Lineage switch is infrequent in acute leukaemia with the literature limited to case reports or small series (Sakaki et al., 2009; Park et al., 2011; Rossi et al., 2012; Della Starza et al., 2016; Rayes et al., 2016; Wu et al., 2017; Hanley et al., 2018; Nomani et al., 2019). The majority of reported cases are switching between B cell acute lymphoblastic leukaemia (B‐ALL) and acute myeloid leukaemia (AML) (Sakaki et al., 2009; Park et al., 2011; Rossi et al., 2012; Della Starza et al., 2016; Rayes et al., 2016) involving lysine methyltransferase 2A (KMT2A) re‐arrangements in paediatric patients (Sakaki et al., 2009; Park et al., 2011; Rossi et al., 2012; Rayes et al., 2016). AML switching to other lineages is extremely uncommon, but has been reported in rare cases (Hanley et al., 2018; Nomani et al., 2019). AML with granulocytic differentiation switching to a megakaryoblastic phenotype has never been described in the English literature. Herein, we report an unusual case of AML with maturation (AML‐M2 by French‐American‐British [FAB] classification) that converted to acute megakaryoblastic leukaemia (AML‐M7 by FAB) by morphology and immunophenotypic profile during induction chemotherapy.

The patient is a 62‐year‐old male who presented with anaemia, thrombocytopenia and marked leucocytosis (179 × 10⁹/l), with markedly increased circulating blasts (64%). Subsequent bone marrow biopsy (biopsy 1) showed 67% blasts (Fig 1, upper panel A–C) that were positive for myeloperoxidase, CD31, CD34 in a small fraction, and negative for E‐Cadherin and factor 8‐related antigen (Fig 1, upper panel C) by immunohistochemistry. Neutrophils and maturing myeloid precursors comprised approximately 20% of the total analysed events. Scattered megakaryocytes were seen, but no increase in dysplastic megakaryocytes was identified in this biopsy. Flow cytometry (Fig 1, upper panel D–F) showed 58% blasts that expressed dim cluster of differentiation 13 (CD13), dim CD33, dim CD11C, CD123, partial CD34, HLA‐DR and aberrant CD7; they were negative for CD41, CD61, CD19, CD10, CD4 and CD14. Chromosomal analysis revealed a normal male karyotype, and fluorescence in situ hybridisation panel was negative for myeloid‐related abnormalities. Next‐generation sequencing (NGS) analysis demonstrated four pathogenic mutations involving nucleophosmin 1 (NPM1; c.860_863dupTCTG on exon 11), Fms‐related tyrosine kinase 3 (FLT3; c.2503G>T on exon 20), DNA methyltransferase 3 alpha (DNMT3A; c.1490G>T on exon 13) and ATM serine/threonine kinase (ATM; c.5558A>T on exon 37) genes. A diagnosis of acute myeloid leukaemia with maturation (AML‐M2) was made, and the patient received standard induction chemotherapy (cytarabine plus daunorubicin; 7 + 3 regimen). At 16 days after the induction, a bone marrow biopsy (biopsy 2) demonstrated 35% blasts and a profound proliferation of dysplastic megakaryocytes, mostly small hypolobate forms (Fig 1, lower panel A, B); the blasts were positive for factor 8‐related antigen (Fig 1, low panel C) and CD31, but were negative for myeloperoxidase, E‐cadherin and CD34 by immunohistochemistry. Flow cytometry (Fig 1, lower panel D–F) demonstrated 24% blasts that were positive for bright CD41 and moderate CD61, in addition to moderate CD13, dim CD33 and CD7; they were negative for CD34 and CD11C at this time. The lineage of the leukaemia cells was apparently converted to megakaryoblastic lineage (AML‐M7). Conventional cytogenetics detected a tetraploid clone in four metaphase cells: 92<4n>,XXYY[4]/46,XY[14]. NGS analysis showed the same mutation profile as the previous analysis with a similar variant allele frequency (VAF) in each mutation (Table 1). Given the lineage switch, the patient received re‐induction with a different therapeutic protocol, high‐dose daunorubicin plus mitoxantrone (7 + 3 regimen). Despite aggressive management, the patient succumbed to severe pancytopaenia and a sudden intracranial haemorrhage 30 days after the induction.

