Sustainability in Pharmaceutical Research and Development: A Perspective on Opportunities for Clinical Pharmacology and Translational Sciences

Abstract

With society's rising concern over resource expenditures, sustainable solutions are sought across industries. The pharmaceutical research & development (R&D) enterprise faces the dual challenge of navigating the complexity of human disease and a highly regulated environment, leading to extensive data generation to ensure patient‐focused development of optimal therapeutics. Through innovative study designs, translational technologies, and model‐informed approaches, clinical pharmacology as a discipline can meaningfully contribute to reducing drug development's carbon footprint and achieving sustainability goals.

The convergence of high material consumption, workflow inefficiencies, and issues of global access lag/drug loss have prompted efforts toward more streamlined, globally representative and patient‐centric drug development. As practices of conservation and efficient use of resources become organizational priorities, an onus is placed on all pharmaceutical scientists to develop and execute concordant strategies. Here, we offer perspectives on championing sustainability within pharmaceutical R&D across translational, clinical, and operational spaces.

TRANSLATIONAL SPACE

Successful translation from animal studies to the clinic remains challenging across many areas of pharmacology and toxicology. This limited mapping to human disease and therapeutic success juxtaposed against ethical concerns on animal use in biomedical research has motivated increased efforts to refine, reduce, and replace animal testing with more predictive constructs. Such New Approach Methodologies (NAMs), including human cell‐based assays, computational models, and other innovative platforms, are expected to gain momentum, reinforced by FDA's NAM roadmap. ¹ Despite enthusiasm for NAMs, substantial technical and evidentiary work is pending, with rigorous qualification being important for more broadly achieving a seamless transition from conventional strategies. Impactful adoption requires clear Contexts of Use, rigorous technical validation, open data infrastructures, and proactive cross‐sector partnership realistically aimed at quantifying and managing (rather than avoiding) uncertainty, to facilitate regulatory acceptance. Through a patient‐focused lens, the clinical pharmacology community offers a unique combination of translational insight and quantitative solutions to advance NAM utility.

Of the available in vitro NAMs, microphysiological systems like organs‐on‐chips constitute perhaps the most advanced representations of human in vivo conditions. Employing multiple human cell types while accounting for physiologic flow rates, physical barriers, and mechanical forces, these systems provide a rich data source for mechanistic models of human biology and pharmacokinetic (PK)/pharmacodynamic (PD) relationships. As an in silico complement to in vitro data generation, physiologically‐based PK (PBPK) and quantitative systems pharmacology (QSP) models can leverage NAM outputs to refine predictions of clinical drug exposure and response, enabling high‐throughput hypothesis testing. For instance, Maass et al. ² combined kidney microphysiological system outputs with a QSP model, predicting the clinical impact of three known nephrotoxic agents on kidney injury molecule‐1, an established renal tubule‐specific biomarker, and propose dosing schedules that optimize benefit‐risk. One can envision similar translational in vitro–in vivo workflows being applied to other scenarios of clinical interest; as the community's involvement in NAMs widens, more use cases demonstrating value may be anticipated.

The movement toward reduced animal testing and increased emphasis on model‐based approaches represents an opportunity to modernize the philosophy of translational science within drug development, particularly in contexts like cell and gene therapies and immuno‐oncology, where translational fidelity demands commitment to multi‐scale systems models of human biology and disease. Leveraging experimental and computational systems that mitigate translational uncertainty and maintain high reusability, R&D costs may be better appropriated to work packages with the highest yield for patients. Clinical pharmacology as a discipline is well‐positioned to advance this sustainability effort in recognizing what in vitro PK/PD data would be most informative for patient trials, strategizing how mechanistic models can support multiple assets within a therapeutic area, and justifying NAM‐derived evidence as a suitable replacement of or supplement to traditional studies.

