Real-time ultrasound elastography in the diagnosis of newly identified thyroid nodules in adults: the ElaTION RCT

Thyroid nodules

Palpable thyroid nodules are common, being detected in about 5–7% of the population. 2 Using ultrasound (US), nodularity of the thyroid can be detected in up to 50% of the population. 3 Approximately 4–7% of thyroid nodules are malignant, and therefore most national guidelines recommend the investigation of nodules larger than 5–10 mm in diameter. 4,5 Clinically impalpable thyroid nodules, commonly incidentally identified on imaging, appear to carry a similar risk of malignancy and should therefore be investigated in the same way as palpable nodules.

Due to the increased use of imaging modalities, such as US carotid duplex, magnetic resonance imaging (MRI) for cervical spinal disease, or whole body imaging such as positron emission tomography–computed tomography (PET-CT), incidental thyroid nodules that are asymptomatic are increasingly being detected and investigated. 6–10 This is resulting in a rapidly increasing burden of investigation of thyroid nodules. 11 In one average-sized hospital, over a period of 5 years, 1412 US-guided fine-needle aspirations were undertaken for the investigation of thyroid nodules – an average of 282 scans per year. 12

Current recommended classification and investigation of thyroid nodules

Definitive investigation of thyroid nodules is by US and fine-needle aspiration cytology (FNAC) according to the 2014 British Thyroid Association (BTA) guidelines. 4 US can detect features that are suggestive of malignancy, including an irregular or microlobulated outline, microcalcifications, hypoechogenicity, extrathyroidal extension and lesions being taller than wide. US alone can predict the risk of malignancy with accuracy varying from 22% to 89%. 13 FNAC remains the best practice for diagnosing thyroid malignancy and the most accurately carried out under US guidance.

The 2014 BTA guidelines recommends a system for classifying the results of FNAC, with subsequent management based on this classification. 4 The recommendations for this study are based on these guidelines, and are as follows:

• Thy 1 – non-diagnostic: BTA guidelines recommend repeat US-guided FNAC, especially if there is suspicion of malignancy, up to two times. If Thy 1 was obtained in all three instances, our recommendation was to surgically remove nodules.

• Thy 2 – benign: guidelines recommend a further benign FNAC if there are suspicious features or indeterminate features on histology or US. While discharge after a single benign FNAC result can be considered in the absence of suspicious features on US, a second US FNAC was requested for the study to allow examination of the false-negative rate of a single Thy 2 policy.

• Thy 3a denotes that a neoplasm is possible, with atypical features present but not enough to place into other categories. A repeat US FNAC is usually recommended. In the situation of repeated Thy 3a, a multidisciplinary team (MDT) discussion and possible surgery should be considered.

• Thy 3f denotes a follicular neoplastic lesion of indeterminate nature, which may be benign or malignant.

• Thy 4 is suspicious of cancer.

• Thy 5 is diagnostic of cancer.

An audit of 1412 consecutive US FNAC from a single institution demonstrated 20% of the FNAC results to be Thy 1, 70%; Thy 2, 5%; Thy 3 (3a and 3f), 5%; Thy 4, 3%; and Thy 5, 2%. 12

The BTA recommendations are to consider offering surgery to remove nodules with Thy 3f, 4 and 5 cytology results to both obtain definitive histological diagnosis and, in some cases, provide definitive treatment.

Current deficiencies in the investigation and diagnosis of thyroid nodules

Technological developments – shear and strain wave elastography

Elastography is a technology that uses low-frequency vibrations to assess the elastic properties – and therefore stiffness – of tissues. Ultrasound elastography (USE) has been proposed as an adjunct to US in the assessment of thyroid nodules. USE combines the diagnostic advantages of US assessment and US-guided FNAC with an additional assessment of the lesion’s stiffness. The aim of this is to increase the accuracy of thyroid cancer diagnosis as malignant thyroid nodules are harder (therefore more stiff) than benign ones. 16 The comparative amount and pattern of soft to hard areas within a nodule indicates the likelihood of malignancy. This offers the potential to aid both nodule selection when faced with multiple nodules (assisting the operator target the nodule with the highest risk of malignancy for sampling) and also may assist the radiologist undertaking the needle biopsy to target the area within a single nodule most likely to harbour malignant cells. Hence, USE may also potentially increase the yield of positive FNAC results. 17 A meta-analysis of 638 patients from 11 studies showed that USE features alone had a pooled sensitivity of 92% [confidence interval (CI) 88 to 96] and specificity of 90% (CI 85 to 95) for identifying malignant thyroid nodules. 18 Most studies have also shown very good correlation between radiologists. 19

