FLT3-ITD Allelic Burden and Acute Promyelocytic Leukemia Risk Stratification

Simple Summary Around 12–38% of acute promyelocytic leukemia (APL) patients carry the FLT3-ITD mutation, which has been associated with several poor-prognosis indicators such as high white blood cell counts, M3v variant morphology, and the bcr3 isoform. We aimed to retrospectively study the impact of FLT3-ITD mutations in APL patients in regard to clinical features, treatment courses, and outcomes. We demonstrate that Sanz high-risk status APL correlates with high FLT3-ITD allelic burdens, with every 1% increase in allelic burden correlating with a 0.6 × 109/L increase in white blood cell count (WBC). The presence of FLT3-ITD was associated with decreased remission rates and higher 5-year mortality from the time of diagnosis. These findings provide novel revelations regarding the features of FLT3-ITD APL, particularly in regard to allelic burden, that warrant further study. Abstract The significance of FLT3-ITD in acute promyelocytic leukemia (APL) is not well-established. We performed a bi-center retrospective study of 138 APL patients, 59 (42.8%) of whom had FLT3-ITD. APL patients with FLT3-ITD had higher baseline white blood cell counts (WBCs) (p < 0.001), higher hemoglobin, (p = 0.03), higher aspartate aminotransferase (p = 0.001), lower platelets (p = 0.004), lower fibrinogen (p = 0.003), and higher incidences of disseminated intravascular coagulation (p = 0.005), M3v variant morphology (p < 0.001), and the bcr3 isoform (p < 0.001). FLT3-ITD was associated with inferior post-consolidation complete remission (CR) (p = 0.02) and 5-year overall survival (OS) of 79.7%, compared to 94.4% for FLT3-WT (wild-type) (p = 0.02). FLT3-ITD was strongly associated with baseline WBCs ≥ 25 × 109/L (odds ratio (OR): 54.4; 95% CI: 10.4–286.1; p < 0.001). High FLT3-ITD allelic burdens correlated with high-risk (HR) Sanz scores and high WBCs, with every 1% increase in allelic burden corresponding to a 0.6 × 109/L increase in WBC. HR APL was associated with a 38.5% increase in allelic burden compared with low-risk (LR) APL (95% CI: 19.8–57.2; p < 0.001). Our results provide additional evidence that FLT3-ITD APL is a distinct subtype of APL that warrants further study to delineate potential differences in therapeutic approach.


Introduction
First described by Norwegian hematologist Leif Hillestad in 1957 through a series of three cases, acute promyelocytic leukemia (APL) was aptly named for its predominance of promyelocytes and deadly coagulopathy characterized by "a very rapid fatal course of only a few weeks' duration" [1]. Today with the advent of all-trans retinoic acid (ATRA) and arsenic trioxide (ATO), APL is highly curable if treatment is initiated promptly. APL is classified as the French-American-British (FAB) subtype M3 of acute myeloid leukemia (AML), classically arising from a balanced reciprocal translocation between chromosomes 15 and 17. The fusion between the promyelocytic leukemia (PML) and retinoic acid receptor alpha (RARα) genes results in the PML-RARα rearrangement t(15;17)(q24;q21), leading to the disruption of the RARα-regulated maturation of myeloid progenitors at the promyelocytic stage [2]. APL comprises less than 10% of AML, with an estimated 0.01/100,000 incidence in Western countries, affecting men and women equally [3].
FMS-like tyrosine kinase 3 (FLT3) is a proto-oncogene implicated in leukemogenesis. The FLT3 ligand binds to the extracellular FLT3 receptor, inducing homodimerization that potentiates a downstream signaling cascade involved in the regulation of the proliferation, differentiation, and apoptosis of early myeloid and lymphoid progenitor cells. The most common FLT3 aberration is a 3-to->400-base-pair in-frame internal tandem duplication (ITD) mutation within exon 14 that leads to the constitutive activation of the FLT3 receptor and ligand-independent autophosphorylation [4][5][6][7][8][9]. FLT3-ITD mutations are found in 20-30% of young adults with AML and are a poor prognostic indicator [9]. Targeted FLT3 inhibitors have been developed such as giltertinib, which was associated with significantly longer overall survival (OS) and higher complete remission (CR) rates compared to those with salvage chemotherapy in relapsed or refractory FLT3-ITD AML [10].

