Sirj Goswami, PhD, CEO, co-founder of InsightRX

Sirj Goswami, PhD

Optimizing drug dosing in oncology has long been a challenge. The current drug development paradigm is oriented toward identifying a maximum tolerated dose (MTD) in early-stage trials, and evaluating the MTD for drug activity and efficacy in Phase II and III trials. In recent years, the limitations of this approach have become clearer.

First, the existing oncology drug development methodology does not adequately evaluate inter-patient variability in treatment response and toxicity. As a result, the dose-exposure-response relationship is rarely well defined for oncology drugs, despite their narrow therapeutic indices. Secondly, the current approach also leaves room for clinically meaningful pharmacodynamic endpoints—such as target inhibition—to be overlooked, despite their potential importance for new classes of medicine. Neglecting the impact of biological modulation may result in poor outcomes for patients.

Most importantly, the classical drug development paradigm does not adequately characterize long-term toxicity. This oversight is problematic, as patients on new biological therapies are expected to undergo treatment for much longer periods of time. Patients may use targeted drugs for months or years, which increases the importance of evaluating long-term tolerability.

 that many approved oncology drugs are inadequately characterized with approved doses that may be too high, the FDA’s Oncology Center of Excellence has launched an initiative to reform the dose optimization and dose selection paradigm in oncology drug development. Known as 

, this initiative recommends a core change in how doses are selected during oncology clinical trials and post-approval studies.

As the design of drug trials is guided by the results of prior research phases, the FDA is recommending a concerted effort to rethink dosing beginning with first-in-human studies. At present, Phase I oncology studies are almost entirely designed to identify and achieve the maximum tolerated dose (MTD), which is unique to oncology drug development. Non-oncology trial designs have a long history of employing dose-ranging approaches that balance efficacy outcomes with toxicity profiles.

In Phase I oncology trials, the objective is to determine the highest tolerable dose, based on the assumption that higher doses will provide greater efficacy (i.e., disease-free survival). In these trials, increasing doses of a drug are sequentially evaluated in small cohorts of patients until a pre-specified rate of dose-limiting toxicity (DLT) is reached. The dose immediately below that which elicited DLT is considered the maximum-tolerated dose (MTD); in most cases, the MTD is then used for Phase II and subsequent Phase III trials, which are used to evaluate drug activity and efficacy. Currently, it is rare for Phase II or III studies of oncology drugs to evaluate more than one dose, although this approach is common in other therapeutic areas.

Trial design flexibility has lagged behind innovation in oncology research. The flourishing of gene therapy, biologics, and anti-body drug conjugates (ADCs) has reinvented how we approach oncology treatment. The promise of more targeted treatment approaches with lower toxicity is a near reality, and will in turn require a much more thoughtful and patient-specific dosing strategy. A true targeted approach requires a molecularly targeted therapy with a tailored dosing regimen that maximizes the benefit/risk ratio for patients receiving the drug.

Likely impacts of Project Optimus on drug development

Clearly, a migration away from MTD-focused study design is long overdue, and the FDA’s guidance has already begun to influence oncology drug development processes. The first impact of Project Optimus concerns imminent drug candidates for regulatory approval. The volume of post-marketing requirements and commitments will increase, placing a greater burden on pharmaceutical companies that have not conducted earlier phase dose optimization. Anecdotally, the pursuit of 

 to further prove and maintain labeling already exceeds two-thirds of all approved drugs. This practice can be expected to grow and become more comprehensive in nature.

In parallel, drugs at various stages of Phase II or III research may require additional testing. Drug candidates in Phase III may be subject to more stringent post-marketing requirements and/or commitments. Phase II drug candidates may require further dose-response analysis to validate the current dose against one or more doses. It remains to be seen how much latitude the agency will grant during this transition period. At the very least, we know the FDA’s go-forward expectation is that the industry will begin conducting randomized dose-ranging studies prior to Phase III.

