Risk of thyroid cancer in a lung cancer screening population of the National Lung Screening Trial according to the presence of incidental thyroid nodules detected on low-dose chest CT
Article information
Abstract
Purpose
This study evaluated thyroid cancer risk in a lung cancer screening population according to the presence of an incidental thyroid nodule (ITN) detected on low-dose chest computed tomography (LDCT).
Methods
Of 47,837 subjects who underwent LDCT, a lung cancer screening population according to the National Lung Screening Trial results was retrospectively enrolled. The prevalence of ITN on LDCT was calculated, and the ultrasonography (US)/fine-needle aspiration (FNA)–based risk of thyroid cancer according to the presence of ITN on LDCT was compared using the Fisher exact or Student t-test as appropriate.
Results
Of the 2,329 subjects (female:male=44:2,285; mean age, 60.9±4.9 years), the prevalence of ITN on LDCT was 4.8% (111/2,329). The incidence of thyroid cancer was 0.8% (18/2,329, papillary thyroid microcarcinomas [PTMCs]) and was higher in the ITN-positive group than in the ITN-negative group (3.6% [4/111] vs. 0.6% [14/2,218], P=0.009). Among the 2,011 subjects who underwent both LDCT and thyroid US, all risks were higher (P<0.001) in the ITNpositive group than in the ITN-negative group: presence of thyroid nodule on US, 94.1% (95/101) vs. 48.6% (928/1,910); recommendation of FNA according to the American Thyroid Association guideline and Korean Thyroid Imaging Reporting and Data System guideline, 41.2% (42/101) vs. 2.4% (46/1,910) and 39.6% (40/101) vs. 1.9% (37/1,910), respectively.
Conclusion
Despite a higher risk of thyroid cancer in the LDCT ITN-positive group than in the ITN-negative group in a lung cancer screening population, all cancers were PTMCs. A heavy smoking history may not necessitate thorough screening US for thyroid incidentalomas.
Introduction
The incidence of thyroid cancer has substantially increased worldwide over the last 40 years [1,2]. Apart from the increased frequency of screening ultrasonography (US), with the advent of recent technological progress, the increased use of new imaging modalities for other indications has been reported to be closely related to the increased incidence of thyroid cancer [3-5].
An incidental thyroid nodule (ITN) refers to a thyroid nodule that was not previously detected or suspected but was identified by an imaging modality. It is one of the most common incidental findings of neck imaging studies [6]. In principle, the American Thyroid Association (ATA) recommends diagnostic thyroid/neck US for all patients with a suspected ITN that was detected on computed tomography (CT), magnetic resonance imaging (MRI), or fluorodeoxyglucose-positron emission tomography [7]. However, in real-world clinical practice, various concerns have been raised with regard to whether and how radiologists should report ITNs detected on these imaging modalities performed for unrelated conditions, considering debates on whether reporting ITNs is beneficial or harmful [8-11]. In this regard, the American College of Radiology (ACR) suggested consensus recommendations for ITNs based on clinicoradiologic findings [6].
Meanwhile, based on the results of the National Lung Screening Trial (NLST), low-dose chest CT (LDCT) has been widely recommended for populations that require lung cancer screening [12,13]. The increased use of LDCT has resulted in increased detection of ITNs, which has subsequently made the issues related to reporting these ITNs more pertinent than before [14]. Some papers have reported the prevalence and malignancy rates of ITNs on chest CT scans with different CT protocols and various populations [15-17]. However, none of them were based on a complete evaluation of US, which is a fundamental step in the analysis of thyroid cancer risk. Moreover, no previous studies have compared the risk for thyroid cancer according to the presence of ITN on LDCT.
Therefore, this study aimed to evaluate the risk of thyroid cancer in a lung cancer screening population according to the presence of ITN detected on LDCT.
Materials and Methods
Compliance with Ethical Standards
This study was approved by the Institutional Review Board of the Seoul National University Hospital (H-1309-026-517). The requirement for informed consent was waived due to the retrospective nature of the study.
