Julie Bullock, PharmD

In January 2023, FDA published the draft guidance “Optimizing the Dosage of Human Prescription Drugs and Biological Products for the Treatment of Oncologic Diseases Guidance for Industry.” While not explicitly stated in the guidance materials, this guidance is being interpreted as the more formal codification of the Agency’s expectations associated with Project Optimus which was launched by the Oncology Center of Excellence (OCE) in February 2022. Key takeaways from the guidance are summarized below:

 Anticipating resistance for certain programs or situations, the Agency has given preemptive ‘no-excuses’ advice in the guidance.

First, the guidance states that it is not addressing selection of first-in-human starting dose and does not address dose optimization for radiopharmaceutical, cellular and gene therapy products, microbiota, or cancer vaccines, which implies that all other therapeutic modalities used in the treatment of cancer will be expected to include dose optimization as part of their development program.

In addition, sponsors developing drugs under FDA expedited programs, including Breakthrough Therapy designation, are also expected to conduct adequate pre-market dose optimization.

Lastly, the guidance is clear that operational hurdles such as the unavailability of suitable formulation strengths will not be adequate justification for a lack of pre-market dose finding.

To mitigate impact on development delays, the Agency points to prior use of seamless development approaches and recommends adequate planning to ensure dosage optimization is built into expeditious clinical development and is conducted in the pre-market setting.

The integrated analysis of all available clinical and non-clinical data to enable dose selection for the next stage of clinical development has been a common FDA/OCE theme since 2022, and it also makes its way into the guidance. This includes analysis of all available translational, clinical pharmacokinetic (PK), pharmacodynamic (PD), safety, and efficacy information to provide a stage-appropriate understanding of dose- and exposure-response (E-R) relationships. The Agency has always encouraged the use of population PK and E-R analyses early in development, and this guidance reiterates the importance of and expectation for these analyses early to justify dose levels selected for evaluation in subsequent development stages. Sponsors should plan for these modeling activities early in development and, at a minimum, plan for exploratory graphical analyses between observed exposure metrics, PD, safety, and activity endpoints.

Specifically, regarding clinical safety information, the guidance reiterates the Agency’s message that when selecting doses sponsors should account for lower grade but persistent symptomatic toxicities which may be difficult to tolerate long-term as patients often receive targeted therapies over extended periods of time. Advice is given to interrogate across dose levels the duration of exposure, the proportion of patients who can receive all planned doses, and dosage modifications or treatment discontinuations due to toxicities. Timing of safety events e.g., early-onset, toxicities which may lessen in severity or not occur with subsequent administration vs cumulative toxicities that emerge after multiple dosing should be considered to design alternative dosing regimens such as step-down or step-up and intermittent dosing. The guidance also encourages including patient reported outcomes (PROs) in early phase dosage finding trials to enhance assessing and understanding of tolerability. For oral drugs, the effect of food on PK and safety should be evaluated early in drug development to support the relative administration of the dosages selected for evaluation in any registration trials with food.

Previously limited to confidential sponsor-FDA communications during the IND stage, we now have a public position that inadequately justified doses may lead to potential clinical holds on trials as ‘FDA may determine that patients are exposed to unreasonable and significant risk, or there is insufficient information to determine risk, or the design of the trial is deficient to meet its stated objectives.’

Includingdose finding in development programs will inherently require more data in early clinical studies than sponsors are used to collecting historically. FDA points to this and mentions that following dose escalation, additional exploration may be needed to gather further safety and activity data to inform which dosages will be further evaluated in a dedicated dose optimization trial. To avoid delays, sponsors should plan for flexibility in dose escalation studies to gather this additional data since it will be uncommon for a dose escalation study in and of itself to gather enough data to determine doses that should be evaluated further for safety and efficacy. The guidance does not expand on specific practicalities on how this additional data should be collected such as recommendations for these early trial designs or how much data is adequate to determine dosages to evaluate further, nor does it mention how tumor heterogeneity commonly observed in early studies should be interpreted. We assume the relative vagueness on these topics is implied because the advice for each of these practicalities would be asset specific.

: Those of us hoping that the guidance would provide more specifics on the dose optimization study designs, number of participants, and endpoints will be left feeling a bit empty. The guidance provides consistent messaging with FDA’s workshops, and publications that were held as early as November 2021. While no one-size-fits-all approach is outlined, the guidance does provide valuable advice regarding adaptive designs, safety monitoring rules, statistical considerations, and incorporating multiple dosages prior to or as part of a registrational trial.

