Path. Res. Pract. I63, 228-23I (I978)

Institute of Oncology (Director: Professor 1. Kiricuta), Cluj-Napoca, Romania

Clustering of Mitoses in Human Cervical Carcinoma

C. D. OLINICI and ELISABETA DAVID

Summary Mitotic nests have been noted both in tissue culture (Cone, 1968) and in human

skin (Rowe and Dixon, 1975). The present paper is concerned with the disposition of mitoses in human tumor tissue. Mitotic indices, cell counts, and mitotic clusters were recorded in 20 carcinomas of the cervix uteri. The significance of the results was verified statistically. In 18 of 20 biopsies, the clustering of mitoses was in excess of chance at the 95% confidence level. Assuming a chance distribution, one would expect only one case (5% of 20 cases) to show significant clustering of mitoses.

The incidence of cases which show mitotic clustering is higher in cervical carcinoma than in normal skin and uninvolved skin of psoriatic patients. The clustering phenomenon might be due to an initiating factor derived from a mitotic cell which propagates to other cells and activates them, or to a local deficiency in inhibitor factors which rever­sibly inhibit mitoses under normal conditions.

Introduction

In I968, Cone reported that mitoses spread sequentially outward from the initial mitotic cells in groups of cultured L-line mouse sarcoma cells linked up as networks by intercellular bridges. Burns et al. (I973) noted mitotic nests in human vaginal smears and in cultures of amphibian tissues. In human skin, the number of clustered mitoses was significantly greater than expected by chance alone (Rowe and Dixon, I 97 5), and it was sug­gested that a mitotic stimulus may radiate from mitotic cells in human epidermis. As this hypothesis has obvious implications for the prolifera­tion of neoplastic cells, we decided to investigate the distribution of mi­toses in human tumor tissue.

Materials and Methods The data reported here are derived from the study of 20 carcinomas of the cervix

uteri. Histologically these were large cell epidermoid carcinomas without significant keratinization, corresponding to the grade II carcinoma of Sidhu et al. (1970).

Clustering of Mitoses 229

Table I Disposition of mitotic figures in human cervical carcinoma

Case Mitotic No. of Xu XI X2 Xa "1..2 index cells!

IO,Ooo[.lm2

1.4 94 3032 628 321 19 1.24 2 1.2 rr6 3197 640 141 22 1. 17

3 1.3 I07 3497 424 79 1.13

4 1.8 16, 3061 601 317 21 1.26

5 1.5 I)L) 21.)85 51.)6 338 81 1.39 6 1.6 I2~ ))16 ),6 142 6 1.17

7 2 . I 141 )242 504 190 4 1.23 8 1.9 r Il) 2764 857 321 58 1.17

9 1.5 101 2633 986 362 19 1.°3 10 1.7 1)4 }lo) 658 123 16 1.12 II 1.5 121 21.) 89 687 319 5 1.16 12 1.1 114 .B35 621 42 2 0.96

13 1.3 1)6 3093 726 175 6 1.08

14 1.2 147 ,) IO 403 84 3 1.18

15 1.8 t2H ,l28 649 176 47 1.27 16 2·3 [[8 ,222 638 117 23 1.16

17 1.4 Il, 21.)49 773 269 9 1.11 18 1.6 148 ,))3 564 96 7 LII

19 1.2 IlL) )525 423 49 3 1.°9 20 1.4 IF 3609 336 51 4 1.17

All operations were performed between 9 a.m. and 1 p.m. Biopsy specimens were immediately fixed in Bouin's fluid, and multiple sections were cut at 6 ftm and stained with hematoxylin and eosin. A 5 X 5 mm grid was placed on a 12.5 X eyepiece. With the 100 X objective and the optic system of our NU-2 (Zeiss, Jena) microscope, this grid projects a square area of 50 X 50 = 2,500 ,um2 onto the microscopic image. Four thousand adjacent field s, corresponding to 10 mm2, were scanned in alternate sections in order to avoid counting the same mitotic figure twice (Rowe and Dixon, 1975). We recor­ded the number of areas with one (XI), twO (X2), and three (X:l) mitotic figures. Conti­guous mitoses were counted separately only when they were in different stages. Otherwise, they were counted as a single mitos is. As suggested by Rowe and Dixon (1975), this probably represents an overcorrection for sister cells.

Statistical calculation fo llowed the procedure of Rowe and Dixon (1975). The following steps were performed:

a) Calculation of the total number of mitoses, which is equal to XI + 2X2 + 3X3. b) Calculation of the mean number of mitoses, which equals the total number of

mitoses (Xl + 2Xt + 3Xa) divided by the total number of scanned areas (4,000). c) Calculation of the variance. which is equal to

(X l + ~X2 + \IXa) - (4,000 )' mean2)

3,999 d) Calculation of X2, namely the variance divided by the mean.

