A neoplasm with FIP1L1-PDGFRA fusion presenting as pediatric T-cell lymphoblastic leukemia/lymphoma without eosinophilia

Abstract

The 2016 World Health Organization (2016 WHO) classification of hematopoietic malignancies classifies neoplasms with a fusion between the FIP1L1 and PDGFRA genes in 4q12 into a group called “myeloid and lymphoid neoplasms with eosinophilia and abnormalities of PDGFRA, PDGFRB or FGFR1 or with PCM1-JAK2”. Neoplasms characterized by this fusion are pluripotent stem cell disorders that can show both myeloid and lymphoid differentiation. They typically occur in adult patients and most are characterized by eosinophilia. We describe identification of a FIP1L1-PDGFRA fusion in a 13-year-old boy who presented with T-lymphoblastic leukemia/lymphoma without eosinophilia. Detection of FIP1L1-PDGFRA driven neoplasms at diagnosis is usually critical for proper treatment, since almost all reported cases responded to tyrosine kinase inhibitors. However, our patient’s leukemia was refractory to standard chemotherapy, and did not show a meaningful response to tyrosine kinase inhibitor therapy. Testing for a FIP1L1-PDGFRA rearrangement is at present limited to patients with idiopathic hypereosinophilia, and we hypothesize that this abnormality may be under-diagnosed in children with acute leukemias.

Introduction

Hematologic malignancies defined in the 2016 WHO classification as “myeloid and lymphoid neoplasms with eosinophilia and abnormalities of PDGFRA, PDGFRB or FGFR1 or with PCM1-JAK2” are myeloid/lymphoid pluripotent stem cell disorders often characterized by sustained eosinophilia (1). Timely identification of these neoplasms is of great clinical importance, since the activated kinases in tumor cells may be therapeutically targeted (2–4).

Neoplasms associated with the FIP1L1–PDGFRA rearrangement are rare, and are primarily reported in adult patients, with a peak of incidence during the fourth decade. There is an unexplained striking male predominance among affected individuals (3,4). Hematologic manifestations in cases with FIP1L1–PDGFRA rearrangements are protean. Many cases present as a myeloproliferative neoplasm manifested by sustained eosinophilia (chronic eosinophilic leukemia, CEL). Clinically, patients often have splenomegaly (~60%), anemia, thrombocytopenia, and symptoms related to release of eosinophilic granules (skin rash and erythema, or pulmonary, gastrointestinal, and cardiac manifestations). The bone marrow may also show mastocytosis (3–6).

Less frequently, FIP1L1–PDGFRA driven neoplasms can present as acute myeloid leukemia (AML) or as T-lymphoblastic leukemia/lymphoma (3–6); patients whose disease presents as T-lymphoblastic lymphoma are often found to also have antecedent or concurrent myeloproliferative features (5).

Pediatric cases of FIP1L1-PDGFRA associated neoplasms are exceedingly rare. Only four pediatric cases of a neoplasm with a PDGFRA rearrangement have been reported to date (7–9), and all the patients presented with prominent eosinophilia. To the best of our knowledge, no pediatric cases of T-lymphoblastic leukemia/lymphoma with a FIP1L1-PDGFRA fusion have been reported to date. Herein, we describe detection of a FIP1L1-PDGFRA fusion in a 13-year- old boy who developed T-lymphoblastic leukemia/lymphoma.

Materials and methods

Cytogenetic and fluorescence in situ hybridization (FISH) analysis

Conventional cytogenetic analysis was performed following standard procedures. Twenty G-banded metaphase cells were examined, and the chromosomal abnormalities were described according to the International System for Human Cytogenetic Nomenclature (ISCN 2016). FISH testing on formalin fixed paraffin embedded tissue was performed at Quest laboratories (San Juan Capistrano, CA).

Immunohistochemistry

Immunohistochemical analysis of CD3 and TdT using the automated Bond system (Leica Biosystems, Newcastle, UK) was performed using manufacturer specification. Immunohistochemical staining for LMO2 was performed courtesy of Dr. Elizabeth Hyjek of the University of Chicago. Immunohistochemical staining for PDGFRA was performed at NeoGenomics Laboratories (Aliso Viejo, CA).

