Advanced cardiac imaging techniques, 18F-fluoro-2-deoxyglucose positron emission tomography (FDG-PET) and cardiac magnetic resonance (CMR) imaging, have shown high accuracy for the detection of cardiac sarcoidosis (CS) in small studies. This meta-analysis was performed to evaluate the performance of FDG PET and CMR in the diagnosis of CS.

PubMed, Embase, and Cochrane Library were searched from the earliest available date of indexing to December 2019 for studies evaluating the accuracy of FDG-PET and CMR for the evaluation of possible CS defined mainly by the Japanese Ministry of Health, Labour and Welfare (JMHW) criteria. For robustness, only studies with a sample size of ≥20 patients were selected and sensitivity analysis excluding outliers was conducted. To calculate the overall performance of the diagnostic accuracy of FDG-PET and CMR, summary receiver operating curve (SROC) and area under the curve (AUC) analyses were done. Heterogeneity was evaluated with the Cochrane Q test and I² test. Possible sources of heterogeneity were further explored by subgroup analyses.

Fourteen studies used FDG-PET and six studies used CMR involving 1491 patients (mean age 53 years, 38% male) fulfilled the inclusion criteria. The pooled sensitivity of FDG-PET for diagnosis of CS was 76.7% (95% CI 71.6–81.2%; Q=4.4, P<0.0001; I²=86.2%) and the pooled specificity was 86.0% (95% CI 82.8–88.8%; Q=37.4, P<0.0001; I²=65.2%). For the CMR, the pooled sensitivity was 94.0% (95% CI 88.8–97.2%; Q=10.59, P=0.06; I²=52.8%) and the pooled specificity was 88.8 (95% CI 85.3–91.3%; Q=80.25, P<0.0001; I²=93.8%). SROC analysis demonstrated good overall accuracy for FDG-PET (AUC value of 0.91, and Q* 0.85) and CMR (AUC value of 0.96, and Q* 0.9) (figure 1). The sensitivity analysis of FDG-PET (11 studies) demonstrated a pooled sensitivity of 86.2% (95% CI 81.3–90.2%; Q=35.3, P=0.0001; I²=71.6%) and pooled specificity of 87.9% (95% CI 84.3–90.9%; Q=16.7, P=0.08; I²=40.2%). Moreover, the sensitivity analysis of CMR (5 studies) showed a pooled sensitivity of 89.3% (95% CI 78.1–96%; Q=7.3, P=0.12; I²=44.9%) and pooled specificity of 77.9% (95% CI 72.1–82.9%; Q=2.5, P=0.6; I²=0%). Meta-regression showed that the age of the patients (coefficient 0.192; P=0.04) and the size of the studies (coefficient 0.035; P=0.028) were the sources of heterogeneity in FDG-PET and CMR, respectively. The subgroup analysis by the age ≥55 years (6 studies and 276 patients) showed improvement in the diagnostic accuracy of FDG-PET with sensitivity of 92.2% (95% CI 86.1–96.2%; Q=18.9, P<0.01; I²=73.5%) and specificity of 85.1% (95% CI 78.4–90.4%; Q=4, P=0.553; I²=0%).

FDG-PET and CMR are useful adjunctive diagnostic tools in the evaluation of CS. Further large multi-center studies are warranted to validate the diagnostic accuracy of FDG-PET and CMR for diagnosis of CS.