Application of the crypt-isolation technique to flow-cytometric analysis of DNA content in colorectal neoplasms

GASTROENTEROLOGY1994;106:100-107

Application of the Crypt-Isolation Technique to FiowCytometric Analysis of DNA Content in Colorectal Neoplasms SHIN-ICHI NAKAMURA,* JUNKO GOTO,* MASAHIKO KITAYAMA,§ and ISAMU KINO§ *Division of Pathology, Central Clinical Laboratory, School of Medicine, Iwate Medical University, Morioka, Japan; *Research Equipment Center and §First Department of Pathology, Hamamatsu University School of Medicine, Hamamatsu, Japan

Background~Aims: Conventional flow-cytometric analysis of colorectal tumor DNA content includes both tumor and stromal cell DNA. To improve the method, we evaluated thq DNA content of purified epithelial cells using the crypt-isolation technique. Methods: Normal and neoplastic crypts were isolated from fresh resected colorectum nonenzymaticaUy. Isolated crypts were digested into single nuclei suspensions and examined by flow cytometry (FCM). Internal controls from normal crypts were used to define diploidy, The neoplastic nuclei in the same individual and mixtures of both normal and neoplastic nuclei in various ratios were analyzed. Results: Tumors having both diploid and aneuploid stemlines were found more frequently than that reported u~singconventional FCM. Near-diploid DNA stemlines with DNA indices ranging between 0.90 and 1.10 were detectable. The cases of advanced cancer in which the DNA index was between 0.90 and 0.95 were associated significantly with Dukes' stage C. Conclusions: This is the first reported analysis of the DNA content of "pure" neoplastic colorectal epithelia using FCM. Multiploid and near-diploid stemlines were more accurately analyzed than that using conventional FCM.

ince the introduction by Hedley et al. of a method for analysis of D N A content in formalin-fixed, paraffinembedded material using flow cytometry (FCM), ~'2 a large number of retrospective studies has been conducted to clarify the relationship between the clinicopathological findings of colorectal cancer and D N A content) -7 Other methods, using fresh solid tumor material processed to single nuclei suspensions by digestion with enzymes or detergents for FCM are now widely used; the methods are rapid and simple with satisfactory low coefficient of variation (CV) values. 8-19 Using any of the above methods, the suspended nuclei obtained are composed of both neoplastic and stromal nuclei. The D N A histograms .using these preparations have peaks composed of stromal cell nuclei and are regarded as diploid stemlines. If neoplastic samples are composed of various D N A ploidies including a diploid

S

stemline, the diploid peak would contain both nonneoplastic and neoplastic cell nuclei. In such cases, diploid stemlines of the neoplasm cannot be identified. In such preparations, near-diploid tumor peaks can be lost in the diploid normal peak if the number of aneuploid cells is smaller than that of normal cells. The opposite can also occur and a near-diploid tumor population may be so large that in swamping the diploid normal peak, the tumor may be analyzed as diploid. By applying the crypt isolation technique, 2°-23 we were able to examine the D N A content of colorectal neoplasms and obtain FCM D N A histograms composed of pure neoplastic cells. Here we report on the efficiency and accuracy of the crypt isolation technique in FCM analysis.

M a t e r i a l s and M e t h o d s Clinical Material Between 1989 and 1992, tissue was taken from 114 patients with colorectal neoplasms undergoing resection of the large intestine at Hamamatsu University Hospital and related city hospitals. The patients consisted of 61 males and 53 females with a median age of 64 years (range, 22-88). Eightyeight patients had adenocarcinomas, 18 patients had tubular or villous adenomas, and 8 patients had colectomies for familial adenomatous polyposis (FAP). Fresh specimens were obtained immediately after surgical excision. Normal mucosa at least 5 cm from the neoplasm was separated from the submucosa with scissors. Flat normal mucosa between multiple adenomas in FAP cases was also sampled. Samples greater than 1 cm 3 were obtained from tubular or villous adenomas of the large intestine. Several pedunculated adenomas less than 1 cm in diameter were cut at their stalks in FAP cases. In centrally ulcerated sporadic carcinomas, samples were routinely obtained both from the area of central ulceration and from the peripheral everted edge. Abbreviations used in this paper: CMFH, calcium-and magnesiumfree Hanks' balanced salt solution; CV, coefficient of variation; DI, DNA index; FAP,familial adenomatous polyposls; FCM, flow cytometry; PI, propidium iodide. © 1994 by the American Gastroenterological Association 0016-5085/94/$3.00