The guidance also states, ‘If safety and efficacy data from multiple dosages will be used to support a marketing application, this approach should be discussed with FDA early in clinical development.’ While the Agency’s message here is unclear, we infer it indicates the Agency’s openness to data from robust dose optimization trials being used to support a potential marketing approval.

An important topic that the guidance does not mention is what specific anti-tumor activity endpoints would be considered adequate to compare dosage in a dose optimization exercise. However, the Agency has previously indicated that it is common to base efficacy assessment in such trials on objective response rate (ORR) (+ duration of response; DoR), as it is an endpoint that can be measured earlier and often with fewer patients than time-to-event outcomes such as progression-free survival and overall survival. The Agency considers achievement of response is a direct measure of drug activity.

 However, in settings where ORR is a poor predictor of the ultimate clinical outcome, other outcomes could be considered.

Subpopulations; Are we muddying the waters? 

The guidance also recommends that population PK and E-R analysis should be used to identify specific subpopulations in which clinically meaningful differences in exposure are expected and to identify potential differences in safety or effectiveness for them. The guidance encourages incorporation of alternative dosages for relevant subpopulations into a registration trial ‘when feasible and appropriate’. In most cases, we consider this recommendation related to but distinct from the key issue under Project Optimus i.e., identifying a dosing regimen/exposure that achieves an appropriate risk-benefit balance in the overall population, to which subpopulations can subsequently be ‘exposure-matched.’ However, the Agency might be looking to potentially identify early on, subpopulations in which exposure-response and therefore risk-benefit balance might be ‘shifted,’ in which case exposure-matching might not be an adequate approach to dose recommendation. Either way, it is unlikely that early clinical trials will be able to (or allowed to) enroll adequately across broad subpopulations to facilitate any meaningful analyses. Instead, the additional variability by those who attempt to do this may cloud the already confusing results seen when using limited data to identify trends in safety and efficacy. Like other therapeutic areas, a more simplistic approach would be to keep the early phase studies as homogenous as possible to detect efficacy and safety signals and identify an optimal dose and slowly broaden inclusion/exclusion criteria as development progresses.

Overall, our impression is that the main messages in the guidance are consistent with and build upon the Agency’s messaging about Project Optimus in the recent past. While the draft guidance might appear lacking in detail, which may be filled out during the comments period, we consider the recent publication by Fourie Zirkelbach et al

, a ‘companion publication’ to the guidance. This publication from FDA authors across the Office of Clinical Pharmacology, Office of Oncologic Diseases, and OCE, provides specific case examples and an understanding of the background/ landscape that potentially formed the basis of the recommendations in the guidance.

, PharmD, senior vice president and head of clinical pharmacology and translational medicine, 

, PhD, associate director, clinical pharmacology, 

, PharmD, director, clinical pharmacology and translational medicine, and 

Fourie Zirkelbach J, Shah M, Vallejo J, et al. Improving Dose-Optimization Processes Used in Oncology Drug Development to Minimize Toxicity and Maximize Benefit to Patients. J Clin Oncol. 2022;40(30):3489-3500. doi:10.1200/JCO.22.00371

Articles

January 2023 guidance from FDA on dosages in oncology provides further context on expectations with Project Optimus.

FDA’s Project Optimus Guidance: Highlights and Tips for Success for Oncology Drug Developers

Insights from modeling and simulation (M&S) can help to overcome critical challenges associated with oncology clinical trials, by quantitatively integrating knowledge and relationships between the disease, drug characteristics, patient populations, and clinical trial parameters. M&S is used to fill in gaps related to limited data and extend the findings from existing trials for different scenarios and expanded patient populations. Such effort produces deeper understanding of a drug candidate’s efficacy and safety profile and can help to streamline the clinical trial and drug development process, thereby reducing patient burden, risk of failure, and time to market.

	A variety of challenges compound oncology drug development. They include:

	•  Patients with cancer who are enrolled in clinical trials are often already very sick and typically have comorbidities and numerous comedications.

	•  For some targeted therapies, clinical trials must enroll only patients with a specific oncologic profile, such as the presence or absence of a desired genetic mutation or other biomarker.

	•  Most oncology drugs are cytotoxic or genotoxic, and thus cannot be studied in healthy volunteers; similarly, targeted agents that can be studied in healthy volunteers may be limited in dose due to potential toxicities.

	Meanwhile, to reduce the patient burden, and streamline the drug development and regulatory approval processes, many oncology clinical trials are limited in scope and duration, and single-arm clinical trials are often leveraged to gain initial accelerated or conditional approval from regulators. This approach limits which patient subpopulations and doses are evaluated in the trial setting-leaving sizable gaps in understanding for drug investigators and regulators. While this approach can speed patient access to the new therapy, it also puts added pressure on investigators, and highlights the importance of M&S tools that can produce greater insight from the limited available trial data.

