www.oncotarget.com Oncotarget, 2019, Vol. 10, (No. 65), pp: 6969-6980 Research Paper Flow cytometric significance of cellular alkaline phosphatase activity in acute myeloid leukemia Laura G. Rico1, Jordi Juncà1,2, Michael D. Ward3, Jolene A. Bradford3 and Jordi Petriz1 1 Functional Cytomics Group, Institut de Recerca contra la Leucèmia Josep Carreras, IJC Campus ICO-Germans Trias i Pujol, Institut Germans Trias i Pujol (IGTP), Universitat Autònoma de Barcelona, UAB, Badalona, Spain 2 Institut Català d’Oncologia, Hospital Germans Trias i Pujol (HGTiP), Badalona, Spain 3 Thermo Fisher Scientific, Eugene, Oregon, USA Correspondence to: Jordi Petriz, email: jpetriz@carrerasresearch.org Keywords: alkaline phosphatase; acute myeloid leukemia; stem cells; leukemic stem cells; CD34 Received: August 20, 2019 Accepted: November 13, 2019 Published: December 10, 2019 Copyright: Rico et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License 3.0 (CC BY 3.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited. ABSTRACT In this prospective hospital-based cohort study that included 43 newly diagnosed patients with acute myeloid leukemia, flow cytometric cellular alkaline phosphatase (ALP) activity within primitive leukemic cells allowed us to identify two groups of patients at diagnosis according to the numbers of leukemic blasts expressing ≥ 12% of ALP+ cells (27 patients, Group A) and less than 12% of ALP+ cells (16 patients, Group B). Differences in outcome for complete response, relapse or treatment resistance, and exitus were statistically analyzed and were significant, when comparing the two groups. The overall survival (OS) and event-free survival (EFS) differences between Group A and B were statistically significant. The survival of Group A patients was significantly shorter than those for Group B. No significant relationship was detected in outcome when comparing ELN prognostic-risk group based on cytogenetic and molecular profile (patients in the favorable, intermediate, and adverse risk groups). Flow cytometric cellular ALP activity at diagnosis may be used to estimate relapses and disease persistence more accurately. The limitations of our study include the small number of patients enrolled and a short follow-up, due to its prospective nature. INTRODUCTION blasts, usually requiring immediate treatment. AML is not a single disease and comprises a heterogeneous group of Flow cytometry immunophenotyping [1] has clonal hematopoietic malignancies with poor prognosis, become one of the mainstream applications for the being the most common type of acute leukemia in adults diagnosis and classification of several hematologic [2–4]. Patients with AML have high overall mortality neoplasms. This technology is indispensable for detection rate at three years, being higher among patients 65 years of leukemic blasts at a single cell level, clonal lineage or older [5]. Current strategies for the treatment of this assignment, identification of aberrant expression of disease, assign prognostically favorable subgroups into antigens, and detection of abnormal rare populations of standard chemotherapy regimens, while unfavorable blasts from normal progenitors, with extreme importance risk patients are generally considered for undergoing in tailoring decision-making. Moreover, functional flow consolidation with allogeneic stem cell transplantation cytometry testing can effectively provide new insights for (ASCT). research and evaluation of disease, and a more complete Alkaline phosphatase (ALP) is expressed at understanding of the complexities and challenges in the very low levels in somatic cells, whereas it is highly analysis of leukemic stem cells. expressed in primitive stem cells. ALP enzyme has a Acute myeloid leukemia (AML) arise from stem dimeric structure capable of binding Zn2+ and Mg2+ cells, causing a rapid overproduction of abnormal myeloid ions at different sites to stimulate or inhibit its catalytic www.oncotarget.com 6969 Oncotarget reaction. In humans, four forms of ALP cDNA have been prognostic-risk group based on cytogenetic and molecular cloned: one of them is widely distributed (liver, bone, profile [16], relevant blast immunophenotyping (CD34/ kidney) [6, 7], one restricted to the intestine [8], one to the CD117/CD123 backbone), initial treatment, and post- placenta [9], and one restricted to teratomas and germ cells remission therapy. The diagnostic performance of the ALP [10]. ALP hydrolyzes phosphate groups and its activity test as a binary classifier system, or the accuracy of the is involved in a broad range of essential physiological method to discriminate two ALP+ populations, was used and pathological processes. Immunohistochemical to determine its predictive value. As shown in Figure 2, expression of ALP has been reported for embryonal Receiver Operating Characteristic (ROC) curve analysis carcinoma and teratoma tumors with two different alkaline (area under the curve (AUC) = 0.768, 95% Confidence phosphatases localized in stem cell populations, as well Interval (CI) = 0.596 to 0.94, P-value < 0.0001) allowed as in embryonic ectodermal cells [10]. Pluripotent stem us to classify two identifiable ALP groups of patients at cells, embryonic stem cells, induced pluripotent stem diagnosis according to the numbers of leukemic blasts cells and embryonic germ cells express highly elevated expressing ≥ 12% of ALP+ cells (27 patients in Group A) ALP [11], as demonstrated by western blot, ELISA, and less than 12% of ALP+ cells (16 patients in Group B). immunohistochemistry, and highly sensitive fluorescent and chemiluminescent substrates [12]. Differences in outcomes between groups Because stem cells lack specific cell surface markers, the identification of this compartment can be difficult, making Overall, 43 patients, with a median age of 63 years cancer stem cells especially elusive [13]. Thus, we undertook (range 26–91), had a median value of 18.33% ALP+ a prospective cohort study based on our previous work [14], blast cells (range 0.26–96.93). Twenty-seven patients in with the objective to investigate whether there might be Group A had a median value of 23.76% of ALP+ blast differences in ALP activity within primitive leukemic cells, cells ranging from 13.83 to 96.63), and 16 patients in and its association with the potential risk of recurrence and Group B had a median