Impact of Microcalcifications on Risk of Malignancy in Thyroid Nodules with Indeterminate or Benign Cytology

Methods

The present study comprised a retrospective review of all cases of US-FNA performed at our institution over a 5-year period, between January 2009 and December 2013. Cases were extracted by review of procedure logs at the radiology department at the South Infirmary Victoria University Hospital and cross-checked with a database of all thyroid FNAs sent to the Department of Cytopathology at Cork University Hospital. Inclusion criterion was any US-FNA of a thyroid nodule during the study period. Cases were excluded where FNA was taken from diffuse enlargement of a thyroid lobe, without a discrete nodule visualized on US. Core needle biopsies were also excluded.

Nearly all US-FNAs were performed by 1 of 3 radiologists, with the majority performed by 1 of these. US was performed with a Siemens Acuscon Antares machine and a linear 13.5-MHz probe. Sonographic features reported included nodule size, consistency, irregularity of margins, presence of coarse calcifications or microcalcifications, and presence of other nodules. In most cases, a laboratory technician was present to assess sample adequacy and to optimally prepare the sample. After multiple passes through the nodule, aspirated material was spread onto glass slides and air-dried. In addition, needle rinse material was sent in Cytolyt solution.

Cytology was reported according to the British Royal College of Pathologists Thy classification system, ²⁵ which is analogous to the Bethesda ²⁶ reporting system. For the purpose of this study, cytology was reclassified according to the following categories: inadequate (corresponding to Bethesda category I), benign (Bethesda II), indeterminate (Bethesda III and IV; including cases suspicious for follicular neoplasm, AUS, or follicular lesion of undetermined significance [FLUS]), and malignant (Bethesda V and VI; including cases suspicious for and positive for malignancy).

The predictive value of sonographic features was correlated with malignancy at final histology for the overall group and separately for the benign and indeterminate cytology categories. For analysis according to cytology category, 2 analyses were performed: 1 according to initial cytology diagnosis and 1 according to the final cytology diagnosis. The final cytology diagnosis was considered to the highest-risk cytology diagnosis from all US-FNAs in nodules undergoing >1 US-FNA, discounting aspirates with inadequate cytology.

Cases undergoing surgery were considered to be malignant if cancer was found within the index (biopsied) nodule by histologic examination. For the purposes of the study, cases of incidental cancer occurring outside the index nodule were considered benign.

Statistical analysis was performed by XLSTAT (version 2013.5.05; Addinsoft, New York, New York). A Student’s t test was used for comparison of 2 groups of normally distributed data. A Fisher’s exact test was performed on 2 × 2 contingency tables. Multivariate analysis was performed with logistic regression analysis.

Ethical approval for the study was obtained by the Cork Clinical Research Ethics Committee.

Results

During the study period, 780 consecutive patients underwent 1159 US-FNAs of the thyroid. Seventeen FNAs taken from 13 diffusely enlarged thyroid lobes without a discrete nodule were excluded. Thus, the final study population consisted of 858 thyroid nodules in 769 patients undergoing 1142 US-FNAs. Mean nodule size was 28.1 ± 15.0 mm; 167 were solid (19.5%). Microcalcification was present in 71 nodules (8.3%) and coarse calcifications in 50 (5.8%). Nineteen nodules (2.2%) had irregular or poorly defined margins.

Initial cytology diagnosis was benign in 518 cases, indeterminate in 210, and malignant in 44; 158 nodules with initial benign cytology underwent repeat US-FNA. In 32 cases, this led to upgrading of the initial benign cytologic diagnosis to a higher-risk category (indeterminate in 31, malignant in 1). In the other 126 cases, repeat FNA was also benign. Forty nodules with initial indeterminate cytology underwent repeat US-FNA. Results of repeat cytology were benign in 22 and indeterminate in 18. For the purpose of the study, the final cytologic diagnosis in these cases was considered indeterminate, being the higher-risk cytology category. Therefore, final cytology diagnosis was benign in 486 cases (56.7%), indeterminate in 241 (28.1%), malignant in 45 (5.2%), and inadequate in 86 (10.0%).

