Journal of Neuro-Oncology46: 157–171, 2000.© 2000Kluwer Academic Publishers. Printed in the Netherlands.

Clinical Study

Preliminary individual adjuvant therapy for gliomas based on the results ofmolecular biological analyses for drug-resistance genes

Satoshi Tanaka1, Hideki Kamitani2, Md Ruhul Amin2, Takashi Watanabe2, Hidehiro Oka1, Kiyotaka Fujii1,Tadashi Nagashima3 and Tomokatsu Hori4

1Department of Neurosurgery, Kitasato University School of Medicine, Sagamihara, Japan;2Division ofNeurosurgery, Institute of Neurological Sciences, Faculty of Medicine, Tottori University School of Medicine,Yonago, Japan;3Department of Neurosurgery, Ichihara Hospital, Teikyo University School of Medicine, Chiba,Japan;4Department of Neurosurgery, Neurological Institute, Tokyo Women’s Medical University, Tokyo, Japan

Key words:glioma, individual adjuvant therapy (IAT), MGMT, MDR1, MRP, GST-π

Comments

The use of chemotherapeutic agents for brain tumor treatment can clearly be made more effective if the likelihoodof drug resistance is known prior to treatment. These experiments suggest that reverse transcription-polymerasechain reaction can provide the type of information needed to evaluate the likelihood of drug resistance. Clearly, thisis a logical and intriguing concept which merits further evaluation.

Dr. P.L. Kornblith (Pittsburgh, PA, USA)

Unfortunately, only a minority of tumors respond to present day cytotoxic drugs and only a minority of patientsmay live longer as a consequence of adjuvant chemotherapy. Furthermore, chemosensitivity among glial tumors isnot predictable by light microscopic analysis, unless by chance, the tumor displays oligodendroglial morphologyin which case up to 70% are drug-sensitive. Consequently, in virtually every clinical situation neuro-oncologistshave no rational basis for making chemotherapy decisions for individual patients. Enter molecular genetics. . .

Perhaps chemosensitive gliomas have a molecular signature that clinicians can use to guide chemotherapy andother treatment decisions. The recent observation that allelic loss of chromosome 1p is statistically associated withchemosensitivity in anaplastic oligodendrogliomas is a first step toward this goal.

Tanaka et al. have enquired whether the expression of drug-resistant genes will predict chemosensitivity andsurvival in patients with several different subtypes of anaplastic glioma. They noted, in particular, that tumorsexpressing low levels of O6-methylguanine-DNA methyltransferase mRNA are more likely to be sensitive to ACNUand that patients harboring gliomas with this molecular characteristic were more likely to have longer survival times.Interestingly, two recent reports, one in Oncogene by Grombacher et al. and another in Carcinogenesis by Nutt et al.,suggest a link between p53 gene status and O6-methylguanine-DNA methyltransferase expression and activity;both groups noticed low expression and activity in cells lacking a functional p53 protein. This raises the intriguingpossibility that p53 status through its effects on DNA repair will be a critical determinant of drug sensitivity inpatients with astrocytic and other subtypes of glioma – a testable hypothesis.

J. Gregory Cairncross (London, Ontario, Canada)

Summary

New adjuvant therapy individualized by the results of reverse transcription-polymerase chain reaction (RT-PCR)for drug-resistance genes has been used to treat malignant gliomas. Protocol studies for malignant gliomas werenot so encouraging in their therapeutic results because of heterogeneity and the various drug-sensitivities of thetumors. Individualization of glioma therapy is recommended. Drug-resistance genes messenger ribonucleic acid

158

(mRNA) expressions were investigated in drug-resistant human glioma cell lines derived from U87MG and 46 frozensamples of retrospectively examined neuroepithelial tumors (12 low grade neuroepithelial tumors, 16 Grade IIIgliomas, 11 glioblastomas, and 7 other malignant neuroepithelial tumors such as medulloblastomas and primi-tive neuroectodermal tumors) by RT-PCR with the specific primers for O6-methylguanine DNA methyltransferase(MGMT), multidrug-resistance gene 1 (MDR1), multidrug-resistance-associated protein (MRP), and glutathione-S-transferase-π (GST-π ). Thirty-seven preliminary individual adjuvant therapies (IAT) based on RT-PCR results,mainly in MGMT expression, were performed on 30 consecutive patients with neuroepithelial tumors. In the retro-spectively examined series, the initial response to 1-(4-amino-2-methyl-5-pyrimidynyl) methyl-3-(2-chloroethyl)-3-nitrosourea hydrochloride (ACNU) was correlated most significantly to the MGMT mRNA expression among 11independent prognostic factors (p = 0.0037) in multivariate logistic regression analysis. In the preliminary IAT,17 of 32 evaluable therapies had a partial or complete response (53.1% response rate). Our IAT based on RT-PCRseemed to be more effective than conventional therapies for malignant gliomas.

Introduction

Malignant gliomas have been treated with surgery,radiation, various chemotherapeutic agents and bio-logical response modifiers such as interferon (IFN)[1–4]. Although many clinical trials have been per-formed using adjuvant therapy for malignant gliomas,there are no results from a considerable number ofpatients that are better than those of the clinical studyof IFN-β in combination with 1-(4-amino-2-methyl-5-pyrimidynyl)methyl-3-(2-chloroethyl)-3-nitrosoureahydrochloride (ACNU) and radiation; IAR therapy [5].The authors reported the results of IAR with hema-tological growth factors that could reduce myelosup-pression and increase the dose of ACNU [6,7]. And wehave also used IFN-β for the maintenance therapy ofmalignant gliomas [8]. These results were encouragingbut there were not so many patients who entered thetrials during these studies. Almost all excellent resultsfrom the therapy, including ours, were not for consec-utive patients and simultaneously showed the limits ofthe protocol study for glioma therapy [1–8].

Pt-compounds such ascis-diamminedichloroplat-inum (cis-platinum, CDDP) andcis-diammine(1,1-cyclobutanedicarboxylato)platinum (II) (carboplatin,CBDCA) are used for primary and recurrent gliomasin some institutes instead of ACNU because of thesevere myelosuppression caused by ACNU and theclinically obvious resistance to it [2–4]. The results ofthis therapy were not so encouraging for gliomas suchas pediatric or congenital brain tumors like germino-mas, Some evidence of the tumors which responded tothese Pt-compounds suggest that it is better to use theminstead of ACNU. Pt-based combination therapy withetoposide or IFN-β has also been tried [2–4].

The difficulty in glioma therapy seems to be due tothe invasive growth that interferes with the true total

removal of the tumors and the heterogeneity of themalignant gliomas [9]. Histological, as well as bio-logical multimodality of gliomas, has been reported[10,11]. Individualization of glioma therapy based on adetailed analysis of the tumor and the patient is neededto overcome heterogeneous tumors. The response of theanti-tumor drug is dependent on the characteristics ofthe tumor such as its histological and biological malig-nancy, the volume of the postoperative residual tumorand the patient’s age, Karnofsky’s performance scale,delivery and metabolization of the drug, and immuno-logical state, etc. [2,5,9–11]. The individualization ofthe therapy means deciding the dose of the drug andthe intensity of the therapy according to these charac-teristics have already been done.

