Molecular Genetics of Diffuse Sclerosing Papillary Thyroid Cancer

Abstract

Context

Diffuse sclerosing papillary thyroid cancer (DSPTC) is rare, with limited data on its molecular genetics.

Objective

We studied the molecular genetics of a cohort of DSPTC.

Methods

DNA was isolated from paraffin blocks of 22 patients with DSPTC (15 females, 7 males, median age 18 years, range 8-81). We performed polymerase chain reaction–based Sanger sequencing and a next-generation sequencing (NGS) gene panel to characterize the genomic landscape of these tumors. We classified genetic alterations to definitely or probably pathogenic. Definitely pathogenic are genetic alterations that are well known to be associated with PTC (e.g., BRAF^V600E). Probably pathogenic are other alterations in genes that were reported in The Cancer Genome Atlas or the poorly differentiated and anaplastic thyroid cancer datasets.

Results

Three tumors were tested only by Sanger sequencing and were negative for BRAF^V600E, HRAS, KRAS, NRAS, TERT promoter, PTEN, and PIK3CA mutations. The other 19 tumors tested by NGS showed definitely pathogenic alterations in 10 patients (52.6%): 2/19 (10.5%) BRAF^V600E, 5/19 (26.3%) CCDC6-RET (RET/PTC1), 1/19 (5.3%) NCOA4-RET (RET/PTC3), 1/19 (5.3%) STRN-ALK fusion, and 2/19 (10.6%) TP53 mutations. Probably pathogenic alterations occurred in 13/19 tumors (68.4%) and included variants in POLE (31.6%), CDKN2A (26%), NF1 (21%), BRCA2 (15.8%), SETD2 (5.3%), ATM (5.3%), FLT3 (5.3%), and ROS1 (5.3%). In 1 patient, the gene panel showed no alterations. No mutations were found in the RAS, PTEN, PIK3CA, or TERT promoter in all patients. There was no clear genotype/phenotype correlation.

Conclusion

In DSPTC, fusion genes are common, BRAF^V600E is rare, and other usual point mutations are absent. Pathogenic and likely pathogenic variants in POLE, NF1, CDKN2A, BRCA2, TP53, SETD2, ATM, FLT3, and ROS1 occur in about two-thirds of DTPTC.

Differentiated thyroid cancer (DTC) is the most common type of thyroid cancer, and its incidence has significantly increased over the last 3 to 4 decades (1, 2). DTC is classified into papillary, follicular, and oncocytic (Hürthle) cell subtypes. Papillary thyroid cancer (PTC) is by far the most common subtype comprising about 80% of all cases of thyroid cancer and accounting for the significant increase in the incidence of DTC (1). The diagnosis of PTC depends on the characteristic features of its cells (3). These include nuclear enlargement, elongation and overlapping, chromatin clearing and margination, irregular nuclear contour, nuclear grooving ,and nuclear pseudo-inclusions (3). Based on further cytological features, additional histopathological components and architectural arrangements, PTC is further classified into several subtypes (3, 4). Although all subtypes share the nuclear features of classic PTC, the histopathological arrangements and additional histopathological features characterize these subtypes (3, 4). For instance, the cells of classic PTC are arranged in branching, randomly oriented papillae with fibrovascular cores lined by PTC cells. In follicular variant PTC, the cells are arranged in follicles but have characteristic nuclear features of PTC. Other subtypes include tall cell, columnar cell, hobnail, solid/trabecular variants, and diffuse sclerosing PTC (DSPTC) (3, 5).

DSPTC is a rare subtype of PTC accounting for 0.7% to 6.6% in different series (5-7). It is more common in young patients (6, 8) and in regions affected by increased radiation exposure (9, 10). It is characterized by dense fibrosis, extensive lymphocytic infiltration, numerous psammoma bodies, foci of squamous metaplasia, and frequent lymphovascular invasion (3, 5, 11). Due to the rarity of this subtype, there are limited data on its clinical and pathological features and outcome (6, 12). Molecular genetics of DSPTC are even much more limited and based on old methods of sequencing a single or a limited number of genetic alterations (6). Therefore, there is a knowledge gap and a need to study more comprehensively the molecular genetics of DSPTC using the currently available high-throughput next generation sequencing (NGS) methods. This was the basis for undertaking this study to characterize the underlying molecular genetics of this rare subtype of PTC.

