Fifteen studies (n=863) were identified for inclusion. Twelve studies met the inclusion criteria, while 13 studies were tabulated; the patient numbers were difficult to provide for these as some were reviews.
The sensitivity and specificity of PET (1 study) for detecting or staging primary or recurrent disease were 87 and 67%, respectively, compared with 47 and 100% for CT (1 study), 94 and 100% for colonoscopy (2 studies) and 97 and 97% for hydrocolonic ultrasound (1 study).
The sensitivity and specificity of PET (4 studies) for diagnosing recurrent tumour versus scar were 92 to 95% and 100%, respectively, compared with 91 and 100% for MRI (1 study), 40 and 50% for immunoscintigraphy (1 study), and 99 and 88% for endoscopic ultrasound (1 study).
The sensitivity and specificity of PET (3 studies) for diagnosing liver metastases were 90 to 95% and 57 to 100%, respectively, compared with the following: 80 to 100% and 14 to 94% for CT (4 studies); 82 to 100% and 99 to 100% for MRI (2 studies); 85 and 98% for CT and/or ultrasound (1 study); 97 to 100% and 9 to 33% for CT portography (1 study); 80 and 90% for ultrasound (1 study); 9 to 47% and 92 to 98% for liver enzymes; 70 and 94% for pre-operative ultrasound; 84 and 97% for surgical exploration; and 97 and 98% for intra-operative ultrasound.