A distinct subgroup of AML resembling the APL immunophenotype is associated with DIC

Abstract

Background

The complexity of acute myeloid leukemia (AML) is increasingly recognized through the identification of distinct subgroups, including those with an APL-like immunophenotype characterized by the absence of CD34 and HLA-DR expression, which is widely recognized as a representative immunophenotype in acute promyelocytic leukemia (APL). This study sought to understand the clinical, molecular, and prognostic differences between AML patients with and without this phenotype.

Methods

This study retrospectively analysed 191 de novo non-M3 AML patients and identified 32 patients with the CD34⁻HLA-DR⁻ phenotype resembling APL-like immunophenotype, considered as the experimental group. Clinical data, including complete blood count, leukemic blasts, coagulation analysis DIC score, and OS, were collected, and immunophenotypic and molecular data were compared between this group and a control group of patients without this immunophenotype.

Results

Patients with the CD34⁻HLA-DR⁻ immunophenotype in the AML cohort had a significantly greater risk of developing disseminated intravascular coagulation (DIC) than did patients in the control group. Additionally, a lower rate of expression of immunophenotypic clusters of differentiation (CD markers) associated with poor prognosis was observed in the CD34⁻HLA-DR⁻ group. At the molecular level, an increased frequency of nucleophosmin 1 (NPM1) mutations and increased expression of the Wilms' tumor 1 (WT1) gene were noted in this subgroup. However, contrary to patients with an expected favourable prognosis, patients in the favourable risk group with the CD34⁻HLA-DR⁻ immunophenotype had significantly shorter overall survival than did patients in the control group.

Discussion

The findings highlight the patients exhibiting the CD34⁻HLA-DR⁻ immunophenotype as a unique AML subgroup with specific clinical and molecular traits, notably a predisposition to DIC, which affecting prognosis. This finding has implications for risk stratification and potential targeted therapies for AML management.

Introduction

AML is a diverse set of complicated diseases that develop in hematopoietic progenitors and have variable prognoses, with overall survival ranging from a few months to several years. In an endeavour to describe physiologically homogenous entities with clinical significance, the World Health Organization (WHO) categorization of AML integrates clinical, morphological, immunophenotypic, cytogenetic, and molecular criteria [1]. Researchers have categorized AML into many subgroups to elucidate its variety. Recently, investigations have been performed to identify the variables influencing AML patients' prognoses and enhance therapeutic approaches. The biological characteristics and outcomes of various subtypes of AML based on distinct immunophenotypes and particularly genetic alterations are less clear, even though refining different subgroups of AML improves our understanding of the underlying mechanism and recognition of favourable or adverse risk.

All-trans retinoic acid (ATRA)-based regimens have been used to successfully treat APL, a subtype of AML with unique morphologic, biological, and clinical characteristics. Due to the significant danger of developing DIC, APL has been considered an emergency [2]. Initial prompt evaluation of APL neurites relied on blast morphology and flow cytometric analysis. APL typically displays a distinct immunophenotype characterized by the absence of CD34 and HLA-DR expression [3]. However, a similar immunophenotypic pattern might be present in non-M3 AML cases that lack the t(15:17) chromosomal translocation, which is distinct from the classic M3 morphology [4–6]. Although uncommon, the “APL-like” immunophenotype pattern has been reported in a subset of patients with de novo AML with a mutation of the NPM1 gene [7]. However, this subtype of AML patients with a specific CD34 and HLA-DR double-negative immunophenotype has not received much attention.

Previous case reports and a few studies have shown that an CD34⁻HLA-DR⁻ immunophenotype is present in the majority of NPM1-mutated AML patients commutated with both TET2 and IDH2 mutations [8]. However, whether such patients can be distinguished from AML patients and whether their clinical features, immunophenotype spectrum, gene mutation profiles, etc., are different from those of other AML patients and ultimately affect prognosis are not yet clear. Therefore, our study included 32 non-M3 AML patients with CD34⁻HLA-DR⁻ immunophenotypes, focused on this type of specific AML patient, and analysed the differences between AML with CD34⁻HLA-DR⁻ and without this phenotype from the above aspects, providing a basis for an in-depth understanding of these patients and ultimately hope to improve the prognosis of this population.

