Risk and Prognostic Factors for Malignant...
Transcript of Risk and Prognostic Factors for Malignant...
Risk and Prognostic Factors for Malignant Glioma
Sara Sjöström
Department of Radiation Sciences, Oncology
Umeå University 2012
Responsible publisher under Swedish law: the Dean of the Medical Faculty
This work is protected by the Swedish Copyright Legislation (Act 1960:729)
Copyright © Sara Sjöström
New Series No. 1536
ISBN: 978-91-7459-521-5
ISSN: 0346-6612
Front cover: The horizon seen from Lövskatan, Nordmaling, Sweden
Elektronisk version tillgänglig på http://umu.diva-portal.org/
Printed by: Print & Media, Umeå University
Umeå, Sweden 2012
To my family
i
Table of Contents
Table of Contents i Abstract iii Abbreviations v Populärvetenskaplig sammanfattning på svenska ix Original papers xii Introduction 1
Glioma 1 Classification 1 Incidence 2 Diagnostics and treatment 3
Risk factors 4 Mobile phone 4 Genetic syndromes and familial aggregation 4 Ionizing radiation 5 Asthma and allergy 5 Low penetrance genes 8
Prognostic factors 10 Virus and immunologic response 11
Viruses 11 Herpesviridae 11
Cytomegalovirus 11 Epstein-Barr virus 11 Varicella zoster virus 12
Adenoviridae 12 Immunologic response 12 Viruses and Cancer 13
EGF/EGFR 13 Function 13 EGF/EGFR and glioma 14
VEGF/VEGFR 16 Function 16 VEGF/VEGFR and glioma 16
DNA repair 18 Mechanisms 18 Base excision repair 18 Mismatch repair 18 Nucleotide Excision Repair 19 Homologous recombination and non-homologous end joining 20 Direct repair 20 DNA-repair gene variants and glioma 21
ii
Genetics and association studies 22 SNPs and association studies 22 Haplotype and haplotype blocks 23
Aims 24 Material 25
Study population 25 Paper I 25
Northern Sweden Health and Disease Study 26 The Malmö Diet and Cancer Study 26 Diet, Cancer and Health 27
Papers II-IV 27 The INTERPHONE study 27
Confirmation 28 Papers II and III 28 Paper IV 28
Methods 29 Paper I 29 ELISA and immunoflourescence 29 Binary logistic regression 29 Papers II-IV 30 Medical records 30 SNP-selection and genotyping 31 Statistical analyses 31
Cox proportional hazards model 31 Correction for multiple testing 32
Results 33 Paper I 33 Papers II and III 34 Paper IV 35
Discussion 37 Paper I 37 Case-control studies 38 Papers II and III 39 Paper IV 41
Conclusions 44 Future perspectives 45 Acknowledgements 47 References 50 Original publications
Paper I
Paper II
Paper III
Paper IV
iii
Abstract
Background: Glioblastoma is the most common and aggressive type of
glioma and associated with poor prognosis. Apart from ionizing radiation
and some rare genetic disorders, few aetiological factors have been identified
for primary brain tumours. Inverse associations to asthma and low IgG levels
for varicella zoster virus have in previous studies indicated that the immune
system may play a role in glioma development. Little is known about
prognostic factors in glioma. Previous studies have shown an association
between age, Karnofsky performance status, O-6-methylguanine-DNA
methyltransferase (MGMT) hypermethylation, and prognosis.
Polymorphisms in different low penetrance genes, such as epidermal growth
factor (EGF), have in some studies been associated with glioma prognosis
The epidermal growth factor receptor, EGFR, is amplified in about 30-50%
of malignant glioma and in 50% of glioblastoma EGFR is mutated
(EGFRvIII), leading to a constitutively active receptor. EGFR overexpression
has been associated with poor prognosis in low grade glioma and anaplastic
astrocytoma. Vascular endothelial growth factor (VEGF) is a key component
in both angiogenesis and the development of malignant tumours. VEGF
overexpression has been found and associated with poorer prognosis in high
grade astrocytoma. Radiotherapy and chemotherapy, is apart from surgery,
the standard treatment for glioblastoma and both these treatments cause
damage to the DNA and therefore DNA repair genes could be hypothesized
to be of prognostic importance.
Material and methods: In paper I we analysed IgG levels for four
different viruses, Epstein-Barr virus (EBV), cytomegalovirus (CMV),
varicella zoster virus (VZV) and adenovirus (Ad), in prediagnostic blood
samples from 197 cases with glioma and 394 controls. The blood samples
were collected from three large cohorts: the Northern Sweden Health and
Disease Study, the Malmö Diet and Cancer Study and the Diet, Cancer and
Health cohort from Copenhagen. IgG levels were measured using Enzyme-
linked immunosorbent assay (ELISA). For EBV, IgG response to both the
nuclear antigen (EBNA1) and the viral capsid antigen (VCA) was measured,
while for VCA, immunoflourescence was used. IgG levels were divided into
quartiles and binary logistic regression was used to compare the quartiles in
cases and controls. All odds ratios were adjusted for age, sex, and cohort. In
papers II-IV, we studied 176 glioblastoma cases from Sweden and Denmark,
originally collected as part of the INTERPHONE study, a large international
study that investigated the association between mobile phone use and glioma
risk. We collected treatment and follow-up data on the cases. Loss of follow-
up, missing treatment data or poor DNA quality led to fewer cases available
iv
for final analyses in the different studies. We genotyped 30 tagging single
nucleotide polymorphisms (SNPs) in EGF, 89 in EGFR, 27 in VEGFR2 and
17 in VEGF. We also studied 1458 SNPs in 136 DNA-repair genes. Hazard
ratios were calculated using Cox regression; the major allele was set as
categorical variable and all HR were adjusted for age, sex, country, and
treatment. For the DNA repair gene results, we adjusted the p-values for
multiple testing. A separate dataset from The M. D. Anderson Cancer Center
(MDACC) was available for confirmation in the EGF/EGFR and
VEGF/VEGFR2 studies, and for the DNA-repair genes, a dataset from
northern UK was used for confirmation.
Results and Discussion: In paper I we found a trend towards higher IgG
VZV levels in controls compared to glioma cases, especially when restricting
the analyses to only include glioma cases with at least 2 years between blood
sample and diagnosis. This finding might indicate that there is an
aetiological and not a disease-related association. These findings confirm
previous findings in early studies of samples taken at diagnosis and support
that a strong immune system can detect and inhibit growth of small cancer
clusters. In EGF, we found seven SNPs in one haplotype block that were
significantly associated with glioblastoma survival. Four of the SNPs were
available for confirmation in the MDACC dataset; however, none of the SNPs
reached statistical significance. In analysing the data further, one
explanation could be that the cases in the MDACC cohort had a lower age at
diagnosis and longer median survival time than the Swedish/Danish cohort.
In EGFR, four SNPs associated with survival were found; however, as 89
polymorphisms were tested this was the expected outcome by chance. In
VEGF and VEGFR2, we found two SNPs associated with glioblastoma
survival, but they could not be confirmed in the MDACC dataset and, due to
multiple testing were considered to be false positives. Among the DNA-
repair genes, we found nine SNPs in three genes - MSH2, RAD51L1 and
RECQL4 - which were significantly associated with glioblastoma survival
after confirmation and adjustment for age, sex, country and treatment. After
adjusting for multiple testing, two SNPs remained significant, one in MSH2
and one in RECQL4.
Conclusions: Our studies provide additional knowledge to the aetiological
and prognostic factors important for glioma, emphasising the possible
importance of immune function mechanisms. We found limited evidence for
the role of genetic variants in glioma progression genes, and some for DNA
repair variants as prognostic factors for glioblastoma survival.
v
Abbreviations
5´UTR 5´Un-Translated Region
A Adenine
Ad Adenovirus
AP-site Apurinic/apyrimidic site
APE1 Apurinic/apyrimidic endonuclease 1
APEX1 APEX nuclease 1
ATM Ataxia telangiectasia mutated
ATP Adenosine triphosphate
ATR Ataxia telangiectasia and Rad3 related
BER Base excision repair
bp base pair
BRCA Breast cancer susceptibility gene
BTBD2 BTB domain containing 2
C Cytosine
CCDC26 Coiled-coil domain containing 26
CCSS Childhood cancer survival study
CD4+ Cluster of differentiation 4, T-helper cell
CD8+ Cluster of differentiation 8, Cytotoxic T Lymphocyte
CDKN2A Cyclin-dependent kinase inhibitor 2A (p16INK4A)
CDKN2B Cyclin-dependent kinase inhibitor 2B
CHAF1A Chromatin assembly factor 1, subunit A
CHEP Utah residents with northern and western Europe ancestry
CI Confidence interval
CMV Cytomegalovirus
CNS Central nervous system
CSA Cockayne syndrome A (ERCC8)
CSB Cockayne syndrome B (ERCC6)
CT Computed tomography
CTL Cytotoxic T lymphocyte
DBS Double strand break
DCLRE1B DNA cross-link repair 1B
DNA Deoxyribonucleic acid
DNA-pk DNA protein kinase
EBNA-1 IgG-anti-Epstein-Barr-Nuclear-Antigen-1
EBV Epstein-Barr virus
EGF Epidermal growth factor
EGFR Epidermal growth factor receptor
EGFRvIII Mutant epidermal growth factor receptor
ELISA Enzyme-linked immunosorbent assay
ErbB v-erb-b erytroblastic leukemia viral oncogene homolog
vi
ERCC1 Excision repair cross-complementing rodent repair deficiency
complementation group 1
EXO1 Exonuclease 1
FEN1 Flap structure-specific endonuclease 1
G Guanine
GGR Global genome repair
Gy Gray
GWAS Genome wide association study
HCMV Human cytomegalovirus
HHRAD23B Human homolog of RAD23 B
HIV Human immunodeficiency virus
HMGA2 High mobility group AT-hook 2
HNPCC Hereditary nonpolyposis colon cancer, Lynch syndrome
HPV Human papilomavirus
HR Homologous recombination
hTERC Human telomerase RNA component
hTERT Human telomerase reverse transcriptase
IDH Isocitrate dehydrogenase
IE1 Immediate-early 1
IF Immunoflourescence
Ifn Interferon
Ig Immunoglobulin
IL Interleukin
JCV John Cunningham virus, JC-virus
kB Kilo base
kU/L Kilo units per litre
LD Linkage disequilibrium
LIG Ligase
LTS Long-term survivors
MAF Minor allele frequency
MAPK Mitogen-activated protein kinase
MDACC MD Anderson Cancer Center
MDSC Malmö Diet and Cancer Study
MGMT O-6-methylguanine-DNA-methyltransferase
MLH MutL-homolog
MMR Mismatch repair
MONICA Monitoring Trends and Determinants in Cardiovascular
Disease
MRE11 Meiotic recombination 11
MRI Magnetic resonance imaging
MRN RAD50/MRE11/NBS1 complex
MSH Melanocyte-stimulating hormone
NBS1 Nijmegen breakage syndrome 1 (nibrin)
vii
NEIL3 Nei endonuclease VIII-like 3
NER Nucleotide excision repair
NF Neurofibromatosis
NHEJ Non-homologous end joining
NK Natural killer
nm Nanometre
No. Number
NSHDS Northern Sweden Health and Disease Study
OR Odds ratio
P3-kinase Phosphoinositide 3-kinase
PARP Poly ADP ribose polymerase
PCNA Proliferating cell nucleus antigen
PFS Progression-free survival
PHLDB1 Pleckstrin homology-like domain family B member 1
PKC Protein kinase C
PLC Phospholipase C
PLK Polo-like kinase
PMS Postmeiotic segregation increased
PNK Polynucleotide kinase
Pol Polymerase
POLD1 Polymerase delta 1
PTEN Phosphatase and tensin homolog
RB Retinoblastoma
RGS22 Regulator of G-protein signalling 22
RNA Ribonucleic acid
RPA Replication protein A
RR Relative risk
RTEL 1 Regulator of telomere elongation helicase 1
SES Socio economic status
SNOMED Systematized nomenclature of medicine
STS Short-term survivors
SV40 Simian virus 40
T Thymine
TCR Transcription coupled repair
TERT Telomerase reverse transcriptase
TFIIH Transcription factor II H
TH T-helper
TNF Tumour necrosis factor
TP53 Tumour protein 53
TSC Tuberous sclerosis
UCSC University of California, Santa Cruz
UK United Kingdom
UV Ultraviolet
viii
VCA Viral capsid antigen
VEGF Vascular endothelial growth factor
VEGFR Vascular endothelial growth factor receptor
VIP Västerbotten Intervention Program
VZV Varicella zoster virus
WHO World Health Organization
XPA Xeroderma pigmentosum complementation group A
XPC Xeroderma pigmentosum complementation group C
XPF Xeroderma pigmentosum complementation group F (ERCC4)
XPG Xeroderma pigmentosum complementation group G (ERCC5)
XRCC X-ray repair cross-complementing
ix
Populärvetenskaplig sammanfattning på svenska
Hjärnan är ett mycket komplext organ som består av flera olika typer av
celler, och tumörer i hjärnan kan därför ha olika ursprung vilket påverkar
växtsätt och aggressivitet. Den vanligaste typen av hjärntumör är gliom, en
typ av tumör som uppstår från gliaceller vilka är hjärnans stödjevävnad.
Gliom kan delas in i fyra grader där grad 4, glioblastom, är den mest
aggressiva formen med dålig prognos. Vi vet idag mycket lite om vad som
orsakar gliom, det finns familjer med ökad risk för att få ett gliom, vissa
ovanliga genetiska sjukdomar som t.ex. Li-Fraumeni och Lynch syndrom ger
en ökad risk för gliom och det finns beskrivet att joniserande strålning ökar
risken för att få en hjärntumör, däremot har inga starka samband mellan
mobiltelefonanvändning och risk att drabbas av hjärntumör hittats. Tidigare
studier har också visat att personer med astma och allergi samt personer
med höga antikroppsnivåer mot det herpesvirus som ger vattkoppor och
bältros (varicella zoster virus) har ett visst skydd mot att utveckla gliom. Idag
känner vi endast till ett fåtal faktorer som påverkar prognosen hos en person
som insjuknar i hjärntumör. Vi vet exempelvis att ålder och allmäntillstånd
vid insjuknande, hur mycket av tumören som går att operera bort och om
genen MGMT är aktiv har betydelse för prognosen. Patienter med en inaktiv
MGMT gen svarar bättre på behandling med Temozolomid, en typ av cellgift
som är en del av standardbehandlingen vid glioblastom.
Vår arvsmassa (DNA) är uppbyggd av 4 baser, adenin (A), tymin (T), cytosin
(C) och guanin (G). Dessa baser binds ihop parvis till två strängar och bildar
på så sätt en dubbelhelix som inrymmer alla våra gener. 99.9% av vår
arvsmassa är identisk, men 0.1 % skiljer sig mellan individer och består till
80 % av variationer i dessa baspar. Den här typen av variationer i olika
gener, som bland annat påverkar celltillväxt, kärltillväxt eller reparation av
DNA-skada, har visat sig ha betydelse för både uppkomst och prognos vid
olika typer av cancer, bl.a. gliom.
I arbete I har vi använt oss av blodprover från olika biobanker för att
undersöka om antikroppsnivåer mot fyra vanliga virus, cytomegalovirus
(CMV), Epstein-Barr virus (EBV), varicella zoster virus (VZV) och
adenovirus (Ad), är associerat till risk att utveckla gliom. Tidigare studier har
visat ett samband mellan höga antikroppsnivåer mot VZV och lägre risk för
gliom. I de studierna togs blodprov först när fallen insjuknat i gliom vilket
gör att både sjukdomen i sig och behandling kan ha påverkat resultatet.
Eftersom vi hade tillgång till prover från biobanker har vi kunnat använda
oss av blodprover som togs innan fallen insjuknade i gliom och jämföra med
friska kontroller. Vi kunde då fastställa samma samband mellan höga
x
antikroppsnivåer mot VZV och minskad risk för gliom. Däremot kunde vi
inte se någon association mellan risk för gliom och antikroppsnivåer för de
andra virusen.
I arbete II-IV har vi studerat 176 fall med glioblastoma från Sverige och
Danmark som insjuknade mellan åren 2000 och 2004 och då deltog i en stor
internationell studie om hjärntumörrisk vid mobiltelefonanvändning
(INTERPHONE). Hos dessa fall hade vi tillgång till behandlingsdata varför
alla våra resultat är kontrollerade för land, kön, ålder och behandling. En
sådan kontroll görs för att minimera risken att dessa faktorer påverkar
resultatet. Vårt syfte var att studera om genetiska variationer i gener som är
inblandade i celltillväxt (EGF, EGFR), kärltillväxt (VEGF, VEGFR2) och
reparation av DNA-skador, och därför viktiga i tumörutveckling, har någon
betydelse för överlevnad hos fall med glioblastom.
En genetisk variation är en variation i ett baspar på en given plats i
arvsmassan, dessa variationer finns naturligt i befolkningen. Om baserna på
den givna platsen i arvsmassan exempelvis är C och G så kan de vara
kombinerade på 3 olika sätt hos olika individer; CC, CG och GG. En av dessa
baser är vanligast förekommande i befolkningen på just den platsen i
arvsmassan och om vi i detta exempel antar att det är basen C så har vi
studerat om en person som har en av de ovanliga baserna, CG, eller två av de
ovanliga baserna, GG, lever kortare eller längre om de får ett glioblastom
jämfört med dem som har två av de vanliga baserna CC.
Vi analyserade 30 genetiska variationer i genen epidermal growth factor
(EGF), 89 i genen epidermal growth factor receptor (EGFR), 17 i genen
vascular endothelial growth factor (VEGF), 27 i genen vascular endothelial
growth factor receptor 2 (VEGFR2) och 1458 genetiska variationer i 136
olika DNA-reparationsgener. I EGF fann vi ett samband mellan 7 variationer
och överlevnad hos glioblastom. Fyra av dessa hade vi möjlighet att även
analysera i en annan grupp bestående av 638 glioblastom från MD Anderson
Cancer Centre (MDACC) i Texas, men där kunde vi inte se något samband
med överlevnad. En förklaring till detta kan vara att fallen från Texas var
yngre och hade en längre medianöverlevnad jämfört med de svenska och
danska fallen och vi hade inte heller tillgång till behandlingsdata för fallen
från Texas. I EGFR hittade vi ett samband mellan 4 genetiska variationer och
överlevnad, men eftersom vi testade 89 så beror detta mest sannolikt på
slumpen då vi använt oss av ett p-värde på 0.05 vilket ger en sannolikhet på
95 % att våra fynd är sanna. I VEGF hittade vi inget samband mellan de
genetiska variationer vi analyserade och överlevnad vid glioblastom. I
VEGFR2 hittade vi ett samband mellan 2 variationer och överlevnad. Dessa
hade vi också möjlighet att analysera i fallen från MDACC, vi fann dock inte
samma samband där och fyndet är mest sannolikt beroende på slumpen. I
det fjärde arbetet studerade vi sammanlagt 1458 genetiska variationer i 136
xi
DNA-reparationsgener. Vi analyserade först i de svenska och danska fallen
och sedan kontrollerade vi positiva fynd i 295 fall från norra England. De
variationer som fortfarande hade ett samband med överlevnad analyserade
vi sedan ånyo i det svenska och danska materialet men kontrollerade denna
gång för land, ålder, kön och behandling. Efter detta fann vi sammanlagt 9
variationer med ett samband till överlevnad, dessa var belägna i generna
MSH2, RAD51L1 och RECQL4.
