Review
A Novel Strategy for the Discovery of Drug Targets: Integrating Clinical Evidence with Molecular Studies
2024 Volume 47 Issue 2 Pages 345-349
Details
2024 Volume 47 Issue 2 Pages 345-349
The mechanisms of several drugs remain unclear, limiting our understanding of how they exert their effects. Receptor affinities have not been comprehensively measured during drug development, and the safety investigations in humans are limited. Therefore, numerous unknown adverse and beneficial effects of drugs in humans persist. In this review, I highlight our achievements in identifying the unexpected beneficial effects of drugs through the analysis of real-world clinical data, which can contribute to drug repositioning and target finding. (1) Through the analysis of real-world data, we found that the anti-arrhythmic amiodarone induced interstitial lung disease, leading to fibrosis. Surprisingly, concurrent use of an anti-thrombin drug, dabigatran mitigated these adverse events. Pharmacological studies using animal models have mimicked this phenomenon and revealed the molecular mechanisms associated with the platelet-derived growth factor-alpha receptors. (2) The antidiabetic dipeptidyl-peptidase 4 inhibitors increased the risk of an autoimmune disease, bullous pemphigoid, which was reduced by the concomitant use of lisinopril. Pharmacological studies using human peripheral blood mononuclear cells have revealed that lisinopril suppressed the skin disorders by inhibiting the expression of cutaneous matrix metalloproteinase 9 in macrophages. (3) The antimicrobial fluoroquinolones increased the risk of tendinopathy, which was reduced by the concomitant use of dexamethasone. However, clinical guidelines have suggested that corticosteroid increases the risk of tendinopathy. Our investigation demonstrated that fluoroquinolones impaired tendon cells through DNA damage by generating reactive oxygen species. In contrast, dexamethasone exhibited an acute beneficial effect on tendon tissue by upregulating the expression of a radical scavenger, glutathione peroxidase 3.
The progress in new drug development has advanced rapidly since the deciphering of the human genome, the blueprint of the human body. This advancement has led to the development of therapeutic drugs for intractable diseases, including cancer and immunological diseases. Nevertheless, we are faced with the challenge of rapidly depleting “targets” for creating novel drugs. Additionally, we confront a significant issue known as the “death valley,” wherein clinical trials struggle to demonstrate efficacy even if the drug is effective in animal models. In an attempt to overcome these challenges in drug discovery, we have recently proposed the concept of “clinical evidence-based drug discovery,” which utilizes patient medical records.
Traditionally, pharmacology has employed adverse drug reactions to make animal models of pathological human conditions. This is because the phenotypes and pathogenic mechanisms observed in these models are similar to those observed in spontaneous human diseases. Recently, a substantial volume of real-world data (RWD), including spontaneous reports of adverse drug reactions and insurance claims, has become available. The statistical analysis of this human data enables the examination of the incidence of adverse events that actually occurred in patients. This facilitates the precise identification of confounding factors (such as concomitant medications) that influence the occurrence of adverse reactions. Identification of these pharmacological drug–drug interactions not only prompt immediate drug repositioning but also contributes to elucidating the mechanisms of adverse drug reactions and discovering new drug targets. Moreover, hypotheses derived from human big data analyses are anticipated to have exceptionally high clinical predictive value.
Here, I present our previously published research reports that demonstrate the linkage between RWD analysis and the discovery of drug targets. Additionally, I present new research strategies and prospects for the future utilization of RWD, encompassing electronic medical records. An outline of the methodologies has been documented and discussed previously.1)
Interstitial Lung Disease (ILD) represents a heterogeneous group of parenchymal lung disorders characterized by varying degrees of inflammation and fibrosis, resulting in the loss of lung function. The incomplete understanding of its etiology and the absence of effective treatment make ILD a condition with a considerable unmet need, necessitating novel approaches to treatment. Because ILD can manifest as an adverse drug effect, we focused on exploring effective treatments for drug-induced ILD.
First, we investigated the association between the use of a particular drug and the incidence of ILD in the U.S. Food and Drug Administration (FDA) Adverse Event Reporting System (FAERS) using disproportionality analysis by calculating the reporting odds ratio (ROR) for each report. Among the numerous drugs that demonstrated a strong association between their use and the emergence of ILD with high RORs, we selected amiodarone due to a substantial number of reported cases for investigating potential confounders for drug-induced ILD. When evaluating the confounding effects of all drug combinations in a population of amiodarone users, several anticoagulants were found to characteristically decrease the risk of ILD. Among them, dabigatran was chosen for further analysis because of its low ROR value and a sufficient reported number of cases. Dabigatran alone did not affect the reported proportion of ILD; however, in combination with amiodarone, it markedly reduced the amiodarone-induced increase in the ROR of ILD.
