Abstract:
Objective:To investigate the predictive value of diffusion-weighted (DW) magnetic resonance imaging (MRI) for invasiveness of hilar cholangiocarcinoma (HC).
Methods:The retrospective case-control study was conducted. The clinicopathological data of 65 HC patients who were admitted to the Sun Yat-sen Memorial Hospital from January 2012 to November 2017 were collected. Patients received DW MRI before treatment, and 2 senior imaging doctors analyzed imaging data and measured the apparent diffusion coefficient (ADC) for the primary lesions of HC. Observation indicators: (1) MRI situations of HC; (2) relationship between ADC and clinicopathological factors; (3) receiver operator characteristic (ROC) curve analysis; (4) treatment and follow-up situations. According to patients′ conditions, treatment plans were done within 2 weeks after MRI and patients underwent radical resection of HC. Follow-up using telephone interview was performed to detect tumor recurrence up to December 2017. Measurement data with normal distribution were represented as
±s, and comparisons between group and among group were respectively analyzed using the t test and one-way ANOVA. Spearman′s rank correlation was performed to analyze the relationship between ADC and clinicopathological factors. ROC curves assessed the diagnostic efficiency of ADC.
Results:(1) MRI situations of HC: MRI and magnetic resonanced cholangio-pancreatography (MRCP) in 65 patients showed varying degrees of soft rattan-like dilations of intrahepatic bile ducts and truncation signs of bile tracts in hepatic port. Of 65 patients, tumors in 23, 7 and 35 patients were respectively pedunculated type, polypoid type and infiltrating type. The pedunculated-type lesions of 23 patients presented as low signal on T1WI and slightly high signal on T2WI; after enhanced scans of MRI, pedunculated-type lesions of 7 patients demonstrated moderate homogenous enhancement in 3 patients, ring-like enhancement with internal liquefaction necrosis in 10 patients and moderate heterogeneous enhancement in 10 patients, respectively. The polypoid-type lesions presented as low signal on T1WI and high signal on T2WI, and moderate homogenous enhancement by enhanced scans of MRI. There were varying degrees of bile duct wall thickness and irregular nodules in the infiltrating-type lesions of 35 patients, showing moderate enhancement by enhanced scans of MRI. All the lesions of 65 patients using DW MRI demonstrated restricted diffusion, showing a clear boundary between lesions and normal surrounding bile ducts or liver tissues; heterogeneous enhancement lesions by MRI scans presented as heterogeneously high signal on DWI and heterogeneously low signal on ADC map, and necrotic area of lesions showed low signal on DWI; homogenous enhancement by MRI scans presented as homogenously high signal on DWI and homogenously low signal on ADC map. (2) Relationship between ADC and clinicopathological factors: ADC was respectively (1.382±0.165)×10
-3mm
2/s, (1.343±0.138)×10
-3mm
2/s, (1.291±0.226)×10-3 mm2/s, (1.111±0.243)×10
-3mm
2/s in stage Ⅰ, Ⅱ, Ⅲ and Ⅳ (TNM staging) and (1.441±0.355)×10
-3mm
2/s, (1.226±0.177)×10
-3mm
2/s, (1.061±0.228)×10
-3mm
2/s in high-differentiated, moderate-differentiated and low-differentiated tumors (pathological grading) and (1.403±0.176)×10
-3mm
2/s, (1.121±0.238)×10
-3mm
2/s in Ki-67 score ≤10% and >10% and (1.115±0.241)×10
-3mm
2/s, (1.347±0.174)×10
-3mm
2/s in HC patients with and without lymph node metastasis, with statistically significant differences in the above indicators (F=4.158, 9.866, t=11.607, 13.464, P<0.05). Results of Spearman′s rank correlation analysis showed that ADC had a negative correlation with TNM staging, pathological grading and Ki-67 score (r=-0.532, -0.522, -0.409, P<0.05). (3) ROC curve analysis: using 1.225×10
-3mm
2/s as a critical value of ADC, the sensitivity and specificity of ADC in the diagnosis of stage Ⅰ-Ⅱ HC and stage Ⅲ-Ⅳ HC were 70.5% and 81.0%, and area under ROC curve was 0.705 (95%CI: 0.62-0.84, P<0.05). Using 1.100×10
-3mm
2/s as a critical value of ADC, the sensitivity and specificity of ADC in the diagnosis of low-differentiated HC and moderate- and high-differentiated HC were 88.2% and 64.3%, and area under ROC curve was 0.814 ［95% confidence interval (CI): 0.69-0.90, P<0.05］. Using 1.243×10
-3mm
2/s as a critical value of ADC, the sensitivity and specificity of ADC in the diagnosis of Ki-67 score ≤ 10% and >10% were 66.7% and 75.0%, and area under ROC curve was 0.783 (95%CI: 0.62-0.90, P<0.05). Using 1.222×10
-3mm
2/s as a critical value of ADC, the sensitivity and specificity of ADC in the diagnosis of lymph node metastasis were 91.3% and 71.4%, and area under ROC curve was 0.873 (95%CI: 0.76-0.94, P<0.05). (4) Treatment and follow-up situations: 65 patients underwent successful radical resection of HC. Thirty-three patients were followed up for 1-24 months. Of 33 patients, 5 had tumor recurrence within 6 months postoperatively, including 4 with ADC < 1.100×10
-3mm
2/s, 13 had tumor recurrence after 6 months postoperatively, and 15 didn′t have tumor recurrence or metastasis, including 1 with ADC <1.100×10
-3mm
2/s.
Conclusions:There are different ADC in different TNM staging, pathological grading, Ki-67 score and with or without lymph node metastasis of HC. ADC of DW MRI can be used as a preoperative imaging predictor for invasiveness of HC.
下载: