Path. Res. Pract. 181, 200-205 (1986)

DNA Flow Cytometry of Cells Obtained from Old Paraffin­Embedded Specimens. A Comparison with Results of Scanning Absorption Cytometry (A Methodological Study)

Introduction

Sophie D. Fossa, Erik Thorud Department of Medical Oncology and Radiotherapy, The Norwegian Radium Hospital, Montebello, Oslo 3

Mohammed C. Shoaib, Erik O. Pettersen Department of Tissue Culture, Norsk Hydro's Institute for Cancer Research, The Norwegian Radium Hospital, Montebello, Oslo 3, Norway

SUMMARY

Suspensions of single cell nuclei were obtained from 31 different samples of 7-9 years old paraffin-embedded bladder cancer biopsies. The DNA content of the ethidium-bro­mide-stained nuclei was analyzed by flow cytometry (FCM). The tumour stemline ploidy, as determined by FCM, was compared with that obtained by Feulgen scanning absorption cytometry (SACM) in imprints obtained from the same biopsy before it was fixed. Fifteen tumours that were diploid by FCM were also diploid by SA CM. All of the 16 tumours with non-diploid DNA-stemlines by FCM were non-diploid by SACM, though minor differen­ces between the ploidy values were occasionally seen when the results of the two methods were compared.

The background activity due to cell debris was considerable and resulted in a mean variation coefficient (CV) of 5.1% for human spleen cells fixed and embedded before preparation for FCM in the same way as the tumour samples. In the tumour samples there was a large and unpredictable variation of the ratio between the DNA values of chicken erythrocytes (internal standard) and diploid cells. In some specimens this ratio would have resulted in incorrect evaluation of the FCM DNA histogram. The following method for evaluation of FCM histograms is therefore proposed: Single peak histograms obtained from paraffin embedded tissue should always be interpreted as representative for a diploid cell population. In FCM histograms from paraffin-embedded tissue with more than one peak, the first peak should be considered as representing diploid cells.

Previous reports2 , 3, 5, 7, 9, 11, 14, 16 have indicated that DNA flow cytometry (DNA FCM) may have significance for diagnosis, prognosis, treatment and follow-up in

patients with various malignancies. However, final proof of the usefulness of this method is still lacking, mainly due to a too short follow-up of the patients. This problem could probably be quickly solved if DNA FCM could be reliably performed in single cell suspensions obtained from

0344-0338/86/0181-0200$3.50/0 © 1986 by Gustav Fischer Verlag, Stuttgart

old paraffin-embedded tumor specimens. This would per­mit correlation of the results of DNA FCM with the known clinical course of the patients.

Recently Hedley et al. 8, Friedlander et aU and Schutte et al. 12 reported that DNA FCM from wax-embedded tissue was possible. These authors used wax-embedded material for making suspensions of nuclei suitable for FCM. The results were compared with those from FCM in single cell suspensions obtained from the tissue before formalin fixa­tion.

Their method has been adopted in our laboratory with slight modifications. The aim of the present work was to study the use and reliability of the results of DNA FCM when performed on cells from old paraffin-embedded tis­sue from human bladder cancer and to investigate the methodological problems encountered with this technique The DNA FCM histograms from cell suspensions obtained from paraffin blocks were compared with DNA histo­grams from scanning absorption cytometry (SACM) done on Feulgen-stained imprints before formalin fixation and embedding obtained from the same biopsies.

Material and Methods

Dewaxing and Rehydration. Sections, lOO!lm thick (1 or 2 pieces depending on the amount of tumour material) were cut from each of 33 paraffin-embedded blocks of tissue from bladder carcinoma. The sections were immersed in xylol for 3 hours to remove the paraffin. The tissue was then rehydrated by 96%, 70%, 30% and 15% alcohol and finally with distilled water (10 minutes for each immersion).

Mechanical Detachment. The sections were mechanically dis­rupted in phosphate buffered saline (PBS) and incubated in a shaker (20U rpm) at 37°C for 90 minutes.

Enzymatic Dispersion and Digestion. The whole tissue mate­rial including both small and large clumps was incubated with 0.4% freshly prepared pepsin (Sigma, pH 1.7) at 37 °C for 1 hour in a shaker placed in a water bath. Tissue clumps were removed by filtration through a 60 !lm nylon mesh. The single cell nuclei suspensions were centrifuged (1000 rpm) for 5 mi­nutes.