DNA extraction and Chromosomal Microarray (CMA) analysis

Genomic DNA was extracted from a frozen lymph node and a bone marrow (BM) aspirate sample with a Qiagen QIAamp DNA Mini Kit (Qiagen Inc., Valencia, CA). CMA analysis of the lymph node sample was performed using the Affymetrix OncoScan FFPE Assay (Affymetrix Inc., Santa Clara, CA), which utilizes Molecular Inversion Probe (MIP) technology to obtain accurate genome-wide copy number and loss-ofheterozygosity (LOH) profiles. CMA analysis of the BM aspirate sample obtained at the time of progression was performed using the Affymetrix CytoScan™ HD arrays (Affymetrix Inc., Santa Clara, CA). For both array platforms, sample preparation, hybridization, washing and labeling procedures were performed following the manufacturer’s recommendations. The data were analyzed with the Chromosome Analysis Suite (ChAS) software from Affymetrix (Affymetrix Inc., Santa Clara, CA).

Molecular testing for sequence abnormalities and abnormal gene fusions

Molecular testing for sequence changes and abnormal gene fusions in tumor cells was performed using the custom Children’s Hospital Los Angeles OncoKids™ panel, which is a primer-based target enrichment (AmpliSeq) next generation sequencing (NGS) based assay designed to detect diagnostic, prognostic, and therapeutic markers across a spectrum of pediatric malignancies. The DNA content of the OncoKids™ panel consists of over 3000 amplicons and covers the full coding regions of 44 tumor suppressor genes and onco-genes, hotspots for mutations in 82 genes, and amplification events in 24 genes. The RNA content includes over 1400 targeted gene fusions associated with acute myeloid and acute lymphoblastic leukemia, childhood sarcomas, pediatric brain tumors and soft tissue tumors (10). The sensitivity (limit of detection) of the assay, established during the “in-house” clinical validation studies, corresponds to 5% variant allele frequency (VAF) for single nucleotide variants (SNVs) and 10% VAF for indels. A total of 20 ng of tumor DNA and RNA, which were isolated from fresh frozen lymph node specimen or from bone marrow aspirate sample, were used for OncoKids™ testing. Sequencing was performed using the 540 chip on the Ion Torrent S5 sequencing platform (Thermofisher Scientific, Waltham, MA), and downstream data analysis was conducted by Ion Reporter (Thermofisher) and an in-house custom software suite.

Results

Case history

The patient was a previously healthy 13-year-old African American boy who presented at an outside institution with an enlarged right supraclavicular lymph node. A CBC performed at the time of diagnosis showed white blood cell count of 1.4 ×10³/uL, hemoglobin of 8.9g/dL, and platelet count of 257 × 10³/uL. A differential showed 52% neutrophils, 31% lymphocytes, 17% monocytes, and 0% eosinophils. A lymph node biopsy and bone marrow (BM) aspirate established a diagnosis of T-lymphoblastic leukemia/lymphoma, with both nodal and marrow disease (30% blasts) present. Notably, there was no evidence of eosinophilia, mastocytosis, or a myeloproliferative neoplasm in the bone marrow at diagnosis. Cytogenetic evaluation showed 46,XY,t(11;14)(p13q11.2)[3]/46,XY[20].

The patient was induced with vincristine, dexamethasone, pegaspargase, daunomycin, and intrathecal cytarabine and methotrexate. A post-induction bone marrow evaluation showed 7% residual blasts. Consolidation therapy was initiated and consisted of cyclophosphamide, cytarabine, mercaptopurine, vincristine, and pegaspargase. A bone marrow evaluation halfway through consolidation therapy showed morphologic remission, but was positive for minimal residual disease (MRD; 0.47%). Toward the end of consolidation, the patient presented with left eye pain, increased tearing, transient blindness, and progressive lymphadenopathy. He was transferred to Children’s Hospital Los Angeles for further management (day +109).

