Assisting the neurologist in diagnosis of CNS malignancies ‐ Current Possibilities and Limits of Cerebrospinal Fluid Cytology and Immunocytochemistry

Abstract Objectives In tumorous impairment of CNS, cytological identification of the neoplastic cells in CSF frequently requires the use of ancillary techniques. Our methods are focused on identifying algorithms that increase the probability of identifying CSF malignant cells. Materials and Methods A total of 1.272 CSF samples from patients with tumorous infiltration of CNS of nonhematologic origin along with 721 samples from patients with hematologic malignancies were analyzed in a complex setting including cytological and immunocytochemical investigations. Results and Discussion In CSF diagnostics we are aware of the limited amount of sample combined frequently with neoplastic oligocytosis. Provided atypical, potentially malignant cells in CSF are found, further investigation(s) should maximize the probability of their identification—an appropriate cytological staining and immunocytochemical panel is to be applied. (i) In cases of known recent malignancy: immunoprofile of the recent neoplasm has been considered in immunocytochemical panel. (ii) In patients with a history of malignancy: The propensity to develop a new different malignancy must be taken into account. (iii) Atypical cells found in the CSF of a patient with a negative history of malignancy: Considering the most frequent clinically silent malignancies, stepwise immunocytochemistry is employed. Three milliliter of initial CSF sample represents the absolute minimum to start with. Conclusions The steps of the laboratory activity targeted on malignancy in the CSF detection can be expected as follows: (i) The sample will be divided for both nonmorphology and cytopathology investigations. (ii) Basic stainings will triage the samples into those with no suspicion of malignancy and the remaining ones. (iii) Special stainings and stepwise immunocytochemistry will be performed in parallel with the nonmorphology investigations.

Tumorous impairment of CNS is the highest priority for the application of CSF cytology (Deisenhammer et al., 2006;Sobek et al., 2012). It is a minimally invasive diagnostic method, yet capable enough to ensure crucial information concerning not only the mere presence of tumorous cells in CSF but also their closer identification which is fundamental for consecutive therapy and for the patient's prognosis. Nevertheless, identification of the neoplastic cells is only partly achievable with cytomorphology and histochemical stainings (Glantz et al., 1998;Kaplan et al., 1990;Twijnstra et al., 1987;Wasserstrom et al., 1982). Ancillary techniques are now a part of guidelines and even routine investigations (Chamberlain et al., 2009;Chandra et al., 2009;Coakham et al., 1984;Weston et al., 2011).
The complex CSF investigation is oriented from the very beginning towards a precise diagnosis, enabling the clinician to start the most appropriate treatment for the patient. The subsequent text describing our laboratory algorithms in the identification of malignancies focuses on daily life diagnostic questions in different settings.

| Processing of cytological samples
In maximizing the diagnostic yield of the cells present, the CSF sample must be processed within 30 min (optimally) to a maximum of no more than 2 hr (some authors concede up to 3 hr (Sobek et al., 2012)) and maintained in-between at 4°C. (To ensure this our laboratory has nonstop operating hours plus transport service.) The common processing of the CSF sample for cytopathology investigation starts with cytocentrifuge (Cyto-Tek ® Sakura, up to 140×g for 5 min) or cytosedimentation slides stained according to May-Grünwald-Giemsa (MGG) and Hematoxylin-Eosin (H&E).
The classical techniques of cytopathology laboratories (alcian blue, mucikarmine, PAS, oil red) are cheap and can be very helpful if purposefully applied. Many of these slides, especially the unmounted MGG, are subsequently usable for immunocytochemistry provided the cells in question are found to be present.
Based on the conventional cytological investigation, if there are cellular elements morphologically suspicious of malignant character found, then it is followed up with immunocytochemical (ICC) identification. The number of microscopic slides that are available for ICC investigations are unfortunately often limited by total sample volume, but optimally 4-5 further slides for ICC should be prepared.

