The reliability of microspectrophotometric and flow cytometric nuclear DNA measurements in adenocarcinomas of the breast

The reliability of microspectrophotometric (MSP) and flow cytometric (FCM) nuclear DNA measurements has been studied in 50 human breast adenocarcinomas. The tumor material was obtained by means of fine-needle aspiration biopsy, and all samples except one were found to be highly representative. The results confirm earlier observations that a good correlation exists between modal value (MV) determined by MSP and DNA index (DI) determined by FCM. However, when tumors were classified into low and high malignant variants according to FCM/DI, FCM/S-phase percentages, and MSP histogram types, the concordance was less pronounced. This was found to be due mainly to the fact that in near-diploid tumors a discrepancy exists between MSP and FCM ploidy, as well as between MSP distribution pattern and the estimated percentages of cells in the S-phase region. Another source of discrepancy was observed in tumors with stemlines in the normal tetraploid region, including cells with highly scattered aneuploid DNA values. These tumors were judged by MSP as aneuploid/high malignant and by FCM as euploid/low malignant. In view of this discrepancy, we conclude that the simple determination of the stemline position by MSP/MV or FCM/DI is not sufficient for adequate cytochemical malignancy grading of breast carcinomas. We suggest that a combination of ploidy and percentage of cells scattered outside the modal peaks is a more sensitive method for optimal cytochemical malignancy grading in breast carcinomas.     

Efficacy of combined image and flow cytometric DNA assessments in human breast cancer: a methodological study based on a routine histopathological material of 2024 excised tumour specimens

In 2024 excised specimens of malignant tumours of the female mammary gland. the nuclear DNA distribution pattern of the neoplastic cells was assessed by means of two procedures. One was image cytometry (ICM); here, all the 2024 samples were assessed. The other was flow cytometry (FCM) where 1336 specimens were analysed. In 829 of the 2024 tumour nodules the results of ICM and FCM could be compared. The efficacy of both techniques alone was about 80%; that of the combination was about 60%. In the ICM procedure the main reason for the reduction of samples was the failure to obtain representative specimens. The losses in the FCM method were due to poor quality of the histograms (too much background noise and too broad coefficients of variation). In addition, in as much as one third of all the cases, no specimens were set aside for FCM assessments. In 16% of the samples, where the results of the ICM assessment could be compared with those of the FCM analyses, completely diverging DNA ploidy patterns were obtained. The discrepancy was caused by differences in the interpretation of the histograms. In addition, the calculations of so-called S-phase fractions from the diploid FCM histograms was found to be associated with methodological errors, further contributing to differences in the DNA assessments by means of ICM and FCM. Nevertheless, it was advantageous to use combined ICM and FCM assessments, particularly in the interpretation of DNA histograms of uncommon types.     

Proliferative characteristics and ploidy of human brain tumors by DNA flow cytometry

We studied nuclear DNA distribution by flow cytometry in 59 human brain tumors. Samples were frozen at -20 degrees C immediately after surgery and unicellular suspensions were obtained with a mechanical dissociation technique. Nuclear DNA was stained with propidium iodide. Normal human brain tissue was used as a diploid reference standard. In 86.3% of benign tumors an unimodal DNA distribution with a DNA index usually within the diploid range was found. Among malignant tumors, 64% had un unimodal DNA distribution with diploid or near-diploid modal DNA content. The remaining 36% showed an additional cell peak with a DNA index ranging from 1.15 to 1.92. The percentage of S-phase cells was higher in malignant (median = 3.8) than in benign tumors (median = 1.9) (p less than .001), without correlation to histological tumor subtype.     

Flow cytometric DNA content of fresh tumor specimens using keratin-antibody as second stain for two-parameter analysis.