To investigate the consequences of amiodarone use and the onset of ILD in clinical records, we analyzed the JMDC insurance claims data. In an amiodarone-treated cohort (n = 1880), a causal association between the use of amiodarone and the onset of ILD was detected with an adjusted sequence ratio of 7.32 in a sequence symmetry analysis. We then divided the amiodarone cohort into two groups: those that received dabigatran (n = 206) and those that did not (n = 1674). These two populations were compared using Kaplan–Meier analysis and Cox proportional hazards modeling. The results from these analyses indicated that the combination of dabigatran significantly decreased the amiodarone-induced incidence of ILD, with a hazard ratio of 0.43 (95% confidential limit [CI]: 0.24–0.77). Concurrently, the decrease in the number of remaining at-risk patients was also slowed in the amiodarone-dabigatran cohort, suggesting that dropouts from long-term amiodarone treatment were avoided by the use of dabigatran. Consequently, the 3-year ILD incidence decreased from 12.4 to 5.8% with dabigatran treatment. These results, derived from real-world patient data, demonstrate that the concomitant use of dabigatran during the treatment of arrhythmia with amiodarone, as an unintended consequence, prevented the onset of drug-induced ILD that leads to pulmonary fibrosis. Notably, a possible role of thrombin in lung fibrosis had been suggested previously.2)
To confirm whether dabigatran mitigates amiodarone-induced ILD in animals, we subjected mice to chronic amiodarone treatment. Subsequently, we monitored the body weight at which drug-induced respiratory insufficiency could be noninvasively detected. Scheduled administration of amiodarone for four weeks resulted in a decreased body weight of mice after several days, which led to gradual death. Dabigatran treatment alone did not affect the natural increase in the body weight of mice; however, when co-administered with amiodarone, it significantly inhibited the amiodarone-induced decrease in body weight, resulting in a decreased death rate. Histologically, dabigatran cotreatment was found to significantly decrease the number of infiltrating macrophages.
Gene expression changes in the lung tissue indicated that our amiodarone-mice model exhibited a chronic inflammatory stage before fibrotic alterations in the lung structure, and that dabigatran specifically prevented proinflammatory changes. Based on the expression data, we sought promising targets for new ILD drugs and identified a substantial increase in the expression of platelet-derived growth factor receptor α (PDGFRα) in the mice lung tissue after amiodarone treatment. Importantly, this upregulation was significantly inhibited by co-treatment with dabigatran. In conclusion, the amiodarone-induced ILD was inhibited by a direct thrombin inhibitor. Investigation of downstream signaling indicated the involvement of fibroblast PDGFRα in the beneficial effect (Fig. 1). Direct inhibition of PDGFRα may serve as a promising strategy for the treatment of ILD in humans.3)
Bullous Pemphigoid (BP) is the most common subepidermal autoimmune blistering disease, characterized by the production of autoantibodies directed against two hemidesmosomal proteins: BP antigens 180 and 230. BP is associated with several disorders, including autoimmune, neurological, and cardiovascular diseases, as well as neoplasms. Moreover, the identification of over 50 drugs reported to elicit drug-induced BP4) prompted us to investigate causal drugs and effective treatments using human clinical data.
In the FAERS analysis of drugs that cause BP, we observed a strong association between all dipeptidyl peptidase 4 (DPP4) inhibitors (vildagliptin, linagliptin, sitagliptin, alogliptin, teneligliptin, anagliptin, and saxagliptin) and the emergence of BP. Additionally, we found that lisinopril decreased the ROR of BP in patients treated with DPP4 inhibitors without having a self-effect on the occurrence of BP.
The causal relationship between the use of DPP4 inhibitors and lisinopril was further analyzed using the American insurance claims data from IBM MarketScan. When evaluating the overall association between the use of DPP4 inhibitors and the onset of BP by estimating the incidence rate ratio (IRR), the DPP4 inhibitor group showed high IRR values for BP, irrespective of their blood glucose-reducing effect. In the DPP4 inhibitor cohort that received lisinopril (n = 34480), a significant reduction in BP incidence was observed with a hazard ratio of 0.53 (95% CI: 0.35–0.81) compared with the propensity score-matched group that did not received lisinopril. Furthermore, the daily and cumulative doses and the DPP4 inhibitor administration period were comparable in each pair in these score-matched cohorts, with or without lisinopril use.
Because BP is an autoimmune skin disease, reproducing the disease state in experimental animals is challenging. Therefore, we initially investigated the mechanism of T-cell differentiation by analyzing the proportions of Th1, Th17, and Treg cells in the presence of DPP4 inhibitors using human peripheral blood mononuclear cells in vitro. However, no distinct changes in these cell populations were observed after treatment with DPP4 inhibitors. Several studies have suggested that the production of matrix metallopeptidase 9 (MMP9) by immune cells is associated with skin blister formation during BP. Therefore, we hypothesized that the molecular mechanisms underlying the bidirectional effects of DPP4 inhibitors and lisinopril on BP are closely related to the regulation of MMP9 function.
