ADAM28, a member of the ADAM family of metalloproteinases, is over-expressed in several human tumors and is related to cell proliferation and metastasis. Our previous study has demonstrated that the expression level of ADAM28 is significantly elevated in patients with relapsed acute lymphoblastic leukemia, which was associated with poor prognosis. However, the impact of ADAM28 on relapse and the prognosis of patients with AML remains unclear.

Aims: To investigate the effect of ADAM28 on the growth and dissemination of leukemia cells and to identify the prognostic significance of ADAM28 levels in patients with AML.

Methods and results

From 2012-2013, 189 de novo AML patients were prospectively enrolled in this study. The expression of ADAM28 in the leukemic cells of AML patients at diagnosis was significantly higher than that in the donor BM cells. No correlations were found between the expression level of ADAM28 and either FAB classification or cytogenetic risk groups. However, the expression levels of ADAM28 differed significantly between patients suffering a relapse and those remaining in CR. Furthermore, the ADAM28 levels in the cerebrospinal fluid (CSF) of patients with central nervous system leukemia (CNSL) were significantly higher than those in patients without CNSL. These data suggested that ADAM28 levels might be related to the incidence of relapse in patients with AML.

We further investigated whether ADAM28 could impact the proliferation, migration and invasiveness of leukemic cells in vitro. Primary AML cells with high ADAM28 expression levels have better proliferation, migration and invasion capacities than those with low ADAM28 expression levels. Knocking out ADAM28with aCRISPR/Cas9 lentivirus significantly inhibited the proliferation, migration and invasion in leukemic cells. The increased expression of ADAM28 lead to more prolific IGFBP-3 degradation and IGF-IR phosphorylation, whereas the ADAM28 knock out cells resulted in significant down-regulation of IGFBP-3 degradation and IGF-IR phosphorylation in leukemic cells.

In a xenotransplantation mice model, primary cells with elevated ADAM28 expression have improved engraftment ability in hematopoietic tissue and enhanced dissemination into nonhematopoietic tissue compared with primary cells with lower ADAM28 expression. Blocking ADAM28 expression in leukemic cells ameliorated AML growth and dissemination after xenotransplantation.

We then analyzed the prognosis of the cohort of AML patients. Patients were divided into a high expression group and low expression group according to the ADAM28 expression cutoff value based on the status of relapse. The cumulative incidence of relapse (CIR) and overall survival (OS) after a 3-year follow-up were used to evaluate the prognosis. The CIR after 3 years was significantly higher in the ADAM28 high expression group (p<0.0001); the higher CIR translated into a significantly worse OS (p=0.001). Moreover, when separately considering the impact of ADAM28 on prognosis within the risk stratifications, patients with high ADAM28 expression levels had a significantly higher CIR and worse OS in the favorable-risk group but not in the intermediate or poor-risk group. Because the patients with favorable risk were predominantly inclined to chemotherapy, the ADAM28 high expression patients presented a significantly higher CIR and worse OS than the low expression patients in the chemotherapy subgroup, whereas the prognosis did not differ significantly with the ADAM28 expression level in patients receiving HSCT.


ADAM28 improved the proliferation, migration and invasion of leukemic cells in which the IGF pathway was involved. High expression of ADAM28 enhanced the growth and dissemination of AML in vivo. ADAM28 expression levels also identified a new subgroup at higher risk for relapse and poor prognosis in favorable-risk AML patients, and this subgroup of patients which were allocated to chemotherapy, might benefit more from HSCT. Combining the expression level of ADAM28 with the existing risk stratification standard may be a more precise and preferable approach in predicting therisk of relapse in AML patients.


No relevant conflicts of interest to declare.

Author notes


Asterisk with author names denotes non-ASH members.