Flow Cytometry of Human Colorectal Tumors: Nuclear Isolation by Detergent Technique

Transcription

1 Alan R. Liss, Inc. Cytometry 6: (1985) Flow Cytometry of Human Colorectal Tumors: Nuclear Isolation by Detergent Technique Steffen E. Petersen The Institute of Cancer Research, Aarhus, Denmark Received for publication June 15,1982; accepted April 10, 1985 A simple one-step technique for detergent isolation and DNA staining of nuclei from mouse colon and from human colorectal tumors was investigated. Nuclear yield increased with treatment time, up to 24 h. There were only minor differences when detergent concentrations from 0 to 0.6% were used. The lowest (0.03%) concentration was most effective. No loss of nuclei was effected by cell lysis and no selectivity was observed for isolation of certain cell-cycle phases or ploidy classes from heterogeneous tumors. The nuclei were stable in the stain-detergent solution for 24 h, but lymphocytes were sensitive to the possible proteolytic activity of one of the two commercial RNase preparations. Of the total number of parenchymal nuclei in mouse liver, as estimated by a stereological method, approximately 60% were isolated by the procedure (approximately 0.9 x 10* nuclei/g tissue). From mouse colon the average nuclear yield was 1.8 x 10s/g, and from human colorectal tumors 0.9 x l@/g (ranges x 10'). Microscopic examination of undissolved tissue fragments from the preparation of tumors and mouse colon showed a high selectivity for isolation of epithelial and neoplastic nuclei, leaving the stroma with its nuclei almost intact. Key terms: Nuclear isolation, detergent technique, tumor dispersal, flow cytometry, colorectal tumors In flow cytometry (FCM) of solid tumors and tissues, the general problem is to obtain representative, wellpreserved, and quantitatively sufficient cell samples (6). For measurements of nuclear DNA content, a number of techniques have been developed to isolate single nuclei using hypotonic media (10,13), pepsin digestion (221, detergent treatment (18,19,20), detergent plus enzyme treatment (21), or other methods (7,9,12). Pepsin and detergent techniques have been used in a great variety of tumors and tissues, giving isolated nuclei well suited for FCM, as indicated by reproducible DNA distributions with low coefficients of variation (CVs). In any isolation technique, however, an intensive tissue disaggregation implies increasing cell damage and loss. Evaluation of possible selectivity in cell or nuclear yield or loss is therefore important, but rather few studies deal comprehensively with these problems. The aim of the present study has been to evaluate a previously described technique for FCM of solid tumors (20) in its application on human colorectal tumors and on mouse colon tissue as a model system. Nuclear yield, possible selectivity in isolating different types of nuclei, and stability of the isolated nuclei have been examined. MATERIALS AND METHODS The design of each experiment is generally given in the results section. The following deals with basic principles. Reagents Stain-detergent solution. Ethidium bromide (EB) 10 mg (BDH Chemicals Ltd., Poole, UK); Trisodium citrate, 1,000 mg; NaC1, 564 mg; add water to 1,000 ml (deionized laboratory water purified by the Milli-Q reagent grade water system, Millipore, Bedford MA); nonionic detergent: Nonidet P-40 (NP-40)(Shell, Carrington, UK) standard concentration in the solution 0.03% vlv. Tonicity: approximately 10% of isotonic saline. TRIS buffer. TRIS, 12.1 g (Sigma 7-9, Sigma Chemical Company, St. Louis, MO); Na EDTA, 370 mg (E. Merck AG, Darmstadt, GFR); water, 920 ml; HC1 1 N (analytic grade), 80 ml; ph 7.4. Supported by grants from The Danish Cancer Society (58177, 14181) and The Danish Medical Research Council ( ) Address reprint requests to Steffen E. Petersen, The Institute of Cancer Research, Radiumstationen, 8000 Aarhus C, Denmark.

