Efficacy of protracted temozolomide dosing is limited in MGMT unmethylated GBM xenograft models

Abstract

Background

Temozolomide (TMZ) is important chemotherapy for glioblastoma multiforme (GBM), but the optimal dosing schedule is unclear.

Methods

The efficacies of different clinically relevant dosing regimens were compared in a panel of 7 primary GBM xenografts in an intracranial therapy evaluation model.

Results

Protracted TMZ therapy (TMZ daily M–F, 3 wk every 4) provided superior survival to a placebo-treated group in 1 of 4 O⁶-DNA methylguanine-methyltransferase (MGMT) promoter hypermethylated lines (GBM12) and none of the 3 MGMT unmethylated lines, while standard therapy (TMZ daily M–F, 1 wk every 4) provided superior survival to the placebo-treated group in 2 of 3 MGMT unmethylated lines (GBM14 and GBM43) and none of the methylated lines. In comparing GBM12, GBM14, and GBM43 intracranial specimens, both GBM14 and GBM43 mice treated with protracted TMZ had a significant elevation in MGMT levels compared with placebo. Similarly, high MGMT was found in a second model of acquired TMZ resistance in GBM14 flank xenografts, and resistance was reversed in vitro by treatment with the MGMT inhibitor O⁶-benzylguanine, demonstrating a mechanistic link between MGMT overexpression and TMZ resistance in this line. Additionally, in an analysis of gene expression data, comparison of parental and TMZ-resistant GBM14 demonstrated enrichment of functional ontologies for cell cycle control within the S, G2, and M phases of the cell cycle and DNA damage checkpoints.

Conclusions

Across the 7 tumor models studied, there was no consistent difference between protracted and standard TMZ regimens. The efficacy of protracted TMZ regimens may be limited in a subset of MGMT unmethylated tumors by induction of MGMT expression.

Temozolomide (TMZ) chemotherapy is an important component of treatment for glioblastoma multiforme (GBM), but the optimal dosing schedule of adjuvant TMZ is unclear. The established treatment regimen for newly diagnosed GBM patients following maximal surgical debulking is TMZ at 75 mg/m² daily for 6 weeks concurrent with radiation followed by adjuvant TMZ monotherapy dosed at 150–200 mg/m² on days 1–5 every 28 days (standard dosing).¹ Alternatively, protracted TMZ regimens (TMZ 150 mg/m² days 1–7 and 15–21 or 100 mg/m² days 1–21 every 28 d) and continuous metronomic dosing (50 mg/m² daily) have been promoted as potential methods to prevent or overcome TMZ resistance.² These regimens can provide more than double the cumulative TMZ exposure per cycle compared with standard adjuvant dosing, which potentially could result in greater tumor cell kill. Moreover, protracted TMZ exposure may deplete cellular levels of the key DNA repair protein O⁶-DNA methylguanine-methyltransferase (MGMT) as a potential means to increase responsiveness to TMZ therapy.^3,4 Despite this promising rationale, a randomized phase III clinical trial (RTOG [Radiation Therapy Oncology Group] 0525), comparing the standard 5/28-day adjuvant TMZ dosing regimen to the protracted 21/28-day dosing regimen, failed to demonstrate a benefit for protracted TMZ therapy. To gain further insight into the potential benefit (or lack thereof) of these different TMZ dosing regimens, the Mayo panel of primary GBM xenografts was used in an intracranial therapy model to evaluate the relative efficacy of standard versus protracted TMZ schedules in 7 different GBM tumor lines with varying levels of inherent TMZ sensitivity.

Materials and Methods

Primary and TMZ-resistant GBM Xenograft Lines

All animal studies were approved by the Mayo Clinic Institutional Animal Care and Use Committee. The establishment and maintenance of the Mayo GBM xenograft panel has been described.⁵ Similarly, a model for acquired TMZ resistance following protracted in vivo TMZ exposure has been described.⁶ The efficacy of different TMZ dosing regimens in orthotopic xenografts was evaluated essentially as described.⁷ Previous pharmacokinetic studies in mice suggest levels of exposure similar to those for humans when TMZ is dosed in mice on a milligram/kilogram basis at one-third the human dose in milligrams per square meter, that is, 50 mg/kg in mice provides a similar exposure as 150 mg/m² in humans.^8–11 Thus, in these mouse studies, standard dosing was 50 or 66 mg/kg on days 1–5 every 28 days, and protracted TMZ dosing was 25 or 33 mg/kg on days 1–5, 8–12, and 15–19 every 28 days. Early studies were performed with the higher dose level for each category (66 and 33 mg/kg), while later studies were performed at the lower dose level (50 and 25 mg/kg) to reduce the toxicity observed with the higher doses.

