Immunohistochemical Determination of Growth Fraction in Human Tumors

Path. Res. Pract. 184, 609-613 (1989)

Immunohistochemical Determination of Growth Fraction in Human Tumors K. Matsumura, T: Tsuji and F. Shinozaki Department of Oral and Maxillofacial Surgery, Yamaguchi University School of Medicine, Ube-shi, Japan

K. Sasaki and M. Takahashi Department of Pathology, Yamaguchi University School of Medicine, Ube-shi, Japan

SUMMARY

The growth fraction in 93 cases of human tumors was estimated by an immunohistochemical staining using Ki-67. The tumors consisted of the following: 14 oral cancers, 14 breast cancers, 9 gastric cancers, 9 uterine cancers, 8 ovarian cancers, 6 colo-rectal cancers, 6 thyroid cancers, 5 esophagus cancers and 22 miscellaneous tumors. Regional labeled cells were predominantly found in the periphery of the tumor nests in squamous cell carcinomas. However, in adenocarcinomas the labeled cells were randomly distributed in tumor cell nests. The Ki-67 labeling index varied greatly from case to case (almost 0 to 50.9% with an average of 17.3%), even within the same organ group. The growth fraction was independent of the histological pattern, although thyroid cancers showed a lower labeling index than other malignant tumors. The labeling indices in benign tumors were lower than those in malignant tumors. The usefulness of this method for the estimation of biological behavior of human tumors is suggested.

Introduction The study of cell kinetics in human tumors has contributed to the understanding of growth characteristics and to the design of effective treatment methods for malignant tumors. Proliferative activity of tumors has been estimated by the frequency of mitotic figures and/or by the tritiated thymidine (3H-TdR) labeling index. However, the former is unreliable because of its low incidence. The latter is laborious and time-consuming, and the established methods by which labeling with DNA precursors is employed are not easily applicable to routine clinical material. DNA flow cytometry offers accurate and relevant information about the human tumors permitting of percentage of cells in S-phase and the ploidy status. © 1989 by Gustav Fischer Verlag, Stuttgart

Recently, a monoclonal antibody, Ki-67, recogmzmg nuclear antigen exclusively present in cycling cells, whereas not in quiescent cells, has been developed 2• It has been shown that the percentage of Ki-67 positive cells in a tumor population corresponds to growth fraction (GF). This means that GF of human tumors can be easily estimated in site by an immunohistochemical method. Extensive data concerning growth fraction of various kinds of human tumors are not yet available. A few reports dealing with the Ki-67 labeling index for limited types of malignant tumors such as lymphoma, breast cancer and bone tumors can be found!,3, 9,14. In this paper, we report growth fraction for various kinds of human tumors and examine the relationship of Ki-67 labeling index to their histological characteristics. 0344-0338/89/0184-0609$3.50/0

610 . K. Matsumura, T. Tsuji, F. Shinozaki, K. Sasaki and M. Takahashi

Material and Methods Specimens: Ninety-three tumor specimens, which were obtained from 56 females and 37 males in Yamaguchi University Hospital. The specimens were as follows: 14 cases of oral and breast cancers, 9 cases of gastric and uterine cancers, 8 cases of ovarian cancers, 6 cases of colo-rectal and thyroid cancers, 6 cases of esophagus cancers and 22 cases of miscellaneous tumors (Table 1). In addition, eight benign tumors were included in this investigation. The ages of the patients ranged from 7 to 82 years with an average of 56.2 years. The specimens were cut into two parts, and one part was subjected to immunohistochemical staining of Ki-67 as previously described 13 . In brief, a sample was embedded in O.T.C. compound (Miles Scientic, U.S.A.) immediately after biopsy or surgical resection, frozen at -80 DC and stored in a freezer (-80 DC) until immunohistochemical staining. The cryostat sections were made at 5 [tm and mounted on slides. The other half of the tissue specimens was fixed in a 10% formalin solution and processed for conventional histological examination. Immunohistochemical staining: The cryostat sections were dried well, fixed in acetone for 10 min at 4 DC and then washed in

