ALBERT

Description: Background Myelodysplastic syndrome (MDS) is a pre-leukemia disease affecting the erythroid, myeloid and megakaryocytic bone marrow production. MDS patients are classified according to the WHO classification of myeloid neoplasms. During the past 15 years management of MDS patients has been stratified according to the International Prognostic Scoring System (IPSS) risk score. Recently a revised version of IPSS has been introduced (IPSS-R). One quarter of LR-MDS in this new IPSS-R were reclassified as having a higher risk and a substantial subset of high risk-MDS (HR-MDS) were reclassified as lower risk. In LR-MDS a differentiation block is observed in the erythroid lineage. The diagnosis and follow up of cytopenias in particular anemia must be the main objective (1). The soluble transferrin receptor (sTfR) directly reflects the erythropoietic activity in individuals without iron defiency and may appreciate ineffective dysplastic erythropoiesis in LR-MDS. In LR-MDS there is also an inverse relationship between EPO level and the degree of anemia but a wide range of EPO levels is found in patients with similar Hb concentrations. Thus the highest EPO levels are found in patients with erythroid hypoplasia in bone marrow. Aims The combination of several biomarkers: Hb, ferritin, EPO and sTfR may be useful in LR-MDS for diagnosis and follow up. Methods A total of 192 patients with LR-MDS were investigated. Median age of the 192 patients was 71 years (21-92) with 56% males, median survival: 54 months, median follow up: 102 months. The stratification according to the WHO criteria and IPSS risk score was realized. Bone marrows were studied and cytogenetic assessment was realized in the same time. Serum concentrations of ferritin, EPO and sTfR has been analyzed by immuno-assays. Hb level was determined on Beckman Coulter apparatus. The follow up of Hb, ferritin, EPO and sTfR was realized every 2 months in patients with supportive care only until the first specific treatment. A multivariate logistic regression analysis to ascertain the correlations between disease progression and studied biological parameters was realized. Results The logistic regression analysis of our results is significant to define a biological evolutive profile of LR-MDS patients with these biomarkers. The combination of these routine parameters may represent a functional erythropoietic follow up in LR-MDS patients (table 1). This biological tool is an easy method to observe the red cell lineage of LR-MDS patients. This combination informs about the progressive ineffective and dysplatic erythropoiesis in LR-MDS patients. The measurement of ferritin which is a correlated parameter in LR-MDS shows the level of iron overload. A normal or high level without inflammation condition excludes an iron deficiency. The EPO level can give a predictive information about the future efficacy of ESA (endogenous EPO 〈 500 U/l). Conclusion With our results and a correlative logistic regression analysis, we can propose a biological scoring system to appreciate the evolutive anemia of LR-MDS progression in patients. In LR-MDS the management of patients may be based on personalized medicine according a risk assessment with IPSS-R, cytogenetics, mutations and HLA typing (2). But an additional biological and functional predictive scoring system informs about the important independent role of dysplasias particularly anemia in LR-MDS patients before to choose a suitable therapy: transfusions, iron chelation, ESA, TGF-ï¢ pathway inhibitors, G-CSF, immun suppressive treatment, lenalidomide, azacytidine, allogeneic HSCT Table 1. Hb ± EPO ±  sTfRDysplastic erythropoiesis without anemia Hb ±  EPO  sTfRStabilized dysplastic erythropoiesis Hb  EPO  sTfRUnstabilized dysplactic erythropoiesis Hb  EPO  sTfRIneffective dysplastic erythropoiesis EPO 〈 500 U/l : ESA may be efficient〉 500 U/l : ESA will be inefficientFerritin level : iron overload References Giagounidis A Management of low-risk myelodysplastic syndromes Hematology Education, 2015, 9 (1), 219-225 Platzbecker U et al Personalized medicine in myelodysplastic syndromes: wishful thinking or already clinical reality? Hematologica, 2015, 100 (5), 568-571 Disclosures No relevant conflicts of interest to declare.