We have paid particular attention to the resistanceof each tumor to anti-tumor drugs. Drug-resistancegenes were one of the most important elements ofthe tumors themselves in determining drug-resistance.The multidrug-resistance gene 1 (MDR1) has beenthought to be associated with resistance to vincristine(VCR), etoposide, and other drugs [12–19]. Recently,multidrug-resistance-associated protein (MRP) hasalso been thought to be important in the resistance tothese drugs [20–23]. Actually, O6-methylguanine DNAmethyl-transferase (MGMT) is one of the most impor-tant genes for ACNU resistance [24–27]. We thinkthat the MGMT gene expression is the most importantdrug-resistance gene in the treatment of gliomas, sinceACNU is the first choice for gliomas of all chemothera-peutic agents in Japan [5,28]. As previously described,some drug-resistance genes, such as glutathione-S-transferase (GST), glutathion, and methalothionein, arethought to be related to the sensitivity of Pt-compounds.Of these molecules, GST-π was reported to havethe closest relationship to tumor resistance to Pt-compounds [14,29–3l].

159

We have applied a recent molecular biological tech-nique to glioma therapy. We investigated the messen-ger ribonucleic acid (mRNA) expression of MGMT,MDR1, MRP and GST-π on frozen human glioma tis-sues by reverse transcription-polymerase chain reac-tion (RT-PCR), and preliminary individual adjuvanttherapy (IAT) based on RT-PCR results has begun forpatients with malignant brain tumors.

In vitro and retrospective clinical study

Materials and methods

Cell linesU87MG human glioma cells which had been pro-vided by American Type Culture Collections weremaintained in Dulbecco’s minimum essential medium(DMEM, GIBCO, Grand Island, NY) with 10% fetalcalf serum (FCS, GIBCO) and antibiotics.

Anti-cancer drugsACNU, VCR, CDDP, CBDCA, IFN-β, and methotrex-ate (MTX) were used forin vitro experiments andclinical trials. All these drugs are commercially avail-able in Japan for clinical use. ACNU was pur-chased from Sankyo Co. Ltd., Tokyo, Japan, VCR andCDDP from Nihon-Kayaku Co. Ltd., Tokyo, Japan,CBDCA from Bristol-Mayers-Squibb Japan, Tokyo,Japan, IFN-β from Toray Co. Ltd., Tokyo, Japan, MTXfrom Japan Lederie, Tokyo, Japan.

Drug-resistant cell linesU87MG human glioma cells were incubated withescalating concentrations of ACNU (0.1, 1, 10µg/ml),VCR (0.1, 1, 10 ng/ml), and CDDP (0.1, 0.2, 0.5,1µg/ml) for 30 days with four or five passages to estab-lish the drug-resistance cell lines, U87AR, U87VR,U87CR. These cells were established for use in RT-PCR as positive controls for MGMT, MDR1, MRP, andGST-π .

Colorimetric 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide (MTT) assayU87MG, U87AR, U87VR, U87CR human glioma cellswere trypsinized and seeded into flat-bottomed 96-well microtiter plates in a medium containing 10%FCS (1× 104 cells/well). After 6 h of incubation at37◦C, fresh medium containing various concentrationsof ACNU, VCR, or CDDP was added to each well.The effect of anti-cancer drugs on cell growth wasdetermined on Day 5 of incubation by monitoring the

number of metabolically active cells using the MTTassay described elsewhere [32]. Briefly, 40µl of MTT(KANTO, Tokyo, Japan) solution (2.5 mg/ml) wasadded to each culture well, the plates were incubatedat 37◦C for 4 h, and MTT formazans were inspectedmicroscopically. Then, 100µl of 10% sodium dodecyl-sulfate (SDS)–0.1 N HCl solution was added to eachwell. After solubilization by overnight incubation, theabsorbance at 570 nm of each well was measuredusing a microplate reader, MPR-4A (Toso, Tokyo,Japan).

Reverse transcription-polymerase chain reaction(RT-PCR)In 46 frozen samples of neuroepithelial tumors (12 lowgrade neuroepithelial tumors, 16 grade III gliomas,11 glioblastomas, and 7 other malignant neuroepithe-lial tumors such as medulloblastomas and primitiveneuroectodermal tumors) together with U87MG,U87AR, U87VR, U87CR cultured cells, MGMT,MDR1, MRP and GST-π mRNA were detected usingRT-PCR. RT-PCR was performed according to themethod previously described [33,34]. Briefly, totalRNA was extracted from frozen tissue specimensor cultured cells weighing about 1 g, homogenizedusing a glass Teflon homogenizer via the guanidiniumthiocyanate–phenol–chloroform single-step extractionmethod with Isogen (Nippon Gene, Toyama, Japan)[35]. After ethanol precipitation, about 100µg oftotal RNA was extracted. Forty microliters of comple-mentary deoxyribonucleic acid (cDNA) solution wassynthesized from 2µg of total RNA with 40 U of areverse-transcriptase, RAV-2, 54 U of a ribonucleaseinhibitor, 16µl each of a 2.5 mM dNTP mixture, 2.0µlof 50 pM random primer, 8.0µl of 5× RAV-2 buffer,and diethyl pyrocarbonate-treated distilled water at(DW) 42◦C for 90 min. All enzymes and buffers forRT-PCR were purchased from Takara, Otsu, Japan.

DNA amplification was carried out with 5.0µlof cDNA synthesized above in a total of 50µlusing 2.5 U of Taq DNA polymerase, TaKaRaExTaq, 5.0µl of 10× Ex Taq buffer, 37.5µl of DW,and 0.2µM of each pair of oligonucleotide pri-mers (MGMT 5′-CTCCTGGGCAAGGGGACGTCT,and 5′-GATGAGGATGGGGACAGGATT; MDR1,5′-TTTTCATGCTATAATGCGAC, and TCCAAGA-ACAGGACTGATGG; MRP, 5′-AGGAGAGATCAT-CATCGATGG, and 5′-GCCTTCTGCACATTCATGG;GST-π , 5′-CATGCTGCTGGCAGATCAG, and5′-CATTCATCATGTCCACCAGG; human β2-microglobulin [β2-MG] was used as an internal

160

control, 5′-TTCTGGCCTGGAGGGCATCC and 5′-ATCTTCAAACCTCCATGATG). The program wascarried out as follows: initial melting for 1 min at 96◦C,35 cycles of melting for 30 sec at 96◦C, annealing for30 sec at 55◦C, elongation for 1.5 min at 72◦C, finalelongation for 7 min at 72◦C, and cooling to 4◦C witha thermal cycler, a Mini Cycler Model PTC-150-16(Funakoshi, Tokyo, Japan). The 354 base pairs (bp)MGMT, 455 bp IFNR, 226 bp MDR1, and 340 bpβ2-MG PCR products (each 10µl) were analyzedby electrophoresis using 3% agarose gel containing0.005% ethidium bromide, and photographed underultraviolet light.

