1977, British Journal of Radiology, 50, 321-328
MAY
1977
The effect of radiation on tumour growth delay, cell survival and cure of the animal using a single tumour system By N. J. McNally, Ph.D., and P. W. Sheldon, M.I.Biol. Cancer Research Campaign Gray Laboratory, Mount Vernon Hospital, Northwood, Middlesex H A6 2RN (Received October, 1976 and in revised form January, 1977) ABSTRACT
The response of a murine tumour to single doses of X rays has been measured using three different assays—animal cure, cell survival in vitro after irradiation in vivo, and tumour growth delay. The dose to cure 50% of the animals, the TCD50, was 79.0 Gy. This was not affected by clamping the tumours to render all the cells hypoxic at the time of irradiation, implying that most of the cells in the tumour were hypoxic already. The enhancement ratio for the hypoxic cell sensitizer Ro-07-0582 was 2.1. The cell survival assay gave an enhancement ratio of 1.6 and an hypoxic fraction of 5%. The discrepancy in the estimates of the hypoxic fraction can be explained by the ability of the naturally hypoxic cells, but not the oxic ones, to recover from potentially lethal damage in vivo. Neither the cell survival assay nor the growth delay assay agreed with the TCD50 assay as to the effect of the hypoxic cell sensitizer, even allowing for recovery from potentially lethal damage. It is doubtful whether the measured survival curve would predict the measured TCD50.
The dose of radiation required to cure an animal of its tumour in the absence of an immune response is determined primarily by the sensitivity of the individual tumour cells to the radiation. The techniques at present available for directly measuring this sensitivity involve excision of tumours immediately after irradiation and assaying cell survival either in vitro (Barendsen and Broerse, 1969), or by injecting the cells into recipient mice (Hewitt and Wilson, 1959; Powers and Tolmach, 1963). If the survival curve parameters measured by such assays were those pertaining to the survival of the clonogenic cells when left in the tumour, one would expect that any agent which modifies the radiation response should do so equally for cell survival assayed directly, animal cure measured in terms of the TCD50 (the dose of radiation to cure 50% of a population of animals of their tumours) and tumour growth delay. This assumes that alterations in tumour growth after irradiation are primarily a reflection of the lethal effects of the radiation on the tumour cells (Thomlinson and Craddock, 1967; McNally, 1974). Previous studies on a rat fibrosarcoma have shown that the modifying effects of oxygen and radiation quality on tumour growth delay and on cell survival in vitro after irradiation in vivo were not the same (McNally, 1973, 1975). It was suggested that assays which involve removal of cells from their normal
environment after irradiation may not accurately reflect the course of events in the undisturbed tumour. In order to study further the relationship between tumour cell survival in situ and in vitro and the effects of modifying agents on the survival, we have used a transplantable tumour designated "MT", chosen because it is possible to measure the effects of radiation on the tumour using all three end-points, cure of the animal, tumour regrowth and cell survival in vitro after irradiation in vivo. We have compared the effects of two modifying procedures on the radiation response of this tumour. These procedures are: clamping the tumour to render all the cells hypoxic and injection of the hypoxic cell sensitizer Ro-07-0582 (Roche Products Ltd) before irradiation to sensitize the naturally hypoxic cells in the tumour. MATERIALS AND METHODS
The anaplastic " M T " tumour arose spontaneously in a WHT/Ht mouse in 1964, in Dr. H. B. Hewitt's animal colony, and has been maintained since then in the same inbred strain of mice. Its volume doubling time at the size used was about one day. To obtain experimental tumours, a tumour of 5-10 mm in diameter was excised aseptically from a mouse, cut into fragments of less than 1 mm 3 and implanted subcutaneously by a fine trochar over the sacral region of the backs of eight-week-old female mice. Tumours reaching a mean diameter of 5.0-6.0 mm between seven and 21 days after implantation were selected for treatment. In some of the cell survival assays tumours up to 8 mm in diameter and some of those which arose as doubles were used. Irradiation was by X rays from a Pantak X-ray set operated at 250 kV and 15 mA (H.V.L. 1.3 mm Cu.) with an additional filter of 0.25 mm Cu. and 1 mm Al. The dose rate at the tumours was 4.3 Gy min -1 (1 gray equals 100 rad). The sacral region of the back of a mouse was irradiated, without anaesthetic, as previously described (Sheldon and Hill, 1977). Tumours could be rendered hypoxic by applying a clamp ten minutes before the start of irradiation. The clamps used were similar to those described by
321
VOL.
50, No. 593 N. J. McNally and P. W. Sheldon
Denekamp and Harris (1975). The hypoxic cell sensitizer Ro-07-0582 [1-(2-hydroxy-3-methoxypropyl)-nitroimidazole] was used at a concentration of 1 mg per g body weight. It was administered by intraperitoneal injection 30 minutes before the start of irradiation. For cell survival studies, tumours were excised at various times after irradiation and single cell suspensions prepared as previously described (McNally, 1972). An aliquot of the cell suspension from an unirradiated tumour was then exposed in aerobic conditions to a dose of 80 Gy of 60Co y rays and 5 X 104 of these "feeder" cells in 3 ml of Eagle's Minimum Essential Medium plus 15% fetal calf serum and antibiotics were pipetted into each of a number of 50 mm plastic tissue-culture Petri dishes. Appropriate numbers of test cells were then added to the Petri dishes. Four replicate plates were used for each tumour cell suspension. The cells were incubated for 11 to 13 days at 37°C in a humidified atmosphere of 5% CO2 in air. Macroscopic colonies were then stained and counted and survival curves were constructed. The plating efficiency was 30 to 50%.
clamp was 79.0 Gy (range of standard error of the mean 78.2 to 79.9 Gy). Applying a clamp ten minutes before irradiation had little effect on the TCD50 (Fig. 1), implying that most of the tumour cells were normally hypoxic (the calculated TCD50 was 77.8 Gy, but this was not significantly different from that for irradiation in the absence of the clamp). If the animals were given 1 mg/g Ro-07-0582, 30 minutes before the start of irradiation, the TCD50 was reduced to 38.0 Gy, giving an "enhancement ratio"* of 2.1 (Fig. 1). Figure 2 shows the times for tumours to grow from a geometric mean diameter of 5.0-6.0 mm (irradiation size) to 10 mm after exposure to various doses of X rays, in the presence or absence of 1 mg/g Ro-07-0582. Each point represents data from at least 12 animals. The error bars represent i 1 standard error of the mean. The drug clearly sensitized the tumours to radiation. Table I shows the enhancement ratio as a function of the dose of radiation, measured from the smooth lines drawn through the * Enhancement ratio = X ray dose without the drug divided by X ray dose with the drug to cure half the mice.
