Introduction

Thyroid nodules are present in up to 50% of the adult population and prevalence increases with age [1]. The incidence of thyroid nodule detection in the general population is even higher according to studies based on thyroid ultrasonography (US) (17–67%) and in autopsy studies (50.5%) [2, 3]. It has been reported that nodules as small as 1–1.5 cm in diameter can be palpated when favorably situated [4, 5]; it is more difficult to palpate a nodule in the posterior portion of the thyroid gland than in the anterior [6]. Other factors, such as the shape of the patient’s neck, also affect detection by palpation.

Detection of nonpalpable thyroid nodules has increased with wider application of US and various other imaging modalities. It has not been established whether a size criterion or special US criteria are needed as guidelines for indicating fine-needle aspiration biopsy (FNAB) of infracentimetric nodules. There is also no consensus on how nonpalpable thyroid nodules should be managed, and contradictory approaches are currently in use. Some clinicians are in favor of US-guided FNAB for nonpalpable thyroid nodules [7–9]. Others propose clinical 6-month or annual follow-up with neck palpation in low-risk patients, if the patient has no family history of thyroid malignancy, head and neck radiation, stable nodule size, or absence of US findings suggesting malignancy [2].

In this study, we examined US features and FNAB results of infracentimetric thyroid nodules for a relatively larger sample group than prior studies—a series of 589 nodules. We compared the cytological findings to the nodules’ US features to assess their value in detecting thyroid malignancy presenting as an infracentimetric nodule, and thus their value for indicating FNAB. We also evaluated the post-thyroidectomy histological findings for the 23 patients who underwent thyroid surgery during follow-up to assess the extent of disease and their relationship with US findings.

Patients and methods

Patients

The study comprised 589 infracentimetric thyroid nodules in 485 (mean age: 46.2±14.9 years, age range: 17–79 years, 87 male, 398 female) patients who underwent FNAB in our interventional radiology department between January 2000 and December 2005. Patients were excluded if they had received radiation therapy to the head and neck region or if there had been a family history of thyroid malignancy.

Methods

US examination of the thyroid and the cervical region was performed using a real-time US scanner with a 7.5-MHz linear transducer (Hitachi EUB-525) by three interventional radiologists. Nodules with maximum dimension less than or equal to 10 mm were included in the study. The data recorded for each patient were number (solitary or multiple) and size of nodules, the echo structure (solid or cystic), echogenicity (hypoechoic, hyperechoic, isoechoic, or mixed), presence or absence of calcification (structure with strong echo and acoustic shadowing), margin character (well-defined or ill-defined), presence or absence of lymphadenopathy (any lymph node with a dimension over 10 mm in short axis) in the cervical region, and size and number (single or multiple) of any accompanying lymphadenopathy.

US-guided FNABs were performed by three different interventional radiologists using a 21-G needle on a 20-ml syringe after informed consent was obtained. Samples were taken from all the nodules detected on US. No criteria for selection or exclusion of a nodule, such as size, were applied. Each nodule was biopsied with two to three needle passes within to collect an aspirate specimen per lesion (Fig. 1). In partially cystic nodules, the needle was directed to the solid region. Immediately after the procedure, the samples were smeared to slides, then fixed in ethanol, and stained with May–Grünwald-Giemsa and hematoxylin-eosin stains.

Fig. 1
figure 1

US-guided FNAB: 21-G needle on a 20-ml syringe was inserted into the nodule. Samples were taken from the nodule with two to three needle passes within the nodule under US guidance

Full size image

Cytological analysis was carried out by two experienced pathologists and the specimens were classified as benign, malignant (papillary carcinoma, medullary carcinoma, anaplastic carcinoma), suspicious for malignancy (papillary carcinoma, medullary carcinoma, papillary carcinoma), follicular neoplasm (follicular adenoma, follicular carcinoma, suspicious for follicular carcinoma), or inadequate cytological specimen. A specimen was considered adequate if there were six or more thyroid cell clusters on each slide. A nodule was considered benign if the specimen exhibited normal thyroid epithelial cells or thyroiditis. A follicular neoplasm was considered separately from the malignant group because it is difficult to differentiate between follicular adenoma and follicular carcinoma; therefore, the nodules classified as malignant excluded follicular carcinomas. Smears showing high cellularity, atypical follicular cells, large nuclei, microacinar formation, and no colloid were considered suspicious for malignancy.