2 Ribonuclease (RNase). Ribonuclease from bovi: pancreas, activity 50 Kunitz unit/mg (Cat. No , BDH Chemicals Ltd. Poole, UK). One milligram RNase was added to 100 ml stain-detergent solution before each preparation. Tissue Handling Biopsy specimens from surgically removed colorectal adenomas and carcinomas were minced with scalpels, washed once in TRIS buffer, and suspended in the staindetergent solution in 10-ml centrifuge vessels under continuous slow mixing at 4 C. Fine-needle biopsies were obtained from tumors by aspiration through a 21-gauge injection needle and suspended in TRIS buffer by flushing of the syringe. Liver tissue,whole colon, and rectum specimens were obtained from 4 to 26-wk-old C3DzF1/ Bom mice. After having been rinsed in TRIS buffer and weighed, the material was processed as the tumor biopsies. The nuclear isolation and staining period was varied from 10 min to 24 h. Prior to FCM, the sample was filtered through a 50-pm nylon filter. In some experiments, the retained, undissolved tissue fragments were collected from the filter, pelleted by centrifugation, and fixed in 10% formaline for histologic examination. Ethanol Fixation of Isolated Nuclei For long-term storage this fixation was performed by centrifuging the isolated and filtered nuclei in the staindetergent solution (250 g for 5 min). The nuclei were resuspended in 0.5 ml of the solution, and ice-cold ethanol 99% was added drop by drop to 10 ml during intensive agitation. The nuclei were stored in ethanol at 8" for up to 3 mo. Prior to analysis they were washed once in TRIS buffer, restained with the standard staining solution, and subjected to 5 s of ultrasound at effects of 20 to 30-watt (Sonifier B12, Bramson Sonic Power Company, Danburg, CT). Flow Cytometry and Nuclei Counting Some analyses of colorectal tumors were performed in a Cytofluorograf 4800A (Bio-Physics Systems, Inc.), using the counting rate as a relative measure for the nuclei concentration of the sample. Other analyses were performed using a microscope-based flow cytometer, where the sample was delivered from a syringe, giving a fixed and known flow, thus providing a direct measure of the nuclei concentration from the counting rate (16). A calculation of the percentage of cells in the GO-G1, S, and G2-M cell cycle phases was performed from the DNA histograms, using a simple model (3). Estimation of the Number of Nuclei per Gram of Mouse Liver Three livers from 6-mo-old CDF mice were weighed before and after being immersed in water, and the specific gravity was calculated from the difference. Five biopsies of 1 mm3 were taken at various depths from each liver, immersion-fixed in 2% glutaraldehyde in cac- NUCLEAR ISOLATION BY DETERGENT 453 odylate buffer and embedded in Vestopal. Shrinkage was estimated from tissue bars of 5-mm length processed with the biopsies. Histological sections, 0.5 pm thick, were stained with 1% toluidine blue. In one section per biopsy, the number of nuclear profiles from parenchymal cells were counted in ten blindly selected squares measuring 70 x 70 pm. A Zeiss projection microscope and a linear magnification of 1,000 x were used. The mean diameter of the nuclei, which appeared almost circular, was estimated from the diameter distribution of approximately 300 nuclear profiles per liver and from tables for evaluation of spheresize distributions in sections (17). The volume density of nuclei was then calculated from the formula: number of spheres per volume = number of sphere profiles per aredmean diameter of spheres (8). About 0.5 g of each liver was also treated with the standard stain-detergent solution. After 24 h, the number of isolated nuclei was counted in a hemocytometer which also was used to score the percentage of nonspherical, presumably stromal, nuclei. RESULTS Yield and Stability of Nuclei from Mouse Colon The effects of NP-40 concentrations of 0.1% and 0.03% and 0% in hypotonic citrate buffer on mouse colon tissue were studied. CDFl mice aged 4 to 22 weeks were used in eight independent experiments. Since age of the mice did not influence the results consistently, these were pooled (Fig. 1A). NP-40 in 0.03% gave the fastest and highest yield of nuclei with a final mean value of 1.8 x 108/g tissue. Without detergent, a slower release of nuclei was obtained, although approaching the 0.03% results at 24 h. Cells isolated by filtering after the mechanical step of the preparation were treated for 1 to 24 h in the 0.03% NP-40 under continuous mixing. No decrease in average number of nuclei was seen, rather a slight increase per gram (from 0.29 x 10' to 