Indeterminate Single Thyroid Nodule: Synergistic Impact of Mutational Markers and Sonographic Features in Triaging Patients to Appropriate Surgery

Abstract

Background:

Patients labeled as having indeterminate thyroid nodular disease following fine-needle aspiration cytology are at risk of non-optimal initial surgery: an overly radical total thyroidectomy, or an unnecessary two-stage operation. The objective of this study was to assess the impact of combining mutational markers and ultrasonographic (US) features preoperatively on predicting the risk of malignancy in patients with indeterminate nodules, thereby offering them a tailored initial surgical intervention.

Methods:

The records of 258 patients who underwent conventional total thyroidectomy for single nodules reported as suspicious for a follicular neoplasm (Bethesda category IV) in a four-year period were reviewed. Main issues addressed included: certain US findings (individually and in combination), mutational markers (BRAF and NRAS), and combinations of both. Correlation of these with malignancy was assessed, as was their ability to predict malignancy. The usefulness of combining the absence of suspicious sonographic features and the absence of mutational markers was also evaluated.

Results:

Among the 258 patients with an indeterminate diagnosis, only 90 lesions were found to be malignant. The sonographic features that correlated significantly with malignancy were irregular margins, microcalcifications, and a “taller than wide” shape. The presence of irregular margins was the feature with the highest positive predictive value. Combinations of two or more features were always associated with predictivity in excess of 90%, and at times at 100%. NRAS mutation was the most common gene alteration. Both BRAF and NRAS mutations were mutually exclusive and correlated significantly with malignancy. Their predictivity of malignancy was high, particularly when combined with suspicious sonographic features (100%). The major limitation of both suspicious sonographic features and/or mutational markers was their low occurrence in malignancy. The absence of both mutational markers and suspicious sonographic features proved extremely useful in tailoring surgical strategy, as it could have ultimately spared 143/258 patients (55%) an overly radical thyroidectomy.

Conclusion:

The preoperative utility of mutational markers and sonographic features in combination has a synergistic impact. It can predict the risk of malignancy with high accuracy, properly triaging patients to appropriate surgery.

Introduction

With the widespread use of imaging studies, thyroid nodules are being frequently encountered by physicians. Compared with the high prevalence of thyroid nodular disease, thyroid cancer is not common. Fine-needle aspiration cytology (FNAC) is the gold-standard modality for determining the nature of a thyroid nodule. However, it fails to yield a conclusive result in a subset of patients, labeling them as having an “indeterminate diagnosis.” In doing so, it compromises the proper triaging of patients to appropriate surgery. This places patients at risk of non-optimal initial surgery: an overly radical total thyroidectomy, or an unnecessary two-stage operation. The aim of this study was to assess the impact of combining gene mutation testing and ultrasonographic (US) features preoperatively on predicting the risk of malignancy in patients with indeterminate nodules, with the goal of offering patients the most appropriate tailored initial surgical intervention.

Materials and Methods

Between January 2009 and July 2013, 11,300 patients underwent total thyroidectomy for various indications at the authors' center. Total thyroidectomy was performed either conventionally or via minimally invasive video-assisted thyroidectomy (MIVAT) or via robot-assisted transaxillary thyroidectomy (RATT), when selection criteria were met for MIVAT and RATT. Among these, 258 underwent conventional thyroidectomy for indeterminate uninodular thyroid disease. An indeterminate cytology includes Bethesda categories III, IV, and V. However, the 258 patients selected for this study had single nodules reported as suspicious for a follicular neoplasm (Bethesda category IV). They included 190 females (73.7%) and 68 males (26.3%). The female:male ratio was 4:1, and the mean age at diagnosis was 44 years (range 14–78 years). The mean sonographically estimated thyroid volume (SETV) was 16 mL (range 2.9–115.7 mL), and the mean nodule size was 23.5 mm (range 6–71 mm). The US features of thyroid nodules and the presence of biochemical thyroiditis were recorded. The US features assessed by gray-scale US were: hypoechogenicity, microcalcifications, irregular margins, and a “taller than wide” shape. Color Doppler US was not applied. Therefore, the pattern of nodular blood flow was not reported. The presence of biochemical thyroiditis was assessed by measurements of serum levels of thyroid antibodies (antithyroglobulin and antithyroid peroxidase antibodies). Serum levels of these antibodies were available, not because they are part of the routine work-up for thyroid nodular disease, but because they were required for research purposes by endocrinologists. None of the patients underwent lymph node dissection or lymphadenectomy through “berry picking,” as none had suspicious lymph nodes sonographically or as an intraoperative finding.