value of 4.99% of ALP+ blast mortality in newly diagnosed patients with AML. cells ranging from 0.26 to 11.98 (P-value < 0.0001, 95% CI: 15.79 – 24.84). RESULTS Differences in outcome comparing the two groups were also analyzed. Seven patients in Group A achieved Patient characteristics and outcomes a complete response (25.9%), in contrast with 11 patients (68.8%) in Group B (P-value = 0.01, 95% CI = 1.34 to Between May 2015 and May 2018, a total of 106 30.99). Twenty-five patients in Group A relapsed or showed patients were diagnosed with acute myeloid leukemia at treatment resistance (92.6%), whereas this occurred only in our unit. After exclusion criteria were applied, 43 eligible 5 patients (31.3%) in Group B (P-value = <0.0001, 95% patients were identified and included in the overall analysis. CI = 0.00 to 0.26). Twenty-two patients in Group A died The characteristics of patients who were included are shown (81.5%), whereas 5 died in Group B (31.3%) (P-value = in Supplementary Table 1. The median age of patients was 0.002, 95% CI = 0.02 to 0.52). Table 2 summarizes the 67 years (range, 26–91 years). Of these, 20 (46.5%) of differences between Group A (≥12% ALP+ blasts cells) and patients were over 70 years of age, and 23 (53.5%) were Group B (<12% ALP+ blasts cells) adjusted for the age, sex, under 70 years of age. Overall, 30 (69.8%) patients were type of AML (de novo or secondary), WHO Classification male, and 13 (30.2%) were female (Table 1). Median of AML (2017 edition), cytogenetic and molecular follow-up for OS and EFS was approximately 12 months. alterations, prognostic-risk group based on cytogenetic The 30-month OS and EFS of all AML patients was 21.3% and molecular profile, relevant blast immunophenotyping and 19% respectively (Figure 1A, 1B). The OS and EFS (CD34/CD117/CD123 backbone), post-remission therapy, differences for adverse-, intermediate- and favorable-risk and outcomes (complete response achievement, relapse or patients were also compared (Figure 1C, 1D). treatment resistance, and exitus). Alkaline phosphatase activity differences in Survival differences between groups AML: identification of two patient groups The overall survival (OS) and event-free survival Differences in ALP+ blast cells at diagnosis were (EFS) differences between Group A and B were analyzed for the entire population. The value of ALP+ statistically significant. The survival for Group A patients blast cells was expressed as percentages of the median was significantly shorter than for Group B. and ranges around the median. Table 1 summarizes The 25-month OS in patients with ≥ 12% of ALP+ the median and range of ALP+ blast cells at diagnosis leukemic cells (Group A) was 9%, and the 10-month EFS according to the age, sex, type of AML (de novo or in the same group was 10%. The 30-month OS and EFS secondary), WHO 2017 classification [15], cytogenetic in patients with <12% of ALP+ leukemic cells (Group B) and molecular alterations, European LeukemiaNet (ELN) was 58% and 60% respectively (Figure 3). www.oncotarget.com 6970 Oncotarget Table 1: Numbers of ALP+ blast cells at diagnosis according to patient characteristics Patients (n = 43), n (%) APL+ blast cells, median (range), % Age at diagnosis <70 years 23 (53.5) 13.83 (1.00 - 96.63) >70 years 20 (46.5) 19.86 (0.26 - 96.63) Sex - Male 30 (69.8) 17.36 (0.26 - 96.63) - Female 13 (30.2) 20.75 (1.62 - 35.91) Type of AML - de novo 30 (69.8) 18.11 (1.20 - 96.63) - secondary 13 (30.2) 20.75 (1.00 - 95.92) WHO 2017 classification - AML with recurrent genetic abnormalities - AML with t(8;21)(q22;q22.1); RUNX1-RUNX1T1 2 (4.7) 49.17 (1.71 - 96.63) - AML with inv(16)(p13.1q22) or t(16;16)(p13.1;q22); CBFB-MYH11 2 (4.7) 21.81 (13.83 - 29.80) - AML with mutated NPM1 8 (18.6) 13.85 (1.00 - 24.71) - AML with biallelic mutations of CEBPA 3 (7.0) 17.09 (6.53 - 21.75) - AML with myelodysplasia related changes 4 (9.3) 19.19 (9.07 - 95.92) - Therapy-related AML 1 (2.3) 54.95 (54.95 - 54.95) - AML not otherwise specified - AML with minimal differentiation 3 (7.0) 15.49 (8.10 - 20.84) - AML without maturation 3 (7.0) 22.89 (2.65 - 23.76) - AML with maturation 8 (18.6) 17.43 (0.26 - 37.77) - Acute myelomonocytic leukemia 4 (9.3) 29.10 (4.96 - 96.63) - Acute monoblastic/monocytic leukemia 5 (11.6) 18.33 (1.20 - 26.91) Cytogenetic alterations - Normal karyotype 20 (46.5) 13.45 (0.26 - 35.27) - inv(16)(p13.1;q22) 2 (4.7) 21.81 (13.83 - 29.80) - Monosomy 7 or del 7q 3 (7.0) 23.76 (15.49 - 54.95) - Trisomy 8 5 (11.6) 35.91 (26.03 - 96.63) - t(8;21)(q22;q22.1) 2 (4.7) 49.17 (1.71 - 96.63) - t(11q23) 1 (2.3) 2.65 (2.65 - 2.65) - Complex karyotype 3 (7.0) 9.53 (4.96 - 20.84) - Others 7 (16.3) 20.75 (8.10 - 95.92) Molecular alterations - NPM1 mutated without FLT3-ITD 5 (11.6) 5.35 (1.00 - 17.89) - FLT3-ITD with or without NPM1 mutated 5 (11.6) 18.97 (1.20 - 26.03) - FLT3-TKD 1 (2.3) 29.8 (29.8 - 29.8) - CEBPA mutated 3 (7.0) 17.09 (6.53 - 21.75) - c-Kit mutated 1 (2.3) 13.83 (13.83 - 13.83) - Wild type 28 (65.1) 21.56 (0.26 - 96.63) Cytogenetic and molecular prognostic-risk group - Favorable 13 (30.2) 13.83 (1.00 - 96.63) - Intermediate 23 (53.5) 22.89 (0.26 - 96.63) - Adverse 7 (16.3) 9.53 (1.62 - 26.03) Relevant LSC markers - CD34+/CD123+/CD117+ 28 (65.1) 20.79 (0.26 - 96.63) - CD34+/CD123-/CD117+ 3 (7.0) 95.92 (2.65 - 96.63) - CD34-/CD123+/CD117+ 9 (20.9) 9.82 (1.00 - 30.39) - CD34-/CD123+/CD117- 3 (7.0) 18.33 (11.98 - 26.91) Initial treatment - CETLAM12<70* 20 (46.5) 11.82 (1.00 - 96.63) - CETLAM12>70† 5 (11.6) 15.49 (0.26 - 35.27) - FLUGAZA clinical assay‡ 11 (25.6) 21.75 (17.64 - 96.63) - Others§ 7 (16.3) 11.98 (1.62 - 30.39) Post remission therapy - Allogeneic SCT 12 (28.0) 23.02 (1.00 - 96.63) - Other 31 (72.0) 17.89 (0.26 - 96.63) Abbreviations: APL, alkaline phosphatase; AML, acute myeloid leukemia; WHO, world health organization; NPM1, nucleophosmin 1; FLT3-ITD, fms-like tyrosine kinase 3 internal tandem duplication; FLT3-TKD, fms-like tyrosine kinase 3 tyrosine kinase domain; SCT, stem cell transplant. * Cytarabine + idarubicin; †Cytarabine + fludarabine; ‡Cytarabine + fludarabine / Azacitidine; §Azacitidine / Decitabine / Cytarabine + idarubicin + quizatinib or placebo / None. www.oncotarget.com 6971 Oncotarget Multivariable analysis model for EFS, age, favorable and intermediate risk, and ALP Group were associated with a significant impact on Univariate and