High-Risk Acute Promyelocytic Leukemia with Unusual T/Myeloid Immunophenotype Successfully Treated with ATRA and Arsenic Trioxide-Based Regimen

Abstract

We describe two patients with acute promyelocytic leukemia (APL) with an unusual immunophenotype with co-expression of myeloperoxidase (MPO) with cytoplasmic CD3 (cCD3) representing myeloid and T-lineage differentiation. Both harbored FLT3-ITD mutations. One additionally had a deletion in the PML gene affecting the primer binding site, thus limiting measurable residual disease (MRD) analysis during follow-up. Both patients achieved durable remission with all-trans retinoic acid (ATRA) and arsenic trioxide (ATO)-based therapy, thus mitigating the need for repetitive conventional chemotherapy cycles and allogeneic stem cell transplantation. Our report highlights the complexity and challenge of diagnosis and management of APL due to the variant immunophenotype and genetics, and underscores the importance of synthesizing information from all testing modalities. The association of the unusual immunophenotype and FLT3-ITD mutation illustrates the plasticity of the hematopoietic stem cell and the pathobiology of leukemia with mixed lineage or lineage infidelity.

Introduction

APL is an oncologic emergency due to the high mortality rate from hemorrhage and disseminated intravascular coagulation. Its exquisite sensitivity to all-trans retinoic acid (ATRA) mandates rapid diagnosis and initiation of therapy. Morphology and immunophenotype allow a preliminary diagnosis in most cases while awaiting confirmation by fluorescent in situ hybridization (FISH) or conventional karyotyping for the chromosomal translocation t(15;17). In the majority of cases, the leukemic promyelocytes express myeloid antigens CD117, CD33 (homogeneous), CD13 (heterogeneous), and myeloperoxidase (MPO), and are negative for CD34 and HLA-DR [1–4]. The blasts in APL consistently lack or have a very dim expression of CD14, CD15, CD11a, CD11b, CD11c, CD18, CD66b, and CD66c [1,5]. This immunophenotype in experienced hands has been reported as 100% sensitive and 99% specific for predicting APL [3,4]. Aberrant expression of T-cell antigen CD2 and of CD34 may be seen in the microgranular variant (M3v) of APL [6,7]. Other variant immunophenotypes have been described including expression of CD15, CD56 [8], and variable expression of CD11b and CD11c [9].

According to the 2008 and 2016 revised World Health Organization (WHO) classification, acute leukemias expressing stringent lineage-defining markers of more than one lineage are characterized as mixed phenotype acute leukemias (MPAL) [8]. A diagnosis of MPAL also requires exclusion of acute leukemia with recurrent cytogenetic abnormalities including t(15;17). Previously, European Group for the Immunological Classification of Leukemia (EGIL), a scoring system based on expression of lineage-specific antigens was applied to define biphenotypic acute leukemia (BAL)[9]; some of the high-scoring antigens were subsequently determined not specific - for example CD79a, which may be present on normal thymocytes, relegating the EGIL system to historical reference. Expression of cytoplasmic CD3 (cCD3) defines T-lineage. The expression of cCD3 in APL is unusual and extremely rare. In a comprehensive analysis of the T-lymphoid genetic program in APL, Chapiro et al. [10] reported cCD3 expression in 2 out of 13 samples from patients with microgranular variant (M3v) of APL, which were selected from a cohort of 36 APL patients. The impact of this phenotype on APL outcome is unknown. Here we describe two patients diagnosed with APL, confirmed for t(15;17) promyelocytic leukemia/retinoic acid receptor α (PML/RARA) by FISH, with a T/Myeloid immunophenotype. Both were treated successfully with ATRA- and arsenic trioxide (ATO)-based therapy, resulting in complete morphological and cytogenetic remission. Molecular remission was confirmed in one patient.

Methods

The morphological evaluation was performed on Wright-Giemsa stained peripheral blood and bone marrow aspirate smears, and hematoxylin and eosin stained core biopsy sections.

