- Email: [email protected]
Thyroid Lesions Visualized on CT
Original Investigations
Thyroid Lesions Visualized on CT: Sonographic and Pathologic Correlation Christopher Lee, MD, Brenna Chalmers, MD, Daniel Treister, BS, Sumita Adhya, MD, Benjamin Godwin, MD, Lingyun Ji, MS, Susan Groshen, PhD, Edward Grant, MD Rationale and Objectives: There are no definitive guidelines for the management of incidental thyroid lesions on computed tomography (CT). The objectives of our study were to assess the association between CT and ultrasound (US) characteristics of thyroid lesions and identify CT predictors of benignity or malignancy. Materials and Methods: Two hundred fifty-nine patients who had undergone both CT of the chest or neck and thyroid US had at least one thyroid lesion visible on CT; both incidentally detected and palpable or symptomatic lesions were included. The CT and US examinations were retrospectively reviewed and lesions characterized. Pathologic results from fine-needle aspiration (FNA) and surgical excision were used to classify lesions as benign or malignant. Thyroid lesions without pathologic correlation were classified as either benign or indeterminate based on US evaluation. Lesions that were cystic, predominantly cystic, solid and cystic with features consistent with a colloid nodule, or stable for at least 1 year were considered benign. Fisher exact tests, Spearman correlations, and logistic regression models were used to evaluate the associations between CT and US characteristics and CT predictors of benignity or malignancy. Multivariable logistic regression analyses were performed to examine the joint associations between the CT features and sonographically assessed benignity. Results: Of 259 patients, 168 (64.9%) patients had incidentally detected thyroid lesions, 49 (18.9%) patients had palpable/symptomatic lesions, and 116 (44.8%) patients underwent surgical resection and/or FNA biopsy. The malignancy prevalence was 1.8% (3 of 168) for incidental thyroid lesions and 10.2% (5 of 49) for palpable/symptomatic lesions. The malignancy prevalence of incidental lesions initially detected on CT was 1.6% (2 of 125). Of the 143 patients without pathologic data, 58 (40.6%) were classified as benign and 85 (59.4%) were categorized as indeterminate based on US evaluation. Statistically significant associations were found between CT and US with regard to lesion number, dominant lesion size, lesion consistency/composition, and associated calcifications. No CT characteristics of thyroid lesions predicted malignancy. However, there were statistically significant associations on multivariate analysis between indeterminate/benign nodules and CT characteristics of smaller lesion size, lower mean attenuation, and homogeneous composition. Conclusions: Recommending sonographic evaluation of all incidentally detected thyroid lesions is likely not the appropriate strategy, given the high prevalence of thyroid incidentalomas, low probability of malignancy, and cost effectiveness of workup. Small, homogeneous, low-attenuation lesions have a high probability of being benign. Key Words: Computed tomography; ultrasound; incidental lesion; thyroid nodule; thyroid malignancy. ªAUR, 2014
I
ncidental lesions of the thyroid are commonly encountered in computed tomography (CT) imaging of the neck and chest. Two separate studies reported a prevalence of up to 17% (1,2). There is considerable variability in the manner of reporting incidentally detected thyroid lesions, including which follow-up recommendations are provided (3,4). This variability is largely due to the fact that there are no established guidelines for the management of incidental thyroid lesions discovered on CT.
Acad Radiol 2014; -:1–7 From the Department of Radiology, Keck School of Medicine of USC, 1500 San Pablo St, 2nd Floor Imaging, Los Angeles, CA 90033 (C.L., B.C., D.T., S.A., E.G.); Department of Radiology, University of Washington School of Medicine, Seattle, WA (B.G.); and Department of Preventive Medicine, Keck School of Medicine of USC, Los Angeles, CA (L.J., S.G.). Received April 8, 2014; accepted August 1, 2014. Address correspondence to: C.L. e-mail: [email protected] ªAUR, 2014 http://dx.doi.org/10.1016/j.acra.2014.08.007
The reported malignancy prevalence of incidental thyroid nodules detected on CT is not insignificant, ranging from 9% to 11% (1,5). However, papillary carcinoma, which accounts for nearly 90% of thyroid cancer cases, has an excellent prognosis (6). In fact, despite increased diagnosis and resection of papillary cancers over the past several decades, there has been essentially no improvement in cancer-specific mortality (6,7). Micropapillary carcinomas measuring <10 mm in diameter have an even better prognosis, with one study reporting no cancer-related deaths during a 10-year surveillance period (8). Thus, although incidental thyroid lesions may harbor malignancy, workup of all thyroid incidentalomas, particularly when small, may result in overdiagnosis and overtreatment of indolent tumors. Given the increasing utilization of thin-collimation multidetector CT, high prevalence of incidental thyroid lesions, and relatively low rate of life-threatening malignancy of these nodules, the potential for unnecessary and costly follow-up is apparent. Although the role of 1
Academic Radiology, Vol -, No -, - 2014
LEE ET AL
sonography to guide further management of thyroid lesions is relatively well established (9,10), the role of CT in prompting sonographic evaluation is still controversial (11). The objectives of our study, therefore, were to assess the association between CT and sonographic characteristics of thyroid lesions, including both incidental and palpable/ symptomatic lesions, and identify CT predictors of thyroid benignity or malignancy. MATERIALS AND METHODS This retrospective Health Insurance Portability and Accountability Act–compliant study was approved by our institutional review board, with waiver of informed consent. A total of 2904 patients underwent thyroid US between July 2004 and July 2011. In the interest of including as many lesions as possible, all clinical indications were included, including incidentally detected and palpable or symptomatic lesions. Of these patients, 419 had also received a CT of the chest or neck. The CT and US examinations of these patients were retrospectively evaluated by a blinded radiologist (6 years of experience) on a digital workstation (Synapse; Fujifilm USA, Cypress, CA), with consensus review with a second radiologist (10 years of experience) performed for equivocal cases. If patients received multiple US and CT studies, the examinations closest to each other temporally were analyzed, with a median time interval of 2.5 months (range, 0–52 months). Patients were included regardless of whether the US preceded or followed the CT. Of the 419 patients, 259 had at least one thyroid lesion visible on CT; these subjects comprised the final study population. Patient demographics are shown in Table 1. CT scans were acquired on various multidetector scanners with a minimum of 10 channels; reconstructed image thickness ranged from 1 to 5 mm. One hundred sixty-three non– contrast-enhanced and 256 contrast-enhanced CTs were included. For contrast-enhanced studies, 80–100 mL of nonionic iodinated contrast (Isovue 370; Bracco Diagnostics, Princeton, NJ) was intravenously administered using a power injector at a rate of 2–4 mL/s. US examinations were performed using an Acuson Sequoia 512 machine with a broadband 8–15 MHz linear-array transducer (Siemens, Mountain View, CA) or Toshiba Aplio XG or MG units with a 7.4–14 MHz linear-array transducer (Toshiba America, Tustin, CA). Thyroid lesions on CTwere characterized based on number of lesions (1 to $5), size of the largest/dominant lesion (maximum axial dimension), and the presence of associated calcifications. Calcifications were classified as coarse, peripheral, or punctate. Coarse calcifications were defined as nonperipheral calcifications measuring at least 3 mm in size, whereas punctate calcifications were defined as calcifications measuring <3 mm. Additionally, lesion consistency was characterized as homogeneous versus heterogeneous, and mean attenuation values of the lesions were determined by placing a region of interest within the lesion (at least half of the area, excluding calcifications when possible). When 2
TABLE 1. Patient Demographics Characteristic Number of females Mean age, years Thyroid lesion presentation Incidental Initially detected on CT Initially detected on US Initially detected on MR Initially detected on PET Palpable/symptomatic Partial thyroidectomy follow-up Unknown Not seen on ultrasound
Data (n = 259) 171 (66.0) 61 14 (range, 18–96) 168 (64.9) 125 (74.4) 34 (20.2) 4 (2.4) 5 (3.0) 49 (18.9) 2 (0.8) 38 (14.7) 2 (0.8)
CT, computed tomography; MR, magnetic resonance; PET, positron emission tomography; US, ultrasound.
