Biomarker Prediction of Efficacy to Vandetanib Plus Gemcitabine in a Phase II Double Blind Multicenter Randomized Placebo Controlled Trial in Locally Advanced or Metastatic Pancreatic Carcinoma.

4104 Background: We investigated the potential of biomarkers to predict efficacy of vandetanib and gemcitabine in patients with locally advanced (N = 41) or metastatic (N = 101) pancreatic cancer in a phase II double-blind multicentre randomised placebo-controlled trial. Methods: All patients were 18y or above, (ECOG = 0-2), with at least 3 mths life expectancy had gemcitabine (1000mg/m2 30min iv wkly for 7 wks, followed by a 1wk break, then cycles of wkly treatment for 3wks with a 1-wk break) and randomly assigned to 300mg/d vandetanib or placebo once daily until disease progression. The primary outcome was overall survival (OS) by intention to treat. A panel of potential biomarkers was tested to predict best survival with vandetanib and gemcitabine. Results: 142 patients were randomised, median FU = 24·9 mths with 131 deaths. The median (95% CI) OS in the 70 gemcitabine-placebo patients was 8·95 (6·55-11·7) mths and 8·83 (7·11-11·6) mths in the 72 gemcitabine-vandetanib patients (HR = 1·21, 95% CI = 0·85, 1·73; log rank X21df = 1·1; P = 0·303). A CTCAE V.4.02 rash grade 2 or above occurred in 4 (6 %) of 70 placebo patients versus 14 (19%) of 72 vandetanib patients. The median OS for the 14 vandetanib patients and with rash was 11·92 (10·89 – NA) mths, 7·76 (4·34 – 11·5) mths for the 58 vandetanib patients and without rash and 8·95 (6·55 – 11·7) mths for the gemcitabine-placebo patients (log rank Χ2 2df = 7·23; P = 0·03). We identified two biomarkers that could select patients for response to vandetanib (JN101, JN102). The biomarker combination was present in 26 patients with median OS of 12.1 (10.9, 16.0) mths versus 8.15 (6.67, 11.7) mths for 23 patients with the same biomarker profile in the placebo group (HR = 0.53 [0.29, 0.97], p = 0.0396). A logistic regression model showed that patients with JN102 were more likely to develop a rash (OR =0.81 [0.713, 0.925] p = 0.002). Conclusions: A two biomarker combination and a rash grade 2 or above may predict response to vandetanib and gemcitabine. This requires prospective evaluation. Clinical trial information: 96397434.