e19158 Background: There is limited data on the prevalence of brain metastases at presentation in NSCLC. Between 1988 and 1997, the Surveillance, Epidemiology, and End Results (SEER) database recorded the use of brain radiation in patients with lung cancer. We developed a predictive clinical model for brain metastases (BM) at presentation based on the use of brain radiation (BRT), which may be used as a surrogate since it is not indicated in NSCLC patients without BM. Methods: Patients with NSCLC diagnosed between 1988-1997, aged 21 or older were subdivided according to radiation data into none or indicated (given, refused or unknown if administrated). Since the database records only initial therapy, patients with indication for radiotherapy were assumed to have had early brain metastases. Frequencies and odds Ratio (OR) were calculated for each demographic factor and the most significant were included in the predictive model. Results: Among the 131,456 patients evaluated, 10,963 (8.4%) had BRT indicated. The most significant predictors in multivariate analysis were age > 60 (HR 0.5; 95% CI 0.46-0.54), non-squamous histology (HR 0.53; 95% CI 0.48-0.59), size > 5 cm (HR 1.27; 95% CI 1.16-1.39), grades III-IV (HR 2.04; 95% CI 1.82-2.27) and N2-3 involvement (HR 2.55; 95% CI 2.34-2.78). The 40,963 patients with known status of all 5 factors were used in the predictive model. The frequency of BM for patients with zero, 1, 2, 3, 4 and 5 factors were 0.7%, 1.8%, 4.7%, 8.9%, 13.7% and 20.2% respectively. Conclusions: Despite the limitations of this study, including indication for BRT as a surrogate for BM, older data and possible bias in the use of brain imaging according to demographic factors, our findings showed that the risk of BM at presentation may be stratified with the use of 5 clinical factors. These results may be used as a baseline for studies on clinical and molecular determinants of BM.
Clinical predictive model for the 1-year remission probability of IgA vasculitis nephritis