CLINICAL SPACE

Clinical trial design, execution, and analysis is considered the most expensive component of drug development, with traditional methodologies often resulting in high resource use for uncertain outcomes. This unfavorable tradeoff calls for innovative solutions to recalibrate data collection around patient needs and data utilization leveraging the totality of evidence. Decentralized trials offer one strategy for reducing patient, administrative, and environmental burden, shifting data collection from the clinical site to the patient's home. An analysis of greenhouse gas emissions of seven clinical trials across phases identified drug product manufacture, packaging and distribution; patient travel; on‐site monitoring visit travel; collection, transport and processing of laboratory samples; and sponsor staff home‐to‐office commuting as principal sources. ³ Notably, patient travel was a consistent hotspot. This points to opportunities for validated patient‐centric sampling technologies like dried blood spots from self‐administered finger sticks for PK/PD measurements. Combined with computationally derived optimal sampling designs, the information content of the data can be maximized, reducing patient burden and inferential uncertainty. Furthermore, wearables like smart watches and other biosensors enable real‐time readouts. Of note, each clinical trial's context is different (e.g., patient population, safety considerations, and validation of measurement technologies/end points) and these technologies are not without environmental impact, necessitating careful consideration for net benefit in the sustainability equation.

The continued recognition of model‐based approaches for the efficient integration and interpretation of clinical data naturally extends to sustainability efforts. Established contexts of model‐informed evidence generation in lieu of dedicated clinical trials include proarrhythmic risk assessment via concentration‐QTc modeling (ICH E14) and PBPK model‐informed labeling for drug–drug interactions (ICH M12 and M15). Timely dose optimization is another established context for sustainability enabled by quantitative clinical pharmacology, exemplified by FDA's Project Optimus for oncology drugs, avoiding high doses when not scientifically justified, decreasing manufacturing burden and safety risks. Also noteworthy is the opportunity to pool geographical regions for evidence generation in multi‐regional clinical trials (ICH E17) based on similar drug‐ and disease‐related intrinsic and extrinsic factors inferred from population pharmacology and drug‐disease models, eliminating redundant regional “bridging” trials and decreasing access lag for medicines worldwide. Through comprehensive evidence synthesis, predictions of drug disposition and disease response enable optimization of benefit–risk, contributing to reduced clinical costs through streamlined clinical development plans and shortened cycle times. ⁴ Such savings of time and cost provide tangible demonstrations of sustainability through model‐informed drug development.

The power of pharmacometric modeling to drive sustainability was recently demonstrated by Centanni et al. ⁵ As an alternative to routine imaging, longitudinal circulating tumor cell and tumor‐derived extracellular vesicle counts were shown to be predictive of overall survival in metastatic colorectal cancer. Model‐based simulations, considering the procedures and patient travels applicable for each scenario, generated hypotheses for reducing treatment monitoring costs by twofold and carbon emissions by over 10‐fold. Given the immense carbon footprint of mid‐ to late‐stage clinical trials, ³ , ⁶ prospective assessments of the range of potential outcomes are needed to improve upon what has historically been an unacceptably high failure rate. To this end, Wojciechowski et al. ⁷ provide a blueprint for evaluating probability of success through clinical trial simulation. Deploying longitudinal exposure‐response models of endoscopic improvement and modified clinical remission from phase IIb data in ulcerative colitis, the authors interrogated various dosing regimens of an investigational agent for responder rates in virtual phase III trials vs. a key competitor. When integrated with similar evaluations for safety, this quantitative appraisal of asset viability increases confidence in Go/No‐Go decisions, decreasing the likelihood of financially and environmentally costly pivotal trial failures. We posit that these model‐based strategies should reduce clinical drug development's long‐term carbon footprint, offering a more sustainable future for pharmaceutical innovation.

Notably, evolving regulatory frameworks support sustainable evidence generation in drug development. Examples include FDA's new default option of one pivotal trial combined with confirmatory evidence for approval of novel products, ⁸ and the Agency's plausible mechanism pathway outlining a path to regulatory approval when randomized trials are infeasible. ⁹ Both initiatives provide calls to action for the clinical pharmacology community to bolster causal evidence synthesis through rigorous translational science and model‐based integration.

