4. Discussion
So far survival analysis has not been performed to evaluate the impact of EVI1 mRNA expression in ALL. In the current study, basing on the EVI1 transcript detected by RQ-PCR in our cohort and further validating by RNA-seq data from TARGET database, we demonstrated that low EVI1 expression at diagnosis predicted poor outcomes in pediatric Ph-negative BCP-ALL patients receiving chemotherapy only.
EVI1 gene encodes an oncogenic transcription factor and aberrant EVI1 expression has been found in various human myeloid malignancies, including AML, MDS, and CML.6 High EVI1 expression predicted poor outcomes in adult and pediatric AML patients as well as ICR-AML patients.9-13 EVI1 could induce MDS in the mouse model, and it was a poor prognostic marker for MDS patients.24,25 In CML patients, high EVI1 expression was related to blast crisis.26 Compared to AML, the prognostic role of EVI1 in ALL remains obscure.
ALL is the most common cancer and the most frequent cause of death from cancer among children. It is a heterogeneous disease with various genetic abnormalities.27 Although the outcomes of pediatric BCP-ALL have improved greatly in the past decades, relapse remains occur in the minority leading to death.28Therefore, searching for new prognostic markers is still necessary in order to guide appropriate treatment selection and improve the overall outcomes. Therefore, the prognostic impact of EVI1 expression in pediatric BCP-ALL needs to be evaluated.
EVI1 expression patterns were evaluated firstly. Several studies have showed EVI1 transcript levels in ALL which tested by PCR or microarray.14,15,18 T-ALL patients and cell lines were found to have no detectable EVI1 transcript by RT-PCR.16 Whereas, Su G et al performed RQ-PCR and found that T-ALL had significantly higher EVI1 levels than BCP-ALL. Furthermore, they showed that the overall EVI1 transcript levels were significantly higher in ALL than AML.15 We also found the similar result in pediatric Ph-negative BCP-ALL patients. That is, they generally had significantly higher levels than adult ICR-AML patients which we reported previously,13 and great variation existed among patients.
The cutoff value for patient grouping usually affected the final comparison result greatly. As for the cutoff value of EVI1 transcript, both the upper limit and the mean±standard deviation values of NBM were ever used in previous studies.13,15 Furthermore, some studies arbitrarily determined.9,14 In the current study, quartile grouping implied that low EVI1 expression was related to low RFS rate, and ROC curve was used to determine the optimal cutoff value.
The survival analysis in our cohort showed that low EVI1 expression independently predicted low RFS and OS rate. The TARGET data for validation showed the similarly negative impact of low EVI1 expression on OS rate though the P value is no less than 0.05. However, a previous in vitro and mice transplantation study demonstrated that EVI1 contributes to the leukemogenic potential and apoptosis resistance of ALL cells, which implied the poor prognostic role of EVI1.14 One reason for this contradictory result is that protein function is usually cell context-specific. Therefore, the in vitro result might be different from that in vivo and cell line could not represent primary leukemic cells completely. Another reason might be the EVI1 level. Just as illustrated by our quartile grouping result, the 2nd to 4th quartile group had similar RFS. Therefore, so-called low or high EVI1 expression must be clearly defined and the in vitro study should correspond to clinical meaningful value in order to clarify the function of EVI1 in ALL.
The percentage of patients belonging to low EVI1 expression were different in our cohort and TARGET cohort for validation, 18.2% and 72.9%, respectively. It might be caused by different case composition. The TARGET ALL Pilot project included comprehensive genomic profiles of nearly 200 high-risk, clinically annotated, BCP-ALL patient cases from Children’s Oncology Group (COG), most of which experienced an early bone marrow relapse (within 4 years of initial diagnosis).
A significant correlation between high EVI1 expression and MLL rearrangement has been demonstrated in both adult and pediatric AML patients.9-12,29 However, we found an association between low EVI1 expression and MLL rearrangement in pediatric BCP-ALL patients. Furthermore, TARGET data and previous reports did not show this relationship in BCP-ALL patients.15,18 These inconsistent results reflected that the association between MLL rearrangement and EVI1 expression were related to leukemia type. Difference in the partner of MLL fusion and all enrolled patients received chemotherapy only in our cohort might also be the reasons. In addition, both our cohort and TARGET data showed that patients with TEL-AML1 fusion gene had high EVI1 expression, and its mechanism remains to be studied.