Clinical, immunophenotypic, and genomic findings of acute myeloid leukemia with RAM immunophenotype: Comparison with other CD56‐positive acute leukemias

Abstract

Background

Acute myeloid leukemia (AML) with RAM immunophenotype is a newly recognized high‐risk AML immunophenotypic subcategory characterized by blasts with bright expression of CD56 and weak to absent expression of CD45, HLA‐DR, and CD38, as first described by the Children's Oncology Group (COG). The relationship between AML‐RAM and other CD56‐positive acute leukemias is unclear. The goal of this study is to characterize the clinicopathological characteristics of AML with RAM phenotype and compare them with other CD56 co‐expressing acute leukemias.

Methods

From a multi‐institutional search, we identified a total of 160 CD56+ acute leukemia cases, including AML‐RAM (n = 28), CD56+ acute undifferentiated leukemia (AUL) (n = 11), CD56+ T‐lymphoblastic leukemia (n = 39), and CD56+ AML (n = 81). We compared the clinical and pathologic findings of these groups.

Results

AML‐RAM patients were significantly younger and presented with significantly higher platelet and white blood cell counts and bone marrow (BM) blast percentages when compared to AUL (p > 0.05) and had higher median BM blast percentages than T‐ALL and CD56+ AML groups (both p < 0.05). Flow cytometry showed significantly brighter expression of CD56 on blasts as compared to other CD56+ AML cases, partial CD34 expression compared to AUL, and AML, weak‐to‐absent CD38 expression compared to all groups, and absent HLA‐DR and terminal deoxynucleotidyl transferase as compared to AUL and T‐ALL (all p < 0.05). The frequency of abnormal karyotypes was significantly higher among RAM when compared to all groups (p < 0.05). Next‐generation sequencing profiles differed among the leukemia groups, with significant enrichment of CBFA2T3::GLIS2 fusions (p < 0.05) and TP53 mutations (p < 0.05) in RAM cases compared to other AML control groups, and U2AF1 (p < 0.05), serine and arginine‐rich splicing factor 2 (p < 0.05), and BCL6 co‐repressor (p < 0.05) mutations compared to AUL. Clinical outcome analysis demonstrated significantly lower 3‐year overall survival of the RAM subgroup (36 months) compared to control groups (p = 0.002).

Conclusion

We find that AML with RAM phenotype occurs primarily in younger ages, with distinct clinicopathological, immunophenotypic, and mutational presentations, and worse prognosis. This diagnosis should be considered in the clinical differential diagnosis of CD56‐positive acute leukemias.

Abbreviations

• AMKL
• acute megakaryoblastic leukemia
• AML
• acute myeloid leukemia
• AML-NOS
• AML-not otherwise specified
• AML-RAM
• AML with RAM phenotype
• AUL
• acute undifferentiated leukemia
• BCOR
• BCL6 co-repressor
• BM
• bone marrow
• CI
• confidence interval
• COG
• Children's Oncology Group
• EFS
• event-free survival
• FISH
• fluorescence in situ hybridization
• FLT3-ITD
• FLT3 internal tandem duplication
• HR
• hazard ratio
• ICC
• International Consensus Classification
• NGS
• next-generation sequencing
• NK-LL
• NK lymphoblastic leukemia
• NPM1
• nucleophosmin-1
• OS
• overall survival
• SRSF2
• serine and arginine-rich splicing factor 2
• TDT
• terminal deoxynucleotidyl transferase
• WBC
• white blood cell count

INTRODUCTION

Acute myeloid leukemia (AML) encompasses a diverse group of aggressive neoplastic disorders and is characterized by clonal proliferation of myeloblasts. AML classification has been categorized based on morphology and immunophenotype in the past; however, currently, AML classification integrates clinical, molecular/genetic, and pathologic criteria [1, 2].

