The establishment of individualized disease course predictions for patients, rather than prognosis for a population of patients with this disorder (1), will increasingly influence clinical decision making. For example, in the treatment of tumors of the parotid and other salivary glands it is important to exclude or confirm malignancy as soon as possible. The reasons are obvious, since surgery is more extensive for malignant tumors than for benign tumors. Using the parotid gland as an example, lateral parotidectomy is performed for benign tumors, whereas malignant tumors require subtotal or radical parotidectomy including primary resection plus reconstruction of the facial nerve. In addition, patients with malignant tumors may benefit from neck dissection for preventing delayed clinical manifestation of occult metastasis even if preoperative diagnostic procedures, such as palpation, sonography, or CT do not provide evidence of lymph node involvement (2). As extended surgery bears a high risk of complications, such as temporary or permanent facial nerve palsy, patients with parotid tumors of uncertain dignity currently undergo inaugural lateral parotidectomy in order to obtain a definitive histopathological diagnosis. In case of malignancy, radical resection of the medial parotid lobe with or without facial nerve resection and neck dissection is performed in a subsequent session. Submandibular gland tumors and those of the small glands within the mucosa are treated analogously. This strategy avoids overtreatment for benign tumors, while patients with definite malignancy are exposed to higher risk of increased morbidity and loss of facial nerve function. This approach also contributes to overhead costs for the health care system.
Clinical indicators of malignancy include facial nerve palsy as a sign of infiltrative growth, clinical loco-regional lymph node metastases, and infiltrative growth to neighboring tissues like the mandible. Other symptoms, such as rapid tumor growth and pain, are weak indicators and not recognized as proofs of malignancy. Many tumors are accessible for incision biopsies; but facial nerve injury is a crucial point, particularly in the large salivary glands. In addition, in the case of the most common benign tumor, the pleomorphic adenoma, it is well known that rupture of the capsule and tumor cell spread leads to multiple recurrences with the potential for malignant transformation (3). In search for novel approaches to pre-therapeutically assess the dignity of salivary gland tumors, preference should be given to minimal invasive procedures such as FNABs and swabs (4).
An increasing number of diagnoses are currently analyzed by cytopathology. This is an inexpensive and quick procedure, but due to the overall rare occurrence of salivary gland tumors, final diagnosis is often committed to a few specialized pathologists (5). Specimens can, furthermore, be analyzed by semi-automated cytometric measurements, yielding quantitative data such as cellular DNA content. DNA ploidy has been shown as a prognostic marker in several tumor entities including the parotid gland (6–8). Up to now, the DNA ploidy of solid carcinomas is determined by image analysis or flow cytometry. Both methods are well described and numerous standardized protocols have been published (9–11). Both methods have, however, certain weaknesses. Rather few cells (150–300) are analyzed by image analysis, and only one parameter (i.e., DNA content) is determined in most protocols; analysis is time-consuming and may give rise to inter-observer variation, since the target cells are selected individually. Flow cytometry, in contrast, allows multiparameter analysis of bulk cell numbers, but physiologic characteristics cannot be attributed to cellular morphology.
In the present study we make use of the laser scanning cytometer (LSC) to determine the DNA ploidy. This microscope-based instrument combines high throughput multiparametric cytometry and morphological analysis. LSC was commercially introduced in the early 1990s; its instrumentation and software are explained elsewhere (12–14). Numerous assays have been developed for different experimental applications, but only few assays are applicable in a routine clinical context (15).
The most important feature of the LSC concerns the recording of the exact position of every object together with the fluorescence data. Among other capabilities this allows 1) verification whether objects are single cells, doublets, debris or artifacts; and 2) documentation of the cell morphology. To this end, the slide can be removed from the stage, stained by conventional cytological methods (Giemsa, Hematoxylin, and Eosin), and followed by re-analysis on the stage (16).
In conclusion, this slide-based design makes it very suitable for analyzing hypocellular specimens such as FNABs. The data can be evaluated in a standardized and objective way, which is extremely important for clinical decision making. We and others have compared DNA ploidy analysis by LSC with image analysis after Feulgen staining or with flow cytometry, and showed a good correlation of both methods (14, 17, 18). It was therefore investigated in this pilot study whether the LSC has the potential to provide preoperative prediction of tumor dignity in solid salivary gland tumors. Cellular DNA was stained by propidium iodide (PI) and cytokeratin by immunofluorescence (19). Similar assays have already been described for specimens from voided urine and other hypocellular samples (20).
