Early Software-Assisted Evaluation Enhances Prediction of Survival in Newly Diagnosed Chronic Graft-versus-Host Disease Patients

Highlight

This multicenter prospective cohort study of 258 newly diagnosed chronic graft-versus-host disease (cGvHD) patients demonstrated that software-assisted (SA) response assessment at 6 months reliably predicts overall survival (OS) and failure-free survival (FFS). SA response correlated well with clinician-based evaluations but was more sensitive in identifying mixed treatment responses, a factor associated with survival outcomes. After 12 months, a substantial proportion of patients could discontinue systemic therapy (ST), underscoring the value of early response monitoring to guide therapeutic strategies.

Study Background

Chronic graft-versus-host disease remains a significant complication following allogeneic hematopoietic stem cell transplantation (HSCT), leading to considerable morbidity and mortality. Despite advances in treatment, reliable early predictors of long-term survival and treatment response are limited. Traditional clinical assessment per NIH consensus criteria guides management but may lack sensitivity in recognizing nuanced response patterns. There is a critical need for standardized tools integrating comprehensive clinical data to enhance early response evaluation and improve prognostication in cGvHD patients.

Study Design

This prospective multicenter observational study enrolled 258 patients newly diagnosed with cGvHD, registered under clinical trial NCT02991846. The primary endpoint was failure-free survival (FFS), with secondary endpoints including overall response rate (ORR) at 6 and 12 months after initiation of best available therapy (BAT) and overall survival (OS). Data collection employed Crosy, a specialized software designed to assemble comprehensive clinical and laboratory data on cGvHD, enabling a standardized, software-assisted (SA) evaluation of treatment response. Parallel clinician-based (CB) assessments following NIH criteria were also conducted at the same time points (6 and 12 months). Among the cohort, 200 patients required systemic therapy (ST), and 167 of these were evaluable for response assessment.

Key Findings

At 6 months post-BAT, the ORR was 62% using SA response criteria, compared to 65% by CB evaluation, with complete response (CR) rates of 19% and 23%, respectively. Partial response (PR) rates were 43% (SA) and 42% (CB), showing substantial concordance (66%) between methods. Multivariate analysis identified the 6-month SA response as a significant independent predictor of OS, outperforming CB evaluation primarily through enhanced detection of mixed responses, which are indicative of heterogeneous disease activity and treatment effect.

At one year, 23% of patients successfully discontinued systemic therapy, with a notably higher discontinuation rate (39%) observed in those maintaining initial ST compared to those switching therapy (9.5%, p < 0.001), suggesting that stable early responders have better treatment trajectories. The 2-year FFS for patients receiving ST was 67%, while the 2-year OS was 85%. Sixteen percent of deaths were directly attributable to cGvHD progression or related comorbidities.

The study supports the reliability of software-assisted cGvHD response assessment as an objective tool for early treatment monitoring and survival prediction. SA evaluation’s ability to detect complex response patterns such as mixed responses has practical implications for optimizing patient management and potentially personalizing therapeutic interventions.

Expert Commentary

Integrating technology-based decision support tools in cGvHD management strengthens clinical assessment by reducing inter-observer variability and capturing subtle changes in disease status. The ability of SA evaluation to independently predict OS highlights its potential role in clinical trials and routine practice. However, further validation in diverse patient populations and real-world settings is warranted. Additionally, exploring the mechanistic basis for differential response patterns identified by SA tools may enhance understanding of cGvHD pathobiology and guide novel therapeutic targets.

Conclusion

This prospective study illustrates that a software-assisted evaluation system for cGvHD response at six months after treatment initiation can robustly predict failure-free and overall survival in newly diagnosed patients. Such tools offer a scalable, objective approach to enhance risk stratification and personalized management in cGvHD, potentially improving long-term patient outcomes.

Funding and ClinicalTrials.gov

The study was conducted under the framework of clinical trial NCT02991846. The original article did not specify funding sources. Further information can be accessed via the trial registration.

References

Olivieri A, Mancini G, Olivieri J, Saraceni F, Guerzoni S, et al. Early software-assisted response predicts survival in a prospective cohort of 258 newly diagnosed cGvHD patients. Bone Marrow Transplant. 2026 Jul 2. PMID: 42393230.

Lee SJ. Advances in chronic graft-versus-host disease biology and therapy. Haematologica. 2020;105(2):271-278. doi:10.3324/haematol.2019.221627.

Filipovich AH, Weisdorf D, Pavletic S, et al. National Institutes of Health consensus development project on criteria for clinical trials in chronic graft-versus-host disease: I. Diagnosis and staging working group report. Biol Blood Marrow Transplant. 2005;11(12):945-956.