0.36 x 10'1, probably due to release of single nuclei from small cell clumps passing the 50-pm filter. In a similar experiment a comparison was done between 0.03% and 0.6% NP-40. Slightly higher isolation rates were still shown with the 0.03% preparation. As with 0.03% NP-40, the 0.6% preparation did not affect the stability of nuclei stored from 1 to 24 h in the staindetergent solution. Yield and Stability of Nuclei from Human Colorectal Tumors Experiments, designed in the same way as the mouse colon experiments, were performed with four human colorectal carcinomas. Results are given in Figure 1B. The maximal nuclear yield per gram of tumor tissue varied from 1.9 x 10' to 0.3 x 10'. In two tumors the highest values were obtained after 1 h of preparation, and in the other two after 24 h. Again, 0.03% NP-40 proved slightly more effective than 0.1% and no-detergent treatment. Cells, isolated mechanically, filtered,

3 454 PETERSEN A NUMBERS OF NUCLEI xio8 PER GRAM TISSUE 100% h NUMBERS OF NUCLEI IN PERCENTAGE OF MAXIMUM FOR EACH TUMOR -'i HOURS I 10 % w- HOURS FIG. 1. Comparison of nuclear yield using the detergent NP-40 in concentrations of 0.1%, 0.036, and 0% on (A) mouse colon tissue. Pooled results from eight independent experiments giving the mean values of the numbers of isolated nuclei per gram of colon tissue from CDF mice aged 4-22 wk; (E3) Mean values of the numbers of isolated nuclei from four colorectal carcinomas in percentage of the single highest value for each tumor. 8 indicates cells isolated only by the mechanical treatment followed by filtration through the 50-pm filter, and 1-h detergent treatment to make the nuclei stainable; * represents the same sample after a further 23 h of storage in the 0.03% detergentstain solution under continuous mixing. All preparations were made in duplicate. Bars indicate standard error of the mean. Table 1 Effects of Preparation Time on Nuclear Isolation From Colorectal Tumors Type of No. of Isolation time tumor tumors 2-5 h 24 h Nuclear yield at Carcinomas h (in % of (74-261) value at 2-5 h)" Adenomas (73-120) Percentage of Carcinomas hyperdiploid cells (mean values) Percentage of S + G2-M cells Carcinomas (mean values) amean value with ranges in parentheses. and thereafter treated with 0.03% NP-40, did not decrease in number by detergent treatment of 1 to 24 hr. Comparison of the nuclear yield after a few hours vs. 24 h of detergent treatment was performed on another 17 tumors (Table 1). The average cell yield was highest at 24 h but with some individual variation, including a few cases with a decrease. No selectivity for certain ploidy classes or cell cycle phases was observed in the 7 and 11 cases, respectively, where these parameters could be evaluated. The above experiments analyzed the stability of already isolated nuclei in the detergent-stain solution but did not deal with possible cell loss in the detergentmediated cell lysis itself. The following experiment analyzed this on nine colorectal tumors, from which cells primarily were isolated by the mechanical step alone. After being minced, the tumor tissue was suspended in PBS and filtered through the 50-pm nylon filter. This gave a suspension of single cells and few small clumps. The cell concentration was counted in a hemocytometer,

4 and aliquots of the cells were fixed by adding 99% cold ethanol to a final concentration of 60%. Aliquots of unfixed cells were stained for 24 h with the standard detergent-stain solution while being continuously mixed. The resulting nuclei concentrations were measured when the samples were analyzed in the flow cytometer, and the histograms were compared with histograms from the ethanol-fixed samples stained with stain-detergent solution for 1 h. The undispersed tumor tissue that remained after the mechanical step was collected from the filter and analyzed after 24 h of standard staining. All samples were made in duplicate, and the mean of the results was used. The following results were obtained: 1. No cell loss was observed in the cell lysis step of the isolation procedure. Thus the mean percentage of nuclei found after 24 h of treatment with the stain-detergent solution was 104% (SD 28%) of the mechanically isolated cells. 2. The mean percentage of cells found after ethanol fixation and the staining and FCM analysis that followed was 45% (SD 18%) of the mechanically isolated cells. 