Impact of Flow Cytometry in Thyroid Cytopathology

Fine-needle aspiration (FNA) is a useful diagnostic tool for preoperative evaluation of thyroid nodules [1,2]. FNA is most accurate for the diagnosis of primary thyroid neoplasms, probably because these are common relative to the other malignancies that involve the thyroid, have well-defined cytomorphologic criteria and also due to the uncertainty that is inherent in many such specimens being well characterized in The Bethesda System for Reporting Thyroid Cytopathology [3,4]. Lymphoma rarely involves the thyroid; 0.3% of all malignancies in the thyroid are lymphoma [5]. Conversely, 0.2% of all non-Hodgkin lymphomas involve the thyroid [6]. Thus, the observation that cytomorphology alone is a poor diagnostic modality for lymphoma in the thyroid [7] is not surprising. The sensitivity of cytomorphology for lymphoma in the thyroid is dubious at best [8,9], and the specificity of morphology alone is thought to be poor, although it must be noted that a comprehensive study has not been published to date.

Given the limitations of establishing a lymphoma diagnosis with cytomorphology, morphological evaluation is commonly augmented with flow cytometry or immunohistochemistry. In two studies, morphology in conjunction with flow cytometry or immunohistochemistry had a positive predictive value of 90% [10,11]. This is comparable to the publications of others who had reported experience with excisional material, such as an open biopsy or lobectomy [12,13]. Regardless, the decision to perform flow cytometry in daily practice may not be straightforward, especially in the setting of Hashimoto thyroiditis. In this article, we review our experience with performing flow cytometry on thyroid FNA and demonstrate how that experience correlated with surgical follow-up at The Johns Hopkins Hospital.

After obtaining institutional review board approval, a Boolean search of the pathology database was conducted for ‘thyroid', ‘fine-needle aspiration' and flow cytometry or immunophenotyping studies from January 1993 to December 2012. A search of the thyroid surgical pathology database was also performed to determine which resected thyroid specimens contained lymphoma. The starting date was chosen because immunohistochemistry and flow cytometry have been used routinely on cytologic material since 1993 [7]. During this period, 11,118 thyroid FNAs were performed. These were read using a custom perl script that excluded cervical lymph nodes and populated searchable fields such as size, location and patient demographics in a Filemaker database (Santa Clara, Calif., USA). The Filemaker database cases were manually reviewed. Of the 11,118 biopsies performed, 49 underwent successful testing with flow cytometry, 13 underwent a failed attempt at flow cytometry and 2 cases did not undergo flow cytometry but were found to have atypical lymphoid cells sufficient for a thyroidectomy that resulted in a lymphoma diagnosis. Seven cases identified by the Boolean search terms were excluded from analysis because they merely contained a suggestion to perform flow cytometry due to contamination of lymphocyte-rich peripheral blood. Thus, the final retrospective review contained 64 cases. The electronic medical records were reviewed for this group, and the presenting symptom/sign, presence or absence of a lymphoma history, follow-up status and surgical follow-up were recorded.

Ultrasound-guided FNA with or without local anesthesia was performed by a radiologist or endocrinologist. Two direct smears were prepared from each needle pass; one smear slide was stained with Diff-Quick® for adequacy evaluation and another smear slide was immediately fixed in 95% alcohol and stained with Papanicolaou stain. The cytologic material was immediately evaluated for adequacy by either a cytotechnologist or cytopathologist as described [14]. Flow cytometric analyses were performed on the rinse material. Cytopathologic diagnosis was made on all cytologic materials by a board-certified cytopathologist. Additional passes were submitted in Hanks balanced salt solution, if possible, for flow cytometry analysis. This material was processed using red cell lysis. Cell suspensions were incubated with combinations of monoclonal antibodies; in most cases the panel included antibodies specific for CD3, CD20, kappa and lambda. Additional markers included CD10, CD34 and CD71. Specimens were analyzed on a BDIS FACSCalibur flow cytometry system. Flow cytometry was interpreted by a board-certified hematopathologist.

