ALBERT

Description: Evidence from clinical investigations and animal models indicate that chronic graft versus host disease (cGVHD) results from defective thymic generation of functional and self-tolerant T cell populations following hematopoietic stem cell transplantation (HSCT). We have previously demonstrated that the rare subset of T cells that naturally express 2 T cell receptors (TCRs) on the cell surface as a result of incomplete allelic exclusion are predisposed to respond to auto- and alloantigens. Dual TCR T cells disproportionately participate in pathologic alloreactivity in HSCT patients and mouse models of acute GVHD. These findings, combined with observations demonstrating that dual TCR T cells represent a physiologic reservoir of unique TCRs that evade negative selection, prompted us to examine the role of thymic selection and dual TCR T cells in cGVHD. To study the role of post-transplant thymopoiesis in generation of potentially pathogenic dual TCR T cells, we used a mouse model of syngeneic bone marrow transplantation into lethally-irradiated recipients. Radiation-induced damage to the thymic stroma was characterized by disruption of thymic architecture and loss of cortical and medullary thymic epithelial cells (TECs). This damage resulted in significantly increased generation of dual TCR T cells following transplantation of congenically-marked syngeneic T cell-depleted bone marrow. Two-fold increased production of dual TCR T cells persisted for at least 20 weeks after transplantation. These data demonstrate the hazard for production of T cells predisposed to pathogenic reactivity in the post-transplant environment, and suggest that dual TCR T cells could be a source of T cells causing cGVHD. To examine involvement of dual TCR T cells in cGVHD, we analyzed peripheral blood samples from patients after allogeneic HSCT (〉 12 months post-transplant) using our previously utilized pair-wise TCRVa labeling flow cytometry approach. Flow cytometry analysis revealed that dual TCR T cells were present at increased frequencies in patients with cGVHD (n = 10, 8.3% + 1.1%, P = 0.028) compared to patients without cGVHD (n = 3, 2.5 + 1.1%) or healthy age-matched controls (n = 5, 1.9 + 0.4%). Dual TCR T cells from patients with cGVHD had an activated CD69+ phenotype as compared to T cells expressing only a single TCR from the same patient. Single-cell TCRa/TCRb sequencing confirmed the increased frequencies of dual TCR T cells specific to activated T cells in patients with cGVHD. Repertoire analysis of TCRs sequenced from single cells indicated that the increase in dual TCR T cells was polyclonal. The single-cell sequencing approach enabled multiplexed examination of T cell lineage-associated transcription factors and cytokines. Single-cell transcriptional profiling demonstrated that dual TCR T cells demonstrated predominantly pro-inflammatory and cytotoxic phenotypes with expression of Tbet and perforin. This is in contrast to T cells expressing only a single TCR from the same patient, or dual TCR T cells from healthy control patients, which had a quiescent phenotype. These data indicate a role for dual TCR T cells in mediating cGVHD. Together, these results suggest that dual TCR T cells may be an important link between post-transplant T cell development and cGVHD. Disclosures No relevant conflicts of interest to declare.