A lymph node and bone marrow biopsy was performed. Pathologic evaluation of the lymph node biopsy showed effacement of the node by lymphoblasts (Figure 1a), with approximately 95% of tumor cells in the sample. The lymphoblasts strongly expressed CD3 and TdT (Figure 1b, c). Flow cytometric immunophenotyping of the nodal lymphoblasts showed positivity for: CD2 (partial), cCD3, CD5, CD7 (variable), CD8 (variable), CD56 (variable), CD13 (partial), CD33 (partial), CD38, CD45, and TdT (partial). The lymphoblasts were negative for surface CD3, CD1a, CD4, CD19, CD20, CD22, CD24, CD14, CD15, CD64, CD117, MPO, HLA-DR, and CRLF2. This confirmed a relapse of the patient’s T-lymphoblastic leukemia/lymphoma, with suspected leukemic infiltrates of the optic nerve based on imaging. Karyotype analysis of the lymph node sample showed a t(11;14)(p13;q11.2) in seven out of twenty analyzed cells (Figure 2). This trans-location is a well characterized recurrent abnormality in T-lymphoblastic leukemia/lymphoma (11); it juxtaposes the LMO2 gene located in 11p13 next to the T-cell receptor α/δ locus in 14q11.2, resulting in LMO2 overexpression. LMO2 protein overexpression was confirmed by immunohistochemistry (Figure 1d), which was strongly positive in >95% of the lymphoblasts. A concurrently performed bone marrow evaluation was positive for T-lymphoblast involvement (3% blasts). CMA analysis was also performed, and it revealed several putative genetic drivers characteristic for T- lymphoblastic leukemia/lymphoma (see below).

Figure 1.

Morphology and immunohistochemistry for the lymph node biopsy sample. 1a. Effacement of the lymph node by abnormal lymphoblasts; 1b, 1c, 1d. Immunohistochemistry showing CD3 (1a), TdT (1b) and LMO2 (1d) expression in leukemia cells.

Figure 2.

Representative karyotype from a lymph node sample showing the t(11;14)(p13;q11.2) (indicated by arrows) which involves the LMO2 locus in 11p13 and the T-cell receptor α/δ locus in 14q11.2.

Surprisingly, CMA analysis also revealed a deletion in the q12 region of chromosome 4 with the breakpoints within the FIP1L1 and PDGFRA genes, predicted to result in an abnormal FIP1L1-PDGFRA fusion and PDGFRA overexpression. This gene fusion was confirmed both by FISH (Quest) and by using a custom NGS-based panel designed to detect DNA mutations and abnormal gene fusions in genes associated with pediatric cancers (OncoKids™). FISH testing of the lymph node sample showed an abnormal signal pattern consistent with the FIP1L1-PDGFR fusion in 98 out of 100 analyzed cells (98%), indicating the presence of this abnormality in the dominant tumor clone. PDGFRA protein overexpression was confirmed by immunohistochemistry (NeoGenomics), with approximately 90% of the lymphoblasts showing moderate expression of PDGFRA.

Patients who have hematologic malignancies with PDGFRA fusions often respond to tyrosine kinase inhibitors (TKIs), and imatinib (400 mg daily) was added to a re-induction therapy of mitoxantrone and high dose cytarabine. This resulted in a partial transient response, with clearance of peripheral blasts, morphologic remission in the bone marrow, and decrease in BM MRD to 0.76% when measured at day +158. However, after this brief clinical and morphologic improvement, the patient presented again with lymphadenopathy (day +187). A bone marrow biopsy at this time showed extensive involvement by T-lymphoblastic leukemia/lymphoma. Genetic testing was repeated on a bone marrow aspirate sample. Cytogenetic analysis revealed a main clone with the same t(11;14)(p13;q11.2) observed previously in the diagnostic specimen, and a subclone which in addition to the t(11;14) showed an interstitial deletion in the long arm of chromosome 9 (Table 1). Similarly, CMA analysis showed the same abnormalities as detected earlier, and also revealed a 38 Mb deletion in the q21.11q31.2 region of chromosome 9 consistent with the 9q deletion noted in the abnormal subclone by karyotype analysis. Testing of the BM sample by the OncoKids™ assay confirmed persistent expression of the abnormal FIP1L1-PDGFRA fusion. Furthermore, evaluation for sequence abnormalities revealed a complex variant affecting the PDGFRA D842 codon, which is a known hot-spot for activating PDGFRA mutations which manifest resistance to imatinib and multiple other TKIs. Focused review of the sequencing results from the previous lymph node sample showed that this mutation was not present at a detectable level (≥10%) before imatinib treatment.

