Value of diffusion-weighted MRI in diagnosis of uterine cervical cancer: a prospective study evaluating the benefits of DWI compared to conventional MR sequences in a 3T environment

Abstract

Background

Imaging of cervical carcinoma remains challenging as local infiltration of surrounding tissues cannot always be discriminated safely. New imaging techniques, like diffusion-weighted imaging (DWI) have emerged, which could lead to a more sensitive tumor detection.

Purpose

To evaluate the benefits of DWI for determination of size, local infiltration, and tumor grading, in patients with primary and recurrent cervical cancer.

Material and Methods

In this prospective, study we enrolled 50 patients with primary (n = 35) and recurrent (n = 15) tumors. All patients underwent 3T magnetic resonance imaging (MRI) including conventional (e.g. T1/T2 ± fs ± contrast) sequences and DWI (b-values of 0, 50, 400, 800 s/mm²). All images were analyzed by three readers with different experience levels (1, 3, 6 years), who compared image quality, tumor delineation, dimensions, local infiltration, lymph node involvement, and quantified ADC values compared to the histopathological grading.

Results

Additional use of DWI resulted in significantly better (P < 0.001) tumor delineation for the least experienced reader, but not for experienced readers. Tumor dimensions were assessed almost equally (P > 0.05) in conventional sequences and DWI. Use of DWI led to an increase in sensitivity of infiltrated adjacent tissue (from 86% to 90%) and detection of lymph node metastases (from 47% to 67%). Quantitative assessment of carcinomas showed lower ADC values (P < 0.001) with significant inverse correlations between different grading levels.

Conclusion

Our study demonstrates the overall benefits using DWI in 3T MRI resulting in a higher reader confidence, sensitivity of tissue infiltration, and tumor-grading for cervical cancer.

Keywords

Magnetic resonance (MR) diffusion cervix genital/reproductive neoplasms – primary pathology

Introduction

Despite new surgical and conservative approaches like total mesometrial resection (1) and use of extended radiochemotherapy (2) cervical carcinoma still remains the third most common malignancy in women worldwide and is characterized by a high mortality rate (3). Imaging of local spread (e.g. involvement of the bladder, rectum, and parametrium) and the detection of lymph node involvement still remains challenging but is inevitable for determination of an adequate therapy even if surgical staging remains the most accurate diagnostic step and is especially recommended for ≤TIIb tumors according to German guidelines for cervical carcinomas (4).

Magnetic resonance imaging (MRI) is the state of the art modality for presurgical staging and follow-up examinations (5,6). A crucial prognostic factor in cervical cancer staging is the infiltration of the parametrium (FIGO IIB), which is a cut-off criterion for extended surgical approach via total mesometrial resection (TMMR), which means the resection of the morphogenetic unit of the uterus, as performed in our institution (7) or Wertheim-Meigs procedure (8), which is still seen as the surgical standard. However, especially in North America, patients with tumor staged FIGO IIB undergo a primary chemotherapy. For MRI, accuracy, sensitivity, and specificity of parametrical infiltration is in the range of 68–95%, 68–79%, and 80–84%, respectively (9), for standard imaging protocols including T1-weighted (T1W) and T2-weighted (T2W) sequences in at least two planes and T1W sequences after injection of contrast media. However, over recent years several new imaging techniques, like DWI and dynamic contrast-enhanced (DCE) imaging have emerged and also been embedded in pelvic examinations (10,11).

DWI hereby serves as a functional technique, which provides information about water mobility, tissue cellularity, and stability of membrane integrity. Present studies show promising and reliable results discriminating cervical carcinoma from healthy tissue and also to evaluate tumor grading (12) in ADC maps. Additionally, previous studies (13) show a beneficial effect using higher magnetic field strengths (3 T) in pelvic examinations with significantly higher image contrast and less motion artifacts probably due to shorter scan times. The purpose of this study was to evaluate possible advantages of DWI compared to standard T1/T2 sequences in 3T MRI regarding local tumor invasiveness, lymph node infiltration, tumor grading, and overall observer confidence.

