|Year : 2019 | Volume
| Issue : 4 | Page : 180-186
Systematic drug repurposing to enable precision medicine: A case study in breast cancer
Krystyna Taylor, Sayoni Das, Matthew Pearson, James Kozubek, Mark Strivens, Steve Gardner
PrecisionLife Ltd., Long Hanborough OX29 8LJ, UK
|Date of Submission||02-Dec-2019|
|Date of Decision||28-Jan-2020|
|Date of Acceptance||05-Mar-2020|
|Date of Web Publication||13-Apr-2020|
PrecisionLife Ltd., Hanborough Business Park, Unit 8B Bankside, Long Hanborough, OX29 8LJ
Source of Support: None, Conflict of Interest: None
Background and Objectives: Precision medicine and drug repurposing provide an opportunity to ameliorate the challenges of declining pharmaceutical R&D productivity, rising costs of new drugs, and poor patient response rates to existing medications. Multifactorial “disease signatures” provide unique insights into the architecture of complex disease populations that can be used to better stratify patient groups, aiding the delivery of precision medicine. Methods: Analysis of a complex disease (breast cancer) population was undertaken to identify the combinations of single-nucleotide polymorphisms that are associated with different disease subgroups. Target genes associated with the disease risk of these subgroups were examined, followed by identification and evaluation of existing active chemical leads as drug repurposing candidates. Results: One hundred and seventy-five disease-associated gene targets relevant to different subpopulations of breast cancer patients were identified. Twenty-three of these genes were prioritized as both promising novel drug targets and repurposing candidates. Two targets, P4HA2 and TGM2, have high repurposing potential and a strong mechanistic link to breast cancer. Conclusions: This study showed that detailed analysis of combinatorial genomic (and other) features can be used to accurately stratify patient populations and identify highly plausible drug repurposing candidates systematically across all disease-associated targets.
Keywords: Artificial intelligence, drug repurposing, genomics, precision medicine
|How to cite this article:|
Taylor K, Das S, Pearson M, Kozubek J, Strivens M, Gardner S. Systematic drug repurposing to enable precision medicine: A case study in breast cancer. Digit Med 2019;5:180-6
|How to cite this URL:|
Taylor K, Das S, Pearson M, Kozubek J, Strivens M, Gardner S. Systematic drug repurposing to enable precision medicine: A case study in breast cancer. Digit Med [serial online] 2019 [cited 2020 May 26];5:180-6. Available from: http://www.digitmedicine.com/text.asp?2019/5/4/180/282372
| Introduction|| |
R&D productivity in the pharma industry has been diminishing for decades. In 2018, R and D returns declined to 1.9%, down from 10.1% in 2010. Drug discovery is costly (at over $2.8B per marketed drug) and lengthy – it takes an average of 12 years to develop and market a new drug. Even then, many drugs benefit only a limited proportion of patients to whom they are prescribed.
Precision medicine promises to deliver better medicines, improved patient outcomes, and lower healthcare costs. It has the potential to benefit millions of patients and save global healthcare systems tens or even hundreds of billions of US dollars/year through new, better targeted therapeutic options, reduced misprescription and overmedication, and better patient compliance.
However, developing novel targeted therapies for each patient subgroup is challenging. Robustly identifying disease causative mutations in druggable targets and developing new medicines to target these have proved difficult to scale, even with the advent of genomic medicine. A more cost-effective approach can be able to identify targets associated with the clinically relevant subgroups of patients with unmet medical needs and search the current formulary to find the drugs that will be effective for each of them.
As of 2018, over 1500 drugs have been approved,, including many safe and effective medicines that hit targets that play roles in multiple diseases. Conventionally, drug repurposing involves identifying a drug candidate that is proven safe in humans, but that was either ineffective for its original indication or that has been approved and launched in another disease area. Someone wishing to repurpose the drug licenses it from the originator, reformulates it if necessary, and takes it through clinical trials in the new indication, before gaining approval and launch. This can be quicker and cheaper than de novo drug development, with average costs of just $300 million and lead times of <7 years.
