Citations

Abstract Introduction Histiocytic neoplasms are rare hematological malignancies that have protean clinical manifestations and can pose significant management challenges. Recently, vemurafenib was approved by the US-FDA for treatment of BRAF-V600-mutant Erdheim-Chester disease (ECD). However, there is a lack of FDA-approved therapies for other histiocytic neoplasms such as Langerhans cell histiocytosis (LCH) and Rosai-Dorfman disease (RDD). Over the last 5 years, immune checkpoint inhibitors such as programmed death-1 (PD-1) and programmed death-ligand 1 (PD-L1) inhibitors have shown significant improvement in outcomes among patients with several hematological and solid organ malignancies. In order to identify appropriate treatment candidates for these therapies, predictive biomarkers have been developed in various cancers. Evidence from solid tumors has suggested a favorable response to checkpoint inhibitor therapy with higher tumor mutational burden (TMB), defined as the number of mutations within a tumor genome. Next generation sequencing (NGS) of various tumors has shown an association between TMB of < 5 mutations/megabase (mut/Mb) to be associated with an absence of benefit from checkpoint inhibitors. In addition, high levels of PD-1/PD-L1 expression and microsatellite instability (MSI) are also correlated with response to therapy. The latter may be a result of somatic or germline alterations in DNA mismatch repair genes. In this study, we report the results for these biomarkers using NGS in patients with histiocytic neoplasms. Methods We utilized TempusTM NGS platform to analyze the tissue specimen of patients with histiocytic neoplasm. The Tempus xO Assay (Tempus Labs; Chicago, IL) combines a 1,711 gene targeted somatic and germline DNA sequencing panel with RNA-sequencing to detect both germline and somatic single nucleotide polymorphisms, indels, copy number variants, and gene rearrangements causing chimeric mRNA transcript expression in a wide array of solid tumor types. The assay utilizes formalin-fixed paraffin-embedded tumor samples and matched blood samples. TMB was calculated and reported as somatic mutations in tumor tissue per million base-pairs or mut/Mb. RNA sequencing was utilized to assess for PD-L1 and PD-1 gene expressions as compared to matched tumor and normal reference sets. Both PD-L1 and PD-1 gene expressions were reported as percentiles. DNA mismatch repair status was predicted by analysis of alterations in five common mismatch repair genes in somatic and germline DNA (MSH2, MSH6, MLH1, PMS2, and EPCAM). If there were no alterations identified in these genes, the mismatch repair status was predicted as microsatellite stable (MSS). Results A total of 13 patients with histiocytic neoplasms were included in the study. The distribution of individual histiocytic neoplasms was as follows: RDD (n=9), ECD (n=3), and LCH (n=1). The median TMB for RDD and ECD patients was 0.17 mut/Mb. For the one patient with LCH, the TMB level was 0.51 mut/Mb. Individual TMB levels are shown in figure 1. PD-L1 and PD-1 expression levels are depicted in table 1. Compared to normal reference sets, the PD-L1 expression was elevated in one patient each with RDD and ECD, and PD-1 expression was elevated in two patients each with RDD and ECD. For both ECD patients with higher PD-1 expression, NGS also showed presence of BRAF-V600E in the tumor tissue. The LCH patient had a low level of PD-L1 and PD-1 expression. For patients where evaluation of DNA mismatch repair was feasible on the tissue specimen (n=4), none showed related somatic or germline alterations. Conclusions In our series, the histiocytic neoplasms RDD, ECD, and LCH demonstrated low levels of TMB. None of the RDD and ECD patients were found to have alterations in DNA mismatch repair genes. Other markers of immunotherapy such as PD-L1/PD-1 expression appeared to be higher in ECD patients with BRAF-V600E, but in only a small subset of RDD patients. The low TMB seen in our study suggests that these histiocytic neoplasms may be less likely to respond to immune checkpoint inhibitors such as anti-PD-1 and anti-PD-L1 agents as compared to the tumors with high TMB. Disclosures No relevant conflicts of interest to declare.