Materials and Methods
Patients 18 years or older treated at the University of Maryland Medical Center and Johns Hopkins Hospital from January 2000 to May 2020 were included. APL was defined by the cytogenetic and/or molecular confirmation of PML-RARα. FLT3-ITD mutations were targeted at the juxtamembrane region of the FLT3 gene (exons 14-15) via primers, amplified, and identified by fluorescent PCR. The FLT3-ITD allelic burden was estimated as a percent ratio of the area under the curve (AUC) of the variant peak divided by the AUC of the wild-type (WT) peak. The ITD insertion length was determined by subtracting the 328-base-pair PCR product of the FLT3 gene from the base pair size of the variant peak. De-identified patient ages, genders, ethnicities, body mass indices (BMIs), laboratory measurements, Sanz and FLT3-ITD statuses, ITD allelic burdens and insertion lengths, induction chemotherapy regimens, and outcomes were inputted into Microsoft Excel. These data were then transferred to Stata, Version 16.1 (StataCorp, College Station, TX, USA). All the statistical analyses were performed and graphs were made using Stata.
The Chi-square test of independence was run for two categorical variables (e.g., Sanz and FLT3-ITD status). Mean differences between two groups (e.g., FLT3-ITD status) were tested using independent-sample t-tests, and mean differences between three or more groups (e.g., Sanz) were tested using one-way analysis of variance (ANOVA), followed by Scheffe tests. Pearson's and Spearman's correlations were used to study the associations of two continuous variables (e.g., allelic burden and WBC). For multiple regression analyses, logistic regression models were used for the associations of independent binary outcomes (e.g., FLT3-ITD status) with independent variables of interest (e.g., Sanz), before and after adjustment for other covariates (e.g., age, gender, and ethnicity). Similarly, linear regression models were used for the association of continuous outcomes (e.g., allelic burden) with independent exposures of interest (e.g., Sanz) before and after adjustment for other covariates. Overall survival (OS) was compared between all groups (FLT3-ITD, FLT3-WT, and Sanz HR/IR/LR) using the log-rank test for the equality of survivor functions and graphed using the Kaplan-Meier method. Cox regression models were used to compare survival between groups after adjustment for covariates of interest.
The Sanz risk status significantly correlated with allelic burden according to one-way ANOVA (F(2, 30) = 12.1, p < 0.0001). According to post hoc Scheffe tests, the allelic burden differed between HR/IR (p = 0.007) and HR/LR (p = 0.001), but not in IR/LR (p = 0.32). These findings persisted when adjusting for center, age, gender, ethnicity, WBC, and platelets via linear regression. HR APL was associated with a 38.5% increased allelic burden compared with LR APL (95% CI: 19.8-57.2; p < 0.001). According to Spearman's Rho, the FLT3-ITD allelic burden was associated with higher WBCs (rs = 0.49, p = 0.03). For every 1% increase in allelic burden, the WBC increased by 0.6 × 10 9 /L. The relationship between the FLT3-ITD allelic burden and WBC can be visualized in Figure 1. Of note, the majority of the reported allelic ratios (33/59 patients) were from those seen after 2012. There was no significant relationship between the FLT3-ITD allelic burden and OS (p = 0.97).  While studies have reported a higher incidence of induction death and inferior CR rates, OS, and RFS in FLT3-ITD APL [6,[9][10][11]15,16], we report no significant differences in the CR duration after induction (p = 0.70) or death during induction (p = 0.13). We found shorter post-consolidation CR durations (p = 0.02) and OS (p = 0.02) in FLT3-ITD APL (Figure 2). The 5-year OS for patients with FLT3-ITD was 79.7% compared to 94.4% for FLT3 wild-type patients. FLT3-ITD was associated with a higher mortality risk, with a hazard ratioof 3.25 (95% CI: 1.14-9.25; p = 0.027). Other studies reported no association between FLT3-ITD and CR/RFS/OS/early death in APL [7,8,11,[17][18][19][20]. The variability of the outcomes may be attributed to the small cohort size, differences in inclusion criteria, varying treatment protocols, and adherence to follow-up.
Pre-arsenic trioxide (ATO) studies showed reduced OS and increased relapse rates with FLT3-ITD [9,15,21] in APL, whereas post-ATO studies demonstrated no prognostic significance of FLT3-ITD [17][18][19][20], suggesting the effect of FLT3-ITD may be mitigated with ATO. Our cohort consisted of 26 FLT3-ITD patients treated with ATO, compared to 32 without ATO, and 41 FLT3-WT patients treated with ATO, compared to 37 without (two patients with unclear induction regimens). When mutually adjusting for FLT3-ITD status, the ATO-containing regimens were not associated with improved OS, with a hazard ratio (95% CI) of 1.13 (0.41-3.16). When stratified by the ATO versus non-ATO regimens, FLT3-ITD was associated with increased mortality in both groups. The hazard ratio (95% CI) was 4.78 (1.01-22.6) for the non-ATO regimens and 3.17 (0.58-17.3) for the ATO-containing regimens (p-value for interaction = 0.46), suggesting no evidence for the impact of ATOcontaining regimens on the prognostic significance of FLT3-ITD in this study. However, the study sample size was modest, making the power relatively low for detecting effect modification. The number of patients treated with gemtuzumab ozogamicin was too small (n = 4) to perform statistical analysis on outcomes.
Biology 2021, 10, x 6 of 9 While studies have reported a higher incidence of induction death and inferior CR rates, OS, and RFS in FLT3-ITD APL [6,[9][10][11][15][16], we report no significant differences in the CR duration after induction (p = 0.70) or death during induction (p = 0.13). We found shorter post-consolidation CR durations (p = 0.02) and OS (p = 0.02) in FLT3-ITD APL (Figure 2). The 5-year OS for patients with FLT3-ITD was 79.7% compared to 94.4% for FLT3 wild-type patients. FLT3-ITD was associated with a higher mortality risk, with a hazard ratioof 3.25 (95% CI: 1.14-9.25; p = 0.027). Other studies reported no association between FLT3-ITD and CR/RFS/OS/early death in APL [7][8]11,[17][18][19][20]. The variability of the outcomes may be attributed to the small cohort size, differences in inclusion criteria, varying treatment protocols, and adherence to follow-up. Pre-arsenic trioxide (ATO) studies showed reduced OS and increased relapse rates with FLT3-ITD [9,15,21] in APL, whereas post-ATO studies demonstrated no prognostic significance of FLT3-ITD [17][18][19][20], suggesting the effect of FLT3-ITD may be mitigated with ATO. Our cohort consisted of 26 FLT3-ITD patients treated with ATO, compared to 32 without ATO, and 41 FLT3-WT patients treated with ATO, compared to 37 without (two patients with unclear induction regimens). When mutually adjusting for FLT3-ITD status, the ATO-containing regimens were not associated with improved OS, with a hazard ratio (95% CI) of 1.13 (0.41-3.16). When stratified by the ATO versus non-ATO regimens, FLT3-ITD was associated with increased mortality in both groups. The hazard ratio (95% CI) was 4.78 (1.01-22.6) for the non-ATO regimens and 3.17 (0.58-17.3) for the ATO-containing regimens (p-value for interaction = 0.46), suggesting no evidence for the impact of ATO-containing regimens on the prognostic significance of FLT3-ITD in this study. However, the study sample size was modest, making the power relatively low for detecting effect modification. The number of patients treated with gemtuzumab ozogamicin was too small (n = 4) to perform statistical analysis on outcomes.