Statistical analysisAmong 46 frozen tumors, 23 tumors were resectedfrom the patients who had been treated with ACNUalone or in combination with other drugs. The resultsof the therapies were judged as 50% or more reductionof the tumor volumes in contrast to enhanced images ofCT scan or magnetic resonance imaging. Some inde-pendent prognostic factors for the results of initialtherapies with ACNU, such as age, sex, Karnofsky’sperformance scales (KPS), histological grades, primaryor recurrence, tumor resection rates, combination drugsand therapies, radiation or not, and mRNA expressions,were statistically analyzed [36]. AllP values reportedwere two-tailed. Difference in the effect of the ther-apy with ACNU to the histological grading was cal-culated by chi-square test. Fisher’s exact probabilitytest was used for binary variables predicting the resultsof the therapies with ACNU including the primary orrecurrence, sex, combination drugs, dose of ACNU≥80 mg/m2 or not, radiation or not, and mRNA expres-sions. Mann–Whitney’sU-test was used for contin-uous variables predicting the results including ages,KPS, and surgical resection rate. Multivariate regres-sion analysis was used to identify variables describedabove predicting the effects of the therapies.

ImmunohistochemistryTwenty formalin-fixed paraffin sections 4µ thicknessfrom the neuroectodermal tumors were prepared andeach of the 15 sections were stained immunohisto-chemically by avidin–biotin complex (ABC) methodwith anti-MGMT (KAMIYA Biol. Co., Seattle, WA),and anti-P-glycoprotein which were the protein prod-ucts of the MDR1 gene (YLEM, Roma, Italy) [17],and anti-GST-π (DBS, Pleasanton, CA) with HistofineSAB-PO (M) kit (Nichirei, Tokyo, Japan).

Results

MTT AssayThough all drug-resistance cells including U87AR,U87VR, and U87CR had slightly slower growth thanU87MG, several times the resistance to ACNU, VCRor CDDP in the applicable cells was observed in theMTT assay described in Figure 1.

RT-PCRThe drug-resistant cell lines derived from U87MGshow each drug-resistance gene as in Figure 2. Theresults of RT-PCR from all 46 samples were mea-sured retrospectively, MGMT mRNA expressions weredetected in 20 (43.5%) samples, MDR1 in 23(50.0%),MRP in 82.9% and GST-π in 80.0% (Figure 3 andTable 1). No significant difference in the positive ratesbetween the histological grades and between the pri-mary and recurrent cases was observed in any of theprimers. Also no significant relationship was evidentbetween both MGMT and GST-π , and MDR1 andMRP, respectively.

The statistical evaluations of 23 patients who hadused ACNU and had evaluable therapeutic results wereperformed using 11 previously described prognosticvariables. In univariate analysis, significant prognos-tic variables for the effects of the therapies usingACNU were surgical resection rate (Mann–Whitney,P = 0.0117), dose of ACNU≥ 80 mg/m2 (Fisher,P =0.0288), and MGMT mRNA expression (Fisher,P =0.0016). Multivariate logistic regression analysis deter-mined that only MGMT mRNA expression indepen-dently predicted failure to the therapy with ACNU(P = 0.0037) (Table 2). The relationship betweenMDR1 or MRP expression and VCR- or etoposide-response and GST-π and CDDP- or CBDCA-responsewere not significant in our experiments because of thesmall number of cases in which these drugs were used.

ImmunohistochemistryOf each of the 15 sections tested by immunohistochem-istry, 7 anti-MGMT, 12 anti-P-glycoprotein, and 13GST-π showed positive staining to some degree. How-ever the MGMT antigen was located in the nucleus ofthe tumor (Figure 4, Upper left), and the P-glycoproteinabundantly in the tumor tissue, including the tumorcytoplasm, infiltrating cells, and vascular endothelialcells (Figure 4, Middle). GST-π was also detectedmainly in the cytoplasm (Figure 4, Lower). As shownin Table 3, there were 12 mismatched cases in the

161

Figure 1. Growth of U87MG in comparison with U87AR (Upper left), U87VR (Upper right), and U87CR (Lower) with various concen-trations of ACNU, VCR and CDDP. U87AR, U87VR, U87CR were established with escalating doses of ACNU, VCR, or CDDP. About10-fold resistances of U87AR, U87VR, or U87CR (•) compared with U87MG (�) to ACNU, VCR, or CDDP, respectively, were provedby MTT assay after 5 days’ culture. Some error bars are not clear from the symbols.

expression between RT-PCR and immunohistochem-istry, 3 for MGMT, 5 for MDR1-PGP, and 4 for GST-π .For MGMT mRNA and protein expressions, 4 resultsfrom immunohistochemistry were not compatible tothe drug sensitivities, and only one from RT-PCR.

Prospective clinical study

Clinical materials and methods

Since May 1997, 37 preliminary individual adjuvanttherapies (IAT) mainly based on the results of RT-PCRfor MGMT have been performed on 30 consecutivepatients with neuroepithelial tumors. The subjects hadbeen operated on and treated after the operation in

the Division of Neurosurgery, Tottori University. Thesecases consisted of 12 grade III gliomas, 11 glioblastomamultiformes, 4 astrocytomas, 2 medulloblastomas anda pineoblastoma. Primary therapies numbered 21 and16 residual therapies were done for recurrence. Agedistribution was from 14 to 84; the mean age was 48.5years-old. The mean Karnofsky’s performance status(KPS) was 57.4%.

All patients underwent preliminary IAT after a possi-ble resection (5–l00% mean 76.7% in resection rate) ofthe tumor, except for 4 patients diagnosed with stereo-tactic biopsy. Immediately after the operation, RT-PCRof at least two specimens from different tumor sites wasperformed with the previously described primers.

At first in the selection of drugs, MGMT mRNA neg-ative cases have been treated with ACNU as the first

162

Figure 2. Three percent agarose-gel electrophoresis of the RT-PCR product amplified from total RNA of U87MG (1), U87AR(2), U87VR (3), and U87CR (4) with the specific primerβ 2MG(A) as an internal control, MGMT (B), MDR1 (C), MRP (D)and GST-π (E), stained with 0.005% ethidium bromide. M, PCRMolecular Weight Markers (A, B, USB, Cleveland, OH), orφ X174DNA/HaeIII digest (C, D, E, TOYOBO, Tokyo, Japan).

Figure 3. Three percent agarose-gel electrophoresis of the RT-PCR product amplified from the total RNA of anaplastic astrocy-toma tissues with the specific primerβ2-MG (A) as an internalcontrol, MGMT (B), MDR1 (C), MRP (D) and GST-π (E), stainedwith 0.005% ethidium bromide. M, PCR Molecular Weight Mark-ers (A, B, USB, Cleveland, OH), orφ X174DNA/HaeIII digest(C, D, E, TOYOBO, Tokyo, Japan).

Table 1. Drug-resistance gene mRNA expression by RT-PCR ingliomas

N MGMT MDR1 MRP GST

Astrocytoma 12 33.3% 41.7% 80.0% 60.0%Grade III gliomas 16 50.0% 56.3% 90.9% 100.0%Glioblastomas 11 45.5% 54.5% 87.5% 87.5%Other neuroepithelial 7 42.9% 42.9% 66.7% 66.7%tumors∗

Total 46 43.5% 50.0% 82.9% 80.0%

∗Other neuropithelial tumors= medulloblastomas and primitiveneuroectodermal tumors.

