Personalized drug discovery from rare diseases to cancer

Abstract

Precision medicine selects individual treatments from a small toolbox of therapies already available. Personalized drug discovery (PDD) brings this concept one step ahead, developing drugs for individual patient’s diseases. Currently, PDD is being developed for rare disorders, where conventional drug discovery paradigms are non-applicable due to small patient sizes, but also for cancer. As a new frontier in precision medicine, PDD should be upscaled in rare and non-rare diseases.

Personalized/precision medicine

Precision medicine (the term often used interchangeably with “personalized medicine”) aims at improving prevention, diagnosis, and treatment of diseases based on the molecular features of the patient and the disease. Started from the foundational document introducing the power of genomics in medical applications¹, it gradually moved towards incorporation of broader tools of molecular analyses^2–4. In cancer, one of the most vivid illustrations of precision medicine is the concept of molecular tumor boards—multi-specialty teams comprising clinical (oncologists, surgeons, pathologists, pharmacists…) and molecular (geneticists, bioinformaticians, molecular biologists…) experts meeting regularly to identify individual treatment opportunities, especially for the cancer patients failing on standard therapies^5–7. Rare neurological disorders are another domain where precision medicine has found wide-spread applications^8,9. Outside of these two areas, translation of precision medicine approaches for complex diseases has so far been less straightforward^3,10. It can be stated that precision medicine is a field in transition, finding more applications in clinical practice upon adoption of additional or better molecular analysis and regulatory tools⁴. Despite these ongoing developments, one element of precision medicine proposed back in 1999 in Francis Collins’ visionary lecture¹, related to the development of new drugs, is still largely missing. Francis Collins wrote: “Ultimately, the real payoff of genetic research will be the development of new gene therapies and drug therapies, but they will generally require many more years of intensive research”. Indeed, although discoveries of molecular basis of diseases constantly feed on new drug developments, the drug discovery itself is largely not part of the precision medicine approach, which rather mainly operates in the framework of choosing from a limited list of therapies, their dosages and possible combinations. Personalized drug discovery (PDD)—which is defined as drug discovery and development tailored to small group of patients (or even individual patients) characterized by the same unique molecular features of disease, rare on the background of larger patient populations—is typically considered as something of the future. While this Perspective argues that PDD is hopefully closer than that, the traditional drug discovery and development strategies do speak strongly against their application in conjunction with precision medicine. Thus, I next briefly overview these current strategies, prior to discussing new paradigms.

Traditional drug discovery and development

The typical model of drug discovery and development uniformly applied across diseases, countries, and economy sectors is a series of standard R&D (research and development) steps comprising target identification and validation, hit identification, hit-to-lead optimization, preclinical ADME (absorption, distribution, metabolism, and excretion) and toxicity assessment, followed by phase I to III clinical trials¹¹. Each of these steps can be viewed as a “filtering” process and is characterized by a given attrition rate, which upon summing up result in the success rate of drug discovery and development to mostly reside in the 1.3–5% zone, depending on the disease type¹². Worryingly, this success rate gradually decreases, which is diagnosed as the “productivity crisis in pharmaceutical R&D”^11–13—despite all the R&D advances and the growing investments. The International Federation of Pharmaceutical Manufacturers & Associations (IFPMA) in their 2024 report estimated the cost of one drug to pass the full path of drug discovery and development to average USD 2.6 billion and to take 10–15 years (ifpma.org/publications/alwaysinnovating-pharmaceutical-industry-facts-figures/). Although other analyses provide a lower estimate of ca. USD 1 billion per drug¹⁴, it is evident that the standard R&D pipeline remains applicable only to large patient groups and cannot be tailored for any individualized drug discovery nor patients representing rare diseases. Apart from the costs that can never be reimbursed due to the small patient populations, the geographically dispersed nature of these populations contributes to the inapplicability of the conventional drug discovery and development model to rare diseases¹⁵. Is there a way out of this limitation?

Rare diseases and the non-applicability of the conventional drug discovery paradigms

Rare diseases/disorders affect less than 200,000 individuals in the US (FDA definition), or less than 5 in 10,000 individuals (EMA definition), which converges on <1 in 1700–2000 individuals. Despite this rarity of each individual rare disease, with over 10,000 conditions identified, they cumulatively affect up to 10% of the population¹⁶, approaching the global diabetes prevalence of 11% and even exceeding it in many, e.g., European, countries¹⁷.

Regulatory agencies since a long while recognized the problem that less than 5% of rare diseases have approved drugs¹⁸. In response, mechanisms have been developed, such as the Orphan Drug Act in the US (1983), similar incentive in the EU (1999), or the Accelerated Approval pathway (1992) adopted by the FDA¹⁹. Although having contributed to development of ca. 900 new drugs since their implementation, these measures are still considered as barely scratching the surface as compared to the need¹⁹. Diseases that are so far most targeted through the Orphan Drug Act mechanisms are HIV infection, multiple myeloma, non-small cell lung carcinoma, and other cancers, with rarer diseases underrepresented¹⁸, recently promoting FDA to adopt a new initiative (fda.gov/news-events/press-announcements/fda-advances-drug-repurposing-address-unmet-medical-needs). An urgent request for new models of drug discovery and development has been well-recognized^15,20, and two approaches are rapidly gaining momentum: gene therapies (gene and RNA-based) and drug repositioning (aka repurposing). Both are discussed in the next section.