Material and Methods

Patients

Our institutional review board approved this study and informed consent was obtained from all patients. Fifty female patients (mean age, 52 years; age range, 30–80 years) with a high clinical suspicion for primary cervical cancer (n = 35) (sonographic or palpable lesion, positive tumor markers, and/or positive histological samples) or recurrence (n = 15) of cervical cancer were prospectively enrolled in this study between March 2010 and February 2013. All patients who were clinically suspected of having cervical cancer and diagnosed by MRI also underwent further clinical evaluation including bimanual pelvic examination, chest radiography, cystoscopy, and sigmoidoscopy. After completion of initial staging 21 patients underwent a total mesometrial resection (TMMR; n = 16) or a lateral extended endopelvine resection (LEER; n = 5) and the resected specimens and lymph-node-bearing tissues were sent for histopathological workup. The decision on therapy was based on a combination of MRI results and clinical FIGO stage. A histopathological sample was obtained for all other patients (n = 29) who did not receive surgical treatment

MRI

All scans were performed on a 3 T MR scanner (Trio Tim, Siemens®, Erlangen, Germany). In order to reduce bowel movement, patients were recommended to fast 4 h prior to imaging. Depending on contraindications patients received either 1 mg glucagon (Glucagen Novo Hypokit®, Novo Nordisk, Mainz, Germany), or 20 mg butylscopolamine (Buscopan®, Boehringer-Ingelheim, Ingelheim, Germany) to prevent artifacts from peristalsis.

A 6-channel body phased array coil (Siemens®, Erlangen, Germany) was used. Routine pelvic MR images including application of contrast agent (Gadovist®, Bayer Healthcare, Leverkusen, Germany; 0.1 mmol/kg/BW) and additional images were acquired as shown in Table 1.

Table 1.

MRI sequences

	Plane	TR/TE	Matrix	b-values	Bandwidth
Conventional MRI sequences (slice thickness: 5 mm)
T2W TSE	Axial	7260 ms/102 ms	512 × 512
T2W TSE	Sagittal	5240 ms/103ms	512 × 512
PD-weighted TSE with fs	Sagittal	3000 ms/38 ms	512 × 512
T1W TSE native and with contrast agent + fs (0.1 mmol/kg/BW)	Axial	800 ms/11 ms	512 × 512
	Sagittal	593 ms/11 ms	512 × 512
Additional MRI sequences (slice thickness: 3 mm)
T2W TSE (high-resolution)	Sagittal	4110 ms/126 ms	192 × 192
DWI (single-shot echo-planar sequence):	Axial	FOV: 220 × 220	128 × 128	0, 50, 400, 800 s/mm²	1628 Hz/Px

Image analysis

Three readers with different experience levels in gynecological MRI (Reader 1, 1 year; Reader 2, 3 years; and Reader 3, 6 years) independently reviewed all sequences. Every reader rated the image quality and tumor delineation of each sequence using a Likert-scaling system in the range of 0–4 (0, insufficient; 1, poor; 2, adequate; 3, good; 4, excellent). Criteria for image quality and tumor delineation were presence of motion artifacts, signal-extinction due to metal implants, SNR, and overall tumor visibility (according to general MR criteria with increased T2 signal, signal reduction in T1, and accumulation of contrast medium).

Largest tumor dimension, which was defined as the maximum antero-posterior (a.p.) and lateral length, was determined in every sequence, after the readers were instructed how to perform this measurement.

Local tumor extension and infiltration of adjacent tissue was rated by each reader for every single sequence, analyzing conventional datasets first and additional DWI datasets afterwards. Sequences were displayed only one after the other, so there was a natural delay of approximately 20 min from analyzing the T2W sequence and the DWI. For each dataset parameters like infiltration of the parametrium, bladder infiltration, rectal infiltration, and lymph node involvement were evaluated (0, not ratable; 1, no evidence; 2, suspicious; 3, highly suspicious) and the radiological T- and N-stage established. Criteria for the diagnosis of parametrial invasion were: disruption of the cervical margin; tumor mass inside the parametrium; thickening of the uterine ligaments; and no definable healthy cervical tissue. Criteria for the diagnosis of lymph node metastasis were: short-axis diameter >10 mm; round shape; and strong contrast medium enhancement (14).

