2090 Prognostic score (REGOSCORE) for survival after Regorafenib (RE) treatment for patients (pts) with pretreated metastatic colorectal cancer (mCRC)

S358 objective response to induction treatment appears to have a maximum benefit in terms of both PFS (HR: 0.40; median 3.0 vs. 2.7 months, p = 0.009) and OS (HR: 0.40; median 24.5 vs. 15.1 months, p = 0.069). This evidence points for the first time to the possibility of exploiting the known link between response to chemotherapy and presence of an active immune response, administering an immunomodulatory treatment as maintenance at the time of maximal response. Conclusions: The results of IMPACT suggest that patients who had an objective response to prior induction chemotherapy benefited most from MGN1703 maintenance. This may allow identification of mCRC patients with increased benefit when treated with an immunomodulator. This information was used to design the phase 3 IMPALA study, currently randomizing patients with objective response after standard induction therapy to maintenance with MGN1703 or standard care. Conflict of interest: Ownership: B. Wittig owns shares of MOLOGEN AG. Board of Directors: A. Zurlo is CMO of MOLOGEN AG. 2090 POSTER Prognostic score (REGOSCORE) for survival after Regorafenib (RE) treatment for patients (pts) with pretreated metastatic colorectal cancer (mCRC) A. Kramar1 , B. Paule2 , P. Burtin3 , D. Tougeron4 , F. Desseigne5 , L.M. Dourthe6 , J. Wallet7 , P.L. Etienne8 , L. Mineur9 , C. Becuwe10 , P. Maes11 , J. Dauba12 , P. Michel13 , A. Hollebecque3 , B. Linot14 , J.F. Morere2 , C. De la Fouchardiere15 , J.M. Phelip16 , T. Andre17 , A. Adenis18 . 1 Centre Oscar Lambret, Head of methodology and biostatistics unit, Lille, France; 2 CHU Paul Brousse, Medical Oncology, Villejuif, France; 3 Institut Gustave Roussy, Medical Oncology, Villejuif, France; 4 University Hospital, Gastroenterology, Poitiers, France; 5 Centre Leon Berard, Medical Oncology, Lyon, France; 6 Clinique Sainte Anne, Medical Oncology, Strasbourg, France; 7 Centre Oscar Lambret, Methodology and Biostatistics, Lille, France; 8 Clinique de l’Armoricaine, Medical Oncology, Saint-Brieuc, France; 9 Institut Sainte-Catherine, Radiation and Medical Oncology, Avignon, France; 10 Polyclinique de Gentilly, Medical Oncology, Nancy, France; 11 Clinique du Bois, Medical Oncology, Lille, France; 12 Mont de Marsan Hospital, Medical Oncology, Mont de Marsan, France; 13 University Hospital, Gastroenterology, Rouen, France; 14 Institut de Cancerologie de l’Ouest, Medical Oncology, ´ Angers, France; 15 Centre Leon Berard, Medical Oncology, Lyon, France; 16 University Hospital, Gastroenterology, Saint-Etienne, France; 17 Saint Antoine Hospital, Medical Oncology, Paris, France; 18 Centre Oscar Lambret, Medical Oncology, Lille, France Background: RE provides a 25% risk reduction of death over placebo and a 51% risk reduction of progression (PD) in refractory mCRC pts (Lancet Oncol 2013). Despite such clinical benefit, half of these heavily pretreated pts get limited benefit from RE, as they present with PD within 2m after RE start. It is thus worth identifying characteristics of pts who benefit or not from RE. Methods: Prognostic factor analyses on survival were performed on the 654pts included in the REBECCA study (NCT02310477), which was designed to evaluate survival and safety associated with RE as used in clinical practice, A backward stepwise regression Cox model was used to identify significant variables. A score was obtained from the sum of the relative weights of the model coefficients. Then the number of categories was reduced from hierarchically defined dummy indicator variables. Categories with similar outcomes were regrouped together (Wald test). Other potential prognostic variables were then added and tested for significance using the likelihood ratio statistic. Results: Median RE duration was 2.2m. Median OS (mOS), and PFS (mPFS) were 5.6m (95% CI: 4.9−6.3), and 2.7m (95% CI: 2.6−2.8), respectively. Total scores (TS) were obtained from the following variables associated with shorter OS: time from initial diagnosis of mets <18m (+2), PS=1 (+2), 2 (+4), <160mg initial RE dose (+1), presence of liver mets (+2), 3+ metastatic sites (+1), and mutated KRAS (+1). TS (0 to 10) were then reduced to 3 prognostic groups: good OS (TS <4, 34%pts, mOS=9.2m), intermediate OS (TS=4 or 5, 42%pts, mOS=5.2m), poor OS (TS=6+, 24%pts, mOS=2.5m). A similar score was built for PFS with the following variables: PS, time from initial diagnosis, initial RE dose, liver mets, timing of mets, and previous bevacizumab exposure. TS were reduced to 3 groups: good PFS (TS <4, 15%pts, mPFS=5m), intermediate PFS (TS=4 to 7, 69%pts, mPFS =2.7m), poor PFS (TS=8+, 16% pts, mPFS=1.4m). Conclusions: Our REGOSCORE prognostic model was able to categorize RE-treated mCRC pts. Before being helpful in making individual therapeutic decisions, the predictive value of this score has to be externally validated in placebo-controlled cohorts. No conflict of interest.

