Should “atypical follicular cells” in thyroid fine-needle aspirates be subclassified?

Abstract

BACKGROUND:

Recently reported Bethesda terminology suggests the use of the term “atypical follicular cells” for thyroid fine-needle aspirates. Previous work has suggested that some types of “atypical follicular cells” have different risks of malignancy.

METHODS:

The author reviewed the results of all thyroid aspirations with surgical resection performed during the past 13 years at our institution, subclassified the “atypical follicular cells,” and compared their relative risk of malignancy.

RESULTS:

During the 13 years in question, a total of 7089 cases were aspirated with 1331 resections. A total of 548 (14%) of all cases were classified as “atypical follicular cells,” and 204 (37%) were resected with an overall risk of malignancy of 25%. The risk of malignancy for atypical follicular cells subclassified as “rule out papillary carcinoma” was significantly higher (38%) than the other atypical cells. The risk of “rule out Hurthle cell neoplasm” was, at 7%, significantly lower than the other cases of atypical follicular cells (P<.001 and P<.02, respectively).

CONCLUSIONS:

Different types of “atypical follicular cells” have significantly different risks of malignancy. This disparity of risk should be communicated by the cytologist. Cancer (Cancer Cytopathol) 2010. © 2010 American Cancer Society.

Fine-needle aspiration (FNA) is an excellent procedure for evaluation of thyroid nodules.1 Recently, some authors have proposed using the term “risk of malignancy” to better classify thyroid aspirates1-3 because the terminology generally used for thyroid aspirates is not consistent in expressing that risk and can be misleading.3-5 The risk of malignancy method has been incorporated into the recent Bethesda System for Reporting Thyroid Cytopathology.1 This classification system proposes the term “atypical follicular cells” for those cases that are difficult to classify in other categories.1 Nevertheless, previous studies have documented some cases that would fit in this category, specifically those cases with focal features of papillary carcinoma, have an elevated risk of malignancy when compared with other “atypical” cases.6-9 The author sought to determine whether the category of “atypical follicular cells” could be subclassified into cases with significantly different risks of malignancy.

MATERIALS AND METHODS

All thyroid fine-needle aspirates interpreted at Baptist Hospital, Miami, Fla, and Homestead Hospital, Homestead, Fla, for the 13-year period October 1996 through November 2009 were reviewed (and the results correlated) with the results of histologic follow-up.

For this study, cases were reclassified according to the current Bethesda System for Reporting Thyroid Cytopathology.1 Within the cases classified as “atypical follicular cells,” 4 distinct subtypes were recognized by the pathologists who diagnosed these cases. “Atypical cells, rule out papillary carcinoma,” had a small number of cells possessing some features of papillary carcinoma.8 Typically, these cells were rare (as few as 20 cells per case), showing focal nuclear enlargement and clearing. Inclusions were not identified. “Atypical cells, rule out Hurthle cell neoplasm,” demonstrated 2 different patterns that have been described previously.10 One pattern was almost exclusively Hurthle cells, but was scant; the other had numerous Hurthle cells, but lacked significant atypia. “Atypical cells, rule out a follicular neoplasm,” had cells that were arranged in a mixed macro and micro follicular pattern, with minimal to no cytologic atypia. “Atypia, NOS,” was a heterogeneous group of cases that typically had rare cells with atypia, making further classification difficult. NOS indicates “not otherwise specified.”

All aspirates were performed by clinicians. Approximately one-third were performed in clinician's offices without imaging studies. Between 2 and 8 passes were made. Two-thirds of aspirates were performed in the radiology department of Baptist Hospital, with the aid of ultrasound guidance and immediate evaluation. Direct smears were made in all cases, and each smear was alcohol-fixed and stained with either Papanicolaou or hematoxylin and eosin (H&E) stain. When sufficient material was obtained, cell blocks were also made. Core needle biopsy was also performed in approximately 400 cases. These core biopsies significantly lowered the nondiagnostic rate, but did not significantly alter the atypia rate.11

Categorical analysis was performed using a 2-tailed Chi-square test. A P value of .05 was considered significant.

RESULTS

During a period of 13 years, a total of 7089 cases were aspirated, with 1331 resections performed (Table 1). A total of 548 (14%) of all cases were classified as “atypical follicular cells,” and 204 (37%) were resected with an overall risk of malignancy of 25%.

Table 1. Risk of Malignancy for Fine-Needle Aspiration of the Thyroid

Diagnosis	No. (%) of Cases	No. (%) of Cases With Histologic Follow-Up	No.(%) Benign Histology	No. (%) Malignant Histology
NOS indicates not otherwise specified.
Benign	3829 (54)	361 (8)	355 (98)	6 (2)
Atypical follicular cells	548 (14)	204 (37)	154 (75)	50 (25)
Suspicious for a follicular neoplasm	343 (5)	179 (52)	126 (70)	53 (30)
Suspicious for a Hurthle cell neoplasm	263 (4)	108 (41)	81 (67)	27 (33)
Suspicious for papillary carcinoma	123 (2)	73 (59)	1 (1)	72 (99)
Suspicious for lymphoma	8 (.1)	2 (25)	1 (50)	1 (50)
Spindle cell neoplasm	3 (.1)	1 (33)	0	1 (100)
Papillary carcinoma	275 (4)	160 (58)	0	160 (100)
Medullary carcinoma	16 (.2)	6 (38)	0	6 (100)
Carcinoma, NOS	10 (.1)	2 (20)	0	2 (100)
Nondiagnostic	1671 (24)	235 (14)	188 (80)	47 (20)
Total	7089	1331 (19)	906 (68)	425 (32)

