Department of Pathology, University of Pittsburgh School of Medicine, Pittsburgh, Pennsylvania
Corresponding author: Yuri E. Nikiforov, MD, PhD, Department of Pathology, University of Pittsburgh, Clinical Laboratory Building, Room 8031, 3477 Euler Way, Pittsburgh, PA 15213; Fax: (412) 802-6799; email@example.com
The first two authors contributed equally to this article.
We thank the staff of the University of Pittsburgh Health Sciences Tissue Bank for providing samples of sporadic thyroid tumors used in this study. We are grateful to Dr. Geraldine Thomas (Imperial College, London, UK) for her valuable assistance and support of the study. We also gratefully acknowledge the confirmation of diagnoses provided by the International Pathology Panel of the Chernobyl Tissue Bank; and we are also grateful to our dosimetry colleagues, including Drs. I. A. Likhtarev, L. N. Kovgan, A. Bouville, and V. Drozdovitch, for their valuable contributions to iodine-I thyroid dose estimates.
In their previous analysis of papillary thyroid carcinomas (PTCs) from an Ukrainian-American cohort that was exposed to iodine-131 (131I) from the Chernobyl accident, the authors identified RET/PTC rearrangements and other driver mutations in 60% of tumors.
In this study, the remaining mutation-negative tumors from that cohort were analyzed using RNA sequencing (RNA-Seq) and reverse transcriptase-polymerase chain reaction to identify novel chromosomal rearrangements and to characterize their relation with radiation dose.
The ETS variant gene 6 (ETV6)-neurotrophin receptor 3 (NTRK3) rearrangement (ETV6-NTRK3) was identified by RNA-Seq in a tumor from a patient who received a high 131I dose. Overall, the rearrangement was detected in 9 of 62 (14.5%) post-Chernobyl PTCs and in 3 of 151 (2%) sporadic PTCs (P = .019). The most common fusion type was between exon 4 of ETV6 and exon 14 of NTRK3. The prevalence of ETV6-NTRK3 rearrangement in post-Chernobyl PTCs was associated with increasing 131I dose, albeit at borderline significance (P = .126). The group of rearrangement-positive PTCs (ETV6-NTRK3, RET/PTC, PAX8-PPARγ) was associated with significantly higher dose response compared with the group of PTCs with point mutations (BRAF, RAS; P < .001). In vitro exposure of human thyroid cells to 1 gray of 131I and γ-radiation resulted in the formation of ETV6-NTRK3 rearrangement at a rate of 7.9 × 10−6 cells and 3.0 × 10−6 cells, respectively.
Thyroid cancer is the most common type of endocrine malignancy, and its incidence has been steadily growing in the United States and many other countries during the last 4 decades. Exposure to ionizing radiation during childhood is a well established risk factor for thyroid cancer. The increased risk of thyroid cancer after radiation exposure was first suggested in 1950 in infants who received external-beam radiation for enlarged thymus. This was confirmed later in multiple studies of patients who had exposure to environmental or medical radiation, including X-ray or γ-radiation as wells as radioiodines, mainly iodine-131 (131I). In the decades after the 1986 Chernobyl accident, the surrounding geographic area experienced a marked increase in the incidence of thyroid cancer among those who were children or young adults at the time of the accident. The post-Chernobyl case-control and cohort studies confirmed previous observations that papillary thyroid carcinoma (PTC) is the predominant type of thyroid cancer associated with radiation exposure[4-6] and that the risk for PTC after exposure increases with dose, with the magnitude of the increase comparable to that after external radiation.[7-10]
Activating mutations in the mitogen-activated protein kinase (MAPK) signaling cascade are common in thyroid cancer and are believed to be essential for tumorigenesis. The most common events include point mutations in the BRAF and RAS genes as well as chromosomal rearrangements involving the RET gene, known as RET/PTC. However, the mutational mechanisms leading to MAPK activation in sporadic and radiation-related PTCs appear to be different. Whereas, in sporadic tumors, point mutations in the BRAF and RAS genes are by far the most common (approximately 60% of all PTCs), in post-Chernobyl or postradiotherapy PTCs, 50% to 80% of tumors typically harbor chromosomal rearrangements of the RET gene known as RET/PTC, whereas point mutations are rare.[12-14] Other chromosomal rearrangements, such as A kinase (PRKA) anchor protein 9-BRAF (AKAP9-BRAF) and rearrangements involving the neurotrophic tyrosine kinase receptor type 1 (NTRK1) and peroxisome proliferator-activated receptor gamma (PPARγ) genes, also are identified more frequently in radiation-associated PTCs.[15, 16] However, a significant proportion of radiation-associated tumors harbor none of the known mutations, suggesting that other, unknown genetic events may occur in these tumors.