RESULTS
The curves relating the probability of local tumour control to the radiation dose and the dose required to control 50% of the tumours (the TCD50) ± o n e standard error of the mean, were derived using a computer programme based on the logit method for maximum likelihood. Animals in which there were no recurrences 80 days after irradiation were scored as cured (Sheldon and Hill, 1977). The calculated curves are shown in Fig. 1 for the three conditions of irradiation used. The TCD50 in the absence of a
gAO '-a E E o
I
A-A
/
d>
O50-
1 mg/g 0582
220
clamped /
7
en
0 /
undamped (°
30
o/
• I
o
CD
E /
On
1 /I
6
o 100
o X-rays only • X-rays +0582
a
i 50
i ° i 60 70 Dose grays
0 i
i
80
90
0
1. Tumour control probability versus X ray dose. Tumours were irradiated either undamped (O ), clamped (A) or undamped 30 minutes after an intraperitoneal injection of 1 mg/g Ro-07-0582 (•). FIG.
i
i
20
i
i
l
i
U0 60 Dose grays
l
80
FIG. 2. The time for tumours to grow from treatment size (5.06.0 mm) to 10 mm diameter versus the X-ray dose. Tumours were irradiated in the absence of (o) or the presence of 1 mg/g Ro-07-0582 (#). The error bars represent standard errors of the mean.
322
MAY
1977
Tumour growth, cell survival and cure experimental points (as in Fig. 2) for three different final diameters, 9, 10 and 12 mm. Above a dose of about 40 Gy there was little difference in the measured enhancement ratios, but below this the enhancement ratio depended on the final size chosen. Figure 3 shows survival curves for tumour cells from tumours irradiated in vivo and then immediately excised and assayed in vitro. Tumours were irradiated either clamped or undamped in animals breathing air, or ten minutes after killing the animals by neck luxation. The clamping procedure was equally effective as killing the animals in rendering the tumour cells hypoxic (Fig. 3). The line through the survival values for hypoxic cells was fitted by eye,
TABLE I ENHANCEMENT RATIOS FOR RO-07-0582 (1 mg/g GIVEN 30 MINUTES BEFORE IRRADIATION) FOR VARIOUS FINAL TUMOUR DIAMETERS CHOSEN FOR THE REGROWTH ASSAY
Drug enhancement ratio final diameter X-ray dose in absence of drug (Gy) 10 20 30 40 50
9 mm
10 mm
1.25 1.22 1.34 1.49 1.72
1.56 1.47 1.49 1.52 1.69
12 mm 1.25 1.30 1.36 1.47 1.69
10'
10'
A Dead • Clamped o Undamped
10
° No drug • 1 mg/g 0582
10'
c o o
c o
CD C
CO
cn c
in
-4
10
e.r=21
-4
0
10
20 Dose grays
10
30
FIG. 3. Survival curves for the cells of tumours irradiated in vivo and then immediately excised and plated in vitro. Tumours were irradiated in dead animals (A), or in live animals clamped (#), or undamped (o ). The line through the points for irradiation of clamped tumours was fitted by eye. The dashed line is a theoretical line obtained from this line by assuming an oxygen enhancement ratio of 2.7. The line through the data points for undamped tumours was drawn assuming an hypoxic fraction of 5 % and the hatched region represents the range from 2 to 15% hypoxic cells.
—J
0
1
e.n=1-6 1
i
10 20 Dose grays
30
FIG. 4. Survival curves for tumour cells irradiated in vivo in the absence (o) or presence of 1 mg/g Ro-07-0582 (•) and immediately plated and assayed in vitro. The data points and line for irradiations in the absence of the drug are redrawn from Fig. 3. The solid line through the points for irradiation in the presence of the drug was drawn assuming the effect of the drug was to decrease the Do for the hypoxic cells by a factor of 1.6. The dashed line was drawn for a factor of 2.1.
323
VOL.
50, No. 593 N. J. McNally and P. W. Sheldon
giving a DQ of 2.6 Gy and an extrapolation number of 30. It is clear that clamping had a marked effect on the survival of the cells to radiation (Fig. 3). Clamping also reduced the scatter, so that it was possible to define the curve for tumours made hypoxic more accurately than that for undamped tumours. By assuming a value for the oxygen enhancement ratio (o.e.r.) it is possible to estimate the fraction of the cells that were hypoxic as determined by this in vitro assay. Assuming a value of 2.7 for the o.e.r. (giving the dashed line in Fig. 3), the solid line through the points for cells irradiated in undamped tumours was drawn assuming an hypoxic fraction of 5%. (In practice the calculated value of the hypoxic fraction is only weakly dependent on the value of the o.e.r. chosen.) The hatched area represents the region between hypoxic fractions of 2 and 15%. Figure 4 shows the effect of the hypoxic cell sensitizer given to undamped tumours at a concentration of 1 mg/g 30 minutes before irradiation. The measured survival values and the calculated line for an hypoxic fraction of 5% in normal tumours have been redrawn from Fig. 3 for comparison. The solid line through the survival values for irradiations in the presence of the drug was drawn assuming the effect of the drug was to decrease the DQ of the survival curve for hypoxic cells by a factor of 1.6 (the enhancement ratio). The dashed line is for an enhancement ratio of 2.1; this value is probably too high to account for the observations (Fig. 4). Dose-modification was assumed because this is the mode of action of the drug in vitro (Asquith et al., 1974). In these calculations it has been assumed that the terminal slope of the survival curve for the naturally hypoxic cells is the same as that of the survival curve for artificially hypoxic cells (Fig. 3). In virtually all those tumours for which it has been possible to measure full survival curves (for example Barendsen and Broerse, 1969; Hill et al., 1971; McNally, 1972) this has been found to be the case. Thus the lines drawn in Figs. 3 and 4 were calculated using only the eye-fitted line through the survival curve for clamped tumours (Fig. 3), the above assumption, with the additional assumption that the sensitizer only affects the DQ to the survival curve for hypoxic cells. In another series of experiments, tumours were excised at various times from one to 24 hours after irradiation and cell survival then assayed in vitro. The results of these experiments will be presented and discussed in the next section. DISCUSSION
Assuming the TCD50 depends primarily on the
radiosensitivity of the individual tumour cells, its high value (79.0 Gy), and the absence of a significant effect of clamping on the TCD 50 (Fig. 1), suggest that the hypoxic fraction of cells in this tumour is very high, well over 50%. If this is the case, it follows that within the accuracy of any of the methods used for measuring tumour response, the effects of radiation on the rather small well oxygenated fraction should be barely detectable, i.e. the tumour would appear to respond as a uniformly hypoxic population. Thus, dose-effect curves based on tumour growth delay should not be expected to demonstrate the biphasic shape indicative of the presence of a comparatively small fraction of hypoxic cells, and the sensitizing drug Ro-07-0582 could be expected to act as a "dose-modifier" as discussed above, giving the same enhancement ratio as that measured using the TCD50 assay. Figure 2 shows that, as expected, the dose-effect curve for tumour growth delay was not biphasic, but that the Ro-070582 enhancement ratio was less than that measured by the TCD50 assay. Even though it was about two at the lowest dose of radiation used in the presence of the drug (10 Gy), a smooth curve representing a constant enhancement ratio of two did not fit the data. (The experimental point corresponding to the dose of 10 Gy in the presence of the drug was determined in a separate experiment whereas all the others were obtained concurrently.) The smooth curve shown in Fig. 2 gave an enhancement ratio of about 1.5, tending toward the value for the TCD50 assay at high doses. The enhancement ratio depended on the final size of the tumour chosen in the regrowth assay (Table I), but no matter what final size was chosen it was low at low radiation doses, increasing toward the TCD50 value at high doses. While the regrowth assay does suggest a cell population with uniform radiation sensitivity, at lower X ray doses it does not give the same drug enhancement ratio as the TCD50 assay. It is possible that only at large doses of radiation do the lethal effects of the radiation on the cells dominate the response in this tumour, with its relatively large TCD50, whereas at lower doses factors such as division delay and alterations in post-irradiation growth kinetics may be significant in determining the pattern of regrowth. The results of the TCD50 assay would lead to the prediction that there should be little effect of clamping the tumour on the survival curve constructed from in vitro assays and that the sensitizer should be essentially dose-modifying. For both predictions this was clearly not the case (Figs. 3 and 4). The hypoxic fraction deduced from the cell