For pelvic examinations in which MRI could detect lesions, averages of three representative (areas with visually highest alteration of diffusion/lowest ADC) polygonal-shaped regions of interest (ROIs) were defined within the lesions on the axial DW images in the slices where the largest lesion was visible, while referring to the corresponding T2W images for verification of the zonal boundaries. The ROIs were then transferred to the corresponding axial ADC map. Additionally three ROIs were defined within MR-morphological healthy cervical stroma. Maximum, mean, and minimum SI values were measured for tumor and normal cervical stroma in T2-TSE, DW images (b50, b400, and b800) and ADC-coefficients (ADC_min, ADC_mean, and ADC_max) were determined.

Histopathological analysis

The histopathological examination of the radical hysterectomy specimen was made in a standardized manner as described previously (1,15,16). The tumor size was determined on the fixed specimen. All tumors were staged and classified according to the WHO and TNM classifications (17,18).

Statistical analysis

Datasets were evaluated on a syngo.via work suite (Siemens®, Erlangen, Germany). Statistical analysis was performed with the data analysis program SPSS® 20 (IBM®, Armonk, NY, USA). Statistical significance was considered for a P value less than 0.05. One-way ANOVA analysis and Mann-Whitney U test were used to compare the differences in ADC values, tumor size, and quality rating of each sequence.

Results

Image quality and tumor delineation

Evaluation of image quality and tumor delineation (Fig. 1) indicated that the reader with the lowest experience (Reader 1) scored the DWI sequence significantly better than the conventional sequences (P < 0.001). In comparison, Reader 2 ranked all sequences almost equally with approximately same values (P = 0.882), whereas Reader 3 scored the conventional MR sequences better than the additional DWI, but not significantly (P = 0.184, Table 2).

Fig. 1.

Assesment of image quality and tumor delineation for different sequences (T2, T1 postcontrast, ADC, and DWI with b = 800) by each reader with different experience (A = Reader 1 with 1 year of experience, B = Reader 2 with 3 years of experience, C = Reader 3 with 6 years of experience).

Table 2.

Evaluation of image quality and tumor delineation in different sequences dependent on reader experience by using a Likert scale.

Sequence	Reader 1	Reader 2	Reader 3
T2W TSE	2.2 ± 0.4	3.0 ± 0.5	2.9 ± 0.9
T1W TSE + CM	2.4 ± 0.6	3.0 ± 0.7	2.8 ± 1.0
DWI (b 800)	2.8 ± 0.5	2.9 ± 0.8	2.6 ± 0.4
ADC	2.7 ± 0.5	3.0 ± 0.7	2.5 ± 1.0

Thus, the FIGO stage of primary cervical cancer was determined according to clinical, pathological, and MRI evaluation (FIGO IB1/2, 6 patients; FIGO IIA, 1 patient; FIGO IIB, 14 patients; FIGO III, 7 patients; and FIGO IV, 7 patients).

Tumor size

The results for tumor size (antero-posterior and lateral length) are shown in Table 3. In conclusion there was no significant difference (P > 0.05) in the mean tumor size comparing conventional sequences and DWI for all three readers. For valid comparison of the sequences we focused on two-dimensional measurements of the conventional planes and DWI, as the latter was only performed in transverse orientation.

Table 3.

Evaluation of maximum tumor dimension (mean ± SD, width × depth in cm) in different sequences dependent on reader experience.

Sequence	Reader 1	Reader 2	Reader 3
CE T1/T2	4.3 ± 2.1 × 3.8 ± 2.1	4.0 ± 1.9 × 3.8 ± 1.9	3.9 ± 2.0 × 3.5 ± 2.2
+DWI	4.3 ± 1.9 × 4.2 ± 2.1	3.8 ± 1.8 × 3.9 ± 2.1	3.8 ± 2.0 × 3.6 ± 2.2

Infiltration of adjacent tissue and lymph nodes

To determine sensitivity and specificity, all tumors suitable for surgical therapy (n = 21) were compared to initial MR findings. The results for sensitivity and specificity are shown in Table 4. Estimating the infiltration of adjacent tissue we found an overall sensitivity for the conventional MRI sequences of 86% for parametrial infiltration (n = 14) with slightly lower values for the inexperienced Reader 1 (P > 0.05). There was a low but not significant (P > 0.05) increase of sensitivity to 90% by comprising additional DWI. Specificity for parametrial invasion amounted to 67% for conventional MRI sequences (T1W/T2W) and 62% for additional use of DWI. For bladder invasion (n = 1) the sensitivity and specificity was 100% for each reader.

