1. Introduction
The diagnostic work-up of acute myeloid leukemia (AML) is an integrated process using cytology, immunophenotyping, cytogenetics as well as the search for specific mutations, as several subtypes are defined by cytogenetic and molecular features in the WHO 2016 classification [
1,
2,
3]. This is especially true for acute promyelocytic leukemia (APL), a subtype of AML defined genetically by the translocation t(15;17) and the promyelocytic leukemia–retinoic acid receptor α (PML-RARA). APL is highly curable if appropriate therapy based on all-trans retinoic acid (ATRA) is timely started, preventing bleeding due to a consumption coagulopathy, which is responsible for a high early mortality rate [
4,
5]. Therefore, there is a need for a rapid diagnostic screening. Results of cytogenetics and molecular tests are not readily available and fluorescence in situ hybridization (FISH) is not available in many laboratories. Therefore, most of the times, the first screening is made by cytology and immunophenotyping in many institutions, since the results of cytogenetics and molecular tests are not immediately available.
Several works have reported that leukemic cells in APL usually do not express CD34 and HLA-DR, but this is also the case in several other types of AML, mainly those presenting the NPM1 mutation [
2,
6,
7,
8]. Recently, some groups [
9,
10] have described a method using multidimensional analysis in radar plots of the Kaluza software, to screen APL compared to non-APL acute myeloid leukemia. However, this method is dependent on a specific software and its reproducibility requires a standardization of antibodies, fluorochromes and analysis.
Our aim was to develop a diagnostic algorithm based on the expression intensity of several antigens examined by multiparametric flow cytometry (MFC) that depends only on internal standardization and could be performed with several antibody combinations and software of analysis, but could reliably discriminate between APL and other types of AML.
2. Material and Methods
2.1. Study Design
We reanalyzed the flow cytometric files of all newly diagnosed cases of de novo AML entering our institution between 2017 and 2019 and where cytogenetics and molecular diagnosis for PML/RARA, nucleophosmin (NPM1), fms-related tyrosine kinase 3 gene internal tandem duplications (FLT3-ITD) and second tyrosine kinase domain (FLT3-TKD) were available. These techniques had been introduced in the routine diagnostic work-up of AML at our Institution before 2017. The diagnosis had been made by WHO 2016 criteria based on peripheral blood counts, bone marrow (BM) cytology, cytogenetics, immunophenotyping and molecular analysis. Cases of APL with variant karyotypes had been excluded.
Cytogenetics had been performed in bone marrow cells by conventional G banding. Screening for nucleophosmin (NPM1), FLT3-ITD and FLT3-TKD mutations had been performed in genomic DNA with labeled primers and examined by fragment analysis. FLT3-ITD/wt allelic burden was calculated as the ratio of the area under the curve of mutant and wild-type alleles (FLT3-ITD/FLT3wt) [
11,
12,
13]. PML-RARA rearrangements were detected by qualitative RT-PCR [
13].
The study was approved by the local Ethics Committee (proc nr. 29661720.6.0000.5404).
2.2. Flow Cytometric Analysis
All BM samples had been collected in EDTA and processed within 24 h after collection using an eight-color platform with at least 2 tubes (
Supplementary Tables S1 and S2). At least 100,000 events had been acquired on a BD FACSCanto II equipment. Analysis was performed in the INFINICYT
TM 1.8 software (Cytognos SL, Salamanca, Spain). The mean fluorescence intensity (MFI) of each antigen tested was assessed for each case and compared with that of normal myeloblasts, obtained from cases where bone marrow MFC had been performed for the diagnostic work-up of peripheral blood cytopenias of unknown origin using the same flow protocol and where no aberrancies had been found. The SSC of the blasts was measured as the ratio between the value obtained for myeloid progenitors and that of the lymphocyte gate, used as internal control.
