Refined Response Criteria Strengthens Prognostic Prediction in Acute GVHD

Reclassified disease response criteria at 28 days of treatment for acute graft-vs-host disease (GVHD) improved prediction of 6-month nonrelapse mortality (NRM) compared with conventional criteria in both first and second lines of therapy, potentially reshaping future clinical trials that evaluate GVHD treatments.¹

In a study published in Blood Advances, investigators developed a refined response (RR) criteria using training and validation sets drawn from 2 large external registries: the Japanese Society for Transplantation and Cellular Therapy (JSTCT) and the Mount Sinai Acute GVHD International Consortium (MAGIC). Specifically, these criteria considered those with GVHD grades 0 or 1 at day 28 who did not require additional systemic therapy as “responders,” and all others as “nonresponders.”

Comparing the RR criteria’s predictive performance with that of the conventional overall response (OR) criteria in first-line treatment, RR yielded higher negative predictive values (NPVs) than OR for 6-month NRM in both the JSTCT (92.0% vs 89.6%; P <.001) and MAGIC (90.9% vs 89.8%; P =.009) validation sets. These differences were even greater in the second line, where RR again outperformed OR in both JSTCT (NPV, 74.5% vs 66.0%; P <.001) and MAGIC (NPV, 86.8% vs 76.1%; P =.004) validation sets. The RR criteria’s superior NPVs translate to increased confidence that a patient who meets the RR definition of response at day 28 is unlikely to die from nonrelapse causes within 6 months, potentially making RR a more dependable end point than OR.

“Our externally validated day 28 RR would serve as a better end point than conventional criteria in future first- and second-line treatment trials,” the authors, Akahoshi et al, concluded the manuscript.¹

The authors specifically attributed these improvements to their reclassification of partial response (PR).

“Classifying persistent but mild skin symptoms as responses and residual lower gastrointestinal GVHD as nonresponses were major drivers in improving the prognostic performance of RR,” added the authors.¹

Data Sources and RR Development

The study included adolescent and adult patients 16 years or older who underwent their first allogeneic hematopoietic cell transplantation (HCT) between 2014 and 2021 in the JSTCT cohort and between 2014 and 2022 in the MAGIC cohort. Patients who received systemic corticosteroids for grade 2 to 4 acute GVHD constituted the first-line treatment cohort, whereas the second-line treatment cohort included patients from the first-line treatment cohort who also received second-line treatment for grade 2 to 4 acute GVHD.

Investigators developed the RR criteria using a training set of patients from the JSTCT registry, a database containing information of patients who underwent HCT at over 300 centers across Japan. Investigators randomly assigned patients 2:1 to either the training or validation set; patients assigned to the training set were subsequently categorized into 10 groups according to GVHD grades, then further by 6-month NRM risk. Validation of the RR criteria involved 4 validation sets from JSTCT as well as MAGIC, where patients were prospectively monitored for acute GVHD at 24 HCT centers in North America, Europe, and Thailand.

Rethinking Response Definitions in Clinical Trials

The short- and long-term effectiveness of emerging treatment candidates for acute GVHD, a common complication following allogeneic HCT, are being evaluated in various clinical trials. Traditionally, these trials assess short-term effectiveness with day 28 OR, comprising complete responses (CRs) and PR, both of which have been shown to correlate with long-term outcomes in the first-line setting.² However, the same correlation has not been consistently observed in long-term or refractory settings.³

Further, PR encompasses a wide range of clinical presentations, often aggregating patients with varying degrees of symptom improvement or resolution, some of whom ultimately experience poor outcomes.¹

As the current standard response criteria can lead to inaccurate representations of disease control, the rationale for this study stemmed from a need for a more accurate end point that could thereby serve as a better surrogate for long-term survival, especially in later lines of treatment. The RR end point improves upon standard criteria by better discriminating patients who are on a favorable trajectory from those at high risk of NRM, which may inform both trial design and clinical decision-making.

The authors noted that while the RR end point has a low specificity which may require adjustment of trial sample size, its high sensitivity may minimize the likelihood of false-positive cases, correctly identifying patients who appear to respond but still have high mortality risk under the conventional OR definition.

“This trade-off could be particularly advantageous when deciding whether new agents tested in small, single-arm trials should proceed to larger, randomized studies,” the authors noted.¹

REFERENCES:

1. Akahoshi Y, Inamoto Y, Spyrou N, et al. Refinement of day 28 treatment response criteria for acute GVHD: a collaboration study of the JSTCT and MAGIC. Blood Adv. 2025;9(18):4640-4653. doi:10.1182/bloodadvances.2025016233

2. MacMillan ML, DeFor TE, Weisdorf DJ. The best endpoint for acute GVHD treatment trials. Blood. 2010;115(26):5412-5417. doi:10.1182/blood-2009-12-258442

3. DeFilipp Z, Kim HT, Spyrou N, et al. The MAGIC algorithm probability predicts treatment response and long-term outcomes to second-line therapy for acute GVHD. Blood Adv. 2024;8(13):3488-3496. doi:10.1182/bloodadvances.2024012561