Liberty on Odaiba, Against the Silhouette of Tokyo Bay: Inspirations for Moving Hypothesis‐Generating Research to Patient‐Centered Therapeutics

There is an art to science, and a science in art; the two are not enemies, but different aspects of the whole. Tokyo is one of my favorite cities in the world. Whenever I can travel there, I steal a little time to ride the automated and driverless Yurikamome from the city center out to Odaiba. On the shores of Odaiba, a striking Statue of Liberty rises against the expanse of Tokyo Bay. Though only one-seventh the size of its New York counterpart, its majesty is amplified by the broader canvas in which it is positioned—the Rainbow Bridge arching gracefully across the water, the Tokyo Tower piercing the skyline, and the bright blue skies by day or shimmering city lights by night that frame its silhouette. While the statue alone is a beautiful work of art, its placement within this panorama transforms it into a scene of resonance, meaning, and awe for the observer (Figure 1). Hypothesis-generating research in clinical pharmacology is much the same. The discipline transcends biology, mathematics, and medicine. Learnings from observational research, associations in safety surveillance databases, simulations of optimal dosing regimens from quantitative pharmacological models, and predictive biomarker signature discoveries from translational research are important and useful, but often modest in isolation—like a statue without its backdrop. Yet when actively positioned within the broader context of pharmacological mechanisms, patient heterogeneity, drug development and regulatory strategy, clinical practice applications, and pathways toward implementation for transforming public health, these exploratory findings acquire depth and value. They invite the scientific community to see beyond immediate observations, reimagine implications for patients, and chart the roadmaps that could transform a hypothesis into evidence that is transformative for practice. It is such an active contextualization of research findings with a big-picture clinical perspective and explicit articulation of roadmaps toward patient-focused implementation that is welcomed in hypothesis-generating research reports in CPT. CPT publishes work across the spectrum of science underlying the Discovery, Development, Regulation, and Utilization (DDRU) of therapeutics. Whereas a research contribution may be primarily rooted in one of these dimensions, a mindful connection to the other dimensions is expected when framing the problem statement and discussing the research findings. For reports aligned with the Discovery dimension, a motivating clinically relevant question and a discussion of the research findings with respect to implications for therapeutics, including future-oriented roadmaps for solving the patient-focused problem at hand, are expected. In this issue, motivated by the clinical problem of the high incidence and severity of mucositis in pediatric patients receiving high-dose methotrexate for acute lymphoblastic leukemia and inadequate understanding of clinical or biological predictors of risk, Zhang and colleagues report a hypothesis-generating reverse translational study.1 Explainable machine learning modeling of pharmacogenomic data from 278 pediatric patients revealed significant enrichment of mucositis-associated genetic variants in IL-6 and WNT/β-catenin signaling pathways, suggesting that polygenic, pathway-level features substantially improve risk prediction compared to clinical variables alone, with associated hypotheses regarding potential therapeutic targets for toxicity mitigation. Whether this pharmacogenomic discovery will impact successful pre-treatment risk stratification, personalized supportive care, or therapeutic targeting for risk mitigation will require substantial efforts to move the generated hypothesis through the second D, R, and U dimensions of the DDRU framework. While precision dosing has been enthusiastically discussed as an untapped opportunity for many years, progress toward evidentiary frameworks and scalable practice-changing innovations has been rather modest. In a randomized clinical trial performed in the clinical practice setting by Kuypers and colleagues, the authors prospectively compared the predictive performance of a fully automated model-informed precision dosing (MIPD) application for tacrolimus dosing in de novo renal allograft recipients, integrated into an electronic patient file, versus physician-based dosing.2 The study found that a fully automated, electronic patient file-embedded MIPD tool for tacrolimus significantly outperformed physician-based dosing during the first 14 days post-transplant. MIPD achieved higher rates of target trough concentrations by Day 8, greater overall time in the target range, and smaller deviations from the target window, with less than 1% of dose suggestions overridden by physicians. These results provide valuable proof-of-principle that automated MIPD in the clinical practice setting can improve dosing accuracy and integration into clinical workflows, warranting evaluation in larger trials for clinical outcome impact. While more work remains ahead of clinical implementation, as duly acknowledged by the authors, research reports of this nature are valuable catalysts for change in the application of computational pharmacology at the patient’s bedside to advance safe and effective use of therapeutics, thereby actively bridging over to the U dimension of DDRU. The discipline of pharmacometrics has made terrific progress over the past two decades in the modeling and simulation of longitudinal measurements of tumor burden to advance oncology drug development. However, application of such models for patient care remains an untapped opportunity. Centanni and colleagues developed longitudinal models of circulating tumor cells and tumor-derived extracellular vesicles and investigated these circulating biomarkers as potential predictors of overall survival in patients with colorectal cancer.3 Using the developed models, the authors performed in silico evaluation of hypotheses for patient monitoring using these circulating biomarkers in lieu of standard imaging assessments and treatment sequencing strategies guided by biomarker-informed assessment of risk level for disease progression. While this manuscript, at its core, is an advanced pharmacometric modeling and simulation exercise, what distinguishes it for the readership of CPT is the