Effectiveness of the cell block technique in diagnostic cytopathology.

BACKGROUND: One of the constraints of the conventional FNA smear is the limited material available for adjuvant diagnostic investigations including immunocytochemistry. The cell block technique employs the retrieval of small tissue fragments from a FNA specimen which are processed to form a paraffin block. It is widely accepted that cell block technique increases the cellular yield and improves diagnostic accuracy. The ability to obtain numerous tissue sections allows for multiple immunostains and other studies to be performed akin to paraffin sections produced in histopathology. AIMS: To determine the effectiveness of the cell block technique by comparing cytomorphological preservation and immunocytochemistry (ICC) stains on paired cell block and conventional fine needle aspiration (FNA) samples.
MATERIALS AND METHODS: In this prospective study, material for both glass slides and cell blocks were collected simultaneously during fine needle aspirates from 47 samples comprising lung and liver masses. Grading of cellularity, morphological preservation, architectural preservation, immunocytochemical staining intensity and presence of background staining on paired FNA smears and cell block samples were compared. Each arm of the paired analysis was performed blindly without knowledge of the grading outcome of the other. The Kappa statistic (κ) was used to measure inter-rater agreement.
RESULTS: The 47 samples evaluated included FNAs from the lung, 24/47 (51%) and liver, 23/47 (49%). The immunocytochemistry stains consisted of 44/47 (94%) CK7; 44/47 (94%) CK20; 18/47 (38%) TTF1; 10/47 (21%) synaptophysin; 10/47 (21%) Hepar-1 and 7/47 (15%) AE1/3. There was no overall agreement in preservation of cytomorphological detail and ICC staining between the two methods. The Papanicolaou-stained conventional FNA smears fared better than the cell block for the evaluation of nuclear and morphologic characteristics. The ICC stains worked better on the cell block samples due to lack of background and aberrant staining.
CONCLUSION: Direct FNA smears and cell blocks complement each other and our results indicate that both are needed in the diagnostic work-up of patients. The cost implications of performing both techniques on all FNA material warrants further evaluation.

Introduction

One of the constraints of the conventional fine needle aspiration (FNA) smear is the limited material available for adjuvant diagnostic investigations. The use of the cell block technique enables the retrieval of small tissue fragments in a fluid specimen which are processed to form a paraffin block. It has been widely accepted that this method of analysis increases the cellular yield and improves diagnostic accuracy.[12] The ability to obtain numerous sections allows for multiple immunostains and other studies to be performed akin to paraffin sections produced in histopathology.[3] The aim of this study was to establish, optimize and implement the cell block technique into the routine diagnostic setting for purposes of immunocytochemistry.

Materials and Methods

This prospective study was performed with approval from the institutional ethics committee.

Forty-seven FNAs were performed by trained personnel from February 2009 to December 2009. Material from lung and liver masses were aspirated for conventional smears and cell block preparation simultaneously. Four slides were made from two needle passes. Two were fixed immediately in alcohol for Papanicolaou (Pap) staining, and two were air-dried for Diff-Quik staining. Samples collected for the cell block were either from an additional dedicated needle aspiration and/or needle rinse.

Cell block preparation

Cell blocks were prepared in accordance with the Shandon Cytoblock Kit (Thermo Shandon Limited, Cheshire, UK). The technique comprises the following: All cell block samples were fixed in Formal-Fixx® (Thermo Shandon Limited) for at least 12 hours (maximum of 48 hours) before processing. Needle rinses were performed in all cases to obtain the highest possible cellular yield. Samples were centrifuged for 1 minute at 2.5 rpm and supernatant discarded. Reagent 2 was added to the cell pellet. The Cytoblock cassette, cytoclip and cytofunnel were assembled. Three drops of Reagent 1 was added into centre of well of Cytoblock cassette and spun (cytospin) at 1500 rpm for 5 minutes on low acceleration. The Cytoblock was removed and 1 drop of Reagent 1 was placed in the centre of the well over the cell button. The Cytoblock cassette was closed and processed in the standard tissue processor as per the processing schedule shown in Table 1.