Staining with Ethidium Bromide. The nuclei were then reincu­bated with freshly prepared 0.1 % RNAse in PBS at 37°C for 1 hour. Smears were prepared by the use of a cytocentrifuge (40U !ll suspension, 400 rpm for 5 minutes) and were stained with haematoxylinleosin for light microscopy (Fig. 1). The rest of the suspensions were centrifuged and pellets were resuspended in ethidium bromide (10 !lg/ml in Tris buffer). Nuclei from chicken erythrocytes which had been fixed in 4% formalin were washed in PBS and added to each sample as an internal standard prior to RNAse treatment. The chicken erythrocytes represented about 10% of the nuclei in the final suspension of 5 mt. The stained nuclei suspensions were kept at 4 °C in darkness overnight.

The flow cytometric measurements were performed as described previously6. Briefly, the flow-cytometer is a laboratory­built high resolution instrument based on a Leitz invertoscope having a Hg-Iamp and fluorescence optics 10, 13. The multichannel analyser (Nuclear Data 66) is linked on line with a central com-

Flow Cytometry in Paraffin-embedded Tissue . 201

puter (Norsk Data 100) programmed to analyse both ploidy and the fractions of cells in the various cell-cycle phases.

Evaluation of the DNA histograms: Initially the ploidy of the various peaks in the tumour FCM histograms was determined by comparison with the site of the peak of diploid (2c) human spleen cells, obtained from a 10 years old paraffin block. These suspen­sions of spleen cells were made as described for the tumour cell suspensions. In each batch 4-8 bladder cancer specimens and one sample of spleen cells were stained. However, using this techni­que we found that 4 of our tumour FCM histograms which had only one peak would have been interpreted as being (hypo-)­tetraploid in the absence of any diploid cell population. These tumours were characterized by SACM as being diploid with only one DNA stemline. All previous FCM experience with human tumours, however, indicates that a diploid peak can always be distinguished, either as a single peak or together with non-diploid cell populations. We always find a diploid peak as the first distin­guishable cell population on the FCM histogram. It was therefore decided that a single peak FCM histogram should always be interpreted as representing a diploid tumour, irrespective of the mathematical ploidy definition based on the human spleen cells from the same batch. Furthermore, in the presence of several peaks, a clearly defined peak having the lowest ploidy in a FCM histogram was considered to be diploid. Table 1 indicates the considerable inter-specimen variation of the ratio between the channel numbers for the internal standard and the lowest ploidy (or single) peak of a FCM histogram presumably comprising dip­ploid cells. The Table thus demonstrates the uselessnes of the internal standard for the definition of diploidy in the different specimens.

In order to make possible a comparison with our previously reported results from SACM5, the ploidy ranges in the present FCM series were defined as 2 c ± 25% and 4 c ± 25% respec­tively for diploid and tetraploid nuclei. Using SACM, tumours having at least 40% cells in the tetraploid range were character­ized as tetraploid. In FCM histograms a tetraploid stemline was judged to be present if a recognizable number of cells were observed in the region between 4 c and 8 c together with a clear peak at 8 C. Otherwise the tetraploid cells detected by FCM were assumed to represent G2 cells belonging to the diploid stemline. No attempt was made to quantitate the fractions of non-GoIG1 cells. If more than 1 non-diploid DNA stemline were found, the highest peak values of the respective methods were compared with one other.

Table 1. Channel number and ratio between chicken erythro­cytes (internal standard) and human diploid cells (= first or single peak of a FCM histogram) from one preparation and staining batch

Channel number Tumour Chicken erythro- Human spleen Ratio specimen number cytes cells

1 20 63 0.32 2 18 64 0.28 3 19 57 0.33 4 19 67 0.28 5 16 64 0.25 6 15 57 0.26 7 16 75 0.21 Spleen cells 16 51 0.31

Mean 17.38 62.25 0.28

202 . S. D. Fossa, E. Thorud, M. C. Shoaib and E. O. Pettersen

Scanning absorption cytometry. All scanning absorption cytometry was done during the years 1974 to 1976, without knowledge to the results of FCM obtained subsequently. Fresh bladder biopsies had been used to make imprints for Feulgen DNA SACM before fixation. The details of preparation, staining and measurement have been described previousll. Briefly, imprint smears were made from fresh biopsies obtained by TUR of bladder tumours. The imprint smears were fixed in a mixture of methanol 85%, formalin 10% and glacial acetic acid 5% after hydrolysis (4 n HCl28 °C, 100 min.). Feulgen-staining was done for 60 minutes with a Schiff reagent prepared with Pararosaniline Base "Merck" No 7601. The slides were washed 6 times in S02 water for 5 minutes each time and mounted in Cargille oil.