January 1994

Crypt Isolation The method of Arai and Kino was used for crypt isolation. 22 Briefly, fresh mucosa/neoplasm was cut into 5-mm cube pieces with a razor, then incubated at 37°C for 30 minutes in calcium- and magnesium-free Hanks' balanced salt solution (CMFH) containing 30 mmol/L ethylene-diaminetetraacetic acid (EDTA). Following this, the tissue was stirred in CMFH. Crypts separated from the lamina propria mucosae or fibrous stroma in 30 to 40 minutes. The isolated crypts were promptly fixed in 70% ethanol and stored at 4°C.

Identification of Crypts Fixed crypts were observed under the dissecting microscope (SZ60, Olympus, Tokyo). Normal and neoplastic crypts showed characteristic three-dimensional features under the dissecting microscope. Histology of the crypts in paraffin-embedded, H & E-stained sections confirmed their origin. The hyperplastic crypts of transitional mucosa were usually intermingled with carcinomatous crypts in the samples taken from the tumor edge. 24'25Transitional crypts were algo separated carefully from cancer crypts and examined independently.

Flow Cytometry Normal, transitional, and neoplastic crypts of each patient were incubated with 0.0125% pepsin (pH 2.0, Sigma Chemical Co., St. Louis, MO) for 5 minutes at 37°C, washed twice with 0.2 mol/L Tris chloride-buffered saline, and syringed through a 27-gauge needle. Under a phase contrast microscope (Diaphot, Nikon, Tokyo), suspended nuclei were observed for shape and counted using a hemocytometer. All samples were prepared to a concentration of 1 × 106 nuclei/ mL of buffer. Both the normal and neoplastic nuclei were stained with propidium iodide (PI) (50 gtg/mL, Sigma Chemi-. cal Co.) coneaining ribonuclease (0.25 mg/mL, Sigma Chemical Co.) in 0.2 mol/L Tris chloride-buffered saline for 30 minutes in darkness at room temperature. After being filtered through 37-~tm nylon mesh (Tokyo Screen, Tokyo), the samples were analyzed on an EPICS PROFILE flow cytometer (Coulter, Hialeah, FL). An argon ion laser was used at 15 m W at a wavelength of 488 nm for excitation. Emission was measured using a 610 nm long-pass filter. Ten thousand nuclei were counted in each sample. After measurement, each sample (mixtures of normal and transitional and of normal and neoplastic nuclei) was stained with PI and analyzed. In this study, dispersed normal epithelial nuclei were added to the neoplastic samples as an extrinsic internal control. Also, appropriate amounts of normal crypts were mixed with neoplastic crypts as an intrinsic internal control in the same cases. The mixture was then digested'with pepsin, stained with PI, and analyzed. No difference of DNA index ~DI) was found between the histograms using the extrinsic and intrinsic controis.

Histogram Interpretation DNA histograms were analyzed using the software package Cytologic (Coulter Corporation). Conforming to the

CRYPT-ISOLATIONTECHNIQUEFOR FCM 101

suggestion of an international convention, 26 we used'normal epithelial cells of the same organ and the same individual as a DNA diploid standard. Samples that had at least one separate G0/G1 peak from a diploid peak were defined as D N A aneuploid. First, a sample from the normal mucosa and from the neoplasm of the patient were measured independently. Next, a mixture of normal and neoplastic tissue (proportion of the mixture; 1:1 to 1:3) was measured. If the three histograms of normal, neoplasm, and normal/tumor mixture all showed one G0/G1 peak, the neoplasm was shown to have a diploid stemline. If the histogram of the neoplasm had one peak and the mixture of normal and neoplasm showed two peaks, the neoplasm must have had an aneuploid peak. To identify which was the diploid peak, the ratio of normal/tumor nuclei was varied to be normal nuclei dominant and then analyzed. When histograms of pure neoplastic crypts showed more than two separate G0/G1 peaks; and the mixture with normal nuclei made one of the peaks higher; then that peak was diploid. These tumors consisted of a diploid and one or more aneuploid stemlines. DI was calculated as the ratio of mean channel number of the aneuploid G0/G1 peaks to that of the diploid one. Near-diploid stemlines were defined as aneuploid when the DI ranged from 0.9 to 1.1, excluding 1.0. The CV for all G0/G1 peaks found in normal and neoplastic samples was calculated as full-width CV. All samples in this study were measured more than twice, and the samples with the lowest CV were used for DNA analysis.