There were 11,118 thyroid FNAs performed in the review period, and the 64 cases (0.6%) that were included in this study were those for which flow cytometry was attempted (n = 62) or where surgery later revealed lymphoma in a cytomorphologically suspicious lesion that did not undergo flow cytometry (n = 2). The demographic data for the patients is shown as table 1. There was a predominance of white (46/64, 71.9%) females (53/64, 82.8%), and the median age was 47.5 years (range 7-85 years). Most of the cases (53/64, 82.8%) presented with thyroid nodules. Of these, 5 were known to be symptomatic. The size of the nodule was known in 14 cases - a smaller number than the total because the practice of recording the nodule size of the needle placement is relatively recent at our institution - and the median size of those cases was 1.8 cm (range 0.7-9.0 cm). All 62 patients in this study either had a morphologically suspicious FNA or a clinical history of hematolymphoid malignancy such as diffuse large B cell lymphoma (DLBCL, n = 3), follicular lymphoma (n = 2), mucosa-associated lymphoid tissue lymphoma (n = 1), precursor T-lymphoblastic lymphoma (n = 1) and leukemia (n = 1). Material explicitly destined for flow cytometry was obtained during the on-site evaluation; diagnostic information was obtained for 49 cases (49/62, 79%). In 13 cases (13/62, 21%), flow cytometry was attempted, but diagnostic information could not be obtained due to inadequate cellularity.

Lymphoma was diagnosed 13 times. The methods used to arrive at this diagnosis are shown in figure 1, and the specific diagnoses are shown in table 2. In 8 cases, both cytomorphology and flow cytometry arrived at the correct diagnosis. Cytomorphology alone made the correct diagnosis twice in the absence of flow cytometric results; flow cytometry made the diagnosis once in the absence of suspicious morphology. Neither flow cytometry nor cytomorphology made the correct diagnosis in 2 sparsely cellular cases. The group in which lymphoma was diagnosed without flow cytometry consisted of 2 cases that showed substantial morphological atypia but did not have dedicated passes available for flow cytometry. One case underwent surgical resection that led to the diagnosis of DLBCL. The diagnosis in the other case was made without flow cytometry using morphology and immunohistochemistry on a cell block. This demonstrated a clonal population sufficient for the diagnosis of large B cell lymphoma. Of the 8 cases in which both flow cytometry and cytomorphology were suggestive of lymphoma, 2 showed that the clonal population consisted of larger cells, but the cellularity was inadequate for further work-up. Both patients underwent resection, resulting in a diagnosis of myeloma for one and mucosa-associated lymphoid tissue lymphoma (MALT) lymphoma for the other. In the remaining 6 cases that were both morphologically and flow-cytometrically suspicious, there was 1 case of precursor T lymphoblastic lymphoma and 5 diagnoses of large B cell lymphoma. None of these had surgical follow-up. The single case of lymphoma diagnosed by flow cytometry in the setting of benign morphology showed immunophenotypic abnormalities that implied plasmacytic differentiation: a loss of CD20 expression and partial loss of CD45 expression. The cellularity was inadequate for further work-up, and this patient underwent resection of a MALT lymphoma with extensive plasmacytic differentiation. The 2 cases that were neither morphologically nor immunophenotypically diagnostic consisted of 1 that was inadequate for morphological analysis and 1 that was marginally adequate for flow; both had surgical resection due to clinical suspicion, and both resections showed DLBCL.

