Inr. .I. Radiation Oncology Bid. Phys., Vol. 23. pp. 55-62 Printed in the U.S.A. All rights reserved.
Copyright
0360-3016/92 $5.00 + .oO 0 1992 Pergamon Press Ltd.
??Biology Original Contribution
INTRINSIC RADIATION SENSITIVITY OF GLIOBLASTOMA MULTIFORME
IN VITRO
ALPHONSE TAGHIAN, FRANCISCO PARDO, M.D.,
M.Sc.,
M.D.,
M.Sc.,
DANIELLE
WILLEM DUBOIS, M.D.
HERMAN
SUIT, M.D.,
D.PHIL.,*
GIOIOSO, B.S., KATHLEEN
AND LEO GERWECK,
TOMKINSON, B.S.,
PH.D.
Edwin L. Steele Laboratory of Radiation Biology, Department of Radiation Oncology, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02 1 I4 Glioblastoma multiforme is one of the most resistant of human tumors to radiation whether used alone or in combination with surgery and/or chemotherapy. This resistance may be caused by one or more of several different factors. These include inherent cellular radiation sensitivity, an efficient repair of radiation damage, an increased number of clonogens per unit of volume, a high hypoxic fraction, high [GSH] concentration, and rapid proliferation between fractions. In the present study, we evaluate the intrinsic radiation sensitivity (surviving fraction at 2 Gy or mean inactivation dose) of malignant human glioma cells in vitro. The in vitro radiation sensitivity of 21 malignant glioma cell lines (early and long term passages) has been measured using colony formation as the end-point of cell viability. The survival curve parameters (SF2 measured and calculated, (Y,fl, 4, ii, and MID) have been determined for single dose irradiations of exponential phase cells (18-24 hr after plating) under aerobic conditions and growing on plastic. The mean SF2 of the 21 cell lines is 0.51 + 0.14 (with a range of 0.19 to 0.76). This value may be compared to the mean SF2 of 0.43-0.47 for SCC, 0.43 for melanoma, and 0.52 for glioblastoma as reported from other authors when using colony formation of cells in exponential phase on plastic. Although glioblastoma is almost invariably fatal, our data demonstrate a very wide range of intrinsic radiosensitivities. These broadly overlap the radiation sensitivities of cell lines from tumors that are often treated successfully. We conclude that standard in vitro measurements of cellular radiation sensitivity (SF,) do not yield values that track in a simple manner with local control probability at the clinical level and that, for at least some of the tumors, other parameters and/or physiological factors are more important. Intrinsic radiosensitivity,
SF2, Malignant glioma.
gliomas comprise more than 40% of central nervous system malignancies (9). Surgery alone (30), or in combination with conventional radiation therapy (28) and/or chemotherapy (3 1) have extended survival only minimally and have not offered a major breakthrough that would more substantially improve therapy. Despite the use of neutrons (16), hypoxic cell sensitizers (misonidazole) ( 15), hyperbaric oxygen (7), hyperfractionation (23), and hypofractionation (3), the median survival time
for glioblastoma multiforme patients in most series is consistently around 9- 12 months with less than lo-20% of patients surviving at 2 years. However, there is evidence that dramatic escalation of dose to - lOO- 110 Gy by brachytherapy technique has yielded an important gain in local control and survival. For example, a 57% survival at 2 years has been reported ( 18). Many radiobiologic and physiologic parameters may contribute to the radiation resistance of these tumors, for example, efficient repair of damage, inherent cellular radiation resistance, large number of clonogenic cells per unit volume of tumor,
Presented at the 33ti Annual Meeting of the American Society for Therapeutic Radiology and Oncology (ASTRO), 4-7 November 1991 in Washington, D.C. * Andres Soriano Professor of Radiation Oncology, Harvard Medical School. Reprint requests to: Alphonse Taghian, M.D., M.Sc. Acknowledgements-The authors greatly appreciate and acknowledge Dr. Wilfred Budach for his critical comments and helpful discussion, Dr. David Louis of the Department of Pathology (Neuropathology) for reviewing the pathology reports
and the histology slides, as well as Dr. Ojemann, Dr. Martuza, Dr. Crowell, Dr. Steichen and Dr. Swearingen of the Department of Neurosurgery at Massachusetts General Hospital for supplying us with surgical specimens. We also acknowledge the excellent secretarial assistance of Ms. Phyllis McNally. This work was supported in part by grant DHHS CA 133 11 awarded by the National Cancer Institute, Dept. of Health and Human Services, by the “Association pour la Recherche sur le Cancer (ARC)” in France and by the Phillip Foundation. Accepted for publication 30 September 199 1.
INTRODUCTION
Malignant
55
56
I. J. Radiation
Oncology
0 Biology 0 Physics
Volume
23, Number
1, 1992
Table 1. In vitro cell survival curve parameters of malignant glioma cell lines No. PE expt. Cell line
Histology
D54MG U251MG U-87MG T98G WF Gl* G9+ Gl9* G25* EOI MMCl+ MMC2 A-2 A-7 HGL4+ HGLS+ HGL9+ HGLl I* HGLlZ* HGLl3* HGLl6*
GBM’ GBM GBM GBM GBM AST3$ GBM GBM AST3$ Olig** GBM GBM GBM GBM GBM GBM GBM GBM GBM GBM GBM
3 3 2 2 2 2 2 2 2 4 3 4 3 3 5 4 3 3 5 3 4
(W)
Do (GY)
fi
MID* (GY)
51 21-85 17-17 34-48 26-29 15-15 8-15 5-10 13-24 II 52-88 23-30 41-48 19-28 21-58 35.5 2 55 44 50 26
1.05 1.39 2.25 1.61 1.92 1.27 1.30 2.86 1.55 1.94 1.57 1.29 1.41 1.56 1.58 1.60 1.39 1.70 1.68 1.49 1.10
10.5 2.8 1.4 8.3 2.3 2.7 5.6 1.77 1.9 2.73 3.1 4.5 6.7 5.9 1.2 2.72 6.01 4.1 3.7 8.3 5.5
2.32 2.00 2.70 3.94 2.66 1.81 2.29 1.81 1.11 3.10 2.60 2.24 2.55 3.02 1.94 2.49 2.74 3.30 3.61 3.72 1.98
,266 .435 ,336 .092 .320 .480 .341 .517 .863 ,231 ,281 .337 .299 ,222 .440 .304 .233 ,164 .064 .073 1 .385
.0449 .0178 .0070 .0292 .0119 .0226 .0244 .0103 .0173 .018l .0242 .029 1 .0217 .0226 .0320 .0232 .0301 .0278 .0450 .0382 .0357
al@ (GY)
SF2
SF2
LXIC
meas
Origin
5.9 24.4 48.0 3.2 26.9 21.2 14.0 50.0 50.0 12.8 11.6 11.6 13.8 9.8 11.0 13.1 7.7 6.0 .81 2.0 11.8
,490 .400 .497 .740 .503 .350 .458 .370 ,190 .590 .520 ,453 ,504 .590 ,390 ,500 ,560 .640 .760 .740 .400
.540 .380 .640 ,882 .468 .326 .450 .400 ,204 ,630 ,590 .592 .534 ,530 .540 .650 .590 .700 .830 .800 .450
Dr. Bigner, Duke Dr. Bigner, Duke ATCC ATCC Dr. Shapiro, MSK, NYC Dr. Ramsay, Cambr., UK Dr. Ramsay, Cambr., UK Dr. Ramsay, Cambr., UK Dr. Ramsay, Cambr., UK Dr. Budach, Essen, Germany Dr. Komblith, NYC Dr. Komblith. NYC MGH++ MGH MGH MGH MGH MGH MGH MGH MGH
* Mean inactivation dose. + Short term cultured cell lines, 5 12 passages. * Glioblastoma multiforme. 8Astrocytoma grade 314. ** Oligodendroglioma grade 3. ++Massachusetts General Hospital.
high hypoxic cell fraction, poor reoxygenation, and a rapid rate of repopulation. The aim of this study is to evaluate the intrinsic cellular radiation sensitivity of cell lines derived from human glioblastoma (GBM) and high grade astrocytoma (AST) as expressed by the surviving fraction at 2 Gy (SF*) and the mean inactivation dose (MID) measured in vitro (lo12) as a factor in the extraordinarily poor clinical results of radiation therapy against GBM.
METHODS
AND MATERIALS
Tumor cell lines Twenty-one cell lines were studied. Twelve of these were in early passage, 9 of the 12 were derived from patients at the Massachusetts General Hospital (MGH). Table 1 shows the origin and the histology of each cell line. Eighteen of these cell lines were derived from glioblastoma multiforme, two from astrocytoma grade 3/4, and one cell line was derived from an oligodendroglioma grade 3 (EO 1). The cell lines G 1, G9, G 19, and G25 were provided by Dr. J. Ramsay (Cambridge, UK). Tumorigenicity of the newly established cell lines was tested in nude mice. All but one showed tumor growth (except HGL 16 recently injected into nude mice and currently in a high passage in vitro). The cell lines were maintained in Dulbecco’s modified Eagle medium supplemented with 10% heat inactivated fetal bovine serum and 1% antibiotics (0.05 mg penicillin/
ml, 0.05 mg streptomycin/ml, and 0.1 mg neomycin sulphate/ml) at 37°C in an atmosphere of 5% CO2 in air. Radiation cell survival assays All cell lines were studied in exponential phase. Single cell suspensions were inoculated on 25 cm* plastic flasks at cell numbers appropriate for colony counting. Heavily irradiated feeder cells were added to the viable cells to bring the total to 40,000-50,000 cells/flask (feeder cells were used in 16 cell lines; for HGL4, HGLS, HGLl 1, HGL 12 and HGL 13, feeder cells provided no advantage in plating efficiency and were not employed). In all experiments, the flasks were seeded at 2 or 3 different cell densities for each of 6 or 7 radiation dose levels. At 18 to 24 hr after plating, the cells were administered single radiation doses under aerobic conditions. The irradiation was performed using a 250 kVp X ray machine with HVL of 0.4 mm Cu at a dose rate of 1.54 Gy/min. The cells were irradiated at room temperature (20-22”C), the flasks placed on a rotating lucite block. After a period of 2 to 3 weeks of incubation at 37°C in an atmosphere of 5% CO2 in air, the cells were fixed with methanol, stained with crystal violet and counted. The number of colonies was determined by counting aggregates containing more than 50 cells. Six to eight replicate flasks were plated for each dose level. Three to five survival curves were performed for 14 cell lines and two survival curves for seven cell lines. Consideration of the technical factors that could influence the results include an increase in cell number between
Intrinsic radiosensitivity of glioblastoma multiforme 0 Table 2. Clinical outcome Cell line