2.3. Statistical Analysis
Descriptive statistics was performed with non-parametric tests. In all AML cases, antigen expression was considered as positive when its MFI was above the 10% percentile of normal myeloid progenitors. The differences among groups were tested by Mann–Whitney and Kruskall–Wallis tests. All variables significantly different in the Mann–Whitney tests were analyzed in linear discriminant analyses [
14,
15]. Variables with a non-normal distribution entered this analysis after logarithmic transformation in order to approximate the data to a Gaussian distribution. In a first step, we ran a discriminant analysis individually. In a second step, we looked for the combination of variables that could best separate APL from non-APL patients in a multivariate discriminant analysis by testing all variables with significant differences between both groups simultaneously in a stepwise procedure. For this purpose, we used the stepwise function of the SPSS software which selects the best variables for an optimized discriminant model in a stepwise manner. The intrinsic stability was evaluated by the jackknife (leave-one-out) method [
14]. Winstat and SPSS 15 softwares were used for the calculations.
3. Results
Only 24 patients with APL and 56 with other AML subtypes fulfilled the inclusion criteria. Regarding the APL patients, the final diagnosis was made by both cytogenetics and molecular biology in 11 cases and only by molecular testing in 12. In only one patient, the diagnosis was based only on the karyotype. The demographic characteristics of the patients are shown in
Table 1. As expected, APL patients were significantly younger and had significantly lower peripheral leukocyte counts than non-APL cases. Among these last, the majority was of the intermediate risk group according to the ELN criteria.
Regarding phenotypic features (
Table 2), 17 of 24 (71%) cases of APL had an increased SSC compared to normal myeloblasts and all were negative for HLA-DR and CD34 by this criterium. APL had an increased expression of CD13 in 23 patients compared with normal myeloid progenitors (
Supplementary Figure S1).
Comparing APL with non-APL cases, expression of CD13, CD64 and CD33 was higher in the former, while that of CD117 showed similar values. CD34 and HLA-DR were expressed in only 18 (32%) and 25 (44%), respectively, of non-APL cases, and there was a positive correlation between expression of CD34 and HLA-DR (0.385; p: 0.007).
The results of the discriminant analysis are presented in
Table 3. In the multivariate analysis, the software suggested the “ratio SSC blast/SSC lymphocyte” and “MFI of HLA-DR” as the best model, which could discriminate APL from non-APL in 91.25% of cases and in 90% after the leave-one-out method (
Figure 1). If “SSC” was not included in the analysis, a model formed by HLA-DR, CD13 and CD64 was obtained, which discriminated 86.1% of the cases and 84.8% cases after the leave-one-out method. The only model containing CD34 was obtained if this marker was examined together with only CD13, discriminating 69.6% of the cases and 69.6% after the leave-one-out method.
Among the patients with APL, five presented the FLT3-ITD mutation and four FLT3-TKD. Both mutations were found in two patients. NPM1 mutation was not found in any of the cases.
The karyotype of non-APL cases could be retrieved in 39/56 cases. Among them, 15 had a normal karyotype, while t(8;21) was found in four, del(5) was found in four, +8 in two, alterations in chromosome 17 in two and several other abnormalities in 12 patients.
Mutation analysis could be retrieved in 48/56 of non-APL cases. In 30 patients, no mutation was found. One was found in seven cases: FLT3-ITD in four cases, NPM1 in two and FLT3-TKD in one patient. Eleven cases presented two mutations: FLT3-ITD and NPM1 were found in 10 cases and FLT3-TKD and NPM1 were seen in one case each. Patients without mutations had lower peripheral leukocyte counts and showed a higher expression of CD34 (p < 0.0001) and HLA-DR (p = 0.008).
4. Discussion
A rapid diagnostic screening for APL is essential for timely starting an appropriate treatment, thus preventing complications due to bleeding. This screening is usually done by cytology and immunophenotyping, or by immunocytochemistry with anti-PML staining [
16] or FISH [
17], which depends on the availability of the appropriate reagents.
Recently, Kárai et al. [
9] described a screening for APL using the multiparametric radar plot analysis of the Kaluza software for immunophenotyping. In this technique, a radar plot was constructed based on three different types of AMLs, including one FAB-M2, FAB-M4 and APL case, and searching for the parameters which best separated the three types of blasts. However, the authors admit that this technology would need a good standardization and the same antibodies and fluorochromes in all laboratories. Gupta et al. [
10] confirmed these results by comparing APL and NPM1
+ AML.
