Value of ancillary studies in the evaluation of fine-needle aspiration specimens: Our experience.

BACKGROUND: The cytological diagnosis of poorly differentiated tumors is challenging because the tumor cells may have morphologically difficult presentations in materials obtained by fine-needle aspiration cytology (FNAC). With the application of FNAC in primary diagnosis of malignant lesions, there has been a significant increase in the use of ancillary studies in the aspirated material. AIMS: We evaluated the value of ancillary studies, namely cell blocks, immunocytochemistry (ICC) and electron microscopy (EM), in the final interpretation of FNAC smears.
MATERIALS AND METHODS: Sixty-nine cases of neoplastic swellings were subjected to FNAC. Material acquired was divided for ICC, consisting of immunoperoxidase staining of direct smears, and/or cellblocks and EM, in addition to routine light microscopy (LM). Correlation with the available histological material with immunohistochemistry and/or pertinent clinical information was used as a "gold" standard.
RESULTS: Five (7.2%) cases were excluded from the study, the material being necrotic or insufficient. Cell blocks were available in 46/64 (71.8%) cases, ICC evaluation was performed in 41/64 cases (64%) and EM studies were done in 57/64 cases (89%). Diagnostic accuracy of LM alone was 32/64 (50%). Cell blocks improved the diagnoses in 8/46 (17%) cases. The ICC data were diagnostic in 18/41 (43.9%) cases, helpful in 8/41 (19.6%) cases and non-helpful in 15/41 (36.5%) cases. EM studies were diagnostic in 22/57 (38.5%) cases, helpful in 18/57 (31.5%) cases and non-helpful in 17/57 (30%) cases. In 34/64 (53.1%) cases, all ancillary techniques (cell blocks, ICC and EM) were applied and their diagnostic accuracy was compared.
CONCLUSIONS: With appropriate case selection, ancillary studies performed on aspirated material can provide useful information in FNAC.

Introduction

Fine-needle aspiration cytology (FNAC) is well established as a safe and accurate diagnostic procedure. Specific typing of tumors can be greatly beneficial for the effective management of patients; a precise diagnosis is not always possible on routine cytomorphological assessment alone.

Various studies have shown a variety of ancillary techniques, such as the use of cell blocks, application of immunocytochemistry (ICC) and the electron microscopic (EM) evaluation, to be of considerable benefit in increasing the diagnostic accuracy.[1-7]

In the present study, we have assessed the usefulness of each of these ancillary techniques on cytological specimens obtained by fine-needle aspiration in the pre-therapeutic diagnosis of various neoplastic lesions.

Materials and Methods

Sixty-nine patients presenting with poorly differentiated malignant neoplasms diagnosed on FNAC at the Pathology Department in a referral hospital were included during a 2-year period. Multiple passes were made in order to increase cellular yield. Computed tomographic/sonographic guidance was resorted to for non-palpable deeper lesions. Immediately following FNAC, a portion of the sample was screened and a preliminary light microscopic (LM) interpretation was given. Additional passes were then made for ancillary techniques, namely ICC, consisting of immunoperoxidase staining of direct smears and/or cellblocks and EM. For LM, air-dried smears were stained by May-Grünwald–Giemsa (MGG), Papanicolaou (Pap) and hematoxylin and eosin (H andE) stains.

Cell blocks

For preparation of cell blocks, the aspirated material was flushed into 10% formalin, pelleted by centrifugation, and the tissue fragment was processed. Five micron sections were cut for H and E, histochemical stains and immunohistochemistry (IHC) studies.

Immunocytochemistry

Immunocytochemistry (ICC) was performed on direct smears and cellblock sections. Air-dried direct smears were fixed in cold acetone, wrapped in aluminium foil and preserved at -20°C for ICC by streptavidin–biotin complex technique. Commercially available antibodies (DAKO) to S100 protein, cytokeratin (CK), carcinoembryonic antigen (CEA), vimentin, desmin, muscle-specific actin (MSA), human melanin black (HMB) 45, chromogranin, CD34 and neuron-specific enolase (NSE) were used in 1:10 dilutions. Appropriate positive and negative controls were used throughout the procedure. The following information was recorded: specimen site, preliminary diagnosis, final diagnosis; types of stains performed; results of the special stains performed either on smear or cell blocks.