multivariate Cox proportional progression/relapse. The only factors that retained their hazards models were used to identify factors that are significance on multivariate analysis were favorable risk associated with OS and EFS in patients with AML and ALP group. (Table 3). ALP expression was a significant predictor of OS on univariate (hazard ratio [HR] = 0.32, 95% CI = DISCUSSION 0.11 to 0.87, P-value < 0.025) analysis. ALP expression was a significant predictor of EFS on univariate ([HR] Acute myeloid leukemia is a heterogeneous clonal = 0.19, 95% CI = 0.17 to 0.52, P-value = 0.0012) and disorder associated with a relatively high early mortality multivariate ([HR] = 0.25, 95% CI = 0.09 to 0.70, P-value rate and poor overall chances for recovery. Even with < 0.0079) analyses. In the univariate Cox model for OS, current treatments, the relative 5-year survival rate for age, favorable and intermediate risk, AML type, and patients with AML is only 5% after the age of 65 [17]. ALP Group were associated with a significant impact on Current diagnosis and medical treatment of AML is survival. The only factor that retained their significance expensive, not just in terms of cost, but also in terms of on multivariate analysis was age. In the univariate Cox the impact on prolonged overall treatment duration for Figure 1: Plots of Kaplan-Meier limit estimates of overall survival and event-free survival curve analysis of acute myeloid leukemia patients at diagnosis. Plots of Kaplan-Meier limit estimates of overall and event-free survival of acute myeloid leukemia ungrouped and grouped patients according to the risk. Plots of Kaplan-Meier limit estimates of overall and event-free survival of acute myeloid leukemia ungrouped patients are shown in (A and B) respectively. Overall and event-free survival differences for adverse-, intermediate- and favorable-risk patients (A, I, and F) are shown in (C and D) respectively. www.oncotarget.com 6972 Oncotarget patients and their families. Despite new therapy options However, the sensitivity of the assays, phenotypic shifts having been developed to treat AML, advancement in at relapse, low frequency of residual leukemic stem cells, understanding the pathogenesis of AML is continuously quality of the marrow aspirate, or the use of peripheral needed. Furthermore, the capability to accurately predict blood, may lead to identifying false-positive and false- survival is often difficult, raising the need for new negative patient groups by current approaches [19]. predictive biomarkers in AML therapeutics. AML comprises a dynamic disease with a very complex The fundamental goal of personalized medicine, landscape of aberrant phenotypic, cytogenetic and aimed at providing patient-specific molecular insights, molecular genetic abnormalities, that importantly confer combines genomic, clinical, and other types of data, to independent biological properties of the leukemic cells. provide strategies aimed at treatment optimization based on In 1970, Clarkson et al. found leukemic cells to individual patient characteristics. In recent years, advances behave as a self-maintaining population [20]. Although in genotypic arrays, digital PCR, and next-generation their data did not preclude some influx of leukemic “stem sequencing, also known as high-throughput sequencing, cells” from an unrecognized precursor compartment, they have played a key role in developing new assays, such as found no evidence to support this hypothesis. On the same those for predicting AML relapse after ASCT, for mutational subject of AML, Bonnet and Dick (1997), demonstrated that screening, risk stratification, guided therapy, or Measurable leukemic cells obtained from patients possess differentiative residual disease (MRD) monitoring. Multiparameter flow and proliferative capacities, as well as the self-renewal cytometry in the clinical laboratory has also become an properties found in leukemic stem cells [21]. However, the essential technology to increase the sensitivity of leukemia heterogeneity of the stem cell compartment supports that detection at the single cell level, especially for identifying the stem cell concept is not necessarily associated with a very low levels of residual disease immediately after specific cellular entity, but rather a function that can be treatment and during follow-up. assumed by numerous diverse cell types. As a result, the Assessment of MRD is today considered cancer stem cell concept cannot be universally applicable, indispensable for forecasting the significance and based on complex evolution, phenotypic heterogeneity and association with treatment outcome and survival [18]. therapeutic challenges of many human cancers, making Figure 2: Receiver operating characteristic (ROC) curves. According to ROC curve analysis, 12% of ALP+ was confirmed as the cut-off point of ALP+ leukemic cell counting for survival outcome of AML patients. ROC curve analysis (area under the curve = 0.768, 95% CI = 0.596 to 0.94, P-value < 0.0001) classified two identifiable ALP groups of patients at diagnosis according to the numbers of leukemic blasts expressing ≥ 12% of ALP+ cells (27 patients in Group A) and less than 12% of ALP+ cells (16 patients in Group B). www.oncotarget.com 6973 Oncotarget Table 2: Differences between Group A and Group B regarding total number of ALP+ blast cells at diagnosis Overall (n = 43) Group A (n = 27) Group B (n = 16) P-value (95% CI) ALP+ blast cells, median (range), % 18.33 (0.26 - 96.63) 23.76 (13.83 - 96.63) 4.99 (0.26 - 11.98) <0.0001 (15.79 - 24.84)**** Age at diagnosis, median (range), years 63.0 (26 - 91) 73.0 (26 - 81) 59.5 (31 - 91) 0.38 (-5.00 - 17.00) Sex, n (%) - Male 30 (69.8) 17 (63.0) 13 (81.3) 0.31 (0.50 - 16.99) - Female 13 (30.2) 10 (37.0) 3 (18.7) Type of AML, n (%) - de novo 30 (69.8) 19 (70.4) 11 (68.8%) 1 (0.22 - 4.93) - secondary 13 (30.2) 8 (29.6) 5 (31.2%) WHO 2017 classification, n (%) - AML with recurrent genetic abnormalities 15 (34.9) 9 (33.3) 6 (37.5) 1 (0.27 - 5.18) - AML with myelodysplasia related changes 4 (9.3) 3 (11.1) 1 (6.3) 1 (0.00 - 7.47) - Therapy-related AML 1 (2.3) 1 (3.7) 0 (0.0) 1 (0.00 - 65.75) - AML not otherwise specified 23 (53.5) 14 (51.9) 9 (56.2) 1 (0.29 - 4.99) Molecular alterations, n (%) - NPM1 mutated without FLT3-ITD 5 (11.6) 1 (3.7) 4 (25.0) 0.06 (0.71 - 441.52) - FLT3-ITD with or without