In our recent study, we performed genotypic analysis of 62 PTCs from a well characterized cohort of Ukrainian individuals (UkrAm) who received 0.008 to 8.6 gray (Gy) of 131I to the thyroid after the Chernobyl accident. That study confirmed that the RET/PTC rearrangement was the most common genetic event in these tumors and identified different trends with dose in PTCs harboring chromosomal rearrangements and point mutations. However, 40% of those tumors had none of the known genetic events and were analyzed further in the current study using RNA sequencing (RNA-Seq) analysis to discover novel genetic events that might occur in radiation-related thyroid cancer. This analysis reveals that the ETV6-NTRK3 chromosomal rearrangement, which was previously unknown in thyroid cancer, is a common genetic event in radiation-related thyroid cancer, but not in sporadic thyroid cancer, and demonstrates that ETV6-NTRK3 rearrangements can be directly induced in human thyroid cells by ionizing radiation in vitro.
MATERIALS AND METHODS
Study Cases and Samples
This study was approved by the University of Pittsburgh, the National Cancer Institute, and the Institute of Endocrinology and Metabolism (IEM) (Kiev, Ukraine) institutional review boards. The radiation-associated PTCs were diagnosed among individuals from the UkrAm study who were aged < 18 years at the time of the Chernobyl accident. Individual radioactivity measurements in thyroid glands were performed within the 2 months after the accident, and individual 131I thyroid doses were estimated based on those measurements, interview data concerning dietary and lifestyle habits, and environmental transfer models.[18, 19] PTC was diagnosed in 104 individuals (age at surgery: range, 14-32 years; mean ± standard deviation, 22.7 ± 5.1 years) between 1998 and 2008 at the Laboratory of Morphology of Endocrine System of the IEM after 4 sequential screenings. Pathologic diagnoses were reviewed by the International Pathology Panel of the Chernobyl Tissue Bank (CTB). Frozen tissue samples were available for 75 PTCs. For 74 PTCs, DNA and/or RNA were extracted at the IEM or at Imperial College (London, UK) and were received through the CTB. Four PTCs from individuals who were exposed to 131I in utero and 8 PTCs that lacked either DNA (n = 3) or RNA (n = 5) were excluded. In addition, a series of 151 consecutive sporadic PTCs (age at surgery: range, 15-97 years; mean ± standard deviation, 45.6 ± 17.7 years) and 92 other PTCs (age at surgery: range, 4-77 years; mean, 44.2 ± 16.7 years) that were previously classified as negative for known genetic alterations were available through the University of Pittsburgh Health Sciences Tissue Bank.
RNA-Seq and Data Analysis
Tumor RNA samples were processed to remove ribosomal RNA using the Ribozero Magnetic Gold kit (Illumina, Inc., San Diego, Calif), followed by library preparation for RNA sequencing using version 2 of the Illumina TruSeq RNA Sample Preparation Kit. Briefly, polyadenylated RNA was fragmented, reverse transcribed, indexed, amplified, and purified to produce individual bar-coded libraries, according to the manufacturer's instructions. The prepared libraries were assessed using a Bioanalyzer and the High Sensitivity DNA kit (Agilent Technologies, Santa Clara, Calif). Paired-end sequencing was performed on the Illumina HiSeq2000 at the High Throughput Genome Center in the Department of Pathology, University of Pittsburgh. The sequence reads obtained were analyzed for gene fusion events using the ChimeraScan[22, 23] and deFuse programs. The predicted fusion events from the 2 programs were integrated and combined with genomic annotation to generate a list of candidate gene fusions. Before the analysis, sequences with low quality (base quality < 13) at both ends of reads were trimmed, and the trimmed reads with < 25 base pairs were removed. Reference human genome (National Center for Biotechnology Information, build 37.1, hg19) and gene annotation databases (Ensembl v69 and University of California-Santa Cruz hg19) were used for the analysis. To reduce false-positive findings, the fusion events detected by both programs were further narrowed down by excluding 1) fusion events between adjacent genes (called as read-throughs), 2) fusion events with no reads spanning the predicted breakpoints, and 3) fusion events predicted to have 5 or more fusion partners and lacking specificity of target regions.
