We've updated our Privacy Policy to make it clearer how we use your personal data. We use cookies to provide you with a better experience. You can read our Cookie Policy here.

Advertisement

Drug Repurposing Strategies, Challenges and Successes

Various pills on a table with a magnifying glass over one, bringing into focus.
Credit: iStock.
Listen with
Speechify
0:00
Register for free to listen to this article
Thank you. Listen to this article using the player above.

Want to listen to this article for FREE?

Complete the form below to unlock access to ALL audio articles.

Read time: 32 minutes

Drug repurposing, also referred to as drug repositioning or reprofiling, is an approach that identifies new clinical applications for approved or investigational drugs outside of the compound’s originally intended uses. This approach commonly results from the serendipitous discovery of off-target effects or newly recognized actions of existing drugs. For example, thalidomide, which was notoriously removed from the market in 1961 after it was found to induce severe congenital disorders, was later repurposed as a treatment for leprosy after clinical case studies suggested potential effectiveness.1 Decades later, it was predicted to have anti-angiogenic and tumor-suppressing properties, which led to the success of thalidomide to treat multiple myeloma patients.2



More recently, the concept that existing drugs have potentially beneficial off-target effects or new usefulness has driven the design of more intentional drug repurposing strategies to identify promising drug candidates. An early example was when azidothymidine (AZT) was identified during the HIV/AIDS crisis as part of an emergency drug screening effort. It was originally developed in the 1960s as a therapeutic for cancers caused by environmental retroviruses, but it never progressed to use in the clinic.3 A collaboration between the National Cancer Institute and the pharmaceutical industry identified AZT as a potent anti-HIV compound,4 and this repurposing effort advanced the drug from in vitro testing to clinical use in 20,000 patients in less than 3 years.3 It became the first U.S. Food and Drug Administration (FDA)-approved drug to manage HIV in 1987.


The COVID-19 pandemic similarly supercharged advancements in drug repurposing, enabled by rapid patient data collection, drug prediction and compound screening. Over 100 unique repurposed drugs were moved into clinical trials for COVID-19 between January 2020 and December 2021, with 4 drugs receiving FDA approval or emergency use authorization and an additional 15 drugs recommended for off-label use by the National Institutes of Health or the Infectious Diseases Society of America.5 In addition to providing a more rapid development timeline in emergency situations, drug repurposing can reduce safety risks and lower costs, and it holds the potential for the discovery of new targets and pathways that can be further exploited to develop novel therapeutics.


The commercial promise of drug repurposing has not been fully realized, primarily due to patentability and market exclusivity challenges. However, the climate may now be changing as there is ever-increasing interest from clinicians, disease-focused foundations and philanthropic organizations that are desperately seeking quick solutions for the patients they represent. There is also great interest from government agencies responsible for military preparedness, biodefense and rapid response to pandemics for the same reasons. Moreover, drug repurposers are now able to use advanced in silico screening enabled by artificial intelligence (AI) and sophisticated tissue- and organ-level in vitro models that more accurately recapitulate human physiology to improve the selection of existing drugs for further pre-clinical testing and, eventually, clinical trials for new indications.


What is drug repurposing?

Many drugs exhibit polypharmacology in that they simultaneously act on multiple targets or disease pathways, regardless of whether they were “targeted” to a specific molecule. Leveraging this phenomenon, drug repurposing reevaluates previously developed compounds for different medical indications and, if successful, implements them in a new clinical context. Drug repurposing can mean repositioning a previously FDA-approved drug or redeveloping an unapproved drug that previously went through some pre-clinical and/or clinical evaluation stages but did not successfully enter the clinic. It also has been used in a more general sense to reposition any chemical, molecule, nutraceutical or even a live biotherapeutic product (probiotic) for a new application for which it was not designed.


Why is drug repurposing useful?

Traditional drug development can be slow, expensive and prone to high failure rates due to poor safety profiles, lack of efficacy or both. Including costs related to out-of-pocket expenses, time and abandoned trials, the average research and development investment for a new chemical entity is more than $2.5 billion per approved product.6 The new products also require 10–15 years of development,7,8 with a likelihood of FDA approval of less than 10% of Phase I candidates.9 Drug repurposing approaches can speed up or side-step some phases of drug development, resulting in potentially faster and cheaper development programs. For example, compounds that have already been evaluated in pre-clinical models and proven to be safe in human Phase I trials can be rapidly tested in humans at similar doses for other indications. Drug repositioning programs cost an average of $300 million, compared with the potential multi-billion-dollar investments for new chemical entities.8,9,10 Repurposing drugs through this mechanism can offer pronounced gains, including saving 6–7 years of time spent in early-stage research.11,12


A drug compound may potently interact with a target of interest or rapidly reverse disease progression in pre-clinical models; however, if the compound is not drug-like or cannot be formulated to behave in a biologically appropriate way in humans, it is unlikely to progress to the clinic. In fact, approximately 10–15% of compounds fail clinically due to poor drug-like properties and pharmaceutical companies have, therefore, focused on improving drug solubility, permeability, protein binding, metabolic stability and in vivo pharmacokinetics (PK) earlier in development.13,14 Thus, another advantage of drug repurposing is that these existing compounds usually incorporate chemical structures that have already been optimized for bioavailability and PK and, in some cases, formulation development has been completed as well. FDA-approved drugs may also have associated publications of structure–activity relationships (SARs) that describe many chemical structures that were evaluated during the development program. This can assist in drug repurposing as well as efforts to develop new and even more effective compounds.

