Cell Tissue Kinet. (1977) 10, 429-436.

A N I N VIVO STATHMOKINETIC STUDY OF CELL PROLIFERATION I N HUMAN GASTRIC CARCINOMA A N D GASTRIC MUCOSA NICHOLAS A . WRIGHT, D A V I Dc . B R I T T O N , G E O R G EB O N E A N D D A V I DR . A P P L E T O N Departments of Pathology, Surgery and Medical Statistics, University of Newcastle upon Tyne (Received I May 1916,revision received 20 December 1976) ABSTRACT

Cell production rates were studied in a group of eleven patients with carcinoma of the stomach using an in uiuo technique with vincristine. In normal pyloric mucosa estimates ranged from 8 to 15 cells/l000 cells/hr, and in the carcinomas from 3 to 24 cells/l000 cells/hr. Because of the very large variation in the data, comparison between individual tumours and normal mucosa is not regarded as being worthwhile at this stage. The implications of these results for previous human in uiuo stathmokinetic experiments are also discussed. The special relevance of cell population kinetics in human tumours is to radiotherapy and cytotoxic chemotherapy; the ultimate aim of oncologists is to be in a position where kinetic parameters in individual human tumours can be measured and used to predict therapeutic responses to a specially designed radiotherapeutic and chemotherapeutic regimen. Viewed in this light, experiments involving tritiated thymidine (3H-TdR) would appear to be of strictly limited usefulness. The technique is time consuming in terms of both technical and quantitative procedures, and a delay of 6-8 weeks in obtaining data which might be needed quite urgently has prompted investigators to seek an alternative, more rapid technique. One of these is the stathmokinetic or metaphase arrest method, in which routine (but thin) paraffin sections can be used, and the method of quantitation merely involves the determination of mitotic indices. The amount of kinetic information is limited when compared with the more sophisticated 3H-TdR methods, but one useful kinetic parameter which should be available is the birth rate (kb) read directly from the slope of the mitotic accumulation graph. The birth rate can be converted to a cell production rate, expressed as cells per 100 or 1000 per hour (Smith, Thomas & Riches, 1974). The potential doubling time ( t p d )can be calculated from the birth rate. From data of this type it might be expected that one could distinguish between slowly and rapidly proliferating tumours, in order to predict which neoplasms would reasonably be expected to respond to radiotherapy or chemoCorrespondance: Dr N. A. Wright, Department of Pathology, Gibson laboratories, Radcliffe Infirmary, Oxford. 429

430

Nicholas A . Wright et al.

therapy. Further, the proliferative rate of the tumour may be compared with that of the native, adjacent tissue of origin, to obtain some idea of the differential rate of cell killing to be expected from radiotherapy. Possibly with these thoughts in mind, several groups of workers have endeavoured to apply stathmokinetic techniques to a variety of human tumours. Using colcemid, Refsum & Berdal (1967) reported a mean tpdof 2.6 days in a group of sixty-one tumours, mainly carcinomas of the mouth and throat; Fulker, Cooper & Tanaka (1971), in vesical carcinoma, found that fpd varied from a mean of 22.2 days in well differentiated tumours down to 2.2 days in undifferentiated carcinomas, and there was a clear negative correlation between the degree of differentiation and the rate of cell proliferation. Also with colcemid, Aherne & Buck (1 97 1 ) reported a mean t,, of 4 . 2 days in a group of five neuroblastomas, and for a group of six nephroblastomas 7.8 days. The vinca alkaloids have also been employed; Meyer & Donaldson (1969) used vinblastine to measure t,, which varied between 6 and 9 days in oral squamous carcinomas, while Camplejohn, Bone & Aherne (1973) biopsied patients with carcinoma of the rectum before and after vincristine, and reported a mean tpd of 192 hr; further, Camplejohn et al. c o n sidered that cell production rates were lower in the tumours than in the adjacent, normal mucosa. Whilst these data seem useful and interesting at face value, in spite of apparently good linearity of metaphase collection, the number of readings in individual cases in most studies is small; although 95% confidence intervals for t,, have not been calculated, these are probably considerable. In this paper we report studies made of eleven patients with carcinoma of the stomach, employing a stathmokinetic technique with vincristine. The adjacent normal mucosa, and hyperplastic and metaplastic gastric mucosa were also studied. Although at first sight the data appear reasonable and indeed useful, closer inspection shows how difficult it is to draw worthwhile conclusions with this technique in its present form. MATERIALS AND METHODS All patients received full information concerning the experimental procedures, and gave informed consent. The experimental procedure was submitted and passed by the Ethical Committee of the Newcastle upon Tyne University Hospitals Group. As a first step, during diagnostic endoscopy, biopsies were taken of the gastric mucosa, and also of the pyloric and fundal mucosa. These biopsies were fixed in 10% neutral buffered formol saline and post-fixed in formol-corrosive. Routine paraffin sections were used to establish and report the diagnosis of malignancy. Once diagnosis was established, serial sections were cut at a thickness of 3 pm for mitotic counting, and stained with haematoxylin and eosin. Immediately prior to surgery, when the patient was anaesthetized, vincristine sulphate (Oncovin, Lilly) was infused rapidly via an intravenous cannula at a dose of 0.045 mg/kg body weight (Wright et al., 1973). A note was made of the time the drug was given. When the stomach was exposed a gastrotomy incision was made, and biopsies of the tumour, and of normal gastric mucosa, were made with pituitary rongeurs, care being taken to biopsy the same portion of the tumour in each case, as far as was possible. Thereafter, during mobilization of the stomach, biopsies were taken at intervals which were convenient to the operative procedure. As far as possible, the period after the blood supply to the stomach had