324
MAY
1977
Tumour growth, cell survival and cure the oxygenated cells so the survivors would have been hypoxic at the time of irradiation, and 19 Gy to clamped tumours in which normally well oxygenated cells would have been rendered hypoxic so they would not have been selectively killed. For cells from the clamped tumours there was no change in survival up to eight hours after irradiation (Fig. 5A). For cells from undamped tumours ("naturally" hypoxic cells) there was a prompt increase in survival by a factor of about four within the first two to three hours, with probably little further increase in survival up to 16 hours later (Fig. 5B). This increase in survival represents recovery from potentially lethal damage (PLD) in the interval between irradiation and excision of the tumours (Hahn and Little, 1972; Little et al., 1973). In this tumour this form of recovery was evidently restricted to cells that were naturally hypoxic rather than those made acutely hypoxic by the application of a clamp. If this were not the case one should also expect to see an increase in survival with time for cells from tumours .•./JVS=0.69 that were clamped at the time of irradiation. As Substituting for S from equation 1 and solving for D Fig. 5A shows, this was not the case. the TCD50, gives:
survival curves was about 5% and the drug enhancement ratio was about 1.6 (instead of 2.1 as in the TCD 50 assay). Bearing in mind these discrepancies between the TCD50 and cell survival assays one can ask whether the measured survival curves would give a TCD50 of about 79.0 Gy. In order to calculate a TCD50 from the survival curves a number of assumptions have to be made. We assume: (a) The cell survival curve at high doses has the form S=ne~DID0 . . . (1) where n is the extrapolation number (b) A tumour recurs if one cell survives, the probability of curing an animal being given by p=e~NS where JV is the number of cells in the tumour capable of repopulating it and S is the surviving fraction after a dose D Gy. If the tumour contains a fraction/of hypoxic cells, then for large single doses, at the TCD50
loge °_^1
(2)
From the survival curves of Fig. 3, Z>o=2.6O Gy, #=30 and/=0.05. The tumours were irradiated at a mean diameter of about 5.5 mm so that a reasonable value for N is probably between 108 and 109 cells per tumour. Taking these as the extreme values gives 49.91 Gy < Z U 55.89 Gy In other words, the survival curves do not predict the measured TCD50. A value of N over 1012 would be needed to give a TCD 50 of 79.0 Gy. Thus, the in vitro assay disagrees with the animal cure assay in regard to the hypoxic fraction, the d^ug enhancement ratio and the TCD50 value. There is, however, an explanation we wish to propose which can, in part, account for these discrepancies.
19 Gy clamped • t
0-O_On _o_0_ _?
O
.«2 I
I
I
I
I
I
t hours c
Effect of potentially lethal damage For the cell survival assay the tumours were removed from the animals immediately after irradiation. It is possible that this rapid change in the cells' environment immediately after irradiation could affect their ability to survive the treatment. We therefore did a further series of experiments as indicated above, in which tumours were irradiated and excised at various times thereafter. Figure 5 shows the surviving fraction of cells assayed in vitro as a function of the time after irradiation that the tumours were excised. The doses were 15 Gy to undamped tumours, a dose large enough to kill all
B o 10
8 15 Gy
-O 2x10 3
1
1
0
2
1
t
i
1
L 6 hours
1
1
8
•
" /
/
•
16
FIG. 5. The surviving fraction of tumour cells assayed in vitro as a function of the time after exposure to a single dose of X rays at which the tumours were excised.
325
VOL.
50, No. 593 N. J. McNally and P. W. Sheldon
t = 6 hours o No drug • 1 mg/g 0582
e.r=1-6 0
20 Dose grays
10
0
30
10 20 Dose grays
30
FIG. 6. Survival curves for tumour cells from tumours excised immediately after irradiation (o ) or six (•) or 24 (A) hours after irradiation. The lines through the six and 24 hour data points were drawn assuming the effect of recovery from PLD was to increase the Do for hypoxic cells by a factor of 1.3 or 1.4.
The effect of recovery from PLD on the enhancement ratio for Ro-07-0582. Tumours were excised six hours after irradiation in the presence or absence of 1 mg/g of the drug. The lines were drawn assuming a factor of 1.3 for recovery from PLD and drug enhancement ratios of 1.6 or 2.1.
Figure 6 shows the effects of recovery from PLD on the survival curves for cells from tumours irradiated in the absence of the sensitizer and excised six or 24 hours later. (The survival curve for cells from tumours excised immediately after irradiation is included for comparison.) The lines were drawn assuming the effect of recovery from PLD was to increase the Do for hypoxic cells by a factor of 1.3 from 2.6 Gy to 3.38 Gy, or by a factor of 1.4 to 3.64 Gy. The factor 1.4 probably represents an upper limit for the effect of recovery from PLD on the survival curve. This recovery had probably reached its maximum by six hours (Fig. 6). While it is possible to draw a line through the zero hour result, parallel to those for six and 24 hours, this is probably not valid for the reasons already given when the results shown in Figs. 3 and 4 were described.
Figure 7 shows the effect of recovery from PLD on the enhancement ratio for Ro-07-0582. The tumours were excised six hours after irradiation with or without injection of 1 mg/g of the drug. The lines were drawn assuming a factor of 1.3 for recovery from PLD and sensitizing enhancement ratios for the drug of 1.6 and 2.1. As Fig. 7 shows, it is clear that recovery from PLD occurred whether or not the sensitizer was present and did not affect the sensitizer enhancement ratio. The value of 2.1 is much too large to account for the observations. The observation that only the naturally hypoxic tumour cells were capable of recovery from PLD, and not the oxic ones made hypoxic by a clamp, can explain the absence of an effect of a clamp on the TCD50 even though the hypoxic fraction was only 5%, if we can make the reasonable assumption that
FIG.-8
natural lyN hypoxic
.oxic x2-7
x2-7 natural lyN hypoxic^/
-8
i
20 Dose
40 grays
60
20 Dose
40 grays
i
60
FIG. 8. Theoretical survival curves for a mixture of oxic and hypoxic cells. The solid lines represent the survival curves for the separate components (oxic cells, naturally hypoxic, or oxic made hypoxic). The dashed lines represent the resultant survival curves obtained by adding the appropriate component curves. A is drawn assuming no recovery from PLD. B is drawn assuming that only the naturally hypoxic cells can recover from PLD. The lines have been drawn using experimentally determined survival curve parameters (see text).