Flow cytometry

Four-color flow cytometry was performed on the BD FACSCantos™ II flow cytometer. Typically 20,000 events were acquired. Peripheral blood and bone marrow sample processing and antibody staining were performed by a standard wash/lyse method per manufacturer’s suggestion. The following antibodies were used for analysis (all obtained from BD Biosciences unless otherwise specified): anti-CD34, CD117, CD13, CD33, CD15, CD16 (BD Pharminogen), CD14, CD11b, CD11c, MPO, CD2, CD3, CD4, CD5, CD7, CD8, CD10, CD19, CD20, CD22, CD79a, kappa/lambda combination (Dako), CD56, CD30, CD1a (Bio-Rad), terminal deoxynucleotidyl transferase (TdT), and HLA-DR. They were conjugated with one of four fluorochromes: fluorescein isothiocyanate (FITC), phycoerythrin (PE), peridinin chlorophyll protein (PerCP), or allophycocyanin (APC). The amount of antibodies used was based on the manufacturer’s instructions. For intracytoplasmic staining (TdT, cytoplasmic CD3, cytoplasmic CD79a, MPO), prior fixation and permeabilization were performed. The analysis was performed using the BD FACS Diva v8.0.1 software. The dim CD45, low or variable side scatter cell population was selected for analysis.

Immunohistochemistry

Immunohistochemical stains were performed on formic acid-decalcified, formalin-fixed, paraffin-embedded, 4 micron thick bone core biopsy sections. Immunohistochemistry was performed on the Bond-III automated immunohistochemistry (IHC) platform (Leica Microsystems) using primary Bond ready-to-use antibodies to CD34, CD2, CD3, CD5, CD7, TdT, and MPO antigens. The tissue bound primary antibodies were visualized using Bond Polymer Refine Detection according to the manufacturer’s instructions.

Cytogenetics

Conventional metaphase karyotyping was performed by GTG-banding, and FISH was performed on interphase cells using dual-color, dual-fusion probes specific for the PML (15q22) and RARA (17q21) loci, that identify the PML/RARA gene fusion.

Quantitative reverse transcription-polymerase chain reaction (qRT-PCR)

qRT-PCR for the PML-RARA mRNA (bcr1, bcr2, and bcr3 isoforms) was performed on pre- and post-treatment samples at a commercial laboratory. RNA was isolated, reverse transcribed into complementary DNA (cDNA), amplified using primers specific for the PML and RARA genes, and subjected to qRT-PCR analysis using primers previously described in the Europe Against Cancer (EAC) guidelines [11] designed to amplify across PML-RARA breakpoints. There are three common breakpoints within the PML gene, bcr1 (intron 6), bcr2 (exon 6), and bcr3 (intron 3). All breakpoints fuse a portion of the PML gene to a consistent breakpoint region within the RARA gene to generate the long, the short, or variable-fusion isoforms. The level of PML-RARA fusions in the patient sample was expressed as copy numbers further normalized to the reference gene ABL1 and resulted in a normalized ratio of PML-RARA transcripts to ABL1 transcripts present in the sample.

Sequencing

Sequencing was performed at the translational genomics laboratory of the University of Maryland School of Medicine. To amplify the PML-RARA fusion gene, a two-step RT-PCR analysis was performed as described by Biondi et al. [12]. The PCR products were electrophoresed on a 2% agarose gel stained with ethidium bromide and visualized under ultraviolet light. Following purification of the PCR products, Sanger sequencing was performed on an automated DNA sequencer [(ABI 3730 DNA analyzer, (Life Technologies, Carlsbad, CA)]. The nucleotide sequence of the different PCR fragments was assessed by using the sense primers M2 and M5 for the M2/R5-R8 and M5/R5-R8 PML-RARA fusion products as described previously [12]. The sequencing data were further analyzed using the human BLAST-like alignment tool (BLAT) search tool from the University of California Santa Cruz (UCSC) genome browser (http://genome.ucsc.edu/cgi-bin/hgBlat?hgsid=662024597_5FG1jOCdchki6XggN5lPRH5oOaIT&command=start).