Histopathology All resected surgical materials were routinely examined on paraffin-embedded, H&E-stained sections and diagnosed histopathologically. The tissue fragments remaining after crypt isolation were fixed with 70% ethanol and examined histologically. These fragments showed some remaining crypts and enabled estimation of the character of the isolated crypts (i.e., normal, transitional, or neoplastic). In some cases, tissue fragments were used for preparing nuclei suspensions by a modified method of Hedley et al., la and analyzed with FCM for comparison with the results of the crypt isolation method. Seventy-six cases of advanced cancer (Dukes' stage B, 40 cases; and Dukes' stage C, 36 cases) were used to assess the relationship between DNA ploidy and Dukes' staging system.

Statistical Analysis Statistical analysis was performed with StatView-II software (Abacus Concepts, Berkeley, CA). Statistical differences between mean values in this study were analyzed using the g2 test.

Results Crypts of normal, adenoma, and edge of cancer were easily separated from lamina propria mucosae following incubation with E D T A . Some t u m o r specimens taken from the central part had a poor harvest o f crypts

102

NAKAMURA ET AL.

GASTROENTEROLOGY Vol. 106, No. 1

Rgure 1. Crypts of background normal mucosa. (A) Isolated normal crypts (original magnification x25). (B) Histology of A. There is no lamina propria mucosae'(H&E; original magnification x400).

because of the intense desmoplastic reaction created by the invading cells. Using the dissecting microscope, normal crypts showed a regular test t u b e - l i k e morphology and measured up to 1 mm in length. (Figure 1A). Transitional crypts were usually more than twice the length of normal crypts (Figure 2A). Isolated crypts from villous adenomas were large epithelial sheets up to 5 mm 2 with an occasional large irregular crypt (Figure 3A). The shape of the crypts in adenocarcinomas depended on differentiation; well-differentiated tumors had large crypts or sheets of epithelium (Figure 4A). In moderately differentiated adenocarcinomas, thick epithelial clusters of various size assumed zig-zag contours that possibly fitted that of the stroma from which they were derived (Figure 5A). Confirmation of the nature of the isolated crypts was performed on paraffin-embedded histological sections (Figures 1B-5B).

No multiple peaks were observed in normal crypt isolates. Crypts from transitional mucosa showed only diploid peaks. The adenomas less than 1 cm in diameter taken from patients with FAP with a tubular morphology and showing mild atypia were diploid. Two of 10 cases of sporadic tubular adenomas larger than 1 cm in diameter showed aneuploid stemlines. Histologically, these adenomas showed moderate atypia without focal cancer. The large villous adenomas in the sporadic cases showed mild to moderate atypia without focal carcinoma confirmed by both routine histological sections and in the tissue fragments used for crypt isolation. Three cases of tubulovillous or villous adenomas had diploid stemline. However, a further 5 cases were DNA aneuploid including one multiploid case. Two had near-diploid stemline (Figure 6; Table 1). There was no relationship between the histological grade of cancer and DNA ploidy (Figure 4C and 5C).

B

"

,

•

l=

Rgure 2. Crypts of transitional mucosa. (A) Crypts were larger and usually more than twice the length of normal crypts (original magnification x15). White arrow indicates normal crypts of the same patient. (B) Histology of transitional crypts (H&E; original magnification xlO0).

January 1994

CRYPT-ISOLATION TECHNIQUE FOR FCM

103

Figure 3. Crypts of villous adenoma. (A) Large epithelial sheets up to 5 mm 2 with occasional irregular crypts (original magnification x22). (B) Histology of isolated crypts (H&E; original magnification x l O 0 ) .