Studies concerning flow cytometric assessed DNA content reveal problems in interpretating DNA histograms of tumor specimens. The main problems are histograms with a broad coefficient of variation in the G0/G1 fraction; a high G2M fraction and samples with a low percentage of tumor cells. Therefore, in the present study, 382 fresh tumor specimens of carcinomas were analysed routinely, double labeled with, on the one hand, propidium-iodide for assessing DNA content and, on the other, a monoclonal keratin-antibody for marking epithelial and tumor cells. Of the 311 tumor samples, using single parameter analysis 165 (54%) were classified as DNA aneuploid and 146 (46%) as DNA "euploid." By double parameter analysis, 224 (72%) samples were keratin positive and 87 (27%) keratin negative and, of the 224 keratin positive tumors, 175 (78%) were DNA aneuploid and 49 (22%) DNA euploid. The DNA histograms of single and double parameter analysis were compared and it was concluded that in 24 cases (11%) keratin labeling was necessary to recognize DNA aneuploidy. In another 23 (10%) cases, keratin labeling was helpful in assessing DNA aneuploidy. Finally when the results of the 311 samples were combined, 215 (68%) were scored as DNA aneuploid and 99 (32%) DNA euploid. Thus the overall gain in assessing DNA aneuploidy using the double labeling technique is 14%. In conclusion, it is shown that keratin labeling on fresh tumor cell suspensions of epithelial tumors is of additional value in establishing DNA content. Because single parameter DNA assessment is adequate in approximately 60% of the tested samples, the double labeling technique can be performed routinely, or after initial single parameter DNA assessment. Histograms having a broad CV and/or a high G2M are good candidates for the double labeling technique. Using this technique, DNA-content assessment becomes more reliable.     

Methodologic aspects of nuclear DNA assessment of gliomas with astrocytic and/or oligodendrocytic differentiation. Correlation of image and flow cytometric studies on paraffin-embedded specimens

A total of 239 samples from paraffin-embedded, formalin-fixed astrocytic and/or oligodendrocytic gliomas from 111 patients were deparaffinized and disaggregated for image cytometric (ICM) and flow cytometric (FCM) DNA assessments. Each measurement technique produced evaluable histograms in about 85% of the samples analyzed. In the 10% that could not be analyzed by FCM, the background counts were too high and the coefficients of variation were too broad for precise evaluation. The failures with ICM were due to a shortage of Feulgen-stained tumor cell nuclei after the deparaffinization and disaggregation procedures. The results obtained were identical in 77% of the samples evaluable by both methods and practically identical (i.e., euploid versus aneuploid) in an additional 18%. The reasons for completely divergent DNA ploidy patterns in 5% of the samples could not be clarified. About 80% of the histopathologically highly malignant gliomas were found to consist of neoplastic cells with an aneuploid or tetraploid nuclear DNA distribution pattern. The results show that cytometric DNA assessments can be reliably performed on paraffin-embedded specimens of gliomas with astrocytic and/or oligodendrocytic differentiation by means of FCM and ICM on deparaffinized and disaggregated specimens.     

DNA and nuclear protein measurement in columnar epithelial cells of human endometrium

Propidium iodide DNA flow cytometry, Feulgen-DNA, and nuclear light green protein scanning cytometry were performed in columnar epithelial cells of normal, nonmalignant human endometrium and endometrial adenocarcinomas. Columnar cells were identified by immunohistochemical staining for cytokeratin 18, an intermediate filament protein specifically present in columnar cell epithelium. DNA measurements derived from flow and scanning cytometry showed comparable results. The DNA content of the G0/G1 fraction of the adenocarcinomas had a considerable overlap with that of normal endometrium, with that of the carcinomas shifted toward higher values. For the carcinomas, no correlation was found with the histological grade, with the exception of the adenosquamous carcinomas. Most of the clinical stage I tumors showed a DNA content in the normal diploid region. Three of the four carcinomas of clinical stage II and higher had an increased DNA content. For the carcinomas, the percentage of cells in the proliferative fraction, as determined from scanning cytometric derived DNA histograms, was comparable to that of normal endometrium, or higher. No correlation was found with the histological grade. Tumors of clinical stage II and higher had intermediate values compared to carcinomas of lower stages. The nuclear protein/DNA ratio of malignant endometrium completely overlapped that of normal endometrium. Within the tumor population, no correlation was found with the histological grade, with the exception of the adenosquamous carcinomas, and clinical stage. Based on the aforementioned parameters, no discrimination could be obtained between normal and malignant endometrium. However, when the DNA content of the G0/G1 fraction was combined with the coefficient of variation of the nuclear protein/DNA ratio, a clear discrimination could be obtained with only two false-positive cases.     

