The feasibility of reduced field-of-view diffusion-weighted imaging in evaluating bladder invasion of uterine cervical cancer

Abstract

Objectives:

Uterine cervical cancer with bladder mucosal invasion is classified as FIGO stage IVA with poor prognosis. MRI can rule out the bladder invasion and skipping cystoscopy may be possible; however, high false-positive rate may be problematic. The purpose of this study is to evaluate the diagnostic performance of reduced field-of-view (FOV) diffusion-weighted imaging (DWI) in evaluating bladder mucosal invasion of cervical cancer.

Methods:

3T MRI including T2WI and reduced FOV DWI in 15 women with histologically proven cervical cancer (two stage IIIB, six stage IVA, seven stage IVB) were retrospectively evaluated compared with cystoscopic findings.

Results:

Cystoscopy revealed mucosal invasion in 13 of 15 cases. The border between the tumor and the bladder wall was unclear on T2WI and clear on reduced FOV DWI in all 15 cases. The diagnosis of mucosal invasion on reduced FOV DWI had a sensitivity of 100%, specificity of 50%, accuracy of 93%, PPV of 93%, and NPV of 100%.

Conclusions:

Addition of reduced FOV DWI may improve the staging accuracy of MRI for cervical cancer in assessing the bladder mucosal invasion.

Advances in knowledge:

Reduced FOV DWI may improve the staging accuracy of cervical cancer in assessing bladder mucosal invasion with high NPV and PPV, which may be helpful for avoiding unnecessary cystoscopy.

Introductory section

Uterine cervical cancer is the fourth most frequent cancer in females globally.¹ The management of cervical cancer depends on FIGO (International Federation of Gynecology and Obstetrics) staging, which is important for choice of appropriate treatment.² Cervical cancer with urinary bladder mucosal invasion classified as FIGO stage IVA represents about 2% of cases, and the prognosis is poor with an estimated 5-year overall survival of 20%.³ Cystoscopy is a diagnostic procedure in evaluating the bladder mucosal invasion of cervical cancer; however, it is relatively invasive examination with risk of complications.⁴ Magnetic resonance imaging (MRI) has a high diagnostic accuracy for ruling out the presence of bladder mucosal invasion, with very high negative predictive value (NPV) of 99.4–100%.^4–6 The absence of bladder mucosal invasion can be diagnosed with sufficient confidence, and skipping cystoscopy based on MRI is safe enough without concern for under-staging. On the other hand, the high rate of false-positive cases on MRI (positive predictive value: PPV of 7–37.5%) resulting from low thresholds for suspicion of bladder invasion to avoid under-staging may be problematic.^4–6

Because bladder muscle involvement alone is not associated with poor survival rate, and the prognosis for patients with evidence of muscle and/or serosal invasion of the bladder on MRI might not differ from that for patients without abnormality on MRI, the differentiation of bladder mucosal invasion from muscle/serosal invasion is important.⁷ Guidelines of the European Society of Urogenital Radiology (ESUR) for cervical cancer using MRI recommended diffusion-weighted imaging (DWI) matched to T2WI in terms of the acquisition plane to allow side-by-side-interpretation.⁸ That is because DWI improves the contrast between the tumor and the adjacent tissue which may facilitate assessment of tumor size and extent.⁸ DWI demonstrates cervical cancer as a high signal intensity mass with low apparent diffusion coefficient (ADC) value. However, relatively low spatial resolution and occasional susceptibility-induced distortion may cause misinterpretation of local extent of cervical cancer such as bladder invasion on conventional DWI.^9–11

Reduced phase direction field-of-view (FOV) technique by using spatially selective phase encoding gradient can offer high quality DWI with improved spatial resolution and less artifacts related to motion and susceptibility compared to conventional DWI with large FOV, and clinical usefulness of reduced FOV DWI in gynecologic tumors has been reported.^12–15 Reduced FOV DWI could enhance the diagnostic ability in the tumor delineation, and for the evaluation of tumor extent of cervical cancer.^13–15 The ESUR guideline mentioned reduced FOV DWI as optional, which is considered a potential alternative to conventional DWI.⁸ The purpose of this study is to evaluate the diagnostic performance of reduced FOV DWI in evaluating bladder mucosal invasion of cervical cancer compared with cystoscopic findings.