DNA aneuploidy was found in 62 (70%) of 88 samples taken from the everted edge of cancer with 30 (34%) of these showing multiploid (including diploid) DNA stemlines. Because of the previously described technical difficulty, only 44 samples from centrally ulcerated areas were analyzed. DNA aneuploidy was found in 35 samples (80%) with 13 cases (30%) showing multiploidy (Table

2). DNA ploidy was concordant between the peripheral edge and central ulceration in 21 (48%) of 44 cases. The mean CV of normal nuclei was .2.94 + 0.48 (range, 1.43-3.86) with that of neoplastic samples 3.48 4- 0.65 (range, 2.03-4.56). Near-diploid stemlines were detected as a small peak or a shoulder on the DNA histogram of mixed normal/

B

Figure 4. Crypts of well-differentiated adenocarcinoma. (A) Large crypts or sheets of cancerous epithelium (original magnification x15). White arrow indicates normal crypts of the same patient. (B) Large glandularlike structures were lined with atypical absorptive cells (H&E, original magnification x200). (C) Left histogram was normal crypts (CV = 2.29). In the middle histogram, the tumor has a main GO/G1 peak (CV = 3.38) with a small shoulder on the left. The right histogram is the mixture of normal and carcinoma nuclei. DI of the carcinoma was 1.20.

~c

4c

~c

4c

104

NAKAMURA ET AL.

GASTROENTEROLOGYVol. 106, No. 1

o

•

•

.

"-~.~

~

•

•

c.

~.~.

~,

~

~"

C

~:

4t

2c

4~

tumor crypt isolates. By changing the ratio of mixture two distinctive peaks could be obtained (Figure 7). Comparison of the crypt isolation technique with conventional technique, FCM showed that neoplasms with distinct aoeuploid peaks were shown by both methods (F'igure 8A). However, when the neoplasm had neardiploid stemline peak, the crypt isolation technique

Figure 5. Crypts of a case of moderately differentiated adenocarcinoma. (A) Thick epithelial clusters (original magnification ×25). (B) A histological section of a tissue fragment after the crypt isolation technique. (H&E, original magnification x40). (C) Left histogram of normal crypts (CV = 3.69). Middle histogram of carcinomatous crypts (CV = 2.91). Right is a mixture of normal and carcinoma. DI of this adenocarcinoma was 1.0.

showed two clear peaks whereas the conventional method disclosed only one peak (Figure 8B). The relationship between tumor ploidy status and Dukes' staging was examined in 76 cases of advanced cancer. All 7 cases with multiple stemlines consisting of a diploid and a hypodiploid stemline (Dis ranging from 0.91 to 0.95) were Dukes' stage C with lymph node involvement (Figure 9). Seven cases had near-diploid stemlines in which Dis were between 1.04 and 1.09, and in these cases, three were Dukes' stage C. Ten of 26 cases with diploid stemlines were Dukes' stage C. In thirty-

Table 1. DNA Ploidy of Colorectal A d e n o m a s in 2 6 Patients

A

B 2C

4C

C

No. of samples with different DNA ploidy

D 2C

4C

2C

4C

Figure 6. FCM DNA histograms of a case of villous adenoma. (A) A histogram of normal crypts (CV = 2.2"T). (B) Villous adenoma (CV = 4.19). (C) A mixture of normal and adenoma nuclei. DI was 1.04. (D) Intrinsic internal standard method. DI was also 1.04.

Colorectal adenoma

No. of samples

Diploid

Aneuploid

Multiploid

Adenoma in FAP Sporadic adenoma Villous adenoma

8

8

0

0

10

8

2

0

8

3

4

1

January 1 9 9 4

CRYPT-ISOLATION TECHNIQUE FOR FCM

Table 2. DNA Ploidy of Colorectal C a r c i n o m a s in 8 8 Patients

105

A

No. of samples with different DNA ploidy (%) Site of sampling

No. of sample

Diploid

Aneuploid

Multiploid

Edge of carcinoma Center of carcinoma

88 44

26 (30) 9 (20)

32 (36) 22 (50)

30 (34) 13 (30) zc

six cases with distinctive aneuploid stemlines with Dis ranging from 1.10 to 2.98, 16 cases were Dukes' stage C. Cases with hypodiploid stemline had significantly higher lymph node metastasis than the other cases (Table 3).

