Challenges in accuracy in molecular genetic diagnosis of childhood AML: case series

Abstract

Survival rate of children with Acute Myeloid Leukemia (AML) improves gradually through cooperative studies. However, the outcome depends on heterogeneous mechanisms. Comprehending the genetic background of pediatric Acute Myeloid Leukemia (AML) is the key to risk stratification. Next Generation Sequencing (NGS) technology uses target panels that may detect additional genetic subsets. The study describes the experience of using NGS for treating pediatric AML patients at an institution. Patients who showed poor outcome aberration were referred to hematopoietic stem cell transplant (HSCT). 11 patients were tested. Aberrations were found in all subjects, mainly only in the NGS panel, indicating referral to HSCT in first remission in 2 cases and helping to outline the genetic features in all cases. The availability of NGS resources has had a therapeutic impact. NGS helped outline the patients’ genetic features and decision for HSCT. NGS is a valuable tool in the precision medicine era and should be widely accessible.

Introduction

Acute myeloid leukemia (AML) is a rare and heterogeneous hematologic malignancy in the pediatric population [1, 2]. The survival rate has improved gradually through cooperative studies, which have standardized treatment practices, however, the outcome depends on heterogeneous factors [2, 3]. Although there is a high remission rate, nearly half of the children will relapse [4] and for certain types of AML, no more than 15% of patients are cured solely with conventional chemotherapy [1].

This heterogeneity is a result of multiple genetic mechanisms [5] and gene rearrangements which are the drivers that predict disease risk. A better understanding of the correlation between genetic information and outcome leads to an improved risk-adapted therapy and thus the identification of genetic aberrations is crucial for improving outcome prediction [6, 7]. Since the third edition of the WHO classification, the importance of cytogenetic aberrations has progressively increased in relation to the morphological classification according to the French-American-British cooperative group (FAB) [8, 9]. New prognostic markers may provide further information to direct future treatment [10].

Karyotype chromosome analysis, fluorescence in situ hybridization (FISH), and polymerase chain reaction (PCR) are the most conventional tools for identifying genetic abnormalities [4, 10]. Despite being rich sources of information, these methods have some limitations. Examples of these are the inability to identify cryptic gene fusions are karyotype analysis, the need for targeted probes in FISH and the need for pre-designed primers for RT-PCR [9–12]. An additional barrier is the fact that low and middle-income countries (LMIC) have a lack of availability of much of diagnostic tests [13–16].

Next Generation Sequencing (NGS) technology uses target panels that may detect additional genetic subsets and thus has the ability to identify previously missed genetic aberrations. Its advent has made sequencing more accessible, and it has been widely applied in precision medicine [16]. The panels are targeted on a subset of genes, enabling refinement of diagnoses, which can impact therapeutic decisions [16–18]. These panels are based on exome sequencing, using a combined DNA and RNA sequencing approach to detect diagnostic, prognostic, and therapeutic markers across malignancies [19]. A single test is able to explore multiple genes [7] and certain aberrations, unusual fusion partners and rare breakpoints are found only in NGS panels [19, 20]; Moreover, NGS may find secondary abnormalities such as TP53 and NRAS [11]. While some panels are specifically designed for pediatric cancers, relatively few studies using this diagnostic method have been conducted in children [10]. Molecular subtypes, defined by a range of genetic abnormalities, present a challenge for accurate classification by current standard-of-care diagnostic tests [19]. Because of that, NGS can be useful to guide therapy choices in pediatric AML treatment.

The study describes the experience of using NGS to guide the treatment of pediatric AML at a Brazilian institution.

Materials and methods

Aims of the study

This is a retrospective observational study that aims to describe the impact on the implementation of a validated NGS panel for pediatric oncology for AML patients, and correlate with the clinical decisions. Between 2018 and 2020, NGS was available as part of a research project.

