Data-driven grading of acute graft-versus-host disease

Area of Science:

• Hematology
• Immunology
• Transplantation Medicine

Background:

• Acute graft-versus-host disease (aGVHD) is a major complication following allogeneic hematopoietic cell transplantation.
• Current aGVHD classification systems have limitations in accurately predicting patient outcomes.
• There is a need for refined methods to assess aGVHD severity and multiorgan involvement.

Purpose of the Study:

• To develop and validate data-driven models for aGVHD classification.
• To compare the performance of data-driven grading against conventional methods.
• To refine the assessment of aGVHD phenotypes and their impact on nonrelapse mortality (NRM).

Main Methods:

• Analysis of large, multicenter patient cohorts (n=3019) to identify aGVHD phenotypes.
• Development and validation of data-driven classification models.
• Comparison of data-driven grading with conventional grading using Akaike information criterion and concordance index.

Main Results:

• Data-driven grading significantly outperformed conventional grading in predictive metrics.
• Refined aGVHD assessment with up to 12 severity grades, revealing distinct NRM associations.
• Intestinal aGVHD was confirmed as a critical factor influencing outcomes.
• An online calculator for principal component-derived grading (PC1) was developed.

Conclusions:

• Data-driven aGVHD grading offers a more accurate and granular assessment of disease severity.
• These models enhance risk stratification and can inform clinical trial design and transplant registries.
• The findings support a move beyond single overall aGVHD severity grades towards more comprehensive evaluation.