Table 4.

Sensitivity and specificity for the detection of parametrial infiltration, metastatic lymph nodes, and bladder infiltration by all three readers by the use of T1/T2 and additional use of DWI.

	Reader1		Reader 2		Reader 3		Overall
	CE T1/T2	+DWI	CE T1/T2	+DWI	CE T1/T2	+DWI	CE T1/T2	+DWI
Sensitivity
Parametrial infiltration (n = 14)	79%	79%	93%	100%	86%	93%	86%	90%
Metastatic lymph nodes (n = 12)	33%	50%	50%	67%	58%	83%	47%	67%
Bladder infiltration (n = 1)	100%	100%	100%	100%	100%	100%	100%	100%
Specificity
Parametrial infiltration (n = 14)	86%	86%	43%	43%	71%	57%	67%	62%
Metastatic lymph nodes (n = 12)	78%	67%	56%	44%	56%	33%	63%	48%
Bladder infiltration (n = 1)	100%	100%	100%	100%	100%	100%	100%	100%

Overall evaluation of metastatic lymph node involvement showed an increase of sensitivity from 47% to 67% and a decrease of specificity from 63% to 48% applying additional DWI. Furthermore we found a slight increase of lymph node metastasis detection accuracy from 53% to 61% using additional DWI (Fig. 2).

Fig. 2.

Comparison of sensitivity and specificity for parametrial invasion and lymph node for conventional MRI sequences and additional DWI.

Use of DWI increased the confidence of metastatic lymph nodes detection from suspicious to highly suspicious in three cases (Readers 2 and 3) and in one/three cases for parametrial invasion (Readers 1 and 3, respectively).

ADC values

The results for measurement of ADC values are shown in Table 5. The differences between normal cervical tissue and tumor tissue were highly significant (P < 0.001) for each ADC value. There was no significant difference between the differentiation grades (G1, n = 4; G2, n = 23; G3, n = 23) for ADC_min in G1–G2 (P = 0.12), G1–G3 (P = 0.06), and G2–G3 (P = 0.73); however, our results indicated lower ADC_min values for poorly differentiated tumors. For ADC_max there was a significant difference between G1 and G2 (P = 0.03) and G1 and G3 (P = 0.03) but no significant difference between G2 and G3 (P = 0.98).

Table 5.

Measurement of ADC values (×10^–3mm²/s, mean ± SD) for tumor-free cervical stroma and different tumor grades.

	Stroma	Tumor overall	G1	G2	G3	Stroma vs. tumor	G1 vs. G2	G1 vs. G3	G2 vs. G3
ADC_min	1.244 ± 0.197	0.719 ± 0.155	0.915 ± 0.400	0.728 ± 0.133	0,685 ± 0,117	<0.001	0.12	0.06	0.73
ADC_mean	1.484 ± 0.18	0.849 ± 0.157	1.091 ± 0.370	0.859 ± 0.133	0.807 ± 0.116	<0.001	0.03	<0.001	0.56
ACD_max	1.704 ± 0.185	0.978 0.175	1.291 ± 0.351	1.003 ± 0.136	0.912 ± 0.136	<0.001	0.03	0.03	0.98

For ADC_mean values there was a significant difference between G1 and G2 (P = 0.03) and G1 and G3 (P = 0.001) but no significant difference between G2 and G3 (P = 0.56) (Fig. 3).

Fig. 3.

ADC values for different differentiation grades of tumor tissue (G1–G3) in comparision to tumor-free cervical stroma (=stroma). Graph A shows results for ADC_min, Graph B results for ADC_mean, and Graph C results for ADC_max.