3. Histograms from the unfixed cells were almost identical with histograms from the fixed cells, although the coefficient of variation for unfixed samples was 17% lower (average CVs of 3.3% vs. 4.0% for diploid peaks, and 4.3% vs. 5.1% for hyperdiploid peaks). In eight tumors a distinct hyperdiploid cell population was found, together with a diploid one. The hyperdiploid cells constituted in unfixed and in fixed samples on the average 48% vs. 42% of the total cell number (p ( 0.005), indicating a slight but consequent loss of hyperdiploid cells during fixation. In one case, the fixed sample showed a small, slightly hyperdiploid peak. Such a peak was not seen in the unfixed sample. This was probably an artefact due to bad preservation of the fixed cells as indicated by a broadening of the hyperdiploid peak. The latter was distinct and narrow in the unfixed sample. 4. In this experiment, the average cell or nuclei gain was 0.9 x 10 cells per gram of tumor tissue (range x 10 ) including the nuclei isolated from the tissue fragments. Recovery of Parenchymal Nuclei from Mouse Liver The reasons for performing this experiment and for choosing liver tissue are given in the discussion. The mean diameter of the parenchymal nuclei was caculated to be 7.98 pm (range ). From an area density of nuclear profiles of X 105/cm2, a specific gravity of 1.08, and no shrinkage of the liver tissue during processing, the mean number of parenchymal nuclei per gram was calculated to be 1.47 x lo8 (range x lo8). Following the detergent treatment of the liver tissue, the mean number of isolated nuclei per gram of liver was 0.86 x 10 (range x lo8). In the hemocytometer the percentage of nonspherical nuclei varied from 1 to 5%. Histological examination of the residual tissue showed a predominance of stromal tissue. This indicates that the isolated nuclei almost exclusively NUCLEAR ISOLATION BY DETERGENT 455 originated from parenchymal cells, and on the average consituted 59% of these. Selectivity in Nuclei Isolation The possible selectivity of the isolation procedure for the nuclei was investigated by microscopy of the residual tissue from the eight experiments with mouse colon tissue previously described and from the preparation of ten human colorectal carcinomas and seven adenomas Crable 2). A high selectivity for isolation of nuclei from epithelial cells and tumor cells was seen, as demonstrated in Figures 2-4. In the experiments with mouse colon tissue a blind comparison was performed of the percentage of partly or totally empty crypts as affected by the various detergent concentrations. With no detergent in the citrate solution only about 50% of the crypts were affected, whereas the percentages for 0.03% and 0.1% NP-40 were on the average 88% and 83%, respectively. For the 17 tumors listed in Table 2, only a semiquantitative estimate of the number of empty crypts could be performed due to the irregular histologic architecture of the tumors. The examination indicated, however, that in most tumors the majority of the tumor cell islands were more or less empty when not centered in larger lumps of fibrous tissue. This selectivity for isolation of tumor nuclei from the surrounding stroma was also indicated by FCM histograms of hyperdiploid tumors, which frequently showed a high percentage of tumor cells in spite of abundant stroma in the biopsies (Fig. 5). Reproducibility of the Sampling and Dispersal Procedures This was investigated by taking two fine-needle biopsies or one fine-needle and one solid tumor biopsy from 15 distinct locations in eight colorectal tumors. A good correspondence between the first and the second biopsy was found, as indicated by a significant correlation (r =.64, p =.01) between the percentage of hyperdiploid cells in the first and the second biopsy. Factors Affecting the Stability of Isolated Nuclei Time. The effects of storage up to 24 h in the staindetergent solution at 4 C have already been described. In 13 biopsies from six tumors, storage was extended to 48 h resulting in an average cell loss of 33%, but little change in resolution as indicated by an average increase of only 3% in the coefficent of variation. No cell population disappeared or was markedly reduced. RNase treatment. Naked nuclei may be sensitive to proteolytic activity. Enzyme preparations are not always pure and may contain some residual proteolytic activity. This was probably the case with an alternative RNase product used (RNase III A, Sigma Chemical Company, Cat. No. R-5125, specified as essentially proteasefree). Whereas the preparation of solid tumors was unaffected, human peripheral lymphocytes, used as a diploid standard in each analysis series, were markedly

