ORIGINAL STUDY

Magnetic Resonance Imaging in the Assessment

of High-Risk Features of Endometrial Carcinoma
A Meta-Analysis
Anna Luomaranta, MD, Arto Leminen, MD, and Mikko Loukovaara, MD

Objective: The aim of this study was to review the available literature on the reliability of
contemporary magnetic resonance imaging (MRI) techniques in the assessment of high-risk
features of endometrial carcinoma, that is, deep myometrial invasion, cervical stromal in-
volvement, and lymph node metastasis.
Methods: The PubMed and Scopus databases were searched for studies published before
March 2014. Studies on plain MRI were excluded.
Results: Fifty-two eligible studies were identified. For the assessment of deep (Q50%)
myometrial invasion (50 studies, 3720 patients), the pooled sensitivity, specificity, positive
predictive value, and negative predictive value were 80.7%, 88.5%, 77.6%, and 89.5%, re-
spectively, by random-effects analysis. For predicting cervical stromal involvement (12 studies,
1153 patients), the pooled values were 57.0%, 94.8%, 68.7%, and 90.5%, respectively. For
lymph node metastasis on a per-patient basis (10 studies, 862 patients), they were 43.5%,
95.9%, 66.3%, and 92.2%, respectively. In a meta-regression analysis, dynamic imaging was
associated with a higher sensitivity in detecting deep myometrial invasion, as compared with
contrast-enhanced imaging (P = 0.021). The improvement by diffusion-weighted imaging
was of a borderline significance (P = 0.057). Significant small-study effects were found for
the sensitivity of MRI in detecting deep myometrial invasion (P G 0.0001) and cervical
stromal involvement (P = 0.049).
Conclusions: Considering the poor-to-moderate sensitivity of MRI in detecting high-risk
features of endometrial carcinoma, patients with negative findings on MRI may not safely
forgo surgical staging unless the findings are confirmed by a backup method. The high
specificities allow the targeting of staging procedures by MRI alone in patients with positive
findings. Compared with contrast-enhanced imaging, dynamic and diffusion-weighted
imaging may be more reliable in the radiological staging of endometrial carcinoma.
Key Words: Endometrial carcinoma, Magnetic resonance imaging, Meta-analysis

Received March 25, 2014, and in revised form May 5, 2014.

Accepted for publication May 5, 2014.
(Int J Gynecol Cancer 2015;25: 837Y842)

A lthough 2 randomized trials failed to observe a survival

benefit from lymphadenectomy in women with clinical stage
lymphadenectomy may be appropriate for patients having tu-
mors with poor outcome or high risk for extrauterine spread.
I endometrial carcinoma,1,2 combined pelvic and para-aortic These include advanced stage carcinomas (stages IIYIV) and

Department of Obstetrics and Gynecology, Helsinki University Central University Central Hospital, PO Box 140, FIN-00029 HUS,
Hospital, Helsinki, Finland. Helsinki, Finland. E-mail: anna.luomaranta@fimnet.fi.
Address correspondence and reprint requests to Anna Luomaranta, The authors declare no conflicts of interest.
MD, Department of Obstetrics and Gynecology, Helsinki Supplemental digital content is available for this article. Direct URL
Copyright * 2015 by IGCS and ESGO citation appears in the printed text and is provided in the HTML
ISSN: 1048-891X and PDF versions of this article on the journal’s Web site
DOI: 10.1097/IGC.0000000000000194 (www.ijgc.net).

International Journal of Gynecological Cancer & Volume 25, Number 5, June 2015 837

Copyright © 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
Luomaranta et al International Journal of Gynecological Cancer & Volume 25, Number 5, June 2015

early carcinomas with lymphovascular invasion, deep (Q50%) and lymph node metastasis, for which variable sensitivity and
myometrial invasion, poor differentiation, or nonendometrioid specificity rates have been quoted by numerous studies. To sum-
histology. Surgical staging is supported in such cases because marize the results of the published studies, we conducted a meta-
it not only helps in directing adjuvant treatments but may also analysis on MRI in endometrial carcinoma.
provide a therapeutic effect.3Y5
During the past 2 decades, interest has developed in METHODS
using magnetic resonance imaging (MRI) for the recognition To identify the studies of interest, we searched the
of high-risk features and tailoring of the surgical treatment of PubMed and Scopus databases with no time limit (last search
endometrial carcinoma. Specifically, MRI has been used to de- on March 1, 2014). The following search terms were used:
tect deep myometrial invasion, cervical stromal involvement, magnetic, MRI, or MR imaging; endometrial, endometrium,

FIGURE 1. Forest plot showing the sensitivity of MRI in detecting myometrial invasion of 50% or greater in
endometrial carcinoma. Pooled effect estimate is from random-effects model.