Sammanfattningsvis så har vi bekräftat att immunförsvaret verkar spela en
viktig roll vid uppkomsten av gliom och vi fann vissa samband mellan
variationer i gener som är av betydelse för tumörprogress och reparation av
skador på DNA och överlevnad hos glioblastom. Genom att hitta faktorer
viktiga för både risk att utveckla hjärntumör och överlevnad hos patienter
med hjärntumör skapar man möjligheter att i framtiden hitta sätt att tidigare
diagnostisera tumören och möjliga angreppspunkter för nya behandlingar.
xii
Original papers
This thesis is based on the following studies, referred to in the text by their
roman numerals:
I. Human immunoglobulin G levels of viruses and associated
glioma risk.
Sjöström Sara, Hjalmars Ulf, Juto Per, Wadell Göran,
Hallmans Göran, Tjönneland Anne, Halkjaer Jytte, Manjer
Jonas, Almquist Martin, Melin Beatrice, Cancer Causes Control
September 2011 ;22(9): 1259-66
II. Genetic variations in EGF and EGFR and glioblastoma outcome.
Sjöström Sara, Andersson Ulrika, Liu Yanhong, Brännström
Thomas, Broholm Helle, Johansen Christoffer, Collatz-Laier
Helle, Henriksson Roger, Bondy Melissa, Melin Beatrice. Neuro
Oncol August 2010;12(8):815-21
III. Genetic variations in VEGF and VEGFR2 and glioblastoma
outcome.
Sjöström Sara, Wibom Carl, Andersson Ulrika, Brännström
Thomas, Broholm Helle, Johansen Christoffer, Collatz-Laier
Helle, Liu Yanhong, Bondy Melissa, Henriksson Roger, Melin
Beatrice. J Neurooncol September 2011 ; 104(2):523-7
IV. DNA-repair gene variants are associated with glioblastoma
survival.
Wibom Carl, Sjöström Sara, Henriksson Roger, Brännström
Thomas, Broholm Helle, Rydén Patrik, Johansen Christoffer,
Collatz-Laier Helle, Hepworth Sara, McKinney Patricia, Bethke
Lara, Houlston Richard, Andersson Ulrika, Melin Beatrice. Acta
Oncologica March 2012, Vol. 51, No. 3 , Pages 325-332
Reprints were made with permission from Elsevier (1-3) and Acta oncologica
(4).
1
Introduction
Glioma
There are several different types of brain tumours that arise from cells with
different origins and function in the brain. The most common type is glioma,
which arises from glial cells, a form of supportive cells in the brain and
spinal cord. There are several different types of glial cells, ependymal cells,
astrocytic cells, microglial cells, oligodendritic cells, Schwann cells and
müller cells.
Classification
Tumours arising from astrocytic cells, astrocytoma, are the most common
type of tumours and they are divided into four grades depending on
malignancy. Grade 1 can be characterized as clinically benign, often curable
by surgery. Grade 2 and 3 are malignant and they often progress to higher
and more malignant grades. The fourth grade, glioblastoma, is the most
aggressive form and has a very poor prognosis. Glioblastoma can either be
primary or secondary; the latter starts as a lower-grade that progress into a
higher-grade astrocytoma and finally into glioblastoma. Differences in
tumour progression and prognosis have been seen between primary and
secondary glioblastoma (Figure 1). Secondary glioblastoma more often has a
tumour protein 53 (TP53) mutation, while primary glioblastoma is more
likely to have epidermal growth factor receptor (EGFR) amplification (1).
Isocitrate dehydrogenase 1 (IDH1) mutations are more frequent in secondary
glioblastoma and have been associated with longer survival (2). Secondary
glioblastomas are rare and account for about 5%.
2
Incidence
The glioma incidence in Sweden in 2009 was 9 per 100,000 person-years
when looking at all regions and including all age groups. The standardized
incidence rate for high grade glioma (Astrocytoma III and IV) for the same
year was about 8 per 100,000 person-years (Figure 1). For all glioma
subtypes, the incidence is higher in men than women.
3
Diagnostics and treatment
There are several different symptoms of brain tumours: epileptic seizures,
altered personality, increasing neurological deficits, and signs of increased
intracranial pressure (headache, nausea) are the most common. Computed
tomography (CT) and magnetic resonance imaging (MRI) are methods used
to examine the brain, with MRI being the method of choice when
investigating a suspicious brain tumour. The first line of treatment for
malignant glioma is surgery. With a glioblastoma diagnosis, however, the
tumour is disseminated, with tumour cells that are not detectable through
MRI spread outside the main tumour bulk, making a microscopic radical
surgery nearly impossible. If possible, macroscopic total surgery is
conducted. If the patient is not suitable for surgery or the tumour location
does not permit radical surgery, a subtotal resection or biopsy is preferred to
verify the diagnosis. Previously, the treatment possibilities for glioblastoma
were very sparse, consisting of radiotherapy and different types of
chemotherapy combinations (CCNU, NOC (CCNU, Procarbacin, Vincristin)).
Today, treatment consists of concomitant radiotherapy, 2 Gray (Gy) per
fraction given daily, Monday to Friday, with a total dose of 60 Gy, and
Temozolomide (3), followed by adjuvant Temozolomide alone.
Figure 2 The age standardized incidence rates per 100,000 person-years in Sweden from 1970 to 2009. Ages 0 - >85 for astrocytoma grades III and IV ( IV=glioblastoma).
0
1
2
3
4
5
6
7
19701971
19721973
19741975
19761977
19781979
19801981
19821983
19841985
19861987
19881989
19901991
19921993
19941995
19961997
19981999
20002001
20022003
20042005
20062007
20082009
Men Women
Figure 2 The age standardized incidence rates per 100,000 person-years in Sweden from 1970 to 2009. Ages 0 - >85 for astrocytoma grades III and IV ( IV=glioblastoma).
0
1
2
3
4
5
6
7
Figure 2 The age standardized incidence rates per 100,000 person-years in Sweden from 1970 to 2009. Ages 0 - >85 for astrocytoma grades III and IV ( IV=glioblastoma).
0
1
2
3
4
5
6
7
19701971
19721973
19741975
19761977
19781979
19801981
19821983
19841985
19861987
19881989
19901991
19921993
19941995
19961997
199819701971
19721973
19741975
19761977
19781979
19801981
19821983
19841985
19861987
19881989
19901991
19921993
19941995
19961997
19981999
20002001
20022003
20042005
20062007
20082009
Men Women
4
Temozolomide is an alkylating agent that works by alkylating/methylating
DNA, usually at the N-7 or O-6 position of guanine, leading to DNA damage
and cell death. Some tumour cells, however, express an enzyme O-6-
methylguanine-DNA-methyltransferase (MGMT) which repairs the DNA
damage by removing the methyl-group, which leads to a poorer response to
Temozolomide. In some tumours, the MGMT gene is silenced through
methylation, promoting Temozolomide response (4). Recently, a monoclonal
antibody towards vascular endothelial growth factor (VEGF), Bevacizumab,
has been approved for second-line treatment of glioblastoma (5).
Risk factors
There are few known risk factors for glioma. Different potential risk factors
have been studied, such as nitrosamines (6, 7), x-ray (8, 9), electromagnetic
fields (10-12), head trauma (13-15), and epilepsy (16-18), with conflicting
results and no clear association with glioma risk. Smoking has also been
studied as a potential glioma risk factor, but no association has been found
(19, 20).
Mobile phone
Mobile phone use has been hypothesised to be a risk factor because of non-
ionizing radiation with radiofrequency close to the brain, but in a big case-
control study, INTERPHONE (21), including 2,765 glioma cases and 2,409
meningioma cases with matched controls from 13 countries, no clear
association was found for either glioma or meningioma after more than 10
years observation. For both meningioma and glioma, a reduced odds ratio
(OR) was found for regular mobile phone users. It has been speculated that
this reduction of risk could be due to recall bias among the participants or
methodological limitations. When looking at glioma cases with the highest
mobile phone use of ≥ 1640 h cumulative call time, an OR of 1.40; 95%
confidence interval (CI) 1.03-1.89 was found (22).
Genetic syndromes and familial aggregation
Studies have shown an association between a number of rare genetic
syndromes such as neurofibromatosis 1 (NF1), NF2, Li-Fraumeni, multiple
harmatoma, tuberous sclerosis 1 (TSC 1), TSC 2, retinoblastoma 1 (RB1),
Turcot syndrome, and brain tumour risk. These syndromes are rare and can
only account for few of the brain tumour cases (23, 24).
5
For first-degree relatives, a twofold increased risk for glioma has been seen
and glioma aggregation has been found in some families (25, 26). To further
study susceptibility loci in families with two or more glioma, an international
study called GLIOGENE (27) was started in 2007.
Ionizing radiation
Studies on ionizing radiation have shown an association to brain tumour
risk. This was first described in a study on Israeli children receiving radiation
therapy for tinea capitis as a prescribed treatment during immigration. They
were treated with doses from 1 Gy up to 6 Gy to the scalp and follow-up
showed an increase in brain tumour incidence for both meningioma (relative
risk (RR) 9.5; 95% CI 3.5-25.7) and glioma (RR 2.6; 95% CI 0.8-8.6). The
risk was dose-dependent (28-30).
Studies on atomic bomb survivors from Hiroshima and Nagasaki have
shown an increase in meningioma incidence among the exposed in
comparison with the unexposed. An increase in total central nervous system
(CNS) malignancy has also been seen among exposed cases (31-33).
In a follow-up study on childhood cancer survivals, the Childhood Cancer
Survival Study (CCSS), children treated with different type of radiotherapy to
the brain showed an increase in brain tumour incidence especially for glioma
and meningioma (34).
Asthma and allergy
Several studies have shown an inverse relationship between glioma and self-
reported atopic disease: asthma, allergy, and eczema (Table 1) (16, 35-42).
Association with immunoglobulin (Ig) E levels and glioma, where cases had
lower post-diagnostic IgE levels than controls, has also been reported (43,
44). It has been hypothesised that this might be caused either by the tumour
itself or different treatments given after diagnosis. In a more resent study of
169 cases with glioma and 520 controls, prediagnostic IgE levels were tested
for association with glioma risk. They found an inverse association with
borderline elevated total IgE levels (25-100 kU/L) and glioma risk (OR 0.63;
95% CI 0.42-0.93) compared with normal IgE levels (<25 kU/L). However
no association was found for elevated IgE (>100 kU/L) and glioma risk (OR
0.98; 95% CI 0.61-1.56) (45). Another study, on prediagnostic IgE levels in
275 cases with glioma and 963 matched controls, found an association
between elevated IgE levels and glioma risk, especially for high levels of IgE
in high-grade glioma (46). A recent study confirmed the inverse association
with IgE levels and glioma risk and also examined IgE levels in blood
6
samples taken more than 20 years before glioma diagnosis and showed that
the association is present at least 20 years before glioma diagnosis (47).
It has also been suggested that the findings of an inverse relationship
between self-reported history of atopic disease could be affected by recall
bias and, because the IgE levels in many of the studies were post-diagnostic,
it could be affected by the tumour itself. This has led to studies of
polymorphisms in asthma-related genes such as interleukin 4 (IL4),
IL4receptor (IL4R), and IL13 (48-51). However, no consistent evidence has
been found for an association between these immune genes and glioma risk.
In one study including 756 cases and 1190 controls, an association with
increased risk was found for the minor allele in a single nucleotide
polymorphism (SNP) in the IL4 gene. A minor allele association with
decreased risk was seen in an IL6 SNP (48). In another study on 110 cases
and 430 controls from Sweden, an inverse risk of glioblastoma was found for
four SNPs in IL4Ralpha and IL13 (49). This was, however, not confirmed in
a larger study on 217 cases and 1171 controls (50). In the latter study, an
association with increased glioma risk with IL4Ralpha polymorphism was
found. In a study on 456 glioma cases and 541 controls, IL4 and IL4R were
not associated with glioma risk, but polymorphisms in IL13 were associated
with case-control status and IgE levels (51).
7
8
Low penetrance genes
The association between single nucleotide polymorphisms in low penetrance
genes and glioma risk has over the years been studied, but often the study
populations have been too small to assure a true association or even detect
an existing association. In the last couple of years, large genome wide
association studies (GWAS) has been conducted and seven chromosomal
loci, in the telomerase reverse transcriptase (TERT), regulator of
telomerase elongation helicase 1 (RTEL 1), coiled-coil domain containing 26
(CCDC26), cyclin dependent kinase inhibitor 2A-cyclin dependent kinase
inhibitor 2B (CDKN2A-CDKN2B), pleckstrin homology-like domain family
B member 1 (PHLDB1), TP53, and EGFR genes, associated with glioma risk
have been identified (Table 2) (52-54). In one study of 1,878 cases and 3,670
controls, later validated in three additional independent datasets of 2,545
cases and 2,953 controls in total, risk loci in TERT, RTEL 1, CCDC26,
CDKN2A-CDKN2B, PHLDB1 were found (52). Two of the loci, in CDKN2B
and RTEL1, were found in another study published at the same time by
Wrensch et al. In the third GWAS, two loci in EGFR were found associated
with risk (54).
In a study by Andersson et al. from 2010 on 725 glioma cases (including 329
glioblastoma cases) and 1,610 controls, a polymorphism in EGFR
intron/exon boundary 7 (rs4947986) was found that was associated with
risk. In the same study, this finding was confirmed in a separate cohort of
713 glioblastoma cases and 2,236 controls from M.D. Anderson Cancer
Center (MDACC) in Texas (55). Recently, large GWAS have identified loci in
EGFR associated with glioma risk. In a pooled study of 1,056 glioblastoma
cases and 2,384 controls, three polymorphisms were found: one in the EGFR
promoter and two in intron 1 (56). In a study of 728 glioma and 1,600
controls two polymorphism associated with risk were found: one located in
the border between intron 1 and exon 2, and one at the border between
intron and exon 7 (rs4947986). The most recent and largest GWAS, on 4,147
glioma cases and 7,435 controls, identified two loci in EGFR, one of them in
intron 1, (rs11979158) and one (rs2252586) in a telomere region to EGFR
with a significance of p=2.08 x 10-8. Both SNPs were located at 7p11.2 and
were significant regardless of tumour grade, EGFR amplification status,
p16INK4A deletion or IDH mutation (54).
CCDC26, located on chromosome 8q24.21, has been associated with low-
grade glioma and oligodendroglioma, but variations in 8q24.21 have also
been associated with other types of cancer such as bladder (57), colorectal
(58), prostate (59) and breast cancer (60). PHLDB1 has also been associated
9
mainly with low-grade glioma (61). Both CCDC26 and PHLDB1 have been
associated with IDH1 mutation. IDH1 and IDH2 mutations have been found
in grade 2 and 3 glioma and also in secondary glioblastoma, and an
association between mutated IDH1 or IDH2 and better prognosis for low-
and high-grade glioma has been found (62). The role of PHLDB1 in
gliomagenesis is mostly unknown. It has been thought to respond to insulin
and enhance Akt (63). The tumour suppressor protein phosphatase and
tensin homolog PTEN is often lost in glioblastoma, which leads to Akt
activation (64), and in a study of astrocytic cells, activation of Akt led to
conversion to glioblastoma (65).
The CDKN2A-B region harbours p14 and p16. P16 is a known tumour
suppressor and is often lost in both copies in glioblastoma, but has also been
linked to other glioma subtypes such as low-grade glioma and
oligodendroglioma (61, 66).
RTEL1 and TERT are both telomerase regulating genes. Telomerase is an
enzyme that regulates the length of a region of repetitive nucleotide
sequences at the end of the chromosomes called telomeres. The length of the
telomeres are crucial for infinite proliferative potential, something acquired
by many malignant tumours giving them a possible immortal phenotype
(67). The telomerase activity is partly regulated by the reverse transcriptase
encoded by the TERT gene. The TERT gene has also been associated with
risk in other cancers such as pancreatic cancer, lung cancer, basal cell
carcinoma, and testicular cancer. In a recent study, two of these risk
polymorphisms (rs2736100, rs2853676) have, together with five other
polymorphisms in the human TERT (hTERT) and human telomerase RNA
component (hTERC) genes, been associated with telomere length at the age
of 60, but not at the age of 50 (68).
RTEL1 has primarily been associated with high-grade glioma and in
glioblastoma, RTEL1 has been associated with survival (61, 66, 69).
In a GWAS on cutaneous basal cell carcinoma, a SNP (rs78378222) in the
3´untranslated region of TP53 was found associated with risk. This
polymorphism was also tested in other cancers and a minor allele association
with risk in prostate cancer, colorectal adenoma, and glioma was found (70).
In a recently published paper, seven low-frequency SNPs at 8q24.21
significantly associated with glioma risk were found. One of the SNPs
(rs55705857) was significant after adjusting for the other six SNPs and was
found to be most significantly associated with oligodendroglioma and glioma
with mutated IDH1 or IDH2 (71).
10
Table 2 Genes associated with glioma risk
Gene Loci Association to glioma subtype
Study (reference)
TERT 5p15.33 High grade glioma Shete (52)
CCDC26 8q24.21 Low grade glioma Oligodendroglioma
Shete (52), Jenkins (71)
CDKN2A-CDKN2B
9p21.3 Low and high grade glioma Oligodendroglioma
Shete (52), Wrensch (53)
RTEL1 20q13.33 High grade glioma Shete (52), Wrensch (53)
PHLDB1 11q23.3 Low grade glioma Shete (52)
EGFR 7p11.2 Reglardless of subtype
Sanson (54)
TP53 Chromosome 17 Stacey (70)
Prognostic factors
Few prognostic factors have been recognised for glioma. The patient’s age
(72) and Karnofsky Performance Status at diagnosis (73) have been found to
affect prognosis as well as grade and histological type. The extent of surgery
affects survival (74), where macroscopically radical resection is favourable
(75). Methylation of the promoter of the MGMT gene has been associated
with prolonged overall survival in glioblastoma patients treated with
Temozolomide (4, 76). The recent discovery and use of Bevacizumab has, in
phase II trials, showed promising results for glioblastoma survival (77). It
has been approved for treatment of glioblastoma in the United States with
the reservation that larger phase III studies confirming the results are
needed; it has not been approved in Europe.