NPM1 mutated 5 (11.6) 3 (11.1) 2 (12.5) 1 (0.09 - 11.26) - NPM1 and FLT3 wild type 33 (76.8) 23 (85.2) 10 (62.5) 0.14 (0.05 - 1.58) Cytogenetic and molecular prognostic-risk group, n (%) - Favorable 13 (30.2) 7 (25.9) 6 (37.5) 0.50 (0.36 - 7.82) - Intermediate 23 (53.5) 17 (63.0) 6 (37.5) 0.12 (0.08 - 1.50) - Adverse 7 (16.3) 3 (11.1) 4 (25.0) 0.39 (0.37 - 20.75) Blasts at diagnosis, median (range), % 45.0 (6.0 - 98.0) 50.0 (18.0 - 98.0) 39.0 (6.0 - 88.8) 0.35 (-7.99 - 23.00) WBC at diagnosis, median (range), n × 109/L 8.5 (0.5 - 249.1) 5.9 (0.7 - 249.1) 13.3 (0.5 - 65.4) 0.93 (-10.80 - 6.60) Relevant LSC markers, n (%) - CD34+/CD123+/CD117+ 28 (65.1) 19 (70.0) 9 (56.2) 0.51 (0.12 - 2.39) - CD34+/CD123-/CD117+ 3 (7.0) 2 (7.5) 1 (6.3) 1 (0.01 - 17.38) - CD34-/CD123+/CD117+ 9 (20.9) 4 (15.0) 5 (31.2) 0.26 (0.45 - 15.67) - CD34-/CD123+/CD117- 3 (7.0) 2 (7.5) 1 (6.3) 1 (0.01 - 17.38) Post remission therapy, n (%) - Allogeneic SCT 12 (28.0) 8 (30.0) 4 (25.0) 1.0000 (0.26 - 7.00) - Other 31 (72.0) 19 (70.0) 12 (75.0) Outcomes, n (%) - Complete response 18 (41.9) 7 (25.9) 11 (68.8) 0.01 (1.34 - 30.99)* - Relapse or Treatment resistance 30 (69.8) 25 (92.6) 5 (31.3) <0.0001 (0.00 - 0.26)**** - Exitus 27 (62.8) 22 (81.5) 5 (31.3) 0.002 (0.02 - 0.52)** Abbreviations: CI, confidence interval; ALP, alkaline phosphatase; n, number; AML, acute myeloid leukemia; WHO, world health organization; NPM1, nucleophosmin 1; FLT3-ITD, fms-like tyrosine kinase 3 internal tandem duplication; WBC, white blood cells; L, liter; LSC, leukemic stem cell; SCT, stem cell transplant. * p-value < 0.05; **p-value <0.01; ****p-value <0.0001. difficult the identification of rare molecular and phenotypic- this compartment, LSCs have unique features based on based putative leukemic stem cells [22, 23]. metabolic response to drugs, cell stress control, and other Aside from antigen expression properties on AML key aspects of stemness and chemotherapy resistance. bulk cells, and that antigen coexpression on leukemic Based on the clonal nature of tumors, and after the stem cells (LSC) makes it difficult to specifically target demonstration that leukemic stem cells from patients www.oncotarget.com 6974 Oncotarget Table 3: Multivariate analyses Overall Survival (OS) Event-Free Survival (EFS) Variable Univariate Multivariate Univariate Multivariate Hazard ratio P-value Hazard ratio P-value Hazard ratio P-value Hazard ratio P-value (95% CI) (95% CI) (95% CI) (95% CI) Age 1.10 (1.00 – 1.10) 0.0014** 1.04 (1.00 – 1.08) 0.0341* 1.00 (1.00 –1.10) 0.039* 1.01 (0.99 – 1.05) 0.21 Sex 1.40 (0.56 – 3.40) 0.48 0.58 (0.24 – 1.92) 0.47 1.00 (0.45 – 2.20) 1.00 - - ALP Group 0.32 (0.11 – 0.87) 0.025* 0.47 (0.15 – 1.34) 0.15 0.19 (0.07 – 0.52) 0.0012** 0.25 (0.09 – 0.70) 0.0079** AML Type 4.20 (1.80 – 9.80) 0.0009*** 1.50 (0.49 – 5.27) 0.43 1.60 (0.71 – 3.50) 0.27 0.48 (0.16 – 1.46) 0.20 Favorable Risk 0.25 (0.09 – 0.70) 0.0086 ** 0.33 (0.07 – 1.69) 0.18 0.24 (0.09 – 0.61) 0.0026 ** 0.23 (0.06 – 0.93) 0.039* Intermediate 2.70 (1.00 – 5.80) 0.042* 0.78 (0.19 – 3.11) 0.72 3.10 (1.40 – 6.80) 0.0053** 1.01 (0.33 – 3.12) 0.99 Risk Adverse Risk 1.60 (0.44 – 5.70) 0.50 - - 1.20 (0.41 – 3.60) 0.72 - - % Blasts 0.99 (0.97 – 1.00) 0.20 1.00 (0.98 – 1.03) 0.76 1.00 (0.98 – 1.00) 0.99 - - WBC 0.99 (0.98 – 1.00) 0.46 1.00 (0.99 – 1.01) 0.94 1.00 (0.99 – 1.00) 0.64 - - Abbreviations: CI, confidence interval; ALP, alkaline phosphatase; AML, acute myeloid leukemia; WBC, white blood cells. * p-value < 0.05; **p-value < 0.01; ***p-value < 0.001. with AML reconstitute the full spectrum of phenotypes to implement a new ALP flow cytometric method test, in the malignant populations that they regenerate in using a novel fluorogenic live cell permeant substrate for transplanted mice, we hypothesized that ALP could help ALP [12], since this stain efficiently visualized pluripotent to identify primitive leukemic cells [14]. Moreover, it has stem cells such as human embryonal carcinoma, murine been demonstrated that ALP activity may be useful for and human embryonic stem cells and induced pluripotent the diagnosis of acute myelomonocytic and monocytic stem cells under fluorescent microscopy [25]. Our leukemia in dogs [24]. In parallel with flow cytometric method uses a no-lyse no-wash approach, offering immunophenotyping, Stokol et al. used a cytochemical opportunities to combine live cell response and functional staining approach for ALP activity, concluding that ALP assessment in combination with cell immunophenotyping activity in canine AML can be especially helpful when (Supplementary Figure 1), while minimizing sample flow cytometric results are inconclusive. We then decided preparation effects on the cell biology as the primary Figure 3: Plots of Kaplan-Meier limit estimates of overall survival and event-free survival curve analysis of acute myeloid leukemia patients, according to the numbers of alkaline phosphatase (ALP) positive leukemic cells at diagnosis. Overall survival after diagnosis was compared between high ALP Group (A) and low ALP Group (B). Patients with ALP+ leukemic cells ≥ 12% achieved shorter OS than those with ALP+ leukemic cells < 12%. According to ROC curve analysis, 12% of ALP+ was confirmed as the cut-off point of ALP+ leukemic cell counting for survival outcome of AML patients. Kaplan-Meier plots of overall survival for grouped acute myeloid leukemia patients are shown in (A). Plots of Kaplan-Meier product limit estimates of event-free survival of grouped patients according to the numbers of ALP positive leukemic cells at diagnosis are shown in (B). www.oncotarget.com 6975 Oncotarget goal [26]. This method can also be used to identify rare significant impact to identify two different groups of cells [27]. Additionally, assays dealing with live kinetic patients. In addition, we have demonstrated that ALP also analysis, like calcium flux and enzyme rate, or assays that has a significant impact to predict complete response, depend on continued active drug transport just prior to relapse or treatment resistance, and exitus, independently analysis, like side population analysis must be done with of variables such as age, sex, type of AML (de novo or live cells, preferably with as little sample manipulation as secondary), WHO Classification of AML (2017 edition), possible. Recent work by the authors [14] showed elevated cytogenetic and molecular