Vishnu, Prakash

Abstract Background: SCT is a standard therapeutic approach for patients with MM and has contributed to improved outcomes. While SCT utilization in MM has increased for all racial-ethnic subgroups over time, population-based studies have repeatedly shown that racial minorities are less likely to receive it. We explored factors that determine SCT utilization among patients within a given race to understand intra-racial disparities and how to address them in order to maximize the use of this imperative treatment among eligible patients. Methods: The 2016 National Cancer Database (NCDB) for MM with patients diagnosed between 2004-2013 was studied. Cases without information on the selected patient or facility characteristics evaluated in this analysis, including SCT use, race, median income, insurance status or facility type of initial MM diagnosis and/or treatment were excluded. The association between utilization of SCT as part of initial therapy of MM and socio-demographic factors was studied stratifying patients into mutually exclusive racial-ethnic subgroups non-Hispanic Whites (NHW), non-Hispanic Blacks (NHB), Hispanics, non-Hispanic Asians (NHA) and others using multivariate GEE logistic regression analysis (clustered on facility ID). Interactions between race and patient/facility characteristics and SCT utilization were tested using similar methods but including all races. Results: Of the 123480 unique MM patients in NCDB diagnosed between 2004-2013, 111799 met the inclusion criteria. Of these, 15021 patients (13.4%) received SCT as part of initial MM therapy. Of those who received SCT, 77.5% were NHW (14.4% of all NHW), 15.1% NHB (10.3% of all NHB), 5.2% Hispanics (13.2% of all Hispanics), 2% NHA (15.5% of all NHA) and <1% others. There was no difference by gender but significantly reduced odds of receiving SCT were noted for every 10-year increase in age at diagnosis for all racial subgroups. NHA were the only subgroup not to have a significant year-wise increase over time in odds of receiving SCT between 2004-2013. A significant increase in SCT utilization was seen with increasing median income (quartiles; 2008-2012 data), for NHW and Hispanics but not for NHB and NHA, while increased education (% population with high-school degree; 2008-2012 data) was associated with a significant increase in SCT use for NHW and NHB, but not among Hispanics or NHA. An increase in great circle distance (GCD) from the treating facility as well as a lower Charlson-Deyo comorbidity index (CCI) was associated with significantly increased odds of receiving SCT for all racial groups. Insurancepayer status significantly affected odds of receiving SCT among all races with those having private insurance to be most likely to receive SCT among NHW, NHB and Hispanics while those with "other government" insurance payer type being most likely among NHA. Among treating facility characteristics, patients of all races were more likely to receive SCT if treated at an academic/researchfacility (including NCI-designated centers) or centers with highest quartile of patientvolume seen. For analysis by geographicalregion, NHW and NHA had the highest odds of receiving SCT in mountain region but there was no impact of geography on SCT for NHB or Hispanics. Odds ratio (OR) of receiving SCT as part of initial MM therapy by socio-demographic and facility characteristics for patients of individual racial subgroups are shown in Table 1. A significant interaction between odds of receiving SCT and various racial subgroups was noted for patient age at diagnosis (p<.001, decreased odds among NHW and NHA), GCD from treating facility (p=0.013, increased odds among NHB and NHA), CCI (p=0.043, decreased odds among NHA) and treatment facility volumes (p=0.046, increased odds among NHA and Hispanics). Conclusions: In the largest analysis exploring socio-demographic factors affecting SCT utilization in MM we note a significant disparity within races for who gets SCT as a part of initial MM management and who does not. While some of these factors are non-modifiable, others including access to healthcare, income, insurance and literacy levels, proximity to treatment center, treating facility type as well as volumes are modifiable and appear to be barriers to SCT among in some races, while not in others. These must be addressed to minimize disparities within racial subgroups, especially the minorities and realize maximal benefit of SCT for MM. Disclosures Ailawadhi: Novartis: Consultancy, Honoraria; Pharmacyclics: Research Funding; Amgen: Consultancy, Honoraria; Takeda: Consultancy, Honoraria. Sher: LAM Therapeutics, Inc: Research Funding.

