This is an early access version, the complete PDF, HTML, and XML versions will be available soon.
Survival Machine Learning Methods Improve Prediction of Histologic Transformation in Follicular and Marginal Zone Lymphomas
by
, , , , , , and
Tong-Yoon Kim
1,2
,
Tae-Jung Kim
3
,
Eun Ji Han
4,
Gi-June Min
2,5,
Seok-Goo Cho
2,5,
Seoree Kim
6,
Jong Hyuk Lee
7,
Byung-Su Kim
8,
Joon Won Jeoung
9,
Hye Sung Won
10 and
Youngwoo Jeon
1,2,* 1
Department of Hematology, Yeouido St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 07345, Republic of Korea
2
Lymphoma and Cell Therapy Research Center, Yeouido St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 07345, Republic of Korea
3
Department of Hospital Pathology, Yeouido St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 07345, Republic of Korea
4
Division of Nuclear Medicine, Department of Radiology, Yeouido St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 07345, Republic of Korea
5
Department of Hematology, Seoul St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 06591, Republic of Korea
6
Department of Oncology, Bucheon St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 14647, Republic of Korea
7
Department of Hematology, Incheon St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 21431, Republic of Korea
8
Department of Hematology, Eunpyeong St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 03312, Republic of Korea
9
Department of Oncology, Daejeon St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 34943, Republic of Korea
10
Department of Oncology, Uijeongbu St. Mary’s Hospital, College of Medicine, The Catholic University of Korea, Seoul 11765, Republic of Korea
*
Author to whom correspondence should be addressed.
Cancers 2025, 17(18), 2952; https://doi.org/10.3390/cancers17182952
Submission received: 11 August 2025
/
Revised: 3 September 2025
/
Accepted: 8 September 2025
/
Published: 9 September 2025
(This article belongs to the Section Clinical Research of Cancer)
Simple Summary
Follicular lymphoma and marginal zone lymphoma are slow-growing non-Hodgkin lymphomas that can sometimes change into a more aggressive type called diffuse large B-cell lymphoma. This change, known as histologic transformation, is linked to worse outcomes, but it is difficult to predict in advance. We used patient data from seven hospitals to train and compare machine-learning models that estimate a risk of transformation. We found that models designed for survival analysis, especially those using advanced machine-learning methods, predicted transformation more accurately than traditional approaches. Adding genetic information further improved accuracy and helped identify important mutations linked to higher risk. These findings could help doctors personalize follow-up schedules and treatments, focusing attention on patients most likely to develop aggressive disease.
Abstract
Background/Objectives: Follicular lymphoma (FL) and marginal zone lymphoma (MZL) are low-grade B-cell lymphomas (LGBCLs) with indolent clinical courses but a lifelong risk of histologic transformation (HT) to aggressive lymphomas, particularly diffuse large B-cell lymphoma. Predicting HT can be challenging due to class imbalances and the inherent complexity of time-dependent events. While there are current prognostic indices for survival, they do not specifically address HT risk. This study aimed to develop and validate survival-based and traditional classification machine-learning models to predict HT in cohorts. Methods: Using a multicenter retrospective dataset (n = 1068), survival models (Cox proportional hazards, Lasso-Cox, Random Survival Forest, Gradient-boosted Cox [GBM-Cox], eXtreme Gradient Boosting [XGBoost]-Cox), and classification models (Logistic regression, Lasso logistic, Random Forest, Gradient Boosting, XGBoost) were compared. The best-performing survival models—XGBoost-Cox, Lasso-Cox, and GBM-Cox—were assessed on an independent test set (n = 92). Model sensitivity was maximized using optimal binary risk cutoff points based on Youden’s index. Results: Survival models showed superior predictive performance than classical classifiers, with XGBoost-Cox exhibiting the highest mean accuracy (85.3%), time-dependent area under the curve (0.795), sensitivity (98%), specificity (83.9%), and concordance index (0.836). Incorporating next-generation sequencing (NGS) data improved model accuracy and specificity, indicating that genetic factors improve HT prediction. Principal component analysis revealed distinct gene mutation patterns associated with HT risk, highlighting DNA-repair genes such as TP53, BLM, and RAD50. Conclusions: This study highlights the clinical value of survival-based machine-learning methods integrated with NGS data to personalize HT risk stratification for patients with FL and MZL.
