Genome-wide association scans of complex multipartite traits like the human face typically use preselected phenotypic measures. Here we report a data-driven approach to phenotyping facial shape at multiple levels of organization, allowing for an open-ended description of facial variation, while preserving statistical power. In a sample of 2,329 persons of European ancestry we identified 38 loci, 15 of which replicated in an independent European sample (n=1,719). Four loci were completely novel. For the others, additional support (n=9) or pleiotropic effects (n=2) were found in the literature, but the results reported here were further refined. All 15 replicated loci revealed distinctive patterns of global-to-local genetic effects on facial shape and showed enrichment for active chromatin elements in human cranial neural crest cells, suggesting an early developmental origin of the facial variation captured. These results have implications for studies of facial genetics and other complex morphological traits.

23Accurate inference of genomic ancestry is critically important in human genetics, epidemiology, and 24 related fields. Geneticists today have access to multiple heterogeneous population-based datasets 25 from studies collected under different protocols. Therefore, joint analyses of these datasets require 26 robust and consistent inference of ancestry, where a common strategy is to yield an ancestry space 27 generated by a reference dataset. However, such a strategy is sensitive to batch artefacts introduced 28 by different protocols. In this work, we propose a novel robust genome-wide ancestry inference 29 method; referred to as SUGIBS, based on an unnormalized genomic (UG) relationship matrix whose 30 spectral (S) decomposition is generalized by an Identity-by-State (IBS) similarity degree matrix. SUGIBS 31 robustly constructs an ancestry space from a single reference dataset, and provides a robust 32 projection of new samples, from different studies. In experiments and simulations, we show that, 33 SUGIBS is robust against individual outliers and batch artifacts introduced by different genotyping 34 protocols. The performance of SUGIBS is equivalent to the widely used principal component analysis 35 (PCA) on normalized genotype data in revealing the underlying structure of an admixed population 36 and in adjusting for false positive findings in a case-control admixed GWAS. We applied SUGIBS on the 37

Chronic myelomonocytic leukemia (CMML) is a clonal disorder with a diagnostic challenge encountered frequently in routine hematology consultations, especially when few dysplastic features in bone marrow and/or no cytogenetics or molecular alterations are detected. Thus, easily applicable parameters in peripheral blood analysis are warranted to aid diagnosis. We studied the utility of flow cytometry analysis of peripheral blood (PB) for the diagnosis of CMML type I. PB samples from 16 CMML type I, age- matched normal (n=10) and reactive monocytosis (>1 x109/L) cases (n=9) were studied. A large panel of monoclonal antibodies was used ( FITC:CD36, CD35, CD15 and CD4, CD16,PCP5.5:CD34, CD163, CD13, PC7: CD117, CD38, CD2,CD7, APC: CD300, CD33, CD11B , CD11C ,CD56 , APCH7: CD14 , HLADR V450, CD64 PE and CD45 OC515) in an 8-color combination, including a backbone of 4 colors in all tubes. At least 1x 10(6) events were acquired by FACSCanto II (BD Biosciences, San Jose, USA). The data was analyzed with the Infinicyt software (Cytognos SL, Spain). Monocytes were selected on the automated population separator plot and confirmed by the expression of CD64 and HLADR. Lymphocytes were used as the internal control. As expected, CMML type 1 patients showed higher absolute monocyte counts in PB than reactive and normal cases (p=0.001), and higher percentage of monocytic cells by flow (p=0.001). CD64bright and CD163 expression were equally valid to identify monocytic population in all three groups (correlation r2 = 1). The majority of the monocytic population belonged to the late stages of maturation, being CD64+CD35+CD14+CD300+ in CMML as well as reactive and normal cases (p=0.14). CMML and reactive cases showed significantly lower levels of early stages of monocytic maturation (CD64+CD35-/+ CD14-CD300-) than normal controls ( p=0.04). The majority of type I CMML (69%) had more than 95% "classical" monocytes (CD14+CD16-) in the monocytic compartment, whereas only 22% of reactive cases and none of the normal cases had such condition (p=0.000). Statistically significant differences among the three groups were also found for "intermediate" (CD14+CD16+) and "non-classical" (CD14lowCD16+) monocytic populations, with the normal cases having higher percentage of "non classical" monocytes, and reactive cases higher "intermediate" (CD14+CD16+) monocytes (12% vs. 5% vs. 1%; -p <0.000-, and 2.6% vs. 3.7% vs. 1.4%; -p=0.01- for "non classical" and "intermediate" monocytes in normal, reactive and CMML cases, respectively). As