The surgical strategy regarding nodal dissection at the center is to perform a selective compartment-oriented nodal dissection based on suspicious sonographic and/or intraoperative findings. Prophylactic central compartment nodal dissection (CCLND) is not performed, even for an established diagnosis of papillary carcinoma, as it has been demonstrated in a prospective randomized controlled study that the addition of prophylactic CCLND does not favorably affect disease-free and overall survival, and is associated with greater morbidity (1). The histologic evaluation of all surgical specimens was conducted by two expert pathologists (F.B. and A.P.). Tumors were classified according to the World Health Organization (WHO) histological criteria (2). BRAF and NRAS sequences were analyzed in all 258 formalin-fixed, paraffin-embedded (FFPE) tissue samples. Tumoral areas were carefully isolated from two 10 μm thick sections, and genomic DNA was purified using a QIAmp DNA Mini Kit (Qiagen) following the manufacturer's instructions. BRAF exon 15 and NRAS exon 2 were amplified according to standard procedures (3), and analyzed by direct Sanger sequencing on a AbiPrism 3130 Genetic Analyzer (Applied Biosystem). The pathologists were blinded to the molecular marker status and ultrasound features when making the histologic diagnosis.

Results

On final histological examination, 90/258 patients (34.9%) had a thyroid carcinoma, and 168/258 patients (65.1%) had a benign nodule. Cases showing malignancy are collectively referred to as the malignant group (MG). Those with benign lesions are referred to as the benign group (BG). Within the MG, 71/90 (78.9%) had a follicular variant of papillary carcinoma, 10/90 (11.1%) had a classical papillary carcinoma, 6/90 patients (6.7%) had a solid variant of papillary carcinoma, and 3/90 patients (3.3%) had a follicular carcinoma. According to the TNM staging system (4), 39, 34, and 17 out of the 90 patients had T1, T2, and T3 tumors, respectively. None had nodal or distant metastases (N0, M0).

The mean age in the BG was 47 years (range 14–75 years), with a female-to-male ratio of 3:1. The mean SETV was 17.3 mL (range 3.4–60.8 mL), and the mean nodule size was 27 mm (range 6–68 mm). The MG had a mean age of 45 years (range 14–78 years), with a female-to-male ratio of 3:1 as well. The mean SETV was 15.7 mL (range 2.9–115.7 mL), and the mean nodule size was 23.5 mm (range 6–71 mm). Both groups were demographically comparable.

Regarding mutational markers, a BRAF mutation was present in 8/258 (3.1%) of cases. A NRAS mutation was present in 31/258 (12.0%) cases. These mutations were mutually exclusive. BRAF mutations were only seen in the MG, equally divided between BRAF^V600E and BRAF^K601E . Thirty-one NRAS exon 2 alterations were identified: 25 (80.6%) Q61R and 6 (19.4%) Q61K mutations. NRAS mutations occurred in 9/168 patients (5.4%) of the BG and in 22/90 patients (24.4%) of the MG. The occurrence of both mutations correlated significantly with malignancy (p < 0.05; Table 1).

Table 1.

