Validated graft-specific biomarkers identify patients at risk for chronic graft-versus-host disease and death

Abstract

BACKGROUNDChronic graft-versus-host disease (cGVHD) is a serious complication of allogeneic hematopoietic cell transplantation (HCT). More accurate information regarding the risk of developing cGVHD is required. Bone marrow (BM) grafts contribute to lower cGVHD, which creates a dispute over whether risk biomarker scores should be used for peripheral blood (PB) and BM.METHODSDay 90 plasma proteomics from PB and BM recipients developing cGVHD revealed 5 risk markers that were added to 8 previous cGVHD markers to screen 982 HCT samples of 2 multicenter Blood and Marrow Transplant Clinical Trials Network (BMTCTN) cohorts. Each marker was tested for its association with cause-specific hazard ratios (HRs) of cGVHD using Cox-proportional-hazards models. We paired these clinical studies with biomarker measurements in a mouse model of cGVHD.RESULTSSpearman correlations between DKK3 and MMP3 were significant in both cohorts. In BMTCTN 0201 multivariate analyses, PB recipients with 1-log increase in CXCL9 and DKK3 were 1.3 times (95% CI: 1.1-1.4, P = 0.001) and 1.9 times (95%CI: 1.1-3.2, P = 0.019) and BM recipients with 1-log increase in CXCL10 and MMP3 were 1.3 times (95%CI: 1.0-1.6, P = 0.018 and P = 0.023) more likely to develop cGVHD. In BMTCTN 1202, PB patients with high CXCL9 and MMP3 were 1.1 times (95%CI: 1.0-1.2, P = 0.037) and 1.2 times (95%CI: 1.0-1.3, P = 0.009) more likely to develop cGVHD. PB patients with high biomarkers had increased likelihood to develop cGVHD in both cohorts (22%-32% versus 8%-12%, P = 0.002 and P < 0.001, respectively). Mice showed elevated circulating biomarkers before the signs of cGVHD.CONCLUSIONBiomarker levels at 3 months after HCT identify patients at risk for cGVHD occurrence.FUNDINGNIH grants R01CA168814, R21HL139934, P01CA158505, T32AI007313, and R01CA264921.

Keywords: Bone marrow transplantation; Transplantation.

Figure 1. Cumulative incidences of cGVHD by high and low biomarker scores in BMTCTN 0201 and 1202 cohorts for PB and BM recipients.

Curves comparing high versus low biomarker scores (above and below the median cutpoint): (A) in PB patients from BMTCTN 0201 cohort, score including CXCL9+MMP3+DKK3, P = 0.002; (B) in BM patients from BMTCTN 0201 cohort, score including MMP3+clinical, P < 0.001; (C) in PB patients from BMTCTN 1202 cohort, score including CXCL9+MMP3+DKK3, P = 0.012; (D) in BM patients from BMTCTN 1202 cohort, score including MMP3+clinical, P = 0.002.

Figure 2. CXCL9, DKK3, and MMP3 circulating concentrations in mice with and without cGVHD before and at diagnosis.

(A) CXCL9 concentrations in HCT mice with and without cGVHD (before and at diagnosis). B10BR irradiated recipient mice received 10 × 10⁶ B6 T cell depleted BM cells with 7 × 10⁵ T cells (cGVHD) or without (No). Serum was collected at day 18 (prediagnosis) and 28 (diagnosis) after HCT. CXCL9 was measured with ELISA (Raybiotech), data are shown as mean ± SEM, Mann-Whitney test, cGVHD (n = 21) and no cGVHD (n = 24) in the prediagnosis group, and cGVHD (n = 21) and no cGVHD (n = 19) in the diagnosis group. (B) Using the same model, DKK3 was measured with ELISA (Raybiotech), cGVHD (n = 21) and no cGVHD (n = 24) in the prediagnosis group, and cGVHD (n = 23) and no cGVHD (n = 20) in the diagnosis group. (C) Using the same model, MMP3 was measured with ELISA (R&D system), cGVHD (n = 4) and no cGVHD (n = 6) in the prediagnosis group, and cGVHD (n = 6) and no cGVHD (n = 6) in the diagnosis group.