Journal of Neuro-Oncology

, Volume 82, Issue 2, pp 133–139

Hypermethylation of the proapoptotic gene TMS1/ASC: prognostic importance in glioblastoma multiforme


    • Department of NeurosurgeryUniversity of Dresden
    • Department of Neurosurgery, Klinikum FuldaAcademic Hospital Philipps University Marburg
  • Gabriele Schackert
    • Department of NeurosurgeryUniversity of Dresden
  • Manel Esteller
    • Cancer Epigenetics LaboratorySpanish National Cancer Centre (CNIO)
Laboratory investigation

DOI: 10.1007/s11060-006-9264-4

Cite this article as:
Martinez, R., Schackert, G. & Esteller, M. J Neurooncol (2007) 82: 133. doi:10.1007/s11060-006-9264-4


The identification of clinical subsets of glioblastomas (GBM) associated with different molecular genetic profiles had opened the possibility to design tailored therapies to individual patients. One of the most intrigued subtypes is the long-term survival (LTS) GBM, which responds better to current therapies. The present investigation on GBM from 50 consecutive GBM displaying classic survival and seven LTS GBM is based on molecular epigenetic, clinical and histopathological analyses. Our aim was to recognize biomarkers useful to distinguish LTS from classic GBM. We analyzed the promoter methylation status of key regulator genes implicated in tumor invasion (TIMP2, TIMP3), apoptosis and inflammation (TMS1/ASC, DAPK) as well as overall survival, therapy status and tumor pathological features. For the first purpose a methylation-specific PCR approach was performed to analyze the CpG island promoter methylation status of each gene. The overall TMS1/ASC methylation rate in the 57 analyzed tumors was 21.05%. Hypermethylation of TMS1/ASC was significantly more frequent in LTS GBM (57.1% vs. 16%, P = 0.029, Fisher’s exact test). DAPK promoter hypermethylation was only observed in the LTS subset (14.3%) whereas TIMP2 and TIMP3 were unmethylated in both GBM collectives. Our results strongly suggest that, compared to classic GBM, LTS GBM display distinct epigenetic characteristics which might provide additional prognostic biomarkers for the assessment of this malignancy.


GlioblastomaLong-term survivalEpigeneticHypermethylationTMS1DAPKTIMP


Long-term survival (LTS) occurs in about 2–5% of all glioblastoma multiforme (GBM) patients and is defined by a median survival time of more than 3 years, approximately three times longer than the average of classic GBM [1]. Understanding this subset of GBM might reveal decisive biological aspects of this malignancy. Clinical characteristics such as younger age, high Karnofsky performance status (KPS) score at diagnosis and a radical extent of surgical resection are associated with a better prognosis [2, 3]. However, the aforementioned characteristics do not explain why GBM patients with classic survival harboring the same constellation of favorable features succumb to the disease rapidly. At molecular level the scenario is not quite different [4, 5]. Thus, overexpression of the protein p53, unaltered copy number of the epidermal growth factor receptor gene (EGFR), combined LOH 1p/19q, loss of chromosome arm 19q and low rate of tumor proliferation were observed in LTS GBM [69], but they are also present in GBM patients dying at the statistically expected time. Based on the former data, there is little doubt that the great efforts made in genetics analysis of survival-related GBM features resulted in a paucity of translation into patient benefit in terms of survival. Perhaps with the exception of the DNA repair gene MGMT (O6-methyl-guanine-DNA methyltransferase) methylation status.

Since we described the strong association between MGMT hypermethylation and positive clinical response to carmustine in a subset of glioblastoma multiforme (GBM) [10], further studies have confirmed this observation and expanded the association between methylation-mediated gene silencing and patients outcome [1115]. Recently, we have observed in the long-term survivors GBM subgroup a significant higher rate of MGMT hypermethylation compared to classic GBM (Martinez et al., submitted). These data agree with previous investigations revealing a different methylation profile in astrocytoma subtypes [16] and human cancers including genes implicated in cell cycle (p16INK4a, p15INK4b), tumor suppression (RB, VHL), DNA repair and genome integrity (MGMT, hMLH1, BRCA1) as well as tumor invasion and apoptosis (DAPK, TIMPs, CDH1) (reviewed in References 17 and 18).