Patient criteria

Pediatric patients (from one to 18 years-old at diagnosis) diagnosed with AML (new or relapsed) in our center had bone marrow aspirate or peripheral blood samples at initial diagnosis (or relapse). All AML patients at the institute were enrolled. The patients that had an NGS analysis only at relapse did not have a first diagnosis sample available for various reasons– timing of first diagnosis out of the NGS availability; first investigation or beginning of the treatment in another hospital; inadequate material. All patients were stratified according to cytogenomic criteria and according to FAB morphologic features. Patients at first diagnosis were treated according to validated protocols as AML BFM 2004 protocol [9, 21] or as part of collaborative study in Brazil, denominated LMAIO 97, based on high-intensive blocks containing cladribine, cytarabine, and daunorubicin, etoposide and mitoxantrone [22]. Routine diagnostic screening included morphological and classical karyotype evaluation. FISH and PCR were not routinely performed due to economic limitations and were only available when there was health insurance coverage. Patients with a genetic aberration deemed to be associated with a poor outcome were referred for hematopoietic stem cell transplantation (HSCT). The relapsed patients received salvage protocol containing idarubicin, cytarabine, filgrastim, with or without idarubicin [23, 24] and were already candidates for HSCT, therefore NGS was performed to better understand the genetic background of the disease. Patients with acute promyelocytic leukemia were not included in this group.

NGS methods

NGS was performed to identify genetic variants in the samples using the Oncomine Childhood Cancer Research Assay^® (OCCRA^®) panel from Thermo Fisher Scientific^® [25]. The ion Torrent Oncomine Childhood Cancer Research Assay is a unique nest-generation sequencing (NGS)– based tool designed for comprehensive genomic profiling of cancer affecting children and young adults. The assay is designed to provide researchers with sensitive and comprehensive sample amplification of relevant DNA mutations and fusion transcripts associated with childhood and young adult cancers in a single NGS run. The panel is comprised of 203 unique genes: 130 key DNA genes, 28 copy number variant targets, an expansive fusion panel of 90 driver genes with multiple partners, and 9 expression genes and controls. These cover the most relevant targets in the vast majority of all childhood and young adult oncology research samples [25]. Total DNA and RNA were extracted from bone marrow or peripheral blood samples using the AllPrep DNA/RNA Mini Kit (QIAGEN^®) according to the manufacturer’s instructions. After extraction, samples were quantified using Nanodrop 2000^® and Qubit 3^® fluorometer (Thermo Fisher Scientific^®). An A260/280 ratio of 1.6–1.8 and 1.8-2.0 was used as a quality control for DNA and RNA, respectively.

Sequencing libraries were prepared from 20 ng DNA and 20 ng RNA to create two pools each. The panel covered a large number of DNA regions (3,069 amplicons) and RNA fusions (1,421 fusion primer pairs). Fusions were analyzed in lon Reporter 5.2 using the AmpliSeq Childhood Cancer Research Panel, w2.1. Only amplicon readings with a forward to reverse primer ratio ≥ 0.6 and ≤ 1.4 were selected. The readings obtained were aligned to the human reference genome hg19/GHCh37. The generated BAM files were analyzed using the Integrative Genomics Viewer (IGV) software. The VCF (Variant Call Format) files were obtained, and the interpretative analysis was performed using the Ion Reporter© software (ThermoFisher Scientific), through a specific platform designed for the OCCRA© panel. Copy number estimation was performed using Thermo Fisher’s “Variability Corrections Informatics Baseline” algorithm [25]. Genetic variants were classified according to type (single nucleotide variants [SNVs], insertion/deletions [InDels], copy number variants [CNVs], fusions, deletions), functional effect (missense, nonsense, synonymous, and frameshift), and clinical significance (pathogenic, likely pathogenic, uncertain significance, likely benign, and benign).

Ethical statement

The families gave their written consent for the samples of bone marrow aspirate or peripheral blood to be stored in the Biobank of the Instituto de Oncologia Pediátrica - IOP/GRAACC/UNIFESP (B-053) for clinical and research purposes. This study was approved by the Committee for Ethics in Research of the Federal University of Sao Paulo, as part as the Pronon scientific research project (Programa Nacional de Apoio à Atenção Oncológica - National Support Program for Oncological Care)– project number 25000.019858/2018-14. The methodology for this study was approved by the Human Research Ethics committee of the Federal University of São Paulo in accordance with the regulations of this committee, approval number 0715P/2021.