| ICC Methodology
• As a fixative, pure methanol is used for 5 min p.a. (according to the bottle min. 99,8%). After that the excess methanol is poured off and slides are air-dried again. Till the phase of actual ICC, the microscopic slides are refrigerated at 2-8°C.
• Before proceeding with ICC, the microscopic slides are rehydrated for 5 min in distilled water.
• The next step is heat-induced epitope retrieval in pH 6,0 or pH 9,0 buffer-depending on the subsequently used antibody-at 95-97°C in water bath for 20 min.
• The slides are then cooled down together with the detection buffer in cold water for 5 min and after that the cooling process continues with just the separate slides in a wash buffer for another 5 min (EnVision FLEX Wash Buffer manufacturer DAKO-thereinafter just the buffer).
• Inhibition of endogenous peroxidase in 3% hydrogen peroxide solution for 20 min.
• Thorough rinse for 3 times 5 min with buffer, application of a primary specific antibody. Concentrated antibodies are diluted according to the ratio recommended by the manufacturer, using dilution solution Primary Antibody Dilutent (manufacturer Diagnostic BioSystems).
Incubation with primary antibody takes place in a moist chamber at room temperature for 30 min.
• Thorough rinse with buffer is followed with a 5 min bath in buffer in a cuvette.
• Then an application of a detection enzyme EnVision FLEX/HRP follows (manufacturer DAKO), and incubation in a moist chamber at room temperature for 30 min.
• Then the slides are rinsed with buffer and washed in distilled water for 5 min.
• The next step is an application of chromogen EnVision FLEX Substrate Working Solution (manufacturer DAKO). Incubation for 5 min, followed by a rinse with distilled water.
• Staining of the nuclei of cells with Hematoxylin (manufacturer DiaPath), for 1-2 min is followed with a rinse in "spring water" for 5 min.

| Methodology notes
There is an exception for slides with antibodies which do not require heat-induced epitope retrieval. These slides start with rehydratation first and continue straight with inhibition of endogenous peroxidase.
Usually these antibodies are applied for longer period of time (overnight) at 2-8°C. Further processing steps are the same.
There is also an exception in fixation process for slides with antibodies which do not tolerate alcohol fixation (e.g., S100, GCDFP-15, etc.)-in those cases formalin is used. Fixed slides are refrigerated, or may be kept at room temperature provided that ICC is done immediately. In the framework of this study there were no other medical examinations in patients or volunteers carried out, only anonymous clinical data were employed.
From each patient whose medical examination results are anonymously used in this manuscript, the authors have an informed consent for research use at their disposal.

| What are the clinical situations requiring identification of neoplastic cells in the CSF?
Generally, there are three different initial arrangements: 1) Recently treated known malignancy. CNS (incl. meningeal) impairment supposed.
2) History of malignancy treated/cured. Signs of CNS impairment present.

symptoms.
To achieve the solution and precise diagnosis, the limited amount of sample-usually no more than 8 ml-is divided into parts for chemical, microbiology, immunology, and cytopathology analyses. Many classical texts and articles dealing with the CSF diagnostics (Chandra et al., 2009;De May, 1996;Weston et al., 2011) stress the initial need of 3-5 ml CSF minimally to perform a valid investigation along with the possible requirements for an additional sample. Considering the fact that the declared minimal invasiveness of the spinal tap is a relative entity, and the sample received in the laboratory represents an extremely valuable source of information, all measures contributing to the optimum utilization should be employed to achieve the available diagnostic maximum.

| How do we approach the three clinical situations mentioned above?
In all three settings, the more introductory information provided, the higher the probability of a precise interpretation.
• Patients' data: Age and gender are always available, race and country of origin (or travelers' history) may be of importance.
o Micromorphology-in case of positive malignancy history, providing both this history and the previous histopathology results including the immunoprofile of the malignancy reported can greatly enhance the efficacy of the confirmatory investigations.
There are of course limits to be considered when designing the investigation steps and in interpreting the results: • Even when an intracranial malignancy is present, the neoplastic cells are not necessarily present in the CSF.
• The past or present malignancy and its treatment and/or concomitant diagnoses can be accompanied with nonmalignant atypical cells.

