Introduction

In thyroid cancer, prognostication and treatment-based decisions are largely dictated by the tumour node metastasis (TNM) staging system.1 2 This classification model is based on tumour size and the presence of extrathyroidal extension, two parameters that are easy to retrieve and usually reproducible between different centres.1 However, there are instances in which these parameters are not sufficient to predict the outcome of the individual patient, and researchers have therefore sought additional prognostic markers to aid in this aspect. Mutations in the promoter region of the telomerase reverse transcriptase (TERT) gene are found in approximately 10%–15% of papillary thyroid carcinomas (PTCs) and follicular thyroid carcinomas (FTCs), and particularly poor-prognosis cases that are associated with adverse clinical features.3–10 Moreover, TERT promoter mutations are recurrently observed in more aggressive tumour types, such as poorly differentiated thyroid carcinomas (PDTCs) and anaplastic thyroid carcinomas (ATCs), thereby solidifying the relationship between this genetic aberration and worse clinical outcome in thyroid cancer.11 12 These mutations might also be able to pinpoint metastatic potential in histologically equivocal ‘follicular thyroid tumours of uncertain malignant potential’ (FT-UMPs).13

To date, there are two established hotspot TERT promoter mutations (denoted C228T and C250T) which are closely positioned and hence both interrogated by Sanger sequencing using a single primer pair.14 15 The mutations are thought to increase TERT gene transcription through the intensified recruitment of various transcription factors, and as TERT encodes the catalytic subunit of telomerase, this mechanism is thought to stimulate telomere lengthening and thus counteracts senescence-induced apoptosis.7 16 In thyroid tumours in general, TERT gene expression is invariably associated to malignant disease and worse clinical outcomes, and therefore constitutes a potential diagnostic and prognostic marker.17–22 In FTCs and FT-UMPs, other genetic mechanisms besides promoter mutations also seem to stimulate TERT mRNA expression, such as aberrant promoter methylation and copy number gain of the TERT gene locus.9 As the presence of TERT mRNA could constitute a marker of worse prognosis seemingly irrespectively of the causative genetic mechanism, we previously investigated whether or not TERT protein expression could constitute a surrogate marker for TERT gene activation in follicular thyroid tumours.23 However, we found no correlation between TERT mRNA levels and TERT immunoreactivity, and to our surprise, nuclear expression of TERT was largely absent. Given the known association between TERT gene output and worse clinical outcomes, we therefore turned our attention to in situ hybridisation (ISH). To our knowledge, this methodology has not previously been attempted in terms of visualising TERT mRNA in formalin-fixated paraffin-embedded (FFPE) follicular thyroid tumour specimen with known TERT promoter genotypes. The potential benefits of ISH include the possibility to detect TERT mRNA expression in clinical routine processed samples, as well as to visualise the specific TERT expressing cell type. As several reports indicate that benign thyroid lesions with concomitant thyroiditis may exhibit TERT mRNA due to constitutive expression in lymphocytes, ISH could in theory be a way to eliminate these false positives from a screening perspective.19 24 25

Material and methodsStudy cohort

The study cohort was retrospectively collected, and consisted of 26 thyroid tumours previously characterised for TERT mRNA expression, TERT promoter mutations, TERT promoter methylation levels and TERT gene copy number.9 Baseline clinical, histopathological and molecular attributes are summarised in table 1. All patients were surgically treated at the Karolinska University Hospital, Stockholm, Sweden between 2013 and 2016, and all tumours were diagnosed using the most recent WHO criteria.26 A total of 7 FTCs, 2 FT-UMPs and a single Hürthle cell carcinoma with established TERT promoter mutations (based on Sanger sequencing results) and TERT mRNA expression (determined by quantitative real-time PCR; qRT-PCR) were included (cases 1–10), as were 16 FTCs absent of TERT promoter hotspot mutations and TERT mRNA expression (cases 11–26). None of the cases exhibited histological evidence of thyroiditis. A single de-identified multinodular goitre sample was included as a non-tumour reference.

Table 1

Summarised histopathological and clinical information

In situ hybridisation

The ISH methodology was carried out using the RNAscope technology (Advanced Cell Diagnostics, CA, USA).27 We used two TERT probes: TERT1; RNAscope Hs-TERT, product no. 605511, targeting homo sapiens telomerase reverse transcriptase (TERT) mRNA transcript variant 1, and TERT2; RNAscope Hs-TERT-O2, product no. 494561, targeting homo sapiens telomerase reverse transcriptase (TERT) mRNA transcript variant 2 (Advanced Cell Diagnostics). As controls, we used a probe against the housekeeping gene Peptidylprolyl Isomerase B (PPIB) (product no. 313901) as positive control and the bacterial RNA sequence Dihydrodipicolinate Reductase (dapB) (product no. 310043) (Advanced Cell Diagnostics) as negative control. Validation of the methodology was performed using two serially sectioned FTC cases with previously established TERT promoter mutations and TERT mRNA expression as well as using mounted HeLa cells (ACD, product no. 310045) using both TERT and control probes (Advanced Cell Diagnostics). All probes were assessed using different methods for upholding the temperature during the pretreatment phase (pressure cooker vs water bath), and optimal signals were retrieved using the RNAscope Target Retrieval Standard for 15 min at 95°C, followed by protease treatment for 30 min. Slides were then processed according to a standardised protocol provided by the manufacturer. The hybridisation was performed using an ACD HybEZ II Hybridization System (Advanced Cell Diagnostics). Diaminobenzidine (DAB) was used as chromogen, and slides were counterstained in hematoxylin.

