Diagnostic Performance of Whole-Body PET/MRI for Detecting Malignancies in Cancer Patients: A Meta-Analysis.

Abstract

BACKGROUND: As an evolving imaging modality, PET/MRI is preliminarily applied in clinical practice. The aim of this study was to assess the diagnostic performance of PET/MRI for tumor staging in patients with various types of cancer. METHODS: Relevant articles about PET/MRI for cancer staging were systematically searched in PubMed, EMBASE, EBSCO and the Cochrane Library. Two researchers independently selected studies, extracted data and assessed the methodological quality using the QUADAS tool. The pooled sensitivity, specificity, diagnostic odds ratio (DOR), positive likelihood ratio (PLR), and negative likelihood ratio (NLR) were calculated per patient and per lesion. The summary receiver-operating characteristic (SROC) curves were also constructed, and the area under the curve (AUC) and Q* estimates were obtained. RESULTS: A total of 38 studies that involved 753 patients and 4234 lesions met the inclusion criteria. On a per-patient level, the pooled sensitivity and specificity with 95% confidence intervals (CIs) were 0.93 (0.90-0.95) and 0.92 (0.89-0.95), respectively. On a per-lesion level, the corresponding estimates were 0.90 (0.88-0.92) and 0.95 (0.94-0.96), respectively. The pooled PLR, NLR and DOR estimates were 6.67 (4.83-9.19), 0.12 (0.07-0.21) and 75.08 (42.10-133.91) per patient and 10.91 (6.79-17.54), 0.13 (0.08-0.19) and 102.53 (59.74-175.97) per lesion, respectively. CONCLUSION: According to our results, PET/MRI has excellent diagnostic potential for the overall detection of malignancies in cancer patients. Large, multicenter and prospective studies with standard scanning protocols are required to evaluate the diagnostic value of PET/MRI for individual cancer types.