MGMT Promoter Methylation Analysis

Methylation specific (MS)–PCR and quantitative (q)MS-PCR were performed on flash-frozen archived flank tumor samples as previously described.¹² The samples used for analysis were either the specific tumor used to establish the intracranial therapy studies or a tumor sample within 1 passage of that used for the intracranial study. Universal methylated DNA (Chemicon) was used as a positive control, and normal human brain DNA was used as a negative control. Relative quantification of MGMT promoter methylation status (qMS-PCR signal) was performed using 7000 SDS 1.1 RQ software (ABI).

Analysis of MGMT mRNA Levels

Optimum Cutting Temperature–embedded (or paraffin-embedded in GBM43) brain from tumor-bearing mice was sectioned and stained with hematoxylin and eosin (H&E). Using the H&E as a template, tumor tissue was scraped from serial sections and processed for mRNA extraction and quantitative real-time (qRT) –PCR for MGMT as previously described.¹²

In vitro Cytotoxicity Assay

Cells were plated in 96-well plates (1500 cells/well) and pretreated with either 0 or 30 μM O⁶-benzylguanine for 1 h prior to treatment with 0 or 30 μM TMZ. Cells were incubated for 6 days, rinsed with phosphate buffered saline (PBS), and then frozen at −80°C. Plates then were processed using a CyQuant cell proliferation assay kit (Invitrogen) according to manufacturer's instructions. Fluorescence was measured at 480 nm excitation and 520 nm emission with a Tecan Infinite F200 plate reader (Männedorf).

MSH6 Immunohistochemistry

MutS homolog (MSH)6 immunohistochemistry (IHC) was performed in formalin-fixed and paraffin-embedded sections that were deparaffinized through 3 changes of xylene and dehydrated with graded ethanol washes. Antigen retrieval was subsequently carried out by steaming for 30 min in 10 mM sodium citrate (pH 6.0), followed by cooling in the buffer for 15 min. Slides were blocked in 5% bovine serum albumin for 1 h, followed by incubation with 1:100 dilution of rabbit polyclonal anti-MSH6 antibody (Epitomics S2269) for 1 h. Slides were stained using a Dako LSAB (labeled streptavidin–biotin) Plus system. Slides were counterstained with hematoxylin and mounted with permanent mounting media. To determine the positivity of MSH6 staining, digital images of 5 high-power fields were obtained for each tumor-bearing animal, and the fraction of area with positive staining was determined with KS400 image analysis software (Carl Zeiss).

Gene Expression Analysis

An Illumina BeadArray, which targets over 47 000 transcripts, was used for gene expression profiling. Total RNA from GBM14 and GBM14TMZ flash-frozen flank tumor xenograft tissue was isolated using the RNeasy kit (Qiagen), followed by quality determination using Agilent software. Sample labeling, hybridization, and scanning were performed by the Mayo Clinic Microarray Shared Facility. The raw data from gene expression intensity values were normalized using R software (http://www.r-project.org/) with the package fastlo.¹³ Gene expression intensity values were logarithm base-2 transformed for all subsequent analyses. Genes differentially regulated in GBM14 and GBM14TMZ with P < .05 were used to perform enrichment analysis for functional significance using MetaCore (GeneGo; http://www.genego.com) as previously described.¹⁴

Statistical Analysis

Cumulative survival probabilities were estimated using the Kaplan–Meier method. A log-rank test was used to compare survival across groups. A pooled survival analysis using Cox proportional hazards regression was performed including all 7 GBM lines, with GBM line included as a covariate in the pooled analysis. Repeated-measures analysis of variance was used to assess treatment-group differences for changes in weight over time. For this analysis, weight was the dependent variable, and time, treatment group, and Time × Treatment Group interaction were the independent variables. Differences in cell survival or gene expression were analyzed using a 2-sample Student's t-test. The criterion for statistical significance was taken as P < .05.

Results

Initial Comparison of TMZ Dosing Regimens

The Mayo GBM xenograft panel comprises over 50 distinct xenograft lines established directly from human GBM samples; MGMT status and in vivo TMZ sensitivity have been defined in 23 of these lines⁷ (unpublished data). From these data, 7 tumor lines (GBM6, GBM8, GBM12, GBM14, GBM39, GBM43, and GBM59) with varying TMZ responsiveness and MGMT methylation status were selected for evaluation in this study. In a previous study, GBM14 was found to have both unmethylated and methylated PCR products in an MGMT MS-PCR assay.⁷ However, in an analysis of archived tumor samples used for the present therapy evaluations, GBM14 was unmethylated by MS-PCR (Fig. 1A), qMS-PCR (Fig. 1B), and pyrosequencing (Supplemental Table 1). Thus, this study evaluated therapy response in 4 MGMT promoter hypermethylated lines (GBM8, GBM12, GBM39, and GBM59) and 3 MGMT promoter unmethylated lines (GBM6, GBM14, and GBM43).