Patients and Methods

DNA Isolation, Polymerase Chain Reaction and Sanger Sequencing

After obtaining an Institutional Review Board approval (ORA # 2130-015) from the Office of Research Affairs of the King Faisal Specialist Hospital & Research Centre, Riyadh, Saudi Arabia, we reviewed the clinical and pathological data of 22 patients with DSPTC who were managed at our institution during the period January 1999 to January 2019. A North American trained, experienced thyroid pathologist (H.A.) reviewed histopathology slides of all cases. The diagnosis of DSPTC was based on dense sclerosis and fibrotic strands, extensive psammoma bodies, lymphatic permeation, and, in most patients, evidence of chronic lymphocytic thyroiditis. Genomic DNA was isolated from formalin-fixed paraffin-embedded (FFPE) tumor tissue using the QIAamp DNA FFPE Tissue Kit (cat. no. 56404, QIAGEN GmbH – Germany) following the manufacturer's instructions. Polymerase chain reaction (PCR) and Sanger sequencing for common somatic point mutations (BRAF^V600E, codons 12/13 and 61 of the HRAS, NRAS, and KRAS, TERT promoter mutations C228T and C250T, PTEN exons 5, 6, 7, 8, and PIK3CA exons 9 and 20) were performed for all 22 patients. The primers, PCR, and sequencing conditions were previously described (13, 14). In 19 patients, this was followed by NGS to confirm Sanger sequencing findings and search for fusion genes and other genetic alterations using a cancer gene panel. In 3 cases, NGS was not done due to a shortage of DNA.

Next Generation Sequencing

Extraction of DNA/RNA From Formalin-Fixed, Paraffin-embedded Tissues

The pathologist (H.A.) examined the pathology slides and the corresponding tumor tissue on the FFPE blocks was labeled and carefully sampled to ensure obtaining tumor tissue and minimize contamination by surrounding nontumor tissue. The RecoverAll Total Nucleic Acid Isolation Kit (Thermo Fisher Scientific, cat no. AM1975) was used for simultaneous DNA/RNA isolation from FFPE tumor tissue using the manufacturer's protocol (https://www.thermofisher.com/order/catalog/product/AM1975#/AM1975) as previously described (15).

Library Construction

To detect sequence alterations and abnormal gene fusions in the tumor tissue, the Oncomine Comprehensive Assay v3 (Thermo Fisher Scientific, cat. no. A35805) was used. Briefly, library preparation followed the default recommended protocol of the manufacturer and we used 2 highly multiplexed primer pools to target 161 genes; of these, 48 genes fully covered 87 hotspot genes, 43 genes for copy number variations, and 51 fusion-related genes (Table 1). These genes and variants cover the most relevant targets in the vast majority of all solid cancers. Simultaneously, the gene fusion targets were interrogated using 2 high multiplexed RNA primer pools. Both DNA and RNA libraries were barcoded and quantified using a TaqMan Quantitation Kit (Thermo Fisher Scientific, Cat # 4468802). Barcoded libraries were pooled and emulsion PCR performed using the Ion Chef (Thermo Fisher Scientific) for automated clonal amplification using the Ion 540 Kit (Thermo Fisher Scientific, A30011) and automated 540 chip loading. A no-template control was included in each batch of library preparations.

Sequencing Data Analysis Using Ion Torrent S5 and S5XL

Sequencing was performed using the 540 chip on the Ion Torrent S5XL. Downstream data analysis used Torrent Suite software version 5.2.1 (Thermo Fisher Scientific). Sequencing was performed by using 200-bp reads and the 540 chip on the Ion Torrent S5/S5XL sequencing platform, and downstream data analysis used Torrent Suite software version 5.2.1 (Thermo Fisher Scientific) with default base-calling parameters. DNA Library reads were aligned within Torrent Suite to Human genome build 19, and binary alignment map (bam) files were exported to Ion Reporter 5.10 (Thermo Fisher Scientific) for variant calling and copy number analysis.

Bioinformatics Analysis

We used the manufacturer's Torrent Suite software (version 5.2.1) for running the base-calling step using the default parameters. The Torrent Suite also reported the basic QC metrics related to the base yield, loading on the chip, and read length. The resulting sequences were then moved to the Ion Reporter System for the alignment and the variant calling steps, using the “Oncomine workflow – DNA and Fusions – Single Sample” (version w3.2). This workflow includes procedures for variant calling, copy number variant detection, and fusion detection.

