Professional Documents
Culture Documents
Corresponding author: Giuseppe Gasparre, PhD, Dipartimento di Scienze Mediche e Chirurgiche, UO Genetica Medica, Pad.11, Policlinico S.Orsola-Malpighi,
via Massarenti 9, 40138, Bologna, Italy (giuseppe.gasparre@gmail.com).
Background: Glioblastoma multiforme (GBM) is the most malignant brain tumor in adults, but its etiology still remains unknown. Recently,
a role of viruses such as cytomegalovirus and JC virus in gliomagenesis has been suggested. Since human papillomavirus (HPV) is con-
sidered the most common oncogenic virus in humans, we evaluated its occurrence in GBM samples.
Material and Methods: Fifty-two formalin-fixed paraffin-embedded primary glioblastoma specimens were retrospectively analyzed. The
presence of HPV genome on tumor DNA was assessed by MY/GP nested PCR. Confirmation of HPV detection was obtained by chromogenic
in situ hybridization (CISH) and immunohistochemistry (IHC) with an antibody directed against the L1 capsidic protein. Finally, univariate
and multivariate proportional-hazards models were used to compare the risk of death among HPV-positive and HPV-negative patients.
Results: Strikingly, viral DNAwas detected after PCR in 12 cases (23%). HPV16 genome was present in 25% infected samples, whereas the
remaining samples tested positive for HPV6. CISH confirmed positivity in all infected samples for which enough material was available.
Moreover, IHC positivity suggested that production of viral proteins from HPV genome is an ongoing process in GBM cancer cells. Finally an
association between HPV infection and a worse prognosis was found in patients upon age stratification with a univariate analysis (HR,
2.10; 95% CI, 1.00–4.44; log-rank P ¼ .045).
Conclusions: HPV infection status may be considered an independent prognostic factor in GBM patients and suggests that prevention
may be considered, should HPV be recognized as a causative agent in gliomagenesis.
Glioblastoma multiforme (GBM) (WHO astrocytoma grade IV) is To this aim, 52 primary GBM cases consecutively referred to the
the most common primary malignant brain cancer in adults1 Bellaria Hospital, Bologna, Italy, were retrieved (27 men and 25
and is associated with a median survival of 12 to15 months from women); their clinical and anagraphical features are summarized
diagnosis.2 Standard treatments consist of total surgical resection in Table 1. Patients ranged in age from 41 to 78 years (mean ¼
whenever possible, followed by a combination of the alkylating 62y) at diagnosis. Standard treatment regimen (TMZ concurrent
cytostatic drug temozolomide (TMZ) and radiotherapy (RT), with or adjuvant to RT) was administered in 27 patients, whereas
which however provide relatively limited improvements.3 The eti- 2 patients were treated with RT alone, and 19 patients did not
ology of GBM is yet unknown, albeit recently a consensus receive any adjuvant therapy. Information about therapy was
towards a possible association with cytomegalovirus (CMV) or not available in 7 cases.
JCV has been proposed.4,5 We show here for the first time that In order to assess the presence of HPV in GBM specimens, DNA
oncogenic human papillomavirus (HPV) is present at a relevant fre- extracted from tumor samples was analyzed by nested PCR with
quency in GBM specimens and associates with a poorer prognosis. the MY/GP primer sets.6 Surprisingly 12 of 52 (23%) cases were
298
Vidone et al.: HPV in prognosis of glioblastoma
Continued
Neuro-Oncology 299
Vidone et al.: HPV in prognosis of glioblastoma
Table 1. Continued
GBM51 M 60 Y Y Y 16 -
GBM52 F 72 N N Y alive 14.7 -
found to be positive for HPV amplification. Sequencing of PCR pro- HR ¼ 0.96; 95% CI. 0.49–1.90) (multivariate Cox model HR ¼ 0.80;
ducts and BLAST alignment were performed to identify the viral 95% CI, 0.38–1.71).
genotype associated with the infection. Interestingly, 3 cases Infection by HPV in cancers of different organs is often asso-
showed 100% of homology with HPV16 genome, the prevalent ciated with either a positive or negative prognostic effect whose
high-risk type in cervical cancer, whereas the remaining 9 cases entity depends on the tissue and the viral type.13 – 17 Therefore,
harbored HPV6, the most relevant low-risk type in benign anogeni- we speculated that HPV may also have an impact in GBM. Kaplan –
tal lesions (Fig. 1A). Meier curves show that patients with HPV-positive GBM had a worse
Confirmation of localization of HPV genome within cancer cells median overall survival than those with HPV-negative cancer. In
was carried out by chromogenic in situ hybridization (CISH) on fact, the 12-month rates of overall survival were 41% and 51%, re-
infected and noninfected samples for which enough material was spectively. Nonetheless, there was no statistically significant differ-
available (Supplementary material, Material and Methods). CISH ence in survival rate between the 2 groups, which was likely due to
ascertained positivity in all infected cases, 2 of which showed an the need for an enlargement of our cohort. A trend for a shorter
intense widespread nuclear positivity (Table 1, Fig. 1B). Surprisingly median overall survival was indeed observed in patients with
in 1 sample, the signal was not restricted to the stromal cancer HPV-positive GBM (7.3 months vs. 11 months) (log-rank P ¼ .19)
cells but was also evident in endothelial cells. Such positivity may re- (Fig. 1D).