Electron microscopy

The sample for EM was flushed into 3% glutaraldehyde in an Eppendorf tube, which was then pelleted by centrifugation. The pellet was then processed through dehydration and embedding in Epon. One-micrometre semi-thin sections were cut and stained with toluidine blue for orientation and study of morphology. Appropriate clusters of cells were selected and the blocks were trimmed accordingly. Ultrathin sections were stained with uranyl acetate and lead citrate and then studied by EM.

Evaluation of results

To evaluate the value of the ancillary techniques, the results were characterized into three groups according to the relative value of the ancillary technique in arriving at a pathological diagnosis, as done by Saleh et al.[1] Correlation of the available histological material with IHC characterization and/or pertinent clinical/radiological information was used as “gold standard.” This assessment compared the initial diagnoses made on smears based on LM features with those made on ICC evaluation and ultrastructural characteristics independently. Ultimately, these diagnoses were correlated with the gold standard. The impact of ancillary studies was assessed by categorizing their results into three groups. The test results were considered “diagnostic (D)” when they were crucial for the final diagnosis; they were considered “helpful (H)” when they confirmed the initial light microscopic diagnosis; while the inconclusive, contradictory or non-contributory results were regarded as “non-useful (N).”

In order to evaluate the tumor types, cases were divided into five separate groups based on predominant cytomorphological features on LM: Carcinomas with difficult diagnosis on LM, spindle cell tumors, pleomorphic sarcomas, round cell tumors and a miscellaneous category [Table 1]. The last group, the miscellaneous category, comprised tumors not fitting into any of the first four groups. This classification was not based on the final histopathological examination (HPE) diagnosis.

Table 1

Tumor groups based on cytomorphological criteria and ancillary studies performed

Data accrued were tabulated to analyze the utility of the ancillary techniques used. Diagnostic accuracy of LM alone, LM and cell blocks, LM, cell blocks and EM, and all ancillary techniques, i.e. LM, cell blocks, ICC and EM together, was calculated in the cases studied by all methods [Table 2].

Table 2

Results of ancillary techniques - Cases evaluated by cell blocks, ICC and EM

Value of ICC and EM in tumor diagnosis was calculated as the proportion of correctly identified tumors by this technique in comparison with the gold standard. For this calculation, the results of diagnostic and helpful groups were combined.

Results

Sixty-four cases were evaluated after exclusion of five cases wherein the material aspirated was either necrotic or insufficient. There were 29 females and 35 males with their ages ranging from 2 to 75 years, having an arithmetic mean of 43.31 years. Forty-five patients underwent surgery and, in four cases, relevant material was obtained after autopsy for HPE, rendering HPE possible in a total of 49 cases. Thirty-six (56%) cases were referred for a primary diagnosis, while 24 (38%) and 4 (6%) cases were evaluated for metastatic and recurrent lesions, respectively. The aspirates from soft tissue swellings (39%) constituted the major bulk, followed by lymph nodes (20%).

Light microscopy

Based on the salient LM features observed on routine FNAC smears, the tumors were classified [Table 1]. On LM, 32/64 (50%) cases were diagnosed accurately (without any ancillary aid) merely on clinical correlation. Accurate diagnosis was possible in cases of melanotic melanomas, neuroendocrine and neural tumors, while in cases of pleomorphic and spindle cell tumors, only a broad range of differential diagnosis was possible. Differentiation of the small cell carcinoma of lung metastatic to lymph node from lymphoma was tricky. Distinction of hepatocelluar carcinoma (HCC) from metastatic malignancies also posed difficulty.

Cell blocks were available in 46 cases. Results of cell block were tabulated as done by Sedigheh et al.[7] Identical smear and cell block results were seen in all cases, and cell blocks improved the diagnostic accuracy in 8/46 (17%) cases (without ICC or EM).

Cell block facilitated accurate diagnosis in a case each of angiosarcoma (breast) and rhabdomyosarcoma (RMS; manifested as small round cell tumor) by enabling reticulin staining [Figure 1a] and better identification of strap cells [Figure 2d], respectively. It also facilitated ICC in these cases. In the case of angiosarcoma, the smears were bloody with only a few spindled cells in contrast to the presence of tissue fragments in the cell blocks. The aspirates from the liver and neural tumors also showed better morphology on cell blocks.