The mitotic index was calculated on 3,000 cells in each case. Cell density was ob­tained by counting the number of nuclei corresponding to 4 fields (10,000 ftm2 ).

23 0 . C. D. Olinici and Elisabeta David

Results Table I summarizes the principal findings concerning the tumors. The

values of mitotic indices range from I. 1% to 2.3%, with an average value of I. 540/0. The cell countslIo.ooo '[lm2 range from 94 to 163 with a mean value of I26.7.

If the number of mitotic clusters were due to chance, then we would expect X2 = I. Values of X2 greater than I.04 are significant at the 5%

level. Only in two cases, namely cases 9 and 12, were the values of X2

lower than I.04. It is thus clear that in 18 of 20 biopsies, the clustering of mitoses is in excess of chance at the 95010 confidence level. Assuming a chance distribution, one would expect only one case (SOlo of 20 cases) to show a clustering of mitoses.

Discussion Our results show that mitotic cells in human cervical carcinoma exhibit

a definite tendency to cluster in nests. The incidence of cases which show mitotic clustering is greater in cervical carcinoma (18 of 20 cases) than in normal skin and in the uninvolved skin in psoriasis (23 of 60 cases, ac­cording to Rowe and Dixon, 1975). The following qualifications should be made, however. a) Cell cycle time can vary even in sister cells, and thus contiguous mitoses could be in different stages of mitosis and still be sister cells. b) The migration of cells following mitosis in carcinoma of the cervix may not be as regular as in normal epidermis, and thus adjacent but noncontiguous mitotic figures could also be sister cells. c) It is well known that the mitotic index is subject to diurnal variation; however, all patients were operated upon during a very limited time of the day.

Data published by Hillemans et al. (1968) have shown that the cell count reaches a maximum in carcinoma in situ, representing the "critical cell number" necessary for infiltration. The frequency of mitoses/cell count is, however, higher in invasive carcinoma than in carcinoma in situ. The study of Chi et al. (1977) demonstrated that foci of dividing cells alternate with foci of non-dividing cells, and they also suggested the existence of proliferating and non-proliferating compartments in carcinoma in situ. It would therefore be interesting to see whether or not there are quantitative differences in the distribution of mitotic figures in these two different stages of cervical neoplasia.

There is no clear explanation of the clustering phenomenon. Cone (1968) and Burns et al. (1973) suppose that an initiating factor, derived from the mitotic cell, could propagate to other cells and activate them. An alternative explanation might be a local deficiency in inhibitor factors

Clustering of Mitoses . .4 31

which reversibly inhibit mitoses (Bell, 1976) under normal conditions. Fi­nally, the influence of tumor architecture, which is difficult to study in the majority of cases, has been shown to have a clear effect on the mitotic index in mouse mammary tumors (Tannock, 1968), and could affect the topographic distribution of mitoses.

Acknowledgement. The autho~s wish to express their gratitude to Dr. Lyon Rowe for helpful comments and to Dr. W. J. Dixon for help with statistical calculation.

References Bell, G. I.: Models of carcinogenesis as an escape from mitotic inhibitors. Science 192,

569-572 (1976) Burns, E. R ., Uyeda, C. K. , and Scheving, L. E.: Mitotic nests in human vaginal smears

and cultures of amphibian tissues. Oncology 27, 92-96 (1973) Chi, C. H ., Rubio, C. A., and LagerlOf, B. : The frequency and distribution of mitotic

figure s in dysplasia and carcinoma in situ. Cancer J9, 1218-1223 (1977) Cone jr., C. D.: Considerations on self-induced mitosis and autosinchrony in sarcoma cell

networks. Cancer Res. 28, 2155-2161 (1968) Hillemans, H. G ., Frohlich, D., and Prestel, E.: Zellzahl und Mitosefrequenz bei Ent­

stehung des Cervixca!"cinoms (zum Carcinoma in situ und Mikrocarcinom der Cervix). Arch. Gynakol. 206, 2')2 -- } 1 9 (1968)

Rowe, L., and Dixon, W. J.: Mitoses in human epidermis. Oncology Jl, 133-138 (1975) Sidhu, G. 5., Koss, L. G. , and Barber, H . R . K.: Relation of histologic factors to the

response of stage I epidermoid carcinoma of the cervix to surgical treatment. Analysis of 1 IS patients. Obster. Gynec. J5 , 329-338 (1970)

Tannock, 1. F. : The relation between cell proliferation and the vascular system in a transplanted mouse mammary tumour. Brir. J. Cancer 22, 258-273 (1968)

Received April 5, 1978 ' Accepted in revised form July 3,1978

Key-words : Cervical carcinoma - Mitoses - Clustering - Statistical calculation - Tumor pathology

Dr. Corneliu D. Olinici, Institute of Oncology, Republicii 34-36, R-3400 Cluj-Napoca, Romania