NEW YORK – A proteogenomic analysis of acute myeloid leukemia has unearthed five proteomics-based subtypes, including two prognostically informative subtypes.
"[O]ur proteogenomics approach led to the discovery of proteomic AML subtypes with clinical relevance and thereby provides a proteomic nosology as a basis for an improved molecular understanding and clinical classification of AML," senior and co-corresponding author Thomas Oellerich, a hematology/oncology researcher affiliated with University Hospital Frankfurt, the German Cancer Research Center, and the Goethe University Frankfurt Cancer Institute, wrote in Cancer Cell on Thursday.
The researchers used mass spectrometry, cytogenetic profiling, RNA sequencing, and/or targeted DNA panel sequencing to look at quantitative proteomic, cytogenetic, gene expression, and DNA mutation profiles in bone marrow biopsy samples from 177 consistently treated individuals with AML. With these data, they flagged five proteomics-based subtypes, which were set against clinical data for the patients to find subtypes corresponding with better or worse survival outcomes — results validated with data for another 75 AML patients.
The paper's integrated analyses "revealed the inter-relationship between the AML proteome and genetic aberrations," the authors noted, "providing insight into the pathogenesis of the disease and suggesting testable therapeutic interventions."
The team noted that AML cases involving tumors from a heterogeneous proteomics-based cluster called C5 tended to have longer-than-usual survival times following intensive induction chemotherapy treatment compared to cases with non-C5 tumor subtypes, for example.
Although the C5 tumors did not share a single protein expression signature, the investigators' subsequent bioinformatic analyses highlighted three tumor sub-clusters within the C5 group. Two of those sub-clusters appeared to be prognostically favorable — including a sub-cluster with dialed down NOTCH4 signaling — while a third sub-cluster was classified as prognostically neutral.
On the other hand, they found that a "Mito-AML," or "C-Mito," subtype — marked by enhanced mitochondrial protein expression and reliance on mitochondrial complex I-dependent respiration — coincided with poorer-than-usual induction chemotherapy treatment responses and overall survival times.
"RNA processing and splicing was one of the dominant proteomic themes that occurred in specific proteomic subgroups similarly to what has been described in the genomic context," the authors explained. "In contrast, the mitochondria-related pathways, the key characteristic of C-Mito, have not been described in previous genomic studies, and we found this molecular feature to be reflected exclusively in the proteome."
The finding was particularly intriguing because the Mito-AML subtype appeared apt to respond to drugs targeting the mitochondrial complex I, the team explained. In preliminary cell line and patient blast experiments they found that Mito-AML cells had enhanced sensitivity to mitochondrial complex I-inhibiting drugs such as mubritinib or rotenone and/or the BCL2 inhibitor venetoclax, which affects mitochondrial respiration.
"Our functional analyses in AML cell lines and primary AML blasts revealed a specific metabolic wiring of Mito-AML," the authors noted, arguing that "it would be very interesting to validate, in future prospective trials, the clinical utility of proteomic signatures defining Mito-AML as predictive biomarkers for venetoclax-based therapies."