838 * 2015 IGCS and ESGO

Copyright © 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
International Journal of Gynecological Cancer & Volume 25, Number 5, June 2015 MRI in Endometrial Carcinoma

uterine, or uterus; and adenocarcinoma, carcinoma, or cancer. a meta-regression module. The variables assessed by meta-
The title and abstract of articles identified in the computerized regression included the following: MRI scan type, number of
search were scanned to exclude any that were clearly irrele- patients, year of publication, and imaging criteria for pathol-
vant. The full text of the remaining articles was read to de- ogic lymph nodes. Funnel plots and Egger test were used to
termine whether they contained information on the topic of assess the possible existence of publication bias. Statistical
interest. The reference lists of relevant articles were searched significance was set for Ps G 0.05. Analyses were performed
for possible missing citation. To avoid overlapping data in using Comprehensive Meta-Analysis version 2 statistical soft-
duplicate studies (articles published by the same group, n = 7), ware (Biostat, Englewood, NJ).
the most recent reports were included. Non-English articles
were not included.
We included studies that evaluated the reliability of RESULTS
contrast-enhanced, dynamic, and diffusion-weighted MRI in Two hundred sixty and 96 records were identified by the
the assessment of myometrial invasion of 50% or greater, PubMed and Scopus database searches, respectively. We identi-
cervical stromal involvement, and/or lymph node metastasis fied 52 articles that met our inclusion criteria (see Table, Supple-
in endometrial carcinoma. Studies on plain MRI imaging and mental Digital Content 1, which shows the characteristics of the
studies in which the type of MRI scan was not reported (n = 17) eligible studies, available at http://links.lww.com/IGC/A224).6Y57
were excluded because our purpose was to assess the perfor- A total of 3816 patients with endometrial carcinoma were in-
mance of contemporary MRI techniques in the diagnostics of cluded in the studies. The median number of patients in the
endometrial carcinoma. Pathological evaluation after surgery studies was 53 (range, 20Y284).
served as the gold standard. Articles that did not differentiate The forest plot of sensitivities of 50 studies (3720 pa-
between cervical mucosal and stromal involvement and articles tients) that evaluated the diagnostic performance of MRI in
that did not define the imaging criteria for lymph node metas- detecting deep myometrial invasion in endometrial carcinoma
tasis were not included (n = 8). When results of different MRI is shown in Figure 1. The pooled sensitivity was 80.7% by
series were reported, the performance assessment of the most random-effects analysis. The pooled sensitivity was lower for
contemporary MRI scan type was prioritized, that is, dynamic the detection of cervical stromal involvement (12 studies, 1153
imaging before contrast-enhanced imaging10,12,13 and diffusion- patients) and lymph node metastasis on a per-patient basis
weighted imaging before contrast-enhanced36 and dynamic im- (10 studies, 862 patients) (57.0% and 43.5%, respectively)
aging.29,33,40,43,49,51,55 When diagnostic indices were reported (Figs. 2, 3). A summary of pooled diagnostic indices is found
separately for the detection of pelvic and para-aortic lymph node in Table 1.
metastasis, those of the former were used for the current study.50 To investigate the impact of study characteristics on
When the indices were given separately for 2 observers, the mean pooled sensitivity, we performed a univariate meta-regression
values were calculated and are reported herein.15,42,43,51,55 analysis. One study in which a mixture of 2 imaging techniques
Because of the significant heterogeneity for all diag- was used was excluded from the meta-regression that evaluated
nostic indices (I2 9 30%), their pooled values were determined the sensitivity of different scan types in detecting deep myomet-
by random-effects analysis. Sensitivity was the main end point rial invasion.52 Studies that reported sensitivities for 2 scan types
of the study. When diagnostic indices were not reported, they separately were included as independent studies for each scan
were determined from the numbers of true and false positive type.10,12,13,29,33,36,40,43,49,51,55 Compared with contrast-enhanced
and negative cases and prevalence, when appropriate. The effect imaging, dynamic imaging was associated with a higher sen-
of different study characteristics on sensitivity was tested with sitivity in detecting deep myometrial invasion (coefficient, 0.612;

