Ethics Statement

The study was established according to the ethical guidelines of the Helsinki Declaration and was reviewed and approved by the Ethics Committee of Norman Bethune First Hospital, Jin Lin University, Changchun, China. Each of the patients provided written informed consent before entering the clinical study. The informed consent form is administered in Chinese language. In the case of patients who could not read standard Chinese, a verbal consent was documented using the following procedure: the investigator or research nurse provided oral explanation of the purpose and the entire procedure of the study to the patient and provided answers to any questions that the patient had raised; the patient provided verbal consent which was recorded in the patient’s chart, including the contents of the discussion, the consent, the time and date; and an independent investigator or research nurse witnessed the verbal informed consent process by signing the records. This procedure was reviewed and approved by the same aforementioned Ethics Committee. All the informed consent forms were kept in the patients’ hospital charts.

Clinical Characteristics of LM Patients in Retrospective Study

In this report, we will describe two studies: a retrospective study and a prospective study. In the retrospective study, there were a total of 88 consecutive LM patients with cancers of non-neurogenic and non-hematologic origins admitted to our hospital from August 2009 to October 2013. All patients underwent CSF cytological analysis by the Thinprep method (1–6 tests per patient) in our hospital. The patients were 31–72- years old (average age, 54 years), including 44 females and 44 males. Of the 88 patients, 71 had a history of confirmed malignancy, including lung adenocarcinoma (31), small-cell lung cancer (19), breast cancer (12), lung squamous cell carcinoma (3), gastric adenocarcinoma (3), breast and lung cancer (1), large-cell lung cancer (1), and hepatocellular carcinoma (1); the remaining 17 were initially diagnosed with LM and primary tumors were further identified in 15 of them, which included lung adenocarcinoma (10), gastric adenocarcinoma (2), small-cell lung cancer (2), and melanoma (1). Primary tumors of the other two patients were not identified and listed as unknown primary tumors. The diagnoses of the aforementioned cancers were confirmed by histopathological examination of primary tumors. The CSF samples were collected by lumbar puncture with a spinal needle being inserted between the lumbar vertebrae L2/3, L3/L4, or L4/L5.

Clinical Data Collection

Clinical Diagnosis of LM

The diagnosis of LM was made according to the National Comprehensive Cancer Network (NCCN) guidelines and American Society of Clinical Oncology (ASCO) guidelines. [11] The diagnosis for LM was made if any one of the following criteria was met: 1) tumor cells were detected in CSF cytological examination; 2) patient had a history of malignancy and the enhanced MRI scans showed diagnostic imaging features of LM, including focal or diffuse leptomeningeal enhancement, ependymal enhancement, and/or metastatic nodules on the spinal cord or cauda equine; 3) patient had a history of malignancy, exhibited LM-related neurological symptoms and the enhanced MRI scan showed imaging features suggestive of LM, including dura mater enhancement, metastasis in subarachnoid nodules, ventricular or parenchymal enhancing nodules, sulcal, folia, or cranial nerve enhancement, and/or communicating hydrocephalus; in addition, the CSF biochemical tests showed abnormal results such as elevated protein or decreased glucose levels, and patient had no history of traumatic brain injury or meningitis caused by pathogenic microorganisms; and 4) Patient had a history of malignancy and exhibited persistent or severe and progressive neurological symptoms typically associated with LM and the MRI scan was negative or inconsistent with the severity of manifestations; and patient had no other diseases that may cause such symptoms and LM-directed treatment effectively reduced the existing symptoms.

Eighty of the 88 patients were diagnosed with LM based on detection of malignant cells in the CSF by the Thinprep method. Of the eight patients with negative CSF results, five patients with small-cell lung cancer exhibited diagnostic imaging features of metastatic spinal tumor and two patients with lung squamous cell carcinoma. One patient with lung adenocarcinoma had symptoms such as severe persistent headache, vomiting and dizziness, elevated protein level in CSF, and imaging features characteristic of metastasis in the cerebral sulci or gyri. Patients had no other diseases that may cause such neurological symptoms. LM-directed treatment effectively reduced the existing symptoms.