- Top of page
- MATERIALS AND METHODS
- LITERATURE CITED
FNABs yielded typically between 5,000 and 50,000 cells. Erythrocytes were effectively destroyed by treatment with lysis buffer. We did not experience problems related to insufficient lysis of erythrocytes. The morphology of the cells was well preserved.
Cover slip removal and HE restaining did not lead to a significant cell loss, since none of 20 regularly relocalized cells within the different cell subsets (typically between 100 and 200 cells per sample) was missing. Even sporadically analyzed sets of up to 50 consecutive cells were relocalized without missing cells.
All tumors were examined by an experienced pathologist (A. Tannapfel) by routine histopathologic diagnosis. A data sheet of all 51 tumors was prepared (Table 1) comparing LSC and routine histopathology. Three of 14 histologically malignant tumors turned out to be DNA diploid by LSC. None of the histologically benign tumors was misdiagnosed as malignant by LSC. Overall, LSC data showed an excellent correlation with histopathology as determined by Fisher's exact test with a positive predictive value of 1.0 and a negative predictive value of 0.93. Sensitivity was 0.79 and specificity was 1.0 (Table 2).
Table 1. Patient Data on LSC and Histopathology
|Patient number||Localization||DI-peaks||5cER||LSC classification||Pathology classification||Histology|
|1||Parotid gland||1.00||0.4||Benign||Benign||Pleomorphic adenoma|
|2||Parotid gland||1.00/2.00||3.6||Benign||Benign||Pleomorphic adenoma|
|3||Parotid gland||1.04/2.08||1.8||Benign||Benign||Pleomorphic adenoma|
|4||Parotid gland||1.00/2.08||2.1||Benign||Benign||Pleomorphic adenoma|
|5||Parotid gland||1.00/2.00||4.7||Benign||Benign||Pleomorphic adenoma|
|6||Parotid gland||1.00/2.00||1.2||Benign||Benign||Benign lymphoepithelial lesion|
|7||Parotid gland||1.00/2.00||1.3||Benign||Benign||Pleomorphic adenoma|
|8||Parotid gland||1.00/2.06||3.3||Benign||Malignant||Squamous cell carcinoma|
|10||Parotid gland||1.00/1.12/2.16||3.3||Malignant||Malignant||Azinic cell carcinoma|
|11||Parotid gland||1.00/2.00||0.9||Benign||Benign||Pleomorphic adenoma|
|12||Parotid gland||1.00/2.00||0.9||Benign||Benign||Benign lymphoepithelial lesion|
|13||Parotid gland||1.00/2.00||0.8||Benign||Benign||Pleomorphic adenoma|
|14||Parotid gland||1.00/2.00||1.6||Benign||Benign||Pleomorphic adenoma|
|15||Parotid gland||1.00/2.00||2.1||Benign||Benign||Pleomorphic adenoma|
|17||Parotid gland||1.00/2.00||0.7||Benign||Benign||Pleomorphic adenoma|
|18||Parotid gland||1.00/2.00||1.4||Benign||Benign||Chronic inflammation|
|19||Parotid gland||1.00/1.62/1.86/3.52||13.5||Malignant||Malignant||Metastasis of melanoma|
|20||Parotid gland||1.00/2.00||0.7||Benign||Benign||Pleomorphic adenoma|
|22||Parotid gland||0.70/1.00/2.00||1.8||Malignant||Malignant||Malignant hidradenoma|
|23||Parotid gland||1.00/2.00||0.9||Benign||Benign||Pleomorphic adenoma|
|24||Parotid gland||1.00/2.00||3.3||Benign||Benign||Pleomorphic adenoma|
|25||Parotid gland||1.00/2.00||1.0||Benign||Benign||Pleomorphic adenoma|
|26||Parotid gland||1.00/2.00||0.8||Benign||Benign||Pleomorphic adenoma|
|27||Parotid gland||1.00/2.00||2.9||Benign||Benign||Benign granular tumor|
|28||Parotid gland||1.00/2.00||2.4||Benign||Benign||Intraglandular lymph node|
|30||Parotid gland||0.56/1.24/1.84/2.46||1.3||Malignant||Malignant||Metastasis of melanoma|
|31||Parotid gland||1.00||2.8||Benign||Benign||Pleomorphic adenoma|
|32||Parotid gland||1.00/1.08/1.48/2.28||1.3||Malignant||Malignant||Mucoepidermoid carcinoma|
|33||Parotid gland||1.00||0.3||Benign||Benign||Intraglandular lymph node|
|34||Parotid gland||1.00||0.1||Benign||Benign||Intraglandular lymph node|
|35||Parotid gland||1.00/1.20/1.66/2.24||5.5||Malignant||Malignant||Squamous cell carcinoma|
|36||Parotid gland||0.36/1.00||2.2||Malignant||Malignant||Squamous cell carcinoma|
|37||Parotid gland||1.00/2.08||3.0||Benign||Benign||Pleomorphic adenoma|
|38||Parotid gland||0.52/1.00/1.40/2.10||0.5||Malignant||Malignant||Squamous cell carcinoma|
|40||Parotid gland||1.00/2.00||1.2||Benign||Benign||Pleomorphic adenoma|
|41||Parotid gland||1.00/2.00||3.6||Benign||Benign||Pleomorphic adenoma|