Table 2. P -value in the effect of ACNU on 23 patients withgliomas

Factor Univariate Multivariateanalysis analysis

Primary/recurrence > 0.9999∗ 0.7580Age 0.4565∗∗∗ 0.9258Sex 0.4136∗ 0.3559Histological grade 0.8580∗∗ 0.8681Karnofsky’s performance scale 0.2103∗∗∗ 0.8030Surgical resection rate 0.0117∗∗∗ 0.0262Dose of ACNU (80 mg/m2 ≤ 0.0288∗ 0.4962or not)

Combination with IFN-β 0.1688∗ 0.5319Combination with platinum > 0.9999∗ 0.9553compounds

Radiation therapy 0.1045∗ 0.3297MGMT mRNA expression 0.0016∗ 0.0037

∗by Fisher’s exact probability test;∗∗by Chi-square test;∗∗∗byMann–Whitney’sU-test.

choice drug. MGMT-positive cases were treated withCBDCA or CDDP instead of ACNU. Usually VCR andetoposide were used in combination with ACNU andplatinum compounds, respectively. Combination ther-apy and the intensity of the therapy including the doseof the drug, with or without radiation, and number ofdrugs used was decided by other factors such as thepatient’s age, KPS, history of therapy, as well as thetumor characteristics including residual tumor volume,histological grading and biological malignancy repre-sented by the growth fraction and invasiveness of thetumor detected by magnetic resonance imaging (MRI).For example, in patients with age over 70 years, KPSunder 40%, WHO grade I or II tumor, and no evalu-able residual tumor after the total resection, they havereceived only one or two drugs as IAT (Table 4). IFN-β

can be used in all cases because it has less adverse

163

Figure 4. Immunohistochemistry on the paraffin sections of glioma tissues stained with anti-MGMT, anti-P-glycoprotein, which is theprotein produced from MDR1 gene, and anti-GST-π monoclonal antibodies. MGMT was expressed abundantly in the glioma cell nucleiof patient No. 10 without mRNA expression from RT-PCR in Table 3 (Upper left) and patient No. 8 with mRNA expression (Upper right).P-glycoprotein expression is described in No. 4 in vascular endothelial cells (Middle left) and in No. 8 in the cytoplasm (Middle right).GST-π was expressed in No. 20 (Lower left) and No. 8 (Lower right). Counterstaining was done with hematoxylin,×400.

164

Table 3. Immunohostochemistry and mRNA expression by RT-PCR of drug-resistance genes∗

No. Age Sex Diagnosis MGMT MDR1 (P-glycoprotein) GST-π

RT-PCR Staining Effect RT-PCR Staining Effect RT-PCR Staining Effect

1 22 M E − − − − + ND2 11 M A + + − + − ND3 14 F A − − R + + + −4 44 M AA − − + + + ND5 51 M AA − − S − + + +6 34 F AA + + R − − + ND7 62 F AA − − S + + − +8 12 F AE + + − + + +9 70 M AA + + + + + ++

10 52 M GM − ++ S + ++ − ++11 72 M GM + − + + + ±12 72 M GM + + + + + ND13 61 F GM − − − + + ++14 12 M MB − − − − + ±15 22 M PB − + S − ++ + ±16 24 F GM + ND − ND − ±17 29 F A + ND + ND + +18 14 M A − ND − ND + ++19 59 M GM − ND − ND + −20 48 M AA − ND + ND + +∗MGMT, O6-methylguanine DNA methyltransferase; MDR1, multidrug-resistance gene 1; MRP, multidrug-resistance-associated pro-tein; GST-π , glutathione-S-transferase-π ; Effect, the response of the applicable drug:+, positive;++, strongly positive;−, nega-tive staining;±, weakly positive; E, ependymoma; A, astrocytoma; AA, anaplastic astrocytoma; AE, anaplastic ependymoma; GM,glioblastoma multiforme; MB, medulloblastoma; PB, pineoblastoma; S, sensitive; R, resistant; ND, not done.

effects than chemotherapeutic agents and an indirectanti-tumor effect was expected especially during main-tenance therapy [37]. The maintenance therapy wasactively performed with IFN-β, ACNU, or CBDCAaccording to the results of the initial preliminary IAT.The effect of preliminary IAT was evaluated at leasttwo months after the beginning of therapy.

Results

The 30 consecutive patients in Table 4, with neu-roepithelial tumors were treated with our preliminaryIAT mainly decided by RT-PCR for MGMT mRNA.Two therapies have been performed with IFN-β alonebecause of the existence of the previously describedfactors in each patient. Seven patients were treatedtwice with IAT. One patient (No. 8 in Table 4) was notevaluable because of abscess formation after the firstoperation, although no residual tumor was detected inthe gross totally resected mass lesion during the sec-ond operation. Other non-evaluable cases arose after100% removal of the tumor in the first operation. Sev-enteen of 32 (56.0%) evaluable therapies, disclosed

a complete response (CR) with 100% tumor regres-sion, or partial response (PR), with over 50% tumorregression evaluated at least 2 months later from thebeginning of IAT. The response rate was 41.7% forglioblastoma multiformes, 59.8% for grade III gliomas,100% for other malignant neuroepithelia tumors and33.3% for astrocytoma. With 27 in 37 (79.0%) ther-apies, patients had gained remissions and followingmaintenance therapies. Since our series included 9recurrent cases, the survival rate could be calculatedby Kaplan–Meier method in spite of the short observa-tion periods. Although comparison of it to the resultsof other studies could not be done because our IAT wasstill preliminary and not controlled, the three years sur-vival rate was very high (Table 5).

The statistical evaluations of 32 therapies whichhad evaluable therapeutic results were performed using12 prognostic variables including primary or recur-rence, age, sex, histological grade, KPS, surgical resec-tion rate, drugs, radiation therapy, performance of theplanned therapy, and MGMT mRNA expression. Forthe effects of the therapy, the mean of KPS of goodresponse group was significantly higher than that ofpoor response group (Fisher,P = 0.0491). Also a

Tab

le4.

Sum

mar

yof

the

patie

nts

trea

ted

bypr

elim

inar

yin

divi

dual

adju

vant

ther

apy

∗

No.