FIGURE 2. Forest plot showing the sensitivity of MRI in detecting cervical stromal involvement in endometrial
carcinoma. Pooled effect estimate is from random-effects model.

* 2015 IGCS and ESGO 839

Copyright © 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
Luomaranta et al International Journal of Gynecological Cancer & Volume 25, Number 5, June 2015

FIGURE 3. Forest plot showing the sensitivity of MRI in detecting lymph node metastasis in endometrial carcinoma.
Pooled effect estimate is from random-effects model.

SE, 0.265; P = 0.021). The improvement by diffusion-weighted Finally, we tested whether publication bias could have
imaging, as compared with contrast-enhanced imaging, was affected the results. In accordance with the funnel plot showing
of a borderline significance (coefficient, 0.366; SE, 0.193; an apparent visual asymmetry (Fig. 4), Egger test revealed the
P = 0.057). The analysis did not suggest a different sensitivity existence of significant small-study effects for the sensitivity of
for dynamic and diffusion-weighted imaging (coefficient, 0.192; MRI in detecting deep myometrial invasion (P G 0.0001). The
SE, 0.266; P = 0.471). The number of eligible studies was too finding of small-study effects by Egger test was of a borderline
small to compare the sensitivities of different scan types in significance for the sensitivity of MRI in detecting cervical
detecting cervical stromal involvement or lymph node metas- stromal involvement (P = 0.049). For the sensitivity of MRI in
tasis (see Table, Supplemental Digital Content 1, available at detecting lymph node metastasis, the test showed no evidence
http://links.lww.com/IGC/A224). of publication bias (P = 0.167).
The number of included patients (per 10) was associ-
ated with a lower sensitivity of MRI in detecting deep myo- DISCUSSION
metrial invasion (coefficient, j0.050; SE, 0.018; P = 0.005), Preoperative MRI use among patients with endometrial
but no significant association existed between the publication carcinoma has increased significantly since the 1990s. For ex-
year and sensitivity of MRI in detecting deep myometrial in- ample, in Ontario, a 10.6-fold increase was noted between 1995
vasion (coefficient, j0.007; SE, 0.025; P = 0.790). Neither and 2005.58 It can be assumed that the upward trend is at least
of these variables was associated with the sensitivity of MRI partly due to the modern practice of risk stratification of patients
in detecting cervical stromal involvement or lymph node me- with endometrial carcinoma, so that only those considered to be
tastasis (P values between 0.305 and 0.882). Adherence to at high risk for lymph node metastasis are subjected to pelvic-
nodal size criteria alone (lymph node size of Q10 mm in minimal aortic lymphadenectomy. The MRI has gained popularity as a
diameter) was associated with a higher sensitivity of MRI in risk-stratification method because of its capability to detect
detecting metastatic lymph nodes (coefficient, 1.504; SE, 0.294; deep myometrial invasion and cervical stromal involvement,
P G 0.0001). both of which are associated with a high rate of pelvic lymph

TABLE 1. Pooled values for diagnostic indices with 95% confidence intervals by random-effects analysis
Deep Myometrial Invasion Cervical Stromal Involvement Lymph Node Metastasis
No. studies 50 12 10
No. patients 3720 1153 862
Sensitivity 80.7% (76.8%Y84.1%) 57.0% (45.9%Y67.4%) 43.5% (31.7%Y56.1%)
Specificity 88.5% (85.3%Y91.1%) 94.8% (92.1%Y96.6%) 95.9% (92.9%Y97.6%)
Positive predictive value 77.6% (73.4%Y81.2%) 68.7% (60.5%Y75.8%) 66.3% (54.8%Y76.1%)
Negative predictive value 89.5% (87.5%Y91.1%) 90.5% (87.7%Y92.8%) 92.2% (88.5%Y94.8%)