Abstract Background: Early integration of Palliative Care (PC) have shown to improve survival in some solid malignancies such as lung and colon cancers. In this study we sought to analyze the PC utilization in acute myeloid leukemia (AML) patients. Methods: Using the National Cancer Data Base (NCDB) Participant User File, we extracted patient-level data of patients with AML reported between 2004 to 2015.Patients who received pain management and/or other palliative procedures were considered as PC utilizers. After excluding the cases with acute promyelocytic leukemia t (15, 17) (given its excellent prognosis), we classified the AML into three groups- AML with genetic abnormalities, therapy related AML and AML with dysplasia. Other co-variates included in the analysis were age, sex, race, origin, Charlson-Deyo comorbidity score, insurance status, household income, educational status, facility type (academic vs non-academic) and year of diagnosis. Multivariate logistic regression was performed to identify the predictors of palliative care utilization. Statistical analysis was done using PC SPSS v24.0. Results: Of 102,778 patients, 3.4% (n=3504) received PC as initial treatment. PC utilization increased significantly from 2004-2006 to 2013-2015 (2.5% vs 3.9%, p < 0.001). Individuals of non-Hispanic origin (as compared to that of Hispanics) are more likely to receive PC as compared to that of non-Hispanics. (3.5 vs 1.8%, p<0.001). The patterns of PC utilization at non-academic centers were much higher than that of academic centers (4.4% vs 2.9%, p<0.001). As compared to that of AML, not otherwise specified subtype (3.3%), therapy-related AML (4%) and AML with dysplasia (4.3%) had higher rates of PC utilization, whereas individuals with AML with genetic abnormalities have lesser rates of PC utilization (1.9%) (p<0.001, all comparisons). Significant predictors of PC on multivariable regression are listed in the Table and forest plot. Conclusions: We noticed a differential utilization of palliative care based on the histological subtype of AML. The lesser utilization of PC in AML with certain genetic abnormalities may be attributed to their better overall survival outcomes. Though we noticed an increased trend in PC utilization over the past decade, the overall PC utilization in AML is markedly low and plagued by origin and socio-economic status. Disclosures No relevant conflicts of interest to declare.

Abstract Background: Langerhans cell sarcoma (LCS) is a rare malignant neoplasm of Langerhans cells characterized by multi-organ involvement and an aggressive clinical course.Given the rarity of the disease, the true epidemiological data on LCS are lacking, and most of our knowledge is obtained from institutional case series. In this study, we utilized two national databases- the Surveillance, Epidemiology, and End Results (SEER) Program database (https://seer.cancer.gov/) and National Cancer Data Base (NCDB) to study the incidence, clinical presentation, and outcomes of LCS. Methods: We calculated LCS incidence (cases/10, 000, 000 individuals) from SEER database using SEER*Stat (v 8.3.4; https://seer.cancer.gov/seerstat/) statistical software. SEER is a population-based registry program of the United States National Cancer Institute that covers approximately 28% of the population. We identified LCS cases that were confirmed histologically using International Classification of Diseases for Oncology edition 3 (ICD-0-3) histology code 9756/3 from SEER 18 (2000-2014) registry. As the disease was mandatory reportable only after 2001, we included cases that were diagnosed after 2001 in the analysis. Demographic patterns, clinical presentation, and overall survival (OS) were calculated utilizing the patient level data from the NCDB Participant User File. NCDB is a joint program by the American College of Surgeons and American Cancer Society that represents approximately 70% of newly diagnosed cancer cases in the nation. All patients aged > 18 years with a diagnosis LCS (ICD-0-3 Code 9756/3) from 2004 to 2015 were included. To ensure accuracy of follow-up, we excluded the patients that were not treated at the same facility of diagnosis from survival analysis. Results: A total of 25 cases of LCS were reported to SEER between 2000 and 2014. The overall incidence of the disease was 0.2 per 10,000,000 [95% CI:0.1-0.3] and was similar among males and females (p=0.33). The incidence of the disease did not differ significantly among different races (whites, blacks and other races) (p=0.56). Between 2004 and 2015, 52 patients with newly diagnosed LCS were identified from NCDB and 60% were males. The median age of diagnosis was 62 (range, 19-90 years). The most common sites of the presentation were connective tissue (29%; n=15), reticuloendothelial system (RES) and hematopoietic system (25%; n=13), skin (19%, n=10), gastrointestinal (10%, n=5), bones and joints (8%, n=4), brain (2%, n=1) and breast (2%, n=1). Site specific details were missing in 27% (n=14) of cases. Out of the 52 patients, twenty patients (39%) received chemotherapy as first-line therapy, while 24 (46%) received surgery and 15 (29%) received radiation therapy. After a median follow up of 18 (range, 1-144) months, 27 patients (52%) had died. The one- year survival rate was 62%, and the median OS was 19 months (Fig. 1A). We also calculated the 1-year disease-specific survival and OS using SEER database, and was 74% and 63%, respectively. After censoring the patients with hematopoietic and RES involvement, the patients who were managed with radiation therapy (median dose 3960 cGy, range 750-6400) had a better OS as compared to those who received no radiation therapy (p=0.03, Fig. 1B). On the contrary, after excluding the patients with hematopoietic and RES as primary site, no significant difference in OS was noted between the patients who were managed with and without surgery (17 vs 21 months; p = 0.75; Fig. 1C). In addition, post-surgical radiation or chemotherapy had no significant benefit in median OS (p=0.25). Among patients with hematopoietic and RES involvement, there was no association between OS and receipt of chemotherapy (p = 0.16; Fig. 1D); however, the number of patients who received chemotherapy was small (n=6). Conclusion: This dual-national registry study shows that LCS is extremely rare, and has a poor prognosis. Although the data regarding optimal management of LCS are limited, our study demonstrates that radiation therapy may offer survival advantage in patients without hematopoietic and RES involvement. Disclosures No relevant conflicts of interest to declare.