It is important to clarify that drug repurposing is often more than simply identifying new uses for a drug that takes advantage of the same mechanism of action defined for the original indication. It is well established that many drugs actually alter the expression or function of multiple molecules and pathways in vivo that extend beyond their single presumed target. It is this multitude of effects that some researchers seek to harness during a drug repurposing program to target previously unknown pathways relevant for novel disease indications. The process of identifying and characterizing these targets can lead to new approaches for targeting particular proteins or other ligands, and these methods can be further exploited for additional indications or for the design of new chemical entities and hence, novel composition of matter intellectual property (IP).


Traditional drug discovery vs. drug repurposing

Drug repurposing can offer advantages over traditional drug discovery in terms of reducing cost, time-to-clinic and risk of failure for a drug product (Table 1). This is facilitated by a shorter pre-clinical stage and a potential reduction in clinical trial requirements if a drug has proven safe in previous Phase I clinical trials. For repurposed drugs, trial requirements may be limited to Phase II and III, which assess the efficacy of the recycled drug for a new indication and ensure that the compound is safe in patients with that particular disease. Although these features make drug repurposing attractive, the IP protections afforded to novel compounds versus repurposed drugs differ, which has significant commercial implications.


Traditional drug discovery frequently results in composition of matter patents that cover the novel chemical entity as well as methods of use patents, which together offer strong protection for the new therapy. In contrast, drug repurposing may only result in a method of use patent that describes the use of the existing drug for a new indication, which offers comparatively weak protection. However, the protection can be stronger and result in a new composition of matter IP in cases where the repurposed drug requires a new formulation or novel delivery mechanism. Commercial protections afforded to repurposed drugs are also limited to three years compared to five years for a new chemical entity. Further exploration of these considerations is covered in the “Remaining challenges in repurposing drugs” section.


Table 1: Comparison of traditional drug discovery and drug repurposing.

Feature

Traditional Drug Discovery

Drug Repurposing

Cost

>$2.5 billion6

<$500 million8,15

Time-to-clinic

10–15 years7,16

3–12 years9

Failure rate

90–95%7

25–70%9

Pre-clinical investigation

1.     Target validation

2.     Compound screen

3.     Lead optimization (SAR, drug-like properties, solubility, etc.)

1.     In silico screening

2.     Activity-based screens

Clinical trial requirements

Phases I–III

Phases II and III

IP protections

Composition of matter and method of use patents

Method of use patents

and composition of matter

in some cases

Commercial protections

New chemical entity protection

(up to 5 years after FDA approval)

New use/formulation exclusivity

(up to 3 years after FDA approval)

On-target vs off-target drug repurposing strategies

There are different flavors of drug repurposing based on whether on-target and off-target strategies are utilized. The on-target approach is the most straightforward as it involves linking a newly identified disease mechanism to a druggable target and then identifying existing compounds that interact with that target. This can be achieved by first compiling a list of potential targets based on the known biology of a disease or new insights from discovery programs. For diseases with many potential targets, the druggability of these targets can be estimated based on known features, such as the availability of a ligand-binding pocket within a protein. The highest value targets can then be matched with existing drugs originally designed for other applications via their defined molecular targets using databases, such as DrugBank,17 and subsequently evaluated for efficacy using in vitro activity-based assays.


In contrast, the off-target strategy seeks to identify known or unknown side effects of existing drugs that can be repositioned as an on-target effect for a different indication. This approach may utilize in silico and activity-based screening to assemble a list of candidate compounds that are expected to bind to a particular target, modulate a molecular signature, modify a gene network or reverse a phenotype. This approach can be more complex than on-target drug repurposing, but it offers potentially more options to identify an appropriate drug candidate as well as the opportunity to discover novel drug mechanisms, targets and pathways.


Drug repositioning approaches

Drug repositioning approaches can focus on identifying promising drugs for repurposing or on a disease state that lacks adequate treatment options. Drug-focused repurposing may expand the application of an existing drug to new indications based on off-label usage, analysis of abandoned drugs, review of drugs pulled from the market due to safety or efficacy issues or recycling of drugs that have reached the end of patent exclusivity. One commonly used approach uses compound screening, although it can be pursued in multiple ways, which may be broadly broken into two categories: in silico screening and activity-based screening. In many cases, the in silico and activity-based screens complement each other, with the in silico approach providing a means for down-selection resulting in prioritization of candidate compounds for activity-based screening. Conversely, the activity-based screens can also provide additional training data for the in silico approaches, potentially benefiting future development programs, in addition to validating predictions.


In silico screening

In silico drug screening can take many forms depending on the available data and development goals. Broadly, in silico screening translates large-format, high dimensional datasets (e.g., transcriptomic, proteomic or metabolomic profiles) into functional knowledge through a systems-level view of patterns and relationships between biological entities (e.g., genes, proteins, metabolites), often at multiple scales. Traditional in silico approaches have focused on evaluating drugtarget interactions by comparing chemical structures of candidate compounds with ones that are known to bind to a target protein or through use of low-throughput docking simulations that require known 3D chemical structures.18 More recently, AI approaches have been used to integrate heterogeneous multi-source data and identify previously unknown relationships between existing drugs and potential protein targets, disease states and phenotypes. These drug repositioning methods implement a range of computational models, including classical machine learning, network propagation and deep learning, to parse data from drug-centric, disease-centric and gene or protein-centric databases.18