In oligodendroglioma, the loss of chromosomes 1p and 19q has been
associated with better prognosis (78). In astrocytic tumours EGFR, plays an
important role in tumour progression. EGFR is amplified in about 30-50 %
of malignant glioma, and this is associated with poorer survival in younger
patient with glioblastoma (age 55-60 years). Glioblastomas often have
11
mutated EGFR with loss in the extracellular part, which renders the receptor
consistently active (79). Low-grade tumours and anaplastic astrocytoma with
overexpression of EGFR are associated with poorer prognosis (80).
Virus and immunologic response
Viruses
Viruses are small infectious agents that are obligate intracellular and depend
on the host cell for replication. There are several different forms of viruses
that can be divided into two groups, RNA viruses and DNA viruses,
depending on the content of the viral genome. The DNA virus family can be
divided into two groups: enveloped, containing Herpesviridae, for example;
and naked capsid, containing Adenoviridae, for example. The virus particle
(virion) contains a nucleic acid genome packed into a capsid (protein coat) or
an envelope (membrane).
Herpesviridae
Herpes viruses are large DNA viruses, 120-200 nanometres (nm) in size.
They are enveloped and contain a double stranded and linear genome that is
surrounded by a capsid and then enclosed by an envelope containing
glycoproteins. In the space between the capsid and the envelope, there are
viral proteins and enzymes that can assist in the initiation of replication.
Cytomegalovirus
Cytomegalovirus (CMV) is a DNA virus and belongs to the Herpesviridae
family. It is widely spread in the population and, in immunocompetent
healthy adults, the infection is mostly asymptomatic or gives mononucleosis-
like symptoms. In infants or immunosuppressive patients, the infection can
be life threatening.
Epstein-Barr virus
Epstein-Barr virus (EBV) is also a member of the Herpesviridae family. It
infects the antibody presenting B-lymphocytes and is very common in the
population. The virus can cause mononucleosis, but the majority of infected
individuals are asymptomatic. After the acute infection, EBV remains latent
in B-lymphocytes.
12
Varicella zoster virus
Varicella zoster virus (VZV) is a DNA virus in the Herpesviridae family, very
similar to the herpes simplex virus but with a smaller genome. A primary
VZV infection causes chicken pox, and, after the infection has resolved, the
virus can enter a latent phase and remain dormant in the nervous system. A
reactivation may occur later in life and cause shingles (81).
Adenoviridae
Adenovirus (Ad) is a DNA virus and belongs to the Adenoviridae virus
family. They are naked capsid viruses about 100 nm in size and contain a 35
kB double stranded and linear genome (82). It causes upper respiratory
infections, mostly in children, but can also cause diarrhoea.
Immunologic response
The immune system is our defence against disease and responds to foreign
factors through two interacting pathways: the innate response (fever,
interferons, cytokines, mononuclear phagocyte system, natural killer (NK)
cells); and the antigen response, which can be divided further into humoral
(antibody) and cell-mediated (T-cell) immunity. Viruses attack host cells that
harbour or replicate the virus, requiring the immune system to eliminate
both the virus and the affected host cell. In the case of a viral infection,
interferons (Ifn) are released as part of the innate response. NK cells kill
virus-infected cells and macrophages can filter the blood from viral particles.
Macrophages also function as antigen-presenting cells, presenting viral
particles on their cell surface, which promotes T helper cell (TH or cluster of
differentiation 4 (CD4+) T cells) activation. Macrophages also release
cytokines, such as tumour necrosis factor (TNF) and interleukins (IL1, IL6),
which induces fever and promote TH cell activation. There are different
types of TH cells. Two of them are TH1, which promotes inflammatory
response, enhances phagocytic killing, and stimulates B cells to start
antibody production (Immunoglobulin (Ig) M and IgG). TH2 produces IL4,
IL5, IL6, and IL10, enhancing IgG production and promoting B-cell
differentiation into antibody-producing plasma cells or memory cells. TH2
can also enhance IgA and IgE production. Activation of CD4+ T cells leads to
activation of both cytotoxic T cells (CTL or CD8+ T cells) and B cells. The
CTL destroy infected cells and B cells produce antibodies (83, 84).
There are different types of immunoglobulins (Ig), which are produced in
different parts of the immune response. IgM is produced in the early
response and indicates a primary viral infection. After two to three days, the
13
production of IgA and IgG starts. IgG remain in the circulation for several
weeks, and can be detected life long for viruses that remains latent in the
body.
Viruses and Cancer
Viruses have been associated with the development of several different types
of cancer. Human papiloma virus (HPV) has been shown to cause cervical
cancer (85) and a subset of cancer in the oropharynx (86).
EBV has been associated with different cancers, such as Hodgkin’s
lymphoma, Burkitt’s lymphoma, nasopharyngeal carcinoma, and human
immunodeficiency virus (HIV) -associated CNS lymphoma (87).
In brain tumours, different viruses have been studied as potential causal
factors. Viral DNA or proteins from Simian virus 40 (SV40) and John
Cunningham virus (JC-virus) have been detected in brain tumour tissue (88,
89). Cytomegalovirus (CMV) gene products have been found in brain tumour
tissue (90-92). However, there are studies that have failed to replicate these
findings (93, 94). In a study by Cobb et al., researchers found an association
between a CMV protein, human CMV (HCMV) immediate early-1 (IE1), and
glioblastoma cell growth, indicating that CMV can affect different pathways
in glioblastoma (95). Adenovirus (Ad) has also been found in brain tumour
tissue (96).
EGF/EGFR
Function
The epidermal growth factor receptor (ErbB1, EGFR) is a transmembrane
tyrosine kinase receptor at the cell surface. Epidermal growth factor (EGF) is
a ligand that binds to one of four receptors in the EGF-receptor family (ErbB
1-4). When a ligand binds to EGFR, a dimerization of one or more of the
receptors in the EGFR family starts leading to the formation of either homo-
dimers, to receptors of the same kind, or hetero-dimers, to different
receptors from the ErbB family. After dimerization, the receptor is activated
and internalized. Activation of EGFR starts a cascade of downstream
signalling molecules and activation of different pathways such as mitogen-
activated protein kinase (MAPK), P13K/Akt, and phospholipase C gamma
(PLCγ), leading to transcription in the nucleus and cell proliferation. This
pathway plays an important role in cell growth, survival, proliferation, and
differentiation (97).
14
EGF/EGFR and glioma
EGF 61 A/G, a polymorphism located 61 bp downstream from the EGF
promoter in the 5´un-translated (5´UTR) region, has been studied in
association with glioma risk and prognosis (Table 3) (98-100). The rare
allele G has been found to be more common in glioblastoma tissue compared
to tumour-free brain tissue. This correlates to an increased EGF expression,
indicating a functional role of that variant (100, 101).
The EGFR gene plays a role in the progression of glioma, and is amplified in
30-50% of malignant glioma. However, no clear association with prognosis
has been found. In a study of 403 glioblastoma cases with available
treatment data, a better prognosis with EGFR expression was found.
However, in the same study astrocytoma with EGFR tumour overexpression
had a significantly worse prognosis. EGFR overexpression was more strongly
associated with survival than EGFR amplification in both astrocytoma and
glioblastoma (44). In a study of 41 glioblastoma cases, with available
treatment data and over the age of 50, no association between prognosis and
EGFR amplification was found (102) and the same results were found in a
larger study of 715 glioblastoma cases (103). A nonsignificant trend towards
less EGFR amplification was seen in a group of glioblastoma with long-term
survival (over 36 months) and available treatment data (104).
Overexpression in astrocytoma and low-grade glioma has been associated
with poorer survival (44, 105). Glioblastoma with amplification often
contains a mutation, with deletion of exon 2-7, leading to loss of the
extracellular part of the EGF-receptor, which renders the receptor to be
constantly active (EGFRvIII) (79).
15
16
VEGF/VEGFR
Angiogenesis is an important part of tumour growth and glioblastomas are
highly vascular tumours. Small cancer clusters depend on their ability to
create blood vessels for the delivery of oxygen and nutrients. Second-line
therapy with the antiangiogenic antibody Bevacizumab targeting vascular
endothelial growth factor (VEGF) has increased in recent years.
Function
The VEGF family consists of six members: VEGF-A, VEGF-B, VEGF-C,
VEGF-D, and the placental growth factor (PlGF). VEGF-A, also referred to as
VEGF, is the molecule mainly involved in tumour angiogenesis. VEGF
expression can be induced by different factors, such as low pH, hypoxia,
chemokines, growth factors, sex hormones, inflammatory cytokines, and also
through loss of tumour-suppressor genes or activation of oncogenes. VEGF
is expressed in elevated levels by nearly all human cancers and binds
primarily to VEGF receptor 2 (VEGFR-2). Endothelial cells express elevated
levels of VEGFR-2 and, when VEGF binds to the receptor, a cascade of
intracellular pathways affecting vascular permeability, mobilization,
proliferation, migration, and survival in the endothelial cells are activated.
VEGF also binds to VEGF receptor 1 (VEGFR-1) which regulates cell growth
induced by VEGFR-2 during angiogenesis (106). VEGF receptor 3 (VEGFR-
3) is activated mainly through the binding of VEGF-C and plays a role in
lymphoangiogenesis (107).
VEGF/VEGFR and glioma
VEGF secreted by tumour cells has a paracrine effects on tumour endothelial
cells and angiogenesis, but VEGF may also have an autocrine effect on cell
viability and tumour growth. Astrocytoma cells express VEGFR-2 and can
also secret VEGF leading to an autocrine effect through the c-Raf/MAPK,
P13K, and PLC/protein kinase C (PKC) pathways (108). VEGF levels have
been found to increase with higher glioma grade (109, 110). In glioblastoma
the tumour vessels are highly disorganized, dilated, and leaky, leading to
hypoxia and high VEGF levels (111). It has been hypothesised that this
hypoxia is one of the mechanisms of radiation resistance in the tumour (112)
and that anti-VEGF therapy or VEGFR-2 blockage could normalize tumour
blood vessels and oxygenate the tumour, which could lead to a possible
window for radiotherapy response (113, 114). However, other studies have
shown a radiation-induced up regulation of VEGF/VEGFR2 in mouse
models, indicating that these factors also play an important role in
17
radiotherapy failure (115). Increased blood flow and reduction of tumour
hypoxia by anti-VEGF therapy might also induce tumour growth but, at the
same time allow improved drug and chemotherapy delivery and effect on the
proliferating tumour cells (116) . Glioma stem-like cells, a type of cell similar
to normal neural stem cells with the ability to self-renew, differentiate, and
proliferate (117, 118), also promote tumour angiogenesis by producing
angiogenetic factors such as VEGF (119). High VEGF levels have been
associated with poor glioblastoma outcome (110, 120) and, in high-grade
astrocytoma, VEGF overexpression has been associated with poor prognosis
(108, 121, 122).
The signalling pathways in human glioma, including EGF/EGFR and
VEGF/VEGFR, are shown in Figure 3 (123).
Figure 3 Key signaling patways in human glioma. Genes that are
inactivated are red and activated are blue. Reprinted with permission
from Biochimica et Biophysica Acta (BBA)-Proteins and Proteomics,
Grzmil et al. 2009. Copyright © Elsevier B.V. All rights reserved.
18
DNA repair
An accurate DNA repair is crucial for cell survival and damage to the DNA is
a function of many effective cancer treatments. Radiotherapy and
chemotherapy often induce damage to the DNA and the tumour’s ability to
repair this damage is a key component in treatment failure. In glioblastoma,
radiotherapy and concomitant Temozolomide are the standard treatment
and both these treatments affect the DNA, with radiotherapy inducing
double strand breaks and Temozolomide acting as an alkylating agent.
Mechanisms
Base excision repair
Base excision repair (BER) responds to small damages in the DNA such as an
oxidised or alkalised base or single strand breaks.
A damaged base is recognised and removed by an enzyme that is a member
of the DNA glycosylase family creating an apurinic/apyrimidic site (AP-site)
(124). Poly ADP ribose polymerase (PARP) and polynucleotide kinase (PNK)
assists in recognition of abasic sites and single strand breaks. The damage
can then be repaired through two different pathways: the short-patch
pathway and the long-patch pathway (125, 126).
In the short-patch pathway, the AP-site is recognised by apurinic/apyrimidic
endonuclease 1 (APE1), which flips out the deoxyribose residue and cleaves
the DNA 5´site, recruiting the next enzyme, DNA polymeraseβ (polβ) (127),
to the site. This enzyme cleaves the 3´site and fills in the gap with a
polymerase domain and then the X-ray repair cross-complementing 1
(XRCC1)-ligase 3 closes the cut (128-131).
In the long-patch pathway the AP-site is filled by DNA polβ, polδ, or polε
through assistance of proliferating cell nucleus antigen (PCNA) protein. The
3´end of the damage strand is then cleaved by flap structure-specific
endonuclease 1 (FEN1) and DNA ligase 1 seals the gap (132-134).
Mismatch repair
Mismatch repair (MMR) is a pathway that repairs mismatched bases, base
deletions, or insertions, contributing to genetic stabilisation.
The mismatch site is first recognized by a melanocyte-stimulating hormone 2
(MSH2)/MSH6 dimer (MutSα) (135, 136) or a MSH2/MSH3 dimer (MutSβ),
19
and then a MutL-homolog 1 (MLH1)/Postmeiotic segregation increased 2
(PMS2) dimer (MutLα) or MLH1/MLH3 dimer (MutLβ) binds to the MutSα,
which leads to excision of the mismatch, resulting in a single strand gap and
recruitment of the exonuclease 1 (EXO1). The created gap is stabilised
through replication protein A (RPA) and then filled by DNA polδ and
proliferating cell nucleus antigen (PCNA), and DNA ligase 1 then seals the
gap (132, 137, 138).
Inactivation of the pathway in humans is associated with hereditary
nonpolyposis colon cancer (HNPCC), also known as Lynch syndrome, and
HNPCC associated cancer in other organs, such as the colon, rectum, uterine
endometrium, stomach, and ovaries. It is also involved in Turcot syndrome,
a rare genetic syndrome that predisposes people for colorectal adenomas and
primary brain tumours (139, 140). It has also been associated with sporadic
cancers in other organs such as breast, prostate, and lung cancer (141, 142).
Nucleotide Excision Repair
Nucleotide Excision Repair (NER) is involved in several different DNA
lesions caused, for example, by chemotherapy or radiotherapy.
NER contains of two different pathways: the global genome repair (GGR)
and transcription coupled repair (TCR).
TCA is a specific repair pathway that finds damage to the DNA that causes
blockage of RNA polymerase II movement. Two TCA specific genes,
Cockayne syndrome A (CSA) and Cockayne syndrome B (CSB), detect the
halted polymerase and remove it, enabling further repair of the damaged
DNA (143). NER starts by xeroderma pigmentosum complementation group
C (XPC) recognising a damaged site in the DNA and binding to human
homolog of RAD23B (HHRAD23B) (144), forming a site for binding of other
DNA repair proteins such as xeroderma pigmentosum complementation
group A (XPA), replication protein A (RPA), transcription factor II H
(TFIIH) (145, 146), and XPG (147). Excision repair cross-complementing
rodent repair deficiency complementation group 1 (ERCC1)-XPF, a
heterodimeric complex, is then recruited to the site and, together with XPG,
cuts the 3´and 5´ends of the damaged DNA. The damaged DNA is then
removed and, with help of PCNA, RPA, and RFS, the gap is filled by DNA
polyδ or ε and sealed by DNA ligase (148, 149).
Defects in these pathways are present in xeroderma pigmentosa, a condition
predisposed to cancer caused by ultraviolet (UV)-light (132, 148, 149) .
20
Homologous recombination and non-homologous end joining
Homologous recombination (HR) and non-homologous end joining (NHEJ)
are two pathways used in double strand break repair (150).
After a double strand break, ataxia telangiectasia mutated (ATM) is activated
through autophosphorylation and a signalling molecule encoded by ATM is
activated. ATM is also one factor involved in protein phosphorylation in both
DNA repair and regulation of the cell cycle (151). The RAD50/MRE11/NBS1
(MRN) complex is phosphorylated by ATM, leading to exposure of the 3´end
of the DNA at the DBS site, a process also involving breast cancer
susceptibility gene 1 (BRCA1) (152). After this, the repair goes through either
the HR pathway or the NHEJ pathway.
In HR, a sister chromatid is acquired and therefore this repair pathway only
occurs in S or G2 of the cell cycle (153). In the HR pathway, RAD51, BRCA2,
and DNA dependent ATPase RAD54 mediate an invasion of the 3´end to a
homologous section of the sister chromatid and DNA-polymerase then uses
the complementary strand and prolongs the 3´end past the damaged site.
After this, DNA-polymerase returns to the damaged strand and DNA ligase
ligands the two strands, resulting in a complex called Holiday junction (154).
NHEJ can work through the whole cell cycle and does not need a
homologous chromosome to function and is therefore major pathway for
DNA double-strand break repair (153, 155). In the NHEJ pathway, the DNA
protein kinase (DNA-pk), consisting of a catalytic subunit and a regulatory
subunit, is the main driver of the pathway. The regulatory subunit
(KU80/KU70 heterodimer) binds to the two double stranded ends (156) and
activates the catalytic subunit (DNA PKCS, which is a member of the
phosphinositide 3-kinase (P3-kinase) family). The catalytic subunit then
binds to and stabilizes the DNA ends, processing or trimming them together
with ATM-activated endonuclease Artemis to allow end joining (157). The
ends are then fused together by the XRCC4/DNA ligase IV complex (158).
This pathway is only possible in higher creatures with lots of non-coding
DNA (132, 159, 160).
Direct repair
There are several different mechanisms for direct repair of DNA damage.
Examples of this are the protein product of the methylguanine-DNA
methyltransferase (MGMT) gene that irreversibly removes alkyl groups
from the O6-position of guanine. MGMT silencing is associated with a better
21
response to Temozolomide treatment and better prognosis in glioblastoma
patients (4).
DNA-repair gene variants and glioma
Genetic variations in different DNA repair genes have been associated with
glioma risk. In a study of 771 glioma cases and 752 cancer-free controls 20
tagging SNPs in Ligase 4 (LIG4) and XRCC4 were studied for association
with glioma risk as well as possible gene-gene interaction. For one LIG4
SNP, researchers found an association with increased glioma risk and, for
one of the XRCC4 SNPs, an association with decreased risk. They also
showed that interactions between these two genes were associated with
glioma aetiology and that effects of one gene might not be strong enough to
be detected if it is linked to another functional gene (161).
In another study, 1,127 tagging and 388 putative functional SNPs in 136 DNA
repair genes were studied for association with glioma risk in 1,013 cases and
1,016 controls. Sixteen SNPs significantly associated with risk were found in
chromatin assembly factor 1, subunit A (CHAF1A), nei endonuclease VIII-
like 3 (NEIL3), ERCC1, RPA3, DNA cross-linked repair 1B (DCLRE1B),
TP53, ataxia telangiectasia and Rad3 related (ATR), POLD1, and
melanocyte-stimulating hormone 5 (MSH5). The strongest and most
significant association with risk was found in CHAF1A (162).