Recurrent genetic abnormalities in specific AML entities define distinct clinicopathologic and prognostic groups [3]. Most recently, the proposed 5th edition of the World Health Organization (WHO) Classification of Haematolymphoid Tumours recognized categories of AML with genetic abnormalities and those defined by differentiation [4]. The International Consensus Classification of Acute Leukemia (ICC) unifies the diagnostic approach to AML based on genetics while maintaining the subcategory of AML-NOS (not otherwise specified) [5]. One of the differences between the two AML classification systems in AML is a defined blast percentage requirement (≥ 10%) for AML with recurrent genetic abnormalities in the ICC 2022. Both classifications now include the rare (∼1% of AML) AML with CBFA2T3::GLIS2 fusion, resulting from a cryptic chromosome 16 inversion (inv(16)(p13.3q24.3)), that typically occurs during infancy and is often associated with non-Down syndrome related acute megakaryoblastic leukemia (AMKL) and RAM phenotype, in contrast to the classic inversion chromosome 16 (CBFB::MYH11 [inv(16)/t(16;16)] rearrangements) which represents a separate, AML entity with a favorable prognosis [4, 5]. In both classifications, AML with RUNX1T3(CBFA2T3)::GLIS2 is a subtype of AML with other defined genetic alterations [4, 5].

The risk stratification of leukemias is mainly based on specific molecular studies and cytogenetic abnormalities [4, 5]. Some subtypes of AML with recurrent molecular abnormalities, such as AML with RUNX1::RUNXT1 and PML::RARA fusions, have distinct immunophenotypic features that may allow the prediction of these abnormalities. The RAM immunophenotype is a newly recognized high-risk AML immunophenotypic subcategory associated with the CBFA2T3::GLIS2 fusion [6, 7, 11, 12].

AML with the RAM immunophenotype (referred to as AML-RAM) was first described in the 2016 Children's Oncology Group (COG) (named after a patient's initials) by Eidenschink Brodersen et al. and characterized by blasts with bright expression of CD56 and weak to absent expression of CD45, HLA-DR, and CD38 [6]. Subsequent COG studies have suggested a high frequency of AMKL among AML-RAM [7, 14]; approximately 60% of AML-RAM (with exceedingly bright CD56) carry the CBFA2T3::GLIS2 fusion, and AML-RAM patients have a worse prognosis [1]. Other recent studies demonstrated a similar phenotype-genotype association: the CBFA2T2::GLIS2 fusion presented in 30%–70% of AML-RAM [7]. These studies further suggested that the CBFA2T2::GLIS2 fusion may be the driver lesion responsible for the poor prognosis but there was no specific survival data regarding the fusion positive versus fusion negative AML-RAM patients.

In this study, we aim to characterize the clinicopathological and genetic characteristics of AML-RAM, particularly assessing any differences between the CBFA2T2::GLIS2 fusion-positive and negative patients and compare with other CD56 expressing acute leukemias including CD56+AML.

METHODS

Diagnostic criteria and patients

Patients were collected from multiple academic institutions (including University of Texas Southwestern Medical Center, the Hospital of the University of Pennsylvania, The University of Texas MD Anderson Cancer Center, St. Jude Children's Research Hospital, Tennessee, Yale School of Medicine, Boston Children's Hospital, Stanford Medical Center, Brigham and Women's Hospital/Dana-Farber Cancer Institute, Massachusetts General Hospital, Seattle Children's Hospital, University of North Carolina, University of Minnesota, University of Utah, University of Pittsburgh, and Weill Cornell Medical Center) using a search for AML with RAM phenotype as well as CD56-positive control cases of AUL, T-ALL, and AML as previously published by Weinberg et al. [9].

The criteria for diagnosing AML with RAM phenotype included flow cytometry findings of having very strong CD56 expression along with weak to absent CD45, HLA-DR, and CD38 expression [5]. Clinical, laboratory, molecular, cytogenetic, and bone marrow (BM) findings with corresponding flow cytometry and ancillary studies were reviewed to confirm that each case met the new ICC 2022 and proposed WHO 5th ed for acute leukemias in the CD56 positive comparison groups. The study was approved by the Institutional Review Boards (IRBs) at all mentioned institutions.