Subjects
In total, 47,837 subjects underwent LDCT for screening purposes from January 2004 to May 2013 at Gangnam Center of Seoul National University Hospital Healthcare System (Seoul, South Korea) in South Korea. A lung cancer screening population for this study was constructed by including the following subjects who were recommended to undergo LDCT for lung cancer screening according to the NLST results: (1) subjects aged between 55 and 74 years old at the time of LDCT; and (2) subjects who were current or former (cessation within 15 years) smokers and had a smoking history of greater than 30 pack-years [12]. Thereafter, the following subjects were excluded: (1) subjects who had a history of lung cancer; (2) subjects who had undergone partial or total thyroidectomy due to thyroid nodules or thyroid cancer; and (3) subjects who had previously been diagnosed with thyroid nodules. Among them, subsequently, a search was conducted for subjects who underwent thyroid US for screening purposes on the same day or within 6 months after the LDCT acquisition to evaluate the US-based risk of thyroid cancer. For subjects with repeated LDCT examinations during the study period, we included data from the first LDCT or the first LDCT with thyroid US. Fig, 1 presents a flow chart of the patients enrolled.
Flowchart of patient enrollment.
LDCT, low-dose chest computed tomography; ITN, incidental thyroid nodule; US, ultrasonography; FNA, fine-needle aspiration.
Low-Dose Chest CT
LDCT was performed with the subject in a supine position and the arms elevated above the head during end-inspiratory breath-holding using a 16-slice multi-detector CT scanner (SOMATOM Sensation 16, Siemens Medical Systems, Munich, Germany) or a 256-slice multi-detector CT scanner (Brilliance iCT, Philips Healthcare, Cleveland, OH, USA). CT images from the lower neck to the upper abdomen were obtained to completely include the lung apex and bases. The detailed scanning parameters were as follows: detector collimation, 0.6-0.75 mm; beam pitch, 0.516-1.2; reconstruction increment, 1.0 mm; rotation time, 0.5 seconds; tube voltage, 120 kVp; tube current, 30-60 effective mAs; and matrix, 512×512. Contrast enhancement was not performed. The axial images were reconstructed with a slice thickness of 3 or 5 mm using standard and lung algorithms and presented in a picture archiving and communication system (Marosis, Infinitt Healthcare, Seoul, Korea). Thyroid shields were not used during the study period. Board-certified radiologists affiliated with our center described the lung nodules, masses, or other abnormalities in the radiologic reports.
Thyroid Ultrasonography and Radiologic Reports
Using high-resolution US machines equipped with 8-14 MHz linear-array transducers, thyroid US examinations were performed by board-certified radiologists who were affiliated with the same center. The scanning protocol in all cases included both transverse and longitudinal real-time imaging of the thyroid nodules, and representative images of the thyroid nodules were stored in the picture archiving and communication system. The radiologists described the character, size, and number of thyroid nodules on the US examination in their radiologic reports, referring to the consensus recommendations and guidelines of those times [18-21].
Analysis of LDCT and Thyroid US for Thyroid Nodules
Blinded to the US results, two radiologists (K.M.K., with 10 years of experience in head and neck radiology and K.N.J., with 15 years of experience in thoracic radiology) retrospectively and independently reviewed the LDCT images with regard to whether the whole thyroid gland was included in the LDCT scan and also for the presence of thyroid nodules under the soft tissue window setting offered by the picture archiving and communication system without window and level adjustment. CT images were displayed in a 1×1 window size of the monitor specified for radiologists.
The presence of a thyroid nodule on LDCT was defined as the observation of any firmly depicted focal feature, such as distinct attenuation, mass effect, or calcification, in the thyroid gland. The type of calcification of the nodule was classified as punctate, rim, or coarse (Fig. 2). The size of the largest nodule was measured. The thyroid nodules on US were retrospectively and independently assessed by two radiologists (H.S., with 13 years of experience in head and neck radiology and J.S.P., with 13 years of experience in head and neck radiology), blinded to the cytopathology reports and LDCT results; they referred to the radiologic and medical reports with regard to the following: (1) the presence of thyroid nodules; (2) the size of the largest thyroid nodule; (3) whether and according to which criteria of the ATA guidelines or Korean Thyroid Imaging Reporting and Data System (K-TIRADS) guidelines were the thyroid nodules indicated for fine-needle aspiration (FNA); and (4) whether FNA or core-needle biopsy (CNB) for the thyroid nodule was performed. Finally, they categorized the results on FNA cytology, CNB pathology, and surgical pathology for the thyroid nodule based on the recent Bethesda System for Reporting Thyroid Cytopathology and the American Joint Committee on Cancer eighth edition (Fig. 3) [7,22-25]. In cases of disagreement between the LDCT and US interpretations, the final decision was made by another radiologist (J.H.K., with 19 years of experience in head and neck radiology). Referring to the radiologic reports and medical records, clinical characteristics including smoking history, cytopathologic results of the thyroid nodule, presence of a lung nodule, and the pathology of the lung nodule were reviewed by a radiologist (H.S.). Participants were followed-up for events that occurred through September 31, 2020.