Fig. 1. Cryostat section of the breast carcinoma is immunostained with Ki-67. The nucleoli are strongly stained with Ki-67 and diffuse nuclear staining was also observed, x 360.

phosphate buffer saline (PBS, Ph 7.4). The slides were exposed to 20% normal horse serum to prevent non-specific binding of the monoclonal antibody to the specimens after blocking of endogeneous peroxidase by 0.3% hydrogen peroxide in methanol. The immunohistochemical method was employed for Ki-67 staining by using a Vectastain ABC kit (Vector Lab., U.S.A.) as previously reported 13 • Briefly, the sections were first incubated with Ki-67 (1: 50 diluted with PBS containing 0.5% bovine serum albumin, Dako, Denmark) for 1 hour at room temperature. The sections were treated with a biotin labeled second antibody and an avidin-biotin-peroxidase complex in order. Peroxidase labeling was developed with a diaminobenzidinehydrogen peroxide (DAB) substrate (Wako Chemical Industries, Osaka, Japan) and the slides were lightly counterstained with hematoxylin. Determination of Ki-67 labeling index: All of the nuclei stained with DAB were regarded as positive for Ki-67, regardless of staining intensity. The labeling index was determined by observing more than 1000 nuclei in areas with a representative histological pattern. Histological examination: Histological diagnoses of the tumors were done on hematoxylin-eosin preparations according to conventional criteria.

Fig. 2. A case of well differentiated squamous cell carcinoma of the tongue seen in a 49-year-old male. The cells in the periphery of tumor cell nests are preferentially stained with Ki-67 and keratinized cells are devoid of the label. The labeling index of Ki-67 is 21.3%, x 200.

Immunohistochemical Determination of Growth Fraction in Human Tumors· 611

Results The nucleoli were strongly stained with Ki-67 and diffuse nuclear staining was also observed. Neither cytoplasm nor cell membrane were labeled in all specimens. Extensive staining was noted on chromosomes in mitotic cells (Fig. 1). Positive cells and negative cells were easily differentiated. The distribution of the Ki-67 labeled cells varied from area to area within a single tumor tissue, and in some cases the labeled cells were found in the restricted portions of the tissue specimen. In squamous cell carcinoma, the Ki-67 positive cells were rarely found in the central portions of the tumor nests and keratinized cells were devoid of the label, whereas the cells in the periphery of the nests were preferentially stained with the antibody (Fig. 2). In adenocarcinomas, the Ki-67 positive cells were randomly distributed along the entire length of the tubular structures (Fig. 3). Stromal elements, such as fibroblasts, were occasionally stained with the antibody.

The GF estimated by immunohistochemical staining of Ki-67 is summarized in Table 1. The frequency of Ki-67 positive cells was not related with either age or sex (Fig. 4). Labeling indices varied greatly from case to case (almost 0 to 50.9%), within the same organ or the same histological group. The overall labeling index in malignant tumors was 18.2%. The labeling indices of the oral cancers ranged from 4.9 to 40.0% with an average of24.1 %. The average labeling indices for cancers of breast, gastric, and uterine were 15.1 %,20.7%, and 25.9%, respectively. In contrast, the average labeling index for benign tumors was 3.9%. In this series, there was a significant difference in the labeling index between malignant and benign tumors. Interestingly, thyroid cancers had significantly low values of the Ki-67 labeling index, compaired with gastric, uterine, ovarian, breast and oral cancers. The difference in the Ki-67 labeling indices of malignant tumors was not significant among involved organs other than the thyroid gland.

Fig. 3. The Ki-67 positive cells are distributed haphazardly along the entire length of the tubular structures in adenocarcinomas. left: A case of serous cyst adenocarcinoma of the ovary seen in a 55-year-old female. The labeling index of Ki-67 is 36.3 %, X 100; right: A case of adenocarcinoma of gall-bladder seen in a 70-year-old male. The labeling index is 31.7%, x 210.

