Circ-Foxo3 is positively associated with the Foxo3 gene and leads to better prognosis of acute myeloid leukemia patients

The bone marrow (BM) samples of experimental group were collected from patients in the Affiliated Hospital of Jiangsu University who were initially diagnosed with acute myeloid leukemia, while the control group were from healthy BM donors or patients with chest trauma. We tested Foxo3 gene in 122 de novo AML patients and 30 control samples. Because some of these samples had run out and new samples were collected, the circ-Foxo3 gene was tested in 116 de novo AML patients and 24 control samples. These patients were accepted standard treatment after diagnosis of AML. The AML patients were classified by FAB classification and the 2008 WHO criteria. The therapeutic regimen and results of laboratory and equipment inspection were recorded in corresponding physician’s order sheet and medical record. All the BM donators had signed the informed consents. This study was approved by the Review Committee of the Ethics Department of the People’s Hospital of Jiangsu University.

The patients, enrolled in this study were all received chemotherapy. The chemotherapy regimens for these patients were selected according to the NCCN (2016) guideline of AML. Patients with acute promylocytic leukemia (APL) can be divided into high risk (WBC count > WBC ≥ 10 × 10⁹/L) and low risk (WBC count≤10 × 10⁹/L) groups. For high risk patients, induction therapy was consisted of oral ATRA 45 mg/m² in divided doses until clinical remission, intravenous daunorubicin 50 mg/m² × 4 days, and cytarabine 200 mg/m² × 7 days. After count recovery, patients received 3 monthly consolidation courses: arsenic trioxide 0.15 mg/kg·day × 5 days per week, for 5 weeks and for 2 cycles, then ATRA 45 mg/m² × 7 days, with daunorubicin 50 mg/m² × 3 days for 2 cycles. For low risk patients, the induction therapy was consisted of oral ATRA 45 mg/m² in divided doses until clinical remission daily and intravenous arsenic trioxide 0.15 mg/kg daily until bone marrow remission. At count recovery, proceed with consolidation courses: intravenous Arsenic trioxide 0.15 mg/kg·day × 5 days per week for 4 weeks every 8 weeks for a total of 4 cycles, and ATRA 45 mg/m²·day for 2 weeks every 4 weeks for a total of 7 cycles [18].

For non-APL patients, who were below 60 years old, induction therapy for them consisted of 4 treatment options: a. Standard-dose cytarabine 100–200 mg/m² continuous infusion × 7 days with idarubicin 12 mg/m² or daunorubicin 60–90 mg/m² × 3 days (for patient ≤45 y). b. Standard-dose cytarabine 200 mg/m² continuous infusion × 7 days with daunorubicin 60 mg/m² × 3 days and cladribine 5 mg/m² × 5 days (for other age groups). c. High-dose cytarabine (HiDAC) 2 g/m² every 12 h × 6 days or 3 g/m² every 12 h × 4 days with idarubicin 12 mg/m² or daunorubicin 60 mg/m² × 3 days. d. Fludarabine 30 mg/m2 IV days 2–6, cytarabine 2 g/m² over 4 h starting 4 h after fludarabine on days 2–6, idarubicin 8 mg/m² IV days 4–6, and G-CSF SC daily days 1–7 [18].

For patients over 60 years old, there are options: a. Standard-dose cytarabine (100–200 mg/m² continuous infusion × 7 days) with idarubicin 12 mg/m² or daunorubicin 60–90 mg/m² × 3 days or mitoxantrone 12 mg/m² × 3 days. b. Lower intensity therapy: low-dose cytarabine or 5-azacytidine, decitabine [18].

In our laboratory, human hematological cell lines (K562, U937, NB4, SHI-1, and HEL) were purchased from American Type Culture Collection Manassas, VA, USA. The cell line information can be queried from two databases: American Type Culture Collection (ATCC) and Deutsche Sammlung von Mikroorganismen und Zellkulturen (DSMZ). The ATCC number of K562, U937 and HEL were CCL-243, CRL-1593.2 and TIB-180. Details can be found on ATCC Official website. The number of SHI-1 and NB4 were ACC 645 and ACC 207, and details can be queried on DSMZ official website. The use of cell lines was approved by the Review Committee of the Ethics Department of the People's Hospital of Jiangsu University. These cell lines were cultured in RPMI-1640. The mycoplasma contamination of cell lines was negative by using quick mycoplasma test kit. Before we conducted this subject, these cell lines were tested by PCR.

