Prevalence of incidental thyroid malignancy on routine 18F-fluorodeoxyglucose PET-CT in a large teaching hospital

Purpose To quantify incidental thyroid pathology including malignancy on routine 18F-FDG PET-CT scans To compare standardised uptake values (SUVmax) in thyroid malignancy subtypes Methods and materials This is a retrospective study of all 18F-FDG PET-CT scans (n = 6179) performed in a teaching hospital between June 2010 and May 2019. RIS database search of reports for the word “thyroid” was performed. Studies with evidence of thyroid uptake were included. Patient age and gender, primary indication for PET scan (malignant or non-malignant), thyroid result on PET (diffuse or focal tracer uptake, SUVmax), ultrasound and FNAC results were recorded. Results Incidental abnormal thyroid tracer uptake as a proportion of all 18F-FDG PET-CT scans was 4.37% (n = 270). Out of region patients (n = 87) whose records could not be obtained were excluded leaving a study group of n = 183. Ninety-four in this group had focal uptake, and 89 had diffuse uptake. Fifty-five patients in the focal group had undergone further investigations. Of these, 30 were thought to be benign on USS alone, and 25 patients underwent USS/FNAC. Thirteen (24%) malignancies were identified (5 papillary, 6 follicular, 1 poorly differentiated thyroid cancer, 1 metastatic malignancy). Mean SUVmax for papillary carcinoma was noted to be 8.2 g/ml, and follicular carcinoma was 12.6 g/ml. Conclusion Incidental abnormal thyroid 18F-FDG PET-CT uptake in PET-CT scans of 4.37% is in keeping with the known limited literature. Rather similar number of patients was noted in the focal and diffuse tracer uptake categories in the final study group. Around quarter of the focal lesions were identified to be malignant, implying focal lesions should always be further investigated.


Introduction
The increasing role of PET-CT in oncological and non-oncological conditions has resulted in an increased detection of PET-CT incidentalomas, commonly involving the thyroid gland (The Royal College of radiologists, 2012; Delivanis & Castro, 2018;Vassiliadi & Tsagarakis, 2011).
The tracer 18 F-FDG 1 used in PET-CT can incidentally accumulate in the thyroid gland, either diffusely or focally. Incidental focal 18 F-FDG uptake within the thyroid gland has previously been found to occur in 1.2-4.3% of all PET-CT scans in patients scanned for an alternative indication (Kao et al., 2012;Soelberg et al., 2012;Cohen et al., 2001;Kang et al., 2003;Chen et al., 2005;Chen et al., 2009;Ho et al., 2011). Patients with focal uptake within the thyroid gland are at a higher risk of malignancy, with studies reporting between 26 and 50% (Kao et al., 2012;Soelberg et al., 2012;Cohen et al., 2001;Chen et al., 2005;Chen et al., 2009;Kim et al., 2005;Chu et al., 2006;Bae et al., 2009). Figures 1 and 2 are two patients in our institution scanned for an alternative indication who subsequently were found to have thyroid malignancy. Further investigation of focal 18 F-FDG uptake with ultrasound and FNAC 2 is recommended because of this increased risk of malignancy (Pencharz et al., 2018;Haugen et al., 2015;Hoang et al., 2015).
PET-CT can deduce semi-quantitative evaluations of glucose metabolism in tissues by measuring standardised uptake value (SUV max ) (Chu et al., 2014). The role of SUVmax in thyroid malignancy is debated as studies have produced conflicting results, and, although reasonable to suggest that malignancy is associated with a higher SUV max value (Kumar et al., 2013), some studies have shown no correlation (Eloy et al., 2009;Are et al., 2007).
Although the usefulness of SUV max in thyroid malignancy is questioned, a cytological diagnosis of focal thyroid FDG-PET incidentalomas is necessary considering the increased risk of malignancy.
The primary aim of our study was to quantify incidental thyroid malignancy on 18 F-FDG PET-CT in patients scanned for an alternative indication. A secondary outcome was to compare mean, median, and range of SUV max according to thyroid malignancy subtypes.

Methods and materials
This is a retrospective study of all 18 F-FDG PET-CT scans performed in a large teaching hospital between June 2010 and May 2019. We performed a CRIS (CDN Radiology Information System) database search, and results were filtered to include the word "thyroid" within the scan reports. All 18 F-FDG PET-CT scans in our institution are double reported by four experienced consultants, two of whom are also experienced in routine thyroid imaging. Studies with no thyroid uptake were excluded. We manually collected patient's data including age and gender, primary indication for PET-CT scan (malignant or non-malignant), thyroid result on PET (diffuse or focal uptake of 18 F-FDG), ultrasound result and result of FNAC. Total number of FNAC diagnosed as thyroid malignancy and mean SUV max values of thyroid malignancy subtypes were collected. Those patients who had tracer uptake but were lost to follow-up were excluded.

