ETV6 fusions from insertions of exons 3-5 in pediatric hematologic malignancies

Current risk classification and treatment of patients with B-cell lymphoblastic leukemia (B-ALL) relies on rapid genomic testing for the identification of subtype-defining, prognostically significant, or potentially targetable alterations, which often occur as fusion events.^1,2 Karyotype and fluorescence in situ hybridization (FISH) are the traditional methodologies used for detecting fusions; however, there is increasing recognition of false positive and false negative results with these tests depending on the structure of the rearrangement. Although FISH has greater resolution than karyotype, it is still limited to targeted aberrations of approximately 100 kb or larger, thus clinically important fusions with false negative FISH results due to relatively small insertions have been observed. ETV6::RUNX1 fusions define one of the most common pediatric B-ALL subtypes comprising 20-25% of cases.³ Although not universally favorable, its status is used in some treatment protocols to classify patients as provisionally low risk.^4,5 These fusions typically result from chromosomal translocations t(12;21)(p13;q22) that are cryptic to karyotyping due to banding pattern similarities of the 12p and 21q chromosomal arms.⁵ As a result, FISH or reverse transcription polymerase chain reaction (RT-PCR) is usually performed for diagnosis. Here we report two pediatric B-ALL cases that were negative for ETV6::RUNX1 rearrangements by FISH, but were deduced through targeted RNA next generation sequencing (NGS) to harbor ETV6::RUNX1 fusions characterized by focal insertions of ETV6 exons 3-5 into RUNX1 with concomitant intragenic deletions of the same exons from ETV6. We further report one pediatric AML case with an ETV6::EP300 rearrangement similarly involving insertion and deletion of ETV6 exons 3-5. Finally, by searching public whole-transcriptome sequencing data, we identify another ETV6::RUNX1-positive B-ALL with the same pattern and characterize the underlying insertion and deletion breakpoints on the DNA level.

Table 1.Patient characteristics and molecular findings.

Cases 1 and 2 were identified within 2 weeks of each other during routine review at Boston Children’s Hospital of a clinically validated targeted RNA NGS assay for detecting fusions based on ArcherDx FusionPlex Heme v2 (FPHeme; IDT, Coralville, IA, USA). Cytogenetics and FISH were performed at Integrated Oncology or partner institutions. Targeted DNA NGS was performed at Brigham and Women’s Hospital. Isoform analysis of RNA sequencing data was performed using the software isoformSR (https://github.com/ht50/isoformSR), as previously described.⁶ Clinical features and molecular findings are summarized in Table 1.

Figure 1.RNA sequencing results. (A) ETV6 exon skipping isoform and ETV6 fusion transcripts detected on targeted RNA sequencing (FPHeme) over historical cohort (N=474) consisting of cases 1-3 (black square, circle, and triangle) and others (gray x’s). Cases 1-3 had outlier expression of the ETV6 isoform junction e2e6 connecting exon 2 to exon 6, consistent with an underlying intragenic deletion of the skipped exons 3-5, as well as fusion breakpoints consistent with insertion of exons 3-5 between exons 2 and 3 of the partner gene RUNX1 (cases 1 and 2) or between exons 1 and 2 of the partner gene EP300 (case 3). Case 1 was also associated with single copy whole gene deletion of ETV6 and accordingly had the highest expressed variant allele frequency (VAF) of e2e6. (B) Clustering by t-distributed stochastic neighbor embedding (t-SNE) of the targeted transcriptional profiles (100 genes) of cases 1 and 2 among all B-cell lymphoblastic leukemia (B-ALL) cases (N=209) colored by subtype including ETV6::RUNX1 fusions (green dots). (C) Screening of public total RNA sequencing data identified one more pediatric ETV6::RUNX1-positive B-ALL case (sample ID: 36; SRR15301257) with ETV6 e2e6 at high VAF (95.3%; not shown), fusion breakpoints consistent with insertion of ETV6 exons 3-5 between RUNX1 exons 2 and 3, and single copy whole gene deletion of ETV6 by targeted DNA next generation sequencing. Split-read analysis of intronic alignments identified 3 sets of expressed DNA breakpoints within split-reads: (i) chr12:11952219+ (ETV6 intron 2) to chr12:12032918+ (ETV6 intron 5) illustrated in the example paired split-read SR1 (light blue), (ii) chr21:36296008-(RUNX1 intron 2) to chr12:11951870+ (ETV6 intron 2) illustrated in the example paired split-read SR2 (yellow), and (iii) chr12:12032916+ (ETV6 intron 5) to chr21:36296069- (RUNX1 intron 2) illustrated in the example paired split-read SR3 (gray). The breakpoints of SR1 implied a deletion of the intervening 80.7 kb segment between the ETV6 intron 2 and intron 5 breakpoints. The breakpoints of SR2 and SR3 implied a focal insertion of the near-exact 81 kb segment of ETV6 (chr12:11951870+ to chr12:12032916+) into RUNX1 intron 2 at breakpoints separated by 62 bp (chr21:36296008- and chr21:36296069-), thus also associated with a duplication of this short 62 bp segment. The inclusion of an additional ~350 bp from intron 2 in the inserted segment compared to the deleted segment suggested some amount of repair/replication within intron 2 during formation of the deletion. In theory, the breakpoints of SR2 and SR3 alternatively could represent a reciprocal translocation with duplication of ETV6 exons 3-5, however single copy loss of the other ETV6 allele made it impossible for the remaining allele to harbor both a reciprocal translocation and a deletion of the same region. The intronic split-reads harboring DNA deletion and rearrangement breakpoints were presumably derived from pre-mRNA, as described in other studies.⁶ Notation - VAF: expressed variant allele fraction. 5’ and 3’: refers to the 5’ and 3’ ends of the split-read alignments. +/-refers to the DNA strand of the alignment; note that ETV6 is transcribed from the plus strand (+) of chromosome 12 whereas RUNX1 is transcribed from the minus strand (-) of chromosome 21.

