Carcinosarcoma of the uterus: MRI findings including diffusion-weighted imaging and MR spectroscopy

Abstract

Background

Recently carcinosarcoma has become regarded as a subset of endometrial carcinoma. Because the clinical course of carcinosarcoma is aggressive with poor prognosis, it should be differentiated from endometrial carcinomas for the appropriate surgical management and adjuvant therapy.

Purpose

To clarify the magnetic resonance imaging (MRI) characteristics of uterine carcinosarcoma including diffusion-weighted imaging (DWI) with apparent diffusion coefficient (ADC) measurement and MR spectroscopy (MRS) with quantitative metabolite evaluation.

Material and Methods

MRI findings of 12 pathologically diagnosed uterine carcinosarcomas obtained on 3T MRI were retrospectively evaluated. The mean and minimum ADCs, and the lipid and choline concentration levels were compared with those of pathologically diagnosed 38 endometrial carcinomas.

Results

The mean and minimum ADCs in carcinosarcomas and endometrial carcinomas were not significantly different. The mean ADC of carcinosarcomas was significantly higher than that of higher grade (G2 and G3) endometrial carcinomas. The choline concentration in carcinosarcomas was significantly lower than that in endometrial carcinomas. High lipid peak was observed in 91% of carcinosarcomas and in 24% of endometrial carcinomas.

Conclusion

Large, exophytic heterogeneous endometrial mass containing strongly enhanced areas, which may exhibit “tumor delivery”, is a suggestive of carcinosarcoma. Relatively high mean ADC and low choline concentration considering its highly malignant nature due to intra-tumoral heterogeneity with necrosis and epithelial cystic components, and the presence of necrosis-associated high lipid peak may be compatible with carcinosarcoma.

Keywords

Genital/reproductive magnetic resonance imaging (MRI)MR spectroscopy MR diffusion/perfusion uterus neoplasms – primary

Introduction

Carcinosarcoma is a very aggressive uterine corpus tumor, and has been considered as the most common subtype of uterine sarcomas (1 –5). Recently carcinosarcoma is regarded as a subset of endometrial carcinoma with sarcomatous differentiation, and has biologic behavior similar to that of high-grade endometrial carcinoma (5 –9). Overall 5-year survival rate for carcinosarcoma is less than 35%, whereas that of endometrial carcinoma is up to 83% (5,10,11). Because the clinical course of carcinosarcoma is aggressive with poor prognosis, it should be differentiated from endometrial carcinomas for the appropriate surgical management and adjuvant therapy (4 –6). Various imaging characteristics of carcinosarcoma have been reported based on the morphological appearances (11 –20). The purpose of this study was to clarify the magnetic resonance imaging (MRI) characteristics of carcinosarcoma including diffusion-weighted imaging (DWI) with apparent diffusion coefficient (ADC) measurement and MR spectroscopy (MRS) with quantitative metabolite evaluation.

Material and Methods

Patients

The institutional review board in our hospital approved this retrospective study, and waived the requirement for written informed consent of patients. We cross-referenced the database of the Department of Obstetrics and Gynecology to identify all patients with histologically proved uterine carcinosarcoma who had undergone MR examinations including DWI and MRS between August 2008 and February 2014. In our institute, MRS is a part of the routine MR protocol for detailed preoperative evaluation of gynecological tumors. A total of 12 women with a mean age of 67 years (age range, 53–80 years) were included in the current study. Histological diagnosis of carcinosarcoma was made by surgical resection (10 lesions) or biopsy (2 lesions): six homologous and six heterologous subtypes. The median lesion size, which was the longest diameter measured by MRI was 93 mm (range, 35–162 mm). We also cross-referenced the database between July 2008 and July 2012, and 38 patients with histologically proven uterine endometrial carcinoma (endometrioid adenocarcinoma) who had undergone MR examinations including DWI and MRS were also included in the study as a comparison group. The comparison group had a mean age of 59 years (age range, 31–77 years). All cases of endometrial carcinoma were histologically diagnosed as endometrioid adenocarcinomas including 24 G1, 11 G2, and three G3 tumors.