Results
Six thousand, one-hundred and seventy-nine (n = 6179) 18 F-FDG PET scans were performed during study period. Three-hundred and forty-two (n = 342) results contained the word "thyroid" of which two-hundred and seventy (n = 270) scans had increased uptake in thyroid gland. Eighty-seven (n = 87) patients were excluded (out of region patients, deaths, lost to follow-up). This left a study group of one-hundred and eighty-three (n = 183) of which one hundred and twenty-seven (n = 127) were females, and fifty-six (n = 56) were males. Ninetyfour (n = 94) PET-CT scans had focal uptake, and eighty-nine (n = 89) had diffuse uptake. In our study group, those with diffuse uptake included seventy female patients (n = 70) and nineteen (n = 19) male patients. Fifty-seven females (n = 57) and thirty-seven male patients (n = 37) had focal uptake in our study group. Fifty patients in the focal group underwent further investigations. Of these, 30 were thought to be benign on USS alone and 25 patients underwent USS/FNAC, where 52% (52%, n = 13/25) were confirmed to be malignancy. Thyroid malignancy subtypes included papillary (n = 5), follicular (n = 6), metastatic malignancy (n = 1) and poorly differentiated carcinoma of the thyroid (n = 1). Patients who had a previous history of thyroid cancer with increased uptake were included in the study (n = 1). Eight (8/25) pathology result confirmed benign lesions including non-neoplastic (n = 2), follicular with Hurthle cell type changes (n = 1), and colloid nodule (n = 5). The remainder of the FNA biopsies were deemed indeterminate (n = 4/25) by the reporting pathologist giving a malignancy positive rate of 62% (13/21) within the confirmed results. Overall, 13 malignancies were identified out of the 55 patients with focal uptake who underwent further assessment (24%). Abnormal tracer uptake in the thyroid gland incidentally as a proportion of all 18 F-FDG PET-CT scans was 4.37%. Abnormal focal tracer uptake as a proportion of all 18 F-FDG PET-CT scans was 2.77%. Mean SUV max in focal malignant lesions ranged from 4 to 35.36 g/ml (Table 1) and in focal benign lesions ranged from 1.6 to 18.2 g/ml (Table 2). Mean SUV max value for papillary carcinoma was 8.2 g/ml, and follicular carcinoma was 12.6 g/ml. Median SUV max for papillary carcinoma was 7.4 g/ml, and follicular carcinoma was 8.2 g/ml. Given the small sample size and significant overlap in the obtained SUV max data for the two groups, a threshold with a clinically valuable power cannot be obtained to differentiate benign and malignant tracer uptake and hence is not stated here. Around a quarter of focal thyroid uptake were malignant; hence, focal lesions should be investigated further, if clinically appropriate. A significant proportion of 18 F-FDG PET-CT scans are performed for known/unknown malignancy, and further assessment of incidental thyroid lesions may not be possible for many reasons. Regardless, it is important to detect thyroid lesions in all patients in order to optimise appropriate clinical decision-making. Some of the patients who underwent USS were deemed benign sonographically despite having focal thyroid tracer uptake. It is recommended that all thyroid tracer focal uptake, if clinically appropriate, are investigated further with an ultrasound and FNAC because of the increased risk of malignancy (Bae et al., 2009;Pencharz et al., 2018;Haugen et al., 2015). Currently, we do not have agreed local guidelines for investigation of focal thyroid uptake on 18 F-FDG PET-CT, but based on current recommendations and findings in this study, it would be our future practice to investigate all focal lesions to undergo FNAC where clinically necessary.  Similar number of diffuse and focal thyroid tracer uptake were found in our study, and this is important to differentiate them as they have different outcomes for patients. Abnormal thyroid tracer uptake, diffuse or focal, also appears to be more in females, and this is comparable with current literature (Stangierski et al., 2014).
Our secondary aim was to compare SUV max across thyroid malignancy histological subtypes. Follicular carcinoma has a higher mean, median and range of SUV max than papillary carcinoma in this study. Soelberg et al. in their meta-analysis found mean SUV max to be 6.9 g/ml in malignant lesions (Soelberg et al., 2012), lower than our findings for both follicular and papillary carcinoma but comparable with our median SUVmax . Our sample size for each malignancy subtype is low, and mean SUV max and median SUV max might be affected by this, so should be interpreted cautiously.
Our study found malignancies with a wide range of SUV max values, implying a higher mean SUV max does not necessitate malignancy. It is debatable if SUV max can differentiate between benign and malignant lesions, but overall it is thought that mean SUV max is lower in benign lesions compared to malignant lesions (Soelberg et al., 2012). It is also unclear if a specific SUV max indicates malignancy and certain cutoffs have been stipulated. In our study, we were unable to confidently identify a cutoff point for SUV max to identify malignancy due to considerable overlap in SUV max values in both benign and malignant focal uptake lesions. This is echoed by the results of two smaller studies, which also showed a large overlap in SUV max value between benign and malignant lesions (Brindle et al., 2014;Agrawal et al., 2015). With the findings in this study and other studies, it is difficult to confidently use mean SUV max value alone in differentiating between benign and malignant thyroid lesions. There is a role for SUVmax in thyroid malignancy, but a specific defined value or range cannot be accurately established in differentiating between benign or malignant lesions.
There are a few limitations to our study that we acknowledge. Our biggest limitation was that we excluded nearly a third of patients with focal thyroid tracer uptake who were lost to our follow-up. Most of those patients were referred to us initially from regional centres, and their follow-up records or results were not available for our reference. Despite this, the study still showed the prevalence of incidental thyroid uptake on 18 F-FDG PET-CT. Another limitation of our study is that only a word search for "thyroid" was done in our reports, and images were not reviewed. As described earlier, all our scans are double reported within a pool of 4 experienced consultants, two of whom specialise in thyroid imaging. Most of these patients undergo scans for known malignancy, and a significant proportion of these scans are reviewed again for MDT requirements. Furthermore, there is a dedicated discrepancy meeting to identify reporting errors. We strongly believe, given the above, that the possibility of missing unreported thyroid lesions in these scans to be very low.

Conclusion
Standardised approach needed for investigation of incidental 18 F-FDG PET-CT focal uptake in the thyroid gland due to high prevalence of malignancy and a combination of ultrasound with FNAC is advised.
There is conflicting evidence at how to utilise SUV max in focal thyroid tracer uptake, but its use in combination with ultrasound and histopathological findings should be sought with further bigger studies. We established that SUV max is relatively higher in follicular carcinoma than papillary carcinoma; however, further research in large patient groups is needed. Roddy et al. European Journal of Hybrid Imaging (2020) 4:21 Page 6 of 8