Discussion
APL therapy differs in Sanz HR versus IR/LR groups, and risk-stratifying APL is important for optimizing outcomes. Genes impacting differentiation are downregulated and genes involved in cellular adhesion, invasiveness, and metastasis are upregulated in FLT3-ITD APL [16]. Given the associations of FLT3-ITD with poor prognostic features and outcomes, FLT3-ITD APL has been proposed to be a distinct subtype of APL [7,9]. The high WBCs, M3v, and bcr3 reported in this study have been well-established in FLT3-ITD APL, whereas the shorter post-consolidation CR duration and OS found here are less well-established.
We found a strong correlation between FLT3-ITD and leukocytosis in APL, suggesting a WBC cutoff of ≥25 × 10 9 /L as an indicator for considering testing for FLT3-ITD in APL. The well-known association between leukocytosis and FLT3-ITD in APL is further expanded upon in this study by the novel revelation of the WBC's relationship with the ITD allelic burden, with every 1% increase in allelic burden equating to a 0.6 × 10 9 /L increase in WBC. Interestingly, the allelic burden was not found to be associated with OS.

Conclusions
Taken together, these results support the importance of additional study of the significance of the FLT3-ITD mutation and ITD allelic burden in APL. More data are required to determine the utility of incorporating FLT3-ITD into risk-adapted treatment algorithms and molecular monitoring. The absence of routine testing for FLT3-ITD in APL, the lack of international standardized FLT3-ITD assays, and the rarity of the disease pose limitations for studies of FLT3-ITD APL [6,13].