1. Clinical outcome according to tumor HER2 status and EGFR expression in advanced gastric cancer patients from the EXPAND study F. Lordick,* Y-K. Kang, P. Salman, S.C. Oh, G. Bodoky, G.P. Kurteva, C.D. Volovat, V. Moiseyenko, A. Sawaki, J.O. Park, V.A. Gorbunova, H. Goette, H. Melezinkova, C. Stroh, M. Moehler *University Cancer Center Leipzig, Leipzig, Germany (presenting author)

2. Introduction In the EXPAND study, the addition of cetuximab (EGFR-antibody) to first-line chemotherapy did not improve clinical outcome in patients with advanced gastric or GEJ cancer 1 High EGFR expression was associated with a survival benefit in advanced NSCLC patients treated with first-line chemotherapy + cetuximab 2 Tumor EGFR expression levels have not been investigated in gastric cancer patients in this treatment setting 1. Lordick F, et al. Lancet Oncol 2013;14:490-9 2. Pirker R, et al. Lancet Oncol 2012;13:33-42

3. Introduction (2) HER2 is a tyrosine kinase receptor in the same family as EGFR, patients with HER2 +ve advanced gastric cancer showed significantly improved OS when treated with trastuzumab (HER2 antibody) added to first-line chemotherapy 1 HER2 +ve status has been associated with poor prognosis in some studies of advanced gastric cancer patients, 2 whilst in other studies no prognostic or predictive value was reported 3 1.Bang Y-J, et al. Lancet Oncol 2010;376:687-97 2. Gravalos C, et al. Ann Oncol 2008;19:1523-29 3. Van Cutsem E, et al. J Clin Oncol 2012;30:2119-27

4. Objectives To assess treatment outcome according to tumor HER2 status (a pre-defined subgroup) and EGFR expression in EXPAND study patients

5. Methods The EXPAND study was a multinational, open-label randomized phase III study in locally advanced unresectable or metastatic gastric or GEJ cancer patients 1 Patients were randomly assigned to 3-week cycles of twice daily (days 1–14) capecitabine 1000 mg/m 2 and iv cisplatin 80 mg/m 2 (day 1) chemotherapy plus weekly cetuximab (day 1) 400 mg/m 2 initial infusion followed by 250 mg/m 2 /week thereafter, or chemotherapy alone The primary endpoint was PFS assessed blind by an independent review committee (IRC) Secondary endpoints included OS and best overall response (IRC) 1. Lordick F, et al. Lancet Oncol 2013;14:490-9

6. Methods (2) Tumor HER2 status was determined according to a scoring system for gastric cancer, 1 primarily by IHC (Dako HercepTest) HER2 +ve tumors were IHC 3+ or IHC 2+ and fluorescence in situ hybridization (FISH) +ve EGFR expression was assessed by IHC (Dako EGFR pharmDx). A continuous scoring system (scale of 0–300) was used to determine the level of EGFR expression 2 Biomarker status was correlated with clinical outcome (exploratory analysis) Statistical analysis was performed with SAS version 9.1.3 or later and R version 2.14.1 1. Lordick F, et al. Lancet Oncol 2013;14:490-9 2. Pirker R, et al. Lancet Oncol 2012;13:33-42

7. Results: HER2 analysis Baseline characteristics in the EXPAND ITT (n=904) and HER2 evaluable populations (n=679) were similar, no differences were noted between the treatment arms Imbalances in baseline characteristics in patients with HER2 +ve vs HER2 –ve tumors were found for sex, tumor location, histology, and peritoneal involvement (Table 1)

8. Table 1. Baseline characteristics Characteristics, n (%) ITT n=904 HER2 +ve n=144 HER2 –ve n=535 Sex Male Female 673 (74) 231 (26) 116 (81) 28 (19) 384 (72) 151 (28) Age, years ≥65 <65 280 (31) 624 (69) 43 (30) 101 (70) 166 (31) 369 (69) Ethnic group White Asian Other 490 (54) 339 (38) 75 (8) 81 (56) 56 (39) 7 (5) 319 (60) 174 (33) 42 (8) ECOG PS* 0 1 465 (51) 438 (48) 87 (61) 56 (39) 284 (53) 251 (47) Tumor location* Gastric GEJ 747 (83) 144 (16) 102 (73) 38 (27) 448 (85) 80 (15) Histological type* Intestinal Diffuse Mixed Unknown 311 (34) 170 (19) 50 (6) 373 (41) 60 (42) 15 (10) 12 (8) 56 (39) 187 (35) 107 (20) 31 (6) 208 (39) Peritoneal mets229 (25)22 (15)137 (26) *Missing data: ECOG PS, 1 HER2 +ve, tumor location, 4 HER2 +ve and 7 HER2 –ve, histological type, 1 HER2 +ve and 2 HER2 –ve.