P/R

Age

Sex

His

tolo

gyO

pe(%

)K

PS

mR

NA

expr

essi

onby

RT-

PC

RT

hera

pyP

erfo

rman

ceof

the

ther

apy

Effe

ctR

emis

sion

TT

P(M

)S

P(M

)M

GM

TM

DR

1M

RP

GS

T

1P

14F

A/A

A50

/100

40/4

0−/−

+/−

+/+

+/−

IAγ

/CE

R−/+

NC

/PR+/+

12/9<

22<

2R

14M

A95

60−

−+

+I

+P

R+

19<

126<

3P

29F

A10

050

++

−+

CI

+N

E−

14<

14<

4P

42F

A0

20+

−+

−C

I−

NC

+6<

13<

5P

70M

AA

100

50+

++

+C

I+

NE

+19

21<

6P

44M

AA

95/7

590

/70−/−

+/−

+/−

+/+

IAR

/IAV

/C+/−

NC

/NC+/+

11/1

227<

7P

51M

AA

95/9

590

/50−/−

−/−

−/+

+/+

AVR

/CI

+/+

PR

/PR+/+

18/6

26<

8P

61M

AA

7540

−+

−+

IAR

−N

E−

23

9R

48M

AA

75/5

070

/20−/−

+/+

+/−

+/+

IA/A

+/+

PR

/NC+/−

5/11

124

10R

44F

AO

100

90−

−−

+IA

VR

+N

E+

16<

99<

11P

67F

AA

030

−−

−+

IAR

+P

R−

23

12P

73F

AA

5060

−+

−+

A+

NC

+6

1013

R49

FA

A75

20+

++

+C

I−

PD

−2

1714

P27

MA

O50

80−

++

+IA

VR

−P

R+

6<

6<

15P

22M

AA

5010

0−

−+

+AV

R+

PR

+2<

2<

16P

70M

AA

7530

+−

+−

PE

−N

C−

2<

2<

17R

72M

GM

75/7

560

/60+

++

+I/C

E+/+

PD

/PR−/+

2/5

2018

R61

FG

M95

80−

−−

+IA

+C

R+

22<

42<

19P

77M

GM

7530

−−

+−

IA−

PD

−2

520

P55

FG

M10

070

++

−+

ICR

+N

E+

1314

21R

44M

GM

7580

−−

−+

AVR

/I/C

+P

D+

918<

22P

59M

GM

7570

−−

−+

IAR

+P

R+

13<

13<

23P

24F

GM

75/5

060

/50+/−

−/−

−/+

−/−

ICR

/IPE

+/+

PR

/PD+/−

8/2

12<

24P

70M

GM

7570

−−

++

IAV

R+

PR

+11<

11<

25P

60M

GM

7530

−+

++

IAR

−P

D−

2<

9<

26P

84M

GM

010

−−

−+

IA−

PD

−1

5<

27P

66M

GM

090

−−

+−

IAV

R+

NC

+2<

2<

28P

22M

PB

5080

−−

++

AVR

+P

R+

20<

20<

29R

23F

M—

40/4

0−

++

−IA

RM

/IAV

R+/+

PR

/PR+/+

7/2

61<

30R

14M

M—

20−

−+

+IA

RM

+P

R+

5<

31<

∗ P/R

,prim

ary

orre

curr

ent;

Ope

,ope

rativ

ere

sect

ion

rate

;KP

S,K

arno

fsky

’spe

rfor

man

cesc

ale;

MG

MT,

O6-m

ethy

lgua

nine

DN

Am

ethy

ltran

sfer

ase;

MD

R1,

mul

tidru

g-re

sist

ance

gene

1;M

RP,

mul

tidru

g-re

sist

ant

asso

ciat

edpr

otei

n;G

ST-

π,

glut

athi

one-

S-t

rans

fera

se;

TT

P,tim

eto

tum

orpr

ogre

ssio

n;S

P,su

rviv

alpe

riod;

AA

,an

apla

stic

astr

ocyt

oma;

GM

,glio

blas

tom

am

ultif

orm

e;A

,ast

rocy

tom

a;P

B,p

ineo

blas

tom

a;A

O,a

napl

astic

olig

oden

drog

liom

a;S

,sur

gery

;I,I

FN

-β

;A,A

CN

U;γ

,γ-k

nife

;R,r

adia

tion;

E,e

topo

side

;V,

vinc

ristin

e;C

,car

bopl

atin

;M,m

etho

trex

ate;

NC

,no

chan

ge;P

R,p

artia

lres

pons

e;P

D,p

rogr

essi

vedi

seas

e;N

E,n

otev

alua

ble;

CR

,com

plet

ere

spon

se.

166

Table 5. Results of preliminary individual adjuvant therapy based on RT-PCR results∗

Histology N Effectiverate (%)

Remissionrate (%)

MTTP(M)

PF rate (%) MSP(M)

Survival rate(%)

1 y 3 y 5 y 1 y 3 y 5 y

All cases 30 53.1 73.0 6 41.1 0 0 14 84.7 67.6 67.6Astrocytomas 4 33.3 100 12 66.7 0 0 14 100 100 100Grade III gliomas 12 53.8 68.8 6 28.6 0 0 10 68.6 57.1 57.1Glioblastomas 11 41.7 61.5 4 42.7 0 0 12 90.0 33.8 0Other neuroepithelial tumors 3 100 100 5 50.0 0 0 31 100 100 100

∗MTTP, median time to tumor progression; PF, progression free; MSP, median survival period; other neuroepithelial tumors=medulloblastomas and a pineoblastoma.

Table 6. P -value in the remission of preliminary IAT on 30patients with gliomas

Factor Univariate Multivariateanalysis analysis

Primary/recurrence > 0.9999∗ 0.9579Age 0.0084∗∗∗ 0.1433Sex 0.7099∗ 0.9552Histological grade 0.2648∗∗ 0.0732Karnofsky’s performance scale 0.0002∗∗∗ 0.1116Surgical resection rate 0.2301∗∗∗ 0.6552Use of ACNU 0.7158∗ 0.0760Use of IFN-β > 0.9999∗ 0.8834Use of platinum compounds > 0.9999∗ 0.0526Radiation therapy 0.1363∗ 0.6506Performance of the planned 0.0011∗ 0.0021therapy

MGMT mRNA expression 0.6788∗ 0.4787

∗by Fisher’s exact probability test;∗∗by Chi-square test;∗∗∗byMann–Whitney’sU -test

significant prognostic variable in univariate analysiswas the practical performance of the planned therapy(Mann–Whitney,P = 0.0346). Multivariate logisticregression analysis determined that no variable inde-pendently predicted the effect of the therapy. On theother hand, for the remission on the patients, signif-icant prognostic variables in univariate analysis wereage (Mann–Whitney,P = 0.0084), practical perfor-mance of the planned therapy (Fisher,P = 0.0011),and KPS (Mann–Whitney,P = 0.0002). Multivariatelogistic regression analysis determined that only per-formance of the planned therapy independently pre-dicted the remission on the patients (P = 0.0068)(Table 6).

Illustrative cases

Case 5. A 61-year old woman who had a recur-rent tumor 20 months after surgery, radiation, andACNU, underwent 95% removal of the right temporal

glioblastoma multiforme Although GST-π mRNA wasdetected, MGMT, MDR1, and MRP were not expressedin a RT-PCR performed immediate after the operation.The patient was treated with 125 mg of intravenousACNU once and 6×106 IU of IFN-β three times a weekfor 4 weeks. She made a good response with tumorreduction 2 months after beginning IAT and now usesmaintenance therapy with IFN-β once every 2 weeksto the present (Figure 5).

Case 18. A 24-year-old woman with primary rightfrontal glioblastoma multiforme had 75% of the tumorresected. Immediate postoperative RT-PCR showedpositive MGMT, and negative MDR1, MRP, andGST-π . We started the preliminary IAT with 500 mgof CBDCA twice at 4 weeks intervals and 6× 106 IUof IFN-β three times a week for 6 weeks in combina-tion with generous radiation therapy. She made a par-tial response, with the tumor reducing 2 months afterthe beginning of IAT and now uses maintenance ther-apy with IFN-β once every 2 weeks to the present(Figure 6).