Patients and methods

Patient cohort

The study cohort included 191 patients with de novo non-M3 AML collected from Zhongnan Hospital of Wuhan University. All AML patients whose bone marrow cytology, flow cytometric, cytogenetic, and next-generation sequencing data were collected as part of routine clinical care from April 2018 to October 2021 were reviewed. The aberrant loss of CD34 and HLA-DR expression was confirmed at the time of diagnosis for all patients whose specimens were used in the present study as the experimental group. AML patients with CD34⁺HLA-DR⁺, CD34⁺HLA-DR⁻ or CD34⁻HLA-DR⁺ phenotype are included in the control group.

Clinical evaluation

The medical records of the patients were studied to determine clinical manifestations, coagulation parameters (prothrombin time (PT), activated partial thromboplastin time (APTT), fibrinogen level, D-dimer level), chemotherapeutic regimen, and patient outcome. DIC was defined according to the International Society on Thrombosis and Haemostasis (ISTH) guidelines using a point scoring system that utilizes PT, fibrinogen levels (FIB), D-dimer levels, and platelet counts to diagnose DIC. A total score ≥ 5 indicated DIC [9]. The definition of genetic risk was based on the 2017 ELN risk stratification [10].

Immunophenotyping

Flow cytometric analysis was performed on a FACSCanto II flow cytometer (BD Biosciences, San Jose, CA), and the data were analysed using FACSDiva software (BD Biosciences). Immunophenotypic data were collected on the expression of the following markers for all patients: CD7, CD11b, CD19, CD13, CD14, CD15, CD33, CD34, CD38, CD56, CD64, CD117, MPO and HLA-DR.

Molecular analysis

Next-generation sequencing (NGS) was performed on an Illumina HiSeq X Ten (Illumina, San Diego, CA) using targeted amplicon sequencing. The sequencing panel utilized included 22 genes recurrently mutated in hematologic malignancies. Data, including CEBPA, DMNT3A, FLT3-ITD, NPM1, TP53, RUNX1, ASXL1, CBL, ETV6, IDH1, IDH2, JAK2, NRAS, SETBP1, SF3B1, SRSF2, TET2, PHF6, U2AF1, EZH2, ZRSR2 and C-kit gene mutations, relatively common in AML for mutations, were collected in this study. Newly diagnosed AML patients were offered the option to voluntarily undergo this genetic panel testing. Four common fusion genes (HOX11, CBFβ/MYH11, AML-ETO, MLL/AF9) were detected with multiplex nested RT‒PCR. The primer sequences, detection methods, and judgment results are described in the literature [11]. Real-time quantitative PCR was used to detect the expression of the WT1 gene, and multiplex nested RT‒PCR was performed with an ABI7500 sequence detector (Applied Biosystems, USA). The genetic data collected in this study were derived exclusively from patients at their initial diagnosis of AML. The above tests were completed at the Zhongnan Hospital of Wuhan University.

Statistical analysis

The percentages of clinical features were calculated from the number of patients with known information for that feature. The continuous variables were described as median with interquartile ranges (IQRs) and categorical variables were described as numbers with percentages. The Pearson χ² test or Fisher’s exact test and the 2-sample t test or the Mann‒Whitney U test were used to calculate the statistical significance for categorical and continuous variables, respectively. The log-rank test was used to analyse Kaplan‒Meier curves. The level for a statistically significant difference was set at 2-sided p < 0.05 for all tests. The SPSS 26.0 software package was used (IBM). The figures were generated using GraphPad Prism 9.3.0 (www.graphpad.com). (P < 0.05*, P < 0.01**, P < 0.001***).