Finally, Project Optimus is expected to make a substantial imprint on all new Phase I study designs, as these serve as the foundation of oncology dose strategies. The FDA’s ultimate goal is to affect a paradigm shift in Phase I study design and the subsequent interplay between Phase I and II research. All appropriate Phase I trials should determine a range of doses to be studied in Phase II. This pivotal change in strategy will have a cascading effect throughout the drug development process and onward, impacting real-world outcomes.

Potential strategies for dose optimization

For patients to fully benefit from the great strides in new oncology treatments, we must architect drug development strategies that include dose optimization but do not unnecessarily delay market entry for critical new drugs. The process of determining the optimal dose will depend on the drug, the target population, and existing clinical data for the drug in question, as well as approved drugs that exhibit a similar mechanism of action.

At the 2013 Conference on Clinical Cancer Research, a panel of experts from the pharmaceutical industry, the FDA, and the American Society of Clinical Oncology 

 on how to both optimize dose selection and provide a better understanding of the relationship between drug exposure and clinical outcomes. The authors proposed to include adequate PK sampling and pharmacodynamic endpoint measurements in Phase I-III. The authors also suggest incorporating adaptive clinical trial designs in Phase II to characterize the relationship between exposure and clinical outcomes, describe the variability in PK/PD, and optimize the dose. Finally, the authors recommend collecting patient-reported outcomes in Phase II and III and using that data, along with PK/PD endpoints, to guide dose optimization.

The authors also discussed how integrating these data elements could improve clinical outcomes. Performing dose comparison studies in Phase II could help companies identify drugs for which they should define dose-escalation or dose-reduction strategies in Phase III trials. Adaptive clinical trial protocols that include criteria for increasing or decreasing the dose under certain conditions, such as when a patient is experiencing severe side effects, may support drug labels that specify appropriate dose-modification strategies.

Currently, the label for most oncology drugs provides only a starting dose that may need to be modified for each patient based on clinical observation. The provision of a more comprehensive dosing dataset—one that includes a drug’s threshold concentration and a peak exposure associated with excess toxicity—would better enable therapeutic drug monitoring in clinical practice. As targeted therapies are often intended to be taken chronically, their tolerability or efficacy may diminish over time, making therapeutic drug monitoring an immensely valuable tool for ensuring patients in chronic treatment maintain an optimal exposure.

Getting prepared for the dose optimization transition

Many pharmaceutical and biopharma organizations are not yet prepared for the impact of Project Optimus. Some are still under the mistaken impression that this transition will incur greater development costs and deliver marginal revenue opportunities. In fact, if companies do not evolve their development strategy, the costs could be substantial. Ignoring the FDA’s guidance will lead to drug approval rejections and even more burdensome post-approval confirmatory commitments.

Beyond a philosophical shift in study design, Project Optimus will require fresh thinking with respect to integrated data, technological infrastructure, and even manufacturing and supply chain optimization. Pivoting to the future under Project Optimus will require companies to overcome three main challenges:

 While the industry is slowly accepting that higher doses are not always congruent with disease reduction, there is a deeply ingrained belief that randomized dose-finding trials are too time-consuming. This is untrue, and this mindset must be overcome. Oncology trials should be designed correctly at the onset, as a pathway to getting life-changing therapies to patients sooner.

 Dose-ranging studies require the deployment of newly available toolsets. The core application set in many pharmaceutical organizations is an amalgamation of disconnected legacy assets that cannot foster real-time, randomized dose-ranging drug trial portfolios. Advancements in machine learning, applied Bayesian statistics, and scalable industry cloud platforms can give pharmaceutical companies the technological infrastructure they need to execute dynamic, adaptive oncology studies.

 Pharmaceutical companies require a greater volume of validated biomarkers, which will allow them to assess a drug’s exposure-response rate more precisely at various dose levels that align more neatly with targeted therapies.

