Abstract
Background/Aim: Studies have shown that the risk of malignancy in follicular neoplasms is as high as 30%. Often, surgery is recommended for such lesions, not for therapeutic purposes but as a diagnostic method, leading to increased hospital costs and related morbidities. Recent studies have suggested that tumor size predicts malignant potential of these follicular neoplasms. Our aim was to identify the impact of nodule size on the risk of malignancy for such lesions. Patients and Methods: A retrospective medical chart review was undertaken for patients who underwent thyroid surgery at a single academic North American Institution. A total of 120 follicular lesions, follicular neoplasms (Bethesda category IV) or follicular lesions of undetermined significance (Bethesda category III) in 110 patients undergoing thyroid surgery were evaluated. Nodule size as measured by ultrasound, fine-needle aspiration cytological results, and final histopathology reports were reviewed. Analysis was performed by classification according to nodule size: <3 cm, ≥3 cm, <4 cm and ≥4 cm. Results: Out of the 120 nodules, 48 (40%) were reported to be malignant on final pathological examination. The malignancy rate in nodules <3 cm and ≥ 3cm was 41% and 37.8%, respectively (p=0.84). When 4 cm was used as the cut-off, the rate in nodules <4 cm and ≥4 cm was 40.6% and 37.5%, respectively (p=0.82). Conclusion: Increased thyroid nodule size does not increase the malignancy rate for follicular neoplasms. Hence, we recommend against routine total thyroidectomy for patients with follicular neoplasms based on the size criteria.
- Nodule size
- follicular neoplasms
- follicular lesion of undetermined significance
- thyroid malignancy
- thyroid cancer
Thyroid nodules are present in 4-7% of the general population, of which about 5% are diagnosed as malignant (1-5). Fine-needle aspiration (FNA) is the most cost-effective and accurate evaluation of thyroid nodules. It is recommended for nodules measuring 1 cm, more as this cutoff has been shown to have high sensitivity and specificity (6). However, there is controversy about the accuracy of FNA cytology for nodules larger than 4 cm due to a high (up to 20%) false-negative rate (7, 8). FNA cytological findings help physicians decide whether to monitor these thyroid nodules or to refer the patients for surgery (9). Since FNA was introduced into clinical practice 50 years ago, the number of patients referred for thyroidectomy has decreased by 25% (10).
One serious limitation is the inability to differentiate follicular adenoma from follicular carcinoma. Capsular invasion with or without additional vascular invasion is necessary to confirm the diagnosis of carcinoma (11). Such indeterminate cytology (including follicular neoplasm) represents 22-42% of FNA cytology (12).
Several studies have shown that approximately 27-52% of lesions diagnosed as follicular neoplasm via FNA cytology often revealed via histopathology to be malignant lesions. The most common malignancy was found to be papillary thyroid carcinoma followed by follicular carcinoma (9, 13). The majority (48-73%) of such nodules are benign, which means that many patients are unnecessarily exposed to surgery for diagnostic purposes.
Thyroid surgeries can be associated with serious complications, including thyroid hormone imbalance, hypo parathyroidism, recurrent laryngeal nerve injury, bleeding, and infection, in addition to incurring increased costs of hospitalization (8).
Improving the ability to predict the risk of malignancy of follicular lesions would help clinicians to make the best decision regarding which patients should be referred to surgery, and the extent of surgery when indicated (11). Hemithyroidectomy, which might be considered for benign lesions, is safer than total thyroidectomy due to the associated lower risk of nerve injury and overall complications (14, 15).
Demographics of 110 patients with 120 follicular neoplasms on FNA cytology.
Many clinical factors, including age, sex, presence of microcalcifications, hypoechoiety, and internal vascularity have been examined to determine their utility in predicting the risk of malignancy in follicular lesions. However, results have been contradictory (16).
Debate exists regarding the usefulness of clinical features, including nodule size, in predicting malignancy. The risk of follicular cancer increases with nodule size, as compared to reduced risk of papillary thyroid cancer in larger nodules (i.e. more than 2 cm) (16).