840 * 2015 IGCS and ESGO

Copyright © 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
International Journal of Gynecological Cancer & Volume 25, Number 5, June 2015 MRI in Endometrial Carcinoma

FIGURE 4. Funnel plot of sensitivity of MRI in detecting myometrial invasion of 50% or greater in endometrial
carcinoma, showing possible small-study effects. P G 0.0001 for the Egger test.

node metastasis and poor outcome. A 20% rate of positive Although the sensitivity of MRI in detecting metastatic
pelvic nodes was reported in patients with deeply invasive lymph nodes may be increased by adherence to strict imaging
endometrial carcinomas of endometrioid histology, compared criteria, the pooled sensitivity of MRI in detecting cervical
with 8% in patients with superficial carcinomas.59 Similarly, the stromal involvement and lymph node metastasis was too low
risk for pelvic lymph node metastasis doubles from 8% to 16% to suggest that MRI can usefully contribute to the decision
when the tumor involves the uterine isthmus or cervix.60 Be- regarding lymphadenectomy in patients with negative find-
sides assessing the risk for metastatic lymph nodes based on ings. The relatively high negative predictive values (990%)
characteristics of the primary tumor, MRI may detect metastatic can be explained by the low prevalence of stage II and stage
lymph nodes directly based on their size and morphology. IIIC carcinomas64; thus, a substantial proportion of these pa-
Sensitivity is the key metric for a test that is used for the tients will be surgically undertreated if the decision of lymph-
risk stratification of endometrial carcinomas because a high adenectomy is based on MRI findings alone. On the other hand,
sensitivity reduces false negative results and helps to avoid as concluded from the specificities and positive predictive values,
incomplete staging procedures in patients who most likely the rate of type 1 error is small, and lymphadenectomy should
benefit from lymphadenectomy. The sensitivity of 80.7% of be considered if cervical stromal or lymph node involvement is
MRI in detecting deep myometrial invasion, or the negative diagnosed by MRI.
predictive value of 89.5%, may not be high enough to allow Standard morphology imaging of endometrial carci-
omission of lymphadenectomy in patients with no or superfi- noma by plain and contrast-enhanced MRI has been replaced
cial myometrial invasion according to MRI. It could be pro- in many centers by dynamic MRI after its introduction as a
posed that a negative finding should be confirmed by another novel imaging technique in the 1990s. Dynamic MRI enhances
method, such as intraoperative frozen section analysis. By the contrast between the primary tumor and normal myometrium
contrast, because of the fairly high specificity (88.5%) and and cervical stroma and improved the staging accuracy of endo-
positive predictive value (77.6%), MRI could serve as a 1-stop metrial carcinoma in some10,13,19,20,34,39 but not all11,12,22,41,55
method of risk stratification when deep myometrial invasion studies. Diffusion-weighted MRI is a functional imaging
is diagnosed by this modality. MRI may be the most useful technique that displays information about water mobility, tissue
under circumstances where risk stratification cannot be based cellularity, and the integrity of cell membranes.65 Its accuracy in
on intraoperative assessment of the uterus, such as in women the assessment of the depth of myometrial invasion has not
desiring fertility-sparing therapy by progestins for complex atyp- been clearly superior to dynamic MRI in comparative stud-
ical hyperplasia or nonmyoinvasive endometrial carcinoma.61 ies.29,33,40,43,49,51,55 Our meta-regression analysis did not sug-
An overlapping disease-specific survival has been found gest a different sensitivity for dynamic and diffusion-weighted
for women with stage I endometrial carcinoma with and without MRI in detecting deep myometrial invasion. However, com-
cervical glandular involvement.62 Accordingly, cervical glandu- pared with contrast-enhanced MRI, dynamic MRI was asso-
lar involvement is no longer included as a staging criterion in the ciated with a higher sensitivity in detecting deep myometrial
International Federation of Gynecology and Obstetrics staging invasion, and the increase in sensitivity by diffusion-weighted
system for endometrial carcinoma.63 This prompted us to ex- MRI was of a borderline significance. Taken together, these
clude articles in which cervical glandular involvement was not data suggest that modern scanning techniques improve the re-
differentiated from cervical stromal involvement in the inter- liability of MRI in the diagnostic workup of patients with endo-
pretation of MRI scans. Regarding lymph node metastasis, we metrial carcinoma. Meta-regression analysis also showed that a
included articles in which the imaging criteria for pathologic higher number of included patients was associated with a lower
nodes were defined. sensitivity of MRI in detecting deep myometrial invasion, which

