Thyroid Follicular Carcinoma: Sonographic Features of 50 Cases

Transcription

1 Neuroradiology/Head and Neck Imaging Original Research Sillery et al. Sonography of Thyroid Follicular Carcinoma Neuroradiology/Head and Neck Imaging Original Research FOCUS ON: John C. Sillery 1 Carl C. Reading 1 J. William Charboneau 1 Tara L. Henrichsen 1 Ian D. Hay 2 Jayawant N. Mandrekar 3 Sillery JC, Reading CC, Charboneau JW, Henrichsen TL, Hay ID, Mandrekar JN Keywords: thyroid follicular adenoma, thyroid follicular carcinoma, sonography, ultrasound DOI: /AJR Received June 15, 2009; accepted after revision July 7, Department of Radiology, Mayo Clinic, 200 First St. SW, Rochester, MN Address correspondence to J. C. Sillery (sillery.john@mayo.edu). 2 Department of Endocrinology, Mayo Clinic, Rochester, MN. 3 Department of Health Sciences Research, Division of Biostatistics, Mayo Clinic, Rochester, MN. AJR 2010; 194: X/10/ American Roentgen Ray Society Thyroid Follicular Carcinoma: Sonographic Features of 50 Cases objective. The purpose of our study was to retrospectively evaluate sonography of thyroid follicular neoplasms for features that would aid in distinguishing follicular carcinoma from follicular adenoma and for any imaging features that distinguish the Hürthle-cell variant of follicular carcinoma from classic follicular carcinoma. MATERIALS AND METHODS. The study cohort consisted of patients with the diagnosis of follicular carcinoma and patients with the diagnosis of follicular adenoma. Fifty patients (25 men and 25 women; median age, 59.5 years) with a diagnosis of follicular carcinoma (27 with classic follicular carcinoma, 22 with Hürthle-cell variant of follicular carcinoma, and one insular variant) in a 6-year period were included. Fifty-two control patients (10 men and 42 women; median age, 46.5 years) were selected from a random sampling of all cases of follicular adenoma during the same time period. Sonograms were reviewed in consensus by four radiologists for various features. All study patients and control patients underwent surgical resection and pathologic analysis of their thyroid follicular neoplasm. The chi-square or Fisher s exact test was used for categorical variables; the Wilcoxon s rank sum test was used for continuous variables. RESULTS. Hypoechoic appearance (82% of follicular carcinoma patients vs 50% of follicular adenoma patients; p < 0.005; odds ratio [OR]), 0.5; 95% CI, ), absence of halo (64% of follicular carcinoma patients vs 42% of follicular adenoma patients; p < 0.05; OR, 0.4; 95% CI, ), absence of cystic change (90% of follicular carcinoma patients vs 69% of follicular adenoma patients; p < 0.05; OR, 0.2; 95% CI, ), greater patient age (median age of 59.5 years for follicular carcinoma patients vs 46.5 years for follicular adenoma patients; p < 0.05), size of the tumor (median size of ml for follicular carcinoma patients vs 5.95 ml for follicular adenoma patients; p < 0.05), and male sex (50% of follicular carcinoma patients vs 19.2% of follicular adenoma patients; p < 0.005; OR, 3.7; 95% CI, ) were more frequently associated with follicular thyroid cancer than with benign adenoma. No significant difference in the prevalence of refractive shadowing, echotexture, visible invasion, lymph node enlargement, adjacent nonfollicular suspicious lesions, vascularity subtype, and calcifications was observed between the two groups. Within the follicular carcinoma subgroup, homogeneous or predominantly homogeneous echotexture (67% of classic follicular carcinoma patients vs 36% of Hürthle-cell variant of follicular carcinoma patients; p < 0.05; OR, 3.5; 95% CI, ) and the presence of calcifications (22% of classic follicular carcinoma patients vs 4% of Hürthle-cell variant of follicular carcinoma patients [multivariate analysis including age]; p < 0.05; OR, 22.9; 95% CI, ) were associated with classic follicular carcinoma. Greater patient age (median age of 53 years for classic follicular carcinoma patients vs 64.5 years for Hürthle-cell variant of follicular carcinoma patients; p < 0.05) was associated with Hürthle-cell variant follicular carcinoma. There was no association between tumor volume, sex, sonographic halo, refractive shadowing, echogenicity, visible invasion, lymph node enlargement, adjacent nonfollicular suspicious lesions, vascularity subtype, and cystic change between the subgroups of follicular carcinoma. CONCLUSION. The sonographic features of follicular adenoma and follicular carcinoma are very similar, but larger lesion size, lack of a sonographic halo, hypoechoic appearance, and absence of cystic change favored a follicular carcinoma diagnosis. Increased patient age and male sex are associated with malignancy. Within the follicular carcinoma subgroup, Hürthle-cell variant of follicular carcinoma is more often seen in older patients with nodules having a heterogeneous appearance and lacking internal calcifications. T hyroid carcinoma is the most common malignancy involving the endocrine glands and is responsible for approximately 1.5% of new cases and 1,500 deaths from cancer annually in the United States [1, 2]. The histopathologic classification of these tumors includes papillary thyroid cancer (60 80%), follicular carcinoma (15 18%), anaplastic carcinoma (3 10%), medullary carcinoma (4 5%), lymphoma (5%), and metastases [3]. 44 AJR:194, January 2010