The results for subclassifying “atypical follicular cells” is shown in Table 2. The mean size of the resected lesions ranged from 2.6 ± 2.3 cm for “NOS,″ 2.0 ± 1.3 cm for rule out follicular neoplasm, 1.8 ± 1.2 cm for rule out Hurthle cell neoplasm, and 2.2 ± 1.6 cm for rule out papillary carcinoma. The risk of malignancy for atypical follicular cells subclassified as “rule out papillary carcinoma” was significantly higher (38%) than the other atypical cells, whereas the risk of “rule out Hurthle cell neoplasm” was significantly lower at 7% than the other cases of atypical follicular cells (P < .001 and P < .02, respectively). The 3 cases diagnosed as “rule out Hurthle cell neoplasm” that proved to be malignant were all scant aspirates composed exclusively of Hurthle cells.

Table 2. Risk of Malignancy for Subclassified Atypical Follicular Cells

Diagnosis	No. (%) of Cases	No. (%) of Cases With Histologic Follow-Up	No.(%) Benign Histology	No. (%) Malignant Histology
NOS indicates not otherwise specified.
Atypia, NOS	55 (1)	15 (27)	11 (73)	4 (27)
Atypia, rule out follicular neoplasm	184 (3)	73 (40)	57 (78)	16 (22)
Atypia, rule out Hurthle cell neoplasm	144 (2)	44 (31)	41 (93)	3 (7)
Atypia, rule out papillary carcinoma	165 (2)	72 (44)	45 (62)	27 (38)
Total	548 (14)	204 (37)	154 (75)	50 (25)

Cases classified as “NOS” did not have a significantly different risk of malignancy when compared with “rule out follicular neoplasm” or “suspicious for follicular neoplasm” (P = .84 and P = .52, respectively).

DISCUSSION

Study results suggest that the diagnosis of “atypical follicular cells” spans a heterogeneous group of cases with significantly different risks of malignancy. The author has found at least 4 different groups of cases within this diagnosis that can be separated based on distinct cytologic criteria, and at least 2 of these groups have significantly different risks of malignancy. These different risks of malignancy may be important for both cytologists and clinicians.

Although a finding of “atypical follicular cells” is separated from other “indeterminant” cases in thyroid FNA by the most recent Bethesda System for Reporting Thyroid Cytopathology,1 whether the overall risk of malignancy for this group of patients is truly different than that of patients with other indeterminant diagnoses is not clear. Other authors have pointed out that the risk of malignancy for patients with a diagnosis of “atypical follicular cells” is similar to that of patients with diagnoses of “suspicious for a follicular” or “Hurthle cell neoplasm.”2, 5 These authors have suggested that patients with these diagnoses should be grouped together into 1 group with a similar risk of malignancy. Where “atypical follicular cells” is not subclassified in the present series, the risk of malignancy is also similar to that of patients with a diagnosis of “suspicious for a follicular” or “Hurthle cell neoplasm.” However, this study shows that certain cases classified as “atypical follicular cells” have a significantly higher or lower risk of malignancy than other cases in this group. The author believes these individuals should be separated from the other patients in the group, and their specific level of risk should be appropriately communicated to their clinicians.

The author's results also have importance for cytologists diagnosing these cases. As the author has previously shown,10 the general criteria for diagnosing Hurthle cell neoplasms is too broad,1 and these criteria can be made significantly more specific with little loss in sensitivity. The study under discussion supports this. In the current series, for cases that did not have cytologic atypia as previously defined,10 risk of malignancy was very low. Therefore, the author suggests cytologists should strive to call these cases “Hurthle cell rich” without atypia and reserve the diagnosis of “atypical, rule out Hurthle cell neoplasm” for those cases where evaluation of atypia cannot be reliably made. In these situations, repeat aspiration may be of value.

Nevertheless, in the current series, 2 patterns of atypical cells with no significant difference in risk of malignancy emerged. Risk of malignancy for these 2 patterns (“NOS” and “rule out follicular neoplasm”) was similar—and was similar to the risk for a diagnosis of “suspicious for a follicular neoplasm.” If the diagnosis of “atypical follicular cells” is restricted to these 2 groups of cases, then the only justification for separating them from each other, or from “suspicious for a follicular neoplasm,” would be when a repeat aspiration could significantly change this risk in 1 group and not in the other.12

In conclusion, the author has shown that, in thyroid FNA specimens, the diagnosis of “atypical follicular cells” can be stratified on the basis of significantly lower and higher risks of malignancy. Cytologists should strive to better classify “atypical follicular cells” and better communicate the true risk of malignancy for each individual patient.

CONFLICT OF INTEREST DISCLOSURES

The author made no disclosures.