There are two elastography methods currently available: strain elastography (STE) [most commonly used with real-time elastography (RTE)] and shear wave elastography (SWE). STE requires application of an oscillating physical pressure by the operator on the target lesion, by exerting light manual pressure on the transducer. By comparing echo signals before and after slight compression, the strain (amount of deformation of tissue) is calculated. This is superimposed on the US image as a colour-coded map (thus providing a qualitative visual map). The harder tissue (e.g. malignant thyroid tissue) is seen as a distinct area of colour (often coded red or blue) differing from normal background tissue. A strain ratio may be calculated from the differences in strain measurements between the nodule and the normal background thyroid (providing a semi-quantitative assessment of the stiffness of the tissue).

Shear wave elastography is generated by a pulse from the US transducer and is a true quantitative measurement of the velocity of sound in tissue; shear waves travel faster in harder (and therefore presumed malignant) tissue. Measured velocities can also be displayed graphically on a colour map or numerically as shear wave indices. SWE gives similar information about tissue hardness of the lesion, and more importantly can quantify the stiffness of abnormal tissue without the need to compare to normal tissue (hence giving a true quantitative measurement as opposed to the semi-quantitative measurement from STE). Both methods of elastography provide a similar basic assessment of the underlying hardness of the tissue and may be used to assess a focal thyroid lesion.

Thus, USE assessment can be qualitative (visual assessment of a colour map), semi-quantitative (strain ratios) or quantitative (shear wave indices). No clear advantage has been identified between these approaches in a meta-analysis of 72 studies with 13,505 patients – sensitivity 84%, 83% and 78%; specificity 81%, 80% and 81%, respectively. 20

Methodological aspects

Rationale for the ElaTION trial

The ElaTION trial is important because of the following factors:

A significant health need:

Thyroid nodules affect a substantial proportion of the population and are increasingly being identified incidentally on routine imaging of the head and neck. There is a need to improve the performance of US-guided FNAC to reduce both the overall number of FNACs and importantly also the diagnostic operations needed to establish a diagnosis. This would reduce the morbidity associated with the procedures (e.g. permanent loss of voice, difficulty swallowing, permanent hypocalcaemia requiring life-long medication), and decrease the inconvenience and anxiety shown to be associated with uncertainty before diagnosis, especially on repeated tests. 25

Considerable potential resource and cost requirements:

Thyroid nodules are very common, affecting up to 70% of the population. They are being detected with increasing frequency due to the rising use of imaging technology. A systematic review of the literature demonstrated that in 168,876 patents, 26% of US scans and 22% of CT scans of the neck performed for non-thyroid causes showed incidental nodules in the thyroid. These data translate to approximately 1900 new patients with thyroid nodules identified over a period of a year in the one hospital in the UK, resulting in a potential cost of investigation with US-FNAC of £683,650 in one institution. Extrapolated to 400 hospitals in the UK, this could potentially cost the NHS over £272M per annum. 6

Decreasing the number of non-diagnostic or unnecessary FNAC and consequent operations might present the opportunity for considerable cost savings to the NHS.

Outstanding issues in FNAC:

There remain some important questions regarding the use of US alone for diagnosis and the need for repetition of US-only guided FNAC. FNAC is subject to sampling and analysis uncertainties, depending on the radiologist’s and cytopathologist’s experiences. FNAC is reported to have a false-negative rate of about 5.2% which has, to date, not been addressed in a large prospective randomised setting. 25 There is therefore a lack of level 1 evidence.

Inclusion criteria

1. Patients with single or multiple thyroid nodules, whether solid or partially cystic (i.e. mixed), undergoing investigation.

2. Patients aged 18 or over.

3. Patients able and willing to give written informed consent.

Exclusion criteria

1. Patients who had undergone previous thyroid FNAC within the last 6 months.

2. Patients with a bleeding diathesis that precluded FNAC (patients currently on warfarin and aspirin therapy were eligible).

3. Patients with a needle phobia.

4. Pregnant patients.

5. Patients with purely cystic nodules or with recent haemorrhage, with no solid component.

6. Thyroid nodules that appeared to have rim calcification or egg shell calcification.

Rationale for choice of inclusion and exclusion criteria

The role of USE in the diagnosis of thyroid nodules remains controversial, particularly whether this technique reduces the number of FNACs that need to be performed to determine a diagnosis of malignancy. In this trial we aimed to reflect the pragmatism of including all adult patients with thyroid nodules.