present a seminal observation of either a route of entry into the brain It has been ascertained that 50 years is usually a more appropri-
or viral spreading within the glial tissue, although functional confir- ate and reproducible cut-off age for clinical subdivision of patients
mations are warranted. In this regard, the possibility of HPV diffusion with GBM into prognostically relevant subsets.18,19 In this group, a
via the bloodstream has already been hypothesized, since the viral statistically significant worsening of the prognosis was observed in
genome has been detected in the blood of patients with cervical HPV-positive patients aged 50 years and older after univariate ana-
and oropharyngeal HPV-associated cancers.7,8 lysis (HR, 2.10; 95% CI, 1.00–4.44; log-rank P ¼ .045; median
To date, there is only 1 report on HPV occurrence in the brain in overall survival ¼ 7.3 vs. 11; Fig. 1E), confirmed by the borderline
association with cortical dysplasia.9 Another recent study reported significance found in multivariate analysis (HR,2.48; 95% CI,
the presence of HPV in temporal arteritis;10 interestingly here both 0.99 –6.24). No statistically significant differences were reported
high- and low-risk types were detected, indicating that the capabil- for age, MGMT methylation status, and treatment between the 2
ity to reach the cerebral districts is not restricted to high-risk HPVs. groups (Fig. 1F).
To assess a functional effect of HPV genome occurrence in GBM Overall, we show here for the first time that GBM may be asso-
samples, we evaluated production of viral proteins by immunohis- ciated with HPV infection and that prognosis is poorer when such
tochemistry (IHC) with an antibody directed against the capsidic infection occurs.
protein L1. Again, cases for which enough material was available HPV is considered the causative agent for nearly all cases of cer-
showed positivity for L1 staining, suggesting production of viral vical cancer and has been associated with an increasing number of
proteins from HPV genome to be an ongoing process in GBM other tumors such as ovarian, oropharyngeal, and lung cancers. Its
cancer cells (Fig. 1C). transforming capabilities are due to the expression of E6 and E7
We next attempted to ascertain whether a correlation could be proteins, which deregulate cell proliferation and apoptosis upon
found between occurrence of HPV infection and prognosis in GBM. binding to p53 and pRb.20 Although functional studies on the in-
To date, few biomarkers have proven to be of prognostic significance volvement of HPV in the origin of GBM were beyond the scope of
with respect to the survival of patients affected by GBM.11 It has this preliminary analysis, we speculate that the virus may be a con-
been reported that the methylation of O6-methylguanine-DNA current cause of gliomagenesis, with additional genetic hits being
methyltransferase (MGMT) gene, which encodes a DNA repair likely required for HPV-related tumorigenesis, such as the activa-
enzyme that catalyzes the transfer of mutagenic and cytotoxic tion of oncogenes.
adducts from O6 guanine in DNA, is associated with improved sen- A relevant issue that deserves further investigation is the deter-
sitivity to chemotherapeutics such as TMZ.12 MGMT methylation mination of the physical state in which HPV exists in a nonepithelial
status had been previously assessed in most of our cases, and its cancer such as GBM. In invasive cervical neoplasms and advanced
promoter had been found to be methylated in 20 cases (Table 1). precancerous lesions, HPV is generally found in an integrated, non-
The evaluation of the effect of MGMT methylation status on patients replicating state characterized by the overexpression of E6 and E7
survival did not result in any significant change in prognosis, either proteins. The production of viral particles occurs only in low-grade
alone or in association with TMZ treatment (univariate Cox model precancerous lesions or nondysplastic tissues and requires a
300
Vidone et al.: HPV in prognosis of glioblastoma
Fig. 1. (A) Pairwise alignment between the HPV L1 protein sequence and the query obtained translating the nucleotide sequence detected by PCR (see
methods). (B) Chromogenic in situ hybridization of a HPV-positive case (left panel) and a HPV-negative control (right panel); the arrow indicates the
nucleus of an endothelial cell showing positivity. (C) Immunohistochemistry staining for L1 capsidic protein. The left panel shows a positive case that
displays a classical nuclear positivity to L1 compared to the negative control (right panel). (D) Kaplan– Meier estimates of the survival among GBM
patients under study according to HPV status. Patients with HPV-positive tumors have a worse median overall survival than patients with HPV negative
tumors (median overall survival 7.3 vs. 11 months; log-rank test P ¼ .19). (E) Kaplan–Meier estimates of the survival among GBM patients according to
HPV status within the subset of patients older than 50 (log-rank P ¼ .045, median overall survival ¼ 7.3 vs. 11). (F) Summary of the clinical
characteristics of HPV-positive and HPV-negative patient subgroups. P values were calculated with t-test or Fisher’s exact test.
Neuro-Oncology 301
Vidone et al.: HPV in prognosis of glioblastoma
pluristratified epithelium.20 Thus, it is likely that in GBM the virus 4. Dziurzynski K, Chang SM, Heimberger AB, et al. Consensus on the role of
may be associated with an abortive infection with transforming human cytomegalovirus in glioblastoma. Neuro Oncol. 2012;14(3):
properties, since viral proteins were detected here by IHC. 246–255.
Although HPV is considered to be an exclusive epitheliotropic 5. Del Valle L, Gordon J, Assimakopoulou M, et al. Detection of JC virus DNA
pathogen, it has been demonstrated that virions are able to bind sequences and expression of the viral regulatory protein T-antigen
to cells derived from several tissues and species.21 Moreover, the in tumors of the central nervous system. Cancer Res. 2001;61(10):
heparan sulphate proteoglycans, which are likely to be the initial 4287–4293.
binding molecules for many HPV types, are also found to be 6. Gravitt PE, Peyton CL, Alessi T, et al. Improved amplification of genital
expressed in astrocytes.22 Whether these molecules may mediate human papillomaviruses. J Clin Microbiol. 2000;38(1):357–361.
302