MRI and MRS measurement

All patients were studied on a 3T superconducting MRI unit (Signa HDx 3T or Discovery MR750, GE Healthcare, Milwaukee, WI, USA) by using body-array torso coils. Axial spin-echo, single-shot echo planar DWI were obtained in all patients with the parallel image-encoding techniques. The parameters of DWI were as follows: repetition time (TR), 6000 ms; echo time (TE), 56.1–64 ms; b factors, 0 and 800 s/mm²; water excitation for fat signal suppression; matrix size, 128 × 192; field of view, 40 × 40 cm; 4 signals acquired; section thickness, 5–8 mm; and section gap, 0–2 mm. Fast spin-echo T2-weighted (T2W) images (TR, 4000–7000 ms; TE, 99.3–100 ms) and spin-echo (TR, 466.7–600 ms; TE, 7.9–9.8 ms) or fast spoiled gradient-recalled echo (TR, 2.9–4 ms; TE, 1.3–1.7 ms) T1-weighted (T1W) images with fat saturation were obtained in all patients. Gadolinium-enhanced spin-echo (TR, 466.7–600 ms; TE, 7.9–9.8 ms) or fast spoiled gradient-recalled echo (TR, 2.9–4 ms; TE, 1.3–1.7 ms) T1W images with fat saturation were obtained in 11 patients with carcinosarcoma.

MRS was performed using point resolved spectroscopic sequence (PRESS) (TR, 2000 ms; TE, 144 ms) with an automated shimming method, and the total measuring time was about 5 min. A single square spectroscopic voxel of interest (VOI) was prescribed within the tumors so as not to contain cystic, necrotic, or hemorrhagic areas as far as possible by referring to all MR images including T2W images, fat-saturated T1W images, and DWI before the administration of contrast medium. The volumes of VOI were 1–8 mL, and outer volume suppression bands were automatically applied at the all six edges of the VOI.

Analysis methods

Two radiologists with 25 and 16 years of experience in body MRI visually evaluated the signal intensity of the solid portions of the tumors on DWI (b = 800 s/mm²), which were classified as high, slightly high, or isointense to low compared with that of normal myometrium. The two reviewers looked at all MR images together, and agreement was reached in consensus after careful individual evaluation. The mean and minimum ADC values (×10^–3mm²/s) of the tumors were measured in a circular region of interest (ROI) in one representative region as large as possible within the tumor from the ADC maps on the workstation (AW 4.2). The ROI was placed on solid portion of the tumors so as not to contain necrotic, cystic, or hemorrhagic areas as much as possible by reference to all MR images including T2W images, fat-saturated T1W images before and after the administration of contrast medium, and DWI.

The presence of choline and lipid peaks in the MR spectra was visually evaluated by the two reviewers. The metabolite concentration level was classified into three classes, in comparison with the average noise level (ANL): two-fold higher than ANL (high), higher than ANL but lower than two-fold of ANL (low), and the same as ANL (none). Quantitative evaluation of the choline and lipid concentration was made by using LCModel (version 6.3-1K, Stephen Provencher Inc., http://www.s-provencher.com/pages/lcmodel.shtml) on the workstation. Lesions with the choline concentration with a percentage standard deviation (%SD) >20% were excluded, because %SD ≤20% has been used as a criterion for estimates of acceptable reliability (21). The Mann–Whitney U test was used to compare the choline and lipid concentration among carcinosarcomas and endometrial carcinomas. A value of P < 0.05 was considered statistically significant.