Results

Between July 2018 and July 2020, 11 patients with AML had samples analyzed using NGS - seven of them had the test performed from diagnosis, while four had it just at relapse time. Age at diagnosis ranged from 1.3 to 17.4 years, with a mean of 6.7 years. There were five females and six males. Four of them had NGS analysis only during relapse. All patients had aberrations found in the panel, most of them only identified by NGS. Table 1 shows the findings and interventions in patients that had the panel performed at diagnosis, while Table 2 shows the cases of relapse. As previously mentioned, these ones did not have the test at diagnosis due to lack of sample of diagnosis, for several reasons.

Table 1.

Abnormalities identified on NGS in patients that had it performed at diagnosis

No	Sex	Age (years)	FAB	Karyotype	Other	NGS	Time	Significance	Intervention due to NGS	Outcome
1	F	5,8	M4	nonevaluable	No	CBFB::MYH11 KIT mutation	Diagnosis	Good prognosis	No HSCT criteria	1st remission
2	M	12,5	M1	46XY, t(14;20)(q24;p12) [20]	RT-PCR negative for BCR-ABL and PML-RARA	NUP98::NSD1 FLT3 mutation	Diagnosis	Poor prognosis	HSCT in 1st remission	1st remission
3	F	6,5	M2	46,XX, add(8)(q24), del(8)(q22), del(21)(q22)[16]/46,XX[6]	No	RUNX1::RUNX1T1	Diagnosis	Good prognosis	No	Death in 2nd relapse, after 1st HSCT
4	F	14,7	M4	46, XX [20]	FISH without aberrations	NPM1 mutation	Diagnosis	Good prognosis	No HCST criteria	1st remission
5	M	1,3	M4	46 XY, inv(16)(p13.1q22)[20]	FISH w/ inv(16) / FISH negative for MYCN	MYCN mutation	Diagnosis	Unknown	No	Induction failure; death in remision due to HSCT complications
6	M	5,1	M2	45, XX [25]	FISH: inv (16) and t(16;16) Unusual	GATA2 mutation CEBPA mutation	Diagnosis	Good prognosis	No HSCT criteria	1st remission
7	F	1,6	M5	46,XX, add(6)(p25)[22]	No	KMT2A-MLLT10	Diagnosis	Poor prognosis	HSCT in 1st remission	1st remission

N– Patient number; FAB– French American British morphologic classification; Other– Other molecular methods; NGS– Next-generation sequencing; Time– Time point when the NGS panel was performed; HSCT–hematopoietic stem cell transplant; FISH - fluorescence in situ hybridization; RT-PCR reverse transcription polymerase chain reaction

Table 2.

Abnormalities identified on NGS in patients that had it performed at relapse

No	Sex	Age (years)	FAB	Karyotype	Other	NGS	Time	Significance	Outcome
8	F	14,5	M0	46, X, -X, add(3)(q27), -5, del(11)(q23), +der(19)t(1;19) (q23;p13), +mar[cp11]//46XY [6]	No	SET::NUP214 ASXL1 mutation PHP6 mutation	Relapse	Rare– uncertain in AML Possible adverse prognostic	Death in 2nd relapse, after 2nd HSCT
9	M	17,4	M2	46, XX [20]	No	NUP214– ABL1	Relapse	Unknown	2nd remission, HSCT
10	M	6,8	M1	46,XY [18]	No	NUP98::NSD1	Relapse	Poor prognosis	2nd remission HSCT
11	M	5,2	M2	47, XY, der(5)t(1;5)(q23;q35), + 6. del(10)(q23), t(11;19)(q23;p13)[11] / 46,XY [9]	FISH: suggestive of 11q23 translocation	KMT2A::MLLT1 ABL2 mutation	Relapse	Intermediate	Death in relapse

N– Patient number; FAB– French American British morphologic classification; Other– Other molecular methods; NGS– Next-generation sequencing; Time– Time point when the NGS panel was performed; HSCT–hematopoietic stem cell transplant; FISH - fluorescence in situ hybridization; RT-PCR reverse transcription polymerase chain reaction

Seven patients (63%) are still alive with a median time from diagnosis of 33.8 months. Causes of death were disease progression (3 cases) and HSCT complications (1 case).