In another study on 373 glioma cases and 365 cancer-free controls, the 18
functional SNPs in DNA repair genes were studied for association with
glioma risk. Researchers found six polymorphisms located in ERCC1,
XRCC1, APEX1, PARP1, MGMT, and LIG1 associated with glioma risk, and,
in an analysis of the cumulative genetic risk, they found a significant gene-
dose association with increasing numbers of unfavourable genotypes of the
same SNPs and glioma risk. They found that MGMT was the predominant
risk factor for glioma and they also found a strong interaction between
exposure to ionizing radiation, PARP1, MGMT, and APEX nuclease 1
(APEX1) (163).
In a study of 119 glioma and 180 cancer-free controls researchers found a 3.5
times increased glioma risk with the G allele in XRCC1 Arg399Gln. In the
same study, they also found a plausible protective effect of heterozygote
genotype VA of PARP1 Val762Ala. This decrease in risk was, however, lost
when combined with the G allele in the XRCC1 locus (164). These results are
in line with the results found on XRCC1 and PARP1 in the previously
mentioned study by Liu et al. (163).
22
In another study of 590 glioblastoma 100 SNPs previously found associated
with glioma in GWAS were studied for association with survival. Two SNPs
in RTEL1 (rs2297440, rs6010620) with a minor allele association with better
prognosis and one SNP in regulator of G-protein signalling 22 (RGS22)
(rs4734443), one in BTB domain containing 2 (BTBD2) (rs11670188), one in
LIG4 (rs73259279), and one on chromosome 3 (rs13099725) with a minor
allele association to shorter survival were found. Polymorphisms were also
studied for association with short-term survivors (STS ≤ 12 months) and
long-term survivors (LTS ≥ 36 months) and an association between one SNP
in LIG4 (rs7325927) and one in BTBD2 (rs11670188) and STS was found.
Additionally one SNP in high mobility group AT-hook 2 HMGA2
(rs1563834), one in VPS8 (rs6765837), two in RTEL1 (rs2297440,
rs6010620), and two SNPs in CCDC26 (rs10464870, rs891835) associated
with LTS were also found. In a survival tree analysis age was found to be the
most important prognostic factor; older patients (> 50 years) homozygote
for the minor allele in LIG4 (rs7325927) had the worst prognosis, whereas
young patients (≤ 50 years) homozygote for the minor allele in RTEL1
(r22297440) and HMGA2 (rs1563834) had the best prognosis. LIG4,
BTBD2, HMGA2, and RTEL1 are all involved in the repair of double strand
breaks (69).
Genetics and association studies
SNPs and association studies
The deoxyribonucleic acid (DNA) strand consists of two long polymers of
nucleotides that run in the opposite direction of each other and form a spiral,
the double helix. A nucleotide is a molecule consisting of a sugar and a
phosphate group that bind together with a nitrogen base, adenine (A),
thymine (T), cytosine (C), and guanine (G). The nucleotides form pairs
where A binds with T and C binds with G (165). Genes consist of sequences
of several nucleotides and each individual has two variants of each gene, one
from the mother and one from the father. The genome consists of 98% non-
coding regions and 99.9% of an individual’s DNA sequence is identical to all
other individuals. In the 0.1 % of the genetic code that differs among
individuals approximately 80 % consist of single nucleotide polymorphisms
(SNPs).
SNPs are variations of a nucleotide which occurs naturally every 200-300
base pair (bp) in the population and creates different variants, alleles, of the
same gene. An individual can be heterozygote (having two different bases,
for example, AT or CG) or homozygote (having two of the same bases, for
23
example, TT, AA, CC, or GG) for a specific SNP. The frequency of different
alleles can vary between different populations, and the lowest allele
frequency at a specific location of the chromosome (loci), the minor allele
frequency (MAF), is the detected frequency of the rare allele in a specific
population. If a variant occurs in less than 1% of the population it is called a
mutation (166). SNPs often occur in the non-coding regions of the DNA, the
introns, but can also occur in the coding regions, the exons, which can lead
to changes in the amino acid sequence and altered protein expression.
These normal variations in the genome can be further studied through
genotyping and used in association studies, where risk or prognosis for a
specific disease can be studied in a population. The linkage disequilibrium
(LD) is the association of alleles in different loci, not necessary on the same
chromosome. LD can be used in association studies to investigate a
candidate gene using fewer SNPs, as alleles in strong LD with each other will
be transmitted together (167). When analysing results from association
studies, one must also keep in mind that the true functional variant might
not be the one measured in the analysis but a nearby allele in LD with the
measured SNP (168).
Haplotype and haplotype blocks
A haplotype is a group of alleles that are close to each other and located on
the same chromosome and therefore tend to be inherited together.
Haplotypes build up blocks, so-called haplotype blocks, where usually a few
common haplotypes, 3-5, cover about 90 % of the haplotype variants that
build up the block (169). In these blocks, there is a strong LD and little
evidence of recombination. Between the blocks, areas of more frequent
recombination can be found, so-called recombination hot spots (170).
Recombination occurs during meioses, the type of cell division creating
gametes (sex cells), when two homologue chromosomes come in contact with
each other and exchanges genetic material (171). Recombination is more
likely to occur for loci located far apart from each other. The sizes of
haplotype blocks can vary and block size has been associated with different
populations For example, European and Asians populations have longer
haplotypes blocks compared to African-Americans and Sub-Saharan
populations (172).
This knowledge of haplotype and haplotype blocks is very useful in
association studies. To cover all common variants in a haplotype and the
surrounding block, usually only a few SNPs, so-called haplotype tagging
SNPs, are needed because of the strong LD among SNPs in the same
haplotype and the limited number of different haplotypes in a block (172).
24
Aims
I. To study the association of IgG levels to varicella zoster virus,
adenovirus, Epstein-Barr virus, and cytomegalovirus and glioma risk.
II. To study if polymorphisms in EGF and EGFR are associated with
glioblastoma outcome.
III. To study if polymorphisms in VEGF and VEGFR2 are associated with
glioblastoma outcome.
IV. To study if polymorphisms in DNA repair genes are associated with
glioblastoma outcome.
25
Material
Study population
Paper I
In paper one, we studied 197 glioma cases and 394 controls originating from
three cohorts, The Northern Sweden Health and Disease Study – The
Medical Biobank (NSHDS), The Malmö Diet and Cancer Study (MDCS), and
the Diet, Cancer and Health cohort from Copenhagen. The glioma diagnosis
was identified through linkage to the cancer register and controls were
collected from living individuals matched for age, gender, and cohort, with
no brain tumour history at the time of case diagnosis. The median age at
blood sample for cases was 55.4 years and for controls 55.3 years, and
male/female ratio was 47.7%/52.3%. For NSHDS and MDSC Systematized
Nomenclature of Medicine (SNOMED) was available (Figure 4).
Figure 4 Glioma subclasses according to Systematized Nomenclature of
Medicine (SNOMED) for the three cohorts.
Study
population
197 cases
394 controls
NSHDS
79 cases
158 controls
MDSC
45 cases
90 controls
Diet Cancer
and Health
73 cases
146 controls
61
Glioblastoma
SNOMED
94403
8
Astrocytoma
Grade 1-2
SNOMED
94003
94203
94213
32
Astrocytoma
Grade 3
SNOMED
94013
14
Oligodendro-
glioma
SNOMED
94503
94513
94603
2
Oligoastro-
cytoma
SNOMED
93823
5
Malignant
glioma
SNOMED
93803
2
SNOMED
not
available
73
SNOMED
not
available
26
Northern Sweden Health and Disease Study
NSHDS is a large population-based cohort consisting 166,000 randomly
selected men and women and blood samples from 278,000 sample
occasions. NSHDS consists of three sub-cohorts, all containing blood
samples and questionnaires: The Västerbotten Intervention Program (VIP),
The Västerbotten Mammary Screening Program, and the Northern Sweden
MONICA (Monitoring Trends and Determinants in Cardiovascular Disease)
Project. In the VIP cohort, all individuals since 1985 have been asked to
participate in a health study the year they turn 40, 50, and 60, and, by June
2006, the cohort comprised 94,630 samples from 74,690 individuals.
Between 1995 and 2006, blood samples have been collected in the Mammary
Screening program (women aged 55-69 are screened for breast cancer with a
mammography every second year), and these have been added to the VIP
cohort. In 2007, there had been 48,000 sampling occasions from 27,500
women. About 50% of the women in the mammography cohort had also
attended VIP. The MONICA Project contains samples from a population-
based screening for risk factors associated with cardiovascular diseases. The
screening was carried out in 1986, 1990, 1994, 1999, and 2004. In all, 14,000
sampling occasions of 9,000 individuals aged 25-64, were collected. 50% of
the individuals are also included in VIP (173-175).
The Malmö Diet and Cancer Study
MDSC, a population-based prospective cohort study, was started in 1991 in
order to study the impact of diet on cancer incidence and mortality. Initially,
all living subjects in Malmö born between 1926 and 1945 were invited, and,
in 1995, the study was extended to include women born 1923-1950 and men
born 1923-1945. The rationale for including a larger group of young women
than men was to enable studies on breast cancer in premenopausal women.
The only exclusion criteria up front were language problems and mental
retardation. The overall participation rate was 40.8% for men and 38.3% for
women and the mean age at 1 January 1991 was 54.9 years for participants
and 54.3 years for non-participants. In summary, men and women between
the ages of 44 and 74 were recruited between 1 January 1991 and 25
September 1996. A total of 28,098 blood samples were collected and stored
in the medical biobank, and regular follow-up in the national registries of
cancer and mortality has been done. In 2001, a study of potential selection
bias between the participants in MDSC (n0=28,098) and the non-
participants (no=40,087) was conducted regarding cancer incidence and
mortality; researchers also conducted a mail health survey looking at
differences in subjective health, socio-demographic characteristics, and
lifestyle between participants and non-participants. They found that
27
mortality was higher among non-participants compared to participants
during the recruitment period and follow-up. A trend towards lower cancer
incidence prior to recruitment but higher cancer incidence during
recruitment was found for non-participants compared to participants. No
significant difference was found for socio-demographic structure for the two
groups (176, 177).
Diet, Cancer and Health
The Diet, Cancer and Health study from Denmark is a population-based
prospective cohort study that started in 1993 in order to study the relations
between dietary factors, lifestyle factors, and incidence of cancer and other
chronic diseases. From December 1993 to May 1997, 80,996 men and 79,726
women were invited to participate by answering a questionnaire on food and
lifestyle. Respondents underwent a physical examination including, for
instance, blood pressure measurement, height, and weight, and blood
samples were collected. Individuals between the ages of 50 and 64 born in
Denmark and with no previous cancer diagnosis registered in the Danish
Cancer Registry were included. In all, 27,179 men and 29,875 women
participated in the study and blood samples were obtained on 57,053
participants. Differences in socioeconomic factors ware found for
participants compared to non-participants, with a higher number of
educated and married persons among participants, who also had a higher
number of persons from higher socioeconomic groups compared to non-
participants (178).
Papers II-IV
In papers two, three, and four, we studied cases with glioblastoma collected
from 2000 to 2004 as part of the INTERPHONE study, an international
case-control study on mobile phone use and risk of developing a brain-
tumour. A total of 108 cases from Sweden and 68 cases from Denmark were
collected and a pathology review was conducted according to the 2007 WHO
criteria. There were 109 men and 67 women. The ages ranged from 20 to 69
and the median age at diagnosis was 56 years and did not differ between the
countries. The mean survival time for the whole group was 15.7 months, 16.1
for the Swedish cases and 15.1 for the Danish cases.
The INTERPHONE study
The INTERPHONE study is a large international case-control study in which
the association between brain tumour risk and mobile phone use was
investigated. The study included 16 study centres from 13 countries:
Australia, Canada, Denmark, Finland, France, Germany, Israel, Italy, Japan,
28
New Zealand, Norway, Sweden, and the UK. Cases were collected between
2000 and 2004 from neurological and neurosurgical facilities or cancer
registries and diagnosis confirmed either histologically or on unequivocal
diagnostic imaging. One control was matched to each case except for
Germany, where two controls were matched to each case. Controls were
matched to cases on age, sex, and region of residence, and for the subjects
from Israel, also on ethnic origin. On cases and controls, detailed
information on mobile phone use was collected through interviews. Proxies
were used when the study subject was too ill to participate or had died.
Information was collected on socio-demographic factors, occupational
exposure to electromagnetic fields and ionizing radiation, medical history,
smoking, medical ionizing radiation exposure, and non-ionizing radiation
exposure. Information on the anatomic location and histological location of
tumours was collected for the cases (22).
Confirmation
Papers II and III
Confirmation on significant SNPs in papers II and III was made using a
dataset from The M. D. Anderson Cancer Center (MDACC), consisting of 638
glioblastoma cases recruited from 1990 to 2008 with follow-up until April
2009. All cases were at least 18 years of age at diagnosis, caucasian, with
primary glioblastomas; 397 were male and 241 were female with a mean age
at diagnosis of 52.3 years. The cases were followed from date of diagnosis
until date of death or date of last contact for the cases still alive. At follow-up,
586 cases were deceased and 52 cases alive. Cases with a history of prior
cancer or who previously had treatment for their glioblastoma were
excluded. No treatment data was available in the confirmation cohort.
Paper IV
Confirmation on significant SNPs in the fourth paper was made using a
cohort containing 295 glioblastoma cases from northern UK. In 97 of the
cases, follow-up data was missing and 77 cases had poor DNA quality,
leaving 121 cases for confirmation. No treatment data was available. The
cohort consisted of 78 male and 43 female cases. The median age at
diagnosis was 55 years and the mean survival time 14.8 months.
29
Methods
Paper I
ELISA and immunoflourescence
Enzyme-linked immunosorbent assay (ELISA) can be used to detect a
substance, for example antigen, in a liquid sample.
We analysed levels of IgG antibodies against adenovirus, CMV, and VZV
using ELISA. For EBV we used ELISA to analyse IgG-anti-Epstein-Barr
Nuclear Antigen-1 (EBNA-1). EBNA-1 is a viral protein that is expressed in
latent EBV infections. It becomes present 2-4 months after the primary
infection and is then present life long.
For EBV, we also analysed IgG against Viral Capsid Antigen (VCA), which
appears within a week of an EBV infection and then is present life-long. For
this analysis we used immunoflourescence (IF), a technique where
antibodies are chemically conjugated to a fluorescent dye. The antibodies
bind to the antigen of interest and can then be measured by different IF
techniques. In our study, a microscope was used and the samples were read
and fluorescence activity graded by an experienced microscopist who was
blinded to case/control status.
Binary logistic regression
Binary logistic regression is a regression model that can be used when
working with binary variables (0/1, dead/alive, healthy/sick,
positive/negative). When calculating logistic regression, you get an odds
ratio (OR) describing how strong the association is for the variables you are
testing.
We studied the association between IgG levels and glioma risk using binary
logistic regression. We divided the control group into four same-size
quartiles using the IgG levels and then divided the cases in the same way,
based on the cut-off values from the control quartiles. We then compared the
quartiles in cases and controls using the lowest quartile as reference. The
odds ratios were adjusted for age at blood sample, sex, and cohort
(Denmark, Malmö, Umeå). We also repeated the test using only cases with at
least two years between blood sample and glioma diagnosis to limit the risk
of disease or treatment affecting our results.
30
Papers II-IV
Medical records
Clinical data was collected by a review of patient journals. Request of journal
copies were sent to the participating hospitals in Sweden and, for the cases
from Stockholm, the data was collected at the hospital. Details on date of
diagnosis, follow-up, surgery, chemotherapy, and radiotherapy were
retrieved from the journals with use of a standardised predefined form.
Treatment data are shown in Table 4. In all data was incomplete for 33 of the
cases. In 11 cases, date of diagnosis was missing and follow-up was missing
in an additional 10 cases. Follow-up date was set as either date of death or,
for living cases, date of last follow-up in patient record or 31 October 2006,
whichever came first. 89.8% of the cases were diseased at follow-up.
Table 4 Treatment data for the Swedish and Danish cases used in papers II-
IV.
Sweden
No. 108
Denmark
No. 68
Total (%) Missing (%)
Total (%) Missing (%)
Gross total
resection
6 (5.5) 1 (1.5)
Yes 26 (24.1) 24 (35.3)
No 76 (70.4) 43 (63.2)
Radiotherapy 7 (6.5) 2 (3)
Yes 90 (83.3) 57 (83.8)
No 11 (10.2) 9 (13.2)
Chemotherapy 13 (12.0) 10 (14.7)
Yes 69 (63.9) 14 (20.6)
No 26 (24.1) 44 (64.7)
31
SNP-selection and genotyping
In papers II and III, we used information from the dbSNP, HapMap, and
SNPper databases. We then used the Haploview software to identify tag
SNPs and LD to cover the genes of interest. In papers II and III, we used a
minimum r2 of o.9 and a minor allele frequency of 5% in a HapMap CEPH
(CEU) population (Utah residents with northern and western Europe
ancestry). In paper IV, over 35,000 SNPs in 141 known DNA repair genes
were selected from an inventory of DNA repair genes done in a previous
study (179). The genes were then updated using the University of California,
Santa Cruz (UCSC) Genome Browser Database and the National Center for
Biotechnology Information (NCBI) Entrez Gene database and SNPs were
selected using dbSNP. To shorten the list, r2>0.8 and MAF>1% in a HapMap
CHEP (CEU) population was used and, after elimination of SNPs unsuitable
for genotyping, 1,127 tagSNPs in 136 genes were selected and we also added
388 known functional SNPs using PupaSuite web-based software and the
Illumna in-house algorithm. R2 is a measurement of how strong the
correlation or LD is between two SNPs. A high r2 indicates a high correlation
and, at the maximum r2 of 1 the two SNPs are in perfect LD with each other.
DNA for genotyping was extracted using conventional methods and
quantified using PicoGreen. Genotype validation was performed on 14 family
trios with available HapMap data and the concordance was 100%. In 90
samples, all SNPs were tested twice for internal concordance, which was
100%. The call rate was 80% in papers II and III.
In paper IV, the samples were genotyped using customised Illumina
GoldenGate Arrays and it was determined that genotyping of a sample failed
if genotypes from less than 80% of all loci were returned and genotyping of a
SNP was considered to have failed if a genotype was returned for less than
95% of all samples. A GenCall score less then 0.25 was set as no-call.
Statistical analyses
Cox proportional hazards model
Cox proportional hazard model is a statistical model for studying survival.
The model explores the changes in hazard (risk) over time when covariates
(factors that might be predictive of outcome) are at baseline, as well as how
the hazard varies due to explanatory covariates. If the covariate is not
directly explanatory, they can be confounding, meaning that they do not
affect the outcome directly but affect the true explanatory covariate. It is
therefore important to identify and control for possible confounding factors
32
in the statistical analyses. The effects of covariates on hazard can be
described as hazard ratios.
