Introduction
Glioblastoma (GBM) is an aggressive brain tumor that, despite multimodal oncological therapy, relapses within 6–9 months (Stupp et al.
2005). A major challenge in developing new treatments is the intricate tumor heterogeneity at the molecular and cellular level (Brennan et al.
2013; Qazi et al.
2017; Lan et al.
2017). Targeted therapies have been sought to address the molecular heterogeneity in GBM, but dozens of clinical trials have failed to demonstrate survival benefit (Touat et al.
2017).
The application of targeted therapies is hampered by the existence of complex intra- and intertumoral heterogeneity in tumor-promoting signaling systems along with tumor evolutionary dynamics leading to acquired resistance to targeted drugs (Szerlip et al.
2012; Sottoriva et al.
2013; Klingler et al.
2015). To circumvent tumor heterogeneity, polytherapeutic approaches combining compounds acting on different targets simultaneously are receiving rising interest (Qazi et al.
2017). Moreover, the dramatic increase in costs for new oncological drugs has increased both the academic and public interest into possibilities in repurposing well-known drugs used for non-oncological indications for their potential anticancer activity (Bertolini et al.
2015; Huang et al.
2018). And as drugs used for decades have established dosing schedules and well-known toxicity profiles, both the time frame and costs to reposition for new indications can greatly be reduced.
Recently, a new treatment approach combining well-known drugs approved for non-oncological indications for polytherapeutic therapy has been suggested in GBM (Kast et al.
2013,
2014). The rationale consists of coordinated undermining of survival paths (CUSP) active in GBM by nine repurposed drugs, termed CUSP9. The concept of simultaneous blockade of multiple signaling pathways aims to prevent cancer cells to escape therapeutic challenges, rendering them susceptible for the cytotoxic effects of temozolomide (TMZ). Due to toxicities, the composition of drugs has been revised, and the current version consists of aprepitant, auranofin, captopril, celecoxib, disulfiram, itraconazole, minocycline, quetiapine, and sertraline (Halatsch et al.
2017).
Since the proposal, the concept of CUSP has been debated among neuro-oncology academics and practitioners (Prados et al.
2015; Purow
2016). More importantly, however, we experience that patients inquire and also adhere to parts or the entire CUSP9 combination along standard-of-care treatments outside of clinical trials within a do-it-yourself approach. Although case reports of patients treated with CUSP9 on a compassionate-use basis (Kast et al.
2014; Halatsch et al.
2017), and a recent registration of a clinical trial (NCT02770378), no experimental data have shown efficacy using the CUSP9 strategy. This prompted us to explore the efficacy of CUSP9 with concomitant TMZ using clinical achievable drug concentrations in patient-derived glioblastoma stem cells (GSCs), which may be responsible for tumor progression and recurrence in GBM (Lan et al.
2017).
Materials and methods
Brain tumor cultures
Glioblastoma biopsies were obtained from 15 informed and consenting patients undergoing surgery for GBM at Oslo University Hospital, Norway, approved by The Norwegian Regional Committee for Medical Research Ethics (REK 2017/167). The IDH mutational status was evaluated by immunohistochemistry and sequencing, and the MGMT-promoter methylation status evaluated by methylation-specific quantitative PCR. Cell cultures were established and maintained in serum-free conditions enriched for bFGF (R&D Systems) and EGF (R&D Systems), as previously described (Vik-Mo et al.
2010). Patient- and GSC culture characteristics are summarized in Online Resource 1. The self-renewal potential of the GSCs was quantified by the total number of cells following serial passages. Differentiation was induced, and cells were fixed and stained, as previously described (Vik-Mo et al.
2010). Images were acquired using Olympus Soft Imaging Xcellence software v.1.1.
Flow cytometry
Cells were suspended in PBS with 2% fetal bovine serum (Biochrom) and stained with directly conjugated antibodies (CD15-PerCP, R&D Systems, CD133-PE, Miltenyi Biotec) according to the manufacturer’s instructions. Cells were washed three times before analysis by flow cytometer LSRII (BD Bioscience). Dead cells were identified by propidium iodine (Thermo Fisher Scientific). Flow Jo software v.10.4.1 was used for data analysis.
qRT-PCR
The qRT-PCR experiments were performed, as previously described (Fayzullin et al.
