The review included a total of 44 studies. Thirty one studies (n=1,841) assessed the performance of EUS for the detection of regional lymph node metastases. Five studies (n=339) assessed the performance of EUS for the detection of celiac lymph node metastases. Seventeen studies (n=943) assessed the performance of CT for the detection of regional lymph node metastases. Five studies (n=254) assessed the performance of CT for the detection of abdominal lymph node metastases. Seven studies (n=437) assessed the performance of CT for the detection of distant metastases. Ten studies (n=424) studies assessed the performance of FDG-PET for the detection of regional lymph node metastases. Nine studies (n=475) assessed the performance of FDG-PET for the detection of distant metastases.
Regional lymph node metastases:
Pooled estimates of sensitivity were EUS 0.80 (95% CI: 0.75 to 0.84), CT 0.50 (95% CI: 0.41 to 0.60) and FDG-PET 0.57 (95% CI: 0.43 to 0.70). Specificity estimates were EUS 0.70 (95% CI: 0.65 to 0.75), CT 0.83 (95% CI: 0.77 to 0.89) and FDG-PET 0.85 (95% CI: 0.76 to 0.95). There were no significant differences in overall diagnostic performance, as measured by DOR, across the three imaging modalities. However, significant differences in sensitivities and specificities were apparent (EUS was more sensitive and less specific than CT and FDG-PET). Regression analysis showed no significant effect on test performance for any study or patient characteristic assessed. Publication bias was considered unlikely.
Celiac and abdominal lymph node metastases:
The pooled estimates of sensitivity and specificity for EUS in the detection of celiac lymph node metastases were 0.85 (95% CI: 0.72 to 0.99) and 0.96 (95% CI: 0.92 to 1.00). The pooled estimates of sensitivity and specificity for CT in the detection of abdominal lymph node metastases were 0.42 (95% CI: 0.29 to 0.54) and 0.93 (95% CI: 0.86 to 1.00). Publication bias could not be assessed.
Distant metastases:
The pooled estimates of sensitivity were CT 0.52 (95% CI: 0.33 to 0.71) and FDG-PET 0.71 (95%CI: 0.62 to 0.79). The corresponding specificity estimates were CT 0.91 (95%CI: 0.86, 0.96) and FDG-PET 0.93 (95% CI: 0.89, 0.97). The overall diagnostic performance of FDG-PET was significantly better than that of CT (relative DOR 2.26, 95% CI: 1.09 to 4.71, p<0.03), although pooled estimates of sensitivity and specificity were not significantly different. Regression analysis showed no significant effect on test performance for any study or patient characteristic assessed. Publication bias was considered unlikely.