Results

Non-M3 AML patients with CD34⁻HLA-DR⁻ immunophenotypes exhibit distinct phenotypic and genetic abnormalities

To understand the features of the 32 specific AML patients whose immunophenotype resembled APL, the general features of this subgroup of AML patients are summarized in Table 1, while the phenotype and genetic characteristics of AML patients with CD34⁻HLA-DR⁻ immunophenotypes were compared with other patients. Interestingly, CD34⁻HLA-DR⁻ AML patients exhibit distinct immunophenotypes and gene alterations. Patients with an CD34⁻HLA-DR⁻ phenotype were significantly more likely to harbour mutations in NPM1 (17/25, 68% vs. 24/129, 18.6%; P < 0.001) and HOX11 (8/29, 27.6% vs. 0/43, 0%; p < 0.001). Correspondingly, 41.5% (17/41) of NPM1 mutation-positive AML patients presented CD34⁻HLA-DR⁻ phenotype, while only 7.1% (8/113) of NPM1 mutation-negative patients presented this phenotype. Besides, in our study, we observed that 14 of 20 patients in the CD34⁻HLA-DR⁻ group with NPM1 positivity had analyzable chromosomal karyotypes, all of which were normal. No statistically significant difference in other common gene mutations was detected between the two groups (Fig. 1A). Apart from the genes mentioned in the figures, other genes in the 22-gene panel were excluded due to their relatively low mutation detection rates and the limited data available in the experimental group, which did not yield significant results.

Table 1.

Characters of AML with CD34-HLA-DR- phenotype (n = 32)

Characters of AML with CD34−HLA-DR− phenotype	Baseline
Age, years median (range)	60.5(41–86)
Gender (female/male)	18/14
Cytogenetics, n (%)
Normal	23(72)
Complex karyotype	1(3)
+ 4	1(3)
+ 21	1(3)
-Y	2(6)
Unknown	4(13)
Risk category (n = 26)
Favorable	14
Intermediate	9
Adverse	3
Outcomes following standard chemotherapy (n = 20)
Complete remission rate (%)	60
Relapse rate (%)	30
Overall survival, months, median	9.4

Fig. 1.

Comparisons of the genetic and phenotypic characteristics of AML patients with and without the CD34⁻HLA-DR⁻ immunophenotype. A The incidences of several common gene mutations and fusion genes, including CEBPA, DMNT3A, FLT3-ITD, NPM1, TP53, c-kit, MLL/AF9, CBFβ/MYH11, AML/ETO and HOX11, in the CD34⁻HLA-DR⁻ and the control group of AML patients are presented. The rate of NPM1 mutation and incidence of HOX11 mutations in the CD34⁻HLA-DR⁻ group were significantly greater than those in the control group. B Immunophenotypic molecular positivity results for the detection of AML patients were collected and are illustrated as an CD34⁻HLA-DR⁻ group versus the control group. The percentages of CD34⁻HLA-DR⁻ patients who were positive for CD64, CD15, CD11b, and CD7 were significantly lower than those of other patients. C Flow cytometry analysis of CD34⁻HLA-DR⁻ and non-CD34⁻HLA-DR⁻ patients revealed that the expression of CD7, CD64, CD15, and CD11b was mostly negative in CD34-HLA-DR- patients

Regarding gene abnormalities, Among the AML patients with CD34-HLA-DR- phenotype, the percentages of patients with AML blast-positive CD64 (14/32, 43.8% vs. 104/159, 65.4%), CD15 (5/32, 15.6% vs. 72/159, 45.3%), and CD11b (2/32, 6.3% vs. 35/1159, 22%) expression were significantly lower than those in the control group. More surprisingly, CD7 was not expressed in any CD34-HLA-DR- patients, while 42.8% (68/159) of other patients tested positive for CD7 (Fig. 1 B, C).

