Cytology and cell‐block immunohistochemistry of circulating tumour cells

Abstract Objective The study set out to assess the feasibility of using ParsortixTM circulating tumour cell (CTC) extraction and CytoFoam Disc cell‐block immunohistochemistry to diagnose metastatic carcinoma from blood samples in a National Health Service district general hospital. Methods Blood samples were taken from 50 patients with metastatic carcinoma and 50 healthy volunteers and processed, using a previously published method, to extract CTCs and collect them in a cell‐block for routine formalin‐fixed paraffin sectioning and immunohistochemistry. The extracted cells were compared with the patients’ routine diagnostic samples. Results The samples from the 50 carcinoma patients showed cytokeratin‐positive cells in 19 cases. In eight of these, the cytokeratin‐positive cells had a similar immunoprofile to the carcinoma in the conventional biopsy or cytology specimen. Some carcinoma patients also had circulating cytokeratin‐positive cells that were probably benign epithelial cells and circulating megakaryocytes. Both of these types of cells were also found in healthy volunteers. Processing and initial examination could be completed in 2 days. The full processing cost was approximately £316 per case. Conclusions CTCs could be extracted from the blood of some patients with metastatic carcinoma and formed into a formalin‐fixed cell‐block for routine paraffin processing and immunohistochemistry. The specificity of this approach is constrained by the observation that some patients with metastatic carcinoma had circulating cytokeratin‐positive cells that were probably benign, and these were also found in healthy volunteers. Circulating megakaryocytes were present in carcinoma patients and healthy volunteers.


| INTRODUC TI ON
There is currently a lot of interest in circulating tumour cells (CTCs), and their potential in liquid-based tumour diagnosis. In this context, a CTC usually refers to a circulating cell from a solid tumour, such as a carcinoma, melanoma or sarcoma, and haematological malignancies are excluded. CTCs are not a recent discovery. There is a description of tumour cells in post-mortem blood from 1869. 1 They were found in a man who had approximately 30 subcutaneous tumours. It is quite likely that this patient actually had a haemato-lymphoid malignancy with a leukaemic component, as the malignant cells in the blood appear to have been much more numerous than our modern experience of CTCs.
There is a more reliable and more detailed description of CTCs in living patients with carcinomas, melanomas and sarcomas from the 1950s, along with a description of other rare benign circulating cells, including macrophages and megakaryocytes, that can mimic CTCs. Crohn's disease, and after benign breast surgery. [3][4][5][6] It has been thought that NCCCs are almost never detectable in healthy subjects.
Recently, we described a method for preparing a cell-block from a very sparsely cellular extract of CTCs and other rare circulating cells, and examining the cells with routine diagnostic methods including formalin-fixed paraffin sections and immunohistochemistry. 7 The study described here sets out to assess the feasibility of using this method to diagnose metastatic carcinoma in the setting of a National Health Service district general hospital in the UK, mainly using methods that are in routine use in diagnostic cellular pathology laboratories. We focused on the type of rare circulating cells recovered, quality of preservation, concordance with the associated diagnostic sample, turn-around times and cost.

