FISH⁺CD34⁺CD38^- cells detected in newly diagnosed acute myeloid leukemia patients can predict the clinical outcome

In this study we included 45 de novo AML patients (25 male, 20 female) with a median age of 42 years (range, 14–71), admitted in the Institute of Hematology, Changhai hospital, between June 2010 and October 2012 (Table 1). Patients with a history of myelodysplastic syndrome or therapy-related AML as well as acute promyelocytic leukemia were not included. The patients had the following cytogenetic abnormalities: t(8; 21) (n = 22), inv(16) (n = 6), trisomy 8 (n = 5), MLL rearrangement (n = 3), trisomy 21 (n = 2), 7q^- (n = 2), 9q^- (n = 2) and t(6;9) (n = 1). (Additional file 1: Table S1) All research samples represented excess bone marrow collected at diagnosis. The present study was approved by the Changhai Hospital Institutional Review Board and signed informed consent was obtained from each patient in accordance with the Declaration of Helsinki.

Induction treatment comprised Daunorubicin (DNR) at 45–60 mg/m²/d or Idarubicin (IDA) at 8-10 mg/m²/d by 30-minute IV infusion on days 1 to 3 and Cytarabine (Ara-C) at 100 mg/m²/d by continuous IV infusion from days 1 to 7 (DA regimen). In patients not reaching CR after 2 courses of DA, a salvage course (FLAG), consisted of five days of treatment with a 30-minute infusion of fludarabine (Flu) 30 mg/m²/d followed, four hours later, by a 4-hour infusion of Ara-C 2 g/m²/d, was given. Granulocyte colony-stimulating factor (G-CSF) 300 μg/day s.c. was administered 12 hours before starting fludarabine, for five days and then continued after the end of therapy until myeloid recovery. Patients reaching CR then received 4 monthly consolidation cycles with Ara-C at 2 g/m²/12 h by 2-hour IV infusion for 3 days, followed by G-CSF starting at day 8 or neutrophile counts less than 1 × 10⁹/L until neutrophil recovery. Patients with CNS disease received triple intrathecal infusions (methotrexate 10 mg, cytarabine 50 mg, dexamethasone 5 mg) twice a week until disappearance of blast cells in the cerebrospinal fluid. Patients were candidates for allogeneic hematopoietic stem cell transplantation (allo-HSCT) in first CR if they had a matched sibling or ≥9/10 HLA allele fully matched unrelated donor. Standard myeloablative or reduced-intensity conditioning regimens were allowed, depending on the patient age and health status. Of all patients, 14 patients received allo-HSCT during the follow-up period.

Mononuclear cells (MNCs) were isolated from bone marrow aspirates, collected at diagnosis, by density gradient centrifugation (Ficoll-Paque, GE Healthcare Life Sciences). The experiments were done on fresh cells. Primary AML cells were washed and resuspended into 100 μl of cold (4°C) Phosphate Buffered Saline solution (PBS) + 2% fetal calf serum (FCS) and incubated for 30 minutes on ice with combinations of fluorescein isothiocyanate-(FITC), Peridinin-Chlorophyll-Protein Complex-Cy5.5- (PerCP-CY5.5), phycoerythrin- (PE), phycoerythrin-Cy7- (PE-CY7), allophycocyanin (APC)-labeled monoclonal antibodies (MoAbs). Anti-CD45 PerCP-CY5.5, anti-CD19 APC, anti-CD7 FITC, anti-CD33 PE-CY7, anti-CD13 PE, anti-CD10 APC, anti-CD34 FITC, anti-HLA-DR PE-CY7, anti-CD117 PE, anti-cCD3 APC, anti-MPO FITC, anti-cCD79a PE, anti-CD14 APC, anti-CD64 FITC, anti-CD2 PE-CY7, anti-CD11c PE, anti-CD15 FITC, anti-CD56 PE-CY7, anti-CD66c PE, anti-CD34 APC, anti-CD11b FITC, anti-CD38 PE-CY7 and anti-CD123 PE MoAbs were all from BD Biosciences (San Jose, CA). After antibody staining, cells were washed with cold PBS, resuspended in 1 ml of cold PBS and stained for 5 minutes with 2 μg/ml propidium iodide (Sigma-Aldrich), enabling exclusion of dead cells. Cells were kept on ice until fluorescence-activated cell sorting (FACS) analysis. Data acquisition was performed using a FACSAira flow cytometery with BD FacsDiVa software (BD Biosciences, Franklin Lakes, NJ, USA) at the Institute of Hematology, Flow Cytometry Core.

To sort the stem cell subpopulation, a total of 1 × 10⁶ primary AML MNCs was stained with the following conjugated antibodies: monoclonal PE-conjugated anti-CD34, FITC-conjugated anti-CD38 and APC-conjugated anti-CD45 (BD Biosciences). Gates were set to detect the CD34⁺CD38^- cells as shown in Figure 1. Cells were sorted using a FACSAria flow cytometery. Cells were kept on ice during the whole procedure. Purity of sorted populations was >99%.

As part of the diagnostics for AML, chromosome analyses (G-banding) and FISH were performed on diagnostic BM samples of all AML patients. For Flow-FISH analysis, 1000 to 3000 cells from CD34 + CD38- cell subpopulation were directly sorted into 20 μL drops of PBS that were placed on a grease-free glass slide. The cells were then fixed with 3:1 methanol-glacial acetic acid (Sigma-Aldrich). Air-dried slides were analyzed by FISH in our laboratory, using probes specific for the patient’s known cytogenetic abnormality (Abbot Molecular).

