Therapy‐Induced Senescence Is a Predictor of Treatment Outcome in Acute Myeloid Leukemia

Introduction: Acute myeloid leukemia (AML) is a heterogeneous disease, for which biologically grounded predictors of outcome remain an urgent clinical need. In addition to tumor cell death, antineoplastic drugs have been shown to mediate a long-lasting growth arrest of vital, metabolically active tumor cells termed therapy-induced senescence (TIS) in various cancer entities and model systems, but predictive investigations on TIS are lacking. We hypothesized here that an ex vivo assessment of TIS in blast samples from AML patients at diagnosis might serve as a biomarker of outcome.

Methods: We established assays to quantitatively detect TIS in a primary culture setting of purified primary human AML specimens (n > 30) assessed through cytochemical and fluorescence-based analysis of senescence-associated β-galactosidase (SA-β-gal) activity and other senescence markers (such as p16^INK4a, H3K9me3) in concert with downregulation of proliferation markers (Ki67, EdU). AML samples were subjected to standard karyotype and molecular genetics analyses, defining risk groups according to the European Leukemia Net (ELN) genetic risk stratification.

Results: We found the individual fractions of senescent AML blasts in response to the standard anti-leukemic agents Daunorubicin and Cytarabine samples to be positively correlated with better disease-free (p= 0.042) and overall survival (p= 0.004) of the patients. Furthermore, a more favorable molecular risk group (p= 0.005), normal karyotype (p= 0.007), and NPM1 mutations (p= 0.018) as well as DNMT3A mutations (p < 0.001) were associated with higher TIS levels in our ex vivo senescence assay.

Discussion: Our study is the first report providing direct evidence that cellular senescence induced in patient-derived AML blasts by chemotherapeutic drugs ex vivo serves as a predictor of patient long-term responsiveness to standard induction therapy. TIS might explain the underlying biology of current para-clinical risk indicators, and also seems to identify subgroups of superior outcome in pre-defined risk groups including patients with DNMT3A-mutant disease. Additional investigations exploring senescence in response to other AML-relevant agents and functional analyses expanding to patient-derived xenograft (PDX) AML models will be presented at the meeting.