Molecular Diagnosis of Hematopoietic Neoplasms

Acute Promyelocytic Leukemia: t(15;17)(q22;q21)/PML-RARA Abnormality

Acute promyelocytic leukemia (APL), or AML-M3 (according to the previous French-American-British [FAB] classification of AML) accounts for 5% to 10% of de novo AML. Patients typically present with symptoms from peripheral blood cytopenias (e.g., anemia, absolute neutropenia, and thrombocytopenia) and have a high propensity for life-threatening coagulopathy. Morphologically, APL consists of a proliferation of hypergranular promyelocytes and myeloblasts; however, a microgranular variant form is also frequently encountered (Mantadakis et al, 2008). Immunophenotypically, the cells are characterized by a lack of CD34 and HLA-DR. APL is genetically defined by the presence of the t(15;17)(q22;q12-21) abnormality, resulting in the fusion of the retinoic acid receptor-α gene, RARα (also called RARA) [17(q21)] with the PML gene [15(q22)] to form a PML-RARA chimeric gene on the derivative chromosome 15q (Grignani et al, 1994; Jurcic et al, 2007).

The RARA gene product is a subunit of a heterodimeric nuclear receptor for the naturally occurring ligand retinoic acid, or vitamin A. The retinoid nuclear receptor pathway is functional in many aspects of normal cell proliferation and differentiation (Chambon, 1996; Collins, 2008; Mark et al, 2009). PML encodes a DNA-binding zinc finger protein and associates with several other proteins in a macromolecular complex, forming discrete nuclear bodies (Dyck et al, 1994; Weis et al, 1994; Reineke & Kao, 2009). This interesting protein appears to have multiple possible cellular functions, including transcriptional regulation, apoptosis, and possibly immune surveillance (Quignon et al, 1998; Wang et al, 1998; Grimwade, 1999; Zhong et al, 2000; Borden & Culjkovic, 2009; Reinecke & Kao, 2009; Gabali, 2013). The hybrid PML-RARα protein is clearly involved in disrupting numerous intracellular processes, primarily resulting in a lack of terminal differentiation of neoplastic myeloid precursors beyond the promyelocytic stage. Nonetheless, the significance of APL as a “therapeutic paradigm” for leukemia resides in the ability of pharmacologic doses of all trans retinoic acid (atRA) to induce leukemic cell differentiation in vivo. When used in combination with cytotoxic chemotherapy, the synergistic effects of atRA induce complete remission in 90% to 98% of patients and prolonged remission in more than 80% of patients (Tallman et al, 1997; 2002; Clavio et al, 2009; Licht, 2009; Sanz et al, 2009; Patel et al, 2012). This remarkable finding has intensified efforts to identify other avenues for “cytodifferentiative” therapy in AML. Even though the effect of atRA would seem superficially related to the interaction with the RARα moiety in the abnormal PML-RARα protein, further investigations have unraveled the basic molecular mechanisms underlying the success of this agent. The normal retinoic acid receptor, when not bound to its retinoic acid ligand, associates with histone deacetylase (HDAC) in a nuclear protein corepressor complex (Guidez et al, 1998; Melnick & Licht, 1999; Lefebvre, 2001; Wei, 2004). This interaction reduces the accessibility of chromatin to transcription factors (and thus locus-specific transcriptional activity), yet is reversible in normal marrow cells upon exposure to physiologic concentrations of retinoic acid. The PML-RARα oncoprotein stabilizes and enhances the state of transcriptional repression by this multiprotein complex; however, therapeutic doses of atRA appear to overcome this abnormal condition and relieve the cellular differentiation block (Melnick & Licht, 1999; Hormaeche & Licht, 2007; Collins, 2008; Licht, 2009). Whereas some evidence suggests that the PML-RARα oncoprotein may also disrupt the normal proapoptotic functions of PML, thus potentiating transformation via pathways involving both partners in the translocation (PML-RARA and RARA-PML), the direct contribution of altered PML to leukemogenesis remains controversial (Strudwick & Borden, 2002; Collins, 2008; Brown et al, 2009). Functional haploinsufficiency of one remaining normal PML gene may also be important in the pathobiology of APL.

The molecular anatomy of the PML-RARA fusion gene and mRNA products in APL is summarized diagrammatically in Figure 76-1, A. The PML gene exhibits breakpoint heterogeneity in that one of three possible break sites can be encountered in any given patient, involving either intron 3 or intron 6, or occurring within exon 6 (Grignani et al, 1994; Gallagher et al, 1995; Reiter et al, 2003). In contrast, the RARA breakpoints are uniformly distributed in intron 2 of the gene. Thus one of three possible PML-RARα chimeric mRNAs can result from this genetic fusion: PML exon 6/RARα exon 3 (long (L)-form, or BCR 1), PML exon 3/RARα exon 3 (short (S)-form, or BCR 3) and PML exon 6Δ/RARα exon 3 (variable [V]-form, or BCR 2). The V-form transcript is unique in that the PML break occurs within exon 6 and a variable proportion of this exon is retained, although additional nucleotides may be added or deleted (Gallagher et al, 1995; Reiter et al, 2003). Notably, an in-frame fusion mRNA is produced in each case of APL, underscoring the critical requirement for the PML-RARα protein in leukemogenesis. The L-form (BCR 1) and S-form (BCR 3) PML-RARα fusions are most commonly found in APL (~45% to 50% of cases each), whereas the V-form (BCR 2) is only rarely encountered (~5% of APL). The strategy for RT-PCR amplification of these PML-RARα transcripts is depicted in Figure 76-1, B, as is a representative diagnostic PCR assay (Fig. 76-1, C). Of note, the L-form (BCR 1) and V-form (BCR 2) type transcripts show alternative splicing out of exons 5 and 6, producing three major amplified products when a PML exon 3 primer is employed.