2 Sonography of Thyroid Follicular Carcinoma Follicular neoplasm (consisting of the combination of high numbers of follicular cells, microfollicular arrangement, and scant or absent colloid) is a cytologic term used to encompass both the benign proliferation of thyroid follicular cells in adenoma and the malignant proliferation in carcinoma. Follicular adenomas are much more common than follicular carcinomas, occupying a histologic (if not biologic) niche between follicular hyperplasia and follicular carcinoma. Unlike carcinomas, adenomas have no vascular or capsular invasion but otherwise have similar cytologic features. In general, when a biopsy specimen of a thyroid nodule reveals a follicular neoplasm, approximately 80 90% of such lesions will be adenomas and 10 20% will be carcinomas [4 6]. Microscopically, most follicular carcinomas are composed of fairly uniform cells forming small follicles of high cellularity containing scant colloid, reminiscent of normal thyroid but lacking the diagnostic features of papillary cancer, which has characteristic nuclear features and often contains psammoma bodies [7 9]. Occasional follicular tumors are dominated (> 75% of cell content) by cells with abundant granular, eosinophilic cytoplasm known as Hürthle cells [10]. A third rare subtype of follicular carcinoma consists of poorly differentiated cells with a characteristic solid infiltrating nestlike appearance (insulae), which commonly contain necrosis, hemorrhage, and vascular invasion, and is known, appropriately, as the insular variant of follicular carcinoma [11]. There is debate among clinicians as to the prognostic significance of the Hürthlecell variant of follicular carcinoma in a patient as opposed to classic follicular carcinoma [12 14], although most studies indicate there is no significant mortality difference between the two subtypes of follicular carcinoma. Invasion of adjacent thyroid parenchyma may be grossly apparent or can be limited to microscopic foci of capsular or vascular invasion [15]. These lesions may require extensive histologic sampling before they can be distinguished from follicular adenoma. Patients with minimally invasive follicular carcinoma have an excellent prognosis. In contrast, patients with follicular carcinoma with extensive vascular invasion have a poorer prognosis, and distant metastases are sometimes present [16]. Vascular invasion is common, with spread to bone (predominantly osteolytic), lungs, and occasionally liver or brain [17]. A noninvasive method of evaluating thyroid nodules identified as containing follicular neoplasia that could reliably differentiate between benign follicular adenomas and malignant follicular carcinomas would be invaluable in avoiding both the risks (including laryngeal nerve injury and hypoparathyroidism) and expense of surgery [18, 19]. Rarely does physical examination help in the differentiation of the benign and malignant thyroid nodule [20], and, although certain clinical features (including male sex, size greater than 4 cm, and solitary nodule) [21 23] may be helpful in risk stratification, a cytologic finding of follicular neoplasia often mandates surgery. Sonography is attractive in the evaluation of palpable thyroid nodules given its high resolution, absence of exposure to ionizing radiation, portability, and ease of use. Several retrospective studies have described the sonographic appearance of follicular neoplasms. Two studies [24, 25] reported no value of sonography in distinguishing follicular carcinoma from follicular adenoma. Two other studies [21, 26] described a total of nine and eight follicular carcinomas, respectively, and combined the findings with the follicular variant of papillary carcinoma to describe hypoechogenicity as a useful feature of malignancy. No study has attempted to describe differentiating between classic follicular carcinoma and the Hürthle-cell variant of follicular carcinoma. The purpose of our study was to retrospectively evaluate sonographic images of thyroid follicular neoplasm for features that would aid in distinguishing follicular carcinoma from follicular adenoma and any imaging features that distinguish the Hürthle-cell variant of follicular carcinoma from classic follicular carcinoma. Materials and Methods Patients The study was approved by our institutional review board. All patients had previously consented to the use of their medical records for the purpose of research. This study was HIPAA compliant. Through a maintained database (Systematized Nomenclature of Medicine [SNOMED]), we retrospectively identified 58 patients who received a diagnosis of follicular carcinoma between 1998 and Eight patients were excluded from the study: In five patients, presurgical sonography was not available, and three patients declined permission to use their medical records for research. A total of 50 patients (25 males and 25 females; median age, 59.5 years; age range, years) were included as cases of follicular carcinoma in the study. The diagnosis of follicular carcinoma was determined by thorough histopathologic analysis of resected hemithyroidectomy or total thyroidectomy specimens. A total of 58 control patients (10 males and 42 females; median age, 46.5 years; age range, years) diagnosed with follicular adenoma were initially selected during the study period from a total of 579 cases of follicular adenoma that resulted in surgical removal (131 males and 448 females) by selecting every tenth case arranged by medical record number. Six patients were excluded from the study: Four patients lacked presurgical sonography, and two patients declined permission to use their medical records for research. Sonography Sonography was available with an