At their core, many computational methods look for similarities or differences between entities of the same type (e.g., drug–drug comparisons) to uncover novel relationships on a large scale. For example, perhaps we have investigational compound A and want to know if there are any FDA-approved drugs that would behave similarly in vivo. It stands to reason that drugs with similar characteristics could exhibit consistent in vivo activity. Based on this assumption, a similarity metric can be calculated between drug A and all FDA-approved drugs based on available data, such as structures, shapes, protein targets, side effects, molecular activities, prior clinical usage and a range of other information (e.g., bioavailability, PK, etc.). Then, the similarity scores can be translated to predictions based on the distribution of similarity scores within a model containing potentially millions of elements (Figure 1).19 Models with increased complexity may incorporate additional concepts, like disease state, with defining characteristics, such as omics signatures, phenotype keywords and existing treatments. When there is little established knowledge about the mechanisms and pathways of a disease condition or phenotype, it can be useful first to generate or collate relevant, high-dimensional transcriptional, proteomic or metabolic datasets to define a disease signature. These data can be used to predict drugs by comparing the input disease signatures to existing signatures in drug perturbation databases.20,21 These approaches are covered in additional detail in the “Drug repurposing methodologies” section.

A toy subnetwork of interactions between drugs, targets and indications, demonstrating how information can be used to identify potential drugs for repurposing.


Figure 1: A toy subnetwork of interactions between drugs, targets and indications. A. The drug network can be constructed based on known relationships between entities with stronger connections given more weight. In this case, drug A is in pre-clinical development, so its connectivity within the network is sparser compared to the FDA-approved drugs B and C. B. It is possible to infer potential relationships across the network based on indirect connectivity. Drugs A and B have known three-dimensional (3D) chemical structures that are very similar, so it is possible to infer that drug A may also interact with a well-established target of drug B (Potential Relationship A). Although the 3D structural similarity between drugs A and C is unknown, they share a target, and therefore, there is an increased likelihood that aspects of their chemical structure are shared (Potential Relationship B). Finally, because drugs A and B are very structurally similar, drug B could be a potential candidate for drug repurposing for indication b, which does not have an FDA-approved treatment (Potential Relationship C). Credit: Technology Networks.


Activity-based screening

Regardless of the in silico screening method performed, it is essential to carry out activity-based screening to determine the biological activity of a compound in relevant assays and disease models. Optimally, testing is performed across multiple scales to capture the effects of a compound on a disease phenotype, determine its effects on cell viability (as an initial estimate of toxicity), evaluate drug binding to the putative target (if known) and assess efficacy in functionally relevant tissue, organ or whole organism disease models if available. Key considerations when selecting the primary assay include the in silico prediction type (e.g., drug–target binding vs. phenotype modulation), how closely the assay models the disease of interest and the time and/or cost of the assay.


When information about the mechanism and targets for a disease are limited, candidate drugs for repurposing can be assessed by tracking changes in the phenotype during and after treatment using a relevant model system. Because this type of screening is typically the most challenging and time-intensive, it is critical to balance the accuracy of the model system with time and cost constraints. The use of small whole-organism models (e.g. zebrafish or Xenopus laevis (Xenopus)) that capture multi-organ drug responses or human microphysiological systems, such as organoids and microfluidic organ-on-a-chip (organ chip) technology, that more closely recapitulate human physiology and disease states can offer a balance between biological relevance and program limitations.22,23,24,25 For example, in a recent study, a human liver chip model was shown to identify drug toxicities significantly more accurately than use of animal models, identifying hepatotoxic drugs with a sensitivity of 87% and a specificity of 100%.26 When executing a drug repurposing initiative with highly defined targets and mechanisms, activity-based screens that evaluate drug–target binding, such as radioligand binding assays, can also be used as an initial evaluation of drug activity on the target of interest, often at much lower cost and effort. After drug–target binding is established, then phenotypical screening approaches can be executed on a narrowed drug candidate list.


Drug repurposing methodologies

Multiple methods have been used to identify existing drugs for repurposing as well as to narrow down drug libraries for further in vitro and in vivo testing. One approach uses a disease-oriented methodology that involves matching of diseases with no treatment or ineffective treatments with drugs that exhibit better therapeutic impact. This can be accomplished using computational models that detect similarities between features of disease states and predict that similar drugs can be used. Another approach is therapeutic target-oriented as it looks to identify new targets for a given indication and then match existing drugs to the novel target. This can be accomplished through experimental discovery of new disease mechanisms and computational approaches that uncover druggable targets related to those mechanisms.


While blinded search or experimental screening methods were primarily used to serendipitously identify effective drugs without prior knowledge in the past, in silico drug repositioning can offer greater speed, reduced cost and the potential to identify novel drug mechanisms. Computational methods that may be used to identify and narrow potential drug candidates include the following (Table 2 and Figure 2):

  • Target-based
  • Structure-based
  • Signature-based
  • Pathway/network-based
  • Knowledge-based
  • Clinical data-based


These efforts also can combine in silico drug prediction with retrospective screening in patient databases to evaluate drugs in specific patient populations before further animal or human testing is performed.27 Although there has been an explosion in methods for drug repurposing in recent years, many of these approaches have not yet been proven out by FDA approval of the drugs that they identified. Thus, additional work is required to evaluate the best methods for finding clinically successful drugs.


Table 2: In silico methods to identify drugs for repurposing.

