The interobserver variability of thyroid fine needle aspirates using The Bethesda System for Reporting Thyroid Cytopathology:

Introduction

Thyroid diseases are among the commonest endocrine disorders worldwide. About 42 million people in India suffer from thyroid diseases.1 Fine needle aspiration cytology (FNAC) is the first‑line diagnostic procedure for evaluating thyroid lesions. It is a simple, rapid, cost-effective test that provides valuable information about the nature of a thyroid nodule and can effectively distinguish between neoplastic and non‑neoplastic lesions of the thyroid therefore guiding appropriate management i.e. follow-up or surgery thus reducing unnecessary surgery for patients with benign disease.4, 3, 2

Reporting of thyroid FNA specimens should follow a standard format that is clinically relevant in order to direct appropriate management.5 The terminology for reporting thyroid FNACs has varied markedly worldwide with use of many reporting formats varying from two category schemes to six or more category schemes with some relying on descriptive phrases instead of categories.4, 3 This lack of uniformity creates confusion amongst referring clinicians in the interpretation of thyroid cytopathology report, thereby affecting definitive clinical management.2

To address terminology and other issues related to thyroid FNA, the National Cancer Institute (NCI) hosted the "NCI Thyroid Fine Needle Aspiration State of the Science Conference", a multidisciplinary conference that took place in Bethesda, Maryland, United States in October 2007,that led to the formation of The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC), a six-category scheme for reporting thyroid cytopathology that recommends each report should begin with a general diagnostic category. Each of the categories has an implied risk of malignancy that links it to a rational clinical management guideline.6

Recently published data support the clinical utility and wide acceptance of TBSRTC by both practicing pathologists and clinicians. In Indian perspective too, many studies have been carried out that have concluded TBSRTC to be an effective reporting system for thyroid FNA. It improves perceptions of diagnostic terminology between cytopathologists and clinicians and also provides clear management guidelines to the clinicians.4, 2 In our institution, reporting of thyroid lesions on FNA is variable and no standardized system of reporting is being followed. The present study was thus carried out to analyse the cytological features of thyroid fine needle aspiration smears and categorise them by The Bethesda System for Reporting Thyroid Cytopathology; and to assess interobserver variability between two independent reporting pathologists using The Bethesda System for Reporting Thyroid Cytopathology.

Materials and Methods

The present study was conducted in the Department of Pathology, Government Medical College, Jammu over a period of three years from Nov.2015 to Oct.2018. It included all patients presenting with thyroid swelling referred to this department from various clinical departments of this hospital and from other health care centers. Non-cooperative and morbid patients were excluded from the study.

Smears were prepared from the sample obtained by aspiration or non-aspiration method and were stained with May-Grunwald-Giemsa (MGG) and Papanicolau (PAP) stains. Stained smears were examined under light microscopy by two independent pathologists in a double blinded fashion. The cytological features were evaluated and reporting was done according to The Bethesda System for Reporting Thyroid Cytopathology (TBSRTC).

The definitions and cytomorphological criteria as described in the, The Bethesda System for Reporting Thyroid Cytopathology atlas were followed by the two reporting pathologists (Pathologist I and Pathologist II). After assessing the adequacy as per TBSRTC adequacy criteria, the thyroid fine needle aspirates were categorized into the six categories of The Bethesda System for Reporting Thyroid Cytopathology.7

Results

A total of 610 patients were included in the study. The thyroid fine needle aspirates were categorized into the six categories of The Bethesda System for Reporting Thyroid Cytopathology by two independent pathologists.

Pathologist I reported 51 cases (8.36%) as Nondiagnostic, 524 cases (85.90%) as Benign, 2 cases (0.33%) as Atypia of Undetermined Significance, 15 cases (2.46%) as Follicular Neoplasm, 1 case (0.16%) as Suspicious for Malignancy, and 17 cases (2.79%) as Malignant while Pathologist II reported 50 cases (8.20%) as Nondiagnostic, 510 cases (83.61%) as Benign, 12 cases (1.97%) as Atypia of Undetermined Significance, 16 cases (2.62%) as Follicular Neoplasm, 6 cases (0.98%) as Suspicious for Malignancy, and 16 cases (2.62%) as Malignant (Table 1) (Figure 6, Figure 5, Figure 4, Figure 3, Figure 2, Figure 1).