Discussion

Imaging local and distant invasiveness of cervical carcinoma with higher magnetic field strengths contributes significantly to optimized staging and is inevitable for presurgical planning and adjustment of therapy (13). Nowadays MR scanners with field strengths of 3T have become widely available in clinical routine.

Emerging from stroke assessment (19) DWI has gained a lot of attention in several imaging approaches like oncologic staging, therapy response, and characterization and detection of inflammation. In our study we analyzed the benefits of additional use of DWI with conventional contrast-enhanced T1/T2 sequences. Interestingly the reader with the lowest experience in gynecological imaging found the additional use of DWI and ADC maps more useful in detection and delineation of tumor tissue. More experienced readers, in our study with 3 and 6 years of experience seemed to prefer conventional MR sequences like high-resolution T2 and contrast-enhanced T1 images over DWI in establishing their diagnosis and delineating tumor boundaries (Fig. 4), which can be explained due to more profound experience with common MR sequences.

Fig. 4.

Large cervical carcinoma; with parametrial infiltration on the left side (white arrow = parametrial infiltration on the left side), which can be sufficiently detected in every sequence (axial plane: A = T2, B = T1CM, C = b800, D = ADC; sagittal plane: E = T2 with marked angulation of the axial plane).

Regarding maximum tumor dimensions there was no significant difference in average size comparing conventional and diffusion-weighted sequences, which makes DWI equal to contrast-enhanced T1/T2 sequences despite reduced anatomical information.

Local tumor spread and lymph node involvement is crucial for disease outcome and therapy strategy in every malignant disease. For MR, size is the main criterion used to differentiate between nodal metastases and benign lymph nodes, and the most widely quoted threshold above which a node is considered malignant is 10 mm in short-axis diameter (14). Other, softer criteria commonly used are, for example, a round shape, central necrosis, soft tissue of the same signal intensity of the tumor within the node, and extracapsular extension of the tumor beyond the nodal capsule. The presence of visible necrosis within the node, seen as pockets of high T2 signal intensity, has a positive predictive value of 100% (20) for nodal involvement. However, in clinical routine these parameters often become secondary, especially in patients who underwent neoadjuvant radiation therapy or chemotherapy, which generally results in morphological tissue changes. Liyanage (21) investigated these standard criteria in cervical carcinoma and found that the sensitivity for detecting nodal metastases on MRI is low with a wide range of 29–86%. This is due to the inability to detect micrometastasis in normal-sized nodes. In light of that new imaging techniques for cancer staging combining nuclear and radiological hybrid techniques like PET/CT or PET/MRI have emerged to improve the assessment of lymph node involvement.

18F-fluorodeoxyglucose (FDG)-PET/CT has been proven to be a useful tool for oncological lymph node staging. Direct comparison of sensitivity in PET/CT and MRI of lymph nodes showed superior results for PET/CT with percentages of 91% vs. 73% and specificity of 100% vs. 83% in a cohort of 11 patients (22). However, there are differences concerning sensitivity with respect to micrometastasis. Leblanc (23) investigated morphological unsuspicious lymph nodes with PET and correlated the results to the histopathological samples with a sensitivity of just 33% and a specificity of 94% for PET/CT. Other issues for PET techniques, especially in the detection of malignomas of the female pelvis, are the close proximity to the bladder and ureters which often cover relevant tumor areas. However, there is evidence that PET is superior to MRI in detection of malignant lymph nodes (24). Even though these modalities become more available they should preferably just come into play when there is suspicion of distant metastasis, according to German guidelines (S3 guidelines, 4). The additional use of DWI in detection of nodal metastasis has been discussed controversially in the literature throughout the last years. We found a significant increase in sensitivity for the detection of lymph node metastasis with a decrease in specificity for all three readers. Over all there is an increase of 9% for diagnostic accuracy (Fig. 5). For the evaluation of the infiltration of surrounding tissue only a slight increase in sensitivity could be found by the additional use of DWI.

Fig. 5.

Metastatic lymph node involvement next to the external iliacal vessels (white circle = metastatic lymph node), which is not suspicious in conventional MRI sequences but does show diffusion restriction (axial plane: A = T2, B = T1CM, C = DWI with b = 800, D = ADC).