* 2015 IGCS and ESGO 841

Copyright © 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
Luomaranta et al International Journal of Gynecological Cancer & Volume 25, Number 5, June 2015

is in agreement with the finding of significant small-study effects gadolinium-enhanced T1-weighted MR imaging.
according to the funnel plot and Egger test. Radiographics. 1999;19:937Y945.
In conclusion, this study summarizes published data on 15. Saez F, Urresola A, Larena JA, et al. Endometrial carcinoma:
the validity of contemporary MRI techniques in the radiological assessment of myometrial invasion with plain and
staging of endometrial carcinoma. Our findings suggest that gadolinium-enhanced MR imaging. J Magn Reson Imaging.
the sensitivity of MRI in recognizing high-risk tumor features is 2000;12:460Y466.
not high enough to allow omission of lymphadenectomy in 16. Zarbo G, Caruso G, Caruso S, et al. Endometrial cancer:
patients with negative findings. The high specificity of MRI preoperative evaluation of myometrial infiltration magnetic
favors surgical staging in patients with positive findings. Com- resonance imaging versus transvaginal ultrasonography.
Eur J Gynaecol Oncol. 2000;21:95Y97.
pared with contrast-enhanced imaging, dynamic and diffusion-
17. Cunha TM, Félix A, Cabral I. Preoperative assessment of deep
weighted imaging may be more reliable in the radiological
myometrial and cervical invasion in endometrial carcinoma:
staging of endometrial carcinoma.
comparison of magnetic resonance imaging and gross visual
REFERENCES inspection. Int J Gynecol Cancer. 2001;11:130Y136.
1. Benedetti Panici P, Basile S, Maneschi F, et al. Systematic pelvic 18. Manfredi R, Mirk P, Maresca G, et al. Local-regional staging
lymphadenectomy vs. no lymphadenectomy in early-stage of endometrial carcinoma: role of MR imaging in surgical
endometrial carcinoma: randomized clinical trial. J Natl Cancer planning. Radiology. 2004;231:372Y378.
Inst. 2008;100:1707Y1716. 19. Nasi F, Fiocchi F, Pecchi A, et al. MRI evaluation of myometrial
2. ASTEC Study Group. Efficacy of systematic pelvic invasion by endometrial carcinoma. Comparison between
lymphadenectomy in endometrial cancer (MRC ASTEC trial): fast-spin-echo T2w and coronal-FMPSPGR
a randomised study. Lancet. 2009;373:125Y136. gadolinium-dota-enhanced sequences. Radiol Med.
3. Chan JK, Cheung MK, Huh WK, et al. Therapeutic role 2005;110:199Y210.
of lymph node resection in endometrioid corpus cancer. A study 20. Nakao Y, Yokoyama M, Hara K, et al. MR imaging in
of 12,333 patients. Cancer. 2006;107:1823Y1830. endometrial carcinoma as a diagnostic tool for the absence
4. Todo Y, Kato H, Kaneuchi M, et al. Survival effect of para-aortic of myometrial invasion. Gynecol Oncol. 2006;102:343Y347.
lymphadenectomy in endometrial cancer (SEPAL study): 21. Chung HH, Kang SB, Cho JY, et al. Accuracy of MR imaging
a retrospective cohort analysis. Lancet. 2010;375:1165Y1172. for the prediction of myometrial invasion of endometrial
5. Mahdi H, Kumar S, Al-Wahab Z, et al. Prognostic impact of carcinoma. Gynecol Oncol. 2007;104:654Y659.
lymphadenectomy in uterine serous cancer. BJOG. 2013; 22. Rockall AG, Meroni R, Sohaib SA, et al. Evaluation
120:384Y391. of endometrial carcinoma on magnetic resonance imaging.
6. Yamashita Y, Mizutani H, Torashima M, et al. Assessment Int J Gynecol Cancer. 2007;17:188Y196.
