Reg3α as a biomarker of GI-cGVHD in two independent cohorts with prognostic significance, irrespective of GI-cGVHD subtype. (A) Plasma concentrations of Reg3α and CXCL9 in patients with (n = 23) and without (n = 88) GI-cGVHD in the UM cohort (n = 111). (B) Plasma concentrations of Reg3α, CXCL9, and ST2 in patients with (n = 63) and without (n = 115) GI-cGVHD in the FHCRC cohort (n = 178). (C) Cohorts combined (n = 289) and dichotomized based on median Reg3α of 72 ng/mL into groups with high levels (≥72 ng/mL; n = 145) and low levels (<72 ng/mL; n = 144). NRM was calculated with relapse as a competing risk from time of sample acquisition (days). (D) Plasma concentrations of Reg3α compared between NIH score of 1 to 3 esophageal (n = 19), upper GI tract (n = 19), lower GI tract (n = 19), and combined (n = 6) GI-cGVHD manifestations in the FHCRC cohort. Plasma concentrations of Reg3α were compared between severe NIH scores of 2 to 3 esophageal (n = 5), upper GI tract (n = 5), lower GI tract (n = 5), and combined (n = 4) GI-cGVHD manifestations in the FHCRC cohort. (E) Proportions of maximum GI score in FHCRC cohort patients with GI-cGVHD divided into extremely high Reg3α levels (≥180 ng/mL; n = 16) and low levels (<180 ng/mL, n = 47). ns, not significant.