Dataset Information

Machine learning-based prediction of one-year mortality after alloHCT identifies the impact of pre-transplant immunity and inflammation.

ABSTRACT: Accurate prediction of mortality after allogeneic hematopoietic stem cell transplantation (alloHCT) is essential for individualized treatment decisions, yet existing clinical risk scores capture only a limited number of variables and show modest predictive performance. In our single-center retrospective analysis, we included data from 909 adult patients with hematologic malignancies undergoing alloHCT. We used 31 features to build machine-learning models to predict death within the first year after alloHCT. These features included established clinical risk factors together with pre-transplant lymphocyte subsets and inflammatory markers. Among four models, a random forest algorithm showed the best performance (AUC = 0.773) and retained good generalizability in an independent test set (AUC = 0.748). SHapley Additive exPlanations (SHAP)-based interpretation of the machine-learning models showed that age together with five easily measurable pre-transplant immunological and inflammatory parameters influenced the outcome: pre-transplant CD4⁺, CD8⁺, and B-lymphocyte counts, albumin, and C-reactive protein (CRP) levels. Based on these features, our random forest approach outperformed established clinical risk scores (HCT-CI, EASIX, rDRI, mGPS) in predicting one-year mortality after alloHCT and more effectively distinguished patients at low and high risk of an adverse outcome. Our study shows that machine-learning-based models can not only predict patient outcomes after alloHCT but also serve as powerful tools for data exploration, confirming the prognostic relevance of pre-transplant inflammation while uncovering the critical role of lymphocyte subsets as previously unknown risk factors. External validation in independent multicenter cohorts will be required to confirm generalizability.

SUBMITTER: Meyer T

PROVIDER: S-EPMC12861908 | biostudies-literature | 2025

REPOSITORIES: biostudies-literature

ACCESS DATA

Publications

Machine learning-based prediction of one-year mortality after alloHCT identifies the impact of pre-transplant immunity and inflammation.

Meyer Thomas T Meyer Robert R Hackenberg Maren M Oelke Daniela D Gengenbach Laura L Rummelt Christoph C Wilcken Hauke H Maas-Bauer Kristina K Wäsch Ralph R Duyster Justus J Bertz Hartmut H Duque-Afonso Jesús J Finke Jürgen J Zeiser Robert R Wehr Claudia C

Frontiers in immunology 20260119

Accurate prediction of mortality after allogeneic hematopoietic stem cell transplantation (alloHCT) is essential for individualized treatment decisions, yet existing clinical risk scores capture only a limited number of variables and show modest predictive performance. In our single-center retrospective analysis, we included data from 909 adult patients with hematologic malignancies undergoing alloHCT. We used 31 features to build machine-learning models to predict death within the first year af ...[more]

PMID: 41635842

Similar Datasets

Project description:We previously observed reduced graft survival for kidney transplants having interstitial fibrosis with subclinical inflammation, but not fibrosis alone, on 1-year protocol biopsy. The current study aimed to determine whether fibrosis with inflammation at 1 year is associated with renal functional decline in a low-risk transplant cohort and to characterize the nature of the inflammation. Subjects were living-donor, tacrolimus/mycophenolate-treated transplant recipients without overt risk factors for reduced graft survival (n=151). Transplants with normal histology (n=86) or fibrosis alone (n=45) on 1-year protocol biopsy had stable renal function between 1 and 5 years, while those having fibrosis with inflammation (n=20) had declining glomerular filtration rate and reduced graft survival. Immunohistochemistry confirmed increased interstitial T-cells and macrophages/dendritic cells in the fibrosis with inflammation group. Gene expression was performed on a subset of biopsies in each group and demonstrated increased expression of transcripts related to innate and cognate immunity in transplants having fibrosis with inflammation. Pathway- and pathological process-specific analyses of microarray profiles revealed that, in fibrosis with inflammation, over-expressed transcripts were enriched for potentially damaging immunological activities including Toll-like receptor signaling, antigen presentation/dendritic cell maturation, interferon gamma-inducible response, cytotoxic T lymphocyte-associated and acute rejection-associated genes. Thus, fibrosis with inflammation in 1-year protocol biopsies is associated with reduced graft survival and function and with a rejection-like gene expression signature even in recipients with no clinical risk for inferior outcome. Early interventions aimed at altering rejection-like inflammation may favor improved long-term KTx survival. We analyzed gene expression from a group of 65 renal transplant recipients. Patient groups were carefully selected to include patients on the same immunosuppression (Prograf-MMF-Pred), transplant type (LD), absence of over post-transplant complications (AR, BK, DGF). For each patient a 1 year protocol biopsy was examined by conventional histology and gene expression. By histology the patients were categorized as histologically normal (n=25, i/cg/ci=0), IF/TA (n=24, i/cg=0, ci>0) and IFTA+i (n=16, cg=0, i/ci>0). This dataset is part of the TransQST collection.

Project description:Introduction. Factors contributing to kidney transplant fibrosis remain incompletely understood—particularly in the absence of acute complications. Methods. Baseline and one-year surveillance biopsies from 15 uncomplicated living donor kidney transplants were subjected to microarray and quantitative RT-PCR (qRT-PCR) analyses in order to examine changes in gene expression patterns over time. Biopsy pairs were purposefully selected from allografts with no history of acute complications and were divided into those that were histologically normal (n = 7) and those that had developed subclinical interstitial fibrosis (n = 8) at 1 year. Results. Compared to the paired baseline specimens, expression levels of 3578 probesets were found altered in all the one-year biopsies studied. A large proportion of the upregulated genes in this transplant-associated profile were functionally linked with inflammation, immunity or response to injury. These included components of inflammation-related signaling pathways (integrin, interferon and TLR) as well as individual mediators of inflammatory and immune responses. An additional 2884 probesets demonstrated altered expression in fibrotic grafts only at 1 year. The gene products in this fibrosis-associated profile were also predominantly linked with inflammation and immune function, suggesting exaggerated inflammatory activity within the fibrotic grafts. qRT-PCR analyses confirmed the predicted expression patterns for selected transcripts from the microarray profiles. Conclusions. Transcriptional profiles of histologically normal living donor renal allografts indicate that there is ongoing injury response and inflammation at 1 year compared to the immediate post-transplant period. Subclinical development of interstitial fibrosis during the first post-transplant year is associated with additional upregulation of inflammation-related genes. Keywords: time course, comparative expression We analyzed gene expression from a group of 15 renal transplant patients. All patients had histologically normal time zero biopsy but while 7 remained histologically normal (TxNorm), 8 developed subclinical interstitial fibrosis (GIF/TA) by 1 year. Patient groups were carefully selected to include patients on the same immunosuppresion therapy, transplant type, biopsy histology and absence of overt post-transplant complications (acute rejection, BK, etc). This dataset is part of the TransQST collection.

Dataset Information

Machine learning-based prediction of one-year mortality after alloHCT identifies the impact of pre-transplant immunity and inflammation.

Publications

Machine learning-based prediction of one-year mortality after alloHCT identifies the impact of pre-transplant immunity and inflammation.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets