[¹⁸F]FDG-PET/CT to prevent futile surgery in indeterminate thyroid nodules: a blinded, randomised controlled multicentre trial

The Efficacy of [¹⁸F]FDG-PET in Evaluation of Cytological indeterminate Thyroid nodules prior to Surgery (EfFECTS) trial was a blinded, randomised controlled multicentre trial performed in all eight academic and seven large community hospitals in the Netherlands (Supplementary Data p3). At all study sites, local investigators and physicians were highly experienced in the multidisciplinary diagnosis and treatment of thyroid nodules and thyroid carcinoma and worked in accordance with national and international guidelines [4, 10]. Adult euthyroid patients in whom diagnostic surgery was scheduled for an indeterminate thyroid nodule, defined as Bethesda III (confirmed on two subsequent FNAC procedures) or Bethesda IV cytology, were eligible for study participation [3]. Bethesda III or IV diagnosis was established by blinded central review by two dedicated thyroid pathologists (AE and BK). Prior to inclusion in the trial, clinical and ultrasound characteristics of the index nodule were considered in a multidisciplinary setting by the local physicians to establish the indication for diagnostic surgery, in accordance with current guidelines [4]. Patients were excluded from study participation if they had contraindications for [¹⁸F]FDG-PET/CT or a higher a priori risk of thyroid malignancy based on their presentation or history (i.e., unexplained stridor, vocal cord paralysis or radiation exposure to the thyroid), if they already underwent any non-routine preoperative diagnostic stratification (i.e., [¹⁸F]FDG-PET/CT or molecular diagnostics) or were unable to undergo randomisation (e.g., patient preference for surgery) [11]. Full eligibility criteria are listed in the study protocol (Supplementary Data). Written informed consent was obtained from all participants prior to any study activity. The study protocol was approved by the Medical Research Ethics Committee on Research Involving Human Subjects region Arnhem-Nijmegen, Nijmegen, the Netherlands. The trial was overseen by a trial steering committee and an independent study safety committee. The funder of the study had no role in its design, data collection and analysis, or writing of this report.

Patients were individually randomly assigned to the [¹⁸F]FDG-PET/CT-driven group or diagnostic surgery group in a 2:1 ratio. To pursue an even distribution of risk factors for differentiated thyroid carcinoma across both arms, stratification was applied by patient sex, age (dichotomised at 45 years), ultrasonographic thyroid nodule size (0–10, 11–20, 21–40, or > 40 mm), Bethesda classification (III or IV), and study site. Randomisation was performed in the trial management system, Castor Electronic Data Capture (Castor EDC, Amsterdam, the Netherlands), which uses a validated variable block randomisation model.

All patients underwent an [¹⁸F]FDG-PET/CT of the neck, acquired by 20 PET/CT scanners at 12 EARL-accredited study sites (Supplementary Table 1) using a standard acquisition and reconstruction protocol in accordance with European Association of Nuclear Medicine (EANM) guidelines [11, 12]. In summary, patients fasted for at least 6 h prior to the injection of the radiopharmaceutical (activity adjusted for patient body weight, time per bed position, and PET/CT scanner sensitivity). Approximately 60 min (range 55–70 min) after intravenous [¹⁸F]FDG administration, patients were scanned from the external acoustic meatus to the aortic arch in a supine position, with at least 2 min per bed position. A low-dose non-contrast–enhanced CT scan was performed. EARL-reconstructed, pseudonymised scans were stored in a central, password-protected online database within the National Biomedical Imaging Archive environment (NBIA, National Cancer Institute, Bethesda, MD, USA). Next, scans were centrally assessed by two independent, experienced nuclear medicine physicians (LG and DV). Any focal [¹⁸F]FDG-uptake within the thyroid that was visually higher than the background [¹⁸F]FDG-uptake of the surrounding normal thyroid tissue and that corresponded to the index thyroid nodule in size and location was considered positive. To support the visual assessment, [¹⁸F]FDG-uptake was quantified using maximum and peak (ø1-cm sphere) standardised uptake values (SUV_max, SUV_peak), using body weight for normalization. In case of a discordant assessment, a third reviewer (WO) was consulted for a consensus meeting. All image analyses were performed using OsiriX Lite DICOM-viewer (Pixmeo SARL, Bernex, Switzerland).

One project team member (EK) combined patient allocation, and the [¹⁸F]FDG-PET/CT result to a preformulated treatment advice. A written report containing only this advice was presented to the patient’s local physician; the patient’s allocation and [¹⁸F]FDG-PET/CT result remained concealed. If present, incidental [¹⁸F]FDG-PET/CT-findings outside the index nodule and with potential diagnostic and/or therapeutic consequences were also reported in an appendix to the report, to be evaluated by the local physician in the context of the patients’ medical history. In the [¹⁸F]FDG-PET/CT-driven group, the treatment advice was based on the [¹⁸F]FDG-PET/CT results. When the index nodule was [¹⁸F]FDG-positive, patients were advised to proceed to the scheduled diagnostic surgery. When the nodule was [¹⁸F]FDG-negative, patients were advised to refrain from surgery and undergo active surveillance of the nodule, which was defined as at least one follow-up ultrasound exam of the neck and outpatient clinic visit after one year. Any additional follow-up visits during study participation were permitted at the discretion of the local physician. The nodule was presumed benign when it remained unchanged in size and appearance on the one-year ultrasound. In case of significant growth (> 50% volume change or > 20% increase in at least two dimensions, excluding cystic components) or changed ultrasound appearance including newly observed suspicious characteristics, further evaluation by repeat FNAC was recommended. Suspicious ultrasound characteristics were defined as a marked hypoechoic solid nodule, irregular shape (i.e., taller-than-wide), irregular margins (i.e., lobulated, infiltrative), and/or presence of microcalcifications.

In the diagnostic surgery group, the treatment advice for all patients was to proceed to the scheduled diagnostic surgery, in accordance with the current international guidelines [4, 10]. In both study groups, the patient and his/her physician were free to deviate from the study treatment advice at any time.

All postoperative patient management was based on the local histopathological diagnosis and current international guidelines [4]. After completion of all study procedures and data collection, all histopathology was centrally reviewed by a dedicated thyroid pathologist (AE). In case of a discordant review, a second central pathologist was consulted for a consensus meeting. Incidentally detected (micro)carcinomas located outside the index nodule were not considered for the main outcome measures.

HRQoL and societal costs were assessed during one year, calculated from the date of the [¹⁸F]FDG-PET/CT scan. Patients were asked to complete the EuroQol 5-dimension 5-level questionnaire (EQ-5D-5L), the iMTA Medical Consumption Questionnaire (iMCQ), and the iMTA Productivity Costs Questionnaire (iPCQ) at 0 (baseline), 3, 6, and 12 months (Supplementary Data p9) [13‐16]. Societal costs (in €) included all direct medical costs for thyroid-related and other healthcare consumption, patient costs (i.e., informal care, travel expenses), and productivity losses. Volumes of healthcare consumption were extracted from individual patient files and the iMCQ. Costs were valued using reference prices and the 2019 reimbursement rates of the Dutch System of Diagnosis-Treatment Combinations, where appropriate and available (Supplementary Data p10). The estimated cost of one partial-body [¹⁸F]FDG-PET/CT scan was €754 [17, 18]. All [¹⁸F]FDG-PET/CT-related costs, including the costs of the scan itself and any additional healthcare consumption for incidental [¹⁸F]FDG-PET/CT findings, were only taken into account for the [¹⁸F]FDG-PET/CT-driven group.

Patients, all local study site personnel, and all pathologists were blinded to [¹⁸F]FDG-PET/CT results and allocation. Central pathologists were additionally blinded to the local cyto- and histopathological diagnoses. Central nuclear medicine physicians were blinded to allocation and all clinicopathological data except for the ultrasonographic size and location of the index nodule. Other study investigators assessing outcomes were blinded to allocation. Patients allocated to the [¹⁸F]FDG-PET/CT-driven group with an [¹⁸F]FDG-negative nodule could inevitably deduce their allocation and [¹⁸F]FDG-PET/CT result from the surveillance advice.