Description: MRD detected by flow cytometry (FC) or PCR has been associated with key outcomes after HCT for ALL. In a prospective multicenter trial (NCT02646839; Pediatric Blood and Marrow Transplant Consortium [PBMTC] ONC1401), we performed a planned sub-study of NGS-MRD to predict outcomes pre- and post-HCT for ALL patients (n=57, median follow-up 523 [range 58-1198] days post-HCT). In addition, we performed analyses of relevant clinical factors to assess their relationship to EFS. The larger trial studying the role of KIR favorable haploidentical vs. other transplantation approaches for children diagnosed with ALL, acute myeloid leukemia and myelodysplastic syndrome will be reported separately. We evaluated baseline blast samples from 74 patients for dominant BCR/TCR rearrangements and to follow MRD by NGS. Dominant rearrangements were identified in 100% of B-ALL patients, 96.8% (61/63) in BCR and 3.2% in TCR gamma. For T-ALL patients, rearrangements were identified in 62.7% (7/11), with the remaining 37.3% being polyclonal. Patients proceeded to HCT only if they were in morphological remission. Pre-HCT NGS-MRD from bone marrow (BM) was highly predictive of EFS (n=29 P=0.027, Figure 1) and although NGS-MRD from peripheral blood (PB) did not reach statistical significance due to decreased sample size (n=27, P=0.17, Figure 2), it trended similarly. In BM NGS-MRD negative patients, relapse was exceptionally low with all events due to transplant related mortality (TRM). There did not appear to be a benefit of acute graft-vs-host-disease (aGVHD) in NGS-MRD- patients (Figure 3) or chronic (c)GVHD, but sample size was a limitation. Pre-HCT, 10% of the BM samples were MRD+ by FC, but 35% were MRD+ by NGS. Direct comparison of NGS-MRD in BM and PB with FC MRD pre- and post-HCT showed improvements positive and negative predictive power (data not shown). Alpha/beta depleted haploidentical grafts had similar outcomes to other stem cell sources (Figure 4) with decreased incidence of GVHD [aGVHD 〉 grade 2: n=1 (3.3%) and extensive cGVHD: n=1(3.3%)]. TBI (total body irradiation) based myeloablative conditioning (TBI/TT [Thiotepa]/CY [Cyclophosphamide], TBI/CY, or TBI/VP16; ± anti-thymocyte globulin [ATG]) and non-TBI reduced toxicity (Flu [Fludarabine]/Mel [Melphalan]/TT; ± ATG) had identical EFS (P= 0.31). TRM was very low 8.7%(n=5) in this population (n=57)) and rescue of relapse was high for the duration of follow up to date, resulting in similar OS for MRD- vs. MRD+ patients (P= 0.15), likely due to rescue with cell/immunotherapy. We next examined the interaction of obesity, using body mass index (BMI) based on height/weight pre-HCT, in the context of NGS-MRD on EFS. The BMI score was converted to a percentile through population norms for age and gender and defined thresholds published by the Centers for Disease Control and prevention (CDC). Lean patients (〈 85th percentile [%]) overall had better survival than the overweight (OW)/obese (85-94%/≥95%) (Figure 5: P=0.016). But among the lean patients, NGS-MRD still had an effect, with NGS-MRD+ patients had poorer EFS. Same was observed with the OW/obese group, where being NGS MRD+ adds further reduction in survival. Overweight/Obese patients who were pre-HCT MRD- had survival closer to the lean/MRD+ patients. We observed decreased EFS by weight category and NGS-MRD (Figure 6: P= 0.02). In conclusion, NGS-MRD was highly predictive for EFS regardless of HCT preparative approach or graft Alpha/Beta Depletion. NGS improved predictive power for MRD positivity compared to FC. NGS was able identify rearrangements suitable for MRD tracking in all B-ALL and in most of T-ALL samples. In BM NGS-MRD negative patients, relapse was exceptionally low. There did not appear to be a benefit of aGVHD or cGVHD in NGS-MRD- patients. Alpha/beta depleted haploidentical grafts had similar outcomes to other stem cell sources with decreased incidence of severe GVHD. We observed decreased EFS by weight category and NGS-MRD. Lean patients had better survival, after HCT, than OW/Obese patients. But among the lean patients, NGS-MRD+ patients had poorer EFS. Disclosures Abdel-Azim: Adaptive: Research Funding. Dvorak:Jazz Pharmaceuticals: Consultancy; Alexion Inc: Consultancy. Bunin:PRA Health Sciences: Other: Immediate family member employed. Walters:Editas Medicine: Consultancy; TruCode: Consultancy; AllCells, Inc: Consultancy. Cairo:Jazz Pharmaceuticals: Other: Advisory Board, Research Funding, Speakers Bureau; Miltenyi: Other: MTA; Osuka: Research Funding. Kitko:Novartis: Consultancy, Honoraria; Mallinckrodt: Honoraria. Jacob:Adaptive Biotechnologies: Employment, Other: shareholder. Wing:Miltenyi Biotec: Employment. Pulsipher:Medac: Honoraria; Miltenyi: Research Funding; Bellicum: Consultancy; Amgen: Other: Lecture; CSL Behring: Membership on an entity's Board of Directors or advisory committees; Adaptive: Membership on an entity's Board of Directors or advisory committees, Research Funding; Jazz: Other: Education for employees; Novartis: Consultancy, Membership on an entity's Board of Directors or advisory committees, Speakers Bureau.

Description: Turbulent fluxes of latent and sensible heat are important physical processes that influence the energy and water budgets of the North American Great Lakes. Validation and improvement of bulk flux algorithms to simulate these turbulent heat fluxes are critical for accurate prediction of lake hydrodynamics, water levels, weather, and climate over the region. Here we consider five heat flux algorithms from three parent model systems; the Finite-Volume Community Ocean Model (FVCOM, with three different options for heat flux algorithm), the Weather Research and Forecasting (WRF) model, and the Large Lake Thermodynamics Model, which are used in research and operational environments and concentrate on different aspects of the Great Lakes’ physical system. The heat flux algorithms were isolated from each model and driven by meteorological data from four over-lake stations within the Great Lakes Evaporation Network (GLEN). The simulation results were then compared with eddy covariance flux measurements from the same GLEN sites. All algorithms reasonably reproduced the seasonal cycle of the turbulent heat fluxes while the original algorithms except for the Coupled Ocean Atmosphere Response Experiment (COARE) algorithm showed notable overestimation of the fluxes in fall and winter. Overall, COARE had the best agreement with eddy covariance measurements. Simulations with the four algorithms other than COARE were improved by updating the parameterization of roughness length scales for air temperature and humidity to match those used in COARE. Agreement between modeled and observed fluxes notably varied according to the geographic locations of the GLEN sites.

Description: Turbulent fluxes of latent and sensible heat are important physical processes that influence the energy and water budgets of the North American Great Lakes. These fluxes can be measured in situ using eddy covariance techniques and are regularly included as a component of lake–atmosphere models. To help ensure accurate projections of lake temperature, circulation, and regional meteorology, we validated the output of five algorithms used in three popular models to calculate surface heat fluxes: the Finite Volume Community Ocean Model (FVCOM, with three different options for heat flux algorithm), the Weather Research and Forecasting (WRF) model, and the Large Lake Thermodynamic Model. These models are used in research and operational environments and concentrate on different aspects of the Great Lakes' physical system. We isolated only the code for the heat flux algorithms from each model and drove them using meteorological data from four over-lake stations within the Great Lakes Evaporation Network (GLEN), where eddy covariance measurements were also made, enabling co-located comparison. All algorithms reasonably reproduced the seasonal cycle of the turbulent heat fluxes, but all of the algorithms except for the Coupled Ocean–Atmosphere Response Experiment (COARE) algorithm showed notable overestimation of the fluxes in fall and winter. Overall, COARE had the best agreement with eddy covariance measurements. The four algorithms other than COARE were altered by updating the parameterization of roughness length scales for air temperature and humidity to match those used in COARE, yielding improved agreement between modeled and observed sensible and latent heat fluxes.