Epidemiology and characteristics of validated mixed phenotype acute leukaemia—A comprehensive nationwide Danish cohort study

Mixed phenotype acute leukaemia (MPAL) is a rare and heterogeneous subtype of acute leukaemia, where the blast cells express both lymphoid and myeloid immunophenotypic markers.1 The knowledge on MPAL is sparse and incidence and characteristics in validated population-based cohorts are lacking.2 For two decades, the diagnostic workup in acute leukaemia has advanced considerably, leading to subsequent changes in classification. The classification of MPAL has changed from the European Group for Immunological Characterization of Acute Leukemia (EGIL) classification, a scoring system based on a broad panel of myeloid, B-lymphoid and T-lymphoid markers, to the World Health Organization (WHO) classification, a scoring system with fewer and more lineage-specific markers, excluding leukaemias that would previously have been classified as MPAL (Tables S1 and S2).1, 3 Studying MPAL epidemiology is challenging due to its heterogeneous presentation, low incidence and evolving classification.2 In this population-based study, we identified and validated adult MPAL patients in Denmark (2000–2022) according to the WHO 2022 (MPALWHO) and the EGIL (MPALEGIL) classifications, and report a population-based incidence rate, individual-level information including immunophenotype, cytogenetics and somatic mutations in validated MPAL patients. The Danish National Pathology Registry (DPR) was searched for bone marrow biopsies performed in Denmark during the study period, using SNOMED codes indicating MPAL or acute undifferentiated leukaemia (AUL) (Figure S1A). Patients were classified into four groups: MPALWHO, MPALEGIL, possible MPAL and AUL (described in Supporting Information). Possible MPAL was assigned to patients with unspecific myeloid and lymphoid surface markers, not fulfilling the WHO 2022 or EGIL criteria for MPAL. Baseline characteristics were reported as medians (with interquartile range [IQR]) for continuous variables and as frequencies for categorical variables. The incidence rate of MPALWHO (per 1 000 000 person-years) was estimated for each year from 2003 to 2022 and for the period 2013–2022 relative to the number of the Danish population ≥18 years in the respective periods. Frequencies of cytogenetic and next-generation sequencing (NGS) abnormalities were presented in bar charts. A Circos plot illustrating clustering patterns of surface markers in MPALWHO and MPALEGIL was designed using the R package circlize. All statistical analyses were performed in R (version 4.2.1). We identified 601 patients in DPR with a SNOMED code indicating MPAL or AUL. Upon validation, 108 patients were included and classified as follows: 76 MPALWHO, 18 MPALEGIL, 7 possible MPAL and 7 AUL (Figure S1B). Table 1 shows baseline characteristics of the 108 MPAL patients. Median age at diagnosis for all patients was 66 years (IQR 46–76) and 51.8% were male. The immunophenotype of MPALEGIL/MPALWHO was 76 B/myeloid, 17 T/myeloid and 1 T/B/myeloid. Clustering patterns of surface markers in MPALWHO and MPALEGIL are shown in Figure S4A,B illustrating the large variation of surface markers in MPAL. In 76 MPALWHO patients, Myeloperoxidase (MPO) was the dominant surface marker clustering most frequently with CD19, CD79a and CD10. CD19 clustered most with CD79a, CD10 and cytoplasmatic CD20; 90.2% of B/myeloid MPALWHO cases expressed MPO and CD19 was expressed in 85.5% of B/myeloid MPALWHO cases. The 14.5% of B/myeloid MPALWHO patients who did not express CD19 expressed strong staining of CD79a and PAX5 together with CD10 (77.8%) or CD22 (22.2%). Cytogenetics were available in 90 patients. The most common cytogenetic aberrations were t(9;22); BCR-ABL1 (n = 12), trisomy 8 (n = 11), monosomy 7 (n = 10), monosomy 13 (n = 8), monosomy 16 (n = 7), monosomy 9 (n = 7), monosomy 20 (n = 7) and trisomy 21 (n = 7). Figure 1A shows cytogenetic aberrations present in ≥2 cases (aberrations <2 cases are listed in Figure S3). Of the 90 patients with available cytogenetic data, 