Types of incidental thyroid nodule calcification (arrows) on low-dose chest computed tomography: punctate (A), rim (B), and coarse (C).
A 71-year-old man (34 packyear smoking history) with an incidental thyroid nodule on lowdose chest computed tomography (LDCT) that was proven to be a benign follicular nodule.
A. LDCT shows a 2.0-cm low-attenuating nodule (arrows) in the left hemithyroid. B. Ultrasonography reveals a 1.8-cm, slightly hypoechoic, well-defined, ovoid solid nodule in the left hemithyroid. It was reported to be a benign follicular nodule (Bethesda category 2) based on fine-needle aspiration results.
Statistical Analysis
Descriptive statistics are presented as number and percentage. For the evaluation of ITN on LDCT, the interobserver agreement was calculated. Concordance was then determined based on the Cohen’s unweighted kappa value as follows: slight agreement, 0.01-0.20; fair, 0.21-0.40; moderate, 0.41-0.60; substantial, 0.61-0.80; and almost perfect, 0.81-1.00 [26]. We used the Fisher exact or Student t-test to compare clinical or imaging features between the ITN-positive and ITN-negative groups as appropriate. We employed SPSS version 19.0 for Windows (IBM Corp., Armonk, NY, USA) for the statistical analysis. Statistical significance was determined by a P-value <0.05.
Results
Radiologic and Clinical Characteristics of ITN on LDCT
In total, 2,329 subjects (female:male [F:M]=44:2,285; mean age, 60.9±4.9 years) were enrolled as the NLST lung cancer screening population for the analysis of the presence of ITN on LDCT. The baseline clinical characteristics according to the presence of ITN on LDCT are summarized in Table 1. In the LDCT ITN-positive group, the proportion of women was larger (7.2% vs. 1.6%, P<0.001), the subjects were older (62.4±4.8 years vs. 60.9±4.9 years, P=0.001) and the incidence of thyroid cancer proven after surgery or reported as Bethesda category 6 on FNA/CNB was higher (3.6% vs. 0.6%, P=0.009) than in the LDCT ITN-negative group. There were no significant differences in the prevalence of ITN on LDCT and the incidence of thyroid cancer according to CT scanners (16-slice vs. 256-slice) and scanning periods (2003-2008 years vs. 2009-2013 years), respectively (Supplementary Table 1). All of the proven thyroid cancers in 18 patients (F:M=1:17; mean age, 60.6±2.7 years; 12 former and six current smokers) were papillary thyroid microcarcinomas (PTMCs) of stage I (n=13; T1aN0, n=4; T1aNx, n=9) or II (n=5; T1aNa, n=4; T1aN1b, n=1) based on American Joint Committee on Cancer eighth edition (Table 2) [25]. None of them showed high-risk features [27], except for one case that was proven on FNA to be associated with lymph node metastasis in the lateral compartment. The incidence of lung cancer in this population was 1.2% (28/2,329). Lung cancer did not occur in any subjects in the LDCT ITN-positive group nor in any subject with proven thyroid cancer. LDCT covered the thyroid gland completely in 2,010 of 2,329 subjects (86.3%).
Baseline characteristics according to the presence of an ITN on LDCT
Pathologic or clinical staging of 18 thyroid cancers
Table 3 shows the CT characteristics of LDCT-detected thyroid nodules according to the presence of proven thyroid cancer. In the LDCT image analysis, ITNs on LDCT were noted in 111 of 2,329 (4.8%) subjects, with multiple ITNs in nine of the 111 subjects (8.1%). The mean size of the largest ITN in 111 subjects was 13.3±8.3 mm (≤10 mm, 46/111 [41.4%]; >10 mm and <15 mm, n=26/111 [23.4%]; ≥15 mm, n=39/111 [35.1%]; range, 1.9 to 41.4 mm). Thirty eight out of the 111 subjects (34.2%) showed calcification (punctate, 8/38 [21.1%]; rim, 6/38 [15.8%]; and coarse, 24/38 [63.2%]) (Fig. 2). There was fair to strong interobserver agreement for the presence of ITN, multiplicity of the ITNs, and calcification in the ITNs on LDCT (kappa value, 0.511, 0.336, and 0.702, respectively).