Sonographic Predictors of Malignancy among Patients Undergoing Surgery

Surgery was performed in 411 cases (47.9%): 170 with initial benign cytology, 161 with indeterminate, 39 with malignant, and 41 with inadequate, of which 11 had undergone ≥2 inadequate US-FNAs. Histology showed malignancy within the index nodule in 78 cases (19.0%). Histologies of cancers were papillary (63: 35 conventional variant, 24 follicular variant, and 4 other variants), follicular (10, including 3 oncocytic variant), medullary (1), poorly differentiated (2), anaplastic (1), and metastatic adenocarcinoma to the thyroid (1). In a further 33 cases, incidental cancer was found outside the index nodule. For the purposes of this study, these cases were considered benign.

Among cases undergoing surgery, 43 nodules had sonographic microcalcifications, of which 18 transpired to be malignant, and 8 cases had irregular margins, of which 4 transpired to be malignant. There was a significant association between final pathologic diagnosis of malignancy and microcalcification (18 of 43 vs 60 of 368; P = .003) and irregular margins (4 of 8 vs 74 of 403; P = .045). Coarse calcifications (P = .44) and solid consistency (P = .31) were not significant ( Table 1 ). Mean nodule size of cases with a final diagnosis of cancer was significantly less than that of benign nodules (29.8 vs 34.4 mm; P = .02).

OPEN IN VIEWER

Table 1.

Sonographic Features According to Final Nodule Histology.

	Malignant Histology (n = 78)	Benign Histology (n = 333)	No Surgery, Repeat FNA a (n = 96)	No Follow-up Surgery or FNA	Total
Microcalcifications	18	25	8	20	71
Coarse calcifications	3	23	7	17	50
Irregular margins	4	4	2	9	19
Solid	23	78	17	49	167

Abbreviation: FNA, fine-needle aspiration.

a

With benign cytology.

Multivariate analysis of predictive factors for malignant histology was performed with input variables of microcalcification and irregular margins. Microcalcifications (OR, 3.73; 95% CI, 1.91-7.30; P < .001) and irregular margins (OR, 4.61; 95% CI, 1.09-19.43; P = .04) were both significant.

Nodules with Benign Cytology

Among 170 cases with initial benign cytology undergoing surgery (including 18 cases where initial benign cytology was upgraded to a higher-risk category by repeat FNA), there were 12 cancers (7.1%). When cases with repeat FNA showing higher-risk cytology were excluded, the false-negative rate of benign final cytology was 5.2% (8 of 152). Histologies of cancers in patients with benign initial cytology included follicular variant papillary carcinoma (FVPC; 8), conventional papillary carcinoma (2), and follicular carcinoma (2).

Among cases with initial benign cytology, 33 had microcalcifications. Among the 170 cases undergoing surgery, 13 had microcalcifications, and 157 did not. The incidence of malignancy was 15.4% (2 of 13) for nodules with microcalcification versus 6.4% (10 of 157) in those without (P = .23). None of the 18 cases with upgraded cytologic diagnosis after initial benign US-FNA had microcalcifications. Therefore, even when cases with upgraded cytologic diagnosis were excluded, the difference in malignancy rates between cases with and without microcalcifications was again not significant (2 of 13 vs 6 of 139; P = .14). These data are also shown in Table 2 , in which cases with changed cytologic diagnosis from initial benign cytology after repeat US-FNA are tabulated in separate rows.

OPEN IN VIEWER

Table 2.

Sonographic Features According to Cytologic Category and Final Histology.