Table 1

Cell block processing schedule

A comparison between the grading of cellularity, morphological and architectural preservation, immunocytochemical staining intensity and presence of background staining were performed on paired FNA smears and cell block samples according to the grading system as shown in Table 2. Each arm of the paired analysis was performed blindly by experienced cytologists without knowledge of the grading outcome of the other.[3] The FNA smears were assessed separately and at different times to that of their paired cell block samples.

Table 2

The grading system

The cellularity, morphology and architecture of each cell block sample was evaluated using the hematoxylin and eosin (H and E) stain and that of the FNA smear was evaluated using the Pap stain and/or Diff-Quik stain especially where the Pap stain was not available, since the best Pap slide containing the most representative material was usually used for split slide conventional immunocytochemistry (ICC). This technique involves destaining Pap-stained FNA smears followed by splitting the slide and restaining for the respective ICC tests.

All ICC assays included the use of appropriately optimized primary antibodies in the presence of positive and negative controls. The latter comprised an additional slide of the test case in which the primary antibody step was omitted and Dako Antibody Diluent (S2022) was applied to the section instead. The Dako Real Envision Detection System (K5007) was used in all assays [Table 3].

Table 3

Primary antibodies: Source and specifications

Statistical evaluation

The data were analyzed using Stata Statistical Software 2007: Release 10, College Station, Tx: StataCorp. The agreement between the two methods of sample preparation were assessed with respect to the quality of definition of cytomorphological characteristics and ICC. The Kappa statistic (κ) was used to measure inter-rater agreement. Kappa values > 0.75 represented excellent agreement beyond chance; values < 0.4 represented poor agreement beyond chance and values between 0.4 and 0.75 represented moderate agreement.[6] P-values ≤ 0.05 were considered to be statistically significant.

Sample size

Using the nQuery Advisor V6 software for calculating the sample size it was determined that, a sample of 41 pairs will have 90% power to detect a difference in proportion of 0.3 when the proportion of discordant pairs is expected to be 0.5. The method of analysis is a McNemar's test of equality of paired proportions with a 0.05 one-sided significance level and the Kappa statistic. To be conservative, as the problem at hand will be tested one-sided, a sample size of between 45 and 50 pairs will be used.

Results

Forty-seven FNA samples were evaluated: 51% (24/47) were FNAs from the lung and 49% (23/47) were from the liver. The immunocytochemsitry stains consisted of 44/47 (94%) CK7; 44/47 (94%) CK20; 18/47 (38%) TTF1; 10/47 (21%) synaptophysin; 10/47 (21%) Hepar-1 and 7/47 (15%) AE1/3.

Cellularity

In the assessment of agreement of cellularity between the two methods of sample preparation, the following were obtained: κ – statistic = -0.0022; P=0.0006.

Morphological preservation

In FNA samples morphology was preserved in all (47/47) samples compared to 94% (44/47) in cell block samples (κ – statistic = -0.02; P=0.00).

Architectural preservation

All FNA samples (47/47) displayed architectural preservation compared to only 47% (22/47) cell block samples.

Immunostaining of cells

In the assessment of agreement of immunocytochemical staining between the two methods of sample preparation the following were obtained: κ- statistic 0.2 and P=0.14 for CK20, TTF-1, Synaptophysin, Hep1 and AE1/3 immunostains.

CK7

CK7 was performed on 44 samples. Of these, 34% (15/44) were negative in the cell block sample and 13.6% (6/44) in the FNA samples (κ-statistic 0.2, P=0.02) [Figures 1 and 2].

Figure 1

FNA CK7 immunostain: FNA Left lung mass showing adenocarcinoma. This figure demonstrates a diffusely positive CK7 immunostain (score 6) with severe background staining (score 3) (ICC, ×100)

Figure 2

Cell Block CK7 immunostain: FNA right lung mass showing adenocarcinoma. This figure emonstrates a diffusely positive CK7 immunostain (score 6) with lack of nonspecific background staining score 0) which is in striking contrast to the FNA sample (IHC, ×400)

Background (BG)/Nonspecific staining in ICC tests

The following Kappa statistic and P-values were obtained for the immunostains: CK 7 (K 0.03; P-value 0.0001), CK20 (K 0.01; P-value 0.00), TTF1 (K 0.00; P-value 0.03) and synaptophysin (K 0.15; P-value 0.03).