of myometrial invasion by endometrial carcinoma: transvaginal 23. Ryoo UN, Choi CH, Yoon JY, et al. MR imaging in
sonography vs contrast-enhanced MR imaging. AJR Am J endometrial carcinoma as a diagnostic tool for the prediction
Roentgenol. 1993;161:595Y599. of myometrial invasion and lymph node metastasis.
7. Ito K, Matsumoto T, Nakada T, et al. Assessing myometrial Cancer Res Treat. 2007;39:165Y170.
invasion by endometrial carcinoma with dynamic MRI. 24. Vasconcelos C, Félix A, Cunha TM. Preoperative assessment
J Comput Assist Tomogr. 1994;18:77Y86. of deep myometrial and cervical invasion in endometrial
8. Kim SH, Kim HD, Song YS, et al. Detection of deep myometrial carcinoma: comparison of magnetic resonance imaging and
invasion in endometrial carcinoma: comparison of transvaginal histopathologic evaluation. J Obstet Gynaecol. 2007;27:65Y70.
ultrasound, CT, and MRI. J Comput Assist Tomogr. 25. Yahata T, Aoki Y, Tanaka K. Prediction of myometrial invasion
1995;19:766Y772. in patients with endometrial carcinoma: comparison of magnetic
9. Minderhoud-Bassie W, Treurniet FE, Koops W, et al. Magnetic resonance imaging, transvaginal ultrasonography, and gross
resonance imaging (MRI) in endometrial carcinoma; visual inspection. Eur J Gynaecol Oncol. 2007;28:193Y195.
preoperative estimation of depth of myometrial invasion.
26. Cabrita S, Rodrigues H, Abreu R, et al. Magnetic resonance
Acta Obstet Gynecol Scand. 1995;74:827Y831.
imaging in the preoperative staging of endometrial carcinoma.
10. Seki H, Kimura M, Sakai K. Myometrial invasion of
Eur J Gynaecol Oncol. 2008;29:135Y137.
endometrial carcinoma: assessment with dynamic MR and
contrast-enhanced T1-weighted images. Clin Radiol. 27. Sanjuán A, Escaramı́s G, Ayuso JR, et al. Role of magnetic
1997;52:18Y23. resonance imaging and cause of pitfalls in detecting myometrial
11. Savci G, Ozyaman T, Tutar M, et al. Assessment of depth of invasion and cervical involvement in endometrial cancer.
myometrial invasion by endometrial carcinoma: comparison Arch Gynecol Obstet. 2008;278:535Y539.
of T2-weighted SE and contrast-enhanced dynamic GRE MR 28. Savelli L, Ceccarini M, Ludovisi M, et al. Preoperative local
imaging. Eur Radiol. 1998;8:218Y223. staging of endometrial cancer: transvaginal sonography vs.
12. Takahashi S, Murakami T, Narumi Y, et al. Preoperative staging magnetic resonance imaging. Ultrasound Obstet Gynecol.
of endometrial carcinoma: diagnostic effect of T2-weighted 2008;31:560Y566.
fast spin-echo MR imaging. Radiology. 1998;206:539Y547. 29. Shen SH, Chiou YY, Wang JH, et al. Diffusion-weighted
13. Joja I, Asakawa T, Shiraiwa M, et al. Endometrial carcinoma: single-shot echo-planar imaging with parallel technique in
multisection dynamic MR imaging using a three-dimensional assessment of endometrial cancer. AJR Am J Roentgenol.
FLASH technique during breath holding. Radiat Med. 2008;190:481Y488.
1999;17:211Y218. 30. Torricelli P, Ferraresi S, Fiocchi F, et al. 3-T MRI in the
14. Lee EJ, Byun JY, Kim BS, et al. Staging of early endometrial preoperative evaluation of depth of myometrial infiltration in
carcinoma: assessment with T2-weighted and endometrial cancer. AJR Am J Roentgenol. 2008;190:489Y495.

842 * 2015 IGCS and ESGO

Copyright © 2015 by IGCS and ESGO. Unauthorized reproduction of this article is prohibited.
Copyright © 2015 by IGCS and ESGO2015