Method

Definition

Target-based

The use of protein data, binding pocket information, and known molecular interactions to identify new targets. It can also be combined with network-based approaches to identify new drug–target relationships.

Structure-based

The use of two dimensional (2D) or 3D chemical structures in combination with known drug activity data to predict novel functions, such as antibiotic activity. This method can also be combined with target-based approaches to assess binding site complementarity between a drug and a target.

Signature-based

The use of gene or other molecular signatures derived from disease or treatment omics data to identify similarities between diseases or drug mechanisms. Similarities can be used to predict drugs for repurposing.

Pathway or network-based

Building a disease model consisting of molecular and cellular interactions to develop and confirm hypotheses for drug repurposing projects. These models sometimes include relationships between entities, such as diseases, cellular processes, proteins and small molecules.

Knowledge-based

The use of automated literature and text mining through natural language processing to search through large bodies of knowledge. Knowledge can be extracted from published papers or reports and patterns identified to link diseases and mechanisms to repurposing candidates.

Retrospective clinical analysis

The use of electronic health records, clinical trial data, post-marketing surveillance data and insurance records to identify relationships between diseases, drugs and side effects. May be used alone or in combination with other methods.


Visual summary of the in silico methods that can be used alone or in combination to identify and assess the potential properties of repurposed drugs.


Figure 2: In silico methods that can be used alone or in combination to identify and assess the potential properties of repurposed drugs. Credit: Technology Networks.


Some success stories for repurposed drugs

Successes in drug repurposing have traditionally relied on a combination of serendipity and astute clinical and/or pharmacological observations to identify potential off-target impacts of existing drugs and these approaches continue to be essential to drug repurposing. In 2015, the widely used analgesic, aspirin, was approved for new uses in cardiovascular disease and colorectal cancer prevention based on retrospective clinical analyses and pharmacological analyses.15 Given the large populations at risk for these diseases and relatively wide safety margin of aspirin, this drug repurposing effort is poised to have huge long-term clinical impact. Although clinical findings continue to be an important aspect of drug repurposing, new computational and screening strategies have also had successes at repositioning old drugs for new purposes. However, most of the compounds identified by these newer approaches are still under evaluation in clinical trials. Therefore, additional time is needed to evaluate their clinical value completely.


Rare diseases represent an unmet patient need in drug development and are particularly well-suited to drug repurposing for both scientific and commercial reasons. Approximately 95% of the 7,000 rare diseases lack an FDA-approved treatment and many have poorly defined mechanisms of disease pathology;9 therefore, predictive repurposing by computational methods may be more useful here than for more common diseases with well-defined targets. Additionally, the Orphan Drug Act provides unique protections for rare disease research programs, including market protection against generic competition for seven years. One example of success in this area is the development of a new treatment for idiopathic multicentric Castleman disease, which involves hyperactivation of the body's immune system that causes uncontrolled organ dysfunction and is currently untreated in the majority of patients. Using a combination of knowledge, signature and pathway-based prediction approaches, as well as data from clinical trials,28 the monoclonal antibody, adalimumab, was predicted to be a potential therapy for these patients by the Every Cure LinkMap. The LinkMap ranked all 3,000 FDA-approved drugs to treat all 12,000 human diseases and generated 36 million evaluations. Originally approved in 2002 for the treatment of inflammatory arthritis, adalimumab was found to control Castleman disease effectively in recent patient case studies29 and a broader clinical trial is underway.


Similarly, in the case of the rare neurodevelopmental disorder, Rett Syndrome, the in silico tool, network models for causally aware discovery (NeMoCAD), was used to predict drugs that shift a disease transcriptomic signature towards a healthy state.30 Using both signature and pathway-based approaches, NeMoCAD correlates an input signature with transcriptional signatures of existing drugs, in addition to analyzing a network comprised of drug–gene and drug–drug interactions, to identify compounds capable of reverting a disease state to a healthy state. The FDA-approved drug vorinostat was predicted to reverse a Rett Syndrome signature in a Xenopus screen and it reduced symptoms in both Xenopus and mouse Rett models.30 A multisite clinical trial to evaluate repurposed FDA-approved drugs for Rett Syndrome, including vorinostat, is currently in development.31


During the COVID-19 pandemic, the same in silico NeMoCAD approach was used to predict drugs for repurposing against COVID-19 and was integrated with retrospective data from medical records to evaluate candidate drugs clinically.27 Together, these methods predicted that only a subset of statin drugs, including simvastatin, would protect against death from COVID-19 and this was confirmed in a retrospective analysis of 4,000 patients.27


The COVID-19 pandemic spurred multiple repurposing efforts, including the rapid development of a SARS-CoV-2 protein interaction network with 332 high-confidence protein–protein interactions between SARS-CoV-2 and human proteins.32 Further, druggable subnetworks within the protein interactome were matched with existing drugs and, amongst several hits, potent antiviral effects were observed when translation elongation mechanisms were targeted.32 Follow-up study of the elongation factor-1A inhibitor, plitidepsin, originally developed for multiple myeloma, demonstrated in vivo efficacy in mouse models of SARS-CoV-2.32 Evaluation of plitidepsin in a Phase I trial suggested a favorable safety profile in COVID-19 patients and encouraging preliminary effects on viral load and inflammatory biomarkers.33 A Phase II study of plitidepsin in immunocompromised COVID-19 patients is underway.