The clinical relevance in detecting the PML-RARα fusion abnormality in cases of suspected APL is evident from the efficacy of administering specific therapy (atRA plus anthracycline-based chemotherapy for induction and consolidation) with a high possibility of long-term remission and survival. To this end, despite the tight correlation between the t(15;17)/PML-RARA and APL morphology, it is recognized that other rare APL-like myeloid leukemias do occur. These morphologic mimics harbor alternative translocations involving the RARA gene with fusion partner genes other than PML. Examples of these variants include the t(11;17)/ZBTB16(PLZF)-RARA, t(11;17)/NUMA1-RARA, t(5;17)/NPM1-RARA, and the t(17;17) STAT5B-RARA acute myeloid leukemias (Melnick & Licht, 1999; Grimwade et al, 2000; Sainty et al, 2000; Zelent et al, 2001; Redner, 2002, Balusu et al, 2011). Notably, ZBTB16(PLZF)-RARA and STAT5B-RARA positive leukemias in particular are not responsive to the differentiating effects of atRA and are associated with less favorable outcome (Jansen & Löwenberg, 2001; Redner, 2002). Molecular assays specific for the common PML-RARα fusion will not detect these variant transcripts. Other morphologic AML subtypes characterized by increased promyelocytes are similarly atRA unresponsive and also require distinction from true PML-RARA positive APL. Several studies have assessed the possible clinical significance of the PML-RARα transcript type in APL patients. Both the S-form (BCR 3) and rare V-form (BCR 2) PML-RARα fusions have been associated with adverse features such as higher presentation white blood cell count, poor atRA response, and possibly shorter remission duration (Vahdat et al, 1994; Gallagher et al, 1995; 1997; Jurcic et al, 2001; Gupta et al, 2004; Patel et al, 2012); however, the independent prognostic value of molecular subclassification in APL has not been definitively established.

Finally, the PML-RARα transcript serves as a valuable molecular disease marker to follow therapeutic response (Fig. 76-1, D). Patients with APL treated with combination chemotherapy and atRA have a relatively favorable prognosis, yet disease relapses occur frequently. The use of qualitative RT-PCR methods to detect the PML-RARα fusion mRNA (with a typical sensitivity of 10^–3 to 10^–4) was initially found to be a very powerful tool for predicting relapse risk in individual patients (Jurcic et al, 2001; Grimwade & Lo Coco, 2002; Lo Coco et al, 2002). The timing, or phase of disease therapy, is an important consideration in this leukemia. Patients evaluated for PML-RARα at the end of induction are often found to be positive, and the predictive value at this time point is not significant. By the end of consolidation therapy, however, PCR positivity is strongly predictive of relapse in patients who have apparently achieved clinical complete remission, in that nearly all such patients will progress to hematologic relapse within months. In contrast, a single PCR-negative measurement at this time point is not necessarily predictive of favorable outcome, in that a significant number of such patients may also suffer relapse. This has led to recommendations for more frequent posttherapy monitoring using sensitive and precise real-time quantitative PCR methods (RQ-PCR). To this end, several APL study groups have reported results of serial PML-RARα monitoring using RQ-PCR techniques, and these efforts have shown a benefit for detecting low-level molecular disease and its value in predicting relapse risk (Grimwade & Lo Coco, 2002; Gallagher et al, 2003; Grimwade et al, 2009; 2010; Borthakur et al, 2011). Furthermore, several studies have validated the concept of “salvaging” or retreating patients with molecular residual disease, with good clinical outcomes. Given an array of additional treatment options for APL, including highly effective second-line intervention with arsenic trioxide (As2O3) and autologous or allogeneic stem cell transplantation, molecular residual disease assessment thus forms an integral aspect of the management of all APL patients (Santamaria et al, 2007; Kohno et al, 2008; Lo Coco et al, 2008; Grimwade et al, 2009; Paschka et al, 2012; Patel et al, 2012). Although no formal guidelines have been firmly established for RQ-PCR monitoring of PML-RARα in APL, proposed common sampling time points for assessment include the end of induction, the end of consolidation, and then every 2 to 3 months for the first year after therapy, because this is the window during which most relapses occur. Bone marrow aspirate samples are preferred for MRD assessment, in that the peripheral blood is less sensitive for transcript detection following treatment onset.