Discussion

Although our clinical study was preliminary becauseof the basis of the selection of the drugs were obscureexcept for MGMT mRNA expression, our responserate of over 50% in 37 consecutive therapies is veryencouraging in comparison with other protocol stud-ies [2–5]. Investigation of the sensitivity of the tumorprior to the use of the drug seems to be ideal for tumortherapy. Some approaches using primary cell cultureshave been used as a sensitivity test for tumor therapy[38,39]. These methods often failed at the first pri-mary cell culture, in spite of the high cost and timeneeded. In addition, thein vitro and in vivo sensitiv-ities are often mismatched with clinical effectiveness

167

Figure 5. A 61-year-old woman (No. 5 in Table 4) who had a recurrent tumor (Upper left) 20 months after surgery, radiation, and ACNU.She underwent 95% removal of the right temporal glioblastoma multiforme (Upper right). Although GST-π mRNA was detected. MGMT,MDR1, and MRP were not expressed in RT-PCR performed immediately after the operation (Lower left). The patient was treated with125 mg of intravenous ACNU once and 6× 106 IU of IFN-β three times a week for 4 weeks. She made a good response with tumorreduction 2 months after the beginning of IAT (Lower right) and current maintenance therapy with IFN-β once every 2 weeks up to thepresent. M,φ X174DNA/HaeIII digest (C, D, E, TOYOBO, Tokyo, Japan). A,β2-MG; B, MGMT; C, MDR1; D, MRP; E, GST-π .

because of problems with the drug-delivery system andother reasons. The results of these tests are not alwaysguaranteed.

We think that our method of RT-PCR is very easyand seems to be universally applicable. Drug-resistancegene expression also can be detected by immunohisto-chemistry on protein level [15,19,21,30]. There werea few cases in which the results of immunohistochem-istry were mismatched with those of RT-PCR in ourseries. The reason why mRNA expression and proteinexpression did not agree in some samples seems to becomplicated. The difference in the selected section of

individual tumors and the conditions of each reactionare thought to be partly the reason. The transcription ofthe gene does not always agree with the translation [l5].At least in MGMT expression, our results showed thatRT-PCR was more sensitive and specific than immuno-histochemistry. Immunohistochemistry should not beused to decide which drugs to prescribe accordingto our results. Our results from immunohistochem-istry were less specific than those from RT-PCR. How-ever, all genes play their roles after translation toprotein. Essentially, the functional activity is moreimportant than mRNA expression. If the clinically

168

Figure 6. A 24-year-old woman with primary right frontal glioblastoma multiforme (Upper left) had 75% of the tumor resected (Upperright). Immediately after surgery RT-PCR showed positive MGMT, and negative MDR1, MRP, and GST-π (Lower left). We startedpreliminary IAT with 500 mg of CBDCA twice at 4 weeks intervals and 6×106 IU of IFN-β three times a week for 6 weeks in combinationwith generous radiation therapy. She made a partial response to tumor reduction 2 months after beginning IAT (Lower right). Currentlymaintenance therapy with IFN-β once every 2 weeks is being continued. M, PCR Molecular Weight marker; A,β2-MG; B, MGMT; C,MDR1; D, MRP; E, GST-π .

acceptable method of drug-resistance gene activitymeasurement, especially of MGMT, is invented, itmay be more useful for IAT than RT-PCR. How-ever the expression in immunohistochemistry does notalways reflect the functional activity. Drug-resistantmRNA expression indicates the potential for drug-resistance if protein expression is absent. RT-PCR ismore rapidly performed after surgical resection of thetumor than immunohistochemistry. Our methods ofRT-PCR take about 6 h from the RNA extraction to thefinal electrophoresis. Immunohistochemistry is usually

performed after formalin fixation, embedding in paraf-fin, and making other preparations. It takes at least3 days following the operation. RT-PCR can be usedfor adjuvant therapy on the day of the operation or 1day after operation. At present, RT-PCR seems to bethe most rapid and reliable method for detecting drug-resistance genes.

MGMT is an enzyme for DNA repair after alky-lation by nitrosoureas [26,40]. It is closely related tothe resistanceto ACNU [24,25]. Some authors reportedthat the expression of MGMT is well correlated to

169

the clinical or experimental resistance of brain tumorsto ACNU [41,42]. We also found equivalent resultsfor cultured cells and clinical samples. MGMT isdifferent from other enzymes because it is a consump-tive. Some authors have tried to repair alkylated DNAby MGMT using this specificity. Excessive alkylatingagents, such as high dose ACNU or 5-(3-3′-dimethyl-1-triazeno)imidazole-4-carboxamate, that can wasteMGMT, have been used for this purpose [43,44].Both therapy results were disappointing because ofsevere side-effects and a low response rate. Naganeet al. documented that ACNU resistance resulting fromthe expression of MGMT in rat glioma cells wassignificantly reduced by the expression of antisenseMGMT RNA [45]. This approach seems logical, but itstill has to be shown inin vitro experiments. For now,it is suggested that ACNU should not be used againsttumors that express MGMT mRNA.

The significance of MDR1, MRP, and GST-πexpression for brain tumors is also controversial[12,14,15,20,30,42]. MDR1 is also included in normalendothelial cells, as shown in our immunohistochem-istry results and it consists of part of the blood–brainbarrier [19]. This fact may explain the drug-resistanceof brain tumor cells, which invade the adjunct brainwith normal vasculature. No evidence on the correla-tion of MDR1, MRP, and GST-π expression or clini-cal response of applicable drugs has been reported onbrain tumors, in contrast to other tumors [13,21,23].Even with other tumors, a correlation between MDR1and MRP expression and clinical response of the appli-cable drug is controversial. Our results from RT-PCRcannot clarify this because of the number of gliomacases, which have been treated with CDDP, VCR oretoposide is too small.

The functional mechanism by which IFN exerts effi-cacy on gliomas remains unknown [18,46,47]. IFNstimulates anti-tumoral effector cells such as naturalkiller cells [37]. The indirect pathway of IFN-β, whichis affected by the host anti-tumor immune mechanismsshould be considered when deciding the indication ofIFN-β, especially the maintenance therapy.

The mismatch of the results of RT-PCR and clinicalresponse seems to occur for these three reasons. (1) Theconditions of RT-PCR for tumor sampling or tech-nical problems, especially in pseudo-negative results.(2) The problem of drug delivery. An effective drugmay not reach the tumor cells due to a blockage, suchas blood–brain barrier or granular septum, coveringthe recurrent tumor tissues [28]. (3) Drug-resistance

by other mechanisms or other drug-resistance genesthan those of the detected genes. For an accountof (1), some previous reports gave a quantitative orsemi-quantitative analysis of the relationships betweendrug-resistance gene expression and clinical responsewith semi-quantitative PCR or a blotting procedure[15,16,22,31,41,42,45]. We decided on the drug onlyby detection in agarose gel electrophoresis of RT-PCRproducts. Indeed, a blotting procedure such as Northernblotting of mRNA or Southern blotting of PCR prod-ucts may give different results. These blotting proce-dures are not applicable when deciding which drugs touse immediately after surgery because of their compli-cated nature and time delay and high cost. Although itis thought that Southern blotting of PCR products is fre-quently more successful than detection by RT-PCR, ourdetection rate for RT-PCR seemed to be very reason-able and suitable for IAT, at least in MGMT expression[18]. It appears that our RT-PCR has a semi-quantitativenature to some extent due to the choice of reactionconditions.