Case Reports

Patient 1

A 29-year-old previously healthy African-American woman presented with new-onset gum bleeding and a petechial rash. White blood cell (WBC) count was 23×10³/mcL (normal range 4.5–11), hemoglobin (Hgb) 4.6 g/dL (normal range 12–17), and platelet count 32×10³/mcL (normal range 150–450). The peripheral blood smear had 92% immature cells with folded or bilobed nuclei, scant to occasionally dense azurophilic cytoplasmic granules, and Auer rods, morphologically typical of microgranular variant of APL (Figure 1A). The blasts co-expressed cytoplasmic MPO, along with cCD3 and cytoplasmic TdT, consistent with a T/myeloid immunophenotype (Figure 1B). The complete immunophenotype is shown in Table 2. The expression of T-lineage antigens (CD2, CD3, CD5, and CD7), TdT, CD34, and MPO was also investigated by immunohistochemical (IHC) stains on the core biopsy section (Figure 1. C-H), which confirmed CD3 expression in a subset of leukemic promyelocytes. The expression of CD3 and other pan T-cell antigens CD2 and CD7 was dimmer in the leukemic promyelocytes when compared to the residual normal T-cells in the sections (Figure 1, E, F and H). The abnormal t(15;17) PML/RARA translocation was identified by metaphase cytogenetics and confirmed by FISH in 92% of nuclei. qRT-PCR identified the bcr3 PML/RARA transcript in peripheral blood (155.568%) and bone marrow (246.369%). The FLT3-ITD mutation was identified by sequencing.

Figure 1.

Patient 1: A) Leukemic promyelocytes in the bone marrow (Wright-Giemsa, 1000x). B) Flow cytometry dot plot showing co-expression of myeloid (cMPO) and T-lymphoid (cCD3*) antigens; the blue dotted line and arrow indicate the dimmer expression of cCD3 in leukemic cells as compared with normal residual T-cells. C) MPO, immunoperoxidase, 2000x. D) TdT, immunoperoxidase, 2000x. E) CD2, immunoperoxidase, 2000x. F) CD3, immunoperoxidase, 2000x. G) CD5, immunoperoxidase, 2000x. H) CD7, immunoperoxidase, 2000x. Note in E, F, and H that the expression of CD2, CD3, and CD7 is dimmer in leukemic cells in contrast to the background normal T-cells. Only normal T-cells are positive for CD5.

Table 2:

Distribution of positive and negative different antigens by flow cytometry and immunohistochemistry

	Flow cytometry			Immunohistochemistry
	CD45 vs. SSC	Positive markers	Negative markers
Patient 1	84.2% with dim CD45 and intermediate side scatter	CD34,CD117(p), CD33,CD13,CD11c,CD38, CD71, cMPO, HLA-DR(p), CD7(p), cCD3,and TdT(p).	CD11b, CD14, CD15, CD16, CD56, CD64, surface CD3, CD2,CD4,CD5, CD8,CD10, CD19, CD20, cCD79a	Blasts express MPO, CD34, CD117, CD2*(dim, subset), CD3 (dim), CD7 (dim), and TdT (subset, dim). Blasts negative for CD5.
Patient 2	89% with dim CD45 and intermediate side scatter	CD34(p), CD117, CD33, CD13, CD38, CD64, CD71, MPO(dim), HLA-DR (p), CD2, and cCD3	CD11b,CD11c, CD14, CD15, CD16, CD56, surface CD3, CD4, CD5, CD7, CD8, CD10, CD19, CD20, cCD79a, TdT	Blasts express MPO, CD117, CD34 (focal), CD2, CD3 (subset, dim), TdT*(subset, dim). Blasts negative for CD5 and CD7.

^*These antigens were dim positive by immunohistochemistry, but negative by flow cytometry.