Nine patients underwent thyroidectomy and lymph node dissection for miscellaneous reasons, in addition to 11 patients of the 13 who had malignant cytology and 3 patients of the 10 who had suspicious malignant cytology. Ten patients with a final histological diagnosis of papillary carcinoma were evaluated for extrathyroidal invasion, presence of metastatic lymph nodes, and multifocality. Statistical analysis was performed using the SPSS software package (SPSS Inc., Chicago, IL, USA). The risk of malignancy for each US feature was calculated using logistic regression analysis. For each US feature studied, the number of malignant nodules with that feature was recorded. Then, the frequency distribution of thyroid malignancy in different groups of US features was compared using the chi-square and Fisher exact tests. P<0.05 was considered significant.

Results

Cytological findings

The 589 nodules were cytologically classified as follows: 473 (80.3%) as benign [443 (75.2%) with benign epithelial cells and 30 (5.1%) with signs of thyroiditis], 13 (2.2%) as malignant, 13 (2.2%) as suspicious for malignancy, 4 (0.7%) as follicular neoplasms, and 86 (14.6%) as inadequate cytological material. All 86 nodules with inadequate cytological material were excluded from the study.

FNAB was repeated for reevaluation of 80 nodules in 66 (13.6%) of the 485 patients, for either follow-up (49 patients) or reassessment after inadequate cytological material collection (17 patients with initial cytological results used for statistical purposes). Of the 80 repeat FNABs, 46 (57.5%) yielded inadequate specimens, 28 (35%) were benign [18 (22.5% of the 80) with benign epithelial cells and 10 (12.5% of the 80) with signs of thyroiditis], 3 (3.8%) were suspicious for malignancy, 1 (1.3%) was malignant, and 2 (2.5%) were follicular neoplastic lesions.

US findings

The mean size of the 503 infracentimetric nodules with adequate cytological specimen was 8.5 mm (range: 3–10 mm). The mean size for benign nodules was 8.5 mm (range: 3–10 mm), malignant nodules was 7.9 mm (range: 5–10 mm), nodules suspicious for malignancy was 8.4 mm (range: 3.5–10 mm), and follicular neoplasms was 9.2 mm (range: 8–10 mm). In the infracentimetric group, analysis revealed that neither the size of the nodules nor the age and gender had any significant effect on malignancy.

The US features of the 503 infracentimetric nodules with adequate cytological specimens were as follows: 129 (25.6%) were solitary whereas 374 (74.4%) were embedded in a multinodular goiter; 463 (92%) were solid whereas 40 (8%) had cystic components; 288 (57.3%) were hypoechoic, 84 (16.7%) were of mixed echogenicity, 71 (14.1%) were isoechoic, and the remaining 60 (11.9%) were hyperechoic; 116 (23.1%) contained coarse calcifications with acoustic shadowing; 26 (5.2%) had irregular margins; and 59 (11.7%) had associated lymphadenopathy [11 (2.2%) of the nodules were accompanied by a solitary lymph node while 48 (9.5%) were accompanied by multiple lymph nodes]. The cytological results and US findings for the nodules are given in Table 1.

Table 1 Characteristics of the patients and cytological results of thyroid nodules, with distribution of their US features
Full size table

In our study, in agreement with previous studies, solidity, hypoechogenicity, solitariness, calcifications, and ill-defined margins were presumed to be positive signs for malignancy [5, 10, 11]. We investigated the presence of lymphadenopathy in the cervical region accompanying thyroid nodules as another possible risk factor as Slapa and colleagues observed the presence of nodal involvement in nonpalpable nodules in two cases [12].

Among single US features only hypoechogenicity (P<0.004) and the presence of lymphadenopathy (P<0.001, odds ratio: 7.067) were significant markers in finding malignancies. Irregular margination and the presence of calcifications were not significant enough as independent features in assessment of malignant cytology. Solidity was not significant by itself, but considering that all of the malignant nodules were solid and hypoechoic when the two features are present in a nodule together, although they are not specific enough, they are significant (P<0.002). When other combinations of certain US features were assessed together, specificity and positive predictive value for malignant cytology was increased especially in the presence of lymph nodes. Solitary thyroid nodules with accompanying lymph nodes had the highest positive predictive value (PPV: 36.3, P<0.0001), detecting four of the malignant nodules. The solid hypoechoic thyroid nodules that were accompanied by lymphadenopathy (P<0.0001) or solitary, hypoechoic solid nodules (P<0.002) were significant. There was no significant difference when thyroid nodules that accompanied a solitary lymph node were compared with thyroid nodules that accompanied multiple lymph nodes (i.e., the number of lymph nodes had no significant meaning). Sensitivity, specificity, predictive value, and relative risk factor in detecting malignant cytology according to US findings are given in Table 2.

