Author’s Disclosures

1. Association of ERCC1 gene expression with outcome in stage II-III esophageal adenocarcinoma patients treated with preoperative platinum-based chemoradiation in a phase II cooperative group study (SWOG S0356). P. Bohanes1, B. H. Goldman2, J. K. Benedetti2, C. Blanke3, L. P. Leichman4, S. Iqbal1, C. R. Thomas5, C. L. Corless5, K. G. Billingsley5, K. D. Danenberg6, P. J. Gold7, and H.J. Lenz1 1University of Southern California, Los Angeles, CA; 2SWOG Statistical Center, Seattle WA; 3University of British Columbia Vancouver, BC, Canada; 4Desert Regional Medical Center, Palm Springs, CA; 5Oregon Health and Science University, Portland, OR;6Response Genetics, Inc, Los Angeles, CA; 7Swedish Cancer Institute, Seattle, WA

2. Author’s Disclosures • P. Bohanes: None • B.H. Goldman: None • L. Leichman: None • C. Blanke: Sanofi-Aventis • S. Iqbal: Sanofi-Aventis • C.R. Thomas: None • C.L. Corless: None • J.K. Benedetti: None • K.G. Billingsley: None • K.D. Danenberg: Response Genetics • P. Gold: None • H.J. Lenz: Sanofi-Aventis, Response Genetics

3. Background • There is no worldwide accepted standard treatment for locally advanced esophageal adenocarcinoma. • Neoadjuvant chemo-radiation prior to surgery is one of the accepted treatment strategies • Platinum agents are often used as the chemotherapy backbone for patients treated with trimodality therapy • In contrast to the growing number of predictive biomarkers for anti-cancer agents, there are no established biomarkers to select patients who will benefit most from chemo-radiation. • Utilization of predictive biomarkers to select therapy should lead to higher cure rates.

4. Background – ERCC1 • ERCC1 has been shown to be a critical gene in DNA repair • NER pathway - recognizes and removes platinum-induced DNA adducts • DSBR pathway- repairs radiation-induced damage (Bohanes et al. Clin Colorectal Cancer. In Press;Ahmad et al. Moll Cell Biol 2008) • A prospective clinical trial demonstrated in advanced NSCLC that customized therapy based on ERCC1 mRNA levels increased the response rate to cisplatin-based chemotherapy administered to patients with low ERCC1 mRNA levels (Cobo et al. J ClinOncol2007)

5. ERCC1 down-regulation sensitizes cells to all platinum compounds Cell viability (%) Younet al. Cancer Res 2004

6. Tumor ERCC1 mRNA Levels

7. ERCC1 in Gastrointestinal Malignancies

8. ERCC1 in Gastrointestinal Malignancies

9. ERCC1 in Gastrointestinal Malignancies

10. ERCC1 in Gastrointestinal Malignancies

11. ERCC1 in Gastrointestinal Malignancies

12. Main Objective • To validate prospectively ERCC1 gene expression (predefined cutoff of 1.7) as a biomarker predicting outcome in patients treated with oxaliplatin-based chemotherapy in combination with radiation in the SWOG S0356 trial.

13. Secondary Objectives • To explore the association between outcome and : • Other baseline mRNA levels of genes involved in DNA repair (XPD, RRM1) and 5-FU metabolism (TS, TP, DPD and GSTP1) • Functional polymorphisms in genes involved in DNA repair (ERCC1, XPD, RAD51, XRCC1, XRCC3) and 5-FU metabolism (TS, MTHFR, GSTP1)

14. SWOG S0356 Treatment Design Tumor biopsy D8 D15 D22 D29 D36 D43 D1 CTX + EBRT Surgery D8 D15 D22 D29 D36 D43 D1 CTX PI 5FU 180 mg/m2/day OHP 85 mg/m2 EBRT 180 cGy/day

15. Inclusion Criteria • Esophageal or gastroesophageal junction adenocarcinoma • Patients > 18 years • Clinical stage II or III; Zubrod PS ≤ 2 • Endoscopic ultrasound only for tumors that do not form a clear mass on CT scan • Pre-tx PET scans mandatory • Thoracic esophagus or gastroesophageal junction • Tumors < 2 cm into the gastric cardia (Siewert I-II) • Standard hematologic/non-hematologic parameters

16. Patient Characteristics February 2005 to August 2008 98 patients registered 6 ineligible patients* 92 eligible for clinical outcome evaluation and this study *2 squamous tumors, 2 met disease, 2 with biopsy/scans performed >28 days from protocol entry