The patient was treated per the APML4 regimen for remission induction with ATRA, ATO, and idarubicin for high-risk APL with dexamethasone prophylaxis for differentiation syndrome [13]. The induction course was complicated by disseminated intravascular coagulation (DIC) and neutropenic fever. Follow-up bone marrow biopsy on day 35 after the initiation of induction chemotherapy (D+35) showed trilineage hematopoiesis without morphological evidence of APL. PML/RARA transcript in the bone marrow had decreased to 2.424%. Subsequently, the patient underwent consolidation therapy with ATRA and ATO and maintenance therapy with ATRA, methotrexate, and mercaptopurine. The PML-RARA transcript was undetectable (<0.001%) on D+139, as was FLT3-ITD. The patient remains in complete remission (CR) at last follow-up and PML-RARA transcript remains undetectable 26 months after diagnosis.

Patient 2

A 50-year-old Caucasian man with no past medical history presented with fatigue, fever, and petechial rash. WBC was 20×10³/mcL, Hgb 7.3 g/dL, and platelet count 11×10³/mcL. On peripheral blood smear, 82% of the WBC were immature with folded or bilobed nuclei, fine azurophilic granules in the cytoplasm, and Auer rods, typical for the microgranular variant of APL (Figure 2A). The blasts co-expressed cMPO and cCD3 by flow cytometry, consistent with a T/myeloid immunophenotype (Figure 2B). The complete immunophenotype is shown in Table 2. The expression of T-lineage antigens (CD2, CD3, CD5, and CD7), TdT, CD34, and MPO was investigated by immunohistochemical (IHC) stains on the core biopsy section (Figure 2. C-H), which confirmed CD3 expression, and additionally expression of CD2 and TdT in a subset of leukemic promyelocytes. The results of IHC are shown in Figure 2. C-H, and Table 2. Conventional cytogenetics identified t(15;17), and FISH confirmed the presence of PML/RARA in 91% of nuclei. However, qRT-PCR did not identify a PML/RARA transcript. Further sequencing analysis of the PML/RARA fusion gene revealed a deletion of exon 5 of the PML gene in its entirety on chromosome 15q. The breakpoints of the deletion were identified as seq[GRCh37] chr15:74324913–74325056 on the GRCh37/hg19 browser. Based on the Human Genome Variation Society (HGVS) (http://varnomen.hgvs.org/), the deletion nomenclature is PML c.1255_1398del p.(Pro419_Glu466del). Molecular studies identified FLT3-ITD and ASXL1 mutations.

Figure 2.

Patient 2: A) Leukemic promyelocytes in the bone marrow (Wright-Giemsa, 1000x). B) Flow cytometry dot plot showing co expression of myeloid (CD33) and T-lymphoid (cCD3) antigens; the blue dotted line and arrow indicate the dimmer expression of cCD3 in leukemic cells as compared with normal residual T-cells. C) MPO, immunoperoxidase, 2000x..D) TdT, immunoperoxidase, 2000x. E) CD2, immunoperoxidase, 2000x. F) CD3, immunoperoxidase, 2000x. G) CD5, immunoperoxidase, 2000x. H) CD7, immunoperoxidase, 2000x H. Note in E and F that the expression of C and CD3 is dimmer in leukemic cells in contrast to the background normal T-cells. Only normal T-cells are positive for CD5 and CD7.

The patient was treated per the APML4 regimen for high-risk APL [13]. His hospital course was complicated with neutropenic fevers and fungal pneumonia, which were successfully treated with intravenous antibiotics and antifungal agents. On D+25, cells with morphological features of APL were absent from the bone marrow aspirate and biopsy. Subsequently, the patient received consolidation and maintenance therapy. Measurable residual disease (MRD) could not be monitored by conventional PML/RARA qRT-PCR due to the deletion of the PML gene, but FISH for t(15;17) was negative at D+28 and D+58, and the FLT3-ITD mutation was undetectable at D+58, consistent with a response to therapy. The patient remains in morphologic and cytogenetic remission at 23 months from his initial diagnosis.

Discussion

We present two patients with an unusual T/myeloid immunophenotype in t(15;17) APL and successful treatment of both with an ATRA and ATO-based regimen. Interestingly, one patient had a deletion in the PML gene (exon 5) that involved the primer binding site and precluded generation of the abnormal PML/RARA transcript. This resulted in discordant FISH and RT-PCR results precluding MRD detection by qRT-PCR.

