Gray zone Bethesda category III – Atypia of undetermined significance lesions of the thyroid: Potential diagnostic issues and image morphometry as a useful adjunct to cytomorphology

INTRODUCTION

The fine-needle aspiration (FNA) technique is a proficient, time-saving, and simple technique that can be used as a first-hand screening tool for thyroid lesions. Reporting of thyroid nodules is done by the universal 6-tiered Bethesda system. Atypia of undetermined significance (AUS) has compromised morphology and is reserved for cases with cytopathological features that are insufficient to be classified under higher Bethesda category but showing more atypia to put under benign category.^[1] The diagnosis of follicular-patterned lesions is challenging for the cytopathologist.

The final histopathology outcome of AUS cases ranges from colloid goiter to carcinoma. Most importantly, there is substantial evidence that the actual risk of malignancy (ROM) of AUS is significantly higher than the estimated incidence. Therefore, it is crucial to identify the peculiar cytopathological features to determine the likelihood of malignancy and guide clinical decision-making.^[2] Nonetheless, cytomorphological assessment is subjective with interobserver variations when reporting thyroid cytopathology especially in AUS cases. Computerized nuclear morphometry is an objective, economical, and consistent tool to detect changes that are too small to be visually perceived and could be used constructively to classify these gray zone lesions of the thyroid. Fine-needle aspiration cytology (FNAC) when corroborated with morphometry can prove beneficial in AUS category and, thus, can aid a cytopathologist in settling cases with diagnostic dilemma.^[3] Literature on this is scarce.^[4,5]

A timely informed and correct cytological diagnosis is fundamental to triaging patients with thyroid nodules and minimizes non-essential surgeries. In the present study, we retrospectively compared the cytomorphological features of histopathology-proven neoplastic and non-neoplastic AUS cases (histopathology proven/clinically stable) and assessed the cytomorphological features contributing to diagnostic pitfalls. Further, nuclear morphometry was done on FNAC smears to assess the role of image morphometry in delineating benign from malignant outcomes in this category.

MATERIAL AND METHODS

The present study is a retrospective study carried out in accordance with the guidelines laid down in the Declaration of Helsinki. The research was carried out according to the Institutional Ethics Guidelines after obtaining due approval of the Departmental Research Ethics Committee. The inclusion and exclusion criteria were outlined as follows: -

RESULTS

A total of 61 patients were included in the study [Figure 1]. There were 47 women and 14 men, with age varying from 20 years to 71 years. This study had 35 neoplastic cases (histopathology proven), risk of neoplasia (RON) being 57.4% and 26 nonneoplastic (on follow-up) cases of category III. These 35 neoplastic cases were distributed among papillary thyroid carcinoma (PTC) (23), follicular adenoma (FA) (6), follicular carcinoma (FC) (3), non-invasive follicular thyroid neoplasm with papillary-like nuclear features (NIFTP) (2), and poorly differentiated thyroid carcinoma (1). The ROM was 47.5%. FNAC smears of AUS with neoplastic and non-neoplastic outcomes revealed a similar trend in cellularity, microfollicle formation [Figure 2], oncocytic change, and fire flares. There was significant conflicting variation for other cytomorphological features. Neoplastic cases displayed prominent intranuclear inclusions (54%) and pseudopapillary clusters (20%) [Figure 3] in signification proportion as compared to non-neoplastic cases that showed these features in a minority of cases constituting 8% and 7%, respectively. Regarding nuclear features, neoplastic cases showed moderate to marked nuclear pleomorphism (20%) compared to non-neoplastic cases, which were monomorphic to mildly pleomorphic. Majority of non-neoplastic cases revealed fine chromatin (96%) and pale chromatin (4%), while in neoplastic cases, 14% showed pale chromatin. None of the non-neoplastic cases exhibited frequent nuclear overlapping, frequent nuclear grooving, or nucleoli, pronouncing that cytopathologist should take utmost care of these features while ascertaining the diagnosis. Non-neoplastic cases exhibited more colloid than neoplastic cases, as shown in Table 1.

Close

Figure 1:

Work flow of current study with histopathological outcome. (AUS: Atypia of undetermined significance, FLUS: Follicular lesion of undetermined significance, NIFTP: Non-invasive follicular thyroid neoplasm with papillary-like nuclear features). This flowchart was created using the Intelligent Diagramming Lucidchart, accessible at: https://www.lucidchart.com (Accessed July 19, 2024).

