FLT3 and NPM1 mRNA Expression-Based Risk Stratification of De Novo Acute Myeloid Leukemia

Introduction:

Prognostication of acute myeloid leukemia (AML) at initial diagnosis relies on the identification of underlying genetic abnormalities as such proposed by European LeukemiaNet (ELN). However, the clinical trajectory of AML is highly variable at least partially due to molecular heterogeneity even within each ELN risk category, making standard guidelines often not straightforwardly applicable. In search of novel prognostication strategies, we explored two publicly available bone marrow RNA-Seq datasets and found significantly inferior overall survival (OS) associated with high-FLT3 and low-NPM1 transcript levels (“FLT3^high/NPM1^low”) compared to low-FLT3 and high-NPM1 transcript levels (“FLT3^low/NPM1^high”) in adult de novo AML. We speculate that the FLT3 and NPM1 gene expression levels may be a novel prognostic biomarker independent of cytogenetic and mutational profiles.

Methods:

Normalized RNA-Seq transcript counts of adult (age ≥ 18 years) de novo AML from The Cancer Genome Atlas AML project (TCGA-LAML; n=157) and the Oregon Health Science University Beat AML project (OHSU; n=230) were analyzed. Transcript level of a gene “X” was classified as “high” and labeled as “X^high” if higher than or equal to the cohort median RPKM (reads per kilobase per million mapped reads), and “low” labeled as “X^low” if lower than the cohort median RPKM. Log-rank p-value and Mantel-Haenszel hazard ratio (HR) are reported with Kaplan-Meier survival analysis unless specified otherwise. Acute promyelocytic leukemia (APL) cases were excluded in our analysis. Patients with FLT3 or NPM1 mutant genotypes were also excluded to eliminate their potential confounding effects to their transcript levels.

Results:

In TCGA cohort, median OS (mOS) was 7 months in the FLT3^high/NPM1^low group vs. 28.1 months in the FLT3^low/NPM1^high group with a hazard ratio (HR) for death of 7.55 (95% CI 2.56-22.3, p=0.0003). Consistently, in OHSU cohort, mOS was 8.3 months in the FLT3^high/NPM1^low group vs. 48.4 months in the FLT3^low/NPM1^high group with an HR for death of 4.20 (95% CI 1.73-10.2, p=0.0015). In both cohorts, mOS of FLT3^high/NPM1^high and FLT3^low/NPM1^low groups were superior to FLT3^high/NPM1^low group, but inferior to FLT3^low/NPM1^high group. The differential OS associated with differential expression of FLT3 and NPM1 was observed even within each cytogenetics-based risk class. For example, patients in the ELN2017 “intermediate” risk class with FLT3^high/NPM1^low exhibited HR for death of 6.99 (95% CI 0.90-54.3, p=0.06), in the OHSU cohort. Differential OS between FLT3^high/NPM1^low and FLT3^low/NPM1^high groups was not associated with TP53 mutation. No significant difference in OS was noted in TP53^high group compared to that of TP53^low group, HR for death of 0.73 (95% CI 0.47-1.13, p=0.16) in TCGA and HR for death of 0.88 (95% CI 0.54-1.43, p=0.61) in the OHSU cohorts. Cox proportional hazard regression analyses revealed male vs. female gender as an additional independent prognostic factor (HR for death 2.809; 95% CI 1.380-6.337, p=0.007), although this was seen only in the OHSU cohort.

Conclusion:

FLT3 and NPM1 transcript level-dependent differential OS even within the established cytogenetics-based risk classes (e.g., ELN) suggest that FLT3 and NPM1 transcriptional profile may represent a global oncogenic state regardless of the underlying genotypic or cytogenetic characteristics. A variety of approaches have been proposed to differentiate RNA-Seq transcript levels into “high” vs. “low”, although there is no consensus to date. In this study, a sample median-based dichotomization approach was used, and our findings were reproducible across two independent datasets. Our findings suggest that the bone marrow FLT3 and NPM1 transcript level may provide supplementary guidance to de novo AML risk stratification strategies when combined with the traditional ELN prognostication guidelines. As an example, patients at ELN “intermediate” risk with FLT3^low/NPM1^high who are clinically suboptimal for consolidative allogeneic stem cell transplantation may defer this option without significantly compromising their survival outcome. Prospective validation study would be warranted.

No relevant conflicts of interest to declare.