Background: Acute myeloid leukemia (AML) demonstrates cell differentiation stages emphasizing the role for hematopoietic maturation arrest during leukemogenesis. Indeed, less differentiated AML stages retain enhanced sensitivity to hypomethylating agents (HMA) plus BCL-2 inhibitor, a phenomenon less observed in monocytic skewed cluster, for which menin inhibitors represent promising intervention. Deconvolution of bulk AML transcriptome allows for better understanding of disease pathogenesis. However, the role of “Immunotranscriptome analysis [i.e. CD34, CD117, HLADR, CD33, CD38, CD11b, BCL2 Gene Expression (GE)]” in identifying maturation arrest remains uncharacterized. In this study, we seek to harmonize marrow immunohistochemistry (IHC) and bulk AML immunotranscriptomic to better discriminate cluster designation. Methods: After IRB approval, 49 AML cases were included for analysis. 6 clusters were designed based on marrow IHC [i.e., Hematopoietic Stem Cell (HSC)-like, Multipotent Progenitor (MPP)-like, Common Myeloid Progenitor (CMP)-like, Granulocytic Monocytic (GMP)-like, Monocyte Progenitor (MP)-like, Granulocyte Progenitor (GP)-like]. Differential immunotranscriptomic for individual GE [i.e. CD34, CD33, CD117 etc.] was examined in our “founder” IHC cohort. Next, discrimination analysis [DA] to detect cluster “membership” was obtained by integrating individual Immunotranscriptome expression into categorical IHC clustering. Descriptive statistics, ANOVA and discrimination procedure were performed with SAS software. Results: Mean age was 65 years (range, 25-92). 22(51%) of cases were male. 19(44%), 22(51%), 2(5%), and 1(2%) were white, African Ancestry (AA), Hispanic and Asians. ELN22 risk was available in 5/49(10.2%), 11/49 (22.4%), 33/49 (67.3%) of favorable, intermediate and adverse groups. 7/49(14.2%), 3/49(6.1%), 10/49(20.4%), 7/49(14.2%), 16/49(33%), and 6/49(12.2%) were HSC, GMP, MP, GP, CMP and MPP-like. CD34; p=0.003, HLA-DR; p=0.01, CD11b; p=0.03 were differentially expressed among clusters. “Stringent” MDS mutations (mut) [i.e. splicing, ASXL1, cohesin, BCOR etc.] were observed in higher proportion of CMP-like cluster (60% vs 21.8%, p=0.01). 7/13 (88%) vs 1/13(12%) of cases with and without MP-like cluster definition harbored KM2TA abnormality, p=<0.0001. By adding GE to our “founder” IHC clustering, 6/6 (100%, p=<0.0001) of pt were allocated in HSC-like disease, 3/3 (100%, p=0.0001) to GMP-like, 6/9 (66.3%. p=0.33) to MP-like, 6/6 (100%, p=<0.0001) GP-like, 2/13 (15.3%) CMP-like and 3/3 (100%, p=0.0001) MPP-like. Conclusions: Our analysis demonstrates that leukemia clusters can be identified based on IHC. GE facilitates robust cluster “membership” for all groups except MP-like and CMP-like. While KM2TA and MDS like mutations were observed significantly higher in MP-like and CMP-like diseases, it is possible that complex epigenetic rewiring [i.e. GE changes induced by KM2TA and MDS mutations] alter Immunotranscriptome ability to enhance cluster designation. Validation of our founder cohort findings is needed.
Read full abstract