Microsatellite Instability in Pediatric and Adult High‐grade Gliomas

Abstract

About 15% of sporadic gastrointestinal and endometrial tumors show the microsatellite instability (MSI) phenotype because of loss of DNA mismatch repair (MMR) function. The incidence of MSI in tumors of the central nervous system still remains controversial. Previous studies reported a particular high frequency of MSI (∼25%) in young patients suffering from high‐grade gliomas. Based on these data and the fact that in different tumor entities MMR deficiency defines a subgroup of tumors with distinct pathogenesis and particular clinicopathological features that may have impact on prognosis and therapy, we screened 624 gliomas from 71 young and 553 adult patients for MMR deficiency by MSI analysis using three highly sensitive diagnostic markers. Alterations of MMR protein expression was examined by immunohistochemistry. A malignant glioma from an adult patient displayed MSI and concomitant loss of nuclear MSH2 and MSH6 protein expression (0.16%; 1/619). No evidence for MSI or loss of MMR protein expression was observed in 71 gliomas from young patients (0%; 0/71) including 41 high‐grade astrocytic tumors. Overall, we observed a much lower incidence of MSI among high‐grade pediatric gliomas than initially reported and suggest that MMR deficiency does not play a major role in the pathogenesis of glial neoplasms.

INTRODUCTION

Genetic instability is a salient feature of most sporadic and hereditary tumors. One particular type of genetic instability termed microsatellite instability (MSI) predominantly affects short repetitive DNA sequences (microsatellites). MSI is a molecular feature of tumors arising in patients with the hereditary Lynch syndrome but also occurring in sporadic colorectal and extracolorectal tumors (1, 18, 33). The MSI phenotype is caused by loss of DNA Mismatch Repair (MMR) function either by germline and somatic mutation in one of the four human MMR genes MSH2, MLH1, MSH6, PMS2 in the hereditary form, or by promoter methylation of MLH1 in the sporadic form (16, 34, 39). Because of MMR‐deficiency, MSI tumors accumulate numerous insertion/deletion mutations not only at non‐coding but also at coding microsatellites thereby leading to translational frameshifts and impaired function of affected proteins. Frameshift mutations in coding repeats of a large number of candidate genes have been identified (9, 21, 26, 36) and some of them (TGFßRII, ACVRII, Bax) were shown to contribute to MSI carcinogenesis (12, 17, 19).

The MSI phenotype itself appears to be closely linked to specific clinico‐histopathological features which have been most extensively studied in colorectal cancers. MSI tumors are often poorly differentiated with a mixed histology containing mucinous and signet‐ring cell areas (13), present with a dense intratumoral lymphocyte infiltration (32), show less distant metastases (5), and appear to have a comparably good prognosis (14, 30). Recent studies indicate that benefit from adjuvant 5‐FU chemotherapy appears to be restricted to non‐MSI cancer patients (7, 27). It is generally believed that coding microsatellite frameshift mutations in MSI target genes and the immunogenicity of frameshift neopeptides derived thereof (23, 28, 29, 31) most likely account for these distinct clinico‐histopathological features of MSI tumors.

For molecular identification of such MSI tumors an international reference microsatellite marker panel has been proposed that consists of two mono‐ and three dinucleotide microsatellites (4). However, increasing evidence suggests that marker panels relying exclusively on mononucleotide repeats appear to be more sensitive for the detection of MMR deficiency and hence have been recommended for MSI classification of colorectal and probably also extracolonic tumors (6, 11). In addition to MSI testing, immunohistochemistry (IHC) for the detection of loss of MMR protein expression has been used to detect sporadic and hereditary MMR deficient tumors. A stepwise application of IHC and MSI analysis for identification of hereditary non‐polyposis colorectal cancer patients has been proposed to maintain maximum sensitivity and specificity at lower costs (10).

MSI gliomas appear to be present in a small subset of MMR‐deficient extracolonic malignancies. In particular, MSI gliomas can arise in a subset of patients with the rare hereditary Turcot’s syndrome. MMR germline mutations in a fraction of these Turcot patients predispose affected individuals to the development of concurrent MSI colorectal tumors and glioblastomas (15, 20, 22). The MSI phenotype has also been detected in sporadic gliomas of young and adult patients although the reported MSI frequencies vary significantly. A survey of the current literature suggests that in adult astrocytic and non‐astrocytic CNS tumors MSI is a rare event with an incidence of 0%–8% (3, 8, 24, 25, 38). However, a high frequency of MSI appeared to be associated with a particular subset of CNS tumors comprising WHO grade III and grade IV pediatric astrocytomas and gangliogliomas (3, 8, 20). So far, no statistically significant correlations between MSI phenotype and gene expression pattern (p53 mutations, EGFR, MDM2, p16) or clinicohistopathological parameters have been identified.

Our goal in the current study was to determine the prevalence of MSI in pediatric high‐grade brain tumors as a basis for subsequent molecular studies.

MATERIALS AND METHODS

Tumor samples.

Overall, 624 glioma samples from 553 individual adult and 71 individual young tumor patients were recruited. Based on the reported high incidence of MSI in pediatric high‐grade gliomas, age at onset of disease (pediatric ≤22 years; adult >22 years), tumor grading (WHO I–IV) and histology (astrocytoma; ganglioglioma) were used as selection criteria. Different types of samples were used for MSI analysis: (i) frozen cell pellets from short‐term primary cultures of brain tumors (n = 33), (ii) genomic DNA from tissue samples (n = 38), (iii) hematoxylin/eosine stained tissue sections (6 µm) from paraffin‐embedded tumors (n = 35), and (iv) paraffin embedded tissue blocks (n = 518). Samples were provided by the following institutions: Institute of Neuropathology, Bonn (n = 388; 261 glioblastomas, 81 astrocytomas grade II–III, 41 ganglioglimas grade I–III, 5 oligodendrogliomas); Division of Neurosurgical Research (n = 179; 140 glioblastomas, 18 astrocytomas grade II–III, 1 ganglioglioma grade III, 9 oligodendrogliomas, 11 mixed gliomas) and Institute of Neuropathology, Heidelberg (n = 8, glioblastoma), Institute of Neuropathology, Zurich (n = 12, glioblastoma), Department of Neurosurgery, Cologne (n = 35;19 glioblastoma, 6 astrocytomas grade III, 3 gangliogliomas grade III, 1 oligodendroglioma grade III, 6 mixed astrocytomas grade III), Institute of Human Genetics, Munich (n = 2; glioblastoma). Informed consent was obtained from all patients included in this study.

MSI analysis and multiplex PCR.

HE‐stained sections from paraffin‐embedded tissues were microdissected in order to enrich for tumor cells (∼90%). Genomic DNA from microdissected tumor samples and cell pellets was isolated using the DNeasy Tissue Kit (Qiagen, Hilden, Germany). MSI status could be determined in 619/624 glioma samples (>99%). For increased screening efficiency, a multiplex polymerase chain reaction (PCR) system was used consisting of three quasimonomorphic mononucleotide microsatellite markers previously described to detect MSI tumors with high sensitivity (11). The following primers were used: BAT25 (sense: 5′‐TCGCCTCCAAGAATGTAAGT‐3′; antisense: 5′‐TATGGCTCTAAAATGCT CTGTTC‐3′; BAT26 (sense: 5′‐TGACT ACTTTTGACTTCAGCC‐3′; antisense: 5′‐AACCATTCAACATTTTTAACCC‐3′), CAT25 (sense: 5′‐CCTAGAAACC TTTATCCCTGCTT‐3′; antisense: 5′‐GAGCTTGCAGTGAGCTGAGA‐3′). PCR products were separated on an ABI Prism 3100 Genetic Analyzer (Applied Biosystems, Darmstadt, Germany) and analyzed with Genescan Sortware (Applied Biosystems). Marker instability was assessed as described (11). High level of MSI (MSI‐H) was scored if at least 2/3 tested markers displayed instability.

Immunohistochemistry.

In addition to MSI analysis, MMR protein expression was examined on 2 µm paraffin‐embedded tissue sections from 236 gliomas (61 young and 175 adult patients) using mouse monoclonal antibodies against MLH1 (G168‐15, BD Pharmingen, Heidelberg, Germany), MSH2 (clone FE11, Calbiochem, Darmstadt, Germany), MSH6 (clone 44, Pharmingen) and PMS2 (clone A16‐4, Pharmingen). Interpretable results were obtained in 223/236 cases (95%). Positive staining of intratumoral endothelial cells served as normal control.

RESULTS AND DISCUSSION

Previously, we identified coding microsatellite frameshift mutations in candidate MSI target genes predicted to be causally implicated in gastrointestinal and endometrial MSI tumorigenesis (2, 35, 36, 37). In the present study we aimed to extend this analysis to human gliomas as systematic coding repeat mutation analysis of candidate MSI target genes in these extracolonic malignancies has not been performed to date. According to literature data a high frequency of high level MSI (MSI‐H) appears to be associated with high‐grade astrocytic and non‐astrocytic tumors in young patients (8, 20, 22). In particular, Alonso et al reported a high frequency of MSI (27%, 12/45) in pediatric high‐grade astrocytomas (3) and concluded that development of pediatric high‐grade astrocytomas may proceed via both MMR‐dependent or MMR‐independent pathways.

Therefore, we determined the MSI status of 77 non‐astrocytic and 547 astrocytic tumors (442 glioblastomas, 105 astrocytomas) (Table 1) including 71 high‐grade pediatric gliomas. Originally, a reference microsatellite marker panel has been recommended for routine MSI testing that includes two mononucleotide (BAT25, BAT26) and three dinucleotide repeats (D2S123, D5S346, D17S250) (4). However, mononucleotide markers seem to provide a higher sensitivity for MSI detection than dinucleotides. Moreover, the quasi‐monomorphic nature of mononucleotide repeats would enable MSI analysis of tumor tissue even in the absence of corresponding normal tissue which applies to most of the glioma samples available for the present study (6). Hence, we chose a panel of three mononucleotide repeat markers (BAT25, BAT26, CAT25) which we had successfully applied for the detection of the MSI phenotype in gastrointestinal tumors (11). Using this triplex PCR system, all 71 gliomas derived from young patients (age at diagnosis ≤22 years)—including 41 high‐grade astrocytic tumors—were classified as microsatellite stable (MSS). Likewise, immunohistochemical staining revealed no evidence for loss of MMR protein expression in this subset. Subsequent MSI screening of the remaining adult tumor specimens (age at diagnosis >22 years) revealed interpretable results in 548/553 cases and identified a single glioblastoma in a 74‐year‐old male patient that displayed instability in all three markers and hence was classified as MSI‐H (Figure 1A). The MSI phenotype in this glioblastoma was confirmed by analyzing additional mononucleotide (BAT40) and dinucleotide markers (D2S123, D5S346, Mfd15) (data not shown). By means of IHC, the glioblastoma cells of this patient showed loss of nuclear expression of MSH2 and MSH6 proteins but retained expression of MLH1 and PMS2 proteins when compared with the control staining pattern of intratumoral endothelial cells (Figure 1B). These IHC results verified MMR deficiency in the tumor and correlated with the MSI phenotype.

Table 1.

Analyzed samples stratified for tumor type and patients’ age.

Tumor type (WHO grade)	Individual patient samples	Patients ≤ 22 years	Patients > 22 years
Astrocytic tumors
Astrocytoma (II)	70	4	66
Anaplastic astrocytoma (III)	35	6	29
Glioblastoma multiforme (IV)	442	35	407
Oligodendroglial tumors
Oligodendroglioma (II)	5	0	5
Anaplastic oligodendroglioma (III)	10	1	9
Mixed gliomas
Oligoastrocytoma (II)	7	0	7
Anaplastic oligoastrocytoma (III)	10	2	8
Mixed neuronal‐glial tumors
Ganglioglioma (I)	32	16	16
Ganglioglioma (II)	7	3	4
Anaplastic ganglioglioma (III)	6	4	2
Total	624	71	553

Figure 1.

MSI analysis and MMR protein immunohistochemistry. A. MSI Analysis. The allele pattern of three mononucleotide markers (BAT25, BAT26, CAT25 in normal and tumor tissue are shown. Deleted nucleotides are depicted by arrows and indicate microsatellite instability in all three markers in the tumor tissue compared with the control tissue. B. Immunohistochemical staining of the MSI tumor. MMR proteins MLH1 and PMS2 are expressed regularly in tumor cell nuclei (brown) whereas loss of expression of MSH2 and MSH6 is indicated by absence of staining in tumor cells (blue nuclei). Endothelial cells express normal levels of these two proteins and served as internal controls.

Our results suggest that the incidence of MSI in pediatric and adult gliomas is extremely low (0.16%; 1/619). This finding differs significantly from the high MSI frequency of these tumors in young patients reported by Alonso et al (24%–27%), Cheng et al (22%; 2/9), and Leung et al (18%; 4/22). The four MSI gliomas described by Leung et al occurred in patients harboring MMR germline mutations, including three patients that developed metachronous colorectal carcinomas and thus matching the criteria of Turcot’s syndrome. However, we cannot exclude a hereditary basis for the single MSI glioblastoma that displayed loss of MSH6 and MSH2 protein expression in the present study.

The discrepancy in MSI frequency between our data and those reported by Alonso et al and Cheng et al are more difficult to reconcile. Similar to our approach, these investigators also used at least two highly sensitive mononucleotide markers as well as stringent criteria for MSI classification. Finally, microdissection of a large fraction of the tumors allowed tumor cell enrichment and excluded sampling bias. Our data on adult glioma patients are well in agreement with a recent study that did not find MSI‐H in 129 sporadic adult glioblastomas (24).

Overall, we present the largest study on MSI classification in adult and pediatric human gliomas. Our results strongly argue in favor of a much lower prevalence of MSI among high‐grade pediatric gliomas than initially reported by Alonso et al and suggest that MMR deficiency does not play a major role in the molecular pathogenesis of glial neoplasms.