Background
The goal of drug repurposing is to find new clinical indications for existing pharmaceuticals that are currently on the market or failed in phase II/III trials. Repurposing is feasible because disease mechanisms are multifactorial and small drug molecules have multiple targets. Drug repurposing is both time and cost effective since the pharmacology and toxicity profile of approved drugs are already established. Approximately 30% of food and drug authority (FDA) applications for repurposed drugs are approved compared with 10% for new drugs [
1,
2]. There is also potential for drug radiotherapy combinations to improve therapeutic ratios and enhance efficacy without increasing toxicity [
3].
Drug and transcriptomic connectivity mapping can identify drug candidates for repurposing. The most widely used method is CMap (connectivity map project) which connects gene expression profiles to drugs based on data obtained from human cell lines treated with FDA approved drugs [
4]. CMap currently has over one million gene expression profiles from multiple cell lines treated with approximately 20,000 compounds. Since the release of the data sets from the library of integrated cellular signatures (LINCS) program, additional connectivity mapping algorithms have been developed such as the Queens University Belfast Accelerated Drug and Transcriptomic Connectivity (QUADrATiC) program [
5]. This software provides an improved and rapid method for calculating connection scores between the LINCS database and FDA approved compounds in order to identify drugs with the potential to reverse the biology or phenotype associated with the genes of interest [
6]. Any positive hit from this algorithm has already been identified as a safe therapeutic and can be progressed into a Phase I/II radiotherapy combination trial.
Computation-based approaches to drug repurposing provide an opportunity to identify novel agents to combine with radiotherapy [
7]. Prostate cancer is the most common malignancy in men with just under 50,000 new cases diagnosed in the UK and 450,000 in Europe each year [
8‐
10]. The local disease is managed with combinations of surgery, radiotherapy and hormones. The presence of hypoxia increases treatment resistance in prostate cancer patients treated with surgery or radiotherapy [
11]. Targeting hypoxia in combination with radiotherapy has not been widely studied in prostate cancer, however, two single arm trials suggest the approach is feasible [
12,
13]. To date the most extensive and convincing evidence for hypoxia modification in combination with radiotherapy comes from studies in head and neck cancer and muscle-invasive bladder cancer [
14‐
16]. The retrospective analysis of hypoxia gene signature biomarkers within clinical trials confirmed patients with hypoxic tumors benefit most from hypoxia modification [
14,
17]. Hypoxia gene signature biomarkers have been derived for multiple cancers and do not necessarily recapitulate across disease sites hence disease site-specific signatures have been developed [
18]. Recently, we derived a gene signature for assessing hypoxia in prostate cancer [
19]. The aim of this study was to investigate whether the transcription network associated with our hypoxia gene signature could be used in QUADrATIC to identify FDA approved drugs for potential repurposing for the treatment of hypoxic prostate cancer.
Discussion
Our study demonstrated how a gene expression network associated with hypoxia can be used in the QUADrATIC software to identify FDA approved drugs with the potential to be repurposed. The candidate drugs selected for in vitro validation, menadione and gemcitabine, showed similar cytotoxicity in normoxia and hypoxia in three cell lines. There was also evidence that the drugs were weak radiosensitizers in normoxia and hypoxia in one of the cell lines studied.
Although menadione and gemcitabine did not demonstrate hypoxia-selective cytotoxicity the drugs had similar efficacy in normoxia and hypoxia. In contrast, many chemotherapeutic agents (e.g., cisplatin, 5-FU and doxorubicin) have reduced cytotoxicity in vitro in hypoxia [
27‐
30]. In general, few studies compared the cytotoxicity of chemotherapeutic agents in normoxia and hypoxia in prostate cancer cell lines. Docetaxel, a first line systemic treatment for prostate cancer, has been shown to have reduced cytotoxicity in hypoxia (1% and 0.1% oxygen) in a range of cell lines [
27]. However, a study in DU145 and 22Rv1 cells showed similar docetaxel cytotoxicity in normoxia and hypoxia (0.5% oxygen) [
31]. It is uncommon for drugs studied in radiotherapy combination trials to have had their efficacy first tested under hypoxia. In future, pre-clinical testing as a justification for trial design should involve in vitro testing in hypoxia as well as normoxia.
As expected, we have shown the hypoxia selective toxicity of tirapazamine. Although previously studied in PC3 cells, this is the first study of tirapazamine in DU145 and LNCaP cells. The IC
50 for PC3 was 5 μM (4 days post-treatment after 24 h at 0.1% O
2), which compares with literature reported IC
50 doses for tirapazamine in PC3 cells of 15 µM [4 h anoxia] and 22 µM (48 h at 3% O
2) [
32,
33]. Tirapazamine has been shown to enhance the effect of castration induced hypoxia by inducing apoptosis and subsequently reducing tumor volume [
34]. Furthermore, hypoxia induces adaptive androgen independence and confers resistance to androgen deprivation therapy (ADT) [
35]. Tirapazamine in combination with ADT has the potential to eliminate hypoxic tumor cells and prevent the development of ADT resistant clones. However, a Phase III trial that randomised head and neck cancer patients to chemoradiotherapy alone or with tirapazamine showed no benefit [
36].