We aimed to develop a diagnostic algorithm based on the intensity of antigen expression and comparing the expression with that of normal myeloid progenitors processed with the same equipment and antibody combinations. Positivity for each marker was defined as above the 10% percentile of the values found for the controls. This approach is feasible in every laboratory, assuming that an internal standardization of instrument settings and antibody combination is made.
We could detect several antigens differentially expressed in APL, such as CD13, CD35, CD64 and HLA-DR. The expression of CD13 and CD64 was highly variable among all cases in both groups but tended to be higher in APL, as also described by Liu at al [
18]. Expression of CD35 was low in all cases. HLA-DR was negative in all our APL cases, but it was positive in only 44% of the non-APL cases. Concerning CD34, all our cases of APL showed a lower expression compared to that of normal myeloid progenitors. Among non-APL cases, expression of this antigen was also decreased compared to normal blasts in 67%. Interestingly, the CD34+ AMLs had a lower peripheral leukocyte count and showed no NPM1 and FLT3 mutation. The ratio “SSC blasts/SSC lymphocytes” was increased in 71% of APL cases, but only in 10% of non-APL cases. Although it has been often reported that SSC is increased in APL, this feature has not been highlighted as a discriminant between this type and other forms of AML [
19].
The best model in the discriminant analysis was based on the ratio between SSC blasts/SSC lymphocytes (high in APL) and a low MFI of HLA-DR, which separated APL from non-APL AML in most cases. MFI of CD34 entered in only one model tested.
Choosing a cut-point is often somehow arbitrary. Moreover, dichotomizing data initially collected as continuous data (positive or negative for an antigen expression) causes an important loss of information, reduces statistical test power, may increase diagnostic uncertainty, introduce bias and lead to oversimplification of the results [
20]. We overcame this problem by analyzing the MFI values, which allows us to extract more valuable information from the expression of the antigens assessed. Thus, expression of HLA-DR alone was able to correctly discriminate 77.5% of the patients. The MFI of CD34 was not useful for this screening, as 67% of the non-APL AMLs also had a low expression of this marker. The introduction of the variable “SSC”, which represents an indirect measure of the cytoplasmic characteristics, increased the screening potential to higher than 90%. The SSC ratio alone could correctly discriminate 85.2% of the patients. By combining SSC with MFI HLA-DR in the discriminant analysis, we achieved a high discrimination rate (91.25%). Since this value dropped only slightly after the leave-one out test, the discriminant analysis showed a good intrinsic stability of the mathematical model. We did not examine any microgranular/hypogranular variant of APL, but our approach could also be suitable to detect this form, as a model not considering SSC was able to discriminate 86.1% of the cases. This must be proven analyzing real cases.
In contrast to the work of Karai et al. [
9], we included 80 patients in the study, thus representing better the dispersion of the variables in the population of AML subtypes. In contrast to other methods, our technique applies variables which can be achieved by different flow cytometric protocols. In summary, our technique permits a simple, rapid and efficient screening of APL from other types of AML based on two standardized flow cytometric parameters.
Author Contributions
V.C.-S. collected and analyzed the data and participated in drafting the paper. K.P. participated in the study design, had confirmed all diagnoses, was responsible for all molecular tests and reviewed the manuscript. M.P. and G.D. were responsible for collection and analysis of all flow cytometric data and diagnosis of the cases. B.D. was responsible for the clinical data and treated all the patients. K.M. participated in the study design, was responsible for the statistical analysis and participated in drafting the paper. I.L.-M. was responsible for the study design, supervised the analysis of all the data, made the critical revision of the text and performed the submission. All authors have read and agreed to the published version of the manuscript.
Funding
The present research received no external funding. I.L.M. and K.M have research grants from the National Council of Technological and Scientific Development—CNPq (305110/2018-7 and 309910/2018-8, respectively).
