The role of liquid-based cytology in the investigation of thyroid lesions

Authors


Dr P. Karakitsos, Department of Cytopathology, University of Athens, Medical School (University General Hospital ‘Attikon’), Athens, Greece
Tel.: +30 2105831952; Fax: +30 2105831942;
E-mail: pkaraki@med.uoa.gr, melima52@otenet.gr

Abstract

Objective:  This study investigates the role of liquid-based cytology by ThinPrep® technique in the detection of thyroid lesions.

Methods:  In all, 252 specimens from 157 patients for pre-operative evaluation of thyroid nodules, prepared by the ThinPrep®, were examined. In all cases thyroidectomy followed the initial cytological evaluation. All cytological diagnoses were correlated to the histological ones.

Results:  According to our findings, a sensitivity of 87.80%, a specificity of 99.50%, a positive predictive value of 97.30%, a negative predictive value of 97.56% and an overall accuracy of 97.52% were observed in fine needle aspiration cytology in correlation to the histological diagnosis after thyroidectomy.

Conclusions:  ThinPrep® technique is a valid method for the pre-operative cytological diagnosis of thyroid nodules, offering the possibility of ancillary techniques, such as immunocytochemical and molecular methods and can, therefore, be potentially complementary to histological evaluation for further investigation of follicular lesions.

Introduction

Fine needle aspiration cytology (FNA) of the thyroid gland is currently considered to be the most accurate and cost-effective method of assessing thyroid nodules.1–3 It is a painless, easy to perform method, often used for the pre-operative evaluation of thyroid nodular lesions. The optimal diagnostic strategy aims to avoid thyroidectomy in patients with benign thyroid disease. It is estimated that the diagnosis of solitary thyroid nodules by FNA may spare as many as 75% of patients the need for further studies or surgery.4

The ThinPrep® (Cytyc Co., Boxborough, MA, USA) technique performed on gynaecological specimens has gained popularity during the last decade, because of high quality of smears and the efficacy of the method. A wide spectrum of applications on non-gynaecological cases has been reported and a diagnostic sensitivity and specificity of over 90% has been observed. Most studies published on thyroid cytology with the use of ThinPrep® technique concern a split-sample method or are based on preparation and estimation of the residual material of FNAs.5–9

The aim of this study was to evaluate the role of liquid-based cytology using the ThinPrep® technique, in order to investigate its accuracy in diagnosing thyroid lesions, using direct sampling from pre-operatively aspirated nodules.

Methods

This current study was carried out initially on 157 patients with a total of 252 thyroid nodules, pre-operatively aspirated under ultrasonographic guidance. All patients were admitted to the 4th Department of Surgery of the University General Hospital ‘Attikon’, during a 2-year period 2004–2005 and were scheduled for thyroidectomy.

A 23 gauge needle was used for the aspiration, and the material collected was transferred to a vial containing fixative (Cytolyt®; Cytyc, Co., Boxborough, MA, USA). The sample was then collected after centrifugation of the vial at 600 g for 10 minutes. In cases of bloody samples, additional Cytolyt® solution washes were necessary, until the sample became clear. The supernatant was then discarded and the material was transferred to a vial with cytopreservative solution (PreservCyt®; Cytyc, Co.). PreservCyt® solution mildly fixes the cells within 30 minutes and then the material is ready to be prepared by the ThinPrep® 2000 Automated Slide Technique.10 From each case one slide was prepared and stained by the Papanicolaou method, while the remaining material was used for additional slides for further diagnostic techniques, if necessary.

The thyroid FNA samplings were evaluated by cytopathologists during the routine diagnostic procedure. All positive cases, as well as those concerning follicular neoplasms, were further discussed and reviewed by all the medical staff of our Laboratory. For practical reasons, as there is no established consensus as to the exact style of reporting the FNA results, the cytological reports were split into 10 categories of diagnostic value, according to Table 1. Histological sections from thyroidectomy specimens were interpreted by histopathologists according to standard diagnostic criteria.11