e20626 Background: It is important to increase our understanding of recruitment barriers in the elderly population to improve their participation in clinical trials. The primary objective of this study was to describe the demographics, treatment, toxicity, and overall survival (OS) of all patients ≥ 65 years who presented to the Phase I clinical trials service at KCI between 1995–2005. Methods: Patient demographics, co-morbidities, tumor type, reason not enrolled on a Phase 1 clinical trial, toxicities, and OS were obtained by retrospective chart reviews. The patients were divided into 3 mutually exclusive groups: patients considered but not enrolled (PC), enrolled but not treated (PE), and treated (PT). OS was compared across these three groups. Results: 216 patients met the study criteria. Of these, there were 114 PC, 7 PE, and 95 PT. Lack of enrollment was due to ineligibility (30%), loss to follow-up (12%), refusal (8%), or no trial available (2%). Median age was 71 years, 63% were males and 87% were Caucasian. 59% had performance status (PS) 1; 16% had PS 2. 66% of patients had a history of cardiovascular disease but renal, liver, hematological diseases were found in less than 7% of patients at baseline. Colorectal (27%), lung (15%) and prostate (8%) were the 3 most common cancers. 54% received a cytotoxic agent, 62% a biologic agent, and 17% combination therapy. The 3 most common Grade 3–4 toxicities were: 15% leucopenia, 11% electrolyte abnormalities, and 4% anemia. The median OS for PC, PE, and PT was 3.9 mos, 2.2 mos, and 8.4 mos, respectively (p < 0.001 between any pair). However, PS 0–1 patients comprised 47%, 75%, and 73% of those 3 patient subgroups, respectively. Conclusions: More than half of the elderly patients initially considered for Phase 1 clinical trials at KCI were not enrolled due to ineligibility, refusal, or lost to follow up. Phase I chemotherapy appears feasible in patients age ≥ 65 years old. OS was significantly longer for patients treated on Phase I trials compared to those who were not, although that may be confounded by PS differences. More detailed OS comparisons will be presented. No significant financial relationships to disclose.

Background: Patients with immune thrombocytopenia (ITP) may require treatment to raise the platelet count before surgery. Intravenous immune globulin (IVIG) is commonly used for this purpose, but eltrombopag, an oral thrombopoietin receptor agonist, may be a good alternative. We compared perioperative eltrombopag or IVIG for the achievement of platelet count targets for major or minor surgery. Methods: We conducted a randomized, non-inferiority trial at 8 centers in Canada.ITP patients with a platelet count &lt;100 x109/L or &lt;50 x109/L before elective major or minor surgery, respectively, were assigned to receive daily oral eltrombopag starting 3 weeks preoperatively or IVIG 1 week preoperatively. Eltrombopag was discontinued 1 week after surgical hemostasis, and a repeat dose of IVIG was permitted up to 1 week after surgical hemostasis. The primary endpoint was the achievement of platelet count targets (90 x109/L for major surgery, or 45 x109/L for minor surgery) immediately before surgery and for 7 days after surgical hemostasis without rescue treatment. Intention-to-treat and per-protocol analyses were done using a one-sided test for the difference in rates of achieving perioperative platelet count targets with a non-inferiority margin of 10% at a significance level of 0.05. Patients were followed for 4 weeks postoperatively. Secondary outcomes were thrombosis, adverse events and treatment satisfaction as measured using a validated patient questionnaire. Results: From 2013 to 2019, 38 patients were assigned to received eltrombopag and 36 patients were assigned to receive IVIG before major (n= 31) or minor (n= 43) surgery. Five patients did not complete the study. By