One hundred randomly chosen epithelial nuclei from each biopsy were measured at the wavelenghth 570 nm with the

SMP 05 (Zeiss Oberkochen, W-Germany) connected with Wang Desk Computer 720 B (objective: Plan: 100/1.25; LD Plan: 401 0.6; measuring field: 0.7 f.tm distance between centres of the measuring fields: 0.5 f.tm. The Feulgen DNA value of each nuc­leus was expressed in arbitrary units, using 30 granulocytes and large lymphocytes from the same imprint smears as internal stan­dard4

•

The DNA stemline value was calculated for each biopsy. Each cell population was characterized as either diploid (2 c ± 25%), tetraploid (4c ± 25%) or octoploid (8c ± 25%). Triploid or hexaploid DNA stemlines had values between these ranges.

For each specimen the tumour stemline ploidy values obtained by DNA FCM were compared with those from SACM. Regres­sion analysis was used for statistical analysis.

A

, , .'­•. ' • • . -

• , •

\,

• ( \

B

Fig. 1. Cell nuclei in human bladder car­cinoma. A; Imprint from fresh biopsy tis­sue (SACM). B: Cell suspension obtained from 9 years old paraffin-embedded tissue (FCM).

c 0 'iii .;; 1

" '-I! • 0

. , " 0 " ~ :'

Qj " " c , ,

c .' III I~ .c , , u , , '- I, I! , , ~

I , I ,

Qj ~ 1 u , • ..... , 0 • '-

" ,

QJ

~ • h-'. ~ ... :z

0 20 40 60 80 100 120

Channel number

Fig. 2. DNA histogram (FCM) in human spleen cell nuclei obtained from a 10 year old paraffin block. Peak 1: Internal standard. Peak 2: Human spleen cell nuclei.

] 600 c III .c u

'­QJ

a. 400 '" Qj u

..... o a; 200 • I' "

"

f, " , I ~ . , I ., , I

• , , , I , , ~

I I I

• , , , I , • I

I I I -(FCM)

A

..Cl E ::J :z . -t-""'; ..

.... -.

100

~ 80

QJ 60 u ::J C

0 40 t-'-QJ

..Cl 20 E ::J :z

0 0

20 40 60 80 Channel numbPr

I

(SACM)

78 % ,..--

20%

I

4c

Ploidy

100 120

B

I

Fig. 3. DNA histograms in a diploid human bladder cancer. A: FCM using a 9 years old paraffin block. Peak 1: Internal stan­dard. Peak 2: Human diploid cell nuclei. Peak 3: Human cancer cell nuclei. B: SACM in imprints (100 nuclei).

Flow Cytometry in Paraffin-embedded Tissue . 203

Results

Two tumour tissue samples yielded an insufficient number of cells for FCM analysis and these suspensions had to be excluded from the final analysis. The mean value (± standard deviation) of the number of cell nuclei obtained for FCM from each of the 31 remaining paraffin blocks, was 18000 (± 11700) (range 2100-46900). Fig. 1 shows the nuclear morphology in the suspensions pre­pared from paraffin-embedded material compared with Feulgen-stained imprints.

The mean variation coefficient for DNA FCM in human spleen cell suspensions was 5.1% (range 2.6-7.9%) (6 different batches) (Fig. 2) .

Figs. 3 and 4 show examples of FCM and SACM histo­grams obtained from a diploid (Fig. 3) or a tetraploid (Fig. 4) bladder cancer. There was good correlation between the tumour ploidy determined by the two methods.

Using the interpretation of the FCM histograms described above, 15 tumours were characterized as diploid

QJ C C III

-5 ~oo '­QJ a.

'0 200 '­QJ

..Cl

E ::J :z

o • o

100 I-

~ ~

80

QJ 60 u

::J C

'5 40 '-QJ

..Cl E ::J 20 :z

0

I

20

• I' .-~\ , \ ; Ii I ' , ' , . . " , , . I

(FCM)

• !, ~, , .. , ' . , , . ~ , ~ \ , ,

~

! I ~ \ . , ...J ' . I

40 60 80 Channel number

(SACM)

67%

5°;' ~

2c 4c

Ploidy

-A

100 120

B -

27% -,...-

1% I'

8e

Fig. 4. DNA histograms in a tetraploid human bladder car­cinoma. A: FCM using an 8 year old block. Peak 1: Internal standard. Peak 2: Human cancer cell nuclei. B: SACM in imprints (100 nuclei).