Patients with bleeding diathesis or needle phobia were excluded on the grounds that it is more risky to perform FNAC in them. FNAC is often avoided during pregnancy in particular since hormonal changes during pregnancy may alter FNAC interpretation. FNAC in nodules with rim or eggshell calcification is technically very difficult and patients with nodules with these characteristics were excluded.

The presence of a cyst or multiple cysts often precludes a USE scan being done because the cyst may not have sufficient amounts of surrounding solid thyroid tissue. Patients with these nodules were excluded from the study. However, to ensure that an accurate and representative picture of current practice and an accurate assessment of the exact usefulness of the technique in routine clinical practice were obtained, we collected anonymised data about those patients, even though they were not randomised.

Primary outcome

The primary outcome measure was the proportion of patients who have a non-diagnostic (Thy 1) FNAC result following the first FNAC compared between the SE-FNAC and US-only guided FNAC.

Secondary outcomes

1. Number of FNACs needed to obtain a definitive diagnosis in each patient.

2. Time from first FNAC to obtaining a definitive diagnosis in each arm.

3. The proportion of patients with a benign histology, compared between arms.

4. The proportion of patients who had a non-diagnostic (Thy 1) cytology result following any FNAC by arm.

5. The proportion of patients who received surgery by arm.

6. Accuracy of first FNAC results and repeated FNAC results, compared between arms.

7. If USE or US without FNAC was as accurate as USE or US with FNAC, compared between arms.

8. Patients that reported anxiety, pain and quality of life [by the hospital anxiety and depression scale (HADS) questionnaire, VAPS and EQ-5D] at baseline and at 3, 6 and 12 months post randomisation.

9. Radiologist survey-completed by radiologists at the end of each procedure to identify whether radiologists found USE had contributed to their decisions, ease of use, and their prediction of malignancy of the nodule using USE or US features alone.

10. Complication rate from any thyroidectomy at 30 days and 6 months post surgery, including haematoma rate and temporary hypocalcaemia at 30 days and vocal cord palsy and permanent hypocalcaemia rate at 6 months.

11. Resource usage for consultation time and diagnostic testing procedures as well as subsequent management including consultations and surgical treatments.

Outcome assessment schedule

Participant reported assessments were made following informed consent and then at baseline (prior to FNAC), and then at 3, 6 and 12 months post randomisation (see Table 1).

Informed consent

The conduct of the trial was in accordance with the principles of good clinical practice (GCP) and applicable regulatory requirements.

The patient’s written informed consent to participate in the trial was obtained prior to performing any trial-related procedure, prior to randomisation and after a full explanation of the study had been given. A PIS was provided to facilitate this process.

If the participant expressed an interest in participating in the trial, they were asked to sign and date the latest version of the Consent Form. The participant had to give explicit consent for the regulatory authorities, members of the research team, and representatives of the sponsor to be given direct access to the participant’s medical records. The investigator (or delegate) then signed and dated the form. Written informed consent was obtained by a trained member of the research team with GCP training, knowledge of the trial protocol and delegated authority from the local principal investigator (PI).

Within the ElaTION trial, consent was usually obtained by an ElaTION research nurse on site. However, consent could also be obtained by the Consultant Radiologist or by a delegated person, for example, specialist registrar, radiographer or sonographer. Details of the informed consent discussions were recorded in the participant’s medical notes. This included the date of discussion, the name of the trial, summary of the discussion, version number of the PIS given to the participant and version number of the consent form signed, and the date consent was received.

At each visit the participant’s willingness to continue in the trial was ascertained and documented in the medical notes. Throughout the trial the participant was given the opportunity to ask questions about the trial. Any new information, that may have been relevant to the participant’s continued participation, was provided.

A separate consent form was signed by participants who were willing to give blood and tissue samples for the purpose of translational research.

The patient’s general practitioner (GP) was notified, with the patient’s consent, and a specimen ‘Letter to GP’ was supplied.

In order to ascertain generalisability, a log was kept of all patients who were potentially eligible and were approached to participate in the study, if they were randomised or not, and the reason for non-randomisation.

Randomisation method and stratification variables

Patients were randomised in a 1 : 1 ratio between USE-guided FNAC and conventional US-guided FNAC.