9. Table 2. HER2 status and treatment outcome HER2 +veHER2 –ve CT n=72 CT + cetuximab n=72 CT n=254 CT + cetuximab n=281 Median OS, mo 95 % CI 14.0 11.3–17.1 13.3 10.9–15.5 9.7 8.6–11.0 9.2 8.1–10.5 Hazard ratio 95 % CI 1.04 0.71–1.53 0.98 0.82–1.19 Median PFS, mo 95% CI 6.9 5.5–8.2 5.6 4.3–6.0 5.5 4.2–5.7 4.6 4.2–5.6 Hazard ratio 95% CI 1.33 0.88–2.01 0.99 0.80–1.23 ORR, % 95 % CI 37.5 26.4–49.7 51.4 39.3–63.3 26.4 21.1–32.3 27.0 21.9–32.6 Odds ratio 95% CI 1.76 0.91–3.42 1.03 0.71–1.52 HER2 status did not seem to be predictive of cetuximab benefit In both treatment arms, patients with HER2 +ve vs HER2 –ve tumors had a longer median OS and a better ORR CT, chemotherapy

10. Figure 1. OS according to HER2 status and treatment Cet, cetuximab; CT, chemotherapy

11. Multivariable selection models A final multivariable selection model of OS included: Asian/ Non- Asian, peritoneal involvement, previous esophagectomy/ gastrectomy, number of metastatic sites, baseline alkaline phosphatise levels, histological subclassification, ECOG PS, HER2 status and treatment group Patients with HER2 –ve (vs HER2 +ve) tumors had a greater risk of death (adjusted HR 1.55, 95% CI 1.24–1.94) In a similar model, patients with HER2 –ve (vs HER2 +ve) tumors had reduced odds of response (adjusted odds ratio 0.48, 95% CI 0.32–0.72)

12. Links between patient subgroups and HER2 status Although not included in the multivariable model, gender and tumor location demonstrated a prognostic OS effect in univariable analysis (data not shown) Exploratory regression-tree analysis (investigating links between patient subgroups and HER2 status), identified GEJ site as the main factor associated with HER2 status: –In GEJ tumors (n=118), 68% were HER2 –ve –In gastric cancer subgroups, 79% of male patients (n=389) and 88% (n=161) of female patients were HER2 –ve Therefore, patient HER2 status and outcome (OS) may be linked via associations with prognostic baseline factors (tumor location, gender)

13. Results: EGFR analysis EGFR tumor expression was evaluable in 774/904 patients; 376 in the CT arm and 398 in the CT + cetuximab arm The EGFR IHC score was low (median 0, range 0–300, 75 th percentile 40)

14. Figure 2. EGFR IHC score and clinical outcome* *Reprinted from Lordick et al 1 with permission from Elsevier (see acknowledgments). Triangles and circles are median times plotted against the midpoint EGFR IHC score in overlapping sliding windows of patient subpopulations each containing 20% of the data (15% in the first and last windows). Windows change in 5% steps (0%-15%,0%-20%,5%-25% etc, and final 80%-100%, 85%-100%) 1. Lordick F, et al. Lancet Oncol 2013;14:490-9 Cet, cetuximab; CT, chemotherapy

15. Results: EGFR analysis continued In patient subgroups defined by a series of cut-off points from an IHC score of 10 upwards (rising incrementally by 10), there was a tendency for improved PFS, and OS when adding cetuximab to CT in patients with high tumor EGFR IHC scores (Figures 3 & 4) Findings were similar for tumor response (data not shown) However, this association was based on a very low number of patients. For example, for EGFR IHC scores  200, there were only 8 patients in CT + cetuximab group (Figure 2) No clear association between EGFR IHC score and patient prognosis was evident (Figure 2)

16. Figure 3. PFS: Hazard ratios in subgroups by EGFR IHC score Cet, cetuximab; CT, chemotherapy

17. Figure 4. OS: Hazard ratios in subgroups by EGFR IHC score Cet, cetuximab; CT, chemotherapy

18. Conclusions In EXPAND study patients, those with HER2 +ve vs HER2 –ve tumors were associated with better outcome irrespective of the treatment The prognostic value of HER2 status in this setting remains controversial and may be dependent on patient baseline and tumor prognostic factors Tumor EGFR expression was generally low Adding cetuximab to chemotherapy failed to improve outcome overall, but may benefit a very small proportion of patients with high EGFR tumor expression

19. Acknowledgments The trial was sponsored by Merck KGaA, Darmstadt, Germany The authors would like to thank patients, the investigators, co-investigators and the study teams at each of the participating centers and at Merck KGaA, Darmstadt, Germany Medical writing assistance was provided by Paul Hoban, Cancer Communications and Consultancy Ltd, Knutsford, UK and funded by Merck KGaA, Darmstadt, Germany Figure 2 was reprinted from Lancet Oncol 2013;14:490-9 Florian Lordick Yoon-Koo Kang, Hyun-Cheol Chung, Pamela Salman, Sang Cheul Oh, György Bodoky, Galina Kurteva, Constantin Volovat, Vladimir M. Moiseyenko, Vera Gorbunova, Joon Oh Park, Akira Sawaki, Heiko Goette, Helena Melezínková, Marcus Moehler, Capecitabine and cisplatin with or without cetuximab for patients with previously untreated advanced gastric cancer (EXPAND): a randomized open-label phase 3 trial (2013) with permission from Elsevier http://www.sciencedirect.com/science/journal/aip/14702045 http://www.sciencedirect.com/science/journal/aip/14702045