The mononuclear cells of AML patients and healthy donors were isolated from bone marrow (BMMC) using the Ficoll-Hypaque gradient. According to the manufacturer’s instructions, Trizol reagent (Invitrogen, Carlsbad, CA, USA) was employed to extract total RNA from BMMC and leukemia cell lines. The cDNA was composed through reverse transcription on the iCycler Thermal Cycler (Eppendorf, Hamburg, Germany) using a reaction mixture, which contains 2 μg of total RNA, dNTPs 10 mM, random hexamers 10 μM, RNAsin 80 units, and 200 units of MMLV reverse transcriptase (MBI Fermentas, Hanover, USA). The system of reverse transcription was incubated for 10 min at 25 °C, 60 min at 42 °C, and then stored at − 20 °C. The primers of Foxo3 were 5′- GCAAGAGCTCTTGGTGGATCATCAA-3′ (forward) and 5′- TGGGGCTGCCAGGCCACTTGGAGAG-3′ (reverse), and the primers of circ-Foxo3 were 5′-GTGGGGAACTTCACTGGTGCTAAG-3′ (forward) and 5′-GGGTTGATGATCCACCAAGAGCTCTT-3′ (reverse). A 20 μL volume of reaction system (20 ng of cDNA, 0.8 μM of primers, 10 μM AceQTMqPCR SYBR Green Master Mix (Takara Shuzo Co, Ltd., Nojihigashi 7–4-38,Kusatsu,Shiga,Japan) and 0.4 μM ROX Reference Dye1(Invitrogen).) was used to perform Real-time RT-PCR. Amplification for circ-Foxo3 was carried out at 95 °C for 30 min, followed by 45 cycles at 95 °C for 5 s, 68.7 °C for 30 s 72 °C for 30 s and 80 °C for 31 s. Meanwhile, amplification for Foxo3 was carried out at 95 °C for 30 min, followed by 45 cycles at 95 °C for 5 s, 63.6 °C for 30s 72 °C for 30 s and 80 °C for 31 s. All reactions were performed on a 7500 Thermo cycler (Applied Biosystems, CA, and USA). Positive and negative controls were included in all tests. Following real-time RT-PCR, a melting curve analysis was carried out to demonstrate the specificity of the PCR product as a single peak. The relative levels of Foxo3 transcript were calculated by the following equation: N_Foxo3 = (E_Foxo3) ^{ΔCT Foxo3 (control-sample)} ÷ (E_ABL) ^{ΔCT ABL (control-sample)}. The parameter efficiency (E) was counted by the formula E = 10^{(− 1/slope) (the slope referred to CT versus cDNA concentration plot)}.

Foxo3 and circ-Foxo3 were detected by high-resolution melting analysis (HRMA) as reported previously. DNA direct sequencing was used for confirming positive samples.

We tested the gene mutation (CEBPA, NPM1, FLT3-ITD, KIT, N/K-RAS and IDH1/2) by next-generation sequencing technology.

All the data analysis was conducted on SPSS 20.0 software package (SPSS, Chicago, IL). Pearson chi-square analysis or variance test was used to distinguish differences in categorical variables, and the differences between two groups of continuous variables were compared by Mann-Whitney U test. In order to make a difference between the expression of AML patients and controls, Receiver operating characteristic curve (ROC) and area under the ROC curve were performed. We plotted ROC curves of AML and healthy people, and next a table of sensitivity and (1- specificity) of circ-Foxo3 and Foxo3 expression was drawn. We used these data to calculating the Yoden index (sensieivity+specificity-1) which could indicate the capacity of true patients and non-patients. The number which corresponding to the largest Yoden index was chosen as the cut-off value to separate AML patients into high and low groups. We used COX regression model, also known as the “proportional hazards model” (Cox model), to test if some common factors had effect on survival outcome, and if they were independent variables, This statistical method can simultaneously analyze the influence of many factors on the survival period. The impact of different Foxo3 and circ-Foxo3 expression on overall survival (OS) of AML patients was analyzed by Kaplan-Meier analysis. For all analyses, two-tailed P-values of 0.05 or less were determined statistically significant.

In this study, we test the expression level of Foxo3 and circ-Foxo3 in control people and de novo AML patients. The Foxo3 expression level (1.0 × 10^− 6-456.234, median 1.193) in de novo AML patients was obviously lower than in control people (0.001–49.528, median 5.619) (P = 0.009). Meanwhile, the circ-Foxo3 expression level (2.8 × 10^− 5-5.761, median 0.1198) was also lower in de novo AML patients than in control people (1.2 × 10^− 5-3.210, median 0.5017) (P = 0.04). The entire scatter diagram was represented at Fig. 1.

The diagnostic value of Foxo3 and circ-Foxo3 expression was analyzed by ROC curve. AML can be remarkably distinguished from controls with an AUC of 0.655 (95% CI: 0.556–0.753; P = 0.009) (Fig. 2a). According to the result of ROC curve analysis, we determined that the cut-off value of Foxo3 expression was at 0.856, the sensitivity and the specificity were 44.7 and 87.7%, respectively. Similarly, the circ-Foxo3 could differentiate AML patients from control people with an AUC of 0.633 (95% CI: 0.523–0.746; P = 0.041) (Fig. 2b). In a similar way, the cut-off value of circ-Foxo3 expression was at 0.233, the sensitivity and the specificity were 62.1 and 75%, respectively.