Acute myeloid leukaemia in this case apparently underwent an intra‐myeloid lineage switch during the induction chemotherapy. Because of phenotypic distinction, a novel de novo clone needs to be excluded for the subsequent leukaemia of different lineage by comparing genomic profiles between the two neoplasms. In the present case, a clonal relationship between pretreatment AML with maturation and subsequent AML with megakaryoblastic differentiation was confirmed by the same somatic mutation profile with similar VAFs between the two (Table 1). The additional karyotypic abnormality seen in the subsequent megakaryoblastic leukaemia likely represents a clonal evolution or may be explained by subclone heterogeneity. The mechanism of lineage switch is currently unclear despite multiple case reports, case series and experimental studies addressing this issue. Hypotheses include theories of common progenitor, dedifferentiation and transdifferentiation (Rossi et al., 2012; Wu et al., 2017; Hanley et al., 2018; Nomani et al., 2019). In these theories, the original lineage of leukaemic cells can be re‐directed to a different lineage via changes in genetic or epigenetic factors within an individual neoplastic cell, or via changes in transcription regulators or microenvironments. Alternatively, the original leukaemia may contain more than one subclone, each with various differentiation potentials, one of which could predominate in the initial presentation. Another subclone with an alternative differentiation potential might emerge after chemotherapy eradicates the dominant leukaemia clone (Wu et al., 2017; Hanley et al., 2018; Nomani et al., 2019). This theory of clonal selection seems feasible to explain lineage switch in the present case, given its close association with induction chemotherapy. In this clonal‐selection theory, the neoplastic progenitor cell in our present case, probably at the stage of committed myeloid progenitor or pluripotent stem cell, generated a repertoire of heterogeneous subclones with various commitments. The subclone with granulocytic differentiation potential initially became dominant by gaining growth advantage, which overwhelmed the other subclones, including that with megakaryocytic differentiation potential. Because of its probably dormant status, the subclone with megakaryocytic differentiation potential might initially exist but be morphologically invisible and immunophenotypically undetectable at the first presentation. When the original dominant clone was eradicated by chemotherapy, the dormant subclone with megakaryoblastic differentiation could proliferate and emerge as the major clone due to its apparent resistance to ongoing chemotherapy.

According to the literature, lineage switch in acute leukaemia predicts refractoriness to conventional chemotherapy, via changes in clonal phenotype, and heralds a dismal clinical outcome (Wu et al., 2017). While a standard treatment for acute leukaemia with lineage switch has not been established, the proposed protocol suggests that the regimens be tailored by adjusting to the converted lineage (Gerr et al., 2010). Although remission may be achieved, relapse, either of the initial or converted lineage often follows a short interval of remission (Rossi et al., 2012; Wu et al., 2017). Along with a deeper understanding of the changes in genomic landscape associated with lineage switch, which may shed light on the pathogenesis underlying this unusual disease process, a novel treatment protocol may be introduced to control this unusual leukaemia with dismal outcome.

In summary, we report a unique case of AML with lineage switch from AML with maturation (AML‐M2) to AML with megakaryoblastic differentiation (AML‐M7). The pathogenesis underlying this lineage switch remains unclear. While a few hypotheses have been proposed for the mechanism, this case underscores the theory of clonal selection, given its close association with chemotherapy.

在急性白血病中，谱系转换很少见，文献仅限于病例报告或小样本报道（Sakaki等，2009； Park等，2011； Rossi等，2012； Della Starza等，2016； Rayes等。 。，2016 ;吴等人，2017 ;汉利。等人，2018 ; Nomani 。等人，2019）。大多数报道的病例是在B细胞急性淋巴细胞白血病（B‐ALL）和急性髓细胞性白血病（AML）之间切换（Sakaki等。，2009年; Park等。，2011 ; 罗西等。，2012 ; Della Starza等。，2016年; Rayes等。，2016）涉及小儿患者赖氨酸甲基转移酶2A（KMT2A）的重排（Sakaki等，2009 ; Park等，2011 ; Rossi等，2012 ; Rayes等，2016）。AML切换到其他谱系极为罕见，但在极少数情况下已有报道（Hanley等。，2018 ; Nomani等。，2019）。在英国文献中从未描述过具有粒细胞分化转换为巨核细胞表型的AML。在本文中，我们报道了一种不寻常的AML成熟病例（法国-美国-英国[FAB]分类为AML-M2），在诱导化疗期间通过形态学和免疫表型特征转变为急性巨核细胞白血病（FAB称为AML-M7）。