Fig. 1.

MGMT status evaluation. Seven xenograft lines were selected for testing based on their MGMT promoter methylation status. Tumor samples used to establish the orthotopic studies were analyzed for MGMT promoter methylation by (A) MS-PCR and (B) qMS-PCR. In the MS-PCR assay, 2 PCR reactions were run specifically for an unmethylated (U) or methylated (M) promoter. In the qMS-PCR assay, a methylation-specific reaction was run and compared with the positive control. Shown are results for each xenograft line (numbers in bold), positive control (PC), negative control (NC), and a water only control. Abbreviation: RQ, relative quantity.

TMZ therapies were evaluated in mice with established orthotopic xenografts. Mice were randomized to three 28-day treatment cycles of placebo, standard TMZ, or protracted TMZ and followed until reaching a moribund state. Both standard and protracted TMZ therapies were associated with a significant prolongation in survival for 6 of 7 tumor lines (Table 1; Supplemental Fig. 1); in GBM6, only standard therapy significantly extended survival compared with placebo (P = .02), while protracted TMZ therapy did not (P = .44) (Fig. 2A). In a pooled analysis of all 7 tumor lines comparing either of the treatments with placebo, both standard therapy (hazard ratio [HR] = 0.09; 95% confidence interval [CI]: 0.06–0.14) and protracted TMZ therapy (HR = 0.09: 95% CI: 0.06–0.14) were superior to placebo therapy (overall P < .001). Moreover, in a direct comparison of protracted TMZ with standard therapy, there was no difference in efficacy (P = .88, HR = 0.98; 95% CI: 0.72–1.32). However, in comparing survival between the 2 TMZ dosing regimens for individual lines, standard dosing was associated with a statistically significantly longer survival than protracted TMZ therapy in GBM14 (Fig. 2B; median survival: 94 days vs 56 days, respectively; P < .001) and in GBM43 (Fig. 2C; median survival: 38 vs 36 days, respectively; P = .004). In contrast, standard TMZ dosing was associated with statistically inferior survival compared with protracted TMZ therapy in GBM12 (Fig. 2D; median survival: 74 vs 107 days, respectively; P = .03 for pooled analysis of 2 experiments). In summary, standard TMZ therapy was associated with superior survival in 2 or 3 MGMT unmethylated tumor lines, while protracted TMZ therapy was associated with superior survival in 1 of 4 MGMT hypermethylated tumor lines.

Table 1.

Comparison of standard vs protracted TMZ therapy

	Placebo		Standarda			Protractedb			Standard vs Protracted	MGMT
	Survivalc	n	Survivalc	n	Pd	Survivalc	n	Pd	Pe	Statusf
GBM6	41	8	56	8	.02	41	8	.44	.09	U
GBM8	40	10	156	9	<.001	176	9	.003	.1	M
GBM12g	16	20	74	20	<.001	107	20	<.001	.03	M
GBM14	20	10	94	10	<.001	56	10	.002	<.001	U
GBM39	34	9	240	10	<.001	200	10	<.001	.96	M
GBM43	29	10	38	10	<.001	36	10	<.001	.0036	U
GBM59	42	11	79	10	<.001	78	10	<.001	.98	M

^aStandard dosing regimen (50 or 66 mg/kg/d on days 1–5 every 28 d for 3 cycles).

^bProtracted dosing regimen (25 or 33 mg/kg/d on days 1–5, 8–12, and 15–19 every 28 d for 3 cycles).

^cMedian survival in days from the time of tumor implantation.

^dP-value comparing treatment vs placebo.

^eP-value comparing standard treatment vs protracted treatment.

^fMGMT promoter methylation status: U, unmethylated, M, methylated.

^gPooled data from 2 experiments.

Fig. 2.

Orthotopic survival assays. Mice with established orthotopic xenografts derived from (A) GBM6, (B) GBM14, (C) GBM43, and (D) GBM12 were treated with placebo, standard TMZ, or protracted TMZ. Fraction of surviving mice is plotted relative to the days from tumor implantation. The P-value comparing standard vs protracted TMZ therapy is shown for each study. The results for GBM12 are pooled from 2 independent studies.