40 patients (44.4%) had a complex karyotype. Targeted-panel NGS for common myeloid variants were available in 30 patients. The most common variants in MPAL patients were RUNX1 (n = 11), TET2 (n = 8), TP53 (n = 8), DNMT3A (n = 7), ASXL1 (n = 7), NRAS (n = 7), SRFF2 (n = 4) and FLT3-ITD (n = 4) (Figure 1B). We identified 94 Danish patients with MPAL diagnosed between 2000 and 2022. All the 94 patients fulfilled the EGIL criteria, but only 76 patients fulfilled the WHO 2022 criteria, corresponding to an MPALWHO incidence rate of 1.20 per 1 000 000 person-years between 2013 and 2022. No MPAL cases could be confirmed according to the EGIL or WHO criteria in the period 2000–2002, primarily due to the limited availability or utilization of diagnostic tests in this period. The diagnostic armamentarium improved during the study period and the use of diagnostic tests increased. This may explain the increasing rate of MPAL cases.4 Moreover, the increasing rate of MPAL coincided with an increasing median age at diagnosis, that is more older patients have been diagnosed in recent years. Most cases of MPAL were validated as MPALEGIL in the period 2003–2007, whereas an increasing number were validated as MPALWHO, when the diagnostic workup was expanded, and the WHO classification was introduced in 2008.1 The EGIL classification is based on less specific markers, classifying more cases as MPAL compared to the WHO classification.5 The WHO classification excludes a range of acute leukaemias with cytogenetic aberrations such as acute myeloid leukaemia (AML) inv.(16)/t(16;16) and AML t(8;21), which may present with aberrant B-cell markers, for example CD19.6 MPALWHO was therefore used to estimate the incidence rate between 2013 and 2022, which was found to be 1.20 per 1 000 000 person-years in Denmark. However, the true incidence might be even higher as in some MPALEGIL cases and possible MPAL cases the essential surface markers according to the WHO classification were not performed. Hence, some MPALEGIL cases could potentially comply with the WHO classification, but the diagnostic flow cytometry report lacked information on more lineage-specific markers. We found a higher incidence rate of MPAL than previously described in a SEER database study including 313 MPAL WHO2008 cases from 2001 to 2011. The SEER database study included children and adults, and the diagnoses were not validated, making a direct comparison with our study difficult. However, they found an incidence of 0.35 cases per 1 000 000 person-years.7 This difference could be explained by the different study periods; the incidence of MPAL increased over time, possibly due to improved diagnostic workup (Figure S2). Furthermore, in our study, we were able to assess the different diagnostic strategies employed by different centres and make a focused reassessment of relevant acute lymphoid leukaemia (ALL) cases (described in Supporting Information). The most common immunophenotype was B/myeloid followed by T/myeloid while T/B/myeloid was very rare, which is in concordance with previous findings.2 A total of 44.4% had complex cytogenetics, which appeared higher than previously described, for example a study including cytogenetics on 92 MPAL patients found complex cytogenetics in 24% and another study including cytogenetics on 76 MPAL patients found complex cytogenetics in 32%.8, 9 t(9;22); BCR-ABL1 was present in 13.3%, which is in line with previous studies.8, 9 We only found KMT2A-rearrangement in one case. This is lower than in a previous study (4 of 92 patients)9 that included paediatric cases where the incidence of KMT2A-rearrangement is known to be higher.10 Most of the trisomies and monosomies reported were part of a complex karyotype. The prognostic impact of trisomies and monosomies is in general unknown in MPAL, but monosomies like -7 and -5 are known to be associated with an inferior outcome in both ALL and AML.11 NGS data were available for 30 patients, and variants were detected in 28 cases. Most patients had >1 variant, which is in line with a previous smaller study reporting NGS data on 