17. Pathologic Response • 26 (28.2%) patients = pCR (centrally confirmed) • 10 patients had either Tin situN0M0 or T1N0M0 Leichman et al.Annual Meeting of the ASTRO, San Diego, 2010

18. Progression-Free Survival (PFS) Overall Survival (OS) 100% 100% 80% 80% 60% 60% 40% 40% 20% 20% 0% 0% 0 0 2 2 4 4 6 6 Years After Registration Years After Registration N Events Median in Months 20.8 92 55 2-year PFS 45.6% N Events Median in Months 92 47 33.7 2-year OS 55.4% Median follow-up of 36.8 months Leichman et al.Annual Meeting of the ASTRO, San Diego, 2010

19. Gene Expression RNA Extracted RNA Laser Capture Micro-dissection Reverse Transcription cDNA PCR with TaqMan® Data Analysis

20. Statistical Considerations • Cox regression used for univariate analyses of biomarker association with outcome • Adjustment for baseline factors did not affect any results • Recursive partitioning used to look for optimal cut points for continuous markers • Also used for building “regression trees” • P values adjusted for the multiple comparisons implied by this technique

21. Genes Analyzed for mRNA Levels 92 patients -> 90 pre-treatment samples -> 55 with sufficient tumor tissue

23. 100% 80% 60% 40% 20% 0% 0 1 2 3 4 5 Years After Registration PFS by ERCC1 mRNA Levels N Median 2-year PFS p* HR (95% CI) ≤1.7 22 NR 67% 1.0 (ref) >1.7 31 14.8 mos 17% 2.97 (1.37-6.45) 0.0058 Median follow-up of 36.8 months

24. 100% 80% 60% 40% 20% 0% 0 1 2 3 4 5 Years After Registration OS by ERCC1 mRNA Levels p* N Median 2-year OS HR (95% CI) ≤ 1.7 22 NR 72% 1.0 (ref) > 1.7 31 22.4mos 37% 2.32 (1.01-5.31) 0.047 Median follow-up of 36.8 months

25. Other Results • An analysis of PFS using a recursive partitioning model found the optimal split for ERCC1 gene expression to be 1.66 (adjusted p=0.04). • ERCC1 mRNA levels were not associated with pCR • None of the other accessed mRNA levels were associated with outcome (either univariate or in recursive partitioning) • DNA-repair: XPD, RRM1 • Platinum detoxification: GSTP1 • 5-FU metabolism: TS, TP, DPD

26. Genotyping • DNA was extracted from blood (Qiagen, CA, USA). • Genotyping was performed by using PCR-RFLP technique. • Performed in 91 patients (out of 92 eligible) • ERCC1 118C>T • ERCC1 8092C>A • GSTP1 Ile105Val • RAD51 135G>C • XPD 156A>C • XPD Lys751Gln • XRCC1 Arg391Gln • XRCC3 Thr241Met • MTHFR 677C>T • MTHFR 1298A>C • TS 3’UTR 6bp+/6bp- • TS 5’UTR VNTR • TS 5’UTR G/C SNP None are significant after adjusting for multiple comparisons

27. Summary • In this trial, using oxaliplatin and protracted-infusion 5FU with radiation, low intratumoral ERCC1 from the primary esophageal cancer predicted for PFS and OS. • Pre-established ERCC1 mRNA cutoff of 1.7 was confirmed in this trial. • This pre-specified cutoff was further validated by using recursive partitioning (optimal cutoff of 1.66). • Genomic polymorphisms analyzed were not associated with outcome in this study.

28. Study Limitations • Small sample size. • Lack of sufficient tumor tissue collection. • No differentiation between clinical stage II and clinical stage III patients

29. Conclusions • ERCC1 mRNA level is a very promising pre-treatment biomarker in patients with localized esophageal and gastroesophageal adenocarcinoma treated with trimodality treatment. • Biomarker studies are feasible within cooperative groups. • Based on these and published data, the SWOG is planning a prospective biomarker-driven clinical trial.

30. Acknowledgments The patients and their families and Investigators who participated in SWOG S0356 Response Genetics: Kathleen D. Danenberg. SWOG Statistical Center: Bryan Goldman. Funded by SWOG award 5-U10-CA058882-18 Pierre Bohanes was partially funded by Cancer & Solidarité Fondation