Automatic analysis of flow cytometric DNA histograms from irradiated mouse male germ cells

An automatic procedure for recovering the DNA content distribution of mouse irradiated testis cells from flow cytometric histograms is presented. First, a suitable mathematical model is developed, to represent the pattern of DNA content and fluorescence distribution in the sample. Then a parameter estimation procedure, based on the maximum likelihood approach, is constructed by means of an optimization technique. This procedure has been applied to a set of DNA histograms relative to different doses of 0.4-MeV neutrons and to different time intervals after irradiation. In each case, a good agreement between the measured histograms and the corresponding fits has been obtained. The results indicate that the proposed method for the quantitative analysis of germ cell DNA histograms can be usefully applied to the study of the cytotoxic and mutagenic action of agents of toxicological interest such as ionizing radiations.     

Nuclear protein content and Ki-67 immunoreactivity in nonneoplastic and neoplastic thyroid cells.

The nuclear DNA content in thyroid tumor cells has been shown to be closely related to the malignant potential of the neoplasm. Besides DNA, nuclear protein (NP) constitutes the major mass of the nucleus. The NP content may vary significantly in relation to the proliferative stage in growing as compared to growth-arrested cells. The increase in NP content associated with the transition from G0 to G1 occurs before the onset of DNA synthesis and may be used to assess growth activity. The nuclear DNA and NP contents were analyzed in 90 nonneoplastic lesions and 75 benign and 62 malignant thyroid tumors. All nonneoplastic specimens were euploid, and 1 of 90 was growth activated. In the group of benign tumors, 59 were euploid, and 16 were aneuploid. Among these there were 5 (9%) of 59 and 6 (38%) of 16 growth-activated specimens, respectively. In the group of malignant tumors 57 of 62 were classified as euploid, and in this group 12 (21%) showed growth activity according to the NP content. Five of 62 were aneuploid, and 3 (60%) of these 5 tumors were growth activated. Evaluation of the growth activity by means of monoclonal antibody Ki-67 was performed on a subgroup of 32 thyroid specimens, both nonneoplastic and neoplastic lesions. Ki-67 immunoreactivity was observed in 0-1.1% cells of 6 nonneoplastic lesions, in 0-3.1% cells of 14 benign cells and in 0.2-3.9% cells of 12 malignant thyroid tumors. Growth activity, as reflected by the NP/DNA ratio and Ki-67 immunoreactivity, appears to be low both in nonneoplastic thyroid lesions and thyroid tumors.     

Comparison of automated and manual techniques for analysis of DNA frequency distributions in bladder washings

Quantitative methods for interpretation of flow cytometry DNA histograms are required for the widespread clinical use of this technology. The usefulness of a histogram analysis technique in this setting requires that it be operator independent, easy to implement in a clinical laboratory, and provide high sensitivity to the desired information. Additionally, the technique must be tolerant of the relatively low signal-to-noise ratios often found in DNA distributions obtained from clinical samples. Among the factors that have been used to assess the malignant potential of tumors are the presence of an aneuploid population, the proportion of hyperdiploid cells, the width of the G1 peak, the DNA index, and the fraction of cells in S. A computer-based method has been developed for extraction of the above-mentioned features from DNA histograms. The program detects peaks in the histogram and uses straight-line fits to the cumulative frequency distribution to define cell population bounds. A test set of 44 histograms compiled from bladder irrigation specimens obtained from patients with a present or past history of transitional cell carcinoma (TCC) was analyzed by five collaborating laboratories forming a Network sponsored by the National Cancer Institute (NCI). This test set was used to evaluate the performance of the computer-based method by comparing results with those of four expert observers. In this preliminary analysis, perfect agreement was found in the detection of aneuploid cell populations by all observers and the computer-based method. Correlation of percent hyperdiploid cell fraction was also excellent.(ABSTRACT TRUNCATED AT 250 WORDS)     

Solid tumor preparation for flow cytometry using a standard murine model

The application of flow cytometry (FCM) to solid human tumors has been hindered by the difficulty in producing high yield, viable, unaltered single cell suspensions. Carcinomas containing a high desmosomal content, such as well-differentiated squamous cell (SCC) cancers of the head and neck (H&N) region, are particularly difficult to prepare. The desire to employ FCM to study cellular DNA parameters of these tumors led to the use of a 3-methylcholanthrene induced murine SCC for the comparative testing of preparative techniques. Dissociation techniques, including mechanical, enucleation, chemical, single and combination enzymes methods, were comparatively tested. Of these, the combination enzyme treatment employing trypsin and collagenase produced the highest cell yields in the shortest time with the highest dye exclusion viability and the least expense. Several fixation systems including glutaraldehyde, paraformaldehyde, acetic acid, and ethanol were comparatively tested using percent of cell loss and quality of the DNA histograms produced as end points. Ethanol-water systems with added fetal calf serum provided minimal cell loss and high quality histograms which were stable for extended periods of time. A murine tumor, closely mimicking the histology of the human tumor of interest, may be used as a model for the determination of optimum techniques of solid tumor preparation for flow cytometric analysis.     