A minimisation procedure, using a computer-based algorithm, was used to avoid chance imbalances in important stratification variables. The stratification variables were:

1. radiologist: as US and USE scans are operator-dependent;

2. solitary nodule versus multinodular: as multiple nodules can affect the utility and accuracy of SE;

3. size of nodule (≤ 4 cm vs. > 4 cm);

4. solid versus mixed solid and cystic nodules.

A random factor was incorporated into the randomisation to reduce predictability and thus avoid selection bias. This means that each patient had a probability of either being minimised, or of receiving the opposite intervention to the one they would have received if they had been minimised.

Randomisation

Once eligibility was confirmed and after written informed consent was obtained, patients were randomised to the trial online using a bespoke trial website, or by telephone directly with the ElaTION Trial Office.

Information was needed on the number of nodules and their nature (solid or mixed) to enable randomisation. If multiple nodules were found, the nodule most suspicious at the time of the US was used for randomisation.

Experimental arm – strain or shear wave elastography in conjunction with ultrasound-guided fine-needle aspiration cytology

Those randomised to the intervention arm underwent USE together with routine US each time they had a diagnostic evaluation and FNAC of their nodule throughout the duration of the trial.

All repeat FNACs had to be undertaken using the same US technique as the first one specified by the randomisation and ideally by the same SE-accredited radiologist.

Control arm – conventional ultrasound-guided fine-needle aspiration cytology

Conventional grey scale and colour Doppler US-guided FNAC for all the FNACs was required until a definitive diagnosis was obtained.

All repeat FNACs were undertaken using the same US technique as the first one specified by the randomisation and ideally by the same USE-accredited radiologist.

Other management at the discretion of local doctors

Apart from the trial treatments allocated at randomisation, all other aspects of patient management were at the discretion of the local doctors, with no other special treatments or investigations and no additional follow-up visits.

Compatibility with other studies

A patient could be part of both ElaTION and another interventional trial, provided the other trial did not affect: (1) the decision to do a US-FNAC and (2) the decision to undertake surgery based on the FNAC result.

Trial schema and trial visit schedule

Follow-up assessments

Follow-up assessments were undertaken for one year from randomisation or until definitive diagnosis had been obtained if not achieved during that first year. This was sufficient time to allow for further FNACs, if required, after the first test. It also allowed sufficient time for any surgery to be undertaken and histological diagnosis to be available.

For both intervention and control arms, diagnosis and management proceeded as follows:

• Thy 1 – repeat FNAC, especially if suspicion of malignancy. If Thy 1 on three FNACs, then the recommendation was made for diagnostic surgery.

• Thy 2 – repeat FNAC within 3–6 months. If two benign (Thy 2) FNAC results were obtained, then the patient could be discharged. Consideration for discharge with a U2 scan and one Thy 2 was allowed, but not recommended for the purposes of this study.

• Thy 3a – repeat FNAC within 3–6 months or discussion with MDT. Surgery may then be advised.

• Thy 3f/4/5 – surgery was necessary.

All repeat FNAC were undertaken using the same US technique as the first one specified by the randomisation and ideally by the same SE-accredited radiologist.

All follow-up data were captured on the relevant case report form (CRF) and returned to the ElaTION Trial Office.

Patient assessment

Patient-reported assessments were performed using commonly used, validated questionnaires:

1. EQ-5D questionnaire completed at: baseline (at recruitment); after surgery; and 3, 6 and 12 months after randomisation;

2. cost collection form for health resource usage, completed at baseline, 3, 6 and 12 months after first FNAC;

3. Visual Analogue Pain Score after every FNAC;

4. HADS questionnaire completed at: baseline, 3, 6 and 12 months after randomisation.

Where possible, patient questionnaires were completed when patients attended hospital appointments. If this was not feasible, the questionnaires were posted by a member of the research team at the site, to the patient for completion at home.

Complication rates

Complication rates, following thyroid operations, were recorded at 30 days and 6 months post surgery.

Data were collected on haematoma rate and temporary hypocalcaemia rate at 30 days and vocal cord palsy and permanent hypocalcaemia at 6 months post surgery only.

Data collection

The data collection comprised the forms detailed in Table 2:

Primary outcome analysis

The primary outcome analysis was the proportion of patients who had a Thy 1 result following their first FNAC. A generalised linear mixed model was used to estimate the risk difference (RD) between the SE-FNAC and US-FNAC groups, with the minimisation variables: nodule multiplicity (solitary vs. multinodular); nodule composition (solid vs. mixed solid and cystic); and nodule size (≤ 4 cm vs. > 4 cm) included as fixed variables; and radiologist included as a random effect.

