AMOXIFEN, a nonsteroidal antiestrogen, has beenwidely used in the treatment of breast cancer.10

It also has actions that are independent of estro-gen,16,18 including cellular phospholipase C– and D–medi-ated inhibition of protein kinase C (PKC) activity.2,8,14,17

The activity of PKC has been linked to the growth rateof glioblastoma multiforme (GBM).5,6 Tamoxifen inhibitsDNA synthesis in GBM in vitro15 and is now being usedin brain tumor therapy.1,7,20

Via its interaction with PKC, tamoxifen may block thecell cycle in its most radiosensitive phase, G2/M.19 There-fore, there is a theoretical basis on which one can proposethat the effects of radiation and tamoxifen therapies aresynergistic and not merely additional. The only work pub-lished so far on the role of tamoxifen as a radiosensitizerin treating GBM21 reported synergy of action; however,there were methodological flaws in that study. No attemptwas made to quantify synergistic, additive, or antagonisticeffects, and only one nonhuman GBM cell line was used.In the present study, we investigated the interaction oftamoxifen and radiation therapies against three GBM celllines of human origin. The combined effect of the two

modalities was quantified using the median-effect prin-ciple and combination index (CI) method of Chou andTalalay.4

Materials and Methods

Cell Culture

Three GBM cell lines, G12, D3, and D19, were established fromprimary tumor by using an Institutional Review Board–approvedprotocol. The tissue culture medium used to propagate cell lines wasRPMI-1640 supplemented with 10% fetal calf serum, 1 mM L-glu-tamine, and 1 mM penicillin/streptomycin. Cell lines were propa-gated at 37˚C in a humidified atmosphere containing 5% CO2.

The cell lines were characterized as GBM by phenotyping andkaryotyping. The karyotype of the G12 cell line was 76,XY,+X,+Y,+1,+2,+3,+5,+5,26,+7,+7,+8,+12,+13,+14,+15,+15,+16,+16,+17,+18,+18,+20,+20,+21,+21,+7mar. The karyotype of the D3cell line was 44, XY, t(1,3)(q25;q26.2),t(1;4)(p13;q23),t(4/12)(q12;q15),add(6)(p11.2),der(7)t(7;20)(q32;q11.2),der(8)t(6;8)(p11.2;p11.2),210,add(10)(q11.2 or q22),214,del(18)(p11.2), der(20)t(8;20)(p11.2;q11.2),der(22)t(10;22)(q22.1;q13.a),0210dmin[9]/44,idem,der(4)(p16)t(1p;4)(p13;q23)[11]. The karyotype of the D19 cellline was 42-45, XX, der(1)t(1;?7)(q26;q22),t(1/12) (p36;q13),

J. Neurosurg. / Volume 90 / March, 1999

J Neurosurg 90:533–536, 1999

Tamoxifen radiosensitization in human glioblastoma celllines

ANDREW M. DONSON, B.SC., MICHAEL D. WEIL, M.D.,AND NICHOLAS K. FOREMAN, M.B.CH.B.

University of Colorado Health Sciences Center, Denver, Colorado; and Department of PediatricOncology, The Children’s Hospital, Denver, Colorado

Object. A combined tamoxifen and radiation therapy is being used in clinical trials to treat glioblastoma multi-forme (GBM). The rationale behind this therapy is that tamoxifen is a radiosensitizer. However, the evidence forthis is weak. The authors, therefore, examined the effect of combined radiation–tamoxifen therapy in three GBMcell lines of human origin.

Methods. The GBM cell lines were exposed to different concentrations (0.3–5 mg/ml) of tamoxifen and subse-quently irradiated at varying doses (0.8–5 Gy). Tumor growth inhibition was measured using a proliferation assay.The interaction of tamoxifen and radiation therapies was quantified using the combination index method, which dis-tinguishes whether a combined antitumor effect is synergistic, additive, or antagonistic. At high doses of tamoxifenor radiation there was significant inhibition of tumor cell proliferation. At low doses of either therapeutic agent,there was little effect. In one cell line, synergism occurred at high doses of tamoxifen and radiation. In the other twocell lines, an additive effect was observed. In only one of the three cell lines was there synergy between tamoxifenand radiation at doses that significantly inhibited tumor proliferation.