Monocytes were obtained from healthy human volunteers and stimulated for seven days with macrophage colony-stimulating factor. In these monocytes, vildagliptin-treated cells showed higher MMP9 expression than their untreated counterparts, in a concentration-dependent manner. Conversely, the increase in MMP9 expression induced by vildagliptin was suppressed by concomitant treatment with lisinopril. Additionally, we observed an increase in the expression of angiotensin-converting enzyme 2 (ACE2) with vildagliptin, which was then suppressed by lisinopril treatment. ACE2 is involved in producing the angiotensin (1–7) peptide that acts on the Mas receptor (MasR), affecting M1/M2 macrophage polarization. Consequently, we evaluated whether DPP4 inhibitors affected the angiotensin (1–7)/MasR axis and whether the MasR inhibitor A779 could inhibit the vildagliptin-induced upregulation of MMP9.
In conclusion, our results, derived from clinical big data mining followed by in vitro experimental validation, demonstrated the effectiveness of lisinopril in reducing the risk of BP associated with the use of DPP4 inhibitors. This suggests that MMP9 and ACE2 expression play important roles in the underlying mechanisms. Furthermore, our findings indicated that the angiotensin (1–7)/MasR axis is involved in this mechanism (Fig. 2). Our findings provide new insights into the pathophysiology of BP and highlight novel drug targets for treating BP.5)
Tendinopathy is characterized by pain, loss of tendon strength, or rupture. Previous studies have reported multiple risk factors, including aging and fluoroquinolone use; however, the therapeutic targets remain unclear. We analyzed self-reported adverse events in FAERS and found that the concomitant use of dexamethasone significantly prevented fluoroquinolone-induced tendinopathy. However, the FDA has warned that the use of corticosteroids increases the risk of fluoroquinolone-induced tendinopathy.6) Given the contradiction between our findings and the FDA’s warning, we decided to further investigate this conflicting result regarding the effect of steroids, with a specific focus on dexamethasone.
Using MarketScan data, Kaplan–Meier analysis and Cox proportional hazards modeling were performed in propensity score-matched cohorts. The results of these analyses indicated that when dexamethasone was prescribed within seven days before the prescription of fluoroquinolone, the concomitant use of dexamethasone mitigated the risk of fluoroquinolone-induced tendinopathy, with a hazard ratio of 0.58 (95% CI: 0.34–0.98). Additionally, when individuals aged ≥65 years were followed up for 180 d from the end of dexamethasone use, dexamethasone users exhibited a lower incidence of spontaneous tendinopathy with a hazard ratio of 0.72 (95% CI: 0.53–0.98). This suggests that dexamethasone prevented the onset of tendinopathy induced by fluoroquinolone and aging.
Rat tendons systemically treated with fluoroquinolone exhibited mechanical fragility, histological changes, and DNA damage. However, co-treatment with dexamethasone attenuated these effects and increased the expression of the antioxidant enzyme, glutathione peroxidase 3 (GPX3), as revealed by RNA sequencing. The primary role of GPX3 was validated in cultured rat tenocytes treated with fluoroquinolone or H2O2. This treatment of tenocytes accelerated senescence, particularly in combination with dexamethasone or viral GPX3 overexpression. These results suggest that dexamethasone prevents tendinopathy by suppressing oxidative stress via the upregulation of GPX3 (Fig. 3). This steroid-free approach to the upregulation or activation of GPX3 may serve as a novel therapeutic strategy for tendinopathy.7)
Several limitations exist in the current strategy. Misclassification of the outcome is a potential issue in data mining of the FAERS and insurance claims databases because the International Classification of Disease 10 (ICD10)-based codes are insufficient for differentiating particular diseases from others with similar symptoms. Future work on the validation of RWD-based algorithms for specific diseases/symptoms will be essential to accurately define the outcome. The low incidence of adverse events posed challenges in conducting cohort studies. For example, in Study 3, we could not analyze the concomitant effects of other corticosteroids in older individuals because of the limited number of total events in the cohorts. Future analyses of multiple databases may help overcome these issues. Another limitation of this study is that drug causing events and those providing benefits may affect multiple cellular functions through their respective targets. Consequently, other pathways, organs, and cells may also affect disease onset. Further in vivo studies to identify the underlying molecular mechanisms could provide valuable insights into addressing this challenge.
The author thanks Drs. Hisashi Shirakawa, Kazuki Nagayasu, Soni Siswanto, Keisuke Nozawa, Haruka Furuta, and Hiroki Yamamoto for their contribution to these analyses and experiments. These studies were supported by JSPS KAKENHI (18KK0216 and 20H00491) and AMED (21ak0101153).
This research is partly supported by Grants from JT, Boehringer-Ingelheim, Ono Pharmaceutical, and Astellas Pharma.
This review of the author’s work was written by the author upon receiving the 2023 Pharmaceutical Society of Japan Award.