Core Binding Factor–Related Acute Myeloid Leukemias: t(8;21)(q22;q22)/RUNX1-RUNX1T1 and Inv(16)(p13q22) or t(16;16)(p13;q22)/CBFB-MYH11 Abnormalities

Acute myeloid leukemias with the t(8;21)(q22;q22) and inv(16)(p13q22) or related t(16;16)(p13;q22) cytogenetic abnormalities together account for approximately 11% to 18% of de novo cases (Schnittger et al, 2007; Cheng et al, 2009). Core binding factor (CBF)–related AML cases are roughly evenly distributed between those with the t(8;21) and cases with inv(16)/t(16;16) (Dombret et al, 2009). Whereas the t(8;21) is generally associated with AML showing maturation (FAB type AML-M2), the inv(16) and t(16;16) are highly correlated with acute myelomonocytic leukemia with abnormal eosinophils (FAB type AML-M4Eo) (Le Beau et al, 1983; Cheng et al 2009). Clinically, these genetically defined AML subtypes are responsive to chemotherapy (especially high-dose cytarabine regimens) and are considered to be prognostically favorable compared with AML in general (Appelbaum et al, 2006; Dombret et al, 2009; Paschka et al, 2012; Patel et al, 2012).

Although different respective leukemia phenotypes derive from these translocations, a remarkable and common pathobiologic feature of both the t(8;21) and inv(16) or t(16;16) leukemias is disruption of the CBF transcriptional regulatory pathway (Speck et al, 1999; Paschka, 2008). CBF is a heterodimeric transcription factor that consists of a DNA binding α-subunit (CBFα) and a peptide-interacting β-subunit (CBFβ), which acts to stabilize CBFα at sites of DNA interaction (Fig. 76-2). In normal cells, CBF acts at “core” enhancer sequences in a number of genes required for proper myeloid and lymphoid cell differentiation and/or maturation. The binding of CBF facilitates the access of other transcription factor complexes to these chromatin sites, in part through increased acetylation of DNA-bound histone proteins by histone acetyltransferases (Lorsbach & Downing, 2001; Yamagata et al, 2005). In human acute myeloid leukemias with the t(8;21), the RUNX1 gene (formerly designated AML1, or CBFA2) encoding an isoform of CBFα is joined to a putative transcription factor, RUNX1T1 (formerly ETO, or MTG8) to form the RUNX1-RUNX1T1 fusion (Downing, 1999; Peterson et al, 2007; Patel et al, 2012). In the case of the inv(16) or t(16;16), the gene producing CBFβ (i.e., CBFB) is juxtaposed to a smooth muscle myosin heavy chain gene MYH11 to form the hybrid CBFB-MYH11 (Liu et al, 1995; Mrózek et al, 2008; Pratcorona et al, 2012). In either case, profound disruption of the normal cellular differentiation program is thought to be central to the causation of acute leukemia. The prominence of CBF pathway alterations in leukemogenesis is additionally emphasized by the finding of RUNX1 gene translocations in several other types of leukemia and myelodysplasia (see Fig. 76-2), including the t(12;21)/ETV6-RUNX1 abnormality, the most common single genetic abnormality in pediatric ALL.

From a mechanistic viewpoint, the RUNX1-RUNX1T1 leukemic fusion protein is thought to act in a dominant negative manner to normal RUNX1 (i.e., CBFα) by recruiting or stabilizing elements of transcriptional repression, including histone deacetylase complexes, at critical DNA sites (Lorsbach & Downing, 2001; Yamagata et al, 2005). Normal gene expression patterns are also disrupted by the altered affinity of the abnormal fusion protein for specific core enhancer sequences, including an increased propensity to bind sites with such sequences in duplicate (Okumura et al, 2008). CBFβ-MYH11 chimeric protein sequesters normal CBFα protein in the cytosol, thereby abrogating heterodimeric CBF assembly in the nucleus, but may also create an abnormally repressive transcriptional complex in a manner analogous to that of RUNX1-RUNX1T1 (Shigesada et al, 2004; Pratcorona et al, 2012). In either case, the resultant inhibition of transcription at key genes alters the genetic program for normal proliferation and differentiation in hematolymphoid stem or progenitor cells.

In keeping with the concept outlined previously that an individual genetic aberrancy is of itself not likely sufficient for carcinogenesis, transgenic mice engineered to conditionally express a RUNX1-RUNX1T1 oncogene develop acute myeloid leukemia only on subsequent exposure to a promoting mutagen (Lorsbach & Downing, 2001; Peterson et al, 2007; Müller et al, 2008). In CBF leukemias, the pathognomonic translocations create a block in cellular differentiation (class II mutations), but affected cells often acquire secondary class I mutations that endow them with enhanced proliferative capacity. Indeed, there is a particularly strong association in CBF leukemias with secondary mutations to the gene encoding the c-KIT tyrosine kinase (stem cell factor receptor), a potent regulator of cell growth. Up to half of CBF leukemias carry KIT mutations; in contrast, approximately 5% of all AML cases show such mutations (Müller et al, 2008). Many of these KIT mutations involve the D816 “hot-spot” amino acid (that is also affected in systemic mastocytosis), but mutations have been described in other exons as well. The presence of KIT mutations in CBF leukemias has been associated with increased relapse risk and, in some studies, poorer overall survival, although the latter finding remains somewhat controversial at present (Cairoli et al, 2006; Paschka et al, 2006; Mrózek et al, 2008; Müller et al, 2008; Faderl et al, 2011). To date, the routine clinical utility of KIT mutation testing in CBF leukemias currently awaits validation by larger trials. Notably, deregulation of c-KIT expression as a consequence of KIT mutations in these leukemias may present a rational therapeutic target, given that the 5-year-survival for CBF leukemia patients, albeit better than for AML in general, does not substantially surpass 50%.