alterations, ELN prognostic- ALP activity in CD34+ cells in highly refractory cancers. risk group based on cytogenetic and molecular profile, This study was made possible by using this new non-toxic relevant blast immunophenotyping (CD34/CD117/CD123 cell-permeant fluorescent ALP for live cells without need backbone), initial treatment, and post-remission therapy to add multidrug resistant transporter inhibitors, taking (Table 2). advantage of one of the most-widely used enzyme-based Despite the cohort’s small number, some risk factors standards, known as the aldehyde dehydrogenase (ALDH) correlated with overall survival and event free survival. assay [28]. Previously, we demonstrated measurement of Specifically, age entered all univariate and multivariate ALP activity with ABCG2 efflux pump activity using this models except with the EFS in the multivariate, and reagent in combination with side population analysis, favorable risk only did not correlated well with OS in the making its use feasible for primitive stem cells expressing multivariate. Intermediate risk correlated well with OS ABC transporters. This can only be done with live healthy and EFS in the univariate. Finally, the ALP group was cells at the point of analysis [14]. significant in the univariate model for EFS and OS, as well Measuring live cell response function and as in the multivariate model for EFS. The ALP group only immunophenotype with the minimum possible sample did not correlate with OS in the multivariate (Table 3). manipulation, may increase the potential for discovery Surprisingly, the percentage of ALP+ blast cells of clinically relevant cell subsets. This area of research in the intermediate risk group (median: 22.89, range: is particularly interesting from a both theoretical and 0.26–96.63) was higher than that of the adverse risk group practical point of view. As far as we know, some (median: 9.53; range: 1.62–26.03), suggesting that the phenotypic characteristics may relate to clinical outcome intermediate risk group may include a relative number in AML patients, i.e, CD34 negativity related with NPM1 of patients with adverse risk (Table 1). Despite that, mutation, CD56 expression in AML with (8;21), or CD2, this result can also be limited by the bias of the cohort’s CD36, CD11b or CD56 positivity in other types of AML. small number, this observation is supported by the OS However, attempts to characterize these associations and EFS Kaplan-Meier analyses according to favorable, have yielded inconclusive results. Given we lack specific intermediate, and adverse definitions (Figure 1C and 1D). markers to identify if a given phenotype is tightly linked Even with these limitations, our results have significant or specifically caused by some mutations, maybe the only clinical and scientifically important implications. way to identify rare cell types is by combining phenotype The outcome of AML treatment is highly variable information with and functional assays. and still not individually predictable [29]. Currently, the At the light of our results, increased cellular ALP most refined predictive data that relate to the prognosis activity in AML with no evidence of aberrant antigen of AML are those derived of cytogenetic and molecular expression is still unclear and should be confirmed in a analysis. However, other most determinant predictive large series of patients. Importantly in the low residual factors may exist, and that they may reflect the degree of disease, MRD evaluation by whatever the method immaturity of the main burden of cells that proliferate in (leukemia-associated immunophenotypes (LAIPs) a given case of AML, and that this degree of immaturity identification vs. “different from normal” approaches) (as judged by a functional analysis) is more prognostically has the value of indicating the clinicians “how things determinant that the cytogenetic or molecular are going”, and that it may have a definitive underlying characteristics of leukemic cells. However, this is only a functional importance is to achieve and maintain a low or hypothesis to verify, and, in fact, we have no explanation negative value for ALP activity, because a marked increase for the apparently contradictory result of the lack of in this biomarker could be interpreted to reflect increased correlation of ALP and cytogenetics. Relative to other pathological functional activity in a not yet detected cancers, AML has a comparatively low level of genetic leukemic but previously fully identified population, the heterogeneity, suggesting that epigenetic heterogeneity emergence of a new and probably more immature and is of primary importance. Thus, epigenetic factors such aggressive type of cells that may in fact determine the as DNA methylation, genomic imprinting, histone clinical course of the patient, or to provide some clues modifications, and expression control by noncoding RNA, for more proliferative advantages related to a given may also play an important role. If confirmed in larger phenotype. series of patients, our results open a new perspective of In this work, we have shown and validated that research in the analysis of outcome in AML. Moreover, ALP expression by blast cells at diagnosis may have a if cellular alkaline phosphatase activity is a prognosticator www.oncotarget.com 6976 Oncotarget at diagnosis, it could help to investigate how to reach the laboratory studies, making possible to dissect human ultimate goal of individualized risk assessment as a guide malignancy. In the next few years, more investigations for therapy decision-making, or to design new novel should shed light on the translational utility of new treatment targets in AML. The development of novel ALP available biological indicators aimed at detecting cell inhibitors and modulators [30] for the treatment of AML function, and discovering in combination with flow are promising approaches to reduce the increased mortality cytometry immunophenotyping, the complex and associated with this heterogeneous disease characterized heterogeneous biology of the stem cell compartment in by a multitude of molecular abnormalities. Since several human pathology. methods are only applicable for MRD detection on a clinical routine basis, cellular ALP could help to better MATERIALS AND METHODS classify or to predict the behavior of the disease, as well as to