To accelerate speed to market and maximize their chances of a successful clinical trial, pharmaceutical and biopharma companies will need to embrace the Project Optimus paradigm of comprehensive dose characterization. They will require the ability to perform patient-specific and population-level drug response simulations in real time, within clinical trials. Companies must be able to characterize each drug’s pharmacokinetics and pharmacodynamics, and to predict dose-specific adverse events and clinical response. Implementing new dose selection and optimization strategies for each phase of a clinical trial will help pharma companies determine dosing that maximizes efficacy, safety, and tolerability—and ultimately, the usability of their products.

, PhD, is the CEO and co-founder of InsightRX, and 

, MBA, is the vice president of Global Biopharma Strategy.

Articles

Understanding Project Optimus

As the shift to value-based care drives increased interest in improving clinical outcomes through individualized care, precision medicine will take center stage, permanently impacting the way drugs are administered to patients. To understand how we got here, it’s important to examine the role of clinical pharmacology and its impact on drug approval and real-world dosing in patients.

The key pillars of clinical pharmacology, pharmacokinetics (PK), and pharmacodynamics (PD), play a pivotal role in precision medicine as it allows us to study drug response variability and determine dosing strategies at both the patient and population level. PK can be defined as the study of how the body processes a drug via the mechanisms of absorption, distribution, metabolism, and excretion. Conversely, PD can be defined as the study of the effects of the drug within the human body as it is processed.

Clinical pharmacology’s role in drug approval

Pharmaceutical companies leverage their understanding of pharmacokinetics and pharmacodynamics to assess drug efficacy, ensure patient safety, and ultimately determine the dosing strategy that goes into the drug label. As part of its regulatory oversight of the pharmaceutical industry, FDA’s Office of Clinical Pharmacology also applies clinical pharmacology principles to validate the efficacy and safety of a drug label’s dosing.

In the early years, FDA’s drug evaluation process focused almost exclusively on pharmacokinetic data. FDA used PK-matching principles—which assume a steady concentration/response relationship—to adjust a drug’s dosage for adult or pediatric populations using criteria such as body weight. However, FDA realized that PK data alone was insufficient. After all, patients who have achieved the same drug concentration levels can still experience varying effects based on their individual disease severity, genetics, or other characteristics.

In recent years, FDA started to link drug concentration to the patient response through PD data. FDA now relies on exposure-response (ER) data as the foundation for its work validating dose justification for the general population, as well as for those with special conditions.

Our understanding of exposure-response is a prerequisite to tailored dosing strategies in the drug label. To reduce patient harm, the approved dose must balance the risk and benefits for each patient. For example, the final approved label dose for one drug designed to reduce the risk of stroke, edoxaban, is individualized based on the patient’s current renal function for patients with mild to moderate renal impairment. For patients with severe renal impairment, it is considered too hazardous to administer.

Some commonly used drugs, such as warfarin, have product label dosage information that is predicated on both PK and PD data. In some instances, a drug’s sponsor requests the removal of an indication for a certain subpopulation based on its own retrospective subset analysis of clinical trial data. For example, a recent oncology drug, cetuximab, now recommends dose individualization based on genetic factors and the type of tumor mutations.

Following drug approval, clinicians may also attempt to leverage PK and PD knowledge to dose-adjust for an individual patient. Precision dosing at the bedside means that a drug’s dosage can be adjusted to patient-specific characteristics—such as body weight, age, gender, and genetics—as well as extrinsic patient-specific factors, such as drug interactions, diet, and behavior.

Surprisingly enough, the first model-informed precision dosing (MIPD) tool using both PK and PD data was developed in 1969, to ascertain the optimal dosing for patients on anticoagulation therapy. Over the past several decades, technological advances such as the ubiquity of EHRs, increased data accessibility, and the emergence of cloud-based infrastructure have enabled adoption of MIPD at scale.

Precision dosing can be implemented as a continuous learning cycle to update our understanding of PK and PD in subpopulations and improve the predictive accuracy of underlying models and algorithms. Practically speaking, clinicians may use precision dosing tools through an EHR-integrated application to determine optimal dosage regimens for their patients.