Institutional Review Board Statement
Approved by the Ethics Committee of the University of Campinas.
Informed Consent Statement
Not applicable.
Data Availability Statement
Data available with authors on demand.
Conflicts of Interest
The authors declare no conflict of interest.
References
- Arber, D.A.; Orazi, A.; Hasserjian, R.; Thiele, J.; Borowitz, M.J.; Le Beau, M.M.; Bloomfield, C.D.; Cazzola, M.; Vardiman, J.W. The 2016 revision to the World Health Organization classification of myeloid neoplasms and acute leukemia. Blood 2016, 127, 2391–2405. [Google Scholar] [CrossRef] [PubMed]
- Bain, B.J.; Béné, M.C. Morphological and immunophenotypic clues to the WHO categories of acute myeloid leukemia. Acta Haematol. 2019, 141, 232–244. [Google Scholar] [CrossRef] [PubMed]
- Haferlach, T.; Schmidts, I. The power and potential of integrated diagnostics in acute myeloid leukemia. Brit. J. Haematol. 2020, 188, 36–48. [Google Scholar] [CrossRef] [PubMed]
- Rego, E.M.; Kim, H.T.; Ruiz-Arguelles, G.J.; Undurraga, M.S.; Uriarte, M.R.; Jacomo, R.H.; Gutiérrez-Aguirre, H.; Melo, R.A.M.; Bittencourt, R.; Pasquini, R.; et al. Improving acute promyelocytic leukemia (APL) outcome in developing countries through networking, results of the International Consortium on APL. Blood 2013, 121, 1935–1943. [Google Scholar] [CrossRef] [Green Version]
- Sanz, M.A.; Fenaux, P.; Tallman, M.S.; Estey, E.H.; Lowenberg, B.; Naoe, T.; Lengfelder, E.; Döhner, H.; Burnett, A.K.; Chen, S.; et al. Management of acute promyelocytic leukemia: Updated recommendations from an expert panel of the European LeukemiaNet. Blood 2019, 133, 1630–1643. [Google Scholar] [CrossRef] [Green Version]
- Mosleh, M.; Mehrpouri, M.; Ghaffari, S.; Saei, Z.; Agaeipoor, M.; Jadali, F.; Satlsar, E.S.; Gholampour, R. Report of a new six-panel flow cytometry marker for early differential diagnosis of APL from HLA-DR negative non-APL leukemia. Scand J. Clin. Lab. Investig. 2020, 80, 87–92. [Google Scholar] [CrossRef] [PubMed]
- Rosainz, M.J.A.; Nguyen, N.; Wahed, A.; Lelenwa, L.C.; Aakash, N.; Schaefer, K.; Rios, A.; Kanaan, Z.; Chen, L. Acute myeloid leukemia with mutated NPM1 mimics acute promyelocytic leukemia presentation. Int. J. Lab. Hematol. 2020, 43, 218–226. [Google Scholar] [CrossRef] [PubMed]
- De Mello, M.R.B.; Albuquerque, D.M.; Pereira-Cunha, F.G.; Albanez, K.B.; Pagnano, K.B.B.; Costa, F.F.; Metze, K.; Lorand-Metze, I. Molecular characteristics and chromatin texture features in acute promyelocytic leukemia. Diagn. Pathol. 2012, 7, 75. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Kárai BHabók, M.; Reményi, G.; Rejtö, L.; Ujfalusi, A.; Kappelmayer, J. A novel flow cytometric method for enhancing acute promyelocytic leukemia screening by multidimensional dot-plots. Ann. Hematol. 2019, 98, 1413–1420. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Gupta, M.; Jafari, K.; Rajab, A.; Wei, C.; Mazur, J.; Tierens, A.; Hyjek, E.; Musani, R.; Porwit, A. Radar plots facilitate differential diagnosis of acute promyelocytic leukemia and NPM1+ acute myeloid leukemia by flow cytometry. Cytom. B 2020. early view. [Google Scholar] [CrossRef]
