Background: Brain metastases are the main determining factor in the failure of patients with locally advanced non-small cell lung cancer (LA-NSCLC) by multimodality treatments. Whether we can use PCI to the patients with NSCLC and how to screen out high-risk patients are still controversial. We have reported a mathematical model, through which we can predict high-risk brain metastases in patients with postoperative LA-NSCLC. The purpose of this study is to verify the accuracy of the mathematical model, using new cases information.
Methods: A total of 196 patients of stage III NSCLC treated with surgical resection were retrospectively analyzed, to verify the consistency between actual and predictive brain metastases.
Results: The median survival time after surgery for all patients was 32.1 months. The one-, two- and three- year survival rate were 84.7%, 63.9%, 51.7%. The incidence rate of brain metastases was 42.3% (83/196). The incidence rate of brain metastases as the first site of recurrence was 28.1% (55/196). Results of accuracy of the mathematical model were sensitivity of 84.3%, specificity of 64.6%, positive predictive value of 63.6% and a negative predictive value of 84.9%, Measure of agreement Kappa value of 0.47 (P <0.001).
Conclusions: The mathematical model can predict brain metastases high risk patients with LA-NSCLC after surgery. It can be used as a basis to screening out patients of high risk brain metastases in future clinical trails about PCI.