A biomarker panel for acute graft-versus-host disease

Abstract

No validated biomarkers exist for acute graft-versus-host disease (GVHD). We screened plasma with antibody microarrays for 120 proteins in a discovery set of 42 patients who underwent transplantation that revealed 8 potential biomarkers for diagnostic of GVHD. We then measured by enzyme-linked immunosorbent assay (ELISA) the levels of these biomarkers in samples from 424 patients who underwent transplantation randomly divided into training (n = 282) and validation (n = 142) sets. Logistic regression analysis of these 8 proteins determined a composite biomarker panel of 4 proteins (interleukin-2-receptor-alpha, tumor-necrosis-factor-receptor-1, interleukin-8, and hepatocyte growth factor) that optimally discriminated patients with and without GVHD. The area under the receiver operating characteristic curve distinguishing these 2 groups in the training set was 0.91 (95% confidence interval, 0.87-0.94) and 0.86 (95% confidence interval, 0.79-0.92) in the validation set. In patients with GVHD, Cox regression analysis revealed that the biomarker panel predicted survival independently of GVHD severity. A panel of 4 biomarkers can confirm the diagnosis of GVHD in patients at onset of clinical symptoms of GVHD and provide prognostic information independent of GVHD severity.

Figure 1

ELISA heat map of discovery set samples. We performed sequential ELISA as described in “Antibody array methods” on samples from 21 GVHD⁻ patients (left panel) and 21 GVHD⁺ patients (right panel) of the discovery set that were analyzed by microarray in Figure S1. Gray indicates that the sample was not assayed for that protein. Levels of PSA-ACT, IL-17, and IL-1β were not detectable and therefore do not appear in the figure. We measured tumor-necrosis-factor-receptor-1 (TNFR1) rather than TNF-α because of its close correlation to TNF-α and documented reproducibility of measurement in previously frozen samples. P values compare GVHD⁺ and GVHD⁻ samples.

Figure 2

Concentrations of 4 individual discriminator proteins, their receiver operating characteristic (ROC) curves, and the composite panel ROC curves. (A) Absolute values on a logarithmic scale of soluble IL-2Rα, TNFR1, HGF, and IL-8 in samples from the training with their medians and ranges. Patients without GVHD (−) versus with GVHD (+): P < .001. (B) Individual ROC curves for IL-2Rα (----), TNFR1 (—), HGF (.-.-.), and IL-8 (.....) and the composite panel (—). (C) ROC curves of the composite panel for the training set (—) and the validation set (- - - -). The cross separates patients into 2 groups based on their predicted probability of GVHD (high and low) with a specificity of 95% in the training set and 94% in the validation set.

Figure 3

Nonrelapse mortality and overall survival stratified by the biomarker panel. (A) The cumulative incidence of nonrelapse mortality (NRM) and overall survival (OS), determined by Kaplan-Meier curve, is plotted according to the predicted probability of acute GVHD: low (—, n = 193) and high (----, n = 89). P = .001 and P = .006 (adjusted for age, donor type, HLA match, and intensity of conditioning) for differences in NRM and OS, respectively. The NRM at 3.5 years is 15% (95% CI, 9%-21%) for the low-risk group and 36% (95% CI, 24%-48%) for the high-risk group. OS at 3.5 years is 53% (95% CI, 45%-63%) for the low-risk group and 33% (95% CI, 22%-48%) in the high-risk group. (B) The cumulative incidence of NRM and OS of the 2 groups is plotted for the validation set: low (—, n = 93) and high (----, n = 49). P values less than .001 and .02 (adjusted as before) for differences in NRM and OS, respectively. The NRM at 3.5 years is 11% (95% CI, 4%-19%) for the low-risk group and 38% (95% CI, 23%-53%) for the high-risk group. OS at 3.5 years is 59% (95% CI, 49%-72%) for the low-risk group and 44% (95% CI, 31%-63%) for the high-risk group.