Sixty-seven studies, reporting 145 data sets were included in the review. The most commonly failed QUADAS criteria were assessment of all participants with the same reference standard, adequate description of the reference standard, and interpretation of the reference standard without knowledge of the index test results.
18 FDG-PET and PET/CT: The pooled, per patient, estimate of sensitivity was 89.7% (95% CI 87.4 to 91.6%) and the corresponding pooled estimate of specificity was 96.8% (95% CI 96.2 to 97.3%); 26 studies, 4,367 participants. The pooled, per lesion, estimate of sensitivity was 86.9% (95% CI 85.3 to 88.3%) and the corresponding pooled estimate of specificity was 97.0% (95% CI 96.5 to 97.5%); 18 studies, 6,779 lesions.
CT Scan: The pooled, per patient, estimate of sensitivity was 72.9% (95% CI 66.6 to 78.6%) and the corresponding pooled estimate of specificity was 94.8% (95% CI 92.4 to 96.6%); seven studies, 723 participants. The pooled, per lesion, estimate of sensitivity was 77.1% (95% CI 72.9 to 81.0%) and the corresponding pooled estimate of specificity was 83.2% (95% CI 79.7 to 86.2%); three studies, 983 lesions.
MRI Scan: The pooled, per patient, estimate of sensitivity was 90.6% (95% CI 86.7 to 93.7%) and the corresponding pooled estimate of specificity was 95.4% (95% CI 93.6 to 96.8%); 15 studies, 1,032 participants. The pooled, per lesion, estimate of sensitivity was 90.4% (95% CI 87.2 to 93.0%) and the corresponding pooled estimate of specificity was 96.0% (95% CI 95.2 to 96.8%); seven studies, 2,874 lesions.
Bone scintigraphy: The pooled, per patient, estimate of sensitivity was 86.0% (95% CI 84.0 to 87.8%) and the corresponding pooled estimate of specificity was 81.4% (95% CI 80.0 to 82.8%); 48 studies, n=4,638 participants. The pooled, per lesion, estimate of sensitivity was 75.1% (95% CI 72.9 to 77.2%) and the corresponding pooled estimate of specificity was 93.6% (95% CI 92.8 to 94.2%); 21 studies, n=6,083 lesions.
The pooled, per patient and per lesion, estimates of sensitivity were similar for 18FDG-PET and MRI and both were significantly higher than those for CT and bone scintigraphy. The pooled, per patient, estimates of specificity were similar for 18FDG-PET, CT and MRI; all were higher than bone scintigraphy. For pooled per lesion estimates, the specificity of CT was lower than that of bone scintigraphy.
Subgroup analyses indicated that PET/CT was more sensitive than PET, axial MRI was more sensitive than whole body MRI, and bone scintigraphy using SPECT was more specific than without SPECT. Full results of subgroup analyses and meta-regression were reported in the article.