Table 1.

Pathology and genetic testing results for tumor samples

	Pathology	Karyotype	Chromosomal Microarray Analysis	Mutations detected by OncoKids DNA Assay	Gene fusions detected by OncoKids RNA assay
Lymph node sample	Lymph node effaced by an infiltrate of leukemic blasts; Immunophenotype (flow cytometry): CD2 partial, cytoplasmic CD3, CD5, CD7 variable, CD8 variable, CD9 partial, CD10 partial, CD13 partial, CD33 partial, CD38, CD56 variable, CD58, CD71 variable, CD123 variable and TdT partial.	46,XY,t(11;14) (p13;q11.2) [7]/46,XY[1]	arr [GRCh37] 1p33(47655598_47783787)x1,4q12 (54309141_55098549) x1,9p24.3p13.3(204737_34431079)hmz,9p21.3p21.3 (21827991_21954953)x0	No clinically significant somatic variants; negative for PDGFRA mutation detected in the bone marrow sample	STIL(1)-TAL1 (2), FIP1L1 (13)-PDGFRA(12)
Bone marrow sample	Sections of bone marrow biopsy show a hypercellular marrow with significantly increast number of lymphoblasts (approximately 50%); Immunophenotype (flow cytometry): CD2 negative, surface and cytoplasmic CD3, CD5, CD7 variable (abnormally dim to moderate), CD8 dim/negative, CD4 negative and partial/dim CD56. The blasts were negative for B cell and myeloid antigens.	46,XY,t(11;14) (p13;q11.2) [3]/46,sl,del(9) (q13q22)[2]/46, XY[15]	ARR[GRCh37] 1 p33(47655598_47783787)x1,4q12 (54280345_55138638)x1, 9p24.3p13.3(192128_34345782) hmz,9 p21.3(21819462_21996603)x0,9q21.11q31.2 (70966261_109046950)x1a	NM_006206 (PDGFRA): c.2522_2527delinsAAG (p.Arg841_lle843delinsLysVal)	STIL(1)-TAL1(2), FIP1L1(13)-PDGFRA(12)

^aSlight difference in the genomic coordinates of the detected copy number segments is due to the use of different array platforms (OncoScan and CytoScanHD) for the two samples.

A final treatment attempt was made by adding Dasatinib to cyclophosphamide, etopiside, dexamethasone, and bortezomib on day +201. Despite this, the patient expired from the progressive disease at day +217 from diagnosis.

CMA analysis

CMA analysis performed on the lymph node sample at the time of first relapse revealed several genetic drivers characteristic for T-lymphoblastic leukemia/lymphoma. A focal deletion was observed in the p33 region of chromosome 1 with breakpoints within the STIL and TAL1 genes (Supplementary Figure S1). The deletion was predicted to result in an abnormal STIL-TAL1 fusion, which has been reported in 9%–26% of T-lymphoblastic leukemia/lymphoma cases (12,13). The fusion results in inappropriate expression of the TAL1 gene, encoding for a basic helix-loop-helix (bHLH) transcription factor involved in controlling cell proliferation, differentiation, and apoptosis during hematopoiesis (12,13). Although FISH studies for the STIL-TAL1 fusion were not performed, inspection of the log2 ratio from the CMA copy number data suggested that approximately 80%–90% of the cells in the sample had this abnormality, which was thus likely present in the major tumor clone. Furthermore, CMA testing showed copy neutral loss of heterozygosity (CN-LOH) in the p24.3p13.3 region of chromosome 9 (34.2 Mb), and a focal deletion in the p21.3 region of chromosome 9 (127 kb), involving the CDKN2A/CDKN2B locus. Together, these 9p abnormalities resulted in a loss of both copies (bi-allelic deletion) of the CDKN2A and CDKN2B genes, which encode p16INK4a/p14ARF and p15INK4b tumor suppressor proteins, respectively. Abnormalities of the short arm of chromosome 9 involving the CDKN2A and CDKN2B genes are very frequent in T-lymphoblastic leukemia/lymphoma, and have been reported in up to 50% of the cases (14). In addition to the abnormalities commonly observed in T-lymphoblastic leukemia/lymphoma, CMA analysis also revealed a deletion in the q12 region of chromosome 4, with the proximal (centromeric) breakpoint within the FIP1L1 gene and the distal (telomeric) breakpoint within the PDGFRA gene (Figure 3). The deletion was predicted to result in an abnormal FIP1L1-PDGFRA fusion and PDGFRA overexpression.