of 13 patients
Histology
and the corresponding
57
A. TAGHIAN el al.
cell lines, histology,
SF,, and MID
SF2
MID
Clinical status Dead from tumor progression 5 months after surgery Dead from disease at 8 months Alive at 36 months Dead from disease at 13 months Dead from disease at 10 months Dead from disease at 26 months Dead from disease at 14 months Dead from disease at 19 months Alive at 14 months Dead from disease at 5 months Treated for an AST grade 2. then 48 months later presented a recurrent GBM and died after 6 months Alive at 12 months Alive at 8 months
A2*
GBM
0.50
2.55
A7 Cl G9 G19 G25 HGL4 HGLS HGL9 HGLll HGL12’
GBM Ast. 314 GBM GBM Ast. 314 GBM GBM GBM GBM GBM
0.59 0.35 0.46 0.37 0.19 0.39 0.50 0.56 0.64 0.76
3.02 1.81 2.29 1.81 1.1 I 1.94 2.49 2.74 3.30 3.61
HGLl3* HGL16
GBM GBM
0.74 0.40
3.72 1.96
* The patient did not receive post-operative RT because of medical problems. + The cell line was established-from the local recurrent GBM. * The patient received a total dose of 90 Gy of protontherapy.
inoculation
and irradiation,
and cell cycle redistribution.
of cases, feeder cells were used and the potential proliferation of test versus lethally irradiated cells could not be distinguished. In view of this, we monitored the change in cell number at various times after inoculation of 100,000 test cells. The cell number increase between 0 and 24 hr was less than 1.1 and the potential error caused by cell multiplicity at the time of irradiation was found to be trivial. The cell cycle distribution at the time of radiation was not evaluated. However, it is highly unlikely that redistribution could have given rise to the large variation in SF2 observed in this study between cell lines. In the majority
Statistical analysis The analysis of variance was performed on the pooled data from all experiments for each cell line, using a software package.* For the calculation of a! and p, 1nSf of all data versus dose were fitted by polynomial regression. For the calculation of Do and ii, 1nSf (the linear part of the curve < 0.1 SF), versus dose was fitted by simple linear regression. All data points were weighted equally. The mean inactivation dose MID was calculated from the LY and p by a previously published method (12). Additional statistical analysis of the data presented in this publication, including a comparison of the relative and absolute SF,‘s among three investigators on the same cell lines, will be presented in another paper. Clinical data The clinical status of 13 patients was known. Table 2 records the status at times after diagnosis as dead of dis-
* StatWorks,
Cricket Software,
Philadelphia,
PA.
ease, or alive. The clinical data concerning the patients Gl, G9, G19, and G25 were generously provided by Dr. Jonathan Ramsay at Cambridge (U.K.). Data are not available on eight patients because the cell lines were established in other laboratories (cf Table 1 for the origin of each cell line). RESULTS
Survival curve parameters The values for survival curve parameters (plating efficiency, Do, ii, (Y,/I, SF2 calculated and SF, measured) as well as the histology and the origin of each cell line are shown in Table 1. The plating efficiency (PE) varied between 2 and 85%. No correlation was found between PE and SF2. Figure 1 presents a plot of the cumulative frequency of the calculated SF2, the MID, and the CYof the 2 1 cell lines. The mean calculated SF2 was 0.5 1 + 0.14 with a range of 0.19 to 0.76 and a CV of 28%. The mean MID was 2.57 + 0.7 1 Gy ranging from 1.11 to 3.94 and a CV of 28%. The mean (Ywas 0.3 14 + 0.18 with a maximum of 0.863 and a minimum of 0.038 and a CV of 56%. The broad cumulative frequency for values of these parameters indicate a very wide range of intrinsic sensitivity for malignant glioma cell lines as measured in vitro. There was no significant difference between the calculated SF2 (0.5 1 + 0.14) and the measured SF2 (0.56 + 0.17) which shows that the linear quadratic model fits the data well. The mean SF2 of five cell lines studied previously (32) was 0.49 compared to 0.5 1 in the present study. The Do was calculated from the linear component of the survival curve down to 10p3. The mean Do was 1.59
I. J. Radiation Oncology 0 Biology 0 Physics
Volume 23, Number 1, 1992
2.57+-0.71
0.0
02
0.4
0.6
0.6
Calculated
10
0.2
00
0.4
SF2
0.6
0.6
1 .o
alpha
Fig. 1. Cumulative frequency of the calculated surviving fraction at 2 Gy (SF,), the (Y(linear quadratic model) and the mean inactivation dose (MID) of the 2 1 glioblastoma and high grade astrocytoma cell lines.
+ 0.41 ranging between 1.05 and 2.86 with a coefficient of variation (CV) of 26%. The mean ii was 4.5 + 2.7 with a minimum of 1.4 and a maximum of 10.5 and a CV of 60%. The mean a//3 ratio was 8.7 + 13.4 (range 0.7750.0). Newly versus well established cell lines Figure 2 demonstrates the cumulative frequency of SF;! in newly established (< 12 passages) versus well-established cell lines. The mean SF2 of newly established cell lines (12) was 0.49 + 0.17 and that of well-established (nine cell lines) was 0.53 -+ 0.10. This difference was not significant (p = 0.533), although the coefficients of variation were 35% and 19% for the newly and well established cell lines, respectively. Histology The pathology reports and/or slides of the nine tumors from which the cell lines were derived and established in our lab were reviewed by Dr. David Louis of the Pathology Department (Neuropathology) at MGH. Eighteen of the 2 1 cell lines were glioblastoma multiforme, the cell lines G 1 and G25 were astrocytoma grade 3/4 and the cell line EO 1 oligodendroglioma grade 3. The average SF2 of glioblastoma multiforme (18 cell lines) was 0.53 + 0.12. SF2 of the two astrocytoma grade 3/4 were 0.35 and 0.19. The oligodendroglioma cell line (EOl) had an SF2 of 0.59.
Clinical outcome Table 2 shows the clinical outcome, histology, SF2, and MID for the tumors from 13 patients. Four patients are alive without evidence of local recurrence (Gl, HGL9, HGL 13, and HGL 16) with follow-up period of 8, 12, 14, and 36 months. The other patients died from disease at 5 (two patients), 8 (two patients), 10, 13, 14, 19, and 26 months (one patient each). Patient HGL12 was treated for an astrocytoma grade 2 in August 1985, then presented with a GBM as local recurrence in June 1989, and died in December 1989. The cell line studied was established from the recurrent tumor. Most patients were treated by some category of surgical resection and post-operative radiation (-60 Gy) with or without chemotherapy. One patient (HGL 13) received 90 Gy by proton therapy. Patient A2 did not receive postoperative radiation therapy because of a medical problem, and died 5 months later from tumor progression. Two patients presented with astrocytoma grade 3/4: (Gl and G25), one is still alive with a follow-up of 36 months (Gl), the other died after 26 months (G25). Three of the nine patients with glioblastoma multiforme are alive with a follow-up period of 8, 12, and 14 months. The other patients survived a mean period of 11.2 months. Figure 3 shows the survival of GBM patients in months, plotted against SF*. There is no evident correlation between SF2 and survival. Patient (Gl), considered as a long-
30-
1.0-
z 5
0.6 -
zo-
s
0.6 -
2
m
lo-
5
?? ??
5 v) 0 0.0
3
o.o! 0.0
.
I
0.2
.
I
.
0.4
1
0.6
-
I
0.6
.
,
? ? alive
?? m0
.,.,.,.I., 0.2 0.4
??
?? dead
?? ??
0.6
0.8
1.0
SF2
1.0
SF2
Fig. 2. Cumulative frequency of SF, of the newly-established versus well-established glioblastoma and high grade astrocytoma cell lines.
Fig. 3. SF, of GBM plotted against the survival (in months) of patients. Two cell lines were not plotted: HGL12 because the cell line was derived from a recurrent GBM that was not irradiated, and A7 because the patient did not receive radiation because of medical problems.
Intrinsic radiosensitivity of glioblastoma multiforme 0 A. TAGHIAN
term survivor (36 months), had SF2 of 0.35. Patient G25 died from disease after 26 months had an SF2 of 0.19. However, patients with relatively high SF2, like HGL9 and HGL13 (0.56 and 0.74), are alive without evidence of disease at 14 and 12 months, respectively. On the other hand, patients with relatively low SF2, like G 19 and HGL4 (0.37 and 0.39) died from disease at 10 and 14 months, respectively.
DISCUSSION In a general review of malignant glioma, Davis (9) emphasized that one of the major causes of failure of radiation treatment of high grade glioma could be a low intrinsic radiation sensitivity. The aim of our study was to evaluate the importance of that radiation sensitivity as measured by loss of colony formation of malignant glioma cells studied in vitro. Several workers have suggested that the initial low-dose region of the radiation survival curves is the most relevant to any consideration of the response of human tumors to the fractionated radiation of conventional therapy ( lo- 13,20). Hence, the interest of most of the recent published studies and the present work, on radiation sensitivity determinations of human tumor cells, has concentrated on parameters relating to the low dose region of the survival curves (a, MID, and SF2) (10-l 3, 20). Results of these determinations were calculated for all the 2 1 high grade brain tumor cell lines. Comparison with glioblastoma cell lines from the literature There are only few published data on the radiosensitivity of glioblastoma or high grade astrocytoma. Allalunis-
59
et al.