Export to PPT

Close

Figure 2:

(a and b) Non-neoplastic. (a) Smear showing microfollicles in a case of adenomatous nodule (MGG, ×100). (b) Aspirate smear showing a pseudo papillary cluster of follicular epithelial cells in colloid goiter (MGG, ×100). (c and d) Neoplastic. (c) Plentiful microfollicles in follicular adenoma (H&E, ×400). (d) Cohesive cluster of cells in papillary thyroid carcinoma. The cluster is avascular and the peripheral cells are irregularly arranged giving rise to vague papillae like arrangement. (MGG, ×400). (MGG: May Grunwald Giemsa stain, H and E: Hematoxylin and Eosin).

Export to PPT

Close

Figure 3:

(a and b) Non-neoplastic. (a). Smear showing cluster and few scattered follicular epithelial cells with occasional intranuclear inclusion and an occasional nuclear groove in lymphocytic thyroiditis (MGG, ×400). (b) Aspirate smear showing follicular epithelial cells with mild pleomorphism (MGG, ×400). (c and d) Neoplastic. (c) Intranuclear inclusion in papillary thyroid carcinoma (MGG, ×400). (d) An occa sional sheet and few scattered follicular epithelial cells with mild pleomorphism in papillary thyroid carcinoma (MGG, ×400). (MGG: May Grunwald Giemsa stain, H and E: Hematoxylin and Eosin).

Export to PPT

Image morphometry

The average time from image acquisition to final output was about 8–10 min for each case with additional 2–3 min required for putting the data into Excel sheets for analysis. A representative case of AUS with benign outcome is shown in Figure 4. Figure 5 demonstrates the image morphometry of an AUS with malignant outcome (PTC). All the nuclear morphometric features were highly significant between the malignant and benign groups of AUS cases. The mean nuclear area, standard deviation of nuclear area (SDNA), nuclear perimeter, diameter, and integrated density in the malignant group were 26.205 ± 8.195 µm² (mean ± standard deviation), 8.642 ± 3.059 µm, 23.988 ± 3.901, 7.308 ± 1.037, and 3460.521 ± 1136.319, respectively. The statistical analysis showed that these values were significantly higher than those observed in the benign group, as depicted in Table 2. P values observed were 0.004, 0.058, 0.293, 0.124, and 0.001 for mean nuclear area, SDNA, nuclear perimeter, diameter, and integrated density using Levene’s test for equality of variances and were < 0.001 using t-test for equality of means.

Close

Figure 4:

Image morphometry in atypia of undetermined significance lesion of thyroid with the outcome of follicular nodular disease (colloid goiter). (a) Captured image is displayed and the calibration is set up. (b) The colored image is converted into a 8-bit gray image. (c) The nuclei are detected by adjusting the gray threshold. (d) The individual nuclei are detected (arrows) and morphometric parameters are measured.

Export to PPT

Close

Figure 5:

Image morphometry in a case of atypia of undetermined significance lesion with the outcome of papillary thyroid carcinoma.

Export to PPT

Table 2: Image morphometric data in Category III AUS lesions of the thyroid.

Parameter	Benign outcome	Malignant outcome	Levene’s test for equality of variances (P-value)
Mean area (µm2), nucleus (±Standard deviation)	16.222 4.533	26.205 8.195	0.004
Mean standard deviation of nuclear area (µm2) (±Standard deviation)	5.441 2.258	8.642 3.059	0.058
Mean perimeter (µm), nucleus (±Standard deviation)	17.965 3.386	23.988 3.901	0.293
Mean nuclear diameter (µm) (±Standard deviation)	5.670 0.818	7.308 1.037	0.124
Mean integrated density, nucleus (±Standard deviation)	2188.860 735.687	3460.521 1136.319	0.001

AUS: Atypia of undetermined significance, µm: Micrometer

DISCUSSION

This paper outlines the specific cytomorphological characteristics of the Bethesda category III, AUS lesions of the thyroid and places emphasis on common diagnostic errors. According to our analysis, significant attention should be paid to substantial intranuclear inclusions, pseudopapillary clusters, and nuclear traits like pale chromatin and nuclear pleomorphism since they connect to a higher category lesion. Although there is some subjectivity in applying the Bethesda system, the AUS instances are the most perplexing. They should not be treated as a waste basket, and before reaching a final diagnosis, weightage should be given to cytomorphological features. Further, the subjectivity inherent to cytomorphological evaluation is non-existent for image morphometry, which was assessed as an additional objective tool in our current work.

Many authors have tried to analyze the risk of malignancy (ROM) in AUS and subclassify the category.^[9-14] The variability in results across different set-ups may be attributed to an element of subjectivity. Results from different centers may differ due to differences in FNA procedures, needle size, ultrasound guidance use, and the cytopathologist’s competence. At many centers, FNAC is done by the radiologist or clinician rather than the pathologist. Differences in smear preparatory methodology are yet another cause. A repeat FNA is a recommendation for an initial diagnosis of category III, and molecular testing should be introduced according to the latest Bethesda. A repetitive diagnosis of category III is seen in 20–28% of thyroid lesions.^[15] Nonetheless, a repeat FNA is yielding with a definitive diagnosis invariably.