Figure 3.

Details of Chromosomal Microarray (CMA) analysis results showing a focal 4q deletion resulting in the FIP1L1-PDGFRA fusion. A 789 kb deletion occurred in the q12 region of chromosome 4 (the approximate position is indicated on the chromosome ideogram on the top), with the centromeric breakpoint within the FIP1L1 gene and the telomeric breakpoint within the PDGFRA gene. Vertical orange dotted lines denote approximate intronic breakpoints on the schematic presentation of the fusion genes (bottom). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

Abnormal gene fusions and sequence abnormalities detected by NGS

The presence of the abnormal gene fusions predicted based on the CMA findings was confirmed by molecular testing of the lymph node sample using the OncoKids™ panel. This analysis showed the presence of both expected fusion transcripts: a transcript with a junction between the first exon of STIL (NM_001048166) and the second exon of TAL1 (NM_003189.5), and a transcript with a junction between exon 13 of FIP1L1 (NM_001134937.1) and exon 12 of PDGFRA (NM_006206.4), with additional sequence inserted at the junction site (Figure 4). The observed FIP1L1 and PDGFRA breakpoints were concordant with previously reported cases of FIP1L1-PDGFRA fusion, in which FIP1L1 breakpoint typically occurred within intron 13, and PDGFRA breakpoint usually mapped within exon 12 (15,16). The breakpoints in our patient correspond to the so-called type A FIP1L1-PDGFRA fusion (15,16), where processing of the abnormal fusion transcript occurs through activation of a cryptic splice site within the FIP1L1 intron 13, so that some of the intronic sequence gets included in the spliced mRNA. In addition to the FIP1L1 intron 13 sequence, the junction region in our patient contained other inserted sequence for which the origin could not be determined, but careful evaluation confirmed that the fusion was in frame, and was thus predicted to generate a functional FIP1L1-PDGFRA chimeric protein. As mentioned above, immunohistochemical staining for PDGFA was positive, demonstrating aberrant protein expression.

Figure 4.

Schematic presentation (top) and the sequence of the FIP1L1-PDGFRA junction region (bottom). The fusion was generated between the FIP1L1 exon 13 (dark green) and the PDGFRA exon 12 (orange). A sequence from the FIP1L1 intron 3 (light green) and a short sequence of unknown origin (grey) were inserted at the breakpoint junction. Sequencing of the junction fragment (bottom) showed that the fusion was “in frame”. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

OncoKids™ analysis was also performed on the BM sample at the time of disease progression after imatinib treatment. In addition to confirming continued expression of the STIL-TAL1 and FIP1L1-PDGFRA fusions, NGS testing at this time revealed a complex sequence variant in exon 18 of the PDGFRA gene: NM_006206.5 (PDGFRA): c.2522_2527delinsAAG (p.R841_Ile843delinsKV). This sequence change was not detected in the lymph node sample before exposure to TKIs (Figure 5), but its presence at the level which was below the limit of detection (5–10%) of the OncoKids™ assay could not be excluded. At the protein level, the detected variant replaces three amino acid residues (arginine, aspartic acid and isoleucine) at positions 841–843 with lysine and valine. The substitution occurs within the activation loop of the PDGFRA kinase, and involves aspartic acid (D) residue at position 842, which is a hot-spot for activating PDGFRA mutations in gastrointestinal stromal tumors (GIST) and other cancers (17,18). The single most common PDGFRA mutation in GISTs is the substitution D842V, however, activating exon-18 mutations frequently consist of in-frame deletions and deletions/insertions including D842 and neighboring amino-acid residues (17,18).