multiple thyroid lesions were identified, only the largest was evaluated. If a subsequent biopsy of a smaller lesion was performed, however, the reviewer was instructed to characterize that lesion instead. Thyroid lesions on US were likewise described according to number of lesions, size of the largest lesion, lesion composition, and the presence of associated calcifications. Lesion composition was classified as solid, cystic, or mixed solid/ cystic. Calcifications were categorized as coarse, peripheral, microcalcifications, and echogenic foci with associated comet-tail artifacts. Coarse calcifications measured at least 3 mm in size, whereas microcalcifications measured less than 3 mm. Only the largest of multiple lesions was described, while ensuring that the lesion was in the same lobe and approximate region as the lesion on US. Once again, however, if a biopsy of a smaller lesion was performed, that lesion was characterized in its place. Thyroid fine-needle aspiration (FNA) results were categorized according to the Bethesda classification (12). Bethesda II and III categories were considered benign, whereas Bethesda IV, V, and VI categories were considered malignant. If surgical excision was performed, the surgical pathology superseded the FNA biopsy and was used to classify a lesion as benign or malignant. Thyroid lesions without pathologic correlation were classified as either benign or indeterminate based on US evaluation. Lesions that were cystic, predominantly cystic, solid and cystic with features consistent with a colloid nodule (anechoic or hypoechoic with internal echogenic foci demonstrating comet-tail artifacts) (13), or stable on follow-up US for at least 1 year were considered benign. All other lesions, including those which were entirely or predominantly solid or contained calcifications, were considered indeterminate. For purposes of statistical analysis, lesions which underwent FNA or surgical resection were also classified as indeterminate. The associations between CT and US characteristics were examined using Fisher exact tests, Spearman correlations, or logistic regression models, as appropriate. The same analysis
Academic Radiology, Vol -, No -, - 2014
methods were used to examine the association between benignity (based on FNA/surgery or US) and the number, size, mean attenuation of lesions, the presence of calcifications, and homogeneous versus heterogeneous consistency on CT. Multivariable logistic regression analyses were performed to examine the joint associations between the CT features and sonographically assessed benignity. A P value of #.05 was considered statistically significant, and a P value of >.05 but #.10 was considered marginally significant; all reported P values were two sided. Statistical analyses were performed with STATA software version 11.0 (StataCorp LP, College Station, TX).
RESULTS Of 259 patients, 257 (99.2%) with thyroid lesions visualized on CT had lesions detected on US; two (0.8%) patients with thyroid lesions on CT had no lesions on US. Of the 259 patients, 19 (7.3%) underwent surgical resection of the thyroid lesion, whereas 97 patients (37.5%) underwent FNA biopsy only. Altogether, there were 9 malignant, 100 benign, and 7 nondiagnostic results (Table 2). Thus, the overall malignancy prevalence in our patient population was 3.5% (9 of 259). Of the 143 patients without pathologic data, 58 (40.6%) were classified as benign and 85 (59.4%) were categorized as indeterminate based on US evaluation. Of 168 patients, 70 (41.7%) patients with incidentally detected thyroid lesions underwent surgical resection and/or FNA biopsy, with 3 (4.3%) malignant, 62 (88.6%) benign, and 5 (7.1%) nondiagnostic results. Two of the incidentally detected malignancies were initially detected on CT, and the third was initially detected on positron emission tomography. Of 49 patients, 26 (53.1%) patients with palpable/ symptomatic lesions underwent resection and/or FNA, with 5 (19.2%) malignant, 21 (80.8%) benign, and 2 (7.7%) nondiagnostic results. One papillary carcinoma was discovered on surveillance US of a patient who had previously undergone a hemithyroidectomy for thyroid cancer. Consequently, the malignancy prevalence was 1.8% (3 of 168) for incidental thyroid lesions and 10.2% (5 of 49) for palpable/symptomatic lesions. The malignancy prevalence of incidental lesions initially detected on CT was 1.6% (2 of 125).
THYROID LESIONS ON CT
TABLE 2. Pathologic Results of Thyroid Lesions Result (n = 116) Incidental lesion FNA biopsy (n = 61) Nondiagnostic (Bethesda I) Benign (Bethesda II) Follicular lesion (Bethesda III) Follicular neoplasm (Bethesda IV) Surgical excision (n = 9) Multinodular goiter Follicular hyperplasia Follicular adenoma Papillary carcinoma Palpable/symptomatic lesion FNA biopsy (n = 20) Nondiagnostic (Bethesda I) Benign (Bethesda II) Surgical excision (n = 8) Colloid nodule Multinodular goiter Papillary carcinoma Medullary carcinoma Anaplastic carcinoma Other lesion FNA biopsy (n = 16) Benign (Bethesda II) Surgical excision (n = 2) Colloid nodule Papillary carcinoma
Number (%)
5 (8.2) 53 (86.9) 2 (3.3) 1 (1.6) 3 (33.3) 2 (22.2) 2 (22.2) 2 (22.2)
2 (10.0) 18 (90.0) 1 (12.5) 2 (25.0) 3 (37.5) 1 (12.5) 1 (12.5)
16 (100.0) 1 (50.0) 1 (50.0)
FNA, fine-needle aspiration. Bethesda II and III lesions were considered benign, whereas Bethesda IV lesion was considered malignant.
er, no association was observed between CT lesion attenuation and US composition. Finally, a significant association was found between associated calcifications on CT versus US (P < .001). Specifically, coarse calcifications on CT correlated with the presence of coarse calcifications on US (P < .001). However, there was a poor correlation of punctate calcifications on CT with microcalcifications or echogenic foci on US (P = .46). A comparison of CT features of benign and malignant thyroid lesions is presented in Table 4. Individual CT characteristics are addressed in the following sections.
Agreement Between CT and US Characteristics
Characteristics of thyroid lesions on CT are summarized in Table 3. Assessment of agreement between CT and US characteristics demonstrated significant correlations between lesion number on CT versus US (r = 0.40, P < .001) and dominant lesion size on CT versus US (r = 0.85, P < .001). There was also a significant association between lesion consistency on CT and solid, cystic, or mixed composition on US (P = .005); 21 of 35 cystic lesions on US appeared homogeneous on CT, whereas mixed and solid lesions were more commonly heterogeneous on CT. Howev-
Number of Lesions on CT
Of 259 patients, 117 (45.2%) patients had a single lesion on CT, 44 patients had two lesions, 22 patients had three lesions, 8 patients had four lesions, and 68 patients had five or more lesions. Five of nine (55.6%) malignant nodules occurred in patients with three or more thyroid lesions. There was no statistically significant association between the number of lesions seen on CT and the proportion of benign or malignant FNA/ surgical diagnoses. However, there was a significant inverse relationship between the number of lesions on CT and 3
Academic Radiology, Vol -, No -, - 2014
LEE ET AL
TABLE 3. Characteristics of Thyroid Lesions on CT CT Characteristic (n = 259 Patients)
Number (%)
Number of lesions 1 117 (45.2) 2 44 (17.0) 3 22 (8.5) 4 8 (3.1) $5 68 (26.3) Size of dominant lesion (mm) 1–4 19 (7.3) 5–9 69 (26.6) 10–19 95 (36.7) 20–100 76 (29.3) Consistency Heterogeneous 165 (63.7) Homogeneous 94 (36.3) Mean attenuation with contrast (HU, n = 174) <50 29 (16.7) 50–99 48 (27.6) 100–199 83 (47.7) >200 14 (8.0) Mean attenuation without contrast (HU, n = 85) <50 42 (49.4) 50–99 38 (44.7) 100–199 4 (4.7) >200 1 (1.2) Associated calcifications* Absent 203 (78.4) Present 56 (21.6) Punctate 10 (25.7) Coarse 39 (15.2) Peripheral 17 (6.6) CT, computed tomography; HU, Hounsfield units. *Some lesions had more than one type of calcification.