| Evaluating the CSF finding
When evaluating the CSF finding, different results can occur. In the case of no suspicious cells present, a negative report is issued and the neurologist has to decide whether the provided diagnosis Provided atypical cells are found, further investigation(s) should maximize the probability of their identification using the best strategy available. While evaluating the morphology and possibly also the immunocytochemical results of CSF slides with atypical cells, many factors are to be considered, mainly: • reactive and regressive changes in cells.
These cytomorphology features overlap largely with those used for malignant cell identification. We start with the broadest general examination possible using the proper fixation while protecting and saving the sample from the very first step (see Materials & Methods section).
Cells in body fluids tend to degrade and lose their immunoreactivity. This is especially true for the hypotonic environment of cerebrospinal fluid. Therefore, emphasis is placed on fast processing and corresponding fixation. The most common protocols use 100% methanol or 4% buffered paraformaldehyde. Such fixed preparations generally retain immunoreactivity for several days when refrigerated. Longer storage (weeks or months) may weaken immunoreactivity (Fowler & Lachar, 2008).
Standards for immunocytochemical procedures develop more slowly than for immunohistochemical procedures. This is undoubtedly due to the obstacles in ensuring the necessary controls. Air-dried preparations are widely used in cytology of aspirations of solid masses.
The smears are created at the sampling point, air-dried, and thus delivered to the laboratory. They are suitable especially for May-Grünwald-Giemsa staining. Polychrome staining methods (Papanicolaou) require wet fixation. Rehydration of air-dried smears has been described in many articles. A cross-sectional study of air-dried smears versus wet fixation was published by Rupinder et al. (2013).
Cytological specimens that are primarily liquid in nature, cavity fluids including CSFs, are usually processed in the laboratory. Air-dried preparations preserve immunoreactivity only for some, usually cytoplasmic antigens. The quality of membrane antigen manifestation can be impaired as demonstrated by Pinheiro et al. (2015).
Being conscious of the preciousness of the CSF sample, we use a proven methanol fixation protocol and refrigerate the reserve preparations. Recently, however, we also test the protocol recommended by Pinheiro et al. (2015). Coating of the preparation with polyethylene glycol after immediate methanol fixation allows storage of the preparations at room temperature. Rehydration can improve the immunoreactivity of air-dried smears (Shidham et al., 2000) but air-drying fixation represents an accepted version for immunocytochemistry (Fulciniti et al., 2008;Knoepp et al., 2013).
A short air-drying step lasting 5 min prior to the 100% methanol fixation was in our protocol since it decreased cell loss in the subsequent procedures. All the illustration cases are processed this way.
Nevertheless, our experience is similar to many other investigators in the necessity to adjust the protocols to the antibody in question and to the cytology material available (Sauter et al., 2016).
After the possibly malignant cells have been found, an appropriate immunocytochemical (mini) panel is to be designed (see Table 1).
In cytological examinations of other body fluids, the cost effectiveness of the diagnostic procedure is usually considered and stepwise diagnostic algorithms are employed. In CSF diagnostics we are aware of the limited amount of sample combined frequently with neoplastic oligocytosis. Irrespective of the exponential increase in antibodies available for routine diagnostics, neither individual antibodies, nor panels provide 100% sensitivity and specificity of detection.
Nevertheless, prudent choice of markers can help. In neoplastic oligocytosis, it can be a helpful strategy to recycle the same scarce cells negative in certain markers to test them for others. It is advisable to ensure the microphotography documentation of the cells exploredany further laboratory test can cause their loss due to detachment from the slide.