The molecular rearrangements underlying the RUNX1-RUNX1T1 fusion are such that the breakpoint-fusion sites invariably involve the same limited set of intron regions in both genes, resulting in a consistent in-frame RUNX1-RUNX1T1 mRNA molecule in each patient with this type of AML, although some degree of variant transcripts derived from alternative splicing and differential promoter usage can also be seen (Zhang et al, 2002; Lafiura et al, 2007; Pratcorona et al, 2012). In contrast, standard RT-PCR detection of the CBFβ-MYH11 chimeric transcript is complicated by the potential for marked breakpoint heterogeneity, mainly in the MYH11 gene, as well as by rare alternate breakage sites described in CBFB. In all, more than 10 CBFB-MYH11 fusion transcript forms have been identified to date (Liu et al, 1995; Viswanatha et al, 1998; Kadkol et al, 2004; Schnittger et al, 2007). Despite this complexity, the CBFB-MYH11 gene fusion can be readily detected by RT-PCR, because approximately 90% of these tumors harbor the so-called “type A” chimeric mRNA, characterized by fusion of CBFB nucleotide 495 to MYH11 nucleotide 1921 to form a transcript of relatively short length (Schnittger et al, 2007). More comprehensive PCR strategies have also emerged to detect the majority of the other rare CBFB-MYH11 fusion types, each of which account for the remainder of cases of inv(16) or t(16;16) AML (Kadkol et al, 2004). RT-PCR analysis for the CBFB-MYH11 abnormality is often advantageous, in that standard cytogenetic interpretation may be uninformative in some cases (Merchant et al, 2004; Monma et al, 2007). In this context, it is possible, although unlikely, that a case with a true non–type A CBFB-MYH11 fusion could be negative by both conventional cytogenetic and molecular analysis, if karyotyping fails to detect a cryptic translocation and RT-PCR technique targets only the type A transcript; this possibility underscores the general importance of maintaining awareness of the limitations of a specific molecular assay, as well as the central role that morphologic assessment must play even in the molecular era. Nevertheless, given the generally more favorable clinical outcome for both of these AML subtypes, molecular diagnosis can play a significant role both in initial identification and disease monitoring following treatment (Pratcorona et al, 2012).

In AML the presence of rearranged MLL gene is considered an unfavorable prognostic indicator and is observed in AML, B-ALL (further discussed below), and myelodysplastic syndrome. The translocated MLL gene is reported in de novo AML, around 3% to 4%, as well as in therapy-related acute myeloid leukemia (t-AML). Numerous partners have been reported to partner with the MLL gene, but in around 80% of cases, the partner will include t(4;11)(q21;q23)/AF4-MLL, t(9;11)(q21;q23)/MLLT3-MLL, or t(11;19)(q23;p13)/MLLT1-MLL. The latter two MLL gene partners are also the most common balanced translocation involving the MLL gene in t-AML, particularly in patients with a previous, relatively short history of receiving topoisomerase II inhibitor therapy. The most common breakpoints reported in the translocated MLL gene are present between exon 5 and 11 of the gene. Some of the translocations/insertions can be too small and cryptic, thus limiting their detectability by conventional karyotyping. FISH analysis, using break-apart probes, is widely used to detect MLL abnormalities and can detect gene amplification or mutant MLL when the partner is unknown. RT-PCR is the most sensitive approach for detecting specific subtypes of MLL rearrangements when the partner gene is known (Ramchandren et al, 2013).

Other Gene Mutations in Acute Myeloid Leukemias

Mutations involving many other genes, including WT1 (see below), MLL, TET2, JAK2, IDH1/IDH2, EZH2, PLK1, NRAS, and KRAS, have been associated with prognostic or biologic importance in AML. In addition, abnormal gene expression patterns of BAALC, ERG, and MN1 have also been tied to prognosis (Baldus et al, 2007; Ernst et al, 2010; Dang et al, 2010; Abbas et al, 2010; Hart et al, 2011; Chotirat et al, 2012; Ernst et al, 2012; Weissman et al, 2012; Benetatos et al, 2013). As this list continues to expand, a major challenge for the diagnostician and clinician will be in the rational interpretation of potentially many interacting genetic factors in order to determine appropriate risk stratification and therapy options for individual AML patients. As suggested from the inherent complexity of simultaneously assessing the three relatively common mutations considered previously (FLT3, NPM1, and CEBPA), the impact of multiple cooperating genetic events requires sophisticated bioinformatics analyses applied to well-designed and sufficiently powered clinical studies. Basically, these mutations are detected by PCR amplification and mutational analysis by sequencing and by comparison of these sequences with the published unmutated sequence (Ramchandren et al, 2013). The optimal strategy for evaluation of gene mutations in AML will thus continue to evolve as additional clinical data emerge (Hatzimichael et al, 2013).