4'c

2C

4C

B

Discussion

Many methods have been introduced to prepare suitable samples of solid tumors for FCM DNA analysis, *a's-*° but all have problems,with contamination by stromal cells. Normal isolated crypts of the human colon have previously bden analyzed using FCM, 19a~ but no attempts have been made to separate malignant epithelial cells from stroma in colorectal cancer. Conventional methods use the contaminating nonneoplastic cells as the internal diploid standard on a DNA FCM histogram. The guideline of the international convention for cytometric standards suggests that the ideal reference cells are diploid cells from the same tissue and same individual, z6 Conventional methods try to conform to this guideline, but there are problems. First, the ratio of neoplastic to nonneoplastic cells in the samples is usually unknown,. Second, if the tumor cell had multiple stemlines including a diploid one, recognition of a tumor diploid stemline is difficult. Third, the proliferative fractions of tumors including S phase cannot be accurately calculated from the histograms because of contaminating stromal cells, t4,27

A 2¢

2C

2C

2¢

2C

2¢

Rgure 7. Crypts of adenocaminoma with near-diploid stemline. (A) Histogram of normal crypts (CV = 2.90). (B) Adenocarcinoma (CV = 3.30). (C) Mixture of normal and carcinoma nuclei with the ratio of 1 to 5. (D) Mixture of normal/cancer, 1:3. (E) Normal/cancer, 1:2, (F) Normal/cancer, 1:1. DI of cancer was 1.09.

d:

zc

4"c

=:

~c

Rgure 8. Comparison of the crypt isolation method and the conventional one. (A) Crypts of villous adenoma. Three h!stograms on the left were obtained by crypt isolation method; normal crypts (CV = 3.61), villous adenoma with two stemlines (/eft small peak, CV = 2.71, right larger peak, CV = 3.27) and mixture of normal/adenoma, left to right. Dis were 1.00 and 1.67. Right histogram was the same case treated by conventional method (left peak, CV = 3.12, right peak, CV = 3.30). DI was 1.67. (B) Crypts of tubulovillous adenoma having a near-diploid stemline; left to right, normal crypts (CV = 2.27), adenoma (CV = 4.19), a mixture of normal/adenoma, 1:3, and a mixture of normal/adenoma, 1:2. DI was 1.08. The conventional method (right histogram) failed to disclose the near-diploid peak (CV = 4.56) and DI was 1.0.

It is our opinion that the crypt isolation technique is an improvement on conventional methods. The diploid standard is normal epithelial cells of crypts from the same large intestine. Abnormal DNA ploidy and proliferative

J

B 4¢

__J

2¢

4'C

2'c

4'c

J

Rgure 9. A case of advanced cancer of the cecum with a diploid and a hypodiploid stemline. Left three histograms were normal crypts (CV = 3:18), cancer and a mixture of normal/cancer nuclei, Cancer histogram has two distinct peaks. Dis of the cancer were 0.91 and 1.0. Right histogram was obtained by conventional method which failed to disclose a hypodiploid peak (CV = 3.12). This case was Dukes' stage C.

106

NAKAMURA ETAL.

GASTROENTEROLOGY Vol. 106, No. 1

Table 3. Relationship Between DNA Ploidy and Dukes' Stage C in 76 Cases of Advanced Colorectal Carcinoma

DI

No. of case

DI < 1.0 DI = 1.0 1.0 < DI =< 1.1 DI > 1.1

7 26 7 36

No. of cases with Dukes' stage C (%) 7 10 3 16

(100) (38) (43) (44)

P value° -0.038 0.018 0.007

aA group with hypodiploid stemline is compared with each of the other three groups with diploid, hyper-near-diploid, and aneuploid stemline.