integrated PACS (ALI, McKesson Medical Imaging Group) for all patients imaged from January 1, 1998, to the present. Patients diagnosed before that time were excluded from the study because retrieval and manipulation of images would be cumbersome. Sonography was performed by trained sonographers using the Acuson Sequoia 512 (Siemens Healthcare). Gray-scale images were obtained with a high-frequency 8 15-MHz linear transducer with compound frequency settings of 7.5 and 10 MHz or a fundamental frequency of 14 MHz, and Doppler images were obtained at fundamental frequencies of MHz. Four radiologists (two with more than 20 years of experience and two with less than 5 years of experience in thyroid sonography) conducted a retrospective consensus review of the 50 patients with follicular carcinoma and 52 control patients with follicular adenoma. Thirteen variables were recorded, including patient age, tumor dimensions, sex, and the following imaging features: presence and completeness of a sonographic halo; presence of refractive shadowing; predominant echotexture, homogeneous or heterogeneous; predominant echogenicity hypoechoic, isoechoic, or hyperechoic relative to the adjacent thyroid parenchyma; presence of visible invasion of adjacent soft tissues at the borders of the follicular neoplasm; presence of abnormal lymph nodes in the neck; presence of other nonfollicular suspicious lesions found in the thyroid at the same examination; presence, amount, and pattern of vascularity using Doppler imaging; cystic change; and calcifications. Halo A halo is defined as a thin rim of decreased echogenicity surrounding the neoplasm and is thought to represent the capsule that surrounds follicular neoplasms in resected specimens. The presence of a well-defined, intact capsule is important for the pathologist to observe and is a hallmark of distinction between follicular adenoma and follicular carcinoma. A thin hypoechoic AJR:194, January

3 Sillery et al. halo was present in approximately half of the 102 studied patients. The converse of the well-defined halo is absence of a halo indicated by indistinct margins around the lesion. Refractive shadowing Refractive shadowing occurs when sound passes from a tissue with one acoustic velocity to tissue with a faster or slower sound velocity. When the ultrasound beam intersects a boundary interface at a critical obliquity, there is refraction of the beam resulting in a change in direction of the sound wave, causing lack of transmission deep in relation to this interface. The result is the appearance of shadowing deep in relation to a specular reflector, thought to be a result of peripheral fibrosis or encapsulation (Fig. 1). Echotexture As a rule, most follicular neoplasms are of fairly uniform echotexture, and this correlates with their macroscopic appearance. Occasionally, some heterogeneity of the echotexture is seen within follicular neoplasms, likely representing different amounts of stromal and follicular tissue within the neoplasm. Uniformity of echotexture was described for each lesion as homogeneous, predominantly homogeneous (if the lesion contained small foci of nonuniform echotexture), or heterogeneous (if more than 10% of the lesion had nonuniform echotexture) (Fig. 2). Echogenicity Analysis of echogenicity of the neoplasm relative to adjacent normal thyroid tissue: hypoechoic (if fewer echoes than the surrounding tissue), isoechoic (if similar echoes to surrounding thyroid tissue), or hyperechoic (if more echoes than surrounding thyroid tissue) was also performed. If there were cystic areas within the neoplasm, only the solid component of the neoplasm was used to describe echogenicity (Fig. 3). Most of the 102 studied patients (n = 67) had a hypoechoic appearance, including five cases of Fig. 1 Sonogram in 67-year-old woman shows refractive shadows from the lateral aspects of follicular adenoma (arrows). follicular carcinoma characterized by the pathologist as widely invasive. Visible invasion Invasion was defined as the obvious extension of the margins of the follicular neoplasm beyond the confines of the thyroid gland proper. Only a single case was identified with visible invasion of the tissues posterior to the thyroid gland. This finding (in a case of Hürthle-cell carcinoma) is a rare but expected finding in highly invasive subtypes of follicular carcinoma. Unfortunately, microscopic invasion is impossible to detect with the current resolution of high-frequency sonography. In this case, both the surgical and pathologic reports confirmed the presence of widespread macroscopic invasion of the adjacent musculature and soft tissue (widely invasive). Cervical lymphadenopathy Images were studied for lymph node enlargement or abnormal nodal architecture, although lymphatic spread is much less common in follicular carcinoma than in papillary carcinoma. Images were examined for the obliteration of the normal hilar fat of the node, obvious enlargement, or increased vascularity measured by Doppler sonography. Nonfollicular suspicious lesions A frequent finding was the presence of other, nonfollicular suspicious lesions in the ipsilateral or contralateral lobes of the thyroid, which, in the absence of the Fig. 2 Echotexture. A, Sonogram in 12-year-old girl shows homogeneous echotexture of follicular carcinoma. B, Sonogram in 60-year-old man shows heterogeneous echotexture of Hürthle-cell carcinoma. A B 46 AJR:194, January 2010