Conclusions. Because synergy could not be demonstrated in all three cell lines at active dosages, the clinical com-bination of tamoxifen and radiation therapies may not be of benefit to all patients.

KEY WORDS • glioblastoma multiforme • radiation therapy • synergism • tamoxifen

T

533

?del(2)(q31),t(2;17)(p13;q11.2),der(3)t(1;3)(p34;q29),t(4;15) (q21.3;q26.1),(4;7)(p12;q22),?add(5)(q13),der(7)del(7)(p12p13)t(1;7)(23;q11.2)del(8)(p23),210,?del(12)(q13.3),213,26[7],der(16)t(16;21)(p11.2;q11.2)[2]222[6],add(22)(q13)[2][cp9]/87,89,idem[cp4]/45,XX,add(1)(p36),der(2)t(2;12)(p25;q13)t(2;?13)(q13;q12),der(4)t(4;7)(p?27;q11.2),t(4;11)(q23;q13),del(7)(p13;p15),add(7)(p22;q22),210,add(11)(?p13),213,+del(17)(p11.2,?add(22)(q13)[cp3].

Morphological Characteristics of GBM Cell Lines

Hematoxylin and eosin staining of the GBM cells revealed thefollowing characteristics. In the G12 cell line there were spindle-shaped cells, fibrillary processes, and spheroid clusters. There weresome giant cells and typical and atypical (ring and triploid) mitoticfigures. A high percentage of G12 cells were in mitosis and the cellshad irregular nuclei (pleiomorphic, prominent multiple nucleoli). Inthe D3 cell line there were polygonal or spindle-shaped cells, somefibrillary processes, and spheroid clusters. There were multinucleategiant cells and typical and atypical (ring and triploid) mitotic fig-ures. The cells had irregular nuclei (pleiomorphic, prominent multi-ple nucleoli). In the D19 cell line there were spindle-shaped cells,fibrillary processes, and spheroid clusters. There were multinucleategiant cells and typical and atypical (ring and triploid) mitotic fig-ures. The cells had irregular nuclei (pleiomorphic, prominent multi-ple nucleoli).

Doubling Time and Saturation Density of GBM Cell Lines

The cell doubling time is taken from the linear phase of growthcurve plots: 2.3 days for G12 cells; 2.8 days for D3 cells; and 6.1days for D19 cells. The saturation density is taken from the plateaupoint of the growth curve: 1.78 3 103 cells/mm2 for G12 cells; 2 3103 cells/mm2 for D3 cells; and 1.27 3 103 cells/mm2 for D19 cells.

Treatment of Cell Lines With Tamoxifen

Tamoxifen was freshly dissolved in dimethyl sulfoxide to pro-vide a 10 mg/ml stock solution. This solution was further diluted intissue culture media to achieve concentrations of 0.3125, 0.625,1.25, 2.5, and 5 mg/ml. Ten thousand cells per well were seeded into24-well plates and incubated for 24 hours. The wells were aspiratedand refed with 1 ml of tamoxifen solution in triplicate wells for eachconcentration. Control wells were refed with tissue culture mediumcontaining no tamoxifen. The cell lines were exposed to tamoxifenfor 24 hours, after which they were irradiated.

Treatment of Cell Lines With Radiation

The cell lines were irradiated using a gamma-beam 150Co source.The wells were aspirated to remove tissue culture media and thenwere irradiated at source-to-target distances of 13, 23, and 32 cm for1 minute, corresponding to doses of 5, 1.6, and 0.8 Gy, respective-ly. The wells were washed with phosphate-buffered saline once andthen refed with tamoxifen at the same concentrations as previouslyadministered. One 24-well plate of tamoxifen-treated cells was notirradiated.