Visualisation and scoring procedure

All slides were evaluated by an endocrine pathologist (unaware of the previous TERT gene screening outcomes) at ×400 magnification using a BX46 Olympus light microscope (Olympus, Tokyo, Japan). Images were captured using a ToupCam Industrial Digital Camera and the ImageView software. Both the cytoplasmic and nuclear compartments were analysed in all cases, and cases were noted as positive if distinct signals were envisioned in subsets of tumour cells. The entire slide was examined in ×400 magnification in order to visualise focal signals, and verified using ×1000 magnification. Negative cases were annotated as such if there was no clear-cut signal in any tumour cells.

Statistical analyses

The statistical analyses (Fisher’s exact test, Mann-Whitney U) were performed using IBM SPSS Statistics V.27. P values<0.05 were considered significant.

ResultsControl experiments

Outcomes of the control experiments are detailed in figure 1. Using mounted HeLa cells (immortalised cervical cancer cell line known to express telomerase), strong and diffuse cytoplasmic housekeeping gene expression was evident, whereas negative controls (bacterial RNA sequence) were devoid of signals.28 Using TERT1 and TERT2 probes, a distinct, dot-like nuclear signal was evident in subsets of the HeLa cells. The cytoplasm of the HeLa cells was not stained. For subsequent analyses of the follicular thyroid tumours, a positive housekeeping control was included for each case, as well as a negative control in each experimental run. All follicular thyroid tumours and the multinodular goitre case stained positive for the housekeeping gene with abundant cytoplasmic expression, while consistently negative for the bacterial RNA sequence. The multinodular goitre sample was absent of signals using both TERT probes.

Figure 1

Control and TERT in situ hybridisation (ISH) signals in HeLa cells. (A) Peptidylprolyl Isomerase B (PPIB) housekeeping gene ISH displaying a strong, predominant cytoplasmic signal. Magnification ×400. (B) ISH probe directed at bacterial Dihydrodipicolinate Reductase (dapB) RNA displaying absent signals, serving as a negative control of the methodology. Magnification ×400. (C,D) TERT1 (C) and TERT2 (D) ISH at ×1000 magnification visualising subsets of HeLa cell with intense nuclear, dot-like signals. Black arrowheads highlight a subset of these signals.

TERT mRNA visualisation in follicular thyroid tumors using ISH

The ISH staining outcomes are detailed in table 2 and illustrated in figures 2 and 3. The majority of the 10 thyroid tumours with established TERT promoter mutations and TERT mRNA expression demonstrated nuclear, dot-like TERT ISH signals in subsets of tumour cells (TERT1; n=6/10; 60%, TERT2; n=7/10; 70%) (figure 2). Cytoplasmic signals were not seen in any case. Two cases were negative using both TERT probes. In five cases, focal positive nuclear signals were retrieved using both TERT probes, whereas the remaining three positive samples were only identified using one of the probes (TERT1 in one case, TERT2 in two cases). Both FT-UMPs with TERT promoter mutations displayed absent signals using the TERT1 probe, but exhibited positive nuclear signals using TERT2. There was no apparent correlation between the level of relative TERT mRNA expression from previous qRT-PCR analyses and ISH outcome (data not shown). All 16 FTCs lacking TERT promoter mutations and TERT mRNA expression were devoid of ISH signals using both TERT probes (figure 3), although four cases displayed various amounts of DAB precipitation, making the scrutinising of these slides somewhat burdensome (data not shown). The tumour stroma (endothelial cells and fibroblasts) was consistently negative for TERT signals in all cases.

Table 2

TERT in situ hybridisation results

Figure 2

Focal, dot-like nuclear signals using both TERT in situ hybridisation probes were seen in the majority of TERT promoter mutated thyroid tumours with previously established TERT mRNA expression, represented here by case 10, a Hürthle cell carcinoma. Magnification ×400, with insets magnified ×1000.

Figure 3

All follicular thyroid carcinomas (FTCs) without TERT promoter mutations and devoid of TERT mRNA as previously interrogated by quantitative real-time PCR were negative for in situ hybridisation signals using both TERT1 and TERT2 probes. Shown here is case 26, an encapsulated angioinvasive FTC, with a negative TERT1 signal (left) while showing a clear cytoplasmic signal using the positive control (right). Magnification ×400.