In papers II-IV, we used the Cox proportional hazard ratio model in SPSS to
study a possible association between different polymorphisms in the
EGF/EGFR and VEGF/VEGFR pathways and genes involved in DNA repair
and glioblastoma outcome. The major allele was set as categorical variable
and all hazard ratios were adjusted for possible confounding factors such as
country, age, sex, and treatment.
Correction for multiple testing
In the forth paper, we analysed a total of 1,458 polymorphisms in 136 DNA
repair genes and possible association with outcome. Because of the large
amount of SNPs tested, we also corrected the results for multiple testing
using a method based on permutation. In this test the SNPs are randomly
assigned to a group and then p-values are calculated; the lowest p-value in
each group is saved and then the procedure is repeated a number of times (in
our study 10,000). The lowest p-values can then be plotted in a graph and
should form a normal distribution curve and its mean value is the lowest
average p-value that can be expected with a random division of the groups.
We used a cut-off of 95%, giving us a 5% false positive rate in the random
test (180).
33
Results
Paper I
In paper I, we studied whether IgG levels of CMV, EBV, Ad, and VZV are
associated with glioma risk in 197 cases compared to 394 controls. These are
very common viral infections in the population in general and we found that
a high number of both cases and controls had antibodies against the viruses,
but there were no significant differences in number of seropositives or the
mean age at blood sample between cases and controls. When comparing the
quartiles in cases and controls, we found no significant difference in Ad,
CMV, or EBV IgG levels. In VZV, we found a trend of lower OR with
increasing quartile. This trend was significant when restricting the analysis
to only cases with at least two years between blood sample and glioma
diagnosis (results shown in Table 5).
Table 5 Cases and controls OR and 95% CI for IgG for VZV. All glioma
cases, and cases with more than 2 years between blood sample and glioma
diagnosis.
Glioma cases versus controls Cases > 2 years between blood sample and glioma diagnosis versus controls
No. (%) OR (95% CI) No. (%) OR (95% CI)
1st quartile
55 (28.1%)
1.0 48 (29.3%)
1.0
2nd quartile
55 (28.1%)
0.85 (0.53-1.36) 47 (28.7%)
0.84 (0.51-1.37)
3rd quartile
43 (21.9%)
0.74 (0.45-1.23) 35 (21.3%)
0.70 (0.41-1.19)
4th quartile
43 (21.9%)
0.68 (0.41-1.13) 34 (20.7%)
0.63 (0.37-1.08)
P-trend 0.06 0.029
34
Papers II and III
In the second paper, we studied the association between polymorphisms in
EGF and EGFR and glioblastoma outcome. In order to make a
comprehensive tagging of the genes, 30 tagging SNPs in EGF and 89 tagging
SNPs in EGFR were chosen and evaluated. An initial screening of outcome
was conducted for each SNP using the homozygotes for the major allele as
categorical variable. All significant HR (p≤0.05) were then adjusted for age,
sex, country, and treatment.
In EGF seven SNPs significantly associated with glioblastoma survival were
found. All seven SNPs had a minor allele association with shorter survival
and were located in haplotype block 4. Two of the SNPs were located in the
promoter region and the other five in different introns (5, 7, 10, 11, 13). The
analyses were also made for the Swedish and Danish cases separately and
the same minor allele association with poor prognosis was seen. The median
survival time for homozygotes for the major allele in six of the seven SNPs
ranged from 14.4 to 15.1 months, and the median survival time for
homozygotes for the minor allele from 8.6 to 8.9 months. In one of the
promoter SNPs (rs6533477), there were no homozygote cases for the minor
allele; in this, an association with shorter survival in heterozygote cases (11.7
months) compared to homozygote cases for the major allele (13.1 months)
was seen.
Four of the SNPs were available for confirmation in the MDACC dataset. The
same trend towards minor allele association with shorter survival was seen
but, this was not significant. This trend was stronger and border-significant
when restricting the analyses to include only subjects over 50 years of age.
This was done to enrich for primary glioblastoma due to differences in
median age at diagnosis (56 in the Swedish/Danish dataset versus 54 in the
confirmation set), and median survival time (12.8 months in the
Swedish/Danish dataset versus 15 months in the confirmation set), which
could indicate more secondary glioblastoma, as the cases were younger and
lived longer in the confirmation set. Hazard ratios in the confirmation
dataset were adjusted only for age and sex, as treatment data was not
available.
In EGFR we found four SNPs significantly associated with glioblastoma
prognosis. All SNPs were located in different intrones (1, 4, 7, 9) and none of
the SNPs were located in the same haplotype block. In two of the SNPs, the
minor allele was associated with longer survival and, in the other two, the
35
minor allele was associated with shorter survival. None of the significant
SNPs were available for confirmation in the MDACC dataset.
In paper III, we studied the association between polymorphisms in VEGF
and VEGFR2 and association with glioblastoma survival. In VEGF, 27
tagging SNPs and, in VEGFR2, 17 tagging SNPs were analysed. No
significant association between any of the VEGF SNPS and glioblastoma
survival was found. In VEGFR2 we found one SNP located in intron 1
(rs1252008) with an association with longer survival for homozygotes for the
minor allele compared to homozygotes for the major allele, HR 0.45; 95% CI
(0.21-0.95). No association with survival was seen for the heterozygotes HR
1.09; 95% CI (0.74-1.62). The SNP was not available for confirmation in the
MDACC dataset.
In another VEGFR2 SNP located in the promoter (rs2071559) we saw an
association between heterozygotes and shorter survival when compared to
homozygotes for the major allele. No significant association with survival
was seen for homozygotes for the minor allele. The SNP was available for
confirmation in the MDACC dataset, but no association with survival was
found.
Paper IV
In paper IV, we studied the association between polymorphisms in DNA
repair genes and outcome in 172 glioblastoma cases from Sweden and
Denmark. First, a screening of all SNPs was conducted in the
Swedish/Danish dataset where a total of 159 glioblastoma were available
after genotyping as 13 had poor DNA quality and an additional 21 had
missing follow-up, leaving 138 cases for the analyses. In all, 142 SNPs
significantly (p<o.o5) associated with outcome were found. In the next step,
the 142 significant SNPs were validated in a separate dataset from northern
UK and 15 SNPs were found significantly associated with outcome. The SNPs
also affected survival in the same way (increased/decreased), as in the
Swedish/Danish dataset. These 15 SNPs were then analysed again in 121
glioblastoma from the Swedish/Danish dataset with available clinical data
and all HR were adjusted for country, sex, and therapy. After this step, nine
polymorphisms were significantly associated with glioblastoma survival
(p<0.05). In the last step, a permutation test was conducted, leaving two
SNPs significantly associated with glioblastoma survival. Both the significant
SNPs were located in the RECQL4 gene, were in strong LD with each other,
and had a minor allele association with longer shorter. When looking at the
seven SNPs significant before the permutation step but not after, two were
located in the MSH2 gene, and five were located in the RAD51L1 gene. In
36
MSH2 the two SNPs were in strong LD with each other and in both SNPs
heterozygotes had a shorter survival than homozygotes for the major allele
(no cases homozygotes for the minor allele were found). The remaining five
polymorphisms significant before the permutation step were all found in
RAD51L1 and were in strong LD with each other, and four of the SNPs were
located in the same haplotype block. The four SNPs on the same haplotype
block had an almost identical distribution of cases between heterozygotes
(32 cases and an additional 3 cases for rs6573805) and homozygotes for the
minor allele (three cases). And three of the SNPs had an identical
distribution in all groups (homozygotes for the major allele, heterozygotes
and homozygotes for the minor allele).
37
Discussion
Paper I
Viruses have been found to play an important role in the development of
different types of cancers (85-87). In astrocytoma products from viruses
such as BK-virus, JC-virus, and Simian virus 40 have been detected in
tumour tissue, but no clear role in glioma development has been found (88,
89, 181). CMV is a common virus in the population that can be reactivated
throughout life and is present in several different parts of the body. CMV
gene products have been found in glioblastoma cells but not in the
surrounding normal brain tissue (90). It has been hypothesised that cancer
cells might activate the immune system which then tries to eliminate the
cells as they are viewed as non-self and, by doing so, creates an inflammation
in the area. CMV might be carried into the inflamed area by infected
monocytes/macrophages and become reactivated and transmitted to tumour
cells, making it hard for the immune system to detect them (182). CMV has
been studied for association with glioma and has an affinity for glial cells
which could allow the virus to affect, for example, tumour progression,
angiogenesis, production of growth factors, and tumour invasiveness (91,
182). In glioma cell lines, CMV has been found to activate telomerase, an
enzyme crucial for cellular immortalization (183). These findings indicate
that CMV might have an oncogenic function and could potentially play a role
in glioma development. However how big the influence of viral proteins are
in glioma progression is still debated (184). Adenovirus is a neurotropic virus
that has been detected in paediatric brain tumours (96). EBV has also been
associated with human cancer and is present in and an important risk factor
for CNS lymphoma in immunosuppressed patients (185). VZV is a
neurotropic virus that remains dormant in the neural ganglia after the
primary infection. VZV can be reactivated and cause Shingles. However
reactivation occurs rarely, approximately once every 20 years.
Previous studies on allergy and asthma support the hypothesis that the
immune system might play an important role in glioma development (43, 44,
46). IgG levels for CMV, EBV, Ad and VZV have previously been studied in
glioma, and an inverse association with glioma risk for having high IgG
levels for VZV has been found (186, 187). These studies were done on
prevalent cases with blood taken at or after diagnosis, making it a possibility
that the findings were due to a confounder, such as the tumour or
antitumour treatment or steroids. Because of this, we used prediagnostic
samples from three different bio banks to minimize the risk of disease or
38
treatment affecting our results. We found that high IgG levels for VZV were
present in a higher percentage of controls compared to cases, especially
when restricting the analyses to only include cases with at least two years
between blood sample and glioma diagnosis. This was done to further limit
the risk of subclinical disease affecting our results. For CMV, EBV, and Ad no
association was found. These findings are in line with previous findings on
IgG levels measured at diagnosis and might suggest that the immune system
plays a role in glioma development. It has been hypothesised that a strong
immune system might be more alert in detecting and destroying small
cancer clusters, which could be one possible explanation of our findings. As
we did not see any association of IgG for the other viral infections, a non-
specific IgG bound protection against glioma seems unlikely. One could
speculate that a specific association with the VZV IgG is more likely. When
comparing these viruses, VZV is the most neurotropic and remains dormant
in the neuroganglia and perhaps even in neurons in the brain, lifelong. One
could therefore speculate that the immune system could be protective
against glioma if VZV has a role in glioma genesis. The negative results for
the other viruses might also depend on the fact that these are common
viruses in the population and larger studies are needed for efficient power to
detect an association in immune response to the other viruses and glioma
risk. Another possibility is that the VZV has special characteristics or
displays viral products on the cell surface of the tumour cell, which might
lead to glioma cells being easier to detect by VZV antibodies compared to
antibodies for other viruses. Association with other viruses that might act as
a confounder is another possibility, but there are no known associations
between VZV and other viruses and IgG is a very virus-specific antibody.
The findings of viral genome and DNA in glioma might be associated in some
way to glioma genesis. However it is important to consider that the virus
might be drawn to the immunosuppressive environment in the tumour; that
is the tumour might be there before the virus and not the other way around.
Even so, viruses drawn to the tumour might be part of the progression of
glioma but not the sole cause. The immune system can be looked upon as a
separate factor in glioma risk, and, as viruses might have a causal
association, the immune system might have a protective role. These, of
course, are two factors that interact and patients that easily get infected or
who more easily get reactivation of these viruses might have a weaker
immune response and in that way be at higher risk for developing glioma.
Case-control studies
We used a case-control study design to evaluate the potential role of IgG
levels for viruses on glioma risk. This method is preferable when studying
39
rare diseases as one can take cases that have already developed the disease
and compare them with disease-free controls for the exposure of interest.
This is much less time consuming than following cohorts exposed to a
specific agent to see if they develop the disease. One problem with this type
of design is that there might be unknown factors (confounders) affecting the
results. To minimize this risk, cases are often matched to controls on
potential confounders such as age, sex, and area of residence.
Papers II and III
In paper II, we studied the association of polymorphisms in EGF and EGFR
and glioblastoma outcome by performing a comprehensive tagging of both
genes. In EGF, we analysed 30 SNPs and found seven SNPs associated with
glioblastoma outcome, all located in haplotype block four and all with a
minor allele association with poor prognosis. In EGF, previous studies have
been focused on a functional SNP in EGF (EGF 61 A/G), in which the minor
allele has been associated with higher EGF expression (100), which is known
to promote tumour angiogenesis and tumour progression (188). This
functional polymorphism is located on the same haplotype block as the seven
SNPs significantly associated with prognosis in our study and they are
therefore likely to be transmitted together.
No clear association with prognosis for the EGF 61 A/G has been found in
previous studies. In a small study of 42 glioblastoma with available
treatment data (surgery, radiotherapy, and chemotherapy) an association
with both poor prognosis and increased tumour recurrence was seen in
heterozygotes or homozygotes for the rare allele G compared with
homozygotes for the common allele A (101). In a study of 197 glioma cases an
association for the EGF 61 A/G and glioma risk was found. A separate
analysis was made for outcome in the glioblastoma cases (n=44), but no
association was found (98). In a study of 209 glioblastoma cases with
available treatment data, no association with either risk or prognosis was
found for the EGF 61 A/G, but some effects on progression-free survival were
seen (100).
The seven SNPs we found significantly associated with survival were located
on the same haplotype block, in strong LD with each other, and had a minor
allele association with poor survival, which makes it possible that this is a
true finding. As EGF 61 A/G is located on the same haplotype block, it is
likely that this polymorphism is the functional variant in the region and, as
EGF 61 A/G in some previous studies has been found to have a minor allele
association with poor prognosis, this could be in line with our results.
40
However, other studies have failed to detect an association with prognosis.
One reason for this could be that these are small studies and, in one study,
treatment data was not available, making it possible that the extent of
surgery or postoperative treatment significantly affected the results. Four of
the significant polymorphisms in our study were available for confirmation
in the MDACC dataset. However, no significant association with prognosis
was seen. When looking at the characteristics of the Swedish/Danish cohort
in comparison to the MDACC cohort, a difference in median age at diagnosis
and median survival time was found, with lower median age at diagnosis and
longer median survival time in the MDACC cohort (54 vs. 56 years and 15 vs.
12.8 months). As age is a known prognostic factor, we decided to also
validate only in cases over 50 years of age in the MDACC cohort and found a
non-significant trend towards minor allele association with survival. No
treatment data was available for the MDACC cohort, making it possible that
the extent of surgery or postoperative treatment affected the results. The
cases in the MDACC cohort were collected between 1990 and 2008, while the
cases in the Swedish/Danish cohort were collected between 2000 and 2004,
making it possible that a larger number of cases in the MDACC cohort
received standard treatment with concomitant radiotherapy and
Temozolomide as first-line treatment. This has proven to be an important
treatment, doubling median survival (3). As most of the cases in the
Swedish/Danish cohort were diagnosed before this finding, fewer patients
received Temozolomide or received it later in the treatment course,
compared with cases in the MDACC cohort. EGFR mutation status was not
available in either the Swedish/Danish cohort or the MDACC cohort, which
might be of importance as a mutated and constantly active receptor reduces
the impact of elevated EGF levels. Not all glioblastoma have mutated EGFR
and the polymorphisms in EGF might be stronger prognostic factors for
those with unmutated receptors. It would be interesting to study these SNPs
in a cohort with known mutation status.
In EGFR we analysed 89 tagging SNPs and found four associated with
glioblastoma outcome, two with poorer and two with better prognosis. None
of the SNPs were available for confirmation in the MDACC dataset.
Activation of EGFR starts a cascade of downstream events that play an
important role in cell proliferation, differentiation, growth, and survival. In
glioma, EGFR has been found to participate in tumour progression and to be
a prognostic factor in low-grade and anaplastic astrocytoma (44, 105). In up
to 50% of all glioma the receptor is amplified and 20-30 % of glioblastoma
with EGFR amplification have a constantly active receptor due to a deletion
of exon 2-7 (EGFRvIII). EGFR amplification has been studied for association
with glioma and glioblastoma without finding a clear association. One of the
polymorphisms with a minor allele association with poorer survival
41
(rs4947986) is located in the exon7/intron border and is close to the
common deletion and, therefore, one could speculate that this SNP or a near
by SNP could have a functional role and be of importance for glioma
progression. In a later study, this polymorphism was associated with glioma
risk (55), and recently two loci in EGFR have been associated with glioma
risk through large GWA studies (52-54, 189), raising further evidence that
EGFR is important in glioma progression. However, as 89 SNPs were tested
and four were associated with survival, this is very close to what we should
find by chance with a p-value of 0.05, and therefore the role of EGFR in
glioblastoma survival is still unclear.
VEGFR2 is one of three receptors in the VEGFR family. Upon activation, this
receptor promotes angiogenesis, proliferation, and cell survival. In
glioblastoma, inhibition of VEGFR2 has been shown to increase cell death
caused by ionizing radiation. This suggests that VEGFR2 might be a factor in
both glioma development and prognosis. Therefore, we analysed 27 tagging
SNPs in VEGFR2 and 17 tagging SNPs in VEGF for association with
glioblastoma survival. We found no significant polymorphisms in VEGF
associated with outcome. In VEGFR2 we found two SNPs significantly
associated with glioblastoma survival, which is near the amount of
significant variants we would expect to find with a significance level of
p=0.05 and neither of the two SNPs had a dose-response pattern that
associated outcome with the major or minor allele. One of the two SNPs was
available for confirmation in an independent dataset with no association
with survival in that cohort. Taken together, all these reservations indicate
that the results are most likely due to chance. In recent years, a monoclonal
antibody against VEGF, Bevacizumab, has shown promising results in
second-line treatment of glioblastoma. This gives some evidence that the
VEGF/VEGFR pathway is of importance in glioblastoma. One reason that we
failed to detect an association with outcome might be that polymorphisms in
these genes are not directly associated with outcome but have an association
with treatment outcome when treated with Bevacizumab; since our cases
were diagnosed in 2000 to 2004 none of them received this treatment.
Paper IV
We studied 1,458 polymorphisms in 136 DNA repair genes for association
with glioblastoma outcome. After confirmation in a separate dataset and
adjusting for age, sex, and therapy we found nine SNPs significantly
associated with glioblastoma outcome (p<0.05) located in MSH2, RAD51L1,
and RECQL4. After adjusting for multiple testing, two of the polymorphisms
remained significant, both located in RECQL4 and both with a minor allele
association with better prognosis. RECQL4 has been suggested to interact
42
with RAD51 and participate in repairing double-strand breaks after
radiation, while MSH2 is part of the mismatch repair pathway. RAD51L1 is a
member of the RAD51 family and is involved in the DNA repair of double
strand breaks.