Probes were denatured and hybridized according to manufacturer’s instructions. For evaluation of the Flow-FISH results, hybridization and deletion signals were scored in 500 interphase nuclei with an Axioscop 20 (Carl Zeiss, Jena, Germany) fluorescence microscope with three single-band-pass filters and one triple-band-pass filter. The images were captured with a digital camera using CytoVision 4.1 software (Applied Imaging Corp., Newcastle, U.K.). Results were expressed as percentage FISH-positive cells relative to number of cells analyzed.

Complete response and relapse rates were defined according to the Cheson criteria [14]. Pair-wise comparisons between patients’ characteristics (covariates) were performed using the Mann–Whitney test or Kruskal-Wallis test for continuous variables and with the Fisher’s exact test for categorical variables. Overall survival (OS) was measured from the date of diagnosis until death or last follow-up and events-free survival (EFS) from the date of complete remission (CR) until death, relapse or last follow-up, respectively. Patients alive in complete remission were censored at the time of last contact. Overall and events-free survival rates were estimated by the Kaplan-Meier method and compared using the log-rank test [15]. The cutoff value for FISH⁺CD34⁺CD38^- was chosen at a percentage that resulted in the largest difference in survival between the two groups defined by that cutoff. In computations of the cumulative incidence of leukemic relapse (CIR), death and relapse were included as competing events; survival was counted as a censored event. The CIR was compared between groups using the method of Gray, with estimation determined by the method of Kalbfleisch and Prentice [16]. Hazard ratios are given with 95% confidence intervals (95% CI). Survival-time data (events-free survival and overall survival) and covariates (age, leukocyte count, karyotype, FLT3 mutations and percentage of CD34⁺CD38^-FISH⁺ cells) were analyzed using the method of backward Cox proportional hazards regression. All calculations were performed using the R 2.7.2 software package.

To distinguish normal and leukemic cells within the CD34⁺CD38^- cell compartment, we established a unique protocol for conducting FISH on flow-sorted cells. Briefly, CD34⁺CD38^- populations from AML patients at diagnosis with FISH-detectable cytogenetic abnormalities were sorted onto glass slides. We could sort as few as 1000 cells on slides, from which at least 500 cells were successfully scored by FISH. Results were calculated as the percentage of FISH-positive cells relative to number of cells analyzed, and the median level was 68.2% (range, 7.8%-89.6%). Results for representative patients (Patient #9 AML1/ETO; Patient #21 CBFβ/MYH11) are illustrated in Figure 1.

The percentage of leukemia initiating cells (FISH⁺CD34⁺CD38^-) was then quantified as the value of the percentage of FISH-positive cells multiplies the percentage of CD34⁺CD38^-cells in the blast (CD45^dimSSC^low). The percentage of FISH⁺CD34⁺CD38^- cells was evaluated in 45 primary AML samples at diagnosis. The median expression of CD34⁺CD38^- was 3.3% in the blast (range, 0.1–52.8%). The FISH⁺CD34⁺CD38^- cells were rare in the newly diagnosed AML patients, constituting a median of only 2.31% (n = 45, range, 0.01%-29.4%) of the total blast cells. According to the LIC level detected by Flow-FISH analysis, patients were categorized into the two groups: low LIC (<1%) and high LIC (≥1%), representing 44.4% and 55.6% of all patients, respectively. The clinical characteristics of the patients are provided in Table 1. Comparison of clinical and laboratory characteristics showed no significant differences between two groups in age, gender, white blood cell (WBC) count, platelet (PLT) count, blast percentage, FAB subtype distribution, FLT3 mutation status and cytogenetics risk at diagnosis (Table 1).

To determine the prognostic impact of detecting LIC load at diagnosis, we analyzed the treatment outcome for 45 patients enrolled in a standardized treatment program (Table 2). The 2-year overall survival for patients with low LIC load was 0.7257[0.4082, 0.8916], compared with 0.1675[0.0323, 0.3947] for the patients with high LIC load (P = 0.0037). And 2-year events-free survival were 0.6723[0.3262, 0.8687] versus 0.1633[0.0328, 0.3825], respectively (P =0 .0018) (Figure 2).

Events, including relapse and death, were more frequent in patients with a high LIC load. Among 21 patients who received CR with a high LIC load, 10 (46.7%) experienced a relapse; among 18 patients who achieved CR with a low LIC load, 2 (11.1%) experienced relapse (P = 0.018). 5 treatment related death (TRD) (23.8%) occurred in the 21 patients who achieved CR with a high LIC load, whereas 2 TRD (5.6%) occurred in the 18 patients who received CR with a low LIC load (P = 0.418). The estimated 2-year cumulative incidence of leukemic relapse for patients with high LIC load was 56.7% ± 5.1% versus 18.0% ± 1.5% for patients with low LIC load (P = 0.021) (Figure 3).

In a univariate analysis of recognized prognostic factors, including age, FLT3 mutation status, leukocyte count, cytogenetics and allo-HSCT, cytogenetics were found to be significantly associated with overall survival (P = 0.01), whereas the leukocyte count, age, karyotype, FLT3 mutation status and allo-HSCT had no impact on events-free survival. We next performed a multivariate Cox proportional hazards model for OS and EFS, a percentage of LIC load of over 1% at diagnosis remained the significant predictor of outcome (P = 0.003 for OS; P = 0.004 for EFS) (Tables 3 and 4).