Cell proliferation in gastric carcinoma

43 1

been seriously compromised was avoided, although during the resection procedure it was inevitable that some decrease occurred in blood supply to the stomach. In the tumour material, at least 2000 nuclei were counted from each biopsy, and the mitotic index determined for each time period. In the adjacent normal mucosa, considerable difficulty was experienced in endoscopy material in obtaining sufficient well-orientated gastric glands in which all cells in the neck, isthmus and pit could be counted to collect 2000 nuclei in each biopsy. In mucosae showing intestinal metaplasia, as far as possible all cells in the ‘crypt’ were included in the count. The results were plotted as mitotic index against time after vincristine, and where appropriate a line was fitted to the data, using least squares and assuming a rectangular age distribution (Wright, Morley & Appleton, 1972). For comparison, the data were also analysed assuming exponential conditions, using the relationship log, (1

+ I,)

=

1 (t, + t ) tP*

Where only two biopsies (before and after vincristine) were available or suitable for counting, then the rate of entry into mitosis, and hence the birth rate, was calculated from the difference between the mitotic index of the two biopsies. Otherwise the birth rate was calculated from the regression line. From this birth rate value, the cell production rate can be conveniently expressed in cells produced per 1000 cells per hour. All the cases studied were adenocarcinomas of the stomach of varying degrees of differentiation. RESULTS

Gastric carcinoma Three cases in the series were excluded because of the presence of anaphases in the postvincristine biopsies. In ten of the eleven cases studied more than two time periods were available, and the results are shown in Fig. 1. The lines are fitted for a rectangular age distribution. The cell production rates with 95% confidence intervals are given in Tables 1 and 2 together with potential doubling times also with 95% confidence intervals; these are calculated assuming that the errors conform to a Gamma distribution. It is apparent that in most cases the confidence limits for the parameters estimated are extremely wide; in several instances the lower limit for the cell production rate is negative; the lower confidence limit for the line is so wide that a line of negative slope could be drawn. Consequently the upper confidence limit for tpdis infinite. The estimates for the cell production rates range from 3.4 to 24.3 cells/1000 cells/hr. In five of the ten cases the lower confidence limit was infinite, and it is difficult to know how to interpret these estimates; it is much more realistic to quote the confidence interval rather than the mean value. In several cases the correlation coefficient is quite high, even over 0.9, but the confidence limits remain wide. This is because of the small number of time intervals available for study (see Discussion). The data do not provide much information about the absolute or relative rates of growth of

Nicholas A . Wright et al.

432

.

6-

0 I

Case 4

5-

3

L

’----:4

2

2 .

Case 2

l -

0 0

30

15

45

0

60

.