Results

Seven of 12 (58%) carcinosarcoma cases showed broad-based exophytic growth (Fig. 1), two cases (17%) showed exophytic growth with a stalk (Fig. 2) and three cases (25%) showed invasive growth (Fig. 3). In addition, four pedunculated tumors (33%) prolapsed through the external os into the vagina as “tumor delivery” (Fig. 2). On the other hand, “tumor delivery” was observed in only one of 38 endometrial carcinomas (3%). High signal intensity hemorrhagic areas on T1W images with fat saturation were observed in five lesions (42%), whereas high signal intensity cystic components on T2W images were observed in eight lesions (67%). Nine of 11 cases with contrast enhancement (82%) had strongly enhanced areas equal to the myometrium, whereas two cases (18%) showed less contrast enhancement in the whole lesion compared to the myometrium. Unenhanced necrotic areas were revealed in nine lesions (82%) on post-contrast T1W images. Only two of 36 endometrial carcinomas with contrast enhancement (6%) had strongly enhanced areas compared to the myometrium. All 12 lesions showed homogeneous or heterogeneous high signal intensity on DWI. The mean ADCs in 12 carcinosarcomas and 38 endometrial carcinomas were 0.91+/−0.20, and 0.85+/−0.13, respectively (P = 0.20) (Table 1) (Fig. 4). By using the F-test, the standard deviations of the mean ADCs of carcinosarcomas and endometrial carcinomas were statistically significantly different (F-statistic was 2.58 and the associated P value was 0.03). The minimum ADCs in carcinosarcomas and endometrial carcinomas were 0.73+/−0.16, and 0.72+/−0.12, respectively (P = 0.85). The mean ADC of carcinosarcoma (0.91+/−0.20) had no significant difference from that (0.89+/−0.12) of well differentiated (G1) endometrioid adenocarcinomas (P = 0.61), and was significantly higher than that (0.78+/−0.10) of higher grade (G2 and G3) endometrioid adenocarcinomas (P = 0.03) (Fig. 4).

Fig. 1.

A 74-year-old woman with uterine carcinosarcoma. (a) Sagittal T2W image shows a large exophytic uterine endometrial mass with heterogeneous signal intensity (arrow) containing multiple high intensity cystic components and irregular-shaped necrotic areas. (b) The mass (arrow) shows strong contrast enhancement equal to the myometrium on post-contrasted T1W image with fat saturation. Unenhanced cystic components and necrotic areas are well visualized. (c) The mass (arrow) shows heterogeneous high signal intensity on DWI (b = 800 s/mm²). (d) ADC mapping image shows intra-tumoral heterogeneity of the mass (arrow). (e) Single-voxel MR spectroscopy image shows bimodal high choline (Cho) and high lipid (Lip) peaks.

Fig. 2.

A 64-year-old woman with uterine carcinosarcoma. (a) Sagittal T2W image shows an exophytic uterine endometrial mass with a stalk (arrows) prolapsing through the external os into the vagina as “tumor delivery”. The mass shows heterogeneous signal intensity containing multiple high intensity cystic components and irregular-shaped necrotic areas. Intracavitary fluid collection is also observed. (b) The mass (arrows) shows heterogeneous contrast enhancement and has strongly enhanced areas equal to the myometrium on post-contrasted T1W image with fat saturation. Unenhanced cystic components and necrotic areas are well visualized. (c) Single-voxel MR spectroscopy image shows bimodal high choline (Cho) and prominent high lipid (Lip) peaks.

Fig. 3.

A 77-year-old woman with uterine carcinosarcoma. (a) Sagittal T2W image shows a huge mass almost replacing the uterine body with heterogeneous signal intensity (arrow). (b) The mass (arrow) contains marked unenhanced central necrotic area, and peripheral viable areas show strong contrast enhancement on post-contrasted T1W image with fat saturation. (c) Single-voxel MR spectroscopy image shows bimodal high choline (Cho) and high lipid (Lip) peaks.

Table 1.

Mean and minimum ADC values of carcinosarcomas and endometrial carcinomas.

	Carcinosarcoma	Endometrial carcinoma
	Carcinosarcoma	Total	G1	G2 + G3
Mean ADC (P value)	0.91+/−0.20	0.85+/−0.13 (0.20)	0.89+/−0.12 (0.61)	0.78+/−0.10 (0.03)*
Minimum ADC (P value)	0.73+/−0.16	0.72+/−0.12 (0.85)	0.76+/−0.10 (0.51)	0.68+/−0.12 (0.52)

The difference between two groups (carcinosarcomas and G2 + G3 endometrial carcinomas) is statistically significant.

Fig. 4.

Scatter plots of the mean ADC values (×10^–3mm²/s) obtained in carcinosarcomas (n = 12) and endometrial carcinomas (n = 38) including 24 G1 and 14 G2 + G3 endometrioid adenocarcinomas. ADC values of carcinosarcomas and G2 + G3 endometrioid adenocarcinomas are significantly different (P = 0.03).

