E. Castanon Alvarez

1485PDOUTPATIENT MANAGEMENT OF CANCER-RELATED PULMONARY EMBOLISM: A PROPENSITY SCORE-MATCHED ANALYSIS OF 803 PATIENTS FROM THE EPIPHANY STUDY

ABSTRACT Aim: Outpatient management of ‘low-risk’ pulmonary embolism (PE) is suggested in clinical guidelines (Evidence 2B) although the definition of ‘low-risk’ varies according to different authors. We aimed here to assess whether home versus hospital management were equivalent to treat ‘low-risk’ cancer-related PE. Methods: EPIPHANY is an observational, multicenter study to assess prognostic factors and patterns of care on cancer-related PE. The definition of low risk used in this study comprised: systolic blood pressure ? 100 mmHg, arterial oxygen saturation≥90%, respiratory rate ?30 breaths/min, pulse ?110 beats/min, no sudden or progressive dyspnea and absence of a clinically relevant hemorrhage. We used propensity score matching to compare the efficacy and safety of outpatient versus inpatient therapy. Results: We included 803 consecutive patients, 390 (48%) of whom were classified as low risk. Ambulatory management was used in 201 (51%) of them, whereas 189 (49%) were treated in hospital. Unadjusted rates of acute complications, 30- and 90- mortality were significantly lower in outpatients (Table 1). However, incidental detection, previous ambulatory status, absence of symptoms and oxygen saturation ≥96% were significantly more frequent in patients treated at home. After adjusting for imbalances, absolute difference estimates for 30- and 90 days mortality were: -0.6% (95% CI, -3.7 to+2.5%) and -2.7% (95% CI, -9.4% to +3.9%), respectively. Rates of venous rethrombosis and major bleeding events were also similar. Table 1. Outpatients (95% CI) Inpatients (95% CI) Short-term complications 2.5% (1- 5.7%) 6.3% (3.7- 10.7%) 30 days mortality 1.4% (0.5- 4.2%) 8.5% (5.2- 13.3%) 90 days mortality 7.4% (4.5- 11.9%) 20.1% (15- 26.3%) 30 days major bleeding 1.8% (0.7- 4.5%) 1.8% (0.7- 4.6%) Venous retrombosis 5.9% (3.4- 9.8%) 6.7% (4.1- 11%) Conclusions: Outpatient management of low-risk cancer-related PE appears to be safe and effective. A randomized trial is needed to confirm these data and assess outcomes in specific subgroups. Disclosure: All authors have declared no conflicts of interest.

4CPS-116 Alectinib hepatic-tolerance in patients with hepatotoxicity background with other anaplastic lymphoma kinase (alk)-inhibitors

Background ALK-inhibitors are indicated in adult patients with ALK-positive advanced NSCLC, with crizotinib being the first choice. Hepatotoxicity has been described for crizotinib and ceritinib. Purpose To describe a case of alectinib hepatic-tolerance in patients with an hepatotoxicity background with other ALK-inhibitors. Material and methods Data were obtained by review of the electronic medical records. Karch-Lasagna, Naranjo and WHO-UMC algorithms have been used. Results A 76-year-old male diagnosed with ALK-positive advanced NSCLC (2016) began crizotinib 250 mg twice daily on 27 October 2016 with basal laboratory hepatic parameters in the normal range. An initial brain and thoracic response was observed but 36 days from the start of crizotinib (3 December 2016) a marked elevation appeared in transaminases ALT 1542UI/L (37.6xULN) and AST 684UI/L (18.5xULN) and a minimal rise in total bilirrubin 2.0 mg/dL (1.67xULN). Crizotinib was discontinued and AST recovered its normal range within 24 days and ALT within 32 days. Then ceritinib 750 mg daily was started (3 January 2017) with frequent evaluations of liver function, showing a progressive increase in transaminases from day +8 until 8 March 2017 with maximum values of ALT 214UI/L (5.2xULN) and AST 128UI/L (3.5xULN). Ceritinib was stopped despite the patient presenting brain and thoracic response. Treatment was changed to pembrolizumab 200 mg every 3 weeks and 4 months’ later was discontinued for brain progression. On 12 July 2017 the patient began alectinib 600 mg twice daily with exhaustive hepatic monitoring. Three months’ later he presented an adequate treatment tolerance, without signs of clinical progression and transaminasesin in the normal range. Karch-Lasagna and Naranjo algorithms established a ‘probable’ relationship between hepatotoxicity and crizotinib/ceritinib. The WHO-UMCalgorithm established this relationship as ‘probable’ to crizotinib and ‘certain’ to ceritinib. In all cases there was a temporal correlation of the facts and an apparent absence of another factor responsible for liver damage. Conclusion Alectinib may be a therapeutic option in patients with ALK-positive NSCLC who have developed hepatic toxicity to other ALK-inhibitors. Further follow-up is needed to ratify this statement. Hepatic toxicity to ALK-inhibitors has frequently a reversible pattern and transaminases appear to be the most sensitive marker. References and/or Acknowledgements (ULN)=upper limit normal No conflict of interest