Detection of ETV6-NTRK3 Fusions by Reverse Transcriptase-Polymerase Chain Reaction
Tumor RNA was reverse transcribed and amplified using the following primers: 5′-CATTCTTCCACCCTGGAAAC-3′ (forward ETV6 exon 4), 5′-AAGCCCATCAACCTCTCTCA-3′ (forward ETV6 exon 5), and 5′-TCCTCACCACTGATGACAGC-3′ (common reverse NTRK3). Polymerase chain reaction (PCR) products were analyzed by agarose gel electrophoresis. The presence of the fusion was confirmed by Sanger sequencing.
Cell Irradiation and in Vitro Induction of ETV6-NTRK3 Fusions
HTori-3 human thyroid cells were grown in RPMI 1640 supplemented with 10% fetal bovine serum. Cells were authenticated using short tandem repeat analysis. For γ-radiation, 1 × 10 HTori-3 cells were exposed to 1 Gy from a cesium-137 source (GammaCell 40 irradiator, Nordion (Kanata, Ontario, Canada)) at a dose rate of 0.58 Gy per minute. For 131I irradiation, 1 × 106 HTori-3 cells in a T25 flask were incubated for 24 hours in 2 mL of culture media in the presence of radio-I131 to deliver 1 Gy of radiation. Calculation of the dose received by a monolayer of cells growing in the T25 flask and exposed to 1 Gy of 131I was performed based on Kernel integrations, as previously described, and was 1.02 Gy per hour per 1 mCi. Two replicates of each experiment were performed. Delivery of radiation was monitored by formation of γH2AX nuclear foci using antiphosphorylated histone H2AX primary antibody (Upstate Biotechnology, Inc., Lake Placid, NY). After γ or 131I irradiation, HTori-3 cells were split into 30 T25 flasks, transferred to T75 flasks for continuous growth, and harvested 9 days after irradiation. RNA was extracted using Trizol (Invitrogen, Carlsbad, Calif), and messenger RNA was purified using the Oligotex mRNA kit (Qiagen, Hilden, Germany). After the reverse-transcription step, multiplex PCR was performed to detect ETV6e4-NTRK3e14 and ETV6e5-NTRK3e14 rearrangements using the sequence-specific primers described above. PCR products were resolved in the agarose gel and detected by Southern blot hybridization with 32P-labeled oligonucleotide probes specific for ETV6e4-NTRK3e14 (5′-ACCATGAAGAAGGTCCCGT-3′) and ETV6e5-NTRK3e14 (5′-AGAATAGCAGGTCCCGTGG-3′).
Univariate analyses were performed using the 2-sample t test. Mutation prevalence data were analyzed using standard logistic regression models as previously described. Briefly, the following model was used to examine the probability of ETV6-NTRK3 rearrangement with 131I dose, D (in Gy), controlling for the effect of age at surgery (a), sex (s), and oblast (province) of residence (oblast [O]):
For some analyses, quartic polynomials or categorical functions of age replaced the a1(a − 25) term. Most analyses used log-linear functions of dose, D, but a few involved log-quadratic functions of dose. Twenty-five years were subtracted from age (at surgery) to aid convergence of fitted models. All tests were 2-sided and were based on the likelihood-ratio test, and confidence intervals (CIs) for the logistic regression analyses were derived from the profile likelihood. Likelihood-ratio tests also were used to assess heterogeneity by endpoint, using an extension of previously described methods. Linear regression analyses were performed using the Stata software package (StataCorp, College Station, Tex), and log-linear logistic regression analyses were conducted using the program Epicure (Risk Sciences International, Toronto, Ontario, Canada).