Secondary outcome analyses

All secondary outcomes (except the accuracy analyses) were adjusted for the minimisation variables.

Binary secondary outcomes (e.g. number of thyroidectomies) were analysed as per the primary outcome. For the proportion of patients with benign histology, all patients randomised to the trial were included in the denominator, to avoid bias if there was an imbalance in the number of thyroidectomies between the groups.

Time from first FNAC to definitive diagnosis was analysed using a Cox-proportional hazards model. Hazard ratios (HRs) were presented as well as the median and interquartile range (IQR). Kaplan–Meier curves were also produced for visual presentation of the comparisons.

The number of FNACs needed to obtain a definitive diagnosis was analysed using an ordinal regression model to obtain an odds ratio (OR).

Quality of life outcome measures included the HADS, VAPS and EQ-5D questionnaires. These were taken at baseline, 3, 6 and 12 months post randomisation. The outcome measures were analysed using repeated measures methods using all available data. The baseline value of the measure, time and minimisation variables as per primary analysis were included in the model as fixed effects, with radiologist included as a random effect.

Accuracy outcomes were presented as tables with corresponding sensitivity and specificity provided alongside the p-value from McNemar’s test.

The radiologist survey results, thyroidectomy complications and serious adverse events (SAEs) were presented in tables and summarised using numbers and percentages.

Missing data and sensitivity analyses

Sensitivity analyses were also performed on the primary outcome measure. These included: a per-protocol analysis, which included only those who receive the test intervention they were randomised to receive; an analysis excluding those centres where there was a cytologist present; an analysis including only those participants who had had a radiologist positive about USE; an analysis including only those radiologists who performed at least: 10 USEs; 20 USEs; 40 SEs; and 80 USEs in the trial; and finally an analysis including only those radiologists who passed their treatment quality assurance.

A sensitivity analysis was also carried out on the accuracy outcomes imputing missing participant final definitive diagnosis (FDD) with the following assumptions: if a participant has at least one FNAC result classified as ‘malignant’, then their FDD will be classified as malignant; and if a participant has a single Thy 2 result, or a Thy 2 result plus a Thy 1 then their FDD will be classified as benign.

Subgroup analyses

Subgroup analyses were planned on the stratification variables used for randomisation. These were: solitary nodule versus multinodular; the size of the nodule (< 4 cm vs. > 4 cm); and solid versus mixed solid and cystic nodules. Tests for interaction were performed to assess whether the intervention effect differs between the strata. The study has not been powered to detect any differences in these subgroups, so any significant results are purely hypothesis generating.

Defining final definitive diagnosis

For the accuracy outcomes, a definition of what would be considered a participant’s final diagnosis was required. The following is what was used as FDD:

Benign – If participant received two Thy 2 FNAC results; or participant received a U2 FNAC assessment with a Thy 2 result; or participant had surgery but the nodule was found to be benign.

Malignant – If the participant obtained a malignant histological diagnosis.

The number of participants and how FDD was obtained is summarised and presented in tables. Assumptions were placed on those missing FDD.

Cost-effectiveness analysis

A cost-effectiveness analysis was planned. However, when the preliminary results became available, it became clear that a cost effectiveness analysis was not appropriate, given there was no difference in primary and secondary outcomes between the two arms, so would be no cost-benefit to the addition of SE. Therefore, the cost-effectiveness analysis was not undertaken.

Ethics approval

ElaTION received full ethical approval from the South Central – Berkshire Research Ethics Committee (REC) on 10 October 2014. The REC reference is 14/SC/1206.

ElaTION was brought under the Health Research Authority (HRA) approval process in August 2016. The HRA assesses governance and legal compliance, and the ElaTION Trial Office was responsible for obtaining this approval.

Local hospitals conducted internal capacity and capability checks to assess the facilities and resources needed to run the trial, in order to give host site permission. The Trial Office provided help to the local Principal Investigator in the process of obtaining trust management approval by supplying the HRA Local Documents Package. The local Principal Investigator was responsible for liaison with the Trust management with respect to locality issues.

Once hospital approval was obtained, the ElaTION Trial Office confirmed that all appropriate site approvals were in place and that the USE accreditation had been completed. When the ElaTION Trial Office, on behalf of the sponsor, verified that all applicable regulatory requirements have been met, the local PI was informed that the study was open at the hospital and potential trial participants could start to be approached. The Trial Office sent the Investigator’s Site File containing all trial materials to the local Principal Investigator.