Comparative Prospective Study

The Thinprep results from the retrospective study

All the 88 patients underwent several of CSF cytological examinations (1–6 times) by the Thinprep method in our hospital. Cancer cells were detected in 80 out of the 88 CSF samples. As shown in Table 1, of the 80 patients with positive CSF results, primary tumors were identified in 78 patients, including 58 adenocarcinomas,16 small-cell lung carcinomas, 1 large-cell lung carcinomas, 1 squamous cell carcinoma, 1 malignant melanoma, and 1 hepatocellular carcinoma. Primary tumors of the other two patients with positive CSF results were not identified. The Thinprep method preserved cytomorphological features with clear background, allowing differentiation of malignant cells of different origins based on their distinct morphological characteristics (Table 1, Fig. 1).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Table 1. Morphological features of different types of tumor cells in CSF specimens (n = 78, samples from all 80 LM patients had positive CSF results except two patients with unknown primary tumors).

https://doi.org/10.1371/journal.pone.0122016.t001

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 1. Different pathological types of tumor cells detected in CSF specimens.

a-n, adenocarcinoma cells; o-q, cancerous squamous cells; r-s, small-cell lung cancer cells; t-u, hepatocellular carcinoma cells; v-x, large-cell lung cancer cells; y-z, malignant melanoma cells. (Thinprep, Pap staining, ×400)

https://doi.org/10.1371/journal.pone.0122016.g001

As shown in Fig. 1a-z, cells examined by the Thinprep method appeared intact, with good preservation of three-dimensional (3D) structures, while cellsprocessed by the Cytospin-WG method appeared shrunk, wrinkled, or deformed. The Thinprep method provided a clearer presentation of subcellular structures and layers, compared to the Cytospin-WG method. In particular, the structures such as nucleus, nucleolus, and chromatin were much more visible with the Thinprep method, allowing a more accurate detection and differentiation of tumor cells. In the Cytospin-WG method, a higher number of cells appeared ruptured or degenerated, and the nucleolus and chromatin structures were rather obscure, making it difficult to detect morphological features of malignant cells or to define their primary origins.

The morphological differences between pathological cell types were mostly observed in subcellular structures including nuclei, chromatin, nucleoli, and vacuoles. The major presentations are as follows:

1. Adenocarcinoma (Fig. 1a-n): scattered (Fig. 1a) or clustered distribution (Fig. 1b); intensely stained cytoplasm with occasionally observed vacuoles (Fig. 1a); most cells had intensely and unevenly stained chromatin and some showed fine granules; most cells had an irregular number of nucleoli, which were irregular in size and stained deep-red(Fig. 1a/b/c/d); some cells showed abnormal nuclear division(Fig. 1e/f/g) and cytoplasmic bridges (Fig. 1c); tumor cells were observable at various mitotic stages (Fig. 1h, i, and j), cell-in-cell arrangement (Fig. 1k) and epithelial like cell distribution were occasionally observed (Fig. 1d). We also observed rimmed structural features in malignant serosal effusion (Fig. 1l), i.e., many tumor cells were distributed in ring-like structure, with giant tumor cells in the center. We also observed mulberry-like structures in which a tumor cell was surrounded by activated and aggregated lymphocytes (Fig. 1m), or tumor cell clusters were surrounded by a great number of lymphocytes (Fig. 1n).
2. Squamous cell carcinoma (Fig. 1o-q): scattered distribution; intensely stained cytoplasm; pink stained keratinized cytoplasm in some cells accompanied by shrunk nuclei (Fig. 1o); intensely stained, flocculated or granular chromatin, some multi-nucleated cells (Fig. 1p); small red-staining nucleoli in most cells located toward the edge of nucleus (Fig. 1p/q).
3. Small-cell lung cancer (Fig. 1r-s): mostly clustered; smaller than other types of tumor cells; greatly increased nucleo-cytoplasmic ratio, nuclei appearing as naked nuclei; single nucleus in most cells; intensely stained coarse granular chromatin; nucleoli not clearly visible.
4. Hepatocellular carcinoma (Fig. 1t-u): scattered distribution, regular shape; clearly defined boundary; intensely stained cytoplasm; intensely and uniformly stained chromatin; a single large, round, and bright red-staining nucleolus found in each of the center of nuclei.
5. Large-cell lung cancer (Fig. 1v-x): scattered distribution; boundary not clearly defined; giant cells (Fig. 1v); single nucleus in most cells; few multi-nuclei cells (Fig. 1w); pale stained cytoplasm with large nuclear-cytoplasimc ratio, nuclei appearing as naked nuclei; intensely stained coarse granular chromatin; obscure red-staining nucleoli in some cells with smaller size (Fig. 1v/x).
6. Malignant melanoma (Fig. 1y-z): scattered or clustered distribution; pseudopodia-like membrane protrusions at the cell periphery; intensely stained nuclei and cytoplasm; some cells multi- nucleolus (Fig. 1y); red-staining nucleolus with irregular size found in the nucleus; some cells multi-nucleated (Fig. 1y); presence of black granular substances in irregular size in the nucleus or cytoplasm (Fig. 1z).