Tritiated Thymidine Uptake

The proliferation rate of the cells was determined by 3H-thymi-dine incorporation. After irradiation, mitomycin C solution wasadded to three wells containing no tamoxifen to provide a final con-centration of 25 mg/ml. Mitomycin C blocks proliferation and, thus,provides a negative control for thymidine incorporation. After 30minutes, each well was pulsed with 1 mCi 3H-thymidine and theplates were incubated at 37˚C for a period of 4 hours prior to cellharvest. The wells were washed with phosphate-buffered saline and1 ml 6% trichloroacetic acid was added to each well. After incuba-tion for 1 hour at 4˚C, the wells were washed with 1 ml cold 6%trichloroacetic acid. The acid precipitate was dissolved overnight in500 ml 0.5 N NaOH and transferred to scintillation vials containing3 ml of ScintiSafe-30%. Incorporated radioactivity was measuredusing a scintillation counter. The dimethyl sulfoxide showed noeffect in the control assays performed.

Determination of Synergism and Antagonism and theConstruction of Isobolograms

The CI was calculated by the Chou–Talalay equation, whichtakes into account both the potency (Dm or IC50) and shape of thedose–effect curve (the m value).4 The general equation for the clas-sic isobologram (CI = 1) is given byCI = (D)l/(Dx)1 + (D)2/(Dx)2

where (Dx)1 and (Dx)2 in the denominators are the doses (or concen-trations) for D1 (tamoxifen) and D2 (radiation) alone that provide x%inhibition, whereas (D)1 and (D)2 in the numerators are the doses oftamoxifen and radiation that also inhibit x% (that is, they are isoef-fective). Combination index values less than 1, equal to 1, and morethan 1 indicate synergism, additive effect, and antagonism, re-spectively.

The (Dx)1 or (Dx)2 (for tamoxifen or radiation) can be readily cal-culated from the median-effect equationDx = Dm[f a /(12f a)]1/m

where Dm is the median-effect dose that is obtained from theantilog of the X-intercept of the median-effect plot, X = logD com-pared with Y = log[f a /(12f a )], and m is the slope of the median-effect plot.

Sources of Supplies and Equipment

The tissue medium was obtained from Gibco (Gaithersburg, MD)and the tamoxifen from Sigma Chemical Co. (St. Louis, MO). The3H-thymidine was purchased from Amersham (Arlington Heights,IL) and the ScintiSafe-30% from Fisher Scientific (Pittsburgh, PA).

Nordian Intl. (Kanata, Ontario, Canada) manufactured the gam-ma-beam 150Co unit. The LS6500 scintillation counter was obtainedfrom Beckman (Fullerton, CA).

Results

Treatment of Cells With Tamoxifen or Radiation Alone

The three GBM cell lines, G12, D3, and D19, with dif-ferent basal rates of proliferation were all inhibited bytamoxifen with Dm values between 1 and 4 mg/ml. Theindividual Dm values obtained from the median-effect plot(X = logD compared with Y = log[f a /(12f a)]) were 3.39mg/ml for G12, 1.68 mg/ml for D3, and 2.24 mg/ml forD19.

All cell lines were inhibited by radiation in a dose-relat-ed manner, with Dm values between 2.5 and 5.5 Gy. Theindividual Dm values from the median-effect plot (X =logD compared with Y = log[f a /(12f a)]) were 5.46 Gyfor G12, 2.6 Gy for D3, and 5.5 Gy for D19.

Treatment of Cells With Both Tamoxifen and Radiation

Combination index values for the effect of both tamox-ifen and radiation in cell lines G12, D3, and D19 areshown in Table 1. The proliferation rates for cells treatedby the combined radiation–tamoxifen therapy are shownin Fig. 1.