Identification and Prevalence of ETV6-NTRK3 Rearrangements in Radiation-Related and Sporadic PTCs
The previous genotyping analysis of 62 post-Chernobyl PTCs from the UkrAm cohort identified driver mutations in 37 tumors (60%), whereas 25 tumors (40%) did not harbor any other mutations previously reported to occur in thyroid cancer. From this group, 2 tumors had a sufficient amount of RNA and were selected for whole-transcriptome (RNA-Seq) analysis to search for novel chromosomal rearrangements. One of these tumors was from a patient who was aged 1 year at the time of the Chernobyl accident and received an estimated 131I dose of 7.5 Gy to the thyroid—among the highest doses in this series. Another tumor was from a patient who was aged 10 years at the time of exposure and received an estimated 131I dose of 0.34 Gy. An analysis of the PTC from the first patient revealed an in-frame fusion event between exon 4 of the ETV6 gene and exon 14 of the NTRK3 gene, which was detected by both programs, ChimeraScan and deFuse. An RNA-Seq analysis of the PTC sample from the second patient did not yield any promising gene fusions involving potential oncogenes. The presence of the ETV6-NTRK3 rearrangement was validated by reverse transcriptase-PCR and was confirmed by Sanger sequencing (Fig. 1).
Upon screening the remaining post-Chernobyl tumors using reverse transcriptase-PCR, 8 additional tumors that were positive for ETV6-NTRK3 fusion were identified, all of which involved exon 4 of ETV6 and exon 14 of NTRK3. Overall, 9 of 62 (14.5%) post-Chernobyl PTCs harbored this rearrangement. One of these tumors also harbored a BRAF V600E mutation (a valine [V] to glutamic acid [E] substitution at codon 600), and another tumor also had the RET/PTC1 rearrangement, whereas the remaining 7 ETV6-NTRK3–positive tumors lacked known common driver mutations (BRAF, RAS, RET/PTC, or paired box gene 8 [PAX8]-PPARγ).
Screening of 151 consecutive sporadic PTCs from the general US population revealed 3 positive tumors, resulting in a prevalence of 2%. None of the 3 ETV6-NTRK3–positive sporadic tumors harbored other common driver mutations known to occur in thyroid cancer. The prevalence of ETV6-NTRK3 in post-Chernobyl PTCs was significantly higher than in sporadic PTCs, including both crude prevalence (P = .01) and prevalence after adjustment for age and sex (P = .019).
Analyses of an additional 92 sporadic PTCs that were selected based on the lack of other known driver mutations identified 4 tumors that were positive for ETV6-NTRK3 (4.3%). Of the 7 ETV6-NTRK3 rearrangements identified in sporadic PTCs, 6 involved the fusion of exon 4 of ETV6 to exon 14 of NTRK3 (ETV6e4-NTRK3e14), and 1 revealed a larger PCR product, which Sanger sequencing indicated was caused by the fusion of exon 5 of ETV6 to exon 14 of NTRK3 (ETV6e5-NTRK3e14) (Fig. 1). None of the 7 patients who had sporadic PTCs that carried an ETV6-NTRK3 rearrangement had a documented history of radiation exposure. Rescreening of post-Chernobyl tumors for ETV6e5-NTRK3e14 rearrangement revealed no additional positive cases. No ETV6-NTRK3 rearrangements were identified in the TPC1 cell line established from PTC.
Exposure-Related Features of ETV6-NTRK3–Positive Tumors
In 62 post-Chernobyl PTCs, the average thyroid 131I dose received was 1.27 Gy. Among the ETV6-NTRK3–positive tumors (n = 9), the average 131I dose was 2.27 Gy (Fig. 2A). The age of these individuals at the time of exposure ranged from 0.5 to 17.2 years (mean age at exposure, 8.1 years), their age at surgery ranged from 14.2 to 35.1 years (mean age at surgery, 23.9 years), and the average time between exposure and surgery was 15.8 years (Table 1). In multivariate analysis (adjusting for age at surgery, sex, and place [oblast] of residence), tumors that harbored the ETV6-NTRK3 rearrangement were associated with increasing 131I dose, albeit with borderline significance (P = .126) (Table 2). The dose response adjusted for these variables is illustrated in Figure 2B.
