Patients

Between January 2001 and December 2007, a total of 346 patients underwent MRI because of biopsy-proven IB1 cervical cancer. Of these patients, 86 patients had cervical cancer (MR-invisible IB1 cancer) that was invisible on MRI, but 260 patients had cervical cancer (MR-visible IB1 cancer) that was visible on MRI. The medical records of 86 patients (age range, 31–74 years; mean, 48.3 ± 11.5 years) with MR-invisible cancer and 260 patients (age range, 23–74 years; mean, 48.9 ± 10.7 years) with MR-visible cancer were reviewed. These cervical cancers were diagnosed by colposcopic biopsy in 80% (276/346 patients) and by conization in 20% (70/346 patients).

All patients underwent bimanual pelvic examination and rectovaginal examination to identify the disease extent. Routine laboratory testing, chest radiography, cystoscopy, and sigmoidoscopy were performed to determine the clinical FIGO staging. Abdominopelvic MRI was conducted before the hysterectomy and was interpreted by 2 radiologists who had experience of gynecologic imaging of >8 years. The time interval between preoperative MRI and hysterectomy ranged from 1.0–34.0 days (mean, 12.7 ± 8.2 days) for MR-invisible IB1 cancers and from 1.0 to 75.0 days (mean, 12.8 ± 9.0 days) for MR-visible IB1 cancers.

All patients were treated with radical hysterectomy, pelvic lymphadenectomy, and additional surgical procedures according to the clinical stage and the surgeons’ decision. Pelvic lymph nodes that were suspicious for metastasis were sent for frozen sectioning during the operation. When a pelvic lymph node was intraoperatively positive for cancer on the frozen section, paraaortic lymph nodes were dissected. Radical hysterectomy and dissection of pelvis without or with paraaortic lymph nodes were performed in both of MR-invisible and -visible IB1 cervical cancer. We did not abandon radical hysterectomy and lymph node dissection, even though pelvic lymph nodes were clinically positive.

The size of the tumor was measured in the hysterectomy specimen, and the size of tumor in the biopsy specimen was not incorporated. The pathologist recorded the size of residual cervical cancer, histologic type and depth of stromal invasion, lymphovascular invasion, parametrial invasion, vaginal invasion, resection tumor margin, and lymph node metastasis.

Adjuvant radiation therapy and concurrent chemoradiation therapy was added to 3.5% (3/86) and 1.1% (1/86) in patients with MR-invisible IB1 cancer and to 27.3% (71/260) and 20.4% (53/260) in patients with MR-visible IB1 cancer. Radiation therapy was performed if there were positive for ≥2 of the following criteria: lymph vascular space invasion, one-third to one-half or more depth of invasion, and ≥4 cm tumor. Concurrent chemoradiation therapy was performed if the cancers were positive for one or more of the following criteria: pelvic lymph node metastasis, parametrial invasion, and positive resection margin. Follow-up computed tomography or MRI was performed once or twice every year after the surgery, and the follow-up period of all cancers ranged from 0.2–132.4 months (median, 53.8 ± 27.0 months). Follow-up period of MR-invisible and -visible IB1 cancers was 0.4–132.4 months (median, 56.9 ± 31.3 months), and 0.2–111.6 months (median, 52.8 ± 25.4 months), respectively.

MRI

All patients underwent preoperative MRI before hysterectomy. MR examinations were performed with 1 of 2 scanners: 1.5T (Signa, GE Medical System, Milwaukee, WI) or 3.0T MR scanners (Intera Achiva 3T; Philips Medical System, Best, The Netherlands). In MRI, the pelvis was scanned first and then the upper abdomen.

Our MR examination of the pelvis was composed of T2-weighted images, T1-weighted images, and dynamic contrast-enhanced images. The T2-weighted images were obtained with the use of a fast or turbo spin echo sequence at the axial, sagittal, and coronal planes. The following were the parameters of the T2-weighted images at these 1 MR units (1.5T and 3.0T): repetition time/echo time, 3300-4575/87-99 and 4177-4292/90 msec; echo train length, 10-12 and 22; slice thickness, 5 and 4 mm; slice gap, 2 and 0.4 mm; matrix size, 512 × 256 and 800 × 698; field of view, 240 and 360 mm; number of excitations, 2 and 3; and scan time, 159-223 and 201-206 seconds, respectively. T1-weighted images were obtained with a fast or turbo spin echo sequence at the axial plane with the following parameters: repetition time/echo time, 467/12 and 437/10 msec; echo train length, 2 and 4; slice thickness, 5 and 5 mm; slice gap, 2 and 2 mm; matrix size, 320 × 192 and 352 × 352; field of view, 240 and 240 mm; number of excitations, 2 and 2; and scan time, 183 and 201 seconds, respectively. Dynamic contrast-enhanced images were obtained with a fast spoiled gradient recalled echo or a 3-dimensional fast field echo sequence at the sagittal or axial plane, with the following parameters: repetition time/echo time, 60/1.4 and 8/4.1 msec; flip angle, 80 and 25 degrees; slice thickness, 5 and 7 mm; slice gap, 2 and 0 mm; matrix size, 256 × 160 and 208 × 166; field of view, 240 and 200 mm; number of excitations, 2 and 1; and scan time, 190 and 200 seconds, respectively.

