MP56-05 THE PREDICTIVE VALUE OF GSTT1 POLYMORPHISMS IN PREDICTING THE EARLY RESPONSE TO INDUCTION BCG THERAPY IN PATIENTS WITH NON-MUSCLE INVASIVE BLADDER CANCER

Introduction: We evaluated the predictive value of glutathione S transferase mu (GSTM1) and theta (GSTT1) polymorphisms in early response to bacillus Calmette-Guerin (BCG) induction therapy in patients with primary non–muscle invasive bladder cancer. Methods: GSTM1 and GSTT1 polymorphisms were analyzed by multiplex polymerase chain reaction using blood genomic DNA from 135 patients with primary non–muscle invasive bladder cancer who were being treated with a single induction course of BCG. BCG nonresponsiveness (early BCG failure) was defined as a tumor recurrence or progression within 12 months after BCG induction therapy. The predictive value of GST polymorphisms was evaluated by Kaplan-Meier analysis and multivariate logistic regression models. Results: Patients carrying a GSTT1-positive genotype demonstrated a higher likelihood of early BCG failure regardless of cigarette smoking. After stratification based on the tumor stage and grade, the high-risk group (T1G3) with a GSTT1-positive genotype showed a 14-fold higher risk of early BCG failure compared with those with a GSTT1-null genotype. In a combined analysis of 2 genes, the GSTT1positive/GSTM1-null genotype had a higher risk of BCG nonresponsiveness compared with the GSTT1-null/GSTM1-null genotype (odds ratio ¼ 4.17, 95% CI: 1.54–11.26). By multivariate logistic regression analysis, the GSTT1-positive genotype was an independent predictor of early BCG failure (odds ratio ¼ 3.67, 95% CI: 1.61–8.38). Kaplan-Meier estimates revealed a significant difference in disease-free survival depending on the GSTT1 genotype (log rank test, P ¼ 0.038). Conclusions: The results of this study suggest that the GSTT1-positive genotype is an independent predictor of early BCG failure. These results can help determine whether patients would benefit from adjuvant BCG treatment or may require more aggressive alternative