CD34⁻HLA-DR⁻ AML patients are more prone to coagulation abnormalities and have a greater risk of developing DIC

With respect to the general clinical manifestations of CD34⁻HLA-DR⁻ patients, we investigated three blood cell traits, namely, white blood cell (WBC) count, haemoglobin, platelet count, presence of leukemic blasts in the bone marrow and expression of WT1 at diagnosis. In our study, although the level of haemoglobin(72.1 [54.4, 88] g/L vs. 81 [59.75, 103.8] g/L), did not differ between these two groups, patients with the CD34⁻HLA-DR⁻ phenotype had a significantly greater disease burden, with a median WBC of 36.07 (6.18, 102.2)*10^9/L, a median BM blast percentage of 83% (57.83, 90.38), compared with 10.82 (3.11, 42.88)*10^9/L for WBCs and 55% (36, 71) for blasts in the control group (p < 0.05 and p < 0.0001, respectively; Fig. 2A, B). In addition, higher WT1 expression was detected in the CD34⁻HLA-DR⁻ group than in the control group (21% [6, 43] vs. 39% [25.75, 64.5], p < 0.001, Fig. 2B).

Fig. 2.

Clinical characteristics of AML patients with an CD34⁻HLA-DR⁻ immunophenotype compared with those control patients. A The white blood cell, haemoglobin, and platelet counts in the peripheral blood of patients are presented. The median WBC count in CD34⁻HLA-DR⁻ patients was significantly elevated. B Results from the coagulation analysis are shown for PT, APTT, FIB, and D-dimer. The median FIB and D-dimer were greater in CD34⁻HLA-DR⁻ patients than in non- CD34⁻HLA-DR⁻ patients. C The percentage of blast cells and the expression of WT1 in the bone marrow were significantly greater in the CD34⁻HLA-DR⁻ group than in the control group. D The incidence of DIC development in CD34⁻HLA-DR⁻ patients was markedly greater than that in the control patients

To test the hypothesis that CD34⁻HLA-DR⁻ patients have default blood clotting functions that are equivalent to those of APL patients, coagulation analysis and DIC calculations were performed. No statistically significant differences were found in the platelet count (48 [28.5, 104.5]*10^9/L vs. 38 [24, 84.5]*10^9/L), PT (13.5 [12.3, 15.15] s vs. 13.6 [12.3, 14.7] s), or APTT (30.16 [27.75, 32.26] s vs. 29.5 [27.2, 32.1] s) between patients with and without CD34⁻HLA-DR⁻ phenotypes (PLT, p = 0.616; PT, p = 0.521; APTT, p = 0.689; Fig. 2A, C). However, the median FIB concentration in the CD34⁻HLA-DR⁻ group was 3.31 (2.38, 4.39) g/L, and that in the control group was 3.98 (3.42, 4.52) g/L. This result is significant at the p = 0.05 level. Notably, 43.75% (14/32) of our patients with CD34⁻HLA-DR⁻ phenotype experienced DIC, while 7.5% (12/159) of our patients without CD34⁻HLA-DR⁻ phenotype experienced DIC (Fig. 2D).

Predictive factors that might affect the presence of the CD34⁻HLA-DR⁻ immunophenotype

Considering the uniqueness of CD34⁻HLA-DR⁻ patients and their clinical features and differential expression of genes, a model to predict factors that might affect the CD34⁻HLA-DR⁻ phenotype was developed by utilizing binary logistic regression. Elements, including NPM1, bone marrow leukemic blasts, and DIC, were eventually applied in this model. All factors were previously shown to be significantly different between CD34⁻HLA-DR⁻ and the control patients (p < 0.05). The results showed that NPM1 mutation (p < 0.0001, OR = 10.177), BM leukemic blast (p = 0.003 < 0.05, OR = 1.044), and DIC (p = 0.005 < 0.05, OR = 6.238) were included in the model to be statistically significant and should be considered risk factors. In particular, the incidence of the CD34⁻HLA-DR⁻ phenotype was approximately 10 times greater in patients with NPM1 mutations than in patients without mutations. The DIC is also a factor that cannot be ignored. Non-M3 AML patients with DIC were 6.238 times more likely to present with an CD34⁻HLA-DR⁻ immunophenotype than patients without DIC. The prediction model is computationally efficient and achieves a high accuracy of 87.4% (Table 2).

Table 2.