4 Sonography of Thyroid Follicular Carcinoma A C detection of the follicular neoplasm, would merit mention in the report and clinical or imaging follow-up. Typically, these were small cystic lesions that were difficult to characterize, although classic findings of incidental papillary cancer were occasionally found. Vascularity A pattern recognition approach to vascularity was used to describe the Doppler appearance of the follicular neoplasms (Fig. 4). In general, many tumors display increased vascularity as a result of their disordered growth. Unfortunately, standardization of sonographic techniques for estimating blood flow do not exist in a quantitative fashion, and only qualitative assessments can be made. Using the method described by Frates et al. [27], follicular neoplasms were classified as type 0, no visible flow; type 1, minimal internal flow without a peripheral ring; type 2, peripheral ring of flow but minimal or no internal flow; type 3, Fig. 3 Echogenicity in three cases of follicular carcinoma. A, Sonogram in 56-year-old woman shows hypoechoic echogenicity. B, Sonogram in 56-year-old man shows isoechoic echogenicity. C, Sonogram in 48-year-old man shows hyperechoic echogenicity. peripheral ring of flow and a small to moderate amount of internal flow; and type 4, extensive internal flow with or without a peripheral ring. Cystic change The appearance of anechoic cystic spaces within the follicular neoplasm, comprising complex cystic change within a lesion, was occasionally seen (21 cases). Cystic change is commonly seen within benign colloid cysts, although it is sometimes described as a feature of papillary carcinoma of the thyroid. B AJR:194, January

5 Sillery et al. To define the degree of involvement, the extent of cystic change was recorded in a semiquantitative way, with mild cystic change defined as cystic spaces occupying 1 5% of the total follicular lesion, moderate cystic change as 6 50% of the neoplasm, and large cystic change as greater than 50% of the total volume of the follicular neoplasm. Presumptively, these are due to areas of cystic necrosis or the presence of degenerated blood products within the lesion because any thyroid neoplasm may undergo hemorrhagic necrosis and contain cystic areas (Fig. 5). A C Calcifications Calcifications are seen only occasionally in resected follicular neoplasms, but their presence was recorded. Intralesional calcifications are characterized by the presence of punctuate areas of increased echogenicity with prominent posterior acoustic shadowing (Fig. 6). Fig. 4 Types of flow pattern. A, Color Doppler sonogram in 58-year-old woman shows no signal within follicular adenoma. This is type 0 flow pattern. B, Color Doppler sonogram in 49-year-old man shows minimal internal flow without peripheral ring in Hürthle-cell carcinoma. This is type 1 flow pattern. C, Color Doppler sonogram in 40-year-old woman shows peripheral ring of flow but minimal or no internal flow within follicular adenoma. This is type 2 flow pattern. D, Color Doppler sonogram in 17-year-old girl shows peripheral ring of flow and a small to moderate amount of internal flow within follicular adenoma. This is type 3 flow pattern. (Fig. 4 continues on next page) B D 48 AJR:194, January 2010

6 Sonography of Thyroid Follicular Carcinoma E Fig. 4 (continued) Types of flow pattern. E, Color Doppler sonogram in 71-year-old man shows extensive flow with or without a peripheral ring within a follicular carcinoma. This is type 4 flow pattern. Statistical Analysis For each case, the sex and age of the patient at diagnosis was recorded as were the results of the preoperative biopsy procedure (if any) and the type of surgical procedure. The size of the follicular neoplasm was recorded in three dimensions and the volume of the encompassing ellipsoid recorded (length width height π / 6). The 102 cases (52 follicular adenoma and 50 follicular carcinoma) were randomized by accession number, blinding the observers to diagnosis, and the variables selected for analysis (sonographic halo, refractive shadowing, echotexture, echogenicity, visible invasion, lymph node enlargement, other nonfollicular suspicious lesions in the thyroid, visible metastases, vascularity subtype, cystic change, and calcifications) were recorded by consensus of all radiologists. Categorical variables of interest such as sonographic halo, refractive shadowing, and calcifications were summarized as frequencies and percentages. Continuous variables, such as age and volume (ml), were summarized as median and minimum maximum. The binary outcome of interest in this study was follicular neoplasm subtype either follicular adenoma or follicular carcinoma. The association between the outcome and categorical predictor variables of interest was assessed using the chi-square test or Fisher s exact test. The association between the outcome and continuous predictor variables of interest was assessed using Wilcoxon s rank sum test because of the non-gaussian distribution of the data. The association between the outcome and categorical predictor variables with ordered categories was assessed using the Cochran-Armitage test for trend. Further analysis was performed using univariate and multivariate logistic regression with follicular neoplasm subtype either follicular adenoma or follicular carcinoma as a binary outcome. Variables having a p value of 0.15 in univariate analysis were considered as candidates in the multivariate logistic regression model. Odds ratios were estimated along with 95% CI. In addition, univariate and multivariate logistic regression analysis also was performed with the identical variables, using the data collected from the follicular carcinoma subgroup and comparing the pathologic descriptors for the subtypes of follicular carcinoma (classic follicular carcinoma, A Fig. 5 Cystic change. A, Color Doppler sonogram in 64-year-old woman shows small amount of internal cystic change (arrows) centrally within follicular adenoma. This finding is statistically more frequent in benign than in malignant disease. B, Color Doppler sonogram in 40-year-old man shows large amount of central fluid containing debris within follicular adenoma. B AJR:194, January