Methods and materials

Study population

The institutional review board in our hospital approved this retrospective study, and waived the requirement for written informed consent from patients. We cross-referenced the database of the Department of Obstetrics and Gynecology and Department of Radiology to identify all patients with histologically proven cervical cancer who had undergone both MRI including reduced FOV DWI and cystoscopy between March 2015 and March 2021. A total of 15 women with a mean age of 67 years (range, 40–92 years) were included in this study. The histopathological diagnoses were squamous cell carcinomas in 13 patients, and adenocarcinomas in 2 patients. The clinical stages of the patients according to the FIGO 2009/2018 were as follows: stage IIIB, 2 patients; stage IVA, 6 patients; stage IVB, 7 patients.

Imaging protocol

MRI was obtained by using 3T superconducting units (Discovery MR750, GE Healthcare, Waukesha, WI, USA) with 32 channel body-array torso coils. In all cases, T₂-weighted images (T2WI) and reduced FOV DWI were acquired in sagittal and oblique axial (short axis of the cervix) planes. Fast spin-echo T2WI was obtained in the sagittal plane with the following parameters: repetition time/echo time (TR/TE) 5000/92.4–104.8 msec; matrix, 352 × 256; FOV, 260–300 mm; section thickness/section gap, 3–6 mm/0.5 mm, number of acquired signals, two; and in the oblique axial plane which was perpendicular to the cervix as short axis plane with the following parameters: TR/TE, 5000/97–104.2 msec.; matrix, 352 × 256; FOV, 260–320 mm; section thickness/section gap, 2–3 mm/0.5–1 mm; number of acquired signals, two. Reduced FOV DWI (FOV optimized and constrained undistorted single shot: FOCUS) of the uterus was obtained with the following parameters: b = 50 and 800 sec/mm ²; number of signals acquired, ten; in the sagittal plane: TR/TE, 4000/55.1–55.4 msec; matrix, 160 × 80; FOV, 240–260 mm x 120–130 mm; section thickness/section gap, 5–6 mm/0 mm, and in the oblique axial plane: TR/TE, 4000/55.1–56.2 msec; matrix, 160 × 80; FOV, 200–260 mm x 100–130 mm; pixel size in x-y plane: 1.50 × 1.50 or 1.63 × 1.63 mm; section thickness/section gap, 5–6 mm/0 mm. A total of 16 slices of reduced FOV DWI were acquired in about 3 min. Conventional DWI of the pelvis was also obtained in the axial plane with the following parameters: TR/TE, 6000-6500/56.1–64 msec; matrix, 128 × 192; FOV, 400 mm; pixel size in x-y plane: 3.13 × 2.08 mm; section thickness/section gap, 5–8 mm/0–2 mm, number of acquired signals, four. The contrast-enhanced study was performed in 9 of 15 patients.

Image analysis

The border between the tumor and the bladder wall was visually evaluated on T2WI and reduced FOV DWI as “clear” or “unclear”. The presence of bladder mucosal invasion was visually evaluated on both sagittal, and oblique axial reduced FOV DWI with corresponding ADC map. Tumor margins were defined as areas with high signal intensity on the reduced FOV DWI and low signal intensity on the ADC map adjacent to the normal organs. The bladder mucosal invasion was regarded as present if all layers of the bladder wall were replaced by the tumor (probable), or nodular or irregular protrusion of the tumor into the lumen was observed (definite) with the consensus of two readers (radiologists with over 20 years of experience in body MRI) after careful individual evaluation. The two readers were blinded to the clinical stages and cystoscopic results but aware that each patient had cervical cancer. The concordance rate of the two readers was examined by κ statistic. The results were compared with the cystoscopic findings to determine the diagnostic accuracy of reduced FOV DWI together with the corresponding ADC map. The mean ADC values (×10⁻³mm²/s) of 15 cervical cancers were measured in a circular region of interest (ROI) from ADC maps on the workstation (Ziostation2, Ziosoft). The circular ROI was placed on the solid portion of the lesions so as not to contain necrotic areas as much as possible by referring all MR images. Contrast-enhanced MR imaging is optional for cervical cancer in the guideline and was performed in only nine cases in this study,⁸ so post-contrast imaging was excluded from the evaluation.