patterns have been reported in mucosa adjacent to colorectal cancer. 28 We found, however, no abnormal proliferative patterns .in the normal background colorectal mucosa of patients with either sporadic colorectal cancer or familial adenomatous polyposis. 23'29 Enker et al. examined 169 samples of normal colorectal mucosa using the crypt isolation technique and found all to be diploid) 9 In this study, all cases of normal crypt nuclei had a narrow G0/G1 peak with low CV and no aneuploid peak. Rather high frequencies (25%-82%) of aneuploidy have been observed in transitional mucosa adjacent to carcinomas. 25'2s Our study permits separation of the transitional drypts from cancer crypts using the dissecting microscope, and all 9 cases of transitional mucosa examined in our study had diploid stemline. Russo et al. reported the greatest frequency (39%) of cases with multiploid DNA aoeuploidy. 3° However, the frequencies of multiple DNA stemlines of colorectal carcinomas in most studies were rather low ( 6 % 10%). 12'14'17 In the peripheral edge and centrally ulcerated areas of cancer, multiple DNA stemlines were observed in 35% and 30% of cases, respectively, in this study. Because of the recognition of both diploid and near-diploid stemlines of carcinoma, rather high frequencies of multiple stemlines were observed using the crypt isolation technique. In larger adenomas, the incidence of DNA aneuploidy increased because of the development of focal carcinoma. H'31 In our study, all tubular adenomas less than 1 cm in diameter sampled from patients with FAP had diploid stemlines, but two cases (20%) of sporadic adenomas larger than 1 cm in size showed aneuploid stemlines. Perrin-Cottier et al. reported that there was no correlation between the ploidy and histopathological type of adenomas. 3~ Contrary to this, we have found a high frequency of aneuploidy (55.6%) in villous adenomas even if they showed only mild to moderate atypia. This is in keeping with their perceived higher risk for developing a malignancy.

Colorectal carcinomas with aneuploid stemlines were reported to have an unfavorable prognosis and tumors with diploid stemlines to have better survival and lower recurrence rates. 4-6'9'~2-~4'tT'Is Other studies, though, have shown no independent association between DNA aneuploidy and prognosis in colorectal cancer) 5't9 The conflicting results between abnormal DNA stemlines, prognosis, and survival may be caused by multiple factors including technical differences and tumor heterogeneity. In conventional studies, a DI value within 10% of the diploid stemline is usually regarded as diploid. 2'9'1~'19 This excludes recognition of near-diploid aneuploid peaks. .5 In our study, an appropriate mixture of control and neoplastic nuclei helped to separate near-diploid peaks from diploid ones. Our results indicate that carcinomas with hypodiploid stemline are associated significantly with Dukes' stage C. Tumoral heterogeneity in DNA ploidy studies of colorectal carcinomas has previously been highlighted, 32'33 and multiple tissue sampling is recommended to give accurate DNA measurements) 6'33 There have been several studies of colorectal cancers in which the sites of sampling are clearly defined. 7'34"35 Advanced colorectal cancer can grow laterally in the mucosa at the peripheral everted edges, but in central ulceration, invasion is vertical into the muscularis propria and subserosa often with extensive fibrosis. The difference between the noninvasive mucosal portions and invasive central areas, may reflect intratumoral variations in the DNA ploidy of cancer cells. The rate o f D N A aneuploid was higher in centrally ulcerated areas than at the peripheral edge (80% vs. 70%). Our incidence of DNA aneuploidy in centrally ulcerated areas is higher than those previously reported. 3e'37 DNA ploidy was concordant between central and peripheral portions in 21 (48%) of 44 cases. Although statistically not significant, our results suggest that some variation in DNA ploidy may accompany tumor invasion. We would therefore recommend that samples should be taken from the centrally ulcerated region. Further detailed prospective studies are being conducted to analyze the relation between clinicopathological status and various DNA ploidies and prognosis of the patients. References 1. Hedley DW, Friedlander ML, Tailor IW, Rugg CA, Musgrove ~,. Method for analysis of cellular DNA content of paraffin-embedded pathological material using flow cytometry. J Histochem Cytochem 1983;31:1333-1335. 2. Hedley DW, Friedlander ML, Taylor IW. Application of DNA flow cytometry to paraffin-embedded archival material for the study of aneuploldy and its clinical significance. Cytometry 1985;6:327333.