Table 1. Iodine-131 Dose and Other Exposure-Related Characteristics by Mutation Type in Post-Chernobyl Papillary Thyroid Cancer
When the ETV6-NTRK3–positive tumors were grouped with tumors that harbored other types of chromosomal rearrangements, ie, RET/PTC or PAX8-PPARγ, rearrangement-positive tumors were associated with a significantly higher dose response compared with tumors that had point mutations (BRAF, NRAS, HRAS) adjusting for age, sex, and oblast (P < .0001) (Table 2). Specifically, the adjusted excess odds ratio (EOR) per Gy for all chromosomal rearrangements was 0.09 (95% CI, −0.24, 0.46) whereas the EOR for point mutations was −3.29 (95% CI, −6.06, −1.38) (Table 2). There was a similar degree of heterogeneity (P < .0001) if adjustment also was made for a possible quadratic term in the dose response for rearrangements.
Histopathologic Features of ETV6-NTRK3–Positive PTCs
The majority of post-Chernobyl PTCs (6 of 9 tumors; 67%) in which the ETV6-NTRK3 rearrangement was identified demonstrated a follicular growth pattern and were classified as the follicular variant of PTC (Fig. 3A). The remaining tumors (3 of 9 tumors; 33%) had a significant papillary component and were classified as the classic papillary type of PTC (Fig. 3B). All 9 tumors had some component of a solid growth pattern, comprising approximately 10% of the examined tumor in 7 PTCs and 20% to 30% of the examined tumor in 2 PTCs (Fig. 3B). Among sporadic PTCs that were positive for ETV6-NTRK3 rearrangement, 4 were the classic papillary type (57%) and 3 were follicular variant PTCs (43%). All 4 classic papillary PTCs had a significant component of follicular growth, but no well defined solid component was observed in any of the sporadic tumors. Among radiation-associated tumors, all were American Joint Committee on Cancer/International Union Against Cancer stage I at presentation; whereas, among sporadic tumors, 5 were stage I and 2 were stage III based on the presence of minimal extrathyroidal extension in patients aged > 45 years.
In Vitro Induction of ETV6-NTRK3 Rearrangements by Ionizing Radiation
The high prevalence of the ETV6-NTRK3 rearrangement in radiation-associated PTCs and its association with high thyroid dose suggested that this rearrangement could be induced by ionizing radiation. To test this possibility, we studied induction of the ETV6-NTRK3 rearrangement in HTori-3 human thyroid cells after exposure to 1 Gy of 131I or γ-radiation. The 131I dose distribution in the monolayer of cells within a T25 flask generated by Kernel integration is illustrated in Figure 4A. Induction of double-stranded DNA breaks by radiation was monitored by the formation of γH2AX nuclear foci (Fig. 4B). Whereas no rearrangements were observed in the unexposed cells, both types of radiation exposure resulted in the generation of ETV6-NTRK3 rearrangements (Fig. 4C).The average rate of ETV6-NTRK3 induction was 7.9 × 10−6 cells per 1 Gy of 131I and 3.0 × 10−6 per 1 Gy of γ-radiation. Both ETV6e4-NTRK3e14 and ETV6e5-NTRK3e14 fusions were induced by ionizing radiation, with the predominance of ETV6e4-NTRK3e14 after both 131I and γ-radiation (Fig. 4D).
In this article, we report for the first time the occurrence of the ETV6-NTRK3 chromosomal rearrangement in thyroid cancer, which was identified as common in PTCs associated with 131I exposure from the Chernobyl accident. Indeed, in this well characterized series of radiation-related thyroid cancers, ETV6-NTRK3 was the second most common rearrangement type after RET/PTC. Moreover, this study demonstrates that the ETV6-NTRK3 rearrangement can be directly induced in human thyroid cells in vitro by exposure to both 131I and γ-radiation.
The fusion between the ETV6 gene on chromosome 12 and the NTRK3 gene on chromosome 15 was first described in congenital fibrosarcoma in 1998. Since then, ETV6-NTRK3 rearrangements have been identified in several other tumor types, including acute myeloid leukemia (AML),[32, 33] chronic eosinophilic leukemia, congenital mesoblastic nephroma, secretory breast carcinoma, and mammary analog secretory carcinoma of the salivary gland. ETV6, also known as TEL, is a transcription factor from the ETS transcription factor family, which is involved in various oncogenic gene fusions resulting from chromosomal translocations, mostly reported in subtypes of AML. NTRK3 is a transmembrane receptor tyrosine kinase, and its ligand is neurotrophin-3,which is primarily involved in neuronal cellular processes. The rearrangement results in fusion of the SAM domain of ETV6, which is required for dimerization, with the tyrosine kinase domain of NTRK3, such that the transcribed product is a constitutively active tyrosine kinase.