intention to treat, 30/38 (78.9%) patients on eltrombopag achieved perioperative platelet count targets compared with 22/36 (61.1%) on IVIG (absolute risk difference, 17.8%; one-sided lower limit of the 95% confidence interval, 0.4%; p=0.005 for non-inferiority). Results were similar in the per protocol analysis (absolute risk difference: 15.8%; one-sided lower limit of the 95% CI, -2.1%; p=0.009). Eltrombopag was superior to IVIG in the intention-to-treat analysis (p=0.047) but not in the per-protocol analysis (p =0.074). One patient developed pulmonary embolism 14 days after minor surgery (skin biopsy) and 7 days after stopping eltrombopag when the platelet count was 135 x109/L. Of 18 patients who underwent splenectomy on study, marked post-splenectomy thrombocytosis (platelets &gt;1,000 x109/L) occurred in 2 (11.1%) patients, both of whom were on eltrombopag. Global treatment satisfaction scores were higher for eltrombopag. Conclusion: Eltrombopag is non-inferior and may be superior to IVIG for achieving target platelet counts perioperatively for patients with ITP. Post-operative thromboprophylaxis should be considered with eltrombopag. Funding Source:This was an investigator-initiated trial funded by Novartis. Disclosures Arnold: Novartis: Honoraria, Research Funding; Rigel: Consultancy, Research Funding; Bristol-Myers Squibb: Research Funding; Principia: Consultancy. Heddle:Novartis: Research Funding. Hsia:Novartis: Honoraria, Membership on an entity's Board of Directors or advisory committees; Pfizer: Membership on an entity's Board of Directors or advisory committees; Amgen: Honoraria; Jansen: Honoraria, Membership on an entity's Board of Directors or advisory committees. Sholzberg:Novartis: Honoraria; Amgen: Honoraria, Research Funding. Lin:Pfizer: Consultancy; Novartis: Research Funding; Octapharma: Research Funding; Amgen: Consultancy. Larratt:Novartis: Honoraria. Amini:Amgen NL: Research Funding; Novartis NL: Research Funding. Schipperus:Novartis: Research Funding. Lim:Pfizer Canada: Consultancy, Honoraria; Portola Pharmaceuticals: Consultancy; Bristol-Myers Squibb: Honoraria. OffLabel Disclosure: Eltrombopag is not licenced for perioperative treatment in patients with ITP.

Multiple myeloma (MM) treatment has advanced significantly over the last 2 decades. In most patients, the disease course has been altered from early fatality to chronic morbidity with multiple lines of treatment. The MM treatment paradigm has shifted toward treating patients before end-organ damage occurs. Thus, timeliness of treatment initiation in this era might improve patient outcomes. This is the first report to our knowledge analyzing disparities and trends in treatment timeliness of patients with MM using the National Cancer Database. Multiple factors affected the timing of treatment initiation in MM and disparities were found. We noted that initiation of treatment was delayed in women (odds ratio [OR], 1.15; 95% CI, 1.1 to 1.2) and blacks (OR, 1.21; 95% CI, 1.14 to 1.28; reference, whites) and in patients diagnosed in more recent years (2012-2015; OR, 1.15; 95% CI, 1.1 to 1.22; reference, 2004-2007). Patients were likely to start treatment earlier if they were age ≥ 80 years (OR, 0.83; 95% CI, 0.76 to 0.9; reference, age < 60 years), were uninsured (OR, 0.81; 95% CI, 0.72 to 0.91; reference, private insurance), had Medicaid (OR, 0.87; 95% CI, 0.79 to 0.95; reference, private insurance), were treated in a comprehensive community cancer program (OR, 0.7; 95% CI, 0.65 to 0.77; reference, community cancer program), lived in a location other than the US Northeast, or had a higher Charlson comorbidity score. Patient education and income levels did not affect time to treatment initiation. Particular aspects of these disparities could be explained by our current health care system and insurance rules, whereas others need to be investigated more deeply.