January 1994

3. Schutte B, Reynders MMJ, Bosman FT, Blijham GH. Row cytometric determination of DNA ploidy level in nuclei isolated from paraffin-embedded tissue. Cytometry 1985;6:26-30. 4. Quirke P, Dixon MF, Clayden AD, Durdey P, Dyson JED, Williams NS, Bird CC. Prognostic significance of DNA aneuploidy and cell proliferation in rectal adenocarcinomas. J Pathol 1987; 151:285-291. 5. Armitage NC, Ballantyne KC, Evans DF, Clarke P, Sheffield J, Hardcastle JD. The influence of tumour cell DNA content of survival in colorectal cancer: a detailed analysis. Br J Cancer 1990; 62:852-856. 6. Witzig TE, Loprinzi CL, Gonchoroff NJ, Reiman HM, Cha SS, Wieand HS, Katzmann JA, Paulsen JK, Moertel CG. DNA ploidy and cell kinetic measurements as predictors of recurrence and survival in stages B2 and C colorectal adenocarcinoma. Cancer 1991; 68:879-888. 7. Kim YJ, Ngoi SS, Godwin TA, DeCosse JJ, Staiano-Coico L. Ploidy in invasive colorectal cancer. Implications for metastatic disease. Cancer 1991;68:638-641. 8. Thornthwaite JT, Sugarbaker EV, Temple WJ. Preparation of tissues for DNA flow cytometric analysis. Cytometry 1980; 1:229237. 9. Tribukait B, Hammarberg C, Rubio C. Ploidy and proliferation patterns in colorectal adenocarcinomas related to Dukes' classification and to histopathological differentiation. A flow-cytometric DNA study. Acta Pathol Microbiol Immunol Scand 1983;91:8995. 10. Vindelev LL, Christensen IJ, Nissen NI. A detergent-tripsin method for the preparation of nuclei for flow cytometric DNA analysis. Cytometry 1983;3:323-327. 11. Sciallero S, Bruno S, Di Vinci A, Geido E, Aste H, Giaretti W. Row cytometric DNA ploidy in colorectal adenomas and family history of colorectal cancer. Cancer 1988;61:114-120. 12. Giaretti W, Sciallero S, Bruno S, Geido E, Aste H, Di Vinci A, d'Amore ESG. DNA flow cytometry of endoscopically examined colorectal adenocarcinomas. Pathol Res Prect 1989;185:589593. 13. Schillaci A, Tirindelli DD, Ferri M, Teodori L, Mauro F, Nicolanti V, Stipa S. Row CYtometric analysis in colorectal carcinoma: prognostic significance of cellular DNA content. Int J Colorsct Dis 1990; 5:223-227. 14. Kouri M, PyrhOnen S, Mecklin J-P, Jarvinen H, Laasonen A, Franssila K, Nordling S. The prognostic value of DNA-ploidy in colorectal carcinoma: a prospective study. Br J Cancer 1990;62:976-981. 15. Hood DL, Petras RE, Edinger M, Fazio V, Tubbs RR. Deoxyribonucleic acid ploidy and cell cycle analysis of colorectal carcinoma by flow cytometry. A prospective study of 137 cases using fresh whole cell suspensions. Am J Clin Pathol 1990;93:615-620. 16. Wersto RP, Liblit RL, Deitch D, Koss LG. Variability in DNA measurements in multiple tumor samples of human colonic carcinoma. Cancer 1991;67:106-115. 17. Meling GI, Rognum TO, Clausen OPF, Chen Y, Lunde OC, Schlichting E, Wiig JN, Hognestad J, Bakka A, Havig (3, Bergan A. Association between DNA ploidy pattern and cellular atypia in colorectal carcinomas. A new clinical application of DNA flow cytometric study? Cancer 1991; 67:1642-1649. 18. Rognum TO, Lund E, Meling GI, Langmark F. Near diploid large bowel carcinomas have better five-year survival than aneuploid ones. Cancer 1991;68:1077-1081o 19. Enker WE, Kimmel M, Cibas ES, Cranor ML, M~lamed MR. DNA/ RNA content and proliferative fractions of colorectal carcinomas: a five-year prospective study relating flow cytometry to survival. J Natl Cancer Inst 1991;83:701-707. 20. Bjerknes M, Cheng H. Methods for the isolation of intact epithelium from the mouse intestine. Anat Rec 1981;199:565-574. 21. Cheng H, Bjerknes M, Amar J. Methods for the determination of

CRYPT-ISOLATION TECHNIQUE FOR FCM 107

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36. 37.

