Skip to main content

Table 1 Recurrent gene fusions in epithelial cancers of different body tissues and their role as clinical biomarkers

From: Landscape of gene fusions in epithelial cancers: seq and ye shall find

Tissue or body region Tumor type Aberration Genetic alteration Diagnostic/prognostic/therapeutic significance Reference
Thyroid gland Papillary thyroid cancer (PTC) (>80 % of thyroid cancers) RET gene fusions Multiple different 5′ partners (most common being CCDC6 (PTC1) and NCOA4 (PTC3)) fuse to 3′ partner RET 10–30 % of PTC cases. RET is an oncogenic receptor tyrosine kinase sensitive to FDA-approved drugs, including vandetanib and cabozantinib [15]
NTRK1 gene fusions 5′ activating gene partners including TPM3, TPR and TGF fuse with 3′ partner NTRK1 5 % of PTC cases. NTRK1 is an oncogenic receptor tyrosine kinase, potentially targetable by kinase inhibitors [15, 16]
ETV6-NTRK3 Chromosomal translocation t(12;15) (p13;q25) generates the fusion, with the dimerization domain of ETS family transcription factor (TF) ETV6 fused to the tyrosine kinase domain of NTRK3. Involves exon 14 of NTRK3, unlike other ETV6-NTRK3 fusions, which involve exon 13 Radiation-associated PTC (14.5 % post-Chernobyl); 2 % of sporadic PTC cases. Second only to RET fusions in prevalence [121]
Radiation-induced PTC AKAP9-BRAF In-frame fusion between exons 1–8 of the AKAP9 gene and exons 9–18 of BRAF protein kinase gene, lacking the auto-inhibitory N-terminal domain Fusion-positive tumors lack BRAF-activating point mutations. Fusion causes constitutive activation of BRAF and downstream MAPK pathways. Thus, a potential target for MEK inhibitors [70]
Follicular thyroid carcinoma (FTC) (10–20 % of thyroid cancers) PAX8-PPARγ Chromosomal translocation t(2;3)(q13;p25) results in chimeric protein involving the DNA-binding domain of the thyroid-specific TF PAX8 fused to PPARγ Fusion-positive FTCs appear to have a significantly better prognosis compared with those lacking this fusion. FTC cells expressing PAX8–PPARγ fusion protein show reduced tumor progression in a mouse xenograft model [22, 122]
Head and neck Pleomorphic adenoma PLAG1 gene fusions Multiple 5′ partners (CHCHD7, CTNNB1, FGFR1, LIFR, TCEA1) fuse to 3′ PLAG1 PLAG1 encodes a zinc finger TF that regulates IGF2 mitogenic signaling pathway [18, 91]
HMGA2 gene fusions HMGA2 is fused with different 3′ partners (including FHIT, NFIB, and WIF1) The fusion retains all the functional domains of HMGA2, and removes the 3′ UTR sequence that contains several inhibitory let7 microRNA binding sites. Absence of the Let-7-regulated 3′ UTR in the fusion transcript results in overexpression of HMGA2 that is sufficient for neoplastic transformation [19]
FGFR-PLAG1 FGFR is the 5′ partner, which, without its kinase domain, provides the promoter to drive the expression of the 3′ partner, PLAG1 This FGFR fusion product does not include the FGFR kinase domain, and therefore is not a target for FGFR inhibitors [91]
Adenoid cystic carcinomas (salivary glands, lacrimal glands, ceruminal glands; also breast) MYB-NFIB Inter-chromosomal gene fusion generating a chimeric transcript comprising almost the entire reading frame of the MYB oncogene fused to the last two exons of NFIB MYB likely provides the oncogenic activity, while NFIB primarily replaces a potentially inhibitory 3′ UTR of MYB [119, 120]
