Page 199 - 2019_01-Haematologica-web

Stem Cell Transplantation
Machine learning reveals chronic graft-versus- host disease phenotypes and stratifies survival after stem cell transplant for hematologic malignancies
Jocelyn S. Gandelman,1,2,3,4 Michael T. Byrne,1 Akshitkumar M. Mistry,3,5 Hannah G. Polikowsky,3,4 Kirsten E. Diggins,2,3 Heidi Chen,6 Stephanie J. Lee,7 Mukta Arora,8 Corey Cutler,9 Mary Flowers,7 Joseph Pidala,10 Jonathan M. Irish2,3,4* and Madan H. Jagasia1,3*
1Department of Medicine, Division of Hematology/Oncology, Vanderbilt University Medical Center, Nashville, TN; 2Department of Cell and Developmental Biology, Vanderbilt University, Nashville, TN; 3Vanderbilt-Ingram Cancer Center, Vanderbilt University Medical Center, Nashville, TN; 4Department of Pathology, Microbiology and Immunology, Vanderbilt University Medical Center, Nashville, TN; 5Department of Neurological Surgery, Vanderbilt University Medical Center, Nashville, TN; 6Department of Biostatistics, Vanderbilt University Medical Center, Nashville, TN; 7Clinical Research Division, Fred Hutchinson Cancer Research Center, Seattle, WA; 8Division of Hematology, Oncology and Transplantation, University of Minnesota, Minneapolis, MN; 9Stem Cell/Bone Marrow Transplantation Program, Dana-Farber Cancer Institute, Boston, MA and 10H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, USA
ABSTRACT
The application of machine learning in medicine has been produc- tive in multiple fields, but has not previously been applied to ana- lyze the complexity of organ involvement by chronic graft-versus- host disease. Chronic graft-versus-host disease is classified by an overall composite score as mild, moderate or severe, which may overlook clin- ically relevant patterns in organ involvement. Here we applied a novel computational approach to chronic graft-versus-host disease with the goal of identifying phenotypic groups based on the subcomponents of the National Institutes of Health Consensus Criteria. Computational analysis revealed seven distinct groups of patients with contrasting clin- ical risks. The high-risk group had an inferior overall survival compared to the low-risk group (hazard ratio 2.24; 95% confidence interval: 1.36- 3.68), an effect that was independent of graft-versus-host disease severity as measured by the National Institutes of Health criteria. To test clinical applicability, knowledge was translated into a simplified clinical prog- nostic decision tree. Groups identified by the decision tree also stratified outcomes and closely matched those from the original analysis. Patients in the high- and intermediate-risk decision-tree groups had significantly shorter overall survival than those in the low-risk group (hazard ratio 2.79; 95% confidence interval: 1.58-4.91 and hazard ratio 1.78; 95% con- fidence interval: 1.06-3.01, respectively). Machine learning and other computational analyses may better reveal biomarkers and stratify risk than the current approach based on cumulative severity. This approach could now be explored in other disease models with complex clinical phenotypes. External validation must be completed prior to clinical application. Ultimately, this approach has the potential to reveal distinct pathophysiological mechanisms that may underlie clusters. Clinicaltrials.gov identifier: NCT00637689.
Ferrata Storti Foundation
Haematologica 2019 Volume 104(1):189-196
Correspondence:
jonathan.irish@vanderbilt.edu or madan.jagasia@vanderbilt.edu
Received: March 17, 2018. Accepted: August 17, 2018. Pre-published: September 20, 2018.
doi:10.3324/haematol.2018.193441
Check the online version for the most updated information on this article, online supplements, and information on authorship & disclosures: www.haematologica.org/content/104/1/189
©2019 Ferrata Storti Foundation
Material published in Haematologica is covered by copyright. All rights are reserved to the Ferrata Storti Foundation. Use of published material is allowed under the following terms and conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode. Copies of published material are allowed for personal or inter- nal use. Sharing published material for non-commercial pur- poses is subject to the following conditions: https://creativecommons.org/licenses/by-nc/4.0/legalcode, sect. 3. Reproducing and sharing published material for com- mercial purposes is not allowed without permission in writing from the publisher.
haematologica | 2019; 104(1)
189
ARTICLE