Case 5

-

30

60

Case 7

L

.

90

Case6

I

0

I

15

I

39

45

I

I

I

30

45

60

Case 7 (rnetastasisl

‘r

_/// 5t

.

4 -

-/:

120

3-

i

‘1

/@

I

0 L,

I

15

I

I

I

I

I

I

60

0

30

60

90

120

4t

Case 9

Case 8

O

--u 0

30

60

90

31/

4 k

2 I

I

I

I

I

I

I

0

15

30

45

60

75

0-L 0 3 0 6 0 9 0

O

60

120

I80

T m e ( r n d after vincristlne

FIG. 1. Mitotic accumulation graphs for human gastric carcinoma after intravenous infusion of vincristine. The lines are fitted assuming a rectangular age distribution.

these tumours. In view of the confidence limits obtained, it would be invidious to compare the birth rates of these tumours one with another, or with the birth rates obtained for control mucosae, particularly in those cases with only two readings (e.g. case 1 1 and all control mucosa: see Tables 1 and 2).

Cell proliferation in gastric carcinoma

433

TABLE1. Kinetic parameters in gastric carcinoma (with 95% confidence limits) Cell production rate (cells/1000 cells/hr)

Case No. 1 2 3 4 5 6 7 7 (metastasis 1) 7 (metastasis 2) 8 9 10 11

8.6 (2-2,15.1) 6.3 (00, 16.9) 8.1 (4.7, 11.5) 24.3 (4.1,45.4) 4.9 (2*8,6.9) 14.9 (1.3,28.6) 3.4 ~ ~ 5 8 . 8 ) 20.8 (00,5296) 27.4 11.4 (00,66.6) 5.5 (0.1, 10.2) 7.6 (00,27.8) 7.0

Potential doubling time (hr) (rectangular age distribution)

Potential doubling time (hr) (exponential age distribution)

116 (66,462) 157 ( 5 9 , ~ ) 123 (87,214) 41 (22,242) 204 (144, 353) 67 (35,766) 290 (17, 00) 48 (19,031 36 88 (15,00) 180 (98, 1072) 132 (36,00) 143

82 (47,325) 109 (41,001 88 (62, 153) 29 (16, 178) 143 (101,247) 48 (25, 550) 204 (12,00) 34 (13,001 62 (10,m) 126 (69,748) 94 (25, m) -

Normal mucosae In all instances where non-neoplastic tissues were analysed, only two post-vincristine biopsies were available. The preoperative endoscopic biopsies were inadequate, partly due to their small size, but mostly because of extreme difficulty in orientation. It is assumed that collection of metaphases was linear over the arrest period, and the birth rate is calculated from the difference between the readings. Confidence limits cannot be calculated in the usual manner, but from extrapolation from the tumour data where only three points were available (see case 10) it is clear that little information has been gathered. The cell production rates for normal pyloric mucosa varied from 8 to 16 cells/l000 cells/hr, with a mean value of 12.7 cells/l000 celldhr. Abnormal mucosae In one case, hyperplastic mucosae of both pyloric and acid-secreting types were available for analysis. These gave values of 30 and 22 cells/l000 cells/hr respectively. In two cases TABLE2. Cell production rates (CPR) and potential doubling time (I,,) in human gastric mucosa Normal mucosa* Case No.

C PR (cells/1000 cells/hr)

3

8.2

4 5 6

16.0 15.1 12.6

I,,

Hyperplastic mucosa

Intestinal metaplasia

C PR (cells/l000 cells/hr)

t,, (hr)

122

30.5. 22.5+

33* 44+

21.5

46

62 66 79

-

-

-

-

-

-

-

33.0

30

(hr)

-

* Pyloric type mucosa. Acid secreting mucosa.

CPR (cells/1000 cells/hr) t,, (hr)

434

Nicholas A . Wright et al.

intestinal metaplasia occurring in pyloric mucosa was analysed and gave values of 2 1 and 33 cells/ 1000 cells/hr respectively.

Hepatic metastases In one case the liver contained several surface metastases; in one larger metastasis three biopsies were obtained, but in a smaller one only two were taken. Cell production rates of 20 and 27 cells/1000 cells/hr were found in these metastases, respectively. For the larger metastasis the confidence intervals are again large (Table 1).