| ME THOD
We recruited 50 patients with metastatic carcinoma together with 50 healthy volunteers, following a protocol approved by an ethics committee. Blood was drawn from each of these subjects. For the first 15 carcinoma patients, this was a volume of 10 mL collected in a single EDTA Vacutainer tube, but for all other carcinoma patients this was increased to 20 mL, collected in two tubes, and also 20 mL for all of the 50 healthy volunteers. The reason for the change from 10 mL to 20 mL was the discovery of probable NCCCs and megakaryocytes in the early samples and it was decided that an additional 10 mL of blood should be taken for Parsortix TM extraction and cytological examination. These samples were processed through a Parsortix™ PR1 device as quickly as possible, either the same day or the next day.
The manufacturer's PX2_ANG_002_SH_90 protocol was used. Two of these devices were available, so that two tubes of blood could be extracted in parallel. The extract from one blood tube was used to form a cell-block, and the extract from the other tube was used for cytology.
The processing through the Parsortix™ device and the formation of a cell-block has been described in detail previously. 7 Briefly, as the blood was processed through the Parsortix™ device, cells larger than approximately 6 μm were retained by the filter cassette. These were then recovered by a phosphate buffered saline back wash of the cassette, yielding 90 μL extract. Next, a 40 μL droplet of plasma was placed at the centre of a clean glass microscope slide. The plasma was derived from the patient's own blood sample. A 12-mm Cytofoam Disc was placed on top of the droplet on the slide and the 90 μL phosphate buffered saline extract containing the recovered cells was added to the centre of the top face of the disc. A further 40 μL of plasma was then added. The surface of the disc was then prodded with a micropipette tip 30 times to encourage mixing of the fluids. A 500-mL plastic histology specimen container was then prepared with a folded paper towel across its floor that was soaked with Following paraffin processing, five cell-block serial sections were placed on one slide, along with suitable positive and negative controls, and immunostained for MNF116 (a broad-spectrum cytokeratin antibody). 7 If this slide showed no positive cells then no further immunohistochemistry was undertaken. If MNF116-positive cells were found, additional immunohistochemistry was performed. If the patient was male, immunohistochemistry for CK7, CK20, TTF1, CDX2, PSMA and PAX8 was performed, and, if female, then for CK7, CK20, TTF1, CDX2, PAX8, oestrogen receptor (ER), WT1 and mammoglobin. In a few healthy volunteer cases, CD31 immunohistochemistry was used to confirm that atypical cytokeratin negative cells were megakaryocytes.
The Parsortix extraction for cytological examination was performed on the second 10 mL EDTA tube of blood, in the same way as for cell-block creation except that the extract was recovered directly onto a glass slide as 3 or 4 small droplets (together 90 μL). The slide was air-dried without being smeared, and then methanol fixed and stained with Speedy-Diff following the manufacturer's instructions.
The slide was then examined for the presence of CTCs, NCCCs and megakaryocytes.
The microscopic appearances and immunoprofile of the extracted circulating cells in the cell-block were compared with the appearances and immunoprofile of the carcinoma in the related conventional biopsy or cytology sample.

| Patients
The patients included 19 males and 31 females with a mean age of 70 years (Table 1). All had metastatic carcinoma. These included 23 cases with metastatic lung carcinoma, five with ovarian carcinoma, four with endometrial carcinoma, four with breast carcinoma, one with colorectal, one with prostatic, one with large bowel carcinoma, one with bladder carcinoma and six with carcinoma of unknown origin.

| Cell blocks
The samples usually arrived in the laboratory between 10 and 20 minutes after being taken. Extraction appeared to be quicker if started immediately after the sample had been taken, and conversely samples that were refrigerated overnight proceeded less quickly.
For the latter, when the extracted sample was examined there were more prominent aggregates of platelets and white blood cells and we believe that these may have interfered with the movement of the blood through the Parsortix TM filter. Samples usually took 3-5 hours to be processed on the device, but there was some variability, the TA B L E 1 Cases with metastatic carcinoma, and circulating cytokeratin-positive cells

| Cytology
As

| Identification of cells in cancer patients
The results are set out in Table 1 (Figures 1 and 2).
For the remaining 11 cases with cytokeratin-positive cells, it was not possible to be certain that these cells were the same as those in the patient's carcinoma. In some cases, this was because there were only scanty cells and this constrained the reliability of the interpretation of the immunostains. However, for some cases the circulating cytokeratin-positive cells appeared to be distinctly different from the primary carcinoma. Their immunoprofile was inconsistent with the carcinoma from the diagnostic specimen, and their morphology was also different. In some cases, the morphology suggested that the cells were probably benign epithelial cells, as they had small regular nuclei. In Case 9, the cell-block showed MNF116-positive cells that were also positive for CK7, CK20 and WT1 (Figure 3). The latter three markers were negative in the patient's clear cell renal carcinoma, and the morphology was different to that of the renal carcinoma. This suggests that the circulating cells were NCCCs, but it is also notable that there is no normal organ that has cells that co-express these markers, further suggesting that there may have been two different types of NCCCs. There also appears to be a mismatch between circulating cells and the carcinoma for cases 4, 30 and 32. For Case 37, the patient had squamous carcinoma, diagnosed without immunohistochemistry, but the circulating cells were positive for CK7 and CK20 which is not typical of squamous carcinoma. It is possible that in some cases a mixture of neoplastic and non-neoplastic cells was obtained. Occasional cells were well preserved ( Figure 4D-F).