A sample is accepted if at least 99% of the target regions are covered with at least 1 read and at least 95% of the target regions are covered with at least 500×, and uniformity is greater than 95%. A variant is reported if it is covered with at least 200 reads with no strand bias. A copy number variant is included if the MAPD <.3 and the confidence at 5% is ≥3 for amplifications. The resulting output of the variant calling step is a VCF file containing a list of the variants. The variants were then annotated using 2 pipelines: (1) the Ion Reporter and OncoReporter pipeline of the manufacturer, and (2) the in-house pipeline. The Ion Reporter pipeline includes basic population databases (1000G, ExAC, dbSNP), clinvar, and well-curated drug databases. Our in-house pipeline integrates extra databases including the in-house gnomAD database, Cosmic, Combined Annotation Dependent Depletion (CADD), and in-house germline database. The filtration step was conducted based on the following parameters: a variant is reported if it is a proven clinically relevant variant (as indicated in the Cosmic, Clinvar, and Ion Reporter hotspot database), not synonymous, its frequency in public and in-house population database is less than 1%, and finally its CADD score larger than 20 if it is exonic. Before the final reporting of the variants, the variants that survived the filtration were evaluated against their relevance to the cancer type, and were also examined using the IGV visualization tool to verify that the alignment and variant calling was in order.

We focused only on reporting Tier I and Tier II variants; hence, we classified the variants to definitely pathogenic (Tier I) with strong clinical significance, and probably pathogenic variants (Tier II) with potential clinical significance. In more specific terms, definitely pathogenic variants are those genetic alterations that are well known to be associated with PTC (e.g., BRAF^V600E). Probably pathogenic variants are alterations in genes that are known to be important in cancer pathogenesis and were reported in the TCGA (16) or poorly differentiated and anaplastic thyroid cancer (17) datasets. Although we did not have normal tissue or blood to test for the possibilities that these variants are germline, we search for these variants in 3 large germline databases (1000 g, ExAC, gnomAD) and in a local database, the Saudi Genome Project (SGP) database. The Saudi Genome project is a national Saudi database in which whole exome sequencing of more than 20 000 individuals has been completed, analyzed, and catalogued as part of a national program aiming to genotype 100 000 individuals (https://shgp.kacst.edu.sa, accessed on 08 December 2022). We excluded variants that have minimal allele frequency >0.01 in any of these databases. We also assessed the pathogenicity of these variants using the American College of Medical Genetics and Genomics (ACMG) criteria (18) and included only those that are pathogenic or likely pathogenic variants.

Results

Clinical Features, Management and Outcome

A total of 22 unrelated patients (15 females, 7 males, median age 18 years, range 8-81) were studied (Table 2). None of these patients had a family history of thyroid cancer or tumor predisposition syndromes. All patients had total thyroidectomy and received I-131 remnant ablation or therapy. Central ±lateral lymph node dissection was performed in 20 cases (91%) and was positive in all of them. Distant metastases at the time of presentation were diagnosed in 9 patients (41%). Fifteen patients (68%) received 1 or more additional surgical or radioiodine-131 (I-131) therapies (Table 2). Ten patients (45.5%) achieved an excellent response, 3 (13.6%) an indeterminate response, 1 (4.5%) a biochemically incomplete response, 5 (22.7%) a structurally incomplete response, 3 (13.6%) unclear statuses and 1 (4.5%) died. The median duration of follow-up was 8.75 years (2.8-19).

Molecular Genetics

In the 3 patients whose tumors were tested only by Sanger sequencing, none was positive for any of the tested mutations (BRAF^V600E, codons 12/13 and 61 of the HRAS, NRAS, and KRAS, TERT promoter mutations C228T and C250T, PTEN exons 5, 6, 7, 8, and PIK3CA exons 9 and 20). The tumors of the 19 patients which were subjected to NGS showed several variants (Table 2). Definitely pathogenic alterations were present in 10 patients (52.6%) as follows (Table 2): 2/19 (10.5%) had BRAF^V600E, 5/19 (26.3%) CCDC6-RET (RET/PTC1), 1/19 (5.3%) NCOA4-RET (RET/PTC3), 1/19 (5.3%) STRN-ALK fusion and 2/19 (10.5%) TP53 mutations (Table 2). Probably pathogenic alterations occurred in 13/19 tumors (68.4%) and included variants in POLE (6/19, 31.6%), CDKN2A (5/19, 26%), NF1 (4/19, 21%), BRCA2 (3/19, 15.8%), SETD2 (1/19, 5.3%), ATM (1/19, 5.3%), FLT3 (1/19, 5.3%), and ROS1 (1/19, 5.3%). In 1 patient, the gene panel showed no alterations. No mutations were found in RAS, PTEN, PIK3CA, or TERT promoter in all patients. The frequencies of alterations in these genes in the current study and in the TCGA and PDTC/ATC datasets are summarized in Table 3 and their frequencies in 1000G, ExAC, gnomAD, and SGP and their predicted pathogenicity according to the ACMG criteria is shown in Table 4.