One other benefit of DWI is the determination of the tumor grade as discussed in previous studies, e.g. Kuang et al. (12). In poorly differentiated tumors water can move less freely due to the increasing cellularity with hydrophobic membrane integrity, extracellular tortuosity, and tissue disorganization, resulting in lower ADC values (25,26). Due to the significant difference between the ADC values of healthy and tumorous tissue, it is possible to differentiate tumor-free and tumorous tissue by quick measurement of ADC values through proper ROI placement. For our MR scanner we found significantly higher ADC values for healthy cervical tissue compared to malignant tissue. Assessment of differentiation levels show that well differentiated tumors (G1) have higher ADC values compared to moderate and poorly differentiated tumors (G2 and G3). Determination of a distinct cutoff value for different tumor grades remains difficult. In our study population, probably due to a relatively inhomogeneous distribution of tumor grades, we could not establish a valid cutoff value. However, it has already been shown that calculated ADC values seem to differ between MR scanners (27), so comparison of ADC values from different vendors has to be done with great caution, if at all. Compared to a study of Kuang et al. (12), we found comparable ADC values for tumor-free tissue and G2 tumors, but slightly higher ADC values for G1 and G3 tumors. Further investigation of ADC values according to the tumor grade in a larger, more homogenous patient cohort with primary cervical carcinomas and thorough histopathological workup might be helpful to define exact cutoff or normalized values for ADC to improve the tumor characterization, which could result in a more sophisticated therapy approach.

However, there are some limitations to this study. One possible limitation could be a minor reader bias based on the algorithm the images where interpreted. Even though the MR sequences were not viewed all together on one display, readers’ decisions could have been slightly influenced by viewing the conventional sequences first and the DWI afterwards (after approximately 20 min). Another limitation certainly is the missing comparison of MR measured tumor diameter to histopathological samples.

In conclusion, our study demonstrates the overall benefits and additional information given by using DWI in a 3T environment, which resulted in a higher reader confidence for lesion delineation, higher sensitivity for prediction of lymph node infiltration, and possibility of MR-based determination of tumor grading.

Footnotes

Acknowledgments

The authors kindly thank the Department of Gynaecology and Obstetrics (University Hospital Leipzig, Prof. Dr. Dr. M. Höckel, Ms Katja Schmidt) for the admission of selected patients with cervical carcinomas and providing insights into clinical procedures and relevant aspects of embryological tumor behavior. Furthermore, the authors thank the technical radiological staff for supervision of MR examinations and patient surveillance.

Declaration of conflicting interests

The author(s) declare no 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

Höckel

Horn

Einenkel

. Laterally extended endopelvic resection: surgical treatment of locally advanced and recurrent cancer of the uterine cervix and vagina based on ontogenetic anatomy. Gynecol Oncol 2012; 127: 297–302.

Milgrom

Kollmeier

Abu-Rustum

. Postoperative external beam radiation therapy and concurrent cisplatin followed by carboplatin/paclitaxel for stage III (FIGO 2009) endometrial cancer. Gynecol Oncol 2013; 130: 436–440.

Jemal

Bray

Center

. Global cancer statistics. CA Cancer J Clin 2011; 61: 69–90.

S3-Leitlinie Diagnostik, Therapie und Nachsorge der Patientin mit Zervixkarzinom. Available at: http://www.awmf.org/uploads/tx_szleitlinien/032-033OLl_S3_Zervixkarzinom_2014-10.pdf (accessed November 2014).

Balleyguier

Sala

Da Cunha

. Staging of uterine cervical cancer with MRI: guidelines of the European Society of Urogenital Radiology. Eur Radiol 2011; 21: 1102–1110.

Bhosale

Peungjesada

Devine

. Role of magnetic resonance imaging as an adjunct to clinical staging in cervical carcinoma. J Comput Assist Tomogr 2010; 34: 855–864.

Höckel

Horn

Hentschel

. Total mesometrial resection: high resolution nerve-sparing radical hysterectomy based on developmentally defined surgical anatomy. Int J Gynecol Cancer 2003; 13: 791–803.

Mangler

Zech

Schneider

. Aspects of therapy for cervical cancer in Germany 2012: Results from a survey of German gynaecological hospitals. Geburtshilfe Frauenheilkd 2013; 73: 227–238.