DISCUSSION The present results are unsatisfactory in indicating absolute values of proliferative rates, and also in any comparison of rates which might be made. The reason for the large confidence intervals lies in the few time periods which, for technical reasons, are available for study. In the absence of information about variation between samples, little can be said with safety about the mean. A major limiting factor in these experiments is the number of biopsies which can be taken. Compared with previously published work (Refsum & Berdal, 1967; Fulker et al., 1971; Camplejohn et al., 1973) in terms of numbers of readings, the present results are considerably more detailed on an individual basis. Consequently, claims made for accurate individual t,, values from these stathmokinetic studies are probably overstated. Of course, the variability between biopsies in the same case (so, for example, case 6) also contribute to the wide confidence limits. There is certainly some difficulty in deciding upon an optimum metaphase arrest period; one must ensure that the period is not long enough to incur errors due to metaphase degeneration (Aherne & Camplejohn, 1972), but at the same time a short arrest period may increase sampling errors. Of practical importance is the duration of the operative procedure, which will limit the time period available for study. In this work, biopsies were obtained until the blood supply to the stomach was compromised. Considering the normal pyloric mucosa, the mean cell production rate of 13 cells/1000 cells/hr does compare reasonably well with the value of 11 cells/1000 cells/hr which can be calculated from the data of Lipkin, Sherlock & Bell (1963). These workers calculated k , for the proliferative region only, but their data can be converted to include all cells from the gland neck to the surface, which was the method used in the present study. It should, however, be remembered that the present confidence limits are likely to be large, and also that linearity of metaphase collection has been assumed. In this latter context, Wright et al. (1973) have shown in the human small intestine that calculations made from two points compare well with those made from multiple biopsies after vincristine. As judged by mean values, cell production rates appear elevated in hyperplastic mucosae, and also in mucosa displaying intestinal metaplasia. This is particularly interesting, since intestinal metaplasia is held to be a premalignant lesion in the stomach, and the results are similar to those obtained by Croft (1972), who relied on mitotic counts for comparison. It is obvious that the gastric blood supply will be compromised during the operation, and this could conceivably lead to an underestimate of the proliferative rate. Camplejohn et al. (1973) found that linearity of metaphase collection was not maintained in excision specimens, and these have not been used in this study. This would not exclude a systematic decrease in

Cell proliferation in gastric carcinoma

43 5

mitotic indices; the apparent agreement of the present data with the 3H-TdR results of Lipkin

et ul. (1963) suggest that proliferative rates are of the correct order, at least in the normal mucosae. While it is reasonable to choose a rectangular or steady state age distribution for the normal and possibly the abnormal mucosae, the tumour presents a problem. At the time these studies were made, the tumours were considerably advanced. The decycling probability is probably large at this time, which would have the effect of lowering the proportion of young cells from that of an exponential age distribution: a high rate of cell loss at mitosis is also probable. These factors would combine to produce a near rectangular age distribution, and hence steady state equations are probably more appropriate. Three cases were excluded because of inadequate metaphase arrest, which was readily detectable by the presence of anaphases and telophases in the late post-vincristine samples. There was no doubt that vincristine was given in these cases, and the observation may only be explained on the basis of different sensitivities of different tumour cells. It is just possible that, in these cases, the duration of mitosis is very long, with the result that no metaphases became arrested. However, in all the other cases studied, metaphase arrest was complete, with no evidence of anaphase escape, as early as 30 min after vincristine injection; consequently the mitotic duration in the cases included is relatively short. It should perhaps be emphasized that, in the cases included, all biopsies taken after vincristine injection showed prophases and metaphases only, and there was no evidence of anaphase escape. Besides the problems itemized above, there are several other sources of error in this technique. There is obviously a good deal of kinetic heterogeneity within a single tumour; Camplejohn (unpublished results) showed that considerable variability in mitotic indices existed between various areas in single neuroblastomas and rectal carcinomas. It is further assumed that vincristine acts ub initio and also that cells are cycling at the same rate, when in fact there is probably a distribution of transit times through the cell cycle. Finally it is assumed that linear regression analysis is the optimal method of fitting the data (Whittal, Appleton & Wright, 1975). A further difficulty concerns the time of onset of metaphase action. In this study, in common with previous investigators, we have been obliged to take the native mitotic index as the first reading; in fact there may be a delay period before the onset of metaphase arrest, which would lead to a systematic underestimate of the cell production rate. It would be easy to say that cell production is higher in hepatic metastases than in the primary tumour; also that because cell production rates are lower in tumours than in the adjacent normal mucosa, carcinoma of the stomach would be expected to be resistant to radiotherapy and phase-specific cytotoxic chemotherapy, which would be expected to kill more normal cells than tumour cells. This latter contention is supported by some results obtained by Hoffman & Post (1967) who demonstrated that labelling and mitotic indices were lower in a gastric carcinoma than adjacent mucosa. However, it is not intended to give these data any more worth than they inherently possess by attempting to compare them with previous results, with the degree of differentiation of the tumour, or even with the normal mucosa, since the very wide confidence intervals do not allow accurate measurement or insight. Indeed, if this technique is to have any future, we must improve technical aspects in order to obtain many more biopsies during the course of the procedure.