for the aberrant antigenic expression, CD56 was most frequently expressed in monocytes of CMML (75% CMML with >20% CD56 positivity) followed by CD2 (31%) and CD7 (6%). CD56 could also be detected in reactive (22%) and normal (7%) cases, whereas CD2 and CD7 were always negative. Though useful, CD56 expression alone could overestimate the diagnosis of CMML type I whereas the utility of CD7 expression was doubtful due to its low frequency in CMMLs. When the pattern of distribution of "classical", "intermediate" and "non classical" monocytes together with the aberrant expression of CD56 and CD2 were considered, only CMML cases met both criteria (>95% of "classical" monocytes among the monocytic compartment in PB, and having an aberrant expression of CD56 or CD2 in >20% of monocytic cells), excluding all the reactive monocytosis and normal cases, although not all CMML (62%) met both criteria. Concerning the antigens analyzed on the monocytes, only CD163 and CD11b had a higher expression in CMML than in reactive or normal cases (p=0.01 & p=0.05). CD64, CD15, CD4, and CD45 were similarly expressed in CMML and reactive, but higher than in normal cases (p<0.03). CD35 was highest (p=0.01) and HLA-DR (p=0.01) and CD33 (p=0.02) lowest in reactive cases, compared to CMML or normal cases. No significant difference was found for the remaining antigens. In spite of these findings, in our opinion, its prospective utility to diagnose CMML or to distinguish it from reactive monocitosis was doubtful. Flow cytometry analysis can be useful to diagnose CMML type I in the routine laboratory by employing a few easily applicable parameters in PB (the pattern of distribution of "classical", "intermediate" and "non classical" monocytes together with the aberrant expression of CD56 or CD2), distinguishing it from reactive monocytosis, and helping to identify patients in whom other studies, as bone marrow assessment, should be performed. Disclosures Diez Campelo: Novartis: Research Funding, Speakers Bureau; Janssen: Research Funding; Celgene: Research Funding, Speakers Bureau. Puig:The Binding Site: Consultancy; Janssen: Consultancy.

Background:Thrombopoietin‐receptor‐agonists (TPO‐RA) are effective treatments of immune thrombocytopenia (ITP). Previous long‐term TPO‐RA clinical trials have shown that thrombotic events occurred in 6% of TPO‐RA‐treated ITP patients, with thrombotic events appearing to be more frequent in patients of older age and having at least 1 general risk factor for thrombosis.Aims:To evaluate the prevalence of venous and arterial thrombosis in patients with primary ITP during treatment with TPO‐RA.Methods:Multicenter retrospective study that included 121 adult primary ITP patients from 19 secondary and tertiary Spanish hospitals who had initiated treatment with Romiplostim (ROM) or Eltrombopag (ELT) as long‐term therapy between January 2012 and December 2014. Information on patient characteristics was collected from medical records to assess and compare risk factors of ITP patients with and without vascular events (VE).Results:A total of 121 patients (median age 63 years, range 19–96 years; 68% chronic phase), initiated TPO‐RA (ROM 54; EPG 67). During a 329.3 patient‐year time under treatment (exposure to ROM and ELT of 161.03 and 168.27 patient‐year, respectively) 15 patients experienced 17 vascular episodes (9 arterial, 8 venous). One patient presented antiphospholipid antibodies, five had been diagnosed with neoplasia, one with vascular peripheral disease, two with hypothyroidism –one of whom also had renal disease‐. Seven events occurred with ROM, and 10 with ELT. The annualized risk was 4.2 and 5.9 VE/100 patient‐years in ROM and ELT treated patients, respectively (median 5.2). Most VE occurred in the first year of TPO‐RA therapy (median 276 days; 5–1183), with a trend toward earlier events under ROM than ELT (127 days vs 360 days, respectively; p = 0.070). In the case of ischemic events the median time to arterial events was 165 vs. 606 days in ROM and ELT treated patients; P = 0.029. There were no significant differences in the 15 patients vs. the 106 patients that did not suffer from vascular events in terms of gender, age, diabetes, hypertension, previous vascular events, nor time on prednisone as 1st line therapy. In patients experiencing VE on TPO‐RA, a significant association with previous splenectomy (53.3% vs. 25.5%, P = 0.026), and chronic phase of the disease (93.3% vs. 64.1%, P = 0.024) compared with those not having such characteristic was detected. Surprisingly previous malignancy significantly associated with VE under TPO‐RA treatment (33% vs. 2.8%, P = 0.000009). All patients were reported to have sustained complete remission of previous neoplasias (1 colon carcinoma; 1 Burkitt lymphoma; 1 bladder cancer; 1 breast cancer; 1 patient with history of thyroid, breast and ovarian tumors) and were not receiving antineoplastic therapies. In multivariate analysis with logistic regression only previous malignancy predicted significantly higher odds of VE (Table 1).Summary/Conclusion:In this study, we describe an annualized risk of 5.2 vascular events/100 patient‐years in TPO‐RA treated patients. Venous and arterial thromboembolism are not frequent complications in ITP patients under TPO‐RA, except in particular settings, such as in splenectomized and chronic patients. Our data revealed a novel association of history of previous neoplasia with thrombotic events in patients with TPO‐RA. These results suggest that a history of prior cancer should be systematically screened before TPO‐RA initiation, and if so, alternative therapy should be considered; however, due to the limited sample size, further research is warranted.image

