Study selection and characteristics
A total of 367 articles were identified from the databases searched. Among them, 51 duplicates were removed and 254 studies were excluded through an initial screening. After a full text assessment for eligibility of the remaining 62 articles, 31 articles with 37 studies, 2577 patients with confirmed or suspected CA were identified for inclusion in this meta-analysis. No additional studies were found through reference screening of the included papers. Figure 1 shows the flow of the database search and literature selection process. The quality of the included studies was regarded as high according to the QUADAS-2 scale (Figure 2). Table 1 details the characteristics of studies included.
Diagnostic performance of noninvasive modalities
The numbers of studies included in the analysis of CMR, SPECT and PET were 8, 20 and 9, respectively. The pooled sensitivity of CMR, SPECT and PET were 0.84 [0.75, 0.90], 0.98 [0.94, 0.99] and 0.78 [0.54, 0.92], respectively. The overall specificities were 0.87 [0.77, 0.93], 0.92 [0.83, 0.97] and 0.83 [0.70, 0.91] for CMR, SPECT and PET, respectively. The AUC values of CMR, SPECT and PET were 0.92 [0.89, 0.94], 0.99 [0.98, 1.00] and 0.86 [0.82, 0.88] (Figures 3-5).
Diagnostic performance of prospective studies
With regard to prospective studies of these detection approaches, the respective overall sensitivities of CMR, SPECT and PET were 0.85 [0.76, 0.91], 0.98 [0.90, 0.99] and 0.85 [0.63, 0.95]. The pooled specificities were 0.89 [0.72, 0.96], 0.87 [0.73, 0.94] and 0.98 [0.68, 1.00] for CMR, SPECT and PET, respectively (Supplementary Figures 1-3). The AUC values of CMR, SPECT and PET were 0.92 [0.89, 0.94], 0.97 [0.96, 0.98] and 0.98 [0.97, 0.99].
Subgroup analysis of SPECT tracers
The numbers of studies using 99mTc-DPD, 99mTc-PYP, 99mTc-HMDP, and 99mTc-aprotinin for SPECT radiotracers were 5, 8, 5, and 2, respectively. Studies using 99mTc-aprotinin were not enrolled in pooled analysis for the inadequate number of studies. Overall results demonstrated that 99mTc-HMDP manifested the highest sensitivity (0.99 [0.83, 1.00]). 99mTc-PYP had the highest pooled specificity (0.95 [0.86, 0.99]). The pooled sensitivity of 99mTc-DPD and 99mTc-PYP reached 0.98 (Supplementary Figures 4-6). The AUC values of 99mTc-DPD, 99mTc-PYP, 99mTc-HMDP were 0.89 [0.86, 0.92], 0.99 [0.98, 1.00], and 0.99 [0.98, 1.00], respectively.
Subgroup analysis of PET tracers
The number of included studies using 11C-PIB, 18F-florbetaben, 18F-flutemetamol, and 18F-NaF for PET tracers were 4, 1, 1, and 3, respectively. Only PET studies utilizing 11C-PIB was included in pooled analysis. It demonstrated a pooled sensitivity of 0.91 [0.81, 0.96], and its pooled specificity was 0.97 [0.81, 1.00] (Supplementary Figures 7). The AUC value of 11C-PIB was 0.98 [0.97, 0.99]. Both the reported sensitivity and specificity of 18F-florbetaben PET for the separation of patients with CA from patients without CA were 100%. The study of 18F-flutemetamol showed a sensitivity of 0.17 with a high proportion of false-negative PET results.
Heterogeneity and publication bias
The I2 values for meta-analysis of CMR were 64 (pooled sensitivity) and 61 (pooled specificity). The respective I2 static for SPECT were 94 and 93. As for PET, the I2 values for pooled analysis of sensitivity and pooled specificity were 84 and 0. Deek’s funnel plot asymmetry tests for publication bias yielded p values of 0.89, 0.88, and 0.06 for CMR, SPECT and PET, which revealed that there may be no potential publication bias in the study.
Sensitivity analysis was conducted to assess the potential influence of single study on the overall results. After omitting each study one by one, the pooled results of CMR, SPECT, PET and the corresponding subgroup analysis remained robust.