Prognostic significance of clinical patient characteristics and blood biomarkers in SR- and SD-aGvHD (REACH2)

Identifying biomarkers of prognostic significance in patients with acute graft-versus-host disease (aGvHD) could facilitate more effective treatment regimens.¹ The phase III REACH2 trial (NCT02913261) enrolled patients aged >12 years with steroid refractory (SR) or steroid dependent (SD) aGvHD who had undergone an allogeneic hematopoietic stem cell transplant. Patients either received ruxolitinib 10 mg twice daily or the best available therapy, which was chosen by the investigator from nine options. Baseline and Day 14 models were created to identify patient characteristics and biomarkers which could influence the likelihood of response at Day 28.

GvHD, immune, and inflammatory biomarkers were identified in patients enrolled in REACH2 and measured at baseline and Day 14 to develop models for response.¹

Study design¹

The study design of REACH2 has been outlined in a previous article. For the exploratory analysis, a total of 29 markers/proteins were assessed at baseline and Day 14 of treatment; however, eight of these were excluded from the end analysis. These markers are shown in Figure 1, along with the measurement method.

Figure 1. Biomarkers and measurement methods in REACH2* 

IL, interleukin.
*Adapted from Socie, et al.¹

All assays were performed in multiple batches with quality control samples to ensure consistency and the batch effect was found to be acceptable on assay validation reports. In the final analysis, six biomarkers were excluded for which the baseline measurement was not within the quantification limit. Interleukin (IL) -1β, IL-12p70, IL-13, IL-2, and IL-4 were below the lower limits, and matrix metalloproteinase-3 was above the upper limits. Tumor necrosis factor and IL-10 were also excluded from additional analysis as no significant differences were found over time.

Results¹

At baseline, a total of 295 patients were evaluated for presence of the remaining 21 biomarkers, with 149 of these patients receiving ruxolitinib and 146 receiving the best available therapy. Of the 295 patients, 242 had available data on at least six of the biomarkers at Day 14.

Significant biomarkers

Of the evaluated biomarkers, eight were found to vary significantly between responders and non-responders at both timepoints. These were IL-8, suppression of tumorigenicity 2, hepatocyte growth factor, regenerating islet-derived protein 3α, IL-6, IL-2Rα, tumor necrosis factor receptor superfamily member 1A, and B-cell marker. An additional four biomarkers were significantly different at Day 14: T-cell immunoglobulin mucin receptor 3, osteopontin, elafin, and CXCL10. Among all biomarkers that differed significantly, plasma concentrations were found to be higher in non-responders than responders.

Important covariates

In the baseline model, important covariates were found include patient age, conditioning regimen, and aGvHD biomarkers. Increasing age was found to decrease the probability of a clinical response. Furthermore, use of myeloablative conditioning and elevated levels of all biomarkers apart from CXCL9 and CXCL11 were also associated with a decreased probability of clinical response;

Important covariates were similar in the Day 14 model. Use of myeloablative conditioning and skin involvement at baseline decreased and increased the probability of clinical response, respectively. The biomarker value at Day 14 was also significant to response and, in both models, ruxolitinib treatment increased the probability of clinical response.

Conclusion 

Overall, aGvHD biomarkers demonstrated prognostic value in both models and immune biomarkers had prognostic value in the Day 14 model.¹ Certain patient characteristics were also found to have prognostic value, such as age and skin involvement. Also, treatment with ruxolitinib increased the probability of response at Day 28 compared with the best available therapy. These characteristics impacted prognostic value independently of each other and could have an additive effect. Further studies on the role of these covariates in the GvHD development may provide further insight.¹

This analysis is limited by its categorization of biomarkers (aGvHD, immune, or inflammatory), as some biomarkers may have multiple functions within GvHD development. However, this categorization was necessary due to the number of baseline patient characteristics and biomarkers included.¹ Further studies will be required to confirm these findings and establish how they translate to clinical practice.