Negative controls (Background/Aberrant staining)

Nonspecific aberrant staining was observed in the FNA-negative controls for CK7, CK20, TTF1, synaptophysin and Hepar-1 immunostains. This phenomenon was not displayed in the corresponding cell block negative controls and in the paired AE 1/3 immunostains.

Discussion

A variety of cell block techniques have been in use for over a century.[7] The use of cell blocks have been widely advocated in the diagnostic work-up of patients with masses amenable to FNA since they provide diagnostic architectural information which complement FNA smears.[7]

In this study, the Pap-stained FNA smear was the better method for routine diagnoses due to superior preservation of nuclear and cytoplasmic characteristics[8] whilst the cell block technique was more suited to immunocytochemical analyses. Our findings would suggest that the cell block samples are best used as an adjunct for ICC and not for primary cytological diagnoses. The degeneration of cells in the cell block samples may be attributed to a delay in immersing the cell block specimen into fixative immediately after collection and variation in FNA technique amongst personnel. This variation in technique may have contributed to either the success or failure of obtaining adequate cell block samples which is largely dependant on the skill of the aspirator and high cellularity of the aspirate.[910]

Due to the pre-fixation lag time of many of the cell block samples, sub-optimal preservation of cellularity (23/47), morphology (41/47) and architecture (28/47) were observed early on in the study. Consequently, there was a poor agreement and statistically significant difference between methods in the assessment of cellularity (κ – statistic = -0.0022; P 0.0006), morphological preservation (κ – statistic = -0.02; P 0.00) and architectural preservation (κ – statistic = 0.00; P 0.00). There were no FNA samples (0/47) that scored zero for cellularity but 8.5% (4/47) of cell block samples were acellular. Of these samples, 4% (2/47) were obtained by rinsing the needle and the hub of the syringe and 4% were obtained by performing one dedicated aspiration for the cell block sample. Low cellularity (score 1) was obtained for 40% (19/47) of cell block samples whereas that of the FNA samples was only 11% (5/47). Of further note is that more (96%) FNA samples (45/47) displayed a higher score (score 2 and 3) for cellularity than cell block samples, 57% (27/47). All FNA samples (47/47) displayed architectural preservation compared to only 47% (22/47) cell block samples. In the majority of the cell block samples (53%) architectural preservation was absent. Nevertheless, it was decided to continue with the use of Formal-Fixx fixative since the remainder of the samples (24/47, 6/47 and 19/47 respectively) displayed optimal preservation illustrated by higher grading scores. Hanley et al.[11] have stated that the preservation of antigenecity of tumor cells is essential for accurate immunocytochemical analyses and the use of an ideal fixative and optimal tissue processing parameters is crucial in this regard. Since optimal preservation was observed in the remainder of the cell block samples, the fixative and the tissue processing schedule used were deemed suitable. The sub-optimal preservation observed could have been due to the pre-fixation time lag.

In our setting, FNA's are predominantly performed by radiology and pathology registrars whose experience and skill varies. Consequently, material for cell block was usually aspirated after 3 to 4 aspirations for the conventional FNA smear which was given priority. This may have contributed to a more traumatized and poorly preserved specimen. In many cases (20/47), a dedicated pass for the cell block samples was not performed. The syringe containing residual FNA sample was merely rinsed into cell block fixative solution. In these cases, 65% (13/20) had suboptimal cellularity (score 0 and 1). In 30/47 cell block samples, a dedicated needle aspiration was placed directly into the vial containing Shandon's Formal-Fixx fixative and 60% (18/30) showed adequate cellularity (score 2 and 3). This demonstrates that a dedicated needle aspiration for cell block improves yield.

Inference drawn from Bardi and Schwartz[12] indicates that the mindset of both patients and aspirators is of equal importance. Some patients did not consent to an extra needle aspiration for the cell block sample while many aspirators were reluctant to perform additional needle aspirations despite patients’ informed consent. This was due either to the patient not being able to withstand the procedure, risk of causing a pneumothorax in patients undergoing lung FNAs, lack of amenability of the mass to FNA, time constraints, inexperience or merely not being willing to do so. Although not statistically evident, samples collected for the cell block technique may have been disadvantaged by not receiving a dedicated aspiration.[9] All of the above could have contributed to the poor cellularity, morphological and architectural preservation of cell blocks[9] and a poor agreement between the two methods of sample preparation.