Description: An obstacle with continued clinical development of CAR T cells is the limited understanding of their biology and mechanisms of anti-tumor immunity. We and others have shown that CARs with a CD28 co-stimulatory domain drive high levels of T cell activation that also lead to exhaustion and shortened persistence. The CD28 domain includes 3 intracellular subdomains (YMNM, PRRP, and PYAP) that regulate signaling pathways post TCR-stimulation, but it is unknown how they modulate activation and/or exhaustion of CAR T cells. A detailed understanding of the mechanism of CD28-dependent exhaustion in CAR T cells will allow the design of a CAR less prone to exhaustion and reduce relapse rates. This led us to hypothesize that by incorporating null mutations of CD28 subdomains (Fig 1A) we could optimize CAR T cell signaling and reduce exhaustion. In vitro, we found mutated CAR T cells with only a functional PYAP (mut06) subdomain secrete significantly less IFNγ, IL6, and TNFα after 24hr stimulation compared to non-mutated CD28 CAR T cells, but greater than the 1st generation m19z CAR. Also, cytotoxicity was enhanced compared to non-mutated CARs (Fig 1B). Using a pre-clinical immunocompetent mouse tumor model, we found the mut06 CAR T cell treated mice had a significant survival advantage compared to non-mutated CD28 CAR T cells (Fig 1C). To examine exhaustion, we ex vivo stimulated CAR T cells with target cells expressing CD19 and PDL1 and found mut06 CAR T cells had increased IFNγ (42%), TNFα (62%) and IL2 (73%) secretion compared to exhausted non-mutated CD28 CAR T cells. This suggests that mut06 CAR T cells are more resistant to exhaustion. To find a mechanistic explanation for this observation we examined CAR T cell signaling. After 24hr stimulation with CD19 target cells mut06 CAR T cells had a significant reduction in pAkt compared to m1928z CAR T cells, which is a critical signaling mediator in the NFAT and NR4A1 transcription factor pathways. Additionally, mut06 had decreased p-NFAT compared to m1928z when examined by western blot. To determine how optimized CAR signaling affected T cell exhaustion we looked at 22 genes that are upregulated when NFAT is constitutively active and overlap with genes identified as important for T cell exhaustion. We found that most of the exhaustion related genes were upregulated in m1928z CAR T cells while they were decreased in m19hBBz. The mut06 CAR T cell gene expression pattern was more similar to m19hBBz with exhaustion related genes downregulated compared to m1928z (Fig 1D). To examine differences in the accessibility of exhaustion related genes we performed ATAC-seq and found NFAT (Nfatc1) and NR4A2 (Nr4a2) had lower chromatin accessibility profiles in mut06 compared to m1928z (Fig 1E). We also found that exhaustion related genes Havcr2 (TIM3), Pdcd1 (PD1), and Lag3 (LAG3) all had greatly reduced chromatin accessibility in mut06 CAR T cells compared m1928z. Overall, these genomic studies support our findings that mut06 optimizes CAR T cell signaling by lowering transcription factors that regulate exhaustion. Figure 1 Disclosures Li: ImmuneBro Therapeutics: Other: sole shareholder . Davila:Atara: Research Funding; Celgene: Research Funding; GlaxoSmithKline: Consultancy; Novartis: Research Funding; Anixa: Consultancy; Bellicum: Consultancy; Adaptive: Consultancy; Precision Biosciences: Consultancy.