A model established by binary logistic regression predicting factors that might affect the presence of the CD34⁻HLA-DR⁻ immunophenotype

Variablesa	P values	OR (95% CI)
NPM1 mutation	< 0.0001***	10.177(3.175–32.620)
BM Blast	0.003**	1.044(1.015–1.074)
DIC	0.005**	6.238(1.731–22.481)

^aR square for this model is 0.433, and the prediction model achieves an accuracy of 87.4%

Patients with CD34⁻HLA-DR⁻ immunophenotype and a favourable risk classification had shorter survival

The results in this chapter indicate that CD34⁻HLA-DR⁻ patients are a special subgroup of AML patients. Therefore, the next chapter discusses the survival of CD34⁻HLA-DR⁻ patients. Unexpectedly, there was no significant difference between the CD34⁻HLA-DR⁻ group and the control group in terms of relapse-free survival (16.2 months vs. 23.3 months) or overall survival (7.5 months vs. 6 months) (Fig. 3A, B). Considering that the NPM1 mutation is closely associated with this subtype and that NPM1 is recognized as a favourable risk factor, we evaluated the overall survival of patients with and without the CD34⁻HLA-DR⁻ phenotype in all favourable prognosis group. However, the survival times of patients with the CD34⁻HLA-DR⁻ phenotype were 4.7 and 36.4 months, respectively, for patients without the CD34⁻HLA-DR⁻ phenotype in the favourable risk group, based on the 2017 ELN risk stratification, and factors included in the ELN algorithm were collected at diagnosis if available (Fig. 3C).

Fig. 3.

Survival analysis of AML patients with and without the CD34⁻HLA-DR⁻ phenotype. A and B The median relapse-free survival (RFS) and overall survival (OS) of patients with CD34⁻HLA-DR⁻ phenotype were 16.2 and 7.5 months, respectively, showing no significant differences in RFS or OS (P > 0.05). C The median OS of patients with the CD34⁻HLA-DR⁻ phenotype in the favourable risk group was 4.7 months, which was significantly shorter than that of patients with the non-CD34⁻HLA-DR⁻ phenotype in the same risk stratification

Discussion

While much recent work has been done to extensively characterize the clinical specificity of various subtypes of AML, relatively little is known about whether certain clinical features are correlated with specific immunophenotypic characteristics and molecular alterations. Here, we first focused on a subgroup of non-M3 AML patients who exhibited an APL-like immunophenotype and who were negative for both CD34 and HLA-DR. In clinical practice, there is another category of acute leukemia patients whose blasts presented with large azurophilic inclusions and prominent Auer rods resembling APL. Among the 32 CD34⁻HLA-DR⁻ AML patients in our study group, only 5 (15.6%) exhibited APL-like cell morphology in their blasts. This finding suggests that not all non-M3 AML patients with CD34⁻ and HLA-DR⁻ immunophenotype concurrently display changes in cell morphology characteristic of APL. Notably, the results showed that this cohort clinically exhibited significantly greater WBC counts, elevated bone marrow leukemic blasts, and thrombotic abnormalities, such as decreased FIB and extremely increased D-dimer, and was even more prone to developing DIC, which may have contributed to the inferior outcomes observed in this group. Surprisingly, their clinical features are similar to those of APL patients. The disease that is most frequently accompanied by DIC is APL; nearly 70% of newly diagnosed APL patients develop DIC. Approximately 17% of non-APL AML patients develop DIC at the time of diagnosis [12]. Concomitant DIC exacerbates the prognosis of patients with hematological malignancies. The development of DIC in non-APL AML patients is associated with increased WBC counts, increased D-dimer levels, and fibrin deposition [13].

The correlations of APL-like immunophenotypic characteristics and molecular alterations indicated that CD34⁻HLA-DR⁻ leukemic blasts in these patients had disparate clusters of differentiation co-expressions, such as lower percentages of CD15-, CD11b-, CD64-, and CD7-positive cells, which are coincidentally associated with poor prognosis in patients with AML [14]. However, further studies will be required to determine whether the co-expression of these clusters of differentiation markers predicts the prognosis of CD34⁻HLA-DR⁻ patients.