Results

Cystoscopy revealed mucosal invasion in 13 of the 15 cases. The border between the tumor and the bladder wall was unclear due to insufficient contrast on T2WI in all 15 cases, which made it difficult to assess the extent of invasion. Whereas the border was clear on reduced FOV DWI in all 15 cases. (Table 1) Median maximum tumor diameter measured on reduced FOV DWI was 70 mm (range 50–150 mm).

Table 1.

Case Summary

Case	Age (years)	Stage	Path.	Size (mm)	T2WI border	Reduced	FOV-DWI border	Cystoscopy invasion	ADC invasion (×10−3mm2/s)
1	75	IVB	SCC	95	unclear	clear	+ (probable)	+	0.67
2	57	IVB	SCC	100	unclear	clear	+	+	0.97
3	62	IVA	SCC	68	unclear	clear	+	+	0.85
4	71	IVB	SCC	60	unclear	clear	+ (probable)	+	0.98
5	92	IVB	SCC	66	unclear	clear	+	+	0.89
6	70	IVA	SCC	75	unclear	clear	+	+	1.01
7	58	IIIB	SCC	70	unclear	clear	+	-	0.87
8	81	IVB	SCC	60	unclear	clear	+	+	0.78
9	81	IVA	SCC	62	unclear	clear	+	+	0.99
10	57	IVA	SCC	100	unclear	clear	+	+	0.79
11	40	IVA	SCC	77	unclear	clear	+	+	1.12
12	60	IVB	AC	70	unclear	clear	+	+	0.72
13	51	IVB	SCC	150	unclear	clear	+	+	0.87
14	74	IIIB	AC	50	unclear	clear	-	-	1.06
15	78	IVA	SCC	87	unclear	clear	+	+	0.98

AC, Adenocarcinoma; Path., Pathologic diagnosis; SCC, Squamous cell carcinoma.

All 13 cases with mucosal invasion on cystoscopy were positive for mucosal invasion (2 probable and 11 definite) on reduced FOV DWI (Figure 1). In the other 2 cases without mucosal invasion on cystoscopy, there were one positive (definite) and one negative cases on reduced FOV DWI (Figures 2 and 3). The diagnosis of bladder mucosal invasion based on reduced FOV DWI had a sensitivity of 100%, specificity of 50%, accuracy of 93%, PPV of 93%, and NPV of 100%. Concordance rate between the two readers was 1.0 for the evaluation of bladder mucosal invasion on reduced FOV DWI.

Figure 1.

A 75-year-old female with cervical cancer (stage IVB, squamous cell carcinoma). (a) Sagittal fast spin-echo T2WI shows a huge cervical mass. The border between the tumor and the bladder wall was unclear (arrow). (b) Sagittal reduced FOV DWI shows the tumor as a high signal intensity mass with clear margin. Involvement of all layers of the posterior bladder wall without nodular or irregular protrusion determined as “probable mucosal invasion” (arrow) was observed. (c) Sagittal ADC map shows the tumor as a low signal intensity mass with clear margin. Involvement of all layers of the posterior bladder wall (arrow) was demonstrated. (d) Cystoscopy revealed bladder mucosal invasion.

Figure 2.

A 58-year-old female with cervical cancer (stage IIIB, squamous cell carcinoma). (a) Sagittal fast spin-echo T2WI shows a large cervical mass. The border between the tumor and the bladder wall was unclear, and nodular protrusion (arrow) was observed. High signal intensity bullous edema was also demonstrated (arrowheads). (b) Sagittal reduced FOV DWI shows the tumor as a high signal intensity mass with clear margin. Involvement of all layers of the posterior bladder wall with nodular protrusion (arrow) determined as “definite mucosal invasion” was observed. (c) Sagittal ADC map shows the tumor as a low signal intensity mass with clear margin. Involvement of all layers of the posterior bladder wall with nodular protrusion (arrow) was demonstrated. (d) Cystoscopy revealed no bladder mucosal invasion, and bullous edema was observed.

Figure 3.

A 74-year-old female with cervical cancer (stage IIIB, adenocarcinoma). (a) Sagittal fast spin-echo T2WI shows a cervical mass. The border between the tumor and the bladder wall was unclear (arrow). (b) Sagittal reduced FOV DWI shows the tumor as a high signal intensity mass with clear margin. No bladder wall involvement (arrow) was observed. (c) Sagittal ADC map shows the tumor as a low signal intensity mass with relatively clear margin. No bladder wall involvement (arrow) was observed. (d) Cystoscopy revealed no bladder mucosal invasion.