2019). The high-capacity cDNA Reverse Transcription Kit, TaqMan Fast Advanced Master Mix, TaqMan oligonucleotide primers and probes [Hs00157674_m1 (GFAP), Hs00801390_s1 (TUBB3), and Hs01009250_m1 (PROM1/CD133)], the ABI Prism Detection System, and software (all from Applied Biosystems) were used according to the manufacturer’s instructions. Human β-Actin [Hs9999999903_m1 (ACTB)] was used as housekeeping gene. The relative gene expression levels were calculated using the standard curve method.
Intracranial transplantation
The National Animal Research Authority approved the animal procedures (FOTS 8318). C.B.-17 SCID female mice (7–9 weeks old, Taconic) were anesthetized with an injection of zolazepam (3.3 mg/mL), tiletamine (3.3 mg/mL), xylazine (0.45 mg/mL), and fentanyl (2.6 μg/mL), and placed in a stereotactic frame (David Kopf Instruments). Cells were prepared and transplanted, as previously described (Vik-Mo et al.
2010). The animals were regularly monitored for signs of distress and killed by cervical dislocation after 15 weeks or earlier if weight loss > 15% or neurological symptoms developed. The brains were harvested and further processed as previously described (Vik-Mo et al.
2010). Images of brain sections were acquired using Axio Scan.Z1 (Carl Zeiss). Processing of images was performed using ImageJ 2.0.
Drugs
Drugs used in this study: aprepitant (Selleck Chemicals, Cat# S1189), auranofin (Santa Cruz Biotechnology, Cat #sc-202476), captopril (Selleck Chemicals, Cat# S2051), celecoxib (Selleck Chemicals, Cat# S1261), copper(II)chloride dehydrate (Sigma-Aldrich, Cat# C3279), disulfiram (Selleck Chemicals, Cat# S1680), itraconazole (Selleck Chemicals, Cat# S2476), minocycline (Selleck Chemicals, Cat# S4226), quetiapine fumarate (Selleck Chemicals, Cat# S1763), sertraline (Selleck Chemicals, Cat# S4052), and temozolomide (Sigma-Aldrich, Cat# T2577). Copper(II)chloride dehydrate (CuCl
2) was added to all wells containing disulfiram (DSF) and corresponding control wells (Skrott et al.
2017). A fixed concentration of 20 µM Cu was used in this study (Twomey et al.
2008). Minocycline was dissolved in H
2O, while all other drugs were dissolved in DMSO for generation of stock solutions and stored according to the manufacturer’s instructions.
Drug concentrations
The clinical plasma concentrations of the individual drugs were obtained from reports of pharmacokinetic evaluations [aprepitant (Azuma and Fukase
2013), auranofin (Gottlieb
1982; Furst and Dromgoole
1984), captopril (Kripalani et al.
1980; al-Furaih et al.
1991), celecoxib (Davies et al.
2000), disulfiram (Johansson
1988,
1992), itraconazole (Heykants et al.
1989; Prentice et al.
1994), minocycline (Macdonald
1973; Agwuh
2006), quetiapine (DeVane and Nemeroff
2001; Jaskiw et al.
2004), sertraline (DeVane et al.
2002), and temozolomide (Ostermann et al.
2004)], and from drug labels by the U.S. Food and Drug Administration (
http://labels.fda.gov).
Cell viability and cell cytotoxicity assay
Cells were plated at 5000 cells/well in a 96-well plate (Sarstedt), cultured for 24 h before adding drugs and further incubated for 72 h. Cell viability was assessed using Cell Proliferation Kit II XTT (Roche) solution, incubated for 24 h before absorbance was analyzed on a PerkinElmer EnVision. Cell survival is reported relative to background corrected negative control of the drug. Cell cytotoxicity was assessed using CellTox™ Green Cytotoxicity Assay (Promega) solution, incubated for 15 min before fluorescence was analyzed on a Perkin Elmer EnVision. Measurements were corrected for background fluorescence, and raw data were scaled with reference to positive (sepantronium bromide) and negative control.