In evaluating the MRS, one of 12 patients with carcinosarcoma was excluded due to suboptimal spectrum. The choline peak was observed in all 11 carcinosarcomas: two low (18%) and nine high (82%) peaks and in all 38 endometrial carcinomas: six low (16%) and 32 high (84%). The choline concentration in 11 carcinosarcomas was 5.99+/−3.36 mM, and was significantly lower than that in 38 endometrial carcinomas (11.29+/−6.30 mM) (P = 0.004) (Fig. 5). The lipid peak was observed in all 11 carcinosarcomas (100%): one low (9%) and 10 high (91%), whereas in 17 of 38 (44.7%) endometrial carcinomas: eight low (21%) and nine high (24%). The lipid concentration in nine of 11 carcinosarcomas (two lesions were excluded because %SD >20%) was 145.39+/−122.65 mM, which had no significant difference from that (209.44+/−188.42 mM) in 11 of 17 endometrial carcinomas (6 lesions were excluded because %SD >20%) with lipid peak (P = 0.73).

Fig. 5.

Scatter plots of the choline concentration (mM) obtained in carcinosarcomas (n = 11) and endometrial carcinomas (n = 38). The choline concentration in carcinosarcomas is significantly lower than that in endometrial carcinomas (P = 0.004).

Discussion

Carcinosarcoma of the uterus is a biphasic malignant neoplasm consisting of a mixture of carcinomatous epithelial and sarcomatous mesenchymal components (1). Recently carcinosarcoma has become considered as a metaplastic carcinoma due to enough evidences of clinical, morphological, pathological, and molecular reasoning (5,6,8,9). The revised FIGO 2009 classified carcinosarcoma together with endometrial carcinoma, distinguishing from other uterine sarcomas (9). However, carcinosarcoma is more aggressive and has a worse prognosis compared to endometrial carcinoma and should be differentiated for the adequate treatment (4 –6). Carcinosarcomas are rarely correctly diagnosed by means of cervicovaginal cytology, dilatation and curettage, or endometrial biopsy, and are often misdiagnosed as endometrial carcinomas (15,18). In addition, the uterine cavity may not be easily accessible for biopsy in some patients such as cases of cervical stenosis (20). Therefore, it would be helpful if the imaging could alert the clinician to the possible diagnosis of carcinosarcoma prior to surgery (20). Various MRI characteristics of uterine carcinosarcoma have been reported in the English literature (11 –20). Tanaka et al. reported that carcinosarcoma typically appeared as a large exophytic, heterogeneous high signal intensity mass on T2W images, and contained strongly enhanced areas within the mass (16). Bharwani et al. reported that radiologic suspicion of carcinosarcoma should increase in the presence of large heterogeneous infiltrative tumors or when tumoral enhancement equaled or exceeded that of myometrium (19). Our results are compatible with their findings, and “tumor delivery”, the prolapse of pedunculated tumor through the external os into the vagina observed in four patients (33%), may be another MRI characteristics of carcinosarcoma reflecting its macroscopic feature (9,22). Kato et al. reported MR findings of four carcinosarcomas and demonstrated that heterogeneous signal intensity on DWI and ADC mapping images reflect complicated tissue components in carcinosarcomas such as necrosis, hemorrhage, and histological difference (18). In our series, the standard deviation of the mean ADC of carcinosarcomas measured at solid components of the tumors was significantly higher than that of endometrial carcinomas possibly reflecting the tissue heterogeneity of carcinosarcoma. However, carcinosarcoma is a high-grade malignant tumor, the mean ADC of carcinosarcomas showed no significant difference compared to that of low-grade, well differentiated (G1) endometrioid adenocarcinomas, and was significantly higher than that of higher grade (G2 and G3) endometrioid adenocarcinomas. This result may also reflect the tissue heterogeneity of carcinosarcoma containing abundant microscopic necrotic areas and epithelial cystic components, which may increase the ADC (1).