In the nine SNPs significant before controlling for multiple testing, strong
LD was seen for SNPs located in the same gene and they also had a nearly
identical distribution of cases heterozygote and homozygote for the minor
and major allele. These also had a clear dose-response pattern with a minor
allele association with either better or poorer prognosis. In validating data in
a separate dataset, adjusting for treatment and multiple testing, we tried to
minimise the risk of false positive findings that multiple tests, small sample
sizes, and rare polymorphisms might render. In doing so, the risk of type 2
errors is important to consider. Polymorphisms that are important factors in
treatment outcome might be eliminated when adjusting for treatment
parameters and correcting for multiple testing with a 10,000 permutation
test is very stringent and might lead to false negative findings.
In a previous study on 590 primary glioblastoma, an association between
four genes involved in the DNA double-strand break repair pathway, LIG4,
BTBD2, HMGA2, and RTEL1, and glioblastoma survival was found (69).
In a study on low-grade glioma, the same 1,458 tagging SNPs in 136 DNA
repair genes were studied for association with outcome. A cohort of 81 low-
grade (grade II) and anaplastic (grade III) glioma was used and significant
findings on a p<0.05 level were analysed in a control cohort of 72 grade II
and III glioma. After confirmation, a total of eight SNPs in five DNA repair
genes were found (ATM, NEIL1, NEIL2 ERCC6, and RPA4) associated with
outcome. After adjusting for treatment, gender, age, and country, one
polymorphism in ERCC6 was found significantly associated with improved
glioma outcome (190).
Several studies have also found an association between different DNA-repair
gene polymorphisms and glioma risk (161-164).
Repair of damaged DNA is crucial for cell survival and gene integrity. This
knowledge is something we take advantage of in treatment of cancer. In
glioblastoma, the standard non-surgical treatment is radiotherapy, which
induces DNA double strand breaks, and concomitant Temozolomide, an
alkylating agent that methylates DNA. The tumour cells’ ability to repair
these damages is an important reason for treatment failure. For example,
MGMT can repair DNA damaged by methylation and MGMT silencing
enables an effective treatment and is an important prognostic factor in
glioblastoma (4). Polymorphisms in different DNA repair genes could
therefore be both a direct prognostic factor involved in progression of the
43
tumour, as well as associated with treatment failure. These pathways could
therefore be important targets for future drugs to optimize treatment
outcome.
DNA repair gene polymorphisms also seem to play an important part in
glioma susceptibility. One of the few known risk factors for glioma is ionizing
radiation and one could hypothesise that people with compromised DNA
repair pathways therefore would be at higher risk of developing glioma.
44
Conclusions Based on the results from these studies, the following conclusions can be
made:
I. High levels of IgG for varicella zoster virus decrease the risk of glioma;
however, IgG levels for adenovirus, cytomegalovirus, and Epstein-Barr virus are not associated with glioma risk.
II. Polymorphisms in EGF are associated with glioblastoma prognosis; however, the association of polymorphisms in EGFR with glioblastoma outcome is more uncertain.
III. Polymorphisms in VEGF and VEGFR2 are not associated with glioblastoma outcome.
IV. Polymorphisms in the DNA repair genes RAD51L1, MSH2, and RECQL4 are associated with glioblastoma outcome.
45
Future perspectives
The idea that a strong immune system could protect against cancer and,
more specifically, glioma is an interesting thought. A validation of the
findings of higher IgG levels for VZV in controls compared to cases would be
an important next step, and to do so in a similar way with blood samples
taken before diagnosis to limit disease and treatment influences on the
results. It would be valuable to do this in a larger cohort as these are
common viral infections therefore confirmation should also be done for IgG
for the viruses where we did not find a significant association to ensure that
the negative findings were not due to small sample sizes. If these are true
findings with only high IgG levels for VZV associated with glioma risk, it is
more likely that it is something specific with either the VZV or the immune
response against this virus and not the immune system as a whole that is
associated with glioma risk, making it a highly interesting area for future
studies.
In recent years, several large GWAS have studied polymorphisms in low-
penetrance genes and their association with glioma risk, and seven risk loci
have been identified. There are fewer studies on polymorphisms in low-
penetrance genes and association with glioma outcome. It would be
interesting to study low-penetrance genes in larger cohorts to find potential
polymorphisms associated with outcome in both low and high-grade glioma,
as robust findings might give an opportunity for prognostic markers and
targeted treatments. EGF/EGFR and DNA repair pathways are highly
interesting for such studies as polymorphisms associated with both risk and
outcome have been found. However, the variants active in glioma
progression might not necessarily be the ones important for outcome, and
when studying outcome, one must consider that the polymorphisms might
not associate directly with glioma/glioblastoma outcome but rather with
treatment outcome, especially in the case of polymorphisms in DNA repair
genes that might be involved in restoring the damage done after antitumour
treatment. They might therefore be potential targets for new treatments or to
enhance treatment effects and also possibly tools to find responders up front.
Although we found no association between VEGF and VEGFR2 and
glioblastoma outcome, these are very interesting genes that should be
studied further. It would be interesting to study these genes in patients
treated with Bevacizumab to see if polymorphisms are associated with
treatment outcome and could be tools to select patients with a high chance to
respond up front.
46
Historically, very little could be done to alter the natural course of the
disease in glioma and especially glioblastoma patients. As we learn more
about the aetiology and prognostic factors for glioma, we come closer to
finding new ways of detecting cases early and new targets for treatment. The
future in glioma treatment and prognosis is promising as many new
discoveries have been made in the aetiology and genetic pathways involved
in glioma development. I am excited to see what lies beyond the horizon for
these patients.
47
Acknowledgements
Det är så många som på olika sätt bidragit och stöttat mig i arbetet med avhandlingen och utan vars hjälp den aldrig hade blivit verklighet. Jag vill speciellt omnämna några;
Till Beatrice Melin, min huvudhandledare och stora förebild, jag förstår inte hur du hinner med allt, jag är säker på att ett dygn för dig måste innehålla fler timmar än för oss andra. Tack för att du alltid ställer upp med allt från kloka råd till leverans av strömkabel till dator, för att du stöttat mig men även utmanat mig, fått mig att gå utanför min ”comfort zone” och därmed växa i arbetet men även som människa.
Till Ulrika Anderson min biträdande handledare, tack för att du tålmodigt
lotsat mig igenom detta arbete, alltid funnits där när jag behövt fråga eller
diskutera något och för att du alltid ser till att alla siffror stämmer.
Björn Tavelin, du har varit ett ovärderligt stöd i all statistik. Tack för att du
alltid tar dig tid att svara på alla mer eller mindre intelligenta statistik och
SPSS frågor.
Carina Ahlgren, Monica Sandström, Sara Huggert Ranta och
Katrin Sundh som på olika sätt hjälpt till med det administrativa. Carina
för ditt oerhörda tålamod och vänliga påminnelser när jag än en gång varit
sen med att lämna in papper eller svara på mail, förlåt.
Barbro Widmark, tack för all hjälp med att jaga journalkopior.
Irene Eriksson, tack för hjälpen med analyserna till arbete 1 och för att du
tog så väl hand om mig och visade hur analyserna går till.
Ulf Hjalmars, tack för din ovärderliga hjälp med arbete 1 under min
mammaledighet och för att du lugnt, vänligt och med stor förståelse mötte
mig och min son, som just innan vårt första möte kring manuset hade fått sin
3 månaders vaccination, och som för ovanlighetens skull skrek oavbrutet och
kräktes på både mig och golvet. Din kommentar, att det inte var konstigt att
han mådde dåligt då han just genomgått livets första trauma, kommer jag
aldrig att glömma.
Carl Wibom, tack för allt jobb med arbete 4 och för alla gånger du på ett
enkelt och pedagogiskt sätt förklarat för mig helt obegripliga saker som t.ex.
permutation.
To all other co-authors, Roger Henriksson, Thomas Brännström,
Helle Broholm, Göran Hallmans, Per Juto, Patrik Rydén,
Christoffer Johansen, Helle Collartz-Laier, Sara Hepworth,
Patricia McKinney, Lara Bethke, Richard Houlston, Göran Wadell
Anne Tjönneland, Jytte Halkjaer, Jonas Manjer, Martin Almquist,
Yanhong Liu, Melissa Bondy, thank you for your work and valuable
comments in the process of writing the papers. Melissa Bondy, thank you
for valuable input on Table of Contents.
48
Thomas Brännström och Helle Broholm tack för att jag fick sitta med
när ni eftergranskade proverna och för att ni tog er tid att visa och förklara.
Göran Hallmans och Per Juto för att ni tog er tid att svara på mina
frågor.
Till arbetsgruppen och VVLL, Soma, Ulrika, Calle, Anna D, Ulf, Karin,
Anna N, Christina, Anne, Ingrid, Camilla och Florentine, för
intressanta diskussioner och för att jag får ta del av ert fantastiska arbete.
Tack Soma för trevliga och stärkande luncher, lycka till!
Till alla mina kollegor och alla andra arbetskamrater på Cancercentrum, tack
för att ni gör det så roligt att gå till jobbet. Till Elisabeth Karlsson och
Göran Edbom verksamhetschef och biträdande verksamhetschef (tidigare
tvärt om) på Cancercentrum, för att jag fått tid till att göra detta arbete och
för att ni alltid uppmuntrar till forskning.
Till specialisterna i lymfomgruppen, Martin, Ann-Sofie, Karin, för att ni
är fantastiska förebilder och för att jag får tillhöra gruppen även om jag
varken springer fort eller långt (men jag har lärt mig gå ganska snabbt).
Till Ann-Sofie Johansson min kliniska handledare, när jag blir stor vill
jag bli som du!
Till mina underbara vänner och rumskamrater Anna, Berivan, Maja, och
Sussi, tack för att ni alltid finns där, för att ni stöttar och peppar när det går
tungt, för fantastisk kurdisk mat och hummus till frysen, kläder till barnen,
sena stockolmskvällar och personliga pepparkakor, för delat TV och
fruktintresse, för mysiga stugkvällar, goda middagar och mycket skratt. Ni är
ovärderliga!
Tack till alla våra vänner för livets alla trevligheter som så väl behövs under
ett sådant här arbete. Till gamla Vännäs/Berghem gänget Klas och Helena,
Sara och Stefan, Lina och André, Joakim och Johanna, Karolina och
Johan, för att även om det nu blir mer sällan är det alltid lika kul när vi
träffas! Till Joakim och Johanna för mysiga och goda middagar med ännu
godare viner. Till Björn och Cicci, för att ni är så trevliga och för att det
alltid är lika roligt och avslappnat när vi träffas. Till min äldsta och bästa
kompis Lillemor och hennes familj, för att du känner mig, för att oavsett
hur länge det går mellan vi träffas så är allt precis som vanligt och för att ni
tagit ert förnuft till fånga och flyttat hem!
Till Jesper och Theresé samt Birgitta och Patrik, för roliga
syskonmiddagar med sång och dans.
Till Lasse och Ing-Britt mina underbara svärföräldrar, tack för att ni alltid
ställer upp med stort och smått. Tack för hämtning på dagis, för el-dragning,
för limpor och saft och tack för mysiga fiskehelger.
Till mina syskon, Sandra, Kristoffer och Anna. För att vi alltid har så
roligt, för att jag inte känner mig lastgammal (annat än när jag ska
programmera videon) och för att jag vet att ni alltid finns där.
49
Sandra, du är en av dem som fått stå ut med mig och min oro denna höst.
Tack för kloka ord, för hjälp med kläder och fest, många samtal och mail,
tabletter och onsdagskillar, för att du alltid förstår precis och vet vad du ska
säga. Hoppas jag kan vara till samma glädje när det är din tur! Till Sandras
Jens, du är en så fantastisk människa att umgås med! Alltid glad och en
riktig busfarbror. Tack för hjälp med att ordna band.
Till alla mina älskade föräldrar, mamma och Göran och pappa och
Kristina. Tack mamma och pappa för att ni alltid trott på mig och stöttat
mig i allt. Tack för att vi alla är en stor familj, för gemensamma firanden,
middagar och mysiga jular. Tack mamma för att du alltid ställer upp på alla
sätt, för att du hjälper till med barnen oavsett om det gäller att hämta på
dagis, följa på barntimmar, ta en extra sjukdag eller vara barnvakt. Du är så
otroligt omtänksam och kärleksfull. Tack pappa för att du hjälpt mig hitta
lugnet i många stunder när pressen tagit över och för att du är värdens bästa
Offa. Till Göran och Kristina mina extra föräldrar, tack för att ni alltid
ställt upp och alltid finns där för mig och tack för all omsorg ni ger barnen.
Tack Göran för att barnen frågar sina kompisar vad deras Göran heter.
Till mormor Ruth och morfar Harald som varit en så stor del av min
uppväxt och trygghet men som inte finns hos mig längre. Jag saknar er!
Till min familj. Till Jonas min bättre hälft och själsfrände. För att du är du,
du känner mig bäst av alla, stöttar mig och står ut med alla mina sidor, du är
fantastisk. Tack för att du finns, jag älskar dig!
Elsa och Vilgot mina små älsklingar och livets mening. Ni är det bästa som
hänt mig och påminner mig om vad som verkligen är viktigt!
Till alla patienter som ställer upp i studier och gör sådant här möjligt. Till
alla patienter och anhöriga som jag dagligen möter, ni gör att jag har
världens bästa jobb!
Tack till ALF, Cancerforskningsfonden och vetenskapsrådet som
bidragit med finansiellt stöd till detta arbete.
50
References
1. Ohgaki H, Kleihues P. Genetic pathways to primary and secondary glioblastoma. Am J Pathol2007 May;170(5):1445-53.
2. Nobusawa S, Watanabe T, Kleihues P, Ohgaki H. IDH1 mutations as molecular signature and predictive factor of secondary
glioblastomas. Clin Cancer Res2009 Oct 1;15(19):6002-7.
3. Stupp R, Dietrich PY, Ostermann Kraljevic S, Pica A, Maillard I, Maeder P, et al. Promising survival for patients with newly diagnosed
glioblastoma multiforme treated with concomitant radiation plus temozolomide followed by adjuvant temozolomide. J Clin Oncol2002 Mar 1;20(5):1375-82.
4. Hegi ME, Diserens AC, Gorlia T, Hamou MF, de Tribolet N, Weller M, et al. MGMT gene silencing and benefit from temozolomide in glioblastoma. N Engl J Med2005 Mar 10;352(10):997-1003.
5. Friedman HS, Prados MD, Wen PY, Mikkelsen T, Schiff D, Abrey LE, et al. Bevacizumab alone and in combination with irinotecan in
recurrent glioblastoma. J Clin Oncol2009 Oct 1;27(28):4733-40.
6. Lee M, Wrensch M, Miike R. Dietary and tobacco risk factors for adult onset glioma in the San Francisco Bay Area (California, USA).
Cancer Causes Control1997 Jan;8(1):13-24.
7. Bunin GR, Kuijten RR, Boesel CP, Buckley JD, Meadows AT. Maternal diet and risk of astrocytic glioma in children: a report from the
Childrens Cancer Group (United States and Canada). Cancer Causes Control1994 Mar;5(2):177-87.
8. Preston-Martin S, Mack W, Henderson BE. Risk factors for gliomas and meningiomas in males in Los Angeles County. Cancer Res1989 Nov 1;49(21):6137-43.
9. Rodvall Y, Ahlbom A, Pershagen G, Nylander M, Spannare B. Dental radiography after age 25 years, amalgam fillings and tumours of the central nervous system. Oral Oncol1998 Jul;34(4):265-9.
10. Feychting M, Forssen U, Floderus B. Occupational and residential magnetic field exposure and leukemia and central nervous system tumors. Epidemiology1997 Jul;8(4):384-9.
51
11. Floderus B, Tornqvist S, Stenlund C. Incidence of selected cancers in Swedish railway workers, 1961-79. Cancer Causes Control1994 Mar;5(2):189-94.
12. Mack W, Preston-Martin S, Peters JM. Astrocytoma risk related to job exposure to electric and magnetic fields. Bioelectromagnetics1991;12(1):57-66.
13. Inskip PD, Mellemkjaer L, Gridley G, Olsen JH. Incidence of intracranial tumors following hospitalization for head injuries (Denmark). Cancer Causes Control1998 Jan;9(1):109-16.
14. Wrensch M, Miike R, Lee M, Neuhaus J. Are prior head injuries or diagnostic X-rays associated with glioma in adults? The effects of control selection bias. Neuroepidemiology2000 Sep-Oct;19(5):234-44.
15. Preston-Martin S, Pogoda JM, Schlehofer B, Blettner M, Howe GR, Ryan P, et al. An international case-control study of adult glioma
and meningioma: the role of head trauma. Int J Epidemiol1998 Aug;27(4):579-86.
16. Schlehofer B, Blettner M, Preston-Martin S, Niehoff D,
Wahrendorf J, Arslan A, et al. Role of medical history in brain tumour development. Results from the international adult brain tumour study. Int J Cancer1999 Jul 19;82(2):155-60.
17. Wrensch M, Lee M, Miike R, Newman B, Barger G, Davis R, et al. Familial and personal medical history of cancer and nervous system conditions among adults with glioma and controls. Am J Epidemiol1997 Apr
1;145(7):581-93.
18. Shirts SB, Annegers JF, Hauser WA, Kurland LT. Cancer incidence in a cohort of patients with seizure disorders. J Natl Cancer Inst1986 Jul;77(1):83-7.
19. Lachance DH, Yang P, Johnson DR, Decker PA, Kollmeyer
TM, McCoy LS, et al. Associations of high-grade glioma with glioma risk alleles and histories of allergy and smoking. Am J Epidemiol Sep 1;174(5):574-81.
20. Mandelzweig L, Novikov I, Sadetzki S. Smoking and risk of glioma: a meta-analysis. Cancer Causes Control2009 Dec;20(10):1927-38.
21. Cardis E, Richardson L, Deltour I, Armstrong B, Feychting M,
Johansen C, et al. The INTERPHONE study: design, epidemiological
52
methods, and description of the study population. Eur J Epidemiol2007;22(9):647-64.
22. Brain tumour risk in relation to mobile telephone use: results
of the INTERPHONE international case-control study. Int J Epidemiol Jun;39(3):675-94.
23. Louis DN, von Deimling A. Hereditary tumor syndromes of
the nervous system: overview and rare syndromes. Brain Pathol1995 Apr;5(2):145-51.
24. Melean G, Sestini R, Ammannati F, Papi L. Genetic insights into familial tumors of the nervous system. Am J Med Genet C Semin Med Genet2004 Aug 15;129C(1):74-84.
25. Malmer B, Haraldsson S, Einarsdottir E, Lindgren P, Holmberg D. Homozygosity mapping of familial glioma in Northern Sweden. Acta Oncol2005;44(2):114-9.
26. Malmer B, Henriksson R, Gronberg H. Different aetiology of familial low-grade and high-grade glioma? A nationwide cohort study of familial glioma. Neuroepidemiology2002 Nov-Dec;21(6):279-86.