Turner et al. ( 1) studied 16 early passage glioblastoma cell lines, using colony formation on plastic as an end-point. They found a mean SF2 of 0.52 f 0.22 with a range from 0.15 to 0.93 and a CV of 42%. Only a few cell lines (1 to 3) have been studied in the other reports. Table 3 presents published values for the radiosensitivity parameters of glioma cell lines. The variation in radiosensitivity cannot be appreciated from the low numbers of cell lines as described in each report, however, when they are grouped together, a quite wide variation is evident which is not greatly different from that observed here (Fig. 4). The mean SF2 calculated from these 15 cell lines from the literature was 0.57 + 0.16 (CV 28%) ranging from 0.3 1 to 0.75 (note that most cell lines were late passage). Figure 4 illustrates the cumulative frequency of SF2 from the literature (Table 3) and from Allalunis-Turner et al. (1) as well as the MGH data. The three curves are almost superimposable. They illustrate a wide range of intrinsic radiation sensitivity: the most radiosensitive being an SF2 of 0.15, the most radioresistant being an SF2 of 0.93 and the mean SF2 for the 52 GBM cell lines was 0.53. Comparison with other tumors Table 4 shows SF:! values of other types of tumors with the number of cell lines studied and the CV of each group. Using colony formation on plastic as the end-point of cell viability, no significant difference was found between the mean SF2 of our malignant glioma cell lines of 0.5 1, and the 0.46 of ovarian cancer (35) the 0.49 of colon cancer (17) the 0.43 of melanoma (26), the 0.44 of the SCC of the head and neck (H&N) (general review 20), and the 0.47 of cervical cancer (the last used primary culture and soft agar method) (8).
Table 3. SF2, (Y,/3 and MID values from the literature on glioblastoma multiforme and high grade astrocytoma cell lines Author Yang et al. (37)
Cell line IN 859 IN 1265 SB
Schultz and Geard (29) Rhaapost et al. (25) Masuda et al. (2 1) Gerweck et al. (14)*
Weischelbaum et al. (36)* Nilsson et al. (22)* Allalunis-Turner et al. (1) Present study
U-87 MG U-138 MG U-373 MG KNS-42 KNS-60 A3 A2 A7 TX-13 U- 118-MG 16 cell lines 21 cell lines
Histology Astrocyt. gr. 3 & 4
Ast. gr. 3 Ast. gr. 4 GBM GBM Ast. high gr. GBM GBM Ast. gr. 4 Ast. gr. 4t Ast. gr. 3t GBM GBM GBM GBM + AST3/4
SF2
0.57 0.73 0.47 0.58 0.46 0.70 0.60 0.75 0.39 0.33 0.72 0.44 0.72 0.31 0.75 0.52 0.51
P
MID
0.31 0.042 0.36 0.20 0.30 0.15 0.20 0.094
0.033 0.053 0.0077 0.036 0.045 0.03 1 0.035 0.033
2.30 3.47 2.53 2.81 2.18 3.30 2.83 3.74
0.086 0.006 0.097 0.559 0.0625 0.314
0.0394 0.0394 0.0319 0.0152 0.04
3.55 2.11 3.76 1.62 3.70 2.6
a
0.026
* No specification of SF2, a, @or MID was in the paper, the calculations of these values were taken from Ref. 11. t The histologic slides of the tumors from which these two cell lines were derived were reviewed by Dr. David Louis. They were diagnosed as glioblastoma multiforme. * SF2 and MID were calculated from cxand /3 published in the paper (22).
I. J. Radiation Oncology 0 Biology0 Physics
60
Fig. 4. Cumulative frequency of SF2 of glioblastoma and high grade astrocytoma. Comparison between data from the literature, data from Allalunis-Turner et al. (1) and our data (MGH).
However,
using the same end-point
and all assays per-
formed by one group, a significant difference was reported by Allalunis-Turner et al. (I) between the SF2 for malignant glioma (0.52) on the one hand and SCC of the uterine cervix (0.27) and adenocarcinoma of the endometrium (0.25) on the other hand. The SF2 values for SCC of H&N (0.43) and sarcoma (0.34) obtained by Weichselbaum et al. (35) and for SCC of H&N (0.32) from Brock et al. (5) are significantly lower then we found for GBM (0.5 I), (the last (5) used primary culture and population growth method on CAM plates). Early versus late passages
Figure 2 illustrates the difference between the early established and late passage GBM cell lines. The mean SF, of the early established was 0.49 + 0.17 ( 12 cell lines) versus 0.53 + 0.10 (nine cell lines) for the late passage lines. This difference was not significant (p = 0.533). The wide distribution of intrinsic radiosensitivity was more evident in the early (coefficient of variation 35%) than in
Volume 23, Number 1, 1992 late passage cell lines (CV 19%). This was comparable with the CV of the 16 early passage GBM cell lines studied by Allalunis-Turner et al. (1) (42%). In a general review (32), the authors pointed out the difference in SF;!between the early and late passages. Using colony formation as the end-point for cells growing on plastic, SF, values for SCC and adenocarcinoma were 0.50 (33 cell lines), 0.49 (20 cell lines) for well-established, versus 0.25 (14 cell lines) and 0.28 (12 cell lines) for primary culture, respectively. It should be noted, however, that the primary culture used the soft agar assay. Also using the soft agar assay, Rofstad (26, 27) studied the radiosensitivity in vitro of melanoma cell lines. The mean SF, values of 18 primary culture cell lines was 0.38 f 0.2 (CV 53%) compared to 0.50 + 0.17 (CV 34%) for 11 wellestablished cell lines. This difference was in the borderline of significance (p = 0.055). In general, the late passage cell lines tend to be less sensitive and with a lower coefficient of variation than the early passage and the primary cultures, however, few published data are available on this point. Histology
The comparison between the SF*‘s of GBM and AST grade 314 is difficult because we studied only two AST grade 3/4 cell lines. The radiation response of three astrocytoma cell lines across grade was studied recently (29). The authors found SF2’s of 0.58 and 0.46 for grade 3 and 4, respectively, in comparison to 0.36 for a single grade 1 cell line. They concluded that the low grade astrocytoma cell lines are more sensitive than the high grade lines in vitro. These data are compatible with the clinical results, but one cannot make conclusions based on findings from a single cell line. Furthermore, in our data we found (G25),
Table 4. Mean SF2 values of several types of tumors, the type of passage, the number of cell line studied as well as the CV of each group
Authors Davidson et al. (8) Brock et al. (5) Weischelbaum al. (35)
Type of tumor
Type of passage
SCC of uterine cervix
Primary culture (Courtenay Mills) Primary culture (population growth) Early passage
SCC from H & N et
Leith et al. (17) Rofstad (26) (general review) Malaise et al. (20) (general review) Allalunis-Turner et al. (1)
SCC from H & N
No. of cell lines
SF, @SE)
cv @)
52
0.47 + 0.18
38
140
0.32 + 0.15
47
24
0.43 * 0.12
29
15 14 16 21
0.46 0.34 0.49 0.43
33 29 23 41
Ovarian cancer Sarcoma Colon cancer *Melanoma
Early passage Early passage Late passage Primary and late passage
see
Late passage
9
0.44 + 0.16
36
SCC of uterine cervix Endometrium
Early passage Early passage
17 19
0.27 f 0.14 0.25 + 0.09
51 37
* The SF,% were calculated from (Yand p published in the paper (26).
+ + f. +
0.15 0.098 0.115 0.18
Intrinsic radiosensitivity of glioblastoma multiforme 0 A. TAGHIAN et al.
an astrocytoma grade 3/4, and (G 19) a glioblastoma multiforme, with SF, of 0.19 and 0.37, respectively. This shows that we may have a high grade AST or a GBM with a relatively low SF2 as measured in vitro. Correlation between SF, and clinical outcome SF2 is being evaluated as a parameter for prediction of the radiation response of tumors (4, 6, 19, 24, 34). The aim of this study was not to test the predictive value of intrinsic radiosensitivity measurement expressed by SF2 for tumor response to radiotherapy, but the value of SF2 as a factor in the poor clinical results of GBM. However, the authors found it interesting to look to the clinical outcome of the patients from which the tumor cell lines were derived. For our small number of glioblastoma multiforme patients, we did not find a correlation between the SF2 measured in vitro and the clinical outcome (Fig. 3); some of whom have only short follow-up. We noticed that the relatively long survival was a patient with an astrocytoma grade 3/4 (Gl) with an SF2 of 0.35, but conclusions cannot, of course, be drawn from such scant clinical data. Nevertheless, the wide range of intrinsic radiation sensitivity of malignant gliomas cell lines found in our study, as well as the Allalunis-Turner study (1) and data from the literature grouped together (Fig. 4) could not explain the quasi-homogeneous fatal clinical outcome of patients with these tumors. Probably mechanisms not evaluated by the in vitro assays, which modify radiation sensitivity are not present in vitro where the cells are under uniform and near optimal metabolic conditions, do operate in vivo and effect large shifts in cellular radiation sensitivity. If
61
so, the lack of concordance between the simple in vitro assay results and clinical outcome is not a cause for surprise. Similar conclusion was found by Bentzen et al. (2) when they found that the dose-control curves of malignant melanoma derived from in vitro radiosensitivity data are significantly steeper than the clinical dose-control curves. Such conclusions would limit the predictive value of current radiation sensitivity assays based on in vitro dosesurvival measurement for at least high grade malignant gliomas.