The aberrations found in NGS were diverse, including fusions (SET::NUP214; CBFB::MYH11; NUP98::NSD1; RUNX1::RUNX1T1; NUP214::ABL1; KMT2A::MLLT1; KMT2A: MLLT10); indels (ASXL; KIT; FLT3; NPM1; CEBPA); CNVs (MYCN; ABL2); and SNVs (GATA 2).

Two patients (No.2 and No.7 Table 1) were referred to HSCT after finding an aberration definitely associated with poor prognosis; one showed NUP98::NSD1 and the other KMT2A::MLLT10. These two patients had no other poor prognostic factor at diagnosis. They underwent hematopoietic stem cell transplantation and did not relapse. A third patient (No.10 Table 2) had a normal karyotype result and did not have good quality material for NGS at diagnosis. The patient relapsed and it was possible to perform sequencing at this time, which showed a translocation that would have indicated HSCT (NUP98::NSD1) in first remission. HSCT was already planned for patient No. 8 (Table 2). The NGS results showed SET::NUP214 and ASXL and PHP6 mutations. Two of the results - patients No. 5 Table 1 and No. 9 Table 2 - showed results with unknown significance of prognosis. In four patients, aberrations with good prognoses were found: No. 1 (Table 1) showed inv(16) (the karyotype was not evaluable); RUNX1::RUNX1T1(No. 3 Table 1); GATA2 mutation (No. 4 Table 1) and NPM1 mutation (associated with normal karyotype) in No. 6 Table 1. Patient No.11 in Table 2 had a fusion related to an intermediate prognosis and a CNV not related to a poor prognosis.

Discussion

During the period when NGS was performed in our institution, all AML patients had a result with new findings, including those who already had an aberration found by another method. Therefore, the use of sequencing was useful for the cases to better stratify their disease and offer appropriate treatment. These findings involved cryptic aberrations that are known prognostic factors, or aberrations that could be seen in other targeted methods that were not available or even some unusual findings, as discussed below.

NUP98::NSD1 fusion, [t(5; 11) (q35.2; p15.4)] is a cryptic rearrangement that may block differentiation. It is found in AMLs with normal karyotype, occasionally associated with additional mutations such as FLT3, NRAS, WT1 and MYC [26, 27]. This was found in 2 different cases (Patients No. 2 Table 1 and No. 10 Table 2), that also differed between themselves, as case No. 2 had an FLT3 associated mutation and No. 10 had no additional findings. The aberration is strongly associated with a poor outcome [27, 28] leading to HSCT is indicated in first remission in these cases due to the high risk of relapse, with or without associated aberrations [29–31]. For this reason, patient No. 2 was referred for HSCT as soon as the aberration was detected, and it was crucial as there was no other evidence of bad outcome, he was in morphologic remission after induction. Unfortunately, one of them (patient No. 10) did not have this information at his first treatment, because the material was not sufficient for the analysis. Otherwise, he would have been transplanted before relapse.

Two different KMT2A rearrangements were identified, whereby the involved partner is essential to determine the risk [1, 9]. Patient No. 11 (Table 2) already had the information of t(11;19)(q23;p13) (KMT2A::MLLT1) in the karyotype and an 11q23 translocation was also seen in FISH (Fig. 1), which is associated with an intermediate prognosis [9]. Furthermore, the patient had ABL2 gain of function, a tyrosine kinase associated with some leukemias, lung and breast cancers with uncertain significance in this AML case [32], the reason why the use of tyrosine kinase inhibitors (TKI) was not considered. This patient relapsed after the first treatment, with a very rapid progression. In case No. 7 (Table 1), KMT2A::MLLT10 was detected in a one-year-old girl with M5 FAB myeloid leukemia. An 11q23 translocation was expected but not found in the karyotype. The KMT2A gene has a wide range of fusion partners, and the specific product of the rearrangement defines the prognostic value. The t(10;11)(p11.2;q23) is associated with a dismal outcome [1, 29], and for this reason she was also considered for HSCT, being in morphologic remission since the induction. After the transplant, she did not relapse.

