| Objective To evaluate the accuracy of multi-slice spiral computed tomography (SCT) imaging in the preoperative assessment of the depth of extramural tumor infiltration and prediction of the completeness of the circumferential resection margin (CRM) and to predict the imaging stage before the operation. Materials and Methods In a prospective study of 39 consecutive patients, All patients were confirmed malignant tumor by fibro-endoscopy and biopsy. Before CT scanning, the necessity and contradictions were told and consents were formally obtained. Preoperative CT was performed 2 to 4 day before the operation carried out. Tumor size ranged from 2 to 12cm (mean 6.8cm). A GE LightSpeed 8MS spiral CT was used. Preparation for the CT scanning required the patient to fast at least 12 hours before the study. Bowel clearance was desperately important by oral 33%MgSO2 solution 100ml mixed with 1000ml water the day before scanning. No oral contrast media was needed. To avoid blurring from descending intestines and the sigmoid, the patient was examined with the bladder distended following the intake of 1000ml water 1 hour prior to the CT scanning. Immediately before scanning, 1000-1500ml air was inflated into the rectum through a 10mm rectal tube. With the patient in the prone and supine position alternately, contiguous slices were obtained at 5mm intervals beginning at the mid part of the sacroiliac joints and ending at the level of the anal canal (ischial tuberosity). A same rapid sequence CT with 2.5mm slices was then performed 40-50sec after bolus injection of 100ml Iopromide 300mg/ml. Data were transformed to the workstation on which observation, analysis and multiple planner reconstruction were performed. 2-4 days after the CT examination, all patients were underwent surgery. Ninety percent of paitents (35/39) were performed by the same surgeon according to the TME principle, as described by Heald. The main principles of this technique involve sharp dissection within the "holy plane" under direct vision in order to remove an intact mesorectal envelope with tumor-free margins and without causing injury to the pelvic autonomic nerves. Curative resection implied a resection with no residual macroscopic tumor, free margins including perirectal mesorectal margins on routine microscopic examination and no evidence of distant metastasis. The resected specimen was fixed in 10% formalin for 48 hours. Dissection was then performed transversely to provide multiple serial 10-15mm sections through the whole tumor and the surrounding mesorectum. Then sections were all wholly embedded step by step in celloidin of different concentration(4%~32%) after dehydrated and degreased sequently, and then cut at 15-20um on a AO sledge microtome, and then routinely processed for haematoxylin and eosin (H-E) staining for subsequent reporting. The sections should keep the tumor tissue and the surrounding mesorectum in intact. The sections were carefully examined microscopically for circumferential resection margininvolvement. If none was found, the extent of extramural spread was measured from the muscularis propria to the outmost part of the tumor. The distance from the outmost part of the tumor or tumor deposit to the resection margin, the pattern of invasion whether pushing or infiltrating, the grade of the tumor, and TNM stage were also documented. The tumor stage according to the TNM classification system and the measured depth of extramural tumor invasion and completeness of circumferential resection margin (CRM) in matched SCT images were compared with histopathologic slices. Results Satisfactory images were obtained in all 39 patients by means of optimized scan techniques. Preoperative SCT correctly indicated the histopathologic tumor stage in 33 of 39 patients in whom comparisons were possible with a diagnostic accuracy of 84.6%. The satisfactory results were probably due to spiral CT with standardized method such as carefully bowel clearance, adequate position and intramural contrast media, techniques that not all conventional studi...
|
PDF Full Text Request