Table 2 Assessment of ultrasound findings for detection of malignant cytology. NA not applicable, NS not significant, OR odds ratio, PPV positive predictive value
Full size table

Surgery group—final histological results

Of the 485 patients, 23 (4.7%) underwent thyroidectomy and lymph node dissection during follow-up. Of these 23, 11 of them had nodules with malignant cytology and 3 of them had nodules with suspicious malignant cytology. Nine patients underwent thyroidectomy due to other miscellaneous reasons (discomfort, enlarged goiter, follicular neoplastic cytology, and others). Two patients with malignant nodules were lost to follow-up because they were admitted to other centers for treatment. Nodules of the 10 of the 11 patients with malignant cytology received a final histological diagnosis of papillary carcinoma while the remaining 1 was diagnosed to be thyroid hyperplasia.

Of the 13 patients who had suspicious malignant cytology, 3 underwent surgery. Two of the three nodules with suspicious malignant cytology were definitively diagnosed as follicular microadenomas and one was identified as nodular hyperplasia. Of the remaining ten patients, three of them received repeat FNABs and four of them were lost to follow-up because they were admitted to other centers for further evaluation.

In seven of the nine patients who had thyroidectomy for other miscellaneous reasons, the thyroid abnormality was identified as nodular hyperplasia. Of the remaining two individuals, one had colloidal goiter and one had follicular adenoma.

None of the nodules of the 23 operated patients was a follicular, anaplastic, or medullary carcinoma.

Patients with a final diagnosis of papillary carcinoma (n=10) were further evaluated for extracapsular extension, involvement of lymph nodes, and multifocality for better evaluation of clinical relevance. Nodal involvement was present in two patients (20%). Four (40%) of the carcinomas were multifocal and two (20%) contained extracapsular growth. No distant metastases were detected by 131I body scan following surgery.

Discussion

The malignancy rates of nonpalpable thyroid nodules, the indolent course of thyroid microcarcinomas, and the frequent detection of occult thyroid cancer in autopsy series raise questions of clinical significance and cost-effectiveness of management [13, 14]. As a natural result of US being widely ordered by endocrinologists, which is gradually becoming an extension of the clinical examination of the thyroid [15, 16], as well as other imaging studies ordered for non-thyroidal purposes (computed tomography and magnetic resonance imaging studies of the head and neck region, US of the parathyroid glands and cervical lymph nodes, and Doppler US of the carotid arteries), nonpalpable and infracentimetric nodules are being detected more often. The important issue in these nonpalpable thyroid nodules is that there is no consensus on their management.

In our series we found a cytological malignancy rate of 2.6% (13/503) within the nodules and a rate of histologically proven malignancy of 2.1% (10/485), which is not high when compared to other series [5, 10, 11]. Postoperative histological findings clearly indicate to us that some of these malignancies are of biological importance. Seventy percent (7/10) of the patients either had one or more of the US features of multifocality, metastatic lymph node presence, or extracapsular involvement. The existing data on the natural course of thyroid microcarcinomas are very contradictory—some series report that they have an indolent course [17, 18] and others report that they have clinical relevance in terms of extrathyroidal invasion, lymph node involvement, and distant metastases [19–21]. We probably would have better insight into this issue if we had a larger series subject to surgery or if all of our patients with suspicious or malignant cytology were operated. Nevertheless, from our present results, it is clear that not all of the thyroid microcarcinomas have an indolent course.