Computed tomography characteristics of the LDCT ITN-positive group according to the presence of proven thyroid cancer
The nodule sizes of four proven thyroid cancer cases out of 111 LDCT-detected thyroid nodules were 4.8 mm, 11.0 mm, 17.0 mm, and 33.0 mm, respectively. The former 4.8-mm calcified ITN was proven on FNA, and the remaining three cases were PTMCs proven surgically. The former 4.8-mm ITN on LDCT was matched with FNA-proven thyroid cancer on thyroid US, and the 11.0-mm and 17.0-mm ITNs on LDCT were matched with suspicious nodules on thyroid US and proven as PTMC surgically (Fig. 4). However, the 33.0-mm ITN on LDCT was not matched with a suspicious nodule on thyroid US (Fig. 5). Compared to the thyroid cancer-negative group, the thyroid cancer-positive group had an isthmic location more frequently (50% vs. 2.8%, P=0.010).
A 60-year-old man (180 pack-year smoking history) with an incidental thyroid nodule (arrows) on low-dose chest computed tomography (LDCT), which was proven to be a papillary thyroid microcarcinoma.
A. LDCT shows an 11-mm low-attenuating nodule in the left hemithyroid. B. Ultrasonography reveals a 13-mm, heterogeneous, ovoid solid nodule (arrows) in the left hemithyroid. It was suspected to be a papillary thyroid carcinoma (Bethesda category 5) based on fine-needle aspiration results. Surgery confirmed the 9-mm papillary thyroid microcarcinoma without extrathyroidal extension.
A 61-year-old man (37 pack-year smoking history) with an incidental thyroid nodule on low-dose chest computed tomography (LDCT), which was proven to be nodular hyperplasia and concurrent papillary thyroid carcinoma.
A. LDCT shows a 33-mm low-attenuating nodule with punctate calcifications (arrows) in the left hemithyroid. B. Ultrasonography reveals a 26-mm ill-defined, isoechoic, predominantly solid nodule (arrows) in the left hemithyroid, which was finally proven to be nodular hyperplasia. C. Unexpectedly, ultrasonography showed another 8-mm low-echoic, suspicious nodule with spiculated margin (arrows) in the left hemithyroid, which was finally confirmed to be a 5-mm papillary thyroid microcarcinoma without extrathyroidal extension during surgery.
US/FNA-Based Risk of Thyroid Cancer According to the Presence of ITN
Thyroid US was optionally performed according to the subjects’ choice. Three hundred eighteen patients did not choose screening thyroid US. For the evaluation of US-based risks of thyroid cancer, we enrolled 2,011 subjects (F:M=44:1,967; mean age, 61.0±4.9 years) who underwent thyroid US on the same day or within 6 months after the LDCT acquisition. In 86.4% (1,737/2,011) of them, LDCT covered the thyroid gland completely. The US/FNA-based characteristics of thyroid nodules according to the presence of ITN on LDCT are summarized in Table 4. Of the 2,011 subjects, 50.9% (1,023/2,011) had ITNs on US. Of 1,023 subjects with ITNs, 85.1% (871/1,023) had small nodules (size ≤1 cm). FNA was recommended in 8.6% (88/1,023) and 7.5% (77/1,023) of cases according to the ATA guidelines and K-TIRADS guidelines, respectively.
Ultrasonography/fine-needle aspiration–based characteristics of thyroid nodules according to the presence of a ITN on LDCT
From a total of 2011 subjects, FNA was recommended in 4.4% (88/2,011) according to the ATA guideline and in 3.8% (77/2,011) according to the K-TIRADS guideline. FNA or CNB was performed in 5.2% (104/2,011). Surgical indications with Bethesda category 4, 5, or 6 on the FNA/CNB results were found in 0.9% (19/2,011) of the subjects. The incidence of surgically proven thyroid cancer was 0.5% (11/2,011). The incidence of surgically proven thyroid cancer or an FNA/CNB Bethesda category 6 result was 0.9% (18/2,011). The incidence of surgically proven thyroid cancer or an FNA/CNB Bethesda category 4, 5, or 6 result was 1.1% (22/2,011).