Background:Thrombopoietin receptor agonists (TPO‐RA) are a well‐established treatment in patients with primary immune thrombocytopenia (ITP). Sustained treatment‐free responses (TFR) after TPO‐RA discontinuation in adult ITP have been reported; yet, to date there are no predictors to identify in which patients this approach is likely to be successful.Aims:To evaluate clinical predictors of TFR in a real world cohort of ITP patients treated with TPO‐RA.Methods:Patients aged >18 years with primary ITP who had initiated TPO‐RA (eltrombopag [EPG] or romiplostim [ROM]) treatment between January 2012 and December 2014 were included in this retrospective, multicenter study from 19 secondary and tertiary Spanish hospitals. Data on patient characteristics, history of disease and previous therapies, TPO‐RA administration, response and discontinuation were collected from medical records.Results:A total of 121 patients with a median age of 63 years (range 19–96 years), 59% females initiated TPO‐RA as long‐term therapy (ROM 54; EPG 67). Sixty‐eight percent of the patients treated with TPO‐RA met criteria for chronic, 16% for newly diagnosed and 16% for persistent ITP. The median time on TPO‐RA treatment was 35.2 months (1 to 67.3 months), and the median follow‐up from start of TPO‐RA until collection of data was 44.9 months (23.8 to 67.5 months). A total of 39 patients (32.2%) switched TPO‐RA during follow‐up. The most frequent cause for switching was lack of efficacy (48.7% of cases ‐in 89.5% the initial TPO‐RA was EPG‐). Due to switching the exposure to both TPO‐RAs was similar during follow‐up; 80 patients received ROM and a similar number was treated with EPG, with total exposure (years) of 161.0, and 168.3, respectively.During follow‐up almost one half of the patients (46.3%, n = 56) tapered‐off the TPO‐RA; 10 out of the 37 cases that discontinued ROM had previously received EPG, and 4 out of the 19 that stopped EPG had switched from ROM. After a median of 432 days (29–1344) under TPO‐RA treatment, 35 patients (28.9%) maintained TFR defined as platelet counts >50x109/l for more than 6 months. In 4 of these 35 cases, TPO‐RAs were reintroduced due to loss of response after a median TFR of 20.5 months. Potential predictors of TFR in the remaining 31 patients (25.6%) with sustained platelet counts in the absence of any agent meant to increase platelet count (median 32 months, 8–217) were analyzed. No specific patient feature (e.g. age, comorbidities, ITP duration, previous treatments), bleeding, previous therapies nor phase of disease seemed to consistently predict for sustained response off therapy. However, univariate analysis (Chi‐square) did identify statistically significant predictors of TFR. Interestingly, while the specific TPO‐RA that was discontinued did not influence the probability to achieve TFR, receiving ROM as first TPO‐RA was positively associated with TFR (P = 0.010), while switching TPO‐RA negatively predicted sustained platelet responses (P = 0.002). In multivariate analysis with logistic regression both variables predicted significantly higher odds of TFR (Table 1).Summary/Conclusion:Platelet response following TPO‐RA cessation is sustained in 25.6% of adult patients with primary ITP. Although published studies have not identified a predictive factor of sustained response after TPO‐RA discontinuation, in this long‐term follow up analysis, ROM as first TPO‐RA being administered, and no need of TPO‐RA switching are factors that positively correlate with the probability to achieve TFR.image