  Acinic cell carcinoma, cystadenocarcinoma, mammary analogue secretory carcinoma of salivary glands (MASC) ETV6-NTRK3 (TEL-TRKC) Chromosomal translocation t(12;15) (p13;q25) generates the ETV6-NTRK3 fusion, with the dimerization domain of the ETS family TF ETV6 fused to the tyrosine kinase domain of NTRK3 This fusion is now considered pathognomonic of MASC [103]
  Mucoepidermoid carcinoma (MEC) in the oral cavity (also lung, cervix and thyroid glands, and clear cell hidradenoma of skin) CRTC1-MAML1 or CRTC3-MAML2 Generated by chromosomal translocation t(11;19)(q14–21;p12–13). The product of the 3′ partner MAML2 acts as a co-activator of NOTCH independent of NOTCH ligand to impart the oncogenic phenotype. The CRTC-MAML2 fusion is restricted to MEC and has been associated with favorable prognosis. [107, 108, 110114, 116, 117]
Midline anatomical structures Nut midline carcinoma (NMC) BRD-NUT 75 % of NMCs express BRD4-NUT fusion proteins, the rest harbor BRD3 or other 5′ partner genes fused to NUT. BRD-NUT fusion proteins contain the N-terminal BET bromodomain, extraterminal domain, and nuclear localization signal fused to the entire coding region of NUT protein that contains a histone acetyltransferase binding domain NMC is a rare but aggressive squamous cell carcinoma originating from midline anatomical structures such as the head, neck or mediastinum (including the bladder, thymus, lung, and skeleton) that is defined by the presence of BRD-NUT fusions. BRD proteins have recently emerged as promising therapeutic targets [104, 105]
Kidney Renal cell carcinoma (RCC) TFE3 gene fusions Translocations at the Xp11.2 breakpoint result in gene fusions involving the TFE3 gene with various 5′ partners (ASPSCR1, PRCC, NONO, CLTC, and SFPQ) 15 % of patients with RCC aged <45 years have this aberration. Fusion-positive RCCs in older patients are more aggressive [20, 87]
ALK fusions In VCL-ALK fusions, the 3′ portion of the ALK transcript encoding the kinase domain is fused in frame to the 5′ portion of VCL Found in pediatric RMC that affects young black individuals with the sickle cell trait. In two independent reports, RMC tumors from three cases of African–American children with sickle cell anemia were found to harbor the VCL-ALK fusion [125, 126]
Non-clear cell renal cell carcinoma (nccRCC) CLTC-TFEB This encodes an in-frame fusion protein containing the conserved bHLH domain of TFEB (similar to other fusions involving TFEB), and is associated with the “MITF high” phenotype Associated with high expression of the anti-apoptotic protein BIRC7, thus potentially sensitive to apoptosis-sensitizing BIRC7 inhibitors that are under development [88]
ACTG1-MITF In this fusion protein the first 118 amino acids of MITF are replaced by the N-terminal 121 amino acids of ACTG1 Although found in only one sample, ectopic expression of the ACTG1-MITF fusion led to cellular transformation, suggesting a potential driver function [87]
Prostate Prostate cancer TMPRSS2-ERG The 5′ partner TMPRSS2 contributes prostate-specific, androgen-inducible upstream regulatory elements fused to the 3′ partner, encoding oncogenic ETS family TF ERG Probably the most prevalent gene fusion in epithelial carcinoma, with 40–50 % of localized prostate cancers found to harbor this fusion across multiple independent cohorts around the world. Associated with prostate carcinogenesis and distinct clinical correlates compared with fusion-negative prostate cancers [7, 8, 4652]