Description: The circadian system regulates numerous physiological processes including adaptive immune system. Here we show that mice deficient for the circadian genes Cry1 and Cry2, (Cry double knockout [DKO]) display an autoimmune phenotype including higher serum IgG concentration than wild type (WT) mice, presence of serum anti-nuclear antibodies, precipitation of IgG, IgM and complement 3 (C3) in glomeruli, and massive infiltrations of leukocytes into the lung and kidney. A large panel of autoantigens demonstrated that the sera of the Cry DKO mice but not the WT mice, had autoantibodies covering most of the specificities reported to be present in patients with SLE, rheumatoid arthritis (RA), multiple sclerosis (MS), Sjögren's syndrome and other autoimmune disorders. Taken together, lost of the CRY circadian protein leds to severe autoimmunity. Furthermore, flow cytometry analysis of lymphoid organs showed lower pre-B cell numbers and higher mature recirculating B cells in the bone marrow as well as increased number of B2 B cells in the peritoneal cavity of Cry DKO mice. The BCR-proximal signaling pathway plays a critical role in peripheral B cell tolerance and activation. Activation of splenic B cells from the Cry DKO mice elicited markedly enhanced and prolonged tyrosine phosphorylation of cellular proteins compared to WT mice, suggesting that a very active BCR signaling pathway may contribute to impaired B cell tolerance in the Cry DKO mice. In summary, our results suggest that B cell development, as well as the intrinsic checkpoints of immune tolerance, are under direct circadian control. Disclosures No relevant conflicts of interest to declare.

Description: Introduction Predicting treatment outcome of acute leukemia has been an important issue. Many factors have been elucidated. We evaluated the impact of donor's availability on treatment outcome including mortality in patients with acute lymphoblastic leukemia (ALL). Methods A total of 294 patients with ALL were evaluated after receiving chemotherapy between year 2001 and July 2014 at our center. Patients were assessed for the need of hematopoietic stem cell transplants (HSCT), availability of HLA sibling match donor and the impact on overall outcome. Indications for transplantation were defined as (1) WBC-〉100,000 for T-ALL, 〉30 for B-ALL with, (2) cytogenetic abnormalities of t(9:22), (4:11) or (1:19), (3) relapse or primary refractory disease or (4) MRD positivity. Patients were divided into 3 categories, group A with an indication for HSCT and available donor (HSCT+/D+), group B with HSCT indication but no available donor (HSCT+/D-) and group C with no indication for HSCT regardless of donor status (HSCT-). Results The median age was 20 (14-63 years). 95 (32%) were female while 198 (68%) were male. 276 (86%) patients were newly diagnosed while 18 (14%) were relapsed. Immuno-phenotype was B for 191 (65%), T for 91 (31%) versus mixed lineage for 12 (4%). 33 (11%) patients were positive for Philadelphia chromosome. Median WBC at diagnosis was 23.2X109/L CNS involvement was positive in 26 (8%) patients. With a median follow-up of 60 months for survivors (range 2 to 116.5 months), 148 (50%) patients showed HSCT+/D+, HSCT+/D- in 79 (27%) and HSCT- in 67 (23%) patients. the 5-year OS for HSCT+/D-, HSCT+/D+ and HSCT- were 29%, 57% and 55% (p value