The relationship between Hashimoto thyroiditis, flow cytometry and the presence of a clone is shown in figure 2. In this series, 1,009 patients who underwent thyroid FNA had cytologic or clinical evidence of Hashimoto thyroiditis. Of these, 33 (3%) had flow cytometry performed, and 3 of these 33 (9%) demonstrated a clonal population. One patient underwent resection that revealed a MALT lymphoma. This case corresponds to case 4 in table 2. The remaining 30 cases underwent flow cytometry and did not show a clone. Five of these patients underwent surgery. Four of the resection specimens showed benign histologic findings and one showed DLBCL. This lymphoma case corresponding with case 12 in table 2 had been signed out as a less than optimal specimen with very scant cellularity. A rapidly enlarging mass was acknowledged in the clinical history. Regardless, considering this case as a false-negative out of the 30 in which both morphology and flow cytometry were benign, yielded a negative predictive value estimate of 97 ± 18%.

Hashimoto thyroiditis is a common finding in thyroid cytopathology that is commonly associated with suspicious epithelial features such as oncocytic change and nuclear clearing, grooves and pseudoinclusions [15]. These changes lead to poor inter-observer reproducibility [16]. However, as we have shown here, a small subset of aspirates with Hashimoto thyroiditis demonstrate lymphocytic atypia and raise the possibility of lymphoma in the thyroid. In such cases, flow cytometry is an invaluable ancillary tool that clearly increases the diagnostic yield. Indeed, the low rate of surgical excision in our institutional experience demonstrates that flow cytometry and cytomorphology often yield summative actionable information without the need for additional diagnostic procedures. The value is perhaps greatest when the diagnostic dilemma is between low-grade lymphoma and Hashimoto thyroiditis, as negative flow cytometry has a relatively high negative predictive value which obviate the need for more invasive procedures if these are not indicated by the imaging or clinical impression. In this study, there are a few cases in which the sensitivity of flow cytometry led to the discovery of a benign clonal population. While the small sample size makes generalizations impossible, the limited data suggest that coexistence of the clone together with other atypical cytometric findings, such as large size, can aid in discerning benign clones from those that may be cause for concern. This series contains only 1 case in which DLBCL was subsequently discovered. While this is a false-negative, it is also an example of how scant cellularity can obscure the diagnosis even with a method as sensitive as flow cytometry. High adequacy standards are necessary in order to definitively determine that a specimen is benign.

In contrast to the distinction between low-grade lymphoma and Hashimoto thyroiditis, the diagnosis of large cell lymphoma is often straightforward morphologically, with practising cytopathologists keenly aware of the morphological features as it is the most common lymphoma [6] of the thyroid. Given that large cell lymphoma is most commonly treated with chemotherapy rather than excision, noninvasive diagnosis is most desirable. As shown in this study, the diagnosis of DLBCL with flow cytometry and cytomorphology is a reliable and actionable diagnosis. The few cases of discordance between flow cytometry and morphology are inadequate to make generalizations. However, given the discovery of lymphoma in these rare instances, the pursuit of additional tissue studies seems prudent. The retrospective nature of this study led to a couple of important limitations. First, most of these cases lacked surgical follow-up. Indeed, only 1 of 7 (14%) diagnosed as lymphoma based on cytology and flow cytometry had surgical follow-up. This is a reflection of how management of lymphoma differs from other thyroid malignancies [17]. Given the increased weight given to cytological diagnoses and the role of chemotherapy and radiation in the management of several cancers, FNA can be the last specimen that certain patients yield [18]. Instead, the observed clinical response of these patients (data not shown) to the appropriate chemotherapy is a natural history confirmation of the diagnoses even though formal outcome modeling is beyond the scope of this paper. Second, only 3% of the patients with Hashimoto thyroiditis had flow cytometry performed on their specimens, so the true incidence of benign clones may have been underestimated. The limited numbers do show that nuanced flow cytometry results, i.e. the finding of a clone on large or otherwise immunophenotypically abnormal cells warrants further attention. However, the risk of harm, from flow cytometry and the initiation of the work-up of a benign clonal population, is not clear from this study.

In conclusion, the combination of cytomorphology and flow cytometry is a useful diagnostic modality for the work-up of lymphocyte-rich fine-needle aspirates from thyroid showing atypical lymphoid populations.