diagnosis for all these uterine sarcomas (ii) Type 2: myometrial epicenter consisted of ESS (n � 11), MMMT (n � 18), LMS (n � 19), and STUMP (n � 2). By convention, in patients with multiple leiomyomas, 2.4. Pathological Correlation. Diagnosis of all malignant the uterine tumor was considered to be single if there were lesions was established at pathological examination in each ≤3 leiomyomas ≤20 mm in diameter. center, with macroscopy, microscopy, and immunohistochemistry. 2.2. MRI Protocol. MRI were performed on 1.5 T systems 2.5. Statistical Analysis. With regard to the number of pa- (Achieva , Philips Healthcare, Netherlands; Aera , Siemens ® ® tients, the following nonparametric tests were used to de- Healthcare, Germany; Signa , GE Healthcare, USA) using a termine p values: Fisher’s exact test for qualitative variables phased array pelvic coil. Protocols included sagittal and axial and Mann–Whitney test for quantitative variables. )e spin echo T2WI and axial spin echo T1WI with or/and reference benign mesenchymal tumor population (leio- without Fat-Sat. myomas) considered was that of Lin’s study [4], taken as a Diffusion-weighted imaging (DWI) was acquired in 37 historical control population used for comparison. )e re- patients in the axial plane using single-shot EPI. b values 2 ceiver operating characteristics (ROC) curve and Youden were 0 and 1000 s/mm . ADC maps were calculated in these index, integrating both sensitivity and specificity, were used 37 patients. to identify the best threshold for ADC. DCE-MRI was performed for 9 of 11 ESS, 16 of 18 A p value< 0.05 was considered statistically significant. MMMT, 15 of 19 LMS, and for all STUMP. DCE-MRI used XLSTAT software was used for all statistical analysis. axial 3D GRE sequence. 20 acquisitions of 72 slices each requiring 23 seconds per acquisition were acquired after 3. Results intravenous injection of 0.1 mmol/kg body weight of gadoterate meglumine (Dotarem , Guerbet, France). 3.1. Correlation between Location within the Uterus and Histological Type. In the type 1 subgroup, among the 27 tumors, there were 9 ESS (33%), 18 MMMT (67%), but 2.3. Image Analysis. MRI were retrospectively analyzed by neither LMS nor STUMP. In the type 2 subgroup, among the three independent radiologists (30, 7, and 4 years of expe- 23 tumors, there were 19 LMS (83%), 2 STUMP (9%), 2 ESS rience respectively), blinded to the final histological (9%), and no MMMT. )is topographic feature was sta- diagnosis. tistically significant (p< 0.001) (Table 1). MRI were reviewed regarding the following items: size of the tumor (greatest diameter), isolated character of the tumor (i.e., single character), topography of the uterine 3.2. Lesion Features. Greater diameters of tumor burdens on lesion (myometrium or endometrium), tumor margins MRI ranged from 10 mm to 240 mm (mean 86 mm, median (blurred or sharp, irregular or regular), presence of intra- 70 mm). tumoral necrosis or hemorrhage, T2 and DWI signal in- In both types, all tumors displayed an intermediate or tensity (with ADC mapping) of the solid component, the high T2 signal (50/50, 100%, p< 0.001) and a high b1000 heterogeneous character of the tumor, vascular enhance- signal (37/37, 100%, p< 0.001) (Table 2). Sometimes, DWI ment profile and angiograms (tumoral vessels), and allowed easy detection of metastatic regional lymph nodes Sarcoma 3 Table 1: Tumor macroscopic location within the uterus on MRI. exclusively with LMS and STUMP on pathological exami- nation. Defining these two radiological types of sarcomas, SSE, MMMT LMS, STUMP corresponding clearly to two different histological categories Number of tumors, % (n) 58 (29) 42 (21) of sarcomas, will have a number of implications during the Type 1, % (n) 93 (27/29) 0 (0/21) diagnostic process and prognostic assessment. )e malig- Type 2, % (n) 7 (2/29) 100 (21/21) nant character of type 1 tumors is easy to diagnose, con- sidering the T2WI/DWI/DCE-MRI triad [3–5, 7]. On these three sequences, disruption of the endometrial-myometrial (Figure 1). Angiographic features assessment showed an interface is usually easy to demonstrate, in addition to the increase in the number of arteries around the tumor and the criteria of malignancy related to each sequence we found presence of pathologic vessels within it. In most cases (41/42, statistically significant in our study. )ese criteria were an 98%), the vascularity was greater in the tumor than in the intermediate signal on T2WI, a high DWI b1000 signal surrounding tissues, some tumor regions containing very increasing compared with b0, and visible intratumoral dense wide, regular, and often parallel vessels (“combed” vessels of malignant type on DCE-MRI. )e greater ease in vessels) (Figure 2). In one case, vessels were not visible, determining the proper diagnosis at radiological step for coinciding with a low-grade ESS. A majority of tumors (44/ type 1 tumors is also found at pathological step. In contrast, 49, 90%, p< 0.001) showed heterogeneity at venous phase the malignant character of type 2 tumors is more difficult as (Table 2). analysis of their contours is more subjective on those three In the type 1 subgroup, all tumors (27/27, 100%) dis- MRI sequences. For type 2 tumors, diagnostic issues are also played signs of local spread: invasion of junctional zone on seen in the pathological process owing to, among other T2WI (Figures 1, 3, and 4), irregular margins on DWI things, the lack of stromal reaction; pathologists circumvent (Figure 1), and disruption of the arcuate arteries sub- this pitfall, talking about “smooth muscle tumors of un- endometrial ring (Figures 5 and 6). certain malignant potential” (STUMP) [6]. Frequency of In the type 2 subgroup (Tables 2 and 3), considering LMS intratumoral T1 Fat-Sat WI hypersignal related to hemor- and STUMP compared with leiomyomas, all tumors dis- rhage was not significantly different between malignant and played an intermediate or high signal on T2WI (21/21, 100%, benign mesenchymal tumors. We believe this is because p � 0.005). A majority of tumors (14/21, 67%) displayed coagulation tumor cell necrosis visible in sarcomas at irregular margins on T2WI, on DWI, and on DCE-MRI pathological examination is by its nature distinct from (disruption of the peripheral vascular ring). Tumor het- hyaline necrosis of ischemic type visible in benign mesen- erogeneity was due to intratumoral hemorrhage (8/20, 40%, chymal tumors [6]. p � 0.755) visible as a high signal on T1WI with Fat-Sat, At the pathological level, potential diagnostic errors are necrosis (17/20, 85%, p< 0.001) visible as areas of non- possible in determining the type of sarcoma. Usually, enhancement at venous phase, and to a lesser extent, to pathological diagnostic is dependent on a good macroscopic myxoid changes (2/21, 9%) visible as large areas with a high examination of the resected gross specimen; this analysis T2 signal (Figures 7 and 8). An isolated tumor (i.e., single makes it possible to identify the most informative areas to character) was common (13/21, 61%). sample for microscopic examination. Extensive use of im- munohistochemistry could lead to rush this crucial step of − 2 3.3. ADC Maps. ADC values ranged from 0.41 × 10 S mm /s macroscopic examination, causing sampling biases. But, − 2 − 2 to 1.3 ×10 S mm /s (mean 0.79 ×10 S mm /s, median according to Hart [2], an endometrial stromal