Forty-seven studies were identified for question 1 and 41 studies for question 2.
Question 1: Mediastinal staging.
CT scan (20 studies, n=3,438): the pooled sensitivity was 0.57 (95% CI: 0.49, 0.66) and the pooled specificity was 0.82 (95% CI: 0.77, 0.86). The authors found evidence of marked heterogeneity between the pooled studies. The overall PPV and NPV were 0.56 (range: 0.26, 0.84) and 0.83 (range: 0.63, 0.93), respectively. The overall prevalence of mediastinal disease was 28% (range: 18 to 50%).
PET scan (18 studies, n=1,045): the pooled sensitivity was 0.84 (95% CI: 0.78, 0.89) and the pooled specificity was 0.89 (95% CI: 0.83, 0.93). The overall PPV and NPV were 0.79 (range: 0.40, 1.00) and 0.93 (range: 0.75, 1.00), respectively. The overall prevalence of mediastinal disease was 32% (range: 5 to 56%).
Combined CT and PET scan (3 studies, n=152): the sensitivity ranged from 0.78 to 0.93, and the specificity ranged from 0.82 to 0.95. The PPV and NPV ranged from 83 to 93% and from 88 to 95%, respectively. The prevalence of mediastinal disease ranged from 32 to 50%.
MRI scan (1 study, n=20): with gadolinium, the sensitivity was 1.0, the specificity was 0.91, and the PPV and NPV were 0.96 and 1.0, respectively. Without gadolinium, the sensitivity was 0.63, the specificity was 1.0, and the PPV and NPV were 1.0 and 0.55, respectively.
EUS (5 studies, n=163): the pooled sensitivity was 0.78 (95% CI: 0.61, 0.89) and the pooled specificity was 0.71 (95% CI: 0.56, 0.82). The overall PPV and NPV were 75% and 79%, respectively. The overall prevalence of mediastinal involvement was 50% (range: 25 to 76%).
Question 2: Clinical evaluation.
Detection of brain metastases (17 studies, n=1,784): the NPV ranged from 0.79 to 1.00; the authors stated that there was significant heterogeneity between the included studies in the pooled estimate (0.94, 95% CI: 0.91, 0.96). For 8 studies that evaluated patients with positive and negative clinical evaluation, the pooled sensitivity was 0.76 (95% CI: 0.64, 0.84) and the pooled specificity was 0.87 (95% CI: 0.74, 0.94). The PPV was 0.54 and the overall prevalence of brain metastases was 13% (range: 0 to 32%).
Detection of abdominal metastases (12 studies, n=1,201): the NPV ranged from 0.82 to 1.0; the authors stated that there was heterogeneity between the included studies in the summary estimate (0.95, 95% CI: 0.93, 0.96). For 3 studies that evaluated patients who had a negative and a positive clinical evaluation, the pooled sensitivity was 0.92 (95% CI: 0.83, 0.97) and the pooled specificity was 0.49 (95% CI: 0.25, 0.74). The PPV ranged from 0.20 to 0.59, and the overall prevalence of abdominal metastases was 10% (range: 0 to 40%).
Detection of bone metastases (7 studies, n=633): the pooled sensitivity was 0.87 (95% CI: 0.79, 0.93) and the pooled specificity was 0.67 (95% CI: 0.40, 0.88). The NPV ranged from 0.70 to 1.0; the authors stated that heterogeneity was detected between the included studies in the summary estimate (0.90, 95% CI: 0.86, 0.93). The PPV ranged from 0.16 to 0.90, and the overall prevalence of bone metastases was 20% (range: 8 to 34%).