Summary of Features Assessed and Their Correlation with Malignancy

	Benign (%)	Cancer (%)	p-Value	OR	CI
Total	168 (65)	90 (35)
BRAF mutation
Positive	0 (0)	8 (8.8)	0.0152	34.72	[1.98–608.93]
Negative	168 (100)	82 (91.2)
NRAS mutation
Positive	9 (5.4)	22 (24.4)	<0.0001	5.71	[2.50–13.05]
Negative	159 (94.6)	68 (75.6)
Hypoechogenicity
Positive	109 (64.8)	53 (58.8)	0.52	0.775	[0.45–1.31]
Negative	59 (35.1)	37 (41.1)
Microcalcifications
Present	17 (10.1)	40 (44.4)	<0.0001	7.10	[3.70–13.63]
Absent	151 (89.9)	50 (55.6)
Irregular margin
Present	2 (1.2)	22 (24.4)	<0.0001	26.85	[6.14–117.34]
Absent	166 (98.8)	68 (75.6)
Taller than wide
Present	9 (5.3)	24 (26.6)	<0.0001	6.42	[2.83–14.55]
Absent	159 (94.7)	66 (73.4)
Thyroiditis
Present	30 (17.8)	15 (16.6)	0.81	0.92	[0.46–1.81]
Absent	138 (82.2)	75 (83.4)

OR, odds ratio; CI, confidence interval.

The US features evaluated in our series (hypoechogenicity, microcalcifications, irregular margins, and a “taller than wide” shape) occurred in 162, 57, 24, and 33 out of 258 nodules, respectively. Hypoechogenicity was present in 53/90 (58.9%) of the MG and in 109/168 (64.8%) of the BG. The correlation of hypoechogenicity with malignancy was not statistically significant (p = 0.52). Microcalcifications were present in 17/168 (10.1%) of the BG and in 40/90 patients (44.4%) of the MG. Irregular margins were present in 2/168 (1.2%) of the BG and in 22/90 (24.4%) of the MG. A “taller than wide” shape was present in 9/168 (5.4%) of the BG and in 24/90 (26.6%) of the MG. The occurrence of all these features with malignancy was statistically significant (p < 0.05; Table 1).

Biochemical thyroiditis was present in 45/258 patients: 30/168 (17.8%) and 15/90 (16.6%) of the BG and MG, respectively. The correlation of biochemical thyroiditis with malignancy was not statistically significant (p = 0.48; Table 1).

The risk of malignancy in patients positive for a mutational marker and one or more suspicious US feature is demonstrated in Table 2. This table also demonstrates the positive predictive value (PPV) of suspicious US features (single and in combinations). Table 3 demonstrates the use of combining the absence of mutational markers and the absence of suspicious US features (single and in combinations) in determining the likelihood of malignancy. The risk of malignancy observed in those negative for both mutational markers and all suspicious US features was only (11.8%).

Table 2.

Rate of BRAF and NRAS Mutations in Patients in the Presence of Suspicious US Features

	BRAF(+)	NRAS(+)	BRAF(+)/NRAS(+)	BRAF(−)NRAS(−)
I. Microcalcifications	(3/3)100%	(12/12) 100%	—	(40/57) 70.1%
II. Irregular margins	(2/2)100%	(3/3) 100%	—	(22/24) 91.6%
III. Taller than wide	—	(5/7) 71.4%	—	(24/33) 72.7%
I + II	(1/1)100%	—	—	(9/9) 100%
I + III	—	(2/3) 66.6%	—	(10/10) 100%
II + III	(1/1)100%	—	—	(4/4) 100%
I + II + III	—	—	—	(2/2) 100%

The last column represents the positive predictive values (PPV) of suspicious sonographic features independent of mutational markers.

US, ultrasound.

Table 3.

Rate of Negative BRAF and NRAS Mutations in Patients in the Absence of Suspicious US Features

	BRAF(−)	NRAS(−)	BRAF(−)NRAS(−)
I. Microcalcifications	(45/196) 22.9%	(39/182) 21.4%	(34/177) 19.2%
II. Irregular margins	(62/228) 27.8%	(49/206) 23.7%	(43/200) 21.5%
III. Taller than wide	(58/217) 26.7%	(49/201) 24.4%	(41/193) 21.2%
I + II	(33/182) 18.1%	(29/170) 17.1%	(25/166) 15.1%
I + III	(33/176) 18.7%	(30/166) 18%	(25/161) 15.6%
II + III	(42/199) 21.1%	(34/184) 18.4%	(28/178) 15.7%
I + II + III	(23/164) 14%	(22/156) 14.1%	(18/152) 11.8%

Discussion

Thyroid nodular disease is common in clinical practice. Its prevalence largely depends on the population being evaluated and the detection method used. The prevalence of palpable thyroid nodules is around 5%, whereas the prevalence of non-palpable lesions (incidentally discovered on imaging studies) is much higher. A prevalence as high as 67% has been reported with the use of high-resolution ultrasonography (5). The risk of malignancy per patient is equal for both palpable and non-palpable nodules. However, it is relatively low (6). The ultimate goal of the diagnostic evaluation of a patient with a thyroid nodule is to determine its nature, consequently providing timely and appropriate treatment.

