Plasma biomarkers of risk for death in a multicenter phase 3 trial with uniform transplant characteristics post–allogeneic HCT

Mohammad Abu Zaid, Juan Wu, Cindy Wu, Brent R. Logan, Jeffrey Yu, Corey Cutler, Joseph H. Antin, Sophie Paczesny and Sung Won Choi

Data supplements

Article Figures & Data


  • Figure 1.

    High ST2 and TIM3 are associated with NRM by 2 years. (A) Absolute values on a logarithmic scale for ST2 and TIM3 at days 28, 100, and 180 with their medians and interquartile range, P < .01 for all comparisons except for TIM3 on day 100 (P = .037). (B) ROC curve for NRM by 2 years with AUC for post-HCT day 28 ST2, IL-6, Reg3α, TIM3, the 4 biomarkers combined, and clinical covariates in the combined model. The biomarker panel improves the AUC from 0.74 for the clinical covariates to 0.80 in the combined model. N, patients who did not die from NRM; Y, patients who died from NRM.

  • Figure 2.

    High ST2 is correlated with future NRM and OS. (A) The cumulative incidence of NRM by 2 years stratified by day 28 ST2 levels (high vs low, median cutoff of 31 ng/mL). (B) Kaplan-Meier curve for OS stratified by day 180 ST2 levels (high vs low, median cutoff of 26.2 ng/mL). CI, confidence interval; N, number.

  • Figure 3.

    Low L-Ficolin is correlated with hepatic VOD. (A) Concentrations of day 28 L-Ficolin (median and interquartile range) for patients who developed VOD (Y) and those who did not (N) (P = .005). (B) ROC curves for day 28 L-Ficolin, ST2, HA, and VCAM, the clinical covariates, and the 4-biomarker panel (model). (C) Cumulative incidence of VOD stratified by day 28 L-Ficolin (high vs low).


  • Table 1.

    Patient characteristics (N = 211)

    VariableTac/SirTac/MtxP value
    Number of patients104107
    Underwent transplantation104 (100)107 (100)
    Age, median, y (range)45 (19-59)41 (13-58).079
    Male sex53 (51)48 (45).41
    Primary malignancy.19
     Acute myelogenous leukemia46 (44)40 (37)
     Acute lymphoblastic leukemia38 (37)53 (50)
     Chronic myelogenous leukemia8 (8)9 (8)
     Myelodysplastic syndrome11 (11)5 (5)
     Acute biphenotypic leukemia1 (1)0 (0)
    Disease status at transplantation
     Acute myelogenous leukemia.58
      1st complete remission37 (80)34 (85)
      2nd complete remission9 (20)6 (15)
     Acute lymphoblastic leukemia.86
      1st complete remission31 (82)44 (83)
      2nd complete remission7 (18)9 (17)
     Chronic myelogenous leukemia.60
      Chronic phase7 (88)7 (78)
      Accelerated phase1 (13)2 (22)
     Acute biphenotypic leukemia
      1st complete remission1
    Karnofsky score.97
     90-100%72 (69)77 (72)
     <90%32 (31)30 (28)
    Recipient-donor CMV status.028
     +/+41 (39)33 (31)
     +/−21 (20)34 (32)
     −/+9 (9)18 (17)
     −/−27 (26)17 (16)
     Missing6 (6)5 (5)
    Recipient CMV status.66
     +66 (63)71 (66)
     −38 (37)36 (34)
    Donor-recipient sex match.77
     Female-male29 (28)30 (28)
    Conditioning regimen
     Cyclophosphamide/total body irradiation85 (82)84 (79).56
     Etoposide/total body irradiation19 (18)23 (22)
  • Table 2.

    Summary of the clinical outcomes of study participants (N = 211)

    OutcomeEvent timeCountTac/SirTac/MtxP value*
    Acute GVHD grade 2-4Never1417170.73
    By day 2822148
    After day 28481929
    Chronic GVHDNever763244.09
    By day 100431
    After day 1001316962
    After day 180955144
    After day 36519109
    After day 730101
    By day 28000
    After day 28231310
    After day 100231310
    After day 18020119
    After day 365954
    After day 730101
    Survival status by 2 y (day 730)Yes (alive)1577681.57*
    No (dead)542826
    By day 28110
    After day 28512328
    After day 100422022
    After day 180271413
    After day 3651275
    After day 730000
    By day 28642
    After day 28862
    Total no. of patientsBiomarker data available211104107
    • * P value from Gray’s test for the cumulative incidence of the outcome, except for OS where the P value is from a log-rank test.

    • Incidence of chronic GVHD was slightly higher in this study compared with the original BMT CTN 0402 study (64% vs 49%).

  • Table 3.

    Correlation between biomarkers and clinical outcomes

    BiomarkerP valueHazard ratioHR lower CLHR upper CL
    Nonrelapse mortality by 2 y*
    Overall survival
    Chronic GVHD§
    • * Proportional hazards model with time-dependent covariates adjusting for treatment group, age, and CMV status.

    • High vs low (see Patients and methods).

    • Proportional hazards model with time-dependent covariates adjusting for treatment regimen.

    • § Proportional hazards model with time-dependent covariates adjusting for treatment group and primary malignancy.

  • Table 4.

    Biomarkers at day 28 and VOD/SOS occurrence

    BiomarkerDay 28 (14 events/197 used)
    Multivariate logistic regression analysis*Biomarker panel*
    P valueAUCP valueAUC
    • * Adjusted for treatment group, performance score, and CMV status.