Smoking And Prognostic Factors In Advanced Non-Small Cell Lung Cancer (Nsclc)

18048 Background: The aim of this prospective, multi-country, observational study (B9E-AA-B004) is to estimate the effect of prognostic factors, including continued smoking during therapy, on treatment outcomes in patients (pts) with advanced NSCLC receiving first- line chemotherapy with a gemcitabine (gem)-platinum combination as part of their routine care. Methods: Major eligibility criteria included: tissue diagnosis of advanced stage IIIB/IV NSCLC not amenable to curative surgery/radiotherapy; no prior chemotherapy; ECOG performance status (PS) 0 or 1; and written informed consent. A predictive model was constructed and validated by splitting the data at random by centre into two datasets in a ratio of 3:1 Construction:Validation. The primary and secondary endpoints are the effect of prognostic factors on survival and selected adverse events (AEs), respectively. The association of smoking with outcomes was tested in the Construction dataset. Results: This interim analysis to assess the effect of prognostic factors on AEs occurred when all pts had completed treatment. 1214 pts were enrolled: 75.1% male; mean age 60.5 yr, range 23–86 yr; 57.1% Stage IV; 66.2% PS 1; 69.4% received gem-cis, 30.5% gem-carb; 25.7% had never smoked, 70.8% had ever smoked and 11.2% continued smoking during therapy. 22.0% of pts had =1 AE. After variable selection in the Construction database (891 pts) the following factors were associated with an AE possibly related to therapy: disease stage (IV vs III, odds ratio (OR) =1.48, p=0.034), weight loss >10% (OR=0.60, p=0.017), age (<70 vs =70, OR=0.66, p=0.046), treatment (gem- carb vs gem-cis, OR=1.5, p=0.04), pain at baseline (present vs absent, OR=1.5, p=0.03), country (OR vs Taiwan ranged from 0.32 (Israel) to 4.2 (Egypt), p<0.0001). Sex (F vs M, OR=0.86) was then added to the model because of its clinical importance. There was a trend towards a higher probability of an AE with continued smoking during therapy (OR=1.4), but this was not statistically significant (p=0.28). Conclusions: This model can be used to improve the prediction of whether patients are likely to experience treatment-related AEs. While the trend was for a greater AE rate in pts who continued to smoke during therapy, this was not proven. [Table: see text]