|43||Submandibular||1.08/1.54/2.16/3.24||10.8||Malignant||Malignant||Metastasis of Merkelcell-carcinoma|
|45||Submandibular||1.00||1.3||Benign||Benign||Chronic sclerosing sialadenitis|
|46||Submandibular||1.00||1.7||Benign||Benign||Chronic sclerosing sialadenitis|
Table 2. Classification of LSC Analyzed FNAB Samples From Solid Salivary Gland Tumors and Their Histopathology
| ||Histology malignant||Histology benign||Sum|
- Top of page
- MATERIALS AND METHODS
- LITERATURE CITED
The setup of our assay includes only one washing and spinning step prior to immobilization of the cells on the slide. This is important because FNABs are hypocellular specimens, and repeated washing steps may reduce cell numbers. All staining steps are performed on the slide after fixation of the cells. Immobilization was stable enough to allow a second staining procedure, i.e., HE, without considerable cell loss.
Some samples were analyzed as multiple aliquots at different times after FNAB preparation. No differences in cytokeratin expression or DNA ploidy after prolonged storage even for up to 18 months were observed. This indicates that cell constituents are very stable once cells are immobilized and fixed on the slide. Specimen stability may be of special value in medicolegal issues since specimens can be used in aliquots for a second analysis and stored as conventional cytological specimens.
Although a relative small number of tumors (N = 51) was analyzed is this study, a broad spectrum of histopathological diagnoses was covered by the benign as well as by the malignant tumors. This broad spectrum is typical for salivary gland tumors and explains why analysis by conventional cytology is difficult. Nevertheless, all tumors diagnosed as benign by routine histopathology showed DNA diploid tumor cells by LSC analysis. Some malignant tumors were DNA diploid as well, but most importantly DNA aneuploidy was restricted to malignancy. In addition to DNA ploidy, the 5cER was evaluated as another sign of malignancy. The 5cER is a long established prognosis-related feature in image cytometry of squamous cell carcinoma of the head and neck (21). It was shown on Feulgen-stained cytological specimens that the 5cER increases with higher grade of malignancy.
Keeping the limitations of an initial study - relative small patient number, heterogeneity of tumors - in mind, the assay described here promises to generate objective preoperative data (DNA ploidy and 5cER) on solid salivary gland tumors. The excellent correlation of LSC classification with routine histopathology and the PPV and NPV, as well as the specificity and sensitivity of the LSC classification, could make this analysis a valuable tool for the clinician and surgical oncologist. If these data were confirmed by a large-scale multi-center study, they could be included in the therapeutic strategy decision-making process. Three out of 14 malignant tumors were missed by LSC (Table 1); these three tumors comprised one squamous cell carcinoma (#8) and two MALT-lymphomas (#9 and #39). This indicates the present limitation of FNAB analysis solely by LSC in case of DNA diploid malignant tumors. DNA diploid tumors could additionally be stained for surface markers, or other innovative techniques such as the detection of DNA strand breaks (22) and of certain other features (23) could be employed. Not only the cytokeratin positive population, but also the cytokeratin negative population can be used for the detection of DNA aneuploid cells or nuclei as exemplified by patient 29, where a mixed malignant tumor was present. Additionally, specimens could be analyzed by conventional cytology after LSC analysis. In conventional cytology, tissue fragments rather than single cells are analyzed. On the single cell level, nuclear to cytoplasmic ratios in HE-stained cells may be used as an additional indicator for the detection of malignant cytokeratin positive DNA diploid cells. In tissue fragments, DNA diploid malignant tumors may exhibit further distinctive features of malignancy, such as growth patterns being not apparent at the single-cell level.