Aberrant or cross antigen expression in leukemic blasts may result from lineage infidelity resulting from genetic aberrations, which may affect transcription factors, cytokines, or growth factors that regulate lineage commitment in hematopoietic cells, or from lineage promiscuity occurring as a consequence of mutations that cause maturational arrest at an early, multilineage potential progenitor stage. Gene expression studies have shown a distinct profile for acute leukemia with a T/myeloid phenotype characterized by silencing of myeloid transcription factor CCAAT-enhancer-binding protein α (C/EBPα) and enhanced expression of Tribbles homolog 2 (TRIB2) [14]. TRIB2 is a direct target of the T-cell commitment factor NOTCH1, implicating aberrantly activated Notch signaling in the pathogenesis [14]. Although large studies with mutational data are lacking in this patient population, NOTCH1 mutations appear to be relatively common in T/myeloid MPAL, along with mutations in DNMT3A, NRAS, KRAS, EZH2, TP53, and RUNX1 genes [15,16]. FLT3-ITD mutations, as seen in our two patients, have also been reported in patients with MPAL [16]. This mutation is more commonly seen in APL, where it is associated with marked leukocytosis, microgranular morphology, expression of CD2, CD34, HLA-DR, and CD11b surface antigens, and a short PML/RARA (BCR3) isoform [17, 18]. Unlike other non-promyelocytic AML, however, in which FLT3-ITD mutations are associated with a high propensity for relapse, the role of FLT3-ITD mutations in APL is controversial [17,19–25], and most multicenter studies suggest that any association with poorer outcomes may be mitigated by the addition of ATO [22,23,25].

A diagnosis of MPAL according to the WHO requires expression of more than one lineage-determining antigens on the blasts and exclusion of recurrent cytogenetic abnormalities seen in AML such as inv(16), t(8;21) and t(15;17). Although cCD3 expression, which establishes T-lineage, is commonly heterogeneous, per WHO, for diagnosis of a mixed T-phenotype, the brightest cCD3-positive blasts should reach the intensity of the normal residual T cells present in the sample [8]. In both cases described in this report, the intensity of cCD3 expression is dimmer than the control T-cells in the specimen (Figures 1B and 2B). The cases do not fulfil either of the two WHO requirements for MPAL. The pattern of expression of cCD3 on the APL cells in our study is similar to that observed by Chapiro et.al [10]; cCD3 in their study was considered positive if the relative fluorescent intensity (RFI) on the leukemic promyelocytes was greater than 2 but lower than the background T-cells within the CD45 leukemic gate. Interestingly, both the cCD3 expressing APL cases in their study expressed preT-cell receptor α, an invariant glycoprotein associated with the β T-cell receptor chain and CD3 protein that is considered specific for the T-cell lineage [26]. Due to the very limited data about cCD3 expression on APL, its clinical significance, and hence of the T-lineage program in APL is not yet known. Diagnosis of MPAL, in contrast, may have profound implications for management. Several retrospective studies have consistently suggested that patients with MPAL have superior outcomes with ALL-type regimens compared to AML-type regimens [27,28], and successful induction chemotherapy is often followed by allogeneic stem cell transplantation [29,30]. Although it is too simplistic to conclude that all patients with MPAL should receive ALL-type therapy and it is impossible to determine whether intensive chemotherapy would result in the same successful outcome for our two patients, ATRA and ATO-based regimens are highly effective for patients with t(15;17) and likely less toxic than intensive chemotherapy [13,31,32]. Paietta and colleagues previously reported on a patient with clinical and morphologic features of APL with co-expression of TdT, T6 (CD1a), and T3 (CD3) demonstrated by double-staining microscopy as well as FACS-analysis. Interestingly, karyotype was +8 at presentation and t(15;17) was seen only at relapse [33]. The authors report that following culture in the presence of GCT-conditioned medium, while the morphologic picture remained dominated by hypergranulated promyelocytes with strong staining for myeloperoxidase, there was no evidence of myeloid maturation, but the proportion of cells expressing T-cell markers T11 (CD2) and T3 (CD3) increased, indicating differentiation along this pathway. Exposure of the patient’s cells to retinoic acid in vitro resulted in marked increase in the expression of T3 along with morphological and immunological evidence of myeloid maturations, analogous to the clinical response seen in our two patients.