3D waterfall plots: a better graphical representation of tumor response in oncology

Compared to a waterfall plot, swimmer plots additionally provides us with information both on the quality of response (date of best response) and quantity (duration of response in time). Nevertheless, the information is given in terms of categorical variables, such as complete or partial response, and does not represent the magnitude of the response and how it numerically changes over time. Furthermore, only those patients who respond are represented in this type of graphic, not reporting data from patients that have progressive or stable disease. In contrast, waterfall plots do provide individual patient information as they show a more accurate estimation of the response (adding relative quantification of tumor size change compared to baseline measures) (3). However, waterfall plots show “static” information since only the best response presented is reported. In order to overcome this caveat, spider plot graphics were designed and they show the individual response of each patient over time. So spider plots, not only allows us to follow the change in tumor dimension according to RECIST criteria, but also to observe the evolution and response duration and indirectly, to determine the length of the best response.

1634PINHIBITOR OF DIFFERENTIATION-1 (ID1) AND ID3 EXPRESSION CORRELATES WITH EPITHELIAL-MESENCHYMAL TRANSITION (EMT)-RELATED PROTEINS (EMTRP) IN NON-SMALL CELL LUNG CARCINOMA (NSCLC)

ABSTRACT Aim: We previously reported the prognostic impact of Id1 in all NSCLC stages and of the combined Id1/Id3 expression in stage III-N2 patients undergoing radical chemo-radiotherapy. Further, we found that Id1 expression in the tumor microenvironment is crucial for the development of liver metastasis from lung cancer in an in-vivo model. In other neoplasms, Id genes are considered to be regulators of EMT-facilitated metastasis. Here we study Id1 and Id3 combined expression in a large NSCLC cohort and its correlation with EMTrp and overall survival (OS). Methods: Three tissue microarrays of 538 NSCLCs including 284 squamous-cell carcinomas (SCC) and 254 adenocarcinomas (AC) were stained by immunohistochemistry using the following antibodies: Sox10, Beta-catenin, E-cadherin, Vimentin, Slug, Podoplanin, S6, ERCC1, Id1 and Id3. Immunoreactivity was H-scored for intensity (range 0 to 3) multiplied by percentage of positive cells and computed by Spearmans Rho correlation test. Results: Among SCC histology a significant positive correlation was observed between the combined expression of Id1 and Id3 and the expression of Sox10, Beta-catenin, E-Cadherin, Slug, Podoplanin and ERCC1. In the AC subset, significant positive correlations were found between Id1/Id3 and Beta-catenin, E-Cadherin, Vimentin, pS6, and ERCC1. As expected, Id1/Id3 expression showed a negative correlation with OS (r= -0.157; p=0.013). Protein SCC AC (n.s; not significant) r p value r p value Sox10 0.146 0.018 n.s. n.s. Beta-Catenin 0.218 0.001 0.161 0.01 E-cadherin 0.128 0.038 0.150 0.017 Vimentin n.s. n.s. 0.147 0.019 Slug 0.187 0.002 n.s. n.s. Podoplanin 0.127 0.039 n.s. n.s. S6 n.s. n.s. 0.150 0.014 ERCC1 0.265 0.001 0.185 0.003 Conclusions: In NSCLC, the combined expression of Id1 and Id3 negatively correlated with OS and positively with several key EMTrp, suggesting a possible key regulation of the EMT process favoring vessel infiltration and distant organ metastasis, respectively. Disclosure: All authors have declared no conflicts of interest.

712PA THREE-STEP STRATEGY OF INDUCTION CHEMOTHERAPY, CHEMO-RADIOTHERAPY AND SURGERY IN LOCALLY ADVANCED PANCREATIC CANCER (LAPC) PATIENTS. ROLE OF A NONLINEAR MIXED EFFECTS MODELING TO PREDICT OUTCOME

