ICAM1, CRK, CD36, IQGAP1 and COX2 genes have been described in the literature as target genes for celecoxib and fluvastatin drugs. Here we analyzed their expression by RT-Q-PCR in a series of PA and GBM, in U87-MG and U118 GBM cells lines, in the initial excised tumor of three patients with PA from which we derived in vitro explants cultures, and in a surgical specimen from the case report (arising from the second surgical resection because mRNA obtained from the initial specimen was of poor quality and not suitable for RT-Q-PCR).

Expression in H/C PA versus cerebellar PA : we confirmed in a larger cohort of tumors that the selected transcripts showed higher ICAM1, CRK, CD36, and IQGAP1 mRNA levels in H/C PA compared with cerebellar PA (p = 0.013, 0.027, 0.035 and 0.027 respectively) (Figure 1a, Additional file 1a). Quantification of COX-2 mRNA revealed no significantly different mRNA levels between these two PA sub-groups.

Expression in PA versus GBM: to go further, quantification of ICAM1, CRK, CD36, IQGAP1 and COX-2 transcripts was also performed in GBM. Results are reported in Figure 1b and revealed higher ICAM1 and CRK mRNA levels in PA compared to GBM samples (p = 0.002 and p < 0.0001 respectively) (see also Additional file 1b). The remaining transcripts were not differentially expressed.

Expression in U87-MG and U118 cell lines, PA-NAV, PA-GAS and PA-PET initial tumor specimen and in the excised tumor of the case report: we quantified the ICAM1, CRK, CD36, and IQGAP1 mRNA expression in our in vitro models and in the surgical specimen of the case report. All transcripts were readily detected (see Additional file 1c for detailed relative expression ratio values).

Overall, these results showed that the target genes ICAM1, CRK, CD36, and IQGAP1 were expressed as transcripts in H/C PA but also, at lower levels, in cerebellar PA and in GBM.

Because of the lack of commercially available cell lines derived from patients with PA, we used two GBM cell lines (U87-MG and U118) expressing the target genes of interest to assess the cytotoxic effect of fluvastatin and celecoxib.

Our results revealed that the two GBM cell lines were sensitive to both drugs. After a 48 h-treatment, IC₅₀ values of fluvastatin were 470 μM for U118 cell line (Figure 2a) and 880 μM for U87-MG cell line (data not shown). IC₅₀ values of celecoxib were 90 μM for U118 (Figure 2a) and 110 μM and U87-MG (data not shown). For the combination analysis, we first simultaneously incubated GBM cells with a concentration of celecoxib next to IC₅₀ values (100 μM) and a concentration of fluvastatin that only slightly affected cell growth when used alone (240 μM). Interestingly, this co-treatment fully suppressed U118 and U87-MG cell growth (up to 99%) as measured by the MTT assay, showing that fluvastatin potentiated the cytotoxic effects of celecoxib in GBM human cells (Figure 2b). GBM cell shrinkage and cell-adhesion-loss observed by microscopy also confirmed the efficacy of such a combination (data not shown).

To further investigate the combination of celecoxib and fluvastatin, growth inhibition assays were performed on the two GBM cell lines after incubation with a range of drug concentrations. Combination index (CI) values were determined on the basis of the Chou and Talalay method for all tested concentrations of chemotherapeutic drugs, using Calcusyn® software (Table 1). In U87-MG cells, the interaction between celecoxib and fluvastatin was synergistic at all concentrations tested (CI<1), except for the highest concentrations that resulted in additive effects (CI=1). It can be noticed that the combination between 100 μM celecoxib and 240 μM fluvastatin was the most synergistic. Similar conclusions were found in U118 cells: only lowest concentrations were antagonistic while others mostly displayed synergistic effects.

Growth inhibition assays were performed on U87-MG and U118 cell lines using the MTT assay after 48 h incubation with a range of concentrations of fluvastatin and celecoxib. The combination index (CI) of fluvastatin and celecoxib was calculated using the median effect method (Chou and Talalay method). CI values less than 1 indicate synergy, CI equal to 1 indicates an additive affect, and CI greater than 1 indicates antagonism between the two drug agents.

To determine whether fluvastatin and celecoxib influence cell proliferation, the cell cycle was analyzed on U87-MG. Upon combination of fluvastatin (240 μM) and celecoxib (100 μM) treatment, cell cycle progression is affected with a cell cycle arrest in G1 (Figure 3a). Then, Ki67 staining was performed and results showed a significant decrease of KI67-positive cells in treated cells in comparison with control cells (p = 0,049) (Figure 3b, A and B). Then, we determined whether cell death was induced by the drug combination, by measuring phosphatidylserine externalization using Annexin V, and propidium iodide (PI) accumulation. As shown in Figure 3c, the co-treatment with fluvastatin (240 μM) and celecoxib (100 μM) for 24 h triggered apoptosis in U87-MG. Same results have been obtained in U118 cells (data not shown).