OPERATIONAL SPACE

Across cycles of data collection and analysis, operational processes provide the interstitium necessary for the timely delivery of clinical pharmacology insights. As the discipline expands its quantitative mindset, increased adoption of script‐based analysis and reporting offers a more sustainable way of working through the efficient generation of reproducible deliverables. The utilization of open‐source computational software such as R (https://www.r‐project.org/) or Julia (https://julialang.org/) fosters both data science literacy and crowdsourced solutions, enabling less redundant computation and avoiding rework in contemporary workflows. Open science frameworks that advocate for precompetitive data sharing and contribute domain‐specific tutorials to facilitate upskilling avoid redundancies in data generation across institutions and promote on‐demand professional development. The Open Systems Pharmacology community (https://www.open‐systems‐pharmacology.org/) is one example, providing qualified source parameters and model structures for PBPK and QSP including but not limited to in vitro–in vivo extrapolation, drug–drug interaction prediction, and construction of virtual populations. Reuse of these repositories presents additional benefits, expediting answers to common clinical pharmacology questions encountered in drug development.

The continuous expansion of biomedical literature and pace of pharmaceutical R&D creates a scenario where knowledge management and task completion can become overwhelming. In the age of artificial intelligence (AI), automated solutions are poised to reduce manual workloads through more efficient execution of repetitive tasks. While typical large language models (LLMs) may achieve some success in reducing time spent on, for example, literature searches, agentic AI frameworks hold promise for greater returns. Built upon LLMs with additional concepts of task specificity and modular orchestration, agent‐based AI deployment can correctly negotiate more complex problems by incorporating domain‐specific expertise through human‐in‐the‐loop architectures. ¹⁰ Within the sphere of clinical pharmacology, such agentic AI workflows may be applied to many recurring processes, for example: (i) synthesis of drug‐ or disease‐specific information from scientific literature and unstructured data sources like clinical notes from electronic health records and medical images; (ii) PK/PD data preparation, visualization, and description ranging from exploratory graphics to execution of formal noncompartmental analyses; or (iii) iterative pharmacometric model building and running of appropriate simulations. The resultant time savings are augmented by built‐in quality control checks, ensuring standardized processes are followed and variability in output is reduced. Once AI agents are in place, they can be recycled within and between projects, freeing time to pursue innovations that advance the most promising therapeutics for patients. Of course, these AI‐enabled efficiency gains are not without environmental costs given the energy expenditures required to power them, emphasizing the need for instilling a culture of responsible and judicious use.

CONCLUDING REMARKS

The global drive toward sustainability is necessary to protect the welfare of future generations. Meaningful opportunities exist across clinical pharmacology and translational sciences to reduce, reuse, and recycle across various facets of pharmaceutical R&D (Figure 1 ). Judicious integration of these opportunities is necessary to enable sustainable and efficient progress toward healthcare solutions, reducing the enterprise's long‐term carbon footprint. Through collaborative efforts, we should be able to find the nexus where cost, speed, and quality meet to produce a sustainable future.

Figure 1.

Major thematic elements of sustainability opportunities for clinical pharmacology and translational sciences in pharmaceutical research and development, aligned to a Reduce/Reuse/Recycle framework. While certain opportunities are categorized in one of these three themes, the lines between them can be subtle. Although some solutions aimed toward increased efficiencies will result in micro‐scale (e.g., trial‐ and/ or analysis‐level and development stage‐level) increases in costs and/ or carbon footprint, their potential macro‐scale benefits (e.g., development program‐level, portfolio‐level, and enterprise‐wide) warrant judicious consideration with a long‐term view to ensure net benefit toward sustainable progress toward healthcare solutions of the future.

FUNDING

No funding was received for this work.

CONFLICTS OF INTEREST

The authors declared no competing interests for this work. As Deputy Editor‐in‐Chief of Clinical Pharmacology & Therapeutics, Karthik Venkatakrishnan was not involved in the review and decision process for this paper.