sarcoma can − 2 0.77 ×10 S mm /s), with a maximum Youden index contain up to 30% of smooth muscle tissue without however (YI � 0.655) suggestive of malignancy for a threshold lesser being included in leiomyosarcomas. Also, according to − 2 than or equal to 0.86 ×10 S mm /s (Figure 9) Blaustein [6], the portion of the epithelial component shall (sensitivity � 73%, specificity � 92%, PPV � 92%, OA � 81%). be at least 10% to have an MMMTdiagnosed; a poor resected Overall accuracy reaches its maximum value (OA � 83%) for gross specimen sampling during macroscopic examination − 2 a threshold of 1.05 ×10 S mm /s (sensitivity � 85%; specif- could lead to miss this component. icity � 80%, YI � 0.653). In our study, for the two radiological types of sarcomas, intermediate signal intensity (i.e., gray) on T2WI is highly suggestive of malignancy. However, intratumoral hemor- 4. Discussion rhage on T1WI with Fat-Sat is not a significant predictor of We defined two radiological types of uterine sarcomas on malignancy, especially for LMS and STUMP. For type 2 sarcomas, intermediate signal intensity on MRI, based on their location within the uterus, matching with two different histological categories of sarcomas. Type 1 T2WI is sensitive but not specific for malignancy. Indeed, sarcomas are of endometrial epicenter; they are protruding degenerated leiomyomas, especially edematous ones, may into uterine cavity and have the potential to invade the display a high T2WI signal. At the pathological level, on subendometrial myometrium macroscopically. In our study, macroscopic examination, broadly speaking, a leiomyoma is these type 1 sarcomas matched almost exclusively with ESS white and stiff, whereas a leiomyosarcoma is gray and and MMMT on pathological examination. Type 2 sarcomas smooth. So it makes sense that these macroscopic features would lead to different signals on T2WI. In our study, there are of myometrial epicenter and stay confined to the myometrium. In our study, these type 2 sarcomas matched were no sarcomas with a low signal on T2WI equal to that of 4 Sarcoma Table 2: Evaluated MRI features for both type 1 and type 2 uterine sarcomas. Intermediate or high T2 signal, % (n) 100 (50/50) High b1000 signal on DWI, % (n) 100 (37/37) Blurred/irregular margins or invasion of junctional zone on T2WI, % (n) 80 (40/50) Adenopathy on DWI, % (n) 39 (14/36) Visible intratumoral vessels or contrast uptake at arterial phase, % (n) 98 (41/42) Tumor heterogeneity at venous phase, % (n) 90 (44/49) Blurred/irregular margins at venous phase, % (n) 78 (38/49) (a) (b) (c) Figure 1: Pelvic MRI of endometrial stromal sarcoma: (a, b, c) axial DWI b1000 showing a large intrauterine tumor with high signal − 2 intensity and lobular/irregular tumor contours (arrow) (a) and low ADC value (0.58 ×10 S mm /s) (b), and an obturator lymph node metastasis (arrowhead) (a). Note the intermediate T2 signal intensity of these two lesions and the blurred contours of the tumor (empty arrow) (c). pelvic muscles. In other words, a low signal intensity on or peritoneal metastases that could be missed on conven- T2WI has a good negative predictive value for malignancy. tional T2WI sequence. )is property is analogous to a Furthermore, in our study, most type 2 sarcomas were found “scintigraphic effect”. within the uterine wall as single tumors. In a poly- In our study, an ADC value lower or equal to − 2 myomatous uterus, a small leiomyosarcoma lost in the major 0.86 ×10 S mm /s is highly suggestive of malignancy benign component could be missed; this malignant focus (sensitivity � 73%, specificity � 92%, PPV � 92%). Our may be the source of a future tumor recurrence. threshold is lower than that found in Namimoto’s study − 2 − 2 In our study, DWI sequence presents several advantages. (1.05 ×10 S mm /s), Lin’s study (1.08 ×10 S mm /s), and − 2 Features in favor of a malignant nature are a high signal )omassin’s study (1.23 ×10 S mm /s) [3–5]. As an illus- intensity at high b values, increasing from b0- to b1000- tration of the benefit of measuring ADC values, the only type weighted sequence, and lobular or irregular tumor contours. 2 tumor that contours were regular and sharp on T2WI Furthermore, DWI allows detection of regional lymph nodes showed a high signal on DWI b1000 sequence, with an ADC Sarcoma 5 Figure 2: Pelvic MRI of endometrial stromal sarcoma: in DCE-MRI, angiogram depicted tumor regions containing dense wide, regular, and parallel vessels (arrowheads). (a) (b) (c) Figure 3: Pelvic MRI of endometrial stromal sarcoma: (a, b, c) sagittal (a) and axial (b) T2WI showing a large intrauterine tumor with heterogeneous intermediate T2 signal with invasion of junctional zone (arrow); and axial T1 Fat-Sat WI (c) showing high signal intensity intratumoral hemorrhage (arrow). − 2 value of 0.84 ×10 S mm /s; it was a low-grade LMS in In our experience, a myxoid leiomyosarcoma may show a pathology. lack of restricted diffusion. However, an infiltrative tumor Uterine myxoid leiomyosarcoma is an infrequent but border is typically present and should be assessed routinely distinct malignant neoplasm. By convention in pathology, a as a diagnostic criterion for malignancy. tumor is designated as myxoid if the neoplasm contains an Angiography on DCE-MRI displays a lower quality than “abundant” myxoid extracellular matrix occupying at least that of conventional angiography in terms of imaging 50% of the tumor area. Morphologic features typical of contrast and spatial resolution. However, in our study, for the two radiological types of sarcoma, DCE-MRI allowed conventional LMS tend to be subtle in the myxoid coun- terpart particularly when the myxoid matrix is extensive [6]. detection of intratumoral vessels at arterial phase and 6 Sarcoma (a) (b) Figure 4: Pelvic MRI of MMMT: (a, b) sagittal (a) and axial (b) T2WI showing a large intrauterine tumor with mixed heterogeneous signal, cystic component showing high signal intensity (arrow). (a) (b) Figure 5: Pelvic MRI of endometrial stromal sarcoma: (a, b) axial DCE-MRI showing tumoral vessels in the lesion at arterial phase and disruption of arcuate arteries subendometrial ring (arrow) (a) and percentage of necrosis at venous phase (arrow) (b). (a) (b) Figure 6: Pelvic MRI of MMMT: (a, b) sagittal DCE-MRI evaluated the presence of tumoral vessels in the lesion at arterial phase (arrow) (a) and percentage of necrosis at venous phase (arrow) (b). disruption of the vascular circle surrounding the tumor, corresponds to subendometrial arcuate arteries network; for features suggestive of malignancy, along with contrast type 2 sarcomas, it corresponds to peritumoral vascular ring. washout. For type 1 sarcomas, this vascular circle At venous phase and on delayed sequences, heterogeneity Sarcoma 7 Table 3: Univariate analysis of MRI-specific features for type 2 uterine sarcomas. LMS, STUMP p Intermediate or high T2 signal, % (n) 100 (21/21) 0.005 High b1000 signal on DWI, % (n) 100 (15/15) 0.015 High signal area on T1WI with fat suppression, % (n) 40 (8/20) 0.755 Tumor heterogeneity at venous phase, % (n) 85 (17/20) <0.001 Indicates statistically significant different values between leiomyomas and leiomyosarcomas or tumors of uncertain malignant potential with p values less than 0.05. (a) (b) (c) Figure 7: Pelvic MRI of leiomyosarcoma: (a, b, c) sagittal (a) and axial (b) T2 and axial (c) T1 Fat-Sat WI