Flow cytometric analysis

Bone marrow aspirates, peripheral blood, and lymph node samples were submitted for multidimensional flow cytometry analysis at each institution. A comprehensive flow cytometric workup was performed at all institutions. The total markers assessed included CD56, CD45, MPO, CD34, CD117, CD13, CD33, CD15, surface CD3, cytoplasmic CD3, CD2, CD4, CD7, CD5, CD1a, CD19, CD20, CD10, CD123, HLA-DR, CD22, CD79a, CD11b, CD64, terminal deoxynucleotidyl transferase (TDT), CD38, CD8. CD16, CD41a, CD57, CD303, and CD36, were assessed in some, but not all institutions. Expression of the surface antigens as well as the physical parameters forward scatter and log SSC were determined for the identified leukemic cell population. As recommended in Bethesda International Consensus recommendations on immunophenotypic analysis, antigen distribution was considered “negative” for antigens not expressed, “positive” for antigens expressed, or “partially expressed” for antigens that are expressed in a subset of the population in question. As recommended, descriptions of antibody fluorescence intensity are “dim” for a uniformly positive population with lower mean fluorescence intensity than a positive normal cell population, “bright” for a uniformly positive population with higher intensity than the control, and “heterogeneous” for antigens that show variable expression or a spectrum of expression intensity, as opposed to a tight uniform cluster of events [20].

Chromosomal and fluorescence in situ hybridization testing

Metaphase analysis, as well as fluorescence in situ hybridization (FISH), were performed on BM samples to detect chromosomal and common targeted genetic abnormalities (including, but not limited to, probes for CBFA2T3::GLIS2, RUNX1::RUNX1T1, PML::RARA, KMT2A (MLL), CBFB::MYH11, +8, 13q−, 5q−/−5, 7q−/−7, or 20q−), respectively. Similar protocols were used by all the participating institutions.

Molecular studies

Genomic DNA was extracted from diagnostic BM and/or peripheral blood specimens and molecular studies were performed using RNA-based fusion panels, RNA sequencing to detect mutations, and next-generation sequencing (NGS) assays with different panels at each institution. Both methods were used to detect common mutations diagnosed in hematolymphoid malignancies. Specific test procedures were different among the institutions, but all methods included AML panels to screen for recurrent mutations: RUNX1, IDH1, IDH2, nucleophosmin-1 (NPM1), NOTCH1, ETV6, TP53, FLT3, JAK2, NRAS, and TET2. The following mutations were present in most but not all panels: CBFA2T3, GLIS2, ASXL1, BRAF, CBL, DNMT3A, EZH2, SETBP1, KMT2A, PHF6, SF3B1, U2AF1, WT1, PTPN11, CEBPA, EZH2, RB1, and BCL6 co-repressor (BCOR). NGS panels were performed in 15 out of 28 AML-RAM cases. The number of AML-RAM cases tested for CBFA2T3::GLIS2 fusion by NGS was 16 out of 28 (4 were detected by FISH and 12 were detected by NGS).

Statistical analysis

Statistical analysis was performed using GraphPad Prism (GraphPad Software) with significance set at a p-value < 0.05 (two-sided). Fisher exact test and Mann-Whitney test were used to compare categorical and numerical variables respectively. Overall survival (OS) was determined from the date of initial diagnosis to death or the last follow-up date, whichever occurs first. Event-free survival (EFS) was calculated from the date of initial diagnosis to the relapse date. Survival probability was analyzed by the Kaplan–Meier method, with differences compared by the log-rank test. Early death or failure to enter remission was considered an event at zero time.

RESULTS

We identified and analyzed a total of 160 CD56+ acute leukemia cases, including 28 AML-RAM, 12 CD56+ acute undifferentiated leukemia (AUL), 39 CD56+ T-lymphoblastic leukemia, and 81 CD56+ AML including 14 cases of AML with t(8;21)(q22;q22.1)/RUNX1::RUNX1T1, one case of acute promyelocytic leukemia with t(15;17)(q24.1;q21.2)/PML::RARA, three cases of AML with t(9;11)(p21.3;q23.3)/MLLT3::KMT2A, one case of AML with inv(3)(q21.3q26.2) or t(3;3)(q21.3;q26.2)/GATA2, MECOM(EVI1), three cases of AML with BCR::ABL1 fusion, 17 cases of AML with mutated NPM1, one case of AML with in-frame bZIP CEBPA mutations, three cases of AML with mutated TP53, 13 cases of AML with myelodysplasia-related cytogenetic abnormalities, three cases of AML with myelodysplasia-related gene mutations, and 23 cases of AML-NOS. One case was classified as myeloid proliferation associated with Down syndrome.