Description: Metastatic disease of the bone is a rare complication of chronic lymphocytic leukemia (CLL), it may be result from richter's transformation or metastatic from non lymphoid malignancies. CLL is the most common form of adult leukemia, with the median age of 70 years at diagnosis [Siegel et al. 2013]. The diagnosis is established by blood counts, blood smears, and immunophenotyping of circulating B-lymphocytes.The result is the increased number of lymphocytes in the peripheral blood, leukocytosis with absolute lymphocytosis, the increase of the lymphnodes, the increase in size of the spleen. The diagnosis of chronic lymphocytic leukemia B requires the presence of Clonal B cells in the peripheral blood at or above 5,000 / ul for at least 3 months. Typing immunophenotypical pathological lymphocytes are positive for surface antigens CD5, CD19, CD23, weakly positive for CD20 and CD22, generally negative FMC7 and CD79b; also expressing surface immunoglobulins. The Rai and Binet staging systems, which are established by physical examination and blood counts, have been recognized as standards for deciding whether to begin treatment. Patients with active or symptomatic disease or with advanced Binet or Rai stages require therapy. For fit patients, chemoimmunotherapy with fludarabine, cyclophosphamide, and rituximab represents the current standard therapy. For unfit patients, treatment with an anti-CD20 antibody (obinutuzumab, rituximab, ofatumumab) plus a milder chemotherapy (Chlorambucil) may be applied. At relapse, if the treatment-free interval exceeds two to three years, the initial treatment may be repeated, if the disease relapses earlier, drugs such as bendamustine (plus rituximab), alemtuzumab, lenalidomide, ofatumumab, ibrutinib, or idelalisib, must be choosen. Patients with a del(17p) or TP53 mutation can be treated with ibrutinib or a combination of idelalisib and rituximab. in relapsing patients with TP53 mutations or del(17p) or patients that are refractory to repeated chemoimmunotherapies, an allogeneic SCT may be considered [Hallek M 2015]. In this article we show a case of a 66-year-old man with CLL and a bone localization. In 2011 diagnosis of CLL, Rai Stage 0, Binet Stage A. Principal characteristics at diagnosis: HB 13.2 g /dl, White Blood Cells 15.800 / mm3, lymphocytes 61%, neutrophils 32%, monocytes 4%, platelets 141.000/mm3; normal hepatic end renal function; flowcytometric immunophenotyping of the peripheral blood revealed B-cell CLL; prognostic factors: CD38 negative, ZAP70 positive, rearrangement of the immunoglobulins mutated; FISH: negative; CT chest / abdomen / pelvis: presence of multiple aorto-pulmonary and axillary adenopathies (max diameter of 2 centimeters); bone marrow biopsy: infiltration of CLL equal to 60% of global cellularity. The patient was only observed until January 2015, when he was hospitalized due to acute anemia, requiring supportive therapy, and right foot pain . So it was decided to re-evaluate the whole disease in order to decide whether to start chemotherapy. The disease was staged again with instrumental and laboratory tests: presence of renal insufficiency, egd and colonoscopy negative, Coombs' test negative, bone marrow biopsy confirmed the diagnosis of chronic lymphocytic with bone marrow infiltration of 90%, abdomen ultrasound showed only moderate splenomegaly. On February, persistence of right foot pain and appearance of swelling, assessed by the orthopedic as a suspected algic and dystrophic syndrome. So he suggested to perform scintigraphy which revealed: pronounced inflammatory osteometabolic reaction of the right tibia/fibula/ankle third distal which could be referred, in the first evaluation, to algic and dystrophic syndrome. However, a local biopsy was performed: localization of chronic lymphocytic leukemia. On March 2015 a total body TC showed 2 nodular calcifications in the right lung lobe, multiple right paratracheal, barety space, aortopulmonary and axillary adenopathies. Prostate size increased. In order to study carefully the liver and prostate lesions, an ultrasound abdomen was performed that documented only enlarged spleen, normal size liver, free of focal disease, increased prostate due to symmetric bilobate hypertrophy . After the second cycle of chemotherapy, prolonged thrombocytopenia, so he continues only with a radiotherapy program. Disclosures No relevant conflicts of interest to declare.