Thus, the in vitro synergistic effects of celecoxib-fluvastatin combination in human GBM cells rely on induction of both apoptosis and cell proliferation decrease.

In this study, because PA cell lines were unavailable for analysis, we made use of a PA explant culture model. Explant culture allows the maintenance of cells in their microenvironment and, as we previously described 25 , it is highly accurate for the study of human brain gliomas because it recapitulates in vivo findings regarding cell migration and cell proliferation.

Both drugs were tested on three PA explant cultures, PA-NAV, PA-GAS and PA-PET with the drug alone (100 μM of celecoxib or 240 μM of fluvastatin) or with the synergistic combination found to be efficient with the GBM cell lines (240 μM of fluvastatin and 100 μM of celecoxib). As observed for GBM cell lines, we observed effects on explants with different levels of degradation depending on the treatment. We have established a 4 level scale: “unaffected” (explant), “affected +”, “affected ++”, “detached” and we recorded the percentage of explants in each state for each condition. This experiment was conducted in the three explant cultures but was quantified in only one, PA-PET, described in Figure 4a.

In untreated controls, explants and cell growth around the explants were “unaffected” (Figure 4b, A). When used alone, 240 μM fluvastatin treatment had little effect on explants but most cells around them mainly became round and lost their adhesion. This state represented the state “affected +” (Figure 4b, D). Celecoxib treatment (100 μM) mainly induced the “affected ++” state: damage to both explants and cell growth (Figure 3b, E). Explants treated with the combination (100 μM of celecoxib and 240 μM of fluvastatin) were totally disrupted and scattered, and represented the “detached” state (Figure 3b, F).

These observations, validated in 3 different PA explant cultures, confirmed that fluvastatin potentiated celecoxib action leading to a massive induction of apoptosis in PA cells.

A 4 year old girl was referred to our department as she developed a cachexia syndrome over several months. A brain computed tomography scan demonstrated three brain lesions: one in the suprasellar area, one in the third ventricle area and one in an infratentorial area, together with major hydrocephaly. Pathological examination of surgical biopsies from the posterior fossa and 3^rd ventricle demonstrated a PA, with KIAA1549-BRAF fusion but no BRAF^V600E mutation.

Complete surgical resection was not indicated and she underwent chemotherapy according to the BB-SFOP protocol 26 from June 2002 to October 2003. After initial stabilization, in 2004, a control MRI (magnetic resonance imaging) demonstrated tumor progression in the 3^rd ventricle and the appearance of medullar metastasis. In January 2008, 4 years after completion of treatment, she demonstrated local tumor progression of 3 lesions together with a major infiltration of the brainstem. The three tumors then grew further and turned into a real H/C tumor. She then received 7 cycles of oral temozolomide 27 that failed to control the disease. In July 2008, she underwent a partial surgical resection. In December 2008, she developed a new local and spinal progression and was treated according to standard chemotherapy published by Packer and colleagues 28 but developed a severe carboplatin allergic reaction after the first carboplatin infusion leading to cessation of the treatment. Because the parents refused standard alternative treatment requiring intraveinous drugs and because there was no short term functional risk, we proposed to initiate in 2009 a new strategy relying on the combination of fluvastatin and celecoxib based on preclinical and previous clinical reports. Celexoxib was administered per os at the dose of 200 mg twice daily as published in several metronomic paediatric protocols 29 30 and fluvastatin per os once daily for 2 weeks every 4 weeks with increasing dosage starting at 2 mg/kg/day to 8 mg/kg/day 31 .

After fluvastatin/celecoxib administration, control MRI performed every 6 months for an 18 month period demonstrated a progressive significant decrease in size of the enhancement area (Figure 5). Because the patient was stabilized, treatment was stopped.

However, eight months after stopping treatment, a degradation of her neurological status was observed. The cerebral and spinal MRI did not show any progression of the disease. Surgery was not feasible and the parents ruled out radiotherapy. Thus in September 2011, because her neurological status worsened, it was decided to initiate a new relevant treatment, at parents’ request, with the combination of irinotecan-bevacizumab, to avoid radiotherapy and aim at a rapid response as recently described 32 . Clinical improvement was noted after a month and she is now able to walk again with a decrease in size of the tumor.