7 Sillery et al. Hürthle-cell variant of follicular carcinoma, and insular variant of follicular carcinoma). All tests were two-sided and a p value of < 0.05 was considered statistically significant. The statistical software used was SAS, version 8.0 (SAS Institute). Fig. 6 Sonogram in 75-year-old woman shows scattered calcifications with posterior acoustic shadowing within follicular adenoma. Results Follicular Carcinoma Follicular Adenoma (Case Control) Comparison The characteristics of the case and control groups appear in Table 1, which shows the clinical and sonographic characteristics of the 50 cases of follicular carcinoma (including follicular carcinoma, Hürthle-cell-variant follicular carcinoma, and insular variant follicular carcinoma) and the 52 cases of follicular adenoma. The carcinoma cases were more likely to be hypoechoic, lack cystic change, have a larger volume, and to be older patients of male sex, whereas the adenoma cases were more likely to have a sonographic halo. Sonographically guided biopsy was frequently performed in both the case and control groups 37 of 50 (74%) in follicular carcinoma and 38 of 52 (73.1%) in follicular adenoma and was typically performed within 1 2 days of the time of diagnostic imaging in our practice. Cytologic findings were typically described as consistent with follicular or Hürthle-cell neoplasm, and in nine of the cases (18%) the pathologist commented on the presence of marked cellular atypia. The median age was 46.5 years in follicular adenoma and 59.5 years in follicular carcinoma, and patient age was statistically significant (Table 2) when data were compared using the Wilcoxon s rank sum test. Tumor volume also was compared using Wilcoxon s rank sum testing and was statistically significant, with median volumes of follicular adenoma being smaller (5.95 ml; range, ml) than follicular carcinoma (11.75 ml; range, ml). Neither finding was surprising given the known risk factors for thyroid carcinoma. Male sex was significantly associated with follicular carcinoma, with 25 of the 50 cases (50%) found in males, an unexpected finding compared with previously described risk factors [7]. Among the control population, 10 (19.2%) of 52 follicular adenoma cases were seen in males, a similar proportion to the sex characteristics of the entire follicular adenoma population during the study period (131/579 [22.6%] of follicular adenoma cases were in males). A statistically significant difference (p < 0.05) was observed for the presence of a sonographic halo by chi-square analysis, TABLE 1: Frequency of Distribution of Clinical and Sonographic Variables for Each Subtype of Follicular Neoplasm Characteristic Follicular Adenoma (n = 52) Follicular Carcinoma (n = 50) Median age (y) (range) 46.5 (15 84) 59.5 (12 77) Median volume (ml) (range) 5.95 ( ) ( ) Male sex Sonographic halo 30 (57.7) 18 (36.0) Refractive shadowing 13 (25.0) 8 (16.0) Echotexture Heterogeneous 19 (36.5) 24 (48.0) Predominantly homogeneous 20 (38.5) 19 (38.0) Homogeneous 13 (25.0) 7 (14.0) Echogenicity Hypoechoic 26 (50.0) 41 (82.0) Isoechoic 24 (46.2) 8 (16.0) Hyperechoic 2 (3.8) 1 (2.0) Visible invasion 0 (0) 1 (2.0) Lymph node enlargement 0 (0) 3 (6.0) Other suspicious lesions 23 (44.2) 25 (50.0) Visible metastases 0 (0) 0 (0) Vascularity a Type 0 2 (4.2) 0 (0) Type 1 2 (4.2) 1 (2.1) Type 2 1 (2.0) 4 (8.3) Type 3 26 (54.2) 23 (47.9) Type 4 17 (35.4) 20 (41.7) Cystic change None 36 (69.2) 45 (90.0) Minimal (1 5%) 4 (7.7) 3 (6.0) Moderate (6 50%) 5 (9.6) 2 (4.0) Large (51 100%) 7 (13.5) 0 (0) Calcifications 3 (5.8) 7 (14.0) Note Except where otherwise indicated, data are number with percentage in parentheses. a Six cases (4 adenoma, 2 carcinoma) did not have vascularity recorded. 50 AJR:194, January 2010

8 Sonography of Thyroid Follicular Carcinoma TABLE 2: Univariate and Multivariate Assessment of Predictors of Histologic Malignancy Predictor OR (95% CI) p (Wald Statistic) Univariate logistic regression analysis Age ( ) Volume ( ) Male sex ( ) Absence of sonographic halo ( ) Refractive shadowing ( ) Echotexture ( ) Hypoechoic echogenicity ( ) Visible invasion of adjacent tissues NA Lymph node enlargement NA Other suspicious lesions within thyroid gland ( ) Vascularity ( ) Absence of cystic change ( ) Calcifications ( ) Multivariate logistic regression analysis Volume (1.001, 1.045) Hypoechoic echogenicity (0.258, 0.705) Absence of cystic change (0.049, 0.587) Note OR indicates odds ratio, NA indicates not available. with the presence of a halo associated with follicular adenoma in 57.7% of cases and with follicular carcinoma in 36% of cases. A halo is defined as a thin rim of decreased echogenicity surrounding the neoplasm and is thought to represent a capsule surrounding the mass. A thin hypoechoic halo was present in approximately 47% of the identified cases. Stated in another fashion, the absence of a halo surrounding a follicular neoplasm was more commonly associated with follicular