We wondered whether epigenetic inactivation of genes previously observed to be silenced in human cancers might distinguish LTS—from classic GBM as well. TMS1/ASC (target of methylation-induced silencing-1) mediates intracellular signaling from apoptotic and inflammatory stimuli and it is hypermethylated in ovarian- and breast cancer and GBM [13, 19, 20]. DAPK (death-associated protein kinase) is a calmodulin-regulated serine/threonine kinase that acts in several apoptotic pathways initiated by Fas, interferon-γ and tumor necrosis factor-α and it is methylated in gastric- esophageal-, bladder-, non-small cell lung cancers and brain metastasis [2125]. TIMP2 and TIMP3 (tissue inhibitors of metalloproteinases 2 and 3, respectively) are inhibitors of matrix metalloproteinases, a group of zinc-binding endopeptidases involved in the degradation of the extracellular matrix, which were found to be hypermethylated in brain tumors, kidney-, colon-, esophageal- and breast cancers [11, 16, 18, 24].

To address this topic, we comparatively studied the CpG island promoter methylation status of TIMP2, TIMP3, TMS1/ASC and DAPK in two GBM populations of classic- and LTS patients. The identification of such, not yet analyzed, epigenetic pattern is of capital importance for the development of targeted therapies for this still incurable malignancy.

Patient and methods

Patient collectives, tumor samples

Primary tumor tissue samples were obtained from 57 GBM patients treated at the Department of Neurosurgery of the University of Dresden, Germany. Of these, 7 patients were LTS GBM (survival time more than 3 years) whereas 50 consecutive GBM patients showed standard survival (≤15 months). Informed consent for samples and data analysis from each patient or the patient’s caretaker was obtained. Survival time was defined as the time lapse from initial surgery to patient’s death or, in LTS cases, the last contact. Standard therapy for all patients in both groups was maximal surgical resection followed by both external fractionated radiotherapy (mean: 58 Gy) and chemotherapy with ACNU, teniposide and/or temozolomide.

Histopathologic and immunohistochemical evaluation

Tissue from different tumor areas was immediately frozen in liquid nitrogen after removal and stored at −80°C. DNA isolation was performed following standard procedures. Representative tumor samples were evaluated by pathologists of the Institute of Pathology, University of Dresden, Germany according to the World Health Organization (WHO) criteria. The immunohistochemical tumor characterization included antibody against glial fibrillary acidic protein (GFAP, Dako, Denmark). The standard avidin-biotin-peroxidase complex method (Merck Tissue Gnost, Merck, Germany) was performed.

Analysis of CpG island promoter hypermethylation by methylation-specific PCR (MSP)

DNA methylation patterns in the CpG islands of TMS1/ASC, DAPK, TIMP2 and TIMP3 were determined by chemical modification of unmethylated, but not the methylated, cytosines to uracil, and subsequent PCR using primers specific for either methylated or the modified unmethylated DNA [26, 27]. DNA (1 μg) was denatured by NaOH and modified by sodium bisulfite. DNA samples were then purified using Wizard DNA purification resin (Promega, Madison, Wis., USA), again treated with NaOH, precipitated with ethanol, and resuspended in water.

For TIMP2, MSP primers that specifically recognized the unmethylated sequence were 5′-GTA ATA AAA TTG TGG TTT GGT TTA AGT TT-3′ (sense) and 5′-TTC TCT CCT CTT TAT CTC AAA AAC ACA-3′ (antisense) and for the methylated sequence were 5′-AAT AAA ATT GCG GTT CGG TTT AAG TTC-3′ (sense) and 5′-CTC TCC TCT TTA TCT CGA AAA CGC G-3′ (antisense). For TIMP3, primer sequences for unmethylated reaction were 5′-TTT TGT TTT GTT ATT TTT TGT TTT TGG TTT T-3′ (sense) and 5′-CCC CCA AAA ACC CCA CCT CA-3′ (antisense) and for the methylated sequence were 5′-CGT TTC GTT ATT TTT TGT TTT CGG TTT C-3′ (sense) and 5′-CCG AAA ACC CCG CCT CG-3′ (antisense). For DAPK, primer sequences for unmethylated reaction were 5′-GGA GGA TAG TTG GAT TGA GTT AAT GTT-3′ (sense) and 5′-CAA ATC CCT CCC AAA CAC CAA-3′ (antisense) and for the methylated reaction were 5′-GGA TAG TCG GAT CGA GTT AAC GTC-3′ (sense) and 5′-CCC TCC CAA ACG CCG A-3′ (antisense). For TMS1/ASC, primers used for the unmethylated sequence were 5′-GGT TGT AGT GGG GTG AGT GGT-3′ (sense) and 5′-CAA AAC ATC CAT AAA CAA CAA CAC A-3′ (antisense) and for the methylated sequence were 5′-TTG TAG CGG GGT GAG CGG C-3′ (sense) and 5′-AAC GTC CAT AAA CAA CAA CGC G-3′ (antisense). Reactions were hot-started at 95°C for 5 min and held at 80°C before addition of 1.25 U of Taq polymerase (Sigma, St Louis, MO, USA). Temperature conditions for PCR were as follows: 35 cycles of 95°C for 30 s, 59°C for 30 s and 72°C for 30 s, followed by one cycle of 72°C for 5 min.

Placental DNA treated in vitro with Sss I methyltransferase (New England Biolabs, Beverly, Mass., USA) was used as positive control for methylated alleles, and DNA from normal lymphocytes was used as negative control for methylated alleles. Controls without DNA were performed for each set of PCR. Ten microliters of each PCR reaction was directly loaded onto non-denaturing 6% polyacrylamide gels, stained with ethidium bromide, and visualized under UV illumination.

Statistical analysis

Student t-test and Chi-square test (with Fisher–Yates corrections when appropriate) were performed to compare differences between groups depending on the analyzed variables. Confidence interval (CI) was obtained through logistic regression. Kaplan–Meier analysis and log-rank test were done to compare survival between groups defined by gene methylation status. Univariate and multivariate analysis of survival time and gene methylation status were estimated by the Kaplan–Meier method and Cox proportional hazard model, respectively. A value of P < 0.05 was considered to be statistically significant. Analyses were performed with the use of SPSS software (version 10, SPSS Inc., Chicago, IL., USA).


Clinical characteristics of the patients

The median age of the LTS GBM patients was 47 years (SD: 10.4 years), similar to the classic GBM group (53 years, SD: 11.4 years, P = 0.60, Fisher–Yates corrected Chi-square test). The median KPS scores at diagnosis in both groups were also alike: 89 in the LTS group and 83 in the standard group (P = 0.70, Fisher–Yates corrected Chi-square test). The median survival time of LTS GBM patients was 53 months (SD: 14.1 months) with 2 patients alive at the time of the last contact, whereas it was 11.5 months in standard GBMs (SD: 5.2 months). The male: female ratio was similar in both groups as well: LTS GBMs 1:1; standard GBMs 1.2:1.

By Kaplan–Meier analysis we observed, as expected, a correlation between age and survival (P = 0.03, Kaplan–Meier analysis, log-rank test, Fig. 1). Concerning cerebral location of tumors, a trend to better outcome was observed in patients harboring a tumor mass not involving the temporal lobe (P = 0.075, log-rank test).
Fig. 1

Kaplan–Meier graph describing the relationship between age and survival rate. Younger GBM patients (independently of the condition classic survivor/long-term survivor) showed a better outcome (P = 0.03, log-rank test)