Description: A Significant Proportion of Young Adult Patients with Post -HCT Relapse Of AML Benefit From Aggressive Salvage And 2nd Cellular Therapy. INTRODUCTION: There is currently no standard of care for patients with AML who relapse following hematopoietic cell transplantation (HCT), and outcomes in these patients are generally poor. Given this fact, there is great variability in practice, and many patients may be palliated in the absence of suitable clinical trials, especially following early relapses. We sought to analyse long-term survival of young adults with AML based on whether or not they received a second cellular therapy (CT) (second transplant or donor lymphocyte infusion [DLI]) following post-HCT relapse. METHODS: We retrospectively analysed data on patients who had received a HCT between 2000 and 2012 and had a post HCT relapse. The patients were stratified by whether or not they had 2nd CT with or without prior chemotherapy. Baseline characteristics and outcomes were compared. RESULTS: Ninety four patients were identified who had relapsed AML following HCT. The median age at transplant for the patients was 27.5 (range 14-58y) years for the whole cohort; 50% were females. Of these, 30 patients received 2nd CT either in the form of DLI (80% for available CT data) or a 2nd HCT. Median in age for both groups was 24 years and there was no significant difference between the 2 groups in good or poor risk cytogenetics. Median time to relapse was significantly lower in the group that did not receive 2nd CT vs the group that did (5.9 vs 18.2 months, p

Description: Introduction: Data on management of upper extremity deep vein thrombosis(UEDVT) in patients with cancer is limited. Patients in this subgroup were excluded from the large clinical trials that showed the efficacy of extended duration low-molecular-weight heparin (LMWH) for cancer-associated thrombosis. Furthermore, risk factors for cancer-associated UEDVT in patients who do not have central lines in situ have not been well defined. The goal of our study was to determine the risk factors for cancer-associated UEDVT and to examine the approach to management of these patients in a real-world setting. Methods: We conducted a retrospective review of 200 consecutive patients who were assessed for cancer-associated UEDVT between January 2010 and June 2014 at a tertiary care center. Outcome measures were recurrent VTE, and major and clinically relevant non-major bleeding. Risk factors for recurrent VTE and bleeding were assessed using multivariable analysis. Results: Median duration of follow-up 11 months. Median age was 61.5 years and 55% were male. Cancer subtypes included colorectal (24%), lung (15%), breast (14%), lymphoma (10%), leukemia (5%), esophageal (4%), pancreatic (4%), head & neck (3%), sarcoma (2%), and others 17%. Metastatic disease at the time of diagnosis was present in 37% of patients and 7% of the study population had a previous history of VTE. Of the study population, 138 (69%) had line-associated UEDVT. Risk factors for UEDVT other than presence of a line after univariate analysis were lung cancer, breast cancer and extrinsic compression of vessels by local tumour on diagnostic imaging. The proportion of patients with UEDVT in the absence of a line according to cancer subtype was as follows: lung cancer (83%), pleural mesothelioma (66%), breast cancer (51%) and head & neck cancer (50%). Of these patients, greater than half had evidence of local mass effect on vessels on diagnostic imaging studies. Of the 138 patients with line-associated UEDVT, 20 (15%) had their line removed within one week of diagnosis for reasons unrelated to thrombosis and 107 (84%) had their line removed after completion of at least 3 months of anticoagulant therapy. Recurrent VTE was documented in 35 patients (17.5%), of which 16 (8%) were UEDVT (PE - 10, lower limb DVT-8, other-2). Recurrent VTE while receiving anticoagulants occurred in 23 (65.7%) of all recurrences and in 10 (62.5%)of UEDVT recurrences. Recurrent VTE occurred in 26 patients with a central line and in 9 patients without a central line. All of the patients with recurrent VTE had solid tumours, and 45% had metastatic disease. Multivariant analysis revealed that male gender(OR 2.42, 95% CI;1.1-5.1,p-value=0.02) and active cancer at the end of follow-up (OR 2.47, 95% CI; 0.1-0.9, p-value=0.04) were the only factors significantly associated with recurrent VTE (Figure 1 and 2). None of the following were significant risk factors for recurrence: type of antineoplastic treatment, accompanying PE, white cell count, initial UEDVT while anticoagulated, cancer stage, previous VTE, number of involved venous segments, removal of line during first week after index event or switching to a different anticoagulant. In the group with UEDVT without a venous catheter, the presence of radiologically proven extrinsic compression of vessels was not statistically associated with recurrent VTE. Patients were treated with LMWH for a median duration of 5 months. Six and 8 patients were switched to rivaroxaban and warfarin, respectively. Clinically relevant non-major bleeding occurred in 24 patients (12%), 61% of the bleeds were gastrointestinal and 83% of the bleeds occurred while receiving anticoagulants. On multivariate analysis, bleeding was significantly associated with ongoing anticoagulation (OR 5.6, 95% CI;1.6-19.3, p-value=0.006) and liver metastasis (OR 7.2, 95% CI;0.9-7.2, p-value=0.05). The use of concomitant clopidogrel or aspirin significantly increased the risk of bleeding (OR 6.6 and 5.5, respectively). Conclusions: While the presence of a venous catheter was the primary risk factor for UEDVT for the majority of our cohort, extrinsic compression of vessels by local tumour appeared to be equally important for certain cancer types. Furthermore, our finding that the majority of recurrent events did not occur in the upper limb suggests that UEDVT may be predictive of overall increased thrombogenic risk rather than just a local effect caused by the line. Disclosures Linkins: Bayer: Honoraria, Research Funding; Pfizer: Honoraria.