proportion of benign cases as assessed by US (P = .002). Patients with more lesions on CTwere more likely to have indeterminate sonographic assessments of the largest/dominant lesion. Lesion Size on CT
The mean maximum diameter of the dominant thyroid lesions on CT was 16 12 mm (range, 2–64 mm). The mean diameter of malignant nodules was 26 19 mm (range, 7–53 mm), whereas the mean diameter of pathologically proven benign nodules was 20 11 mm (range, 5–64 mm), a difference that was not statistically significant (P = .19). In addition, there was no statistically significant association between dominant lesion size on CT and proportion of benign or malignant FNA/surgical diagnoses. However, there was a significant inverse relationship between lesion size and benignity as determined with US (P < .001). Increasing lesion size on CTwas associated with a decreasing rate of sonographically assessed benignity. 4
Lesion Consistency on CT
The characterized lesions on CT were heterogeneous in 165 patients (63.7%) and homogeneous in 94 cases (36.3%). Malignant nodules were, similarly, slightly more commonly heterogeneous. Once again, there was no statistically significant association between lesion consistency on CT and pathologic benignity or malignancy. However, there was a significant association between lesion consistency and proportion of benign cases as assessed with US (P < .001). Homogeneous lesions on CT were more often associated with benign sonographic evaluations. Lesion Attenuation on CT
The average CT attenuation of thyroid lesions was 93 77 Hounsfield units (HU; range, 11–1140 HU). For patients with noncontrast studies, the average lesion attenuation was 65 105 HU, and for those with contrast-enhanced studies, the average attenuation was 107 54 HU. There was no statistically significant association between mean lesion attenuation and proportion of benign or malignant FNA/surgical diagnoses. There was, however, a significant inverse relationship between lesion attenuation on contrast-enhanced studies and sonographic benignity (P = .011). As the mean attenuation increased, the proportion of benign cases as determined with US decreased. There was no statistically significant association between CT attenuation on noncontrast scans and benignity as assessed by US, although a slight trend was observed between increasing lesion attenuation and indeterminate assessments. Lesion Calcifications on CT
Of 259 patients, 56 (21.6%) patients demonstrated thyroid lesions with associated calcifications visible on CT. Only one of nine (11.1%) malignant nodules exhibited calcifications on CT. Once again, there was no statistically significant association between the presence of calcifications and pathologic benignity or malignancy. However, there was a marginally significant inverse relationship between associated calcifications on CT and proportion of benign cases per sonographic evaluation (P = .068). Accompanying calcifications were more often associated with indeterminate sonographic assessments. Further analyses of specific types of calcifications on CT did not yield any improvements in statistical significance. Multivariable logistic regression analyses revealed that dominant lesion size, mean attenuation, and consistency were independently associated with the proportion of benign cases as determined with US (P < .001 for lesion size, P = .044 for mean attenuation, and P = .005 for consistency). After adjusting for these three variables, no significant association was found between sonographic benignity and number of lesions or the presence of associated calcifications.
Academic Radiology, Vol -, No -, - 2014
THYROID LESIONS ON CT
TABLE 4. Comparison of CT Features of Benign Versus Malignant Thyroid Lesions
Parameter Age, years Sex Female Male Number of lesions 1 2 3 4 $5 Size of lesion (mm) 1–4 5–9 10–19 20–100 Lesion consistency Heterogeneous Homogeneous Mean lesion attenuation with contrast (HU) <50 50–99 100–199 >200 Mean lesion attenuation without contrast (HU) <50 50–99 100–199 >200 Associated calcifications Present Coarse and peripheral Coarse Punctate Peripheral
Malignant (n = 9)
Benign (n = 100)
65.0
62.1
6 (66.7) 3 (33.3)
67 (67.0) 33 (33.0)
4 (44.4) 0 2 (22.2) 0 3 (30.3)
36 (36.0) 19 (19.0) 8 (8.0) 5 (5.0) 32 (32.0)
P Value .53* .62y
.99z
.62z 0 4 (44.4) 0 5 (55.6)
0 17 (17.0) 42 (42.0) 41 (41.0)
6 (66.6) 3 (33.3) n=7
78 (78.0) 22 (22.0) n = 70
2 (28.6) 2 (28.6) 3 (42.9) 0 n=2
7 (10.0) 19 (27.1) 36 (51.4) 8 (11.4) n = 30
1 (50.0) 1 (50.0) 0 0 n=9 1 (11.1) 1 (11.1) 0 0 0
16 (53.3) 13 (43.3) 1 (3.3) 0 n = 100 18 (18.0) 1 (1.0) 11 (11.0) 4 (4.0) 2 (2.0)
.43y
.14z
.99z
.51y
CT, computed tomography; HU, Hounsfield units. *Wilcoxon rank–sum test. y Fisher exact test. z Trend test.
DISCUSSION Incidental thyroid lesions are commonly encountered on routine CT scans of the chest or neck; however, no definitive guidelines exist for the reporting and workup of these incidentalomas. The ability to identify benign or otherwise low-risk findings on CT would help prevent unnecessary workup. Therefore, we sought to describe the CT features of thyroid lesions, compare them with sonographic characteristics, and identify CT predictors, if any, of thyroid benignity or malignancy. Our study demonstrated a good agreement between CT and US with regard to number of lesions, dominant lesion size, lesion consistency/composition, and the
Figure 1. A 40-year-old man with palpable left neck mass. Contrast-enhanced axial computed tomography image (a) showed a homogeneous 7-mm nodule (arrow) within the left thyroid lobe. A large lymph node (arrowheads) was also present adjacent to the left internal jugular vein. Corresponding ultrasound image 8 days later (b) demonstrated a 17-mm cystic and solid nodule (open arrows) with microcalcifications (arrowheads). Surgical excision revealed medullary carcinoma with extrathyroidal and nodal spread.
presence of associated calcifications. In addition, we observed significant associations between lesion number, size, consistency, mean attenuation, and associated calcifications on CT and benignity as determined by US. Despite good correlation of lesion size between CT and US, we observed consistent underestimation of size on CT compared to US (Fig 1), which has been confirmed by other investigators and attributed to a variety of technical and anatomic factors (5,11). Interestingly, only coarse calcifications demonstrated good correlation between CT and US. On the other hand, punctate calcifications on CT were most commonly associated with no visible calcifications on US, whereas microcalcifications on US—a finding suggestive of malignancy (9,14)—were most commonly associated with no associated calcifications on CT (Fig 2). This may be partly due to the sonographers’ inability to prospectively identify tiny calcifications associated with a lesion. Additionally, microcalcifications on US are likely too small to be resolved on CT. As stated previously, we detected significant associations between lesion number, size, consistency, mean attenuation, and associated calcifications on CT and benignity as determined by US. Multivariable logistic regression analysis further showed that smaller lesion size, lower mean attenuation, and homogeneous consistency were significantly associated with 5
LEE ET AL
Figure 2. A 57-year-old woman with incidental thyroid nodule on CT. Contrast-enhanced axial CT image (a) demonstrated an ill-defined heterogeneous nodule without obvious calcifications (arrows) within the right thyroid lobe. Corresponding ultrasound image (b) showed a cystic and solid nodule with multiple echogenic foci/microcalcifications, with occasional comet-tail artifacts (arrowhead) suggesting colloid. Fine-needle aspiration biopsy revealed a colloid nodule compatible with Bethesda II lesion. CT, computed tomography.