Case 1 was from a 4-year-old male with newly-diagnosed B-ALL with unrevealing genetic analysis, including normal karyotype and negative FISH for ETV6::RUNX1 and other subtype-defining rearrangements. FPHeme subsequently revealed an ETV6::RUNX1 fusion with standard breakpoints connecting ETV6 exon 5 (NM_001987.5) to RUNX1 exon 3 (NM_001754.5), together with an atypical reciprocal RUNX1::ETV6 fusion with rare breakpoints connecting RUNX1 exon 2 to ETV6 exon 3 and abnormally high expression (quantified by split-reads sequencing the junction) of an in-frame ETV6 exon-skipping isoform junction connecting exon 2 to exon 6 (i.e., skipping exons 3-5), satisfying previously established criteria for “outlier expression” relative to other FPHeme cases (Figure 1A). We have previously shown outlier expression to be a sensitive and specific marker for underlying intragenic deletions in other genes, thus this case was most consistent with DNA deletion of ETV6 exons 3-5.⁶ Despite its unusual aspects, the targeted gene expression profile clustered with B-ALL cases harboring typical FISH-positive ETV6::RUNX1 fusions (Figure 1B). Theoretically, the observed fusion breakpoints could represent either a reciprocal translocation or a focal insertion; however, FISH analysis showed loss of one ETV6 signal in 68.5% of cells, similar to the 66% blast estimate by flow cytometry and confirmed to represent a single-copy, whole-gene ETV6 deletion by targeted DNA sequencing, making it impossible for the single remaining ETV6 allele to harbor both deletion of ETV6 exons 3-5 and a reciprocal translocation to RUNX1. The overall findings, therefore, implied an intragenic deletion of ETV6 exons 3-5 with associated insertion between RUNX1 exons 2 and 3 (Figure 2). Case 2 was from a 6-year-old female with newly diagnosed B-ALL and similar findings of a normal karyotype and negative FISH for ETV6::RUNX1 and other fusions. FPHeme again revealed ETV6::RUNX1 and RUNX1::ETV6 fusions with the same exon structure as case 1, outlier expression of the same ETV6 exon-skipping isoform, and a gene expression profile clustering with typical cases of ETV6::RUNX1 B-ALL. Given this recurrent pattern of insertions and deletions of intragenic exons, we retrospectively searched historical FPHeme cases (n=474) for outlier expression of any ETV6 exon-skipping isoform as a potential marker for cryptic insertions. We identified one additional case with outlier expression involving skipping of ETV6 exons 3-5 (case 3), but did not identify outlier expression involving skipping of other exons.