the oxic cells would not have been able to recover from PLD even if the clamp had not been applied during irradiation. Figure 8 shows theoretical survival curves for a mixture of oxic and hypoxic cells. The survival curve parameters used to draw these curves were those derived from the results of the present experiments. Figure 8A shows the survival curves when the naturally hypoxic cells do not recover from PLD. The parameters used were: D$ for hypoxic cells 2.6 Gy, oxygen enhancement ratio 2.7, survival curve extrapolation number 30, hypoxic fraction 5%. As Fig. 8A shows, application of a clamp displaces the resultant survival curve upwards relative to that for undamped cells by an amount proportional to the hypoxic fraction. This displacement occurs at all levels of survival. In Fig. 8B, as well as being protected by hypoxia, the naturally hypoxic cells (but not the oxic ones) are assumed to be capable of recovery from PLD. The survival curve parameters are the same as in Fig. 8A except that the DQ for the naturally hypoxic cell survival curve, relative to that for oxic cells made hypoxic, has been increased by a factor of 1.3 to 3.38 Gy, to
allow for recovery from PLD. The resultant survival curve when a clamp is applied is obtained by adding the "oxic X 2.7" curve and the "naturally hypoxic" curve as in Fig. 8A, except that this time once survival is significantly less than that indicated by the point P (Fig. 8B), the survival curve for naturally hypoxic cells will dictate the response regardless of whether or not a clamp is applied. The presence of the clamp will only increase the dose of radiation beyond which the naturally hypoxic cells dominate the response (Fig. 8B). The value of the dose of radiation to give a surviving fraction represented by the point P in Fig. 8B can be calculated using the simple model already outlined. It will be determined by the equation where r is the factor by which the Do for naturally occurring hypoxic cells must be increased over that for oxic ones made hypoxic by the clamp, due to recovery from PLD. For the survival curves shown in Figs. 4, 6 and 8B, D has a value of 35 Gy. Thus, since the TCD50 was 79.0 Gy, one would not expect it to be affected by clamping the tumours.
327
i
VOL.
50, No. 593 N. J. McNally and P. W. Sheldon
Two discrepancies remain between the two (direct) methods of assaying cell survival (TCD 50 and in vitro assay). First, the enhancement ratio for Ro-07-0582 was larger in the TCD 5 0 assay (2.1) than in the cell survival assay (1.6); this difference is probably independent of recovery from PLD (Fig. 6). Second, the TCD50 was much larger than the cell survival curves would predict. If we now allow for recovery from PLD so that the value for DQ in equation 2 is increased by a factor 1.3 from 2.6 Gy to 3.38 Gy, then for values of N between 108 and 109 cells per tumour, 64.88 Gy^D ^72.67 Gy This is still less than the measured TCD50. A factor of 1.4 for the effect of recovery from PLD on the Do, which probably represents an upper limit for the experimental observation (Fig. 6), would give a TCD50 in the range 69.89 Gy to 78.29 Gy. The calculations of the TCD50 based on equation 2 have assumed that all naturally hypoxic cells in the tumour are capable of repopulating it. This is unlikely to be the case, so that even assuming a factor of 1.4 for the effect of recovery from PLD, it is doubtful if the measured survival curves would predict the highTCD 50 . These discrepancies have, therefore, not been fully accounted for. The drug might be having more effect in the TCD50 assay due to direct toxicity on the tumour cells. However, by six hours its concentration in the tumours will have become quite low (Flockhart, personal communication), and this time of contact with the cells did not affect the enhancement ratio measured in vitro (Fig. 7). Also, giving the drug after irradiation had no effect on theTCDso, suggesting no cytotoxic action (Sheldon, unpublished data). The large TCD50 could be due to a greater ability of the hypoxic cells in the tumour to absorb radiation damage as sublethal when left in situ than when plated in vitro. It is known that the degree of intercellular contact between cells in vitro can affect their ability to absorb sublethal damage (Durand and Sutherland, 1972). It is clear from these results that the fate of irradiated cells in this tumour depends critically on the immediate post-irradiation environment. It follows that, without knowing the effects of this environment, incorrect interpretations of the results may be made. From a practical point of view it may not matter whether one assumes either that the hypoxic fraction is large or that the hypoxic cells are capable of recovery from PLD. However, in a
fractionated course of radiotherapy, or in circumstances where drugs which could inhibit the cells' ability to recover from PLD are used in conjunction with radiation, the use of a single assay of radiation response alone could lead to erroneous conclusions. Without the use of several different methods of assay in the same tumour, the details of its radiation response, for instance the role of recovery from potentially lethal damage, may not be elucidated. ACKNOWLEDGMENTS
We thank Miss A. Walder, Miss A. Marriott and Mrs. S. Bull for production and care of the animals; Miss S. Hill and Mrs. J. de Ronde for expert technical assistance and Dr. J. F. Fowler for helpful discussions. REFERENCES ASQUITH, J. C , WATTS, M. E., PATEL, K., SMITHEN, C. E.,
and ADAMS, G. E., 1974. Electron-affinic sensitization. V. Radiosensitization of hypoxic bacteria and mammalian cells in vitro by some nitroimidazoles and nitropyrazoles. Radiation Research, 60,108-118. BARENDSEN, G. W., and BROERSE, J. J., 1969. Experimental
radiotherapy of a rat rhabdomyosarcoma with 15 MeV neutrons and 300 kV X rays. I. Effect of single doses. European Journal of Cancer, 5, 373-391. DENEKAMP, J., and HARRIS, S., 1975. Tests of two electron-
affinic radiosensitizers in vivo using regrowth of an experimental carcinoma. Radiation Research, 61,191-203. DURAND, R. E., and SUTHERLAND, R. M., 1972. Effects of
intercellular contact on repair of radiation damage. Experimental Cell Research, 71, 75-80. HAHN, G. M., and LITTLE, J. B., 1972. Plateau phase
cultures of mammalian cells. Current Topics in Radiation Research Quarterly, 8, 39-83. HEWITT, H. B., and WILSON, C. W., 1959. A survival curve
for mammalian leukaemia cells irradiated in vivo. British Journal of Cancer, 13, 69-75. HILL, R. P., BUSH, R. S., and YEUNG, P., 1971. The effect of
anaemia on the fraction of hypoxic cells in an experimental tumour. British Journal of Radiology, 44, 299-304. LITTLE, J. B., HAHN, G. M., FRINDEL, E., and TUBIANA, M.,
1973. Repair of potentially lethal radiation damage in vitro and in vivo. Radiology, 106, 689-694. MCNALLY, N. J., 1972. Recovery from sublethal damage by hypoxic tumour cells in vivo. British Journal of Radiology, 45,116-120. MCNALLY, N. J., 1973. A comparison of the effects of radiation on tumour growth delay and cell survival. The effect of oxygen. British Journal of Radiology, 46, 450455. MCNALLY, N. J., 1974. Tumour growth delay and cell survival in situ. British Journal of Radiology, 47, 510-511. MCNALLY, N. J., 1975. A comparison of the effects of radiation on tumour growth delay and cell survival. The effect of radiation quality. British Journal of Radiology, 48, 141-145. POWERS, W. E., and TOLMACH, L. J. 1963. A multicom-
ponent X ray survival curve for mouse lymphosarcoma cells irradiated in vivo. Nature, 197, 710-711. SHELDON, P., and HILL, S. A., 1977. The effect of hypoxic
cell radiosensitizing drugs on local control by single doses of X rays of a transplanted anaplastic tumour in mice. British Journal of Cancer (in press). THOMLINSON, R. H., and CRADDOCK, E. A., 1967. The gross
328
response of an experimental tumour to single doses of X rays. British Journal of Cancer, 21 ,\ 08-125.