The toxicity of protracted TMZ therapy was compared with standard treatment in an analysis of weight loss. In both tumor lines where protracted TMZ therapy was associated with a worse survival, there was no significant difference in the weight loss observed between the 2 treatment regimens in either the GBM14 (P = .38) or the GBM43 (P = .26) experiment (Fig. 3). Consistent with the assumption that tumor burden, and not toxicity, accounted for the early mortality of the dose-dense-treated GBM14 tumors, tumor sections stained by hematoxylin and eosin were reviewed by a neuropathologist (C.G.), and all but 1 mouse in the standard dosing regimen were found to have appreciable tumor burden (Supplemental Fig. 2). There were several mice that did not have adequately preserved brain sections and were found dead in their cages (5 placebo-treated, 2 protracted TMZ [both of these mice died after the first dose of TMZ], and 3 standard TMZ–treated mice). A re-analysis of the survival data for GBM14 including only those mice with histologic proof of recurrent tumor confirmed a significantly decreased survival in the protracted TMZ–treatment group compared with the standard TMZ–treatment group (P = .001). Thus, these data suggest that the inferior survival for GBM14 and GBM43 mice treated with protracted TMZ was due to reduced efficacy and not excessive toxicity.

Fig. 3.

Body weight analysis. Serial body weight measurements from mice treated with placebo, standard TMZ, or protracted TMZ are shown for (A) GBM14 and (B) GBM43. The survival of these mice is shown in Fig. 2B and C, respectively.

Analysis of Tumor Specimens From TMZ Schedule Experiments

MGMT directly repairs cytotoxic O⁶-methylguanine (O⁶MG) lesions induced by TMZ, and MGMT expression is mechanistically linked to TMZ resistance in multiple preclinical models.¹⁵ Therefore, archived intracranial tumor samples for GBM6 (where benefit existed with only standard TMZ), GBM12 (where protracted TMZ dosing was superior), GBM14, and GBM43 (where standard dosing was superior) from the original therapy studies (see Fig. 2) were analyzed for MGMT mRNA expression levels by qRT-PCR (Fig. 4A). In comparison with MGMT levels of T98 glioma cells (MGMT unmethylated and TMZ resistant), MGMT levels in GBM12 tumors were quite low in placebo-treated mice (mean ± SD: 0.0043 ± 0.0057, n = 3) and in mice treated with either standard TMZ dosing (0.00059 ± 0.00048, n = 3, P = .3 relative to placebo) or protracted TMZ dosing (0.049 ± 0.082, n = 3, P = .4 relative to placebo). In contrast, mean MGMT expression was low in GBM14 mice treated with placebo (0.014 ± 0.012, n = 2) and standard dosing (0.18 ± 0.18, n = 2, P = .3 relative to placebo) but significantly elevated with protracted TMZ dosing (0.81 ± 0.32, n = 3, P = .04 relative to placebo). For the protracted TMZ treatment, this represents a 58-fold increase in MGMT in comparison with placebo-treated mice. In GBM43, MGMT level was significantly upregulated in both the standard TMZ group (11.89 ± 6.86, n = 5, P < .05) and the protracted TMZ group (13.65 ± 5.31, n = 5, P < .01) compared with the placebo group (3.17 ± 1.53, n = 5). Finally, in GBM6, MGMT levels were significantly elevated in the protracted TMZ group (7.88 ± 3.74, n = 7) compared with either the placebo group (3.77 ± 1.97, n = 6, P = .04) or the standard TMZ group (4.40 ± 2.13, n = 7, P = .05). Collectively, these data suggest that MGMT elevation with protracted TMZ therapy may account for the reduced benefit in protracted TMZ treatment in the unmethylated tumors.

Fig. 4.

MGMT and MSH6 expression. (A) Optimum Cutting Temperature–embedded or paraffin-embedded brain specimens from GBM12, GBM14, GBM6, and GBM43 orthotopic xenografts (treated in Fig. 2) were serially sectioned, and tumor was isolated from slides and processed for human-specific MGMT qRT-PCR analysis. Each symbol represents the MGMT expression level for an individual tumor, relative to a T98 control sample. *P< .05 and **P< .01 compared with placebo group. (B) Immunohistochemistry staining of MSH6 with paraffin-embedded brain sections from GBM14 xenografts treated with placebo (n = 2), standard TMZ (n = 4), and protracted TMZ (n = 3).