The MR scan range of the upper abdomen was between the lower lung and aortic bifurcation. T2-weighted axial images at 1.5T and 3.0T were obtained with fast or turbo spin echo sequences with the following parameters: repetition time/echo time, 2400/102 and 2430/80 msec; echo train length, 18 and 24; slice thickness, 8 and 8 mm; slice gap, 2 and 2 mm; matrix size, 256 × 192 and 256 × 180; field of view, 340 and 380 mm; number of excitations, 1 and 1; and scan time, 57 (19 × 3) and 45 sec (15 × 3), respectively. T1-weighted axial images were also obtained with fast spoiled or fast field gradient echo sequences with the following parameters: repetition time/echo time, 140/4.2 and 190/2.3 msec; flip angle, 70 and 75 degrees; slice thickness, 8 and 8 mm; slice gap, 2 and 2 mm; matrix size, 256 × 128 and 224 × 180; field of view, 340 × 255 and 380 × 300 mm; number of excitations, 1 and 1; and scan time, 28 (14 × 2) and 72 sec (18 × 4), respectively.

Data analysis

MR-invisible cancer was defined as a cervical cancer that was not seen on either the T2-weighted images or contrast-enhanced T1-weighted images ( Figure 1 ). MR-visible cancer was defined as a cervical cancer that was slightly hyperintense on T2-weighted images and the lesion was poorly enhanced on contrast-enhanced T1-weighted images as compared with adjacent normal cervical tissue ( Figure 2 ). Postbiopsy inflammation was considered to be present when a cervical lesion was hyperintense on T2-weighted images, but the lesion enhancement was equal to or higher than normal tissue enhancement on contrast-enhanced MR images.

MR images in a 60-year-old woman with an MR-invisible IB1 cervical cancer

A, T2-weighted and B, contrast-enhanced sagittal images show no residual cancer but a tissue defect ( arrow ) after conization.

MR , magnetic resonance.

Park. Postoperative MR-invisible IB1 cervical cancer. Am J Obstet Gynecol 2014 .

MR images in a 60-year-old woman with an MR-visible IB1 cervical cancer

A, T2-weighted and B, contrast-enhanced sagittal images show a small cervical cancer ( arrow ) after colposcopic biopsy.

MR , magnetic resonance.

Park. Postoperative MR-invisible IB1 cervical cancer. Am J Obstet Gynecol 2014 .

MR-invisible and MR-visible cancers were compared in terms of age, biopsy type (conization or colposcopic biopsy), histologic type, and squamous cell carcinoma (SCC) antigen. These cancers were also compared with respect to postoperative tumor size, depth of stromal invasion, lymphovascular invasion, parametrial invasion, vaginal invasion, and lymph node metastasis. MR-invisible and -visible cancers were correlated with the hysterectomy specimen to assess how many cases had histologically no residual tumor in the cervix.

Recurrent tumor was recorded on the follow-up computed tomography (CT) or MRI. The recurrence-free 5-year survival rate was calculated. MR-invisible and -visible cancers were compared in terms of recurrent tumor and recurrence-free or overall 5-year survival rate.

Statistical analysis

Mann-Whitney test was used to compare age, size of tumor, SCC antigen, and depth of stromal invasion because of non-Gaussian distribution.

Fisher exact test was used to compare types of biopsy, types of cancer histology, postoperative residual tumor, lymphovascular invasion, parametrial invasion, vaginal invasion, lymph node metastasis, and recurrent tumor.

Odds ratios and 95% confidence intervals were also obtained with the approximation of Woolf. When at least 1 value was zero, 0.5 was added to each value to make the calculation possible.

Kaplan-Meier survival curves were plotted to compare recurrence-free and overall 5-year survival rates.

Statistical analyses were performed with SPSS for Windows (version 20.0; SPSS Inc, Chicago, IL). A probability value of < .05 was considered statistically significant and was shown up to the thousandths, which were rounded up from the ten-thousandths.