carcinoma. Because there was a fairly uniform (and nonsignificant) distribution of the three echotextures, the homogeneous and predominantly homogeneous subgroups were combined and compared with the heterogeneous subgroup. No statistical association was observed with echotexture and the likelihood of follicular carcinoma. The presence of a hypoechoic pattern within the follicular neoplasm relative to the surrounding thyroid parenchyma was significantly associated with follicular carcinoma; 41 of the 50 cases of follicular carcinoma were hypoechoic compared with 26 of the 52 cases of follicular adenoma. A single case of visible invasion of adjacent tissues and only three cases of lymph node enlargement were observed in follicular carcinoma. The size of the sample population precluded adequate statistical analysis of these two subgroups, which yielded insignificant results. The distribution of the finding of other suspicious nodules in the thyroid gland was nearly equal between the two subgroups and not statistically significant. Metastasis, either adjacent to soft tissue or adjacent to lymph nodes, might be an expected finding in widely invasive follicular carcinoma. No cases of visible metastases were recorded at the time of data collection, nor were they mentioned in the corresponding surgical and radiologic reports at the time of diagnosis. The vascularity subtype was similar in distribution between the two subgroups, and not statistically significant. Two of the cases of follicular carcinoma and four of the cases of follicular adenoma did not have color Doppler images recorded at the time of imaging and were excluded from the analysis of vascularity. Because comparison of the subtypes of cystic change failed to reveal any statistically relevant association, the lesions showing any cystic change were combined, yielding a significant association between any degree of cystic change and a benign follicular adenoma. Put another way, the absence of cystic change within a follicular neoplasm was more common with follicular carcinoma than follicular adenoma. There was no statistically significant association between the presence of calcifications and the pathologic diagnosis. Follicular Carcinoma Subgroup Comparison (Follicular Carcinoma Hürthle-Cell Variant of Follicular Carcinoma) Because there was only a single case of the insular subtype of follicular carcinoma within the study population, this was excluded from subgroup analysis. Therefore, the follicular carcinoma subgroup comparison consisted of 27 cases characterized by histology as classic follicular carcinoma and 22 cases of Hürthle-cell variant of follicular carcinoma (n = 49) (Table 3). By univariate analysis, the Hürthle-cell subgroup was more likely to be older and have heterogeneous echotexture, whereas multivariate analysis combining age and the presence of calcifications showed an association between follicular carcinoma and the presence of calcifications within the lesion (Table 4). Five of the cases of follicular carcinoma were characterized by the pathologist as widely invasive. An attempt was made to differentiate this small population from the remainder of the subgroup, but because the population size was too small to make any valid statistical comparison with the data, the cases were compared with the minimally invasive subtypes by observation only. No difference was readily discernible between the subgroups. Discussion Follicular thyroid lesions can account for up to 29% of all fine-needle aspiration (FNA) cytologic diagnoses [28]. FNA of follicular neoplasms presents a challenge to the cytopathologist and can be considered a gray zone of thyroid FNA cytology. Included in this group are follicular adenomas and follicular carcinomas, and because of overlapping cytologic features, distinguishing benign from malignant follicular lesions is difficult. The risk of carcinoma on the basis of a cytologic diagnosis of follicular neoplasm was 10% in our study. As a result, many patients diagnosed with a follicular lesion by FNA undergo surgical excision for what is ultimately determined to be benign disease. Historically, alternative diagnostic imaging, including nuclear medicine, has not been of value in distinguishing benign and malignant nodules [29, 30]. In most clinical cases of suspected follicular neoplasm encountered in day-to-day practice, the sonographic features cannot confi- AJR:194, January

9 Sillery et al. TABLE 3: Frequency of Distribution of Clinical and Sonographic Variables for Each Subtype of Follicular Carcinoma Characteristic dently prospectively distinguish follicular carcinoma from follicular adenoma. In this current large series of follicular neoplasms, we attempted to determine the sonographic features that would distinguish benign from malignant neoplasms. Features that were associated with increased risk of a follicular neoplasm being a carcinoma were increasing lesion volume, lack of a sonographic halo, hypoechoic appearance, absence of internal cystic change, increasing patient age, and male sex. That patient age was observed to increase the risk of follicular carcinoma in this study was not surprising. It is well known that Classic Follicular Carcinoma Subtype (n = 27) Hürthle-Cell Variant of Follicular Carcinoma Subtype (n = 22) Median age (y) (range) 53 (12 75) 64.5 (41 77) Median volume (ml) (range) 9.0 ( ) ( ) Male sex Sonographic halo 13 (48.1) 5 (22.7) Refractive shadowing 6 (22.2) 2 (9.1) Echotexture Heterogeneous 9 (33.3) 14 (63.6) Predominantly homogeneous 13 (48.1) 6 (27.3) Homogeneous 5 (18.6) 2 (9.1) Echogenicity Hypoechoic 20 (74.1) 20 (90.9) Isoechoic 6 (22.2) 2 (9.1) Hyperechoic 1 (3.7) 0 (0) Visible invasion 0 (0) 1 (4.5) Lymph node enlargement 1 (3.7) 2 (9.1) Other suspicious lesions 14 (58.1) 10 (45.4) Visible metastases 0 (0) 0 (0) Vascularity a Type 0 0 (0) 0 (0) Type 1 0 (0) 1 (4.5) Type 2 1 (3.7) 2 (9.1) Type 3 15 (55.6) 8 (36.4) Type 4 11 (40.7) 9 (40.9) Cystic change None 24 (88.9) 20 (90.9) Minimal (1 5%) 2 (7.4) 1 (4.5) Moderate (6 50%) 1 (3.7) 1 (4.5) Large (51 100%) 0 (0) 0 (0) Calcifications 6 (22.2) 1 (4.5) Note Except where otherwise indicated, data are number with percentage in parentheses. a Two cases of Hürthle-cell carcinoma did not have any vascularity recorded. there are physiologic and structural alterations in almost all organ systems with aging. In addition, metabolic events induce genetic damage over the human lifespan, increasing the risk of chromosomal mutation and the induction of neoplasm. Tumor size recently has been described as a risk factor for follicular carcinoma [31, 32], with the large number of cases in our series supporting an association with greater tumor volume. In general, most benign tumors grow slowly over a period of years, and the same can be inferred about follicular adenoma. Most carcinomas, conversely, grow rapidly, and neoplasms in the thyroid are probably not an exception. Alternatively, however, follicular carcinoma may be derived from progressive dedifferentiation from a preexisting follicular adenoma, and therefore the increased tumor size could reflect the stepwise progression over considerable time as the tumor transforms from circumscribed adenoma to invasive carcinoma. Male sex was an unanticipated but not unexpected finding in our study because the finding of thyroid nodules in a male patient is known to be a more concerning finding than in a female patient [1] (in whom benign nodules are more frequent). The presence of a sonographic halo surrounding follicular neoplasms is a frequent finding and was more frequently associated with benign follicular neoplasia in our study (57.7% benign, 36.0% malignant). In all likelihood, the sonographic halo corresponds with the well-defined capsule that surrounds many follicular adenomas and whose continuity is important for the pathologist to describe as a feature of benignity. Hypoechoic lesions previously have been reported as a risk factor for malignancy in follicular neoplasms [33, 34], and this conclusion was substantiated in our study. Presumptively, the presence of decreased echogenicity (in the absence of cystic change) within a thyroid nodule implies that the follicular cells are undergoing rapid, disordered growth with a loss of the normal orderly arrangement of follicles in normal thyroid parenchyma, which renders the tissue less echogenic than adjacent normal parenchyma. The absence of internal cystic change can be hypothesized as secondary to the rapid proliferation of malignant cells, which do not undergo autolysis (and degenerative cystic change) as in a benign adenoma. To the best of our knowledge, our study is the first in the radiologic literature to describe and differentiate the findings in classic follicular carcinoma and Hürthle-cell variant follicular carcinoma. We found that classic follicular carcinoma is more likely in younger patients whose lesions had homogeneous echotexture and internal calcifications. It is widely acknowledged that FNA is not reliable in distinguishing benign from malignant follicular neoplasms. However, 18% of FNAs of follicular carcinoma in our series showed marked cellular atypia, a feature not seen with the follicular adenoma cases. The current standard practice of recommending resection of FNA-proven follicular neoplasms to exclude malignancy should con- 52 AJR:194, January 2010

10 Sonography of Thyroid Follicular Carcinoma TABLE 4: Univariate and Multivariate Assessment of Predictors of Follicular Carcinoma Subtype Predictor OR (95% CI) p (Wald Statistic) Univariate logistic regression analysis Age ( ) Volume ( ) Male sex ( ) Absence of sonographic halo ( ) Refractive shadowing ( ) Echotexture ( ) Echogenicity ( ) Visible invasion of adjacent tissues NA Lymph node enlargement ( ) Other suspicious lesions within thyroid gland ( ) Vascularity ( ) Absence of cystic change ( ) Calcifications ( ) Multivariate logistic regression analysis Age (0.836, 0.964) Calcifications (1.999, ) Note OR indicates odds ratio, NA indicates not available. tinue in most cases. The sonographic findings of hypoechoic echogenicity, lack of halo, lack of internal cystic change, and more often occurring in a larger tumor in an older male patient would all potentially point to a follicular carcinoma diagnosis and encourage a patient to undergo surgical resection. Conversely, the presence of a nodule with isoechoic or hyperechoic echogenicity, a well-defined halo, and small size in a young female patient might be reassuring findings and could perhaps permit clinical and sonographic follow-up when a patient would be reluctant to consider undergoing thyroid resection. References 1. Ries LAG, Eisner MP, Kosary CL, et al., eds. SEER cancer statistics review, , National Cancer Institute Website. seer.cancer.gov/ csr/1975_2002. Published November Accessed September 19, Jemal A, Siegel R, Ward E, et al. Cancer statistics, C A Cancer J Clin 2006; 56: Hedinger C, Williams ED, Sobin LH. The WHO histological classification of thyroid tumors: a commentary on the second edition. Cancer 1989; 63: Stolf BS, Santos MM, Simao