A New Composite End Point for Acute GVHD: The MAGIC Response

A newly developed composite response metric, the MAGIC Composite Response (MCR), has been shown to more accurately predict long-term outcomes for patients with acute graft-vs-host disease (GVHD) compared with the current gold standard of clinical response alone (CRO).¹The MCR system, which integrates both clinical symptoms and serum biomarkers, may serve as a superior surrogate end point for future clinical trials, according to a study published in Blood Advances.

Acute GVHD remains a significant source of morbidity and mortality following allogeneic hematopoietic cell transplantation (HCT).²For decades, response to treatment has been assessed primarily by the reduction in clinical symptoms by day 28 (D28) post treatment initiation. However, these clinical criteria have limitations, including a modest ability to predict nonrelapse mortality (NRM) and a failure to account for the disproportionate influence of gastrointestinal (GI) tract involvement on the risk of NRM.

To address these limitations, the Multidisciplinary Approach to GVHD International Consortium (MAGIC) previously developed the MAGIC Composite Score (MCS), a grading system that integrates clinical and biomarker data at the onset of GVHD to predict treatment response and NRM.¹ The new study, a follow-up to this work, sought to determine if this composite approach would also be valuable in assessing treatment response on D28.

Methods and Analysis

The study analyzed data from 1135 adult and pediatric patients who underwent HCT between 2014 and 2023 and received systemic treatment for acute GVHD. The patient data were divided into a training cohort (n = 826) and a validation cohort (n = 309) to account for evolving trends in HCT, such as increased use of posttransplant cyclophosphamide(PTCy). The researchers used a classification and regression tree (CART) model with 6-month NRM as the primary end point to identify responder and nonresponder groups.

The MCR system was defined by a patient's status at D28 as either MCS 0 or MCS 1, which the study identified as a response.Patients with complete resolution of symptoms and low-risk biomarkers (Ann Arbor 1) were classified as MCS 0, experiencing a very low 6-month NRM of 3.7%. Patients with low 6-month NRM were classified as MCS 1. This composite approach was then directly compared to CRO. The study did not specify a particular drug but referred only to "systemic treatment.”

Key Findings and Clinical Implications

In the validation cohort, MCR demonstrated superior predictive accuracy for 6-month NRM compared with CRO (AUC, 0.77 vs 0.69; P =.014). This improvement was attributed to MCR's ability to correctly reclassify both clinical responders and nonresponders.

A significant number of patients (13% of clinical responders in the validation cohort) who were considered responders by clinical criteria alone were reclassified as nonresponders by MCR. This subgroup experienced a 5-fold higher NRM than MCR responders (34.3% vs 6.8%, P <.001). Most of these reclassified patients had high biomarker levels at D28, indicating ongoing subclinical disease activity despite symptomatic improvement.

Conversely, MCR reclassified a larger group of patients (30% of clinical nonresponders) from nonresponders to responders. This group experienced a 6-fold lower NRM than MCR nonresponders (7.6% vs 50.7%, P <.001). These patients typically had unresolved mild skin and/or upper GI symptoms but low biomarker levels, suggesting that their clinical symptoms were not reflective of a high risk of death.

The study found that the MCR metric was consistently superior to CRO in predicting NRM across various time points and in patient subsets mirroring different clinical trial designs. It also showed superiority over using D28 biomarkers alone.

“Importantly, the MCR system includes patients as responders with low biomarkers and mild but stable symptoms, who have favorable outcomes, even though they would be classified as nonresponders by clinical response criteria alone. Thus, a composite status integrating clinical symptoms and biomarkers after 28 days of treatment better predicts long-term outcomes than conventional criteria that rely only on changes in clinical symptom severity,” study authors concluded.

FAQs

What is acute graft-vs-host disease (GVHD), and why is assessing treatment response important?

Acute GVHD is a severe complication that can occur after an allogeneic hematopoietic cell transplantation (HCT), where transplanted immune cells attack the recipient's tissues. It is a significant cause of illness and death. Accurately assessing a patient's response to treatment is crucial for predicting long-term outcomes, guiding further therapeutic decisions, and evaluating the effectiveness of new treatments in clinical trials.

What are the limitations of current methods for assessing acute GVHD treatment response?

Currently, the "gold standard" for assessing acute GVHD treatment response in clinical trials relies on changes in clinical symptoms observed on day 28 (D28) after treatment initiation. However, these "clinical response only" (CRO) criteria have notable limitations. They provide only a modest prediction of nonrelapse mortality (NRM) because they treat all affected organs equally, even though gastrointestinal (GI) tract involvement has a disproportionately higher impact on NRM risk. This means that clinical improvements might not always reflect a true reduction in the risk of serious complications or death.

How does the MAGIC Composite Response (MCR) system perform compared to the CRO approach?

In a validation cohort, the MCR system demonstrated superior predictive accuracy for 6-month NRM compared with CRO. The MCR had a higher area under the curve (AUC; 0.77 vs 0.69; P =.014) and showed improved negative and positive predictive values. This means MCR is better at identifying both patients who will and will not experience NRM. The MCR also created a greater separation in NRM between responders and nonresponders, indicating its enhanced ability to stratify patient risk.

This article was created with assistance from NotebookLM. The content has been reviewed and edited by Targeted Oncology staff. For more information on the extent and nature of AI usage, please contact us.

REFERENCES:

1. Akahoshi Y, Portelli J, Katsivelos N, et al. The MAGIC Composite Response: A Novel Endpoint Integrating Clinical and Biomarker Parameters for Acute GVHD. Blood Adv. 2025; bloodadvances.2025017116. doi:10.1182/bloodadvances.2025017116.

2. Zeiser R. Advances in understanding the pathogenesis of graft-versus-host disease. Br J Haematol. 2019 Dec;187(5):563-572. doi: 10.1111/bjh.16190. Epub 2019 Oct 6.