Table 1.   Cytological categories and main cytological criteria
Cytological entity/categoryCytological criteria
 1. Benign lesion, non-neoplastic lesion
  goitre, follicular hyperplasia,   papillary hyperplasia, cystic   degeneration, haemorrhage
Polymorphous pattern
Cells in (monolayered) sheets or double-layered clusters
Microfollicular arrangement rare
Papillary- like formation rare
Regular nuclei
Mild to moderate anisonucleosis
Reactive and inflammatory cells
Stromal cells sometimes present
Colloid always present
 2. Hashimoto’s thyroiditis
  nodule and/or dominant nodule   consistent with Hashimoto’s   thyroiditis
Pattern depending on the site, stage, and severity of the disease
Follicular and reactive cells
Moderate to marked anisonucleosis
Sheets of Hürthle (oxyphilic) cells
Lymphocytes in several stages (mature and immature)
Colloid: few droplets or absent
 3. Hashimoto’s thyroiditis and    hyperplastic goitre (coexistent) Sheets of Hürthle cells
Lymphocytes in several stages (mature and immature)
Colloid present or not, depending on the site of the FNA
 4. Suspicious in favour of    hyperplastic benign lesion, Double-layered microfollicular arrangement of epithelial cells, without atypia
Pseudopapillary arrangement without peripheral palisading, suggesting papillary hyperplasia
Colloid present or not
 5. Suspicious for neoplastic lesion    (in favour of malignancy) Not specific atypical pattern
Double-layered microfollicular arrangement of epithelial cells, without atypia
Pseudopapillary arrangement without typical pattern of a papillary carcinoma
Anisonucleosis, necrosis without other indications
Absence of colloid
 6. Follicular neoplasmMonomorphous pattern of microfollicular clusters
Double-layered or multilayered microfollicular arrangement of epithelial cells, overlapping
Nuclear polymorphism, atypia, coarse chromatin and prominent nucleoli
A spherule may or may not retain the colloid content
Colloid usually present in the microfollicles or out of them in oval-shaped dense droplet
 7. Hürthle cell neoplasmSheets, clusters and single Hürthle(oxyphilic) cells
Monomorphous cellular population
Occasionally distinct or prominent nucleoli
Intranuclear cytoplasmic inclusion rare
Colloid sometimes present
 8. Papillary carcinoma (classic or    follicular variant of papillary    carcinoma) Crowded clusters of cells
Papillary formation of groups, with endothelial cells, true fibrovascular cores
Microfollicles present, in the follicular variant of papillary carcinoma
The external contour of groups tends to be smooth, with peripheral palisading of nuclei, in tall cell variant of papillary carcinoma
Longitudinal nuclear grooves and ground glass appearance
Intranuclear inclusions rare
Multinucleated giant cells commonly encountered
Psammoma bodies rare
Colloid sparse inspissated and focally located, when present
 9. Medullary carcinomaHeterogeneous cellular population
Spindle or plasmacytoid cells
Amyloid
10. InadequateLess than six clusters of ten or more well preserved epithelial cells

For the specimen adequacy, at least six clusters of ten or more cells were required.6,12 All cytological diagnoses were correlated to the histological ones. The histological result of thyroidectomy sampling is considered as the final diagnosis. For the purpose of this study, results of cytological and histological diagnosis were compared, in order to evaluate the role of ThinPrep® liquid-based cytology technique.

Results

Correlation between the cytological and the final histological diagnosis (thyroidectomy) is presented in Table 2 and summarized in Table 3.

Table 2.   Correlation between ThinPrep® cytological and thyroidectomy histological diagnosis
HistologyCytology
Cyt cat. 1Cyt cat. 2Cyt cat. 3Cyt cat. 4Cyt cat. 5Cyt cat. 6Cyt cat. 7Cyt cat. 8Cyt cat. 9Cyt cat. 10Total
  1. *Other than lymphocytic/Hashimoto’s thyroiditis and adenomatous non-neoplastic nodules. Grave’s disease (two cases), iodine effect changes (three cases), phagocytic granuloma (one case), non-specific thyroiditis (two cases), necrotic and asbestotic pattern (one case) were included in this group.

  2. Adenoma with atypia.

  3. In a Hashimoto’s thyroiditis background.

Goitre and other benign lesions*121315000007137
Lymphocytic/Hashimoto’s thyroiditis4276000000037
Lymphocytic/Hashimoto’s thyroiditis and nodular goitre206301000012
Adenomatous nodules (not neoplasms)40040000008
Follicular adenomas20011000004
Follicular Ca00000400004
Hürthle cell adenomas00010010002
Hürthle cell Ca00001000001
Papillary microCa6 3 3230000320
Papillary Ca0000200230025
Medullary Ca00000000101
Parathyroid adenoma00010000001
Total13933161775123110252
Table 3.   Summarized correlation between ThinPrep® cytological and thyroidectomy histological diagnosis
HistologyCytology
Benign lesionsSuspicious, in favour benign lesionsSuspicious for neoplasmNeoplasmsInadequateTotal
Benign lesions, thyroiditis included17412017194
Benign neoplasms231107
Malignant neoplasms00328031
Papillary microCa12230320
Total1881773010252

According to our findings, ten unsatisfactory specimens were obtained (3.96%). Among them three papillary microcarcinomas were diagnosed by histological examination. For statistical reasons categories 5 (suspicious for neoplastic lesion, in favour of malignancy), and 6,7,8,9 (neoplasms) were considered as a single category. Furthermore, category 3 (Hashimoto’s thyroiditis and hyperplastic goitre coexistence) was included, because sometimes these conditions may coexist (the so-called nodular type of Hashimoto’s thyroiditis).13