showing a large uterine tumor developed within the myometrium, showing heterogeneous intermediate T2 signal with irregular margins (arrows) and intratumoral hemorrhage in high T1 Fat-Sat WI signal (arrowhead) (c). due to intratumoral necrosis could be assessed, a feature each histological subtype, namely ESS, MMMT, and LMS. typically found in malignant tumors, due to mismatch be- Other series included few LMS, and a majority of ESS in tween cell proliferation and angiogenesis. )omassin’s study [5]. For the first time, our study dis- Interestingly, as previously reported in the literature tinguishes two radiological types of uterine sarcomas, which [8–11], in our study, sarcomas’ angiograms were different match with two different macroscopic aspects in pathology from those of carcinomas. Angiograms depicted tumor and two distinct histological categories. Furthermore, dif- regions containing very dense wide, regular, and often ficulties in diagnosis are different between these two ra- parallel vessels (“combed” vessels) in sarcomas. For carci- diological types. )us, MRI reading should begin with locating tumor epicenter within the uterus (protruding into nomas, angiograms usually depict a network of fine multiple and tortuous intratumoral vessels. the endometrial cavity versus myometrium) to classify in To our knowledge, our series of uterine sarcomas is the type 1 or 2. )en, MRI features of malignancy should be largest in the literature and includes a similar number of sought, the triad T2WI/DWI/DCE-MRI being the most 8 Sarcoma (a) (b) (c) (d) Figure 8: Pelvic MRI of leiomyosarcoma: (a, b, c, d) axial DCE-MRI without injection (a), at arterial phase (b), venous phase (c), and late phase (d). Note the disruption of the peripheral vascular ring (arrow) and heterogeneity due to central necrosis (arrowhead) (c). 1.0 0.8 0.6 0.4 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1-specificity Figure 9: ROC (receiver operating characteristic) curve of ADC value for the risk of being malignant. Youden index reaches its maximum − 2 value (YI � 0.655) for a threshold lesser than or equal to 0.86 ×10 S mm /s (arrow) (sensitivity � 73%; specificity � 92%). efficient way; hence, we here propose a diagnostic algorithm scoring system using the features found to be significant in (Figure 10). our univariate analysis, as has already been done for other Diagnosis of malignancy is easy to assert on MRI for type mesenchymal tumors [12, 13]. 1 tumors. In contrast, radiologists’ conclusions should be Our study has a few limitations. Our study population is circumspect for type 2 tumors, for which difficulties remain the largest in the literature for such rare tumors; however, multivariate analysis was impractical. )e retrospective in distinguishing between benign and malignant processes (i.e., leiomyoma versus leiomyosarcoma). For type 2, in aspect of this multicenter study explains variable scanning particular, several informative MRI sequences are required techniques, sometimes without DWI and/or DCE-MRI. to have many arguments, thereby reducing the risk of di- Pathology procedures might vary between institutions. )e agnostic errors and allowing suitable treatment to be per- criterion we used for malignancy assessment was initial formed. Considering this issue for type 2 tumors, in order to histopathology report conclusions, notwithstanding follow- select patients with very likely malignant subtypes preop- up data. eratively, especially to distinguish atypical leiomyomas from In the uterine smooth muscle tumors group, myomas leiomyosarcomas, it may be feasible to design an MRI proportion is higher than 99.5%, and leiomyosarcomas Sensitivity Sarcoma 9 (i) DWI: hyperintensity (i) T2 intermediate Endometrial increasing from b0 to b1000 DCE-MRI (ii) Tumor contours: epicenter (ii) Lymphadenopathy detection Arterial phase: disruption of disruption of type 1 on b1000 (“scintigraphic arcuate arteries subendometrial ring junctional zone effect”) Location ADC ≤ 0.86 DCE-MRI within the Leiomyosarcoma (i) Arterial phase: uterus (i) DWI b1000: hyperintensity suspicious intratumoral (i) T2 intermediate (ii) Tumor contours: vessels or marked 0.86 < ADC ≤ 1.23 Myometrial irregular/blurred vs (ii) Tumor contours: contrast uptake of the STUMP epicenter regular/sharp irregular/blurred tumor Cellular myoma type 2 (malignant) vs (iii) Lymphadenopathy detection (ii) Venous or late phase: regular/sharp (benign) on b1000 (“scintigraphic tumor contours ADC > 1.23 effect”) irregular/blurred vs Cellular myoma regular/sharp Edematous myoma Figure 10: Diagnostic algorithm to assess malignant MR features and to suggest pathological hypotheses. For type 2 tumors, the radiologist needs ADC values only if the lesion has a high b1000-weighted signal. malignancy will strongly incite the pathologist to fulfill the frequency is less than 0.5%. Given this epidemiological imbalance, constitution of a matched comparison group of widest possible sampling of the resected gross specimen in 21 myomas for our 21 LMS/STUMP would have yield a order to find the small malignant focus lost in the major sample selection bias due to the limited number of tumors. benign and/or fibrous component. )is focus will be the In order to avoid this bias, we decided to consider a series of source of a future tumor recurrence. )is is for producing myomas, which have been previously validated in the lit- the most accurate diagnosis possible. Even if pathology erature, namely Lin’s series [4]. MRI features of these report is reassuring, such MRI features will provide a strong myomas were accurately described by Lin aiming at com- argument to plan a long-term follow-up to check for evi- parison with LMS/STUMP. )e number of myomas in this dence of relapse. reference series is 25, almost identical to the number of LMS/ STUMP in our study. )is similarity allowed an optimum Abbreviations statistical matching. )e principle of using a reference series validated in the literature is acceptable since it is admitted in ADC: Apparent diffusion coefficient meta-analyses and in epidemiological studies, such as the use DCE: Dynamic contrast-enhanced of the Framingham cohort in vascular medicine for example. DWI: Diffusion-weighted imaging ESS: Endometrial stromal sarcoma Fat-Sat: Fat saturation 5. Conclusion LMS: Leiomyosarcoma We have identified several MRI features specific to the MMMT: Carcinosarcoma/malignant mixed Mullerian different histological subtypes of uterine sarcomas, whose tumor base is firstly tumor macroscopic location within the uterus: MRI: Magnetic resonance imaging endometrial or myometrial epicenter. STUMP: Smooth muscle tumors of uncertain malignant In addition to intellectual interest for radiologists with potential the view to propose a pathological diagnostic hypothesis, T1WI: T1-weighted imaging this MRI reading approach would allow suitable patient T2WI: T2-weighted imaging. management to be performed. An endometrial neoplasm (type 1 tumor), which features Data Availability of malignancy are easy to detect, will be treated by wide surgical excision; the first hypothesis would be carcinoma. )e data used to support the findings of this study are In contrast, for myometrial neoplasms (type 2 tumors), available from the corresponding author upon request in to discriminate between leiomyoma and sarcoma is not that order to protect patient privacy. simple, even in pathology, some pathologists talking about “tumors of uncertain malignant potential” (STUMP). Disclosure Hence, for type 2 tumors, a combination of several MRI signs suggestive of malignancy, if present, will all the more Preliminary results of this study have previously been compel physicians to propose surgical treatment instead of presented at European Congress of Radiology (ECR) Annual follow-up. Moreover, in patients with comorbidities that Meeting 2020 as a poster (https://epos.myesr.org/poster/esr/ increase anesthetic risk, presence of such MRI signs will ecr2020/C-04290). provide a valid argument supporting surgery during tumor boards, especially now that prosecution of medical profes- Conflicts of Interest sionals is becoming increasingly common. Concerning histological diagnosis, such MRI features suggestive of )e authors declare no conflicts of interest. 