Cells were plated at 500 cells/well in 96-well plate (Sarstedt), cultured for 24 h before adding drugs and further incubated for 10 days. After 10 days, the spheres were stained using Thiazolyl Blue Tetrazolium Bromide (Sigma-Aldrich) 4 h prior to image acquisition and counting using an automated colony counter (GelCount, Oxford Optronics). Spheres > 30 µM in diameter were included in the final analysis, and results are reported relative to negative control.
Wnt-pathway activity (luciferase assay)
The GSCs were stably transfected with a luciferase reporter containing a synthetic 7xTCF-responsive promoter (7TFP was a gift from Roel Nusse, Addgene plasmid 24308). The lentiviral Renilla luciferase reporter was used as control (Amsbio, Cat# LVP370). The cells were plated at 20,000 cells/well in a 96-well plate before adding the respective drugs of CUSP9. To boost WNT/β-catenin signaling, 10 mM LiCl was added. The cells were incubated for 24 h before luciferase activity was quantified using the Dual-Glo Luciferase Assay System (Promega) according to the manufacturer’s protocol.
Statistical considerations
Data analysis and graphic presentation were undertaken using GraphPad Prism 7.0 and Microsoft Excel 14.7.3. Dose–response curves were fitted on the basis of a four-parameter sigmoidal logistic fit function defined by maximal and minimal cell survival, slope, and inflection point (EC50). In the curve fitting, the maximal cell survival was fixed to 100%, the minimal cell survival was allowed to float between 0% and 75% and slope between 0 and − 2.5. For drugs not reducing any cell survival, the constraints were removed. Statistical analyses were performed using paired sample t test or one-way ANOVA corrected for multiple comparisons using Dunnett’s test, as stated when the analysis was applied. Correlation analysis was undertaken using Spearman correlation coefficient. A p value < 0.05 was chosen to represent significance for the statistical analyses.
Discussion
Using clinically achievable concentrations, we provide, in this study, experimental data of a functional combination effect utilizing a coordinated pharmacological blockade by nine well-known drugs approved for non-oncological indications together with TMZ (the CUSP9 strategy) in patient-derived GSC cultures from both primary and recurrent GBMs.
As some GBM patients already supplement the conventional treatment with the CUSP9 strategy, we conducted this study with a clinical focus. However, mirroring clinical practice in preclinical studies is challenging. One fundamental aspect is the determination of drug concentrations that are achievable within the tumor of the patients (Liston and Davis
2017). What drug concentrations to pursue preclinically to reflect a clinical situation are not well defined (Smith and Houghton
2013). It has been suggested that preclinical drug levels can be decided using
Cmax as an upper reference limit to mirror a clinical situation and remove potential off-target effects of individual drugs (Liston and Davis
2017). In this study, we initially determined the dose–response relationships to each drug spanning the therapeutic range to investigate the inhibitory effects of individual drugs in clinically achievable concentrations. This led us to more carefully investigate a combined effect by reducing dominant effects of single drugs in CPCs. However, CPCs and
Cmax varies by different dosing schedules and routes of administration (Liston and Davis
2017). In this study, we decided to pursue concentrations in the lower end of reported clinical values at standard dosing regimens for the individual drug intended use. The adaptation of our experimental conditions to clinical plasma concentrations may, however, not reflect therapeutic drug concentrations achievable intratumorally or within the brain parenchyma. Although the drugs in CUSP9 are designed based on the properties of the drugs to cross the blood–brain barrier (Kast et al.
2014), penetrability and brain tissue levels of the individual CUSP9 drugs are unclear, as are intracerebral concentrations of most anticancer agents (Pitz et al.
2011). Moreover, when interpreting the results on combined effects, it is important to consider the limitations of the artificial stable drug exposure in vitro, which do not reflect the complex pharmacokinetics in patients–a complexity that increases by orders of magnitude when adding up to ten drugs in combination in vivo.
Different assays capturing different evaluations of cell death create more robust data when reporting drug efficacy in preclinical studies (Begley and Ellis
2012). In this study, the evaluation of overall efficacy after exposure to CUSP9 w/TMZ using cell viability and cytotoxicity readouts displayed a very good correlation. In selected drug responses (e.g., AUR and ITZ, Fig.