Repurposing can help identify therapies especially for areas of unmet medical need in complex diseases such as asthma, amyotrophic lateral sclerosis, dementia, and breast cancer. Breast cancer, for example, is a highly heterogeneous disease with variations in prognosis, treatment response, underlying disease mechanisms, and tumor biology. It is currently the leading cause of cancer-related mortality in women, with approximately one in eight women being diagnosed with the disease at some point in their lifetime. Patients are currently classified into several different molecular subtypes, including human epidermal growth factor receptor 2 (HER2)-positive and triple-negative breast cancer (TNBC).
Greater understanding of underlying HER2-positive disease mechanisms has led to the development of HER2-targeted therapies, such as trastuzumab and lapatinib, generating vast improvements in patient survival as a result. However, although breast cancer treatment has a more personalized approach than some other diseases, subtypes of breast cancer patients – such as those with TNBC – do not respond to these targeted hormonal therapies, and up to 50% of HER2-positive breast cancer patients still go on to develop metastases. Hence, there is still a significant need for greater personalization of treatment strategies in breast cancer therapy to increase patient response rates and overall survival.
Delivering a personalized medicine requires a detailed understanding of the population architecture of a disease and the accurate stratification of its patients. For complex, multifactorial diseases, such as breast cancer, this means finding combinations of features (disease signatures) that accurately describe disease subgroups rather than just finding single mutations in genes.
| Methods|| |
Revealing this level of detail requires a change in analytical approach. Most complex diseases are caused by complex interactions among many genes, and even the most important genomic loci have small effect sizes in isolation. High-order epistatic interactions are commonly implicated in disease etiology and pleiotropic effects. Existing analytical methods, such as genome-wide association studies (GWAS), are however limited to finding singular, or at most pairs, of disease-associated single-nucleotide polymorphisms (SNPs). They are unable to fully reveal the disease mechanisms for complex, multifactorial disease.
PrecisionLife is a massively scalable multi-omics (genomic, proteomic, transcriptomic, and phenotypic) association platform that enables the hypothesis-free detection of high-order phenotype-associated combinations at genome-wide study scale. It analyzes GWAS datasets that have been prepared using standard techniques on distributed graphics processing unit instances, applying a prefiltering step and a six-stage automated analysis workflow:
- Mining – Find all (or the substantial majority of the) distinct n-combinations of SNP genotypes and/or other types of features found in the cases but not in the controls (or vice versa in case the study is focusing on protective factors)
- Permutations – Repeat mining of, e.g., 1000 random permutations of all individuals using the same mining parameters
- Network analysis – Find networks of distinct n-combinations sharing one or more SNP genotypes
- Network validation – Find networks from all n-combinations and from all random permutations using the same parameters, compare null hypothesis, and determine P value with false discovery rate (FDR) correction to eliminate random observations
- Network annotation – Annotate networks with semantic graph containing SNP IDs, genes, pathways, druggable targets, pharmacogenetic interactions, epigenetic modifications, and other features
- Reclustering – Correlate validated networks sharing specific features, which may simply be common SNP genotypes or more complex biological hypotheses involving lifestyle factors, pathways, and others available in the specific disease datasets.
The underlying analytical mining platform has been validated in multiple disease populations.
Hence, PrecisionLife can find and statistically validate combinations of features (typically 3–10 features in combination known as “signatures”) that together are strongly associated with a specific disease diagnosis or other clinical phenotype (e.g., fast disease progression or therapy response).
Genotype data of 547,197 SNPs from 11,088 breast cancer cases and 22,176 controls (1:2 case–control ratio) were obtained from the UK BioBank (ICD10 code C50). An age-matched control set was created of randomly selected healthy females who did not have any history of cancer. The PrecisionLife platform took less than an hour to identify high-order combinatorial genomic signatures (up to 5 SNP genotypes in combination, using an FDR of 0.05) in this dataset.