Description: Introduction: It is extremely significant to predict treatment outcomes using a biomarker. The prognostic significance of CD20 expression is controversial. In this study, we attempted to evaluate the impact of CD 20 expression on treatment outcomes in large cohort of adolescents and adults patients with acute lymphoblastic leukemia (ALL) treated with chemotherapy only in one single center. Methods: A total of 172 adolescent and adult patients with newly diagnosed B precursor ALL receiving CALGB based chemotherapy protocol between year 2001 and July 2014 at our Centre were evaluated for the expression of CD20 and it is impact on treatment outcomes. Anti-CD20 monoclonal antibody use were not part of the chemotherapy protocol at our Centre. Patients underwent hematopoietic stem cell transplantation were excluded from the analysis, a total of 83 patients were finally included in the analysis. CD20 expression at diagnosis of 20% in blast population was used as cutoff to stratify patients as CD20 positive (CD20+) versus CD20 negative (CD20-) group. Results: Median age for all patients is 21 years (range, 14-62), 41% (n=34) were females. Median WBC at diagnosis was 8 (range, 0.4-244), 28% (n=23) had WBC 〉30. Philadelphia (Ph) chromosome was negative for 75 patients (86%) while positive for 12 patients (14%). Day 28 blasts 5 (n=5) and missing for 4 patients (5%). A median follow-up of 3.7 years for survivors (range, 1-11) showed 29 patients (35%) had CD20+ versus 54 patients (65%) was CD20-. Comparison of patient's characteristics between the two groups were similar for gender, age, median WBC at diagnosis, median LDH, Ph chromosome, CNS involvement and primary refractoriness at day 28 of chemotherapy. The cumulative incidence of relapse at 1 year for CD20+ was 42% versus 15% (p=0.05) for + CD20- (Figure 1) while 5 years relapse rate was 55% versus 58%, respectively, suggesting higher incidence of 1 year relapse in CD20+ group. The overall survival for CD20+ was 39% versus 50% for CD20- (P=0.18). In Univariate analysis for One-year relapse, CD20 expression was significant factor (p=0.02, HR=2.97). Multivariate analysis confirmed CD20 as an independent adverse risk factor for relapse at 1 year (p=0.02, HR 2.97). For 5-years relapse, male gender (p=0.03, HR 2.1) and age 〉30 years (p=0.05, HR2.38) were found to be independent factors but not CD20 expression. Conclusion: The present study showed that the overall incidence of CD20 expression in B-ALL is 35%. Also, it suggested that in non-HSCT patients, CD20 expression at a diagnosis is associated with higher risk of early relapse at one year, which was confirmed by multivariate analysis. Further study is strongly warranted to confirm the prognostic role of CD20 and to evaluate the importance of anti-CD20 monoclonal antibodies in a prospectively designed study. Figure 1. Cumulative incidence of relapse at 1 year (A) and at 5 years (B) Figure 1. Cumulative incidence of relapse at 1 year (A) and at 5 years (B) Figure 2. Figure 2. Figure 3. Figure 3. Disclosures Kim: Bristol-Myers Squibb: Consultancy, Research Funding; Novartis Pharmaceuticals: Consultancy, Research Funding.