Background: Recently a range of alternative novel therapies have been developed to improve treatment options for patients with Hemophilia A. One approach is to generate a Factor VIII (F.VIII) mimetic molecule using a humanised bispecific antibody, as was done for Hemlibra. Taking advantage of Kymab's fully human antibody discovery platform, we describe the selection and optimisation of an FVIII mimetic common light chain (CLC) bispecific antibody which can similarly catalyse the generation of Factor Xa (FXase) and normalise the activated partial thromboplastin time (aPTT).Aims: To generate a functionally active F.VIII mimetic bispecific antibody for Haemophilia A treatment Methods: F.IX and F.X. binding antibodies were generated by immunizing Kymouse ® , which contain the full human immunoglobin repertoire, with F.IX or F.X, respectively. Isolated F.IX and F.X specific arms were co-expressed as 2-heavy-2-light-chain (2H2L) bispecific antibodies. Purified 2H2L bispecifics were screened using a high-throughput chromogenic FXase assay. The light chain of a promiscuous F.IX arm was chosen to generate transgenic mice expressing this bespoke common light chain (CLC) in the Kymouse ® background. By immunizing these transgenic mice with F.X, F.X binding antibodies containing the CLC were recovered. The heavy chains of these F.X binding antibodies were co-expressed with the heavy/light chains of the chosen F.IX arm as CLC bispecific antibodies. One biologically active CLC bispecific antibody was identified by functional assays, and chosen for further optimization. The optimization of the lead bispecific antibody, KY1049, was achieved by data mining of next generation sequencing information using Kymab's IntelliSelect ® bioinformatic platform, coupled with site-specific mutagenesis. Results: More than 8,000 2H2L bispecifics were screened by a chromogenic FXase assay to select the most active and versatile F.IX arm (Figure 1A). More than 400 F.X heavy chains subsequently isolated from the bespoke CLC Kymouse ® were screened to identify a highly active CLC bispecific (Figure 1B). Further optimisation of the molecule was carried out to iteratively increase FVIII mimetic activity by deep data-mining of heavy chain NGS sequence data, or site-specific mutagenesis. The combinatorial optimization process resulted in a highly functional CLC bispecific, KY1049 with comparable F.VIII mimetic activities to a sequence-identical analogue of Hemlibra (Figure 1C). Summary/Conclusion:Our bispecific antibody discovery platform consisting of four-chain matrix screening, common light chain transgenic mouse technology, B cell network analysis and site-specific mutagenesis was applied to develop KY1049, a potent FV.III mimetic bispecific antibody, which shows equivalent activity to Hemlibra in vitro and holds promise as a future therapeutic in Haemophilia A.

Introduction: Diffuse large B-cell lymphoma (DLBCL) is one of the most common malignant neoplasms in elderly patients, potentially curable when optimum treatment is administered. The combination of rituximab with CHOP chemotherapy (R-CHOP) is considered standard for these patients, but randomized studies published to date are limited to the range of age from 60 to 80 years, so that in patients over this age treatment election is not so clear, usually opting for palliative treatment or a "full" treatment at a reduced dose. This retrospective study is primarily aimed to analyze the influence of the type of treatment and comorbidity scales in overall survival (OS) of a large series of patients >80 years with aggressive B-cell lymphoma. Methods: Eligible patients were aged ≥ 80 years, diagnosed of DLBCL, follicular lymphoma grade 3B or transformed lymphoma. The main patient characteristics were obtained retrospectively from the medical records, including a complete geriatric assessment (CGA, "comprehensive geriatric assessment") and the Charlson comorbidity index. The Ethics Committee of the University Hospital of Salamanca approved the study. Results: 288 patients from 19 GELTAMO hospitals were registered in the study, of which 234 (60% women) were evaluable and have been included in this preliminary analysis. The median age was 84 years (80-94) and the vast majority (94%) were DLBCL. According to the Charlson index, 65% of patients were low-intermediate risk, and according to CGA, 63% of patients were considered "fit". A higher proportion (60% v 44%, p = 0.03) of patients with low or intermediate comorbidity index were treated with a curative intent (CHOP +/- rituximab), as compared with patients with high or very high index. With a median follow up of 41 (range 9-142) months, the median OS was 11.5 months (33% estimated at 3 years). The median OS for patients treated with R-CHOP-like (N=96) was 35.3 months, significantly better (p <0.001) than those achieved with CHOP-like (n=23, 7.9 months), R-CVP (n=20, 6.9 months) or cyclophosphamide- prednisone +/- vincristine (n=69, 6.2 months). Charlson comorbidity index and CGA scale also had a significant influence on OS (median of 14.6 vs. 6.1 months for patients with low or intermediate versus high or very high risk, p = 0.006; and 18 vs 6.6 months for patients "fit" versus "non-fit", p = 0.006). In the multivariate analysis, treatment with R-CHOP-like (RR = 0.4; 95% CI: 0.3-0.6) and IPI <3 (RR = 0.4; 95% CI: 0.3-0.6) had an independent positive influence on OS. Conclusions: In patients over 80 years with DLBCL, treatment with R-CHOP-like was associated with the best results in terms of OS. Therefore, its administration must be considered whenever possible. Disclosures Sancho: CELLTRION, Inc.: Research Funding.