Figure 5.

Integrated Genome Viewer (IGV) screenshots showing the D842 molecular mutation in the BM sample. The mutant reads are indicated with the arrows in yellow. The mutation results in replacing the nucleotides between positions c.2522 and c.2527 (black arrows) with nucleotides “AAG”, which at the protein level replaces three amino acid residues (arginine, aspartic acid and isoleucine) at positions 841–843 with lysine and valine.

Both samples (the lymph node sample from first relapse and the BM sample from the time of progression) were also evaluated for mutations in genes associated with T-lymphoblastic leukemia/lymphoma and other hematologic malignancies (including NOTCH1, IL7R, JAK1, JAK3, STAT5, FBXW7, GATA3, WT1, AKT, FLT3, PTEN, EZH2, EED, DNMT3A, SUZ12 and others), but no acquired, disease associated mutations were found.

Discussion

We describe a case of a FIP1L1–PDGFRA neoplasm that in multiple ways challenges the current knowledge about hematologic malignancies associated with this fusion. One unusual feature of our case is the patient’s young age. Pediatric cases of FIP1L1–PDGFRA associated neoplasms are rare, and to our knowledge only four cases have been reported in the literature. All pediatric patients described to date had a typical disease presentation as CEL (7–9), and the diagnosis was established through extensive workup for unexplained eosinophilia. Our patient is the first pediatric case of a rare presentation of a FIP1L1–PDGFRA associated neoplasm as T-lymphoblastic lymphoma. Furthermore, he presented without eosinophilia or other clinical or morphologic features commonly seen in these diseases, and the abnormal fusion was detected unexpectedly by routine CMA testing. We confirmed this finding with multiple ancillary methods including FISH and NGS, and showed immunohistochemical over-expression of PDGFRA.

A FIP1L1-PDGFRA fusion gene usually results from a cryptic interstitial deletion in 4q12. In the majority of reported cases, the karyotype was reported as normal or as showing non-specific secondary abnormalities (2), and there is limited information about associated molecular mutations in these neoplasms. Notably, our patient showed the presence of several abnormalities that are well characterized driver mutations in primary T-lymphoblastic leukemia/lymphoma, including the t(11;14)(p13;q11.2), STIL-TAL1 fusion and homozygous CDKN2A/2B deletion. Cases of T-ALL with TAL1 fusions have been reported to occasionally co-occur with LMO2 overexpression (19), such as seen in the current case.

In addition to the patient’s clinical presentation and histologic and immunophenotyping findings, the presence of these abnormalities further suggested the diagnosis of primary T-lymphoblastic leukemia/lymphoma, making the detection of the FIP1L1-PDGFRA fusion even more surprising. This unusual constellation of genetic findings thus raises a question about the most appropriate classification for this disease, which was ultimately designated as “T-lymphoblastic leukemia/lymphoma with FIP1L1-PDGFRA fusion”. FIP1L1-PDGFRA driven malignancies can in rare cases present as T-lymphoblastic leukemia/lymphoma, but consistent with this being stem-cell disorders, in the majority of such cases eosinophilia and/or other abnormalities involving myeloid cells are clearly recognizable (5). Unfortunately, no additional material from our patient is available to test for the presence of the FIP1L1-PDGFRA fusion in other cell lineages, but through the entire course of his disease, the patient had no clinical or hematologic signs of myeloid involvement. The only genetic change observed in leukemia cells which was characteristic for myeloid malignancies and had also been associated with mixed-phenotype leukemia was the 9q deletion (20–22). Although this deletion was only detected as a sub-clonal abnormality at the time of progression after imatinib treatment, its presence may potentially provide some evidence of myeloid differentiation. Interestingly, a FIP1L1-PDGFRA fusion has recently been reported in a patient who apparently presented with acute lymphoblastic leukemia of B-lineage, but no details were provided about his clinical, histologic or immunophenotyping findings (23).

