Patients
The local ethical institutional review board approved the study. Informed consent was waived for this retrospective study. A total of 320 consecutive patients diagnosed with rectal cancer and then treated in our hospital between January 2019 and June 2019 were enrolled in this retrospective study. Selection criteria included the following: ①a biopsy-confirmed primary rectal carcinoma, ②treatment by surgical resection, ③a preoperative rectal MRI with good image quality, ④operation within two weeks after MRI. Exclusion criteria included the following: ①patients with MRI images with motion artifacts or poor image quality (n = 30), ②patients who received radiotherapy, neoadjuvant treatment or/and palliative treatment (n = 100), ③an interval between MRI and surgery higher than 2 weeks (n = 20), ④patients with the previous history of other pelvic surgery (n = 60).
The collected clinical and imaging data included the following: patient age, gender, Body Mass Index (BMI), T stage, the tumor location with regard to APR (MRI and intraoperative findings), the degree of bladder filling, the orientation of the uterus, pelvic effusion, and the distance from seminal vesicle/uterus to the rectum, the height of tumor (MRI and colonoscopy), and the distance of AV-APR.
MRI-Sequence Acquisition
MRI was performed on a 3.0 Tesla (T) MRI scanner (MAGNETOM Skyra, Siemens Healthcare, Erlangen, Germany) using a phased-array body coil while patients were placed in a supine position. Before scanning, intestinal cleaning was performed by enema administration with 20 ml of glycerin. MRI scan sequences included: ①high-resolution oblique axial T2-weighted image (T2WI) without fat saturation, ②sagittal T2WI without fat saturation, ③axial T1-weighted image (T1WI), ④axial diffusion weighted images (b = 0, 1000 s/mm2), and ⑤gadolinium contrast-enhanced T1WI with fat saturation (axial, sagittal and coronal planes). The scanning parameters of sagittal T2WI were as follows: repetition time/echo time [TR/TE]: 5000/106 ms, field of view [FOV]: 23 cm, section thickness: 5 mm, number of slices: 23 slices, voxel: 0.7*0.7*5.0 mm, total time: 157s.
Radiologist and Colorectal Surgeons Revaluation Strategy and Anatomic Measurements
Radiologists: MR images were independently reviewed by two gastrointestinal radiologists (** and **) with more than 5 years' working experience in rectal MRI. The two readers were blinded to any pathological results of the patients and achieved a unified standard to evaluate the MRI features through discussion before the independent analysis. If there were discrepancies between the two readers on qualitative parameters, a final decision was reached by a third reader (** with 11 years of experience in imaging diagnosis). We divided APR into two categories: definitely visible (the thin hypointense linear structure attached to the anterior rectal wall is definitely visible) and probably visible (the thin hypointense linear structure is probably visible or not definitely visible) (Fig. 2A, 2B) [19]. For all the patients rated with a definitely visible APR, the distances of AV-APR were measured in the sagittal T2W images as a line from the APR to the anal verge along the direction of the rectum (Fig. 3A) [4]. For all the patients, the height of tumor was measured as the distance from the inferior tumor margin to the anal verge (Fig. 3B). The distances from seminal vesicle/uterus to rectum were measured as a line from the junction of APR and seminal vesicle/uterus to the junction of APR and rectum along the direction of APR (Fig. 3C). The tumor location with regard to APR (MRI) was assigned to the following categories: ①above the APR(the distal end of the tumor reaches above the height of the APR), ②straddle the APR(the distal end of the tumor reaches below the height of the APR and the proximal end of the tumor reaches above the height of the APR), ③below the APR(the proximal end of the tumor reaches below the height of the APR). According to the angle between the axis of the uterus and the axis of the axial plane on the sagittal T2WI, uterine orientation was categorized as follows: anteversion, perpendicular, or retroversion. A distended bladder was defined as the bladder wall showing no folds on both sagittal and axial T2WI [4], which means that the bladder was filling.
Colorectal Surgeons: The two surgeons were blinded to any MRI results of the patients. For all the patients, the height of tumor was measured by colonoscopy. And the tumor location with regard to APR (intraoperative findings) was also assigned to the following categories: ①above the APR, ②straddle the APR, ③below the APR. If there were discrepancies between the two reviewers, a final decision was reached by consensus..
Statistical Analysis
For continuous variables, the normality test was carried out by One-Sample Kolmogorov-Smirnov test. Data conforming to normal distribution were reported as mean ± standard deviation, and data not conforming to normal distribution were reported as median and quartile. The Chi-square test or Fisher’s exact test was carried out for assessing categorical variables, as appropriate. The Independent-Samples T test was carried out for assessing continuous variables. A univariate logistic regression analysis was performed to identify the independent factors associated with the visualization of the APR on MRI. A binary logistic regression analysis was used to establish a combined model and calculate the odds ratio (OR) and 95% confidence interval (CI) between these potential influencing factors and the APR. Moreover, the nomogram analysis and the receiver operating characteristic curve (ROC curve) were performed. Intraclass correlation coefficients (ICC) were used to evaluate the differences in the distance of AV-APR between the two radiologists. The Pearson correlation coefficient was used to characterize the agreement between pairs of measurements of the tumor height by colonoscopy and MRI. The consistency check of a diagnostic test (Kappa statistics) was used to evaluate the value of MRI in the diagnosis of the tumor location with regard to the APR. Statistical analyses were performed using SPSS version 19.0 for windows (SPSS Inc., Chicago, Illinois, USA) and R software (version 3.4.3). All P values < 0.05 were considered statistically significant.