MAY
1977
The effect of radiation on tumour growth delay, cell survival and cure of the animal using a single tumour system By N. J. McNally, Ph.D., and P. W. Sheldon, M.I.Biol. Cancer Research Campaign Gray Laboratory, Mount Vernon Hospital, Northwood, Middlesex H A6 2RN (Received October, 1976 and in revised form January, 1977) ABSTRACT
The response of a murine tumour to single doses of X rays has been measured using three different assays—animal cure, cell survival in vitro after irradiation in vivo, and tumour growth delay. The dose to cure 50% of the animals, the TCD50, was 79.0 Gy. This was not affected by clamping the tumours to render all the cells hypoxic at the time of irradiation, implying that most of the cells in the tumour were hypoxic already. The enhancement ratio for the hypoxic cell sensitizer Ro-07-0582 was 2.1. The cell survival assay gave an enhancement ratio of 1.6 and an hypoxic fraction of 5%. The discrepancy in the estimates of the hypoxic fraction can be explained by the ability of the naturally hypoxic cells, but not the oxic ones, to recover from potentially lethal damage in vivo. Neither the cell survival assay nor the growth delay assay agreed with the TCD50 assay as to the effect of the hypoxic cell sensitizer, even allowing for recovery from potentially lethal damage. It is doubtful whether the measured survival curve would predict the measured TCD50.
The dose of radiation required to cure an animal of its tumour in the absence of an immune response is determined primarily by the sensitivity of the individual tumour cells to the radiation. The techniques at present available for directly measuring this sensitivity involve excision of tumours immediately after irradiation and assaying cell survival either in vitro (Barendsen and Broerse, 1969), or by injecting the cells into recipient mice (Hewitt and Wilson, 1959; Powers and Tolmach, 1963). If the survival curve parameters measured by such assays were those pertaining to the survival of the clonogenic cells when left in the tumour, one would expect that any agent which modifies the radiation response should do so equally for cell survival assayed directly, animal cure measured in terms of the TCD50 (the dose of radiation to cure 50% of a population of animals of their tumours) and tumour growth delay. This assumes that alterations in tumour growth after irradiation are primarily a reflection of the lethal effects of the radiation on the tumour cells (Thomlinson and Craddock, 1967; McNally, 1974). Previous studies on a rat fibrosarcoma have shown that the modifying effects of oxygen and radiation quality on tumour growth delay and on cell survival in vitro after irradiation in vivo were not the same (McNally, 1973, 1975). It was suggested that assays which involve removal of cells from their normal
environment after irradiation may not accurately reflect the course of events in the undisturbed tumour. In order to study further the relationship between tumour cell survival in situ and in vitro and the effects of modifying agents on the survival, we have used a transplantable tumour designated "MT", chosen because it is possible to measure the effects of radiation on the tumour using all three end-points, cure of the animal, tumour regrowth and cell survival in vitro after irradiation in vivo. We have compared the effects of two modifying procedures on the radiation response of this tumour. These procedures are: clamping the tumour to render all the cells hypoxic and injection of the hypoxic cell sensitizer Ro-07-0582 (Roche Products Ltd) before irradiation to sensitize the naturally hypoxic cells in the tumour. MATERIALS AND METHODS
The anaplastic " M T " tumour arose spontaneously in a WHT/Ht mouse in 1964, in Dr. H. B. Hewitt's animal colony, and has been maintained since then in the same inbred strain of mice. Its volume doubling time at the size used was about one day. To obtain experimental tumours, a tumour of 5-10 mm in diameter was excised aseptically from a mouse, cut into fragments of less than 1 mm 3 and implanted subcutaneously by a fine trochar over the sacral region of the backs of eight-week-old female mice. Tumours reaching a mean diameter of 5.0-6.0 mm between seven and 21 days after implantation were selected for treatment. In some of the cell survival assays tumours up to 8 mm in diameter and some of those which arose as doubles were used. Irradiation was by X rays from a Pantak X-ray set operated at 250 kV and 15 mA (H.V.L. 1.3 mm Cu.) with an additional filter of 0.25 mm Cu. and 1 mm Al. The dose rate at the tumours was 4.3 Gy min -1 (1 gray equals 100 rad). The sacral region of the back of a mouse was irradiated, without anaesthetic, as previously described (Sheldon and Hill, 1977). Tumours could be rendered hypoxic by applying a clamp ten minutes before the start of irradiation. The clamps used were similar to those described by
321
VOL.
50, No. 593 N. J. McNally and P. W. Sheldon
Denekamp and Harris (1975). The hypoxic cell sensitizer Ro-07-0582 [1-(2-hydroxy-3-methoxypropyl)-nitroimidazole] was used at a concentration of 1 mg per g body weight. It was administered by intraperitoneal injection 30 minutes before the start of irradiation. For cell survival studies, tumours were excised at various times after irradiation and single cell suspensions prepared as previously described (McNally, 1972). An aliquot of the cell suspension from an unirradiated tumour was then exposed in aerobic conditions to a dose of 80 Gy of 60Co y rays and 5 X 104 of these "feeder" cells in 3 ml of Eagle's Minimum Essential Medium plus 15% fetal calf serum and antibiotics were pipetted into each of a number of 50 mm plastic tissue-culture Petri dishes. Appropriate numbers of test cells were then added to the Petri dishes. Four replicate plates were used for each tumour cell suspension. The cells were incubated for 11 to 13 days at 37°C in a humidified atmosphere of 5% CO2 in air. Macroscopic colonies were then stained and counted and survival curves were constructed. The plating efficiency was 30 to 50%.
clamp was 79.0 Gy (range of standard error of the mean 78.2 to 79.9 Gy). Applying a clamp ten minutes before irradiation had little effect on the TCD50 (Fig. 1), implying that most of the tumour cells were normally hypoxic (the calculated TCD50 was 77.8 Gy, but this was not significantly different from that for irradiation in the absence of the clamp). If the animals were given 1 mg/g Ro-07-0582, 30 minutes before the start of irradiation, the TCD50 was reduced to 38.0 Gy, giving an "enhancement ratio"* of 2.1 (Fig. 1). Figure 2 shows the times for tumours to grow from a geometric mean diameter of 5.0-6.0 mm (irradiation size) to 10 mm after exposure to various doses of X rays, in the presence or absence of 1 mg/g Ro-07-0582. Each point represents data from at least 12 animals. The error bars represent i 1 standard error of the mean. The drug clearly sensitized the tumours to radiation. Table I shows the enhancement ratio as a function of the dose of radiation, measured from the smooth lines drawn through the * Enhancement ratio = X ray dose without the drug divided by X ray dose with the drug to cure half the mice.