		Calcifications
Cytology	Undergoing Surgery, n	Micro	Coarse	Irregular Margins	Solid
Inadequate (n = 87)	41
Benign	33	5	1	0	5
Malignant	8	2	0	1	2
Benign a (n = 486)	152
Benign	144	11	11	2	31
Malignant	8	2	0	0	2
Indeterminate a (n = 210)	161
Benign	135	8	8	2	37
Malignant	26	5	1	0	12
Malignant a (n = 44)	39
Benign	7	1	0	0	3
Malignant	32	9	3	3	9
Changed cytologic diagnosis	18
Benign	14	0	2	0	2
Malignant	4	0	0	0	0

Abbreviations: FNA, fine-needle aspiration; US, ultrasonography.

a

Cases where cytologic diagnosis changed after repeat US-FNA are excluded from the benign, indeterminate, and malignant categories and shown in separate rows.

There were 10 thyroid nodules with initial benign cytology with irregular sonographic margins. Two underwent surgery, and both transpired to be benign. Two further cases underwent follow-up US-FNA with benign cytology. Due to small number of cases, statistical analysis of impact of irregular margins on malignancy rates was not performed.

Discussion

While cytology remains the diagnostic test of choice for thyroid nodules, recent improvements in US have led to increasing interest in sonographic features as predictors of malignancy. In this regard, several features have been shown to predict malignancy risk—including microcalcifications,^3,16-19 hypoechogenicity,^3,16,17,19 irregular margins,^3,16-18 and taller-than-wide shape.^17,19 The major drawback of sonographic predictors is that the sensitivity of any single sonographic feature is low; therefore, sonographic assessment cannot replace the need for cytology. Nevertheless, sonographic features may be useful in selecting nodules for US-FNA. ²⁷ Recently, there is increasing interest in determining whether sonographic features may predict malignancy risk in thyroid nodules with indeterminate cytology. Currently, most of these cases are recommended for surgical excision for definitive diagnosis or for follow-up with repeat US-FNA.^9,27

In the present study, microcalcifications and irregular margins were highly predictive of malignancy in the overall cohort. There was a significant association between microcalcification and malignancy among nodules with indeterminate cytology but not among nodules with benign cytology. The sensitivity and PPV of microcalcifications for malignancy was low across the entire group. However, the specificity of microcalcifications was high across the entire group.

The association between microcalcification and malignancy in thyroid nodules with indeterminate cytology has been reported by other authors. Lee ²¹ reported microcalcification to be predictive of malignancy among thyroid nodules with initial FNA diagnosis of AUS, while Yoo ²² reported microcalcification to be predictive of malignancy among nodules with initial FNA diagnosis of AUS or FLUS. Similar findings were also reported by Gweon. ¹⁵ An important difference between these studies and ours was that these studies were limited to patients with AUS/FLUS, whereas indeterminate cytology in our study also included cases suspicious for follicular neoplasm. Including all cases with indeterminate cytology in a single group is a potential weakness of our study. However, series that are limited to AUS/FLUS may be subject to bias due to interinstitutional variation in the use and interpretation of this category.^13,28 This is illustrated by the high malignancy rates in the studies of Yoo and Lee (33.7%-40.2%),^21,22 which are higher than the advised malignancy rate of 5% to 10% as recommended by the National Cancer Institute for this category ⁹ and much higher than the malignancy rate of 16.3% for cases with indeterminate cytology in the present study. The malignancy rate among surgical cases reported by Gweon was even higher (81.9%). ¹⁵

The false-negative rate of benign FNA for thyroid malignancy in surgical series is reported to be between 2% and 10%,^2,3,6-8,23 but rates as high as 21% have been reported. ²⁹ Lower false-negative rates (0.3%-1%) are reported in series where cases with benign cytology not undergoing surgery are presumed to be benign.^30,31 However, the latter method of calculation may be flawed due to the slow growth rate of many thyroid cancers such that lack of growth even after a significant period of observation does not exclude cancer. ¹⁰ In the present study, we reported a false-negative rate based only on cases with surgical correlation of 5.2%. When analyzed according to initial FNA results, the false-negative rate was 7.1%. This not-insignificant false-negative rate suggests that cytology is not gold standard in excluding malignancy and that even nodules with benign cytology warrant some form of follow-up. Due to the very large volume of thyroid nodules with benign cytology, identification of risk factors for malignancy that might target cases that need further follow-up would be of clear benefit. Kwak et al reported the presence of a single adverse sonographic feature to be associated with increased risk of malignancy in nodules with benign initial FNA. ²⁴ In contrast, Wharry et al did not find sonographic features to discriminate between benign and malignant nodules >4 cm in size. ²³