The combinatorial signatures produced can be used to more accurately estimate a patient's disease risk, explain disease mechanisms, and identify novel disease targets. In this study, we used the signatures discovered to identify drug repurposing candidates for key disease-associated targets. Given a series of disease-associated SNPs, we mapped these to the reference genome to identify disease-associated and clinically relevant target genes. We then used a semantic knowledge graph derived from multiple public and private data sources to annotate the targets, testing the 5Rs principles of drug discovery, and forming strong and testable hypotheses for their mechanism of action and impact on disease phenotype as illustrated in [Figure 1].,
|Figure 1: Combinatorial signatures provide a detailed insight into the causes and potential intervention points for complex diseases. The example shows a six single-nucleotide polymorphism disease signatures associated with significantly reduced risk of developing breast cancer, mediated through interaction of the gene marked (6) with the insulin receptor INSR, a key activator of the oncogene phosphatidylinositol 3-kinase|
Click here to view
A series of heuristics were then applied on the identified genes to find targets and candidate drugs with the highest potential for repurposing on the basis of correlation to disease, existing disease indications, relevant tissue expression, acceptable safety profiles, and patent scope.
| Results|| |
Generating combinatorial signatures and annotating the relevant gene targets with biomedical data provide unique insights into their mechanism of action and impact on patient phenotype as shown in [Figure 1]. Merging the signatures to better analyze the overall population structure, we identified more than 174,000 unique disease signatures in the breast cancer population that captures different combinations of SNP genotypes. Clustering of these SNPs by co-occurrence in patients provides detailed insights into the architecture of this complex disease population [Figure 2].
|Figure 2: PrecisionLife analysis results showing multiple new mutations and targets associated with subpopulations of breast cancer patients. Single-nucleotide polymorphisms for genes found by standard genome-wide association studies (Plink 1.9 - FGFR2, CCDC170, and CCDC91) are shown colored red, yellow, or green. Novel disease-associated single-nucleotide polymorphisms that can only be identified using a combinatorial approach are shown in gray. Single-nucleotide polymorphisms are clustered by co-occurrence in patient cases – closer single-nucleotide polymorphisms co-occur more frequently together in cases. Yellow lines indicate that single-nucleotide polymorphisms are in linkage disequilibrium|
Click here to view
We found 175 risk-associated genes that are relevant to different patient subpopulations. These genes were annotated and analyzed using a druggability heuristics. We identified 23 gene targets as high scoring repurposing candidates. Two of these targets, P4HA2 and TGM2, were identified as having high repurposing potential and have already been investigated in the context of breast cancer.
P4HA2 scored highly in the analysis. It encodes an enzyme that plays a role in collagen synthesis, catalyzing the formation of crucial 4-hydroxyproline residues that are involved in collagen helix formation and stabilization. Collagen deposition in breast cancer increases cancer cell development and growth, and inhibiting P4HA2 may prove beneficial in breast cancer through a reduction in collagen synthesis and deposition.
Aspirin decreases the expression of P4HA2, resulting in decreased collagen deposition. Aspirin is well studied, with a wealth of pharmacokinetic and toxicology data at high and low dose, and has a simple molecular structure [Figure 3], meaning that it interacts with a wide variety of biological targets. It was originally licensed as a nonselective cyclooxygenase-2 inhibitor; however, it also modulates several different transcription factors and pathways implicated in cancer, including NFĸB, PIK3CA, AMPK, and mTORC1. Aspirin reduces P4HA2 activity through two different mechanisms and also enhances the levels of an miRNA called let-7 g, which binds and suppresses the expression of P4HA2. In addition, the promoter of P4HA2 has three NFĸB-binding sites and aspirin inhibits NFĸB expression, resulting in a concomitant decrease in P4HA2 activity. The benefits of this aspirin-induced reduction in collagen deposition were observed in a model of hepatocellular carcinoma, wherein the inhibition of P4HA2 resulted in a reduction in tumor growth.
|Figure 3: Molecular structures of aspirin (a), cystamine (b), and disulfiram (c)|
Click here to view
There is however conflicting evidence so as to whether aspirin is effective in both reducing the risk of breast cancer and improving disease survival after diagnosis., A greater understanding of the mechanisms behind aspirin's antitumor effect and stratification of the population into more clinically relevant subsets may indicate groups of patients who are more likely to respond to aspirin treatment. Our results identified a subgroup of patients with a gene signature that indicates aberrant P4HA2 expression, for whom administration of aspirin is more likely to be effective.
TGM2 also scored highly for repurposing potential. TGM2 encodes an enzyme (transglutaminase 2 [TG2]) involved in postranslation modification of proteins, facilitating their crosslinking. High TG2 expression has been associated with increased tumor growth and invasion in several different cancer types through the activation of phosphatidylinositol 3-kinase/Akt and other cell survival pathways. In breast cancer, TG2 is upregulated compared to normal epithelial tissue and increasing expression is correlated with higher tumor stage. Furthermore, it was also shown that TG2 interacts with interleukin-6 (IL-6), facilitating IL-6 mediated inflammation, tumor aggressiveness, and metastasis in a mouse model of breast cancer. Hence, repurposing a TG2 inhibitor in breast cancer could be therapeutically beneficial in a specific subtype of patients.