Sponsorship

Sponsorship was provided by the University of Birmingham upon signing of the Clinical Study Site Agreement with each trial site.

ElaTION was developed by the Institute of Head and Neck Studies and Education (InHANSE) team and the Birmingham Clinical Trials Unit (BCTU) and funded by the Health Technology Assessment (HTA) programme of the NIHR (12/19/04).

The University of Birmingham was the trial ‘sponsor’.

There were no specific arrangements for compensation made in respect of any SAEs occurring through participation in the trial, whether from the side effects listed, or others yet unforeseen. Hospitals selected to participate in this trial provided clinical negligence insurance cover for harm caused by their employees and a copy of the relevant insurance policy or summary was to be provided to the University of Birmingham, upon request.

No liability claims were submitted against the ElaTION trial.

Regulations

The trial was conducted in compliance with the principles of the Declaration of Helsinki (1996), the principles of GCP and in accordance with all applicable regulatory requirements including but not limited to the Research Governance Framework for Health and Social Care and the Medicines for Human Use (Clinical Trials) Regulations 2004, as amended in 2006 and any subsequent amendments.

Confidentiality of personal data

All data were handled in accordance with the UK Data Protection Act 1998 and subsequent amendments.

Participant name was not included on any CRF used in ElaTION. The participant’s initials, date of birth and trial identification number were used for identification.

Personal data and sensitive information required for the ElaTION trial were collected directly from trial participants and hospital notes. Participants were informed about the transfer of this information to the ElaTION Trial Office at the BCTU and InHANSE and asked for their consent. The data were entered onto a secure computer database, either directly via the internet using secure socket layer encryption technology or indirectly from paper by BCTU staff.

All personal information received in paper format for the trial was held securely and treated as strictly confidential according to BCTU policies. All staff involved in the ElaTION trial (clinical, academic, BCTU, InHANSE) shared the same duty of care to prevent unauthorised disclosure of personal information. No data that could be used to identify an individual will be published. Data are stored on a secure server at BCTU under the provisions of the Data Protection Act and/or applicable laws and regulations.

In-house data quality assurance

Independent Trial Steering Committee

The Trial Steering Committee (TSC) provided independent supervision for the trial, providing advice to the Chief and Co-Investigators as well as the Sponsor on all aspects of the trial and affording protection for patients by ensuring the trial was conducted in compliance with the protocol, GCP and the applicable regulatory requirements.

Data Monitoring and Ethics Committee

During the study, interim analyses of safety and outcome data were supplied, in strict confidence, to an independent Data Monitoring and Ethics Committee (DMEC) along with any other analyses that the committee requested.

Long-term storage of data

Archiving will be authorised by the BCTU on behalf of the Sponsor following submission of the end of trial report.

Principal Investigators are responsible for the secure archiving of essential trial documents (for their site) as per their NHS Trust policy. All essential documents will be archived for a minimum of 5 years after completion of the trial.

Destruction of essential documents requires authorisation from the BCTU on behalf of the Sponsor.

Adverse events

The collection and reporting of data on adverse events (AE) and SAEs was done in accordance with International Conference on Harmonisation Guidelines for Good Clinical Practice (ICH GCP) and the Research Governance Framework 2005.

No investigational medicinal products (IMPs) were used as part of this trial. As all of the surgical techniques being tested in this trial were used as standard practice, there were no (serious) AEs which would be anticipated as a unique consequence of participation in the trial. Any trial-related SAEs, which required immediate reporting, were reported on a trial-specific SAE form.

Other outcomes, which may also be considered safety outcomes, but which were anticipated outcomes for this group of patients, were captured on the routine follow-up CRFs (30-day and 6-month Post-Operative CRFs), these included:

• vocal cord palsy

• temporary or permanent hypocalcaemia

• haematoma

• infection

• re-operation due to surgical complications.

Serious adverse events

Notification of deaths

All deaths were reported to the BCTU on the SAE Form irrespective of whether the death was related to disease progression or an unrelated event. If a participant died, any post-mortem findings were provided to the BCTU with the SAE form. The BCTU reported all deaths to the DMEC for continuous safety review.

Pharmacovigilance responsibilities

Notification of serious breaches of GCP and/or the protocol

A ‘serious breach’ is a breach which is likely to affect to a significant degree:

1. the safety or physical or mental integrity of the participants of the trial; or

2. the scientific value of the trial.

The BCTU on behalf of the Sponsor notified the REC in writing of any serious breach of:

1. the conditions and principles of GCP in connection with the trial; or

2. the protocol relating to the trial, as amended from time to time, within 7 days of becoming aware of that breach.