sonographically assessed benignity. Unfortunately, because of considerable overlap in size and attenuation of benign and malignant lesions, useful threshold values could not be determined from our data. There was no significant association between the various CT features and benignity or malignancy based on surgical resection and/or FNA. This may be due to the small number of documented malignancies, which may have underpowered the analyses. Our results are somewhat concordant with previously published studies. Shetty et al. reported that benign thyroid lesions were smaller than malignant or potentially malignant lesions (5). Similarly, we demonstrated that smaller lesions on CT were more likely to have benign sonographic assessments. Yoon et al. (2) observed that malignant lesions more frequently demonstrated calcifications and mean attenuation values >130 HU on contrast-enhanced CT. In our study population, calcifications and increased attenuation on CT were associated with indeterminate sonographic evaluations. On the other hand, six of seven pathologically proven malignancies on contrast-enhanced CT demonstrated a mean attenuation of <130 HU. This discrepancy may be due to variations in contrast concentration, volume, and scanning delays between the two patient populations. 6
Academic Radiology, Vol -, No -, - 2014
In our patient population, the malignancy prevalence of patients with thyroid lesions visualized on CT was 3.5%. When only including patients with incidental thyroid lesions on CT, the malignancy prevalence dropped to 1.6%. This rate is less than that reported by two prior publications which also studied incidental thyroid lesions on CT. Shetty et al. (5) reported an 11.3% prevalence of malignant or potentially malignant lesions; however, only 3.9% represented definite malignancies, a prevalence closer to ours. Yoon et al. (2), on the other hand, observed a 9.4% malignancy prevalence, but their study population included patients with palpable or symptomatic lesions. Therefore, it is likely that the malignancy prevalence of incidentally detected lesions is closer to 2%. The true malignancy prevalence may be much lower, as sonographic workup of a CT incidentaloma likely selects for lesions that have a higher probability of being malignant. Three of the nine malignant lesions in our study were papillary carcinomas that measured <10 mm on CT. Furthermore, five of the malignant lesions occurred in patients with multinodular glands. These observations underscore the fact that small thyroid lesions or multinodular glands may harbor a significant percentage of thyroid malignancies, a notion that has been asserted by previous authors (9,15). Of course, small cancers, particularly when demonstrating papillary histology, have a high probability of representing indolent tumors (8). Thus, the low prevalence of malignancy and relatively high proportion of small indolent neoplasms combine to create the management dilemma of working up incidental thyroid lesions at high expense but conceivably little impact on cancer-specific mortality. Recently, a three-tiered system for workup of incidental thyroid nodules encountered on cross-sectional imaging was proposed, recommending sonographic evaluation of thyroid nodules with evidence of local invasion or nodal spread, nodules of any size in patients <35 years of age, and nodules $1.5 cm in those $35 years (16,17). The three-tiered system would have significantly reduced benign biopsies in our cohort, because the mean age was 61 years and more than a third of the lesions in our study were <1 cm. However, three papillary carcinomas which manifested as subcentimeter nodules on CT would have been missed, with one of the lesions representing an incidentally detected papillary carcinoma with extrathyroidal spread (Fig 3). All three tumors measured at least 10 mm on sonography; CT underestimated the size of the malignant nodule by more than one-third. Of interest, if we were to also recommend follow-up US of all heterogeneous lesions, regardless of size, we would have only missed one 8-mm, localized, papillary carcinoma. There were several limitations to our study. First, it was a retrospective analysis, and only a fraction of patients with thyroid USs had corresponding CTs. The time intervals between CT and US varied widely, although this is likely inconsequential for most thyroid lesions, including papillary carcinomas. CT scanners and acquisition parameters, reconstructed slice thicknesses, and contrast protocols also varied. Despite our best efforts, the largest characterized lesion on
Academic Radiology, Vol -, No -, - 2014
THYROID LESIONS ON CT
have a high probability of being benign, although this was not confirmed pathologically, possibly because of small sample size. Although small lesions on CT may represent advanced disease, recommending sonographic evaluation of all incidentally detected thyroid lesions is likely not the appropriate strategy, given the high prevalence of thyroid incidentalomas, low probability of malignancy, and considerable cost of diagnosis and treatment.
REFERENCES
Figure 3. A 62-year-old woman with incidental thyroid nodule on CT. Contrast-enhanced axial CT image (a) showed a 9-mm nodule without obvious calcifications (arrow) in the left thyroid lobe. Corresponding ultrasound image from the same date (b) revealed a 22-mm nodule with a central coarse calcification (arrowhead). FNA biopsy demonstrated papillary carcinoma, and surgical excision 1 month later revealed papillary carcinoma with extrathyroidal extension. Neither the CT nor the ultrasound demonstrated evidence of extrathyroidal spread. CT, computed tomography.
CT may not have corresponded to the largest characterized lesion on US. Moreover, sonographic assessments of benignity may have been imperfect. We did not exclude patients with history of thyroid malignancy, other preexisting thyroid disease, history of head and neck radiation, and family history of thyroid cancer. Despite our broad inclusion criteria, our prevalence of thyroid malignancy was lower than in previous studies (2,5). Finally, the number of malignant nodules was small, which may have underpowered the analyses. In conclusion, despite good agreement between CT and US descriptions of thyroid lesions, US remains the optimal modality for characterizing these lesions. Our results suggest that small, homogeneous, low-attenuation lesions on CT
1. Yousem DM, Huang T, Loevner LA, et al. Clinical and economic impact of incidental thyroid lesions found with CT and MR. Am J Neuroradiol 1997; 18:1423–1428. 2. Yoon DY, Chang SK, Choi CS, et al. The prevalence and significance of incidental thyroid nodules identified on computed tomography. J Comput Assist Tomogr 2008; 32:810–815. 3. Hoang JK, Riofrio A, Bashir MR, et al. High variability in radiologists’ reporting practices for incidental thyroid nodules detected on CT and MRI. Am J Neuroradiol 2014; 35:1190–1194. 4. Lehnert BE, Sandstrom CK, Gross JA, et al. Variability in management recommendations for incidental thyroid nodules detected on CT of the cervical spine in the emergency department. J Am Coll Radiol 2014; 11: 681–685. 5. Shetty SK, Maher MM, Hahn PF, et al. Significance of incidental thyroid lesions detected on CT: correlation among CT, sonography, and pathology. Am J Roentgenol 2006; 187:1349–1356. 6. Davies L, Welch HG. Current thyroid cancer trends in the United States. JAMA Otolaryngol Head Neck Surg 2014; 140:317–322. 7. Hoang JK, Choudhury KR, Eastwood JD, et al. An exponential growth in incidence of thyroid cancer: trends and impact of CT imaging. Am J Neuroradiol 2014; 35:778–783. 8. Ito Y, Miyauchi A, Inoue H, et al. An observational trial for papillary thyroid microcarcinoma in Japanese patients. World J Surg 2010; 34:28–35. 9. Frates MC, Benson CB, Charboneau JW, et al. Management of thyroid nodules detected at US: Society of Radiologists in US consensus conference statement. Radiology 2005; 237:794–800. 10. Cooper DS, Doherty GM, Haugen BR, et al. Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid 2009; 19:1167–1214. 11. Ahmed S, Horton KM, Jeffrey RB, Jr, et al. Incidental thyroid nodules on chest CT: review of the literature and management suggestions. Am J Roentgenol 2010; 195:1066–1071. 12. Cibas ES, Ali SZ. The Bethesda system for reporting thyroid cytopathology. Am J Clin Pathol 2009; 132:658–665. 13. Ahuja A, Chick W, King W, et al. Clinical significance of the comet-tail artifact in thyroid US. J Clin Ultrasound 1996; 24:129–133. 14. Kim EK, Park CS, Chung WY, et al. New sonographic criteria for recommending fine-needle aspiration biopsy of nonpalpable solid nodules of the thyroid. Am J Roentgenol 2002; 178:687–691. 15. Papini E, Guglielmi R, Bianchini A, et al. Risk of malignancy in nonpalpable thyroid nodules: predictive value of ultrasound and color-Doppler features. J Clin Endocrinol Metab 2002; 87:1941–1946. 16. Nguyen XV, Choudhury KR, Eastwood JD, et al. Incidental thyroid nodules on CT: evaluation of 2 risk-categorization methods for work-up of nodules. Am J Neuroradiol 2013; 34:1812–1817. 17. Hobbs HA, Bahl M, Nelson RC, et al. Incidental thyroid nodules detected at imaging: can diagnostic workup be reduced by use of the Society of Radiologists in Ultrasound recommendations and the three-tiered system? Am J Roentgenol 2014; 202:18–24.