Methodologic aspects of DNA assessment by means of image cytometry in tumors of the salivary glands. A comparison between the results obtained using sections and cytospin preparations from the same paraffin-embedded specimens

The image cytometric nuclear DNA assessments on paraffin-embedded tissue sections and on Cytospin preparations of disaggregated specimens from the same cases were compared in 98 salivary gland tumors, including 21 acinic cell carcinomas, 29 mucoepidermoid carcinomas, 21 adenocarcinomas and 27 adenoid cystic carcinomas. The histogram type (diploid, tetraploid or aneuploid) and the number of cells with DNA values greater than 2.5c (expressed in relative units) were considered as variables in the correlation. A high correlation between the results in different specimens was found in acinic cell carcinomas, mucoepidermoid carcinomas and adenocarcinomas; the histogram type and the number of cells with DNA values greater than 2.5c were essentially the same between specimen types in these three tumor entities. The cases of adenoid cystic carcinomas showed a considerably lower degree of correlation: in 8 of the 27 cases, the Cytospin preparations yielded diploid histograms, while the tissue sections yielded aneuploid histograms. The number of cells with DNA values greater than 2.5c was notably lower in the Cytospin preparations from adenoid cystic carcinoma; the reasons for this exceptional behavior of the cells of adenoid cystic carcinoma are discussed. These findings demonstrate that paraffin-embedded specimens of different tumor entities, even from the same organ, can be affected differently by disaggregation procedures. While retrospective studies on disaggregated paraffin-embedded specimens can yield reliable results, comparative assessments using both DNA analysis techniques, as in this study, should be performed before a large number of cases is evaluated.     

DNA assessment in thin sections of lymphomas by densitometric and stereological methods: comparison with imprint and flow cytometric results

The nuclear DNA content was estimated in 2 microns sections of 18 lymphoma cases by two methods: (1) Feulgen densitometry using QTM 900 with correction by Bins' procedure which allows size-independent DNA distributions; (2) stereological unfolding as proposed by Cruz-Orive giving sphere-size distributions. A general correlation was found between results and DNA measurements obtained by imprint and flow cytometric techniques in the same specimens. When histologic DNA profiles were compared to cytologic histograms, a high correlation was found between the distribution of ploidy classes by correspondence analysis. However two highly proliferating lymphomas were erroneously classified as aneuploid. Conversely, sphere-size distributions allowed the identification of the majority of aneuploid lymphomas but failed to recognize proliferating ones. It appears that when cytologic specimens are not available, densitometric studies on sections may provide valuable information on DNA content, with complementary data obtained from stereological procedures.     

Cell kinetic analysis of mixed populations using three-color fluorescence flow cytometry.

The development of antibodies to DNA-incorporated thymidine analogs has in turn led to the development of flow cytometric techniques for rapidly measuring cell kinetics parameters. More recently, these techniques have been applied to clinical tumor material. One problem with such measurements has been the difficulty of distinguishing malignant cells from coexistent normal cells in the biopsy material. In the present study, the feasibility of selecting out the desired malignant cell population for kinetic analysis from a mixture of cells was tested in vitro. An anticytokeratin antibody was used to discriminate between a mixture of tumor cells (cytokeratin positive) and normal cells (cytokeratin negative). The cell lines chosen for the study, A549 human lung carcinoma cells and Chinese hamster ovary (CHO) cells, were pulse labeled with iododeoxyuridine (IdUrd) and sampled every hour up to 16 hours. Selecting out cells from the mixture required the application of three-color fluorescence flow cytometry, which was carried out using the fluorochromes FITC (fluorescein isothionate, green fluorescence, IdUrd-DNA antibody), PE (phycoerythrin, orange fluorescence, cytokeratin antibody), and PI (propidium iodide, red fluorescence, DNA). This allowed single laser excitation. The staining procedure involved incubation with the IdUrd antibodies (specific antibody plus FITC-conjugated second antibody) followed by the cytokeratin antibodies (specific antibody plus PE-conjugated second antibody) and lastly by the DNA stain containing RNase. Two analysis methods of the IdUrd/DNA cytograms were applied: a mid-S window analysis and a relative movement (RM) analysis. Results of the analyses for cells selected out of mixtures were compared with results of cells stained and analyzed separately. A clear separation of the two cell lines could be obtained on the basis of orange fluorescence (cytokeratin content) despite a large overlap of their DNA histograms. By gating on high or low orange fluorescence, almost pure populations of the individual cell types could be selected out for further kinetic analysis. Little difference was seen, with both the mid-S and RM analyses, between cells gated from mixtures or stained separately. It is concluded that this technique is feasible for use on clinical material, provided good cell suspensions can be obtained, leading to the hope of increasing the accuracy of kinetic measurements on human tumors.     