All US/FNA-based risks were significantly higher in the LDCT ITN-positive group than in the LDCT ITN-negative group. In particular, the size of the thyroid nodules was larger (14.9±8.7 mm vs. 5.5±3.7 mm, P<0.001) and the nodules were more frequently larger than 1 cm (58.4% [59/101] vs. 4.9% [93/1,910], P<0.001) in the LDCT ITN-positive group than in the LDCT ITN-negative group (Fig. 3). In addition, FNA was more frequently recommended (ATA guideline, 41.2% [42/101] vs. 2.4% [46/1,910]; K-TIRADS guideline, 39.6% [40/101] vs. 1.9% [37/1,910]; P<0.001, respectively) and thyroid nodules with Bethesda category 4, 5, or 6 on FNA/CNB results also occurred more frequently (6.9% [7/101] vs. 0.6% [12/1,910], P<0.001).
For the four patients with surgically proven thyroid cancer in the LDCT ITN-positive group, ITNs on LDCT were matched with suspicious nodules on thyroid US in three cases (Fig. 4), but the ITN on LDCT was not matched with a suspicious nodule on thyroid US in the remaining case (Fig. 5).
Discussion
This study showed that in a screening population for lung cancer, the prevalence of ITN on LDCT was 4.8% (111/2,329), which was much lower than the prevalence of ITN on US (50.9% [1,023/2,011]). The majority of the ITNs (64.9%, 72/111) on LDCT were <1.5 cm in size, and most of the ITNs (85.1%, 871/1,023) on US were ≤1 cm. FNA was not recommended according to the ATA guideline for the majority of the subjects (58.4%, 59/101) in the LDCT ITN-positive group, and FNA was not recommended for most subjects with ITN on US (91.4%, 935/1,023). When referring to the ACR consensus recommendations that target only nodules ≥1.5 cm in subjects ≥35 years, a US evaluation would not have been recommended for 64.9% (72/111) of subjects with ITN on LDCT; two out of the four patients with thyroid cancer in the LDCT ITN-positive group would not have undergone US, and the diagnosis would probably have been missed [6]. There were 18 PTMCs in the study population (0.8%, 18/2,329), and the incidence of thyroid cancer was higher in the LDCT ITN-positive group than in the LDCT ITN-negative group. All the US/FNA-based risk factors for thyroid cancers, including the size of the thyroid nodule, recommendation for FNA, and surgical indication based on FNA/CNB results, were present at higher levels or frequencies in the LDCT ITN-positive group than in the ITNnegative group.
This is the first study to evaluate the clinical significance of ITN detected on LDCT, which was performed in a lung cancer screening population for whom LDCT was recommended according to the NLST results. The correlation between LDCT and US for thyroid nodules could be analyzed because the data were collected in 2004-2013 at a health-care center in South Korea. South Korea is reported to have a very high incidence of thyroid cancer, which might be related to the abundant screening examinations [3]. However, even in South Korea, there has been a marked decrease in screening programs for thyroid US, resulting in a decrease in thyroid operations since 2014 due to mass media presentations discouraging screening thyroid US [28,29]. The data in this study are from the period when screening US was the most popular in South Korea.
Whereas the prevalence of ITNs on contrast-enhanced chest CT has reached about 25% [30], the prevalence of ITNs on LDCT in this study was only 4.8%; although this was slightly more frequent than the 2.8% reported by Lee et al. [15] in a study using LDCT that included a study population without consideration of whether LDCT was recommended according to the results of the NLST study. This low prevalence of ITN on LDCT may be related to the lack of intravenous contrast and the lower radiation dose of LDCT compared to that of contrast-enhanced chest CT. The present study reported a malignancy rate of 3.6% (4/111) in subjects with ITN on LDCT in the lung cancer screening population; this rate lies somewhere within the previously reported range of 0%-11% in CT and MRI studies [6,31]. Compared with a study using the NLST population, our study showed a higher incidence of thyroid cancer (0.8% [18/2,329] vs. 0.1% [22/17,309]). In addition, whereas all cancers in our study were PTMCs, all papillary thyroid carcinomas in the previous study were larger than 2 cm [16]. The difference in the sensitivity of detecting ITN on LDCT and the difference in the US examination rate might be important causes of these discrepancies.
