AGA Abstracts Disease duration of IBD at diagnosis of HLH was 8.1 years (range 0.1-34). Twenty five patients were under thiopurines in monotherapy (n=17), in combination with steroids (n= 3), anti-TNF (n=4) or both (n=1); one patient received anti-TNF in monotherapy and one patient was on steroids, tacrolimus and mycophenolate mophetil for a liver transplant. At the time of diagnosis, immunosuppressive drugs were discontinued in all patients. The mean time between the first symptoms and diagnosis was 14 days (range 3-48). The clinical features were: fever (39-40°C) (27/27), rash (9/27), lymphadenopathy (11/27), splenomegaly (11/26), dyspnea (8/27), neurological disorders (5/27). Bicytopenia was observed in 23/27 patients. The mean fasting triglycerides was 3.4 mmol / L (range 1.3 and 6.3 mmol / L) and the serum ferritin 12 496 mg / L (range 746 and 99 393). Hemophagocytosis was observed on bone marrow aspiration in 19/23 (83%) patients. A pathogen was identified as the precipitating cause of HLH in all subjects: CMV (n = 10), EBV (n = 10), HHV1 (n = 2), parvovirus B19 (n = 2), BK (n = 2), Klebsiella oxytoca and Enterobacter aerogenes (n = 1). Lymphoproliferative disorders [B lymphoma (n = 3) and T-cell lymphoma (n = 1)] were associated in 4 cases. A specific treatment for HLH was administered in 18 patients: immunoglobulins (n = 10), steroids (n = 11), etoposide (n = 7), cyclosporine (n = 1). Six patients died (22%), in whom HLH was related to EBV infection (n=4), CMV infection (n = 1) and HHV1 infection (n=1). At the latest follow-up (average 27 months, range 0.1 to 100), 3 patients were in clinical remission without treatment, 4 patients underwent surgical resection, 3 patients received 5ASA, 1 patient steroids and 5 patients anti-TNF therapy. Thiopurine was restarted in 4 patients. Conclusion: In all IBD patients under immunosuppres- sants, a febrile cytopenia should alert to the possibility of an HLH. Early diagnosis, identifica- tion of the causative agent and appropriate treatment could reduce the mortality of this life threatening conditions unfamiliar for the gastroenterologists. Tu1138 Efficacy of Vedolizumab in Crohn's Disease by Prior Treatment Failure in Gemini II, a Randomized, Placebo-Controlled, Double-Blind, Multicenter Study Stephen B. Hanauer, Brian G. Feagan, Donald G. MacIntosh, Jing Xu, Catherine Milch, Irving Fox, Michael D. Smyth Background: Vedolizumab (VDZ), an investigational, gut-selective antagonist of the α4β7 integrin, has been studied in the treatment of moderately to severely active Crohn's disease (CD) in patients (pts) in whom 1 or more therapies has failed (NCT00783692). In an exploratory analysis, we evaluated the efficacy of VDZ in subgroups of pts who had failed treatment with specific classes of drug therapy. Methods: Eligible adult pts had a CD Activity Index (CDAI) of 220-450; CRP .2.87 mg/dL, fecal calprotectin .250 mg/g, or ileocolonoscopy with ulceration consistent with CD; and intolerance or inadequate response to purine antimetabolites, anti-TNF, and/or corticosteroids (CS). After 2 induction doses of VDZ (wks 0 and 2), pts with clinical response ( ≥70-point decrease in CDAI from baseline) to VDZ at wk 6 were randomized 1:1:1 to VDZ 300 mg IV every 8 wks (Q8W), VDZ 300 mg IV every 4 wks (Q4W), or placebo (PBO) for 46 wks. The 2 primary endpoints of the induction trial were clinical remission (CDAI ≤150) and CDAI-100 response ( ≥100-point decrease from baseline CDAI) at 6 wks. The primary endpoint of the maintenance trial was clinical remission at 52 wks. Secondary endpoints were CDAI-100 response and CS-free remission at wk 52, and durable clinical remission (clinical remission at ≥80% of study visits including wk 52). Key endpoints were assessed in pts categorized by previous failure to prior treatment with an anti-TNF; an immunomodulator but not an anti-TNF; or CS only. The study was not powered for these subgroup analyses; the purpose was to analyze the trend of treatment effect within subgroups. Results: VDZ efficacy was generally consistent with the overall study results in the induction (Table 1) and maintenance (Table 2) trials. More pts achieved clinical remission with VDZ vs PBO in all subgroups at wk 6. At wk 52, VDZ-treated pts had improvements in clinical remission, CDAI-100 response, and CS-free clinical remission, compared with PBO, across the 3 subgroups. Although a greater proportion of pts achieved clinical remission with VDZ in the immunomodulator and CS-only failure subgroups than the anti-TNF failure subgroup, the treatment differences