The FLT3-like Gene Expression Signature Predicts Response to Quizartinib in Wild-Type FLT3 Acute Myeloid Leukemia: An Analysis of the Pethema Quiwi Trial

Background The identification of predictive biomarkers is crucial for guiding treatment decisions in acute myeloid leukemia (AML). Previously, we identified a FLT3-like gene expression signature in FLT3 wild-type AML patients, which clustered a variable proportion of FLT3 wild-type patients with FLT3-ITD and TKD mutated patients. The QUIWI trial was a randomized, placebo-controlled, phase II study preliminary showing a significant increase in overall survival (OS) in wild-type FLT3 AML patients treated with the FLT3 inhibitor quizartinib (Quiza) plus standard chemotherapy. This preplanned correlative study was designed to assess the value of the FLT3-like signature to predict responses to Quiza. Methods We performed RNA sequencing (RNAseq) analysis on a subset of patients from the clinical trial. RNA was extracted using standard methods, followed by assessment of nucleic acid integrity (TapeStation) and quantification (Qubit). Total mRNA sequencing was performed using polyA RNAseq with TruSeq technology. A total of 206 adequate samples from bone marrow and peripheral blood were sequenced (161 from FLT3-ITD negative enrolled in QUIWI; and 55 from FLT3-ITD positive patients who were screen failure of the QUIWI trial by this reason). The sequences were aligned to the GRCh37 reference genome using the Hisat algorithm. Gene expression quantification was performed using the Bioconductor workflow, and gene expression estimates (FPKM) were obtained. The gene expression estimates were log2 normalized. Finally, those genes mapping to the original FLT3-like signature (595 genes) were selected for downstream analysis. Clustering was based on the hierarchical method, using standard euclidean distance metrics. OS was defined as time from start of screening to death. Event-free survival (EFS) was defined as time from randomization to failure to achieve CR/CRi after 1 or 2 cycles, death in CR/CRi, or relapse (whichever occurred the first). Relapse-free survival was defined as time from randomization to disease relapse or death by any cause. Results Among the total 206 patients, a cluster of 54.37% (N=112) was enriched in FLT3-mutant cases (71.11% of cases, Figure 1A). This subgroup comprised 49.67% of wild-type FLT3 cases (N=80), hereafter FLT3-like patients. In the group of not FLT3-like patients, no differences were identified between the placebo and Quiza group in the total number of deaths (Fisher's p-value, 0.63), EFS (cox p-value, 0.83; HR 1.07 [0.56-2.06]), RFS (cox p-value 0.76; HR 0.88 [0.38-2.01]) and OS (cox p-value, 0.62; HR 1.22 [0.55-2.67]). Among FLT3-like patients, significant differences were identified in the total number of deaths (Fisher's p-value, 0.004), EFS (cox p-value, 0.009; HR 0.45 [0.25-0.82], RFS (cox p-value 0.01, HR 0.37 [0.18-0.79]) and OS (cox p-value 0.01, HR, 0.41 [0.20-0.84]) ( Figure 1B). No statistically significant association was observed between the FLT3-like pattern and the ELN-17 classification: 30.4% were low risk, 40.5% intermediate risk, 29.1% high risk. Conclusion The FLT3-like gene expression signature successfully identified a subset of patients who derived the most benefit from Quiz, while patients without the FLT3-like signature did not demonstrate a benefit compared with placebo. These findings support the use of the FLT3-like signature as a potential biomarker to identify those wild-type FLT3 AML patients who may benefit from Quiz, providing a valuable insight for personalized treatment in AML.