Cellient Atlas - Thyroid FNA

INTRODUCTION

Andrew H. Fischer, M.D

Note: The cell block images presented here were generated using a manual method during development of the automated instrument.

The Papanicolaou Society of Cytopathology recently proposed a uniform thyroid FNA classification system to minimize ambiguity between pathologists and clinicians [1]. In this chapter, we will use their proposed classification:

  • Unsatisfactory/non-diagnostic
  • Benign (includes hyperplastic/adenomatoid nodule and lymphocytic thyroiditis)
  • Cellular lesion, cannot rule out follicular neoplasm
  • Follicular neoplasm
  • Suspicious for malignancy (specify type)
  • Malignant (including papillary thyroid carcinoma, medullary carcinoma, anaplastic carcinoma and other tumors)

Conventional thyroid FNA cytology preparations provide the following diagnostic information:

  1. Nuclear morphology
  2. Microfollicular vs. macrofollicular arrangement, and amount of colloid.
  3. Character of colloid (watery vs. dense)
  4. Presence of macrophages with hemosiderin
  5. Presence or absence of lymphocytes
  6. Variation in cell morphology from group to group

As described below, nuclear morphology has to be examined in order to diagnose or exclude papillary thyroid carcinoma. Whether a lesion is microfollicular or macrofollicular relates directly to the amount of colloid and the proportion of follicular cells to colloid. In general, follicular type neoplasms are microfollicular with scant dense colloid [2]. Papillary thyroid carcinomas may be either micro or macrofollicular [2]. Benign follicular nodules are generally macrofollicular and very commonly show cystic degeneration [2, 3]. The cystic degeneration in benign follicular nodules appears as though it has a component of venous stasis, and exhibits edematous, watery colloid together with hemosiderin. Papillary thyroid carcinoma also often shows macrophages with hemosiderin [2]. However, such cystic degeneration is reportedly very uncommon in follicular neoplasms [2, 3], and this has also been our experience. Therefore, once papillary thyroid carcinoma is excluded (based on the absence of the characteristic nuclear features described below), the presence of hemosiderin is very strong evidence in favor of a benign follicular lesion. Lymphocytes, often with plasma cells and fragments of germinal centers, are seen in lymphocytic thyroiditis, but they are rare in follicular neoplasms [2], even follicular neoplasms arising in the setting of lymphocytic thyroiditis. Lymphocytes and plasma cells are, however, relatively common in papillary thyroid carcinoma. Benign follicular lesions will typically show some variation in the cytologic features from group to group, rather than a monomorphic population.

Cell blocks provide all of the diagnostic information listed above, with the exception of very watery colloid. In our experience, the presence of hemosiderin in a follicular-type lesion is virtually always associated with watery colloid. Solid colloid fragments tend to be more abundant in the cell block than in the corresponding ThinPrep®.

Core biopsy can be synergistic with FNA for accurate evaluation of thyroid nodules [4, 5]. Likewise, combining cell blocks with FNA has been found to be useful, disclosing diagnostic patterns that may not be evident in monolayers [6, 7]. Kern found that 1/5 of thyroid FNA’s had diagnostic material only in the cell block sections and not in smears [8]. In the following, the diagnostic synergy between monolayer preparations such as ThinPreps® and cell blocks are illustrated. ThinPrep will present the smaller fragments that are easy to study without histologic sectioning. The remaining largest fragments are captured in the cell block and enhance the ability to see how cell morphology varies over larger-scale distances, useful for the following diagnostic challenges:

  • Lymphocytic thyroiditis with Hürthle cell change vs. Hürthle cell neoplasm
  • Lymphocytic thyroiditis with reactive follicular cells vs. papillary thyroid carcinoma
  • Benign follicular lesion with reactive follicular cells vs. papillary thyroid carcinoma
  • Microfollicular groups in a goiter vs. follicular neoplasm.

This chapter illustrates some of these advantages. In general, non-neoplastic Hürthle cell changes, or focal reactive-type nuclear clearing suggestive of papillary thyroid carcinoma, will show a spectrum of changes that blend over the span of a few follicles with normal-appearing follicular cells. In cell blocks, one may have the opportunity to see how the reactive changes are related spatially to areas of lymphocytic destruction of the follicles, or areas in a benign colloid nodule that have become denuded to evoke the reparative reaction resembling papillary thyroid carcinoma. In contrast, neoplastic Hürthle cells or papillary thyroid carcinoma will show an abrupt (« clonal-appearing ») transition when compared to benign follicular cells, without any apparent relation to a lymphocytic infiltrate.

