# De novo and quiescent cGVHD are distinguishable in a prognostic biomarker panel

## TL;DR

This study identifies a panel of biomarkers that can distinguish between different forms of chronic graft-versus-host disease after stem cell transplants.

## Contribution

The study introduces a new panel of ten cytokines and chemokines for predicting distinct subtypes of cGVHD.

## Key findings

- BAFF, CCL4, CXCL9, and sRAGE levels differ between de novo cGVHD and tolerant patients.
- Elevated IL-6, IL-17A, PAI-1, and others are linked to quiescent cGVHD.
- The results highlight the biological diversity of cGVHD and the need for multi-biomarker approaches.

## Abstract

Chronic graft-versus-host disease (cGVHD) remains the most significant long-term complication after allogeneic hematopoietic stem cell transplantation (allo-HSCT), despite increasing insights into its pathogenesis. The development of reliable prognostic biomarkers is essential for identifying patients at high risk of developing cGVHD which may benefit from pre-emptive intervention. However, valid biomarkers remain elusive, and cGVHD is typically treated after clinical onset only, when irreversible manifestations such as ocular involvement may already be present. In this exploratory study, we identified ten cytokines and chemokines with potential prognostic value for predicting subsequent cGVHD. Using bead-based multiplex analysis, we assessed serum samples from 60 adult allo-HSCT recipients at day +90 and day +180 post-transplant to identify proteins distinguishing patients who later developed cGVHD from those who remained tolerant. Significant differences were found in the serum levels of BAFF, CCL4, CXCL9, and sRAGE between patients with de novo cGVHD and those without GVHD. In contrast, elevated IL-6, IL-17A, PAI-1, IL-10, CX3CL1, CXCL1, and CCL4 levels were prognostic for quiescent cGVHD compared with patients with resolved acute GVHD only. These findings underscore the biological heterogeneity of cGVHD and the limited value of single-biomarker approaches, emphasizing the need to consider distinct clinical subgroups and prior disease courses in future predictive models.