Abstract Introduction Immune thrombocytopenia (ITP) is an acquired disorder characterized by immune-mediated platelet destruction along with decreased platelet production. Adults with ITP are treated initially with glucocorticoids, intravenous immunoglobulin, and/or anti-D, but as many as 80% of patients (pts) will eventually relapse and become steroid-refractory. There is no universally agreed upon 2nd line therapy for ITP and consensus guidelines recommend either rituximab (R) or splenectomy (S) as acceptable, potentially curative options. Long-term responses to 2nd line S have been reported around 60%, and around 25% for R. It is unknown whether long-term outcome in the 3rd line setting is influenced by sequence of therapy. We evaluated the impact of sequence of 2nd and 3rdline therapy with R and/or S. Methods We identified all pts with ITP treated from 1990 through 2015 at the three Mayo Clinic (Arizona, Florida, and Minnesota) sites that were treated with R and/or S. Patients were excluded who were <18 years of age at the time of second-line treatment. Those with secondary ITP that received specific treatment for an infectious, autoimmune, or malignant condition causing thrombocytopenia were excluded. Patients with Evan's syndrome were included if the treatment was for thrombocytopenia, and not primarily for hemolytic anemia. For this analysis, secondary ITP was defined as ITP associated with hematologic malignancy. Hematologic malignancies did not require specific therapy at the time of treatment for ITP. Patients with uncontrolled infections or uncontrolled autoimmune disorders were excluded from analysis. We collected data from medical records regarding demographics, medical history, treatment history, response to treatment, and relapse. Medical history was captured for all concurrent diagnoses that could be considered associated with ITP including infectious and autoimmune conditions. Primary end-points were freedom from relapse after 2nd and 3rd line treatment with R or S. We used Kaplan-Meier method to estimate freedom from relapse after 2nd and 3rd line treatments. We also evaluated response to treatment, as defined by a platelet count >30,000/mm3 according to consensus guidelines. We performed three separate analyses for the entire cohort, primary ITP, and secondary ITP (separating out those with a concurrent diagnosis of malignancy). Results We identified 222 patients with a diagnosis of ITP, of which 191 had primary and 31 had secondary ITP. Treatment characteristics were similar in all groups except that pts treated initially with S were younger at diagnosis than pts treated with R (49 vs 60 years, P=0.003) and at time of 2ndline treatment (51.5 vs 61 years, P=0.018). For the overall cohort, patients treated with S as second-line therapy were more likely to achieve CR (86.6% vs 44%, P<0.0001), and had a significantly higher freedom from relapse at 5y (53.57% v. 14.96%, P<0.0001). Patients treated with S followed by R (S to R) had a trend towards higher 2y freedom from relapse than those treated with R followed by S (R to S) (69.48% v. 58.43%, P=0.3264). These data remained similar when primary ITP and secondary ITP were separated (data not shown). Conclusions These data support that S provides superior FFR at 5 years compared to R as second line treatment and suggests that, when needed, splenectomy followed by rituximab as opposed to rituximab followed by splenectomy might have a slight advantage. Disclosures No relevant conflicts of interest to declare.

Abstract Background: Therapeutic advancements have led to significant improvement in outcomes for MM patients. Revised diagnostic criteria from International Myeloma Working Group focus on initiating treatment before the onset of end-organ damage. Thus, timeliness of initial therapy can significantly improve patient morbidity and outcomes. Such an analysis has not been reported for MM. Methods: We analyzed patients diagnosed with MM between 2004-2015 from the National Cancer Database (NCDB). Patients who received systemic treatment any time within the first year of diagnosis were included; those treated later than one year were considered possibly initially diagnosed with smoldering myeloma and hence, excluded. Time to initial treatment was divided into centiles with the 4 centiles analyzed being 0-7 days, 8-18 days, 19-37 days and >37 days. We performed univariate and multivariate analyses to compare sociodemographic and clinical factors influencing patient distribution across these timeliness categories and trends over time. Results: A total of 53,665 patients were included with 15,282 in 0-7 days, 12,462 in 8-18 days, 12,723 in 19-37 days and 13,198 in the >37 days treatment categories. Distribution across these centiles over time is shown in Figure. Univariate analyses showed a significant difference in distribution across the timeliness categories by year of diagnosis, patient age, gender, race/ethnicity, education level, insurance type, comorbidity score, treatment facility type, geographical location and distance to treating facility (all p<0.001) but no effect of income level. Multivariate analysis showed that there was a significantly higher likelihood of being in 4th centile (>37 days to treatment) for women (p<0.001), African Americans (p<0.001) or Hispanics (p=0.03; reference: Whites) or diagnosed in more recent years (p<0.001). Patients were less likely to be in the 4th centile if they were ≥80 years (HR 0.89, 95% CI 0.81-0.98, p=0.02), had >0 Charlson comorbidity score (p<0.001), were uninsured (p=0.02) or had Medicaid (p=0.03) as compared to having private insurance, were in non-community treatment facilities or geographical location other than Northeast (p<0.001). Patient education, income level or distance from treating facility was no longer significant. Conclusions: Several factors were found to affect timeliness of treatment in MM. We noted relatively faster time to initial treatment in elderly, uninsured or those who were treated in academic/comprehensive cancer programs but a delay in treatment for women and racial/ethnic minorities among other factors. These factors need to be addressed for equitable access and treatment practices and to realize improved outcomes for all. Figure. Figure. Disclosures Ailawadhi: Amgen: Consultancy; Celgene: Consultancy; Janssen: Consultancy; Takeda: Consultancy; Pharmacyclics: Research Funding.