612 . K. Matsumura, T. Tsuji, F. Shinozaki, K. Sasaki and M. Takahashi Table 1. Ki-67 Labeling Index of Human Tumors Site of Tumors Oral cancer Breast cancer Gastric cancer Uterine cancer Ovarian cancer Colo-rectal cancer Thyroid cancer Esophagus cancer Lung cancer Malignant lymphoma Hepatoma M. F. H. Renal cell cancer Nephroblastoma Gall bladder cancer Malignant meningioma Astrocytoma Benign tumor

No. of Cases 14 14 9 9 8 6 6 5 3 2 2 2 1 1 1 1 1 8

Mean LI (%)*

24.1 15.1 20.7 25.9 18.6 22.5 3.9 16.7 19.7 6.9 4.7 31.0 2.2 10.0 31.7 6.3 3.0 3.9

0

50 Range of LI (%)

4.9-40.0 1.0-34.9 4.3-40.0 14.7-39.0 3.5-36.9 3.6-50.9 0 -14.3 16.8-24.7 13.2-25.6 5.7- 8.1 3.1- 6.3 19.1-42.8

itO' '-' If)

40

•

...J ...J

• 0

w 30

()

I-

20

o

If)

~

,...

coI ~

o~

00 0

w

>

e

o o • 00

eo.

o·

0

•

0

• •

o

•

Q

o

• 0

•• 0

•

• 0 .~~

0:>

•

10

.IJ

0

• 0 0

0 •• 0

•

0

00 0

~.

00 10 20 30 40 50 60 70 80 90

AGE (YEARS)

o

-10.6

" The result of t test in the labeling index shows significant difference between malignant and benign tumors (p < 0.05). Thyroid cancers have significantly low values of Ki-67 labeling index compared with oral, breast, gastric, uterine, ovarian, colorectal and esophagus cancers (p < 0.05). LI: Labeling index, M.F.H.: Malignant fibrous histiocytoma.

Discussion

It is important to measure tumor proliferating activity not only for understanding the biological characteristics of tumor growth but also for cancer treatment4, 8. In the case of human tumors, however, there are methodological problems: the method using DNA precursors is laborious. Actually the precursor labeling procedure cannot be widely applied to routine clinical material. The development of a simple and rapid method for estimating proliferation activity is necessary for clinical use of cell kinetic data. A monoclonal antibody, Ki-67, which reacts with an antigen exclusively expressed in cycling cells has been produced and now provides GF of tumor cell populations. The GF is one of the most important parameters indicating tumor growth activity. The Ki-67 labeling index correlated well with the frequency of mitosis in breast cancer 1,9. As is commonly known, the fraction of S phase is smaller than GF. However, a close relationship between Ki-67 and BrdDrd labeling indices was demonstrated in human tumors 13 , as well as in experimental animal tumors 6• These observations would support the feasibility of this approach in determining the proportion of proliferating cells in a tumor population. As shown in this investigation, the growth fraction was much smaller in human solid tumors than in in vitro cultured cells such as HeLa cells in which GF is usually higher than 0.9 12 . The Ki-67 labeling index varied greatly from case to case even within a group of the same histological type, and also from area to area within a single

Fig. 4. The relationship of Ki-67 labeling indices to age and/or sex (0: female, .: male). The frequency of Ki-67 positive cells is not related with either age or sex. Six thyroid cancers and 8 benign tumors are eliminated from this figure.

tumor tissue specimen. These observations suggest the complexity of tumor cell proliferation, especially in human cases. A relationship of the labeling index to morphological features was not demonstrated. In general, tumor growth activity was independent of histological patterns and might be influenced by other factors. This was reported in the study of breast carcinoma using Ki-67 1,9, and the same conclusion was also introduced by autoradiographic studies for carcinomas of the stomach 11, the uterine cervix 10 , the colon? and the breast9 • McDonald suggested that the variability in the labeling index reflects inherent proliferative activity of individual tumors 5 • As expected from biological behavior, however, thyroid cancers which show relatively good prognosis had low GF, as compared to other malignant tumors. The labeling index was significantly different between benign and malignant tumors. It may be appropriate to consider that GF represents biological activity of the tumors. The nucleus was diffusely stained with Ki-67, and the nucleolus was stained more densely. Chromosomes reacted strongly with the antibody. This was supported by a flow cytometric bivariate analysis of the expression of the Ki-67 reactive antigen in HeLa cells. The level of the antigen increased with the progression of the cell cycle and was maximum at G2M phase 12 • In cases of squamous cell carcinoma, labeled cells were frequently in the peripheral area of the tumor nests but rarely in the central area where keratinized cells were noted. The regional difference in the frequency of labeled cells suggests that tumor cells are produced in the periphery of tumor nests and emigrate to the inner portion of the nests. As a consequence, a tumor grows if cell loss is smaller than cell production. This proliferating pattern imitates that of normal stratified squamous epithelium. In contrast to squamous cell carcinomas, progenitor cells in adenocarcinomas were scattered throughout the tumor nests. It is hypothesized that