application of the developed models to pursue innovative solutions for clinically relevant questions. A noteworthy innovative element is the authors’ pursuit to quantify the potential impact of their proposed biomarker-based monitoring strategy on environmental sustainability, illustrating how the science of clinical pharmacology has the potential to reduce the carbon footprint of our healthcare system. As acknowledged by the authors, while the findings of this research indicate the potential for integrating liquid tumor biomarkers into individualized treatment and monitoring strategies, validation in clinical trials is needed to bridge over to the U dimension of DDRU for the envisioned application of these biomarkers in clinical practice. As an integrative bridging discipline, clinical pharmacology is the cornerstone of advancing safe and effective dosing and utilization of therapeutics in all patients, including, importantly, understudied populations with limited data. Two research articles in this issue of CPT highlight this opportunity—one in the context of optimal dosing of monoclonal antibodies in pregnancy and another in the setting of optimizing dosing in an infectious disease with complex pathophysiology in LMIC geography. Duvnjak and colleagues tackle the problem of dosage optimization of monoclonal antibodies in pregnancy.4 The authors developed a physiologically contextualized population pharmacokinetic model of the anti-α4β7 integrin monoclonal antibody vedolizumab in pregnant patients with inflammatory bowel diseases. Recognizing the multifactorial determinants of changes in pharmacokinetics in pregnancy, the authors evaluated the use of serum albumin dynamics as a potential hemodilution biomarker and physiologically motivated time-varying covariate alongside gestational age in their population model of vedolizumab pharmacokinetic data in 39 pregnant patients. Exposure to vedolizumab decreased throughout pregnancy, with an estimated decrease in trough serum concentrations of approximately 50% by the end of the third trimester—a magnitude of potential clinical concern. Motivated by the therapeutic dilemma discovered by their model-based analyses, the authors leveraged the developed model for hypothesis-generating simulations of optimal dosing regimens, with a proposal that progressive shortening of the inter-dose interval during pregnancy is expected to be necessary to maintain individual efficacious pre-pregnancy trough concentrations. To facilitate potential future evaluation, the authors offer a nomogram-like graphic and discuss opportunities for further research to integrate the tool into an MIPD implementation framework. While the core of this research is population pharmacokinetic modeling and simulation, several aspects are noteworthy, making this stand out as potentially paradigm-changing research in our discipline. First, the patient-focused motivation for this research is clearly articulated upfront. Second, there are notable innovative elements advancing the science of clinical pharmacokinetic modeling in pregnancy with a good balance of physiological contextualization and pragmatism. Third, the authors act upon the observations to offer potential solutions for therapy optimization, balanced importantly with careful acknowledgment of limitations and residual uncertainties, and responsible forward-looking commentary on what it would take to ultimately move the needle at the U end of DDRU. Finally, the presented research is on a specific monoclonal antibody but can have valuable implications for the therapeutic modality at large, as discussed by the authors. Personally, I see this as an inspiration and invitation for our research community to explore the development of platform models on meta-data through pre-competitive interdisciplinary collaborations. Reinforced by the progress of the ICH E21 guideline development efforts, this research further opens doors to reimagine implications for the second D and the R dimensions of DDRU. Chu and colleagues present population pharmacokinetic-pharmacodynamic analyses of a Phase II randomized clinical trial evaluating liposomal amphotericin B as monotherapy or in combination with oral miltefosine in patients with post-kala-azar dermal leishmaniasis in India and Bangladesh.5 Of note, the formulated mathematical models to describe the clinical trial data are deeply rooted in a careful understanding of the unique features of the pathophysiology of the disease and considerations of the drug’s mechanism of action, laying the foundation for credible hypothesis generation. Recognizing that the mononuclear phagocyte system (MPS) is crucial in the distribution of liposomal drugs as well as the pathophysiology of leishmaniasis, the authors characterized the interaction between liposomal amphotericin B pharmacokinetics, the MPS, and parasite dynamics. As the model incorporated saturable uptake of liposomal amphotericin B by the MPS, a key insight is the disconnect between systemic drug concentration-time profiles and those at the site of action. This is notably evident following multiple dosing and associated parasite clearance, which reduces the capacity of the MPS compartment. The developed model informed simulations of alternative lower daily dose regimens of amphotericin B projected to achieve similar parasite clearance with lower cumulative plasma exposures. While the hypotheses generated from this research need confirmation in clinical trials to enable evidence-based implementation, this work is a fine example of hypothesis-generating research for driving progress toward safer and more cost-effective LMIC public health solutions. We are in an era of endless possibilities in reverse translational research for hypothesis generation in biomedical sciences, enabled by advances in measurement technologies, access to public health databases, and advanced data analytics. Associations observed in uncontrolled databases, as in the setting of pharmacovigilance reporting systems, can particularly be misleading, representing a common concern with disproportionality analyses of safety signals, as cautioned by Khouri and colleagues.6 As a clinical pharmacologist trained in both the biological and quantitative anchors of the discipline, I firmly believe that it is our responsibility to integrate the foundational principles of our science in reverse translational research design and the associated data analytical pipelines. This is crucial to embed causality, increase fidelity, and