CONCLUSION The average SF2 for the GBM cell lines studied here was in the upper range of intrinsic radiosensitivities as reported in the literature, however, there was extensive overlap with SF2’s for cells from tumors often treated effectively. Our results demonstrate a wide range of intrinsic radiation sensitivity that is not compatible with the almost invariably fatal outcome of glioblastoma multiforme and high grade malignant gliomas, if comparisons of measured SF2s have any absolute value. From these data, we conclude that either standard measurements of cellular radiation sensitivity (SF,) in vitro of GBM do not yield values that track in a simple manner with the local control probability at the clinical level, or the intrinsic radiation sensitivity of malignant gliomas is not the most important factor in the determination of the in vivo radiation resistance of these tumors and is probably overshadowed by other factors like stem cell fraction, proliferation, hypoxia, or others.
REFERENCES I. Allalunis-Turner, M.; Pearcey, R.; Urtasun, R. Inherent radiosensitivity testing of human tumor biopsy specimens. 38th Annual Meeting of the Radiation Research Society, New Orleans, Louisiana, April 7- 12, 1990; Abstract Cn- I, p. 67. 2. Bentzen, S.; Thames, H.; Overgaard, J. Does variation in the in vitro cellular radiosensitivity explain the shallow clinical dose-control curve for malignant melanoma? Int. J. Radiat. Biol. 57: 117- 126; 1990. 3. Bleehen, N. M.; Wilshire, P. N.; Plowman, P. N.; Watson, J. V.; Gleave, J. R.; Holmes, A. E.; Lewin, W. S.; Treip, C. S.; Hawkins, T. D. A randomized study of misonidazole and radiotherapy for grades 3 and 4 cerebral astrocytoma. Br. J. Cancer 43: 436-442; 198 1. 4. Bristow, R.; Hardy, P.; Hill, R. Comparison between in vitro radiosensitivity and in vivo radioresponse of murine tumor cell lines 1: parameters of in vitro radiosensitivity and endogenous cellular glutathione levels. Int. J. Radiat. Oncol. Biol. Phys. 18: 133-145; 1990. 5. Brock, W.; Peters, L. Radiosensitivity of head and neck squamous cell carcinomas and local tumor control (Abstract). 25th Paterson Symposium, 3-5 December, 1990, Paterson Institute for Cancer Research, Manchester, UK. 6. Brock, W.; Maor, M.; Peters, L. Predictors oftumor response to radiotherapy. Radiat. Res. 8(Suppl.): S290-S296; 1985. 7. Chang, C. H. Hyperbaric oxygen and radiation therapy in
8.
9.
10.
11.
12.
13.
14.
the management of glioblastoma. NC1 Monogr. 46: 163169; 1977. Davidson, S.; West, C.; Roberts, S.; Hendry, J.; Hunter, R. Radiosensitivity testing of primary cervical carcinoma: evaluation of intra- and intertumour heterogeneity. Radiother. Oncol. 18: 349-356; 1990. Davis, L. Presidential address: malignant glioma-a nemesis which requires clinical and basic investigation in radiation oncology. Int. J. Radiat. Oncol. Biol. Phys. 16: 1355-1365; 1989. Deacon, J.; Peckham, M. J.; Steel, G. The radioresponse of human tumors and the initial slope of the cell survival curve. Radiother. Oncol. 2: 3 17-323; 1984. Fertil, B.; Malaise, E. P. Inherent cellular radiosensitivity as a basic concept for human tumor radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 7: 621-629; 198 1. Fertil, B.; Dertinger, H.; Courdi, A.; Malaise, E. P. Mean inactivation dose: a useful concept for intercomparison of human cell survival curves. Radiat. Res. 99: 73-84; 1984. Fertil, B.; Malaise, E. P. Intrinsic radiosensitivity of human cell lines is correlated with radioresponsiveness of human tumors: analysis of 101 published survival curves. Int. J. Radiat. Oncol. Biol. Phys. 11: 1699-1707; 1985. Gerweck, L.; Kornblith, P.; Burlett, P.; Wang, J.; Sweigert, S. Radiation sensitivity of cultured human glioblastoma cells. Radiology 125: 231-234; 1977.
62
I. J. Radiation Oncology 0 Biology0 Physics
15. Green, S. P.; Byar, D. P.; Strike, T. A.; Alexander, E.; Brooks, W. H.; Burger, P. C.; Hunt, W. E.; Mealey, J.; Odom, G. L.; Paoletti, P.; Pistenma, D. A.; Ransohoff, J.; Robertson, J. T.; Selker, R. G.; Shapiro, W. R.; Smith, K. R. Randomized comparisons of BCNU, streptozotocin, radiosensitizer, and fractionation of radiotherapy in the post-operative treatment of malignant glioma (study 7702). Proc. Amer. Sot. Clin. Oncol. 3: 260; 1984. 16. Laramore, G. E.; Diener-West, M.; Griffin, T. W.; Nelson, J. S.; Griem, M. L.; Thomas, F. J.; Hendrickson, F. R.; Griffin, B. R.; Myrianthopoulos, L. C.; Saxton, J. Randomized neutron dose searching study for malignant gliomas of the brain: results of an RTOG study. Int. J. Radiat. Oncol. Biol. Phys. 14: 1093-l 102; 1988. 17. Leith, J.; Faulkner, L.; Papa, G.; Quinn, P.; Michelson, S. In vitro radiation survival parameters of human colon tumor cells. Int. J. Radiat. Oncol. Biol. Phys. 20: 203-206; 1991. 18. Loeffler, J.; Alexander, E.; Wen, P.; Shea, M.; Coleman, N.; Kooy, H.; Fine, H.; Nezdi, L.; Silver, B.; Riese, N.; Black, P. Results of stereotactic brachytherapy used in the initial management of patients with glioblastoma. JNCI 82: 19 181921; 1990. 19. Malaise, E. P.; Fertil, B.; Chavaudra, N.; Guichard, M. Distribution of radiation sensitivities for human tumor cells of specific histological types: comparison of in vitro to in vivo data. Int. J. Radiat. Oncol. Biol. Phys. 13: 617-624; 1986. 20. Malaise, E. P.; Fertil, B.; Deschavanne, P.; Chavaudra, N.; Brock, W. Initial slope of radiation survival curves is characteristic of the origin of primary and established cultures of human tumor cells and fibroblasts. Radiat. Res. 111: 3 19333; 1987. 21. Masuda, K.; Aramaki, R.; Takaki, T.; Wakisaka, S. Possible explanation of radioresistance of glioblastoma in situ. Int. J. Radiat. Oncol. Biol. Phys. 9: 255-258; 1983. 22. Nilsson, S.; Carlsson, J.; Larsson, B.; Ponten, J. Survival of irradiated glia and glioma cells studied with a new cloning technique. Int. J. Radiat. Biol. 37: 267-279; 1980. 23. Payne, D. G.; Simpson, W. J.; Keen, C.; Platts, M. E. Malignant astrocytoma. Hyperfractionated and standard radiotherapy with chemotherapy in a randomized prospective clinical trial. Cancer 50: 2301-2306; 1982. 24. Peters, L.; Brock, W.; Johnson, R.; Meyn, R.; Tofilon, P.; Milas, L. Potential methods for predicting tumor radiocurability. Int. J. Radiat. Oncol. Biol. Phys. 12: 459-467; 1986.
Volume 23, Number 1, 1992 25. Raaphorst, G. P.; Feeley, M.; Da Silva, V.; Danjoux, C.; Gerig, L. A comparison of heat and radiation sensitivity of three human ghoma cell lines. Int. J. Radiat. Oncol. Biol. Phys. 17: 615-622; 1989. 26. Rofstad, E. Radiation biology of malignant melanoma. A review article. Acta Radiol. 25: l-10; 1986. 27. Rofstad, E.; Wahl, A.; Brustad, T. Radiation sensitivity in vitro of cells isolated from human tumor surgical specimens. Cancer Res. 47: 106-l 10; 1987. 28. Sheline, G. E. Radiation therapy of primary tumors. Cancer 39: 873-881; 1977. 29. Shultz, C.; Geard, C. Radioresponse of human astrocytic tumors across grade as a function of acute and chronic irradiation. Int. J. Radiat. Oncol. Biol. Phys. 19: 1397- 1403; 1990. 30. Stage, W. S.; Stein, J. J. Treatment of malignant astrocytomas. Am. J. Roentgenol. 120: 7-18; 1974. 31. Stenning, S. P.; Freedman, L. S.; Bleehen, N. M. An overview of published results from randomized studies of nitrosoureas in primary high grade malignant glioma. Br. J. Cancer 56: 89-90; 1987. 32. Suit, H.; Baumann, M.; Skates, S.; Convery, K. Clinical interest in determinations of cellular radiation sensitivity. Int. J. Radiat. Biol. 56: 725-737; 1989. 33. Suit, H.; Zietman, A.; Tomkinson, K.; Ramsay, J.; Gerweck, L.; Sedlacek, R. Radiation response of xenografls of a human squamous cell carcinoma and a glioblastoma multiforme: a progress report. Int. J. Radiat. Oncol. Biol. Phys. 18: 365373; 1990. 34. Tucker, S.; Thames, H. The effect of patient-to-patient variability on the accuracy of predictive assays of tumor response to radiotherapy: a theoretical evaluation. Int. J. Radiat. Oncol. Biol. Phys. 17: 145-155; 1989. 35. Weichselbaum, R.; Rotmensch, J.; Ahmed-Swan, S.; Beckett, A. Radiobiological characterization of 53 human tumor cell lines. Int. J. Radiat. Biol. 56;5: 553-560; 1989. 36. Weichselbaum, R. R.; Epstein, J.; Little, J. B. In vitro cellular radiosensitivity of human malignant tumors. Europ. J. Cancer 12: 47-51; 1976. 37. Yang, X.; Darling, J. L.; McMillan, T. J.; Peacock, J. H.; Steel, G. G. Radiosensitivity, recovery and dose-rate effect in three human glioma cell lines. Radiother. Oncol. 19: 4956; 1990.