Bipat

Glas

van der Velden

. Computed tomography and magnetic resonance imaging in staging of uterine cervical carcinoma: a systematic review. Gynecol Oncol 2003; 91: 59–66.

10.

Tamai

Koyama

Saga

. Diffusion-weighted MR imaging of uterine endometrial cancer. J Magn Reson Imaging 2007; 26: 682–687.

11.

Beddy

Moyle

Kataoka

. Evaluation of depth of myometrial invasion and overall staging in endometrial cancer: comparison of diffusion-weighted and dynamic contrast enhanced MR imaging. Radiology 2012; 262: 530–537.

12.

Kuang

Ren

Zhong

. The value of apparent diffusion coefficient in the assessment of cervical cancer. Eur Radiol 2013; 23: 1050–1058.

13.

Kataoka

Kido

Koyama

. MRI of the female pelvis at 3T compared to 1.5T: evaluation on high-resolution T2-weighted and HASTE images. J Magn Reson Imaging 2007; 25: 527–534.

14.

Carrington BM. Cervical Cancer. In: Mak ASK, Hulse PA, Carrington BM, editors. MRI Manual of pelvic cancer. 2nd edn. London: CRC Press, 2011: 261.

15.

Horn

Fischer

Raptis

. Tumor size is of prognostic value in surgically treated FIGO stage II cervical cancer. Gynecol Oncol 2007; 107: 310–315.

16.

Kurman

Amin

. Protocol for the examination of specimens from patients with carcinomas of the cervix: A basis for checklists. Cancer Committee, College of American Pathologists. Arch Pathol Lab Med 1999; 123: 55–61.

17.

Wells

Ostor

Crum

Epithelial tumours and related lesions of the uterine cervix. In: Tavassoli

Devilee

(eds). World Health Organization Classification of Tumours. Pathology and Genetics of Tumours of the Breast and Female Genital Organs, Lyon: IARC Press, 2003, pp. 262–279.

18.

Sobin LH, Gospodarowicz MK, Wittekind C, editors. TNM-classification of malignant tumors. London: Wiley-Blackwell, 2009: 206–211.

19.

Le Bihan

Breton

Lallemand

. MR imaging of intravoxel incoherent motions: application to diffusion and perfusion in neurologic disorders. Radiology 1986; 161: 401–407.

20.

Yang

Lam

. Comparison of dynamic helical CT and dynamic MR imaging in the evaluation of pelvic lymph nodes in cervical carcinoma. Am J Roentgenol 2000; 175: 759–766.

21.

Liyanage

Roberts

Rockall

. MRI and PET scans for primary staging and detection of cervical cancer recurrence. Womens Health (London Engl) 2010; 6: 251–267.

22.

Reinhardt

Ehritt-Braun

Vogelgesang

. Meta-static lymph nodes in patients with cervical cancer: detection with MR imaging and FDG PET. Radiology 2001; 218: 776–782.

23.

LeBlanc

Gauthier

Querleu

. Accuracy of 18-Fluoro-2-deoxy-D-glucose positron emission tomography in the pretherapeutic detection of occult para-aortic node involvement in patients with a locally advanced cervical carcinoma. Ann Surg Oncol 2011; 18: 2302–2309.

24.

Signorelli

Guerra

Montanelli

. Preoperative staging of cervical cancer: is 18-FDG-PET/CT really effective in patients with early stage disease? Gynecol Oncol 2011; 123: 236–240.

25.

Downey

Riches

Morgan

. Relationship between imaging biomarkers of stage I cervical cancer and poor-prognosis histologic features: quantitative histogram analysis of diffusion-weighted MR images. Am J Roentgenol 2013; 200: 314–320.

26.

Liu

Sun

. Grading of uterine cervical cancer by using the ADC difference value and its correlation with microvascular density and vascular endothelial growth factor. Eur Radiol 2013; 23: 757–765.

27.

Kıvrak

Paksoy

Erol

. Comparison of apparent diffusion coefficient values among different MRI platforms: a multicenter phantom study. Diagn Interv Radiol 2013; 19: 433–437.