The mutation of NPM1 is one of the most common recurrent driver mutations in AML and occurs in approximately 30% of patients [15]. AML with mutated NPM1 is now recognized as a distinct entity in the World Health Organization classification and leads to a good prognosis [16]. Here, we showed that 68% of CD34⁻HLA-DR⁻-phenotype AML patients harboured NPM1 mutations, supporting similar findings reported in previous studies [7, 17]. The absence of CD34 and HLA-DR in these patients indicates that the leukemic cells are arrested at a slightly later stage of myeloid differentiation, closer to the promyelocyte stage than the blast stage [18]. APL is effectively treated with ATRA and arsenic trioxide (ATO). Interestingly, 2 recent studies have shown that treatment of NPM1-mutated AML cells with ATRA and ATO induces selective degradation of mutant NPM1, causing differentiation and/or apoptosis in leukemic cells [19, 20].

The HOX11 gene is rearranged in 3–5% of pediatric and up to 30% of adult patients with T-ALL by recurrent t (10;14) (q24; q11); however, its role in AML is less known. Previous in vitro studies have shown that HOX11 immortalizes hematopoietic precursors, transforms bone marrow cells, and inhibits their differentiation. HOX11-induced leukemias represent a distinct oncogenic group with specific genetic alterations, including mutations in the WT1 [21] and PHF6 [22] tumour suppressor genes. In our study, the CD34⁻HLA-DR⁻ cohort indeed exhibited an increased percentage of patients with HOX11 gene fusion and elevated WT1 mutation. However, until recently, very little was known about the specific mechanisms that mediate leukemia generation downstream of HOX11.

Another study revealed that HOX11 may lead to abnormal expression of target genes in leukemia, thereby affecting the sensitivity of tumour cells to chemotherapy drugs. There were only 8 AML patients with positive expression of the HOX11 gene in the CD34⁻HLA-DR⁻ group, so the results are not necessarily representative of this group. It is necessary to continue to increase the sample size and cooperate with multiple centres to further clarify the mechanism by which the HOX11 gene participates in regulating biological behaviour and affects the treatment and prognosis of AML patients, even those with specific subtypes of AML.

In summary, these characteristic genetic alterations, distinct immunophenotypes, and clinical characteristics uniquely distinguish CD34⁻HLA-DR⁻ phenotype AML from other AML classes. However, the overall survival and relapse-free survival of patients in the CD34⁻HLA-DR⁻ phenotype group were not significantly different from the control AML group. Although elevated WBC counts, leukemic blasts, WT1 expression, and DIC probability in the CD34⁻HLA-DR⁻ cohort are associated with inferior outcomes, NPM1 mutations were detected in 68% of these patients, and mutations of this gene in AML patients have been proven to improve patient outcomes [15]. These confounding factors may have led to the lack of a significant difference in patient outcomes between the CD34⁻HLA-DR⁻ phenotype group and the control group. However, 53.8% (14/26 patients) of CD34⁻HLA-DR⁻ patients are categorized into favourable risk groups, and they exhibit a shorter overall survival than other patients at favourable risk.

Together with these findings, our results suggest that although AML patients with an CD34⁻HLA-DR⁻ immunophenotype are mostly classified into favourable risk groups at diagnosis, their outcomes are unsatisfactory. The characteristic clinical features of APL patients require extra attention and could be exploited in the development of future treatment strategies. Future studies are needed to fully elucidate the mechanisms that underlie these distinct subgroups of AML and to understand how the presence of different gene abnormalities may affect prognosis and response to therapy.

Authors’ contributions

PL and FZ designed the study. Data curation was done by LL. Analysis of data is done by PL. The original manuscript was written by PL and LL. Supervision was done by FZ. All authors agreed to be accountable for all aspects of the work and approved the submitted version.

Funding

2019 Science and Technology Innovation Cultivation Fund of Zhongnan Hospital of Wuhan University ZNLH201902.

Data availability

The data and material are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Clinical data and specimens were acquired with the informed consent of patients, and the study was approved by the Institutional Ethics Committee Board of Zhongnan Hospital of Wuhan University, China.

Consent for publication

Given that this manuscript does not contain any images that divulge patients' personal information, the statement regarding patients' consent for release is not applicable to this study.

Competing interests

The authors declare no competing interests.