10 Sarcoma histopathological correlates,” Diagnostic and Interventional Acknowledgments Imaging, vol. 97, no. 3, pp. 347–353, 2016. )e authors thank Dr. Catherine Sciot (Paris, France), Pr. Bruno Deval (Paris, France), Dr. Salim Benabadji (Paris, France), and Dr. Eveline Tamby (Paris, France). References [1] T. Van Den Bosch, A. Coosemans, M. Morina, D. Timmerman, and F. Amant, “Screening for uterine tu- mours,” Best Practice & Research Clinical Obstetrics & Gy- naecology, vol. 26, no. 2, pp. 257–266, 2012. [2] M. Yoonessi and W.-R. Hart, “Endometrial stromal sarco- mas,” Cancer, vol. 40, pp. 898–906, 1977. [3] T. Namimoto, Y. Yamashita, K. Awai et al., “Combined use of T2-weighted and diffusion-weighted 3-T MR imaging for differentiating uterine sarcomas from benign leiomyomas,” European Radiology, vol. 19, no. 11, pp. 2756–2764, 2009. [4] G. Lin, L.-Y. Yang, Y.-T. Huang et al., “Comparison of the diagnostic accuracy of contrast-enhanced MRI and diffusion- weighted MRI in the differentiation between uterine leio- myosarcoma/smooth muscle tumor with uncertain malignant potential and benign leiomyoma,” Journal of Magnetic Res- onance Imaging, vol. 43, no. 2, pp. 333–342, 2016. [5] I. )omassin-Naggara, S. Dechoux, C. Bonneau et al., “How to differentiate benign from malignant myometrial tumours using MR imaging,” European Radiology, vol. 23, no. 8, pp. 2306–2314, 2013. [6] C. Zaloudek and M. R. Hendrickson, “Mesenchymal tumors of the uterus,” in Blaustein’s Pathology of the Female Genital Tract, R. J. Kurman, Ed., pp. 561–615, Springer, New York, NY, USA, 5th edition, 2001. [7] A. Sahdev, S. A. Sohaib, I. Jacobs, J. H. Shepherd, D. H. Oram, and R. H. Reznek, “MR imaging of uterine sarcomas,” American Journal of Roentgenology, vol. 177, no. 6, pp. 1307–1311, 2001. [8] A. Goto, S. Takeuchi, K. Sugimura, and T. Maruo, “Usefulness of Gd-DTPA contrast-enhanced dynamic MRI and serum determination of LDH and its isozymes in the differential diagnosis of leiomyosarcoma from degenerated leiomyoma of the uterus,” International Journal of Gynecological Cancer, vol. 12, no. 4, pp. 354–361, 2002. [9] T. Ohguri, T. Aoki, H. Watanabe et al., “MRI findings in- cluding gadolinium-enhanced dynamic studies of malignant, mixed mesodermal tumors of the uterus: differentiation from endometrial carcinomas,” European Radiology, vol. 12, no. 11, pp. 2737–2742, 2002. [10] C. Lagergren, A. Lindbom, and G. Soderberg, “)e blood vessels of osteogenic sarcomas: histologic, angiographic and microradiographic studies,” Acta Radiologica, vol. 55, no. 3, pp. 161–176, 1961. [11] S. G. Silverberg, F. J. Major, J. A. Blessing et al., “Carcino- sarcoma (malignant mixed mesodermal tumor) of the uterus,” International Journal of Gynecological Pathology, vol. 9, no. 1, pp. 1–19, 1990. [12] S. H´elage, M.-P. Revel, M. Alifano, A. Mansuet-Lupo, and D. Vadrot, “A simple computed tomography scoring system to predict histological malignancy of solitary fibrous tumors of the pleura,” British Journal of Medicine and Medical Research, vol. 5, no. 10, pp. 1301–1308, 2015. [13] S. Helage, M. P. Revel, M. L. Chabi et al., “Solitary fibrous tumor of the pleura: can computed tomography features help predict malignancy? A series of 56 patients with http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Sarcoma Hindawi Publishing Corporation

Uterine Sarcomas: Are There MRI Signs Predictive of Histopathological Diagnosis? A 50-Patient Case Series with Pathological Correlation

Loading next page...
 
/lp/hindawi-publishing-corporation/uterine-sarcomas-are-there-mri-signs-predictive-of-histopathological-iCmVqENfE0

References (15)

Publisher
Hindawi Publishing Corporation
Copyright
Copyright © 2021 Siegfried Hélage et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
ISSN
1357-714X
eISSN
1369-1643
DOI
10.1155/2021/8880080
Publisher site
See Article on Publisher Site

Abstract

Hindawi Sarcoma Volume 2021, Article ID 8880080, 10 pages https://doi.org/10.1155/2021/8880080 Research Article Uterine Sarcomas: Are There MRI Signs Predictive of Histopathological Diagnosis? A 50-Patient Case Series with Pathological Correlation 1 1 1 1 Siegfried He ´lage , Ste ´phanie Vandeventer, Jean-Noe ¨l Buy, Corinne Bordonne ´, 2 3 4 5 Pierre-Alexandre Just, Denis Jacob, Michel Ghossain, Pascal Rousset, and Elisabeth Dion Department of Radiology, Hoˆtel-Dieu de Paris (AP-HP), 1 Place Du Parvis Notre-Dame, Paris 75004, France Department of Pathology, Hoˆpital Cochin (AP-HP), 27 Rue Du Faubourg Saint-Jacques, Paris 75014, France Department of Gynecological Surgery, Clinique Bizet, 21 Rue Georges Bizet, Paris 75016, France Department of Radiology, CHU Hoˆtel-Dieu de France, Beirut, Lebanon Department of Radiology, Centre Hospitalier Lyon Sud, Hospices Civils de Lyon, Pierre Be´nite, France Correspondence should be addressed to Siegfried He´lage; siegfriedhelage@yahoo.fr Received 6 September 2020; Revised 18 April 2021; Accepted 25 May 2021; Published 1 July 2021 Academic Editor: Eugenie S. Kleinerman Copyright © 2021 Siegfried Helage et al. )is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Purpose. To make clear distinction between two radiological types of uterine sarcomas. Methods. 50 preoperative MRI were analyzed retrospectively, blinded to histopathology: 11 endometrial stromal sarcomas (ESS), 19 leiomyosarcomas (LMS), 18 carcinosarcomas/malignant mixed Mullerian tumors (MMMT), and 2 smooth muscle tumors of uncertain malignant potential (STUMP). Results. According to their locations, two radiological types of sarcomas were identified: type 1: intracavitary (ESS, MMMT) and type 2: intramyometrial (LMS, STUMP). In both types, all tumors displayed intermediate T2-weighted signal (p< 0.001) and high diffusion-weighted imaging (DWI) b1000 signal (p< 0.001). Dynamic contrast-enhanced (DCE) MRI showed intratumoral pathologic vessels (98%) and heterogeneity at venous phase (p< 0.001). In the type 1 subgroup, all tumors displayed local spread: invasion of junctional zone on T2-weighted imaging (T2WI), irregular margins on DWI, and disruption of arcuate arteries subendometrial ring on DCE-MRI. In the type 2 subgroup, all tumors displayed irregular margins on T2WI, DWI, and DCE-MRI. Tumor heterogeneity was due to necrosis (p< 0.001). Most commonly the tumor was single (61%). In both types, −3 2 apparent diffusion coefficient (ADC) lesser than or equal to 0.86 ×10 mm /s (sensitivity � 73%, specificity � 92%) was suggestive of malignancy. Conclusion. It may be feasible to get close to