Recent studies suggest that tumor size predicts malignant potential in follicular neoplasms of the thyroid (17). This is consistent with the American Thyroid Association guidelines which recommend total thyroidectomy for follicular lesions larger than 4 cm due to increased risk of malignancy (18).
Our objective was to identify the impact of size on the risk of malignancy of such lesions with indeterminate diagnosis. This could provide vital information for surgeons in counseling patients regarding the extent of surgery needed.
Patients and Methods
Clinical patients. This retrospective study was approved by the Tulane University Medical Center Institutional Review Board (140492-1). The medical records of a total of 944 patients who underwent thyroid surgery at Tulane University hospital between 2006 and 2012 were retrospectively reviewed. One hundred and ten patients were found to have 120 follicular lesions on FNA cytological examination carried out at our Institution. All nodules included were ≥1 cm in size.
Clinical and FNA cytology features of malignant and benign nodules.
Data analysis. Age, gender, and race were reported. Using different size criteria (<3 cm, ≥3 cm, <4 cm, ≥4 cm), preoperative FNA biopsy and surgical histopathological results of all follicular lesions were reviewed and documented. Follicular lesions included follicular neoplasms (Bethesda category IV) and follicular lesions of undetermined significance (FLUS) (Bethesda category III). Ultrasonographic features of these lesions, including size, calcifications, echogenicity, and internal vascularity, were also documented.
Statistical analysis. The pathological features of follicular lesions of different size categories (<3 cm vs. ≥3 cm, and <4 cm vs. ≥4 cm) were documented. The p-value for the significance of differences was calculated by two-sided Student's t-test for the continuous variables and by two-sided Fisher's exact test for the categorical variables. Logistic regression analysis was used to test the association between size and different pathological categories. The p-values were considered statistically significant if less than 0.05.
Results
Clinical demographics. A total of 944 patients were screened, out of these 110 were found to have follicular lesions upon FNA cytological examination. Patients' demographics and pathology subsets are summarized in Table I. Patients diagnosed with benign histopathology tended to be older, with a mean age of 53.5±12.4 (SD) (N=67) years than those with malignant histopathology (mean age of 46.78±13.6; N=43; p=0.008). The malignancy rate for patients ≤45 years vs. those older than 45 years was 52.8% vs. 32.4% (p=0.06).
Association between follicular neoplasm size and malignancy.
As expected, the majority of patients were female (79.1%). There was no significant difference in gender or race distribution between those with benign vs. those with malignant lesions (Table I). Among all follicular lesions, 48 (40%) were found to be malignant on histopathology. Out of the 48 carcinomas, 23 (47.9%) were papillary carcinoma, 17 (35.4%) were follicular variant of papillary carcinoma, and eight (16.7%) were follicular carcinoma.
Size of lesion. The malignancy rate in nodules <3 cm and ≥3 cm was 40.9% (34/83) and 37.8% (14/37), respectively (p=0.84). When 4 cm was used as the cutoff, the malignancy rate in lesions <4 cm and ≥4 cm was 40.6% (39/96) and 37.5% (9/24) respectively (p=0.82) (Table II). In addition, our results showed that there was no significant difference in the final benign pathology rates of indeterminate FNA cytology between nodules <4 (59.4%) cm and ≥4 cm (62.5%) in size.
Sonographic features. Additional analysis showed that there was no significant difference between malignant and benign lesions with regard to ultrasonographic features, echogenicity, calcification, or internal vascularity (Table II).
Discussion
This analysis of 110 patients with 120 clinically-relevant thyroid nodules suggests that follicular lesions of larger size are not associated with a higher risk of malignancy. The malignancy rate in nodules <3 cm and those ≥3 cm was 40.9% and 37.8%, respectively (p=0.84), while in those <4 cm compared to those ≥4 cm, the rate of malignancy was 40.6% and 37.5%, respectively (p=0.82).
