201PEXPLORATORY ANALYSIS OF CIRCULATING FREE DNA TO IDENTIFY BIOMARKERS PREDICTIVE OF OUTCOME IN GALAXY-1, A LARGE RANDOMIZED PHASE IIB STUDY OF GANETESPIB IN COMBINATION WITH DOCETAXEL VERSUS DOCETAXEL ALONE IN ADVANCED NSCLC (NCT01348126)

Aim: NSCLC is comprised of several distinct molecular subtypes driven by somatic oncogenic mutations, and several of these (eg, EML4-ALK, BRAF, KRAS, KIT) confer dependence on Hsp90. Inhibition of Hsp90 leads to client protein degradation and tumor apoptosis. In the clinic, the Hsp90 inhibitor ganetespib (G) has single agent activity in various molecularly defined cancers (eg, EML4-ALK, mut KRAS, HER2 +, mut BRAF). Circulating free DNA (cfDNA), present at low levels in plasma, allows detection of somatic mutations by deep sequencing. The aim of this work was to determine the mutational spectrum in cfDNA from patients (pts) enrolled in the GALAXY-1 trial to uncover plasma-borne somatic mutations predictive of clinical outcome to G-based therapy. Methods: Ultra-deep sequencing was performed using plasma samples isolated from whole blood of adenocarcinoma pts enrolled in GALAXY-1. CfDNAwas isolated from 1ml of plasma collected at baseline using Qiagen kits and quantified by AQ real-time PCR. As only 3 of 105 samples (3%) met the desired cut-off of 1.7ng/µl cfDNA (10ng template) for direct sequencing, samples were concentrated by SpeedVac®; 36 samples (34%) met the requirement of 10ng template DNA, while the remaining samples had <10 ng template DNA. DNA from the total plasma cfDNA pool (n = 105) was enriched using a PCR-based approach (Ion AmpliSeq TM Cancer Hotspot Panel). Results: cfDNA targeted sequence analysis of the first 105 pts identified somatic mutations in several genes commonly mutated in NSCLC (eg, EGFR, PIK3CA, MET), as well as germline mutations and single nucleotide variants (SNVs) in genes not commonly seen in NSCLC (eg, KIT). Preliminary results suggest that PTEN and KIT mutations may be predictive of response to G + D. Analysis of cfDNA from remaining samples, including paired samples collected during cycle 1 and at end-of-treatment, is ongoing. Conclusions: Ultra-deep re-sequencing of somatic mutations in circulating cfDNA represents a new approach to exploring tumor heterogeneity and biomarkers of treatment outcome. The early analysis of cfDNA in GALAXY-1 identified G sensitive subgroups. Final results will be presented at the meeting.