The identification of the specific PML-RARA transcript at diagnosis in patients with APL is essential for subsequent MRD analysis and disease monitoring by qRT-PCR. Whereas the breakpoint on the RARA gene is always in intron 2, breakpoints on the PML gene can vary: intron 6 for bcr1 (55%), variable points on exon 6 for bcr2 (5%), and intron 3 for bcr3 (40%). The dual color, dual-fusion probe sets for FISH can detect all three variants of the PML-RARA translocation in 98% of cases [34]. The rare cases with complex translocations or cryptic translocations from submicroscopic insertions of the PML gene into RARA may be undetectable by FISH, but can be identified RT-PCR [35–38]. It is exceedingly rare for RT-PCR to be negative in cases of APL that are identified by conventional cytogenetics or FISH, and this is usually the result of suboptimal sample quality or complex genomic fusion sequences that include a third gene or alternative splicing of RARA exon 2 [39]. Uncommonly, this may be due to mutations in either derivative chromosome that may affect primer binding sites [40]. Additional sequencing studies of the PML/RARA fusion gene are required to resolve these rare cases with a discordant result by RT-PCR. For patient #2 in this report, a deletion in the exon 5 of the PML gene abrogated the internal forward primer binding site, precluding amplification of the fusion transcript and limiting our ability for MRD testing.

In conclusion, the unique pathological features of the two patients presented in this report add to our understanding of stem cell plasticity, clarify the dilemma of classifying acute leukemia when encountering the unusual T/Myeloid immunophenotype with t(15;17), and increase awareness of interpreting the rare situation with discordant FISH (positive) and RT-PCR (negative) results. Our report highlights the importance of integrating multiple modalities of laboratory investigation for disease classification and careful consideration of disease biology prior to initiation of chemotherapy. Based on our experience, t(15;17) APL with aberrant expression of CD3 behaves more similar to APL than MPAL, and the T/myeloid phenotype does not abrogate the good prognosis conferred by t(15;17) when an ATRA and ATO-based regimen is utilized.

Table 1.

Patient characteristics

	Age (years)/sex	WBC at diagnosis (/mcL)	Platelets at diagnosis (/mcL)	Immunophenotype	Karyotype	Oncoprotein	Myeloid mutations	Outcome
Patient 1	29/F	23,000	32,000	CD34, CD117, HLA-DR, CD38, CD71, CD33, CD11b, CD11c, CD13, CD7, cMPO, cTDT, cCD3	46,XX,t(15;17)(q24q21)[20]. nuc ish(PML,RARA)x3,(PML con RARA)x2[183/200]	PML/RARα bcr3 isoform	FLT3-ITD (p.Glu604_Phe605ins6. C.1794_1811dup)	Ongoing CR (22 months after diagnosis)
Patient 2	50/M	20,000	11,000	CD34 (partial), CD117, HLA-DR, CD38, CD71, CD64, CD33, CD13, CD11c, cMPO (dim), cCD3	46,XY,t(15;17)(q24;q21)[20]. nuc ish(PML,RARA)x3,(PML con RARA)x2[183/200],(RARAx2)(5′ RARA sep 3′RARAx1)[179/200]	--	FLT3-ITD (p.Tyr599_Pro606dup. c.1795_1818dup), ASXL1 (p.Pro779Leu)	Ongoing CR (19 months after diagnosis)

ASXL1 = additional sex combs like 1; c = cytoplasmic; CR = complete remission; F = female; FLT3-ITD = FMS-like tyrosine kinase 3 receptor internal tandem duplication; M = male; PML/RARα = promyelocytic leukemia/retinoic acid receptor alpha; TDT = terminal deoxynucleotidyl transferase; WBC = white blood cells