Drug repurposing approaches have also identified treatments against aggressive cancers for which the standard of care has not changed over multiple decades. By integrating signatures from cancer biopsies and existing drugs, tricyclic antidepressants and related inhibitors of G-protein coupled receptors were predicted to be active against small cell lung cancer (SCLC) and other high-grade neuroendocrine tumors that metastasize early.34,35 The predicted compounds were shown to induce apoptosis in SCLC cells that are chemonaive and chemoresistant as well as in human xenografts and endogenous tumors in mice.34 Using a similar approach, drugs were predicted to reverse disease-associated gene expression changes for breast, liver and colon cancers.36 The in silico potency of predicted drugs was shown to correlate with the efficacy of those drugs in pre-clinical models of these cancers.36 These developments have been translated into the development of a commercial technology (Artificial Intelligence for Drug Discovery), which utilizes public and private data combined with AI, machine learning and deep learning approaches to predict new drugs and targets.


There have also been successful efforts to use deep learning models to represent molecular structures in more useful ways and thereby improve their predictive capabilities across large chemical libraries. A significant advance in antibiotics discovery was made recently using a deep neural network model to predict antibiotic activity across molecules that are structurally distinct from traditional antibiotics.37 The model maps molecules into continuous vectors and compares these maps to a database of mapped compounds with known antibiotic properties. Using this approach, an enzyme inhibitor of c-Jun N-terminal kinase, halicin, which was originally developed to treat diabetes, was found to exhibit broad-spectrum antibiotic activity in mice. Halicin is currently undergoing additional preclinical development in the commercial sector in preparation for clinical trials.


This work also has been extended to identify additional antibiotics, especially against challenging pathogens such as Gram-negative and treatment-resistant bacteria.38,39 By expanding the neural network model to include both antibiotic activity and human cell cytotoxicity data for > 39,000 previously characterized compounds, the authors were able to predict the potential antibiotic efficacy and toxicity for > 12 million compounds.39 From this, a new structural class of antibiotics with favorable toxicology and chemical properties was identified that demonstrated activity against methicillin-resistant Staphylococcus aureus (MRSA) and vancomycin-resistant enterococci in murine skin and thigh infections.39


In addition to in silico approaches, innovative experimental model systems have been implemented to accelerate drug repurposing. A human organ-on-a-chip microfluidic culture model of the lung alveolus (lung alveolus chip) was lined by human alveolar epithelium interfaced with lung microvascular endothelium and respectively exposed to air and fluid flow while both experienced cyclic breathing-like deformations. This model system led to the discovery that innate immune responses to viral infection are influenced by physical forces.25 Analysis of this mechanism resulted in identification of the receptor for advanced glycation end products (RAGE) as a potential new target for treatment in patients infected with influenza and existing drugs were identified using a shared target-based method. The drug azeliragon, a RAGE inhibitor that demonstrated safety but not efficacy in Phase III clinical trials for Alzheimer’s disease and diabetic nephropathy, was shown to suppress host-damaging inflammatory responses to viral infection in this model.25 These organ chip data were included in an investigational new drug (IND) application to the FDA to initiate Phase II trials of azeliragon in patients with COVID-19.


Human organoids composed of organ-specific stem cells have also been increasingly used for drug repurposing, especially for rapid screening programs during the COVID-19 pandemic. Self-organizing organoids more accurately recapitulate the 3D tissue microenvironment compared to 2D cell culture systems and, thus, can be more predictive of drug function in vivo. Since SARS-CoV-2 primarily infects the respiratory tract, a human lung organoid model was generated and used to screen FDA-approved drugs for repurposing.40 Imatinib, mycophenolic acid and quinacrine dihydrochloride were found to inhibit entry of SARS-CoV-2 in the lung organoid, as well as a second organoid model of the colon.40 Imatinib was included in multiple COVID-19 clinical trials, and mortality, duration of mechanical ventilation and length of intensive care unit stay were lower for patients treated with imatinib.41,42 However, it is hypothesized that these outcomes are primarily due to imatinib’s attenuation of vascular leakage under inflammatory conditions and not its antiviral properties.41


Philanthropic organizations that desire near term positive outcomes and are less focused on commercial success have pursued drug repurposing with the goal of providing patients in developing countries with critically needed drugs for pennies a day. One example of such efforts is the repurposing of the existing antimalarial drug amodiaquine by first testing in human lung airway chips (lined by bronchial airway cells interfaced with microvascular endothelium) infected with pseudotyped SARS-CoV-2, which demonstrated superior performance of amodiaquine over other antivirals.22 After identification and testing in this human-relevant system, and confirmation in studies with hamsters infected with infectious SARS-CoV-2, the drug was included in ANTICOV clinical trials in Africa in 2021 with the goal of treating mild and moderate cases of COVID-19 early.43 The ANTICOV trials investigated multiple repurposed drugs and helped address the relative neglect of clinical research for outpatient treatment in this region, aiming to avoid overwhelming fragile and overburdened health systems.43


Drug repurposing also can be used as a short cut to enable discovery of new drugs, with candidates from drug repurposing programs serving as launching points for design of improved, but related, compounds. A recent example of this approach was in the development of Paxlovid, which is comprised of two compounds: the new COVID-19 antiviral nirmatrelvir and the repurposed HIV antiviral ritonavir. Nirmatrelvir originated from an antiviral developed in 2003 for the treatment of severe acute respiratory syndrome (SARS) and medicinal chemistry was used to improve its oral bioavailability in 2020.44 The second component, ritonavir, is not active against SARS-CoV-2, but is used to prevent the breakdown of nirmatrelvir before it can effectively act on SARS-CoV-2’s main protease. Paxlovid received emergency use authorization in December 2020 and became the first oral antiviral pill to be FDA-approved for COVID-19 in 2023.