There was a poor agreement in the CK7 immunostain which was the only test that displayed a statistically significant difference (P 0.02) between methods. CK7 was performed on 44 samples. Of these, 34% (15/44) were negative in the cell block sample and 13.6% (6/44) in the FNA samples. In the assessment of background (BG) / nonspecific staining in immunostains, a statistically significant difference in this discordance was obtained for CK 7 (K 0.03; P-value 0.0001), CK20 (K 0.01; P-value 0.00), TTF1 (K 0.00; P-value 0.03) and synaptophysin immunostains (K 0.15; P-value 0.03). In all cases, more cell block samples displayed no background staining (score 0) than FNA samples. Non specific aberrant staining was not observed in the corresponding cell block negative controls and in the paired AE 1/3 immunostains. A possible explanation for this phenomenon is that not all antigens are susceptible to anoxic degradation or diffusion just as not all antigens are equally affected by fixation.[13] The FNA samples most affected by background and anomalous staining were some liver FNA samples, samples containing proteinaceous debris and predominantly necrotic or very thickly smeared samples[9] indicating that the problem was intrinsic.[14] Kung et al.[15] made a similar observation with regard to stronger staining intensity, lack of nonspecific staining and lack of aberrant staining in cell block samples, especially with cytokeratin stains.

Discrepant results were obtained for one sample- a liver FNA. Immunocytochemistry performed on this smear demonstrated tumor cell positivity for CK7 and synaptophysin whilst CK20 was negative. This cytokeratin profile is observed in 56% of neuroendocrine carcinomas of the lung.[16] The same panel of tests performed on the corresponding cell block sample were negative. Although all 3 tests were repeated the results remained unchanged. A possible explanation could be sub-optimal preservation of antigenecity of tumor cells in this sample which was somehow rendered more susceptible than others.

Future research in this department would benefit from exploring the use of 10% neutral buffered formalin (NBF) as the fixative of choice in preparing cell block samples for immunohistochemistry (IHC) with a decrease in the time lapse between sample collection and fixation. An ad-hoc committee on Immunohistochemistry Standardization, affiliated with the College of American Pathologists have discouraged the use of non-formalin-based fixatives and or alternative fixation methodologies.[17] This is due to the performance data of the IHC assay using other fixatives being limited and extrapolation from existing data being unreliable.[17] Axe et al.[18] produced cell blocks from FNA of liver which were fixed in 10% NBF with optimal preservation of cellularity and successful IHC thus enabling subclassification of the tumor.

With the use of the Shandon Cytoblock Cell Block Preparation System it was possible to capture small groups of cells. An improved technique of cell block preparation and cell capture has been devised by Varsegi and Shidham[2] which may be beneficial incorporating into the current method. Using the Varsegi technique increases the chance of capturing individually scattered cells with the use of Histogel (Thermo Shandon), thus preventing the histotechnologist from cutting too deeply into the block and risk missing the area with the cells of interest. Although not statistically evident, samples collected for the cell block technique may have been disadvantaged by not receiving an initial dedicated aspiration which may have compromised cellularity. In this regard, obtaining material from multiple dedicated needle aspirations for cell block should be explored to improve yield.

The role of cell block preparation in diagnostic cytopathology is without doubt of immense significance as it allows for multiple special investigations and consequently a more refined cytological diagnosis. The methodology could be improved on by modifying techniques and reducing the limitations posed in this study, viz. exploring the use of 10% neutral buffered formalin (NBF) as the fixative of choice in preparing cell block samples for immunohistochemistry (IHC), with a decrease in the time lapse between sample collection and fixation and standardization of FNA technique among personnel. Direct FNA smears and cell blocks complement each other[7] and our results indicate that both are needed in the diagnostic work-up of patients; the former to assess morphology, and the latter for optimal immunocytochemistry results. In resource constrained settings, the cost implications of performing both conventional and blocked smears on all FNA material warrants further evaluation.