14 MPAL patients.12 Furthermore, known variants associated with unfavourable prognosis in AML such as RUNX1, TP53, ASXL1 and FLT3-ITD were detected in MPAL in 36.7%, 26.7%, 23.3% and 13.3% of the cases respectively.13 FLT3-ITD(4), EZH2(3), SF3B1(2), SETBP1(2), CSF3R(1), CALR(1), CBL (1), IDH1(1) and FLT3-TKD(3) were exclusively found in B/myeloid MPAL, and PTPN11(1) was only found in T/myeloid. However, FLT3, EZH2 and CSF3R variants were found in T/myeloid MPAL cases in a previous study.9, 14 In another study, SF3B1, SETBP1 and IDH1 variants were also exclusively detected in B/myeloid MPAL, but only in 1–2 cases.14 Nevertheless, one limitation of using only the diagnostic myeloid NGS panel is that potentially lymphoid variants are not reported. Furthermore, the myeloid NGS panels used in the five Danish health administrative regions demonstrated minor variations depending on the diagnostic site and period; however, the most frequently detected variants in our cohort were consistently included across all panels. We were able to retrieve the entire medical history of all MPAL patients by use of the Danish civil registration number. The free and unlimited access to health services offered during time to the individual patient linked to the CPR-number, allowed us to recognize earlier diseases not reported at diagnosis of MPAL. Surprisingly, 27.8% of our cohort had previous cancers. Previous cancers in MPAL have only been described in a few cases earlier but may have been overlooked in studies based on register data.15 More details and further investigations regarding this subgroup of MPAL patients with previous cancers including previous treatment are needed. This study presents novel knowledge on MPAL, reporting a population-based incidence rate of validated MPALWHO of 1.20 per 1 000 000 person-years in the period 2013–2022. Our findings highlight the complexity and heterogeneity of MPAL, including high rates of complex cytogenetics, multiple NGS variations with frequent unfavourable AML-associated variants and a high proportion of patients with prior cancer. Major discrepancies between the classifications were observed, highlighting the need for consensus regarding the classification in both clinical practice and research. Our study emphasizes the importance of access to individual-level data, including flow cytometry results, cytogenetic analyses and pathology reports, which are essential for accurate validation of MPAL. A substantial proportion of cases recorded as MPAL in DPR did not meet the WHO 2022 diagnostic criteria upon review. This finding highlights a key limitation of registry-based studies in the context of rare diseases such as MPAL, as these cohorts may include both true MPAL cases and patients more appropriately classified as having ALL or AML. More knowledge on the genetic landscape of MPAL is needed, as well as the clinical courses and outcomes in validated population-based cohorts. Conception and design: L-MC, KD, TCE-G, KG, DTK and MTS. Provision of patient's material: MTS, HBO, DLH, ALTS, MSE and KG. Collection and assembly of data: L-MC, HBO, DLH, ALTS, DTK and MTS. Data analysis and interpretation: L-MC, JFJ, DTK, KD, TCE-G, KG, RFB, MSE and MTS. Manuscript writing: All authors. Final approval of manuscript: All authors. Accountable for all aspects of the work: All authors. We thank Mette Klarskov, Department of Haematology, Rigshospitalet, for assistance with locating the old cytogenetic analyses. This project is funded by the Danish Cancer Society R368-A21409 and Sawmill Owner Jeppe Juhls and Wife Ovita Juhls. HBO has received research funding from Jazz Pharmaceutical, and consultancy fees from Abbvie, Daiichi Sankyo, Takeda and Sanofi. ALT served on consulting/advisory boards for AbbVie, Servier, Otsuka and Daiichi Sankyo. The project was approved by the Danish Health Research Ethics Committee (reg. no. 2304766) and the North Denmark Region (reg. no 2022–015509). Data are not publicly available due to General Data Protection Regulation (GDPR) restrictions. Data S1. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.