The low risk of thyroid cancer in the lung cancer screening population can be attributed to the following factors. First, the lung cancer screening population had a higher proportion of men than women, and men are known to be less susceptible to thyroid cancer than women are. Second, current tobacco smoking might have an inverse association with thyroid cancer risk [1,32]. Even when ITNs were malignant, they tended to be indolent. Most of the malignant ITNs on LDCT in this study were early-stage PTMCs, which was consistent with the previous findings that most ITNs on thyroid US were small papillary cancers [33]. Some doctors have argued that "incidental" and "non-incidental" thyroid PTMCs are different entities [34]. Many papers have recently reported that indiscriminate US evaluations for ITN due to the physician’s concern about the risk of malignancy was potentially harmful because the risk for poor outcomes without an evaluation was minimal and an intensive work-up of the ITN tended to result in unwanted benign thyroid nodule operations [35,36]. In this regard, we think that US for thyroid nodules may not be necessary for this lung cancer screening population, especially for the ITN-negative group. In the ITN-positive group, there was no specific CT feature for malignancy due to the limited number of thyroid cancer cases, except that an isthmic location was more frequent in thyroid cancer-positive group than in the thyroid cancer-negative group. Even for the ITN-positive group, a US evaluation may not be generally recommended because the thyroid cancers detected were mostly early-stage PTMCs. However, a US evaluation may be considered for subjects with other clinical risk factors for thyroid cancer or a personal concern about thyroid cancer. In the absence of strict criteria for this population, we advocate for the adoption of the ACR consensus recommendations that have a triage system for ITNs detected on various imaging modalities, rather than the ATA guidelines that do not [6,7].
There were some limitations to this study. First, since this was a retrospective study using data from 2004 to 2013, it was not possible to analyze the various clinical risk factors of thyroid cancer, such as a family history of thyroid cancer and a history of head and neck irradiation, because of the lack of information. However, as mentioned, the authors suggest that this analysis of the correlation between LDCT and thyroid US may not be possible with a prospective study at present. Second, during the 10-year study period, changes in the consensus recommendations and guidelines caused some inconsistencies in the US evaluation of thyroid nodules. However, it is unlikely that there were any serious changes in the lexicon that would have caused significant differences in the management of thyroid nodules. Third, this study did not compare correlations of the nodule size with findings between LDCT and thyroid US. The thyroid nodules detected on LDCT might not have corresponded to the nodules identified on US or biopsied, as presented in Fig. 5. Finally, this study was conducted at a single health center in an Asian iodine-rich area. The results should be generalized through studies with various populations.
In conclusion, in a lung cancer screening population, the risk of thyroid cancer was higher in the LDCT ITN-positive group than that in the ITN-negative group. However, all proven cancers were early-stage PTMCs.
Notes
Author Contributions
Conceptualization: Seo H, Kim JH. Data acquisition: Seo H, Jin KN, Park JS, Kang KM, Lee EK, Kim JH. Data analysis or interpretation: Seo H, Jin KN, Park JS, Kang KM, Lee EK, Lee JY, Yoo RE, Kim JH. Drafting of the manuscript: Seo H, Jin KN, Park JS, Kang KM, Lee EK, Kim JH. Critical revision of the manuscript: Lee JY, Yoo RE, Kim JH. Approval of the final version of the manuscript: all authors.
Ji-hoon Kim serves as Editor for the Ultrasonography, but has no role in the decision to publish this article. All remaining authors have declared no conflicts of interest.
Acknowledgements
We thank Jungju Lee for her assistance in data collection.
Supplementary Material
Supplementary Table 1. Incidence of ITN on LDCT and the incidence of thyroid cancer according to CT scanners and scanning periods (https://doi.org/10.14366/usg.22111).
Supplementary Table 1.
Incidence of ITN on LDCT and the incidence of thyroid cancer according to CT scanners and scanning periods
References
Article information Continued
Notes
Key point
In lung cancer screening, although the presence of an incidental thyroid nodule on low-dose chest computed tomography increased thyroid cancer risk, only papillary thyroid microcarcinomas were identified. A heavy smoking history may not necessitate thorough screening ultrasonography for thyroid incidentalomas in a lung cancer screening population.