204 . S. D. Fossa, E. Thorud, M. C. Shoaib and E. O. Pettersen

BC

•

6C

•

u 4C LL

2C cases)

OL-______ ~ ______ _L ______ _L ______ ~~

o 2C 4C 6C BC

SACM

Fig. 5. DNA stemline ploidy values determined by FCM and SACM. Using the described modification for evaluation of FCM: R:l 0.984; p < 0.05.

by FCM, 13 as tetraploid, 2 as hexaploid and 1 as octo­ploid. In all diploid and all but one non-diploid tumours the same DNA stemline was also identified by SACM. One tetraploid tumour (by FCM) was hexaploid by SACM. If the discrimination had been done between diploid and non-diploid tumours, all diploid and non-diploid tumours would have been identified similarly by both methods.

When the DNA stemline values from FCM were com­pared with those from SACM, a significant correlation was found (R = 0.984, p < 0.05, Fig. 5).

Discussion

The variation coefficient for human spleen cells was higher than that usually observed in FCM histograms from fresh tissue. A high background activity was found due to debris. Uncontrolled variation in the condition of the original tissue blocks may represent an additional reason for the high variation coefficient and the consider­able background activity. This also caused difficulties in the quantitation of the cell cycle phase fractions in the individual tumour suspensions. In our experience this background activity could not be compensated for by the methods generally used for correction of the background level in FCM histograms. Based on our preliminary experi­ence we therefore conclude that cell suspensions, from paraffin blocks, so far, can only serve to determine tumour ploidy with an acceptable degree of certainty and not for quantitation of the cell cycle phase fractions.

During the preparation of cell suspensions from paraf­fin-embedded material, cell destruction and production of

cell debris are common features. A considerable percen­tage of tissue samples do not yield a cell suspension suit­able for FCM. In the present study, 2 of 33 specimens could not be analyzed. Thereby an uncontrollable selec­tion favouring analysable tumour cell suspensions cannot always be avoided.

As was also pointed out by Hedley et al. 8, one of the main problems when performing DNA FCM in paraffin­embedded tissue, was the determination of the diploid reference. Uncontrolled variations in fixation and hand­ling of the tissue prior to block preparation are thought to be the main sources for these variations. As a result the stainability of suspended cells varied from sample to sam­ple causing large variations of the channel number for the diploid cells. For FCM histograms from fresh material human spleen cells are generally used to define the diploid region, though human lymphocytes may result in falsely low diploid values15. However, if paraffin-embedded spleen lymphocytes are used as standard cells as in the present study they actually represent an external standard, which is inevitably handled differently from the tumour cells, e. g. with respect to duration of formalin fixation prior to paraffin embedding. Our experience in the present study (Table 1) is that the use of such external standard cannot be recommended, even if chicken erythrocytes are used additionally as a kind of internal standard. Uncon­trollable differences between the stainability of the spleen cells and the diploid cells from the tumour samples were frequently observed. These variations could only be over­come by use of the described modification assuming that a diploid peak always is recognizable in human tumour cell suspensions and represents the lowest ploidy or the only peak of the FCM histogram. Admittedly, FCM histograms with a hypodiploid stemline (alone or in combination with hypopolyploid stemlines) which appear occasionally in special tumour types would not be interpreted correctly using the described method.

Recently Hedley et aU and Schutte et al. 12 came to the same conclusions: No external standard can be used to define the diploid value if FCM is performed in cells from paraffin-embedded tissue. Diploid normal cells and/or diploid tumour cells from the same paraffin section are the only cells which can serve to establish the diploid refer­ence.

When using the above interpretation of the FCM histo­grams - the first recognizable peak being the diploid one -the correlation was excellent between the DNA stemline values obtained by FCM and SACM. However, small dif­ferences could not be avoided in individual cases. The pre­sent correlation coefficient was of the same order as that obtained by Hedley et al.8 and Schutte et alY when com­paring the results of FCM from fresh material with that from paraffin-embedded tissue.

Although refinement of the technique is necessary, elipe­c.ally ifthe cell cycle phas<i fractions are to be estimated, we find the overall agreement between stemline ploidy determinations by FCM and SACM results to be promis­ing. DNA FCM in single cell suspension from old paraffin blocks opens a possible new method of investigating the clinical significance of ploidy levels in human cancer.