Abstract BACKGROUND: Glanzmann's thrombasthenia (GT) is an inherited platelet disorder (IPD) that leads to clinically significant bleeding. Platelets from patients with GT can show qualitative or quantitative defects of platelet membrane glycoprotein (GP) IIb/IIIa complex. Most GT are caused by autosomal recessive genetic defects in ITGA2B and ITGB3 (genes for GPIIb and GPIIIa, respectively) with rare cases showing an autosomal dominant pattern. Accurate diagnosis of GT requires a constellation of both phenotypic and genetic studies. Here we report a unique case of autosomal dominant GT resulted from thorough phenotypic and genetic studies. PATIENTS AND METHODS: A 19 year-old woman was recently evaluated for life-long history of easy bruising and severe menorrhagia that was only partially responsive to medroxyprogesterone. She has severe thrombocytopenia first recognized shortly after birth with a platelet count around 20x109/L. Platelet size was normal. She was presumed to have immune-mediated thrombocytopenia (ITP) since early infancy, but lacking responsiveness to the conventional treatments of ITP including immunomodulation, splenectomy and thrombopoietin receptor agonists. Family history was negative for a bleeding diathesis. Both von Willebrand factor antigen and activity were within normal range. Bone marrow aspirate and biopsy with associated chromosome studies were all normal. Platelet surface glycoprotein assessment by flow cytometry using antibodies to GP IIb, IIIa, Ia, Ib-a, VI, IX, and whole exome sequencing (WES) utilizing a predefined list of 53 clinically significant genes* related to genetically IPDs were performed. Due to thrombocytopenia platelet aggregation studies could not be performed. METHODS/RESULTS: Platelet surface expression of GPIIb (CD41) and GPIIIa (CD61) were markedly decreased suggestive of a variant of GT (Figure). WES performed with Illumina HiSeq 2500 sequencing system by using Agilent SureSelelct CRE kit V1 to capture and target the exonic regions showed presence of a heterozygous mutation in ITGB3 gene (c2213T>G; p.Leu738Arg). By in silico prediction modeling, this mutation is likely to be pathogenic and results in the substitution of hydrophobic leucine with hydrophilic arginine in the transmembrane domain of the β3 subunit of alpha IIb/beta 3 integrin (αIIbβ3), the platelet receptor that binds to fibrinogen. Interestingly, GPVI is also decreased which may be associated with GPIIb and GPIIIa deficiency or due to accelerated shedding since no GPVI mutation was identified. CONCLUSION: We describe a patient with GT associated with a novel heterozygous autosomal dominant mutation in ITGB3 gene with substitution of leucine with arginine (c2213T>G; p.Leu738Arg). This deletion caused a low expression of αIIbβ3 integrin on her platelets surface and severe thrombocytopenia. This case underscores the importance of an integrated phenotyping and genotyping approach to render a definitive diagnosis of an IPD. *Platelet exome gene list: ABCG5, ABCG8, ACBD5, ACTN1, ANKRD26, ANO6, AP3B1, BLOC1S3, BLOC1S6, CYCS, DTNBP1, ETV6, FLI1, FLNA, GATA1, GFI1B, GP1BA, GP1BB, GP6, GP9, HOXA11, HPS1, HPS3, HPS4, HPS5, HPS6, ITGA2B, ITGB3, LYST, MASTL, MPL, MYH9, NBEAL2, ORAI1, PSRX1, P2RY12, PLA2G4A, PRKACG, RASGRP2, RBM8A, RUNX1, STIM1, STX11, STXBP2, TBXA2R, TBXAS1, THPO, TUBB1, UNC13D, VIPAS39, VPS33B, VPS45, WAS Figure Figure. Disclosures No relevant conflicts of interest to declare.