Although determining the nature of a thyroid nodule cannot be achieved solely by US, US can display features that may aid in predicting the nature of the nodule. Thyroid nodules are usually evaluated by a gray-scale and color Doppler US. Many sonographic features have been evaluated for their ability to predict the nature of a thyroid nodule. These include size, echogenicity, margin definition, shape/orientation, the presence and extent of a halo, the presence and type of calcifications, the pattern of blood flow, and measures of tissue elasticity (US elastography) (7). Nodule size is not predictive of malignancy. The likelihood of cancer in a thyroid nodule is the same, irrespective of the size measured on US (6,7). As for the other features, the literature has demonstrated inconsistency in their predictive values, and the combination of findings improves the positive predictive value of US to some extent (6 –10). In the present series, the vascularity of nodules was not evaluated. It has been demonstrated that there is no additional value of color Doppler US compared to gray-scale US in predicting the nature of a thyroid nodule (11 –13). The features whose presence significantly correlated with malignancy included microcalcifications, irregular margins, and a “taller than wide” shape. On the other hand, hypoechogenicity did not significantly correlate with malignancy in the present study, as it occurred in 58.9% and 64.8% of malignant and benign nodules, respectively. The PPV of irregular margins, “taller than wide” shape, and microcalcifications were 91.6%, 72.7%, and 70.1%, respectively. The sensitivities and PPV of US features are demonstrated in Tables 1 and 2, respectively. Microcalcifications are defined as punctuate tiny foci that are too small to induce posterior acoustic shadowing. A “taller than wide” nodule is a nodule with an anteroposterior diameter greater than its transverse diameter, and was predictive of malignancy. It is believed that this non-parallel orientation is a reflection of the centrifugal growth pattern of malignancy across normal tissue planes. In the present series, the PPV of ultrasonography improved dramatically when findings were combined. Combining at least two features always achieved a likelihood of malignancy of 100% (Table 2). However, a major limitation to the utility of US alone in diagnosing malignancy is that sonographic features with a high PPV have low occurrences in malignancy, and thus are of low sensitivity.

FNAC establishes a reliable diagnosis in 70–80% of instances. However, the remaining 20–30% of cases are labeled as being indeterminate for malignancy (14). An indeterminate diagnosis collectively includes Bethesda categories III, IV, and V. Labeling patients as having an indeterminate diagnosis is troublesome, as it places them at risk of a non-optimal initial surgical intervention, that is, an overly radical total thyroidectomy, or an unnecessary two-stage operation. This particularly applies to Bethesda category IV, whose risk of malignancy lies in the gray zone between those of Bethesda categories III and IV. The 15–30% risk of malignancy associated with lesions suspicious for a follicular/Hürthle cell neoplasm could be considered neither low nor high enough to make a diagnostic lobectomy or a total thyroidectomy, respectively, acceptable routine options.

In an attempt to resolve the dilemma of an indeterminate diagnosis, many centers turned to molecular diagnostics. Gene mutation markers (mutational markers) are one of several molecular tools used. Gene mutations associated with thyroid cancer can be detected in both fine-needle aspirates and surgical specimens. In the present series, the mutational analysis was performed on surgical specimens from areas clearly identified as malignant. Archived cytological smears were not available to compare concordance of results. Nevertheless, absolute concordance between mutational analysis results obtained with histological and cytological samples has been demonstrated (15). BRAF mutations are the most common gene alterations in thyroid cancer, and are typically seen in papillary carcinomas but rarely in follicular carcinomas (1.4%) (16). Many mutational variants of BRAF exist; the most common one is BRAF^V600E followed by BRAF^K601E . BRAF mutations are highly specific for malignancy. They correlate with a malignant outcome in almost 100% of instances (17). RAS point mutations are also common in thyroid cancer. However, they are also found in 20–40% of benign lesions. RAS mutations can occur in all three RAS genes (NRAS, HRAS, and KRAS). NRAS mutations are the most common genetic alterations detected in indeterminate nodules (16). Despite occurring in a considerable proportion of benign lesions, it has been demonstrated that RAS-mutated benign lesions are prone to malignant transformation and would benefit from surgery (14). Based on what has been mentioned, and taking cost-effectiveness into account, the mutational analysis was limited to BRAF and RAS mutations. However, as mutational panels are considered rule-in tests, including additional mutations and/or rearrangements would be useful.