Personalized drug discovery (PDD) for rare disorders

Gene replacement, gene editing, and RNA-based antisense oligonucleotides (ASO) represent examples of gene therapies rapidly developing for rare disorders^21–23. An example of gene replacement is the treatment developed for the aromatic L-amino acid decarboxylase (AADC) deficiency, where a functional copy of the AADC gene is delivered to patients’ brains, restoring dopamine production and patients’ conditions²⁴. Gene editing has shown promising results in the mouse model of alternating hemiplegia of childhood—a pediatric disorder originating through missense ATP1A3 variants²⁵, but the most illustrative example is approval of Casgevy, a CRISPR/Cas9 gene-editing technology, for the treatment of sickle-cell disease²⁶. Development of ASO-based treatments is currently gaining momentum for several disorders amenable to this approach, with one of the best-known examples being Nusinersen—a splice-switching ASO restoring SMN2 protein expression in Spinal Muscular Atrophy²⁷. Current challenges of gene therapies relate to the route of their delivery (often requiring packaging into adeno-associated viruses or lentiviruses that need to be injected to the diseased tissue such as CNS), exacerbating costs (https://intuitionlabs.ai/articles/gene-therapy-pricing-economics), and applicability of the available approaches only to a subset of genetics-driven medical conditions^28,29. Efforts to overcome some of these hurdles are being made, with an example of the non-profit N-Lorem project that offers free individualized ASO-based therapy development for ultra-rare disease patients in the US³⁰, but the long-term sustainability and the upscaling potential of this model need to be verified.

Drug repositioning is, in several aspects, complementary to gene therapy^21,31,32. Ideally started by elucidation of the molecular deficiency at the core of a disease, it employs high-throughput screening (HTS) of libraries of approved drugs or clinical drug candidates, with the hope to identify a drug that can be repurposed for the new indication³³. In the ideal scenario, the repurposed drug will rapidly progress to the clinics with relatively low costs and standard delivery routes. An example of successful drug repurposing is the development of zinc acetate treatment for a rare pediatric encephalopathy caused by pathogenic missense mutations in GNAO1. In this case, understanding of the molecular etiology of the disease, followed by assay development, HTS, and drug validation in cellular and insect models of the disease were published in 2022, followed by mouse safety and first-in-human application in 2024, demonstrating strong improvement of the patient conditions, laying the ground for a larger clinical trial completed in 2025. As a result, this safe, cheap, and highly feasible therapy is applied in patients world-wide, although different pathogenic variants respond to the treatment with varying efficiencies^34–36. This pipeline is being replicated to other rare disorders^21,37. A limitation of drug repositioning is imposed by the limited number of available drugs (ca. 5000, the number that can be increased to ca. 20,000 if clinical drug candidates are included, https://go.drugbank.com/stats)—numbers that are miniscule as compared to the library sizes of hundreds of thousands to millions of compounds used in HTS-based de novo drug discovery. However, it can be argued that the biologically and clinically validated chemical diversity subspace is by far more relevant as the source of new drug indications than the rest of the essentially endless chemical diversity universe³⁸. While proper translation of the drug repositioning campaigns, especially when run in academic settings, to clinical developments is in need³⁹, numerous successful examples of drug repurposing⁴⁰, especially in orphan drug designation areas⁴¹, do argue for the upscaled implementation of this approach. A better integration of drug screening, drug testing, and clinical translation—the integration that will benefit from partnerships among academia, hospitals, industry, patient associations, and regulatory authorities⁴²—is needed to increase the productivity of drug repositioning.

PDD for cancer

Cancer, despite all the efforts, remains a major health burden, with >10 million people dying from cancer annually on the global scale (who.int/en/news-room/fact-sheets/detail/cancer), and the global economic loss from the disease is estimated as >$0.5 trillion annually⁴³. Progress in fighting cancer is perplexingly modest. When comparing the period of 2006–2010 to the period of 1981–1985, only a 5% decrease of cancer-driven mortality was recorded in countries with medium-high HDI (human development index; a 17% decrease was seen in countries with very high HDI). This is in opposition to a more significant drop by 19% and 53%, respectively, when analyzing mortalities due to cardiovascular diseases⁴⁴. Clearly, novel approaches to combat cancer must be developed, and cancer PDD should be viewed from this angle.