Discussion

According to the recent guidelines for the staging of uterine cervical cancer with MRI, T2WI is essential; however, tumors sometimes are not well delineated on T2WI and the use of DWI is also recommended.^7,8 In our study, because T2WI showed insufficient contrast between the bladder wall and the tumor in all cases, it was difficult to assess the degree of bladder invasion in detail. In the previous reports, loss of fatty layer was used as one of the criteria for positive bladder invasion.^4–6 Although this resulted in a high negative predictive value (99.4–100%), there were many false positive cases and PPV was 7–37.5%.^4–6 Rockwall et al⁴ reviewed the MR images including T1WI and T2WI of 112 patients with cervical cancer by using a 5-point invasion score to determine bladder invasion (1 = no invasion, 5 = definite invasion). A confidence score of 3 (tumor abuts and/or indents the bladder over a significant area: probable serosal involvement) or above was used to identify patients with possible bladder involvement. MRI was suspicious for bladder invasion in 14, and one had invasion confirmed at cystoscopy. The diagnosis of bladder involvement on MRI had a sensitivity of 100%, specificity of 88%, NPV of 100% and PPV of 7%. Kim et al⁵ reviewed the MR images including T1WI, T2WI, and post-contrast T1WI of 257 patients with cervical cancer by using a 3-point invasion score to determine bladder mucosal invasion (no, suggestive, and definite). A total of 16 patients had definite or suggestive invasion on MRI, and six patients had actual bladder invasion confirmed by cystoscopy. The diagnosis of bladder involvement on MRI had a sensitivity of 100%, specificity of 96%, NPV of 100% and PPV of 37.5%. Jeong et al⁶ reviewed the MR images including T1WI, T2WI, and post-contrast T1WI of 562 patients with cervical cancer by using the criteria: MRI findings of wall irregularity with heterogeneous signal, enhancement with thickening and nodularity, loss of fat plane or mass protruding into the bladder lumen were interpreted as positive invasion. 50 patients demonstrated bladder invasion on MRI included 12 true positive and 38 false negative cases, whereas 512 patients demonstrated no bladder invasion on MRI included three false-positive and 509 true-negative cases. The diagnosis of bladder involvement on MRI had a sensitivity of 88.0%, specificity of 93.1%, NPV of 99.4% and PPV of 35.6%. These results with low PPVs may suggest that unnecessary cystoscopy could be avoided by reducing the number of false-positive cases with more accurate MR examination. Omitting the cystoscopy may reduce the likelihood of possible complications, save costs, and shorten the time to start treatment.^4–6

Although DWI is recommended in the ESUR guideline for cervical cancer,⁸ evaluation of bladder invasion by DWI has not been reported. In the current study, we used reduced FOV DWI, which is considered to have higher spatial resolution with less artifacts, instead of conventional DWI.^12–15 Hwang et al¹³ reported that reduced FOV DWI of cervical cancer showed significantly better image quality in terms of anatomic detail and lesion conspicuity with fewer artifacts compared to full FOV DWI (p < 0.05). Chen et al¹⁵ also reported that reduced FOV DWI of cervical cancer provided significantly better image quality and lesion conspicuity than full FOV DWI (p < 0.001). Takeuchi et al¹⁴ reported that reduced FOV DWI could define the tumor margins and demonstrate local extension of cervical cancer significantly more accurate than T2WI (p < 0.001). In the current study, the bladder mucosal invasion was clearly demonstrated on reduced FOV DWI with high NPV (100%) and PPV (93%). There was no false-negative case and only one false-positive case. The false-positive case was determined as definite mucosal invasion on reduced FOV DWI, but cystoscopy showed bullous edema suggestive of muscular invasion and no mucosal invasion. The cause of the false-positive result may have been overestimation of submucosal invasion due to the limitation of spatial resolution.

Since the retrospective nature and relatively small population are limitations to the current study, further prospective studies in larger populations to support the results are needed. In addition, only cases in which cystoscopy was ordered by gynecologists were included in this study, which may have been a bias. We could not compare reduced FOV DWI with conventional DWI because of the different imaging directions, and that was another limitation.

Conclusion

Addition of reduced FOV DWI may improve the staging accuracy of cervical cancer in assessing bladder mucosal invasion and may be helpful for avoiding unnecessary cystoscopy.