Results from the prospective, comparative study between the Thinprep and Cytospin-WG methods

Morphology.

In the comparative study, the typical features of malignancy such as large size, intense staining, and thickened nuclear membrane of tumor cells were visible in Cytospin-WG preparations (Fig. 2b, d, f, h, j, and l). Abnormal nuclear division (Fig. 2m), cytoplasmic bridges (Fig. 2n), and tumor cells encircled by lymphocytes were also occasionally observed (Fig. 2o). However, compared with the Thinprep results, most chromatin and nucleoli were not presented with clarity (Fig. 2a-l). With the WGC method, cells appeared shrunk and deformed (Fig. 2b/d/f); but in the same subject with the Thinprep method, cells had a full, regular cell morphology (Fig. 2a/c/e). Nipple-like protrusions were commonly observed at the cell periphery with the WGC method (Fig. 2b and j), but not seen in the same subject with the Thinprep method (Fig. 2a and i). Many cells on the glass slide appeared degenerated or ruptured with the WGC method (Fig. 2b and l), but not with the Thinprep method (Fig. 2a and k). This may have resulted from air exposure during dry-fixing and slide preparation, which may be one of the reasons for the low diagnostic rate of Cytospin-WG.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. Side-by-side comparisons of CSF cytology results between the Thinprep plus Papanicolaou stain method (a, c, e, g, i and k) and the Cytospin-WG method (b, d, f, h, j and l) in patients with LM from various solid tumors.

Representative slides from identical CSF samples are shown for the same samples using the two methods in paired photos: a and b, gastric adenocarcinoma origin; c-h, lung adenocarcinoma origin; i and j breast adenocarcinoma origin; k and l, small-cell lung cancer origin. m-p, the typical features of malignancy of tumor cells seen by the Cytospin- WG stain method.

https://doi.org/10.1371/journal.pone.0122016.g002

During Cytospin preparation, cellular components may adhere to the filter paper, resulting in cell loss. Since tumor cells were randomly distributed on glass slides as single cells or in clusters, it would be difficult to determine their numbers and density. However, in CSF samples of equal volume, the number of tumor cells was remarkably different between patients, indicating different tumor cell density between patients. In the comparative study, samples prepared by Cytospin method had a significantly fewer number of tumor cells than those prepared by the Thinprep method. For example, in a case of breast cancer, only two tumor cells were detected in one out of three slides prepared by Cytospin method, whilst over 10 tumor cells were detected in a single slide prepared by the Thinprep method (Fig. 2i-j). In samples that tested negative in Cytospin but positive in Thinprep, only a small number of tumor cells were detected in Thinprep, either as single cells or in small clusters. In these samples, either no abnormal cells were detected in slides prepared by Cytospin, or there were few abnormal cells, but they could not be definitively categorized as tumor cells for the lack of clear presentation of chromatin and nucleoli and other malignant characteristics.

The WG staining is more suitable for hematopoietic cell staining, and is one of the best staining methods for blood and bone marrow smears. Historically, LM diagnosis and treatment have mostly been focused on LM from hematologic malignancies, especially meningeal leukemia. Therefore, at present, the WG stain is routinely used in many institutions in China for CSF cytological examination. Results in this study showed that most solid tumor cells were stained pink or purple-red by the WG staining, and some nucleoli appeared gray in a purple-red background(Fig. 2p), or intensely stained in blue-purple with obscure presentation (Fig. 2d/j). The cellular structures were not presented with sufficient clarity. Only a few samples were stained with clarity (Fig. 2p). In summary, the Cytospin method was less sensitive than the Thinprep in cell collection, which may be one of the major factors causing false negative results, especially in patients who had a low tumor cell density in CSF.

The Pap staining has long been recognized as a suitable stain for malignant tumors of epithelial origin such as cervical cancer and urinary tract tumors [12]. Since solid tumor-caused LM often originates from primary tumors of epithelial origin, Pap staining may be a more appropriate staining method for CSF specimens from these patients. Results from this study showed that in most tumor cells stained by Pap stain, the cytoplasm was stained blue or blue-green; nuclear chromatin was stained blue-purple or dark blue; and nucleoli were stained red or purple. The structural features of chromatin and nucleoli were also presented with clarity. Cellular structures were presented with high clarity and high contrast. Various images of abnormal nuclear divisions and the process of cell division were observed with clarity.