AG Computational Hematology and Immunotherapy
The Computational Hematology working group valorizes Big Data sources for the improvement and understanding of immune therapies in Hematology and its complications. We apply Artificial Intelligence (AI) approaches, such as Machine Learning and advanced mathematical models to improve our understanding of the underlying disease mechanisms and the acuity of existing models. Beyond the improved understanding of disease, the application of the developed models within clinical decision support systems may improve treatments and quality in patient care with respect to individual therapy outcomes. We develop these advanced models using data generated at our home institution as well as in collaboration with a strong network of national and international collaborators (e.g. EBMT, Harmony Alliance, XplOit).
Currently, our medical focus areas cover several facets of allogeneic hematopoietic stem cell transplantation, such as Graft-versus Host Disease, infectious complications, immune reconstitution and late effects. Allogeneic hematopoietic stem cell transplantation (HCT) is one of the main therapies for intermediate- and high-risk patients with hematologic malignancies (e.g. acute myeloid leukemia, AML). Although HCT is a routine treatment, there are several associated complications, which can impair the clinical outcome:
Infectious complications may arise in form of viremia (e.g. human cytomegalovirus, CMV; Epstein-Barr-virus, EBV; conventional- and multi-drug resistant bacteria and fungal infections). Often these complications relate to demanding conditioning therapies before transplantation, which impair patient’s immunity. We develop advanced kinetics models of infectious complications in order to identify risk patients and improve anti-infectious prophylaxis strategies for better patient outcome.
Acute and/or chronic graft-versus-host disease (GVHD), are today still the major cause of early morbidity and mortality after HCT and have a complex pathophysiology. Our focus in understanding GVHD focuses on Machine Learning-based algorithms to improve classification and the early identification of risk patients.
Immune reconstitution is influenced by several pre- (e.g. disease, age, donor), post- and transplant conditions (stem cell source, conditioning regimen) itself. Important complications after HCT (e.g. infections, relapse, GVHD) may result at least in part to impaired or dysregulated immune reconstitution. In this context we focus on the impact of several factors (e.g. in vivo T cell depletion, infections) and their effects on the process of immune reconstitution.
Late effects after HCT may cover a broad spectrum of morbidity and require intensive follow-up to ensure the results of the HCT procedure. The understanding of this complex condition requires interdisciplinary approaches that need a sound and broad data basis.
- We have developed and successfully validated the first Machine Learning based grading system for acute GVHD on a cohort of over 3000 patients with GVHD from several German transplant centers (Essen, Hamburg, Hannover, Berlin, Heidelberg). Our results were presented as oral presentation at the 25th Annual Meeting of the European Hematology Association: doi: 10.1097/HS9.0000000000000404
- The use of additional T cell depleting therapies along with the conditioning chemotherapy prior to allogeneic stem cell transplantation significantly reduces the incidence of acute and chronic GVHD but may also come along with adverse effects, such as prolonged immunosuppression and infections and eventually increased relapse rates. We have developed a model for optimizing the dosage of anti-T-Lymphocyte globulin as GVHD prophylaxis that may be used for improving patient outcome (Turki et al., Am. J of Transpl. 2019)
- We also investigate the impact of CMV viremia after allogeneic hematopoietic stem cell transplantation. For immunocompromised patients after transplantation, CMV causes important morbidity and mortality. However, there is also evidence that early cytomegalovirus replication may reduce leukemic relapse. The complexity of CMV’s impact on clinical outcome challenges traditional models of disease. Our aim is to better understand the CMV’s impact on individual patients using kinetics- and Machine Learning models for individualized patient outcome prediction.