Fifty-one pilocytic astrocytomas (PA) and 10 glioblastomas (GBM) were included in this study. Among the 51 PA, 27 were located in the cerebellum, 17 in the H/C location (optic pathway), 2 in the cerebral hemisphere, 3 in the medulla and 2 in the brainstem. BRAF status (BRAF^V600E mutation and KIAA1549-BRAF fusion) was known for 38/51 47 : 2/38 displayed BRAF^V600E mutation and 29/38 PA displayed KIAA1549-BRAF fusion. Only one patient was diagnosed with neurofibromatosis type 1 (NF1). Seven pilomyxoid astrocytomas were included in this study: 2/7 from the cerebellum and 5/7 were from the H/C region.

Forty-four PA and 10 GBM were collected at our hospital (Assistance Publique-Hôpitaux de Marseille, Marseille, France) and 7 PA samples were obtained from the Department of Pathology, University of Cambridge. Mean age at diagnosis was 7 years for PA (range: 1 year to 19 years, and a median age of 6 years) and mean age at diagnosis was 60 years for GBM (range: 44 years to 73 years, and a median age of 59 years).

In addition, three PA specimens from posterior fossa location, obtained from 3 additional young patients (6, 9 and 10 years old), were also used for explant culture. BRAF status was also known for these tumor samples: none of them displayed BRAF^V600E mutation but they all had KIAA1549-BRAF fusion gene.

Tumor specimens were obtained after written consent and according to a protocol approved by the local institutional review board and ethics committee and conducted according to national regulations. All frozen samples were stored in the Assistance Publique-Hôpitaux de Marseille tumor bank (authorization number 2008–70). Histological review of the frozen samples (DFB) confirmed the neoplastic nature of the tissue and demonstrated lack of normal residual tissue in samples used for RT-Q-PCR techniques.

Total RNA was extracted using TRI Reagent (Sigma-Aldrich, Paris, France), an improved version of the single-step total RNA isolation reagent developed by Chomczynski and Sacchi 48 , according to the manufacturer’s instructions. RNA was analyzed on the spectrophotometer Nanodrop and Agilent 2100 bioanalyzer (Agilent Technologies, Massy, France). Only samples with no evidence of ribosomal peak degradation and RIN values ranging between 8.0 and 10.0 were considered as high quality intact RNA. Before use, RNA samples were treated with 1U ribonuclease-free deoxyribonuclease (Roche Applied Science, Meylan, France) at 37°C for 20 min.

Total RNA (1 μg) DNA-free was reverse-transcribed into cDNA using 1 μg of random hexamers and Superscript II reverse transcriptase as recommended by the manufacturer (Invitrogen Life Technologies, Cergy Pontoise, France).

All PA and GBM samples were processed for the RT-Q-PCR experiment using a LightCycler 480 (Roche Applied Science) and the LightCycler 480 SYBR Green I Master Mix (Roche Applied Science). The relative expression ratio of the target mRNA and reference RNA (18S, GAPDH, β-actin) was calculated using Q-PCR efficiencies and the crossing point Cp deviation of a tumor sample versus normal adult human brain (Agilent Technologies) used as a control tissue 49 . Results are expressed as median (interquartile range). Forward and reverse primers for each gene are listed in Table 2.

Forward (F)/Reverse (R) primers and size of corresponding amplified fragment for each gene are listed. PCR conditions were as follows: 10 min at 95°C, followed by 45 cycles of 15 s at 95°C, 30 s at 60°C for CD36, ICAM1, IQGAP1, CRK and COX-2, 10 min at 95°C, followed by 30 cycles of 15 s at 95°C, 30 s at 67°C for 18S and 10 min at 95°C, followed by 45 cycles of 15 s at 95°C, 20 s at 65°C for GAPDH and β-actin.

The human U87-MG and U118 GBM cell lines (American Type Culture Collection, Rockville, MD, USA) were cultured in Dulbecco’s modified Eagle’s medium (DMEM) supplemented with 10% foetal calf serum (FCS), 50 U/ml penicillin, 50 μg/ml streptomycin and 5 mM sodium pyruvate (all purchased at Invitrogen Life Technologies) and they were maintained at 37°C in a 5% CO2 and 95% air atmosphere.

Celecoxib (Sigma-Aldrich) was reconstituted in dimethyl sulfoxide (DMSO) (Sigma-Aldrich) and fluvastatin (Sigma-Aldrich) in sterile water then diluted in culture media before use.