The sensitivity and specificity for TERT ISH to detect cases with underlying TERT promoter mutations and TERT mRNA expression was 60% and 100%, respectively, for TERT1 and 70% and 100%, respectively, for TERT2. Accepting positive signals with any of the two probes, the sensitivity rose to 80%. The positive predictive value was 100% for either probe, whereas the negative predictive values were 80% (TERT1), 84% (TERT2) and 89% (any of the two probes). There was a strong correlation between the visualisation of TERT mRNA expression using ISH and underlying TERT promoter mutations as well as evident TERT mRNA expression as interrogated via qRT-PCR (Fisher’s exact p<0.001). Moreover, there was no significant difference in the age of the tissue blocks selected for ISH analyses between mutation-positive (2248 days) and mutation-negative groups (2234 days) (Mann-Whitney U, p=0.75) (table 1). Also, the two tumours with established TERT promoter mutations and TERT mRNA expression that were negative on ISH using both TERT probes were not the two oldest cases among the mutation-positive tumours (data not shown).

Discussion

TERT gene expression is invariably associated with poorer patient outcomes in thyroid cancer, and this dysregulation has in turn been coupled to underlying TERT promoter mutations as well as alternate genetic mechanisms leading to increased TERT gene output.3–5 7 9 As TERT mRNA expression correlates poorly to TERT protein expression in follicular thyroid tumours, there are potential clinical benefits to develop a method that will correctly pinpoint TERT mRNA expression in clinical samples.23 Although qRT-PCR could be considered in this aspect, recent improvements of the ISH technique have made this method attractive for clinical purposes.27 First, not all institutions have the ability to collect and analyse fresh-frozen tissues, and second, TERT mRNA expression is recurrently reported in lymphocytes.19 25 29 Visualisation of TERT by ISH is performed on clinical routine FFPE material, and also enables the pathologist to detect spatial and tissue-specific distribution patterns that qRT-PCR cannot.

In our series, TERT signals were exclusively found in the nuclear compartment. This was true for the thyroid tumours as well as for the HeLa cells used as controls, suggesting a true biological role for nuclear TERT mRNA. Previous observations support our findings, in which TERT mRNA seems to aggregate in the nucleus when analysing various cancer cell lines as well as malignant melanomas.30 31 As we could not detect a nuclear signal in the negative controls, we do not suspect the findings of nuclear-specific TERT to be a false-positive observation. Moreover, as all cases exhibited a strong cytoplasmic housekeeping gene signal, this implies that RNA levels were intact even after 24–48 hours of formalin fixation. Therefore, the absence of cytoplasmic TERT signals is most likely not a consequence of poor RNA quality, but rather implies a biological phenomenon worthy of attention. Indeed, TERT mRNA molecules could potentially exhibit non-conventional roles besides acting as a ribosomal template for translation.

Interestingly, TERT ISH exhibited perfect specificity, as all TERT mRNA devoid FTCs were negative on ISH using both probes. Therefore, the method could potentially be of value for clinical screening purposes. In theory, the detection of TERT signals in a histologically confirmed FTC could therefore imply an underlying TERT gene aberrancy, which in turn is strongly associated with worse patient outcome.9 Moreover, studies using preoperative fine-needle aspiration biopsy material could also be valuable, as the expression of TERT could imply a clinically more burdensome tumour in need of more extensive interventions. However, TERT-positive cases only exhibited nuclear signals in small subsets of tumour cells, thereby forcing the pathologist to scrutinise the whole slide using high power (×400 or ×1000) magnification. This could be time-consuming from a clinical perspective, conferring a risk of failing to detect the area of positivity and misclassifying the tumour as negative. Indeed, our experience herein suggests that TERT-positive cells aggregate in small clusters and might be hard to detect if not the entire slide is investigated carefully. In a way, our findings thus bear similarities to current PD-L1 immunohistochemical scoring algorithms, in which very low cut-offs for positivity have been recommended, and several heterogeneous expression patterns have been reported.32 As of this, modern pathologists are thus getting acquainted to scoring principles based on regional and variable intensity across a tissue slide, and not only diffuse and unequivocal expression patterns. Indeed, the previous findings of TERT promoter mutational spatial heterogeneity in follicular thyroid tumours adds to the complexity of TERT visualisation for clinical purposes, and future studies will possibly need to address the potential benefit of multi-section analyses in highlighting focal TERT dysregulation.33 34

Take home messages

TERT promoter mutations and TERT mRNA expression are prognostic markers of relevance in follicular thyroid tumours.

TERT in situ hybridisation correctly pinpoints TERT aberrancies in the majority of cases by visualising exclusive nuclear signals.

The methodology could be of potential value for clinical screening purposes and might imply unconventional roles for nuclear TERT mRNA in thyroid cancer development.

Data availability statement

All data relevant to the study are included in the article. The authors confirm that the data supporting the findings of this study are available within the article.

Ethics statementsPatient consent for publication

Not required.

Ethics approval

Granted by the Swedish Ethical Review Authority (approval number 2015-959-31).

Acknowledgments

We acknowledge Dr Mensur Dzabic and Ms Semra Öz Baloglu at the Department of Pathology and Cytology, Karolinska University Hospital, for their continuous support of the study.