Major Translocation Fusion Gene Abnormalities in B Cell ALL

The t(9;22)(q34;q11.2)/1BCR-ABL1 is found in approximately 3% to 4% of childhood B-lineage ALL, but it occurs in 20% to 25% of adult B-ALL (Armstrong & Look, 2005). Typically, these patients present with markedly elevated lymphoblast counts and other adverse clinical features. Of the two common break-fusion events associated with the BCR-ABL1 fusion gene, the majority of cases of pediatric B-ALL (80% to 90%) have BCR breakpoints situated in the minor breakpoint cluster region (m-BCR), with production of an e1-a2 type chimeric BCR-ABL1 mRNA (p190). The e1-a2 mRNA type is also found in many adult BCR-ABL1 positive B-ALL, but approximately one-third of adult cases alternatively demonstrate e13-a2 or e14-a2 transcripts, characteristic of major breakpoint cluster region (M-BCR) disruption in the BCR gene (p210). The structure and molecular diagnostic aspects of the BCR-ABL1 gene fusion are presented in greater detail in the subsequent section on CML. The presence of the BCR-ABL1 in childhood ALL is an independently poor prognostic factor, placing these patients among the very highest at risk for primary treatment failure and relapse (Fletcher et al, 1991; Gaynon et al, 1997; Jones & Saha, 2005; Yanada et al, 2009). The majority of BCR-ABL1 positive ALL patients are candidates for more aggressive therapy, including early allogeneic hematopoietic stem cell transplantation. The introduction of the tyrosine kinase inhibitor imatinib mesylate may offer additional therapeutic benefit in the treatment of this disease; in adult Ph+ALL, imatinib in combination with conventional chemotherapy results in a very high (95%) rate of initial complete remission, although durable responses are often difficult to maintain (Jones & Saha, 2005; Gökbuget & Hoelzer, 2009; Ribera et al, 2009; Vrooman & Silverman, 2009; Yanada et al, 2009; reviewed in Gruber et al, 2009; Shabbir & Stuart, 2010). Studies of the efficacy of imatinib in childhood Ph+ ALL are promising (Jones & Saha, 2005; Vrooman & Silverman, 2009), so in 2013, the U.S. Food and Drug Administration (FDA) approved the use of imatinib to treat children newly diagnosed with Ph+ALL. Molecular analysis of the BCR-ABL1 transcript by the RT-PCR technique is critical for predicting risk in childhood ALL, as well as providing a marker for residual disease monitoring. Deletion of IKZF1, which codes for the transcription factor Ikaros, has been identified as a genetic change present in many cases of BCR-ABL1 ALL (Mullighan et al, 2008; 2009; Iacobucci et al, 2009; Martinelli et al, 2009), and Ikaros alterations appear to be an important cooperative lesion in this subclass of ALL, strongly influencing the aggressive behavior of BCR-ABL1 ALL (Martinelli et al, 2009; Mullighan et al, 2009).

B-ALL characterized by the t(1;19)(q23;p13.3) abnormality are uncommon, accounting for less than 5% of pediatric cases and only rare occurrences in adults (Armstrong & Look, 2005). This translocation is strongly associated with “pre-B” immunophenotypic features, typified by the presence of cytoplasmic IgM production in the tumor cells. Following its initial recognition in childhood ALL, the t(1;19) was associated with poor outcome. However, it has become apparent that more intensive treatment protocols largely negate the adverse effects of this genotype, resulting in stable remissions for most of these individuals. Recognition of this translocation is thus required for appropriate therapy (Schultz et al, 2007). The t(1;19) most frequently occurs as an unbalanced form of translocation (with resultant loss of some genetic material) and produces the chimeric TCF3 (formerly E2A)-PBX1 gene fusion, although balanced translocations with the same fusion product also occur (Hunger et al, 1991; Paulsson et al, 2007). The TCF3 gene encodes a series of basic helix-loop-helix transcription factors, which are involved in myocyte and B-lymphocyte development. PBX1 is a DNA-binding transcription factor that is not normally expressed in lymphoid cells. The fusion TCF3-PBX1 protein replaces the DNA binding and protein dimerization motifs of TCF3 with the DNA binding region of PBX1, predisposing the progenitor B cell to neoplastic proliferation. The location of genomic breakpoints within TCF3 and PBX1 introns is consistent in essentially all cases, resulting in the formation of a single TCF3-PBX1 transcript (Hunger et al, 1991; Wiemels et al, 2002a), with only few reported variants (Paulsson et al, 2007). This fusion mRNA is readily amenable to RT-PCR detection and can also serve as a disease-specific marker for sensitive detection in the bone marrow. Of note, rare cytogenetically detectable ALL cases with t(1;19) lack the TCF3-PBX1 fusion and appear to be associated with an inferior clinical outcome (Privitera et al, 1992; Prima & Hunger, 2007); these variant t(1;19)(6q23;p13.3) translocations instead create reciprocal oncogenic fusion proteins, DAZAP1/MEF2D and MEF2D/DAZAP1 (Prima & Hunger, 2007).