The choline peak on MRS may reflect the metabolic activity of tumor cells, and tends to show higher peaks in malignant tumors than in benign pathologies due to the enhancement of phospholipid basement membrane turnover associated with tumor cell proliferation in malignant tumors (23,24). Takeuchi et al. reported that choline concentration in malignant uterine corpus tumors was significantly higher than that in benign lesions, and concluded that MRS with quantitative evaluation of choline concentration can provide helpful information in distinguishing benign and malignant uterine corpus tumors (25). However, Takeuchi et al. also reported that high-grade malignant uterine tumors may often show massive necrosis and may cause low choline concentration due to the decrease in viable tumor cells, and the presence of high lipid peaks may be useful in distinguishing uterine sarcomas from benign leiomyomas (26). Lipid peaks arising from triglycerides and cholesterol esters in neutral lipid droplets may be observed both in viable and necrotic areas of the tumor, and are considered to be important biomarkers in the diagnosis of malignant tumors (27). In particular, demonstration of high lipid peak may be helpful in evaluating highly heterogeneous malignant tumors containing areas of viable tumor cells, cystic components, necrosis, or hemorrhage such as high-grade sarcomas (26,27). In gynecological tumors, some investigators reported elevated lipid resonances in malignant tumors such as endometrial carcinomas, uterine sarcomas, cervical cancers, and malignant ovarian tumors (26,28 –30). In our series the choline concentration in carcinosarcomas was significantly lower than that in endometrial carcinomas possibly due to the decrease in viable tumor cells reflecting intra-tumoral heterogeneity of carcinosarcoma with necrosis and epithelial cystic components (1). The presence of a high lipid peak is a suggestive finding of high-grade malignancy even in tumors showing low choline concentration due to their high necrotic tendency, and may be helpful for the diagnosis of carcinosarcoma (26).

There were several limitations in our study. We compared MR findings of carcinosarcomas including DWI with ADC measurement and MRS with quantitative metabolite evaluation, with those of endometrioid adenocarcinomas, whereas other high-grade endometrial carcinomas such as clear cell carcinoma, serous adenocarcinoma, and undifferentiated carcinomas were not included in this study. Because the clinical courses of these high-grade endometrial carcinomas are also aggressive with poor prognosis like carcinosarcoma, differentiating carcinosarcoma from these high-grade endometrial carcinomas on imaging is not clinically required. The retrospective nature and the relatively small population were further limitations of our study, and prospective studies with larger populations are needed to support our results.

In conclusion, a large, exophytic heterogeneous mass in the endometrial cavity containing strongly enhanced areas equal to the myometrium, which may prolapse through the external os into the vagina as “tumor delivery”, is a suggestive of carcinosarcoma. The relatively high mean ADC and low choline concentration considering its highly malignant nature may be observed in carcinosarcoma due to intra-tumoral heterogeneity with necrosis and epithelial cystic components, and the presence of high lipid peak may be suggestive of its high-grade malignant nature of carcinosarcoma.

Footnotes

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Wells M, Palacios J, Oliva E, et al. Mixed epithelial and mesenchymal tumours. In: Kurman RJ, Carcangiu ML, Herrington CS, et al. eds. WHO Classification of Tumors of Female Reproductive Organs, 4th ed. Lyon: International Agency for Research on Cancer, 2014:148–152.

Silverberg

Major

Blessing

. Carcinosarcoma (malignant mixed mesodermal tumor) of the uterus. A Gynecologic Oncology Group pathologic study of 203 cases. Int J Gynecol Pathol 1990; 9: 1–19.

Brooks

Zhan

Cote

. Surveillance, epidemiology, and end results analysis of 2677 cases of uterine sarcoma 1989–1999. Gynecol Oncol 2004; 93: 204–208.

Artioli

Wabersich

Ludwig

. Rare uterine cancer: carcinosarcomas. Review from histology to treatment. Crit Rev Oncol Hematol 2015; 94: 98–104.

Cantrell

Blank

Duska

. Uterine carcinosarcoma: A review of the literature. Gynecol Oncol 2015; 137: 581–588.

George

Lillemoe

Twiggs

. Malignant mixed müllerian tumor versus high-grade endometrial carcinoma and aggressivevariants of endometrial carcinoma: a comparative analysis of survival. Int J Gynecol Pathol 1995; 14: 39–44.

McCluggage

. Uterine carcinosarcomas (malignant mixed Mullerian tumors) are metaplastic carcinomas. Int J Gynecol Cancer 2002; 12: 687–690.

Kernochan

Garcia

. Carcinoasarcomas (malignant mixed Müllerian tumor) of the uterus: advances in elucidation of biologic and clinical characteristics. J Natl Compr Canc Netw 2009; 7: 550–557.