Most of the ETV6-NTRK3 fusions that occur in various tumor types involve the fusion point initially identified in congenital fibrosarcoma, ie, between exon 5 of ETV6 and exon 13 of NTRK3. A shorter variant, in which exon 4 of ETV6 is fused to NTRK3, has been identified in patients with of AML and chronic eosinophilic leukemia.[32, 34] Here, we report the occurrence of ETV6-NTRK3 rearrangements in radiation-related and sporadic PTCs, with fusion points that differ from those previously identified in other tumor types because they lack exon 13 of NTRK3.
Our results provide several lines of evidence that link ETV6-NTRK3 rearrangements in PTC with radiation exposure. First, this rearrangement is observed with significantly greater prevalence in PTCs among patients who were exposed to 131I from the Chernobyl accident than in PTCs arising in the general US population. Second, among post-Chernobyl PTCs, we observed a borderline statistically significant association between 131I dose and the prevalence of ETV6-NTRK3 rearrangement. Finally, our in vitro experiments demonstrated the induction of both types of ETV6-NTRK3 fusions in human thyroid cells by radiation. Both 131I and γ-radiation were efficient at inducing this rearrangement in cultured human cells. This suggests that, in addition to patients exposed to 131I after the Chernobyl accident, thyroid cancers that develop after external-beam radiation therapy may also harbor these rearrangements.
Among post-Chernobyl tumors in this study, ETV6-NTRK3–positive PTCs arose in individuals who were exposed to an average 131I dose of 2.3 Gy, which is higher than the average dose received by patients who develop PTCs driven by other oncogenes. The average age of patients with PTCs that were positive for ETV6-NTRK3 was about 8 years at the time of the accident and 24 years at the time of surgery, resulting in an average time between exposure and thyroid surgery of 16 years. The UkrAm cohort includes individuals who have been followed since 1998; therefore it remains unknown whether ETV6-NTRK3 may also be common in tumors that developed < 12 years after the accident. We note that a sharp increase in thyroid cancer incidence in the area surrounding the Chernobyl nuclear power plant was observed as early as 4 to 6 years after the accident.
Phenotypically, both radiation-associated and sporadic tumors harboring ETV6-NTRK3 rearrangement were either the follicular variant of PTC or classic papillary cancer. It is noteworthy that the presence of a solid growth pattern was a microscopic feature observed only in radiation-associated PTCs that carried this rearrangement. We and others previously demonstrated a common presence of solid growth pattern in post-Chernobyl tumors compared with sporadic PTCs and the association of this growth pattern with RET/PTC3 rearrangement.[40, 41] The results from this study suggest that ETV6-NTRK3 may represent another type of chromosomal rearrangement associated with the solid growth pattern of PTC in patients exposed to radiation.
Findings in this study provide additional evidence for an association between specific types of mutations and etiologic factors implicated in the development of thyroid cancer. Previous studies have indicated that thyroid cancers developing after exposure to ionizing radiation have a high prevalence of chromosomal rearrangements like RET/PTC and AKAP9-BRAF and a low prevalence of point mutations.[12-14] The current results extend evidence supporting such an association, adding ETV6-NTRK3 to the list of fusions that preferentially occur in patients exposed to radiation and that can be induced by radiation in vitro.
The association between ETV6-NTRK3 and radiation exposure of the thyroid gland identified in this study raises a possibility that these fusions observed in other cancer types may also be related to radiation exposure. This is particularly plausible for AML, which has a strong dose-dependent relation with environmental or medical irradiation.
In summary, we report here the occurrence of ETV6-NTRK3 rearrangement in papillary thyroid cancer and demonstrate that this is a common event in thyroid tumors associated with 131I radiation exposure. Moreover, we demonstrate that this rearrangement can be directly induced by 131I or γ-radiation in vitro and, thus, may represent a novel molecular mechanism contributing to the development of radiation-induced thyroid cancer.
This work was supported by National Institutes of Health (NIH) grant R01 CA88041 and in part by the Intramural Research Program of the NIH, National Cancer Institute.