Overall, 20% to 25% of pediatric ALL patients are characterized by the presence of the t(12;21)(p13;q22) abnormality, establishing this lesion as the most common recurrent translocation in this patient group (Armstrong & Look, 2005). This genetic finding is essentially absent among adults with B-ALL. The t(12;21) is cytogenetically cryptic in almost all cases (Karrman et al, 2006), instead often being suggested by an apparent deletion of the short arm of one chromosome 12. Although patients with t(12;21) ALL are found distributed among different clinical risk groups, in general these individuals are between the ages of 2 and 7 and have diploid karyotype tumors (Rubnitz et al, 1997a; Forestier & Schmiegelow, 2006; Forestier et al, 2007). The t(12;21) results in the joining of the 12p locus ETV6 gene (also known as TEL) with the RUNX1 (or AML1, CBFA2) gene on 21q to form the chimeric gene fusion ETV6-RUNX1 on the derivative chromosome 12 (Romana et al, 1995; Zelent et al, 2004). Pediatric patients with ETV6-RUNX1 positive ALL have a favorable outcome, with excellent disease-free remissions approaching 80% (similar to the aforementioned high-hyperdiploid group with specific chromosomal trisomies) (Shurtleff et al, 1995; Borkhardt et al, 1997; Rubnitz et al, 1997a; 1997b; Forestier et al, 2008). However, reports of a substantial late relapse rate among some ETV6-RUNX1 ALL patients have refocused attention on more clearly defining the risk characteristics of this group, even though many such relapsed patients remain responsive to “salvage therapy” protocols (Harbott et al, 1997; Seeger et al, 1998; Ford et al, 2001; Forestier et al, 2008). The molecular pathology of ETV6-RUNX1 ALL is related to disruption of normal CBF activity, leading to enhanced transcriptional repression of many target genes, as similarly discussed for the RUNX1-RUNX1T1 abnormality in AML (Fig. 76-3) (Zelent et al, 2004). In almost 90% of ETV6-RUNX1 ALL cases, the remaining ETV6 allele is partially or completely deleted in a variable proportion of tumor cells, suggesting that this event may be an important secondary event for leukemogenic transformation, through complete loss of normal ETV6 function (Raynaud et al, 1996; Tsuzuki et al, 2007; Wiemels et al, 2008). At the DNA level, ETV6 and RUNX1 breaksites in ALL occur most often in the same genomic regions of ETV6 intron 5 and RUNX1 intron 1 (Romana et al, 1995; Von Goessel et al, 2009), with evidence of some propensity for breakpoint clustering. A second minor variant with breakpoints situated in RUNX1 intron 2 is also described. The in-frame transcribed ETV6-RUNX1 mRNA thus encompasses a novel region joining ETV6 exon 5 to RUNX1 exon 2 (most cases) or ETV6 exon 5 to RUNX1 exon 3 (fewer cases) (Nakao et al, 1996a; Von Goessel et al, 2009). Additional complexity is present in this gene fusion owing to alternative splicing out of RUNX1 exons 2 and/or 3, with the possibility of detecting up to four ETV6-RUNX1 amplified transcript forms by RT-PCR in any given case. Molecular (RT-PCR) or FISH detection of the ETV6-RUNX1 is frequently employed in pediatric ALL, given the inability to diagnose the t(12;21) by standard cytogenetics because of its cryptic nature. The chimeric mRNA further represents a highly specific molecular marker for residual disease assessment.

Translocations involving the “mixed lineage leukemia” (MLL) gene located on chromosome 11q23 are observed in 5% to 9% of pediatric ALL cases overall and are disproportionately represented in infants younger than 1 year of age (Kaneko et al, 1986; Bartolo & Viswanatha, 2000; Armstrong & Look, 2005; Silverman, 2007; Chowdhury & Brady, 2008; Ramchandren et al, 2013). The presence of MLL gene translocations is associated with a poor outcome for these very young children and is considered an adverse finding in older children as well, particularly in the setting of a poor prednisone response (Pui et al, 2003; Chowdhury & Brady, 2008). Infant MLL-positive leukemias are of B-lineage, lack CD10 positivity, and demonstrate aberrant expression of some myeloid-associated markers, particularly CD13 and CD33 (Chen et al, 1993; Burmeister et al, 2009). Clinically, these patients have high initial leukocyte counts, organomegaly, and a tendency for involvement of the central nervous system. Translocations involving 11q23/MLL also occur in de novo AML and secondary (therapy-related) AML, and in either case are characterized by monocytic morphologic features (i.e., FAB AML-M4 or M5 types). In de novo adult AML, MLL gene rearrangements, specifically the t(9;11) MLLT3-MLL abnormality, may not represent a high-risk tumor genotype. In contrast, secondary AML arises in subsets of both pediatric and adult patients who have had prior dose-intensive chemotherapy with DNA topoisomerase II inhibitors (e.g., epipodophyllotoxins and anthracyclines); these tumors occur abruptly within a few months to years after exposure and have a very aggressive treatment-resistant course (Godley & Larson, 2008).