The details of demographic, clinical, and laboratory characteristics are summarized in Table 1.

TABLE 1 Comparison of clinical presenting characteristics of acute myeloid leukemia (AML)-RAM, acute undifferentiated leukemia (AUL), T-ALL, and CD56+ AML patients. Values with asterisks are significant at p < 0.05.

Patient characteristics	RAM	AUL	T-ALL	CD56+ AML
Total	28	11	39	81
Media Age; years; (range)	2 (1.1-39)	29 (13-56) *	24 (11-54) *	56 (28.5-66.5) *
Median WBC x109/L	11.27	5.5 *	11.6	19.9 *
Median platelet count x109/L	80	51 *	89.5	55 *
Median hemoglobin, g/dL	9.1	9.8	9.5	8.5
BM cellularity, %	90	85	90	90
Median BM blasts, %	90	87	77 *	70 *

AML with RAM phenotype patient cases

AML-RAM patients (N = 28) presented at a significantly younger age of 2 years when compared to the non-RAM CD56 + AML cases (Median age AUL: 30 years, T-ALL: 24 years, CD56+ AML: 56.5 years; all p < 0.05). Also, there were no patients with age < 10 years with CD56 expression who were anything other than RAM. The male-to-female ratio was 1.3:1. None of the patients included in the study had a previous history of malignancy. The most common presentations included fever, loss of appetite, weight loss, and bony pain. Two (7%) AML-RAM patients were presented with central nervous system manifestations at the time of diagnosis, including headaches and seizures in one patient and a paraspinal mass in the other patient. Skin involvement manifested by facial skin involvement was present in one AML-RAM patient (4%).

Peripheral blood testing of RAM patients showed a median white blood cell (WBC) count of 11.27 × 10³/µL, median hemoglobin 9.1 g/dL, and median platelet count of 80 × 10³/µL. Bone marrow evaluation in these patients showed a median blast count of 90% and cellularity of 90%. In all 28 RAM cases, the blasts expressed bright CD56, had weak CD34 and CD38, and lacked CD45 and HLA-DR as seen in figure 4. Additional performed markers included CD41a and CD61 in seven RAM cases. Two out of eleven (18.2%) CBFA2T3::GLIS2 fusion-positive RAM cases in our study had a non-Down Syndrome megakaryoblastic leukemia immunophenotype.

Metaphase analysis showed that 71% of RAM cases had a complex karyotype. The most common pathogenic mutations detected in RAM patients were NPM1 (14.2%), FLT3 internal tandem duplication (FLT3-ITD) (10.7%), serine and arginine-rich splicing factor 2 (SRSF2) (7.2%), and BCOR loss (3.6%). CBFA2T3::GLIS2 fusion was detected in (11/16) 32% of RAM patients.

All AML-RAM patients received AML-type intensive chemotherapy, and 14/28 patients (50%) underwent allogeneic hematopoietic stem cell transplantation. The median overall follow-up time was 58.3 months (range 0.1–86.6 months). Twelve patients achieved complete remission. Eight out of 28 patients relapsed after AML chemotherapy, while 5 patients never achieved remission and died within 5 months of the initial diagnosis (Figure 1).

[IMAGE OMITTED. SEE PDF]

AML-RAM with CBFA2T3::GLIS2 fusion showed a trend but no significant difference in OS and EFS as compared to fusion-negative patients (albeit lack of significance, p = 0.05) (Figure 2). AML-RAM patients showed a significantly higher frequency of extramedullary leukemia as compared to other CD56-positive leukemias (10.72% vs. 1.23%, p < 0.05). A detailed comparison between CBFA2T3::GLIS2 fusion positive AML-RAM bearing cases (n = 11) and CBFA2T3::GLIS2 fusion negative AML-RAM cases (n = 5) is shown in Tables S1 and S2.

