ALBERT

Description: Background: T-cell acute lymphoblastic leukemia (T-ALL) is a rare aggressive neoplasm accounting for ∼20% of all ALL cases. It is more common in adults than in children, although the incidence decreases with older age. Subclassification of T-ALL cases according to WHO is so far solely based on the immunophenotype. Although a number of common molecular aberrations have been described in T-ALL, a molecular classification of T-ALLs so far is missing. Aim: (1) Molecular subclassification of T-ALL cases based on whole genome sequencing (WGS) and whole transcriptome sequencing (WTS) data. (2) Analysis of correlations between aberrations, subgroups and other parameters. Methods: WGS and WTS were performed in 114 patients. For WGS, 151bp paired-end reads were generated on NovaSeq 6000 and HiSeqX machines (Illumina, San Diego, CA). For WTS, 101 bp paired-end reads were produced on a NovaSeq 6000 system with a yield between 35 and 125 million paired reads per sample. SPSS (version 19.0.0, IBM Corporation, Armonk, NY) was used for statistical analysis. Results: The cohort comprised 114 T-ALL cases (29% female, 71% male) with a median age of 37 years (range 1 - 91 years). Based on mutations (mut), translocations and gene expression, the cases were subdivided into six different groups: group 1 (n=20) was defined by overexpression (oex) of TLX1 (by t(10;14)(q24;q11)/TRAD-TLX1, n=17; t(7;10)(q34;q24)/TRB-TLX1, n=2; inv(10)(q23q24), n=1) and was correlated to a high frequency of mut in NOTCH1 (19/20, 95%, p=0.011 compared to the other T-ALL cases), PHF6 (11/20, 55%, p=0.04) and BCL11B (5/20, 25%, p=0.004). Group 2 (n=9) showed TLX3 oex by t(5;14)(q35;q32)/BCL11B-TLX3 (n=8) or t(5;8)(q35;q24) (n=1). Mut in WT1 (5/9 = 56%, p

Description: Background: Machine Learning (ML) offers automated data processing substituting various analysis steps. So far it has been applied to flow cytometry (FC) data only after visualization which may compromise data by reduction of data dimensionality. Automated analysis of FC raw matrix data has not yet been pursued. Aim: To establish as proof of concept an ML-based classifier processing FC matrix data to predict the correct lymphoma type without the need for visualization or human analysis and interpretation. Methods: A set of 6,393 uniformly analyzed samples (Navios cytometers, Kaluza software, Beckman Coulter, Miami, FL) was used for training (n=5,115) and testing (n=1,278) of different ML models. Entities were chronic lymphatic leukemia (CLL) 1103 (training) and 279 (testing), monoclonal B-cell lymphocytosis (MBL, 831/203), CLL with increased prolymphocytes (CLL-PL, 649/161), lymphoplasmacytic lymphoma (LPL, 560/159), hairy cell leukemia (HCL, 328/88), mantle cell lymphoma (MCL, 259/53), marginal zone lymphoma (MZL, 90/28), follicular lymphoma (FL, 84/16), no lymphoma (1211/291). Three tubes comprising 11 parameters per tube were applied. Besides scatter signals analyzed antigens included: CD3, CD4, CD5, CD8, CD10, CD11c, CD19, CD20, CD22, CD23, CD25, CD38, CD45, CD56, CD79b, CD103, FMC7, HLA-DR, IgM, Kappa, Lambda. Measurements generated LMD files with 50,000 rows of data for each of the 11 parameters. After removing the saturated values (≥ 1023) we produced binned histograms with 16 predefined frequency bins per parameter. Histograms were converted to cumulative distribution functions (CDF) for respective parameters and concatenated to produce a 16x11 matrix per each tube. Following the assumption of independence of parameters this simplification of concatenating CDFs represents the same information as if they were jointly distributed. The first matrix-based classifier was a decision tree model (DT), the second a deep learning model (DL) and the third was an XGBoost (XG) model, an implementation of gradient boosted decision trees ideal for structured tabular data (such as LMD files). The first set of analyses included only three classes which are readily separated by human operators: 1) CLL, 2) HCL, 3) no lymphoma. The second set included all nine entities but grouped into four classes: 1) CD5+ lymphoma (CLL, MBL, CLL-PL, MCL), 2) HCL, 3) other CD5- lymphoma (LPL, MZL, FL), 4) no lymphoma. The third set included each of the nine entities as its own class. Results: Analyzing the three classes from the first set (CLL, HCL, no lymphoma) the models achieved accuracies of 94% (DT), 95% (DL) and 96% (XG) when including all cases. By analysis of cases with prediction probabilities above 90%, DT now reached 97%, DL 97% and XG 98% accuracy, whilst losing 38%, 8% and 6% of samples, respectively. We further observed that accuracy was also dependent on the size of the pathologic clone, which is in line with the experiences from human experts with very small clones (≤ 0.1% of leukocytes) representing a major challenge regarding their correct classification. Focusing on cases with clones 〉 0.1% but considering all prediction probabilities accuracies were 96% (DT), 97% (DL) and 98% (XG), with loss of 5% of samples for each model. Considering cases only with prediction probabilities 〉 90% and clones 〉 0.1% accuracies were 97% (DT), 99% (DL) and 99% (XG) whilst losing 38%, 9% and 9% of samples, respectively. Further analyses were performed applying the best model based on results above, i.e. XG. Analyzing four classes in the second set of analyses (CD5+ lymphoma, HCL, other CD5- lymphoma, no lymphoma) and considering cases only with prediction probabilities 〉 95% and clones 〉 0.1% accuracy was 96% while losing 28% of samples. In the third set of analyses with each entity assigned its own class and again considering cases only with prediction probabilities 〉 95% and clones 〉 0.1% accuracy was 93% while losing 28% of samples. Conclusions: This first ML-based classifier using the XGboost model with transforming FC matrix data to concatenated distributions, is capable of correctly assigning the vast majority of lymphoma samples analyzing FC raw data without visualization or human interpretation. Cases that need further attention by human experts will be flagged but will not account for more than 30% of all cases. This data will be extended in a prospective blinded study (clinicaltrials.gov NCT4466059). Disclosures Heo: AWS: Current Employment. Wetton:AWS: Current Employment.