is really predicting the outcome in a given patient. Study design and patient population Furthermore, it would be also very interesting to study the comparative value of monitoring MRD regularly with flow This was a prospective cohort study of patients cytometric ALP, as truly negative non-detectable MRD newly diagnosed with acute myeloid leukemia and status has strong clinical relevance in AML, making it an completed follow-up. From May 2015 through May essential tool in the overall strategy adopted to treat AML. 2018, 106 patients were diagnosed with AML at the The limitations of our study include the small hematological cytometry unit at Germans Trias i Pujol number of patients enrolled and a short follow-up, due to Hospital (HGTiP) in Badalona (Spain) that covers an area its prospective nature. We are fully aware that standard risk of approximately 700,000 inhabitants. After applying assessment markers have been well validated in numerous the exclusion criteria, 43 patients (male 30, female 13; large studies and suggest that findings in this small set median age 63 years, range 26–91) were included in our of 43 patients may not be valid in a larger population. study. Most of the patients were initially treated according Despite this small number, this study offers evidence to the Spanish CETLAM protocols, based on the use of to the hypothesis that intrinsic cellular ALP activity at idarubicin or fludarabine with cytarabine. A total of 12 out diagnosis may be used to estimate relapses and disease of 43 patients underwent allogeneic stem cell transplant persistence more accurately. Larger studies will be needed (ASCT) following chemotherapy. The patient’s risk to determine whether ALP activity in primitive leukemic (favorable, intermediate, adverse) was evaluated according cells is associated with the potential risk of recurrence and to ELN 2017 recommendations, based on cytogenetic mortality in newly diagnosed patients with AML. and/or molecular abnormalities [16]. Details of patient Since the discovery of the hybridoma technology, population and statistics are provided in Table 1 and Table many monoclonal antibodies have been available for 2 respectively. All patients enrolled in this study provided Figure 4: Representative flow cytometric study of the alkaline phosphatase activity in a patient of the ALP ≥ 12% group. Alkaline phosphatase positive cells are represented in combination with CD34 staining at diagnosis (upper row) and after relapse (lower row). Reference contour plots for two bone marrow aspirates are compared in the same patient, displaying high levels of alkaline phosphatase activity in combination with CD34 staining. The statistics in the region represents percentage of the gate. www.oncotarget.com 6977 Oncotarget their informed consent in accordance with the Declaration large numbers of red blood cells that are found in unlysed of Helsinki. All procedures were in accordance with blood or bone marrow from detection. DCV was excited at the internal protocols of our laboratory, which were 405 nm and its emission was collected using the following authorized by the HGTiP Clinical Investigation Ethical filter combination: 413 LP, 495 DLP, and 440/50 BP in the Committee, in agreement with current Spanish legislation. VL2 detector. SSC was detected using the violet laser 405 nm with Flow cytometry methods a 405/10 nm bandpass filter. APLS was detected with the blue laser 488 nm excitation and a 530/30 nm bandpass All EDTA-anticoagulated bone marrow (n = 38) filter in the BL1 (Blue Laser) detector. For the Attune™ and blood (n = 5) samples were prepared using a modified NxT upgraded with the yellow laser kit, PE was excited at previously developed method [27] aimed at avoiding 561 nm and its emission was collected using the following the lysis of erythrocytes during sample preparation, filter combination: 595 LP, 600 DLP, and 585/16 BP in the which can result in unwanted damage to leukocytes, YL1 detector. PECy5 was detected with the yellow laser and conceivably to leukemic cells. As most enzyme 561 nm excitation and a 695/40 nm bandpass filter in the functions are performed at 37°C, we first established YL3 detector. PECy7 was detected with the yellow laser 561 an optimal Alkaline Phosphatase Live Stain (APLS, nm excitation and a 780/60 nm bandpass filter in the YL4 Thermo Fisher Scientific) stable loading time (t = 20 min) detector. APLS, PE, PECy5, PECy7 and DCV fluorescence and temperature (T = 37°C) to measure ALP enzymatic are displayed on a logarithmic scale. Representative contour activity [14]. This dye is a cell-permeable fluorescent plot analysis for two bone marrow aspirates at diagnosis and substrate for ALP that is non-toxic to cells. Importantly, relapse, used to calculate ALP+ cell numbers in combination ammonium chloride- and paraformaldehyde-based lysing with CD34 staining is shown in Figure 4. solutions impair and almost completely abrogate ALP ALP+ cell numbers were determined in parallel to staining, are not recommended for detection of ALP+ cells flow cytometry immunophenotyping of leukemia cells (Supplementary Figure 1). The KG-1a cell line, derived performed at the HGTiP clinical hematology laboratory. from Human Caucasian bone marrow acute myelogenous Standard immunophenotyping was performed on a leukemia [31], was used as a positive control for highly FC500 (Beckman Coulter) flow cytometer until October expressing ALP+ cells (Supplementary Figure 2). 