[IMAGE OMITTED. SEE PDF]

AML-RAM compared to AUL patients

Patients classified as AUL included five patients with NK lymphoblastic leukemia (NK-LL) and six patients with CD56+ AUL. At diagnosis, patients with AML-RAM presented at a significantly younger age (p = 0.01) as compared with CD56+ AUL and NK cases, had higher WBCs (p < 0.05) and platelets (p = 0.05) but had no differences in hemoglobin, BM cellularity or percent blasts (Table 1). Immunophenotyping showed blasts had a brighter expression of CD56 in AML-RAM versus AUL cases (p < 0.05). Additionally, CD34, CD38, and TDT were significantly brighter in AUL cases (p < 0.05). Myeloid markers were overall less frequently expressed in AUL cases as compared to AML-RAM (p < 0.05). Cytogenetic analysis showed an abnormal karyotype in 45% of AUL cases as compared to AML-RAM (71%, p < 0.05). The most common alterations in AUL were RUNX1::RUNX1T1 (27%) and BCOR (18%). Both groups received AML chemotherapy regimens. The median follow-up time for AUL patients was 65.9 months (range 0.33–192). There was no significant difference in the OS rate between AML-RAM and AUL patients (Figure 3, hazard ratio [HR] 0.6; confidence interval [CI]: 0.21–1.65; p > 0.05). However, EFS was significantly better among the AUL compared to the AML-RAM group (HR 0.18; CI: 0.05–0.64; p < 0.05) as shown in Figure 3.

[IMAGE OMITTED. SEE PDF]

AML-RAM compared to CD56+ T-ALL patients

Patients with CD56+ T-ALL were presented with an older median age of 24 years (range 11–54 years) than AML-RAM cases (p < 0.05). As compared with AML-RAM patients, T-ALL patients presented with similar WBC count, hemoglobin, platelet count, and BM cellularity (all p  >  0.05) but had a significantly lower median blast percentage (77%, p < 0.05). AML-RAM cases showed a significantly brighter CD56 expression and dimmer CD38 and TDT expression as compared with CD56 positive T-ALL (all p < 0.05) shown in table 3. Abnormal karyotypes were detected in 60% of T-ALL cases and the most detected mutations included TP53 (5%), SRSF2 (5%), FLT3-ITD (3%), and BCOR loss (3%). All T-ALL patients were treated with T-ALL regimens. The median follow-up time for T-ALL patients was 34.4 months (range 1–132) and a median relapse-free time was 15 months (range 0–132) as seen in table 2. There was no significant difference in OS (HR 0.86; CI: 0.39–1.87; p > 0.05), or EFS (HR 0.40; CI: 0.14–1.12; p >  0.05) between AML-RAM and T-ALL patients, (Figure 3).

TABLE 2 Comparison of cytogenetics and molecular findings of acute myeloid leukemia (AML)-RAM, acute undifferentiated leukemia (AUL), T-ALL, and CD56+ AML patients. Values with asterisks are significant at p < 0.05.

Abnormal karyotype (%N)	RAM	AUL	T-ALL	CD56+ AML
Abnormal karyotype	71 (20/18)	45 (5/11)*	60 (23/39)	68 (55/81)
Molecular Abnormality (%)
CBFA2T3::GLIS2	32 (9/28)	ND	ND	ND
FLT3-ITD	10.7 (3/28)	ND	2.6 (1/39)	6.2 (5/81)
NPM1	14.2 (4/28)	ND	ND	21 (17/81)
TP53	3.6 (1/28)	ND	5.2 (2/39)	6.2 (5/81)
RUNX1T1::RUNX1	17.8 (5/28)	27 (3/11)	2.6 (1/39)*	17.2 (14/81)
U2AF1	3.6 (1/28)	ND	2.6 (1/39)*	1.2 (1/81)
SRSF2	7.2 (2/28)	ND	5 (2/39)	5 (4/81)
BCOR	3.6 (1/28)	18 (2/11)	2.6 (1/39)	1.2 (1/81)
IDH2	3.6 (1/28)	2.6 (1/39)	5 (2/39)	2.4 (2/81)