Abstracts 2091 POSTER A Japanese cohort study of first-line chemotherapy for metastatic colorectal cancer: The EMERaLD study K. Takeda1 , T. Amano2 , Y. Ohashi3 . 1 Osaka City Genenal Hospital, Medical Oncology, Osaka, Japan; 2 Hokkaido University Hospital, Clinical Research and Medical Innovation Center, Sapporo, Japan; 3 Chuo University, Integrated Science and Engineering for Sustainable Society, Tokyo, Japan Background: Observational cohort studies are crucial in establishing databases to assess the clinical status of patients. We planned a cohort study to investigate the first line standard chemotherapy and metastatic colorectal cancer in Japan. The overall survival at the two-year follow-up and the safety data were evaluated in this large cohort study. Material and Methods: A large cohort study of first-line chemotherapy in patients with metastatic colorectal cancer was conducted in order to establish a database for use in future clinical practice in Japan. Patients treated with first-line chemotherapy including oxaliplatin and bevacizumab and those who started treatment in or after January 2010 were eligible. The primary objective was to evaluate overall survival, and the secondary outcomes were response rate, progression-free survival, and safety. Subgroup analyses for overall survival were also performed. This study was sponsored by the Public Health Research Center Foundation, Comprehensive Support Project for Oncology Research (CSPOR). Clinical trial registration: UMIN000006392 Results: A total of 1353 patients were recruited from 132 centers in Japan between October 2010 and September 2011, and 1146 (84.7%) pieces of informative data were analyzed during a two-year follow-up period. The characteristics of these patients were as follows: male/female, 694/452; median age, 65 years (range, 26−89); Eastern Cooperative Oncology Group performance status (0/1/2/3), 959/168/15/4; site of primary tumor (colon/rectum/others), 633/504/9; site of metastatic disease (redundantly) (liver/lung/others), 696/338/565; and KRAS status (wild-type/mutant/ unknown), 444/321/381. All patients were treated with bevacizumabcontaining chemotherapy. Platform regimens were capecitabine plus oxaliplatin (CapeOX), 5-FU plus oxaliplatin (FOLFOX), and others, in 617, 494, and 35 patients respectively. The two-year survival rate for all patients was 59.2% (95% confidence interval [CI], 55.6–62.6%), and subgroup analyses showed that the two-year survival rates were 59.7% (95% CI, 54.5–64.4%), 55.0% (95% CI, 49.6–60.1%), and 79.8% (95% CI, 58.0–91.1%) for CapeOX, FOLFOX, and others, respectively. There were no differences among these three subgroups. Grade 3/4 adverse events related to bevacizumab included hypertension (4.5%), thromboembolism (2.1%), digestive tract perforation (1.9%), bleeding (0.78%), and proteinuria (0.34%). There were two treatment-related deaths, associated with thromboembolism and bleeding. Conclusions: Standard combination chemotherapy with bevacizumab was effective with tolerable toxicity in patients with metastatic colorectal cancer in Japan. Conflict of interest: Other Substantive Relationships: Chugai Pharmaceutical Co., Ltd., Yakult Honsha Co., Ltd. 2092 POSTER Updated analysis: Observational cohort study of 1st line bevacizumab combined with chemotherapy in metastatic colorectal cancer (HGCSG0802): Comparison of infusional FU/oxaliplatin(OX)+BV and oral FU/OX+BV H. Nakatsumi1 , S. Yuki1 , Y. Kawamoto2 , T. Muranaka2 , K. Hatanaka3 , T. Kato4 , T. Meguro4 , M. Nakamura5 , I. Iwanaga6 , M. Uebayashi6 , M. Tateyama7 , K. Eto8 , M. Kudo9 , S. Kato10 , H. Okuda11 , S. Sogabe12 , K. Miyashita13 , Y. Sakata14 , N. Sakamoto1 , Y. Komatsu2 . 1 Hokkaido University Hospital, Gastroenterology and Hepatology, Sapporo, Japan; 2 Hokkaido University Hospital, Cancer Center, Sapporo, Japan; 3 Hakodate Municipal Hospital, Gastroenterology, Hakodate, Japan; 4 Hokkaido Gastroenterology Hospital, Internal Medicine, Sapporo, Japan; 5 Sapporo City General Hospital, Gastroenterology, Sapporo, Japan; 6 Japanese Red Cross Kitami Hospital, Gastroenterology, Kitami, Japan; 7 Tomakomai Nisshou Hospital, Internal Medicine, Tomakomai, Japan; 8 Tomakomai City Hospital, Gastroenterology, Tomakomai, Japan; 9 Sapporo Hokuyu Hospital, Gastroenterology, Sapporo, Japan; 10 Iwamizawa Municipal General Hospital, Gastroenterology, Iwamizawa, Japan; 11 Keiyukai Sapporo Hospital, Medical Oncology, Sapporo, Japan; 12 Kushiro Rosai Hospital, Medical Oncology, Kushiro, Japan; 13 Aiiku Hospital, Gastroenterology, Sapporo, Japan; 14 Misawa City Hospital, CEO, Misawa, Japan Background: A few reports have shown no deference between the efficacy of infusional FU and that of oral FU (Capecitabine/S-1) for colorectal cancer, and some studies have reported the non-inferiority between infusional