The pathobiology underlying MLL gene translocations in these different types of de novo and secondary acute lymphoid and myeloid leukemias is becoming better understood. The MLL protein is a large (~430 kD) DNA binding protein, which is characterized by several functional domains, indicating its role as a multifunctional transcriptional factor with histone methyltransferase activity (Hess, 2004; Li et al, 2005; Krivtsov et al, 2007; Wang et al, 2009). MLL is known to regulate several HOX (homeobox) genes involved in the coordination of proper skeletal development during mammalian embryogenesis (Yu et al, 1995; Soshnikova & Duboule, 2008). Several HOX genes also appear to be central to the development and maintenance of myeloid and lymphoid progenitor cells, thus establishing a tight link for MLL in the genetic control of normal hematopoiesis as well (Hess, 2004; Ernst et al, 2004a; 2004b; McGonigle et al, 2008). MLL gene translocations produce an altered functional form of the protein; in each instance, the N-terminal “A-T hook” DNA binding motif of MLL is retained in the chimeric protein, typically with fusion to a portion of another transcription factor (Waring & Cleary, 1997; Chowdhury & Brady, 2008; Ramchandren et al, 2013). Accumulating data indicate that abnormal expression of MLL fusion proteins can constitutively deregulate certain HOX genes, leading to hematopoietic progenitor cell immortalization and predisposition to leukemogenesis (Armstrong et al, 2003; Hess, 2004; Ono et al, 2009). Indeed, studies using chromatin immunoprecipitation to identify portions of the genome bound by MLL have further detected its presence at numerous promoter regions, suggesting a more global role for MLL in transcriptional regulation.

Some additional key biologic insights have emerged from the structural analysis of MLL gene breakpoints in different subsets of leukemias. The majority of de novo AML with MLL translocations have breaksites situated 5′ in the MLL breakpoint cluster region, whereas the preponderance of secondary AML occur in a more distal (3′) segment of this locus (Stissel-Broeker et al, 1996; Zhang et al, 2006). Infant ALL cases strikingly also tend to associate with the 3′ genomic region, suggesting a commonality in pathogenesis with secondary AML. This 3′ “hotspot” area contains a strong consensus binding site for DNA topoisomerase II, as well as an adjacent “scaffold attachment region,” which is highly susceptible to cleavage by endonucleases. Although the presence of the topoisomerase II site presents an attractive explanation for the basis of developing secondary AML (e.g., via double strand break induction by topoisomerase II inhibitors and subsequent MLL translocation rearrangements), the hypersensitive 3′ scaffold site of the gene has received further attention, because it too appears to be particularly prone to double-strand breaks in experimental models (Stissel-Broeker et al, 1996). The latter MLL region is in fact selectively targeted for cleavage during cell death in response to diverse apoptogenic stimuli (e.g., following cellular stress or DNA damage) (Betti et al, 2001; 2003; 2005; Sim & Liu, 2001; Greaves & Wiemels, 2003a). These investigations imply that exposure to agents that can induce this site-specific MLL breakage potentially set the stage for MLL gene fusions, leading to the very aggressive forms of leukemia observed in infants and after distinct types of chemotherapy. In turn, because the causal etiology is already identified for secondary AML with MLL translocations, current studies in infant ALL seek to discover possible exogenous or naturally occurring transplacental factors that may predispose developing fetal hematopoietic stem cells to MLL gene breakage and illegitimate recombination (Spector et al, 2005; Pombo-de-Oliveira & Koifman, 2006).

The MLL gene is involved in translocation events with more than 60 known partner genes (Liu et al, 2009); however, the t(4;11)(q21;q23)/MLL-AF4 abnormality is the most frequently encountered in infant ALL (75% of cases). MLL gene locus breakpoints are distributed over an 8.3-kb region of DNA encompassing exons 5 to 11 (Chowdhury & Brady, 2008). This marked breakpoint heterogeneity may also occur in the translocated gene (e.g., AF4), resulting in the possibility of several MLL hybrid gene products, typically one of which is expressed in any given patient with this leukemia. RT-PCR method can be used to identify individual MLL fusion gene transcripts, such as the MLL-AF4 chimeric mRNA; however, the combination of multiple rearranging gene partners in these leukemias and the potential for substantial genomic breakpoint variability requires comprehensive and technically challenging laboratory approaches, such as multiplex PCR assays or, in the research setting, techniques such as long-distance inverse PCR (Pallisgaard et al, 1998; Andersson et al, 2001; Meyer & Marschalek, 2009; Meyer et al, 2009). MLL gene locus rearrangements have also been successfully detected by Southern blot analysis, although the FISH technique now provides a rapid and highly efficacious alternative for this purpose (Cuthbert et al, 2000; Cavazzini et al, 2006). FISH analysis can provide very rapid confirmation of MLL gene breakage, implying a translocation event, but it cannot give information regarding the particular translocation present. Thus RT-PCR, even with the limitations described, can serve a useful purpose in the diagnosis and monitoring of these leukemias, especially because the fusion gene type can be specifically and more sensitively identified in contrast to other analytic methods. Furthermore, mutation due to partial tandem duplication involving the MLL gene (MLL-PTD) is only detected by RT-PCR.