In addition to using signature-based, network-based, and retrospective clinical methods for repurposing drugs during the COVID-19 pandemic, numerous drug repositioning efforts used molecular structure-based approaches to target the viral surface spike protein. However, since surface binding sites on the spike protein readily mutate, on-going efforts against COVID-19 and related viruses are focused on identifying highly conserved, internal sites within the spike protein. Molecular dynamics simulations and AI-based blind docking are being used at the Wyss Institute for Biologically Inspired Engineering to identify a potential binding site, predict existing compounds with strong binding affinity and then apply medicinal chemistry to design novel chemical compounds with improved activities based on the molecular structure of the existing compounds.


Medicinal chemistry was also used in combination with a drug repurposing program, which identified an existing compound (SNC80) as a lead candidate to induce slowing of cell, tissue and organ physiology and metabolism to enhance organ preservation.45 SNC80 is a delta opioid receptor agonist that was originally developed as a non-addictive analgesic but never progressed to the clinic. To establish if its originally defined mechanism is responsible for physiological slowing, the distal basic nitrogen was removed using medicinal chemistry, which resulted in a 1,000-fold reduction in its activity at the delta opioid receptor.45 Even without the delta opioid activity, this analog with a novel structure retained its physiological slowing effects, demonstrating the potential power of polypharmacology in drug repurposing.


Collectively, these stories highlight the wide range of approaches and applications that have been pursued for drug repurposing with each instance demonstrating multiple scientific and commercial pathways for therapeutic development. Despite the significant challenges associated with drug repositioning, including misaligned commercial incentives and lack of adequate patent protection for innovations, this approach has contributed to important developments for patients and holds even greater promise for the future.


Remaining challenges in repurposing drugs

Drug repurposing approaches can offer gains in terms of total development cost, time to market and failure rate due to safety concerns; however, repurposed drugs can still fail at later phases of development due to lack of efficacy in clinical populations, similar to novel compounds.15 For pharmaceutical companies, repurposed drugs also present a commercial risk due to challenges of patentability and market exclusivity for the repositioned drug.15,46 In some cases, repurposed indications are discovered by the original drug developer in pre-clinical studies or as a side effect during clinical trials and patented. A famous example of this was when development of the hypertensive drug, sildenafil, was pivoted to the development of Viagra as a treatment for impotence after clinical trials revealed potent off-target effects.47 Furthermore, the literature contains many speculated uses for investigational drugs, creating prior art that can interfere with patenting.


Novel FDA-approved drugs are legally and commercially protected by composition of matter patents and new chemical entity status; however, after these protections expire, multiple manufacturers can make generic versions of the drug. Although the FDA offers three years of market exclusivity for the new use of previously marketed drugs, pharmaceutical companies may have limited ability to recoup the costs of developing the drug for additional indications due to this limited time window and off-label use that devalues the product.5,15 In fact, no generic drug has been successfully repurposed through to FDA approval without modification of its dose or dosing schedule, updating the route of administration or combining the drug with a second repurposed compound, each of which provide additional patent protection.46 Thus, without IP and/or market exclusivity protections, a company typically cannot justify investing the research and development necessary to get approval for the new indication. Measures to incentivize drug repurposing, potentially through extended exclusivity periods and royalty agreements with generic drug manufacturers, could improve the attractiveness of this path.


Drug repurposing is also hindered by barriers to accessing data from failed trials and the resulting shelved compounds. Although there is information about most FDA-approved drugs to support repurposing efforts, compounds that do not make it to the clinic often “disappear” when their development is abandoned, with trial data and results left unpublished.48 Wider sharing of industry-generated pre-clinical and clinical compounds in large libraries and their associated data from human clinical trials would greatly enhance opportunities for drug repurposing. The lack of published negative outcomes also impacts the performance of in silico models, which benefit from more complete and accurate biological information, including knowledge of what types of compounds do not revert particular diseases or fail to target a given protein or pathway reliably. Furthermore, many in silico models are built on the premise that similar compounds will have similar properties and downstream effects. Although this approach is well-supported and offers a good starting point for investigation, it can miss very novel mechanisms and lead to compound discovery within well-trodden pharmacological spaces.19 Despite significant advances in experimental screening models, bioengineered systems still have limitations on their complexity and don’t capture all aspects of in vivo biology; they also have restricted lifespans, which limits analysis of effects on chronic conditions and sometimes lack appropriate immunomodulation.49 


In this article, we explore the evolution of drug repurposing from a completely serendipitous occurrence to a more formalized endeavor that involves advanced computational approaches and bioengineered test beds. Although scientific and commercial challenges remain, the range of potential applications – from rare diseases to infectious disease emergencies to philanthropic efforts – provides hope for patients with few desirable treatment options.