2016, and on a Navios™ flow cytometer (Beckman Briefly, our staining strategy used Vybrant™ Coulter) from this date onwards. Blast cells were DyeCycle™ Violet stain (DCV, Thermo Fisher Scientific), identified using cell surface protein markers according a low cytotoxicity permeable DNA-specific dye that can to the ELN recommendations [16]. MPO/CD45/CD34/ be used for DNA content cell cycle and stem cell Side CD117/CD123/Myeloid/T-Lymphoid/B-lymphoid and Population analysis by flow cytometry [32]. DCV can be megakaryocyte (MK) markers were used for typifying excited with violet 405 nm laser light and can be used for the blastic population through the identification of simultaneous measurement with APLS, which is excited LAIPs (whenever possible), or through the detection with blue 488 nm laser light and its emission can be of abnormalities in the different-from-normal cellular collected using a standard FITC filter (for example 530/30 distribution in association with a low expression of CD45 nm). Subsequently, cells were stained with fluorophore- and low SSC. A minimum of 20,000 events were studied at conjugated monoclonal antibodies for 20 min at room diagnosis. A marker was considered as positive when was temperature as recommended by the manufacturer (PE- expressed by ≥20% of blastic cells. Assessment of MRD CD123, PECy5-CD34 and PECy7-CD117, Sysmex was performed on bone marrow aspirates with a minimum GmbH). The improved phycoerythrin (PE) signal using sensitivity of 0.1%. LAIPs identified at diagnosis were 561 nm excitation together with the fact that there is carefully looked for, and shifts in the relative scatter no need for color compensation between FITC and distribution of the different-from-normal pattern were also PE under 488 nm and 561 nm excitation, results in the taken into account. improved immunophenotypic analysis of leukemic cells by multicolor flow cytometry [33]. Statistical methods Stained cells were diluted with Hank’s Balanced Salt Solution (HBSS) (1000 µL final volume) prior to We used receiver operating characteristic (ROC) sample acquisition. All cell measurements were done analysis, area under the curve (AUC), sensitivity and using the Attune™ Acoustic Focusing Cytometer and the specificity to measure prognostic accuracy of our test in Attune™ NxT Acoustic Focusing Cytometer (Thermo predicting EFS. The optimal cut-off value was identified Fisher). Samples were acquired at 25–100 µL/min sample according to the Youden’s Index to classify patients into rates, and a minimum of 100,000 DCV+ events were two groups [34]. Differences between categorical variables collected per sample when possible. Threshold levels in each group were compared using two-sided Fisher’s were set empirically using a Violet-Side Scatter (SSC) vs. exact tests. Differences between continuous variables in DCV-H dual parameter plot to eliminate debris and the each group were compared using two-sided Wilcoxon www.oncotarget.com 6978 Oncotarget rank sum tests. EFS and OS rates were estimated by the REFERENCES method described by Kaplan and Meier [35]. EFS was defined as the time from diagnosis until progression, 1. Cossarizza A, Chang HD, Radbruch A, Akdis M, Andrä I, death, or the last follow-up date. OS was defined as the Annunziato F, Bacher P, Barnaba V, Battistini L, Bauer WM, time from diagnosis until death or the last follow-up date. Baumgart S, Becher B, Beisker W, et al. Guidelines for the Patients that underwent ASCT were censored. Differences use of flow cytometry and cell sorting in immunological in survival between each group were tested for statistical studies. Eur J Immunol. 2017; 47:1584–797. https://doi. significance using the two-sided log-rank test with the org/10.1002/eji.201646632. [PubMed] Bonferroni method to adjust P value. Univariate analysis 2. Ley TJ, Miller C, Ding L, Raphael BJ, Mungall AJ, using the Cox proportional hazards model was performed Robertson A, Hoadley K, Triche TJ Jr, Laird PW, Baty JD, to investigate the impact of % ALP at diagnosis on EFS Fulton LL, Fulton R, Heath SE, et al, and Cancer Genome and OS, adjusting for the following variables: age, sex, Atlas Research Network. Genomic and epigenomic ELN prognostic-risk group, blast percentage, white blood landscapes of adult de novo acute myeloid leukemia. N cell (WBC) count and AML type (de novo vs, secondary). Engl J Med. 2013; 368:2059–74. https://doi.org/10.1056/ Factors prognostic for EFS and OS with a P-value < 0.5 NEJMoa1301689. [PubMed] in the univariate analysis were studied in a multivariate 3. Papaemmanuil E, Gerstung M, Bullinger L, Gaidzik VI, analysis. A P-value < 0.05 was considered statistically Paschka P, Roberts ND, Potter NE, Heuser M, Thol F, Bolli significant. R Studio version 1.1.463 (https://www.R- N, Gundem G, Van Loo P, Martincorena I, et al. Genomic project.org/) was used for all statistical analysis. Classification and Prognosis in Acute Myeloid Leukemia. N Engl J Med. 2016; 374:2209–21. https://doi.org/10.1056/ CONCLUSIONS NEJMoa1516192. [PubMed] 4. Döhner H, Weisdorf DJ, Bloomfield CD. Acute Myeloid In this study we have found that increased cellular Leukemia. N Engl J Med. 2015; 373:1136–52. https://doi. alkaline phosphatase activity in leukemic cells at diagnosis org/10.1056/NEJMra1406184. [PubMed] was significantly associated with a higher risk of relapse, 5. Walter RB, Kantarjian HM, Huang X, Pierce SA, Sun or treatment resistance, and mortality. Due to the small Z, Gundacker HM, Ravandi F, Faderl SH, Tallman MS, number of patients enrolled in this study, this finding Appelbaum FR, Estey EH. Effect of complete remission requires further investigation. and responses less than complete remission on survival in acute myeloid leukemia: a combined Eastern Cooperative Author contributions Oncology Group, Southwest Oncology Group, and M. D. Anderson Cancer Center Study. J Clin Oncol. 2010; 28:1766– L.G.R., and J.P. performed the experiments. L.G.R., 71. https://doi.org/10.1200/JCO.2009.25.1066. [PubMed] J.J., M.D.W., J.A.B., and J.P performed data interpretation, 6. Seargeant LE, Stinson RA. Evidence that three structural and drafted the manuscript. J.P. conceived of the study, genes code for human alkaline phosphatases. Nature. 1979; participated in its design, conducted most experiments, 281:152–54. https://doi.org/10.1038/281152a0. [PubMed] and revised the manuscript. 