Genotype/Phenotype Correlation

There was no clear genotype/phenotype correlation (Table 2); 10 patients (45.5%) achieved an excellent response (4 with no mutation by Sanger sequencing and/or NGS, 1 BRAF^V600E, 2 RET/PTC1, 1 STRN-ALK, and 2 probably pathogenic variants, Table 2). Three (13.6%) were in an indeterminate status (2 had RET/PTC1 and 1 probable variant, Table 2), 1 patient (4.5%) had a biochemically incomplete response (RET/PTC1), 5 patients (22.7%) continue to have a structurally incomplete response (1 had no mutation by NGS, 2 had probably pathogenic variants, 1 had RET/PTC3 and 1 had TP53 mutation, Table 2). One patient (4.5%) with probably pathogenic variants died and 3 patients (13.6%) had unclear status (2 probably pathogenic variants and 1 negative by Sanger sequencing).

Discussion

In this study, we evaluated DSPTC, a rare subtype of PTC seen mostly in young patients. Due to its rarity, data on its clinical and pathological features are limited (6). Data on its molecular genetics are scarce and selective, using old methods of sequencing (17). Our results suggest that DSPTC is an aggressive form of PTC with a common occurrence of lymph nodes and distant metastasis. Persistent/recurrent disease is common but disease-specific mortality is rare. Its molecular genetics are also significantly different from the classic PTC, with rare occurrence of BRAF^V600E mutation and a common occurrence of fusion genes, especially RET/PTC1. Several other pathogenic and probably pathogenic genetic alterations have also been found. However, we could not detect a clear genotype/phenotype correlation, probably due to the small sample size.

Data on the underlying molecular genetics of DSPTC are limited and investigated mostly individual mutations using Sanger sequencing (summarized in Table 5) (19-24). To our knowledge, no study has used NGS to comprehensively evaluate several genetic alterations in the same samples. Joung et al screened 37 DSPTC for BRAF^V600E, NRAS codon 61, HRAS codons 12/13/61, and KRAS codon 12/13 point mutations as well as RET/PTC1, RET/PTC3, and PAX8/PPARγ rearrangements using reverse transcription real-time polymerase chain reaction (RT-PCR) (19). Out of 37 cases of DSPTC, 17 (46%), 6 (16%), and 9 (24%) tested positive for the mutually exclusive RET/PTC1, RET/PTC3, and BRAF^V600E, respectively. Advanced-stage disease, including T4 and distant metastases and persistent disease were significantly associated with RET/PTC3, while remission was higher in tumors with RET/PTC1 (19). In our study, RET/PTC3 was rare, occurring only once in a 12-year-old girl with stage T4N1bM1 tumor and was treated with total thyroidectomy and bilateral neck dissection and three I-131 therapies but continued to have evidence of structural disease.