7
Thyroid Lesions Visualized on CT: Sonographic and Pathologic Correlation Christopher Lee, MD, Brenna Chalmers, MD, Daniel Treister, BS, Sumita Adhya, MD, Benjamin Godwin, MD, Lingyun Ji, MS, Susan Groshen, PhD, Edward Grant, MD Rationale and Objectives: There are no definitive guidelines for the management of incidental thyroid lesions on computed tomography (CT). The objectives of our study were to assess the association between CT and ultrasound (US) characteristics of thyroid lesions and identify CT predictors of benignity or malignancy. Materials and Methods: Two hundred fifty-nine patients who had undergone both CT of the chest or neck and thyroid US had at least one thyroid lesion visible on CT; both incidentally detected and palpable or symptomatic lesions were included. The CT and US examinations were retrospectively reviewed and lesions characterized. Pathologic results from fine-needle aspiration (FNA) and surgical excision were used to classify lesions as benign or malignant. Thyroid lesions without pathologic correlation were classified as either benign or indeterminate based on US evaluation. Lesions that were cystic, predominantly cystic, solid and cystic with features consistent with a colloid nodule, or stable for at least 1 year were considered benign. Fisher exact tests, Spearman correlations, and logistic regression models were used to evaluate the associations between CT and US characteristics and CT predictors of benignity or malignancy. Multivariable logistic regression analyses were performed to examine the joint associations between the CT features and sonographically assessed benignity. Results: Of 259 patients, 168 (64.9%) patients had incidentally detected thyroid lesions, 49 (18.9%) patients had palpable/symptomatic lesions, and 116 (44.8%) patients underwent surgical resection and/or FNA biopsy. The malignancy prevalence was 1.8% (3 of 168) for incidental thyroid lesions and 10.2% (5 of 49) for palpable/symptomatic lesions. The malignancy prevalence of incidental lesions initially detected on CT was 1.6% (2 of 125). Of the 143 patients without pathologic data, 58 (40.6%) were classified as benign and 85 (59.4%) were categorized as indeterminate based on US evaluation. Statistically significant associations were found between CT and US with regard to lesion number, dominant lesion size, lesion consistency/composition, and associated calcifications. No CT characteristics of thyroid lesions predicted malignancy. However, there were statistically significant associations on multivariate analysis between indeterminate/benign nodules and CT characteristics of smaller lesion size, lower mean attenuation, and homogeneous composition. Conclusions: Recommending sonographic evaluation of all incidentally detected thyroid lesions is likely not the appropriate strategy, given the high prevalence of thyroid incidentalomas, low probability of malignancy, and cost effectiveness of workup. Small, homogeneous, low-attenuation lesions have a high probability of being benign. Key Words: Computed tomography; ultrasound; incidental lesion; thyroid nodule; thyroid malignancy. ªAUR, 2014
I
ncidental lesions of the thyroid are commonly encountered in computed tomography (CT) imaging of the neck and chest. Two separate studies reported a prevalence of up to 17% (1,2). There is considerable variability in the manner of reporting incidentally detected thyroid lesions, including which follow-up recommendations are provided (3,4). This variability is largely due to the fact that there are no established guidelines for the management of incidental thyroid lesions discovered on CT.
Acad Radiol 2014; -:1–7 From the Department of Radiology, Keck School of Medicine of USC, 1500 San Pablo St, 2nd Floor Imaging, Los Angeles, CA 90033 (C.L., B.C., D.T., S.A., E.G.); Department of Radiology, University of Washington School of Medicine, Seattle, WA (B.G.); and Department of Preventive Medicine, Keck School of Medicine of USC, Los Angeles, CA (L.J., S.G.). Received April 8, 2014; accepted August 1, 2014. Address correspondence to: C.L. e-mail: [email protected] ªAUR, 2014 http://dx.doi.org/10.1016/j.acra.2014.08.007
The reported malignancy prevalence of incidental thyroid nodules detected on CT is not insignificant, ranging from 9% to 11% (1,5). However, papillary carcinoma, which accounts for nearly 90% of thyroid cancer cases, has an excellent prognosis (6). In fact, despite increased diagnosis and resection of papillary cancers over the past several decades, there has been essentially no improvement in cancer-specific mortality (6,7). Micropapillary carcinomas measuring <10 mm in diameter have an even better prognosis, with one study reporting no cancer-related deaths during a 10-year surveillance period (8). Thus, although incidental thyroid lesions may harbor malignancy, workup of all thyroid incidentalomas, particularly when small, may result in overdiagnosis and overtreatment of indolent tumors. Given the increasing utilization of thin-collimation multidetector CT, high prevalence of incidental thyroid lesions, and relatively low rate of life-threatening malignancy of these nodules, the potential for unnecessary and costly follow-up is apparent. Although the role of 1
Academic Radiology, Vol -, No -, - 2014
LEE ET AL
sonography to guide further management of thyroid lesions is relatively well established (9,10), the role of CT in prompting sonographic evaluation is still controversial (11). The objectives of our study, therefore, were to assess the association between CT and sonographic characteristics of thyroid lesions, including both incidental and palpable/ symptomatic lesions, and identify CT predictors of thyroid benignity or malignancy. MATERIALS AND METHODS This retrospective Health Insurance Portability and Accountability Act–compliant study was approved by our institutional review board, with waiver of informed consent. A total of 2904 patients underwent thyroid US between July 2004 and July 2011. In the interest of including as many lesions as possible, all clinical indications were included, including incidentally detected and palpable or symptomatic lesions. Of these patients, 419 had also received a CT of the chest or neck. The CT and US examinations of these patients were retrospectively evaluated by a blinded radiologist (6 years of experience) on a digital workstation (Synapse; Fujifilm USA, Cypress, CA), with consensus review with a second radiologist (10 years of experience) performed for equivocal cases. If patients received multiple US and CT studies, the examinations closest to each other temporally were analyzed, with a median time interval of 2.5 months (range, 0–52 months). Patients were included regardless of whether the US preceded or followed the CT. Of the 419 patients, 259 had at least one thyroid lesion visible on CT; these subjects comprised the final study population. Patient demographics are shown in Table 1. CT scans were acquired on various multidetector scanners with a minimum of 10 channels; reconstructed image thickness ranged from 1 to 5 mm. One hundred sixty-three non– contrast-enhanced and 256 contrast-enhanced CTs were included. For contrast-enhanced studies, 80–100 mL of nonionic iodinated contrast (Isovue 370; Bracco Diagnostics, Princeton, NJ) was intravenously administered using a power injector at a rate of 2–4 mL/s. US examinations were performed using an Acuson Sequoia 512 machine with a broadband 8–15 MHz linear-array transducer (Siemens, Mountain View, CA) or Toshiba Aplio XG or MG units with a 7.4–14 MHz linear-array transducer (Toshiba America, Tustin, CA). Thyroid lesions on CTwere characterized based on number of lesions (1 to $5), size of the largest/dominant lesion (maximum axial dimension), and the presence of associated calcifications. Calcifications were classified as coarse, peripheral, or punctate. Coarse calcifications were defined as nonperipheral calcifications measuring at least 3 mm in size, whereas punctate calcifications were defined as calcifications measuring <3 mm. Additionally, lesion consistency was characterized as homogeneous versus heterogeneous, and mean attenuation values of the lesions were determined by placing a region of interest within the lesion (at least half of the area, excluding calcifications when possible). When 2
TABLE 1. Patient Demographics Characteristic Number of females Mean age, years Thyroid lesion presentation Incidental Initially detected on CT Initially detected on US Initially detected on MR Initially detected on PET Palpable/symptomatic Partial thyroidectomy follow-up Unknown Not seen on ultrasound
Data (n = 259) 171 (66.0) 61 14 (range, 18–96) 168 (64.9) 125 (74.4) 34 (20.2) 4 (2.4) 5 (3.0) 49 (18.9) 2 (0.8) 38 (14.7) 2 (0.8)
CT, computed tomography; MR, magnetic resonance; PET, positron emission tomography; US, ultrasound.