AML-RAM compared to CD56+ AML patients

Patients in the CD56+ AML group presented with a significantly older median age of 56 years (range 28.5–66.5, p < 0.05), significantly higher WBC (p < 0.05), but with similar hemoglobin, platelet count, and BM cellularity (all p > 0.05) as compared with AML-RAM cases. BM blast percentage was significantly lower than AML-RAM cases (p < 0.05). CD56 expression was partial in CD56+ AML cases compared to the strong bright expression in AML-RAM (all p < 0.05). However, CD56+ AML cases showed more frequent CD38 and CD13 expression (p = 0.018). Molecular analysis of CD56+ AML cases showed frequent FLT3-ITD (6%), NPM1 (21%), RUNX1::RUNX1T1 (17%), NPM1 mutation (21%), and TP53 mutation (6%). Chromosomal analysis showed an abnormal karyotype in 68% of the CD56+ AML cases. The median follow-up time was 18.19 months (range 1–88). Median survival time was significantly lower in CD56+ AML cases compared to AML-RAM cases (HR 0.58; CI: 0.3-0.9; p < 0.05), while there was no significant difference in the EFS among the two groups (p > 0.05) (See tables 2 and 3 and Figure 4).

[IMAGE OMITTED. SEE PDF]

DISCUSSION

AML arises from the accumulation of genetic abnormalities including gene mutations and chromosomal alterations within hematopoietic stem cells, progressing from a pre-leukemic clone to the acquisition of a proliferative advantage and impaired hematopoietic differentiation [28]. Different genetic alterations are associated with distinct AML subtypes, and thus new targeted treatments are being implemented and studied for a personalized treatment approach [28, 32]. Expression of CD56 in AML has been associated with worse clinical outcomes, shortened OS, and failure to achieve complete remission [2, 23, 24, 25]. However, newer studies have suggested that the worse prognosis may be restricted to a unique subgroup with a specific immunophenotype characterized by blasts with bright expression of CD56 and weak to absent expression of CD45, HLA-DR, and CD38 [2]. Pardo et al. compared three cohorts of AML with different intensities of CD56 antigen expression and different cytogenetic abnormalities [8]. The 5-year EFS and 5-year OS were significantly worse in the subgroup with the brightest CD56 intensity (mean linear MFI for the entire cohort was 2040) that also had a high percentage of CBFA2T3::GLIS2 fusion transcript, compared to the other two cohorts with lower CD56 intensity and different characteristic genetic aberrations (One with a mean MFI of 121 harboring RUNX1::RUNXT1, and another with a mean MFI 228 harboring 11q23/KMT2A(MLL) rearrangement [8, 31, 33, 34].

We have confirmed that the immunophenotypic approach of combining bright CD56 expression along with dim to negative CD45, HLA-DR, and CD38 is helpful in the rapid diagnosis of a rare, high-risk subtype of AML with poorer prognosis and OS defined as AML with RAM phenotype [5, 15, 16, 29]. We also confirm that AML with RAM phenotype is distinct as compared to other CD56 co-expressing leukemia control groups, with unique clinicopathological, immunophenotypic, and mutational presentation. AML with RAM phenotype shows worse prognosis, OS, and EFS and is a prognostic indicator for frequent relapse rate and treatment failure even after hematopoietic stem cell transplantation when compared to other CD56 positive AML subtypes. However, the similarities in presenting overlap in WBC, hemoglobin, and blast percentage between AML with RAM phenotype and control groups would preclude the use of these variables in differentiating between CD56-positive leukemias at first diagnosis (See Tables 2 and 3)

TABLE 3 Immunophenotypic features of acute myeloid leukemia (AML)-RAM, acute undifferentiated leukemia (AUL), T-ALL, and CD56+ AML patients. Values with asterisks are significant at p < 0.05.

Immunophenotype	RAM	AUL	T-ALL	CD56+ AML
CD56	+B	+ *	+P *	+P *
CD34	-D	+	+P	V
CD33	+	-P *	-P *	+
CD13	-P	—*	-	+ *
HLA-DR	—	V *	V *	+P *
CD117	V	-P *	-P *	V
MPO	V	-	—*	-P *
TDT	-P	+ *	+ *	-
CD38	-D	+ *	+P *	+P *
CD45	-D	+D	+ *	+ *