- Saskias PhD project is focusing on the immune reconstitution of hematopoietic stem cell transplantation recipients. The first aim of her project is to characterize the reconstitution of immune cells for patients with acute clinically relevant reactivation of viruses, such as CMV, EBV or Herpes-Simplex virus (HSV). Her analysis should allow us to understand to what extent, virus reactivations influence the reconstitution and especially in which cell subsets (e.g. T cells, B cells and NK cells). We hypothesized that in case of viral infection the reconstitution of the innate and adaptive immunity is delayed. It is unclear, if the analyzed viruses will show similar effects on the immune system. This first analysis is conducted with retrospectively collected virology and flow cytometry data. In a second step, prospective data from patients receiving a “virus-specific T-cell therapy”, which is established in the transplant-production facility of our clinic, will also be analyzed concerning the reconstitution of immune cells. Afterwards analyzed data should be compared to patients, without this promising therapeutic approach. Via this comparison possible differences on the cellular level should be demonstrated and characterized. Both parts of Saskia’s thesis are well integrated in the context of our group’s research. The results will be integrated in mathematical models or machine-learning approaches, with the objective to find a better care for patients. Saskia presented her preliminary results at several scientific meetings. She received the best abstract award of the DAG-KBT meeting 2019 in Berlin (oral presentation) and a travel grant for the 25th Annual Meeting of the European Hematology Association (doi: 10.1097/HS9.0000000000000404). Furthermore, her research results were presented at 45th Annual Meeting of the European Group for Blood and Marrow Transplantation (EBMT) 2019 and the 2019’ Annual Meeting of the German Society for Haematology and Oncology (DGHO): https://doi.org/10.1038/s41409-019-0559-4, https://doi.org/10.1159/000502425
Dr. med. Dr. phil. Amin Turki is a senior physician at the Department of Bone Marrow Transplantation with a long-standing experience in the diagnosis and treatment of immunosuppressed and transplanted patients. He leads the research group “Computational Hematology and Immunotherapy” and has been BMT project responsible in the on-going E-Health project “XplOit- Semantic support for predictive modelling in predictive medicine”. Dr. Turki studied Medicine and Philosophy at the Universities of Münster, Paris Descartes, Ecole Normale Supérieure and Harvard. He received his clinical training at the University Hospitals of Cologne and Essen.
Dr. rer. nat. Andrej Klassen is a postdoctoral fellow at the Department of Bone Marrow Transplantation. He studied mathematics at the Universität Duisburg-Essen. His PhD and a joint research project at the Alpen-Adria-Universität Klagenfurt was focusing on methods of Quasi-solution for parameter identification problems in non-reflexive Banach spaces.
Saskia Leserer is a PhD Student at the faculty of biology working in the group "Computational Hematology and Immunotherapy". Besides her experience in the field of biology she is also familiar with chemistry. Saskia Leserer studied Chemical Biology (B.Sc. and M.Sc.) at the Technical University Dortmund and started her PhD studies in April 2018. During her PhD phase in this group she is expanding her knowledge in fields like statistics and bioinformatics.
Theresa Graf is a bioinformatician with focus on data analysis and machine-learning algorithms. She is part of the research group "Computational Hematology and Immunotherapy". Theresa Graf studied Bio- und Nanotechnology (B.Sc.) at the South Westphalia University of Applied Sciences and completed her Master’s degree in bioinformatics and systems biology at the Justus Liebig University Gießen.
Aleksandra Pillibeit is responsible for the documentational work in the group "Computational Hematology and Immunotherapy". She is involved in the E-health Project "XplOit" as well as the Harmony Alliance.
Former group members:
- Dr. rer. nat. Evren Bayraktar was a senior mathematician and contributed to our group (2018-2019) with his expertise in mathematical modelling and machine leaning.
- Rashit Bogdanov was a resident in internal medicine and supported our group (2019-2020) with his general knowledge in medicine, especially in hematology.