Image and flow cytometric analysis of DNA content in breast cancer. Relation to estrogen receptor content and lymph node involvement

A comparative flow-cytometric and image-cytometric study was performed on 166 human breast cancers. Parallel measurements of 67 cases showed a good correlation between the DNA indices measured with each of the techniques. However, minor ploidy abnormalities were detected with flow cytometry. Only with this technique about 70% of the tumors appeared to be aneuploid. DNA profiles obtained with image cytometry frequently contained DNA values above the modal G2 value of the tumor. At least part of these broadly scattered DNA profiles resulted from multiploid tumors, as was found with flow cytometry. The mean percentage of cells with DNA contents exceeding the modal G0G1 value in image-cytometric DNA profiles (non- G0G1 cells) appeared to be highest in aneuploid lymph-node-positive (N+) tumors and lowest in near-diploid lymph-node-negative (N-) tumors. Near-diploid tumors were more frequently (70%) estrogen receptor positive (ER+) than were aneuploid tumors (50%), whereas highly aneuploid tumors with low or negative ER contents tended to be N+. Tumors with a high percentage of non- G0G1 cells value were predominantly estrogen-receptor negative (ER-).     

Qualitative and quantitative changes in nuclear DNA and phenotypic gene expression in human malignant skin tumors during their progression.

Qualitative and quantitative changes in nuclear DNA and phenotypic expression of human malignant skin tumors were examined during the course of progression. The numerical abnormalities of chromosomes demonstrated by interphase cytogenetics using the chromosome-specific in situ hybridization technique, were also used to reveal qualitative DNA changes in malignant tumor cells. For the analysis of the quantitative changes in nuclear DNA, fluorescence cytophotometry was used on the DAPI-stained tumor cells isolated from the paraffin-embedded sections. To survey abnormal gene expression in malignant tumor cells, lectin histochemistry for different sugar residues, immunohistochemical staining of HLA-DR, and in situ hybridization for H-ras, c-myc, N-myc or v-fos were used. The results showed that: 1) in one case of squamous cell carcinoma with invasion, the number of chromosomal abnormalities was much greater in the invasive than in non-invasive parts, with marked topographical heterogeneities; 2) the DNA-ploidies were largely shifted to the higher side with aneuploid stem-lines and polyploid cells in the invasive parts of all malignant tumors; 3) the expression of HLA-DR was induced at the invasive fronts of malignant melanomas; 4) the GS-I specific sugar residue(D-galactose) appeared in all extra-mammary Paget's cells; and 5) expression of "oncogenes" was found in about 60% of all malignant tumors examined. Thus, the progression of malignancy is accompanied by both qualitative and quantitative changes in nuclear DNA, resulting in abnormal gene expression.     

Different opinions on classification of DNA histograms produced from paraffin-embedded tissue

Flow cytometric DNA ploidy determination has been regarded as an objective prognostic parameter in several types of human cancer. To test whether DNA histograms are similarly interpreted, a series of flow cytometric DNA histograms was posted to six investigators working in the field for independent classification. The histograms were produced from paraffin-embedded adrenal adenomas or non-neoplastic tissue and had several different patterns. Only 44% of the histograms were similarly classified by all investigators, and 85% by five of the six participants, when DNA ploidy was evaluated. Different criteria for tetraploidy existed, and also some uncertainty in classifying peridiploid and small aneuploid peaks. It is concluded that lack of consensus on histogram classification may result in widely varying percentages of DNA aneuploid tumors found even if the data are similar. Until general agreement is reached on the definition of DNA aneuploidy and its subclasses, classification of DNA histograms is variable and subjective.     