DF, et al. Class distinction between follicular adenomas and follicular carcinomas of the thyroid gland on the basis of their signature expression. Cancer 2006; 106: Smith J, Cheifetz RE, Schneidereit N, Berean K, Thomson T. Can cytology accurately predict benign follicular nodules? Am J Surg 2005; 189: Carpi A, Nicolini A, Gross MD, et al. Controversies in diagnostic approaches to the indeterminate follicular thyroid nodule. Biomed Pharmacother 2005; 59: Goldstein RE, Netterville JL, Burkey B, Johnson JE. Implications of follicular neoplasms, atypia, and lesions suspicious for malignancy diagnosed by fine-needle aspiration of thyroid nodules. Ann Surg 2002; 235: LiVolsi VA, Asa SL. The demise of follicular carcinoma of the thyroid gland. Thyroid 1994; 4: Albores-Saavedra J, Carrick K. Where to set the threshold between well differentiated and poorly differentiated follicular carcinomas of the thyroid. Endocr Pathol 2004; 15: Giorgadze T, Rossi ED, Fadda G, Gupta PK, Livolsi VA, Baloch Z. Does the fine-needle aspiration diagnosis of Hürthle-cell neoplasm/follicular neoplasm with oncocytic features denote increased risk of malignancy? Diagn Cytopathol 2004; 31: Liska J, Altanerova V, Galbavy S, Stvrtina S, Brtko J. Thyroid tumors: histological classification and genetic factors involved in the development of thyroid cancer. Endocr Regul 2005; 39: Khafif A, Khafif RA, Attie JN. Hürthle cell carcinoma: a malignancy of low-grade potential. Head Neck 1999; 21: Bhattacharyya N. Survival and prognosis in Hürthle cell carcinoma of the thyroid gland. Arch Otolaryngol Head Neck Surg 2003; 129: Sanders LE, Silverman M. Follicular and Hürthle cell carcinoma: predicting outcome and directing therapy. Surgery 1998; 124: Gardner HA, Ducatman BS, Wang HH. Predictive value of fine-needle aspiration of the thyroid in the classification of follicular lesions. Cancer 1993; 71: Singer PA, Cooper DS, Daniels GH, et al. Treatment guidelines for patients with thyroid nodules and well-differentiated thyroid cancer. Arch Intern Med 1996; 156: Lupoli GA, Fonderico F, Colarusso S, et al. Current management of differentiated thyroid carcinoma. Med Sci Monit 2005; 11:RA368 RA Castro MR, Gharib H. Continuing controversies in the management of thyroid nodules. Ann Intern Med 2005; 142: Hay ID, Grant CS, Taylor WF, McConahey WM. Ipsilateral lobectomy versus bilateral lobar resection in papillary thyroid carcinoma: a retrospective analysis of surgical outcome using a novel prognostic scoring system. Surgery 1987; 102: Nix P, Nicolaides A, Coatesworth AP. Thyroid cancer review. Part 1. Presentation and investigation of thyroid cancer. Int J Clin Pract 2005; 59: Raber W, Kaserer K, Niederle B, Vierhapper H. Risk factors for malignancy of thyroid nodules initially identified as follicular neoplasia by fine-needle aspiration: results of a prospective study of one hundred twenty patients. Thyroid 2000; 10: Schlinkert RT, van Heerden JA, Goellner JR, et al. Factors that predict malignant thyroid lesions when fine-needle aspiration is suspicious for follicular neoplasm. Mayo Clin Proc 1997; 72: Tuttle RM, Lemar H, Burch HB. Clinical features associated with an increased risk of thyroid malignancy in patients with follicular neoplasia by fine-needle aspiration. Thyroid 1998; 8: Koike E, Noguchi S, Yamashita H, et al. Ultrasonographic characteristics of thyroid nodules: prediction of malignancy. Arch Surg 2001; 136: Rago T, Di Coscio G, Basolo F, et al. Combined clinical, thyroid ultrasound and cytological features help to predict thyroid malignancy in follicular and Hürthle cell thyroid lesions: results from a series of 505 consecutive patients. Clin Endocrinol (Oxf) 2007; 66: Lin JD, Hsueh C, Chao TC, Weng HF, Huang BY. Thyroid follicular neoplasms diagnosed by highresolution ultrasonography with fine-needle aspiration cytology. Acta Cytol 1997; 41: Frates MC, Benson CB, Doubilet PM, Cibas ES, Marqusee E. Can color Doppler sonography aid in the prediction of malignancy of thyroid nodules? AJR:194, January

11 Sillery et al. J Ultrasound Med 2003; 22: Weber AL, Randolph G, Aksoy FG. Radiologic evaluation of the neck: the thyroid and parathyroid glands: CT and MR imaging and correlation with pathology and clinical findings. Radiol Clin North Am 2000; 38: Ashcraft MW, Van Herle AJ. Management of thyroid nodules. Part I. History and physical examination, blood tests, x-ray tests, and ultrasonography. Head Neck Surg 1981; 3: FOR YOUR INFORMATION 30. Ashcraft MW, Van Herle AJ. Management of thyroid nodules. Part II. Scanning techniques, thyroid suppressive therapy, and fine-needle aspiration. Head Neck Surg 1981; 3: Raparia K, Min SK, Mody DR, Anton R, Amrikachi M. Clinical outcomes for suspicious category in thyroid fine-needle aspiration: patient s sex and nodule size are possible predictors of malignancy. Arch Pathol Lab Med 2009; 133: Mark your calendar for the following ARRS annual meetings: May 2 7, 2010 Manchester Grand Hyatt San Diego, San Diego, CA May 1 6, 2011 Hyatt Regency Chicago, Chicago, IL April 29 May 4, 2012 Vancouver Convention Center, Vancouver, BC, Canada April 14 April 19, 2013 Marriott Wardman Park Hotel, Washington, DC 32. Mihai R, Parker AJ, Roskell D, Sadler GP. One in four patients with follicular thyroid cytology (THY3) has a thyroid carcinoma. Thyroid 2009; 19: Blankenship DR, Chin E, Terris DJ. Contemporary management of thyroid cancer. Am J Otolaryngol 2005; 26: Carling T, Udelsman R. Follicular neoplasms of the thyroid: what to recommend. Thyroid 2005; 15: AJR:194, January 2010