histological type of a uterine sarcoma based on our topographic classification into two radiological subgroups, corresponding to two kinds of diagnostic difficulties. Advances in knowledge. MRI signs suggestive of histopathological malignancy are identifiable, considering the triad T2WI/DWI/DCE-MRI, easily for type 1 but −3 2 less easily for type 2; the threshold value for ADC is 0.86 ×10 mm /s. sarcomas and myometrial sarcomas [3–5]. MR features 1. Introduction predictive of histological malignancy have not been much Uterine sarcomas are very rare neoplasms with poor evaluated, especially a combination of several signs. prognosis. )ey account for only 1 to 3% of gynecological Considering myometrial tumors, pathological distinc- tumors and 3 to 7% of uterine tumors. )e 5-year survival tion between benign and malignant types is often difficult to rate in advanced stages is less than 10% [1, 2]. In the lit- such an extent that there are borderline types termed erature, radiological series include very few patients. Besides, “STUMP” for which only tumor progression during follow- authors never make a distinction between endometrial up will allow physicians to conclude [6]. 2 Sarcoma )e aim of the present observational study was to define intrauterine and extrauterine extension (adenopathy, peri- two types of uterine sarcomas based on tumor macroscopic toneal effusion, peritoneal nodules of carcinomatosis). location within the uterus, found to be closely related to An intermediate T2 signal was defined as a signal be- pathological subtypes of uterine sarcomas. )us, we assessed tween pelvic muscle and adipose tissue, and a T2 signal was MR features predictive of histological malignancy for each of considered high if similar to that of bladder urine. these two location types. Lastly, we pointed out the value of A high b1000 signal intensity was defined as a higher DCE-MRI, especially concerning angiograms analysis, to signal than the myometrium. For diffusion analysis, we determine the malignant character of the tumor and his- recorded the ADC values of each tumor by placing a circular tological differential diagnoses. region of interest (ROI) on the ADC map in the area with the lowest ADC, avoiding hemorrhagic, necrotic, or cystic portions within the lesion by referring to T1-weighted, T2- 2. Materials and Methods weighted, and contrast-enhanced images. Several ROI were placed on the maps, and the lowest value of mean ADC was 2.1. Study Population. After institutional ethical board ap- recorded. proval, our study population included retrospectively 50 DCE-MRI evaluated the presence of tumoral vessels at women, aged 20–95 years (mean age 60) who underwent arterial phase and percentage of necrosis at venous phase MRI for various uterine masses before surgery, in 6 medical and later phases. centers in France, from October 2009 to July 2016. We defined two radiological types of uterine sarcomas: All patients underwent surgery (colpohysterectomy with or without bilateral adnexectomy). )e diagnosis of all tu- (i) Type 1: endometrial epicenter or protruding into the mors was proved at pathology. endometrial cavity Pathological diagnosis for all these uterine sarcomas (ii) Type 2: myometrial epicenter consisted of ESS (n � 11), MMMT (n � 18), LMS (n � 19), and STUMP (n � 2). By convention, in patients with multiple leiomyomas, 2.4. Pathological Correlation. Diagnosis of all malignant the uterine tumor was considered to be single if there were lesions was established at pathological examination in each ≤3 leiomyomas ≤20 mm in diameter. center, with macroscopy, microscopy, and immunohistochemistry. 2.2. MRI Protocol. MRI were performed on 1.5 T systems 2.5. Statistical Analysis. With regard to the number of pa- (Achieva , Philips Healthcare, Netherlands; Aera , Siemens ® ® tients, the following nonparametric tests were used to de- Healthcare, Germany; Signa , GE Healthcare, USA) using a termine p values: Fisher’s exact test for qualitative variables phased array pelvic coil. Protocols included sagittal and axial and Mann–Whitney test for quantitative variables. )e spin echo T2WI and axial spin echo T1WI with or/and reference benign mesenchymal tumor population (leio- without Fat-Sat. myomas) considered was that of Lin’s study [4], taken as a Diffusion-weighted imaging (DWI) was acquired in 37 historical control population used for comparison. )e re- patients in the axial plane using single-shot EPI. b values 2 ceiver operating characteristics (ROC) curve and Youden were 0 and 1000 s/mm . ADC maps were calculated in these index, integrating both sensitivity and specificity, were used 37 patients. to identify the best threshold for ADC. DCE-MRI was performed for 9 of 11 ESS, 16 of 18 A p value< 0.05 was considered statistically significant. MMMT, 15 of 19 LMS, and for all STUMP. DCE-MRI used XLSTAT software was used for all statistical analysis. axial 3D GRE sequence. 20 acquisitions of 72 slices each requiring 23 seconds per acquisition were acquired after 3. Results intravenous injection of 0.1 mmol/kg body weight of gadoterate meglumine (Dotarem , Guerbet, France). 3.1. Correlation between Location within the Uterus and Histological Type. In the type 1 subgroup, among the 27 tumors, there were 9 ESS (33%), 18 MMMT (67%), but 2.3. Image Analysis. MRI were retrospectively analyzed by neither LMS nor STUMP. In the type 2 subgroup, among the three independent radiologists (30, 7, and 4 years of expe- 23 tumors, there were 19 LMS (83%), 2 STUMP (9%), 2 ESS rience respectively), blinded to the final histological (9%), and no MMMT. )is topographic feature was sta- diagnosis. tistically significant (p< 0.001) (Table 1). MRI were reviewed regarding the following items: size of the tumor (greatest diameter), isolated character of the tumor (i.e., single character), topography of the uterine 3.2. Lesion Features. Greater diameters of tumor burdens on lesion (myometrium or endometrium), tumor margins MRI ranged from 10 mm to 240 mm (mean 86 mm, median (blurred or sharp, irregular or regular), presence of intra- 70 mm). tumoral necrosis or hemorrhage, T2 and DWI signal in- In both types, all tumors displayed an intermediate or tensity (with ADC mapping) of the solid component, the high T2 signal (50/50, 100%, p< 0.001) and a high b1000 heterogeneous character of the tumor, vascular enhance- signal (37/37, 100%, p< 0.001) (Table 2). Sometimes, DWI ment profile and angiograms (tumoral vessels), and allowed easy detection of metastatic regional lymph nodes Sarcoma 3 Table 1: Tumor macroscopic location within the uterus on MRI. exclusively with LMS and STUMP on pathological exami- nation. Defining these two radiological types of sarcomas, SSE, MMMT LMS, STUMP corresponding clearly to two different histological categories Number of tumors, % (n) 58 (29) 42 (21) of sarcomas, will have a number of implications during the Type 1, % (n) 93 (27/29) 0 (0/21) diagnostic process and prognostic assessment. )e malig- Type 2, % (n) 7 (2/29) 100 (21/21) nant character of type 1 tumors is easy to diagnose, con- sidering the T2WI/DWI/DCE-MRI triad [3–5, 7]. On these three sequences, disruption of the endometrial-myometrial (Figure 1). Angiographic features assessment showed an interface is usually easy to demonstrate, in addition to the increase in the number of arteries around the tumor and the criteria of malignancy related to