Benign category was the largest category followed by Nondiagnostic category. Benign follicular nodule was the predominant subcategory followed by chronic lymphocytic thyroiditis as reported by both pathologists. Papillary Thyroid Carcinoma was the most common malignancy reported by both the pathologists in our study (Table 1).

The two pathologists were in agreement in 548 cases (89.84%) and disagreed in 62 cases (10.16%). Diagnostic agreement was highest in the Benign category (95.68%), followed by Malignant category (87.5%), Follicular Neoplasm (75%), Nondiagnostic (68%). No agreement was seen in AUS category and Suspicious for Malignancy category (0%, 0% respectively) between the two reporting pathologists (Table 2).

Discordant cases of Nondiagnostic category by one pathologist were categorized in Benign category by other pathologist. Discordant cases in Benign category were put into Non diagnostic, AUS, Follicular neoplasm and Suspicious for Malignancy categories by other pathologist. Of the two cases in AUS category by one pathologist, there was disagreement in both the cases with one case reported in Benign category and other case in Suspicious for Malignancy category by other pathologist. Of the 15 cases in Follicular neoplasm category by one pathologist, there was disagreement in 3 cases with 2 cases reported in Benign category and 1 case in Malignant category by other pathologist. Of one case in Suspicious for Malignancy category by one pathologist, there was complete disagreement with the case being reported in Malignant category by other pathologist. Of the 17 cases in Malignant category by one pathologist, there were 3 discordant cases, 2 cases in Benign category and 1 case in Suspicious for Malignancy category by other pathologist (Table 2).

The interobserver variability was statistically assessed using Cohen's kappa coefficient with observed value of 0.628; 62.8% (95% Confidence Interval 54.0-71.6) i.e. actual agreement between the two pathologists beyond chance was 62.8% (moderate) in our study.

Figure 1

Photomicrograph from a case of Colloid Nodule showing abundant colloid with few follicular cells (MGG 40 X).

Figure 2

Photomicrograph from a case of Lymphocytic Thyroiditisshowing sheet of follicular cells infiltrated by mature lymphocytes along with numerous lymphoid cells in the background (MGG 400 X)

Figure 3

Photomicrograph from a case of AUS showing few clusters as well as singly scattered follicular cells with mild anisonucleosis and focal microfollicle formation (PAP 400 X).

Figure 4

Photomicrograph from a case of Follicular Neoplasm showing follicular cells arranged in crowded clusters and microfollicles (PAP 400 X).

Figure 5

Photomicrograph from a case of Papillary Thyroid Carcinoma (Bethesda Malignant category) showing multiple complex papillae with peripheral palisading of the tumor cells (PAP 100 X).

Figure 6

Photomicrograph from a case of Papillary Thyroid Carcinoma showing optically clear nuclei with intranuclear grooves (PAP 400 X).

Discussion

TBSRTC has been widely adopted in the United States and in many places worldwide and has been endorsed by the American Thyroid Association.8, 7 The distribution of cases into various TBSRTC categories by the two pathologists in our study were compared with other studies as shown in Table 3.

The frequency of Nondiagnostic interpretations varies notably from laboratory to laboratory (range, 3-34%).7 The findings of our study (8.36%, 8.20%) are consistent with this range and are comparable with other studies.9, 4 As most thyroid nodules are benign, a benign result is the most common FNA interpretation (approximately 60-70% of all cases).7 The cases in benign category in our study are higher than this range (85.90%, 83.61%) and also from study by Jo et al.10 but are comparable with other studies.11, 9, 4, 2 Being the only tertiary care center of our province, it caters to a large number of patients on both direct and referral basis, so a large population representative of general population is encountered that could be a reason for higher number of cases in benign category.