Cystamine is an allosteric inhibitor of TG2, causing the formation of a disulfide bond between two cysteine residues, diminishing TG2's catalytic activity. Moreover, although cystamine has not yet been trialed in breast cancer patients, anin vitro study has found that inhibiting TG2 expression resulted in reduced breast tumor growth compared to controls. However, trials in humans demonstrate that cystamine can cause a range of dose-limiting side effects. Conversely, disulfiram, a drug approved for the treatment of chronic alcoholism, has a comparable molecular structure to cystamine [Figure 3]. Palanski andKhosla demonstrated that disulfiram has the same activity as cystamine in vitro, with comparable inhibitory constants when assessed experimentally. Disulfiram has a more favorable pharmacokinetic profile than cystamine; it can be administered orally with a maximum dose of 500 mg/day and is reasonably well tolerated in patients.,
| Discussion|| |
Different diseases may share common pathways, and drugs that affect genes in these pathways could therefore treat a variety of disease indications. Mapping existing drugs onto the genetic and metabolic signatures [Figure 4] indicates the areas where there are already good clinical options, and also where off-label use of existing therapeutics with good safety and tolerability profiles, with acceptable routes of administration, could have potential. For a given patient, their specific combination of SNPs will in large part determine which drug or combination of drugs is likely to benefit them personally.
|Figure 4: Graph of existing drug options for key targets identified as being relevant to disease subpopulations in a breast cancer population|
Click here to view
Both of the targets identified from this study, TGM2 and P4HA2, have strong mechanistic links to breast cancer and are targeted by approved drugs with favorable pharmacokinetic and toxicity profiles. These two targets demonstrate the potential of this approach to identify repurposing candidates that have potential to be effective in specific subgroups of breast cancer patients.
| Conclusions|| |
The combination of large quantities of patient genotype, phenotype, and clinical data and improved data analytics methods have the potential to usher in a new era of affordable personalized medicine, lowering the cost of care, and identifying the best drugs for individual patients, thereby giving them the best possible outcome. In a time of rising drug costs and squeezes on healthcare budgets, this step could be crucially important for the future of healthcare.
The analysis of multifactorial, multi-omic datasets with the PrecisionLife platform identifies disease-associated combinations of features, provides an important improvement in analytical capability that will be central to the delivery all aspects of precision medicine. This will enable the development of more detailed insights and personalized medicine strategies, with the potential to target specific subtypes of diseases with the greatest unmet need, such as TNBC.
We would like to acknowledge UK BioBank for providing us access to the data under application number 44288. Special thanks to Gert Moller, Chief Analytics Officer of PrecisionLife Ltd., who developed some of the novel technologies which are the foundation of PrecisionLife's platform and to the rest of the PrecisionLife team.
Financial support and sponsorship
Conflicts of interest
All authors are employees of PrecisionLife Ltd. (formerly RowAnalytics Ltd.), which developed the PrecisionLife analytics platform.
| References|| |
van Norman GA. Drugs, devices, and the FDA: Part 1: An overview of approval processes for drugs. JACC Basic Transl Sci 2016;1:170-9.
Schork NJ. Personalized medicine: Time for one-person trials. Nature 2015;520:609-11.
Kinch MS, Haynesworth A, Kinch SL, Hoyer D. An overview of FDA-approved new molecular entities: 1827-2013. Drug Discov Today 2014;19:1033-9.
Mullard A. 2018 FDA drug approvals. Nat Rev Drug Discov 2019;18:85-9.
Nosengo N. Can you teach old drugs new tricks? Nature 2016;534:314-6.
Pernas S, Tolaney SM. HER2-positive breast cancer: New therapeutic frontiers and overcoming resistance. Ther Adv Med Oncol 2019;11:1758835919833519.
Mellerup E, Andreassen OA, Bennike B, Dam H, Djurovic S, Jorgensen MB, et al
. Combinations of genetic variants associated with bipolar disorder. PLoS One 2017;12:e0189739.
