Optimal Preoperative Examination of Colorectal Liver Metastases: Assessment of the diagnostic performance of CT, PET/CT, MRI and Intraoperative Ultrasound

Abstract

Background: Metastases from colorectal cancer are in many cases confined to the liver. Metastases are frequent and in many cases decisive for the prognosis. Recent developments have led to an increasing number of available treatment options. However, only surgical resection of all the CRLM has the potential of long-term survival or even cure. The choice of a suitable treatment is therefore of great importance and is usually based on imaging findings. Liver imaging in patients with CRLM must ensure the identification of all CRLM with the highest possible accuracy. However, several imaging modalities are available and the optimal diagnostic approach has yet to be determined. Purpose: Our 1st study evaluated the contribution of CE-IOUS as last imaging modality prior to liver resection for the staging of CRLM and its impact on operation strategy. In the 2nd study we evaluated the preoperative diagnostic performance of CT, MRI and PET/CT in patients with CRLM scheduled to liver resection. In the 3rd study, a novel method for rgPET/CT and its impact on the preoperative diagnostic performance for the detection of CRLM was investigated. Materials and methods: In the 1st study, 86 patients underwent 97 open liver resections for CRLM. The patients were retrospectively evaluated. Preoperative CT was available for all patients and MRI was available in 72%. CE-IOUS findings were compared with preoperative findings and the initial planned operation strategy with the finally performed operation. In the 2nd study, 46 patients scheduled for liver resection for suspected CRLM were prospectively included. Multiphase CT, Gd-EOB-DTPA-enhanced MRI with diffusion-weighted imaging and PET/CT were performed in all patients. Each examination was evaluated by two independent and blinded readers. The diagnostic performance for CRLM was determined for each modality and compared with each other. In the 3rd study, 43 patients with both PET/CT and rgPET/CT available were included from the same patient cohort as patients from the 2nd study. The preoperative impact of rgPET/CT on the diagnostic performance of PET/CT for CRLM was evaluated. McNemar test was used to identify significant differences between the modalities in the 2nd and 3rd study. Mann–Whitney test was used for comparisons of SUV data in the 3rd study. The reference standard was histopathological confirmation or if not available follow-up in all studies. Results: In the 1st study, CE-IOUS changed the initial operation strategy in 30%, resulting in a smaller resection in 11%, larger resection in 13% and 5% were found inoperable. In 17 patients additional 31 CRLM were identified. In the 2nd study, MRI had the highest sensitivity both overall and for CRLM <10 mm (P<0.001). The overall sensitivity and PPV on per-lesion basis were 68%/89% for CT, 90%/82% for MRI and 61%/97% for PET/CT. For CRLM <10 mm it were 16%/54% for CT, 74%/64% for MRI and 9%/57% for PET/CT. In the 3rd study, the combined sensitivity and PPV of standard PET/CT + rgPET/CT was 68%/94% which was significantly higher than for standard PET/CT (P=0.002) or rgPET/CT (P=0.031) alone. Conclusion: CE-IOUS should be performed prior to liver resection for CRLM to optimize operation strategy. MRI has the highest sensitivity for CRLM, particularly for CRLM <10 mm. PET/CT with respiratory-gating has the potential to improve the diagnostic performance of standard PET/CT for CRLM.