epithelial cell kinetic parameters of human colonic epithelium isolated from surgical and biopsy specimens. Gastroenterology 1984;86:78-85. Arai T, Kino I. Morphometrical and cell kinetic studies of normal human colorectal mucosa. Comparison between the proximal and the distal large intestine. Acta Pathol Jpn 1989;39:725730. Nakamura S, Kino I, Baba S. Nuclear DNA content of isolated crypts of background colonic mucosa fr£)m patients with familial adenomatous polyposis and sporadic colorsctal cancer. Gut 1993;34:1240-1244. Rlipe MI, Branfoot AC. Abnormal patterns of mucus secretion in apparently normal mucosa of large intestine with carcinoma. Cancer 1974;34:282-290. Wersto RP, Greenebaum E, Deitch D, Kersbergen K, Koss LG. Deoxyribonucleic acid ploidy and cell cycle events in benign colonic epithelium peripheral to carcinoma. Lab Invest 1988; 58:218-225. Hiddemann W, Schumann J, Andreeff M, Barlogie B, Herman CJ, Leif RC, Mayall BH, Murphy RF, Sandberg AA. Convention on nomenclature for DNA cytometry. Cytometry 1984;5:445-446. Nsien E, Steinberg WM, Wilkinson DS, Rhame JG, Henry JP. Single-parameter DNA flow cytometric analysis of normal-appearing colonic mucosa does not predict the presence of colonic neoplasia. Am J Gastroenterol 1991;86:1477-1481. Ngoi SS, Staiano-Coico L, Godwin TA, Wong RJ, DeCosse JJ. Abnormal DNA ploidy and proliferative patterns in superficial colonic epithelium adjacent to colorectal cancer. Cancer 1990;66:953-959. Nakamura S, Kino I, Baba S. Cell kinetics analysis of background colonic mucosa of patients with intestinal neoplasms by ex vivo autoradiography. Gut 1988;29:997-1002. Russo A, Bazan V, Plaja S, Leonardi P, Bazan P. Patterns of DNAploidy in operable colorectal carcinoma: a prospective study of 100 cases. J Surg Oncol 1991;48:4-10. Perrin-Cottier M, Jouffre C, Sabido O, Maubon I, Barth~l~my C, Pasteur X, Audigier JC, Laurent JL. Row cytometric analysis of DNA content in colorectal polyps resected at endoscopy. Anal Quant Cytol Histol 1991;13:132-138. Sasaki K, Hashimoto T, Kawachino K, Takahashi M. Intratumoral regional differences in DNA ploidy of gastrointestinal carcinomas. Cancer 1988; 62:2569-2575. Lindmark G, Glimelius B, P~hlman L, Enblad P. Heterogeneity in ploidy and S-phase fraction in colorectal adenocaminomas. Int J Colorect Dis 1991;6:115-120. Quirke P, Dyson JED, Dixon MF, Bird CC, Joslin CAF. Heterogeneity of colorectal adenocarcinomas evaluated by flow cytometry and histopathology. Br J Cancer 1985;51:99-106. Scott NA, Grande JP, Weiland LH, Pemberton JH, Beart RW, Lieber MM. Row cytometric DNA patterns from colorectal cancers--how reproducible are they?. Mayo Clin Proc 1987;62:331337. Silvestrini R, Faranda A, Costa A. DNA ploidy and cell kinetics in human colorectal carcinomas. J Surg Oncol Suppl 1991;2:4-8. Dean PA, Vemava III AM. Row cytometric analysis of DNA content in colorectal carcinoma. Dis Colon Rectum 1992;35:95-102.

Received March 31, 1993. Accepted July 7, 1993. Address requests for repflnts to: Shln-ichl Nakamure, M.D., Division of Pathology, Central Clinical Laboratory, School of Medicine, Iwate Medical University, 19-1 Uchimaru, Morioka 020, Japan. , The authors thank Drs. Jonathan P. Sheffield and lan C. Talbot, Department of Pathology, St. Mark's Hospital, for their Instructive suggestions; Michikazu Samejlma, Associate Professor, Department of Physiology, Hamamatsu University School of Medicine, for statistical analysis; and Tuneji Noguchl for photography.