Figure 1

(a) Reticulin staining in cell block preparation of angiosarcoma, outlining pattern of tumor cells (Reticulin, ×400); (b) angiosarcoma: Weibel Palade bodies seen within the cells (arrows) (EM, ×67,000); (c) FNAC thyroid, spindle cell tumor (MGG, ×400); (d) medullary carcinoma thyroid: Membrane-bound neurosecretory granules (arrows) (EM, ×28,000)

Figure 2

(a) Ultrasound scan abdomen of a 2-year-old female child showing large retroperitoneal mass; (b) round to plump spindled cells (arrow) with hyperchromatic round nuclei (MGG, ×400´); (c) rhabdomyoblasts demonstrating strong positivity for myoglobin (ICC, ×100); (d) cell block preparation showing strap-shaped cells with abundant acidophilic cytoplasm (arrow) (H and E, ×400´)

Immunocytochemical evaluation

ICC was done in 41/64 (64%) cases. Of these, immunostains were performed on smears in 5 cases and on cell blocks in the remaining 36 cases. Results were diagnostic in 18/41 (43.9%) cases, helpful in 8/41 (19.6%) cases and non-helpful in 15/41 (36.5%) cases. ICC facilitated accurate diagnosis in 63.4% cases [Table 3].

Table 3

Value of electron microscopy and immunocytochemistry

Among 16 cases of carcinomas with difficult diagnosis on LM, immunostaining helped to reach a conclusive diagnosis in 5 cases and provided additional information in 4 cases. The major role of ICC in this subgroup was to help differentiate an epithelial malignancy from a lymphoma [Figure 3].

Figure 3

(a) FNAC cervical lymph node, poorly differentiated carcinoma (MGG, ×400); (b) positive immunostaining for leucocyte common antigen in tumor cells (ICC, ×400´)

In five of nine cases of spindle cell tumors, a definitive diagnosis was reached at with immunostaining. Desmin and MSA positivity improved the diagnosis in myogenic tumors. Endothelial markers helped to confirm the diagnosis in a case of angiosarcoma with spindle cell morphology.

In the study of pleomorphic sarcomas, S-100 positivity was a valuable indicator of Schwannian and melanocytic differentiation. One case of carcinosarcoma, presenting as a pleomorphic sarcoma, was diagnosed on positive immunostaining for CK and vimentin. Despite round cell tumors forming a small subgroup, immunostaining was diagnostic in four cases. Based on a comprehensive diagnosis of embryonal RMS [Figure 2] in a young child, reached at with the help of EM and ICC, clinicians were able to initiate chemotherapy preoperatively.

EM correlation was available in 57/64 (89%) cases. Results were diagnostic in 22 (38.5%) cases, helpful in 18 (31.5%) cases and non-helpful in 17 (29.8%) cases. EM facilitated accurate diagnosis in 70.2% cases [Table 3].

EM was helpful in the subclassification of carcinomas where 9 of 19 cases were correctly identified. Ultrastructurally, diagnostic features could be detected in liver, gastrointestinal, ovarian, lung and breast carcinomas. This technique helped us to differentiate metastatic small cell undifferentiated carcinoma from lymphoma.

EM confirmed the LM diagnosis in 9 of 15 cases of spindle cell tumors. These included two cases of angiosarcoma, where demonstration of Weibel Palade bodies [Figure 1a, b] helped to clinch the diagnosis. In two cases of spindle cell tumors of the thyroid, neurosecretory granules ranging in size from 80 to 150 nm were diagnostic of medullary thyroid carcinoma [Figure 1c, d]. In five cases, neural differentiation was proven by ultrastructural findings of complex arrangement of long cell processes with scattered desmosome like structures and prominent focally reduplicated basal lamina [Figure 4]. EM was inconclusive in two spindle cell tumors, a case of fibromatosis and synovial sarcoma.