Fig. 1.

a and b– case No.10. Karyotype 47, XY, der(5)t(1;5)(q23;q35), + 6. del(10)(q23), t(11;19)(q23;p13). FISH presented 11q23 rearrangement. He also had KMT2A::MLLT1 fusion, associated with ABL2 mutation in NGS

There were also cases where the NGS supported the consideration for better outcome and reinforced the decision of sparing HSCT in first remission. For instance, there was a CBFB::MYH11 fusion associated with a KIT indel was found in a case without an informative karyotype (No. 1 Table 1), and the NGS was crucial in understanding the AML features in this case. AML with inv(16)(p13;q22) or t(16;16)(p13;q22) is known to be associated with a good outcome [9, 11]. She had associated KIT c.1255_1257delGAC; p.Asp419del. It is noteworthy that the presence of KIT mutation has not shown a worse impact in the prognosis, except for mutations on exon 17 in the KIT gene [33].

Also, there was a NPM1 mutation without concomitant FLT3 mutation in normal karyotype, revealing a favorable genetic background (Case No. 4 Table 1) [1, 34]. A gain of function in GATA2 associated with biallelic CEBPA mutation and normal karyotype is related to a good outcome [35, 36], as seen in case No. 9.

The translocation of the SET::NUP214 was found in one complex case (Case No. 8, Table 2) and it is an unexpected finding in myeloid leukemia, a fusion protein that can inhibit cell apoptosis [37]. Although it is a rare translocation, it has been described in some cases of AML, especially in minimally differentiated myeloid leukemia as well as in undifferentiated leukemia [37–39]. This rearrangement is mainly associated with T-ALL and it has been described as chemotherapy and steroid resistant [37, 40]. In addition, she presented with mutations on ASXL1 and PHF6, which may be associated with a worse outcome [41]. Even before the NGS result, this patient had several indications for HSCT: complex karyotype; absence of remission after induction [29] and FAB phenotype associated with poor prognosis [42]. At her first diagnosis, NGS was not available. The NGS findings did not change the therapeutic decision, however, they reinforced the conclusion that this was a case with a poor prognosis. In short, NGS provided further information about the complexity of the disease.

There was a surprising finding of NUP214::ABL1 rearrangement (Case No. 9 Table 2) which is not usually described in AML, but in some cases of T ALL and B ALL [43–44]. As it is an active tyrosine kinase, it is a potential TKI target [45]. It is a cryptic fusion [43, 44], as it was seen in this case with a normal karyotype, that did not have a good sample for analysis at diagnosis. There was one case in literature showing an AML patient with this fusion, who was treated with conventional chemotherapy for AML [46]. This was the second published case with this fusion in AML. The use of TKI was even considered. However, as no previous evidence was found to support its use and there was a good response for the salvage therapy followed by HSCT, this strategy stayed on standby in case of progression [47].

On the other hand, some patients had an unfortunate progression despite a favorable genetic background, such as patient No. 4 with RUNX1::RUNX1T1 fusion [1]. Interestingly, the classic translocation t(8;21)(q22;q22.1) was not detected on karyotypic evaluation, either at the initial diagnosis or at relapse, despite an addition in 8q24, a deletion in 8q22, and a deletion in 21q22 (Fig. 2). It was not considered as complex karyotype at first, as the it can be defined as 3 or more aberrations without recurring translocations, as the t(8,21) [1], considering the NGS finding of RUNX1::RUNX1T1. However, it was not noted in the karyotype, which would be the classical definition. Presenting this unfortunate evolution, perhaps the complex karyotype definition could have been further discussed.