RESULTS
The curves relating the probability of local tumour control to the radiation dose and the dose required to control 50% of the tumours (the TCD50) ± o n e standard error of the mean, were derived using a computer programme based on the logit method for maximum likelihood. Animals in which there were no recurrences 80 days after irradiation were scored as cured (Sheldon and Hill, 1977). The calculated curves are shown in Fig. 1 for the three conditions of irradiation used. The TCD50 in the absence of a
gAO '-a E E o
I
A-A
/
d>
O50-
1 mg/g 0582
220
clamped /
7
en
0 /
undamped (°
30
o/
• I
o
CD
E /
On
1 /I
6
o 100
o X-rays only • X-rays +0582
a
i 50
i ° i 60 70 Dose grays
0 i
i
80
90
0
1. Tumour control probability versus X ray dose. Tumours were irradiated either undamped (O ), clamped (A) or undamped 30 minutes after an intraperitoneal injection of 1 mg/g Ro-07-0582 (•). FIG.
i
i
20
i
i
l
i
U0 60 Dose grays
l
80
FIG. 2. The time for tumours to grow from treatment size (5.06.0 mm) to 10 mm diameter versus the X-ray dose. Tumours were irradiated in the absence of (o) or the presence of 1 mg/g Ro-07-0582 (#). The error bars represent standard errors of the mean.
322
MAY
1977
Tumour growth, cell survival and cure experimental points (as in Fig. 2) for three different final diameters, 9, 10 and 12 mm. Above a dose of about 40 Gy there was little difference in the measured enhancement ratios, but below this the enhancement ratio depended on the final size chosen. Figure 3 shows survival curves for tumour cells from tumours irradiated in vivo and then immediately excised and assayed in vitro. Tumours were irradiated either clamped or undamped in animals breathing air, or ten minutes after killing the animals by neck luxation. The clamping procedure was equally effective as killing the animals in rendering the tumour cells hypoxic (Fig. 3). The line through the survival values for hypoxic cells was fitted by eye,
TABLE I ENHANCEMENT RATIOS FOR RO-07-0582 (1 mg/g GIVEN 30 MINUTES BEFORE IRRADIATION) FOR VARIOUS FINAL TUMOUR DIAMETERS CHOSEN FOR THE REGROWTH ASSAY
Drug enhancement ratio final diameter X-ray dose in absence of drug (Gy) 10 20 30 40 50
9 mm
10 mm
1.25 1.22 1.34 1.49 1.72
1.56 1.47 1.49 1.52 1.69
12 mm 1.25 1.30 1.36 1.47 1.69
10'
10'
A Dead • Clamped o Undamped
10
° No drug • 1 mg/g 0582
10'
c o o
c o
CD C
CO
cn c
in
-4
10
e.r=21
-4
0
10
20 Dose grays
10
30
FIG. 3. Survival curves for the cells of tumours irradiated in vivo and then immediately excised and plated in vitro. Tumours were irradiated in dead animals (A), or in live animals clamped (#), or undamped (o ). The line through the points for irradiation of clamped tumours was fitted by eye. The dashed line is a theoretical line obtained from this line by assuming an oxygen enhancement ratio of 2.7. The line through the data points for undamped tumours was drawn assuming an hypoxic fraction of 5 % and the hatched region represents the range from 2 to 15% hypoxic cells.
—J
0
1
e.n=1-6 1
i
10 20 Dose grays
30
FIG. 4. Survival curves for tumour cells irradiated in vivo in the absence (o) or presence of 1 mg/g Ro-07-0582 (•) and immediately plated and assayed in vitro. The data points and line for irradiations in the absence of the drug are redrawn from Fig. 3. The solid line through the points for irradiation in the presence of the drug was drawn assuming the effect of the drug was to decrease the Do for the hypoxic cells by a factor of 1.6. The dashed line was drawn for a factor of 2.1.
323
VOL.
50, No. 593 N. J. McNally and P. W. Sheldon
giving a DQ of 2.6 Gy and an extrapolation number of 30. It is clear that clamping had a marked effect on the survival of the cells to radiation (Fig. 3). Clamping also reduced the scatter, so that it was possible to define the curve for tumours made hypoxic more accurately than that for undamped tumours. By assuming a value for the oxygen enhancement ratio (o.e.r.) it is possible to estimate the fraction of the cells that were hypoxic as determined by this in vitro assay. Assuming a value of 2.7 for the o.e.r. (giving the dashed line in Fig. 3), the solid line through the points for cells irradiated in undamped tumours was drawn assuming an hypoxic fraction of 5%. (In practice the calculated value of the hypoxic fraction is only weakly dependent on the value of the o.e.r. chosen.) The hatched area represents the region between hypoxic fractions of 2 and 15%. Figure 4 shows the effect of the hypoxic cell sensitizer given to undamped tumours at a concentration of 1 mg/g 30 minutes before irradiation. The measured survival values and the calculated line for an hypoxic fraction of 5% in normal tumours have been redrawn from Fig. 3 for comparison. The solid line through the survival values for irradiations in the presence of the drug was drawn assuming the effect of the drug was to decrease the DQ of the survival curve for hypoxic cells by a factor of 1.6 (the enhancement ratio). The dashed line is for an enhancement ratio of 2.1; this value is probably too high to account for the observations (Fig. 4). Dose-modification was assumed because this is the mode of action of the drug in vitro (Asquith et al., 1974). In these calculations it has been assumed that the terminal slope of the survival curve for the naturally hypoxic cells is the same as that of the survival curve for artificially hypoxic cells (Fig. 3). In virtually all those tumours for which it has been possible to measure full survival curves (for example Barendsen and Broerse, 1969; Hill et al., 1971; McNally, 1972) this has been found to be the case. Thus the lines drawn in Figs. 3 and 4 were calculated using only the eye-fitted line through the survival curve for clamped tumours (Fig. 3), the above assumption, with the additional assumption that the sensitizer only affects the DQ to the survival curve for hypoxic cells. In another series of experiments, tumours were excised at various times from one to 24 hours after irradiation and cell survival then assayed in vitro. The results of these experiments will be presented and discussed in the next section. DISCUSSION
Assuming the TCD50 depends primarily on the
radiosensitivity of the individual tumour cells, its high value (79.0 Gy), and the absence of a significant effect of clamping on the TCD 50 (Fig. 1), suggest that the hypoxic fraction of cells in this tumour is very high, well over 50%. If this is the case, it follows that within the accuracy of any of the methods used for measuring tumour response, the effects of radiation on the rather small well oxygenated fraction should be barely detectable, i.e. the tumour would appear to respond as a uniformly hypoxic population. Thus, dose-effect curves based on tumour growth delay should not be expected to demonstrate the biphasic shape indicative of the presence of a comparatively small fraction of hypoxic cells, and the sensitizing drug Ro-07-0582 could be expected to act as a "dose-modifier" as discussed above, giving the same enhancement ratio as that measured using the TCD50 assay. Figure 2 shows that, as expected, the dose-effect curve for tumour growth delay was not biphasic, but that the Ro-070582 enhancement ratio was less than that measured by the TCD50 assay. Even though it was about two at the lowest dose of radiation used in the presence of the drug (10 Gy), a smooth curve representing a constant enhancement ratio of two did not fit the data. (The experimental point corresponding to the dose of 10 Gy in the presence of the drug was determined in a separate experiment whereas all the others were obtained concurrently.) The smooth curve shown in Fig. 2 gave an enhancement ratio of about 1.5, tending toward the value for the TCD50 assay at high doses. The enhancement ratio depended on the final size of the tumour chosen in the regrowth assay (Table I), but no matter what final size was chosen it was low at low radiation doses, increasing toward the TCD50 value at high doses. While the regrowth assay does suggest a cell population with uniform radiation sensitivity, at lower X ray doses it does not give the same drug enhancement ratio as the TCD50 assay. It is possible that only at large doses of radiation do the lethal effects of the radiation on the cells dominate the response in this tumour, with its relatively large TCD50, whereas at lower doses factors such as division delay and alterations in post-irradiation growth kinetics may be significant in determining the pattern of regrowth. The results of the TCD50 assay would lead to the prediction that there should be little effect of clamping the tumour on the survival curve constructed from in vitro assays and that the sensitizer should be essentially dose-modifying. For both predictions this was clearly not the case (Figs. 3 and 4). The hypoxic fraction deduced from the cell