Figure 4

(a) Computed tomographic scan of pelvis showing soft tissue tumor involving right gluteal region; (b) spindle cell sarcoma with necrotic background (Pap, ×100); (c) positive cytoplasmic immunostaining for S100 (ICC, ×400); (d) malignant peripheral nerve sheath tumor: Complex arrangement of long cell processes with prominent focally reduplicated basal lamina (arrow) (EM, ×40,000)

The pleomorphic sarcomas comprising 13 cases in this study showed a spectrum of ultrastructural differentiation. In eight of these, the histogenetic cell of origin was correctly identified by EM. These included a case of leiomyosarcoma where prominent myofibrils with associated focal densities were seen. Two neuroendocrine tumors were correctly identified by the presence of membrane-bound neurosecretory granules [Figure 5]. Demonstration of premelanosomes helped to confirm the diagnosis of metastatic amelanotic melanoma by EM in two cases [Figure 6a, b]. Malignant fibrous histiocytoma (MFH) was confirmed by the presence of histiocyte like cell with irregular clumped nuclear chromatin and phagolysosomes [Figure 6c, d].

Figure 5

(a) FNAC pancreatic mass, small round cell tumor (MGG, ×400); (b) membrane-bound neurosecretory granules (EM, ×67,000)

Figure 6

(a) Pleomorphic sarcoma, (MGG, ´400), inset shows positive cytoplasmic immunostaining for HMB45 in cells (ICC, ×400); (b) malignant melanoma, same case as in (a), premelanosomes seen as ovoid bodies with centrally running bands, finer transverse striations (arrow) (EM, ×67,000); (c) pleomorphic sarcoma (MGG, ×400), inset showing positive cytoplasmic immunostaining for vimentin in cells (ICC, ×400); (d) malignant fibrous histiocytoma, same case as in (c), histiocyte like cell with irregular clumped nuclear chromatin, few RERs and phagolysosomes (arrow) (EM, ×36,000)

Comparison of ancillary studies, ICC and EM

A comparison of diagnostic accuracy was carried out in 34/64 (53.1%) cases evaluated by all methods [Table 2]. With the help of routine cytology, a definitive diagnosis could be achieved in 16/34 (47%) cases. Cell blocks, without the application of any special histochemical or IHC stains, improved the LM appearance and confirmed diagnoses in two cases of HCC and one RMS. Diagnostic accuracy of LM and ICC together was 23/34 (67.6%). Diagnostic accuracy of LM and EM was 26/34 (76.4%). Diagnostic accuracy of LM, ICC and EM together was 32/34 (94.1%).

EM was inconclusive in 8 cases in this study, while ICC was not helpful in 11 cases. Both the techniques were complementary to each other and enabled the most accurate and precise diagnosis as brought out by 94% results by both methods in this study.

Discussion

Preoperative diagnosis of neoplasms by FNAC has presented a diagnostic challenge and a type-specific diagnosis can only be achieved with the help of ancillary studies.[12] Routine cytomorphology helped to accurately diagnose 50% of cases; also, further investigations in the form ancillary studies, i.e. the panel of antibodies for ICC and specific features to be looked for in EM, were guided by the LM diagnosis.

The main application of cell blocks was in performing ICC and histochemical stains, where required. Cell blocks had the advantage of providing more tissue in hemorrhagic aspirates and also a simultaneous assessment of the histological architecture in cases of RMS, neural tumors and liver aspirates. This is in concordance with the results reported by other authors.[8-11] Cell blocks in ICC offer a number of advantages. Kung et al.[12] showed that FNAC cell blocks gave better results than standard histological sections from the same case. In our study, however, we found that not much difference existed between the cell block and histological sections. In some cases, histological sections gave better IHC staining results. We attribute this to insufficient tissue availability on the respective blocks in these cases. We found cell blocks to be most suitable for ICC due to ease of morphological interpretation, standardized comparison with surgical pathology material, minimal background staining and expected immunostaining patterns. We are in agreement with other authors[1213] on the advantages of ICC staining on cell blocks versus direct smears. The main restraining factors that we faced with ICC done on smears were in the storage of smears and the limited number of smears available for testing. Domagala et al.[13] have cited other disadvantages, namely high and not necessarily specific staining that can result from disrupted cells and membrane fragments sticking to the slides, lack of parallel samples of same cells for additional or control tests and increased cost due to use of considerable volume of antibodies to cover large areas of less cellular smears. The use of cell block ICC is a reliable and technically unsophisticated aid in the cytological examination of tumors. Success depends on having obtained sufficient cellular material on aspiration.