27. Malmer B, Adatto P, Armstrong G, Barnholtz-Sloan J, Bernstein JL, Claus E, et al. GLIOGENE an International Consortium to Understand Familial Glioma. Cancer Epidemiol Biomarkers Prev2007
Sep;16(9):1730-4.
28. Ron E, Modan B, Boice JD, Jr., Alfandary E, Stovall M,
Chetrit A, et al. Tumors of the brain and nervous system after radiotherapy in childhood. N Engl J Med1988 Oct 20;319(16):1033-9.
29. Werner A, Modan B, Davidoff D. Doses to brain, skull and
thyroid, following x-ray therapy for Tinea capitis. Phys Med Biol1968 Apr;13(2):247-58.
30. Sadetzki S, Chetrit A, Freedman L, Stovall M, Modan B, Novikov I. Long-term follow-up for brain tumor development after childhood exposure to ionizing radiation for tinea capitis. Radiat Res2005 Apr;163(4):424-32.
31. Preston DL, Ron E, Yonehara S, Kobuke T, Fujii H, Kishikawa M, et al. Tumors of the nervous system and pituitary gland associated with atomic bomb radiation exposure. J Natl Cancer Inst2002 Oct 16;94(20):1555-63.
53
32. Sadamori N, Shibata S, Mine M, Miyazaki H, Miyake H, Kurihara M, et al. Incidence of intracranial meningiomas in Nagasaki atomic-bomb survivors. Int J Cancer1996 Jul 29;67(3):318-22.
33. Shibata S, Sadamori N, Mine M, Sekine I. Intracranial meningiomas among Nagasaki atomic bomb survivors. Lancet1994 Dec 24-31;344(8939-8940):1770.
34. Armstrong GT, Stovall M, Robison LL. Long-term effects of radiation exposure among adult survivors of childhood cancer: results from the childhood cancer survivor study. Radiat Res Dec;174(6):840-50.
35. Schwartzbaum J, Jonsson F, Ahlbom A, Preston-Martin S, Lonn S, Soderberg KC, et al. Cohort studies of association between self-reported allergic conditions, immune-related diagnoses and glioma and
meningioma risk. Int J Cancer2003 Sep 1;106(3):423-8.
36. Hochberg F, Toniolo P, Cole P, Salcman M. Nonoccupational
risk indicators of glioblastoma in adults. J Neurooncol1990 Feb;8(1):55-60.
37. Ryan P, Lee MW, North B, McMichael AJ. Risk factors for tumors of the brain and meninges: results from the Adelaide Adult Brain
Tumor Study. Int J Cancer1992 Apr 22;51(1):20-7.
38. Schlehofer B, Blettner M, Becker N, Martinsohn C, Wahrendorf J. Medical risk factors and the development of brain tumors.
Cancer1992 May 15;69(10):2541-7.
39. Cicuttini FM, Hurley SF, Forbes A, Donnan GA, Salzberg M,
Giles GG, et al. Association of adult glioma with medical conditions, family and reproductive history. Int J Cancer1997 Apr 10;71(2):203-7.
40. Wiemels JL, Wiencke JK, Sison JD, Miike R, McMillan A,
Wrensch M. History of allergies among adults with glioma and controls. Int J Cancer2002 Apr 1;98(4):609-15.
41. Brenner AV, Linet MS, Fine HA, Shapiro WR, Selker RG, Black PM, et al. History of allergies and autoimmune diseases and risk of brain tumors in adults. Int J Cancer2002 May 10;99(2):252-9.
42. Schoemaker MJ, Swerdlow AJ, Hepworth SJ, McKinney PA, van Tongeren M, Muir KR. History of allergies and risk of glioma in adults. Int J Cancer2006 Nov 1;119(9):2165-72.
54
43. Wiemels JL, Wiencke JK, Patoka J, Moghadassi M, Chew T, McMillan A, et al. Reduced immunoglobulin E and allergy among adults with glioma compared with controls. Cancer Res2004 Nov 15;64(22):8468-73.
44. Wrensch M, Wiencke JK, Wiemels J, Miike R, Patoka J, Moghadassi M, et al. Serum IgE, tumor epidermal growth factor receptor expression, and inherited polymorphisms associated with glioma survival. Cancer Res2006 Apr 15;66(8):4531-41.
45. Calboli FC, Cox DG, Buring JE, Gaziano JM, Ma J, Stampfer M, et al. Prediagnostic plasma IgE levels and risk of adult glioma in four
prospective cohort studies. J Natl Cancer Inst Nov 2;103(21):1588-95.
46. Schlehofer B, Siegmund B, Linseisen J, Schuz J, Rohrmann S, Becker S, et al. Primary brain tumours and specific serum immunoglobulin
E: a case-control study nested in the European Prospective Investigation into Cancer and Nutrition cohort. Allergy Nov;66(11):1434-41.
47. Schwartzbaum J, Ding B, Johannesen TB, Osnes LT, Karavodin L, Ahlbom A, et al. Association Between Prediagnostic IgE Levels and Risk of Glioma. J Natl Cancer Inst Aug 1;104(16):1251-9.
48. Brenner AV, Butler MA, Wang SS, Ruder AM, Rothman N, Schulte PA, et al. Single-nucleotide polymorphisms in selected cytokine genes and risk of adult glioma. Carcinogenesis2007 Dec;28(12):2543-7.
49. Schwartzbaum J, Ahlbom A, Malmer B, Lonn S, Brookes AJ, Doss H, et al. Polymorphisms associated with asthma are inversely related to glioblastoma multiforme. Cancer Res2005 Jul 15;65(14):6459-65.
50. Schwartzbaum JA, Ahlbom A, Lonn S, Malmer B, Wigertz A, Auvinen A, et al. An international case-control study of interleukin-4Ralpha, interleukin-13, and cyclooxygenase-2 polymorphisms and glioblastoma risk. Cancer Epidemiol Biomarkers Prev2007 Nov;16(11):2448-54.
51. Wiemels JL, Wiencke JK, Kelsey KT, Moghadassi M, Rice T,
Urayama KY, et al. Allergy-related polymorphisms influence glioma status and serum IgE levels. Cancer Epidemiol Biomarkers Prev2007 Jun;16(6):1229-35.
52. Shete S, Hosking FJ, Robertson LB, Dobbins SE, Sanson M, Malmer B, et al. Genome-wide association study identifies five susceptibility loci for glioma. Nat Genet2009 Aug;41(8):899-904.
55
53. Wrensch M, Jenkins RB, Chang JS, Yeh RF, Xiao Y, Decker PA, et al. Variants in the CDKN2B and RTEL1 regions are associated with high-grade glioma susceptibility. Nat Genet2009 Aug;41(8):905-8.
54. Sanson M, Hosking FJ, Shete S, Zelenika D, Dobbins SE, Ma Y, et al. Chromosome 7p11.2 (EGFR) variation influences glioma risk. Hum Mol Genet Jul 15;20(14):2897-904.
55. Andersson U, Schwartzbaum J, Wiklund F, Sjostrom S, Liu Y, Tsavachidis S, et al. A comprehensive study of the association between the EGFR and ERBB2 genes and glioma risk. Acta Oncol Aug;49(6):767-75.
56. Schwartzbaum JA, Xiao Y, Liu Y, Tsavachidis S, Berger MS, Bondy ML, et al. Inherited variation in immune genes and pathways and glioblastoma risk. Carcinogenesis Oct;31(10):1770-7.
57. Kiemeney LA, Thorlacius S, Sulem P, Geller F, Aben KK, Stacey SN, et al. Sequence variant on 8q24 confers susceptibility to urinary
bladder cancer. Nat Genet2008 Nov;40(11):1307-12.
58. Tomlinson I, Webb E, Carvajal-Carmona L, Broderick P, Kemp Z, Spain S, et al. A genome-wide association scan of tag SNPs
identifies a susceptibility variant for colorectal cancer at 8q24.21. Nat Genet2007 Aug;39(8):984-8.
59. Yeager M, Orr N, Hayes RB, Jacobs KB, Kraft P, Wacholder S,
et al. Genome-wide association study of prostate cancer identifies a second risk locus at 8q24. Nat Genet2007 May;39(5):645-9.
60. Easton DF, Pooley KA, Dunning AM, Pharoah PD, Thompson D, Ballinger DG, et al. Genome-wide association study identifies novel breast cancer susceptibility loci. Nature2007 Jun 28;447(7148):1087-93.
61. Simon M, Hosking FJ, Marie Y, Gousias K, Boisselier B, Carpentier C, et al. Genetic risk profiles identify different molecular etiologies for glioma. Clin Cancer Res Nov 1;16(21):5252-9.
62. Yan H, Parsons DW, Jin G, McLendon R, Rasheed BA, Yuan W, et al. IDH1 and IDH2 mutations in gliomas. N Engl J Med2009 Feb 19;360(8):765-73.
63. Zhou QL, Jiang ZY, Mabardy AS, Del Campo CM, Lambright DG, Holik J, et al. A novel pleckstrin homology domain-containing protein enhances insulin-stimulated Akt phosphorylation and GLUT4 translocation in adipocytes. J Biol Chem Sep 3;285(36):27581-9.
56
64. Choe G, Horvath S, Cloughesy TF, Crosby K, Seligson D, Palotie A, et al. Analysis of the phosphatidylinositol 3'-kinase signaling pathway in glioblastoma patients in vivo. Cancer Res2003 Jun 1;63(11):2742-6.
65. Sonoda Y, Ozawa T, Aldape KD, Deen DF, Berger MS, Pieper RO. Akt pathway activation converts anaplastic astrocytoma to glioblastoma multiforme in a human astrocyte model of glioma. Cancer Res2001 Sep 15;61(18):6674-8.
66. Jenkins RB, Wrensch MR, Johnson D, Fridley BL, Decker PA,
Xiao Y, et al. Distinct germ line polymorphisms underlie glioma morphologic heterogeneity. Cancer Genet Jan;204(1):13-8.
67. Shay JW, Bacchetti S. A survey of telomerase activity in
human cancer. Eur J Cancer1997 Apr;33(5):787-91.
68. Melin BS, Nordfjall K, Andersson U, Roos G. hTERT cancer
risk genotypes are associated with telomere length. Genet Epidemiol May;36(4):368-72.
69. Liu Y, Shete S, Etzel CJ, Scheurer M, Alexiou G, Armstrong G,
et al. Polymorphisms of LIG4, BTBD2, HMGA2, and RTEL1 genes involved in the double-strand break repair pathway predict glioblastoma survival. J Clin Oncol May 10;28(14):2467-74.
70. Stacey SN, Sulem P, Jonasdottir A, Masson G, Gudmundsson J, Gudbjartsson DF, et al. A germline variant in the TP53 polyadenylation signal confers cancer susceptibility. Nat Genet Nov;43(11):1098-103.
71. Jenkins RB, Xiao Y, Sicotte H, Decker PA, Kollmeyer TM, Hansen HM, et al. A low-frequency variant at 8q24.21 is strongly associated with risk of oligodendroglial tumors and astrocytomas with IDH1 or IDH2 mutation. Nat Genet Oct;44(10):1122-5.
72. Siker ML, Wang M, Porter K, Nelson DF, Curran WJ,
Michalski JM, et al. Age as an independent prognostic factor in patients with glioblastoma: a Radiation Therapy Oncology Group and American College of Surgeons National Cancer Data Base comparison. J Neurooncol Aug;104(1):351-6.
73. Curran WJ, Jr., Scott CB, Horton J, Nelson JS, Weinstein AS, Fischbach AJ, et al. Recursive partitioning analysis of prognostic factors in three Radiation Therapy Oncology Group malignant glioma trials. J Natl Cancer Inst1993 May 5;85(9):704-10.
57
74. Ammirati M, Vick N, Liao YL, Ciric I, Mikhael M. Effect of the extent of surgical resection on survival and quality of life in patients with supratentorial glioblastomas and anaplastic astrocytomas. Neurosurgery1987 Aug;21(2):201-6.
75. Hentschel SJ, Sawaya R. Optimizing outcomes with maximal surgical resection of malignant gliomas. Cancer Control2003 Mar-Apr;10(2):109-14.
76. Esteller M, Garcia-Foncillas J, Andion E, Goodman SN, Hidalgo OF, Vanaclocha V, et al. Inactivation of the DNA-repair gene MGMT
and the clinical response of gliomas to alkylating agents. N Engl J Med2000 Nov 9;343(19):1350-4.
77. Vredenburgh JJ, Desjardins A, Herndon JE, 2nd, Marcello J,
Reardon DA, Quinn JA, et al. Bevacizumab plus irinotecan in recurrent glioblastoma multiforme. J Clin Oncol2007 Oct 20;25(30):4722-9.
78. Smith JS, Perry A, Borell TJ, Lee HK, O'Fallon J, Hosek SM, et al. Alterations of chromosome arms 1p and 19q as predictors of survival in oligodendrogliomas, astrocytomas, and mixed oligoastrocytomas. J Clin Oncol2000 Feb;18(3):636-45.
79. Nagane M, Lin H, Cavenee WK, Huang HJ. Aberrant receptor signaling in human malignant gliomas: mechanisms and therapeutic implications. Cancer Lett2001 Jan;162 Suppl:S17-S21.
80. Aldape KD, Ballman K, Furth A, Buckner JC, Giannini C, Burger PC, et al. Immunohistochemical detection of EGFRvIII in high
malignancy grade astrocytomas and evaluation of prognostic significance. J Neuropathol Exp Neurol2004 Jul;63(7):700-7.
81. Kinchington PR, St Leger AJ, Guedon JM, Hendricks RL.
Herpes simplex virus and varicella zoster virus, the house guests who never leave. Herpesviridae Jun 12;3(1):5.
82. Cupelli K, Stehle T. Viral attachment strategies: the many faces of adenoviruses. Curr Opin Virol Aug;1(2):84-91.
83. Delves PJ, Roitt IM. The immune system. First of two parts. N
Engl J Med2000 Jul 6;343(1):37-49.
84. Delves PJ, Roitt IM. The immune system. Second of two parts. N Engl J Med2000 Jul 13;343(2):108-17.
58
85. Bosch FX, Manos MM, Munoz N, Sherman M, Jansen AM, Peto J, et al. Prevalence of human papillomavirus in cervical cancer: a worldwide perspective. International biological study on cervical cancer (IBSCC) Study Group. J Natl Cancer Inst1995 Jun 7;87(11):796-802.
86. D'Souza G, Kreimer AR, Viscidi R, Pawlita M, Fakhry C, Koch WM, et al. Case-control study of human papillomavirus and oropharyngeal cancer. N Engl J Med2007 May 10;356(19):1944-56.
87. Maeda E, Akahane M, Kiryu S, Kato N, Yoshikawa T, Hayashi N, et al. Spectrum of Epstein-Barr virus-related diseases: a pictorial review.
Jpn J Radiol2009 Jan;27(1):4-19.
88. Del Valle L, Gordon J, Assimakopoulou M, Enam S, Geddes JF, Varakis JN, et al. Detection of JC virus DNA sequences and expression of
the viral regulatory protein T-antigen in tumors of the central nervous system. Cancer Res2001 May 15;61(10):4287-93.
89. Huang H, Reis R, Yonekawa Y, Lopes JM, Kleihues P, Ohgaki H. Identification in human brain tumors of DNA sequences specific for SV40 large T antigen. Brain Pathol1999 Jan;9(1):33-42.
90. Cobbs CS, Harkins L, Samanta M, Gillespie GY, Bharara S, King PH, et al. Human cytomegalovirus infection and expression in human malignant glioma. Cancer Res2002 Jun 15;62(12):3347-50.
91. Cobbs CS, Soroceanu L, Denham S, Zhang W, Britt WJ, Pieper R, et al. Human cytomegalovirus induces cellular tyrosine kinase signaling and promotes glioma cell invasiveness. J Neurooncol2007 Dec;85(3):271-80.
92. Scheurer ME, Bondy ML, Aldape KD, Albrecht T, El-Zein R. Detection of human cytomegalovirus in different histological types of gliomas. Acta Neuropathol2008 Jul;116(1):79-86.
93. Sabatier J, Uro-Coste E, Pommepuy I, Labrousse F, Allart S, Tremoulet M, et al. Detection of human cytomegalovirus genome and gene
products in central nervous system tumours. Br J Cancer2005 Feb 28;92(4):747-50.
94. Lau SK, Chen YY, Chen WG, Diamond DJ, Mamelak AN, Zaia
JA, et al. Lack of association of cytomegalovirus with human brain tumors. Mod Pathol2005 Jun;18(6):838-43.
95. Cobbs CS, Soroceanu L, Denham S, Zhang W, Kraus MH.
Modulation of oncogenic phenotype in human glioma cells by
59
cytomegalovirus IE1-mediated mitogenicity. Cancer Res2008 Feb 1;68(3):724-30.
96. Kosulin K, Haberler C, Hainfellner JA, Amann G, Lang S,
Lion T. Investigation of adenovirus occurrence in pediatric tumor entities. J Virol2007 Jul;81(14):7629-35.
97. Yarden Y, Sliwkowski MX. Untangling the ErbB signalling
network. Nat Rev Mol Cell Biol2001 Feb;2(2):127-37.
98. Costa BM, Ferreira P, Costa S, Canedo P, Oliveira P, Silva A, et al. Association between functional EGF+61 polymorphism and glioma risk. Clin Cancer Res2007 May 1;13(9):2621-6.
99. Pinto GR, Yoshioka FK, Clara CA, Santos MJ, Almeida JR,
Burbano RR, et al. Association study of an epidermal growth factor gene functional polymorphism with the risk and prognosis of gliomas in Brazil. Int J Biol Markers2009 Oct-Dec;24(4):277-81.
100. Vauleon E, Auger N, Benouaich-Amiel A, Laigle-Donadey F, Kaloshi G, Lejeune J, et al. The 61 A/G EGF polymorphism is functional but is neither a prognostic marker nor a risk factor for glioblastoma. Cancer
Genet Cytogenet2007 Jan 1;172(1):33-7.
101. Bhowmick DA, Zhuang Z, Wait SD, Weil RJ. A functional polymorphism in the EGF gene is found with increased frequency in
glioblastoma multiforme patients and is associated with more aggressive disease. Cancer Res2004 Feb 15;64(4):1220-3.
102. Rich JN, Hans C, Jones B, Iversen ES, McLendon RE, Rasheed BK, et al. Gene expression profiling and genetic markers in glioblastoma survival. Cancer Res2005 May 15;65(10):4051-8.
103. Ohgaki H, Dessen P, Jourde B, Horstmann S, Nishikawa T, Di Patre PL, et al. Genetic pathways to glioblastoma: a population-based study. Cancer Res2004 Oct 1;64(19):6892-9.
104. Krex D, Klink B, Hartmann C, von Deimling A, Pietsch T, Simon M, et al. Long-term survival with glioblastoma multiforme. Brain2007 Oct;130(Pt 10):2596-606.
