509PDPRIMARY TUMOUR LOCATION (PTL) AS A PROGNOSTIC AND PREDICTIVE FACTOR IN ADVANCED COLORECTAL CANCER (ACRC): DATA FROM 2075 PATIENTS (PTS) IN RANDOMISED TRIALS

ABSTRACT Aim: Variations in tumour biology and outcomes depending upon PTL have been reported in aCRC. We tested effects of PTL in two phase III randomised trials. Methods: We studied 2075 pts, from FOCUS (1st-line; n = 1390; Lancet 370: 143-52) and PICCOLO (2nd-line; n = 685; Lancet Oncol 14:749-59). We compared: (1) right colon (RC) vs [left colon (LC) or rectum] and (2) LC vs rectum. Association of PTL with RAS/RAF, AREG/EREG and MMR was assessed where available. PTL was tested as a prognostic factor, then for predictive utility by testing PTL/treatment interactions on OS and PFS for: 1st line FU vs doublet (FOCUS); 1st-line irinotecan doublet vs oxaliplatin doublet (FOCUS); 2nd line irinotecan (Ir) +/- panitumumab (Pan) (KRAS-wt pts, PICCOLO). Results: PTL was RC in 575 (28%), LC in 801 (39%) and rectum in 699 (34%) pts. RC tumours had more BRAF mutations (n = 1136, 22% vs 6%, p Interaction test p-values for treatment comparisons Comparison Interaction Test p-values RC vs LC/rectum LC vs Rectum OS PFS OS PFS 1st line doublet vs single-agent FU (1334 pts) 0.39 0.84 0.48 0.8 1st line OxFU doublet vs IrFU doublet (452 pts) 0.99 0.67 0.18 0.50 2nd line Ir +/- Pan KRAS-WT (450 pts) 0.35 0.13 0.63 0.46 2nd line Ir +/- Pan KRAS/BRAF-WT (341 pts) 0.72 0.89 0.83 0.19 Conclusions: We confirm that RC tumours are biologically distinct, and have worse outcomes with 1st-line therapy. We did not find PTL to be predictive for the benefit of the drugs under test in these trials, so cannot recommend its use for selection of therapy. Better objective predictive biomarkers are required. Disclosure: All authors have declared no conflicts of interest.