In order to analysis the correlation of Foxo3 and circ-Foxo3, Pearson analysis was conducted on the expression of Foxo3 and circ-Foxo3 in cell lines (K562, U937, NB4, SHI-1, and HEL). The result showed that Foxo3 and circ-Foxo3 were positively corrected (R = 0.98, P < 0.0021) (Fig. 3a). Besides, Pearson analysis was also conducted between these two genes in AML patients. Results revealed that they are positively corrected as well (R = 0.63, P < 0.0001) (Fig. 3b).

Based on the cut-off value of 0.856 of Foxo3 expression, AML patients was divided into two groups, high expression group (Foxo3 ^high) and low expression group (Foxo3 ^low). The high level was over 0.856 while the low expression was below. In the same way, according to the expression of circ-Foxo3, at the cut-off value of 0.233, the patients were also separated into two groups, circ-Foxo3 ^high and circ-Foxo3 ^low group.

After analyzing the clinical data, it displayed that no statistical significance were exhibited in sex, age, white blood cells (WBC), hemoglobin (HB), platelets (PLT), BM blast, Karyotype classification, WHO classification and other seven gene mutations (CEBPA, NPM1, FLT3-ITD, N/K-RAS, IDH1/2, DNMT3A, U2AF1 etc.) between Foxo3 ^high and Foxo3 ^low groups (Table.2).

In this study, total median follow-up time of the patients we tested the Foxo3 expression was 8.5 months. We carried out Kaplan-Meier survival analysis on Foxo3 ^high and Foxo3 ^low patients. It was found that there was a trend that Foxo3 ^high patients’ survival time (95% CI, 15.75–29.42 months, median value, 10 months) was longer than Foxo3 ^low group (95% CI, 10.54–22.46 months, median value, 7 months) (P = 0.192) (Fig. 4). In different classifications, this trend could be observed as well. In patients without M3 patients (non-M₃ patients), the Foxo3 ^high patients’ OS time (95% CI, 12.90–26.74 months, median value, 9 months) was longer than Foxo3 ^low patients (95% CI, 5.09–12.08 months, median value, 4 months) (P = 0.002) (Fig. 5a). Besides, in normal Karyotypic patients, the overall survival time of Foxo3 ^high patients (95% CI, 12.23–36.33 months, median value, 7 months) was significantly longer than Foxo3 ^low patients (95% CI, 5.31–16.24 months, median value, 6 months) (P = 0.034) (Fig. 5b). Furthermore, in intermediate and adverse (non-favorable) risk groups, Foxo3 ^low patients (95% CI, 4.22–10.69 months, median value, 3.5 months) had shorter survival time than Foxo3 ^high patients (95% CI, 10.91–25.13 months, median value, 6 months) (P = 0.004) (Fig. 5c). In the same way, the leukemia free survival time (LFS) of different classifications of patients was also analyzed. We found that in non-M3, non-favorable and normal Karyotypic groups, patients with high level of Foxo3 expression had longer LFS time than low level patients (Fig. 6).

Univariate and multivariate analyses (COX regression Model) were also conducted, applying age (≤ 60 y vs. > 60 y), sex (male vs. female), WBC (≥ 30 × 10⁹/L vs. < 30 × 10⁹/L), HB (< 110 g/L vs. ≥110 g/L), PLT (100 × 10⁹/L vs. 100 × 10⁹/L), karyotype classifications (favorable vs. intermediate vs. poor), gene mutations (mutant vs. wild-type) and Foxo3 expression status (high vs. low) as covariates. The univariate analysis showed that NPM1, Age, Karyotypic classification and WBC count were independent risk factors of AML patients (Exp > 1). After multivariable analysis, it figured out that Karyotypic classification, Foxo3 expression and Age were factors that affected AML prognosis. Among them, Foxo3 expression was a protective factor. Besides, Karyotypic classification and age were adverse prognosis factors either (Exp < 1) (Table.3).

We also studied the clinical data of patients with high and low level of circ-Foxo3 expression. Results showed that patients with high level of circ-Foxo3 expression survived longer (95% CI, 11.28–25.16 months, median value, 8 months) than low level of that (95% CI, 11.93–24.07 months, median value, 7 months), although this was out of statistical significance (P = 0.762). More than that, the difference of overall survival time between high and low level of circ-Foxo3 expression in different patients group also had no statistical significance. Multivariate analysis (COX regression model) was also conducted, applying age (≤ 60 y vs. > 60 y), sex (male vs. female), WBC (≥ 30 × 10⁹/L vs. < 30 × 10⁹/L), HB (< 110 g/L vs. ≥110 g/L), PLT (100 × 10⁹/L vs. 100 × 10⁹/L), karyotype classifications (favorable vs. intermediate vs. poor), gene mutations (mutant vs. wild-type) and Foxo3 expression status (high vs. low) as covariates. The result demonstrates that, in our study, these covariates above did not affect the prognosis of patients.