In the present study, although there was a somewhat higher incidence of malignancy among cases with benign cytology that underwent surgery and that had microcalcifications than among cases without microcalcification, the difference was not significant. However, the number of cases with benign cytology and microcalcifications undergoing surgery was low. In consideration that the expected likelihood of malignancy in cases with benign cytology is low, it is thus likely that our study may have been underpowered to find a significant impact on malignancy risk for microcalcifications in this group. Furthermore, the majority of thyroid nodules with benign cytology and microcalcifications (20 of 33) did not undergo surgery, with a high proportion of these (14) not even undergoing follow-up FNA. These observations raise the question of whether we should have been more rigorous in arranging follow-up FNA or even in considering surgery in cases with benign initial cytology but with adverse radiologic features. Clearly, the management of thyroid nodules with benign cytology and suspicious sonographic features is a topic that warrants further study.

Another possible explanation for failing to find a significant association between microcalcifications and malignancy among cases with benign cytology is the high proportion of FVPC in this group. In the present study, two-thirds of cancers occurring in patients with initial benign cytology were FVPC. We previously reported FVPC to be more likely to show lower-risk preoperative cytology than that of conventional papillary carcinoma.^2,32 Recently, Kim et al reported microcalcification and other adverse sonographic features to be less common in FVPC than in conventional papillary carcinoma. ³³ Thus, it is possible that the type of malignancy that may be present in thyroid nodules with benign cytology may also be one that is less likely to display sonographic features of malignancy also. More data are awaited to confirm this.

Among the weaknesses of our study were its retrospective nature and the noninclusion of other sonographic features predictive of malignancy, such as marked hypoechogenicity, taller-than-wide shape, and internal vascularity, as these findings were not always consistently documented in US reports. In addition, we did not subdivide cases with indeterminate cytology into AUS/FLUS cases (Bethesda III) and follicular neoplasm cases (Bethesda IV), due to our systematically introducing this subdivision only in 2011. However, having a single indeterminate category may have minimized bias due to variability of usage and interpretation of the AUS and FLUS categories. Finally, it is possible that the inclusion of histologies other than conventional papillary carcinoma may have lowered the sensitivity of microcalcifications. However, as histologic type of malignancy is determined only after surgery, we believed it appropriate to include all histology types in the analysis. However, strengths include the analysis of impact of microcalcification on different cytology categories separately and inclusion in the main analysis only cases with surgical correlation, thereby minimizing bias from inclusion of cases without histologic confirmation of benignity.

Conclusion

Sonographic microcalcifications may help predict for malignancy in thyroid nodules. In particular, among nodules with indeterminate cytology, the presence of microcalcifications may be useful in identification of cases at increased risk of malignancy and thus in selection of cases for surgery. The predictive value of microcalcifications among cases with benign cytology in the present study was not significant; however, most patients with benign cytology did not undergo surgery, and many did not undergo follow-up FNA. Thus, we cannot discount the possibility that a higher malignancy rate may have been identified had such cases been subject to more rigorous follow-up.

Author Contributions

Catherine Brophy, data acquisition; drafting; final approval; accountability; James Stewart, data acquisition; drafting; final approval; accountability; Neil O’Donovan, data acquisition; critical revision; final approval; accountability; Julie McCarthy, data acquisition; critical revision; final approval; accountability; Matthew Murphy, concept and design; critical revision; final approval; accountability; Patrick Sheahan, concept and design; drafting; final approval; accountability.

Disclosures

Competing interests: None.

Sponsorships: None.

Funding source: None.