Purpose: The first aim of the review (aim 1) was to obtain the diagnostic performance values of (18)F-FDG PET for the detection and staging of liver metastases in patients with colorectal cancer (CRC), the second aim (aim 2) was to compare PET and conventional imaging modalities, and the third aim (aim 3) was to evaluate the impact of PET on patient management. The incidence of extrahepatic disease (EHD) detected by PET is also reviewed.
Methods: A comprehensive search was performed on PubMed/MEDLINE for studies evaluating PET and PET/CT in CRC patients with liver metastases up to June 2014. For inclusion PET had to have been performed prior to surgery, there had to be at least 18 patients in the study, and the reported data had to allow calculation of 2 × 2 contingency tables (for aim 1). A total of 18 studies were eligible for at least one of the three intended subanalyses including a total of 1,059 patients. Pooled sensitivity, specificity and accuracy and the corresponding 95 % confidence intervals were derived from the contingency tables on a patient basis (patient-based analysis, PBA) and a lesion basis (lesion-based analysis, LBA) for eight studies.
Results: Pooled sensitivity and specificity of PET on PBA were both 93 %. Corresponding values for LBA were 60 % and 79 %, respectively. Areas under the summary ROC were 0.97 for PBA and 0.67 for LBA. Regarding aim 2, PET had a slightly lower sensitivity than MRI and CT on PBA (93 %, 100 % and 98 %, respectively) and LBA (66 %, 89 % and 79 %, respectively) but appeared to be more specific than MRI and CT (86 %, 81 % and 67 %, respectively). PET findings resulted in changes in the management of a mean of 24 % of patients. The mean incidence of PET-based EHD was 32 %.
Conclusion: This meta-analysis suggests that FDG PET/CT is highly accurate for the detection of liver metastases on a patient basis but less accurate on a lesion basis. Compared to MRI, PET is less sensitive but more specific and affects the management of about one-quarter of patients.