The combined tamoxifen–radiation therapy resulted inmultiphasic interactions in all three cell lines. In cell lineG12, there was synergism (CI , 1) at high dose ranges (5mg/ml tamoxifen and 5 Gy radiation). At lower concen-trations of tamoxifen and radiation (1.25 mg/ml tamoxifenand 1.6–0.8 Gy radiation), an antagonistic effect (CI . 1)was observed. Cell lines D3 and D19 behaved in a differ-ent fashion from cell line G12. In both D3 and D19 celllines, there was synergism at low tamoxifen and radiationdoses (0.31 mg/ml tamoxifen and 0.8 Gy radiation). Highdoses of tamoxifen (5 mg/ml) produced additive effects

A. M. Donson, M. D. Weil, and N. K. Foreman

534 J. Neurosurg. / Volume 90 / March, 1999

(CI = 1) and high radiation and low tamoxifen doses (5 Gyradiation and 0.31–0.62 mg/ml tamoxifen) producedantagonistic effects (CI . 1).

Discussion

Radiation is still the most effective modality, after sur-gery, for the treatment of the majority of brain tumors.True radiosensitization can be defined as a synergisticinteraction between a drug and radiation therapy. If achemotherapeutic agent has a synergistic effect whencombined with radiotherapy, it may be of particular value.

Over the past decades, various authors have claimedsynergism, additive effects, or antagonism as the effects ofmultiple drugs. There is still controversy as to the mean-ing of these terms. This is partly due to the lack of a theo-retical basis that would permit rigorous and quantitativeassessment of the effects of drug combinations. Chou andTalalay4 formulated a simple equation to evaluate syner-gism, additive effects, or antagonism for drugs, whichdoes not require the mechanism of action to be known.This method takes into account both the potencies of eachdrug and combinations of these drugs and the shapes oftheir dose–effect curves. Since its formulation, this meth-od has been used for quantitation of the combined effectof up to three antitumor drugs3 and the combined effect ofradiation and antitumor drug therapies.9 The action oftamoxifen and radiation are ideally suited to this typeof analysis. Only one study in which the interactions oftamoxifen and radiation were evaluated in GBM has beenpublished. Zhang, et al.,21 claimed to have demonstrated asynergistic effect of tamoxifen with radiation therapy, al-though no attempt was made to quantify this effect. Avalid criticism of this omission was made by Olsen in thecommentary accompanying this paper.

In this study we use the formula described by Chou andTalalay4 to produce CI values for the combined effect oftamoxifen and radiation therapies in treating glioblastomacell lines. We found a multiphasic synergism in the com-bined tamoxifen–radiation treatment in all three cell linesstudied. In one of the three cell lines, the synergism oc-curred at high concentrations of tamoxifen and radiation,

which resulted in a strong antitumor effect. In the remain-ing two cell lines, the effect of high doses of tamoxifenand radiation provided only additive effects. In these twocell lines, synergism occurred at low doses of tamoxifenand radiation. The levels of antitumor activity at thesedoses, however, have little impact on tumor proliferation.Therefore, when high doses of tamoxifen are used, com-bination therapy may be of synergistic value in treatingsome GBMs but of additive value only in treating others.

J. Neurosurg. / Volume 90 / March, 1999

Tamoxifen radiosensitization in GBM cells

535

TABLE 1Combination index values for the combination of radiation ther-

apy and 24-hour administration of tamoxifen at various doseand effect levels in three GBM cell lines*

Cell Line & Concentration of Tamoxifen (mg/ml)Radiation

Level (Gy) 0.3125 0.625 1.25 2.5 5.0

G12 cell line0.8 1.71 2.60 1.62 1.26 0.621.6 0.62 3.63 1.34 1.45 0.585.0 1.00 1.46 0.88 0.55 0.26

D3 cell line0.8 0.68 0.90 1.21 1.47 —1.6 1.12 0.83 0.97 1.28 0.875.0 21.04 5.55 2.04 1.67 1.29

D19 cell line0.8 0.40 0.83 0.62 0.95 0.981.6 0.69 1.26 0.87 0.70 0.895.0 4.23 6.33 1.69 1.55 0.91

* — = not calculated.

