Relation of tumor pathologic and molecular features to outcome after surgical resection of localized primary gastrointestinal stromal tumor (GIST): Results of the intergroup phase III trial ACOSOG Z9001.

Abstract

10006 Background: Adjuvant therapy with imatinib mesylate increases recurrence-free survival (RFS) in primary GIST (Lancet 2009). However, the effect of tumor pathologic and molecular factors on outcome in patients treated with or without adjuvant imatinib is uncertain. Methods: 713 patients with a KIT-expressing, ≥3 cm, primary GIST were randomized postoperatively in a double-blind manner to 1 year of 400 mg imatinib or placebo daily. Patients underwent serial (q3 months x8 then q6 months x6) cross-sectional imaging. Mitotic rate (mitoses per 50 high power fields) and KIT and PDGFRA mutation status were determined by central pathologic analyses. Tumor mitotic rate, size, location, and mutation status were available in 513 patients. Results: Tumor size was <5 cm in 41% of patients, 5-10 cm in 34%, and ≥10 cm in 25%. Tumor location was stomach (65%), small bowel (31%), rectum (1%), or other (3%). Mitotic rate was high (≥5) in 39% of patients. The most common (68%) mutation occurred in KIT exon 11. At a median follow-up of 20 months, 2 year RFS was 74% in the placebo arm vs. 91% in the imatinib arm. On multivariate analysis of the placebo arm, RFS was worse with high mitotic rate (p<0.0001, HR 17.1), tumor size (p<0.01, HR 1.7), small bowel location (p<0.05, HR 2.1), and KIT exon 11 mutation (p<0.05, HR 3.0). On multivariate analysis of the imatinib arm, RFS was worse with high mitotic rate (p<0.001, HR 7.4), tumor size (p<0.01, HR 2.9), and small bowel (p<0.05, HR 3.1) or rectal (p<0.05, HR 5.4) location. In the placebo and imatinib arms respectively, 2 yr RFS was 65 vs. 91% (p<0.0001) for KIT exon 11 mutation and 76 vs. 100% (p<0.01) for PDGFRA mutation. Notably, there was only 1 event in the 28 total patients with a PDGFRA D842V mutation. Patients with a KIT exon 9 mutation assigned to placebo (n=13) had a 1 yr RFS of 80% vs. 100% on the imatinib arm (n=22), but overall RFS was not different. There was no statistical difference in RFS between arms in the 64 patients with wild-type KIT and PDGFRA. Conclusions: Tumor pathologic features and mutation status are both prognostic and predictive for RFS after surgical resection of primary GIST. Author Disclosure Employment or Leadership Position Consultant or Advisory Role Stock Ownership Honoraria Research Funding Expert Testimony Other Remuneration Novartis, Pfizer Novartis, Pfizer Novartis, Pfizer Pfizer