[Pulmonary nodule: a bayesian approach]

A solitary pulmonary nodule is a common clinical problem. It is usually detected incidentally. The prevalence of solitary pulmonary nodule (SPN) in the lung cancer screening study varies from 8 to 50% (with a prevalence of malignant nodule from 1 to 13%). The bayesian approach can help us to identify promptly malignant nodule in order to treat them surgically and to avoid surgery for benign nodules. Therefore, it is needed to estimate the probability of cancer (Pca) in the SPN. Likelihood ratio (LR) for overall prevalence of malignancy and for different clinical and radiological information (age, smoking exposure, symptoms, cancer history, nodule size, spiculation, calcification, location, growth...) can be obtained from the literature. The odds of cancer-malignancy (odds ca) can be calculated by multiplying all of these LRs together. The Pca = odds ca/1+odds ca. Using this bayeasian approach, the probability of cancer based on an abnormal or normal fluoro-2-deoxy-D-glucose-positron emission tomography (FDG-PET) scan has been estimated. Sensitivity, specificity, positive predictive value and negative predictive value of PET scan are respectively about 90%, 83%, 92% and 90%. Moreover, the LR for malignancy are higher with an abnormal PET scan when compared to most clinical and radiological LRs. Today, the Bayesian approach of SPN must include PET scan.