Using RT-PCR and DNA sequencing, Sheu et al studied 7 cases of DSPTC and found 2 carried RET/PTC1, 1 harbored RET/PTC3, and none had BRAF^V600E mutation (21). All 7 samples expressed the RET tyrosine kinase domain, but none expressed the extracellular domain suggesting the presence of other types of RET/PTC rearrangements (21). Lim et al studied BRAF^V600E mutation in 2947 conventional PTC, 98 DSPTC, and 85 follicular variant PTC (22). The prevalence of BRAF^V600E mutation was 73%, 61%, and 40%, respectively. Although BRAF^V600E was associated with several aggressive histopathological features in conventional PTC, it was not associated with any clinicopathological factors in DSPTC and follicular variant PTC (22). This high rate of BRAF^V600E in DSPTC from South Korea is not shown in studies from other countries. In the study by Sheu et al described above, none of the 7 German patients with DSPTC had BRAF^V600E (21). In a study from the United States assessing the value of molecular testing for thyroid nodules in pediatric patients using multiplex qualitative PCR followed by bead array cytometry, 8 DSPTC were included and 4 (5%) of them were positive for RET/PTC1 (23). The other 4 were negative for all 17 genetic alterations tested, including BRAF^V600E, HRAS, KRAS, and NRAS genes, RET/PTC1, RET/PTC3, and PAX8/PPARG (23). In another study from Canada that aimed to correlate the BRAF^V600E and TERT promoter mutations with the histopathological architecture, 7 DSPTC were included; only 2 were positive for BRAF^V600E and none for TERT promoter mutations (24). In our study, only 2/19 (10.5%) had BRAF^V600E, 5/19 (26.3%) had CCDC6-RET (RET/PTC1)), 1/19 (5.3%) had NCOA4-RET (RET/PTC3)), 1/19 (5.3%) had STRN-ALK) fusions, and 2/19 (10.5%) had TP53). We also found several other genetic alterations (Table 2) in genes that were reported in the TCGA data for the well-differentiated PTC (16) and in the Memorial Sloan Kettering Cancer Center study of poorly differentiated and anaplastic thyroid cancers (17) (Table 3). Although these alterations passed the thorough bioinformatics analysis of these internationally well-accepted datasets, we cautiously labeled them as “probably pathogenic”. To exclude the possibility that they are germline variants, we assessed their frequencies in several international and local germline databases and excluded those with minimal allele frequency >0.01. In fact, the majority were not reported in these databases or had extremely low frequencies (Table 4). We also assessed their pathogenicity using American College of Medical Genetics and Genomics criteria (18) and included only those with pathogenic or likely pathogenic pathogenicity score. These variants are all in cancer-related genes. These genes are tyrosine kinase genes (NF1, FLT3, ROS1), DNA repair genes (POLE, BRCA2) or cell cycle regulators (CDKN2A, BRCA2, ATM). All of these features strongly support their pathogenic role in DSPTC.

Our study is the first to use a comprehensive NGS gene panel for the molecular characterization of DSPTC. It reaffirms previous single or limited gene studies in that BRAF^V600E is rare, and RET/PTC rearrangements are the most common genetic alterations in DSPTC. It also shows for the first time additional genetic variants that we cautiously labeled as probably pathogenic but have a strong basis that they are likely pathogenic. Admittedly, our study is a small study with only 22 cases and showing mainly that DSPTC has a much lower rate of BRAF^V600E mutations than classic PTC. Further collaborative studies involving larger samples from multiple centers are needed to better characterize the clinical and molecular features of this rare subtype of PTC. By limiting our analysis to well-known thyroid cancer genes and those that were reported in the TCGA and poorly differentiated and anaplastic thyroid cancer datasets, we may have missed some other genetic alterations. However, these databases are well analyzed and comprehensive, justifying their use as references for the current analysis.

In summary, DSPTC is an aggressive type of PTC affecting mostly young patients and characterized by frequent lymph node and distant metastases. Fusion genes, especially CCDC6-RET are the most common genetic alterations. BRAF^V600E is rare and other usual somatic point mutations in RAS, TERT promoter, PTEN, and PIK3CA are absent. However, other pathogenic or likely pathogenic variants occur in more than two-thirds of patients, including TP53, POLE, NF1, CDKN2A, BRCA2, SETD2, ATM, FLT3, and ROS1.

Acknowledgment

We would like to acknowledge our colleagues in the Section of Endocrinology, Department of Medicine, and the Department of Molecular Oncology for their support.

Funding

No specific grant but supported by general funding of the Research Centre of the King Faisal Specialist Hospital and Research Centre, Riyadh, Saudi Arabia.

Author Contributions

M.A.: Undertook the major part in laboratory work. B.A.: Patient samples preparation and organization of the work. A.A.: Collected clinical data. A.A.: collected clinical data. H.A.H.: examined the pathology and selected and prepared samples for DNA extraction. A.K.M.: Supervised and analyzed laboratory work. M.A.: Bioinformatics analysis. Y.S.: participated in writing and review of the final manuscript. A.S.A.: conceptualization, designing the project, analysis, and writing.

Disclosures

All authors have no conflict of interest to declare.

Data Availability

Some datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

References

Abbreviations

 
• DSPTC

  diffuse sclerosing papillary thyroid cancer

 
• DTC

  differentiated thyroid cancer

 
• NGS

  next-generation sequencing

 
• FFPE

  formalin fixed paraffin embedded

 
• PTC

  papillary thyroid cancer

 
• RT-PCR

  reverse transcription real-time polymerase chain reaction

 
• TCGA

  The Cancer Genome Atlas