The Sponsor was notified immediately of any case where the above definition applied during the trial conduct phase.

Adherence to randomised test

Adherence to the participant’s randomised test was high. For the initial FNAC assessment, adherence was 99% or higher for both the USE-FNAC and US-FNAC groups (see Table 5). The adherence dropped slightly for FNAC assessments following the initial FNAC, with a larger drop observed in the USE-FNAC group compared with the US-FNAC group, but adherence was still good (85% or above) by the third FNAC.

Adherence to final definitive diagnosis protocol definition

Out of 982 participants randomised, 688 reached a FDD. There were 346 participants randomised to the USE-FNAC group who had a FDD, of which 276 had a benign FDD and 70 had a malignant diagnosis. In the US-FNAC group 342 participants had a FDD, 263 had a benign FDD and 79 had a malignant FDD. Table 6 shows how these final definitive diagnoses were obtained.

Primary analysis

The primary outcome is the proportion of participants with a Thy 1 result following the first FNAC post-randomisation. The sample size was based on a Thy 1 rate of 10% in the US-FNAC arm and a 4% Thy 1 rate in the USE-FNAC arm.

The Thy 1 rate following the first FNAC observed in the trial was higher than anticipated in both the USE-FNAC and US-FNAC groups (19% vs. 16% respectively). However, there was no evidence of a difference between the groups found (RD 0.03; 95% CI, −0.007 to 0.066; p = 0.11) (see Table 7).

Sensitivity analyses

There were a number of sensitivity analyses carried out to assess the effect, if any, on inferences including: adherence to the test; there being a cytologist present at the centre; radiologist opinion of elastography being positive; and radiologist passing on treatment quality assurance.

The per-protocol analysis result, which included only those who adhered to randomised test, supported the primary analysis (RD: 0.029; 95% CI, −0.007 to 0.065; p = 0.11). The only centres with a cytologist present were King’s College Hospital and St. Peter’s Hospital. However, two participants recruited from St. Peter’s Hospital had their FNAC performed at Ashford Hospital, which did not have a cytologist present. Again, the results from this analysis excluding cases where a cytologist was present, whilst seeing a slight drop in Thy 1 results (17% vs. 14% for USE-FNAC and US-FNAC respectively) saw no evidence of a difference between the groups (see Report Supplementary Material 1: Table 5).

Similarly, the analyses including only those radiologists who felt elastography helped above conventional US in determining malignancy and the radiologists who passed on treatment quality assurance supposed the primary analysis (see Report Supplementary Material 1: Table 6).

We also performed an analysis to assess whether the accuracy of using elastography increased with the number of SEs. Whilst the percentage of Thy 1 rates decreased in the USE-FNAC group (see Table 8), they also decreased in the US-FNAC group. The differences between USE and US remained broadly the same as per the primary analysis, despite the volume of USE scans done, and the direction of effect was in favour of US-FNAC.

Thy 1 result following any FNAC

When considering the number of Thy 1 results following any FNAC performed, a borderline significant difference was observed in favour of US-FNAC when compared with USE-FNAC (RD: 0.041; 95% CI 0.002 to 0.081; p = 0.04) (see Table 9). However, the number of follow-up FNACs would be influenced by the result of the first FNAC result, so this result is to be treated with caution as it is subject to bias. Indeed, the Thy 1 rate appears to increase with successive FNACs, and that occurs for both USE and US guided. This suggests that as non-Thy 1 cases are excluded, more Thy 1 cases remain in the pool, and therefore represent a high proportion of the pool. It also suggests that a large proportion of the Thy 1 cases remain Thy 1 despite successive FNAs.

Thyroidectomies and benign histology

Patients in the USE arm received fewer thyroidectomies than those in the US-FNAC group (37% vs. 40% respectively), but this was not statistically significant (see Report Supplementary Material 1: Table 7). Although not powered sufficiently, there was no evidence of a difference in the rate of benign histology between the groups in the trial (RD: −0.007; 95% CI −0.04 to 0.03; p = 0.70).

Final definitive diagnosis

There were 688/982 (70%) participants who had a FDD. Details of how these were reached can be found in Report Supplementary Material 1: Table 8. For the 294 without a FDD, 102 (35%) had a Thy 2 result following first FNAC and of these, 76 did not receive a follow-up FNAC (see Report Supplementary Material 1: Table 9).