each sequence we found presence of pathologic vessels within it. In most cases (41/42, statistically significant in our study. )ese criteria were an 98%), the vascularity was greater in the tumor than in the intermediate signal on T2WI, a high DWI b1000 signal surrounding tissues, some tumor regions containing very increasing compared with b0, and visible intratumoral dense wide, regular, and often parallel vessels (“combed” vessels of malignant type on DCE-MRI. )e greater ease in vessels) (Figure 2). In one case, vessels were not visible, determining the proper diagnosis at radiological step for coinciding with a low-grade ESS. A majority of tumors (44/ type 1 tumors is also found at pathological step. In contrast, 49, 90%, p< 0.001) showed heterogeneity at venous phase the malignant character of type 2 tumors is more difficult as (Table 2). analysis of their contours is more subjective on those three In the type 1 subgroup, all tumors (27/27, 100%) dis- MRI sequences. For type 2 tumors, diagnostic issues are also played signs of local spread: invasion of junctional zone on seen in the pathological process owing to, among other T2WI (Figures 1, 3, and 4), irregular margins on DWI things, the lack of stromal reaction; pathologists circumvent (Figure 1), and disruption of the arcuate arteries sub- this pitfall, talking about “smooth muscle tumors of un- endometrial ring (Figures 5 and 6). certain malignant potential” (STUMP) [6]. Frequency of In the type 2 subgroup (Tables 2 and 3), considering LMS intratumoral T1 Fat-Sat WI hypersignal related to hemor- and STUMP compared with leiomyomas, all tumors dis- rhage was not significantly different between malignant and played an intermediate or high signal on T2WI (21/21, 100%, benign mesenchymal tumors. We believe this is because p � 0.005). A majority of tumors (14/21, 67%) displayed coagulation tumor cell necrosis visible in sarcomas at irregular margins on T2WI, on DWI, and on DCE-MRI pathological examination is by its nature distinct from (disruption of the peripheral vascular ring). Tumor het- hyaline necrosis of ischemic type visible in benign mesen- erogeneity was due to intratumoral hemorrhage (8/20, 40%, chymal tumors [6]. p � 0.755) visible as a high signal on T1WI with Fat-Sat, At the pathological level, potential diagnostic errors are necrosis (17/20, 85%, p< 0.001) visible as areas of non- possible in determining the type of sarcoma. Usually, enhancement at venous phase, and to a lesser extent, to pathological diagnostic is dependent on a good macroscopic myxoid changes (2/21, 9%) visible as large areas with a high examination of the resected gross specimen; this analysis T2 signal (Figures 7 and 8). An isolated tumor (i.e., single makes it possible to identify the most informative areas to character) was common (13/21, 61%). sample for microscopic examination. Extensive use of im- munohistochemistry could lead to rush this crucial step of − 2 3.3. ADC Maps. ADC values ranged from 0.41 × 10 S mm /s macroscopic examination, causing sampling biases. But, − 2 − 2 to 1.3 ×10 S mm /s (mean 0.79 ×10 S mm /s, median according to Hart [2], an endometrial stromal sarcoma can − 2 0.77 ×10 S mm /s), with a maximum Youden index contain up to 30% of smooth muscle tissue without however (YI � 0.655) suggestive of malignancy for a threshold lesser being included in leiomyosarcomas. Also, according to − 2 than or equal to 0.86 ×10 S mm /s (Figure 9) Blaustein [6], the portion of the epithelial component shall (sensitivity � 73%, specificity � 92%, PPV � 92%, OA � 81%). be at least 10% to have an MMMTdiagnosed; a poor resected Overall accuracy reaches its maximum value (OA � 83%) for gross specimen sampling during macroscopic examination − 2 a threshold of 1.05 ×10 S mm /s (sensitivity � 85%; specif- could lead to miss this component. icity � 80%, YI � 0.653). In our study, for the two radiological types of sarcomas, intermediate signal intensity (i.e., gray) on T2WI is highly suggestive of malignancy. However, intratumoral hemor- 4. Discussion rhage on T1WI with Fat-Sat is not a significant predictor of We defined two radiological types of uterine sarcomas on malignancy, especially for LMS and STUMP. For type 2 sarcomas, intermediate signal intensity on MRI, based on their location within the uterus, matching with two different histological categories of sarcomas. Type 1 T2WI is sensitive but not specific for malignancy. Indeed, sarcomas are of endometrial epicenter; they are protruding degenerated leiomyomas, especially edematous ones, may into uterine cavity and have the potential to invade the display a high T2WI signal. At the pathological level, on subendometrial myometrium macroscopically. In our study, macroscopic examination, broadly speaking, a leiomyoma is these type 1 sarcomas matched almost exclusively with ESS white and stiff, whereas a leiomyosarcoma is gray and and MMMT on pathological examination. Type 2 sarcomas smooth. So it makes sense that these macroscopic features would lead to different signals on T2WI. In our study, there are of myometrial epicenter and stay confined to the myometrium. In our study, these type 2 sarcomas matched were no sarcomas with a low signal on T2WI equal to that of 4 Sarcoma Table 2: Evaluated MRI features for both type 1 and type 2 uterine sarcomas. Intermediate or high T2 signal, % (n) 100 (50/50) High b1000 signal on DWI, % (n) 100 (37/37) Blurred/irregular margins or invasion of junctional zone on T2WI, % (n) 80 (40/50) Adenopathy on DWI, % (n) 39 (14/36) Visible intratumoral vessels or contrast uptake at arterial phase, % (n) 98 (41/42) Tumor heterogeneity at venous phase, % (n) 90 (44/49) Blurred/irregular margins at venous phase, % (n) 78 (38/49) (a) (b) (c) Figure 1: Pelvic MRI of endometrial stromal sarcoma: (a, b, c) axial DWI b1000 showing a large intrauterine tumor with high signal − 2 intensity and lobular/irregular tumor contours (arrow) (a) and low ADC value (0.58 ×10 S mm /s) (b), and an obturator lymph node metastasis (arrowhead) (a). Note the intermediate T2 signal intensity of these two lesions and the blurred contours of the tumor (empty arrow) (c). pelvic muscles. In other words, a low signal intensity on or peritoneal metastases that could be missed on conven- T2WI has a good negative predictive value for malignancy. tional T2WI sequence. )is property is analogous to a Furthermore, in our study, most type 2 sarcomas were found “scintigraphic effect”. within the uterine wall as single tumors. In a poly- In our study, an ADC value lower or equal to − 2 myomatous uterus, a small leiomyosarcoma lost in the major 0.86 ×10 S mm /s is highly suggestive of malignancy benign component could be missed; this malignant focus (sensitivity � 73%, specificity � 92%, PPV � 92%). Our may be the source of a future tumor recurrence. threshold is lower than that found in Namimoto’s study − 2 − 2 In our study, DWI sequence presents several advantages. (1.05 ×10 S mm /s), Lin’s study (1.08 ×10 S mm /s), and − 2 Features in favor of a malignant nature are a high signal )omassin’s study (1.23 ×10 S mm /s) [3–5]. As an illus- intensity at high b values, increasing from b0- to b1000- tration of the benefit of measuring ADC values, the only type weighted sequence, and lobular or irregular tumor contours. 