105. Andersson U, Guo D, Malmer B, Bergenheim AT, Brannstrom T, Hedman H, et al. Epidermal growth factor receptor family (EGFR, ErbB2-4) in gliomas and meningiomas. Acta Neuropathol2004 Aug;108(2):135-42.
60
106. Rahimi N, Dayanir V, Lashkari K. Receptor chimeras indicate that the vascular endothelial growth factor receptor-1 (VEGFR-1) modulates mitogenic activity of VEGFR-2 in endothelial cells. J Biol Chem2000 Jun 2;275(22):16986-92.
107. Herold-Mende C, Steiner HH, Andl T, Riede D, Buttler A, Reisser C, et al. Expression and functional significance of vascular endothelial growth factor receptors in human tumor cells. Lab Invest1999 Dec;79(12):1573-82.
108. Knizetova P, Ehrmann J, Hlobilkova A, Vancova I, Kalita O,
Kolar Z, et al. Autocrine regulation of glioblastoma cell cycle progression, viability and radioresistance through the VEGF-VEGFR2 (KDR) interplay. Cell Cycle2008 Aug 15;7(16):2553-61.
109. Schmidt NO, Westphal M, Hagel C, Ergun S, Stavrou D, Rosen EM, et al. Levels of vascular endothelial growth factor, hepatocyte growth factor/scatter factor and basic fibroblast growth factor in human
gliomas and their relation to angiogenesis. Int J Cancer1999 Feb 19;84(1):10-8.
110. Zhou YH, Tan F, Hess KR, Yung WK. The expression of PAX6, PTEN, vascular endothelial growth factor, and epidermal growth factor receptor in gliomas: relationship to tumor grade and survival. Clin Cancer Res2003 Aug 15;9(9):3369-75.
111. Plate KH, Breier G, Weich HA, Risau W. Vascular endothelial growth factor is a potential tumour angiogenesis factor in human gliomas in vivo. Nature1992 Oct 29;359(6398):845-8.
112. Rampling R, Cruickshank G, Lewis AD, Fitzsimmons SA, Workman P. Direct measurement of pO2 distribution and bioreductive enzymes in human malignant brain tumors. Int J Radiat Oncol Biol Phys1994 Jun 15;29(3):427-31.
113. Winkler F, Kozin SV, Tong RT, Chae SS, Booth MF,
Garkavtsev I, et al. Kinetics of vascular normalization by VEGFR2 blockade governs brain tumor response to radiation: role of oxygenation, angiopoietin-1, and matrix metalloproteinases. Cancer Cell2004 Dec;6(6):553-63.
114. Gorski DH, Beckett MA, Jaskowiak NT, Calvin DP, Mauceri HJ, Salloum RM, et al. Blockage of the vascular endothelial growth factor stress response increases the antitumor effects of ionizing radiation. Cancer Res1999 Jul 15;59(14):3374-8.
61
115. Solberg TD, Nearman J, Mullins J, Li S, Baranowska-Kortylewicz J. Correlation between tumor growth delay and expression of cancer and host VEGF, VEGFR2, and osteopontin in response to radiotherapy. Int J Radiat Oncol Biol Phys2008 Nov 1;72(3):918-26.
116. Jain RK, Duda DG, Clark JW, Loeffler JS. Lessons from phase III clinical trials on anti-VEGF therapy for cancer. Nat Clin Pract Oncol2006 Jan;3(1):24-40.
117. Galli R, Binda E, Orfanelli U, Cipelletti B, Gritti A, De Vitis S, et al. Isolation and characterization of tumorigenic, stem-like neural
precursors from human glioblastoma. Cancer Res2004 Oct 1;64(19):7011-21.
118. Singh SK, Clarke ID, Terasaki M, Bonn VE, Hawkins C, Squire J, et al. Identification of a cancer stem cell in human brain tumors. Cancer
Res2003 Sep 15;63(18):5821-8.
119. Folkins C, Shaked Y, Man S, Tang T, Lee CR, Zhu Z, et al.
Glioma tumor stem-like cells promote tumor angiogenesis and vasculogenesis via vascular endothelial growth factor and stromal-derived factor 1. Cancer Res2009 Sep 15;69(18):7243-51.
120. Flynn JR, Wang L, Gillespie DL, Stoddard GJ, Reid JK, Owens J, et al. Hypoxia-regulated protein expression, patient characteristics, and preoperative imaging as predictors of survival in adults with glioblastoma multiforme. Cancer2008 Sep 1;113(5):1032-42.
121. Oehring RD, Miletic M, Valter MM, Pietsch T, Neumann J, Fimmers R, et al. Vascular endothelial growth factor (VEGF) in astrocytic
gliomas--a prognostic factor? J Neurooncol1999;45(2):117-25.
122. Jo J, Schiff D, Purow B. Angiogenic inhibition in high-grade gliomas: past, present and future. Expert Rev Neurother Jun;12(6):733-47.
123. Grzmil M, Hemmings BA. Deregulated signalling networks in human brain tumours. Biochim Biophys Acta Mar;1804(3):476-83.
124. Parikh SS, Mol CD, Slupphaug G, Bharati S, Krokan HE, Tainer JA. Base excision repair initiation revealed by crystal structures and binding kinetics of human uracil-DNA glycosylase with DNA. EMBO J1998
Sep 1;17(17):5214-26.
125. Frosina G, Fortini P, Rossi O, Carrozzino F, Raspaglio G, Cox LS, et al. Two pathways for base excision repair in mammalian cells. J Biol Chem1996 Apr 19;271(16):9573-8.
62
126. Matsumoto Y, Kim K, Bogenhagen DF. Proliferating cell nuclear antigen-dependent abasic site repair in Xenopus laevis oocytes: an alternative pathway of base excision DNA repair. Mol Cell Biol1994 Sep;14(9):6187-97.
127. Sobol RW, Horton JK, Kuhn R, Gu H, Singhal RK, Prasad R, et al. Requirement of mammalian DNA polymerase-beta in base-excision repair. Nature1996 Jan 11;379(6561):183-6.
128. Kubota Y, Nash RA, Klungland A, Schar P, Barnes DE, Lindahl T. Reconstitution of DNA base excision-repair with purified human
proteins: interaction between DNA polymerase beta and the XRCC1 protein. EMBO J1996 Dec 2;15(23):6662-70.
129. Thompson LH, Brookman KW, Jones NJ, Allen SA, Carrano
AV. Molecular cloning of the human XRCC1 gene, which corrects defective DNA strand break repair and sister chromatid exchange. Mol Cell Biol1990 Dec;10(12):6160-71.
130. Caldecott KW, McKeown CK, Tucker JD, Ljungquist S, Thompson LH. An interaction between the mammalian DNA repair protein XRCC1 and DNA ligase III. Mol Cell Biol1994 Jan;14(1):68-76.
131. El-Khamisy SF, Masutani M, Suzuki H, Caldecott KW. A requirement for PARP-1 for the assembly or stability of XRCC1 nuclear foci at sites of oxidative DNA damage. Nucleic Acids Res2003 Oct 1;31(19):5526-33.
132. Alexander BM, Pinnell N, Wen PY, D'Andrea A. Targeting
DNA repair and the cell cycle in glioblastoma. J Neurooncol May;107(3):463-77.
133. Lindahl T, Wood RD. Quality control by DNA repair.
Science1999 Dec 3;286(5446):1897-905.
134. Lindahl T. New class of enzymes acting on damaged DNA.
Nature1976 Jan 1-8;259(5538):64-6.
135. Palombo F, Gallinari P, Iaccarino I, Lettieri T, Hughes M, D'Arrigo A, et al. GTBP, a 160-kilodalton protein essential for mismatch-
binding activity in human cells. Science1995 Jun 30;268(5219):1912-4.
136. Drummond JT, Li GM, Longley MJ, Modrich P. Isolation of an hMSH2-p160 heterodimer that restores DNA mismatch repair to tumor cells. Science1995 Jun 30;268(5219):1909-12.
63
137. Jiricny J. The multifaceted mismatch-repair system. Nat Rev Mol Cell Biol2006 May;7(5):335-46.
138. Kolodner RD, Marsischky GT. Eukaryotic DNA mismatch
repair. Curr Opin Genet Dev1999 Feb;9(1):89-96.
139. Hegde MR, Chong B, Blazo ME, Chin LH, Ward PA, Chintagumpala MM, et al. A homozygous mutation in MSH6 causes Turcot
syndrome. Clin Cancer Res2005 Jul 1;11(13):4689-93.
140. Hamilton SR, Liu B, Parsons RE, Papadopoulos N, Jen J, Powell SM, et al. The molecular basis of Turcot's syndrome. N Engl J Med1995 Mar 30;332(13):839-47.
141. Peltomaki P. Role of DNA mismatch repair defects in the
pathogenesis of human cancer. J Clin Oncol2003 Mar 15;21(6):1174-9.
142. Kolodner RD, Tytell JD, Schmeits JL, Kane MF, Gupta RD, Weger J, et al. Germ-line msh6 mutations in colorectal cancer families. Cancer Res1999 Oct 15;59(20):5068-74.
143. Hanawalt PC. Subpathways of nucleotide excision repair and their regulation. Oncogene2002 Dec 16;21(58):8949-56.
144. Sugasawa K, Ng JM, Masutani C, Iwai S, van der Spek PJ, Eker AP, et al. Xeroderma pigmentosum group C protein complex is the initiator of global genome nucleotide excision repair. Mol Cell1998 Aug;2(2):223-32.
145. Li RY, Calsou P, Jones CJ, Salles B. Interactions of the transcription/DNA repair factor TFIIH and XP repair proteins with DNA lesions in a cell-free repair assay. J Mol Biol1998 Aug 14;281(2):211-8.
146. Guzder SN, Habraken Y, Sung P, Prakash L, Prakash S. Reconstitution of yeast nucleotide excision repair with purified Rad proteins, replication protein A, and transcription factor TFIIH. J Biol Chem1995 Jun 2;270(22):12973-6.
147. Sugasawa K, Okamoto T, Shimizu Y, Masutani C, Iwai S, Hanaoka F. A multistep damage recognition mechanism for global genomic
nucleotide excision repair. Genes Dev2001 Mar 1;15(5):507-21.
148. Sancar A. Mechanisms of DNA excision repair. Science1994 Dec 23;266(5193):1954-6.
64
149. Sancar A, Lindsey-Boltz LA, Unsal-Kacmaz K, Linn S. Molecular mechanisms of mammalian DNA repair and the DNA damage checkpoints. Annu Rev Biochem2004;73:39-85.
150. Khanna KK, Jackson SP. DNA double-strand breaks: signaling, repair and the cancer connection. Nat Genet2001 Mar;27(3):247-54.
151. Morrison C, Sonoda E, Takao N, Shinohara A, Yamamoto K, Takeda S. The controlling role of ATM in homologous recombinational repair of DNA damage. EMBO J2000 Feb 1;19(3):463-71.
152. Zhong Q, Chen CF, Li S, Chen Y, Wang CC, Xiao J, et al. Association of BRCA1 with the hRad50-hMre11-p95 complex and the DNA damage response. Science1999 Jul 30;285(5428):747-50.
153. Sonoda E, Hochegger H, Saberi A, Taniguchi Y, Takeda S. Differential usage of non-homologous end-joining and homologous
recombination in double strand break repair. DNA Repair (Amst)2006 Sep 8;5(9-10):1021-9.
154. Karpenshif Y, Bernstein KA. From yeast to mammals: Recent
advances in genetic control of homologous recombination. DNA Repair (Amst) Aug 10.
155. Moore JK, Haber JE. Cell cycle and genetic requirements of
two pathways of nonhomologous end-joining repair of double-strand breaks in Saccharomyces cerevisiae. Mol Cell Biol1996 May;16(5):2164-73.
156. Walker JR, Corpina RA, Goldberg J. Structure of the Ku heterodimer bound to DNA and its implications for double-strand break repair. Nature2001 Aug 9;412(6847):607-14.
157. Ma Y, Pannicke U, Schwarz K, Lieber MR. Hairpin opening and overhang processing by an Artemis/DNA-dependent protein kinase complex in nonhomologous end joining and V(D)J recombination. Cell2002
Mar 22;108(6):781-94.
158. Gu J, Lu H, Tippin B, Shimazaki N, Goodman MF, Lieber MR. XRCC4:DNA ligase IV can ligate incompatible DNA ends and can ligate
across gaps. EMBO J2007 Feb 21;26(4):1010-23.
159. Lieber MR. The mechanism of double-strand DNA break repair by the nonhomologous DNA end-joining pathway. Annu Rev Biochem;79:181-211.
65
160. Lieber MR. The mechanism of human nonhomologous DNA end joining. J Biol Chem2008 Jan 4;283(1):1-5.
161. Liu Y, Zhou K, Zhang H, Shugart YY, Chen L, Xu Z, et al.
Polymorphisms of LIG4 and XRCC4 involved in the NHEJ pathway interact to modify risk of glioma. Hum Mutat2008 Mar;29(3):381-9.
162. Bethke L, Webb E, Murray A, Schoemaker M, Johansen C,
Christensen HC, et al. Comprehensive analysis of the role of DNA repair gene polymorphisms on risk of glioma. Hum Mol Genet2008 Mar 15;17(6):800-5.
163. Liu Y, Scheurer ME, El-Zein R, Cao Y, Do KA, Gilbert M, et al. Association and interactions between DNA repair gene polymorphisms and adult glioma. Cancer Epidemiol Biomarkers Prev2009 Jan;18(1):204-14.
164. Yosunkaya E, Kucukyuruk B, Onaran I, Gurel CB, Uzan M, Kanigur-Sultuybek G. Glioma risk associates with polymorphisms of DNA
repair genes, XRCC1 and PARP1. Br J Neurosurg Oct;24(5):561-5.
165. Hood L, Galas D. The digital code of DNA. Nature2003 Jan 23;421(6921):444-8.
166. Kruglyak L, Nickerson DA. Variation is the spice of life. Nat Genet2001 Mar;27(3):234-6.
167. Palmer LJ, Cardon LR. Shaking the tree: mapping complex disease genes with linkage disequilibrium. Lancet2005 Oct 1;366(9492):1223-34.
168. Andersson U, McKean-Cowdin R, Hjalmars U, Malmer B. Genetic variants in association studies--review of strengths and weaknesses in study design and current knowledge of impact on cancer risk. Acta
Oncol2009;48(7):948-54.
169. Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, et al. The structure of haplotype blocks in the human genome. Science2002 Jun 21;296(5576):2225-9.
170. Jeffreys AJ, Kauppi L, Neumann R. Intensely punctate
meiotic recombination in the class II region of the major histocompatibility complex. Nat Genet2001 Oct;29(2):217-22.
171. Neale MJ, Keeney S. Clarifying the mechanics of DNA strand exchange in meiotic recombination. Nature2006 Jul 13;442(7099):153-8.
66
172. Wall JD, Pritchard JK. Haplotype blocks and linkage disequilibrium in the human genome. Nat Rev Genet2003 Aug;4(8):587-97.
173. Sundstrom P, Svenningsson A, Nystrom L, Forsgren L.
Clinical characteristics of multiple sclerosis in Vasterbotten County in northern Sweden. J Neurol Neurosurg Psychiatry2004 May;75(5):711-6.
174. Hallmans G, Agren A, Johansson G, Johansson A, Stegmayr
B, Jansson JH, et al. Cardiovascular disease and diabetes in the Northern Sweden Health and Disease Study Cohort - evaluation of risk factors and their interactions. Scand J Public Health Suppl2003;61:18-24.
175. Pukkala E, Andersen A, Berglund G, Gislefoss R, Gudnason V, Hallmans G, et al. Nordic biological specimen banks as basis for studies of cancer causes and control--more than 2 million sample donors, 25 million
person years and 100,000 prospective cancers. Acta Oncol2007;46(3):286-307.
176. Manjer J, Carlsson S, Elmstahl S, Gullberg B, Janzon L, Lindstrom M, et al. The Malmo Diet and Cancer Study: representativity, cancer incidence and mortality in participants and non-participants. Eur J Cancer Prev2001 Dec;10(6):489-99.
177. Berglund G, Elmstahl S, Janzon L, Larsson SA. The Malmo Diet and Cancer Study. Design and feasibility. J Intern Med1993 Jan;233(1):45-51.
178. Tjonneland A, Olsen A, Boll K, Stripp C, Christensen J, Engholm G, et al. Study design, exposure variables, and socioeconomic
determinants of participation in Diet, Cancer and Health: a population-based prospective cohort study of 57,053 men and women in Denmark. Scand J Public Health2007;35(4):432-41.
179. Wood RD, Mitchell M, Lindahl T. Human DNA repair genes, 2005. Mutat Res2005 Sep 4;577(1-2):275-83.
180. Lystig TC. Adjusted P values for genome-wide scans. Genetics2003 Aug;164(4):1683-7.
181. Rollison DE, Helzlsouer KJ, Alberg AJ, Hoffman S, Hou J,
Daniel R, et al. Serum antibodies to JC virus, BK virus, simian virus 40, and the risk of incident adult astrocytic brain tumors. Cancer Epidemiol Biomarkers Prev2003 May;12(5):460-3.
67
182. Soderberg-Naucler C. HCMV microinfections in inflammatory diseases and cancer. J Clin Virol2008 Mar;41(3):218-23.
183. Straat K, Liu C, Rahbar A, Zhu Q, Liu L, Wolmer-Solberg N, et
al. Activation of telomerase by human cytomegalovirus. J Natl Cancer Inst2009 Apr 1;101(7):488-97.
184. Cobbs CS. Evolving evidence implicates cytomegalovirus as a
promoter of malignant glioma pathogenesis. Herpesviridae;2(1):10.
185. Blumenthal DT, Raizer JJ, Rosenblum MK, Bilsky MH, Hariharan S, Abrey LE. Primary intracranial neoplasms in patients with HIV. Neurology1999 May 12;52(8):1648-51.
186. Wrensch M, Weinberg A, Wiencke J, Miike R, Barger G,
Kelsey K. Prevalence of antibodies to four herpesviruses among adults with glioma and controls. Am J Epidemiol2001 Jul 15;154(2):161-5.
187. Wrensch M, Weinberg A, Wiencke J, Miike R, Sison J, Wiemels J, et al. History of chickenpox and shingles and prevalence of antibodies to varicella-zoster virus and three other herpesviruses among adults with glioma and controls. Am J Epidemiol2005 May 15;161(10):929-
38.
188. Abramovitch R, Marikovsky M, Meir G, Neeman M. Stimulation of tumour growth by wound-derived growth factors. Br J
Cancer1999 Mar;79(9-10):1392-8.
189. Rajaraman P, Melin BS, Wang Z, McKean-Cowdin R,
Michaud DS, Wang SS, et al. Genome-wide association study of glioma and meta-analysis. Hum Genet Aug 11.
190. Berntsson SG, Wibom C, Sjostrom S, Henriksson R,
Brannstrom T, Broholm H, et al. Analysis of DNA repair gene polymorphisms and survival in low-grade and anaplastic gliomas. J Neurooncol Dec;105(3):531-8.