Mammary carcinoma. Comparison of nuclear DNA content from in situ and infiltrative components

Twenty-eight mammary carcinomas were analyzed with respect to their nuclear DNA content. Ten of the carcinomas were entirely in situ (noninfiltrative) while 18 showed areas of both infiltrative and noninfiltrative growth. The DNA content of individual tumor cells was measured in sections from the original paraffin-embedded specimens. In the tumors that had noninfiltrative as well as infiltrative zones, DNA analyses were performed in both areas. Comparison between the DNA patterns obtained from these different areas of the same tumor showed very close agreement. Both groups of tumors (those with and those without areas of invasion) contained some cases that showed a euploid DNA pattern and some cases that showed an aneuploid pattern. Furthermore, analysis of the DNA content of regional lymph node metastases in seven of the invasive cases did not show an increased aneuploidy in the metastases. The results suggest that, in mammary carcinomas, invasive and noninvasive tumors cannot be distinguished by DNA analysis and that tumor progression does not seem to be associated with a significant alteration of the nuclear DNA content.     

Morphometry, densitometry and pattern analysis of plastic-embedded histologic material from urothelial cell carcinoma of the bladder.

An image analysis method of grading histologic sections of bladder carcinoma was tested. The method was new in four respects. First, for fixation of the biopsies a coagulant fixative was used. Second, 2-microns plastic sections were used to ensure the reproducibility of nuclear imaging. Third, a new stereologic approach was used for calculation of the nuclear volume and DNA content. Fourth, for the classification rule the morphometric, densitometric and texture features were used in concert. The IBAS 2000 instrument was used for the measurements. Texture analysis of the chromatin patterns was performed using Markovian texture features. Using discriminant analysis, of 22 parameters, 2 morphometric, 2 densitometric and 3 texture features were selected for the classification rule. With them, 89% of the bladder carcinomas were correctly classified into the three grades. All grade III tumors were classified correctly. Among the features tested, the densitometry of the DNA had the highest F values. All of the grade III tumors and 45% of the grade II tumor group had DNA histograms indicating aneuploidy. This study showed that plastic-embedded material is well suited to morphometry and densitometry and can be used for quantitative grading of bladder carcinoma.     

Detection of underlying characteristics of nuclear chromatin patterns of thyroid tumor cells using texture and factor analyses

BACKGROUND: Aspiration biopsy cytology of thyroid tumors has been used more frequently in recent times to differentiate between malignant and benign lesions. Chromatin patterns of the tumor cell nuclei are one of most important factors for cytologic diagnosis. The interpretation of nuclear chromatin patterns is subjective and more difficult than that of nuclear size or shape. In the present report, we investigated how to detect underlying chromatin characteristics of benign and malignant thyroid tumor cells by means of texture and factor analyses. METHODS: We employed a computer-aided system in which light microscopy was combined with an image processor and monochrome camera. Using this system, 100 randomly selected cells in a Papanicolaou stained, aspiration biopsy cytologic smear in each case of 39 benign and malignant thyroid tumor cases were digitized. We applied two-dimensional and higher texture analyses with the use of co-occurrence and run-length matrices to analyze the chromatin patterns. Factor analysis was used to determine whether a large number of independent variables actually measured one or more underlying common variables. RESULTS: According to parameters with high factor-loading values, the morphologic chromatin characters were classified into three categories according to heterogeneity, contrast, and homogeneity of chromatin patterns. On the basis of analyses with these morphologic categories, nuclei of papillary carcinoma showed higher contrast of chromatin patterns than did those of the benign group. Moreover, there was a variety of contrasting chromatin patterns among cells in each papillary carcinoma case in comparison with the benign group. In contrast, follicular carcinomas showed a significant difference in the standard deviation of factor 3, which indicated more monotonous chromatin patterns among cells in each follicular carcinoma case than in each benign case. CONCLUSION: We believe that this technique, using texture and factor analyses, is useful in the detection of underlying characteristics of nuclear chromatin patterns in aspiration biopsy cytology.