Meta-analysis: the detection of pancreatic malignancy with positron emission tomography.

BACKGROUND Several factors contribute to the high mortality of pancreatic cancer, including limitations of diagnostic imaging. AIM To perform a meta-analysis to assess the diagnostic accuracy of Fluro-deoxy-glucose positron emission tomography with computed tomography compared with computed tomography alone. METHODS Articles were identified through a MEDLINE search and bibliography review. Summary estimates and receiver operating curves were calculated using Meta-Test 0.6. Publication bias and heterogeneity were assessed with a funnel plot and chi-squared test. RESULTS The summary estimate and 95% confidence interval for sensitivity and specificity were as follows: computed tomography 81% (72-88%) and 66% (53-77%), PET after a positive computed tomography 92% (87-95%) and 68% (51-81%), PET after a negative computed tomography 73% (50-88%) and 86% (75-93%) and PET after an indeterminate computed tomography 100 and 68%. The area under the summary receiver operating curve was 0.82 for computed tomography and 0.94 for PET. There was no heterogeneity or publication bias. CONCLUSIONS Our results suggest that although adding Fluro-deoxy-glucose positron emission tomography to the diagnostic work-up may enhance the diagnosis of pancreatic malignancy, its usefulness will vary depending upon the pretest probability of the patient, the results of computed tomography and the provider's testing thresholds. Further evaluation using a well-designed prospective study with a cost-effectiveness analysis is needed to clarify the appropriate role of Fluro-deoxy-glucose positron emission tomography.