Cytotoxic effect of celecoxib and/or fluvastatin on U87-MG and U118 cell lines was evaluated by assessing cell metabolic capacity, which reflects viability, using the MTT kit (3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide, Sigma-Aldrich). The assays were conducted in quadruplicate with blank controls containing culture media only.

U87-MG and U118 cell lines (3.10³ cells/well) were seeded in 96-well plates. After 48 hr (subconfluency), cells were treated with serial concentrations of fluvastatin (30; 60; 120; 240; 365; 490; 610; 730; 850; 975 μM), celecoxib (0; 26; 52; 65; 78; 91; 104; 117; 131; 183; 210; 236; 260; 288; 314 μM) and their combinations in 100 μl of culture media. In both cell lines, the concentrations of fluvastatin in combined treatments tested were 240 and 480 μM and the concentrations of celecoxib were 50, 80, 100, 130 and 260 μM.

At 48 h after treatment, 10 μl of MTT reagent (1/10) was added to each well and incubated for 4 h at 37°C. Then, the reduced formazan crystals were dissolved in iso-propanol and absorbance was measured at 562 nm on a microtiter ELISA plate reader. The cell growth inhibitory activity was obtained by subtracting the absorbance of the blank controls and expressed as percentage of cell growth inhibition as compared to untreated controls (medium and drug diluents).

The IC₅₀ values of both drugs for the 2 GBM cell lines were determined. Then, synergistic interaction between fluvastatin and celecoxib was analysed using the combination index (CI) values that were calculated with the Calcusyn software based on the Chou and Talalay method 50 . The CI theorem provides quantitative definition for additive effects (CI=1), synergisms (CI<1) and antagonisms (CI>1) in drug combinations.

Cell cycle analysis was performed by PI staining of permeabilized cells and flow cytometry (FACS Calibur; BD Biosciences). A total of 10 000 events were counted for each sample. Data were analyzed with FlowJo software (Celeza GmbH, Olten, Switzerland) choosing the Dean-Jet-Fox model analysis.

Quantification of cell proliferation was performed by KI67 staining. After permeabilization, cells were incubated with KI67 antibody (1/25) (Dako, Glostrup) for 30 min at 4°C. Then, cells were incubated with the secondary antibody (anti-mouse IgG FITC, 1/100) (Jackson immunoresearch, West Grove, USA). Cells were analysed by flow cytometry (FACS Calibur; BD Biosciences) and data were analyzed using CellQuest Pro analysis software.

Apoptotic cells were quantified by Annexin V/PI double staining assay using the FITC Annexin V Apoptosis Detection Kit (BD Biosciences, Le Pont de Claix, France) as recommended by the manufacturer. The cells were analysed by flow cytometry using a FACS Calibur flow cytometer (BD Biosciences) within 1 hour and data were analyzed by CellQuest Pro analysis software.

Three PA samples named PA-NAV, PA-GAS and PA-PET were collected after surgery in DMEM supplemented with 10% FCS, 50 U/ml penicillin, 50 μg/ml streptomycin and 5 mM sodium pyruvate (all purchased at Invitrogen Life Technologies). Tumors were processed as previously described 25 . Briefly, tissues were washed, dissected, automatically sectioned using a McIlwain tissue chopper (Campden Instruments, Loughborough, England) and cut into 500-μm³ pieces in DMEM 10% FCS, and plated on glass coverslips (12-mm diameter) precoated with poly-(L)-lysine (10 μg/ml, Sigma-Aldrich). The explant pieces were maintained in DMEM 10% FCS. Medium was supplemented with 0.4% methylcellulose (Sigma-Aldrich). Explant cultures were incubated at 37°C in a 5% CO2 and 95% air atmosphere during maximum 40 days and were fed every 3 days. Expression of GFAP and A2B5 were systematically analyzed on culture explants by immunofluorescent staining, as previously described 25 in order to confirm the glial nature of cultured cells.

Cytotoxic effect of celecoxib and fluvastatin and synergistic interaction between both drugs were tested on PA explants from PA-NAV, PA-GAS and PA-PET. Briefly, after 10 days of culture, explants were treated with fluvastatin (240 μM), celecoxib (100 μM) and their combination (celecoxib 100 μM + fluvastatin 240 μM). At 48 h after treatment, explant cultures behavior was analyzed by phase contrast microscopy (Leica).

The association of the results of RT-Q-PCR with diagnosis (H/C PA versus cerebellar PA and PA versus GBM) and KI67 quantification was assessed by the non parametric Mann–Whitney test using IBM SPSS PASW statistics 17.0. A p value <0.05 was considered significant.