The percentage of participants who reached FDD was the same in each group: 346/493 (70%) in the USE-FNAC group and 342/489 (70%) in the US-FNAC group.

Time to final definitive diagnosis

No difference was observed in the trial between the groups when examining time to FDD (HR: 0.94; 95% CI 0.81 to 1.10; p = 0.45) (see Table 10, Figure 3).

FIGURE 3.

Time to FDD Kaplan–Meier Plot (unadjusted HR included in the figure).

Number of FNACs until final definitive diagnosis

The median number of FNAC required to reach FDD was higher in the USE-FNAC group (median: 2.0; IQR 1.0–2.0) than the US-FNAC group (median: 1.0; IQR 1.0–2.0). The odds of having more FNACs in the USE-FNAC was 1.10 when compared with the US-FNAC group holding all other variables constant, however, this was not statistically significant (see Report Supplementary Material 1: Table 10).

Hospital and Depression Scale

Patients having USE were generally slightly more anxious and reported more depression at baseline (see Report Supplementary Material 1, Table 11). There appeared to be some divergence between the arms in both anxiety and depression at 6 months in favour of USE-FNAC (see Appendix 1, Figures 4 and 5), and there was evidence of an interaction between test and assessment on the depression scale (see Report Supplementary Material 1: Table 12).

However, there was no evidence of an overall difference between USE-FNAC and US-FNAC in either the anxiety subscale [mean difference (MD): −0.18: 95% CI −0.43 to 0.07; p = 0.16] or the depression subscale (MD: −0.006; 95% CI −0.27 to 0.26; p = 0.97).

EQ5D-5L

EQ5D-5L was included as a quality-of-life outcome measure predominantly for the health economic analysis. The statistical analyses showed no differences between the groups at any assessment point (see Report Supplementary Material 1: Table 13), nor when looking at overall assessments (see Report Supplementary Material 1: Table 14).

Visual Analogue Pain Scale

If greater accuracy was obtained by using USE-FNAC compared with US-FNAC, then this might have resulted in participants experiencing less pain from the procedure.

Slightly less pain was observed following the first FNAC in the USE-FNAC group (see Table 11), however, not significantly so. The direction of the pain score favours US-FNAC for the third and extra FNAC; however, numbers are smaller and the differences still remain insignificant.

Accuracy of first FNAC results compared to second FNAC

For those with a FDD of malignant, there were only 13 participants with a first and second FNAC that was not a Thy 1 result (and therefore could be included in the analysis). Due to this, of those included in the analysis, all 13 have a malignant diagnosis on their second FNAC (see Table 12). The sensitivity was 0.54 following the first FNAC, compared with 1.00 for the second FNAC. Sensitivity analyses were performed where assumptions were placed on both Thy 1 results and those without a FDD. These can be found in Report Supplementary Material 1: Table 15.

For those with a benign FDD, there is some evidence that the specificity is better on the first FNAC (specificity = 0.89) compared with the second FNAC (specificity = 0.84), p = 0.03 (see Table 13). The sensitivity analyses for benign nodules are presented in Report Supplementary Material 1: Table 16.

Agreement between first FNAC and second FNAC

Agreement between the first FNAC and second FNAC was examined in those who had at least two FNACs, and in the first instance looking at only those who had a diagnostic result. The result of the first FNAC changed on the second FNAC in 72/330 (21.8%) cases. Importantly in 13% of the cases, it changed from benign to malignant, and in 8.8% it changed from malignant to benign (see Table 14). Secondary analyses of agreement between the first and second FNACs were performed (see Report Supplementary Material 1: Table 17).

Accuracy of USE/US without FNAC and accuracy of USE/US with FNAC

Surgery complications

Out of 982 participants in the trial, 379 received surgery. Complications were reported at 30 days and 6 months post surgery. There were two deaths incorrectly reported on the surgery complication forms: one participant died due to cancer; the other due to a bowel obstruction. Overall, the surgical complications were very low (see Table 21). Further details on the complications can be found in Report Supplementary Material 1: Table 23.

Serious adverse events

As anticipated, the number of SAEs was very low. Thirteen out of the 982 participants (1%) experienced an SAE (see Report Supplementary Material 1: Table 24). There were marginally more SAEs in the USE-FNAC group than the US-FNAC group (10 in 9 participants vs. 4 in 4 participants respectively), though not significantly so (p = 0.26). Details on the SAEs by group can be found in Report Supplementary Material 1: Table 25.

Patient and public representatives

Judith Taylor (Thyroid Cancer Alliance)

Oversight Committees