TP53 Mutations Negatively Impact Survival of Acute Myeloid Leukemia Patients Treated with Standard Doses of Azacitidine

Validated therapies for older pts with AML could rely on intensive or low-intensity strategies. Patient selection for these options remains controversial. There is currently no validated biomarker which can been used to guide therapeutic decision. TP53 mutations which are known to negatively impact AML pts outcome when treated with ICT, have been recently described as a positive prognosis factor for blast clearance with a 10-days regimen of decitabine (Welch, NEJM 2016). To date, it remains unclear whether AML pts with TP53 mutation represent a clinically homogeneous group. Several classification systems of p53 mutant, derived from in vitro or in vivo data, have been validated in solid tumors and aggressive lymphomas as predictors of p53 mutant functional impact or patient outcome. We retrospectively evaluated the impact of TP53 mutational status on the outcome of a real-world cohort of pts, treated frontline with standard doses of azacitidine (AZA). We further hypothesized that functional characterization of TP53 mutations could define a subgroup of pts with specific outcome with AZA

From Jan 2007 to Dec 2016, we identified 279 AML pts enrolled in the regional cancer network ONCOMIP registry, treated frontline with AZA. Median age was 76 yrs (45-93), karyotype was adverse in 135 pts (49.1%), including 54 pts with -17 or del17p (19.4%). AML was secondary to MDS in 71 pts (25.4%), to MPN in 24 (8.6%) and therapy related in 46 pts (16.5%). Pts received a median of 6 cycles (1-67). Overall, 54 pts obtained CR/CRi (19.4%) and median OS was 10.6 months (95%CI ,9.7-12.1). For 224 pts with an available bone marrow baseline DNA sample, TP53 mutations were screened with next-generation sequencing on an Illumina® MiSeq sequencer. Sequencing results were filtered with the IARC TP53 mutations database and a variant allele frequency (VAF) >10%, strengthening the specificity of the data of this cohort. Of the 224 analyzed cases, 55 cases (24.6%) contained TP53 mutations. Response rates did not significantly differ between TP53mut (21.8% CR/CRi) and TP53wt (17.8% CR/CRi, p=.50) nor between pts with TP53mut and/or -17/del17p (19.1% CR/CRi) and pts without TP53 abnormality (18.6%CR/CRi, p=.93). Median OS was 7.9 months in pts with TP53mut and 12.6 months in TP53wt (p<.0001). With regards to the group of 109 pts with adverse karyotype, response rates did not significantly differ between TP53mut pts (20.8% CR/CRi) and TP53wt (14.3%, p=.37) and median OS was 7.9 months for TP53mut pts versus 9.6 for TP53wt pts (p=.02)

Among the 55 pts with TP53mut, 53pts had adverse cytogenetics (96.4%), 16 pts had secondary AML to MDS or MPN (29.1%) and 13 had t-AML (23.6%). This subgroup included 49 cases (89%) with single TP53 mutation (missense n=42, nonsense n= 3, frameshift n=4) and 6 cases (11%) with 2 mutations (2 pts with missense and frameshift mutations and 4 pts with 2 missense mutation). We further characterized TP53mut pts with 3 validated classification systems. Due to dominant negative effect of TP53 mutation, for pts with >1 TP53 mutation, we selected the mutation with the predicted highest impact:

• 15 pts had disruptive mutations (i.e. missense mutation in L2/L3 helix of the DNA binding domain or truncating mutation) versus 40 pts with non-disruptive mutations (Poeta M, NEJM 2007), which was not associated with clinical response (25% in CR/CRi vs 27.9% in failure; p=1.00) nor with 6mOS (46.7% vs 55%, respectively; p=.79)

• Mutant p53 transactivation activity assessed with a 0-100 evolutionary score (Neskey D, Cancer Research 2015), was not associated with response (median score of 79.3[28-90] in CR/CRi vs 73.3 [49-96] in failure, p=1.00) nor with OS (HR 1.01; 95% CI, 0.99-1.03, p=.51).

• Relative fitness score (on a log2 scale) which was recently reported as a proxy of p53 mutant in vitro and in vivo cell proliferation advantage (Kotler E, Molecular cell 2018) was not associated with response (median score in CR/CRi of 0.094 [-0.79-0.58] vs 0.094 [-2.52-0.84] in failure, p=.68) nor with OS (HR 0.75; 95% CI, 0.45-1.22, p=.24)

Overall, the response rate was not influenced by the TP53mut status, but median OS was negatively impacted by the TP53mut status in the entire cohort and in the sub-group of pts with adverse karyotype. None of the mutant p53 classification systems validated in other neoplasms succeed in identifying a subset of AML pts who specifically benefit from AZA suggesting a rather homogenous functional impact of TP53 mutations in this setting