- Van Dongen, J.J.; Macintyre, E.A.; Gabert, J.A.; Delabesse, E.; Rossi, V.; Saglio, G.; Gottardi, E.; Rambaldi, A.; Dotti, G.; Griesinger, F.; et al. Standardized RT-PCR analysis of fusion gene transcripts from chromosome aberrations in acute leukemia for detection of minimal residual disease. Report of the BIOMED-1 Concerted Action: Investigation of minimal residual disease in acute leukemia. Leukemia 1999, 13, 1901–1928. [Google Scholar] [CrossRef]
- Thiede, C.; Steudel, C.; Mohr, B.; Schaich, M.; Schäkel, U.; Platzbecker, U.; Wermke, M.; Bornhäuser, M.; Ritter, M.; Neubauer, A.; et al. Analysis of FLT3-activating mutations in 979 patients with acute myelogenous leukemia: Association with FAB subtypes and identification of subgroups with poor prognosis. Blood 2002, 99, 4326–4335. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Boissel, N.; Renneville, A.; Biggio, V.; Philippe, N.; Thomas, X.; Cayuela, J.-M.; Terre, C.; Tigaud, I.; Castaigne, S.; Raffoux, E.; et al. Prevalence, clinical profile, and prognosis of NPM mutations in AML with normal karyotype. Blood 2005, 106, 3618–3620. [Google Scholar] [CrossRef] [PubMed]
- Lorand-Metze, I.; Ribeiro, E.; Lima, C.S.P.; Batista, L.S.; Metze, K. Detection of hematopoietic maturation abnormalities by flow cytometry in myelodysplastic syndromes and its utility for the differential diagnosis with non-clonal disorders. Leuk. Res. 2007, 31, 147–155. [Google Scholar] [CrossRef] [PubMed]
- Irazusta, S.P.; Vassallo, J.; Magna, L.A.; Metze, K.; Trevisan, M. The value of PCNA and AgNOR staining in endoscopic biopsies of gastric mucosa. Pathol. Res. Pract. 1998, 194, 33–39. [Google Scholar] [CrossRef]
- Alayed, K.M.; Medeiros, L.J.; Phan, D.; Ojiaku, C.; Patel, J.; Yap, J.P.V.; McCord, Y.; Woods, J.S.; Konoplev, S.; Bueso-Ramos, C.E.; et al. Immunostaining for Rapid Diagnosis of Acute Promyelocytic Leukemia with the Tetramethylrhodamine-5-Isothiocyanate–Conjugated Anti–Promyelocytic Leukemia Monoclonal Antibody PG-M3. Arch. Pathol. Lab. Med. 2013, 137, 1669–1673. [Google Scholar] [CrossRef] [PubMed]
- Rashidi, A.; Fisher, S.I. FISH-negative, cytogenetically cryptic acute promyelocytic leukemia. Blood Cancer J. 2015, 5, e320. [Google Scholar] [CrossRef] [PubMed] [Green Version]
- Liu, M.; Weng, X.; Gong, S.; Chen, H.; Ding, J.; Guo, M.; Hu, X.; Wang, J.; Tang, G. Flow cytometric analysis of CD64 expression pattern and density in the diagnosis of acute promyelocytic leukemia: A multi-center study in Shanghai, China. Oncotarget 2017, 8, 80625–80637. [Google Scholar] [CrossRef] [PubMed]
- Silva, V.C.; Pagnano, K.B.; Duarte, G.B.O.; Pellegrini, M.D.B.; Duarte, B.K.L.; Metze, K.; Lorand-Metze, I. Developing a Simple Algorithm Based on Multiparameter Flow Cytometry for Fast Screening of Acute Promyelocytic Leukemia. Blood 2020, 136 (Suppl. 1), 25–26. [Google Scholar]
- Metze, K. Dichotomizing Continuous Prognostic Factors Can Cause Paradoxical Results in Survival Models. J. Am. Coll. Surg. 2011, 212, 132–134. [Google Scholar] [CrossRef] [PubMed]
| Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations. |
© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).