Positron emission tomography

Published Date: 19th July 2016

Publication Authors: Malhotra P

​

Introduction 

Positron Emission Tomography (PET) is an important tool in the evaluation of pulmonary nodules. A number of risk prediction tools attempt to discriminate between benign and malignant nodules. Herder et al previously demonstrated an improvement in performance of the Mayo risk prediction tool when information regarding FDG avidity was incorporated. However, the Mayo model was developed for nodules identified on CXR, and may over-estimate malignant risk in smaller pulmonary nodules detected incidentally on CT scan. More recently a new risk prediction algorithm was described by McWilliams (Brock tool) derived and validated in CT screening populations. We sought to validate the Brock tool in a UK population, and to assess the effect of modulating this formula to include information about nodule FDG avidity.

Methods

Patients with pulmonary nodules were retrospectively identified, and malignant probability calculated using the Brock model. A modified risk prediction tool was developed incorporating a function of the SUV to the Brock tool risk.

Results

244 patients with pulmonary nodules were identified. The prevalence of malignancy in the overall population was 40.6% (33.2% lung cancer, 7.4% metastases). The area under the receiver operator curve (AUC ROC) for the Brock model was 0.868 (95% CI 0.825-0.911) for the whole cohort. 139 patients were subsequently investigated with PET-CT. When these nodules were considered separately, the AUC for the Brock model was 0.775 (95% CI 0.695-0.854). Using a composite tool incorporating the Brock model and SUV, the AUC was 0.935 (95% CI 0.895-0.975, p < 0.001 compared to standard Brock model). By comparison, the AUC for the Herder prediction tool was 0.916 (95% CI 0867-0.964).

Conclusion

The Brock model performs well in a UK population, but accuracy is significantly improved by incorporating the SUV on PET scans. This modified Brock model performs very similarly to the model described by Herder et al.

Al-Ameri, A; Malhotra, P et al. (2015).  The added value of positron emission tomography to a clinical prediction model for pulmonary nodules . Lung Cancer. 87 (Suppl 1), S15.

 

« Back