5 456 PETERSEN FIG. 2. A) Mouse colon shown before detergent treatment, but after mincing and washing in TRIS buffer. Original magnification x 50. Hematoxylineosin (HE) staining. B) The tissue remnant collected from the 50-pm filter after 24 h of detergent treatment and staining. Most crypts appear without epithelial cells. The stroma architecture appears preserved and lymph follicles still contain lymphocytes. Original magnification x 50; HE staining. C) A segment of B demonstrates that most of the stroma nuclei are retained. Original magnification x 125. Table 2 Microscopic Znvestigation of Residual Tumor Tissue Percentage of "emptied" crypts < 25% 25-75% > 75% Total Adenomas Carcinomas affected, with about 90% cell loss after 24 h, increased background, and broadened peaks. To investigate this problem, further experiments were performed. The deleterious effect of the Sigma RNase was found in the half standard concentration (0.5 mgl dl), but not in 0.1 mgldl, whereas the usual BDH RNase had no such effect in five times the standard concentration. The deleterious effect of the Sigma RNase was completely prevented by addition of 0.2% human serum to the lymphocytes. During 4 y of using the BDH RNase from various lots a similar problem has never been seen. Ethanol fixation. Isolated nuclei from 22 biopsies from ten tumors were analyzed unfixed and after fixation and storage in ethanol for periods ranging from 1 d to 3 mo. Aggregation of nuclei was frequently observed after fixation, resulting in cell loss in the 50-pm filter. This aggregation was minimized by vigorous shaking of the sample after restaining, and could be avoided by ultrasound. In the resulting histograms the average change (increase or decrease) in fluorescence of the hyperdiploid cells relative to the near-diploid cells was 5% with a maximum of 10%. In all samples a small but significant relative loss of near-diploid nuclei was seen. No significant change was seen in the coefficient of variation of the peaks. DISCUSSION These experiments represent various approaches to one of the central problems in FCM of solid tumors and tissues: Which cells are actually analyzed, and which are lost? Or, in other words: How representative is the sample introduced into the flow cytometer? Few papers deal with these problems, and a simple figure giving the number of cells obtained per gram of tissue is only rarely reported for the various isolation procedures (4,7,12,14,15,19). It is of interest to know whether one technique yields significantly more cells than another.

6 NUCLEAR ISOLATION BY DETERGENT 457 FIG. 3. A tubulary adenoma (A) before and (B) after staining and detergent treatment for 20 h. The stroma appears well preserved, and the tubuli are without epithelial lining apart from an area below the center. Original magnification x 80; HE staining. This was one reason for comparing the effect of various detergent concentrations, which have been used by other investigators (5,19,20). Within this range, the detergent concentration seemed to be of limited importance, although the lowest concentration gave the fastest and the highest final nuclear yield. There is no obvious explanation for this, since no loss of already isolated nuclei was found with differing concentrations. The stability of already isolated nuclei when stored in the stain-detergent solution was found for the colorectal tumors even after 48 h; when some nuclear loss appeared, this did not seem to be selective in respect of cell ploidy or cellcycle phase. These experiments, however, have not ruled out that some ploidy classes or other cell populations could be selectively destroyed by the cell lysis step itself and therefore would not appear in the FCM analysis at all, even after very short staining periods. The experiment comparing only mechanically isolated and alcohol-fixed tumor cells with the same cells after hypotonic detergent treatment showed that this treatment did not give any cell loss. The histograms of isolated nuclei were almost identical with the histograms of the fixed cells, although of a better quality. Still, the initial mechanical isolation step may give some cell loss, but hardly especially selective. The last experiment in these quantitative aspects of the nuclei isolation procedure dealt with the recovery of the epithelial cells in the tissue. We knew that the number of isolated nuclei from mouse colon and liver was in the range of 1-2 x 10' per gram and for colorectal tumors x 10' per gram. However, little information on the nuclear density in normal tissues was available for comparison. For human tumors a frequently mentioned figure, lo9 cells per gram, seems to be just a rough estimate from average cell size. If this figure were representative, it would indicate a cell recovery of only about lo%, which would be unsatisfactory. Techniques for evaluating cell density in heteroge- neous tissues without using an isolation procedurewith possible cell loss-may be very sophisticated and time-consuming. The stereological technique performed