A CBFA2T3::GLIS2 fusion transcript where a CBFA2T3 gene in frame fuses with a region of GLIS2 gene resulting from a recurrent cryptic inversion of chromosome 16 [inv (16)(p13.3q24.3)] has been described in pediatric non-Down Syndrome AMKL (non-DS AMKL) patients and in about 8% of pediatric cytogenetically normal AML [17, 22]. This fusion has been associated with poor prognosis and short EFS [18]. In our study, we identified 68% (10/28) AML-RAM patients with CBFA2T3::GLIS2 who appear to have worse clinical course; conversely, no control CD56+ AML cases harbored this fusion. Two out of the seven (28%) CBFA2T3::GLIS2 positive RAM cases in our study were non-Down Syndrome AMLs with a megakaryoblastic leukemia immunophenotype. In the reported COG cohort, 63% of AML-RAM patients harbored the CBFA2T3::GLIS2 chimeric fusion gene [19]. Another AML-RAM cohort found CBFA2T3::GLIS2 in three of seven (43%) patients [10]. Previous studies showed that CBFA2T3::GLIS2 oncogene directly up-regulates CD56 expression, inferring the association between RAM phenotype and this peculiar fusion [17]. In addition to that, the OS was shorter in RAM cases harboring this mutation, so appropriate testing for fusion after identifying the RAM phenotype would better stratify the risk and might qualify for a more intensive chemotherapy regimen or might perhaps qualify for targeted inhibitors in the context of for clinical trials [21].

Recognizing the unique features of AML with RAM-phenotype entity can help differentiate it from other rare CD56-expressing acute leukemias and aid in the appropriate management, such as AUL. AUL is a rare acute leukemia of ambiguous lineage that shows no lineage differentiation evidence and was shown to have distinct mutational profiles including PHF6, SRSF2, RUNX1, ASXL1, and BCOR [27]. In our study, we incorporated AUL and NK-LL into one group, and our data showed that RAM patients were significantly younger, had higher WBC, and had higher platelets count (all p < 0.05) at presentation. Immunophenotypically, they can be distinguished by the brighter CD34, CD38, and TDT expression in AUL patients. A case series by Weinberg et al. showed that a cohort of NK-LL patients presented at a young age but had a good prognosis in comparison to CD56-positive leukemias [26]. The molecular profile was comparable to T-ALL involving NOTCH1 mutations [2].

Our study showed that AML-RAM differs from T-ALL early at presentation with the unique signature of brighter CD56 expression but dimmer CD38 and TDT expression; however, other clinicopathologic findings were similar. Gupta et al. showed that T-ALL can present with a lack of immaturity markers such as CD34, TDT, and HLA DR [30, 38]. Hence, there is a need for a comprehensive approach utilizing different diagnostic modalities for the correct identification and stratification of each entity.

In conclusion, we confirm that AML with RAM phenotype should be considered in the clinical differential diagnosis of CD56-positive acute leukemias which is challenging. The addition of flow cytometry markers such as CD41 and CD61 might be helpful in identifying megakaryoblastic differentiation in AML-RAM, and testing for CBFA2T3::GLIS2 cryptic fusion and gene mutation studies, including TP53, should be performed for accurate classification. Thorough consideration of the distinct clinical presentation and outcomes between cases expressing an antigen can also support evidence demonstrating how patients would respond to targeted therapy [13, 35].

CONFLICT OF INTEREST STATEMENT

The authors declare no conflict of interest.

FUNDING INFORMATION

The authors received no specific funding for this work.

DATA AVAILABILITY STATEMENT

Data are available upon request.

ETHICS STATEMENT

The authors have confirmed ethical approval statement is not needed for this submission.

PATIENT CONSENT STATEMENT

The authors have confirmed patient consent statement is not needed for this submission.

CLINICAL TRIAL REGISTRATION

The authors have confirmed clinical trial registration is not needed for this submission.

References

Arber DA, Brunning RD, Le Beau MM. Acute myeloid leukaemia (AML) and related precursor neoplasms. In: Swerdlow SH, Campo E, Harris NL, Jaffe ES, Pileri SA, editors. WHO classification of tumours of haematopoietic and lymphoid tissues. 4th ed. Lyon, France: IARC Press; 2008:110–144.

Jaffe ES, Harris NL, Stein H, Vardiman JW. Pathology and genetics of tumours of haematopoietic and lymphoid tissues. 3rd ed. Lyon, France: IARC Press; 2001:75–105.

Hwang S. Classification of acute myeloid leukemia. Blood Res. 2020;55(S1):S1–S4. [DOI: https://dx.doi.org/10.5045/br.2020.S001]