Abstract Introduction Isolated myeloid sarcoma is a rare form (<1%) of acute myeloid leukemia presenting as extramedullary tumor. Contemporary clinical data are mostly limited to institutional case series. Using the National Cancer Database, the largest public cancer registry covering >70% of all newly diagnosed cancers in the United States, we determined the patterns of anatomical presentation and clinical outcome of myeloid sarcoma. Methods We identified patients with a histologically confirmed diagnosis of isolated myeloid sarcoma from 2004-2013 using International Classification of Diseases for Oncology version 3 (ICD-O-3) code: 9930. To allow at least 1 year of follow-up, only patients diagnosed from 2004-2012 were included in the survival analysis using Kaplan-Meier estimates. Results A total of 746 patients were included in the study. The median age of patients was 59 years (range, 41 to 73) and 56% were males. The anatomical distribution and median overall survival of patients are depicted in the Table. The top 3 most common sites of presentation were connective/soft tissues (31.3%), skin/breast (12.3%), and digestive system (10.3%). Compared to other races, Blacks were more likely to have presentation in bones/joints (11.8% vs 6.3% in others), lymph nodes/spleen (22.1% vs. 9%), and less likely in skin/breast (4.4% vs. 13.8%). Asians were more likely to present with cardiopulmonary/mediastinal disease as compared to other races (13.6% vs. 4.2%). According to outcomes, we can categorize the patients into 3 groups: good (median overall survival >30 months: reproductive and digestive systems), intermediate (median overall survival 15-30 months: head/neck and kidney/bladder/retroperitoneum/adrenal), and poor (median overall survival <15 months: nervous system, connective/soft tissue, and bones/joints). There was no significant difference in overall survival between males and females (P =0.06). Among the races, Blacks had the worst overall survival (P =0.02; Figure). Conclusions This is the largest registry-based study on isolated myeloid sarcoma in the United States. Isolated myeloid sarcoma has a diverse anatomic clinical presentation and the overall survival varied significantly according to sites of presentation and racial subgroups. The results of our study may aid the prognostication of patients for treatment decision making and in the understanding of the biological differences by anatomic sites of presentation. Table Anatomical distribution and median overall survival of isolated myeloid sarcoma Table. Anatomical distribution and median overall survival of isolated myeloid sarcoma Figure Overall survival by sex and race in isolated myeloid sarcoma Figure. Overall survival by sex and race in isolated myeloid sarcoma Disclosures Al-Kali: Onconova Therapeutics, Inc.: Research Funding; Celgene: Research Funding.