Copyright
0360-3016/92 $5.00 + .oO 0 1992 Pergamon Press Ltd.
??Biology Original Contribution
INTRINSIC RADIATION SENSITIVITY OF GLIOBLASTOMA MULTIFORME
IN VITRO
ALPHONSE TAGHIAN, FRANCISCO PARDO, M.D.,
M.Sc.,
M.D.,
M.Sc.,
DANIELLE
WILLEM DUBOIS, M.D.
HERMAN
SUIT, M.D.,
D.PHIL.,*
GIOIOSO, B.S., KATHLEEN
AND LEO GERWECK,
TOMKINSON, B.S.,
PH.D.
Edwin L. Steele Laboratory of Radiation Biology, Department of Radiation Oncology, Massachusetts General Hospital, Harvard Medical School, Boston, MA 02 1 I4 Glioblastoma multiforme is one of the most resistant of human tumors to radiation whether used alone or in combination with surgery and/or chemotherapy. This resistance may be caused by one or more of several different factors. These include inherent cellular radiation sensitivity, an efficient repair of radiation damage, an increased number of clonogens per unit of volume, a high hypoxic fraction, high [GSH] concentration, and rapid proliferation between fractions. In the present study, we evaluate the intrinsic radiation sensitivity (surviving fraction at 2 Gy or mean inactivation dose) of malignant human glioma cells in vitro. The in vitro radiation sensitivity of 21 malignant glioma cell lines (early and long term passages) has been measured using colony formation as the end-point of cell viability. The survival curve parameters (SF2 measured and calculated, (Y,fl, 4, ii, and MID) have been determined for single dose irradiations of exponential phase cells (18-24 hr after plating) under aerobic conditions and growing on plastic. The mean SF2 of the 21 cell lines is 0.51 + 0.14 (with a range of 0.19 to 0.76). This value may be compared to the mean SF2 of 0.43-0.47 for SCC, 0.43 for melanoma, and 0.52 for glioblastoma as reported from other authors when using colony formation of cells in exponential phase on plastic. Although glioblastoma is almost invariably fatal, our data demonstrate a very wide range of intrinsic radiosensitivities. These broadly overlap the radiation sensitivities of cell lines from tumors that are often treated successfully. We conclude that standard in vitro measurements of cellular radiation sensitivity (SF,) do not yield values that track in a simple manner with local control probability at the clinical level and that, for at least some of the tumors, other parameters and/or physiological factors are more important. Intrinsic radiosensitivity,
SF2, Malignant glioma.
gliomas comprise more than 40% of central nervous system malignancies (9). Surgery alone (30), or in combination with conventional radiation therapy (28) and/or chemotherapy (3 1) have extended survival only minimally and have not offered a major breakthrough that would more substantially improve therapy. Despite the use of neutrons (16), hypoxic cell sensitizers (misonidazole) ( 15), hyperbaric oxygen (7), hyperfractionation (23), and hypofractionation (3), the median survival time
for glioblastoma multiforme patients in most series is consistently around 9- 12 months with less than lo-20% of patients surviving at 2 years. However, there is evidence that dramatic escalation of dose to - lOO- 110 Gy by brachytherapy technique has yielded an important gain in local control and survival. For example, a 57% survival at 2 years has been reported ( 18). Many radiobiologic and physiologic parameters may contribute to the radiation resistance of these tumors, for example, efficient repair of damage, inherent cellular radiation resistance, large number of clonogenic cells per unit volume of tumor,
Presented at the 33ti Annual Meeting of the American Society for Therapeutic Radiology and Oncology (ASTRO), 4-7 November 1991 in Washington, D.C. * Andres Soriano Professor of Radiation Oncology, Harvard Medical School. Reprint requests to: Alphonse Taghian, M.D., M.Sc. Acknowledgements-The authors greatly appreciate and acknowledge Dr. Wilfred Budach for his critical comments and helpful discussion, Dr. David Louis of the Department of Pathology (Neuropathology) for reviewing the pathology reports
and the histology slides, as well as Dr. Ojemann, Dr. Martuza, Dr. Crowell, Dr. Steichen and Dr. Swearingen of the Department of Neurosurgery at Massachusetts General Hospital for supplying us with surgical specimens. We also acknowledge the excellent secretarial assistance of Ms. Phyllis McNally. This work was supported in part by grant DHHS CA 133 11 awarded by the National Cancer Institute, Dept. of Health and Human Services, by the “Association pour la Recherche sur le Cancer (ARC)” in France and by the Phillip Foundation. Accepted for publication 30 September 199 1.
INTRODUCTION
Malignant
55
56
I. J. Radiation
Oncology
0 Biology 0 Physics
Volume
23, Number
1, 1992
Table 1. In vitro cell survival curve parameters of malignant glioma cell lines No. PE expt. Cell line
Histology
D54MG U251MG U-87MG T98G WF Gl* G9+ Gl9* G25* EOI MMCl+ MMC2 A-2 A-7 HGL4+ HGLS+ HGL9+ HGLl I* HGLlZ* HGLl3* HGLl6*
GBM’ GBM GBM GBM GBM AST3$ GBM GBM AST3$ Olig** GBM GBM GBM GBM GBM GBM GBM GBM GBM GBM GBM
3 3 2 2 2 2 2 2 2 4 3 4 3 3 5 4 3 3 5 3 4
(W)
Do (GY)
fi
MID* (GY)
51 21-85 17-17 34-48 26-29 15-15 8-15 5-10 13-24 II 52-88 23-30 41-48 19-28 21-58 35.5 2 55 44 50 26
1.05 1.39 2.25 1.61 1.92 1.27 1.30 2.86 1.55 1.94 1.57 1.29 1.41 1.56 1.58 1.60 1.39 1.70 1.68 1.49 1.10
10.5 2.8 1.4 8.3 2.3 2.7 5.6 1.77 1.9 2.73 3.1 4.5 6.7 5.9 1.2 2.72 6.01 4.1 3.7 8.3 5.5
2.32 2.00 2.70 3.94 2.66 1.81 2.29 1.81 1.11 3.10 2.60 2.24 2.55 3.02 1.94 2.49 2.74 3.30 3.61 3.72 1.98
,266 .435 ,336 .092 .320 .480 .341 .517 .863 ,231 ,281 .337 .299 ,222 .440 .304 .233 ,164 .064 .073 1 .385
.0449 .0178 .0070 .0292 .0119 .0226 .0244 .0103 .0173 .018l .0242 .029 1 .0217 .0226 .0320 .0232 .0301 .0278 .0450 .0382 .0357
al@ (GY)
SF2
SF2
LXIC
meas
Origin
5.9 24.4 48.0 3.2 26.9 21.2 14.0 50.0 50.0 12.8 11.6 11.6 13.8 9.8 11.0 13.1 7.7 6.0 .81 2.0 11.8
,490 .400 .497 .740 .503 .350 .458 .370 ,190 .590 .520 ,453 ,504 .590 ,390 ,500 ,560 .640 .760 .740 .400
.540 .380 .640 ,882 .468 .326 .450 .400 ,204 ,630 ,590 .592 .534 ,530 .540 .650 .590 .700 .830 .800 .450
Dr. Bigner, Duke Dr. Bigner, Duke ATCC ATCC Dr. Shapiro, MSK, NYC Dr. Ramsay, Cambr., UK Dr. Ramsay, Cambr., UK Dr. Ramsay, Cambr., UK Dr. Ramsay, Cambr., UK Dr. Budach, Essen, Germany Dr. Komblith, NYC Dr. Komblith. NYC MGH++ MGH MGH MGH MGH MGH MGH MGH MGH
* Mean inactivation dose. + Short term cultured cell lines, 5 12 passages. * Glioblastoma multiforme. 8Astrocytoma grade 314. ** Oligodendroglioma grade 3. ++Massachusetts General Hospital.
high hypoxic cell fraction, poor reoxygenation, and a rapid rate of repopulation. The aim of this study is to evaluate the intrinsic cellular radiation sensitivity of cell lines derived from human glioblastoma (GBM) and high grade astrocytoma (AST) as expressed by the surviving fraction at 2 Gy (SF*) and the mean inactivation dose (MID) measured in vitro (lo12) as a factor in the extraordinarily poor clinical results of radiation therapy against GBM.
METHODS
AND MATERIALS
Tumor cell lines Twenty-one cell lines were studied. Twelve of these were in early passage, 9 of the 12 were derived from patients at the Massachusetts General Hospital (MGH). Table 1 shows the origin and the histology of each cell line. Eighteen of these cell lines were derived from glioblastoma multiforme, two from astrocytoma grade 3/4, and one cell line was derived from an oligodendroglioma grade 3 (EO 1). The cell lines G 1, G9, G 19, and G25 were provided by Dr. J. Ramsay (Cambridge, UK). Tumorigenicity of the newly established cell lines was tested in nude mice. All but one showed tumor growth (except HGL 16 recently injected into nude mice and currently in a high passage in vitro). The cell lines were maintained in Dulbecco’s modified Eagle medium supplemented with 10% heat inactivated fetal bovine serum and 1% antibiotics (0.05 mg penicillin/
ml, 0.05 mg streptomycin/ml, and 0.1 mg neomycin sulphate/ml) at 37°C in an atmosphere of 5% CO2 in air. Radiation cell survival assays All cell lines were studied in exponential phase. Single cell suspensions were inoculated on 25 cm* plastic flasks at cell numbers appropriate for colony counting. Heavily irradiated feeder cells were added to the viable cells to bring the total to 40,000-50,000 cells/flask (feeder cells were used in 16 cell lines; for HGL4, HGLS, HGLl 1, HGL 12 and HGL 13, feeder cells provided no advantage in plating efficiency and were not employed). In all experiments, the flasks were seeded at 2 or 3 different cell densities for each of 6 or 7 radiation dose levels. At 18 to 24 hr after plating, the cells were administered single radiation doses under aerobic conditions. The irradiation was performed using a 250 kVp X ray machine with HVL of 0.4 mm Cu at a dose rate of 1.54 Gy/min. The cells were irradiated at room temperature (20-22”C), the flasks placed on a rotating lucite block. After a period of 2 to 3 weeks of incubation at 37°C in an atmosphere of 5% CO2 in air, the cells were fixed with methanol, stained with crystal violet and counted. The number of colonies was determined by counting aggregates containing more than 50 cells. Six to eight replicate flasks were plated for each dose level. Three to five survival curves were performed for 14 cell lines and two survival curves for seven cell lines. Consideration of the technical factors that could influence the results include an increase in cell number between
Intrinsic radiosensitivity of glioblastoma multiforme 0 Table 2. Clinical outcome Cell line
