Response: CEBPA promoter hypermethylation in a subset of myeloid/T-lymphoid leukemias with a distinct gene expression profile

Response

With great interest we have read the letter by Terriou and coworkers, who have performed an extensive survey of CEBPA CpG promoter hypermethylation in T-ALL as well as in a selection of immature acute myeloid leukemia (AML) cases. This study relates to our previous work in which we identified hypermethylation of the proximal CEBPA promoter in a small subset of AMLs with low CEBPA mRNA.¹  It should be noted that the patient group we described was not primarily defined by CEBPA promoter hypermethylation. Instead, those cases were studied because they exhibited a unique gene expression profile. In addition to this most discriminating feature, the leukemic blasts coexpressed CD34 and T-cell antigens with myeloid markers, frequently carried NOTCH1 mutations, and expressed the transforming gene TRIB2.^1,2  Using a predictive gene expression signature in an independent cohort of AML, we identified genetically and immunophenotypically similar cases with silenced CEBPA.

In the present study, Terriou et al found CEBPA promoter hypermethylation in 4/54 AML cases. These 4 cases all revealed one or more T-cell characteristics (cCD3, CD7, and/or TCRG rearrangement) and may therefore resemble the leukemias we described. In T-ALL, the investigators found CEBPA promoter hypermethylation more frequently, in 37/99 cases. Importantly, however, only a minority of those T-ALLs with methylated CEBPA coexpressed myeloid surface markers (CD13 and/or CD33) and CD34. The findings of Terriou et al therefore indicate that CEBPA proximal promoter hypermethylation in combination with an immature myeloid/T-lymphoid immunophenotype is generally rare in acute leukemia, and can be found in a small percentage of AML as well as in a small fraction of T-ALL. It will be important to apply gene expression profiling to these cases to fully assess their relationship with the leukemias that we described previously.

Together, these observations highlight the challenges in classifying particular acute leukemias as AML, ALL, or a unique entity. We believe that studies as the one performed by Terriou et al will be instrumental in obtaining a better understanding of this topic. Gene expression profiling applied to these AML and T-ALL cohorts will clearly be of added value. Combining data from multiple research groups will be a requirement to address important questions regarding prognosis, and should help to evaluate according to which protocols these relatively small patient groups are best treated.