Abstract Introduction Plasmacytomas comprise approximately 3% of all plasma cell malignancies. There is a paucity of large studies assessing clinical features and outcomes of this relatively uncommon disease. Our objective was to describe the patterns of clinical presentation and survival of plasmacytomas using the National Cancer Data Base (NCDB). Methods This is a retrospective study of patients with histologically confirmed diagnosis of plasmacytoma from 2000-2011 using International Classification of Diseases for Oncology version 3 (ICD-0-3) code: 9930. Patients who had bone marrow involvement were excluded from the analysis. Plasmacytomas were grouped into two broad categories: extramedullary plasmacytoma (P-EM) and bone plasmacytoma (P-bone) based on their anatomical locations. Overall survival was analyzed using Kaplan-Meier estimates. Results A total of 6,939 patients were included in the study (median age 64 years; range 18-90 years). Approximately 62% of these patients were males and 46% patients were ≥ 70 years. Racial/ethnic distribution of the disease was as follows: 76% Whites, 13% Blacks, 7% Hispanics, and 4% others. The anatomical distribution and survival of patients is depicted in the Table. P-EM comprised 30% of the patients, with remaining 70% presenting as P-bone. Among P-EM, most common sites of presentation were upper aero-digestive tract (41%) and connective/soft tissue (20%). After a median follow up of 65 months, the overall survival of P-EM was significantly better than P-bone (113 vs. 78 months; Figure). Based on outcomes, we can categorize the P-EM patients into 3 groups: good (median overall survival > 120 months: upper aero-digestive tract, lymph nodes, endocrine and digestive system), intermediate (median overall survival 90-120 months: nervous system, eye/orbit, pulmonary), and poor (median overall survival <90 months: connective/soft tissue, cardiac/mediastinum, genitourinary, and hepatobiliary system). Males had better overall survival than females in P-bone (84 vs. 67 months; p<0.0001), but had no significant survival difference in P-EM (109 vs. 116 months; p=0.7). In the P-EM group, median overall survival was worst for Blacks (98 months vs. Whites- 108 months, Hispanics- not reached, others-145 months). Conversely, for P-bone, Blacks had a better overall survival than Whites (Blacks- 99 months, Whites- 72 months, Hispanics- 156, others-84 months). Patients who were treated at an academic center had better overall survival than patients treated at other/community centers both in case of P-EM (130 vs. 108 months) and P-bone (107 vs. 69 months). Conclusions This is the largest registry-based study on plasmacytoma in the United States. Plasmacytomas have varied clinical presentation, along with significantly different survival based on sites of presentation, race, sex, and treatment facility. Our study results point toward important differences in disease biology based on location and may aid in assessing prognosis for planning treatment. Table. Anatomical distribution and median overall survival of plasmacytomas. Table Overall survival based on location (P-EM: extramedullary plasmacytoma; P-bone: bone plasmacytoma) Table. Overall survival based on location (P-EM: extramedullary plasmacytoma; P-bone: bone plasmacytoma) Figure Figure. Disclosures Kumar: Noxxon Pharma: Consultancy, Research Funding; Kesios: Consultancy; Skyline: Honoraria, Membership on an entity's Board of Directors or advisory committees; Onyx: Consultancy, Research Funding; BMS: Consultancy; Array BioPharma: Consultancy, Research Funding; Sanofi: Consultancy, Research Funding; Celgene: Consultancy, Research Funding; Millennium: Consultancy, Research Funding; Glycomimetics: Consultancy; AbbVie: Research Funding; Janssen: Consultancy, Research Funding.

6581 Background: While AC is recommended for all patients with stage IIA, IIB, and IIIA NSCLC, its use and benefit among the elderly population is unclear. Methods: We identified patients with stage IIA, IIB, or IIIA NSCLC using the National Cancer Database from 2006-2014. Patients were divided into age groups ≤65, 66-70, 71-75, 76-80, > 80 years. Trends in AC use, factors influencing AC administration, and outcomes associated with AC were studied. Results: Out of 27368 patients, 13464 received AC and 13904 did not. 11% had stage IIA disease, 50% had stage IIB disease and 39% had stage IIIA disease. AC use was lower with increasing age (49% age ≤65 vs. 3% age > 80 (P < 0.0001)). Temporal use of chemotherapy for each age group was unchanged from 2006-2014. In multivariate logistic regression analysis, factors predictive of lower AC use were older age, Medicare or Medicaid insurance, academic center, and higher Charlson Comorbidity Index (CCI). Stage IIIA (OR 1.4, 95% CI 1.3-1.6) and Stage IIB (OR 1.3, 95% CI 1.2-1.4) had a higher chance of receiving AC than stage IIA. In a Cox proportional hazard model, younger age, female sex, academic center, private insurance, higher income, lower CCI, West coast center and AC were associated with better outcomes. When stratified by age and stage, AC was still associated with better survival (Table). Conclusions: AC utilization in stages IIA, IIB and IIIA NSCLC remains low, despite its association with improved survival in the elderly. Future clinical trials may be needed focused on elderly patients to establish the best regimen to optimize outcomes. [Table: see text]

PURPOSE Time to event is an important aspect of clinical decision making. This is particularly true when diseases have highly heterogeneous presentations and prognoses, as in chronic lymphocytic lymphoma (CLL). Although machine learning methods can readily learn complex nonlinear relationships, many methods are criticized as inadequate because of limited interpretability. We propose using unsupervised clustering of the continuous output of machine learning models to provide discrete risk stratification for predicting time to first treatment in a cohort of patients with CLL. PATIENTS AND METHODS A total of 737 treatment-naïve patients with CLL diagnosed at Mayo Clinic were included in this study. We compared predictive abilities for two survival models (Cox proportional hazards and random survival forest) and four classification methods (logistic regression, support vector machines, random forest, and gradient boosting machine). Probability of treatment was then stratified. RESULTS Machine learning methods did not yield significantly more accurate predictions of time to first treatment. However, automated risk stratification provided by clustering was able to better differentiate patients who were at risk for treatment within 1 year than models developed using standard survival analysis techniques. CONCLUSION Clustering the posterior probabilities of machine learning models provides a way to better interpret machine learning models.