| Ad (1) Recently treated malignancy
With a known diagnosis we always try to get the previous biopsy and

| Ad (2) History of malignancy
Recently, two features have become obvious with advanced therapy of malignancies ( Figure 2). First, recurrences occur after longer and longer periods in many tumors, other than those known in the past to behave this way, for example, melanoma. Breast or kidney carcinomas recur after a period previously considered improbable for such a turn in the disease. Second, (subsequent) primaries become more and more common due to the immunocompromising effects of the efficient oncology therapy. The practical impact is that the long known and diagnosed malignancy that was considered cured can become the source of recent disease. This should be exploited diagnostically and ruled out as described before. To trace back a previous biopsy is not always an easy task. Malignancies treated successfully years ago may escape inclusion into recent clinical data. A second primary in a known past malignancy represents an important differential diagnosis.

| Ad (3) Suspicious malignant cells found in the CSF sample of a patient with entirely negative history of tumorous disease
The broadest differential diagnostic judgments together with the limited sample represent the most challenging reality in these situations ( Figure 3). Three milliliter of initial CSF sample represents the absolute minimum to start with, considering the statistical data of the most frequent clinically silent malignancies (lung, breast, kidney, This question is addressed in previous studies (Fowler & Lachar, 2008). In fact, reporting of rare or even isolated suspicious cells is anchored in standardized reporting recommendations such as the Bethesda system for reporting cervical cytology or the Bethesda system for reporting thyroid FNAB (Ali & Cibas, 2010;Nayar & Wilbur, 2015).
The diagnostic success is more frequently achieved in solid metastatic malignancies (representing 60% of all malignancies in the CSF). In hematology (30% of all malignancies in the CSF) flow cytometry has been confirmed as a method of choice superior to immunocytopathology by many studies (Chamberlain et al., 2009;Chandra et al., 2009;Kaplan et al., 1990;Weston et al., 2011).
Nevertheless, it faces frequently the same problem of sample volume. Primary malignancies in the brain represent only 10%; as they are usually deeply located they may frequently exhibit negative CSF.
In investigating the CSF sample with the neoplastic oligocytosis, recycling of the preserved cells can be successfully used. Both formulation of such results and the clinical interpretation and application must be done with strict awareness of the quantitative limit and thus possibly limited validity. It has been described thoroughly and also beautifully illustrated by Perske et al. that no single morphological parameter is sufficient to detect neoplastic lymphocytes. Taking into account a combination of cell size and irregular shape of cell and nucleus, however, may improve the diagnostic accuracy (Perske, Nagel, Nagel, & Strik, 2011). Meningeosis neoplastica occurs in 5%-10% of the most frequent malignancies, lung and breast cancer, melanoma, and diffuse large B-cell lymphoma. The diagnosis opens the way to the intrathecal therapy (Strik & Prömmel, 2010).

| CONCLUSION
Close cooperation between the diagnostic laboratory and clinicians will result in a quick and accurate diagnosis, enabling appropriate treatment.
Necessary conditions on the laboratory side are: • continuous availability (nonstop service) • equipment with modern technologies.
• staffing with experienced specialists.
On the part of clinical specialists it is necessary to ensure in particular: • timely delivery of the sample in a nondegraded quality and quantity required for diagnosis (in tumor diagnostics the initial minimum for cytopathology only requires 3-5 ml, additional sample requirements are not excluded) When these introductory conditions are fulfilled, the steps in the laboratory activity targeted on malignancy in the CSF detection can be expected as follows: • The sample will be divided for both nonmorphology and cytopathology investigations • Basic stainings will triage the samples into those with no suspicion of malignancy and the remaining ones • Special stainings and stepwise immunocytochemistry will be performed in parallel with the nonmorphology investigations • The final report will be signed out in the shortest time possibleideally on the same day as the completion of the cytological, resp. immunocytochemical investigation.
Cooperation and close contact among the team members throughout the entire process contributes greatly both to the desired results as well as to the greatest benefit for the patient.

ACKNOWLEDGMENTS
Authors would like to express special thanks to Jitka Doležalová for editorial support and to Ondřej Duchoň for Patient database operations.