2 tumor that contours were regular and sharp on T2WI Furthermore, DWI allows detection of regional lymph nodes showed a high signal on DWI b1000 sequence, with an ADC Sarcoma 5 Figure 2: Pelvic MRI of endometrial stromal sarcoma: in DCE-MRI, angiogram depicted tumor regions containing dense wide, regular, and parallel vessels (arrowheads). (a) (b) (c) Figure 3: Pelvic MRI of endometrial stromal sarcoma: (a, b, c) sagittal (a) and axial (b) T2WI showing a large intrauterine tumor with heterogeneous intermediate T2 signal with invasion of junctional zone (arrow); and axial T1 Fat-Sat WI (c) showing high signal intensity intratumoral hemorrhage (arrow). − 2 value of 0.84 ×10 S mm /s; it was a low-grade LMS in In our experience, a myxoid leiomyosarcoma may show a pathology. lack of restricted diffusion. However, an infiltrative tumor Uterine myxoid leiomyosarcoma is an infrequent but border is typically present and should be assessed routinely distinct malignant neoplasm. By convention in pathology, a as a diagnostic criterion for malignancy. tumor is designated as myxoid if the neoplasm contains an Angiography on DCE-MRI displays a lower quality than “abundant” myxoid extracellular matrix occupying at least that of conventional angiography in terms of imaging 50% of the tumor area. Morphologic features typical of contrast and spatial resolution. However, in our study, for the two radiological types of sarcoma, DCE-MRI allowed conventional LMS tend to be subtle in the myxoid coun- terpart particularly when the myxoid matrix is extensive [6]. detection of intratumoral vessels at arterial phase and 6 Sarcoma (a) (b) Figure 4: Pelvic MRI of MMMT: (a, b) sagittal (a) and axial (b) T2WI showing a large intrauterine tumor with mixed heterogeneous signal, cystic component showing high signal intensity (arrow). (a) (b) Figure 5: Pelvic MRI of endometrial stromal sarcoma: (a, b) axial DCE-MRI showing tumoral vessels in the lesion at arterial phase and disruption of arcuate arteries subendometrial ring (arrow) (a) and percentage of necrosis at venous phase (arrow) (b). (a) (b) Figure 6: Pelvic MRI of MMMT: (a, b) sagittal DCE-MRI evaluated the presence of tumoral vessels in the lesion at arterial phase (arrow) (a) and percentage of necrosis at venous phase (arrow) (b). disruption of the vascular circle surrounding the tumor, corresponds to subendometrial arcuate arteries network; for features suggestive of malignancy, along with contrast type 2 sarcomas, it corresponds to peritumoral vascular ring. washout. For type 1 sarcomas, this vascular circle At venous phase and on delayed sequences, heterogeneity Sarcoma 7 Table 3: Univariate analysis of MRI-specific features for type 2 uterine sarcomas. LMS, STUMP p Intermediate or high T2 signal, % (n) 100 (21/21) 0.005 High b1000 signal on DWI, % (n) 100 (15/15) 0.015 High signal area on T1WI with fat suppression, % (n) 40 (8/20) 0.755 Tumor heterogeneity at venous phase, % (n) 85 (17/20) <0.001 Indicates statistically significant different values between leiomyomas and leiomyosarcomas or tumors of uncertain malignant potential with p values less than 0.05. (a) (b) (c) Figure 7: Pelvic MRI of leiomyosarcoma: (a, b, c) sagittal (a) and axial (b) T2 and axial (c) T1 Fat-Sat WI showing a large uterine tumor developed within the myometrium, showing heterogeneous intermediate T2 signal with irregular margins (arrows) and intratumoral hemorrhage in high T1 Fat-Sat WI signal (arrowhead) (c). due to intratumoral necrosis could be assessed, a feature each histological subtype, namely ESS, MMMT, and LMS. typically found in malignant tumors, due to mismatch be- Other series included few LMS, and a majority of ESS in tween cell proliferation and angiogenesis. )omassin’s study [5]. For the first time, our study dis- Interestingly, as previously reported in the literature tinguishes two radiological types of uterine sarcomas, which [8–11], in our study, sarcomas’ angiograms were different match with two different macroscopic aspects in pathology from those of carcinomas. Angiograms depicted tumor and two distinct histological categories. Furthermore, dif- regions containing very dense wide, regular, and often ficulties in diagnosis are different between these two ra- parallel vessels (“combed” vessels) in sarcomas. For carci- diological types. )us, MRI reading should begin with locating tumor epicenter within the uterus (protruding into nomas, angiograms usually depict a network of fine multiple and tortuous intratumoral vessels. the endometrial cavity versus myometrium) to classify in To our knowledge, our series of uterine sarcomas is the type 1 or 2. )en, MRI features of malignancy should be largest in the literature and includes a similar number of sought, the triad T2WI/DWI/DCE-MRI being the most 8 Sarcoma (a) (b) (c) (d) Figure 8: Pelvic MRI of leiomyosarcoma: (a, b, c, d) axial DCE-MRI without injection (a), at arterial phase (b), venous phase (c), and late phase (d). Note the disruption of the peripheral vascular ring (arrow) and heterogeneity due to central necrosis (arrowhead) (c). 1.0 0.8 0.6 0.4 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1-specificity Figure 9: ROC (receiver operating characteristic) curve of ADC value for the risk of being malignant. Youden index reaches its maximum − 2 value (YI � 0.655) for a threshold lesser than or equal to 0.86 ×10 S mm /s (arrow) (sensitivity � 73%; specificity � 92%). efficient way; hence, we here propose a diagnostic algorithm scoring system using the features found to be significant in (Figure 10). our univariate analysis, as has already been done for other Diagnosis of malignancy is easy to assert on MRI for type mesenchymal tumors [12, 13]. 1 tumors. In contrast, radiologists’ conclusions should be Our study has a few limitations. Our study population is circumspect for type 2 tumors, for which difficulties remain the largest in the literature for such rare tumors; however, multivariate analysis was impractical. )e retrospective in distinguishing between benign and malignant processes (i.e., leiomyoma versus leiomyosarcoma). For type 2, in aspect of this multicenter study explains variable scanning particular, several informative MRI sequences are required techniques, sometimes without DWI and/or DCE-MRI. to have many arguments, thereby reducing the risk of di- Pathology procedures might vary between institutions. )e agnostic errors and allowing suitable treatment to be per- criterion we used for malignancy assessment was initial formed. Considering this issue for type 2 tumors, in order to histopathology report conclusions, notwithstanding follow- select patients with very likely malignant subtypes preop- up data. eratively, especially to distinguish atypical leiomyomas from In the uterine smooth muscle tumors group, myomas leiomyosarcomas, it may be feasible to design an MRI proportion is higher than 99.5%, and leiomyosarcomas Sensitivity Sarcoma 9 (i) DWI: hyperintensity (i) T2 intermediate Endometrial increasing from b0 to b1000 DCE-MRI (ii) Tumor contours: epicenter (ii) Lymphadenopathy detection Arterial phase: disruption of disruption of type 1 on b1000 (“scintigraphic arcuate arteries subendometrial ring junctional zone effect”) Location ADC ≤ 0.86 DCE-MRI within the Leiomyosarcoma (i) Arterial phase: uterus (i) DWI b1000: hyperintensity suspicious intratumoral (i) T2 intermediate (ii) Tumor contours: vessels or marked 0.86 < ADC ≤ 1.23 Myometrial irregular/blurred vs (ii) Tumor