of 13 patients
Histology
and the corresponding
57
A. TAGHIAN el al.
cell lines, histology,
SF,, and MID
SF2
MID
Clinical status Dead from tumor progression 5 months after surgery Dead from disease at 8 months Alive at 36 months Dead from disease at 13 months Dead from disease at 10 months Dead from disease at 26 months Dead from disease at 14 months Dead from disease at 19 months Alive at 14 months Dead from disease at 5 months Treated for an AST grade 2. then 48 months later presented a recurrent GBM and died after 6 months Alive at 12 months Alive at 8 months
A2*
GBM
0.50
2.55
A7 Cl G9 G19 G25 HGL4 HGLS HGL9 HGLll HGL12’
GBM Ast. 314 GBM GBM Ast. 314 GBM GBM GBM GBM GBM
0.59 0.35 0.46 0.37 0.19 0.39 0.50 0.56 0.64 0.76
3.02 1.81 2.29 1.81 1.1 I 1.94 2.49 2.74 3.30 3.61
HGLl3* HGL16
GBM GBM
0.74 0.40
3.72 1.96
* The patient did not receive post-operative RT because of medical problems. + The cell line was established-from the local recurrent GBM. * The patient received a total dose of 90 Gy of protontherapy.
inoculation
and irradiation,
and cell cycle redistribution.
of cases, feeder cells were used and the potential proliferation of test versus lethally irradiated cells could not be distinguished. In view of this, we monitored the change in cell number at various times after inoculation of 100,000 test cells. The cell number increase between 0 and 24 hr was less than 1.1 and the potential error caused by cell multiplicity at the time of irradiation was found to be trivial. The cell cycle distribution at the time of radiation was not evaluated. However, it is highly unlikely that redistribution could have given rise to the large variation in SF2 observed in this study between cell lines. In the majority
Statistical analysis The analysis of variance was performed on the pooled data from all experiments for each cell line, using a software package.* For the calculation of a! and p, 1nSf of all data versus dose were fitted by polynomial regression. For the calculation of Do and ii, 1nSf (the linear part of the curve < 0.1 SF), versus dose was fitted by simple linear regression. All data points were weighted equally. The mean inactivation dose MID was calculated from the LY and p by a previously published method (12). Additional statistical analysis of the data presented in this publication, including a comparison of the relative and absolute SF,‘s among three investigators on the same cell lines, will be presented in another paper. Clinical data The clinical status of 13 patients was known. Table 2 records the status at times after diagnosis as dead of dis-
* StatWorks,
Cricket Software,
Philadelphia,
PA.
ease, or alive. The clinical data concerning the patients Gl, G9, G19, and G25 were generously provided by Dr. Jonathan Ramsay at Cambridge (U.K.). Data are not available on eight patients because the cell lines were established in other laboratories (cf Table 1 for the origin of each cell line). RESULTS
Survival curve parameters The values for survival curve parameters (plating efficiency, Do, ii, (Y,/I, SF2 calculated and SF, measured) as well as the histology and the origin of each cell line are shown in Table 1. The plating efficiency (PE) varied between 2 and 85%. No correlation was found between PE and SF2. Figure 1 presents a plot of the cumulative frequency of the calculated SF2, the MID, and the CYof the 2 1 cell lines. The mean calculated SF2 was 0.5 1 + 0.14 with a range of 0.19 to 0.76 and a CV of 28%. The mean MID was 2.57 + 0.7 1 Gy ranging from 1.11 to 3.94 and a CV of 28%. The mean (Ywas 0.3 14 + 0.18 with a maximum of 0.863 and a minimum of 0.038 and a CV of 56%. The broad cumulative frequency for values of these parameters indicate a very wide range of intrinsic sensitivity for malignant glioma cell lines as measured in vitro. There was no significant difference between the calculated SF2 (0.5 1 + 0.14) and the measured SF2 (0.56 + 0.17) which shows that the linear quadratic model fits the data well. The mean SF2 of five cell lines studied previously (32) was 0.49 compared to 0.5 1 in the present study. The Do was calculated from the linear component of the survival curve down to 10p3. The mean Do was 1.59
I. J. Radiation Oncology 0 Biology 0 Physics
Volume 23, Number 1, 1992
2.57+-0.71
0.0
02
0.4
0.6
0.6
Calculated
10
0.2
00
0.4
SF2
0.6
0.6
1 .o
alpha
Fig. 1. Cumulative frequency of the calculated surviving fraction at 2 Gy (SF,), the (Y(linear quadratic model) and the mean inactivation dose (MID) of the 2 1 glioblastoma and high grade astrocytoma cell lines.
+ 0.41 ranging between 1.05 and 2.86 with a coefficient of variation (CV) of 26%. The mean ii was 4.5 + 2.7 with a minimum of 1.4 and a maximum of 10.5 and a CV of 60%. The mean a//3 ratio was 8.7 + 13.4 (range 0.7750.0). Newly versus well established cell lines Figure 2 demonstrates the cumulative frequency of SF;! in newly established (< 12 passages) versus well-established cell lines. The mean SF2 of newly established cell lines (12) was 0.49 + 0.17 and that of well-established (nine cell lines) was 0.53 -+ 0.10. This difference was not significant (p = 0.533), although the coefficients of variation were 35% and 19% for the newly and well established cell lines, respectively. Histology The pathology reports and/or slides of the nine tumors from which the cell lines were derived and established in our lab were reviewed by Dr. David Louis of the Pathology Department (Neuropathology) at MGH. Eighteen of the 2 1 cell lines were glioblastoma multiforme, the cell lines G 1 and G25 were astrocytoma grade 3/4 and the cell line EO 1 oligodendroglioma grade 3. The average SF2 of glioblastoma multiforme (18 cell lines) was 0.53 + 0.12. SF2 of the two astrocytoma grade 3/4 were 0.35 and 0.19. The oligodendroglioma cell line (EOl) had an SF2 of 0.59.
Clinical outcome Table 2 shows the clinical outcome, histology, SF2, and MID for the tumors from 13 patients. Four patients are alive without evidence of local recurrence (Gl, HGL9, HGL 13, and HGL 16) with follow-up period of 8, 12, 14, and 36 months. The other patients died from disease at 5 (two patients), 8 (two patients), 10, 13, 14, 19, and 26 months (one patient each). Patient HGL12 was treated for an astrocytoma grade 2 in August 1985, then presented with a GBM as local recurrence in June 1989, and died in December 1989. The cell line studied was established from the recurrent tumor. Most patients were treated by some category of surgical resection and post-operative radiation (-60 Gy) with or without chemotherapy. One patient (HGL 13) received 90 Gy by proton therapy. Patient A2 did not receive postoperative radiation therapy because of a medical problem, and died 5 months later from tumor progression. Two patients presented with astrocytoma grade 3/4: (Gl and G25), one is still alive with a follow-up of 36 months (Gl), the other died after 26 months (G25). Three of the nine patients with glioblastoma multiforme are alive with a follow-up period of 8, 12, and 14 months. The other patients survived a mean period of 11.2 months. Figure 3 shows the survival of GBM patients in months, plotted against SF*. There is no evident correlation between SF2 and survival. Patient (Gl), considered as a long-
30-
1.0-
z 5
0.6 -
zo-
s
0.6 -
2
m
lo-
5
?? ??
5 v) 0 0.0
3
o.o! 0.0
.
I
0.2
.
I
.
0.4
1
0.6
-
I
0.6
.
,
? ? alive
?? m0
.,.,.,.I., 0.2 0.4
??
?? dead
?? ??
0.6
0.8
1.0
SF2
1.0
SF2
Fig. 2. Cumulative frequency of SF, of the newly-established versus well-established glioblastoma and high grade astrocytoma cell lines.
Fig. 3. SF, of GBM plotted against the survival (in months) of patients. Two cell lines were not plotted: HGL12 because the cell line was derived from a recurrent GBM that was not irradiated, and A7 because the patient did not receive radiation because of medical problems.
Intrinsic radiosensitivity of glioblastoma multiforme 0 A. TAGHIAN
term survivor (36 months), had SF2 of 0.35. Patient G25 died from disease after 26 months had an SF2 of 0.19. However, patients with relatively high SF2, like HGL9 and HGL13 (0.56 and 0.74), are alive without evidence of disease at 14 and 12 months, respectively. On the other hand, patients with relatively low SF2, like G 19 and HGL4 (0.37 and 0.39) died from disease at 10 and 14 months, respectively.
DISCUSSION In a general review of malignant glioma, Davis (9) emphasized that one of the major causes of failure of radiation treatment of high grade glioma could be a low intrinsic radiation sensitivity. The aim of our study was to evaluate the importance of that radiation sensitivity as measured by loss of colony formation of malignant glioma cells studied in vitro. Several workers have suggested that the initial low-dose region of the radiation survival curves is the most relevant to any consideration of the response of human tumors to the fractionated radiation of conventional therapy ( lo- 13,20). Hence, the interest of most of the recent published studies and the present work, on radiation sensitivity determinations of human tumor cells, has concentrated on parameters relating to the low dose region of the survival curves (a, MID, and SF2) (10-l 3, 20). Results of these determinations were calculated for all the 2 1 high grade brain tumor cell lines. Comparison with glioblastoma cell lines from the literature There are only few published data on the radiosensitivity of glioblastoma or high grade astrocytoma. Allalunis-
59
et al.