Fig. 2.

case No.4. Karyotype 46,XX, add(8)(q24),del(8)(q22),del(21)(q22). The NGS panel revealed RUNX1::RUNX1T1 fusion

Case number 5 (Table 1) was found to have inv(16) in the classic molecular evaluation, which may indicate a good outcome [9, 11]. Nevertheless, additional mutations may influence the prognosis [1]. This case had induction failure. When the NGS results were available, MYCN CNA was identified. This finding was not confirmed in FISH and is not conclusive.

In summary, the availability of the NGS resource improved the stratification of patients for clinical decisions, in some cases with findings that could be reached with other technologies that were not available, like FISH and PCR, and other cases where it probably wouldn’t be seen even with them. Two of the patients were referred to HSCT solely due to the NGS result. In a patient with a normal karyotype, NPM1 was identified, providing insight into a favorable genetic background. Additionally, this resource helped outline a case where the karyotype was non-evaluable. Finally, it provided more detailed information about challenging cases, such as one with a complex karyotype, the genetic profile of two cases with normal karyotype that relapsed, and additional aberrations present even in cases with an expected favorable outcome.

Brazil is a continental size country and, as a LMIC, a heterogenous context. Some resources of high technology can be provided in some situations and in others there are still major challenges in diagnosis, follow up, and specific and support treatment, with a great challenge on centralizing and standardizing strategies countrywide. With limitations on availability of transplant centers and higher mortality due to toxicity than high income countries, decision and timing for referring to HSTC need all the possible attention [15, 48]. Although the NGS technology is more affordable than previous sequencing methods [16], it is still a distant reality in LMIC situations. Most centers lack access to NGS, as well as other methods such as PCR and FISH. Sadly, these cases may remain without sufficient genetic information. A global effort to increase the availability of refined diagnostic tools is a relevant matter that requires broader discussions. This study demonstrates both the feasibility and the utility of incorporating comprehensive somatic-germline genomic analysis into pediatric oncology care.

The study had limitations, as it is a retrospective description of cases within a limited period, resulting in a small cohort of patients of a rare and heterogeneous disease, making the cases with huge difference between themselves and making it complex to analyze. Survival curves were not performed, and some patients did not have NGS results at first diagnosis. During this period, the minimal residual disease was not a strategy feasible to be implemented, that could also be a tool for reevaluation and decision [49]. As a cohort of genetically and clinically well-annotated childhood and adolescent AML, we have produced accurate subtyping by molecular methods, clearly demonstrating the future role of NGS for improved genetic risk-directed stratification compared to previous standard-of-care approaches.

In an era in which genomics is driving enormous scientific progress and demonstrating the potential for precision medicine, this study endorses the clinical advantage of introducing NGS as a first-line diagnostic tool in childhood AML. While accurately detecting the range of clinically relevant cytogenetic abnormalities, it identified an expanded list of genetic abnormalities which may define novel subtypes or co-operating genetic abnormalities in larger collaborative studies. Although the cost and infrastructural requirements of NGS have been limiting factors for many countries, the rapidly expanding list of genetic tests required for accurate diagnosis makes NGS a viable option for some healthcare providers.

In conclusion, NGS is a valuable tool for diagnostic refinement in the precision medicine era. This report showed that it provided additional information for AML.

Author contributions

Conception/Design and development of methodology: SRCT, FTG, MGCP, JTG, IDO, ED, BNR; Acquisition of data: SRCT, FTG, AVLS, FOM; Analysis and interpretation of data: SRCT, FTG, AVLS, FOM; Writing of manuscript: FOM; Review and/or revision of manuscript: AVLS, SRCT, FTG; Medical support: AVLS, NSS, RFCG, VGZ, MLML.

Funding

The study was supported by Pediatric Oncology Institute-Grupo de Apoio ao Adolescente e à Criança com Câncer/Federal University of Sao Paulo (IOP-GRAACC /UNIFESP) and Pronon (Programa Nacional de Apoio à Atenção Oncológica = National Support Program for Oncological Care). project number 25000.019858/2018-14.

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethical approval

This project was approved by the Ethical committee of Federal University of São Paulo in accordance with the regulations of this committee, approval number 0715P/2021, CAAE: 48326921.4.0000.5505.

Competing interests

The authors declare no competing interests.