Three lines of PDD in cancer should be considered: personalized gene therapy, personalized drug repositioning, and personalized immunotherapy. Of these, the latter is currently being developed most actively, with approaches such as cancer vaccines and CAR-T cells being inherently personalized. The foundational mRNA technology developed for personalized cancer vaccines (and “repurposed” against COVID-19) was recognized in 2023 with Nobel Prize in Physiology or Medicine to Katalin Karikó and Drew Weissman. In addition to mRNA-based, cancer vaccines can be DNA-based, peptide-based, or cell-based^45–47. In the latter case, patient’s antigen-presenting cells such as dendritic cells are first isolated, transfected with the desired antigens, and propagated before reintroduction to the patient, as is the case with the so far only one personalized vaccine given FDA approval: Sipuleucel-T against advanced prostate cancer⁴⁸. Multiple other cancer vaccines are in clinical trials against pancreatic cancer, melanoma, glioblastoma, and other cancers^45–47,49. Cancer vaccines can be “shared” (targeting multiple patients) or personalized (raised against tumor-associated or tumor-specific antigens identified for a given patient)^45–47. Production of personalized vaccine can be fast (around 2 months), which is crucial for the application of this PDD approach in cancer.

CAR-T (chimeric antigen receptor—T-cell) therapy is another field of personalized immunotherapy, where patient-derived T-cells are isolated, genetically modified to target the patient’s cancer cells, amplified, and injected back to the patient^50,51. Several CAR-T therapies have been approved by FDA to treat B-cell malignancies, while effectiveness of the CAR-T approach for other blood malignancies and solid tumors is still to be demonstrated^51,52. High costs, limited application scope, and side effect risks are among the current limitations of these revolutionary anti-cancer developments.

While the CAR-T technology can be considered as a gene therapy targeting immune cells, gene therapies targeting cancer cells are also being developed. An illustrative example is Gendicine, a recombinant adenovirus expressing functional TP53—the gene most frequently mutated in cancers—approved in 2003 in China for the treatment of head and neck carcinoma⁵³. While many other developments are underway⁵⁴, they are currently not personalized, but instead targeted at many cancer patients. The potential of personalization of anticancer gene therapies is, however, clear, and will hopefully be exploited in the near future. Although drug repositioning for cancer has been widely studied, the list of drugs officially approved for oncology indications remains short⁵⁵, with many repositioning drugs, such as metformin or clofazimine, still staying in the investigational / off-label status^55–57.

PDD for cancer and other diseases: the near future

Challenges ahead of PDD in cancer and other diseases are numerous, including high costs of personalized therapy development, time needed for such development, and regulatory hurdles. However, rapid progress along multiple possible lines of PDD permits a degree of optimism on the chances of PDD to find its way towards daily clinical practice in the near future. At the ending-section of this Perspective, I would like to emphasize the steps of PDD that need to be elaborated for that.

1. Rapid assessment of the “molecular portrait” of the disease, with the combination of genomics and other relevant omics tools, following the already well-established principles of molecular tumor boards^5–7, with the goal of identifying molecular drivers and molecular vulnerabilities of the disease.

2. Rapid establishment of HTS-ready personalized in vitro disease models, such as those utilizing patient-derived tumor organoids or tumor-on-chip in cancer^58,59, iPSC-derived neuronal precursors in neurological disorders⁶⁰, or mutant protein-based assays^34,37.

3. HTS of approved drugs/clinical grade drug candidates^34,61,62, of ASOs or other gene therapies^28,30, of personalized immunotherapies^47,51, or other modalities, utilizing the models established above.

4. Hit validation in orthogonal assays, where possible, including animal models^29,34,35,63.

5. Rapid clinical translation, involving cooperation among different stakeholders.

Time, costs, feasibility, and regulatory practices are the critical elements that need to be optimized in order for PDD to become reality. Timewise, individualized drug development for GNAO1 encephalopathy took us ca. 2 years from the beginning to the first-in-human clinical application^34,35. Although much better than 10-15 years with conventional drug discovery and development, the duration of the cycle is still too long for cancer applications, where patients may not survive (or the tumor may markedly evolve) over this period. Further shortening of the personalized drug development cycle is thus in need. For other PDD approaches, such as CAR-T generation, time is less of a concern than the costs that can be intolerable for countries without appropriate health insurance or with low-income patients. Similarly, massive further technical developments are in need to advance PDD for smooth and efficient applications. Finally, these novel developments must be accompanied by modifications to regulatory practices that are, as of today, not fully adapted to personalized treatments⁴. Continuing implementation of the N-of-1 trials and elaboration/standardization of the funding scheme of such trials is an important step towards regulatory and financial simplification of personalized drug discovery^64–66.

Altogether, the obstacles listed above are likely to be manageable, which should pave the way for the advancement of PDD in the near future, to rare and common diseases such as cancer. A massive paradigm shift from the current model of drug discovery and development, that many consider to be approaching its natural expansion limits, is expected to emerge as a result of PDD becoming more of a regular practice. I would like to conclude that advancement of PDD should witness a “democratization” of the drug discovery and development, shifting it from the highly expensive and lengthy process to highly individualized, affordable, and speedy models, offering novel solutions to the currently poorly treatable diseases, such as rare neurological disorders and cancer.

Data availability

No datasets were generated or analyzed during the current study.

Declarations

Competing interests

The authors declare no competing interests.