7 458 PETERSEN FIG. 4. An infiltrating adenocarcinoma (A) before and (B) after staining and detergent treatment for 22 h. No visible carcinoma cells are left in the preserved stroma. Original magnification x 40; HE staining. on the very homogeneous mouse liver tissue with almost spherical nuclei appeared to be sufficiently simple and reliable. The figures found are in accordance with information in the literature (1,19) and indicate a satisfactory recovery above 50%. It is hard to believe that the nuclear density in human colorectal tumors, with frequent necrosis and fibrotic areas, is substantially higher than in mouse liver. The average yield of 0.9 x lo8 nuclei per gram of tumor tissue is about twice as high as figures obtained by enzymatic tumor digestion of transplantable mouse tumors (14,15), and orders of magnitude higher than the nuclear yield reported for a technique using tissue homogenization (12). A trypsin technique on human mammary tumors yielded about 30 times fewer cells and a simple mechanical disaggregation about six times fewer (7). The experiments presented here give evidence, although circumstantial, that the hypotonic detergent isolation technique gives a reasonably quantitative, stable and reproducible recovery of epithelial and tumor nuclei from the tissues investigated. In spite of the above stated, the microscopic investigation of the residual tissue after nulear isolation for 24 h showed a marked selectivity for cells of epithelial nature. As shown in Table 2, not all epithelial nuclei were actually dispersed; but from Figures 2-4 it is clear that most of the stroma nuclei were retained. For the mouse colon even the lymphocytes in the submucosal lymph follicles seemed to be retained. If a more thorough mechanical mincing had been performed, e.g., by the use of tissue homogenizator (12), it is possible that more of the stromal nuclei would be released, but not without greater loss of epithelial nuclei. For analysis of tumor DNA ploidy, this selectivity seems to be a major advantage? making it possible to analyze tumors with a dominating stroma component. This selectivity should be borne in mind when comparing other methods of dispersal, especially enzymatic. The preference for epithelial cells, using the same detergent solution on whole mouse bladder, has recently been described (2), and it was also noted in our laboratory in cases of cervical dysplasia (11).

8 NUCLEAR ISOLATION BY DETERGENT A u z a I V \ v) 4 4 Lu v 1000 LL 0 s u 3 z CHANNEL NUMBER v) RELATIVE FLUORESCENCE FIG. 5. The histogram and full section through the corresponding half of the biopsy from an adenocarcinoma with large amounts of stroma. The first peak from the left constitutes 16% of the cells and represents diploid or near-diploid cells, as determined from a separately run diploid standard. The large peak represents a near-triploid tumor cell population and the following peak a near-tetraploid population. All five biopsies from the tumor had a similar histologic pattern, and all histograms were dominated by hyperdiploid cells. Original magnification X 25; HE staining. Under the standard conditions the isolated nuclei appeared remarkably stable, but obviously they are sensitive to various deleterious factors, including enzymes. The experience with another RNase preparation showed that washed lymphocytes were sensitive to a factor that probably had proteolytic activity, since it was totally blocked by small amounts of serum. The effect was not observed on solid tissue samples, which probably contain sufficient antiproteolytic activity to avoid such effect. It has been the only event of this kind in our laboratory, but its occurrence stressed the importance of having a standard cell suspension for preparation with each analysis series. For this, washed lymphocytes, which are stored frozen (Petersen, manuscript in preparation) seem well suited. The isolated nuclei may be stabilized by ethanol fixation, and stored for long periods, which makes the tech- nique more versatile, especially when applied to clinical samples. Finally, it should be pointed out that the present investigation only deals with tissues and tumors that are not affected by any sort of antineoplastic treatment. If cells are more or less damaged by such treatment prior to the isolation of the nuclei, their stability may be affected, giving an unknown loss. Indication of such effects have been observed in our laboratory after hyperthermic treatment of cell cultures. The risk of a possible increased cell loss of partly damaged cells should, however, be considered in any isolation procedure, and may especially complicate the evaluation of cell kinetic parameters from DNA distributions. In conclusion, the present method seems to be a simple and reliable technique for preparation of nuclei from colorectal tissue and tumors for FCM. Furthermore, it