To identify gene expression changes in response to hypoxia the cells were exposed to 1% O
2 because the HIF-1 transcription factor is stabilised and changes in gene expression occurs. However, the level of O
2 at which significant resistance to radiation is observed is < 0.13% hence the in vitro experiments were performed at 0.1% O
2 [
37]. The physiological level of oxygen in the normal prostate is 3.4–3.9% but oxygen levels in prostate tumor tissue are in the range of 0.3–1.2% [
38]. Normoxia in this study is 21% O
2 which is supraphysiological compared to the level of oxygen in the prostate gland. However, in vitro cell lines are routinely established and cultured in the laboratory under these conditions and have adapted to grow at 21% O
2 and therefore the difference in gene expression may not reflect in vivo changes.
The OERs for the DU145 and PC3 cell lines exposed to 0.1% O
2 for 24 h were low over the dose ranges studied but similar to those previously reported for DU145 and PC3 with the exception of PC3 transfected with mir210 inhibitors giving an OER of ~ 2 [
39‐
42]. Furthermore, in vitro studies have reported varying OERs depending on the duration under hypoxia with OER decreasing with time of exposure under hypoxia for the same cell lines [
41].
The guidelines for preclinical and early phase assessment of radiosensitizers state that relatively low SER values in the range 1.2–1.5 may indicate a useful effect, particularly if sensitization occurs at clinically relevant doses of radiation [
43]. In the DU145 cell line, the SER for menadione under hypoxia was 1.15 which is comparable to the SER of nimorazole in the head and neck cancer cell lines FaDu (SER 1.14) and UMSCC47 (SER 1.13) [
44]. The SER of 1.27 for gemcitabine in normoxia was comparable to previously reported values in the range of 1.1–3 [
45]. This normoxic radiosensitization with gemcitabine is comparable to the radiosensitizing effects of other chemotherapeutic agents such as 5-FU [
46].
Menadione, also known as vitamin K
3, is a quinone and synthetic vitamin that can be converted into active vitamin K
2 in the body. Menadione induces the production of reactive oxygen species (ROS) through redox cycling and disrupts the interaction between HIF-1a and its coactivator p300 thus inhibiting HIF-1a transcriptional activity [
47‐
49]. Apatone (menadione and vitamin C) has shown prostate cancer antitumor activity in vitro and the toxicity profile was favorable in a phase I/IIa study [
50,
51]. PSA velocity and PSA doubling time decreased in 15 of 17 patients suggesting value in progressing apatone into the Phase III setting.
Gemcitabine is a chemotherapeutic agent used to treat several cancers, it is a nucleoside analogue that is incorporated into the DNA and inhibits DNA synthesis resulting in cell death. Nucleoside analogues such as gemcitabine are considered as potential radiosensitizers because they inhibit the repair of radiation induced DNA damage [
52]. In muscle invasive bladder cancer evidence from phase I/II trials supports the concurrent administration of gemcitabine and radiotherapy as a bladder preservation strategy [
53,
54]. In the breast cancer cell line MDA-MB-231, the SER for gemcitabine and radiation under hypoxia was 1.59 and under normoxia 1.70 [
55]. The radiosensitizing effects of gemcitabine in breast cancer are greater than the effects reported in this study for prostate cancer. However, the radiosensitizing effect of gemcitabine is greater under normoxia agrees with our findings.
A limitation of our study is that radiosensitizing effects were weak and only observed in DU145 cells. This cell line is derived from a central nervous system metastasis of primary prostate adenocarcinoma [
56]. In comparison, the PC3 cell line is characteristic of neuroendocrine-like prostate cancer, which represents less than 2% of cases and is biologically distinct from the more common adenocarcinoma subtype [
57,
58]. Both patients had prior treatment with hormonal therapy before cell lines were derived. Nonetheless, they are the most commonly used prostate cancer cell lines. Interestingly, a study investigating the radiosensitizing effect of vorinostat in prostate cell lines reported a radiosensitizing effect under normoxia and hypoxia in the DU145 cells but no effect in PC3 cells [
59]. A second limitation is that effects were only studied in cell lines grown as monolayers, an in vitro spheroid model may incorporate physiological hypoxia into the model. Although it is worth noting that, the in vitro data utilized by the QUADrATIC connectivity mapping software was obtained from monolayer cultured under 21% O
2.
Two approaches have been employed to identify FDA-approved drug for repurposing: in silico analytics and experimental screening studies [
2]. In a high-throughput oxygen consumption screen, atovaquone was shown to reduce oxygen consumption [
60]. Atovaquone is an FDA approved anti-malarial with a similar chemical structure to menadione. Atovaquone was shown to reduce tumor hypoxia and increase radiosensitivity at pharmacological concentrations in spheroids in vitro and in vivo [
60]. Interestingly, the drug did not alter the radiosensitivities of hypoxic cells grown as monolayers suggesting that atovaquone affects the tumor microenvironment rather than increasing the intrinsic radiosensitivity of cells [
60]. Atovaquone is currently being tested in phase I clinical trial for its ability to alleviate tumor hypoxia in lung cancer (NCT02628080).
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.