Abstract ABSTRACT Background: In 2014, the American Society of Hematology (ASH) established a practice improvement module (PIM) incorporating quality metrics for management of diffuse large B cell lymphoma (DLBCL). Such PIMs have allowed physicians to monitor the quality of care in their practice. We implemented a DLBCL quality improvement initiative (QII) at our institution in January, 2015. In an appraisal of this initiative, we reviewed the ASH PIM metrics and included several others to assess adherence to guidelines for treatment of DLBCL and to examine the need for institutional improvement. Methods: Patients who were newly diagnosed with DLBCL and received treatment at our institution from January 2006 through December 2017 were identified. Electronic medical records were reviewed for documentation of ASH PIM quality measures (e.g., key pathologic features of DLBCL, lymphoma staging, screening for hepatitis B virus (HBV) infection in patients receiving rituximab-based chemotherapy, etc). We also reviewed the proportion of patients who had assessment of prognosis by revised International Prognostic Index (r-IPI) score, testing for hepatitis C (HCV), HIV viral infections, chemotherapy education, and the addition of rituximab in the treatment regimen of CD20+ DLBCL. Results: Following implementation of the QII, our institution saw improvement in most quality metrics. Particularly significant were improvements in reporting of key molecular features (45.45% to 91.6%, P < 0.0001), screening for HBV (41.82% to 91.67%, P < 0.001) and HIV infections (33.94% to 87.5%, P < 0.0001). All patients post-QII had a PET-CT scan for staging of DLBCL and there was a significantly lower use of bone marrow biopsy (61.2% to 33.33%, P = 0.011). Conclusion: Implementation of a quality initiative and employing standardized metrics can aid in improving institutional quality of care for patients with newly diagnosed DLBCL, and allows opportunity to build and ensure better adherence to evolving national and professional patient care guidelines. Figure. Figure. Disclosures No relevant conflicts of interest to declare.

Abstract Background: DLBCL is the most common adult NHL and despite significant advancements, outcomes in ethnic minorities remain suboptimal. Factors responsible for these disparities in ethnic minorities, especially variability in clinical and pathological features with evaluation of patient-level data have not been studied. Methods: A multi-institutional database of DLBCL patients seen at Mayo Clinic in Florida, University of Miami and University of Southern California between 1996-2016 was developed. Patient demographics, clinical disease characteristics and details of DLBCL immunophenotype [CD10, MUM1, BCL6 (Han's classification), Ki 67, MYC, BCL2) were reviewed. Comparisons between Hispanic and Non-Hispanic patients were made by Chi-square, Fischer's exact and Signed rank tests, where applicable. Patients with missing data on a certain feature were not included in the statistical analysis of that specific characteristic. Results: A total of 966 DLBCL patients with 69 Hispanics and 752 non-Hispanics were identified. These included 52% males (N=36) and 48% females (N=33) in Hispanics versus 58% males (N=436) and 42% females (N=316) in Non-Hispanics (p=0.35). Median age at diagnosis was significantly lower at 61 years (range 27-89) in Hispanics versus 66 in non-Hispanics (p=0.014). Extranodal disease was seen significantly more commonly in Hispanics (74%) than in non-Hispanics (60%) (p=0.037). There was no significant difference in disease stage at presentation between the two cohorts (p=0.189) although the proportion of patients with stage III and IV disease was higher among Hispanics (80.4%) as compared to non-Hispanics (68.5%). Similarly there was no significant difference in presence of primary CNS DLBCL at diagnosis between the two groups (p=0.74). GCB phenotype per the Hans' classification was seen in 59% (N=29) in both, Hispanics and Non-Hispanics (N=241) (p=0.99). Similarly, the median ki67 index of 80% was seen in both cohorts (p=0.48). MYC was positive in 59% in Hispanics versus only 37.5% in non-Hispanics (p=0.024). BCL2 expression by itself and also combined BCL2 and MYC double expression were not significantly different between the two groups (p=0.58 and p=0.11, respectively). Median overall survival (OS) was 22.5 months in Hispanics compared 54.9 months in Non-Hispanics (p<0.0001) (Figure 1). Conclusion: Survival outcome disparities in racial and ethnic minorities have been previously reported from population-based analyses. Lack of patient-level clinical data makes these analyses hypothesis generating, but not conclusive. We confirmed some prior findings including younger age at diagnosis for Hispanics with DLBCL. We also reported greater incidence of extranodal as well as stage III or IV disease at the time of presentation among Hispanics which may contribute to poor outcomes in this population. From DLBCL pathology standpoint, we found increased expression of MYC among Hispanics but no other significant immunohistochemical differences including double expression or non-GCB phenotype. The inferior OS despite uniform treatment practices at large Cancer Centers suggest multifactorial influences on patient outcomes which may at least in part include some of the differences we report here. Disclosures Ailawadhi: Pharmacyclics: Research Funding; Janssen: Consultancy; Takeda: Consultancy; Celgene: Consultancy; Amgen: Consultancy.