multiple thyroid lesions were identified, only the largest was evaluated. If a subsequent biopsy of a smaller lesion was performed, however, the reviewer was instructed to characterize that lesion instead. Thyroid lesions on US were likewise described according to number of lesions, size of the largest lesion, lesion composition, and the presence of associated calcifications. Lesion composition was classified as solid, cystic, or mixed solid/ cystic. Calcifications were categorized as coarse, peripheral, microcalcifications, and echogenic foci with associated comet-tail artifacts. Coarse calcifications measured at least 3 mm in size, whereas microcalcifications measured less than 3 mm. Only the largest of multiple lesions was described, while ensuring that the lesion was in the same lobe and approximate region as the lesion on US. Once again, however, if a biopsy of a smaller lesion was performed, that lesion was characterized in its place. Thyroid fine-needle aspiration (FNA) results were categorized according to the Bethesda classification (12). Bethesda II and III categories were considered benign, whereas Bethesda IV, V, and VI categories were considered malignant. If surgical excision was performed, the surgical pathology superseded the FNA biopsy and was used to classify a lesion as benign or malignant. Thyroid lesions without pathologic correlation were classified as either benign or indeterminate based on US evaluation. Lesions that were cystic, predominantly cystic, solid and cystic with features consistent with a colloid nodule (anechoic or hypoechoic with internal echogenic foci demonstrating comet-tail artifacts) (13), or stable on follow-up US for at least 1 year were considered benign. All other lesions, including those which were entirely or predominantly solid or contained calcifications, were considered indeterminate. For purposes of statistical analysis, lesions which underwent FNA or surgical resection were also classified as indeterminate. The associations between CT and US characteristics were examined using Fisher exact tests, Spearman correlations, or logistic regression models, as appropriate. The same analysis
Academic Radiology, Vol -, No -, - 2014
methods were used to examine the association between benignity (based on FNA/surgery or US) and the number, size, mean attenuation of lesions, the presence of calcifications, and homogeneous versus heterogeneous consistency on CT. Multivariable logistic regression analyses were performed to examine the joint associations between the CT features and sonographically assessed benignity. A P value of #.05 was considered statistically significant, and a P value of >.05 but #.10 was considered marginally significant; all reported P values were two sided. Statistical analyses were performed with STATA software version 11.0 (StataCorp LP, College Station, TX).
RESULTS Of 259 patients, 257 (99.2%) with thyroid lesions visualized on CT had lesions detected on US; two (0.8%) patients with thyroid lesions on CT had no lesions on US. Of the 259 patients, 19 (7.3%) underwent surgical resection of the thyroid lesion, whereas 97 patients (37.5%) underwent FNA biopsy only. Altogether, there were 9 malignant, 100 benign, and 7 nondiagnostic results (Table 2). Thus, the overall malignancy prevalence in our patient population was 3.5% (9 of 259). Of the 143 patients without pathologic data, 58 (40.6%) were classified as benign and 85 (59.4%) were categorized as indeterminate based on US evaluation. Of 168 patients, 70 (41.7%) patients with incidentally detected thyroid lesions underwent surgical resection and/or FNA biopsy, with 3 (4.3%) malignant, 62 (88.6%) benign, and 5 (7.1%) nondiagnostic results. Two of the incidentally detected malignancies were initially detected on CT, and the third was initially detected on positron emission tomography. Of 49 patients, 26 (53.1%) patients with palpable/ symptomatic lesions underwent resection and/or FNA, with 5 (19.2%) malignant, 21 (80.8%) benign, and 2 (7.7%) nondiagnostic results. One papillary carcinoma was discovered on surveillance US of a patient who had previously undergone a hemithyroidectomy for thyroid cancer. Consequently, the malignancy prevalence was 1.8% (3 of 168) for incidental thyroid lesions and 10.2% (5 of 49) for palpable/symptomatic lesions. The malignancy prevalence of incidental lesions initially detected on CT was 1.6% (2 of 125).
THYROID LESIONS ON CT
TABLE 2. Pathologic Results of Thyroid Lesions Result (n = 116) Incidental lesion FNA biopsy (n = 61) Nondiagnostic (Bethesda I) Benign (Bethesda II) Follicular lesion (Bethesda III) Follicular neoplasm (Bethesda IV) Surgical excision (n = 9) Multinodular goiter Follicular hyperplasia Follicular adenoma Papillary carcinoma Palpable/symptomatic lesion FNA biopsy (n = 20) Nondiagnostic (Bethesda I) Benign (Bethesda II) Surgical excision (n = 8) Colloid nodule Multinodular goiter Papillary carcinoma Medullary carcinoma Anaplastic carcinoma Other lesion FNA biopsy (n = 16) Benign (Bethesda II) Surgical excision (n = 2) Colloid nodule Papillary carcinoma
Number (%)
5 (8.2) 53 (86.9) 2 (3.3) 1 (1.6) 3 (33.3) 2 (22.2) 2 (22.2) 2 (22.2)
2 (10.0) 18 (90.0) 1 (12.5) 2 (25.0) 3 (37.5) 1 (12.5) 1 (12.5)
16 (100.0) 1 (50.0) 1 (50.0)
FNA, fine-needle aspiration. Bethesda II and III lesions were considered benign, whereas Bethesda IV lesion was considered malignant.
er, no association was observed between CT lesion attenuation and US composition. Finally, a significant association was found between associated calcifications on CT versus US (P < .001). Specifically, coarse calcifications on CT correlated with the presence of coarse calcifications on US (P < .001). However, there was a poor correlation of punctate calcifications on CT with microcalcifications or echogenic foci on US (P = .46). A comparison of CT features of benign and malignant thyroid lesions is presented in Table 4. Individual CT characteristics are addressed in the following sections.
Agreement Between CT and US Characteristics
Characteristics of thyroid lesions on CT are summarized in Table 3. Assessment of agreement between CT and US characteristics demonstrated significant correlations between lesion number on CT versus US (r = 0.40, P < .001) and dominant lesion size on CT versus US (r = 0.85, P < .001). There was also a significant association between lesion consistency on CT and solid, cystic, or mixed composition on US (P = .005); 21 of 35 cystic lesions on US appeared homogeneous on CT, whereas mixed and solid lesions were more commonly heterogeneous on CT. Howev-
Number of Lesions on CT
Of 259 patients, 117 (45.2%) patients had a single lesion on CT, 44 patients had two lesions, 22 patients had three lesions, 8 patients had four lesions, and 68 patients had five or more lesions. Five of nine (55.6%) malignant nodules occurred in patients with three or more thyroid lesions. There was no statistically significant association between the number of lesions seen on CT and the proportion of benign or malignant FNA/ surgical diagnoses. However, there was a significant inverse relationship between the number of lesions on CT and 3
Academic Radiology, Vol -, No -, - 2014
LEE ET AL
TABLE 3. Characteristics of Thyroid Lesions on CT CT Characteristic (n = 259 Patients)
Number (%)
Number of lesions 1 117 (45.2) 2 44 (17.0) 3 22 (8.5) 4 8 (3.1) $5 68 (26.3) Size of dominant lesion (mm) 1–4 19 (7.3) 5–9 69 (26.6) 10–19 95 (36.7) 20–100 76 (29.3) Consistency Heterogeneous 165 (63.7) Homogeneous 94 (36.3) Mean attenuation with contrast (HU, n = 174) <50 29 (16.7) 50–99 48 (27.6) 100–199 83 (47.7) >200 14 (8.0) Mean attenuation without contrast (HU, n = 85) <50 42 (49.4) 50–99 38 (44.7) 100–199 4 (4.7) >200 1 (1.2) Associated calcifications* Absent 203 (78.4) Present 56 (21.6) Punctate 10 (25.7) Coarse 39 (15.2) Peripheral 17 (6.6) CT, computed tomography; HU, Hounsfield units. *Some lesions had more than one type of calcification.
