INTRODUCTION: Wilms' tumor gene 1 is a "pan-leukemic marker" for evaluating measurable residual disease (MRD) in leukemia. The prognostic value in prediction of relapse following chemotherapy or allogeneic hematopoietic stem cell transplantation (allo-HSCT) has been recognized. However, there lack relevant guidelines for WT1 gene quantification as a MRD assessment for patients with acute myeloid leukemia (AML) undergoing allo-HSCT. The optimal monitoring time and threshold of WT1 expression level for MRD assessment following allo-HSCT are still unclear.

METHODS: We retrospectively investigated the dynamic change of WT1 expression level in bone marrow in 151 adult patients with AML following allo-HSCT. All patients were divided into two subgroup including complete remission (CR) and non-remission (NR), which was based on the disease status before HSCT. WT1 expression levels were compared between two subgroups. Meanwhile, Receiver operating characteristic (ROC) curve analysis was performed on WT1 expression levels at different time points and the highest value of WT1 before relapse-related intervention to find the cutoff values for relapse prediction. Furthermore, the cutoffs were tested by cumulative incidence of relapse (CIR), overall survival (OS) and disease-free survival (DFS), which were analyzed through the Kaplan-Meier method.

RESULTS: Compared with complete remission (CR) subgroup, non-remission (NR) subgroup has a trend of a higher expression baseline and a rapidly increasing from 2 to 3 months after allo-HSCT. ROC curve analysis results showed that 0.6% was the optimal cutoff value for relapse prediction at 3 months after allo-HSCT. The cutoff level 0.6% was effective in identifying patients in CR and NR group. Among the patients with CR status (n=131), patients with WT1≥0.6% had a significantly higher 5-year cumulative incidence of relapse (CIR, 27.3% vs. 6.0%; p=0.006), than patients with WT1<0.6%. Meanwhile, among the patients with NR status before allo-HSCT (n=20), patients with WT1≥0.6% had a significantly higher 5-year cumulative incidence of relapse (CIR, 75.7% vs. 40.0%; p=0.021), than patients with WT1<0.6%.Although 0.6% was effective in both group, there was no significant difference in OS between the patients with WT1≥0.6% and <0.6% in CR group (p>0.05). Moreover, ROC analysis was performed using the highest value of WT1 before relapse-related intervention, which showed that 2.25% was the upper threshold for predicting relapse after transplantation. The upper cutoff 2.25% was only effective in CR group. In CR group, patients with WT1≥2.25% had a significantly higher 5-year cumulative incidence of relapse (CIR, 100% vs. 7.1%; p<0.001), than patients with WT1<2.25%. In NR group, there was no significant difference in the 5-year CIR between patients with WT1≥2.25% and patients with WT1<2.25% (CIR, 75.0% vs. 53.8%; p=0.215). Results showed that the cutoff 0.6% was the optimal threshold for NR patients, while the cutoff 2.5% or 1.2% might be the optimal threshold for CR group in relapse prediction. Because the intervention guided by higher cutoff value would decrease the transplant-related mortality which might be caused by more aggressive relapse intervention in CR group.

CONCLUSION: Our results suggest that the relapse predictive WT1 level is different in patients with different risk stratification before HSCT and the 3th months after allo-HSCT is the most meaningful time point for WT1-based MRD monitoring and relapse intervention. Different intervention thresholds should be used according to the pre-transplant disease status. Patients in NR before transplant should strictly use 0.6% as the threshold, while for patients in CR can be appropriately widened.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.