Out of 242 adequate specimens:

  • 1In all, 187 were histologically diagnosed as benign non-neoplastic lesions. One of them was reported as a neoplasm on ThinPrep® smear (false positive case). Out of 12 cases reported as suspicious in favour of benign lesion (category 4) five proved to be goitres and seven were hyperplastic nodules (three of them in a lymphocytic thyroiditis background).
  • 2Two out of four follicular adenomas were missed (false negative cases) and one case of dominant nodule in a Hashimoto’s thyroiditis was reported as follicular neoplasm (false positive case).
  • 3All four follicular carcinomas were correctly diagnosed (two insular carcinomas, one follicular carcinoma and one poorly differentiated carcinoma).
  • 4Out of three cases of Hürthle cell tumours (two adenomas and one carcinoma) one adenoma was reported as suspicious in favour of benign lesion and the Hürthle cell carcinoma as suspicious for malignancy.
  • 5Out of 17 histologically diagnosed papillary microcarcinomas, only three were detected by FNA.

Twelve of the above cases, reported as benign lesions, were incidentally detected by histology, representing true negative cases. Three cases concerning aspirated nodules of 6–8 mm diameter were reported as suspicious of malignancy and verified by histology, representing true positive diagnoses. Two cases reported as suspicious in favour of benign lesion, concerning aspirated nodules of 16 and 21 mm and representing coexisting hyperplastic nodules, were counted as true negatives, when comparing cytological to histological diagnosis.

  • 6All papillary carcinomas and one medullary carcinoma were correctly diagnosed.

Accepting that what is cytologically reported as suspicious, even in favour of benign lesion, alerts the clinician and that all adenomas have to be histologically confirmed to rule out vascular and/or capsular invasion, then according to our findings the ThinPrep® technique had a sensitivity of 87.80%, a specificity of 99.50%, a positive predictive value of 97.30% and a negative predictive value of 97.56%. The overall accuracy was 97.52%.

Moreover, out of 49 histologically diagnosed cases of lymphocytic and Hashimoto’s thyroiditis, 39 were correctly reported on ThinPrep® cytology, giving sensitivity, specificity, positive predictive value, negative predictive values and overall accuracy of 79.59%, 100%, 100% and 95.07% respectively.

Discussion

For several years, most cytopathologists preferred direct conventional smears, rather than smears prepared from material rinsed in vials. During the last decade however, it has been shown that as sufficient diagnostic criteria have been established for thin layer procedures, the latter has proved to be an effective innovation for many laboratories.14,15

In order to evaluate the potential role of direct to vial ThinPrep® cytology of thyroid FNAs, all patients included in this study had undergone a planned surgical operation, according to the surgical standards.16

Since 1995 several investigators have postulated that the diagnostic accuracy with ThinPrep® in thyroid FNA is almost equivalent to that of the conventional smear preparation,9,15 with the sensitivity usually mentioned of approximately 80% and a specificity ranging from 60% to 100%.5,6,8 Follicular lesions are considered to be the bane of pathologists and the main source of false reports by cytopathologists.12,17 Tulecke and Wang suggested that ThinPrep® technique can be used to distinguish high risk (microfollicular lesions) from low risk (macrofollicular or mixed lesions).18 Evaluating the architectural arrangement in our material and according both to the previous study and our experience on conventional smears, most benign follicular lesions, 125 out of 130, were true negative. Nonetheless, two cases of macrofollicular adenomas were missed in ThinPrep® smears because they revealed monolayered sheets of thyrocytes with moderate anisonucleosis and abundant colloid. The arrangement of the epithelial component and the high ratio of colloid/cellularity was the reason for false negative diagnosis. On the other hand, one false positive diagnosis of follicular neoplasm, which proved to be the dominant nodule in Hashimoto’s thyroiditis was due to the microfollicular arrangement of thyrocytes, and the absence of lymphocytes and colloid. Additionally, one adenoma was reported as suspicious in favour of benign lesion and one with atypia as suspicious of malignancy. The latter revealed marked anisonucleosis, prompting us to alert the clinician. All four cases of follicular carcinoma revealed atypical thyrocytes with no evidence of colloid on ThinPrep® smear (Figure 1), and with cytomorphological features similar to the ones of conventional cytology.

Figure 1.

 Follicular carcinoma (Category 6). Multilayered microfollicular arrangement, anisokaryosis, overlapping, dense droplet of colloid (×40).

One Hürthle cell adenoma was reported as suspicious in favour of benign lesion because the presence of oxyphilic cells lacking nucleoli did not exclude Hashimoto’s thyroiditis with no lymphocytes sampled (Figure 2). Although actually there are no large series of such neoplasms with documented criteria on the ThinPrep® smears, we observed that the presence of nucleoli was decisive for establishing an accurate diagnosis (Figure 3). In any case, as stated in previous studies diagnosis of follicular and Hürthle cell neoplasms is difficult by cytology alone either on conventional or liquid-based cytology.5,6,9,12

Figure 2.