Case 3 was from a 5-year-old female with relapsed acute myeloid leukemia (AML) with detected in-frame reciprocal fusions ETV6::EP300 connecting exon 5 of ETV6 to exon 2 of EP300 (NM_001429.4) and EP300::ETV6 connecting exon 1 of EP300 to exon 3 of ETV6 (Figure 1A). The predicted chimeric protein contained the helix-loop-helix (HLH) domain of ETV6 inserted between the nuclear localization sequence and transactivation domains of EP300 and retained the EP300 chromatin modification region (Figure 2). Outlier expression of the ETV6 isoform skipping exons 3-5 was again observed, consistent with single-copy intragenic deletion (Figure 1A). Per report, interphase FISH analysis showed two strong ETV6 signals and one weak ETV6 signal. Meta-phase FISH analysis showed that the weak ETV6 signal was located on a small G-size chromosome. EP300 FISH was not performed. Based on these data, we hypothesize that the weak ETV6 signal represented partial binding of the ETV6 probe to ETV6 exons 3-5 inserted into the EP300 gene on chromosome 22. Another case of ETV6::EP300, with unspecified structure, has been reported in the literature.⁷ Finally, we analyzed ETV6 exon-skipping isoforms in public total RNA sequencing data from a pediatric ALL cohort, revealing one ETV6::RUNX1 case (sample ID: 36) with outlier expression of the isoform skipping exons 3-5, comprising 3.4% (1/29) of ETV6::RUNX1-positive B-ALL in the cohort, where it again co-occurred with the only instance of a reciprocal fusion connecting RUNX1 exon 2 to ETV6 exon 3.^6,8 Targeted DNA NGS showed single copy loss of ETV6, thus the exon-skipping isoform was expressed at a high variant allele fraction (95.3%; data not shown), similar to case 1 (Figure 1A). Intronic split-read analysis identified three sets of expressed DNA breakpoints derived presumably from pre-mRNA, compatible with deletion of 80.7 kb between ETV6 introns 2 and 5 (chr12:11,952,220 to chr12:12,032,917) and insertion of a near-exact 81.0 kb segment (chr12:11,951,870 to chr12:12,032,916) into RUNX1 intron 2 at breakpoints separated by 62 bp (chr21:36,296,008 and chr21:36,296,069) (Figure 1C). Per report, FISH analysis described a non-classical ETV6::RUNX1 rearrangement, where the translocated 5’ part of ETV6 to the derived 21 demonstrated a smaller green signal than usual, together with an absence of the small extra red signal that ordinarily represents the 5’ part of RUNX1 translocated to the derived 12. The FISH positivity of this case, in contrast to cases 1-2, might be related to differences in FISH probes or ETV6 insertion sizes, which theoretically could be as small as ~30 kb versus ~81 kb in this case.

Figure 2.Schematics of ETV6 insertions. Structure of the rearrangements detected in cases 1 and 2 (left side) and case 3 (right side) with the predicted chimeric protein products.

An ETV6::RUNX1 fusion cryptic to FISH but detected by RT-PCR, which was designed to amplify any fusion connecting ETV6 exon 5 to RUNX1 exon 3 or exon 4 (including the insertion fusions described here), was previously reported in a pediatric B-ALL, although underlying genomic structure was not determined.⁹ Importantly, in the absence of RNA sequencing or RT-PCR, such FISH-negative cases would likely remain unclassified and, in some clinical protocols, may lead to unintended higher risk stratification and more intensive treatment regimens.^3,4 Similarly, although uncommon in pediatric AML, ETV6 rearrangements are important to identify given their association with adverse risk regardless of fusion partner; chromosome 12p abnormalities/ETV6 rearrangements are accordingly an indication in pediatric AML for allogeneic hematopoietic stem cell transplantation in first remission in the current Children’s Oncology Group AAML1831 clinical trial (clinicaltrials gov. Identifier: NCT04293562).^7,10 The mechanistic consequences of insertion events will require elucidation. The gene expression profiles of cases 1-2 were similar to typical cases of ETV6::RUNX1 positive B-ALL in our historical cohort, suggesting functional similarity. An ETV6::RUNX1-like gene expression profile, which may also confer a favorable prognosis, has been described as a provisional entity in otherwise unclassified B-ALL and has been associated with inactivating ETV6 variants and shown to be mediated by microsatellite enhancers ordinarily bound by ETV6.^2,11 Detection and screening for exon-skipping isoforms may be an effective way of recognizing fusions resulting from concurrent deletion and insertion of intragenic genomic material, which otherwise may be mistaken for typical balanced reciprocal translocations in standard short read sequencing data. Indeed, targeted RNA sequencing of case 3 at a partner institution reported the ETV6::EP300 fusion as connecting exons 1-5 of ETV6 to exons 2-31 of EP300. The deletion-insertion pattern has also been described in the context of YAP1::KMT2A rearranged sarcomas although screening our cohort did not reveal outlier expression of any KMT2A exon-skipping isoforms in either KMT2A-rearranged or non-rearranged hematologic cases.¹² In summary, clinically relevant ETV6 fusions from focal insertions of ETV6 may be more common in childhood leukemia than previously recognized, particularly in FISH-negative cases. Larger studies are necessary to determine their true frequency, assess clinical significance, and inform the use of NGS fusion testing in B-ALL.

Footnotes

• Received January 19, 2023
• Accepted June 19, 2023

Correspondence

H. K. TSAI - Harrison.Tsai@childrens.harvard.edu

Disclosures

No conflicts of interest to disclose.

Contributions

SM and HKT designed the study, performed data analysis, and wrote the manuscript. SKT, YP, LBS, and MHH contributed data. All authors edited the manuscript.

Data-sharing statement

Select data is available upon request to the corresponding author.

Funding

Acknowledgments

SKT is a scholar of the Leukemia and Lymphoma Society and holds the Joshua Kahan Endowed Chair in Pediatric Leukemia Research. This work includes two cases from clinical trials (clinicaltrails gov. Identifier: NCT02670525 and NCT03020030).