ICC allowed diagnostic results to be given in 43.9% cases and additional information was provided in 19.6% cases, thus helping in definitive or accurate diagnosis to be made in 63.5% of cases, similar to results of other authors.[38111214-21] In our study, ICC was limited to a panel of antibodies to determine the cell line of differentiation, namely carcinoma, lymphoma, melanoma, sarcoma, restricting the use of cell-specific antigens to a limited number of cases only. Inconclusive results were obtained because of negative immunostaining and lack of specific markers, which may classify tumors, especially for mesenchymal tumors and those with a divergent differentiation. Common causes of false-negative results included poor antigen preservation, insufficient antigen retrieval, denaturation or prolonged fixation or low antibody concentrations. Conversely, common causes of false-positive results included low specificity of antibody, necrotic cells or dried preparation or inappropriate fixation. Mitteldorf et al.[14] found ICC to be non-contributory in their study, mainly due to misleading interpretation, uncharacteristic profile and inconclusive immunostaining.

Ultrastructural studies enabled us to determine the characteristics of various tumors with great precision. By performing separate passes by needle aspiration for EM, we obtained sufficient material and observed performances similar to those of other authors.[4821-24] Table 4 summarizes the usefulness of various ultrastructural studies on cytology samples as reported by other authors.[146822-24] While most of the studies were not restricted to any particular site like in the present, the studies by Neil et al.[6] and Brahmi et al.[8] were restricted to FNAC of mediastinal masses and malignant round cell tumors, respectively. The main difficulty of EM as applied to FNAC is ensuring that the material observed on thin sections corresponds to the neoplastic cells examined with LM.[25] This difficulty was essentially overcome by comparing the ultrastructural features with semi-thin sections and LM features. The other difficulty faced was the stringent processing schedules and the expertise required in the interpretation of the material. Results were inconclusive in our study in 17 (30%) cases because of either the presence of necrotic material or poorly preserved morphological appearances on EM. This was probably due to stringent processing schedules wherein delay of transfer of material from glutaraldehyde to buffer may have distorted morphology of cells, rendering them difficult to interpret ultrastructurally. Besides, the interpretation of EM often requires more judgement and expertise.[16]

Table 4

Comparative study of the usefulness of EM in FNAC

Case selection for EM study is the focal point of analysis for routine work. The cases which benefited the most were small, round, undifferentiated cancer cells, spindle cell tumors, identification of the primary site of poorly differentiated metastatic deposits and cases wherein the presence of more than one cell type was seen on LM. Similar results have been reported in literature.[101125-29]

In recent years, ICC has taken over as the first technique for further evaluation of poorly differentiated tumors. This is because ICC is easier to interpret, sampling is less of a problem, and in routine practice, most pathologists do not have access to an electron microscope and have little experience in interpreting ultrastructural features.[2630] However, there remains a definitive role for EM as a limited panel of broad-spectrum CKs cannot distinguish between metastatic neoplasms of unknown primaries. Brahmi et al.,[8] in their study of the contributions of EM and ICC as adjuncts in cytodiagnosis, concluded that while ICC is the ancillary method of choice in the cytological diagnosis, EM contributes to the diagnosis and improves diagnostic accuracy. Zbieranowsky and Bedard[15] have discussed the value of EM and ICC in preoperative diagnosis of schwannomas and they found EM to be more useful in predicting the cell of origin. Dar et al.[16] compared the relative usefulness of EM and ICC in tumor diagnosis in 10 years of retrospective analysis and found that EM should be retained for selected cases in which the results of ICC might be predictably ambiguous or otherwise unhelpful. Similar results were reported in a study of pediatric round cell tumors by Akhtar et al.[25]

Conclusions

This study analyzed the usefulness of ancillary studies in cytodiagnosis. Cell block preparations had the advantage of histological interpretation and in the performance of immunostains. With the ongoing discovery of newer and more type-specific markers, ICC is invaluable for the reliable identification and classification of tumor cells in cytology. Our results in which EM was decisive in 38.5% of cases confirm the important role of EM in the diagnosis of FNAC specimens. For inclusion of aspirates in EM to be productive, the requirements are stringent selection of cases, impeccable processing and expertise in evaluation.

Concurrent use of EM and ICC helps to avoid the shortcomings associated with each of these techniques and enables the most accurate and precise diagnosis as brought out by 100% results in the cases evaluated by both methods in this study.