1), the tetrazolium-based cell viability assay displayed a bimodal dose–response pattern with a paradoxical increase in cell viability before the inhibitory effects occurred. This effect was similarly found for the same drugs when tested in CPCs (Fig.
2a). This may suggest a growth stimulatory effect; however, the inhibitory effects of these drugs evaluated by the sphere-forming assay (Fig.
2c, d) suggest that the response rather reflected a cellular stress response increasing the metabolic activity of the cells. A cellular stress response is biased in cell-based assays where metabolic activity is used as a surrogate marker for cell viability. This observation is in accordance with reported inaccuracies in using tetrazolium-based cell viability assays when interpreting subtle differences in cell viability evaluations (Sims and Plattner
2009; Stepanenko and Dmitrenko
2015). It further points to the importance of using different readouts for more accurate evaluation of drug efficacy in preclinical studies. For the sphere-forming assay, cells were incubated for a longer time. This could explain some of the more pronounced effects that were observed in this experimental setup.
Although not described as a key target of the CUSP9 combination, we found a significant reduction of Wnt-signaling activity by CUSP9 w/TMZ treatment, suggesting that this pathway may play a role in the combined treatment effect. In concentrations tenfold higher than clinically achievable in commercially available GBM cell lines, celecoxib and quetiapine have been shown as inhibitors of canonical Wnt-signaling in GBM (Sareddy et al.
2013; Wang et al.
2017). In this study, using CPCs, we found no individual effects of the drugs comprised in the CUSP9, and thus, the inhibition of the signaling pathway was related to a combined effect, which suggests a significance of the Wnt-signaling pathway as a mediator of the combined effect of CUSP9 w/TMZ. This finding, however, requires further studies exploring both the heterogeneity between patient-derived GSC cultures and exploring the entire spectra of expected key signaling target pathways.
In this study, the efficacy of the CUSP9 combination was evaluated using patient-derived GSC cultures from both primary and recurrent GBMs. Compared to commercially available GBM cell lines grown in serum, the GSC model system is recognized as a superior representation of GBM as it can recapitulate the cellular spectrum and malignant phenotype of GBM upon xenotransplantation, and preserve genomic feature of the parent tumor (Lee et al.
2006; Vik-Mo et al.
2010; Davis et al.
2016; Rosenberg et al.
2017; Lan et al.
2017). Experimental models that faithfully recapitulate the human disease are essential for preclinical studies; however, we acknowledge that selection of the aggressive GSC population underestimates the complexity in drug responses compared to the situation in vivo. Interestingly, we found that the recGBM cultures displayed resistance to TMZ, consistent with cultures being derived from recurrent tumors after TMZ treatment. This finding supports the external validity of the presented drug sensitivity data. Despite demonstrating a combination effect, we have not delineated whether all, some, or only a few of the drugs are required for the observed effect. However, patients following the CUSP9 strategy aim to utilize a combination of all drugs; therefore, detailed elucidation of whether only some of the drugs are required for the observed combination effect was not the scope of the current investigation. The use of patient-derived cultures from both treatment-naïve and pretreated tumors suggests that the combined effect can be found in several cultures sampled from a genetically heterogeneous GBM population. And as in vitro sensitivity to the standard-of-care, TMZ, a GSC gene signature, and the ability of GSC to expand as tumorspheres are independent predictors of patient outcome (Laks et al.
2009; Sandberg et al.
2013; D’Alessandris et al.
2017), a growing body of experimental data suggests the clinical relevance of using the GSC model system in preclinical GBM research.
In summary, using clinically achievable drug concentrations, we have added preclinical experimental data of a combined effect utilizing the CUSP9 strategy with TMZ in patient-derived GSCs, which supports clinical assessment of this approach. However, predicting response in individual cultures will require further profiling of GSCs. As some patients adhere to the CUSP9 treatment strategy outside of clinical trials within a do-it-yourself approach, we emphasize the importance of providing experimental data and trials with systematic follow-up of new treatment approaches consisting of drugs available for patients outside of specialized neuro-oncology treatment centers for adequate delineation of efficacy and toxicity.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.