该患者是一名62岁的男性，患有贫血，血小板减少和明显的白细胞增多症（179×10 ⁹/ l），循环爆炸次数显着增加（64％）。随后的骨髓活检（活检1）显示出67％的胚泡（图1，A–C上部），其中髓过氧化物酶，CD31，CD34阳性的比例很小，而E-钙黏着蛋白和因子8相关抗原的阴性（图1）。 1，上图C）通过免疫组织化学。中性粒细胞和成熟的髓样前体占全部分析事件的约20％。可见散在的巨核细胞，但在该活检中未发现增生异常的巨核细胞增加。流式细胞仪（图1，上图D–F）显示58％的胚细胞表达分化分化13（CD13），昏暗CD33，昏暗CD11C，CD123，部分CD34，HLA-DR和异常CD7的暗团；它们对CD41，CD61，CD19，CD10，CD4和CD14呈阴性。染色体分析显示出正常的男性核型，并且荧光原位杂交组对髓样相关异常呈阴性。下一代测序（NGS）分析显示了四个致病突变，涉及核磷脂1（NPM1 ;外显子11上的c.860_863dupTCTG），Fms相关酪氨酸激酶3（FLT3 ;外显子20上c.2503G> T），DNA甲基转移酶3α（DNMT3A ;外显子13上的c.1490G> T）和ATM丝氨酸/苏氨酸激酶（ATM; 外显子37）基因上的c.5558A> T。诊断为具有成熟的急性髓细胞白血病（AML-M2），患者接受标准诱导化疗（阿糖胞苷加柔红霉素； 7 + 3方案）。诱导后第16天，骨髓活检（活检2）显示出35％的胚泡和发育异常的巨核细胞的大量增殖，多数为小次叶状（图1，下部A，B）；通过免疫组织化学，原始细胞的8因子相关抗原（图1，C小图）和CD31呈阳性，而髓过氧化物酶，E-cadherin和CD34呈阴性。流式细胞仪（图1，下图D–F）显示24％的胚细胞，除了中度CD13，中度CD33和CD7外，对明亮的CD41和中度CD61呈阳性。他们此时对CD34和CD11C阴性。白血病细胞的谱系显然转变为巨核细胞谱系（AML-M7）。常规细胞遗传学在四个中期细胞中检测到四倍体克隆：92 4n，XXYY [4] / 46，XY [14]。NGS分析显示与以前的分析相同的突变谱，每个突变中的变异等位基因频率（VAF）相似（表1）。在改变血统后，患者接受了另一种治疗方案的重新诱导，即大剂量柔红霉素加米托蒽醌（7 + 3方案）。尽管采取了积极的治疗措施，但患者在诱导后30天仍然屈服于严重的全血细胞减少症和突然的颅内出血。NGS分析显示与以前的分析相同的突变谱，每个突变中的变异等位基因频率（VAF）相似（表1）。根据血统的改变，患者接受了另一种治疗方案的重新诱导，即大剂量柔红霉素加米托蒽醌（7 + 3方案）。尽管采取了积极的治疗措施，但患者在诱导后30天仍然屈服于严重的全血细胞减少症和突然的颅内出血。NGS分析显示与以前的分析相同的突变谱，每个突变中的变异等位基因频率（VAF）相似（表1）。根据血统的改变，患者接受了另一种治疗方案的重新诱导，即大剂量柔红霉素加米托蒽醌（7 + 3方案）。尽管采取了积极的治疗措施，但患者在诱导后30天仍然屈服于严重的全血细胞减少症和突然的颅内出血。

在这种情况下，急性髓性白血病显然在诱导化疗期间经历了髓内谱系转换。因为表型的区别，一个新颖的从头通过比较两个肿瘤之间的基因组图谱，需要排除克隆，以免随后发生不同谱系的白血病。在本例中，成熟的预处理AML与随后的巨核细胞分化AML之间存在克隆关系，这是由于两者之间具有相同的体细胞突变谱和相似的VAF所致（表1）。在随后的巨核细胞白血病中看到的其他核型异常可能代表克隆进化，或可以由亚克隆异质性解释。尽管有多个案例报告，案例系列和针对此问题的实验研究，但谱系转换的机制目前尚不清楚。假设包括共同祖细胞，去分化和转分化的理论（Rossi et al。，2012; Wu等。，2017 ; Hanley等。，2018 ; Nomani等。，2019）。在这些理论中，白血病细胞的原始谱系可以通过单个赘生性细胞内遗传或表观遗传因子的变化或转录调节剂或微环境的变化而重新定向为不同谱系。或者，原始白血病可能包含一个以上的亚克隆，每个亚克隆具有不同的分化潜能，其中一种可能在最初的表现中占主导地位。化疗根除白血病的优势克隆后，可能会出现另一个具有替代分化潜能的亚克隆（Wu等人，2017; Hanley等。，2018 ; Nomani等。，2019）。考虑到克隆选择与诱导化学疗法的密切联系，这种克隆选择理论似乎可以用来解释本例中的谱系转换。在这种克隆选择理论中，在我们目前的情况下，可能是在定型的骨髓祖细胞或多能干细胞阶段，肿瘤祖细胞产生了具有各种作用的异源亚克隆库。具有粒细胞分化潜能的亚克隆最初通过获得生长优势而占优势，这使其他亚克隆不堪重负，包括具有巨核细胞分化潜能的亚克隆。由于其可能处于休眠状态，可能最初存在具有巨核细胞分化潜能的亚克隆，但在首次出现时在形态学上是不可见的，并且在免疫表型上无法检测到。

根据文献，急性白血病的谱系转换通过克隆表型的变化预测了对传统化疗的难治性，并预示了令人沮丧的临床结果（Wu et al。，2017）。虽然尚未建立具有谱系转换的急性白血病标准治疗方法，但拟议的方案建议通过调整转化谱系来调整治疗方案（Gerr等人，2010）。尽管可以实现缓解，但初始谱系或转化谱系的复发通常都在很短的缓解间隔后进行（Rossi等，2012 ; Wu等，2017）。随着对与谱系转换相关的基因组格局变化的更深入了解，这可能为这种异常疾病过程的潜在发病机理提供了线索，可以引入新的治疗方案来控制这种异常白血病，结果令人沮丧。

总而言之，我们报告了一种独特的AML案例，其谱系从具有成熟的AML（AML-M2）到具有巨核细胞分化（AML-M7）的AML。这种谱系转换的潜在发病机制仍不清楚。尽管已经为该机制提出了一些假设，但鉴于其与化学疗法的密切联系，本案例强调了克隆选择的理论。