between PBO and VDZ were generally similar across the 3 subgroups. Patients with prior anti-TNF failure had higher rates of adverse events (90% for VDZ and 83% for PBO) than pts without prior anti- TNF failure (81% and 77%, respectively). Conclusions: The point estimates for the efficacy of VDZ in inducing clinical remission in pts with prior failure to anti-TNF, immunosuppres- sive agents, or CS, were consistently higher than those for PBO at wk 6. In pts with a clinical response at wk 6, patients in the VDZ groups had numerically higher rates vs PBO in maintaining clinical remission, CDAI-100 response, and CS-free clinical remission at wk 52, regardless of the type of prior treatment that had failed. Table 1. Clinical Remission and CDAI-100 Response at Week 6 *These subgroup analyses were prespecified to demonstrate internal consistency but were not fully powered nor adjusted for multiple testing. Therefore P values are not provided. S-772 AGA Abstracts Table 2. Clinical Remission, CDAI-100 Response, Durable Clinical Remission, and Corticosteroid-free Clinical Remission at Week 52 *These subgroup analyses were prespecified to demonstrate internal consistency but were not fully powered nor adjusted for multiple testing. Therefore P values are not provided. **Overall ITT population: PBO, N=82; VDZ Q8W, N=82; VDZ Q4W, N=80. Tu1139 Potential Synergism Between Anti-TNF and Thiopurine Therapy: Increased Thiopurine Metabolites by Anti-TNF Jessica L. Yoon, Brian L. Huang, Sundip S. Karsan, Gil Y. Melmed, Andrew Ippoliti, Marla Dubinsky, Manreet Kaur, Minh Nguyen, Puja Vora, Phillip Fleshner, Dermot P. McGovern, Stephan R. Targan, Eric A. Vasiliauskas, David Q. Shih Background: Thiopurines (azathioprine/AZA and 6-mercaptopurine/6MP) and antibodies (infliximab, adalimumab, and certolizumab) targeting tumor necrosis factor (TNF) have therapeutic efficacy in IBD. The SONIC Study showed that combination therapy (thiopurines + anti-TNF) resulted in higher rates of clinical remission and mucosal healing. The greater efficacy of combination therapy may be due to suppression of immunogenicity. We hypothe- size that enhanced benefit of combination therapy may also be due to interaction between the 2 drug classes. Aim: To determine whether addition of anti-TNF to AZA/6MP changes thiopurine metabolite levels. Methods: A retrospective chart review of 522 IBD patients treated at the Cedars-Sinai IBD center with AZA/6MP was performed. We identified 34 patients who were started on anti-TNF after having been ≥6 weeks of stable dose of AZA/ 6MP. Of these, 9 patients remained on the same dose of AZA/6MP and had metabolites measured between 8 and 30 weeks after starting anti-TNF. 6MMP preferential metabolizers were defined as patients with a 6TGN level ,230 and 6-MMP level ≥4000 pmol/8×10^8 RBC. Student t-test was performed and significance was set at p ≤0.05. Harvey-Bradshaw Index (HBI), biochemical parameters (WBC, ALT, ESR, CRP) and thiopurine metabolite levels were assessed. Results: Of the 9 patients (6 CD; 3 UC) on the same thiopurine dose before and after anti-TNF (infliximab n=7, adalimumab n=2), 4 patients exhibited 6MMP preferential metabolism. Initiation of anti-TNF led to increased 6TGN levels in 7 of 9 patients (Pre: 189±106 to Post: 284±106 pmol/8×10^8 RBC, P=0.03). The increased 6TGN level after anti-TNF led to reduced WBC from a pre-anti-TNF level of 7.3±2.0 to post-anti-TNF level of 4.8±2.0 x1000/μL, p=0.02. An increase in 6MMP levels was observed in 4 of 9 patients (3421±3770 to 5758±5744 pmol/8×10^8 RBC); in those 4, mean 6MMP levels increased from 6551 to 8460 pmol/8×10^8 RBC. In particular, three of the 4 patients who exhibit 6MMP preferential metabolism had 6MMP levels above the hepatoxic threshold (.5700 pmol/8×10^8 RBC) with consequent abnormal ALT elevations: 18 to 465; 25 to 35; and 14 to 54 U/L. Addition of anti-TNF to thiopurine reduced disease activity in all 9 (mean HBI reduction of 5±3.4) and reduced inflammatory markers in all except 1 patient: mean CRP and ESR reduction of 0.73±1.2 mg/DL and 7.2±8.1 MM, respectively. Conclusion: The greater efficacy of combination anti-TNF and thiopurine may be due to increased 6TGN levels, not just suppression of immunogenicity. In preferential metabolizers, preemptive AZA/6MP dose-splitting should be considered to maintain the benefits of 6TGN increase but avoid excess 6MMP production. Laboratory parameters should be followed during anti- TNF induction period among patients continuing thiopurines. A prospective study is needed to confirm our finding.