Multivariate logistic regression analysis has theadvantage for the evaluation of non-controlled retro-spective analysis [48,49]. MGMT mRNA expressionwas proved to be significantly correlated to the resultsof the adjuvant therapy using ACNU by this method inour retrospectively examined series. As a result of themultivariate analysis in the patients treated by prelimi-nary IAT, age and KPS must be considered for the indi-vidualization of the glioma therapy in addition to thedrug resistance and sensitivity detection. Simultane-ously, the practical performance of the planned therapyovercoming the side effects of the drugs and therapiesby the recent medical techniques such as hematopoi-etic growth factors seems to be very important [6,7].

In conclusion, our preliminary IAT by RT-PCR sug-gested a simple and beneficial glioma therapy. In partic-ular, in the choice of the main drug for adjuvant therapy,ACNU should be selected for tumors without MGMTmRNA expression. More detailed methods of individ-ualization and a randomized study should be done topromote the universal acceptance of this new methodfor glioma therapy.

References

1. Arcicasa M, Roncadin M, Bortolus R, Bassignano G, Boz G,Franchin G, De Paoli A, Trovo MG: Results of three consec-utive treatments for malignant gliomas. Ten-year experienceat a single institution. Am J Clin Oncol 17: 437–443, 1994

170

2. Boiardi A, Silvani A, Milanesi I, Botturi M, Broggi G: Pri-mary glial tumor patients treated by combining cisplatin andetoposide. J Neuro-Oncol 11: 165–170, 1991

3. Boiardi A, Silvani A, Milanesi I, Botturi M, Broggi G: Car-boplatin combined with carmustine and etoposide in thetreatment of glioblastoma. Ital J Neurol Sci 13: 717–722,1992

4. Silvani A, Milanesi I, Munari L, Broggi G, Botturi M,Boiardi A: Intratumoral beta interferon and systemicchemotherapy. Preliminary data in GBM patients. J Neuro-surg Sci 34: 257–259, 1990

5. Yoshida J, Kajita Y, Wakabayashi T, Sugita K: Long-termfollow up results of 175 patients with malignant glioma:importance of radical tumor resection and postoperativeadjuvant therapy with interferon, ACNU and radiation. ActaNeurochir (Wien) 127: 55–59, 1994

6. Tanaka S, Tabuchi S, Watanabe K, Akatsuka K, Numata H,Hokama Y, Hori T: Preventive effects of interleukin 1βfor ACNU-induced myelosuppresion in malignant braintumors: the experimental and preliminary clinical studies.J Neuro-Oncology 14: 159–168, 1992

7. Tanaka S, Nagashima T, Matsuno A, Tanaka H, Kamitani H,Hori T: The efficacy of granulocyte-colony stimulating fac-tor on the adjuvant therapy for malignant gliomas. Jpn JNeurosurgery 6: 827–832, 1997 (Jpn)

8. Tanaka S, Taniura S, Matsumoto S, Kamitani H, Oka H,Fujii K, Nagashima T, Watanabe T, Hori T: Long-termhuman interferon-β maintenance therapy for malignantgliomas. Int J Immunotherapy 17: 39–49, 1999

9. Weingart J, Brem H: Biology and therapy of glial tumors.Curr Opin Neural Neurosurg 5: 808–812, 1992

10. James CD, Olson JJ: Molecular genetics and molecular biol-ogy advances in brain tumors. Curr Opin Oncol 8: 188–195,1996

11. Janus TJ, Kyritsis AP, Forman AD, Levin VA: Biology andtreatment of gliomas. Ann Oncol 3: 423–433, 1992

12. Becker I, Becker KF, Meyermann R, Hollt V: The multidrugresistance gene MDR1 is expressed in human glial tumors.Acta Neuropathol Berl 82: 516–519, 1991

13. Kartner N, Riordan JR, Ling V: Cell surface P-glycoproteinassociated with multidrug resistance in mammalian celllines. Science 221: 1285–1288, 1983

14. Kirches E, Oda Y, Von Bossanyi P, Diete S, Schneider T,Warich Kirches M, Dietzmann K: Mdr1 mRNA expressiondiffers between grade III astrocytomas and glioblastomas.Clin Neuropathol 16: 34–36, 1997

15. Nabors MW, Griffin CA, Zehnbauer BA, Hruban RH,Phillips PC, Grossman ST, Brem H, Colvin M: Multidrugresistance gene (MDR1) expression in human brain tumors.J Neurosurg 75: 941–946, 1991

16. Noonan KE, Beck C, Holzmayer TA, Chin JE, Wunder JS,Andrulis IL, Gazdar AF, Willman CL, Griffith B,von Hoff DD, Roninson IB: Quantitative analysis of MDR1(multidrug resistance) gene expression in human tumors bypolymerase chain reaction. Proc Natl Acad Sci USA 87:7160–7164, 1990

17. Riordan JR, Deuchars K, Kartner N, Alon N, Trent J,Ling V: Amplification of P-glycoprotein genes in multidrug-resistant mammalian cell lines. Nature 316: 817–819, 1985

18. Tishler DM, Weinberg KI, Sender LS, Nolta JA, Raffel C:Multidrug resistance gene expression in pediatric primi-tive neuroectodermal tumors of the central nervous system.J Neurosurg 76: 507–512, 1992

19. Toth K, Vaughan MM, Peress NS, Slocum HK, RustumYM: MDR1 P-glycoprotein is expressed by endothelial cellsof newly formed capillaries in human gliomas but is notexpressed in the neovasculature of other primary tumors.Am J Pathol 149: 853–858, 1996

20. Abe T, Hasegawa S, Taniguchi K, Yokomizo A, Kuwano T,Ono M, Mori T Hori S, Kohno K, Kuwano M: Possi-ble involvement of multidrug-resistant-associated protein(MRP) gene expression in spontaneous drug resistance tovincristine, etopside and adriamycin in human glioma cells.Int J Cancer 58: 860–864, 1994

21. Endo K, Maehara Y, Kusumoto T, Ichiyoshi Y, Kuwano M,Sugimachi K: Expression of multidrug-resistance-associated protein (MRP) and chemosensitivity in humangastric cancer. Int J Cancer 68: 372–377, 1996

22. Narasaki F, Matsou I, Ikuno N, Fukuda M, Soda H, Oka M:Multidrug resistance-associated protein (MRP) gene expres-sion in human lung cancer. Anticancer Res 16: 2079–2082,1996

23. Tomonaga M, Oka M, Narasaki F, Fukuda M, Nakano R,Takatani H, Ikeda K, Teranishi K, Matsuo I, Soda H,Cowan KH, Kohno S: The multidrug resistance-associatedprotein gene confers drug resistance in human gastric andcolon cancers. Jpn J Cancer Res 87: 1263–1270, 1996