Mutational inactivation of MSH6, associated with loss of MSH6 protein expression, also has been linked to development of TMZ resistance.^16,17 Therefore, MSH6 IHC was performed on the intracranial tumor samples from mice treated in the GBM14 experiment. As seen in Fig. 4B, there was no appreciable difference in MSH6 expression levels among placebo-, standard TMZ–, and protracted TMZ–treated tumors. Thus, although the number of animals available for analysis was limited, these data suggest that the reduced efficacy of protracted TMZ therapy in GBM14 could be related to significantly elevated MGMT expression.

Modeling-acquired TMZ Resistance in GBM14

The development of TMZ resistance in GBM14 with TMZ therapy was modeled in a separate study in which a mouse with an established GBM14 heterotopic tumor was treated with escalating doses of TMZ until the originally responsive tumor was unaffected by 5 consecutive daily doses of 120 mg/kg TMZ; this derivative line is referred to as “GBM14TMZ.” To assess the extent of TMZ resistance in this model, orthotopic xenografts were established from GBM14TMZ, and mice were randomized to therapy with a single cycle of 66 mg/kg TMZ daily × 5 days. While the parental GBM14 line was significantly responsive to TMZ therapy (median survival: 186 d for TMZ treatment vs 33 d for placebo treatment, P < .001), GBM14TMZ was markedly less responsive to TMZ therapy (median survival: 45 d for TMZ treatment vs 26 d for placebo treatment, P < .001; Fig. 5A). As with the parental tumor, GBM14TMZ can be propagated by serial heterotopic tumor passage and thus can provide an unlimited amount of tumor tissue for molecular and functional analyses.

Fig. 5.

Model for acquired TMZ resistance in GBM14. (A) Mice with established orthotopic xenografts derived from GBM14 or GBM14TMZ were treated with placebo or TMZ (66 mg/kg/d, days 1–5). Fraction of surviving mice is plotted relative to the days from tumor implantation. (B) Flank tumor tissue from 3 GBM14 and 3 GBM14TMZ heterotopic xenografts were analyzed for MGMT mRNA by qRT-PCR. The mean ± SD of measured MGMT level relative to a control sample derived from T98 glioma cells is shown. (C) Short-term explant cultures from GBM14TMZ were seeded in 96-well plates and treated with graded concentrations of TMZ with and without 30 μM O⁶BG. Six days later, cells were stained and quantitated in a CyQuant assay. Mean relative fluorescence ± SEM is plotted for 3 independent experiments.

The potential role of MGMT overexpression as a mechanism of resistance was evaluated in GBM14TMZ. Quantitative RT-PCR was performed on tumor samples derived from GBM14 and GBM14TMZ flank tumor specimens, respectively, using human-specific primers for MGMT and glyceraldehyde 3-phosphate dehydrogenase. Consistent with the intracranial GBM14 study, MGMT mRNA levels were significantly elevated in GBM14TMZ (MGMT mean ± SD: 0.11 ± 0.02 relative to T98 positive control) compared with parental GBM14 (2.2E-4 ± 3.8E-5; P = .001; Fig. 5B); this represents a 500-fold increase in MGMT mRNA expression levels. The functional significance of MGMT elevation in GBM14TMZ also was assessed using short-term explant cultures in an in vitro cell survival assay. Treatment with 30 μM TMZ alone had essentially no effect on cell survival (absorbance relative to untreated control of 1.12 ± 0.12, P = .29), while co-incubation of TMZ with the specific MGMT inhibitor O⁶-benzylguanine resulted in a significant reduction in cell number at 30 μM (absorbance of 1.12 ± 0.12 vs 0.17 ± 0.02, respectively; P < .001; Fig. 5C). Collectively, the qRT-PCR data and the reversal of resistance with O⁶-benzylguanine mechanistically linked elevated MGMT expression to the development of TMZ resistance in GBM14TMZ.

Gene expression profiling was performed on GBM14 and the derivative GBM14TMZ line to gain additional insight into potential mechanisms of TMZ resistance in this line. Using 3 independent flank tumor samples from both the parental and the resistant subline, the expression levels associated with each gene probe were compared using Student's t-test. For those with a significant difference in expression (P < .05), genes were ranked by the extent of fold-change between GBM14 and GBM14TMZ, and the top 25 genes that were increased or decreased in the resistant GBM14TMZ line are listed in Tables 2 and 3, respectively. Notably, the DNA repair gene MGMT was the most highly elevated gene (3.58 log2-fold increase) in GBM14TMZ compared with parental GBM14. Further analysis of the expression data was performed on genes with a significant (P < .05) difference in expression between the GBM14 and GBM14TMZ lines using the Metacore software package. This analysis highlighted potential dysregulation of several functional networks, including regulation of mitosis, G2/M and S phases, and DNA damage checkpoints (Fig. 6). Although validation of these altered networks is beyond the scope of the current paper, these data provide interesting potential hypotheses about additional resistance mechanisms that may contribute to the TMZ resistance found in GBM14TMZ.