Due to the significant complications associated with thyroid surgeries, researchers are seeking preoperative findings that can be helpful in predicting malignancy of nodules diagnosed as follicular lesions, especially given the fact that up to 80% of follicular lesions eventually turn out to be benign (13, 17). Several recent studies have shown the ability of immunohistological markers such as galectin-3, HBME1, and cytokeratin-19 to improve the preoperative sensitivity/specificity in differentiating benign from malignant cases in such indeterminate nodules. However, they have not been widely accepted in clinical practice for several reasons, including their operator-dependent nature, differences in analytical methods, and the overlap between follicular adenomas and differentiated thyroid carcinomas (19, 20).
Debate also persists on the usefulness of size in predicting the risk of malignancy of thyroid nodules. Many studies reported a higher risk of malignancy in nodules >2 cm (16, 21), >3 cm (22), and >4 cm (9, 23). However, this was negated by other studies, which reported no association between size and malignancy risk in follicular lesions (19, 24) (Table III).
Regarding indeterminate nodules, the American Thyroid Association (ATA) recommends total thyroidectomy for lesions larger than 4 cm, when the biopsy is suspicious for papillary carcinoma or shows marked atypia, in patients with family history of thyroid carcinoma, and in patients with a history of radiation exposure (recommendation rating: A). For all other nodules, the ATA recommends thyroid lobectomy (recommendation rating: C) (18).
It has been documented that the incidence of papillary thyroid cancer decreases, and that of follicular cancer increases as nodule size increases (16). Nodules larger than 4 cm have been documented to have a higher rate of false-negative FNA cytology, which supports the recommendation for surgical resection (16).
The Bethesda System for Reporting Thyroid Cytopathology subclassifies indeterminate FNA cytology into FLUS and follicular neoplasm. Follicular neoplasm is differentiated from FLUS by nuclear atypia, trabecular pattern, loss of colloid or mitosis (20). The estimated rates of malignancy are 5-15% and 15-30% for FLUS and follicular neoplasm, respectively (25). Williams et al. reported similar rates of malignancy with 7.0% (14/199) of the FLUS and 21.4% (21/98) of the follicular neoplasms being malignant (p=0.0005) (20). Our results showed a higher malignancy rate of 65% (13/20) for follicular neoplasms compared to 35% (35/100) for FLUS (p=0.023). In addition, Williams et al. reported that follicular neoplasm has a 40% rate of malignancy when larger than 4 cm (20).
Additionally, patients younger than 45 years had a higher cancer rate as compared to those who were older (52.8% vs. 32.4%, p=0.06) in this study. This is consistent with the study by Ozluk et al. who reported a higher cancer rate in patients <40 vs. ≥40 years (64.2% vs. 34.2%) (26). This is contrary to prior studies that have shown that increasing age is associated with a higher cancer rate (27), and to those which documented no association with the cancer rate at all (28).
Along with other clinical factors, molecular biology is playing an increasingly significant role in personalizing the management plan for thyroid nodules due to its ability to predict the patient's tumor behavior (29).
Recently, a gene-expression classifier (Afirma) was found to be highly beneficial in distinguishing between benign and malignant nodules in follicular lesions. This gene-expression classifier was reported to have a sensitivity of 90% for both FLUS and follicular neoplasm. For FLUS, the specificity and negative predictive value was 53% and 95%, respectively. For follicular neoplasm, the respective value were 49% and 94% (30).
In addition, a number of studies documented that BRAF mutational analysis increased the sensitivity of FNA biopsy for papillary thyroid cancer, with sensitivity of 15-84% and specificity of 97.3-100% (31).
We acknowledge certain limitations to the present study, some of which are: i) This study was performed retrospectively; ii) data are from a single Institution; iii) small sample size; iv) lack of genetic testing; v) questioned reliability of diagnosing follicular lesion according to different pathologists. These findings should be investigated by larger, randomized, multicenter trials.
Conclusion
These data suggest that risk of malignancy of follicular neoplasms is not influenced by lesion size, and extent of surgery should not be indicated based on size per se.
- Received November 13, 2014.
- Revision received December 1, 2014.
- Accepted December 4, 2014.
- Copyright© 2015 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved
