Abstract
We investigated two polymorphisms of the epidermal growth factor receptor promoter as potential risk factors and prognostic markers for glioblastoma. The −216T allele (which results in a 30% higher activity) was more frequent in the patients compared with the control population (224/376 = 59.6% vs 165/352 = 46.8%; p = 0.0006) corresponding to an odd ratio of 1.67 (1.24; 2.25). A modest difference in median survival was also associated with the TT genotype.
Glioblastoma (GBM) is the most frequent and most malignant glioma, with a median survival of 12 to 15 months.1,2 Most GBMs are sporadic, and little is known about the environmental and genetic risk factors. However, critical steps of GBM development have been identified over the two last decades, involving genetic alterations such as 10q loss, p53 mutation, p16 deletion, PTEN mutation, EGFR amplification, and rearrangement of the extracellular domain.3,4 In this setting, functional polymorphisms involving these genes may represent risk factors or prognostic factors.5
Recently, a common polymorphism in the epidermal growth factor receptor (EGFR) promoter region was associated with altered promoter activity and EGFR expression.6 Indeed, the replacement of G by T at position −216 resulted in a higher binding efficiency of the Sp1 transcription factor, leading to 30% higher promoter activity.
In this study, we analyzed the role of the −216G/T and −191C/A EGFR polymorphism as a prognostic factor and a risk factor for GBM in a large case-control study.
Methods.
Patient selection.
Inclusion criteria were histologic diagnosis of GBM (World Health Organization grade IV) without previous history of glial tumor, age ≥18, clinical data and follow-up available on the database of the Salpêtrière Hospital established prospectively since 1997, blood DNA available, and written informed consent obtained.
Control DNA arose from unrelated healthy Caucasian volunteers.
Polymorphism identification.
Polymorphisms at −216 and −191 positions were identified by nested PCR followed by direct bidirectional sequencing, using the following primers: EGFRF1: 5′tcgcatctcctcctcctc3′, EGFRR1: 5′gacacgcccttaccttcttt3′, EGFRF2: 5′ctcctcctcctctgctcctc3′, EGFRR2: 5′gctctcccgatcaatactgg3′. The first PCR (EGFRF1/EGFRR1), which amplified a fragment of 446 pb, was carried out as follows in 5% dimethyl sulfoxide, 50 mM KCl, and 30 mM Tris-HCl for 5 minutes at 95 °C followed by 30 seconds at 95 °C, 30 seconds at 63 °C, and 1 minute at 72 °C for 30 cycles and 10 minutes at 72 °C. The PCR product was diluted (1:50,000) in the same buffer for the nested PCR (EGFRF2/EGFRR2), which was performed as follows: 5 minutes at 95 °C, followed by 30 seconds at 95 °C, 30 seconds at 61 °C, and 1 minute at 72 °C for 30 cycles and 10 minutes at 72 °C. The 326-pb fragment was purified and sequenced bidirectionally.
EGFR amplification of tumor DNA was detected by real-time PCR with an internal probe as previously described.7
Statistical analysis.
The independence of alleles (Hardy–Weinberg equilibrium) was ensured using the χ2 test at 1 df for each polymorphism. The χ2 test (or Fisher exact test when one subgroup was <5) was used to compare the genotype distribution between the two groups (GBM and controls), and the odds ratio was calculated. Overall survival was defined as the time between the diagnosis and death or the last follow-up. The survival curves were obtained by Kaplan–Meier methods and compared by log-rank test.
Results.
We analyzed DNA from 188 patients with GBM and 176 control subjects. The frequencies of the EGFR variants for both the control and the GBM population are reported in figure 1. In the control population, the proportion of the different genotypes met the Hardy–Weinberg equilibrium at −216 (p = 0.1) and at −191 locus (p = 0.13). The GBM population met the Hardy–Weinberg equilibrium at the −216 locus (p = 0.09) but not at the −191 locus (p = 0.0037). We then compared the genotype distribution between the control and the GBM population: At the −216 locus, the TT genotype tended to be overrepresented and the GG genotype underrepresented in the GBM population (p = 0.055). At the −191 locus, the overrepresentation of AA genotype and the underrepresentation of AC genotype in the GBM group were significant (p = 0.002). We then compared the number of alleles in both populations. As shown in figure 1, the number of T alleles was higher in the GBM population vs the controls (224/376 = 59.6% vs 165/352 = 46.8%; p = 0.0006). The odds ratio was 1.67; intraclass coefficient = (1.24; 2.25).
Figure 1. Comparison of (A) genotypes and (B) allele frequencies in the glioblastoma and control groups, showing (A) an overrepresentation of AA genotype and the underrepresentation of AC (p = 0.002), a trend for overrepresentation of TT genotype and underrepresentation of GG in the GBM population (p = 0.055); (B) T allele frequency is much higher in the GBM population compared to the controls (224/376 = 59.6% vs 165/352 = 46.8%; p = 0.0006).
The frequencies of the haplotypes deduced from figure 2 were as follows for the GBM group: GC = 125/344 (36.3%), TC = 205/344 (59.6%), GA = 11/344 (3.2%), and TA = 3/344 (0.9%); these differed from the control group: GC = 150/316 (47.5%), TC = 144/316 (45.5%), GA = 19/316 (6.0%), and TA = 3/316 (1.0%) (p = 0.002).
Figure 2. Link between −216G/T and −191C/A polymorphism in GBM and control patients. The frequencies of the haplotypes (B) are derived from the upper table (A) as follows: number of GC alleles= 2(GG/CC) + (GG/AC) + (GT/CC) = 2 × 29 + 4 + 63 = 125; number of AT alleles = 2(AA/TT) + (AA/GT) + (AC/TT)= 0 + 1 + 2 =3, etc.
One hundred eighty-four tumor samples were available for EGFR analysis. EGFR amplification was present in 56 of 184 (30.4%). There was no correlation between EGFR amplification status in tumor DNA and allelic status on both the −216 and the −191 locus, being present in 10 of 35 GG (28%), 26 of 78 GT (33%), 20 of 71 TT (28%), and 1 of 3 AA (33%), 6 of 22 AC (27%), and 49 of 159 CC (31%).
We analyzed the impact of −216T allele on survival according to different hypotheses: recessive (TT vs GG + GT), dominant (TT + GT vs GG), or intermediate effect (TT vs GT vs GG). The same analysis was conducted for −191 locus. Only a small difference was noted when comparing the TT with GG + GT population (figure 3, A; median survival = 18.3 vs 14.4 months; p < 0.05; log-rank test). EGFR amplification in tumor DNA was not a prognostic factor in this series (figure 3, B).
Figure 3. (A) Overall survival according to genotypes TT and GG + GT (p < 0.05). (B) Overall survival according to the presence of EGFR gene amplification in the tumor (p = NS).
Discussion.
Because the EGFR pathway is a critical pathway for glioma tumorigenesis, functional polymorphisms involving EGFR may affect glioma development. In this case-control study, −216T allele, which has a higher activity for the Sp1 transcription factor, resulting in a 40% higher EGFR gene transcription,6 is clearly more frequent in the GBM population vs the control population. Both populations were of Caucasian origin, and this difference cannot be explained by ethnic variability. Moreover, such a high frequency of T allele has not been reported in any ethnic group, T allele representing 34% in Caucasian populations vs 30% in African and 8.7% in Asian populations.6 Among ethnic groups, Caucasian populations showed the highest haplotype diversity, and our results (figure 2) are in accordance with previous data based on a sample of 22 Caucasians.6 In this perspective, it is also of interest that Asian populations, which have the lowest −161T allele frequency,6 also have a lower incidence of glioma compared with the United States and Europe, as recently shown by a Korean multicenter study.8
The −191C/A did not meet the Hardy–Weinberg equilibrium in the GBM population, with a relative overrepresentation of AA homozygotes and underrepresentation of AC heterozygotes vs the control population. However, as no functional analysis of the −191C/A has been performed so far, the significance of this result needs confirmation.
We also investigated the prognostic value of these polymorphisms. Only the comparison of TT homozygotes with GG + GT GBM population showed a mild favorable effect on survival, which cannot be retained because of the multiplicity of tests, unless confirmed by an independent study. It is also important to note that EGFR amplification is not a prognostic factor in GBM in this study. These data need to be confirmed and validated by EGFR mRNA-level analysis.
Footnotes
-
This article was previously published in electronic format as an Expedited E-Pub on August 2, 2006, at www.neurology.org.
Supported by the Délégation à la Recherche Clinique (AP-HP; grant no. MUL 03012) and the Ligue Nationale contre le Cancer, comité d’Ille et Vilaine.
Disclosure: The authors report no conflicts of interest.
Received November 30, 2005. Accepted in final form April 24, 2006.
Disputes & Debates: Rapid online correspondence
NOTE: All authors' disclosures must be entered and current in our database before comments can be posted. Enter and update disclosures at http://submit.neurology.org. Exception: replies to comments concerning an article you originally authored do not require updated disclosures.
- Stay timely. Submit only on articles published within the last 8 weeks.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- 200 words maximum.
- 5 references maximum. Reference 1 must be the article on which you are commenting.
- 5 authors maximum. Exception: replies can include all original authors of the article.
- Submitted comments are subject to editing and editor review prior to posting.