Description: Background: The pathogenesis of chronic lymphoproliferative disorder of NK-cells (CLPD-NK) is poorly understood. Mutations in the JAK/STAT pathway (especially STAT3) are found in 30% of patients, but the genetic or exogenous drivers responsible for other cases are unknown. Here we comprehensively define the genetic drivers of CLPD-NK by integrated genome and transcriptome sequencing (WGS/WTS) of a large cohort of cases and complementary functional analyses. Methods: We studied 63 CLPD-NK patients (M/F: 42/21, median age: 71 years [35-89 y]) by WGS, WTS and flow cytometry. A validation cohort of 67 patients (M/F: 43/24, median age: 64 years [7-91]) was analyzed by targeted sequencing. To study the role of CCL22 in CLPD-NK pathogenesis, we examined internalization of the CCL22 receptor, CCR4, and cell chemotaxis in response to exogenous wild type (wtCCL22) or mutant CCL22 (mtCCL22: L45R, P79R, IL87_88NF) in CCR4-expressing Ba/F3 cells (Ba/F3-CCR4). To examine potential autocrine/paracrine activity, we exchanged supernatants of Ba/F3-CCR4-wtCCL22 and -mtCCL22 cells and examined CCR4 expression. To examine the in vivo effects of the mutations on proliferation and phenotype, GFP-tagged empty vector, wtCCL22, or mtCCL22-transduced NK-92 cells were engrafted into IL15-transgenic NOD.Cg-Prkdcscid Il2rgtm1Wjl Tg(IL15)1Sz/SzJ (NSG) mice. Human NK-92 cells isolated from spleens of moribund mice were analyzed by WTS and immunophenotyping. Results: WGS of 63 CLPD-NK identified STAT3 mutations in 18 (29%) cases, with mutually exclusive CCL22 mutations (mtCCL22) in 14 (22%) patients. WGS of 4500 hematological malignancies showed that mtCCL22 were only found in CLPD-NK. Recurrent co-mutations in both groups were found in ATM (n=3), FAS (n=2) and TET2 (n=5). Of the remaining patients, 23/31 had one or more mutation including epigenetic regulators (n=12), signaling components (n=7) or TP53 (n=4). Our findings of CCL22 mutations were confirmed in an independent validation cohort with STAT3 mutations in 19/67 (28%) and mtCCL22 in 13/67 (19%). CCL22 mutations were clustered at the conserved leucine 45 and proline 79 residues (Fig. 1A). Sequencing of purified CD3+ T and CD56+ NK cells showed that the mtCCL22 were somatic mutations acquired by the CD56+ NK population. mtCCL22 defined a subgroup of CLPD-NK, with high NCAM1 (CD56) positivity by flow cytometry (p