1.        Barnhill RL, McDougall AC. Thalidomide: Use and possible mode of action in reactional lepromatous leprosy and in various other conditions. J Am Acad Dermatol. 1982;7(3):317-323. doi:10.1016/S0190-9622(82)70118-5

2.        Singhal S, Mehta J, Desikan R, et al. Antitumor activity of Thalidomide in refractory multiple myeloma. N Engl J Med. 1999;341(21):1565-1571. doi:10.1056/NEJM199911183412102

3.        Hirsch MS. Azidothymidine. J Infect Dis. 1988;157(3):427-431. doi:10.1093/infdis/157.3.427

4.        Fischl MA, Richman DD, Grieco MH, et al. The efficacy of Azidothymidine (AZT) in the treatment of patients with AIDS and AIDS-related complex. N Engl J Med. 1987;317(4):185-191. doi:10.1056/NEJM198707233170401

5.        Greenblatt W, Gupta C, Kao J. Drug repurposing during the COVID-19 pandemic: Lessons for expediting drug development and access. Health Aff. 2023;42(3):424-432. doi:10.1377/hlthaff.2022.01083

6.        DiMasi JA, Grabowski HG, Hansen RW. Innovation in the pharmaceutical industry: New estimates of R&D costs. J Health Econ. 2016;47:20-33. doi:10.1016/j.jhealeco.2016.01.012

7.        Roessler HI, Knoers NVAM, van Haelst MM, van Haaften G. Drug repurposing for rare diseases. Trends Pharmacol Sci. 2021;42(4):255-267. doi:10.1016/j.tips.2021.01.003

8.        Nosengo N. Can you teach old drugs new tricks? Nature. 2016;534(7607):314-316. doi:10.1038/534314a

9.        Biotechnology Innovation Organization, Informa Pharma Intelligence, QLS Advisors. Clinical development success rates and contributing factors 2011–2020. February 2021. Accessed November 27, 2023. https://www.bio.org/clinical-development-success-rates-and-contributing-factors-2011-2020

10.      Deloitte Center for Health Solutions. Seize the digital momentum: Measuring the return from pharmaceutical innovation 2022. January 2023. Accessed January 5, 2024. https://www2.deloitte.com/content/dam/Deloitte/uk/Documents/life-sciences-health-care/deloitte-uk-seize-digital-momentum-rd-roi-2022.pdf

11.      Mullard A. Drug repurposing programmes get lift off. Nat Rev Drug Discov. 2012;11(7):505-506. doi:10.1038/nrd3776

12.      Mullard A. Bank tests drug development waters. Nat Rev Drug Discov. 2014;13(9):643-644. doi:10.1038/nrd4412

13.      Kola I, Landis J. Can the pharmaceutical industry reduce attrition rates? Nat Rev Drug Discov. 2004;3(8):711-716. doi:10.1038/nrd1470

14.      Sun D, Gao W, Hu H, Zhou S. Why 90% of clinical drug development fails and how to improve it? Acta Pharm Sin B. 2022;12(7):3049-3062. doi:10.1016/j.apsb.2022.02.002

15.      Pushpakom S, Iorio F, Eyers PA, et al. Drug repurposing: Progress, challenges and recommendations. Nat Rev Drug Discov. 2018;18(1):41-58. doi:10.1038/nrd.2018.168

16.      Paul SM, Mytelka DS, Dunwiddie CT, et al. How to improve R&D productivity: the pharmaceutical industry’s grand challenge. Nat Rev Drug Discov. 2010;9(3):203-214. doi:10.1038/nrd3078

17.      Wishart DS, Feunang YD, Guo AC, et al. DrugBank 5.0: A major update to the DrugBank database for 2018. Nucleic Acids Res. 2018;46(D1):D1074-D1082. doi:10.1093/nar/gkx1037

18.      Luo H, Li M, Yang M, Wu FX, Li Y, Wang J. Biomedical data and computational models for drug repositioning: A comprehensive review. Brief Bioinform. 2021;22(2):1604-1619. doi:10.1093/bib/bbz176

19.      Hodos RA, Kidd BA, Shameer K, Readhead BP, Dudley JT. In silico methods for drug repurposing and pharmacology. WIREs Syst Biol Med. 2016;8(3):186-210. doi:10.1002/wsbm.1337

20.      NIH LINCS Program. lincsproject.org. Updated January 5, 2024. Accessed February 14, 2024. https://lincsproject.org/.

21.      Davis AP, Wiegers TC, Johnson RJ, Sciaky D, Wiegers J, Mattingly CJ. Comparative toxicogenomics database (CTD): Update 2023. Nucleic Acids Res. 2023;51(D1):D1257-D1262. doi:10.1093/nar/gkac833

22.      Si L, Bai H, Rodas M, et al. A human-airway-on-a-chip for the rapid identification of candidate antiviral therapeutics and prophylactics. Nat Biomed Eng. Published online 2021. doi:10.1038/s41551-021-00718-9

23.      Sperry MM, Novak R, Keshari V, et al. Enhancers of host immune tolerance to bacterial infection discovered using linked computational and experimental approaches. Adv Sci. Published online 2022. doi:10.1101/2021.09.30.462576

24.      Dasgupta Q, Jiang A, Wen AM, et al. A human lung alveolus-on-a-chip model of acute radiation-induced lung injury. Nat Commun. 2023;14(1):6506. doi:10.1038/s41467-023-42171-z

25.      Bai H, Si L, Jiang A, et al. Mechanical control of innate immune responses against viral infection revealed in a human lung alveolus chip. Nat Commun. 2022;13(1):1928. doi:10.1038/s41467-022-29562-4

26.      Ewart L, Apostolou A, Briggs SA, et al. Performance assessment and economic analysis of a human Liver-Chip for predictive toxicology. Commun. Med. 2022;2(1):154. doi:10.1038/s43856-022-00209-1