Description: Introduction: Various forms of aplastic anemia (AA)/bone marrow failure syndromes (IBMFs) show significant clinical and molecular heterogeneity with significant clinical overlap and are often diagnosed based on established clinical and pathological criteria. While 〉70 genes have been identified in patients with AA/IBMFs, most cases have are labelled as idiopathic with no identifiable genetic abnormlaity found. Precise detection of genetic abnormalities in these patients may assist in more accurate molecular diagnosis in these patients, proper counseling, cancer surveillance and personalized clinical intervention. Method: As part of the Saudi Human Genome Project, we developed a comprehensive 405 gene panel encompassing all known Mendelian hematological disorders (hemolytic anemias, aplastic anemias/bone marrow failure syndromes, coagulation disorders) using the Ion Torrent AmpliSeq technology. Patients who met the clinical diagnosis of aplastic anemia/bone marrow failure syndrome were enrolled into this study. Peripheral blood samples were subjected to this next-generation sequencing analysis. Results: We validated the Saudi Mendeliome assay using 642 samples with known mutations across various medical specialties. We then tested 37 patients with AA/IBMFS using this Proton-Ion sequencing platform. Mutations were identified in 7/37 (19%) of patients, followed by whole exome sequencing (WES) in those patients without identifiable mutations. Conclusion: Compared with clinical WES and/or whole genome sequencing (WGS), which are still expensive, time consuming and difficult to interpret, this novel and comprehensive targeted gene panel is more economical (〈 $150), faster (3-4 weeks), upgradable (by spiking in newly identified AA/IBMFs genes) and can be used to genotype patients with acquired aplastic anemia/bone marrow failure syndromes and guide their management. Second tier testing using WES/WGS is recommended for cases without identifiable mutations. Disclosures No relevant conflicts of interest to declare.

Description: Introduction: Immune Thrombocytopenia (ITP) is a disease of immune-mediated destruction of platelets and suppression of platelet production. ITP has been historically treated with corticosteroids and/or immune globulins as first-line agents. There are several second-line treatments available, should patients fail to respond to initial therapy or relapse after it is tapered. These include eltrombopag, romiplostim, rituximab and splenectomy. This study utilized a national electronic health record (EHR) database to begin to explore the real world treatment patterns of the aforementioned second-line (index) therapies. Methods: Utilizing the Optum EHR database, we identified patients who initiated their first second-line treatment (i.e. the index treatment) with eltrombopag, romiplostim, rituximab or splenectomy from Jan. 1, 2009 to Sep. 30, 2016 for primary or unspecified ITP. Patients included in the analysis had the following characteristics: 18 years or older; previous treatment with corticosteroids and/or immune globulin products; active in the database for at least 6 months prior to and 12 months post initiation of the index treatment. Outcomes that were evaluated after initiation of the index treatment included: (1) Duration of therapy for eltrombopag and romiplostim; (2) Proportion of patients who started a subsequent line of treatment after their index treatment; (3) Treatment free duration between the end of the index treatment and start of a subsequent line of treatment; and (4) Proportion of patients using a first-line medication (corticosteroids and/or immune globulin) during treatment with eltrombopag and romiplostim. Chi-square and t-tests were used for statistical analysis. Results: 2,047 patients met the inclusion criteria and used an index treatment as follows: eltrombopag, N=110 (5.4%); romiplostim, N=189 (9.2%); rituximab, N=1488 (72.7%); splenectomy, N=260 (12.7%). The mean age was 60.8 years (standard deviation [SD]: 17.4), with 52.4% female and mean Charlson comorbidity score of 2.1 (SD: 2.1). Treatment duration was 481 days for eltrombopag versus 346 days for romiplostim (p=0.033). The proportion of patients who started a subsequent line of treatment after their index treatment ranged from 41% for rituximab to 49% for splenectomy (p=0.071). Treatment free duration between the end of the index treatment and start of a subsequent treatment ranged from a mean of 248 days for romiplostim to 575 days for splenectomy (p