9 460 PETERSEN appears to have the advantage of a certain selectivity for isolation of the neoplastic nuclei from solid tumors making the analysis feasible also for tumors with a large stroma component. ACKNOWLEDGMENTS The author wants to thank Anne Lise Madsen for excellent technical assistance. We thank the Department of Pathology, Aarhus County Hospital for help in preparing the photomicrographs and the Electron Microscopy Laboratory for Diabetes Research, Aarhus University for help to the stereological estimation of nuclear density in mouse liver. LITERATURE CITED 1. Adler CP, Ringlage WP, Bohm N: DNA content and cell number in heart and liver of children. Pathol Fks Pract 172:2540, Ayres PH, Schol HM, Hudson JL: A rapid method for preparation of urinary bladder epithelium for flow cytometric analysis. J Urol 131: , Baisch H, Beck HP: Comparison of cell kinetic parameters obtained by flow cytometry and autoradiography. In: Biomathematics and Cell Kinetics, Valleron A-J, MacDonald PDM (eds): ElsevierNorth-Holland Biomedical Press, Amsterdam, 1978, pp Beck HP: Is the composition of the cell suspension measured by flow cytometry representative for the composition of the tumor? Abstract Combined IX International Conference on Analytical Cytology and Cytometry and the VIth International Symposium on Flow Cytometry, Elmau, West Germany, 1982, p Bichel P, Frederiksen P, Kjaer T, Thornmesen P, Vindelav LL: Flow microfluorometry and transrectal fine-needle biopsy in classification of human prostatic cancer. Cancer , Brattain MG Tissue disaggregation. In: Flow Cytometry and Sorting, Melamed MR, Mullaney PF, Mendelsohn ML (eds): John Wiley & Sons, New York, 1979, pp Chassevent A, Daver A, Bertrand G, Coic H, Geslin J, Bidabe M- C1, George P, Larra F: Comparative flow DNA analysis of different cell suspensions in breast carcinoma. Cytometry , DeHoff RT: Measurement of number and average size in volume. In: Quantitative Microscopy, DeHoff RT, Rhines FN (eds): Mc Graw- Hill, New York, 1968, pp Hoshino T, Gray JW, Nomura K: Flow cytometry of isolated nuclei prepared from 9L rat brain tumor. Lab Invest 41:72-76, Fried J, Perez AG, Clarkson BD: Rapid hypotonic method for flow cytofluorometry of monolayer cell cultures: Some pitfalls in staining and DNA analysis. J Histochem Cytochem 26: , Jakobsen A, Kristensen PB, Poulsen HK: Flow cytometric classification of biopsy specimens from cervical intraepithelial neoplasia. Cytometry 4: , Koss LG, Wolley RC, Schreiber K, Mendechi J Flow-microfluorometric analysis of nuclei isolated from various normal and malignant human epithelial tissues. J Histochem Cytochem 25: , Krishan A Rapid flow cytofluorometric analysis of mammalian cell cycle by propidium iodide staining. J Cell Biol 66: , Noel JS, Zucker RM, Wu N-C, Demaray SY: The dissociation of transplantable tumors. J Histochem Cytochem 25: , Pallavicini MG, Falstad LJ, Dunbar C: Solid KHT tumor dispersal for flow cytometric cell kinetic analysis. Cytometry , Petersen SE: Setting up and running a microscope-based flow cytometer. Cytometry 3: , Rose PE: Improved tables for the evaluation of sphere size distributions including the effect of section thickness. J Microsc 118: , Taylor Iw: A rapid single step staining technique for DNA analysis by flow microfluorometry. J Histochem Cytochem 28: , Thornthwaite JR, Sugarbaker EV, Temple WJ Preparation of tissues for DNA flow cytometric analysis. Cytometry 1: , Vindelav LL: Flow microfluorometric analysis of nuclear DNA in cells from solid tumors and cell suspensions. Virchows Arch B Cell Pathol24: , Vindelav LL, Christensen IJ, Nissen NI: A detergent-trypsin method for the preparation of nuclei for flow cytometric DNA analysis. Cytometry 3: , Zante J, Schumann J, Barlogie B, mhde W, Buchner T: New preparation and staining procedures for specific and rapid analysis of DNA distributions. In: Pulse-Cytophotometry 11, Ghde W, Schumann J, Buchner T (eds): European Press, Gent, Belgium, 1976, pp