24. Mineura K, Izumi I, Watanabe K, Kowada M: Influenceof O6-methylguanine-DNA methyltransferase activity onchloroethylnitrosourea chemotherapy in brain tumors. Int JCancer 55: 76–81, 1993

25. Nagane M, Asai A, Shibui S, K, Matsutani M, Kuchino Y:Expression of O6-methylguanine-DNA methyltransferaseand chloroethylnitrosourea resistance of human braintumors. Jpn J Clin Oncol 22: 143–149, 1992

26. Tano K, Shiota S, Collier J, Foote RS, Mitra S: Isolation andstructural characterization of a cDNA clone encoding thehuman DNA repair protein for O6-alkylguanine. Proc NatlAcad Sci USA 87: 686–690, 1990

27. Wu Z, Chan CL, Eastman A, Bresnick E: Expression ofhuman O6-methylguanine-DNA methyltransferase in Chi-nese hamster ovary cells and restoration of cellular resis-tance to certain N-nitroso compounds. Mol Carcinog 4: 482–488, 1991

28. Hori T, Muraoka K, Saito Y, Sasahara K, Inagaki H, InoueY, Adach S, Anno Y: Influence of mode of ACNU adminis-tration on tissue and blood drug concentration in malignanttumors. J Neurosurg 66: 372–378, 1987

29. Hara A, Sakai N, Yamada H, Tanaka T, Kato K, Mori H,Sato K: Induction of glutathione-S-transferase, placentaltype in T9 glioma cells by dibutyryladenosine 3′,5′-cyclicmonophosphate and modification of its expression by nat-urally occurring isothiocyanates. Acta Neuropathol (Berl)79: 144–148, 1989

30. Hara A, Yamada H, Sakai N, Hirayama H, Tanaka T, MoriH: Immunohistochemical demonstration of the placentalform of glutathione-S-transferase, a detoxifying enzyme inhuman gliomas. Cancer 66: 2563–2568, 1990

171

31. Kim W-J, Kakehi Y, Kinoshita H, Arao S, Fukumoto M,Yoshida O: Expression patterens of multidrug-resistance(MDR1), multidrug resistance-associated protein (MRP),glutathione-S-transferase-π (GST-π ) and DNA topoiso-merase II (TOPO II) genes in renal cell carcinomas and nor-mal kidney. J Urol 156: 506–511, 1996

32. Mosmann T: Rapid colorimetric assay for cellular growthand survival: application to proliferation and cytotoxicityassays. J Immunol Meth 65: 55–63, 1983

33. Wada C, Kasai K, Kameya T, Ohtani H: A general transcrip-tion initiation factor IID, overexpressed in human lungs andbreast carcinoma and rapidly induced with serum stimula-tion. Cancer Res 52: 307–313, 1992

34. Kawasaki ES: Amplification of RNA. In: Innis MA,Gelfand DH, Sninski JJ et al. (eds). PCR Protocols, A Guideto Methods and Applications. San Diego, Academic Press,1990, pp 21–27

35. Chomczynski P, Sacchi N: Single-step method of RNA iso-lation by acid guanidinium thiocyanate–phenol–chloroformextraction. Anal Biochem 162: 156–159, 1987

36. Takakura K, Nomura K (eds): Relative Survival Rate inBrain Tumor Registry of Japan. The Committee of BrainTumor Registry of Japan, 1996, pp, 67–75

37. Nagai M, Arai T, Watanabe K: Immunological state ofpatients with malignant glioma and its alteration by inter-feron therapy. Biotherapy (Tokyo) 2: 242–247, 1988

38. Kubota T, Sasano N, Abe O, Nakao I, Kawamura E, SaitoT, Endo M, Kimura K, Demura H, Nagura H, Ogawa N,Hoffman RM: Potential of histoculture drug-response assayto contribute to cancer patient survival. Clin Cancer Res 1:1537–1543, 1995

39. Rosenblum ML, Gerosa MA, Wilson CB, Barger GR,Pertuiset BF, de Tribolet N, Dougherty DV: Stem cell stud-ies of human malignant brain tumors. Part 1: Developmentof the stem cell assay and its potential. J Neurosurg 58: 170–176, 1983

40. Silber JR, Mueller BA, Ewers TG, Berger MS: Compari-son of O6-methylguanine-DNA methyltransferase activityin brain tumors and adjacent normal brain. Cancer Res 53:3416–3420, 1993

41. Mineura K, Yanagisawa T, Watanabe K, Kowada M,Yasui N: Human brain tumor O(6)-methylguanine-DNAmethyltransferase mRNA and its significance as an indicator

of selective chloroethylnitroso urea chemotheraphy. Int JCancer 69: 420–425, 1996

42. Nagane M, Shibui S, Oyama H, Asai A, Kuchino T, NomuraK: Investigation of chemoresistance-related genes mRNAexpression for selecting anticancer agents in successful adju-vant chemotherapy for a case recurrent glioblastoma. SurgNeurol 44: 462–468, 1995

43. Ikeda J, Aida T, Sawamura Y, Abe H, Kaneko S, KashiwabaT, Kawamoto T, Mitsumori K, Saitoh H: Phase II studyof DTIC, ACNU, and vincristine combined chemotherapyfor supratentorial malignant astrocytomas. Neurol Med Chir(Tokyo) 36: 555–559, 1996

44. Nomura K, Watanabe T, Nakamura O, Ohira M, ShibuiS, Takakura K, Miki Y: Intensive chemotherapy withautologous bone marrow rescue for recurrent malignantgliomas. Neurosurg Rev 7: 13–22, 1984

45. Nagane M, Asai A, Shibui S, Nomura K, Kuchino Y: Appli-cation of antisense ribonucleic acid complementary to O6-methylguanine-deoxyribonucleic acid methyltransferasemessenger ribonucleic acid therapy of malignant gliomas.Neurosurgery 41: 434–441, 1997

46. Gresser I: Antitumor effects of interferon. Adv Cancer Res16: 97–140, 1972

47. Rosenblum MG, Yung WKA, Kelleher PJ, Ruzika F,Steck PA, Borden EC: Growth inhibitory effects ofinterferon-β but not interferon-α on human glioma cells:correlation of receptor binding, 2′,5′-oligoadenylate syn-thetase and protein kinase activity. J Interferon Res 10:141–151, 1990

48. Kondziolka D, Flickinger JC, Bissonette DJ, Bozik M,Lunsford LD: Survival benefit of stereotatic radiosurgery forpatients with malignant glial neoplasms. Neurosurgery 41:776–785, 1997

49. Barker II FG, Chang SM, Gutin PH, Malec MK,McDermott MW, Prados MD, Wilson CB: Survival andfunctional status after resection of recurrent glioblastomamultiforme. Neurosurgery 42: 709–723, 1998

Address for correspondence:Satoshi Tanaka, Department ofNeurosurgery, Kitasato University School of Medicine, 1-15-1Kitasato, Sagamihara-City, Kanagawa, 228-8555, Japan, Tel. andFax: +81-42-778-9337; E-mail: s1422.tanaka@nifty.ne.jp