Cell blocks allow the follicle size to be judged in a manner that complements ThinPrep slides. For a follicle to be considered large, the follicular cells need to be arranged 2-dimensionally over a fairly broad distance; if the follicles are small, the cells will not present in a broad 2-dimensional flat sheet. Surprisingly, it is difficult to find a consensus on the definition of « macrofollicular » [9]. Eight to 10 follicular cell diameters appears to be a reasonable definition for the macrofollicular colloid fragments [9], or for the diameter of flat, 2-dimensional macrofollicular sheets. In tissue sections that cut perpendicular to a macrofollicular group, macrofollicles will appear as a linear arrangement of at least 8 follicular cells. If the section cuts in the plane of a macrofollicular group, one can occasionally encounter a solid-appearing mass of cells analogous to a tissue section that cuts tangentially right through the ductal cell layer in a breast biopsy. Just as surgical pathologists can recognize such tangential sectioning and distinguish this from ductal hyperplasia, one can recognize tangential sectioning through macrofollicular groups. This is illustrated below. Microfollicles by definition have a high proportion of cells to colloid, and the cells cannot lie within a broad 2-dimensional sheet. In cell block sections, microfollicles will not show long linear arrangements of cells. Renshaw et al., found a significant gray zone among experts in the distinction between microfollicular and macrofollicular groupings on cytology preparations [9]. One of the major gray areas relates to the problem of disaggregated follicular cells artifactually piled on top of each other. This artifactual piling up of cells appears easier to diagnose in tissue sections. Another problem area includes those cases in which there are mixtures of microfollicles and macrofollicles. We and others have noted that benign goiterous nodules tend to show flattened cytoplasm in the cells of the microfollicles. The more robust cells of a follicular neoplasm tend to show mostly cuboidal, or even columnar, follicular cells lining the microfollicles [7]. These features are evident in cell block sections and are illustrated below.

Benign follicular nodules tend to show an admixture of cytologically different types of follicular cells. Follicular neoplasms are more likely to be uniform from follicle to follicle. While the appearance of the follicles in a cytology preparation can be compared from group to group, this approach to judging « clonality » may at times be difficult. For example, it can be impossible to know whether the FNA needle sampled areas outside of a neoplasm. Cell blocks can often allow assessment of the morphologic features of truly adjacent follicles over a span of several hundred microns, as illustrated below.

There are some differences in the appearance of nuclei in direct smears, ThinPreps®, and cell blocks. One advantage of cell blocks is that they provide a common platform that can be shared with the surgical pathologists who will ultimately diagnose a resection specimen. We illustrate the utility of this approach for trying to classify the FNAs of problematic follicular-patterned lesions with partially developed nuclear features of papillary thyroid carcinoma [3].

Papillary thyroid carcinoma is an easy diagnosis if the nuclear features are well-developed and the sample is well-preserved. The best diagnostic features of papillary thyroid carcinoma are finely dispersed chromatin, and nuclear envelope irregularity [2]. Interestingly, papillary thyroid carcinoma is the first example of cancer where the diagnostic nuclear features are known to be directly related to the genes that cause it (reviewed in [10]). A spectrum of nuclear envelope irregularity can be present [11]. The chromatin of a papillary thyroid carcinoma is dispersed into very fine particles, and since much of the chromatin is smoothly adherent to the nuclear envelope, even a tiny fold will cause the nuclear envelope-associated chromatin to appear as a linear aggregate or longitudinal groove. Intranuclear cytoplasmic inclusions are particularly characteristic of papillary thyroid carcinoma, but may also be seen in medullary thyroid carcinoma. This intranuclear cytoplasmic inclusion appears as a sharp demarcation which comes from the nuclear envelope-associated heterochromatin, giving it a distinct dark purple border with the cytoplasm. Intranuclear cytoplasmic inclusions are very uncommon in follicular neoplasms [2].

In contrast to papillary thyroid carcinoma, follicular neoplasms tend to show compact aggregates of heterochromatin. As a rough way of distinguishing the papillary carcinoma nuclei from follicular carcinoma nuclei, we have found it useful to note that papillary thyroid carcinomas tend to show more linear aggregates of chromatin (representing very shallow folds of the nuclear envelope) compared to rounded or clumped aggregates of chromatin. These features are illustrated below.

Finally, cell blocks provide a stable platform for immunohistochemistry. Immunohistochemistry for calcitonin is recommended for confirming medullary thyroid carcinoma [1]. Immunohistochemical distinction of papillary thyroid carcinoma from follicular neoplasms has not been widely successful [12]. In general, it appears that when cells develop papillary thyroid carcinoma-like nuclear changes, the cells will show markers of papillary thyroid carcinoma, including cytokeratin 19 and HBME positive staining [13, 14]. The distinction between papillary thyroid carcinoma and reactive changes in such cases is determined by the larger-scale architectural features that are sometimes evident in cell blocks. The future development of markers that can distinguish the diagnostic categories of follicular lesions on FNA will likely require a stable platform for immunohistochemistry or molecular studies. Paraffin-embedded cell blocks provide such a platform. We illustrate one novel application of CD34 immunostaining that suggested the specific diagnosis of an insular-type of poorly differentiated thyroid carcinoma.

CytologyStuff