Fusions involving other ETS family genes, including ETV1, ETV4, ETV5, ELK4, and FLI1 5′ partners include androgen-inducible genes such as TMPRSS2, SLC45A3, and FLJ35294, and androgen-repressed C15ORF21, or housekeeping genes such as HNRPA2B1 and DDX5, fused to multiple 3′ oncogenic ETS family TF genes Together these represent 10–20 % of localized prostate cancers [7, 8, 24, 5356]
RAF gene fusions (SLC45A3-BRAF and ESRP1-RAF1) SLC45A3 is a prostate-specific, androgen-inducible gene fused upstream to gene encoding N-terminal-truncated BRAF, resulting in constitutive activation of this potent oncogene Although rare, BRAF/RAF1 fusions represent therapeutic targets [82, 83]
TMPRSS2-SKIL, SLC45A3-SKIL, MIPEP-SKIL, PIPOL1-SKIL, ACPP-SKIL, HMGN2P46-SKIL 5′ partners TMPRSS2, SLC45A3, and ACPP contribute prostate-specific, androgen-inducible upstream regulatory elements fused to 3′ partner SKIL, a negative regulator of SMAD SKIL fusions are observed in 1–2 % of prostate cancers and potentially upregulate the TGF-β pathway [101]
TBXLR1-PIK3CA, ACPP-PIK3CB Index cases with PIK3CA/B fusions show outlier expression of PIK3CA/B. ACPP imparts androgen-responsive expression to PIK3CB PIK3CA fusions may be responsive to PIK3CA inhibitors [83]
GRHL2-RSPO2 Index cases with RSPO2 fusions/rearrangements show outlier expression of RSPO2 RSPO2 is an agonist of the Wnt pathway and therefore may be responsive to porcupine inhibitors [83, 142]
Lung Lung cancer ALK gene fusions (most commonly EML4-ALK, but also TFG-ALK) EML4-ALK fusion encodes the N-terminal portion of EML4 fused to the intracellular portion of ALK, always retaining the tyrosine kinase domain EML4-ALK fusion is reported in 3–7 % of patients with NSCLC in different cohorts. ALK-fusion-positive lung cancers are sensitive to the FDA-approved kinase inhibitor crizotinib [10, 5963, 65, 66]
ROS1 gene fusions Multiple 5′ partners such as TPM3, SDC4, SLC34A2, CD74, EZR, LRIG3, and GOPC fused to ROS1. All of the fusion proteins retain the kinase domain of ROS1 2 % of lung cancer samples in one study [9, 43, 58, 65]
RET gene fusions Multiple isoforms of KIF5B-RET and CCDC6-RET. All of these products retain the kinase domain of RET Lung cancer cases with RET fusions may be candidates for FDA-approved RET inhibitor therapies such as vandetanib and cabozantinib [6469]
Mammary gland Breast cancer ETV6-NTRK3 (TEL-TRKC) Chromosomal translocation t(12;15) (p13;q25) generates ETV6-NTRK3 fusion, with the dimerization domain of the ETS family TF ETV6 fused to the tyrosine kinase domain of NTRK3 Almost 100 % of secretory breast carcinomas. ETV6-NTRK3 chimeric protein activates the IRS1 adapter protein, RAS-MAP kinase and PI3K-AKT pathways, and suppresses TGF-β signaling. ETV6-NTRK3-expressing cells and tumors are sensitive to the IGIFR/INSR kinase inhibitors BMS-536924 and BMS-754807 (currently in clinical trials) [21, 142]
MAST1 and MAST2 gene fusions 5′ partners including ZNF700, NFIX, and TADA2A fused to MAST1. ARID1A and GPBP1L1 fused to MAST2 serine/threonine kinase. All MAST fusions encode contiguous open reading frames, some retaining the canonical serine/threonine kinase domain, all retaining the PDZ domain and a 3′ kinase-like domain 3 % of breast cancer samples in one study [86]
NOTCH gene fusions SEC16A-NOTCH1, SEC22B-NOTCH2, NOTCH1 exon 2–exon 28 (intramolecular rearrangement) NOTCH fusions retain the NOTCH intracellular domain, which mediates downstream NOTCH signaling. The SEC16A-NOTCH1 fusion retains the γ-secretase cleavage site and shows sensitivity to γ-secretase inhibitors compared with SEC22B-NOTCH2, which loses this site [86]