Table 2.

Top 25 upregulated genes in GBM14TMZ vs GBM14

Rank	Gene Symbol	Gene Name	Transcript	Function (UniProtKB/Swiss-Port Classification)	Pa	Log2 FCb
1	MGMT	O6 methylguanine-DNA methyltransferase	NM_002412.2	DNA repair	.000020	3.58
2	CT45-3	Cancer/testis antigen 45-3	NM_001017435.1	Unknown	.000108	2.62
3	CT45-4	Cancer/testis antigen 45-4	NM_001017436.1	Unknown	.000040	2.55
4	CT45-5	Cancer/testis antigen 45-5	NM_001007551.2	Unknown	.000073	2.29
5	EDG2	Lysophosphatidic acid receptor 1 (LPAR1)	NM_001401.3	G protein-coupled receptor	.000007	2.01
6		Unknown	Hs.524902	Unknown	.001850	1.88
7	COBLL1	COBL-like 1	NM_014900.3	Unknown	.000203	1.65
8	CT45-1	Testis antigen family 45-1	NM_001017417.1	Unknown	.003657	1.63
9	GLRA2	Glycine receptor, alpha 2	NM_002063.2	Neurotransmitter-gated ion channel	.008723	1.57
10	SNCAIP	Alpha-synuclein interacting protein	NM_005460.2	Protein-protein interaction	.009129	1.49
11	DIO2	Deiodinase, iodothyronine, type II (DIO2)	NM_001007023.1	Deiodinase activity	.000384	1.40
12	LOC115749	Chromosome 12 open reading frame 56	XM_056680.7	Unknown	.000006	1.38
13	MAGEA1	Melanoma antigen family A, 1	NM_004988.3	Directs expression of antigen MZ2-E	.000214	1.36
14		Unknown	Hs.571502	Unknown	.000744	1.35
15	FLJ22746	Family with sequence similarity 124B	NM_024785.2	Unknown	.000341	1.26
16	C20orf103	Chromosome 20 open reading frame 103	NM_012261.2	Unknown	.007849	1.24
17	ASCL1	Achaete-scute complex-like 1 (Drosophila)	NM_004316.2	Transcription factor	.014055	1.18
18	PDPN	Podoplanin	NM_001006625.1	May regulate cell migration	.003098	1.14
19		Unknown	Hs.48372	Unknown	.008660	1.05
20	TCF12	Transcription factor 12	NM_207040.1	Transcription factor	.002951	1.02
21	WISP1	Wnt1 inducible signaling pathway protein 1	NM_003882.2	Wnt signaling downstream regulator	.011494	1.01
22	ROBO2	Roundabout, homolog 2 (Drosophila)	NM_002942.1	Receptor (axon guidance)	.000593	0.99
23	TOX	Thymus high mobility group box protein	NM_014729.2	Chromatin assembly; transcription	.003650	0.97
24	TSPAN12	Tetraspanin 12	NM_012338.3	Regulator of cell surface receptor signaling	.005783	0.96
25	SULF1	Sulfatase 1	NM_015170.1	Sulfatase activity	.023802	0.95

^aP-value comparing gene expression in GBM14TMZ vs GBM14.

^bLog₂ fold-change in gene expression between GBM14TMZ and GBM14.

Table 3.