proportion of benign cases as assessed by US (P = .002). Patients with more lesions on CTwere more likely to have indeterminate sonographic assessments of the largest/dominant lesion. Lesion Size on CT
The mean maximum diameter of the dominant thyroid lesions on CT was 16 12 mm (range, 2–64 mm). The mean diameter of malignant nodules was 26 19 mm (range, 7–53 mm), whereas the mean diameter of pathologically proven benign nodules was 20 11 mm (range, 5–64 mm), a difference that was not statistically significant (P = .19). In addition, there was no statistically significant association between dominant lesion size on CT and proportion of benign or malignant FNA/surgical diagnoses. However, there was a significant inverse relationship between lesion size and benignity as determined with US (P < .001). Increasing lesion size on CTwas associated with a decreasing rate of sonographically assessed benignity. 4
Lesion Consistency on CT
The characterized lesions on CT were heterogeneous in 165 patients (63.7%) and homogeneous in 94 cases (36.3%). Malignant nodules were, similarly, slightly more commonly heterogeneous. Once again, there was no statistically significant association between lesion consistency on CT and pathologic benignity or malignancy. However, there was a significant association between lesion consistency and proportion of benign cases as assessed with US (P < .001). Homogeneous lesions on CT were more often associated with benign sonographic evaluations. Lesion Attenuation on CT
The average CT attenuation of thyroid lesions was 93 77 Hounsfield units (HU; range, 11–1140 HU). For patients with noncontrast studies, the average lesion attenuation was 65 105 HU, and for those with contrast-enhanced studies, the average attenuation was 107 54 HU. There was no statistically significant association between mean lesion attenuation and proportion of benign or malignant FNA/surgical diagnoses. There was, however, a significant inverse relationship between lesion attenuation on contrast-enhanced studies and sonographic benignity (P = .011). As the mean attenuation increased, the proportion of benign cases as determined with US decreased. There was no statistically significant association between CT attenuation on noncontrast scans and benignity as assessed by US, although a slight trend was observed between increasing lesion attenuation and indeterminate assessments. Lesion Calcifications on CT
Of 259 patients, 56 (21.6%) patients demonstrated thyroid lesions with associated calcifications visible on CT. Only one of nine (11.1%) malignant nodules exhibited calcifications on CT. Once again, there was no statistically significant association between the presence of calcifications and pathologic benignity or malignancy. However, there was a marginally significant inverse relationship between associated calcifications on CT and proportion of benign cases per sonographic evaluation (P = .068). Accompanying calcifications were more often associated with indeterminate sonographic assessments. Further analyses of specific types of calcifications on CT did not yield any improvements in statistical significance. Multivariable logistic regression analyses revealed that dominant lesion size, mean attenuation, and consistency were independently associated with the proportion of benign cases as determined with US (P < .001 for lesion size, P = .044 for mean attenuation, and P = .005 for consistency). After adjusting for these three variables, no significant association was found between sonographic benignity and number of lesions or the presence of associated calcifications.
Academic Radiology, Vol -, No -, - 2014
THYROID LESIONS ON CT
TABLE 4. Comparison of CT Features of Benign Versus Malignant Thyroid Lesions
Parameter Age, years Sex Female Male Number of lesions 1 2 3 4 $5 Size of lesion (mm) 1–4 5–9 10–19 20–100 Lesion consistency Heterogeneous Homogeneous Mean lesion attenuation with contrast (HU) <50 50–99 100–199 >200 Mean lesion attenuation without contrast (HU) <50 50–99 100–199 >200 Associated calcifications Present Coarse and peripheral Coarse Punctate Peripheral
Malignant (n = 9)
Benign (n = 100)
65.0
62.1
6 (66.7) 3 (33.3)
67 (67.0) 33 (33.0)
4 (44.4) 0 2 (22.2) 0 3 (30.3)
36 (36.0) 19 (19.0) 8 (8.0) 5 (5.0) 32 (32.0)
P Value .53* .62y
.99z
.62z 0 4 (44.4) 0 5 (55.6)
0 17 (17.0) 42 (42.0) 41 (41.0)
6 (66.6) 3 (33.3) n=7
78 (78.0) 22 (22.0) n = 70
2 (28.6) 2 (28.6) 3 (42.9) 0 n=2
7 (10.0) 19 (27.1) 36 (51.4) 8 (11.4) n = 30
1 (50.0) 1 (50.0) 0 0 n=9 1 (11.1) 1 (11.1) 0 0 0
16 (53.3) 13 (43.3) 1 (3.3) 0 n = 100 18 (18.0) 1 (1.0) 11 (11.0) 4 (4.0) 2 (2.0)
.43y
.14z
.99z
.51y
CT, computed tomography; HU, Hounsfield units. *Wilcoxon rank–sum test. y Fisher exact test. z Trend test.
DISCUSSION Incidental thyroid lesions are commonly encountered on routine CT scans of the chest or neck; however, no definitive guidelines exist for the reporting and workup of these incidentalomas. The ability to identify benign or otherwise low-risk findings on CT would help prevent unnecessary workup. Therefore, we sought to describe the CT features of thyroid lesions, compare them with sonographic characteristics, and identify CT predictors, if any, of thyroid benignity or malignancy. Our study demonstrated a good agreement between CT and US with regard to number of lesions, dominant lesion size, lesion consistency/composition, and the
Figure 1. A 40-year-old man with palpable left neck mass. Contrast-enhanced axial computed tomography image (a) showed a homogeneous 7-mm nodule (arrow) within the left thyroid lobe. A large lymph node (arrowheads) was also present adjacent to the left internal jugular vein. Corresponding ultrasound image 8 days later (b) demonstrated a 17-mm cystic and solid nodule (open arrows) with microcalcifications (arrowheads). Surgical excision revealed medullary carcinoma with extrathyroidal and nodal spread.
presence of associated calcifications. In addition, we observed significant associations between lesion number, size, consistency, mean attenuation, and associated calcifications on CT and benignity as determined by US. Despite good correlation of lesion size between CT and US, we observed consistent underestimation of size on CT compared to US (Fig 1), which has been confirmed by other investigators and attributed to a variety of technical and anatomic factors (5,11). Interestingly, only coarse calcifications demonstrated good correlation between CT and US. On the other hand, punctate calcifications on CT were most commonly associated with no visible calcifications on US, whereas microcalcifications on US—a finding suggestive of malignancy (9,14)—were most commonly associated with no associated calcifications on CT (Fig 2). This may be partly due to the sonographers’ inability to prospectively identify tiny calcifications associated with a lesion. Additionally, microcalcifications on US are likely too small to be resolved on CT. As stated previously, we detected significant associations between lesion number, size, consistency, mean attenuation, and associated calcifications on CT and benignity as determined by US. Multivariable logistic regression analysis further showed that smaller lesion size, lower mean attenuation, and homogeneous consistency were significantly associated with 5
LEE ET AL
Figure 2. A 57-year-old woman with incidental thyroid nodule on CT. Contrast-enhanced axial CT image (a) demonstrated an ill-defined heterogeneous nodule without obvious calcifications (arrows) within the right thyroid lobe. Corresponding ultrasound image (b) showed a cystic and solid nodule with multiple echogenic foci/microcalcifications, with occasional comet-tail artifacts (arrowhead) suggesting colloid. Fine-needle aspiration biopsy revealed a colloid nodule compatible with Bethesda II lesion. CT, computed tomography.