Conclusions

1. DNA FCM for determination of the stemline ploidy can be reliably performed in cell suspensions obtained from paraffin embedded tumour tissue although internal standardisation of the samples is at present not possible. 2. Essentially there is no need for any internal (chicken erythrocytes) or external (human spleen cells) standard if the following assumptions are made: - A single peak in a FCM histogram from human tumours

always represents a diploid population. - If more than one peak is recognized the first one repre­

sents a the diploid (sub-)-population.

Acknowledgement: The study was financially supported by The Norwegian Cancer Society

References

1 Barlogie B, Raber M, Schumann J, Johnson T, Drewinko B, Swartzendruber D, Gohde W, Andreef M, Freireich E (1983) Flow Cytometry in Clinical Research. Cancer Res 43: 3982-3997

2 Bichel P, Frederiksen P, Jkrer T, Thommesen P, Vindel0v L (1977) Flow Microfluorometry and Transrectal Fine-Needle Biopsy in the Classification of Human Prostatic Carcinoma. Cancer 40: 1206-1211

3 Buchner T, Hiddemann B, Wormann B, Kleinemeier B, Schumann J, Gohde W, Ritter J, Muller K, Bassewitz von D, Roessner A, Grundmann E (1985) Differential Pattern of DNA­Aneuploidy in Human Malignancies. Path Res Pract 179: 310-317

4 Fossa SD, (1975) Feulgen-DNA-Values in Transitional Cell Carcinoma of the Human Urinary Bladder. Feulgen-DNS-Werte in den Obergangsepithelien des menschlichen Harnblasen-kar­zinoms. Beitr Path 155: 44-55

Flow Cytometry in Paraffin-embedded Tissue . 205

5 Fossa SD, Kaalhus 0, Scott-Knudsen P (1977) The Clinical and Histopathological Significance of Feulgen DNA-values in Transitional Cell Carcinoma of the Human Urinary Bladder. Eur J Cancer 13: 1155-1162

6 Fossa SD, Thorud E, Shoaib C, Pettersen E, H0ie J, Scott­Knudsen 0 (1984) DNA Flow Cytometry in Primary Breast Cancer. Acta Pathol Scand 92: 475-480

7 Friedlander ML, Hedley DW, Taylor IW, Russel P, Coates AS, Tattersall MHN (1984) Influence of Cellular DNA Content on Survival in Advanced Ovarian Cancer. Cancer Res 44: 397-400

8 Hedley D, Friedlander M, Taylor I, Rugg C, Musgrove E (1983) Method for Analysis of Cellular DNA content of Paraffin­embedded Pathological Material Using Flow Cytometry. J His­tochem Cytochem 31: 1333-1335

9 Hedley DW, Friedlander ML, Taylor IW (1985) Application of DNA Flow Cytometry to Paraffin-Embedded Archival Mate­rial for the Study of Aneuploidy and Its Clinical Significance. Cytometry 6: 327-333

10 Lindmo T, Steen HB (1979) Characteristics of a simple high-resolution flow cytometer based on a flow configuration. Biophys J 28: 33-44

1 Rognum T, Thorud E, Elgjo K, Brandtzaeg P, 0rjasreter H, Nygaard K (1982) Large-Bowel Carcinomas with Different Ploidy Related to Secretory Component, IgA, and CEA in Epithelium and Plasma. Br J Cancer 45: 921-923

12 Schutte B, Reynders MMJ, Bosman FT, Blijham GH (1985) Flow Cytometric Determination of DNA Ploidy Level in Nuclei Isolated From Paraffin-Embedded Tissue. Cytometry 6: 26-30

13 Steen HB, Lindmo T (1979) Flow Cytometry: a High-Resol­ution Instrument for Everyone. Science 204: 403-404

14 Tribukait B, Gustafson H, Esposti P (1982) The Significance of Ploidy and Proliferation in the Clinical and Biological Evalua­tion of Bladder Tumours: a Study of 100 Untreated Cases. Br J Urol 54: 130-135

15 Wolley RC, Herz F, Koss LG (1982) Caution on the Use of Lymphocytes as Standards in the Flow Cytometric Analysis of Cultured Cells. Cytometry 2: 370-373

16 Wolley R, Schreiber K, Koss L, Karas M, Sherman A (1982) DNA Distribution in Human Colon Carcinomas and its Rela­tionship to Clinical Behavior. JNCI 69: 15-22

Received December 13, 1984 . Accepted in revised form September 9, 1985

Key words: DNA flow cytometry - Paraffin-embedded tissue - Scanning absorption cytometry

S. D. Fossa, General Department, The Norwegian Radium Hospital, Oslo 3, Norway