ABSTRACT Aim: The optimal strategy for patients with LAPC remain a therapeutic challenge. A growing evidence suggests that both, pts with borderline resectable and unresectable tumors may benefit from a multimodal approach aimed at improving resectability and survival times. In the present work our experience after a long-term follow-up period is reported. Methods: From December 2005 to July 2011, 67 histologically confirmed LAPC, endoscopic ultrasound (EUS) staged T3-4 and/or N+ were retrospectively analysed. They received induction gemcitabine/oxaliplatin- based chemotherapy followed, in case of radiological response or stable disease, by chemo-radiotherapy (50.4 Gy concurrently with daily capecitabine and weekly oxaliplatin). Salvage surgery was performed when technically feasible. We have applied a nonlinear mixed effects (NLME) modeling to evaluate the impact of dynamic changes in tumor size, Ca- 19.9, neutrophil-lymphocyte ratio and platelet-lymphocyte ratio on the clinical outcome of these patients. Results: The median age was 63 years (range 35-85). Male to female 36/31. Thirty eight pts (57%) completed the whole program (group A), whereas 27 (40%) received chemo and radioterapy but were not elegible for surgery (Group B). Two pts (3%) progressed after induction chemotherapy (Group C). EUS staged T4 or N+ was found in 20 (30%) and 24 (36%) respectively. Toxicity profile was mild, with no grade 4 toxicity being documented. On an intent to treat basis, R0 resection rate was 57%; ypT0ypN0 were observed in 11 (29%) of patients. Among resected patients, local and distant failure rates were 5% and 55%, respectively. The liver was the most frequent site of relapse 21(43%) pts. After a median follow up of 23 months (range 4 to 102), median PFS was 21, 10 and 1 month in groups A, B and C respectively (p = Conclusions: Our date suggest that this three-step strategy is feasible and active in LAPC patients. The NMLE population modeling will be presented at the meeting Disclosure: All authors have declared no conflicts of interest.

645PINDUCTION CHEMOTHERAPY, CHEMO-RADIOTHERAPY AND SURGERY IN LOCALLY ADVANCED GASTRIC CANCER (GC) PATIENTS: LONG TERM RESULTS FROM A SINGLE INSTITUTION

ABSTRACT Aim: Multimodal therapy is the standard of care in locally advanced esophagogastric adenocarcinomas. However, most trials include both, GEJ and GC. We aimed to specifically rule out in gastric cancer patients the R0 rate, pathological response degree, patterns of recurrence and long-term outcomes when treated with induction chemotherapy (ICT), followed by chemoradiotherapy (CRT) and salvage surgery. Methods: Patients (pts) with CT scan and endoscocopic ultrasound (EUS) T2-4 and/or N+ M0 GC were retrospectively analyzed. The neoadjuvant strategy consisted of 3-4 cycles of chemotherapy followed by CRT (weekly chemotherapy concurrently with daily external beam radiotherapy up to 45 Gy). Surgery was scheduled 4 to 6 weeks after the end of CRT. Pathological response was graded according to the Becker criteria. We have applied a nonlinear mixed effects (NLME) modeling to evaluate the impact of dynamic changes in tumor size, neutrophil lymphocyte ratio (NLR) and platelet lymphocyte ratio (PLR) on the clinical outcome of these patients. Results: From November 2004 to July 2013, 60 pts [M/F: 42/18, median age 60 (range 36-76), T2/T3/T4a/T4b: 2/24/26/8; N0/N + , 6/54; Diffuse/Intestinal: 35/25] were retrospectively analyzed. Thirty-four and 54 patients were stages by EUS as T4 or N +, respectively. Median preoperative NLR and PLR were both 2.43 and 148. Forty-nine pts underwent radical surgery (total/partial gastrectomy: 33/16) with an R0 resection rate of 92%. Pathological response according to Becker criteria was grade Ia (12%), grade Ib (39%), grade II (22%) and grade III (2%). pN0 was achieved in 26 pts (53,1%). Median harvested nodes were 13. Median DFS for pN0 and pN+ pts was NR and 23 months, respectively (p = 0.01), whereas median OS for pN0 and pN+ was NR and 30 months, respectively (p = 0.009). Among resected pts, distant failures were reported 22 pts, most of them in the peritoneal cavity (55%). After a median follow-up of 30 months (range 4-95 months) the 5-year overall and disease-free survival, were 31% and 39%, respectively. Conclusions: Our data suggest that a three-step strategy is feasible and active in locally advanced gastric cancer patients. The NMLE population modeling to evaluate the impact of dynamic markers on the clinical outcome will be presented at the meeting. Disclosure: All authors have declared no conflicts of interest.