Histopathologic evaluation

Routine haematoxylin–eosin (HE) sections of formalin-fixed and paraffin embedded tissue demonstrated a high cellularity in all cases (Fig. 2). Gemistocytic tumor cells demonstrating mitosis were evidenced. Pleomorphic cells containing atypical nuclei with numerous atypical mitotic figures were found as well. The lesions demonstrated areas of pseudopalisading necrosis and microvascular proliferation. Additionally, necrotic vessels and large necrotic tumor foci were found. In large areas of the specimens, a network of reticulin fibers was present between tumor cells. An oligodendroglial-like cell component was found only in one LTS GBM sample. Pleomorphic giant multinucleated tumor cells were only occasionally detected. By immunohistochemical analysis the expression of GFAP was detected in all cases in a variable degree.
Fig. 2

Photomicrograph showing staining of fraction of tumor cells from a LTS GBM patient for hematoxylin-eosin (A) and glial fibrillary acidic protein, GFAP (B). Necrosis, microvascular proliferation and mitoses are evidenced. Original magnification is (×100 and ×200 for HE and GFAP, respectively

Methylation analysis of TMS1/ASC promoter

Methylation of the TMS1/ASC promoter was observed in 12 out of the 57 studied GBM (overall rate: 21.05%). TMS1/ASC hypermethylation was more frequently detected in LTS GBM (4 of 7, 57.1%), than in classic GBM (8 of 50 cases, 16%, P = 0.0297, two-tailed Fisher’s exact test). The distribution of methylation of TMS1/ASC is described in Table 1. Illustrative examples of the methylation analysis of TMS1/ASC in both classic- and LTS GBM are shown in Fig. 3. Significant correlations between TMS1/ASC promoter methylation and age, sex or treatment status were not evidenced (all P > 0.05). Concerning outcome, no statistically significant association was found between methylation of TMS1/ASC and overall survival of all 57 analyzed patients (P = 0.21, Kaplan–Meier analysis, log-rank test, Fig. 4). Applying multivariate analysis of survival time and TMS1/ASC promoter methylation status (Cox proportional hazard model) no significant differences were observed between both GBM subsets.
Table 1

Characteristics of classic- and long-term survival glioblastoma patients in relation to the methylation status of the TMS1/ASC promoter


TMS1/ASC methylated

TMS1/ASC unmethylated


Classic GBM


Classic GBM

Age (years)


3 (42.86%)

2 (4%)

1 (14.28%)

9 (18%)



3 (6%)

2 (28.57%)

15 (30%)


1 (14.28%)

3 (6%)


18 (36%)



2 (28.57%)

3 (6%)

2 (28.57%)

20 (40%)


2 (28.57%)

5 (10%)

1 (14.28%)

22 (44%)
Fig. 3

Up: Genomic map of TMS1/ASC. CpG sites are indicated as vertical bars. Exons are represented as open boxes. Down: analysis of TMS1/ASC CpG island promoter methylation status in GBMs by the methylation-specific PCR assay. Molecular weight markers are shown in the left. The presence of a visible PCR product in those lanes marked “U” indicates the presence of unmethylated genes; the presence of product in those lanes marked “M” indicates the presence of methylated genes. Water (H2O) was used as negative PCR control. GBM1-4: classic glioblastoma samples. GBM5-7: long-term survivor GBM samples. TMS1 is methylated in tumors GBM4, GBM5 and GBM6
Fig. 4

Kaplan–Meier analysis of survival distribution for patients with methylated and unmethylated TMS1/ASC. GBM patients with methylated TMS1/ASC showed a better outcome but it did not reach significance (P = 0.21, log-rank test)

Methylation status of DAPK, TIMP2 and TIMP3 promoters

DAPK was hypermethylated only in one LTS GBM (14.3%) which also showed methylation of TMS1/ASC, whereas all classic GBM showed unmethylated DAPK. All tumors in both GBM subsets were observed to harbor unmethylated TIMP2 and TIPM3 promoters.


Since GBM still represents a clinical challenge with a dismal outcome, many efforts were made to identify biomarkers with prognostic significance. To date, only few genetic parameters were observed in GBM patients with prolonged survival. Nevertheless, the long-term survival phenomenon in GBM provides a unique opportunity to get insights into the molecular characteristics underlying such favorable prognosis. We previously observed MGMT to have such impact in LTS GBM (Martinez et al., submitted). To further identify epigenetic features distinguishing the LTS subset we aimed to characterize the methylation status of key regulator genes concerned in GBM invasion (TIMP2, TIMP3), apoptosis and immune-inflammatory response (TMS1/ASC, DAPK).

TMS1/ASC is localized to chromosome 16p11.2–12.1 and encodes a bipartite adaptor molecule containing an N-terminal PYRIN domain and a C-terminal CARD (caspase recruitment domain) and mediates intracellular signaling from apoptotic stimuli. Overexpression of TMS1/ASC promotes the initiation phase of a p53-independent apoptotic pathway by coupling death receptors at the cell surface and activating caspase-1 [19]. Recently, TMS1/ASC (through the PYRIN domain) was observed to harbor a links to inflammatory pathways since it activates NF-κβ family members, which are transcription factors implicated in both initiation and resolution of the inflammatory response [28]. Epigenetic silencing of TMS1/ASC might be advantageous for cancer cells due to the loss of ability to induce apoptosis and to activate the NF-κβ family transcription factors that control expression of proteins involved in cancer recognition by cytolitic T cells [2931].

In our sets of GBM patients, we have found that methylation of TMS1/ASC gene promoter was present in 57.1% of LTS GBM, distinguishing this collective from classic GBM, which showed such characteristic in only 16% of the cases (P = 0.0297, two-tailed Fisher’s exact test). The overall methylation rate of this gene in the whole GBM series was 21.05%, a percentage which is in the same range as described in ovarian cancer [20] but lower than in breast cancer [19]. As far as GBM, the series of Stone et al. [13] is the only previous investigation analyzing 23 GBM with a methylation rate of 43%. It would be very interesting to know whether LTS patients were also included in that study.

We have previously defined the methylation status of MGMT in the patients of the present series as a part of a previous analysis (Martinez et al., submitted). Strikingly, we observed in all LTS GBM with methylated TMS1 the simultaneous hypermethylation of MGMT, whereas it did not occur in the classic GBM subset (only 3 classic GBM harbored methylation of both TMS1/ASC and MGMT, P = 0.080, two-tailed Fisher’s exact test). These results might further support the hypothesis of different epigenetic signatures of both subsets, as previously observed for MGMT alone.

CpG island methylation of the DAPK promoter occurred in 14.3% of the LTS tumors, whereas an unmethylated status characterized classic GBM. TIMP2 and TIMP3 were found to be unmethylated in both GBM subsets. The observed rate of DAPK methylation contrast with those previously reported (between 10–29%) in non-small cell lung cancers and neoplastic gastric epithelia [23, 32] and widely differ from that observed (up to 90%) in brain metastasis [25]. These data may suggest that DAPK promoter hypermethylation is a tumor-specific epigenetic event. Concerning TIMP, our results are in accordance with those from Uhlmann et al. [16] who did not observe hypermethylation of TIMP3 in astrocytomas WHO grades II–IV, but diverge from those from Bachmann et al. [33] reporting TIMP3 to be hypermethylated in brain tumors, although the type of brain tumors investigated was not further specified.

Taken together, our data underscores the increasing interest that tight methylation-mediated silencing of key genes with glioma subtypes and pinpoints that LTS GBM, rather than an artificial statistical entity, probably harbor a distinct (epi-) genetic profile compared to classic GBM. The identification of further predictive biomarkers for LTS GBM may provide a platform for tailored therapy to individual patients on the basis of molecular phenotypes. We believe that assessment of the methylation status of well established genes such as MGMT and also perhaps TMS1/ASC in GBM might provide new molecular tools for the clinical management of these patients.

Copyright information

© Springer Science+Business Media, LLC 2006