7. McKenna MJ, Hamilton TA, Sussman HH. Comparison of human alkaline phosphatase isoenzymes. Structural ACKNOWLEDGMENTS AND FUNDING evidence for three protein classes. Biochem J. 1979; 181:67–73. https://doi.org/10.1042/bj1810067. [PubMed] The authors are very grateful to Sara Vergara and Minerva Raya, for kindly providing samples used in 8. Hass PE, Wada HG, Herman MM, Sussman HH. Alkaline this study. The authors are also very grateful for their phosphatase of mouse teratoma stem cells: immunochemical advice and technical support to Sergio Ramón, Víctor and structural evidence for its identity as a somatic gene Querol, Clara Streiff, Paola Paglia, Luca Tiengo and Lluís product. Proc Natl Acad Sci U S A. 1979; 76:1164–68. Sanz from Life Technologies for all his comments and https://doi.org/10.1073/pnas.76.3.1164. [PubMed] discussions on earlier work in this research field. Jordi 9. Terao M, Mintz B. Cloning and characterization of a cDNA Petriz also acknowledges the financial support from The coding for mouse placental alkaline phosphatase. Proc Natl Obra Social La Caixa-Fundació Josep Carreras and The Acad Sci U S A. 1987; 84:7051–55. https://doi.org/10.1073/ Generalitat de Catalunya. pnas.84.20.7051. [PubMed] 10. Berstine EG, Hooper ML, Grandchamp S, Ephrussi B. CONFLICTS OF INTEREST Alkaline phosphatase activity in mouse teratoma. Proc Natl Acad Sci U S A. 1973; 70:3899–903. https://doi. M.D.W. and J.A.B. work for Thermo Fisher org/10.1073/pnas.70.12.3899. [PubMed] Scientific, which is in the business of selling flow 11. Takahashi K, Tanabe K, Ohnuki M, Narita M, Ichisaka T, cytometers and flow cytometry reagents. Tomoda K, Yamanaka S. Induction of pluripotent stem cells www.oncotarget.com 6979 Oncotarget from adult human fibroblasts by defined factors. Cell. 2007; 23. Nguyen LV, Vanner R, Dirks P, Eaves CJ. Cancer stem cells: 131:861–72. https://doi.org/10.1016/j.cell.2007.11.019. an evolving concept. Nat Rev Cancer. 2012; 12:133–43. [PubMed] https://doi.org/10.1038/nrc3184. [PubMed] 12. Singh U, Quintanilla RH, Grecian S, Gee KR, Rao MS, 24. Stokol T, Schaefer DM, Shuman M, Belcher N, Dong L. Lakshmipathy U. Novel live alkaline phosphatase substrate Alkaline phosphatase is a useful cytochemical marker for for identification of pluripotent stem cells. Stem Cell Rev the diagnosis of acute myelomonocytic and monocytic Rep. 2012; 8:1021–29. https://doi.org/10.1007/s12015-012- leukemia in the dog. Vet Clin Pathol. 2015; 44:79–93. 9359-6. [PubMed] https://doi.org/10.1111/vcp.12227. [PubMed] 13. Jordan CT, Guzman ML, Noble M. Cancer stem cells. N 25. Martí M, Mulero L, Pardo C, Morera C, Carrió M, Engl J Med. 2006; 355:1253–61. https://doi.org/10.1056/ Laricchia-Robbio L, Esteban CR, Izpisua Belmonte JC. NEJMra061808. [PubMed] Characterization of pluripotent stem cells. Nat Protoc. 14. Rico LG, Juncà J, Ward MD, Bradford J, Petriz J. Is alkaline 2013; 8:223–53. https://doi.org/10.1038/nprot.2012.154. phosphatase the smoking gun for highly refractory primitive [PubMed] leukemic cells? Oncotarget. 2016; 7:72057–66. https://doi. 26. Petriz J, Bradford JA, Ward MD. No lyse no wash flow org/10.18632/oncotarget.12497. [PubMed] cytometry for maximizing minimal sample preparation. 15. Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, Methods. 2018; 134–135:149–63. https://doi.org/10.1016/j. Stein H, Thiele J, editors. WHO classification of tumours of ymeth.2017.12.012. [PubMed] haematopoietic and lymphoid tissues (Revised 4th edition). 27. Fornas O, Garcia J, Petriz J. Flow cytometry counting of International agency for research on Cancer (IARC). Lyon; CD34+ cells in whole blood. Nat Med. 2000; 6:833–36. 2017. 4002 p. https://doi.org/10.1038/77571. [PubMed] 16. Döhner H, Estey E, Grimwade D, Amadori S, Appelbaum 28. Corti S, Locatelli F, Papadimitriou D, Donadoni C, Salani S, FR, Büchner T, Dombret H, Ebert BL, Fenaux P, Larson Del Bo R, Strazzer S, Bresolin N, Comi GP. Identification of RA, Levine RL, Lo-Coco F, Naoe T, et al. Diagnosis and a primitive brain-derived neural stem cell population based management of AML in adults: 2017 ELN recommendations on aldehyde dehydrogenase activity. Stem Cells. 2006; from an international expert panel. Blood. 2017; 129:424–47. 24:975–85. https://doi.org/10.1634/stemcells.2005-0217. https://doi.org/10.1182/blood-2016-08-733196. [PubMed] [PubMed] 17. De Kouchkovsky I, Abdul-Hay M. ‘Acute myeloid 29. Ossenkoppele G, Schuurhuis GJ. MRD in AML: does it leukemia: a comprehensive review and 2016 update’. Blood already guide therapy decision-making? Hematology Am Cancer J. 2016; 6:e441. https://doi.org/10.1038/bcj.2016.50. Soc Hematol Educ Program. 2016; 2016:356–65. [PubMed] [PubMed] 30. Kuang Y, Miki K, Parr CJ, Hayashi K, Takei I, Li J, Iwasaki 18. Gupta S, Devidas M, Loh ML, Raetz EA, Chen S, Wang M, Nakagawa M, Yoshida Y, Saito H. Efficient, Selective C, Brown P, Carroll AJ, Heerema NA, Gastier-Foster JM, Removal of Human Pluripotent Stem Cells via Ecto- Dunsmore KP, Larsen EC, Maloney KW, et al. Flow-cytometric Alkaline Phosphatase-Mediated Aggregation of Synthetic vs. -morphologic assessment of remission in childhood acute Peptides. Cell Chem Biol. 2017; 24:685–694.e4. https://doi. lymphoblastic leukemia: a report from the Children’s Oncology org/10.1016/j.chembiol.2017.04.010. [PubMed] Group (COG). Leukemia. 2018; 32:1370–79. https://doi. 31. Koeffler HP, Billing R, Lusis AJ, Sparkes R, Golde DW. org/10.1038/s41375-018-0039-7. [PubMed] An undifferentiated variant derived from the human acute 19. Ravandi F, Walter RB, Freeman SD. Evaluating myelogenous leukemia cell line (KG-1). Blood. 1980; measurable residual disease in acute myeloid leukemia. 56:265–73. https://doi.org/10.1182/blood.V56.2.265.265. Blood Adv. 2018; 2:1356–66. https://doi.org/10.1182/ [PubMed] bloodadvances.2018016378. [PubMed] 32. Petriz J. Flow cytometry of the side population (SP). Curr 20. Clarkson B, Fried J, Strife A, Sakai Y, Ota K, Okita T. Protoc Cytom. 2013; Chapter 9:Unit9.23. https://doi. Studies of cellular proliferation in human leukemia. 3. org/10.1002/0471142956.cy0923s64. [PubMed] Behavior of leukemic cells in three adults with acute 33. G Rico L, Juncà J, Ward MD, Bradford J, Petriz J. Yellow- leukemia given continuous infusions of 3H-thymidine green laser-based flow cytometry for CD34+ progenitor for 8 or 10 days. Cancer. 1970; 25:1237–60. https:// cell counting. Cytometry A. 2018; 93:172–76. https://doi. doi.org/10.1002/1097-0142(197006)25:6<1237::AID- org/10.1002/cyto.a.23299. [PubMed] CNCR2820250602>3.0.CO;2-7. [PubMed] 34. Youden WJ. Index for rating diagnostic tests. Cancer. 1950; 21. Bonnet D, Dick JE. Human acute myeloid leukemia is 3:32–35. https://doi.org/10.1002/1097-0142(1950)3:1<32::AID- organized as a hierarchy that originates from a primitive CNCR2820030106>3.0.CO;2-3. [PubMed] hematopoietic cell. Nat Med. 1997; 3:730–37. https://doi. 35. Kaplan EL, Meier P. Nonparametric Estimation from org/10.1038/nm0797-730. [PubMed] Incomplete Observations. J Am Stat Assoc. 1958; 53:457– 22. Blau HM, Brazelton TR, Weimann JM. The evolving concept 81. https://doi.org/10.1080/01621459.1958.10501452. of a stem cell: entity or function? Cell. 2001; 105:829–41. https://doi.org/10.1016/S0092-8674(01)00409-3. [PubMed] www.oncotarget.com 6980 Oncotarget