FIG. 1. Graphs displaying the combined effects of tamoxifenand radiation on GBM cell lines G12 (upper), D3 (center), andD19 (lower). Proliferation rate curves are shown for each cell lineirradiated at 0, 0.8, 1.6, and 5 Gy after a 24-hour exposure totamoxifen at 0, 0.3125, 0.625, 1.25, 2.5, and 5 mg/ml. Values arethe mean of three assays (standard error , 10% of the mean). Cpm= counts per minute; mitoC = mitomycin C.

The attractiveness of tamoxifen as a radiosensitizer isreduced if its effect is only additive. Tamoxifen has toxic-ity, and may cause deep vein thrombosis with a risk of pul-monary embolism.12 When such toxicity does occur, thecompletion of radiation therapy may be delayed. More-over, there is a risk that low-dose tamoxifen may antago-nize radiation treatment, as shown in two of the three celllines. Therefore, it could be argued that, because a syner-gistic effect is not uniformly seen with the combination ofradiation and tamoxifen, perhaps such therapy should bedelivered sequentially rather than concurrently.

The tamoxifen concentrations used in this study areachievable clinically. In a high-dose tamoxifen protocol,480 mg/day administered for 6 days to 17 patients result-ed in a mean plasma concentration of 2.4 mg/ml.13 Tamox-ifen has been shown to accumulate in brain metastases atmore than a 10-fold higher concentration than in serum.11

Therefore, it may be possible to achieve an accumulationin tumor tissue of at least 25 mg/ml, fivefold higher thanthe highest concentration used in our study.

There were other criticisms of the study performed byZhang, et al.2, made in the commentaries by Olsen andVertosick. A single immortalized cell line of nonhumanorigin (C6) was used, and, hence, it is difficult to drawbroad conclusions. Even if human GBM lines are used, itis important to use more than one because there may beconsiderable variation in the responsiveness to a particu-lar agent or combination of agents. This is evident in thevaried response of the three human GBM cell lines ob-served in this study.

Further basic research into the effect of tamoxifen and,perhaps, its analogs on cell cycling in GBM may providea more rational approach to the timing of chemotherapy inrelationship to radiotherapy. The effect of tamoxifen ther-apy on the complex phospholipase C and PKC secondmessenger system needs further clarification.

Conclusions

The significant finding of this study is that, althoughtamoxifen and radiation can be synergistic at clinicallyrelevant doses, this is not invariably the case. Tamoxifenshould not be considered a radiosensitizer in all formsof GBM.

References

1. Baltuch G, Shenouda G, Langleben A, et al: High dose tamox-ifen in the treatment of recurrent high grade glioma: a report ofclinical stabilization and tumour regression. Can J Neurol Sci20:168–170, 1993

2. Cabot MC, Zhang Z, Cao H, et al: Tamoxifen activates cellularphospholipase C and D and elicits protein kinase C transloca-tion. Int J Cancer 70:567–574, 1997

3. Chou TC, Motzer RJ, Tong Y, et al: Computerized quantitationof synergism and antagonism of taxol, topotecan and cisplatinagainst teratocarcinoma cell growth: a rational approach to clin-ical protocol design. J Natl Cancer Inst 86:1517–1524, 1994

4. Chou TC, Talalay P: Quantitative analysis of dose-effect rela-tionships: the combined effects of multiple drugs or enzyme in-hibitors. Adv Enzyme Regul 22:27–55, 1984

5. Couldwell WT, Antel JP, Yong VW: Protein kinase C activity

correlates with the growth rate of malignant human gliomas:Part II. Effects of glioblastoma mitogens and modulators ofprotein kinase C. Neurosurgery 31:717–724, 1992