In the present series, the presence of a BRAF mutation was restricted to papillary carcinomas, with a PPV of 100%. BRAF^V600E and BRAF^K601E were equally common. However, a major limitation to its diagnostic utility, as inferred from the present series, is its low occurrence (3.1% of all indeterminate nodules and 8.8% of all malignancies). An NRAS mutation was the most common gene alteration in the present series of indeterminate nodules. Its occurrence correlated significantly with a malignant outcome, and its PPV was 70.9%. These results can be inferred from Table 1.

In the present study, the detection of any combination of a mutation and a suspicious US feature was strongly predictive of malignancy. Correlation with a malignant outcome was often at 100% (Table 2). The synergistic impact of adding mutational markers to suspicious gray-scale US features in predicting malignancy becomes evident when comparing the present results with the scoring system proposed by Russ et al. (9). Applying the diagnostic modality in the present study achieves a predictivity equivalent or higher than that of a TIRADS score of 5 with only one suspicious US feature instead of requiring more than two. On the basis of the present findings, indeterminate nodules that are positive for such a combination can be treated with total thyroidectomy as the initial surgical approach. This is particularly applicable to cases that are considered clinically significant malignancies for which total thyroidectomy would be favored over any other treatment modality according to the American Thyroid Association (ATA) guidelines (18). On the other hand, being negative for both mutations and suspicious US features (152 in the present series) was associated with only a 11.8% risk of malignancy (18/152; Table 3). Furthermore, of the 18 patients with malignancy, only nine patients (six and three with T2 and T3 tumors, respectively) required postoperative radioactive iodine ablation therapy. On the basis of the reduced likelihood of malignancy, particularly high-risk malignancies in this group of patients with lesions suspicious for follicular neoplasms, lobectomy could have been both diagnostic and therapeutic in 94.1% (143/152). It is also believed that with improvements of diagnostic modalities and consequent further risk reduction in this category of lesions, diagnostic lobectomy could even be avoided. Observation could become a viable option in view of the fact that thyroid cancers grow slowly and most patients have an excellent prognosis. In summary, abiding to the diagnostic modality used in this study and, consistent with ATA guidelines on the management of differentiated thyroid cancer, 55% of patients could have been spared a total thyroidectomy, which could be considered as an overly radical procedure for this condition. These 55% of cases are the sum of 52% with a benign lesion and 3% with a malignancy, in which a total lobectomy could be considered as adequate treatment.

To conclude, the utility of innovative diagnostic modalities such as mutational markers complements standard evaluation of thyroid nodules labeled as “indeterminate.” Furthermore, a synergistic impact is obtained when molecular markers are combined with combinations of suspicious sonographic findings. Improving diagnostic accuracy in this subset of nodules aids in providing patients with the optimal initial surgical intervention.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

References

Viola

, Materazzi

, Valerio

, Molinaro

, Agate

, Faviana

, Seccia

, Sensi

, Romei

, Piaggi

, Torregrossa

, Sellari-Franceschini

, Basolo

, Vitti

, Elisei

, Miccoli

. 2015. Prophylactic central compartment lymph node dissection in papillary thyroid carcinoma: clinical implications derived from the first prospective randomized controlled single institution study. J Clin Endocrinol Metab, 100:1316–1324.

World Health Organization Classification of Tumors 2004 Pathology and Genetics of Tumors of Endocrine Organs. IARC Press, Lyon.

