MRI potentially changes liver resection planning



Gadoxetic acid-enhanced MRI (EOB-MRI) is superior to contrast-enhanced CT (CECT) for the preoperative detection of small colorectal liver metastases (CRLM) in patients previously treated with chemotherapy, according to research published in the November 2017 issue of HPB1. This finding also led to the conclusion that the improved diagnostic performance of EOB-MRI may alter the surgical plan in close to half of patients scheduled for liver resection.

Investigators in Toronto from the University Heath Network lead by Dr. Kartik Jhaveri prospectively compared the use of EOB-MRI and CECT for preoperative detection of CRLM in 51 patients treated with preoperative chemotherapy and who would be considered for liver resection.

Two independent blinded radiologists reviewed the CT and MRI images with at least four weeks between review of the two modalities. The readers characterized lesions using a 5-point scoring system (from Score 1-definitely metastasis to Score 5-definitely benign). Forty-one patients underwent hepatic resection with a mean interval of 35 days following the last imaging examination. The liver specimens were examined by gross inspection, and representative sections were sent for histopathological evaluation, which served as the reference standard.

The biggest differences between MRI and CT were found for small lesions ≤ 1.0 cm, while lesions of this size accounted for 38% (58/151) of the CRLM diagnosed on postoperative histopathology. For example, the sensitivity of MRI was found to be significantly higher than CT for the detection of small CRLM for both readers (R1, 93% vs. 29%; R2, 79% vs. 60%; p < 0.001), with no significant difference in sensitivity for larger lesions (> 1 cm) between the two modalities. In addition, the number of small lesions (≤ 1.0 cm) diagnosed as indeterminate was significantly lower on MRI compared to CT (combined result for both readers was 7% vs. 33%; p < 0.001).

The higher sensitivity of MRI led to the detection of 41 and 38 additional metastases for readers 1 and 2, respectively, compared to CT. Furthermore, 55 and 25 lesions for readers 1 and 2 that were scored as benign or indeterminate on CT were diagnosed as metastases on MRI. Based on these differences in diagnostic performance between the two modalities, MRI could have potentially changed the surgical plan in 45% of patients.


1. Jhaveri K et al. HPB 2017; 19(11):992-1000