Turner et al. ( 1) studied 16 early passage glioblastoma cell lines, using colony formation on plastic as an end-point. They found a mean SF2 of 0.52 f 0.22 with a range from 0.15 to 0.93 and a CV of 42%. Only a few cell lines (1 to 3) have been studied in the other reports. Table 3 presents published values for the radiosensitivity parameters of glioma cell lines. The variation in radiosensitivity cannot be appreciated from the low numbers of cell lines as described in each report, however, when they are grouped together, a quite wide variation is evident which is not greatly different from that observed here (Fig. 4). The mean SF2 calculated from these 15 cell lines from the literature was 0.57 + 0.16 (CV 28%) ranging from 0.3 1 to 0.75 (note that most cell lines were late passage). Figure 4 illustrates the cumulative frequency of SF2 from the literature (Table 3) and from Allalunis-Turner et al. (1) as well as the MGH data. The three curves are almost superimposable. They illustrate a wide range of intrinsic radiation sensitivity: the most radiosensitive being an SF2 of 0.15, the most radioresistant being an SF2 of 0.93 and the mean SF2 for the 52 GBM cell lines was 0.53. Comparison with other tumors Table 4 shows SF:! values of other types of tumors with the number of cell lines studied and the CV of each group. Using colony formation on plastic as the end-point of cell viability, no significant difference was found between the mean SF2 of our malignant glioma cell lines of 0.5 1, and the 0.46 of ovarian cancer (35) the 0.49 of colon cancer (17) the 0.43 of melanoma (26), the 0.44 of the SCC of the head and neck (H&N) (general review 20), and the 0.47 of cervical cancer (the last used primary culture and soft agar method) (8).
Table 3. SF2, (Y,/3 and MID values from the literature on glioblastoma multiforme and high grade astrocytoma cell lines Author Yang et al. (37)
Cell line IN 859 IN 1265 SB
Schultz and Geard (29) Rhaapost et al. (25) Masuda et al. (2 1) Gerweck et al. (14)*
Weischelbaum et al. (36)* Nilsson et al. (22)* Allalunis-Turner et al. (1) Present study
U-87 MG U-138 MG U-373 MG KNS-42 KNS-60 A3 A2 A7 TX-13 U- 118-MG 16 cell lines 21 cell lines
Histology Astrocyt. gr. 3 & 4
Ast. gr. 3 Ast. gr. 4 GBM GBM Ast. high gr. GBM GBM Ast. gr. 4 Ast. gr. 4t Ast. gr. 3t GBM GBM GBM GBM + AST3/4
SF2
0.57 0.73 0.47 0.58 0.46 0.70 0.60 0.75 0.39 0.33 0.72 0.44 0.72 0.31 0.75 0.52 0.51
P
MID
0.31 0.042 0.36 0.20 0.30 0.15 0.20 0.094
0.033 0.053 0.0077 0.036 0.045 0.03 1 0.035 0.033
2.30 3.47 2.53 2.81 2.18 3.30 2.83 3.74
0.086 0.006 0.097 0.559 0.0625 0.314
0.0394 0.0394 0.0319 0.0152 0.04
3.55 2.11 3.76 1.62 3.70 2.6
a
0.026
* No specification of SF2, a, @or MID was in the paper, the calculations of these values were taken from Ref. 11. t The histologic slides of the tumors from which these two cell lines were derived were reviewed by Dr. David Louis. They were diagnosed as glioblastoma multiforme. * SF2 and MID were calculated from cxand /3 published in the paper (22).
I. J. Radiation Oncology 0 Biology0 Physics
60
Fig. 4. Cumulative frequency of SF2 of glioblastoma and high grade astrocytoma. Comparison between data from the literature, data from Allalunis-Turner et al. (1) and our data (MGH).
However,
using the same end-point
and all assays per-
formed by one group, a significant difference was reported by Allalunis-Turner et al. (I) between the SF2 for malignant glioma (0.52) on the one hand and SCC of the uterine cervix (0.27) and adenocarcinoma of the endometrium (0.25) on the other hand. The SF2 values for SCC of H&N (0.43) and sarcoma (0.34) obtained by Weichselbaum et al. (35) and for SCC of H&N (0.32) from Brock et al. (5) are significantly lower then we found for GBM (0.5 I), (the last (5) used primary culture and population growth method on CAM plates). Early versus late passages
Figure 2 illustrates the difference between the early established and late passage GBM cell lines. The mean SF, of the early established was 0.49 + 0.17 ( 12 cell lines) versus 0.53 + 0.10 (nine cell lines) for the late passage lines. This difference was not significant (p = 0.533). The wide distribution of intrinsic radiosensitivity was more evident in the early (coefficient of variation 35%) than in
Volume 23, Number 1, 1992 late passage cell lines (CV 19%). This was comparable with the CV of the 16 early passage GBM cell lines studied by Allalunis-Turner et al. (1) (42%). In a general review (32), the authors pointed out the difference in SF;!between the early and late passages. Using colony formation as the end-point for cells growing on plastic, SF, values for SCC and adenocarcinoma were 0.50 (33 cell lines), 0.49 (20 cell lines) for well-established, versus 0.25 (14 cell lines) and 0.28 (12 cell lines) for primary culture, respectively. It should be noted, however, that the primary culture used the soft agar assay. Also using the soft agar assay, Rofstad (26, 27) studied the radiosensitivity in vitro of melanoma cell lines. The mean SF, values of 18 primary culture cell lines was 0.38 f 0.2 (CV 53%) compared to 0.50 + 0.17 (CV 34%) for 11 wellestablished cell lines. This difference was in the borderline of significance (p = 0.055). In general, the late passage cell lines tend to be less sensitive and with a lower coefficient of variation than the early passage and the primary cultures, however, few published data are available on this point. Histology
The comparison between the SF*‘s of GBM and AST grade 314 is difficult because we studied only two AST grade 3/4 cell lines. The radiation response of three astrocytoma cell lines across grade was studied recently (29). The authors found SF2’s of 0.58 and 0.46 for grade 3 and 4, respectively, in comparison to 0.36 for a single grade 1 cell line. They concluded that the low grade astrocytoma cell lines are more sensitive than the high grade lines in vitro. These data are compatible with the clinical results, but one cannot make conclusions based on findings from a single cell line. Furthermore, in our data we found (G25),
Table 4. Mean SF2 values of several types of tumors, the type of passage, the number of cell line studied as well as the CV of each group
Authors Davidson et al. (8) Brock et al. (5) Weischelbaum al. (35)
Type of tumor
Type of passage
SCC of uterine cervix
Primary culture (Courtenay Mills) Primary culture (population growth) Early passage
SCC from H & N et
Leith et al. (17) Rofstad (26) (general review) Malaise et al. (20) (general review) Allalunis-Turner et al. (1)
SCC from H & N
No. of cell lines
SF, @SE)
cv @)
52
0.47 + 0.18
38
140
0.32 + 0.15
47
24
0.43 * 0.12
29
15 14 16 21
0.46 0.34 0.49 0.43
33 29 23 41
Ovarian cancer Sarcoma Colon cancer *Melanoma
Early passage Early passage Late passage Primary and late passage
see
Late passage
9
0.44 + 0.16
36
SCC of uterine cervix Endometrium
Early passage Early passage
17 19
0.27 f 0.14 0.25 + 0.09
51 37
* The SF,% were calculated from (Yand p published in the paper (26).
+ + f. +
0.15 0.098 0.115 0.18
Intrinsic radiosensitivity of glioblastoma multiforme 0 A. TAGHIAN et al.
an astrocytoma grade 3/4, and (G 19) a glioblastoma multiforme, with SF, of 0.19 and 0.37, respectively. This shows that we may have a high grade AST or a GBM with a relatively low SF2 as measured in vitro. Correlation between SF, and clinical outcome SF2 is being evaluated as a parameter for prediction of the radiation response of tumors (4, 6, 19, 24, 34). The aim of this study was not to test the predictive value of intrinsic radiosensitivity measurement expressed by SF2 for tumor response to radiotherapy, but the value of SF2 as a factor in the poor clinical results of GBM. However, the authors found it interesting to look to the clinical outcome of the patients from which the tumor cell lines were derived. For our small number of glioblastoma multiforme patients, we did not find a correlation between the SF2 measured in vitro and the clinical outcome (Fig. 3); some of whom have only short follow-up. We noticed that the relatively long survival was a patient with an astrocytoma grade 3/4 (Gl) with an SF2 of 0.35, but conclusions cannot, of course, be drawn from such scant clinical data. Nevertheless, the wide range of intrinsic radiation sensitivity of malignant gliomas cell lines found in our study, as well as the Allalunis-Turner study (1) and data from the literature grouped together (Fig. 4) could not explain the quasi-homogeneous fatal clinical outcome of patients with these tumors. Probably mechanisms not evaluated by the in vitro assays, which modify radiation sensitivity are not present in vitro where the cells are under uniform and near optimal metabolic conditions, do operate in vivo and effect large shifts in cellular radiation sensitivity. If
61
so, the lack of concordance between the simple in vitro assay results and clinical outcome is not a cause for surprise. Similar conclusion was found by Bentzen et al. (2) when they found that the dose-control curves of malignant melanoma derived from in vitro radiosensitivity data are significantly steeper than the clinical dose-control curves. Such conclusions would limit the predictive value of current radiation sensitivity assays based on in vitro dosesurvival measurement for at least high grade malignant gliomas.