Recent publications of group members
Turki AT, Klisanin V, Bayraktar E, Kordelas L, Trenschel R, Ottinger H; Steckel NK; Tsachakis-Mück N, Leserer S, Ditschkowski M, Koldehoff M., Fleischhauer K and Beelen DW, “Optimizing anti-T-lymphocyte globuline dosing to improve long-term outcome after unrelated hematopoietic cell transplantation for hematologic malignancies”, American Journal of Transplantation, 2019 Oct 09. https://doi.org/10.1111/ajt.15642
Alashkar F, Vance C, Herich-Terhürne D, Turki AT, Schmitz C, Bommer M, Hüttmann A, Dührsen UC, Vogel U, Röth A, “Serologic Response to Meningococcal Vaccination in Patients with Cold Agglutinin Disease (CAD) in the Novel Era of Complement Inhibition”, Vaccine, 2019 Sep 24. (Epub ahead of print.) https://doi.org/10.1016/j.vaccine.2019.09.033
Turki AT, Bayraktar E, Basu O, Kehrmann J, Tzalavras A, Yi J, Liebregts T, Benkö T, Beelen DW, Steckel NK, “Ileostomy for severe, steroid-resistant graft-versus-host disease of the gastrointestinal tract”, Annals of Hematology, Jul 23. (Epub ahead of print.) https://doi.org/10.1007/s00277-019-03754-3
Turki AT et al. “Machine Learning-based acute GVHD grading” Abstract , 24th Congress of European Hematology Association (EHA) in Amsterdam, Netherlands, June 13-16, 2019 https://doi.org/10.1097/01.HS9.0000561280.44375.d2
Alashkar F, Oelmüller M, Herich-Terhürne D, Turki AT, Schmitz C, Vance C, Dührsen U, Röth A, “Immunosuppressive therapy (IST) in adult patients with acquired aplastic anemia (AA): A single-center experience at the West German Cancer Center over the past 15 years”, European Journal of Hematology, Apr 11. (Epub ahead of print.) https://doi.org/10.1111/ejh.13235
Leserer S et al. “Cytomegalovirus reactivation kinetics and peak titers as novel predictors of survival and relapse after allogeneic cell Transplantation for hematologic malignancies” Abstract, 45th Annual Meeting of the European Group for Blood and Marrow Transplantation (EBMT) in Frankfurt, Germany, March 24-27, 2019. https://doi.org/10.1038/s41409-019-0559-4
Turki AT, Lamm W, Schmidt C, Alashkar F, Beelen DW, Liebregts T, “Platelet number and sustained graft function predict intensive care survival in patients with allogeneic stem cell transplantation”, Annals of Hematology, 2019, 98(3), 811.
Turki AT, Lamm W, Liebregts T, Dührsen U “R-ICE chemotherapy with or without autologous transplantation for elderly patients with relapsed or refractory aggressive B-cell lymphomas“, Oncology Research and Treatment, 2018; 41(9):534-538.
Wohlfarth P, Turki AT, Steinmann J, Fiedler M, Steckel NK, Beelen DW, Liebregts T, “Microbiological diagnostic work-up of acute respiratory failure with pulmonary infiltrates following allogeneic hematopoietic stem cell transplantation: findings in the era of molecular and biomarker-based assays”, Biol. Blood and Marrow Transplantation (2018). https://doi.org/10.1016/j.bbmt.2018.03.007
Weiler G, Schwarz U, Rauch J, Rohm K, Lehr T, Theobald S, Kiefer S, Götz K, Och K, Pfeifer N, Handl L, Smola S, Ihle M, Turki AT, Beelen DW, Rissland J, Bittenbring J and Graft N, “XplOit: An Ontology-Based Data Integration Platform Supporting the Development of Predictive Models for Personalized Medicine”, Studies in Health Technology and Informatics; 247:21-25. 2018
Turki AT, Tsachakis-Mück N, Leserer S, Yi JH et al "Early Cytomegalovirus Reactivation and Donor Constellation Reduce Early and Late Relapse Incidence of Acute Myeloid Leukemia in a Long-Term Study", Abstract 3413 at the Annual Meeting of the American Society of Hematology, ASH 3413 , to be published in Blood
Turki AT et al. “A new Data Integration Platform for the Development of Predictive Models in Stem Cell Transplantation (XplOit)” Abstract A430, 44th Annual Meeting of the European Group for Blood and Marrow Transplantation (EBMT) in Lisbon, Portugal, March 18-21, 2018, to be published in Bone Marrow Transplantation