Over time, patient data can be aggregated and stored in a database and subsequently analyzed to provide new insights into a drug’s behavior in patient subpopulations that are not typically studied by drug manufacturers. By implementing precision dosing as a continuous learning cycle informed by analytics and model updates, we will be able to expand our cumulative clinical knowledge of the drug and democratize such knowledge for providers.

FDA’s guidance on overcoming precision dosing challenges

In 2019, FDA invited hospital representatives, software developers, and practicing precision-dosing physicians to a working session on how to best promote the widespread adoption of precision dosing. They discussed a variety of challenges to implementing an evidence-based dosing regimen in hospital or clinic setting, from current drug labeling practices to incomplete clinical trial populations, insufficient predictive models for major drug clearance mechanisms, and an overall lack of real-world data.

In two subsequent papers that resulted from the session, researchers and FDA scientists 

First, drug labels should be more dynamic, and should use broader language to encourage the use of exposure-response data within the studied range of safe, effective doses in the clinic. By enrolling more local patients and including adequate measures to characterize the patient, disease, and drug, global clinical trials can differentiate responders from non-responders and connect the trial results to local patient population characteristics. This will help improve the approval process and promote precise labeling of drugs.

Secondly, to combat the lack of data to inform precision dosing approaches, pharmaceutical companies should be encouraged to use multiple drug doses or titration-based dosing during clinical trials, as well as to enroll understudied and more diverse populations. FDA’s recent initiative in oncology with 

 is a reflection of this broader trend. The goal of Project Optimus is to educate, innovate, and collaborate with companies, academia, professional societies, international regulatory authorities, and patients to move forward with a dose-finding and dose optimization paradigm across oncology that emphasizes selection of a dose or doses that maximizes not only the efficacy of a drug, but also its safety and tolerability as well.

To broaden our understanding of drug response, real-world data sources can also provide adequate safety and efficacy information to help researchers predict precise doses for certain patient subgroups not represented in trial data, such as children and pregnant women. FDA has already created a repository of deidentified subject-level data to help the industry advance its understanding of drug behavior in different patient sub-populations. In addition, researchers recommended that the FDA and health plans should assess and share known phase III trial-RWD drug efficacy-safety gaps for major diseases and treatments.

To increase industry support for precision dosing approaches, healthcare leaders also proposed increasing training on quantitative modeling and simulations to show clinicians and pharmacists how precision dosing improves patient outcomes. Precision dosing at the bedside can also be incentivized with increased reimbursement, particularly under value-based models that reward the provision of high-value, individualized care. In the meantime, FDA and pharmaceutical companies should develop better dosing for relevant older drugs—as well as newer drugs—and translate this knowledge to prescribers and patients.

FDA and industry leaders also proposed solutions for how to address the industry-wide lack of clarity regarding regulatory pathways for clinical decision support (CDS) tools and encouraged software companies to design drug dosing CDS systems to support the needs of patients, physicians, and prescribers.

There is no question that quantitative approaches such as pharmacometrics and machine learning techniques will serve as the quantitative basis for personalized drug dosing, although many questions still remain as to how these approaches will be incorporated into the bedside and in clinical trials. As we usher in this new era of drug development—the age of individualized medicine—many researchers are calling on clinical specialist societies to create a high-priority drug-disease target list to highlight when precision dosing is likely to result in improved patient outcomes.

One thing is certain: we can be sure that individualized medicine is here to stay. Over time, the embrace of precision dosing will reduce costs, empower clinicians, and optimize drug delivery for every population segment, helping patients to realize the best possible clinical outcomes.

, PhD, CEO, Createrna Science and Technology, former Director of FDA’s Division of Pharmacometrics, and 

Continued embrace of precision dosing will reduce costs and optimize clinical outcomes.

Sirj Goswami, PhD

Articles

Understanding Project Optimus

Understanding FDA’s Perspective on Precision Dosing