contours: contrast uptake of the STUMP epicenter regular/sharp irregular/blurred tumor Cellular myoma type 2 (malignant) vs (iii) Lymphadenopathy detection (ii) Venous or late phase: regular/sharp (benign) on b1000 (“scintigraphic tumor contours ADC > 1.23 effect”) irregular/blurred vs Cellular myoma regular/sharp Edematous myoma Figure 10: Diagnostic algorithm to assess malignant MR features and to suggest pathological hypotheses. For type 2 tumors, the radiologist needs ADC values only if the lesion has a high b1000-weighted signal. malignancy will strongly incite the pathologist to fulfill the frequency is less than 0.5%. Given this epidemiological imbalance, constitution of a matched comparison group of widest possible sampling of the resected gross specimen in 21 myomas for our 21 LMS/STUMP would have yield a order to find the small malignant focus lost in the major sample selection bias due to the limited number of tumors. benign and/or fibrous component. )is focus will be the In order to avoid this bias, we decided to consider a series of source of a future tumor recurrence. )is is for producing myomas, which have been previously validated in the lit- the most accurate diagnosis possible. Even if pathology erature, namely Lin’s series [4]. MRI features of these report is reassuring, such MRI features will provide a strong myomas were accurately described by Lin aiming at com- argument to plan a long-term follow-up to check for evi- parison with LMS/STUMP. )e number of myomas in this dence of relapse. reference series is 25, almost identical to the number of LMS/ STUMP in our study. )is similarity allowed an optimum Abbreviations statistical matching. )e principle of using a reference series validated in the literature is acceptable since it is admitted in ADC: Apparent diffusion coefficient meta-analyses and in epidemiological studies, such as the use DCE: Dynamic contrast-enhanced of the Framingham cohort in vascular medicine for example. DWI: Diffusion-weighted imaging ESS: Endometrial stromal sarcoma Fat-Sat: Fat saturation 5. Conclusion LMS: Leiomyosarcoma We have identified several MRI features specific to the MMMT: Carcinosarcoma/malignant mixed Mullerian different histological subtypes of uterine sarcomas, whose tumor base is firstly tumor macroscopic location within the uterus: MRI: Magnetic resonance imaging endometrial or myometrial epicenter. STUMP: Smooth muscle tumors of uncertain malignant In addition to intellectual interest for radiologists with potential the view to propose a pathological diagnostic hypothesis, T1WI: T1-weighted imaging this MRI reading approach would allow suitable patient T2WI: T2-weighted imaging. management to be performed. An endometrial neoplasm (type 1 tumor), which features Data Availability of malignancy are easy to detect, will be treated by wide surgical excision; the first hypothesis would be carcinoma. )e data used to support the findings of this study are In contrast, for myometrial neoplasms (type 2 tumors), available from the corresponding author upon request in to discriminate between leiomyoma and sarcoma is not that order to protect patient privacy. simple, even in pathology, some pathologists talking about “tumors of uncertain malignant potential” (STUMP). Disclosure Hence, for type 2 tumors, a combination of several MRI signs suggestive of malignancy, if present, will all the more Preliminary results of this study have previously been compel physicians to propose surgical treatment instead of presented at European Congress of Radiology (ECR) Annual follow-up. Moreover, in patients with comorbidities that Meeting 2020 as a poster (https://epos.myesr.org/poster/esr/ increase anesthetic risk, presence of such MRI signs will ecr2020/C-04290). provide a valid argument supporting surgery during tumor boards, especially now that prosecution of medical profes- Conflicts of Interest sionals is becoming increasingly common. Concerning histological diagnosis, such MRI features suggestive of )e authors declare no conflicts of interest. 10 Sarcoma histopathological correlates,” Diagnostic and Interventional Acknowledgments Imaging, vol. 97, no. 3, pp. 347–353, 2016. )e authors thank Dr. Catherine Sciot (Paris, France), Pr. Bruno Deval (Paris, France), Dr. Salim Benabadji (Paris, France), and Dr. Eveline Tamby (Paris, France). References [1] T. Van Den Bosch, A. Coosemans, M. Morina, D. Timmerman, and F. Amant, “Screening for uterine tu- mours,” Best Practice & Research Clinical Obstetrics & Gy- naecology, vol. 26, no. 2, pp. 257–266, 2012. [2] M. Yoonessi and W.-R. Hart, “Endometrial stromal sarco- mas,” Cancer, vol. 40, pp. 898–906, 1977. [3] T. Namimoto, Y. Yamashita, K. Awai et al., “Combined use of T2-weighted and diffusion-weighted 3-T MR imaging for differentiating uterine sarcomas from benign leiomyomas,” European Radiology, vol. 19, no. 11, pp. 2756–2764, 2009. [4] G. Lin, L.-Y. Yang, Y.-T. Huang et al., “Comparison of the diagnostic accuracy of contrast-enhanced MRI and diffusion- weighted MRI in the differentiation between uterine leio- myosarcoma/smooth muscle tumor with uncertain malignant potential and benign leiomyoma,” Journal of Magnetic Res- onance Imaging, vol. 43, no. 2, pp. 333–342, 2016. [5] I. )omassin-Naggara, S. Dechoux, C. Bonneau et al., “How to differentiate benign from malignant myometrial tumours using MR imaging,” European Radiology, vol. 23, no. 8, pp. 2306–2314, 2013. [6] C. Zaloudek and M. R. Hendrickson, “Mesenchymal tumors of the uterus,” in Blaustein’s Pathology of the Female Genital Tract, R. J. Kurman, Ed., pp. 561–615, Springer, New York, NY, USA, 5th edition, 2001. [7] A. Sahdev, S. A. Sohaib, I. Jacobs, J. H. Shepherd, D. H. Oram, and R. H. Reznek, “MR imaging of uterine sarcomas,” American Journal of Roentgenology, vol. 177, no. 6, pp. 1307–1311, 2001. [8] A. Goto, S. Takeuchi, K. Sugimura, and T. Maruo, “Usefulness of Gd-DTPA contrast-enhanced dynamic MRI and serum determination of LDH and its isozymes in the differential diagnosis of leiomyosarcoma from degenerated leiomyoma of the uterus,” International Journal of Gynecological Cancer, vol. 12, no. 4, pp. 354–361, 2002. [9] T. Ohguri, T. Aoki, H. Watanabe et al., “MRI findings in- cluding gadolinium-enhanced dynamic studies of malignant, mixed mesodermal tumors of the uterus: differentiation from endometrial carcinomas,” European Radiology, vol. 12, no. 11, pp. 2737–2742, 2002. [10] C. Lagergren, A. Lindbom, and G. Soderberg, “)e blood vessels of osteogenic sarcomas: histologic, angiographic and microradiographic studies,” Acta Radiologica, vol. 55, no. 3, pp. 161–176, 1961. [11] S. G. Silverberg, F. J. Major, J. A. Blessing et al., “Carcino- sarcoma (malignant mixed mesodermal tumor) of the uterus,” International Journal of Gynecological Pathology, vol. 9, no. 1, pp. 1–19, 1990. [12] S. H´elage, M.-P. Revel, M. Alifano, A. Mansuet-Lupo, and D. Vadrot, “A simple computed tomography scoring system to predict histological malignancy of solitary fibrous tumors of the pleura,” British Journal of Medicine and Medical Research, vol. 5, no. 10, pp. 1301–1308, 2015. [13] S. Helage, M. P. Revel, M. L. Chabi et al., “Solitary fibrous tumor of the pleura: can computed tomography features help predict malignancy? A series of 56 patients with

Journal

SarcomaHindawi Publishing Corporation

Published: Jul 1, 2021

There are no references for this article.