Inclusion of patients
This was a retrospective,
institutional review board approved study.
Patients were included by performing a search in our single-institution radiology database looking for patients with a recent diagnosis of rectal carcinoma (any T and N stage) scheduled for surgery (i.e.
anterior resection or total mesorectal excision – TME –),
with or without neoadjuvant therapy,
between January 2012 and December 2013.
In this time interval,
we included all patients who both underwent thin-section rectal MRI for local tumor staging and contrast-enhanced CT for distant metastases staging before surgery.
Surgical exploration was used as reference standard to assess the exact position of the inferior margin of tumors with respect to the APR.
Exclusion criteria were represented by general contraindications to MRI (e.g.
ferromagnetic metal fragments or implanted medical devices).
Renal insufficiency (GFR <30ml/min) was considered an exclusion criteria for CT but not for rectal MRI,
since the latter is commonly performed without intravenous injection of paramagnetic contrast media.
MRI and CT protocols
High-spatial-resolution pelvic MRI examinations were performed with a 1.5T MRI scanner (Signa™ HDxt,
WI) and a pelvic phased-array coil,
according to a standardized protocol [4,
The following parameters were applied for two-dimensional T2-weighted fast-recovery fast spin echo (FSE) sequences: repetition time (TR)/ echo time (TE) 4000–6000/80–160,
echo-train length 24–49,
flip angle 90°,
bandwidth 50 kHz,
field-of-view (FOV) 18–36 cm,
number of excitations 4–6.
FSE T2-weighted sequences (slice thickness 4 mm,
interslice gap 0.4 mm) with a large field of view were first oriented in the three orthogonal planes (sagittal,
without fat saturation,
enabling identification of the primary tumor.
High-resolution oblique axial and coronal scans (slice thickness 3-3.5 mm,
interslice gap 0.3 mm) were further oriented perpendicular and parallel to the long axis of the tumor,
in order to avoid misinterpretation due to partial volume effects.
In the case of low rectal cancers,
high-spatial-resolution coronal images were performed to optimally show the levator muscles,
the sphincter complex,
as well as the intersphincteric plane.
A Diffusion-Weighted Imaging (DWI) sequence (b 1000 s/m2,
bandwidth 250 kHz,
slice thickness 4.0 mm,
interslice gap 0.4 mm,
number of excitations 6) was acquired in the axial plane for aiding the identification of the inferior edge of tumors.
Patients did not receive bowel preparation or spasmolytic agents.
Intravenous paramagnetic contrast medium was not administered.
CT examinations were performed with a 64-slice CT scanner (LightSpeed VCT,
Contrast-enhancement was produced by i.v.
injection of iodinated contrast medium with an iodine concentration ranging between 350 and 370mg/mL (iobitridol,
The iodine load was 1.5 mg per kg of body weight.
The flow rate was set at 3.2–3mL/s with an automatic injector and CT acquisition was started in the portal phase 45s after contrast media bolus detection in the upper abdominal aorta,
using a bolus-tracking software.
In some cases a CT with water enema (CT-WE) was performed,
using a previously described technique .
Large bowel cleansing was obtained with a low fiber diet for 3 days before the CT-WE,
and oral administration of 2000 mL of an isotonic non-absorbable electrolyte solution containing poly- ethylene glycol (Isocolan,
Italy) was given the afternoon before examination.
At the time of examination,
large bowel lumen was distended by the trans-rectal introduction of 1500–2000mL of tap water with the patient placed on the CT table.
To reduce abdominal discomfort of patients and to avoid motion artifacts during the CT acquisition,
bowel hypotonia was obtained by the intravenous injection of 2 mL hyoscine-N-butylbromide 20 mg/mL (Buscopan,
CT images were acquired with the patient in supine position.
Bowel wall enhancement was produced with the same technique described for conventional CT examinations.
All MRI and CT studies were reviewed in consensus on a dedicated workstation (ADW 4.6,
GE Medical Systems,
WI) by two abdominal radiologists (MR and FP) with 5 and 10 years of experience in abdominal imaging,
blinded to the location of the masses.
They had to define the extra- or intra-peritoneal location of tumor’s inferior edge with respect to the APR.
This anatomical structure was defined according to Brown et al.
On MR midsagittal images the APR was identified as a thin hypointense linear structure coursing along the superior aspect of bladder (men) or uterus (women) to reach its attachment onto the anterior rectal wall (Figure 1B),
as previously described [4,
On FSE T2-weighted axial images,
the peritoneum attaches in a V-shaped manner onto the anterior rectal wall,
an appearance that has been defined as “seagull” sign .
DWI axial sequences were helpful to confirm the level of the inferior edge of tumors .
On CT images the APR was identified as a thin slightly hyperdense linear structure,
surrounded by hypodense fat tissue,
coursing as previously described on MR images (Figure 1A).
Curved planar reformations were helpful to identify the correct location of the APR (Figure 2A).
The two readers defined the quality of identification of the APR according to a 4-point confidence scale: 0,
not visible; 1,
The distance from the inferior edge of tumors to the anal verge was measured in millimeters by means of an electronic digital caliper both on MRI and the CT images (Figure 2B).
This measurement required two or more interconnecting straight lines for an approximate total length on MR images and the use of multi-planar reconstructed images on CT examinations.
The distance between the anal verge and the APR was registered too (Figure 2B).
Standard of reference
We used surgical exploration as reference standard to assess the exact position of the inferior margin of tumors with respect to the APR: as previously described [4,
the limit between intra- and extra-peritoneal rectum has been defined intra-operatively as the site where the peritoneal serosa leaves the lateral walls of rectum merging inferiorly toward the midline to cover the anterior surface (i.e.
Categorical variables were expressed as number and percentage,
while continuous variables as mean and standard deviation.
The normal distribution of MRI and CT measurements was assessed by means of the D'Agostino-Pearson test.
Mann-Whitney test was used to assess the presence of a significant difference between MRI and CT measurements,
while their degree of correlation was assessed by the Spearman's rank test.
The diagnostic performance of MRI and CT in determining the extra- vs intra-peritoneal location of rectal cancers was calculated using tables with positive cases corresponding to the extra-peritoneal location and negative cases to the intra-peritoneal location.
as well as positive and negative predictive values were calculated.