The primary outcome was the fraction of patient management that was considered unbeneficial one year after the [¹⁸F]FDG-PET/CT scan. Unbeneficial management was defined as futile diagnostic surgery for histopathologically benign nodules (including hyperplastic nodules, follicular adenoma, and Hürthle cell adenoma) and/or unjustified surveillance for histopathological malignant or borderline tumours. According to novel insights arising during the trial, nodules diagnosed as noninvasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP) or follicular tumour of uncertain malignant potential (FT-UMP) are considered benign yet potentially premalignant (i.e., borderline) lesions, for which surgery is considered justified [19, 20]. Following broad acceptance of these insights, we added a refinement concerning these borderline tumours to the study protocol during the trial. As histopathology was not reviewed until trial completion, this modification did not in any way influence trial execution and primary endpoints. The results for the primary outcome for a strictly benign-malignant differentiation are reported in the Supplementary Data (p16).

Secondary outcomes of the trial included the differences in surgical complication rates, one-year HRQoL expressed in quality-adjusted life years (QALYs), and societal costs (in €) between both strategies. The diagnostic accuracy of [¹⁸F]FDG-PET/CT (whole-group analysis) was estimated: [¹⁸F]FDG-positive or [¹⁸F]FDG-negative nodules confirmed as malignant or borderline tumours on histopathology were considered true-positive or false-negative, respectively; [¹⁸F]FDG-positive or [¹⁸F]FDG-negative nodules confirmed as benign on histopathology and those that remained unchanged on the one-year ultrasound were considered false-positive or true-negative, respectively. Finally, the number of incidental [¹⁸F]FDG-PET/CT findings with diagnostic and/or therapeutic consequences in the scanned area (descriptive whole-group analysis), implementability of [¹⁸F]FDG-PET/CT (i.e., diagnostic confidence) defined as the fraction of patients not reassured by a negative [¹⁸F]FDG-PET/CT result, and survival were assessed. Per protocol, the follow-up for all endpoints was set at one year after [¹⁸F]FDG-PET/CT. Whenever available and relevant, data beyond one year of follow-up (censored 1 October 2021) are presented.

The trial was designed to have 80% power to detect a reduction in unbeneficial management from ~ 75% to ~ 40% at a significance level of 0.05. At least 90 evaluable patients with nodules > 10 mm were required (2:1 allocation). After correction for 82.7% expected nodule size > 10 mm and 15% estimated data-attrition, the sample size was set at 132 patients [5, 7].

After half of the anticipated patients of the diagnostic surgery group were recruited, the study safety committee conducted a predetermined interim analysis and reported no objections to safe continuation of the trial.

The applied descriptive statistics were mean ± standard deviation or median and interquartile range for continuous variables and absolute numbers and relative frequencies (%) for categorical variables. Intention-to-treat analysis was performed. Categorical primary and secondary outcomes were compared between allocated groups using Pearson’s chi-squared or Fisher’s exact tests, where appropriate. We adjusted for the stratifying variables using binary logistic regression; the corrected p values are reported together with an adjusted odds ratio and their 95% confidence intervals (CI) [21].

Sensitivity, specificity, and negative and positive predictive value (NPV, PPV) were calculated using the traditional formulas. 95% CIs were calculated using the β-distribution (Clopper-Pearson interval). For EQ-5D-5L, iMCQ, and iPCQ questionnaires, we used multiple imputation to account for possible selectively missing values. To estimate HRQoL, we calculated Dutch utility scores from the EQ-5D-5L and estimated the mean one-year QALYs as the area under the utility curves (Supplementary Data p8). One-year societal costs were estimated as the mean sum of [volume x costs] for all components. QALYs and costs are presented as mean and 95% CI and compared using independent samples T-tests with unequal variances. In the analysis of QALYs and costs, we adjusted for the stratifying variables and additionally adjusted for possible influences of the unevenly distributed malignancy/borderline rate (see Results, and Supplementary Tables 4 and 5), using a generalized linear model (GLM). The local benign/malignant histopathological diagnosis was included in the GLM as a covariate, as this diagnosis determined the patients’ postoperative course of treatment and thus contributed to the costs and perceived HRQoL. The adjusted means, p values, and mean differences are presented.