sonographically assessed benignity. Unfortunately, because of considerable overlap in size and attenuation of benign and malignant lesions, useful threshold values could not be determined from our data. There was no significant association between the various CT features and benignity or malignancy based on surgical resection and/or FNA. This may be due to the small number of documented malignancies, which may have underpowered the analyses. Our results are somewhat concordant with previously published studies. Shetty et al. reported that benign thyroid lesions were smaller than malignant or potentially malignant lesions (5). Similarly, we demonstrated that smaller lesions on CT were more likely to have benign sonographic assessments. Yoon et al. (2) observed that malignant lesions more frequently demonstrated calcifications and mean attenuation values >130 HU on contrast-enhanced CT. In our study population, calcifications and increased attenuation on CT were associated with indeterminate sonographic evaluations. On the other hand, six of seven pathologically proven malignancies on contrast-enhanced CT demonstrated a mean attenuation of <130 HU. This discrepancy may be due to variations in contrast concentration, volume, and scanning delays between the two patient populations. 6
Academic Radiology, Vol -, No -, - 2014
In our patient population, the malignancy prevalence of patients with thyroid lesions visualized on CT was 3.5%. When only including patients with incidental thyroid lesions on CT, the malignancy prevalence dropped to 1.6%. This rate is less than that reported by two prior publications which also studied incidental thyroid lesions on CT. Shetty et al. (5) reported an 11.3% prevalence of malignant or potentially malignant lesions; however, only 3.9% represented definite malignancies, a prevalence closer to ours. Yoon et al. (2), on the other hand, observed a 9.4% malignancy prevalence, but their study population included patients with palpable or symptomatic lesions. Therefore, it is likely that the malignancy prevalence of incidentally detected lesions is closer to 2%. The true malignancy prevalence may be much lower, as sonographic workup of a CT incidentaloma likely selects for lesions that have a higher probability of being malignant. Three of the nine malignant lesions in our study were papillary carcinomas that measured <10 mm on CT. Furthermore, five of the malignant lesions occurred in patients with multinodular glands. These observations underscore the fact that small thyroid lesions or multinodular glands may harbor a significant percentage of thyroid malignancies, a notion that has been asserted by previous authors (9,15). Of course, small cancers, particularly when demonstrating papillary histology, have a high probability of representing indolent tumors (8). Thus, the low prevalence of malignancy and relatively high proportion of small indolent neoplasms combine to create the management dilemma of working up incidental thyroid lesions at high expense but conceivably little impact on cancer-specific mortality. Recently, a three-tiered system for workup of incidental thyroid nodules encountered on cross-sectional imaging was proposed, recommending sonographic evaluation of thyroid nodules with evidence of local invasion or nodal spread, nodules of any size in patients <35 years of age, and nodules $1.5 cm in those $35 years (16,17). The three-tiered system would have significantly reduced benign biopsies in our cohort, because the mean age was 61 years and more than a third of the lesions in our study were <1 cm. However, three papillary carcinomas which manifested as subcentimeter nodules on CT would have been missed, with one of the lesions representing an incidentally detected papillary carcinoma with extrathyroidal spread (Fig 3). All three tumors measured at least 10 mm on sonography; CT underestimated the size of the malignant nodule by more than one-third. Of interest, if we were to also recommend follow-up US of all heterogeneous lesions, regardless of size, we would have only missed one 8-mm, localized, papillary carcinoma. There were several limitations to our study. First, it was a retrospective analysis, and only a fraction of patients with thyroid USs had corresponding CTs. The time intervals between CT and US varied widely, although this is likely inconsequential for most thyroid lesions, including papillary carcinomas. CT scanners and acquisition parameters, reconstructed slice thicknesses, and contrast protocols also varied. Despite our best efforts, the largest characterized lesion on
Academic Radiology, Vol -, No -, - 2014
THYROID LESIONS ON CT
have a high probability of being benign, although this was not confirmed pathologically, possibly because of small sample size. Although small lesions on CT may represent advanced disease, recommending sonographic evaluation of all incidentally detected thyroid lesions is likely not the appropriate strategy, given the high prevalence of thyroid incidentalomas, low probability of malignancy, and considerable cost of diagnosis and treatment.
REFERENCES
Figure 3. A 62-year-old woman with incidental thyroid nodule on CT. Contrast-enhanced axial CT image (a) showed a 9-mm nodule without obvious calcifications (arrow) in the left thyroid lobe. Corresponding ultrasound image from the same date (b) revealed a 22-mm nodule with a central coarse calcification (arrowhead). FNA biopsy demonstrated papillary carcinoma, and surgical excision 1 month later revealed papillary carcinoma with extrathyroidal extension. Neither the CT nor the ultrasound demonstrated evidence of extrathyroidal spread. CT, computed tomography.
CT may not have corresponded to the largest characterized lesion on US. Moreover, sonographic assessments of benignity may have been imperfect. We did not exclude patients with history of thyroid malignancy, other preexisting thyroid disease, history of head and neck radiation, and family history of thyroid cancer. Despite our broad inclusion criteria, our prevalence of thyroid malignancy was lower than in previous studies (2,5). Finally, the number of malignant nodules was small, which may have underpowered the analyses. In conclusion, despite good agreement between CT and US descriptions of thyroid lesions, US remains the optimal modality for characterizing these lesions. Our results suggest that small, homogeneous, low-attenuation lesions on CT
1. Yousem DM, Huang T, Loevner LA, et al. Clinical and economic impact of incidental thyroid lesions found with CT and MR. Am J Neuroradiol 1997; 18:1423–1428. 2. Yoon DY, Chang SK, Choi CS, et al. The prevalence and significance of incidental thyroid nodules identified on computed tomography. J Comput Assist Tomogr 2008; 32:810–815. 3. Hoang JK, Riofrio A, Bashir MR, et al. High variability in radiologists’ reporting practices for incidental thyroid nodules detected on CT and MRI. Am J Neuroradiol 2014; 35:1190–1194. 4. Lehnert BE, Sandstrom CK, Gross JA, et al. Variability in management recommendations for incidental thyroid nodules detected on CT of the cervical spine in the emergency department. J Am Coll Radiol 2014; 11: 681–685. 5. Shetty SK, Maher MM, Hahn PF, et al. Significance of incidental thyroid lesions detected on CT: correlation among CT, sonography, and pathology. Am J Roentgenol 2006; 187:1349–1356. 6. Davies L, Welch HG. Current thyroid cancer trends in the United States. JAMA Otolaryngol Head Neck Surg 2014; 140:317–322. 7. Hoang JK, Choudhury KR, Eastwood JD, et al. An exponential growth in incidence of thyroid cancer: trends and impact of CT imaging. Am J Neuroradiol 2014; 35:778–783. 8. Ito Y, Miyauchi A, Inoue H, et al. An observational trial for papillary thyroid microcarcinoma in Japanese patients. World J Surg 2010; 34:28–35. 9. Frates MC, Benson CB, Charboneau JW, et al. Management of thyroid nodules detected at US: Society of Radiologists in US consensus conference statement. Radiology 2005; 237:794–800. 10. Cooper DS, Doherty GM, Haugen BR, et al. Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid 2009; 19:1167–1214. 11. Ahmed S, Horton KM, Jeffrey RB, Jr, et al. Incidental thyroid nodules on chest CT: review of the literature and management suggestions. Am J Roentgenol 2010; 195:1066–1071. 12. Cibas ES, Ali SZ. The Bethesda system for reporting thyroid cytopathology. Am J Clin Pathol 2009; 132:658–665. 13. Ahuja A, Chick W, King W, et al. Clinical significance of the comet-tail artifact in thyroid US. J Clin Ultrasound 1996; 24:129–133. 14. Kim EK, Park CS, Chung WY, et al. New sonographic criteria for recommending fine-needle aspiration biopsy of nonpalpable solid nodules of the thyroid. Am J Roentgenol 2002; 178:687–691. 15. Papini E, Guglielmi R, Bianchini A, et al. Risk of malignancy in nonpalpable thyroid nodules: predictive value of ultrasound and color-Doppler features. J Clin Endocrinol Metab 2002; 87:1941–1946. 16. Nguyen XV, Choudhury KR, Eastwood JD, et al. Incidental thyroid nodules on CT: evaluation of 2 risk-categorization methods for work-up of nodules. Am J Neuroradiol 2013; 34:1812–1817. 17. Hobbs HA, Bahl M, Nelson RC, et al. Incidental thyroid nodules detected at imaging: can diagnostic workup be reduced by use of the Society of Radiologists in Ultrasound recommendations and the three-tiered system? Am J Roentgenol 2014; 202:18–24.
7