AUS/FLUS usage varies from 1% to 22% of thyroid FNAs with practical upper limit of 10%.7 AUS cases and Follicular Neoplasm cases reported in our study are comparable with studies in Table 3. Suspicious for Malignancy (SFM) diagnoses account for approximately 3% (range 1.0-6.3%) of all thyroid FNAs. As with any indeterminate diagnosis, this category should be used judiciously so that patients are managed as appropriately as possible.7 The SFM cases in our study as reported by Pathologist I are lesser than the lower limit of the given range (0.16%) but are comparable with study by Laishram et al.,11 while are 0.98% i.e. ~1% as reported by Pathologist II which are comparable with other studies.4, 2 A malignant thyroid FNA diagnosis accounts for approximately 5% (range, 2-16%) of all thyroid FNAs.7 Malignant cases (Category VI) in our study are within this range (2.79%, 2.62%) and comparable with studies in Table 3.

The two pathologists were in agreement in 548 cases (89.84%) and disagreed in 62 cases (10.16%) which is consistent with other studies.15, 14, 13, 12 Cohen's kappa score of 0.628 is comparable to kappa score in a study by Awasthi et al.12 In a study by Salillas et al.,14 the strength of agreement was very good with a kappa statistic of 0.90. In a study by Pathak et al.,15 unweighted Cohen's kappa score for consultant and SR was 0.7517 (strong), between consultant and JR was 0.5907 (moderate).

Diagnostic agreement was highest in the Benign category (95.68%), followed by Malignant category (87.5%) in our study. In a study by Bhasin et al.,13 maximum degree of agreements were noted in nondiagnostic (100%), malignant (100%) and benign (93.87%) categories. In a study by Awasthi et al.,12 there was absolute agreement on the cases in ND/US and AUS/FLUS categories followed by 94.8% concordance in benign category. Lesser number of cases reported under ND/US category in above two studies could be reason for absolute agreement in ND category in their studies. In a study by Salillas et al.,14 the category under which maximum degree of agreement was noted was malignant (100%, k=0.61) followed by benign (k=0.60).

In our study, no agreement was seen in AUS (0%) category and SFM (0%) category between the two reporting pathologists. In a study by Bhasin et al.,13 maximum disagreement was noted in AUS/FLUS category that is consistent with our study. Of 7 discordant cases in study by Salillas et al.,14 3 were SFM and 2 AUS which is consistent with our study. In a study by Pathak et al.,15 poor interobserver agreement level (K=0.1301) was observed in the AUS/FLUS category which is consistent with our study. In a study by Kocjan et al.,16 there was poor agreement for Thy3a (κ = 0.11) and Thy4 (κ = 0.17) categories which are similar to AUS and SFM categories respectively of Bethesda system and comparable to our study.

Low reproducibility for AUS/FLUS has been reported, variability in criteria used for AUS/FLUS is responsible for significant interobserver and inter-institutional variation in making diagnoses.15 The reproducibility of AUS/FLUS category is at best only fair.7 Significant interobserver variability of thyroid FNA though well established has been reported to be smaller (although still significant) for "benign" and "malignant" categories. AUS/FLUS category is one with the most interobserver variability among the cytopathologists. Differing threshold levels in applying the diagnostic criteria as noted in one of the studies could be a reason for this variability in our study too.17

Conclusion

The findings of our study are consistent with other published studies in literature. The systematic reporting according to TBSRTC has led to clear interpretation of the thyroid FNAC report. The use of this uniform terminology by two reporting pathologists revealed a moderate interobserver agreement that favours its use because of its relative ease of reproducibility. It also provides malignancy risk and management guideline for each category. Thus, the present study encourages the use of TBSRTC as a standardized reporting system in our institution and elsewhere for effective communication among pathologists and clinicians with regard to thyroid FNAC reporting and management.

Conflict of interest

The authors have no conflicts of interest to declare.

Source of funding

The study was not funded by any source.