6. Couldwell WT, Uhm JH, Antel JP, et al: Enhanced proteinkinase C activity correlates with the growth rate of malignantgliomas in vitro. Neurosurgery 29:880–887, 1991

7. Couldwell WT, Weiss MH, DeGiorgio CM, et al: Clinical andradiographic response in a minority of patients with recurrentmalignant gliomas treated with high-dose tamoxifen. Neuro-surgery 32:485–490, 1993

8. Horgan K, Cooke E, Hallett MB, et al: Inhibition of protein ki-nase C mediated signal transduction by tamoxifen. Importancefor antitumor activity. Biochem Pharmacol 35:4463–4465,1986

9. Leonard CE, Chan DC, Chou TC, et al: Paclitaxel enhanced invitro radiosensitizer of squamous carcinoma cell lines of thehead and neck. Cancer Res 56:5198–5204, 1996

10. Lerner LJ, Jordan VC: Development of antiestrogens and theiruse in breast cancer: eighth Cain memorial award lecture. Can-cer Res 50:4177–4189, 1990

11. Lien EA, Solheim E, Ueland PM: Distribution of tamoxifen andits metabolites in rat and human tissues during steady-statetreatment. Cancer Res 51:4837–4844, 1991

12. McClay EF, Mastrangelo MJ, Bellet RE, et al: Combinationchemotherapy and hormonal therapy in the treatment of malig-nant melanoma. Cancer Treat Rep 71:465–469, 1987

13. Millward MJ, Lien EA, Robinson A, et al: High-dose (480 mg/day) tamoxifen with etoposide: a study of a potential multi-drugresistance modulator. Oncology 51:79–83, 1994

14. O’Brian CA, Liskamp RM, Soloman DH, et al: Inhibition ofprotein kinase C by tamoxifen. Cancer Res 45:2462–2465,1985

15. Pollack IF, Randall MS, Kristofik MP, et al: Effect of tamoxifenon DNA synthesis and proliferation of human malignant gliomalines in vitro. Cancer Res 50:7134–7138, 1990

16. Rowlands MG, Parr IB, McCague R, et al: Variation of the inhi-bition of calmodulin-dependent cyclic AMP phosphodiesteraseamongst analogues of tamoxifen; correlations with cytotoxici-ty. Biochem Pharmacol 40:283–289, 1990

17. Su HD, Mazzei GJ, Vogler WR, et al: Effects of tamoxifen, anonsteroidal antiestrogen, on phospholipid/calcium-dependentprotein kinase and phosphorylation of its endogenous substrateproteins from rat brain and ovary. Biochem Pharmacol 34:3649–3653, 1985

18. Sutherland RL, Murphy LC, Foo MS, et al: High-affinity anti-oestrogen binding site distinct from the oestrogen receptor. Na-ture 288:273–275, 1980

19. Terasima T, Tolmach LJ: X-ray sensitivity and DNA synthe-sis in synchronous populations of He La cells. Science 140:490–492, 1963

20. Vertosick FT, Selker RG, Pollack IF, et al: The treatment of in-tracranial malignant gliomas using orally administered tamoxi-fen therapy: preliminary results in a series of “failed” patients.Neurosurgery 30:897–903, 1992

21. Zhang W, Yamada H, Sakai N, et al: Enhancement of radiosen-sitivity by tamoxifen in C6 glioma cells. Neurosurgery 31:725–730, 1992

Manuscript received April 22, 1998.Accepted in final form September 28, 1998.This work was supported by Delta Delta Delta Sorority, Denver,

Colorado.Address reprint requests to: Nicholas K. Foreman, M.B.Ch.B.,

Department of Pediatric Oncology, The Children’s Hospital, 1056East 19th Avenue, Denver, Colorado 80218. email: Foreman.Nicholas@TCHDen.org.

A. M. Donson, M. D. Weil, and N. K. Foreman

536 J. Neurosurg. / Volume 90 / March, 1999