 Suspicious in favour of benign lesion (Category 4). Double-layered groups of epithelial cells with moderate anisokaryosis and absence of colloid (×10).

Figure 3.

 Hürthle cell tumour (Category 7). A group of oxyphilic cells with distinct and prominent nucleoli (×40).

In all cases of papillary carcinoma cytomorphological criteria12,19 were present in different combinations. It has been demonstrated that the diagnostic features remain essentially the same for papillary carcinoma whether the specimen is prepared with a ThinPrep® technique or with the conventional direct smear method.8,20 True papillary tissue fragments with a fibrovascular stromal core were almost always observed. However, in some cases the stromal component was not tightly connected to the epithelial cells as in conventional smears (Figures 4 and 5).

Figure 4.

 Papillary carcinoma (Category 8). Crowded clusters of cells, papillary formation of groups (×10).

Figure 5.

 Papillary carcinoma (Category 8). Longitudinal nuclear grooves and an intranuclear inclusion (×40).

The rate of papillary microcarcinomas in thyroidectomy specimens may rise to 30%.4 The term is defined as a tumour smaller than 1 cm.11 As papillary microcarcinoma is an incidental finding during histological examination of a thyroidectomy specimen (and therefore cannot be previously diagnosed by FNA), the pre-operative diagnosis of this neoplasm should not be expected. In the current study, three cases, whereby a papillary microcarcinoma was diagnosed by cytology, no other malignant lesion (apart from this papillary microcarcinoma) was found on histological examination.

Only one case of medullary carcinoma was included in this study. Cytomorphological features were similar to the ones on direct smears.21 Atypical cells, either isolated or in small clusters, with eccentric nuclei, were immunopositive for calcitonin (Figures 6 and 7).

Figure 6.

 Medullary carcinoma (Category 9). A group of cells with eccentric nuclei and coarse chromatin (×40).

Figure 7.

 Medullary carcinoma (Category 9). Calcitonin positive spindle and plasmacytoid cells (×10).

Most authors note that lymphocytic and Hashimoto’s thyroiditis may lead to overestimation of direct smears because of cellular atypia and absence of colloid.20–23 On the other hand, the presence of lymphocytes may be missed or considered to derive from the peripheral blood. In a retrospective study, Scurry and Duggan pointed out that in suspicious for malignancy cases because of marked cellular pleiomorphism, lymphocytes were yet present. Cochand-Priollet et al. stated that the presence of reactive cells with few lymphocytes in the background makes the distinction of follicular adenoma from Hashimoto’s thyroiditis very difficult. Studies based on conventional cytology show that adequate sampling of the thyroid is important.23 In our study, most cases of lymphocytic/Hashimoto’s thyroiditis revealed the presence of lymphocytes in a clear background, requiring meticulous screening under high power magnification. As the cytological pattern in Hashimoto’s thyroiditis depends on the stage of the disease and the heterogeneity of the material is often obvious, the advantage is that by the ThinPrep® technique different cell populations are represented on the slide prepared.

In some studies colloid was not detectable on ThinPrep® slides. However, in most of our benign cases colloid was observed, although fragmented, (Figure 8), while in neoplasms, when present, it appeared as droplet-like material, located within the follicular groups (Figure 1).

Figure 8.

 Benign lesion (Category 1). Monolayered sheet of epithelial cells and colloid (×40).

One parathyroid adenoma, an incidental case, was included in our series. It is generally admitted that it represents one of the most difficult tumours to diagnose, also on conventional smears.12

Finally, although ten inadequate specimens were included and simply mentioned in this report, they have not been counted in the statistics because they do not meet the necessary requirements. They are important, as we encounter them in everyday clinical practice; moreover, they alert clinicians to the fact that additional cytological material is necessary to establish a safe diagnosis. The lower rate (3.96%) of inadequate samples in comparison to others previously published5,8 is likely to be due to the use of the direct-to-vial technique.

In conclusion, direct-to-vial preparation of smears using the ThinPrep® technique is a valid method for the pre-operative cytological diagnosis of thyroid nodules. The method is a safe and user-friendly office technique, easily learnt and thereafter performed by short-term training of personnel in daily practice, because, as has previously been shown, the cytomorphological diagnostic criteria present minor differences, compared to the ones used in conventional cytology specimens.

As only one slide is sufficient for diagnosis, the ThinPrep® technique reduces the workload of the cytopathologists and offers the possibility of an accurate application of immunocytochemical and molecular techniques, in slides with the same cellularity as those in which the morphological diagnosis is made.

Ancillary