Our tertiary care facility holds multidisciplinary meetings fortnightly where pathologists and radiologists participate, and cases are discussed to determine the best further course of action wherever required. Our study emphasizes better cytomorphological criteria that support malignancy or benignity and avoiding diagnostic pitfalls that can lead to under or overdiagnosis and plan patient management accordingly. In our study, 23 out of 61 cases turned out to be PTC (37.7%), followed by FA (6) and FC (3), NIFTP (2), and one case of poorly differentiated thyroid carcinoma. In our study, most neoplastic cases displayed intranuclear pseudo-inclusions (54%), prominent nucleoli (46%), and pseudopapillary clusters (papillaroid clusters of follicular epithelial cells lacking fibrovascular cores) and moderate to marked pleomorphism in 20% of cases. Majority of nonneoplastic cases revealed fine chromatin (96%) and pale chromatin (4%), while in neoplastic cases, 14% showed pale chromatin. None of the non-neoplastic cases exhibited frequent nuclear overlapping, frequent nuclear grooving or nucleoli pronouncing that cytopathologist should take utmost care of these features while ascertaining the diagnosis. While smears show pleomorphism, intranuclear pseudo-inclusions, pale chromatin, and nuclear grooves, the diagnosis may be TBSRTC category IV or TBSRTC category V based on cell quantity. Probably, one should focus both on cell quantity and degree of atypia to label a case TBSRTC category V.

Non-neoplastic cases exhibited more colloids as compared to neoplastic cases. FNAC smears of AUS with neoplastic and non-neoplastic outcomes revealed a similar trend in microfollicle formation, oncocytic change, and fire flares.

Image morphometry can be used as an adjunctive modality for making pre-operative assessments and its importance in gray zone areas of AUS cases in thyroid cytopathology which need additional ancillary techniques cannot be overemphasized. It is an inexpensive objective tool.^[16] It is precise and objective in comparison to the manual assessment of nuclear features which have subjectivity. In our study, image morphometry was able to delineate the malignant and benign outcomes of AUS cases. The RON and ROM observed were 57.4% and 47.5%, respectively. The average time from image acquisition to final output was about 8–10 min for each case with additional 2–3 min required for putting the data into Excel sheets. Compared to molecular testing, it is cheaper, less time-consuming, and not much demanding technically. Doing it on the cytology smears which show intact and well-preserved cells yields superior results than when done on histopathology sections.^[17]

Our observations on image morphometry in Category III AUS thyroid lesions are in line with Kang et al.,^[4] who performed morphometry in 84 cases of AUS which had histologic follow-up and noted a significant difference between the benign and malignant groups. Moreover, the nuclear irregularities in AUS/FLUS lesions may be helpful in differentiating benign from malignant thyroid lesions. Artacho-Pérula et al.^[5] studied 55 patients with FA, FC, and PTC by image analysis. An increase in nuclear size was noted for carcinoma as compared to adenoma. This increase achieved greater significance when volume-weighted mean nuclear volume was evaluated.

The most updated 2023 Bethesda System divides AUS into AUS with Nuclear Atypia and AUS-Other.^[6] The results of our study and the inferences drawn thereby are not affected by utilizing the new Bethesda III terminology. The recent 2022 WHO classification of thyroid neoplasms has many new/alternative terminologies, most characteristic one being thyroid FND, thereby avoiding terms such as multinodular goiter, hyperplastic nodule, neoplastic, or adenomatous hyperplasia.^[7] It is pertinent to note that the updated terminology of the latest TBSRTC is in line with the recent 2022 WHO classification of thyroid neoplasms.

The potential shortcomings or limitations of the present study include a relatively small sample size and a short duration of clinical follow-up which was 2 years. Nevertheless, the study is reproducible and paves way for building up of diagnostic algorithmic models especially in resource-limited set-ups based on robust morphometric findings observed in the study, as depicted in Table 2.

SUMMARY

The present study emphasizes on better cytomorphological criteria to classify the lesions of the thyroid either benign or malignant and prioritize cases that should be put in the Bethesda category III AUS when it is utmost needed for better management of patients. To reiterate, this category should not to be treated as a waste basket. Objective analysis using image morphometry and its utility as an adjunct to cytomorphological assessment in differentiating benign versus malignant outcomes in AUS lesions of the thyroid are demonstrated.