Top 25 downregulated genes in GBM14TMZ vs GBM14

Rank	Gene Symbol	Gene Name	Transcript	Function (UniProtKB/Swiss-Port Classification)	Pa	Log2 FCb
1	RPS4Y1	Ribosomal protein S4, Y-linked 1	NM_001008.3	Protein synthesis	.000203	−4.91
2	EIF1AY	Eukaryotic translation initiation factor 1A, Y-linked	NM_004681.2	Protein biosynthesis	.000003	−3.82
3	A2M	Alpha-2-macroglobulin	NM_000014.4	Protease inhibitor and cytokine transporter	.035084	−2.82
4	TGFBI	Transforming growth factor, beta-induced	NM_000358.1	Cell-collagen interactions	.009982	−2.40
5	C14orf78	AHNAK2 nucleoprotein 2	XM_290629.6	Unknown	.002865	−2.08
6	SERPINE1	Serpin peptidase inhibitor, clade E	NM_000602.1	Glioma cell invasion and mobility	.022639	−1.70
7	ITGA3	Integrin, alpha 3	NM_002204.1	Cell migration	.009520	−1.65
8	LMO2	LIM domain only 2 (rhombotin-like 1)	NM_005574.2	Hematopoietic development	.000380	−1.63
9	NOVA1	Neuro-oncological ventral antigen 1	Hs.31588	RNA-binding protein	.015658	−1.61
10	GYPC	Glycophorin C	NM_002101.3	An integral membrane glycoprotein	.000028	−1.57
11	ZNF14	Zinc finger protein 14	NM_021030.2	Transcriptional regulation	.039802	−1.56
12	VGF	VGF nerve growth factor inducible	NM_003378.2	Cell-cell interaction	.020937	−1.55
13	NPNT	Nephronectin	NM_001033047.1	Functional ligand of integrins	.003712	−1.49
14	CTSK	Cathepsin K	NM_000396.2	Extracellular matrix degradation	.001217	−1.47
15	COL7A1	Collagen, type VII, alpha 1	NM_000094.2	Basement membrane organization	.028794	−1.47
16	TGM2	Transglutaminase 2	NM_004613.2	Catalyzes the cross-linking of proteins	.024484	−1.44
17		Unknown	Hs.544637	Unknown	.038929	−1.44
18	TNC	Tenascin C	NM_002160.1	Glioma-associated- extracellular matrix antigen	.002279	−1.44
19	LRRFIP1	Leucine rich repeat (in FLII) interacting protein 1	NM_004735.2	Transcriptional repressor	.031952	−1.40
20	OLFM1	Olfactomedin 1	NM_014279.2	May have an essential role in nerve tissue	.004807	−1.37
21	ADCY8	Adenylate cyclase 8 (brain)	NM_001115.1	Calcium-stimulable signaling	.001180	−1.37
22	SPP1	Secreted phosphoprotein 1	NM_000582.2	Acts as a cytokine	.030730	−1.36
23	MUC1	Mucin 1, cell surface associated	NM_001018021.1	Cell adhesion and cell signaling	.041100	−1.36
24	MYBPC1	Myosin binding protein C	NM_002465.2	May modulate muscle contraction	.004820	−1.35
25	MCART1	Mitochondrial carrier triple repeat 1	NM_033412.1	Unknown	.036059	−1.29

^aP-value comparing gene expression in GBM14TMZ vs GBM14.

^bLog₂ fold-change in gene expression between GBM14TMZ and GBM14.

Fig. 6.

Pathway analysis. Top ranked 10 functional pathways revealed by GeneGo Metacore analysis of differentially regulated genes with P < .05. Pathways are ranked based upon P-value, bars represent inverse log of the P-value.

Discussion

Multiple clinical TMZ dosing regimens have been described, but the most efficacious regimens have not yet been fully defined. In comparison with the standard regimen of TMZ given for 5 days every 28 days, several groups have explored more protracted dosing schedules that allow greater cumulative TMZ exposure with tolerable side effects.² Alternatively, others have evaluated more dose-intense regimens that deliver the same cumulative TMZ dose as standard regimen but in a shorter time frame.^18,19 While the latter regimens are associated with significantly elevated hematologic toxicities and are not clinically used, there is significant interest in using protracted dosing regimens to enhance the efficacy of TMZ-based therapies. In a direct comparison between standard and protracted adjuvant TMZ regimens in newly diagnosed GBM patients, the RTOG 0525 clinical trial demonstrated no significant difference in overall survival for patients treated with either regimen.²⁰ Similarly, in a pooled analysis of all 7 tumor lines evaluated in this study, there was no difference in survival for mice treated with standard or dose-dense regimens. In an evaluation of individual tumor lines, protracted TMZ therapy provided superior survival benefit compared with standard dosing in only 1 of 7 xenograft lines, while the same protracted TMZ regimen was inferior in 2 xenograft lines. Thus, while both regimens were equally efficacious in the majority of tumors tested, there are select tumors with unique sensitivities to one or the other dosing regimen.