The clinical importance of establishing the diagnosis of FIP1L1-PDGFRA associated neoplasms stems from their exquisite sensitivity to TKIs. Most patients respond well to imatinib, and a dose of 100 mg daily is usually sufficient to achieve a complete and lasting remission (6). Good response to imatinib has even been documented in patients presenting with more aggressive forms of the disease, such asAML or T-lymphoblastic lymphoma (5). In contrast, our patient had an incomplete, transient response, similar to what has been observed in blast crisis of BCR-ABL positive CML, in which responses to imatinib can be short lived. Given the presence of multiple oncogenic drivers in this case, it is not surprising that agents targeting one molecular pathway would be ineffective. A recent NGS based study of cases with the FIP1L1-PDGFRA fusion identify a very small number of co-occurring driver mutations (20%) present at diagnosis (24).

Primary or acquired resistance to TKIs has rarely been described in FIP1L1-PDGFRA associated neoplasms, and can occur as a result of a further mutation, such as T674I or D842V (25,26). In our patient, a mutation affecting the D842 hot-spot was evident in the dominant leukemic clone after a very brief exposure to imatinib, suggesting that a small subclone of leukemia cells carrying this resistance mutation may have already been present before treatment. While the PDGFRA T674I mutation confers resistance to imatinib and dasatinib but appears to retain sensitivity to sorafenib, nilotinib and midostaurin (26), mutations involving the D842 residue seem to be associated with resistance to tolerated doses of all tested, currently available TKIs (25). D842 mutations have been extensively studied in gastrointestinal stromal tumors (GIST), where they show resistance to inhibition by TKIs which are currently approved as the first-line and second-line treatment of advanced GIST (27). More recently, crenolanib besylate has shown efficacy in a phase I/II clinical trial in advanced GIST patients with a PDGFRA D842V activating mutation, and this drug is now being tested in larger studies for relapsed/refractory GIST and other PDGFRA- and FLT3- mutated cancers (28). Although in our patient, detection of a PGFRA fusion did not ultimately improve the outcome, this may have been related to his advanced disease stage and likely existence of a resistance mutation in a small subclone of tumor cells prior to TKI therapy. It is plausible that addition of TKIs to the standard chemotherapy would have been more effective if instituted at the time of diagnosis, especially if the choice of TKI utilized could have been informed by sensitive molecular monitoring for resistance mutations.

Even though it is increasingly recognized that not all patients with FIP1L1-PDGFRA associated neoplasms have eosinophilia, our case further expands the spectrum of possible clinical and hematologic presentations of these diseases, and suggests that malignancies harboring these fusions may be under-recognized, including within the pediatric population. Considering the treatment implications, we suggest that screening for PDGFRA rearrangements should not be reserved only for patients with unexplained eosinophilia, but should also include other pediatric acute leukemias without readily detectable and well characterized molecular drivers.

This case also illustrates the utility of comprehensive multi-modal genetic testing in acute leukemia. Only combined use of conventional cytogenetic evaluation, CMA and NGS-based panel testing allowed complete and accurate genetic characterization of the patient’s disease. Furthermore, each of the utilized testing modalities revealed important driver mutations that would have been difficult to identify by other techniques: karyotyping detected the t(11;14) involving LMO2 and the T-cell receptor α/δ locus, CMA revealed submicroscopic deletions associated with the STIL-TAL1 and FIP1L1-PDGFRA fusions, and NGS confirmed previously identified abnormalities but also identified an activating PDGFA mutation associated with TKI resistance. In addition to routinely used cytogenetic analysis and FISH evaluation, broader availability and utilization of CMA analysis and NGS-panel testing will thus be critical in order to fully capture clinically actionable genetic abnormalities in pediatric acute leukemias.