Temporal clinical trajectories and Day+14 landmark stratification of infection risk after CD19 CAR T-cell therapy
5 affiliations
- 1Department of Hematology, The First People’s Hospital of Yunnan Province, Affiliated Hospital of Kunming University of Science and Technology, Yunnan Province Clinical Research Center for Hematologic Disease, Kunming, China
- 2Yunnan Provincial Clinical Medical Center for Blood Diseases and Thrombosis Prevention and Treatment, The First People’s Hospital of Yunnan Province, Kunming, China
- 3Yunnan Atherosclerosis Cooperation Base of Chinese and Western Medicine, The First People’s Hospital of Yunnan Province, Kunming, China
- 4The Third Affiliated Hospital of Kunming Medical University, Yunnan Cancer Hospital, Peking University Cancer Hospital Yunnan, Yunnan, China
- 5The Second Affiliated Hospital of Kunming Medical University, Yunnan, China
Abstract
BackgroundInfection after CD19-directed chimeric antigen receptor T-cell (CAR-T) therapy develops over a changing clinical course. We examined two related questions: how completed Day+100 registry data characterize the infection-associated clinical phenotype, and whether information available by Day+14 can stratify subsequent infection risk through Day+100.MethodsWe analyzed 3,302 adults with large B-cell lymphoma using a 70:30 development/test partition. An all-history classifier used completed Day+100 records. The corrected Day+14 cohort included 2,876 patients with observation extending beyond the landmark; models used information available by Day+14. Supplemental outer cross-validation repeated development, and logistic recalibration used development cross-fitted scores. A separate common-feature model was evaluated in 465 pediatric and young-adult patients with B-cell acute lymphoblastic leukemia.ResultsInfection was recorded in 735/3,302 adults (22.3%). The all-history test AUROC was 0.724 (95% CI, 0.685–0.765). In the corrected Day+14 test set (862 patients; 113 infections), the dynamic-model AUROC was 0.669 (95% CI, 0.616–0.716) and AUPRC was 0.217 (95% CI, 0.165–0.286). At score cutoff 0.25, the low-score group contained 160/862 patients (18.6%) and 5 infections (3.1%; 95% CI, 1.0–7.1%). Its raw and recalibrated Brier scores were 0.198 and 0.110, compared with 0.114 for the development-prevalence reference. In the cross-disease evaluation, the common-feature model achieved an AUROC of 0.663.ConclusionsCompleted records characterize infection-associated clinical trajectories. Day+14 information identifies a low-score subgroup for prospective evaluation of outpatient monitoring, with its coverage, event rate and calibration defining the conditions for that evaluation.
Related
- Feature-token clinical transformer for 24-h landmark mortality risk stratification in critically ill patients with heart failure: development and multicenter external validation
- Joint Dynamic Trajectories of CEA and CA19-9 After Gastric Cancer Surgery: Prognostic Associations and a 12-Month Landmark Analysis
- Integrating 72-h BUN/Cr trajectories with interpretable machine learning for 28-day mortality after acute stroke: a dual-center landmark study.
- Beyond Landmark Error: Agreement Between AI-Assisted and Manual Cephalometric Tracing at the Level of Clinical Measurements: A Systematic Review
- Temporal trends in the clinical presentation, management, and outcomes of children with congenital cytomegalovirus infection.
BibTeX
@article{yang2026temporal,
title = {{Temporal clinical trajectories and Day+14 landmark stratification of infection risk after CD19 CAR T-cell therapy}},
author = {Liuyang Yang and Liuyang Yang and Liuyang Yang and Liyu Shan and Liyu Shan and Shuai Feng and Shuai Feng and Shuai Feng and Renbin Zhao and Renbin Zhao and Renbin Zhao and Yan Wen and Yan Wen and Yan Wen and Xuezhong Gu and Xuezhong Gu and Xuezhong Gu and Zengzheng Li and Zengzheng Li and Zengzheng Li and Xiangmei Yao and Xiangmei Yao and Xiangmei Yao and Yajie Wang and Yajie Wang and Yajie Wang},
journal = {Frontiers in Cellular and Infection Microbiology},
year = {2026},
month = oct,
doi = {10.3389/fcimb.2026.1933555},
url = {https://doi.org/10.3389/fcimb.2026.1933555},
note = {CC BY},
}
RIS
TY - JOUR
TI - Temporal clinical trajectories and Day+14 landmark stratification of infection risk after CD19 CAR T-cell therapy
AU - Liuyang Yang
AU - Liuyang Yang
AU - Liuyang Yang
AU - Liyu Shan
AU - Liyu Shan
AU - Shuai Feng
AU - Shuai Feng
AU - Shuai Feng
AU - Renbin Zhao
AU - Renbin Zhao
AU - Renbin Zhao
AU - Yan Wen
AU - Yan Wen
AU - Yan Wen
AU - Xuezhong Gu
AU - Xuezhong Gu
AU - Xuezhong Gu
AU - Zengzheng Li
AU - Zengzheng Li
AU - Zengzheng Li
AU - Xiangmei Yao
AU - Xiangmei Yao
AU - Xiangmei Yao
AU - Yajie Wang
AU - Yajie Wang
AU - Yajie Wang
PY - 2026
DA - 2026/10/23
JO - Frontiers in Cellular and Infection Microbiology
DO - 10.3389/fcimb.2026.1933555
UR - https://doi.org/10.3389/fcimb.2026.1933555
AB - BackgroundInfection after CD19-directed chimeric antigen receptor T-cell (CAR-T) therapy develops over a changing clinical course. We examined two related questions: how completed Day+100 registry data characterize the infection-associated clinical phenotype, and whether information available by Day+14 can stratify subsequent infection risk through Day+100.MethodsWe analyzed 3,302 adults with large B-cell lymphoma using a 70:30 development/test partition. An all-history classifier used completed Day+100 records. The corrected Day+14 cohort included 2,876 patients with observation extending beyond the landmark; models used information available by Day+14. Supplemental outer cross-validation repeated development, and logistic recalibration used development cross-fitted scores. A separate common-feature model was evaluated in 465 pediatric and young-adult patients with B-cell acute lymphoblastic leukemia.ResultsInfection was recorded in 735/3,302 adults (22.3%). The all-history test AUROC was 0.724 (95% CI, 0.685–0.765). In the corrected Day+14 test set (862 patients; 113 infections), the dynamic-model AUROC was 0.669 (95% CI, 0.616–0.716) and AUPRC was 0.217 (95% CI, 0.165–0.286). At score cutoff 0.25, the low-score group contained 160/862 patients (18.6%) and 5 infections (3.1%; 95% CI, 1.0–7.1%). Its raw and recalibrated Brier scores were 0.198 and 0.110, compared with 0.114 for the development-prevalence reference. In the cross-disease evaluation, the common-feature model achieved an AUROC of 0.663.ConclusionsCompleted records characterize infection-associated clinical trajectories. Day+14 information identifies a low-score subgroup for prospective evaluation of outpatient monitoring, with its coverage, event rate and calibration defining the conditions for that evaluation.
SN - 2235-2988
ER -