27.      Sperry MM, Oskotsky TT, Marić I, et al. Target-agnostic drug prediction integrated with medical record analysis uncovers differential associations of statins with increased survival in COVID-19 patients. PLoS Comput Biol. 2023;19(5). doi:10.1371/journal.pcbi.1011050

28.      The ROADMAP Project. everycure.org/roadmap. Published 2023. Accessed December 5, 2023. https://www.everycure.org/roadmap

29.      Mumau M, Irvine A, Ma C, et al. A Translational approach to identifying and targeting TNF signaling in idiopathic multicentric Castleman disease. Blood. 2023;142(Supplement 1):3921-3921. doi:10.1182/blood-2023-188076

30.      Novak R, Lin T, Kaushal S, et al. Target-agnostic discovery of Rett syndrome therapeutics by coupling computational network analysis and CRISPR-enabled in vivo disease modeling. bioRxiv. Published online 2022. doi:10.1101/2022.03.20.485056

31.      Bonvissuto D. VUMC receives $13 million grant to coordinate multisite clinical trial for Rett syndrome treatment. VUMC Reporter. January 9, 2024.Accessed January 17, 2024. https://news.vumc.org/2024/01/09/vumc-receives-13-million-grant-to-coordinate-multisite-clinical-trial-for-rett-syndrome-treatment/

32.      Gordon DE, Jang GM, Bouhaddou M, et al. A SARS-CoV-2 protein interaction map reveals targets for drug repurposing. Nature. 2020;583(7816):459-468. doi:10.1038/s41586-020-2286-9

33.      Varona JF, Landete P, Lopez-Martin JA, et al. Preclinical and randomized Phase I studies of plitidepsin in adults hospitalized with COVID-19. Life Sci Alliance. 2022;5(4):e202101200. doi:10.26508/lsa.202101200

34.      Jahchan NS, Dudley JT, Mazur PK, et al. A drug repositioning approach identifies tricyclic antidepressants as inhibitors of small cell lung cancer and other neuroendocrine tumors. Cancer Discov. 2013;3(12):1364-1377. doi:10.1158/2159-8290.CD-13-0183

35.      Wang J, Byers LA. Teaching an old dog new tricks: Drug repositioning in small cell lung cancer. Cancer Discov. 2013;3(12):1333-1335. doi:10.1158/2159-8290.CD-13-0790

36.      Chen B, Ma L, Paik H, et al. Reversal of cancer gene expression correlates with drug efficacy and reveals therapeutic targets. Nat Commun. 2017;8(1):16022. doi:10.1038/ncomms16022

37.      Stokes JM, Yang K, Swanson K, et al. A deep learning approach to antibiotic discovery. Cell. 2020;180(4):688-702.e13. doi:10.1016/j.cell.2020.01.021

38.      Liu G, Catacutan DB, Rathod K, et al. Deep learning-guided discovery of an antibiotic targeting Acinetobacter baumannii. Nat Chem Biol. 2023;19(11):1342-1350. doi:10.1038/s41589-023-01349-8

39.      Wong F, Zheng EJ, Valeri JA, et al. Discovery of a structural class of antibiotics with explainable deep learning. Nature. Published online December 20, 2023. doi:10.1038/s41586-023-06887-8

40.      Han Y, Duan X, Yang L, et al. Identification of SARS-CoV-2 inhibitors using lung and colonic organoids. Nature. 2021;589(7841):270-275. doi:10.1038/s41586-020-2901-9

41.      Aman J, Duijvelaar E, Botros L, et al. Imatinib in patients with severe COVID-19: A randomised, double-blind, placebo-controlled, clinical trial. Lancet Respir Med. 2021;9(9):957-968. doi:10.1016/S2213-2600(21)00237-X

42.      Duijvelaar E, Schippers JR, Smeele PJ, et al. Long-term clinical outcomes of COVID-19 patients treated with imatinib. Lancet Respir Med. 2022;10(4):e34-e35. doi:10.1016/S2213-2600(22)00052-2

43.      Drugs for Neglected Diseases initiative. anticov.org. Updated November 17, 2023. Accessed December 18, 2023. https://anticov.org/about-anticov/

44.      Halford B. The path to Paxlovid. ACS Cent Sci. 2022;8(4):405-407. doi:10.1021/acscentsci.2c00369

45.      Sperry MM, Charrez B, Fotowat H, et al. Identification of pharmacological inducers of a reversible hypometabolic state for whole organ preservation. eLife. 2024;13:RP93796. doi: 10.7554/eLife.93796.1

46.      Novac N. Challenges and opportunities of drug repositioning. Trends Pharmacol Sci. 2013;34(5):267-272. doi:10.1016/j.tips.2013.03.004

47.      Begley CG, Ashton M, Baell J, et al. Drug repurposing: Misconceptions, challenges, and opportunities for academic researchers. Sci Transl Med. 2021;13(612). doi:10.1126/scitranslmed.abd5524

48.      Krishnamurthy N, Grimshaw AA, Axson SA, Choe SH, Miller JE. Drug repurposing: A systematic review on root causes, barriers and facilitators. BMC Health Serv Res. 2022;22(1):970. doi:10.1186/s12913-022-08272-z

49.      Loewa A, Feng JJ, Hedtrich S. Human disease models in drug development. Nat. Rev. Bioeng. 2023;1(8):545-559. doi:10.1038/s44222-023-00063-3