Dynamic forecasting module for chronic graft-versus-host disease progression based on a disease-associated...

Predicting chronic graft-versus-host disease (cGVHD) progression has been challenging due to its dynamic nature and the lack of reliable real-time monitoring tools, necessitating substantial investments of time and financial resources for effective management. Consequently, identifying appropriate immune cell subsets or molecules as prognostic or predictive biomarkers for cGVHD is essential.

Building on the pivotal role of B-cell homeostasis in cGVHD progression, we integrated spectral flow cytometry with advanced machine learning algorithms to systematically analyze the relationship between B cells and cGVHD progression. Leveraging the identification of a distinct B-cell subpopulation, we developed cGPS (cGVHD Progress Score), a user-friendly tool based on marker distribution. To validate cGPS, we conducted both retrospective and prospective multi-center studies involving 91 patients (25 non-GVHD and 66 cGVHD cases).

We identified a distinct B-cell subpopulation characterized by CD27⁺CD86⁺CD20^-, which can precisely distinguish cGVHD. Leveraging this discovery, we developed cGPS. The retrospective study highlighted the predictive power of cGPS, achieving an impressive area under the curve (AUC) of 0.98 for identifying non-GVHD patients prone to cGVHD and 0.88 for predicting disease progression in cGVHD patients. Notably, the prospective study highlighted cGPS's effectiveness, as it accurately predicted all instances of cGVHD development or progression within an average of three-month observation window.

These findings validate cGPS as a highly effective and dynamic B cell-based tool for monitoring cGVHD progression, offering a crucial solution for prognosis and prediction of treatment effectiveness. The multicenter approach applied to both retrospective and prospective studies strengthen the reliability and adaptability of our findings. We are confident that cGPS is a highly competitive tool with great potential for clinical application.

This work was supported by grants from the National Key Research and Development Program of China, Stem Cell and Translational Research (2022YFA1105000, 2022YFA1104100); the National Natural Science Foundation of China (82430050, 32130046, 82270230, 81970109); Key Research and Development Program of Guangdong Province (2023B1111050006); Guangdong Basic and Applied Basic Research Foundation (2023B1515020119); Key Scientific and Technological Program of Guangzhou City (2023B01J1002); Pioneering talents project of Guangzhou Development Zone (2021-L029); the Shenzhen Science and Technology Program (KJZD20230923114504008).