Abstract Background: Racial disparities in outcomes of cancer patients have been reported. Access to comprehensive cancer centers is associated with improved overall survival (OS) but racial/ethnic minorities may have a disparate access to such care. While the impact of treatment facility volume on outcomes has been evaluated, outcomes of centers with minority-predominant patient population have not been studied. We compared demographic profiles, facility level data and OS of patients treated at minority-predominant facilities to facilities that treated predominantly non-Hispanic Whites (NHW) with non-DLBCL. Methods: The National Cancer Database (NCDB) was used to identify all non-DLBCL patients diagnosed between 2004 and 2015. "Minority-treating facilities" were defined as facilities in the top decile by proportion for initial treatment of non-Hispanic African-Americans (NHAA), Hispanics and other races. We performed univariate and multivariate analyses to compare sociodemographic and clinical factors influencing outcomes between minority treating and non-minority treating facilities. A subgroup analysis stratified by race/ethnicity was also conducted to study the effect of treating facilities on the outcome of NHWs and minorities separately. Results: Of 1339 total facilities, 123 (9.1%) qualified as minority treating. Of 207,239 eligible patients in NCDB, 18,719 (9.03%) received treatment at the minority-treating facilities and of these, 11,190 (~60%) belonged to the minority races. Overall, 4.5% (6,988/156,664) NHWs and 30% (11,190/37,639) minorities received treatment at the minority-treating facilities. Several demographic and facility level characteristics were significantly different among the patients treated at minority-treating facilities as compared to non-minority treating facilities. Overall, significantly higher number of patients in minority-treating hospitals had lower income and education, had Medicaid coverage or lack of insurance. The OS of patients in minority treating facilities was significantly worse as compared to non-minority facilities (Figures). On multivariate analysis, patients who received treatment at minority-treating facilities were at 10% (HR=1.10, 95% CI: 1.06-1.14 p<0.001) higher risk of mortality as compared to those treated at the non-minority treating facilities. On multivariate analysis, NHAA (30% increased risk) and 'other races' (9% increased risk) were at significantly higher risk of mortality as compared to the NHW (Table 2). To study the effect of treatment at minority-treating facilities on OS among the patients of same race/ethnicity group, a multivariate analysis was also run separately for NHW and racial minorities. The NHWs who received treatment at minority-treating facilities were at 13% higher risk of death (HR=1.13, 95% CI: 1.08-1.19 p<0.001) as compared to NHWs who were treated at non-minority treating facilities. Similarly, the racial minorities who received treatment at minority treating facilities were at 8% higher risk of death (HR=1.08, 95% CI: 1.03-1.19 p=0.003) as compared to those who received treatment at the nonminority treating facilities. Conclusions:Outcomes of patients who received treatment at minority treating facilities was significantly worse than those at non-minority treating facilities. This was true for NHWs and racial minorities separately as well. Several demographic and facility level characteristics were significantly different in the two groups however OS remained worse after adjusting for them. Causes of poor outcomes at minority-treating facilities must be analyzed to mitigate them and improve outcomes for all. Figure. Figure. Disclosures Ailawadhi: Takeda: Consultancy; Celgene: Consultancy; Amgen: Consultancy; Janssen: Consultancy; Pharmacyclics: Research Funding.