CONCLUSION The average SF2 for the GBM cell lines studied here was in the upper range of intrinsic radiosensitivities as reported in the literature, however, there was extensive overlap with SF2’s for cells from tumors often treated effectively. Our results demonstrate a wide range of intrinsic radiation sensitivity that is not compatible with the almost invariably fatal outcome of glioblastoma multiforme and high grade malignant gliomas, if comparisons of measured SF2s have any absolute value. From these data, we conclude that either standard measurements of cellular radiation sensitivity (SF,) in vitro of GBM do not yield values that track in a simple manner with the local control probability at the clinical level, or the intrinsic radiation sensitivity of malignant gliomas is not the most important factor in the determination of the in vivo radiation resistance of these tumors and is probably overshadowed by other factors like stem cell fraction, proliferation, hypoxia, or others.
REFERENCES I. Allalunis-Turner, M.; Pearcey, R.; Urtasun, R. Inherent radiosensitivity testing of human tumor biopsy specimens. 38th Annual Meeting of the Radiation Research Society, New Orleans, Louisiana, April 7- 12, 1990; Abstract Cn- I, p. 67. 2. Bentzen, S.; Thames, H.; Overgaard, J. Does variation in the in vitro cellular radiosensitivity explain the shallow clinical dose-control curve for malignant melanoma? Int. J. Radiat. Biol. 57: 117- 126; 1990. 3. Bleehen, N. M.; Wilshire, P. N.; Plowman, P. N.; Watson, J. V.; Gleave, J. R.; Holmes, A. E.; Lewin, W. S.; Treip, C. S.; Hawkins, T. D. A randomized study of misonidazole and radiotherapy for grades 3 and 4 cerebral astrocytoma. Br. J. Cancer 43: 436-442; 198 1. 4. Bristow, R.; Hardy, P.; Hill, R. Comparison between in vitro radiosensitivity and in vivo radioresponse of murine tumor cell lines 1: parameters of in vitro radiosensitivity and endogenous cellular glutathione levels. Int. J. Radiat. Oncol. Biol. Phys. 18: 133-145; 1990. 5. Brock, W.; Peters, L. Radiosensitivity of head and neck squamous cell carcinomas and local tumor control (Abstract). 25th Paterson Symposium, 3-5 December, 1990, Paterson Institute for Cancer Research, Manchester, UK. 6. Brock, W.; Maor, M.; Peters, L. Predictors oftumor response to radiotherapy. Radiat. Res. 8(Suppl.): S290-S296; 1985. 7. Chang, C. H. Hyperbaric oxygen and radiation therapy in
8.
9.
10.
11.
12.
13.
14.
the management of glioblastoma. NC1 Monogr. 46: 163169; 1977. Davidson, S.; West, C.; Roberts, S.; Hendry, J.; Hunter, R. Radiosensitivity testing of primary cervical carcinoma: evaluation of intra- and intertumour heterogeneity. Radiother. Oncol. 18: 349-356; 1990. Davis, L. Presidential address: malignant glioma-a nemesis which requires clinical and basic investigation in radiation oncology. Int. J. Radiat. Oncol. Biol. Phys. 16: 1355-1365; 1989. Deacon, J.; Peckham, M. J.; Steel, G. The radioresponse of human tumors and the initial slope of the cell survival curve. Radiother. Oncol. 2: 3 17-323; 1984. Fertil, B.; Malaise, E. P. Inherent cellular radiosensitivity as a basic concept for human tumor radiotherapy. Int. J. Radiat. Oncol. Biol. Phys. 7: 621-629; 198 1. Fertil, B.; Dertinger, H.; Courdi, A.; Malaise, E. P. Mean inactivation dose: a useful concept for intercomparison of human cell survival curves. Radiat. Res. 99: 73-84; 1984. Fertil, B.; Malaise, E. P. Intrinsic radiosensitivity of human cell lines is correlated with radioresponsiveness of human tumors: analysis of 101 published survival curves. Int. J. Radiat. Oncol. Biol. Phys. 11: 1699-1707; 1985. Gerweck, L.; Kornblith, P.; Burlett, P.; Wang, J.; Sweigert, S. Radiation sensitivity of cultured human glioblastoma cells. Radiology 125: 231-234; 1977.
62
I. J. Radiation Oncology 0 Biology0 Physics
15. Green, S. P.; Byar, D. P.; Strike, T. A.; Alexander, E.; Brooks, W. H.; Burger, P. C.; Hunt, W. E.; Mealey, J.; Odom, G. L.; Paoletti, P.; Pistenma, D. A.; Ransohoff, J.; Robertson, J. T.; Selker, R. G.; Shapiro, W. R.; Smith, K. R. Randomized comparisons of BCNU, streptozotocin, radiosensitizer, and fractionation of radiotherapy in the post-operative treatment of malignant glioma (study 7702). Proc. Amer. Sot. Clin. Oncol. 3: 260; 1984. 16. Laramore, G. E.; Diener-West, M.; Griffin, T. W.; Nelson, J. S.; Griem, M. L.; Thomas, F. J.; Hendrickson, F. R.; Griffin, B. R.; Myrianthopoulos, L. C.; Saxton, J. Randomized neutron dose searching study for malignant gliomas of the brain: results of an RTOG study. Int. J. Radiat. Oncol. Biol. Phys. 14: 1093-l 102; 1988. 17. Leith, J.; Faulkner, L.; Papa, G.; Quinn, P.; Michelson, S. In vitro radiation survival parameters of human colon tumor cells. Int. J. Radiat. Oncol. Biol. Phys. 20: 203-206; 1991. 18. Loeffler, J.; Alexander, E.; Wen, P.; Shea, M.; Coleman, N.; Kooy, H.; Fine, H.; Nezdi, L.; Silver, B.; Riese, N.; Black, P. Results of stereotactic brachytherapy used in the initial management of patients with glioblastoma. JNCI 82: 19 181921; 1990. 19. Malaise, E. P.; Fertil, B.; Chavaudra, N.; Guichard, M. Distribution of radiation sensitivities for human tumor cells of specific histological types: comparison of in vitro to in vivo data. Int. J. Radiat. Oncol. Biol. Phys. 13: 617-624; 1986. 20. Malaise, E. P.; Fertil, B.; Deschavanne, P.; Chavaudra, N.; Brock, W. Initial slope of radiation survival curves is characteristic of the origin of primary and established cultures of human tumor cells and fibroblasts. Radiat. Res. 111: 3 19333; 1987. 21. Masuda, K.; Aramaki, R.; Takaki, T.; Wakisaka, S. Possible explanation of radioresistance of glioblastoma in situ. Int. J. Radiat. Oncol. Biol. Phys. 9: 255-258; 1983. 22. Nilsson, S.; Carlsson, J.; Larsson, B.; Ponten, J. Survival of irradiated glia and glioma cells studied with a new cloning technique. Int. J. Radiat. Biol. 37: 267-279; 1980. 23. Payne, D. G.; Simpson, W. J.; Keen, C.; Platts, M. E. Malignant astrocytoma. Hyperfractionated and standard radiotherapy with chemotherapy in a randomized prospective clinical trial. Cancer 50: 2301-2306; 1982. 24. Peters, L.; Brock, W.; Johnson, R.; Meyn, R.; Tofilon, P.; Milas, L. Potential methods for predicting tumor radiocurability. Int. J. Radiat. Oncol. Biol. Phys. 12: 459-467; 1986.
Volume 23, Number 1, 1992 25. Raaphorst, G. P.; Feeley, M.; Da Silva, V.; Danjoux, C.; Gerig, L. A comparison of heat and radiation sensitivity of three human ghoma cell lines. Int. J. Radiat. Oncol. Biol. Phys. 17: 615-622; 1989. 26. Rofstad, E. Radiation biology of malignant melanoma. A review article. Acta Radiol. 25: l-10; 1986. 27. Rofstad, E.; Wahl, A.; Brustad, T. Radiation sensitivity in vitro of cells isolated from human tumor surgical specimens. Cancer Res. 47: 106-l 10; 1987. 28. Sheline, G. E. Radiation therapy of primary tumors. Cancer 39: 873-881; 1977. 29. Shultz, C.; Geard, C. Radioresponse of human astrocytic tumors across grade as a function of acute and chronic irradiation. Int. J. Radiat. Oncol. Biol. Phys. 19: 1397- 1403; 1990. 30. Stage, W. S.; Stein, J. J. Treatment of malignant astrocytomas. Am. J. Roentgenol. 120: 7-18; 1974. 31. Stenning, S. P.; Freedman, L. S.; Bleehen, N. M. An overview of published results from randomized studies of nitrosoureas in primary high grade malignant glioma. Br. J. Cancer 56: 89-90; 1987. 32. Suit, H.; Baumann, M.; Skates, S.; Convery, K. Clinical interest in determinations of cellular radiation sensitivity. Int. J. Radiat. Biol. 56: 725-737; 1989. 33. Suit, H.; Zietman, A.; Tomkinson, K.; Ramsay, J.; Gerweck, L.; Sedlacek, R. Radiation response of xenografls of a human squamous cell carcinoma and a glioblastoma multiforme: a progress report. Int. J. Radiat. Oncol. Biol. Phys. 18: 365373; 1990. 34. Tucker, S.; Thames, H. The effect of patient-to-patient variability on the accuracy of predictive assays of tumor response to radiotherapy: a theoretical evaluation. Int. J. Radiat. Oncol. Biol. Phys. 17: 145-155; 1989. 35. Weichselbaum, R.; Rotmensch, J.; Ahmed-Swan, S.; Beckett, A. Radiobiological characterization of 53 human tumor cell lines. Int. J. Radiat. Biol. 56;5: 553-560; 1989. 36. Weichselbaum, R. R.; Epstein, J.; Little, J. B. In vitro cellular radiosensitivity of human malignant tumors. Europ. J. Cancer 12: 47-51; 1976. 37. Yang, X.; Darling, J. L.; McMillan, T. J.; Peacock, J. H.; Steel, G. G. Radiosensitivity, recovery and dose-rate effect in three human glioma cell lines. Radiother. Oncol. 19: 4956; 1990.