Two prespecified subgroup analyses for the primary outcome were performed: one for nodules > 10 mm (ultrasonographic largest diameter) and one for Hürthle cell nodules (defined as Bethesda IV HCN/SHCN cytology) and non-Hürthle cell nodules (defined as Bethesda III AUS/FLUS and Bethesda IV FN/SFN cytology).

Data collection was performed using Castor EDC (Castor EDC, Amsterdam, the Netherlands). Statistical analysis was performed using SPSS Statistics version 26 (IBM Corp, Armonk, NY, USA). This trial is registered with ClinicalTrials.gov NCT02208544 (5 August 2014).

After one year of follow-up, patient management had been unbeneficial in 38 of 91 (42% [95% CI, 32–53%]) patients in the [¹⁸F]FDG-PET/CT-driven group, compared to 34 of 41 (83% [95% CI, 68–93%]) patients in the diagnostic surgery group (p < 0.001, OR 0.1 [95% CI, 0.1–0.4]). These were all futile diagnostic surgeries for histopathologically benign nodules. There was no unjustified surveillance: no malignancies or borderline tumours were observed in patients under surveillance. [¹⁸F]FDG-PET/CT-driven management avoided surgery for 25 of 63 (40% [95% CI, 28–53%]) benign nodules. In comparison, only one of 35 (3% [95% CI, 0–15%]) patients in the diagnostic surgery group did not undergo the recommended surgery and was considered benign on ultrasound follow-up (p = 0.002, OR 26.9 [95% CI, 3.3–219.0]) (Table 2).

Sensitivity, specificity, NPV, PPV, and benign call rate of [¹⁸F]FDG-PET/CT were 94.1% (95% CI, 80.3–99.3%), 39.8% (95% CI, 30.0–50.2%), 95.1% (95% CI, 83.5–99.4%), 35.2% (95% CI, 25.4–45.9%), and 31.1% (95% CI, 23.3–39.7%), respectively (Table 3). Two of 132 (1.5%) [¹⁸F]FDG-PET/CT scans were false-negative (both in the diagnostic surgery group). In both cases, the corresponding index nodules had caused extensive debate during the blinded interpretation of the histopathology (i.e., benign or malignant diagnosis). One was a 15 mm, RAS-mutated, non-invasive neoplasm with uncommon spindle cell metaplasia, which was ultimately classified as a papillary thyroid carcinoma (PTC, TNM pT1b). The other was a 32 mm, predominantly cystic, non-invasive neoplasm with a solid component of 8 mm. It was only considered malignant (follicular variant of PTC, TNM pT2) after detection of an ETV6-NTRK3 fusion during the central review of the histopathology (details provided in the Supplementary Data p18).

No difference in the surgical complication rate was observed between both groups. Still, following the reduction in diagnostic surgeries, the rate of new levothyroxine suppletion-dependent hypothyroidism due to partial thyroidectomy procedures was only 6% (5/86) in the [¹⁸F]FDG-PET/CT-driven group as compared to 17% (7/41) in the diagnostic surgery group (p = 0.07, OR 0.3 [95% CI, 0.1–1.1]). Other surgical complications infrequently occurred (Table 4).

EQ-5D-5L questionnaires were completed by 69 of 91 (76%) patients in the [¹⁸F]FDG-PET/CT-driven and 29 of 41 (71%) patients in the diagnostic surgery group (p = 0.54). Perceived HRQoL during the first year after the [¹⁸F]FDG-PET/CT scan was similar in both groups. Adjusted for the stratifying variables and malignancy/borderline rate, a mean of 0.793 (95% CI, 0.753–0.833) QALYs was estimated in the [¹⁸F]FDG-PET/CT-driven group, as compared to 0.725 (0.651–0.799) QALYs in the diagnostic surgery group (p = 0.11). The adjusted mean societal costs during the first year were significantly lower in the [¹⁸F]FDG-PET/CT-driven group than the diagnostic surgery group: €14,800 (95% CI, + €12,600– + €17,000) as compared to €21,700 (+ €16,800– + €26,600) per patient, respectively, with an adjusted mean difference of − €6,900 (-€12,100– − €1,600, p = 0.01) (Table 5 and Supplementary Data p8-10).