Prenatal Origins of Childhood Leukemias

Major advances continue to be made in understanding the molecular basis of the pediatric acute leukemias. These efforts now span the realms of basic science, translational and clinical research, and, increasingly, population-based methods. Studies of concordant ALL occurring in monozygotic twins first demonstrated that the leukemias arising in each paired affected individual were genetically identical based on molecular analysis of the genomic breakpoint sequences of involved MLL, ETV6-RUNX1, or clonotypic immunoglobulin or T cell receptor gene rearrangements (summarized in Greaves et al, 2003b; Taub & Ge, 2004). Notably, the propensity to develop concordant leukemias is variable for different leukemic subtypes in twin studies, approaching 100% for MLL translocation ALL and only 10% for ETV6-RUNX1 ALL. These findings suggest a variable transformation potential for these different gene fusions in vivo, raising the issues of both the nature and timing of secondary genetic insults. The results of these investigations have been corroborated using similar experimental techniques to “backtrack” leukemia-specific DNA abnormalities to archived neonatal blood samples of individual (nontwin) children of different ages, with genetically defined subtypes of ALL or AML (Gale et al, 1997; Wiemels et al, 1999; Fasching et al, 2000; Yagi et al, 2000; Wiemels et al, 2002a; Greaves & Wiemels, 2003a; Greaves et al, 2003b; Taub & Ge, 2004; Ross, 2008). Together, these data strongly suggest that many pediatric ALL (and some AML) have origins in utero; that is, prenatal acquisition of potentially leukemogenic translocation events, or other genetic abnormalities, can occur during the time of active fetal hematopoiesis, but in most cases, additional, later-occurring genetic alterations in susceptible individuals are required in order to produce overt leukemia. The rapid evolution of MLL-associated ALL versus the longer latency (i.e., older age at onset) of ETV6-RUNX1 ALL indicates that both the biologic effects of the primary (predisposing) genetic lesion, as well as the timing and type of promoting insult(s) are complex, although within defined subtypes of leukemia, it may now be possible to better elucidate the molecular pathways resulting in tumorigenesis. Not all types of pediatric ALL, however, can be traced back to a prenatal clonal cell origin. The retrospective neonatal blood data for the t(1;19)/TCF3-PBX1 ALL subgroup instead support the concept that the majority of these patients develop this genetic aberration in the postnatal period (Wiemels et al, 2002a).

Prospective RT-PCR and FISH screening of blood samples from healthy newborn children have subsequently revealed a detection frequency of 1% for the ETV6-RUNX1 abnormality and 0.2% for the RUNX1-RUNX1T1 fusion transcript, markedly higher than the incidence of either of these leukemia subtypes in childhood (Mori et al, 2002; Olsen et al, 2006). These data demonstrate that the occurrence of at least some chromosomal translocation fusion genes is a common event in utero; however, the presence of such an abnormality is clearly insufficient to cause overt leukemia in the vast majority of individuals. Although the translocation-type acute leukemias are karyotypically characterized in most instances by an apparent “single genetic anomaly,” it is becoming evident that these tumors also conform to the “two-hit” hypothesis of Knudson proposed for solid tissue cancers (Knudson, 1992; Schindler et al, 2009; Ramchandren et al, 2013), as discussed in further detail in Chapter 75. In this concept, the initiating translocation event must be stable enough in a long-lived progenitor hematopoietic cell to confer some advantage in survival or proliferation; the acquisition of one or more subsequent genotoxic events in a susceptible individual would then lead to complete malignant transformation. The fate of such silent “translocation-positive” cells is thus an important parameter, not only for understanding disease pathogenesis but also in a practical setting for the appropriate interpretation of molecular diagnostic assays in the context of detecting minimal residual disease.

Intriguingly, studies using genome-wide association have identified certain low-penetrance germline variants of IKZF1 (as well as other genes) that may increase susceptibility to the development of childhood ALL (Papaemmanuil et al, 2009). Even though much research remains directed at the basic molecular and cell biology of childhood acute leukemias, molecular epidemiologic methods will play an increasing role in identifying the interplay between environmental factors and host genetic polymorphisms, which underlie the predisposition to this disease in some individuals (Clavel et al, 2005; Spector et al, 2005; Lafiura et al, 2007; reviewed in Pui et al, 2008). Indeed, the relatively brief period of potential environmental exposure in utero may lend itself to the detection of specific risk factors in a manner impossible for adult-onset leukemias (Kim et al, 2006; Ross, 2008); mitigation strategies to decrease the risk of childhood leukemia is thus a goal of these efforts.