News
Article
GVHD Treatment Response at Day 14 as Effective as 28-Day Model, Say Researchers
Author(s):
The group showed that their 14-day model using an easily measurable calculation was able to predict long-term mortality as well as the traditional model using responses from 2 weeks later.
Although 28-day response rates are often used as a primary end point for clinical trials of graft-vs-host disease (GVHD), researchers of a new study say using 14-day data may be a more attractive option.
Publishing their findings in Transplantation and Cellular Therapy, using data from patients in the Mount Sinai Active GVHD International Consortium, the group showed that their 14-day model incorporating an easily measurable calculation was able to predict long-term mortality as well as the traditional model using responses from 2 weeks later.
“The use of the standard clinical response at day 28 as a decision end point has 2 significant limitations,” detailed the researchers. “First, the approval of ruxolitinib for steroid-resistant GVHD has changed practice so that it is increasingly difficult to continue experimental treatments unless a response is rapidly achieved. Second, day 28 [overall response rate] is used to determine the success of experimental treatments for both high- and low-risk GVHD despite markedly different treatment goals for these subgroups of patients.”
They tested the models among 1144 patients who received treatment for acute GVHD after transplantation between 2015 and 2021. Two-thirds of these patients were used as a training set to create the model and the remaining one-third was used as the validation group.
This study compared 14- and 28-day models for mortality prediction in GVHD | Image Credit: Dzmitry-stock.adobe.com
The researchers created their Mount Sinai model using a recursive partitioning algorithm, categorizing patients in the validation group as favorable or unfavorable based on GVHD grade at onset and on day 14. The researchers found that the model was able to predict 12-month outcomes more effectively than the traditional 28-day model. The 14-day model separated patients with a greater difference in nonrelapse mortality rates (NMR) than the 28-day model (55% and 9% vs 48% and 11%, respectively). The model was also more effective at correctly classifying NRM events among patients considered unfavorable (57% vs 48%).
The researchers noted the classification of grade 2 GVHD at the time of response assessment as a key difference between the 2 models. With the 14-day model, patients with grade 1/2 GVHD at the start of treatment who still had grade 2 disease at day 14 were considered favorable, with a 12-month NRM rate of 13%. Patients were only considered unfavorable if their GVHD became more severe. Similarly, patients with grade 3/4 GVHD at the start of treatment whose disease had a partial response and was reduced to grade 2 at day 14 had a 12-month NRM of 31% and were still considered unfavorable.
From there, the researchers integrated the D14 MAGIC algorithm probability (MAP) biomarker score into their model, finding even more improved predictive value. MAP validated score predicts NRM by combining serum concentrations of suppression of tumorigenicity 2 and regenerating islet-derived 3-α. The combined model identified 3 distinct groups of patients with markedly different rates of NRM: good (8%), intermediate (35%), and poor (76%). Area under the curve, sensitivity, positive and negative predictive value, and net benefit in decision curve analysis were all approved with the model compared with the traditional 28-day model.
“The inclusion of biomarkers in the D14 MAGIC model proved an even better surrogate for long-term outcomes and may allow clinicians and investigators to pursue a variety of treatment goals,” explained the researchers. “For example, investigators may choose to use achievement of both MAGIC good and intermediate status to define success in trials that test investigational agents for high-risk GVHD. Alternatively, investigators may prefer the achievement of good status in trials that de-escalate treatment in low-risk GVHD because NRM is already favorable in these patients.”
The researchers added that the integrated model may help guide providers in continuing therapy for patients whose mild GVHD has not changed over 2 weeks of treatment.
Reference
Spyrou N, Akahoshi Y, Kowalyk S, et al. A day 14 endpoint for acute GVHD clinical trials. Transplant Cell Ther. Published online February 4, 2024. doi: 10.1016/j.jtct.2024.01.079