324
MAY
1977
Tumour growth, cell survival and cure the oxygenated cells so the survivors would have been hypoxic at the time of irradiation, and 19 Gy to clamped tumours in which normally well oxygenated cells would have been rendered hypoxic so they would not have been selectively killed. For cells from the clamped tumours there was no change in survival up to eight hours after irradiation (Fig. 5A). For cells from undamped tumours ("naturally" hypoxic cells) there was a prompt increase in survival by a factor of about four within the first two to three hours, with probably little further increase in survival up to 16 hours later (Fig. 5B). This increase in survival represents recovery from potentially lethal damage (PLD) in the interval between irradiation and excision of the tumours (Hahn and Little, 1972; Little et al., 1973). In this tumour this form of recovery was evidently restricted to cells that were naturally hypoxic rather than those made acutely hypoxic by the application of a clamp. If this were not the case one should also expect to see an increase in survival with time for cells from tumours .•./JVS=0.69 that were clamped at the time of irradiation. As Substituting for S from equation 1 and solving for D Fig. 5A shows, this was not the case. the TCD50, gives:
survival curves was about 5% and the drug enhancement ratio was about 1.6 (instead of 2.1 as in the TCD 50 assay). Bearing in mind these discrepancies between the TCD50 and cell survival assays one can ask whether the measured survival curves would give a TCD50 of about 79.0 Gy. In order to calculate a TCD50 from the survival curves a number of assumptions have to be made. We assume: (a) The cell survival curve at high doses has the form S=ne~DID0 . . . (1) where n is the extrapolation number (b) A tumour recurs if one cell survives, the probability of curing an animal being given by p=e~NS where JV is the number of cells in the tumour capable of repopulating it and S is the surviving fraction after a dose D Gy. If the tumour contains a fraction/of hypoxic cells, then for large single doses, at the TCD50
loge °_^1
(2)
From the survival curves of Fig. 3, Z>o=2.6O Gy, #=30 and/=0.05. The tumours were irradiated at a mean diameter of about 5.5 mm so that a reasonable value for N is probably between 108 and 109 cells per tumour. Taking these as the extreme values gives 49.91 Gy < Z U 55.89 Gy In other words, the survival curves do not predict the measured TCD50. A value of N over 1012 would be needed to give a TCD 50 of 79.0 Gy. Thus, the in vitro assay disagrees with the animal cure assay in regard to the hypoxic fraction, the d^ug enhancement ratio and the TCD50 value. There is, however, an explanation we wish to propose which can, in part, account for these discrepancies.
19 Gy clamped • t
0-O_On _o_0_ _?
O
.«2 I
I
I
I
I
I
t hours c
Effect of potentially lethal damage For the cell survival assay the tumours were removed from the animals immediately after irradiation. It is possible that this rapid change in the cells' environment immediately after irradiation could affect their ability to survive the treatment. We therefore did a further series of experiments as indicated above, in which tumours were irradiated and excised at various times thereafter. Figure 5 shows the surviving fraction of cells assayed in vitro as a function of the time after irradiation that the tumours were excised. The doses were 15 Gy to undamped tumours, a dose large enough to kill all
B o 10
8 15 Gy
-O 2x10 3
1
1
0
2
1
t
i
1
L 6 hours
1
1
8
•
" /
/
•
16
FIG. 5. The surviving fraction of tumour cells assayed in vitro as a function of the time after exposure to a single dose of X rays at which the tumours were excised.
325
VOL.
50, No. 593 N. J. McNally and P. W. Sheldon
t = 6 hours o No drug • 1 mg/g 0582
e.r=1-6 0
20 Dose grays
10
0
30
10 20 Dose grays
30
FIG. 6. Survival curves for tumour cells from tumours excised immediately after irradiation (o ) or six (•) or 24 (A) hours after irradiation. The lines through the six and 24 hour data points were drawn assuming the effect of recovery from PLD was to increase the Do for hypoxic cells by a factor of 1.3 or 1.4.
The effect of recovery from PLD on the enhancement ratio for Ro-07-0582. Tumours were excised six hours after irradiation in the presence or absence of 1 mg/g of the drug. The lines were drawn assuming a factor of 1.3 for recovery from PLD and drug enhancement ratios of 1.6 or 2.1.
Figure 6 shows the effects of recovery from PLD on the survival curves for cells from tumours irradiated in the absence of the sensitizer and excised six or 24 hours later. (The survival curve for cells from tumours excised immediately after irradiation is included for comparison.) The lines were drawn assuming the effect of recovery from PLD was to increase the Do for hypoxic cells by a factor of 1.3 from 2.6 Gy to 3.38 Gy, or by a factor of 1.4 to 3.64 Gy. The factor 1.4 probably represents an upper limit for the effect of recovery from PLD on the survival curve. This recovery had probably reached its maximum by six hours (Fig. 6). While it is possible to draw a line through the zero hour result, parallel to those for six and 24 hours, this is probably not valid for the reasons already given when the results shown in Figs. 3 and 4 were described.
Figure 7 shows the effect of recovery from PLD on the enhancement ratio for Ro-07-0582. The tumours were excised six hours after irradiation with or without injection of 1 mg/g of the drug. The lines were drawn assuming a factor of 1.3 for recovery from PLD and sensitizing enhancement ratios for the drug of 1.6 and 2.1. As Fig. 7 shows, it is clear that recovery from PLD occurred whether or not the sensitizer was present and did not affect the sensitizer enhancement ratio. The value of 2.1 is much too large to account for the observations. The observation that only the naturally hypoxic tumour cells were capable of recovery from PLD, and not the oxic ones made hypoxic by a clamp, can explain the absence of an effect of a clamp on the TCD50 even though the hypoxic fraction was only 5%, if we can make the reasonable assumption that
FIG.-8
natural lyN hypoxic
.oxic x2-7
x2-7 natural lyN hypoxic^/
-8
i
20 Dose
40 grays
60
20 Dose
40 grays
i
60
FIG. 8. Theoretical survival curves for a mixture of oxic and hypoxic cells. The solid lines represent the survival curves for the separate components (oxic cells, naturally hypoxic, or oxic made hypoxic). The dashed lines represent the resultant survival curves obtained by adding the appropriate component curves. A is drawn assuming no recovery from PLD. B is drawn assuming that only the naturally hypoxic cells can recover from PLD. The lines have been drawn using experimentally determined survival curve parameters (see text).