644PINDUCTION CHEMOTHERAPY FOLLOWED BY CHEMORADIOTHERAPY IN LOCALLY ADVANCED ESOPHAGOGASTRIC ADENOCARCINOMA. HAS THE LOCATION OF THE PRIMARY TUMOR ANY INFLUENCE ON PATIENTS' OUTCOME? A RETROSPECTIVE ANALYSIS

ABSTRACT Aim: Multimodal therapy is the standard of care in locally advanced esophagogastric adenocarcinoma. However, most trials include both, gastric (GC) and gastroesophageal junction (GEJ) cancers. We aimed to rule out whether patients (pts) with either locally advanced GEJ (group A) or GC (group B) have different outcomes when treated with a neoadjuvant approach based on induction chemotherapy (ICT), chemoradiotherapy (CRT) and surgery. Methods: EUS-T2-4 and/or N+ M0 GEJ or GC adenocarcinoma patients were scheduled to receive preoperative therapy (3-4 cycles of chemotherapy followed by concurrent CRT). Surgery was scheduled 4 to 6 weeks after the end of CRT. Pathological response was graded according to the Becker criteria. Patterns of recurrence, progression-free survival (PFS) and overall survival (OS) were also evaluated. Results: From november 2004 to July 2013, 87 pts (27 in group A, 60 in group B) met the specified criteria. Pts characteristics were; Group A; M/F; 23/4; EUS-T3 52%, T4 40,7%, N+ 95%, median age 56 (31-81); Diffuse/Intestinal 55%/44%. Group B; M/F; 42/18; EUS-T3 40%, T4 56%, N+ 95%; median age 60.5 (36-76); Diffuse/Intestinal 57%/42%. Seventy-four pts underwent surgery (group A/ B; 92.6%/82%). Pathological response (Group A/B); Grade Ia-Ib according to Becker criteria (72%/51%), pN0 (60.3%/53%). R0 resection (Group A/B); (96%/94%). Among resected pts we analysed the impact of tumor location (GC vs GEJ) on known prognostic factors such as Lauren classification and degree of pathological response regarding PFS and OS. In group A, median OS and PFS did not difer according to Lauren neither classification nor pN status. In contrast, median PFS varied according to Becker response (Ia-b: Not Reached; II-III: 13m; p = 0.02). In group B, we found differences in PFS and OS according to; a)- Lauren Classification; Intestinal GC vs Diffuse GC pts [median PFS; NR vs 13.6m; p = 0.028, median OS: NR vs 15.3m; p = 0.026]. b)pN status (-/+) [median PFS; NR vs 18m; p Conclusions: When treated with ICT followed by CRT, GEJ and GC pts have different outcomes in terms of pathological response degree, patterns of recurrence and survival times. Disclosure: All authors have declared no conflicts of interest.

182PTARGETING THE EXPRESSION OF THE INHIBITOR OF DIFFERENTIATION (ID) GENES IN TUMOR MICROENVIRONMENT (TME) AND TUMOR CELLS IS CRUCIAL IN THE PREVENTION OF LIVER METASTASIS (LM) FROM LUNG CANCER

ABSTRACT Aim: Around 30% of non-small cell lung cancer (NSCLC) patients present LM during the disease course causing a negative clinical impact on survival and quality of life. The expression of certain genes in cancer cells and also TME gene expression might be crucial for allowing tumor cells to spread to the liver. According to this hypothesis Id1 and Id3 genes, part of the signature that facilitates breast cancer cells to disseminate to the lungs, might be determinant for NSCLC LM development. Methods: First we validated Id1 as a significant and independent prognostic factor among patients with adenocarcinoma of the lung. Recently, we have communicated the development of a mouse model of LM from lung cancer in which the only absence of Id1 expression in TME of Id1-knockout (KO) animals was sufficient to avoid LM formation by 50%. Now we aimed to evaluate the additional impact of Id1 and Id3 genes deficiency in tumor cells on LM generation. Therefore, 2 cohorts including 23 mice were compared; Id1 wild-type (WT) female mice (n = 12) vs. KO female animals (n = 11). In both groups of mice 500,000 Lewis Lung Carcinoma cells (LLC) either Id1 homozygously deficient (Id1-/-) and Id3 WT (Id3 + /+) or Id1-/- and Id3 heterozygously deficient (Id3 + /-), generated through gene silencing, were intrasplenically injected. Thereafter, both groups of mice were weekly monitored with FDG-micro-positron emission tomography scans (mPET) for LM formation. Animals were sacrificed (and tissues microscopically analyzed) by the time LM were developed and clinical deterioration was evident, according to our Animal Ethics Committee approved protocol. Mice genotype Positive PET at week 2 p value WT 41.7% KO 0% 0.02 Microscopic LM at necropsy WT 83.3% KO 36.4% 0.03 Results: LM were significantly more likely to appear among Id1 WT mice compared to Id1 KO mice with a relative risk of 2.29 (IC95% 1.01-5.22). Results are summarized in Table 1 Conclusions: In this model, the absence of Id1 in TME of KO mice in combination with the lack of Id1/Id3 expression in intrasplenically injected lung cancer cells showed to be a more significantly efficient strategy to prevent LM formation. Disclosure: All authors have declared no conflicts of interest.