| Circulating cytokeratin-positive cells in healthy volunteers
Circulating MNF116-positive cells were found in the blood of 12 of the 50 healthy volunteer samples ( Figure 4A-C). Of these 12, there were six that also showed positivity for CK7 and one that was CK7 negative but showed positivity for CK20. The remaining five were only positive for MNF116. One of these five seemed to show morphological evidence of squamous differentiation ( Figure 4B).

| Circulating megakaryocytes in healthy volunteers
Circulating megakaryocytes were found in 8 of the 50 healthy volunteer samples ( Figure 4D-F). These were detected by morphological examination and therefore micro-megakaryocytes ( Figure 4F) would have tended not to be recognised as megakaryocytes.

| D ISCUSS I ON
The study set out to assess the feasibility of using this method to diagnose metastatic carcinoma. While cost and turnaround times are potentially an improvement on alternatives such as an ultrasound and computed tomography-guided core biopsy, the sensitivity was suboptimal. Many of the patients with metastatic carcinoma did not have detectable CTCs using this method. In addition, the specificity of this approach is constrained by the observation that some patients with metastatic carcinoma had circulating cytokeratin-positive cells that were probably benign NCCCs and these were also found in healthy volunteers. It might be possible to increase the volume of the blood sample and this is likely to increase the yield of circulating epithelial cells, but in order to increase the volume substantially, one would probably have to resort to a recirculation method such as diagnostic leukapheresis. In this method, up to 2.5 L of blood may be sampled 8 and returned to the patient. However, such a specialist technique would increase the complexity and cost of the method described above, and so make it a less compelling alternative to routine biopsy methods. In some patients with clinically suspected metastatic malignancy, there is no obvious target mass for biopsy on imaging and so leukapheresis might be a useful strategy. Another approach would be to screen all carcinoma patients to find those that had CTCs, and for those patients use repeat testing to monitor treatment and progression.
The study has provided some unexpected new insights into rare circulating cells. Firstly, in patients with metastatic carcinoma it appears that some circulating cytokeratin-positive cells were probably benign background circulating epithelial cells, and these were also found in healthy volunteers. Secondly, circulating megakaryocytes were commonly seen in patients with metastatic carcinoma, and these were also seen, albeit less frequently, in healthy volunteers.
As mentioned above, we have known that NCCCs can be found in some diseases, but most studies have found them to be absent in healthy volunteers. One study using CellSearch ® (an immunomagnetic recovery method) found one circulating epithelial cell per 7.5 mL of blood in 5.5% of 145 healthy women, 9 but no sample had more than one cell. Although our method did not attempt to count the Another study of metastatic breast carcinoma patients 12  suggesting that a mesothelial origin is unlikely.
It is well known that megakaryocytes can be seen in the peripheral blood of patients with myeloproliferative disorders. However, it is less well appreciated that they can also be found in the peripheral blood of patients with carcinoma and non-neoplastic disorders, and also healthy subjects. 14,15 They seem to be more common in patients with advanced malignancy than early stage malignancy and benign conditions. Megakaryocytes are commonly seen in pulmonary capillaries. 16 It is possible that other researchers have found megakaryocytes while looking for CTCs but have not recognised them.
Images of megakaryocyte-like cells have been published in at least one previous paper on CTCs. 13 In the current study, we also found that megakaryocytes were more common in patients with carcinoma than in healthy volunteers. Many of the megakaryocytes appeared to have lost their cytoplasm. While this could be due to the physical trauma of being trapped by the Parsortix TM filter, similar features were seen when enzymatic streptolysin O haemolysis was used to separate the megakaryocytes from other blood cells. 14 In summary, this study shows that CTCs can be extracted from the blood of some patients with metastatic carcinoma, and that routine formalin-fixed cell-block immunohistochemistry can be used to demonstrate that these CTCs have features similar to those of the tumour biopsy. Processing and examination can be done in 2 days, if prioritised, in a National Health Service district general hospital laboratory mainly using standard techniques developed for routine biopsy specimens. The processing cost is approximately £316 per case. This study also indicates that benign cytokeratin-positive cells and megakaryocytes are present in the blood of patients with metastatic carcinoma and in healthy volunteers. A follow-on study is planned that will attempt to capture CTCs in patients with large cell lymphoma.
The Parsortix CTC extraction method described above is not approved for clinical use.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.