Nam-Goong et al. (2004) analyzed the findings of 317 nodules that were nonpalpable and ranged in size from 0.2 to 1.5 cm and found the rate of malignant cytology to be 13% (42/317) [10]. Papini et al. (2002) found a rate of 7% [11] and Leenhardt et al. (1999) a rate of 5.2% [11] of malignant cytology in infracentimetric nodules. The differences between these studies and ours could be due to the fact that we did not categorize nodules with suspicious malignant cytology and follicular neoplasms as malignant, and besides it should be noted that the criteria for admission and pathological assessment used in these different studies are very heterogeneous. However, it is striking that their post-operative histological findings are in agreement with our results. Nam-Goong et al. reported that, of the 36 patients who had postoperatively confirmed malignancies, 25 (69%) had either extrathyroidal extension or regional lymph node involvement and 39% had multifocal tumors [10]. Papini et al. reported that 35.5% of the malignancies were accompanied by extrathyroidal extension and 19.4% by regional lymph node involvement in a group of 31 patients [5].

Several US features have been used in the past to increase the number of malignancies detected by FNAB and in line with previous studies; we have considered and evaluated solidity, hypoechogenicity, solitary nature, calcification, and ill-defined margins as signs of malignancy [5, 10, 11]. However, in our series, only hypoechogenicity was a significant single factor in assessing malignancy; but on the other hand, it was not specific enough.

We have used here a new US criterion for detection, the presence of accompanying lymphadenopathy. It is questionable whether it is cost effective and worth the investment of time to examine the neck area for cervical lymph nodes of every patient by US presenting only with thyroidal symptoms, but this proved to be the most independently significant factor that we assessed. The probability of coincidentally detecting lymphadenopathy due to non-thyroid-related causes could be of concern as four patients with thyroid malignancy in our study had lymphadenopathy with benign pathology, but it should be noted that two metastatic lymph nodes were detected in our study as a result of the post-operative pathological examination.

According to the previous studies there is no single US feature that is highly predictive for malignancy, and so a combination of features should be used [22]. Nam-Goong et al. found solidity, hypoechogenicity, and punctuate calcification on US to be significantly related to malignancy [10]. Papini et al. reported that irregular nodule margins, intranodular vascular spots, and microcalcifications were associated with malignancy [5]. Leenhardt et al. found the same relationship in solid and hypoechoic nodules [11]. In our study, leaving lymphadenopathy presence and hypoechogenicity aside, other features required a combination to be significantly associated with malignancy. If we had performed FNABs only on hypoechoic and solid nodules, all of the malignant nodules would be detected. It would help us to decrease the number of unnecessary FNABs in management, but this would only help us to narrow the group approximately to half (275/503). If additional criteria to a basis of hypoechoic and solid nodules were to be used–such as the presence of calcifications or the presence of lymphadenopathy–8 of the 13 malignancies would be detected, seriously narrowing the group (82/503, odds ratio: 8.995, P<0.0001) (Fig. 2). We believe that we have identified a new US parameter, presence of lymphadenopathy, which could be used to reinforce indications for interventional procedures carried out.

Fig. 2
figure 2

Histologically proved papillary carcinoma of the subcentimetric (9 mm) thyroid nodule. a Solitary hypoechoic nodule with ill-defined borders. Note the punctuate calcification and (b) accompanying lymph node, increasing the suspicion of malignancy

Full size image

In our study, there was a high incidence of inadequate specimen collection (57.5%) in repeat FNABs. This was possibly due to the smaller size of the nodules needing repeat FNABs (mean: 4.4 mm, range: 2.5–6 mm, while the mean size of the nodules with adequate cytologic material was 8.5 mm)

A major deficit of our study is that the population was a referred sample, so it cannot necessarily be applied to the general population. It is not possible to evaluate the exact nature and prognosis of thyroid microcarcinomas as it would not be ethical to conduct a randomized prospective trial of observation vs standard thyroid cancer care, as discussed by Topliss in his commentary [23]. With the data that we present here it would not be rational to offer observation of a nodule that is hypoechoic, solitary, and solid or with any other significant accompanying US features, based only on size. The problem lies in limiting the pool of people receiving FNAB.

Another shortage of our study is that there were few malignant nodules in the study. Statistics based on a higher number of malignancies would have been more powerful in detecting the significance of the US features of possible malignancy.

To conclude, it is not feasible to burden the health system and economy by scanning every single thyroid nodule as thyroid cancers do not have a high incidence or high malignancy rates (which, according to the American Cancer Society, represents 1.5% of all new cancers and represented 0.23% of all cancer deaths in 2001) [24]. Overall, the low malignancy rate of subcentimetric nodules in this study confirms this conclusion. Additional studies are necessary to establish an appropriate strategy for managing nonpalpable thyroid nodules, keeping in mind their biological significance but not turning them into an epidemic.