Bycroft C, Freeman C, Petkova D, Band G, Elliott LT, Sharp K, et al
. The UK Biobank resource with deep phenotyping and genomic data. Nature 2018;562:203-9.
Morgan P, Brown DG, Lennard S, Anderton MJ, Barrett JC, Eriksson U, et al
. Impact of a five-dimensional framework on R and D productivity at AstraZeneca. Nat Rev Drug Discov 2018;17:167-81.
Boucher J, Kleinridders A, Kahn CR. Insulin receptor signaling in normal and insulin-resistant states. Cold Spring Harb Perspect Biol 2014;6. pii: a009191.
Yang J, Nie J, Ma X, Wei Y, Peng Y, Wei X. Targeting PI3K in cancer: Mechanisms and advances in clinical trials. Mol Cancer 2019;18:26.
PrecisionLife Ltd. Available from: https://precisionlife.com/. [Last accessed on 2019 Dec 02].
Xiong G, Stewart RL, Chen J, Gao T, Scott TL, Samayoa LM, et al
. Collagen prolyl 4-hydroxylase 1 is essential for HIF-1α stabilization and TNBC chemoresistance. Nat Commun 2018;9:4456.
Xiong G, Deng L, Zhu J, Rychahou PG, Xu R. Prolyl-4-hydroxylase α subunit 2 promotes breast cancer progression and metastasis by regulating collagen deposition. BMC Cancer 2014;14:1.
Wang T, Fu X, Jin T, Zhang L, Liu B, Wu Y, et al
. Aspirin targets P4HA2 through inhibiting NF-κB and LMCD1-AS1/let-7g to inhibit tumour growth and collagen deposition in hepatocellular carcinoma. EBioMedicine 2019;45:168-80.
Henry WS, Laszewski T, Tsang T, Beca F, Beck AH, McAllister SS, et al
. Aspirin suppresses growth in PI3K-mutant breast cancer by activating AMPK and inhibiting mTORC1 signaling. Cancer Res 2017;77:790-801.
Frisk G, Ekberg S, Lidbrink E, Eloranta S, Sund M, Fredriksson I, et al
. No association between low-dose aspirin use and breast cancer outcomes overall: A Swedish population-based study. Breast Cancer Res 2018;20:142.
Elwood PC, Pickering JE, Morgan G, Galante J, Weightman AL, Morris D, et al
. Systematic review update of observational studies further supports aspirin role in cancer treatment: Time to share evidence and decision-making with patients? PLoS One 2018;13:e0203957.
Tatsukawa H, Furutani Y, Hitomi K, Kojima S. Transglutaminase 2 has opposing roles in the regulation of cellular functions as well as cell growth and death. Cell Death Dis 2016;7:e2244.
Szondy Z, Korponay-Szabó I, Király R, Sarang Z, Tsay GJ. Transglutaminase 2 in human diseases. Biomedicine (Taipei) 2017;7:15.
Oh K, Ko E, Kim HS, Park AK, Moon HG, Noh DY, et al
. Transglutaminase 2 facilitates the distant hematogenous metastasis of breast cancer by modulating interleukin-6 in cancer cells. Breast Cancer Res 2011;13:R96.
Stamnaes J, Pinkas DM, Fleckenstein B, Khosla C, Sollid LM. Redox regulation of transglutaminase 2 activity. J Biol Chem 2010;285:25402-9.
Oh K, Lee OY, Park Y, Seo MW, Lee DS. IL-1β induces IL-6 production and increases invasiveness and estrogen-independent growth in a TG2-dependent manner in human breast cancer cells. BMC Cancer 2016;16:724.
Dubinsky R, Gray C. CYTE-I-HD: Phase I dose finding and tolerability study of cysteamine (Cystagon) in Huntington's disease. Mov Disord 2006;21:530-3.
Palanski BA, Khosla C. Cystamine and disulfiram inhibit human transglutaminase 2 via an oxidative mechanism. Biochemistry 2018;57:3359-63.
Mutschler J, Dirican G, Gutzeit A, Grosshans M. Safety and efficacy of long-term disulfiram aftercare. Clin Neuropharmacol 2011;34:195-8.
[Figure 1], [Figure 2], [Figure 3], [Figure 4]