Development of protracted TMZ regimens was based on the premise that protracted TMZ schedules will suppress MGMT DNA repair activity. MGMT repairs cytotoxic O⁶MG lesions through a covalent attack on the O⁶-methyl group by a cysteine sulfhydryl group in the active site. The resulting methylated MGMT protein is inactive and must be replaced.²¹ Since MGMT is consumed in this repair reaction, protracted daily TMZ dosing can lead to reduced MGMT activity if demand for MGMT repair activity outstrips the synthesis of new MGMT molecules, and the resulting accumulation of unrepaired O⁶MG lesions could lead to improved efficacy in tumor lines that express MGMT protein. However, while clinical studies have demonstrated robust suppression of MGMT activity in peripheral blood mononuclear cells (PBMCs) during protracted TMZ treatment schedules, there was only marginal suppression of MGMT activity in paired patient tumor biopsy samples.^3,19,22 Although this may reflect simply heterogeneity of drug delivery within tumors compared with PBMCs, a previous study from our lab demonstrated increased MGMT expression and activity following TMZ treatment of several GBM xenografts, including the GBM43 tumor line used in the current study.²³ Similarly, protracted TMZ dosing was associated with a marked elevation in MGMT expression in all 3 unmethylated tumor lines (GBM6, GBM14, and GBM43), and a similar effect was seen with standard TMZ dosing in GBM43 in the current study. Several studies of paired patient specimens from initial diagnosis and failure after alkylating therapy have demonstrated robust increases in MGMT protein expression or repair activity or loss of MGMT promoter hypermethylation in recurrent tumor specimens.^24–28 Collectively, these data suggest that potential suppression of MGMT activity through protracted TMZ dosing may be offset by increased MGMT expression in at least a subset of GBM tumors.

In the current study, protracted TMZ dosing was significantly less effective than standard dosing only in MGMT unmethylated tumor lines. Despite a 33% lower cumulative TMZ dose with each treatment cycle, standard TMZ dosing provided statistically significant prolongation in survival compared with protracted TMZ therapy in the MGMT unmethylated GBM14 and GBM43 lines; and in GBM6, standard but not protracted TMZ therapy was associated with a statistically significant survival benefit compared with placebo. While the extent of survival benefit with protracted TMZ therapy in the latter 2 lines is limited, these data suggest that protracted TMZ therapy may be specifically ineffective in MGMT unmethylated tumors. Ongoing studies in our laboratory are investigating whether the observed increase in MGMT expression in GBM14 reflects simply a selection pressure for elevated MGMT expression during the multiple cycles of treatment or is a more dynamic process that is occurring specifically in response to continuous DNA methylation damage. Consistent with the latter possibility, a pathway analysis comparison between GBM14 and GBM14TMZ suggests significant activity of both sp1 and p53 transcription factors (unpublished data), and both of these transcription factors have been linked to regulation of MGMT expression.^29–32 In an analysis of functional ontologies, cell cycle control within S, G2, and M phases of the cell cycle and DNA damage checkpoints were significantly enriched. Although beyond the scope of the current study, these data are consistent with a hypothesis that MGMT overexpression in recurrent GBM14 tumors following protracted TMZ therapy might be related to a more robust activation of DNA damage-responsive pathways associated with protracted TMZ exposure.

The direct comparison of 2 clinically important treatment regimens in multiple primary GBM xenograft models may provide important insight into completed or ongoing clinical studies of different TMZ dosing schedules. Two studies in GBM patients have been published, and results from both studies are consistent with decreased benefit associated with protracted TMZ dosing schedules. First, in a randomized phase II comparison of procarbazine/lomustine/vincristine and TMZ chemotherapy in recurrent GBM, patients were further randomized in the TMZ arm to either standard (5 d every 28 d) or protracted (21 d every 28 d) TMZ therapy. In this comparison, standard TMZ treatment was associated with superior progression-free survival compared with treatment with either protracted TMZ or procarbazine/lomustine/vincristine.³³ Second, in a study of adjuvant TMZ dosing regimens in newly diagnosed GBM, patients randomized to continuous low-dose daily TMZ (50 mg/m²/d) had inferior overall and progression-free survival compared with patients randomized to higher-dose therapy given every other week (150 mg/m²/d, days 1–7, 15–21).³⁴ Finally, the RTOG has completed a large randomized phase III trial in newly diagnosed GBM patients comparing standard (5 d every 28 d) or protracted (21 d every 28 d) adjuvant TMZ therapy. Although the final results have not yet been published, the initial analysis suggested no significant difference in overall survival for the 2 treatment regimens. The concordance of the results from these clinical studies with our xenograft study suggests that the primary GBM xenograft model, used at the Mayo Clinic and elsewhere, is a robust preclinical therapy evaluation platform that can effectively model novel therapeutic strategies. Moreover, because tumor samples from before and after therapy can be readily attained, these model systems can be used to interrogate potential mechanisms of sensitivity and resistance to treatment.

Conclusions

There was not an appreciable difference in efficacy between protracted and standard TMZ dosing in a panel of primary GBM xenografts. The efficacy of protracted therapy may be specifically limited in MGMT unmethylated tumors due to induction of MGMT expression.

Funding

This work was supported by the Mayo Foundation, the National Cancer Institute (grant nos. CA108961, CA127716, and CA141121) and the Brain Tumor Funders Consortium.