Nikiforov

, Steward

, Robinson-Smith

, Haugen

, Klopper

, Zhu

, Fagin

, Falciglia

, Weber

, Nikiforova

. 2009. Molecular testing for mutations in improving the fine-needle aspiration diagnosis of thyroid nodules. J Clin Endocrinol Metab, 94:2092–2098.

Edge

, Byrd

, Compton

, Fritz

, Greene

, Trotti

(Eds) 2010. AJCC Cancer Staging Manual. Seventh edition. Springer, New York.

Dean

, Gharib

. 2008. Epidemiology of thyroid nodules. Best Pract Res Clin Endocrinol Metab, 22:901–911.

Frates

, Benson

, Charboneau

, Cibas

, Clark

, Coleman

, Cronan

, Doubilet

, Goellner

, Hay

, Hertzberg

, Intenzo

, Jeffrey

, Langer

, Larsen

, Mandel

, Middleton

, Reading

, Sherman

, Tessler

, Society of Radiologists in Ultrasound. 2005. Management of thyroid nodules detected at US: society of radiologists in ultrasound consensus conference statement. Radiology, 237:794–800.

Moon

, Baek

, Jung

, Kim

, Kwak

, Lee

, Na

, Park

, Korean Society of Thyroid Radiology (KSThR), Korean Society of Radiology. 2011. Ultrasonography and the ultrasound-based management of thyroid nodules: consensus statement and recommendations. Korean J Radiol, 12:1–14.

Frates

, Benson

, Doubilet

, Kunrevther

, Contreras

, Cibas

, Orcutt

, Moore

Jr , Larsen

, Margusee

, Alexander

. 2006. Prevalence and distribution of carcinoma in patients with solitary and multiple thyroid nodules on sonography. J Clin Endocrinol Metab, 91:3411–3417.

Russ

, Leboullex

, Leenhardt

, Hegedus

. 2014. Thyroid incidentalomas: epidemiology, risk stratification with ultrasound and work up. Eur Thyroid J, 3:154–163.

10.

Seo

, Na

, Kim

, Yoon

. 2015. Ultrasound-based stratification for malignancy in thyroid nodules: a four-tier categorization system. Eur Radiol, 25:2153–2162.

11.

Rago

, Vitti

, Chiovato

, Mezzo

, De Liperi

, Miccoli

, Viacava

, Bogazzi

, Martino

, Pinchera

. 1998. Role of conventional ultrasonography and color flow-doppler sonography in predicting malignancy in “cold” thyroid nodules. Eur J Endocrinol, 138:41–46.

12.

Moon

, Kwak

, Kim

, Son

, Kim

. 2010. Can vascularity at power Doppler US help predict thyroid malignancy?. Radiology 255:260–269.

13.

Rosario

, da Silva

, Borges

, Calsolari

. 2015. Is Doppler ultrasound of additional value to gray-scale ultrasound in differentiating malignant and benign thyroid nodules?. Arch Endocrinol Metab, 59:79–83.

14.

Nikiforov

, Yip

, Nikiforova

. 2013. New strategies in diagnosing cancer in thyroid nodules: impact of molecular markers. Clin Cancer Res, 19:2283–2288.

15.

Capelli

, Marfisi

, Puccetti

, Saragoni

, De Paolo

, Zaccaroni

, Chiadini

, Gagliardi

, Ferretti

, Zoli

, Uilvi

. 2015. Role of BRAF molecular analysis in the management of papillary thyroid carcinoma: analysis of cytological and histological samples. Cytopathology, 26:297–301.

16.

Parangi

, Suh

. 2014. The role of genetic markers in the evaluation and management of thyroid nodules. Surg Clin N Am, 94:515–528.

17.

Yip

. 2015. Molecular markers for thyroid cancer diagnosis, prognosis, and targeted therapy. J Surg Oncol, 111:43–50.

18.

Haugen

, Alexander

, Bible

, Doherty

, Mandel

, Nikiforov

, Pacini

, Randolph

, Sawka

, Schlumberger

, Schuff

, Sherman

, Sosa

, Steward

, Tuttle

, Wartofsky

. 2016. 2015 American Thyroid Association management guidelines for adult patients with thyroid nodules and differentiated thyroid cancer. Thyroid, 26:1–133.