436

Nicholas A . Wright et al. REFERENCES

AHERNE,W.A. & BUCK,P. (1971) The potential cell population doubling time in neuroblastoma and nephroblastoma. Brit. J. Cancer, 25, 29 1. AHERNE,W.A. & CAMPLEJOHN,R.S. (1972) On correcting the error due to metaphase degeneration in stathmokinetic studies. Exp. Cell Res. 74, 498. CAMPLEJOHN, R.S.,BONE, G. & AHERNE,W.A. (1973) Cell proliferation in human rectal carcinoma: a stathmokinetic study. Europ. J. Cancer, 9, 577. CROFT,D.N. (1972) Cell loss from gastric mucosa and its relevance to human gastric disease. Arch. Franc. Mot. appl. Dig.61, 3 10. FULKER,M.J., COOPER,E.H. & TANAKA, T. (1971) Proliferation and ultrastructure of papillary transitional cell carcinoma of the human bladder. Cancer, 27, 71. HOFFMAN, J. & POST,J. (1967) In uivo studies of DNA synthesis in human normal and tumour cells. Cancer Res. 27, 898. LIPKIN,M., SHERLOCK, P. & BELL,B. (1963) Cell proliferation kinetics in the intestinal tract of man. 11. Cell renewal in stomach, ileum, colon and rectum. Gastroenterology, 45, 72 1. MEYER,J.S. & DONALDSON, R.C. (1969) Growth kinetics of squamous cell carcinoma in man. Arch. Path. 87, 479. REFSUM,S.B. & BERDAL,P. (1967) Cell loss in malignant turnours in man. Europ. J. Cancer, 3,235. SMITH,R.S., THOMAS,D.B. & RICHES,A.C. (1974) Cell production in tumour isografts measured using vincristine and colcemid. Cell Tissue Kinet. 7 , 529. WHITTAL,P., APPLETON,D.R. & WRIGHT,N.A. (1975) The,robustness of the stathmokinetic technique. In: Mathematical Models in Cell Kinetics (ed. by A. J. Valleron). European Scientific Press. WRIGHT,N.A., MORLEY,A.R. & APPLETON,D.R. (1972) Variation in the duration of mitosis in the crypts of Lieberkuhn of the rat: a cytokinetic study using vincristine. Cell TissueKinet. 5 , 35 1. WRIGHT, N.A., WATSON,A.J., MORLEY,A.R., APPLETON,D.R., MARKS,J. & DOUGLAS,A.P. (1973) The cell cycle time in flat (avillous) mucosa of the human small intestine. Gut, 14,603.

An in vivo stathmokinetic study of cell proliferation in human gastric carcinoma and gastric mucosa.

Cell Tissue Kinet. (1977) 10, 429-436. A N I N VIVO STATHMOKINETIC STUDY OF CELL PROLIFERATION I N HUMAN GASTRIC CARCINOMA A N D GASTRIC MUCOSA NICHO...
460KB Sizes 0 Downloads 0 Views