the oxic cells would not have been able to recover from PLD even if the clamp had not been applied during irradiation. Figure 8 shows theoretical survival curves for a mixture of oxic and hypoxic cells. The survival curve parameters used to draw these curves were those derived from the results of the present experiments. Figure 8A shows the survival curves when the naturally hypoxic cells do not recover from PLD. The parameters used were: D$ for hypoxic cells 2.6 Gy, oxygen enhancement ratio 2.7, survival curve extrapolation number 30, hypoxic fraction 5%. As Fig. 8A shows, application of a clamp displaces the resultant survival curve upwards relative to that for undamped cells by an amount proportional to the hypoxic fraction. This displacement occurs at all levels of survival. In Fig. 8B, as well as being protected by hypoxia, the naturally hypoxic cells (but not the oxic ones) are assumed to be capable of recovery from PLD. The survival curve parameters are the same as in Fig. 8A except that the DQ for the naturally hypoxic cell survival curve, relative to that for oxic cells made hypoxic, has been increased by a factor of 1.3 to 3.38 Gy, to
allow for recovery from PLD. The resultant survival curve when a clamp is applied is obtained by adding the "oxic X 2.7" curve and the "naturally hypoxic" curve as in Fig. 8A, except that this time once survival is significantly less than that indicated by the point P (Fig. 8B), the survival curve for naturally hypoxic cells will dictate the response regardless of whether or not a clamp is applied. The presence of the clamp will only increase the dose of radiation beyond which the naturally hypoxic cells dominate the response (Fig. 8B). The value of the dose of radiation to give a surviving fraction represented by the point P in Fig. 8B can be calculated using the simple model already outlined. It will be determined by the equation where r is the factor by which the Do for naturally occurring hypoxic cells must be increased over that for oxic ones made hypoxic by the clamp, due to recovery from PLD. For the survival curves shown in Figs. 4, 6 and 8B, D has a value of 35 Gy. Thus, since the TCD50 was 79.0 Gy, one would not expect it to be affected by clamping the tumours.
327
i
VOL.
50, No. 593 N. J. McNally and P. W. Sheldon
Two discrepancies remain between the two (direct) methods of assaying cell survival (TCD 50 and in vitro assay). First, the enhancement ratio for Ro-07-0582 was larger in the TCD 5 0 assay (2.1) than in the cell survival assay (1.6); this difference is probably independent of recovery from PLD (Fig. 6). Second, the TCD50 was much larger than the cell survival curves would predict. If we now allow for recovery from PLD so that the value for DQ in equation 2 is increased by a factor 1.3 from 2.6 Gy to 3.38 Gy, then for values of N between 108 and 109 cells per tumour, 64.88 Gy^D ^72.67 Gy This is still less than the measured TCD50. A factor of 1.4 for the effect of recovery from PLD on the Do, which probably represents an upper limit for the experimental observation (Fig. 6), would give a TCD50 in the range 69.89 Gy to 78.29 Gy. The calculations of the TCD50 based on equation 2 have assumed that all naturally hypoxic cells in the tumour are capable of repopulating it. This is unlikely to be the case, so that even assuming a factor of 1.4 for the effect of recovery from PLD, it is doubtful if the measured survival curves would predict the highTCD 50 . These discrepancies have, therefore, not been fully accounted for. The drug might be having more effect in the TCD50 assay due to direct toxicity on the tumour cells. However, by six hours its concentration in the tumours will have become quite low (Flockhart, personal communication), and this time of contact with the cells did not affect the enhancement ratio measured in vitro (Fig. 7). Also, giving the drug after irradiation had no effect on theTCDso, suggesting no cytotoxic action (Sheldon, unpublished data). The large TCD50 could be due to a greater ability of the hypoxic cells in the tumour to absorb radiation damage as sublethal when left in situ than when plated in vitro. It is known that the degree of intercellular contact between cells in vitro can affect their ability to absorb sublethal damage (Durand and Sutherland, 1972). It is clear from these results that the fate of irradiated cells in this tumour depends critically on the immediate post-irradiation environment. It follows that, without knowing the effects of this environment, incorrect interpretations of the results may be made. From a practical point of view it may not matter whether one assumes either that the hypoxic fraction is large or that the hypoxic cells are capable of recovery from PLD. However, in a
fractionated course of radiotherapy, or in circumstances where drugs which could inhibit the cells' ability to recover from PLD are used in conjunction with radiation, the use of a single assay of radiation response alone could lead to erroneous conclusions. Without the use of several different methods of assay in the same tumour, the details of its radiation response, for instance the role of recovery from potentially lethal damage, may not be elucidated. ACKNOWLEDGMENTS
We thank Miss A. Walder, Miss A. Marriott and Mrs. S. Bull for production and care of the animals; Miss S. Hill and Mrs. J. de Ronde for expert technical assistance and Dr. J. F. Fowler for helpful discussions. REFERENCES ASQUITH, J. C , WATTS, M. E., PATEL, K., SMITHEN, C. E.,
and ADAMS, G. E., 1974. Electron-affinic sensitization. V. Radiosensitization of hypoxic bacteria and mammalian cells in vitro by some nitroimidazoles and nitropyrazoles. Radiation Research, 60,108-118. BARENDSEN, G. W., and BROERSE, J. J., 1969. Experimental
radiotherapy of a rat rhabdomyosarcoma with 15 MeV neutrons and 300 kV X rays. I. Effect of single doses. European Journal of Cancer, 5, 373-391. DENEKAMP, J., and HARRIS, S., 1975. Tests of two electron-
affinic radiosensitizers in vivo using regrowth of an experimental carcinoma. Radiation Research, 61,191-203. DURAND, R. E., and SUTHERLAND, R. M., 1972. Effects of
intercellular contact on repair of radiation damage. Experimental Cell Research, 71, 75-80. HAHN, G. M., and LITTLE, J. B., 1972. Plateau phase
cultures of mammalian cells. Current Topics in Radiation Research Quarterly, 8, 39-83. HEWITT, H. B., and WILSON, C. W., 1959. A survival curve
for mammalian leukaemia cells irradiated in vivo. British Journal of Cancer, 13, 69-75. HILL, R. P., BUSH, R. S., and YEUNG, P., 1971. The effect of
anaemia on the fraction of hypoxic cells in an experimental tumour. British Journal of Radiology, 44, 299-304. LITTLE, J. B., HAHN, G. M., FRINDEL, E., and TUBIANA, M.,
1973. Repair of potentially lethal radiation damage in vitro and in vivo. Radiology, 106, 689-694. MCNALLY, N. J., 1972. Recovery from sublethal damage by hypoxic tumour cells in vivo. British Journal of Radiology, 45,116-120. MCNALLY, N. J., 1973. A comparison of the effects of radiation on tumour growth delay and cell survival. The effect of oxygen. British Journal of Radiology, 46, 450455. MCNALLY, N. J., 1974. Tumour growth delay and cell survival in situ. British Journal of Radiology, 47, 510-511. MCNALLY, N. J., 1975. A comparison of the effects of radiation on tumour growth delay and cell survival. The effect of radiation quality. British Journal of Radiology, 48, 141-145. POWERS, W. E., and TOLMACH, L. J. 1963. A multicom-
ponent X ray survival curve for mouse lymphosarcoma cells irradiated in vivo. Nature, 197, 710-711. SHELDON, P., and HILL, S. A., 1977. The effect of hypoxic
cell radiosensitizing drugs on local control by single doses of X rays of a transplanted anaplastic tumour in mice. British Journal of Cancer (in press). THOMLINSON, R. H., and CRADDOCK, E. A., 1967. The gross
328
response of an experimental tumour to single doses of X rays. British Journal of Cancer, 21 ,\ 08-125.