Prat

. FIGO staging for uterine sarcomas. Int J Gynaecol Obstet 2009; 104: 177–178.

10.

Ferguson

Tornos

Hummer

. Prognostic features of surgical stage I uterine carcinosarcoma. Am J Surg Pathol 2007; 31: 1653–1661.

11.

Tirumani

Ojili

Shanbhogue

. Current concepts in the imaging of uterine sarcoma. Abdom Imaging 2013; 38: 397–411.

12.

Worthington

Balfe

Lee

. Uterine neoplasms: MR imaging. Radiology 1986; 159: 725–730.

13.

Shapeero

Hricak

. Mixed müllerian sarcoma of the uterus: MR imaging findings. Am J Roentgenol 1989; 153: 317–319.

14.

Sahdev

Sohaib

Jacobs

. MR imaging of uterine sarcomas. Am J Roentgenol 2001; 177: 1307–1311.

15.

Ohguri

Aoki

Watanabe

. MRI findings including gadolinium-enhanced dynamic studies of malignant, mixed mesodermal tumors of the uterus: differentiation from endometrial carcinomas. Eur Radiol 2002; 12: 2737–2742.

16.

Tanaka

Tsunoda

Minami

. Carcinosarcoma of the uterus: MR findings. J Magn Reson Imaging 2008; 28: 434–439.

17.

Teo

Babagbemi

Peters

. Primary malignant mixed mullerian tumor of the uterus: findings on sonography, CT, and gadolinium-enhanced MRI. Am J Roentgenol 2008; 191: 278–283.

18.

Kato

Kanematsu

Furui

. Carcinosarcoma of the uterus: radiologic-pathologic correlations with magnetic resonance imaging including diffusion-weighted imaging. Magn Reson Imaging 2008; 26: 1446–1450.

19.

Bharwani

Newland

Tunariu

. MRI appearances of uterine malignant mixed Müllerian tumors. Am J Roentgenol 2010; 195: 1268–1275.

20.

Genever

Abdi

. Can MRI predict the diagnosis of endometrial carcinosarcoma? Clin Radiol 2011; 66: 621–624.

21.

Provencher S. LCModel & LCMgui user’s manual (LCModel version 6.3-1K), 2014. Available at: http://s-provencher.com/pub/LCModel/manual/manual.pdf.

22.

Kanthan

Senger

. Uterine carcinosarcomas (malignant mixed müllerian tumours): a review with special emphasis on the controversies in management. Obstet Gynecol Int 2011; 2011: 470795–470795.

23.

Glunde

Jacobs

Bhujwalla

. Choline metabolism in cancer: implications for diagnosis and therapy. Expert Rev Mol Diagn 2006; 6: 821–829.

24.

Payne

Schmidt

Morgan

. Evaluation of magnetic resonance diffusion and spectroscopy measurements as predictive biomarkers in stage 1 cervical cancer. Gynecol Oncol 2010; 116: 246–252.

25.

Takeuchi

Matsuzaki

Harada

. Differentiation of benign and malignant uterine corpus tumors by using proton MR spectroscopy at 3T: preliminary study. Eur Radiol 2011; 21: 850–856.

26.

Takeuchi

Matsuzaki

Harada

. Preliminary observations and clinical value of lipid peak in high-grade uterine sarcomas using in vivo proton MR spectroscopy. Eur Radiol 2013; 23: 2358–2363.

27.

Delikatny

Chawla

Leung

. MR-visible lipids and the tumor microenvironment. NMR Biomed 2011; 24: 592–611.

28.

Celik

Hascalik

Sarac

. Magnetic resonance spectroscopy of premalignant and malignant endometrial disorders: a feasibility of in vivo study. Eur J Obstet Gynecol Reprod Biol 2005; 118: 241–245.

29.

Mahon

Cox

Dina

. (1)H magnetic resonance spectroscopy of preinvasive and invasive cervical cancer: in vivo-ex vivo profiles and effect of tumor load. J Magn Reson Imaging 2004; 19: 356–364.

30.

Cho

Lee

. In-vivo proton magnetic resonance spectroscopy in adnexal lesions. Korean J Radiol 2002; 3: 105–112.