Incidental [¹⁸F]FDG-PET/CT findings with diagnostic or therapeutic consequences were reported for 22 of 132 (17%) [¹⁸F]FDG-PET/CT scans (Table 4, Supplementary Data p26-27). These included 21 [¹⁸F]FDG-positive thyroid incidentalomas in 19 (14%) patients, for which 13 additional FNAC procedures were performed. Eleven of 21 (52%) incidentalomas were surgically resected. Two ipsilateral incidentalomas were malignant. In four patients, their initially scheduled hemithyroidectomy was extended to a total thyroidectomy to include a contralateral incidentaloma; all were histopathologically benign: one follicular adenoma and three hyperplastic nodules. These total thyroidectomy procedures in 4 of 132 (3%) of patients are considered overtreatment due to the [¹⁸F]FDG-PET/CT.

Diagnostic confidence in [¹⁸F]FDG-PET/CT was high: only one of six patients who underwent surgery (three during and three after study follow-up) despite advised surveillance (Fig. 1) was not fully reassured by the negative [¹⁸F]FDG-PET/CT result. The main reason for surgery in all six patients, however, was not the fear or suspicion of cancer but increasing compressive symptoms causing discomfort. Noncompliance to the surveillance advice did not change the one-year therapeutic yield in the [¹⁸F]FDG-PET/CT-driven group, as patient crossover between surgical and non-surgical management occurred in both directions (Fig. 1). Based on theoretical full compliance to the given treatment advice, a maximum 41% reduction in futile diagnostic surgeries for benign nodules (i.e., 26 [¹⁸F]FDG-negative nodules of 63 benign nodules) was estimated following full implementation of [¹⁸F]FDG-PET/CT (p = 0.86).

Subgroup analysis of nodules > 10 mm (128/132 patients, excluding two 10 mm nodules from each group) demonstrated similar therapeutic yield and diagnostic accuracy as compared to the main results (Supplementary Data p12).

Subgroup analysis of the 101 non-Hürthle cell nodules (60 AUS/FLUS and 41 FN/SFN) and 31 Hürthle cell (HCN/SHCN) nodules was performed. The malignant/borderline rate was 17% (10/60) in Bethesda III as compared to 33% (24/72) in Bethesda IV nodules (p = 0.03), of which 37% (15/41) in FN/SFN and 29% (9/31) in HCN/SHCN nodules (p = 0.50).

In non-Hürthle cell nodules, the fractions of unbeneficial management and prevented surgeries for benign nodules after one year were 37% (95% CI, 25–49%) and 48% (95% CI, 33–63%) in the [¹⁸F]FDG-PET/CT-driven group, as compared to 85% (95% CI, 68–95%) (p < 0.001) and 0% (95% CI, 0–18%) (p < 0.001) in the diagnostic surgery group (Table 6). Sensitivity, specificity, NPV, PPV, and benign call rate in non-Hürthle cell nodules were 92.0% (95% CI, 74.0–99.0%), 50.0% (95% CI, 38.3–61.7%), 95.0% (95% CI, 83.1–99.4%), 37.7% (95% CI, 25.6–51.0%), and 39.6% (95% CI, 30.0–49.8%), respectively (Table 3). Therapeutic yield and diagnostic accuracy were similar in AUS/FLUS and FN/SFN nodules. In Hürthle cell nodules, [¹⁸F]FDG-PET/CT showed a benign call rate of only 3.2% (1/31). Consequently, [¹⁸F]FDG-PET/CT-driven management was not contributory to improve the diagnostic workup: the fractions of unbeneficial management and prevented surgeries for benign Hürthle cell nodules were low and similar in both study groups (p = 1) and included one patient in each group who declined the advised surgery (Table 6). [¹⁸F]FDG-PET/CT-driven management avoided significantly more futile surgeries in non-Hürthle cell nodules (48% [95% CI, 33–63%]) than in Hürthle cell nodules (13% [95% CI, 2–40%]) (p = 0.02).