Immunohistochemical Determination of Growth Fraction in Human Tumors· 613 the pattern of cell proliferation is different between squamous cell carcinomas and adenocarcinomas. Immunohistochemical investigation using Ki-67 provides useful information on the growth of human tumors. Further study would be necessary to further clarify the growth characteristics of human tumors and the clinical relevance of GF. References 1 Barnard NJ, Hall PA, Lemoine NR, Kadar N (1987) Proliferative index in breast carcinoma determined in site by Ki-67 immunostaining and its relationship to clinical and pathological variable. J Pathol152: 287-295 2 GerdesJ, Schwab W, Lemke H, Stein H (1983) Production of a mouse monoclonal antibody reactive with a human nuclear antigen associated with cell proliferation. Int J Cancer 31: 13-20 3 GerdesJ, Dallenbach F, Lennert K, Lemke H, Stein H (1984) Growth fractions in malignant non-Hodgkin's lymphomas (NHL) as determined in site with the monoclonal antibody Ki-67. Haematol Oncol 2: 365-371 4 Hart S, Livingston RB, Murphy WK, Barlogie B, Gehan EA, Bodey GP (1976) Neoplasia, kinetics, and chemotherapy. Seminar Oncol 3: 259-270 5 MacDonald I (1951) Biological predeterminism in human cancer. Surg Gynecol Obstet 92: 443

6 Mendelsohn ML (1975) The cell cycle in malignant and normal tissues. In: Hampton JC (Ed.) The Cell Cycle in Malignancy and Immunity. Technical Information Center 293-314 7 Meyer JS, Prioleau PG (1981) S-phase fractions of colorectal carcinomas related to pathologic and clinical features. Cancer 48: 1221-1228 8 Meyers JS, Freidman E, McCrate M, Bauer W (1983) Prediction of early course of breast carcinoma by thymidine labeling. Cancer 51: 1879-1886 9 MucgurrinJF, Doria MI, Dawson PJ, Karrison T, Stein HO, Franklin WA (1987) Assessment of tumor cell kinetics by immunohistochemistry in carcinoma of breast. Cancer 59: 1744-1750 10 Sasaki K (1977) Measurement of tritiated labeling index by incubation in vitro of surgically removed cervical cancer. Gann 68: 307-313 11 Sasaki K, Takahashi M, Ogino T, Okuda S (1984) An autoradiographic study on the labeling index of biopsy specimens from gastric cancers. Cancer 54: 1307-1309 12 Sasaki K, Murakami T, Kawasaki M, Takahashi M (1987) The cell cycle associated change of the Ki-67 reactive nuclear antigen expression. J Cell Physiol133: 579-584 13 Sasaki K, Matsumura K, Tsuji T, Shinozaki F, Takahashi M (1988) Relationship between labeling indices of Ki-67 and BrdUrd in human malignant tumors. Cancer 62: 989-993 14 Vollmer E, Roessner A, Gerdes J, Mellin W, Stein H, Chong-Schachel S, Grundmann E (1986) Improved grading of bone tumors with the monoclonal antibody Ki-67. J Cancer Res Clin Onco1112: 281-282

Received October 10, 1988 . Accepted in revised form January 11, 1989

Key words: Ki-67 - Growth fraction - Labeling index - Biological behaviour of human tumors Kouji Matsumura, D. D. S., Department of Oral and Maxillofacial Surgery, Yamaguchi University School of Medicine, Ube-shi 755, Japan