broaden generalizability of resulting insights. In this context, two publications in this issue of CPT provide a springboard for reflection. The first example is the work of O’Hanlon and colleagues in the context of the identification of predictors of severe neutropenia during cancer chemotherapy with docetaxel.7 Leveraging the semi-mechanistic Friberg model of myelosuppression to predict the time course of neutropenia, the authors develop a model-informed approach to assigning labels for severe neutropenia for machine learning analyses on real-world data in electronic health records. The predictive performance of the model-informed approach was superior to the observational naïve labeling approach. The developed methodology addresses the common issue of sparsity and lack of completeness of longitudinal observational data in electronic health records that limit labeling methods that rely solely on sparse observed data, where the true nadir neutrophil count may not have been observed and recorded. While primarily methodological in focus, this research is representative of innovation that has potential for application in other contexts of hypothesis generation in precision medicine. It is a reminder that data, whether collected in the laboratory, in clinical trials or in healthcare systems, must be placed in an appropriate pharmacological context for scientifically credible interpretation. It is also a reminder of the value of intersectional innovation across disciplines—in this case, between pharmacometricians, real-world evidence scientists, and data scientists. It is a call to educate and collaborate without boundaries to responsibly bring data to life—an invitation to ASCPT members to forge collaborations and build bridges across the three Networks and constituent communities of our Society. The second example is the work of Cortés-Ríos and Jusko that illustrates the power of a mechanistic systems mindset when gleaning insights for hypothesis generation from published data.8 Through mechanism-based pharmacokinetic-pharmacodynamic modeling of published data on aldosterone dynamics and response to pharmacological modulation of the system, the authors generate deep insights into the dynamics and regulation of the renin–angiotensin–aldosterone system (RAAS), providing a quantitative platform to inform trial designs for aldosterone synthase inhibitors and an invitation to continue enhancements of the presented model. This work illustrates how the first principles of our discipline continue to advance our learning of fundamental human physiology—in this case, even over 125 years following discovery of the renin–angiotensin–aldosterone system. Looking ahead, advanced analytical frameworks like Knowledge Graphs and Agentic Artificial Intelligence platforms are rapidly enabling efficient mining of multimodal data, including but not limited to data in research publications, for knowledge management and generation of insights. When embracing these opportunities, mindful integration of the foundational principles of pharmacology is necessary in designing and pressure-testing such computational workflows. The clarity and awe I have experienced on Odaiba’s shore during my visits to the beautiful country of Japan, where perspective transforms sight into possibilities, metaphorically resonates with hypothesis-generating research in clinical pharmacology—the opportunity to deliberately but responsibly carry ideas across the bridge from discovery to utilization of patient-centered therapeutics. Vital to this in the context of publishing original research in CPT is the importance of acknowledging assumptions, limitations, and areas for future research in a rich discussion of research findings. A manuscript may not traverse every step of the virtuous circle of Ideation ➔ Hypothesis Generation ➔ Verification ➔ Confirmation ➔ Implementation ➔ Ideation. However, it is important to discuss what it would take to progress along this circle, offering roadmaps for next steps and interdisciplinary collaboration. Realizing that a picture is worth a thousand words, CPT offers authors the opportunity to submit graphical abstracts as part of their manuscripts, providing a valuable vehicle for enhancing the impact of their research. I encourage our authors to take this opportunity and unleash the artists within them as they disseminate their scientific research. The year 2026 marks my tenth year as a member of the CPT Editorial Team. I am grateful to Drs. Scott Waldman and Andre Terzic for trusting me and inviting me to join the team as an Associate Editor in 2016. Drs. Piet van der Graaf and Kathleen Giacomini continued their unwavering mentorship and support over the past 6 years, for which I am immensely grateful. I will always cherish warm memories of their exceptional leadership. Stepping into my role as Deputy Editor-in-Chief of CPT in January 2026, I am filled with renewed energy and optimism for the discipline of clinical pharmacology and its impact on transforming therapeutics and enhancing public health on a global scale. I consider myself very fortunate to have the privilege to serve CPT under the visionary leadership of our Editor-in-Chief, Dr. Karen Rowland Yeo.9 I have known Karen for over 20 years. She has been an inspiring role model for rigorous science, a tireless catalyst for patient-focused implementation, an authentic mentor, and a courageous leader. We share complementary backgrounds in quantitative translational sciences with a common purpose that is deeply rooted in a passion to translate the science of clinical pharmacology to advances in healthcare for all patients. I look forward to continuing my journey with CPT, guided by Karen’s bold vision for the Journal, learning from our amazing multidisciplinary, interprofessional Editorial Team of Associate Editors and Editors-in-Training. I am grateful to Alethea Gerding, Alaina Webster, and Aimee Greene for everything they do for CPT—always with a smile, curiosity, and a can-do mindset. The author dedicates this editorial, with gratitude, in loving memory of his mother, Shantha Venkatakrishnan—an inspiring educator, valued mentor, literary scholar, and soulful poet—whose trust, guidance, and unwavering support were instrumental in enabling his pursuit of higher education in biomedical sciences and nurturing his passion for scientific writing, often embracing science as an art. No funding was received for this work. The author declared no competing interests for this work.