EML4-ALK EML4 exon 13 fused to ALK exon 20, similar to NSCLC fusions One exon array profiling study reported EML4-ALK fusions in 2.4 % of breast carcinomas (5 of 209). One EML4-ALK fusion was detected in inflammatory breast cancer [82, 124]
Stomach Gastric cancer RAF gene fusions AGTRAP-BRAF: N-terminal protein AGTRAP fused to the C-terminal kinase domain of BRAF. SND1-BRAF: 5′ SND1 gene fused to BRAF, found in GTL16 gastric cancer cell line Both these fusions retain the kinase domain of BRAF, indicating potential responsiveness to RAF/MEK inhibitors [28, 82]
  CLDN18-ARHGAP26 CLDN18 on 3q22.3 fused to ARHGAP26 on 5q31.3. The fusion protein loses the PH domain of ARHGAP26, but retains the Rho-GAP and SH3 domains 3 % of Southeast Asian gastric cancers [27]
CD44-SLC1A2 Fusion involving adjacent genes (lying in opposite orientations on chromosome 13p) 1–2 % of gastric cancers [85]
Gut Colorectal cancer (CRC) EIF3E-RSPO2, PTPRK-RSPO3 Both these fusion proteins retain the functional domain of the R-spondins, which are known to be agonists of the canonical Wnt/β-catenin signaling pathway Recurrent fusions involving R-spondin family genes, EIF3E-RSPO2 (two cases) and PTPRK-RSPO3 (five cases) were detected by RNA sequencing of 68 “microsatellite stable” subtype CRC samples [85]
LACTB2-NCOA2 The fusion disrupts expression of NCOA2, which is an inhibitor of the Wnt/β-catenin pathway. This loss-of-function fusion thus represents a novel oncogenic mechanism in a subset of CRC Found in 6 of 99 (6.1 %) CRC cases [103]
VTI1A-TCF7L2, RP11-57H14.3- TCF7L2 Gene fusion involving activator of Wnt/β-catenin signaling pathway. VTI1A-TCF7L2 fusion lacks the TCF4 β-catenin-binding domain VTI1A-TCF7L2 was found in 3 of 97 CRCs. A screen for TCF7L2 fusion transcripts revealed its presence in more than 80 % of CRCs, 29 % of normal colonic mucosa, and 25–75 % of normal tissues from other organs. Thus, TCF7L2 fusion transcripts are neither specific to cancer nor to the colon or rectum. TCF7L2 fusion transcripts represent “read through” events [32, 33]
Skin Melanoma BRAF and RAF1 gene fusions Diverse N-terminal proteins fused to the BRAF/RAF kinase domain Seen in 3 % of melanomas; fusions retain the kinase domain of BRAF, indicating potential responsiveness to RAF/MEK inhibitors [82]
Other, non- recurrent aberrations RB1-ITM2B, PARP1-MIXL1, RECK-ALX3, TMEM8B-TLN1, CCT3-C1orf61, GNA12- SHANK2, ANKHD1-C5orf32 11 novel gene fusions were identified in 6 different patient samples, including both inter- and intra-chromosomal events. These fusions encode putative dominant-negative proteins (RB1, PARP1), and a truncated inhibitor of tumor invasion and metastasis (RECK) [81]
Central nervous system Gliomas PTPRZ1-MET The fusion involves translocation of introns 3 or 8 of PTPRZ and intron 1 of MET Found only in grade III astrocytomas (1/13; 7.7 %) or secondary GBMs (3/20; 15.0 %) [71]
Pilocytic astrocytoma BRAF/RAF1 gene fusions KIAA1549-BRAF, FAM131B-BRAF, SRGAP3-RAF1 Most frequently observed in pediatric brain tumors. Approximately 80 % of pilocytic astrocytomas and other low-grade gliomas harbor the KIAA1549-BRAF gene fusion [7280]
  1. FDA Food and Drug Administration, FTC follicular thyroid carcinoma, GBM glioblastoma multiforme, MASC mammary analog secretory carcinoma of salivary glands, MEC mucoepidermoid carcinoma, nccRCC non-clear-cell renal cell carcinoma, NMC NUT midline carcinoma, NSCLC non-small-cell lung carcinoma, PTC papillary thyroid cancer, RCC renal cell carcinoma, RMC renal medullary carcinoma, TF transcription factor