	Today, there is increased acceptance of M&S results and growing encouragement from regulatory agencies to use M&S tools. FDA Commissioner Scott Gottlieb recently included “the more widespread use of modeling and simulation, the greater use of real-world evidence in the pre- and post-market setting, and the adoption of better tools for collecting and evaluating more real-time safety information after products are approved” among the new scientific domains that have been introduced into the development and review process. Furthermore, the European Medicines Agency (EMA) just upgraded the reach of M&S within that agency.

	Specifically, advanced M&S approaches continue to improve oncology clinical trials by helping investigators to:

	•  Plan, inform, analyze, and extend clinical trial data and conduct standalone virtual trials to provide supplemental understanding, taking into account inherent patient-to-patient heterogeneity in terms of their response to therapy and tolerability to treatment.

	•  Develop objective response data to show how drug-mediated, tumor-growth inhibition impacts both overall survival and other relevant surrogate endpoints (such as objective response rates, progression-free survival, disease-free survival, and patient-reported outcomes).

	•  Quantitatively evaluate and compare the efficacy-safety profile of a new drug against the standard of care or other existing treatment options.

	•  Assess go/no-go decisions, support comparator-effectiveness studies, and streamline regulatory filings.

	Poor or ill-informed dose selection is often to blame for failed trials, delays, and denials of regulatory submissions, and changes in dosing post-approval. M&S is being increasingly leveraged to improve dose escalation and determination of first-in-human (FIH) doses, to predict and analyze variable dose-response, and to optimize dose-regimen decisions-more broadly and more comprehensively than is possible using the traditional approach of hypothesis testing in a limited trial setting. M&S can be used to both interpolate and extrapolate existing trial data related to specific tested doses, compare with existing treatment options leveraging publicly-available data from competitor compounds, and thus investigate other possible doses and dosing strategies without the need for additional human subjects or dedicated studies.

	Historically, oncology drug development involving traditional cytotoxic agents relied on maximizing toxicity, using the maximum tolerated dose (MTD) as the key indicator for maximizing treatment efficacy. Today’s newer biologic agents and immuno-oncology therapies are often able to provide an efficacious dose well before drug levels have become toxic. Thus, clinical trials for such novel oncology agents must work to identify optimal doses and dosing frequency below the MTD to achieve the needed efficacy with better tolerability for the patient. This considerably complex undertaking is greatly enhanced using M&S.

	Due to polypharmacy, patients with cancer are at risk of multiple drug-drug interactions (DDIs) and it is impossible for clinical investigators to conduct an endless array of dedicated studies and trials to identify and understand all potential DDIs that may arise for a given oncology agent. Instead, investigators are increasingly turning to quantitative modeling techniques, which can produce rational, data-driven predictions about how the drug’s absorption, distribution, metabolism, and excretion profile will impact different DDI combinations across many different simulated patient cohorts. The resulting insight can be used to inform drug labeling, and also help guide the inclusion or exclusion of patients taking DDI-implicated drugs during clinical trial design and post-marketing studies.

	Today, promising work is also underway using modeling techniques to identify potentially advantageous DDIs-that is, specific combinations of investigational and approved drugs that may be able to improve clinical outcomes or tolerability.

Predicting drug activity in virtual patient cohorts

	M&S lets investigators leverage relationships elucidated in the existing trial data to predict the drug-exposure impacts, clinical efficacy, and toxicity of investigational therapies in specific modeled patient sub-populations. For instance, such virtual patient cohorts may be structured by common variants such as age, gender, ethnicity, and weight, but also by more complex characteristics such as the presence or absence of specific genetic mutations or other biomarkers, specific comorbidities and coprescribed medications, organ impairment, and vulnerable patient groups (e.g., pediatric or geriatric patients and pregnant women).

Using modeling to assess the competitive landscape

	As the number of oncology agents continues to grow, drug developers must demonstrate not just how an investigational drug performs, but how it compares to other available therapies and those under development. Such comparative effectiveness studies typically look at how the investigational therapy stacks up in terms of clinical effectiveness and safety profile, and also potential complications, tolerability, dosing strategies, and potential DDIs-all factors that could reduce long-term adherence to therapy and clinical outcomes for patients.

	Modeled and predicted results help to extend available data produced in actual clinical trials, and answer key questions about how the drug performs in larger, virtual patient populations. This helps to produce the strongest case for drug developers to present to regulators and payers regarding how the drug is likely to perform in heterogeneous patients under real-world conditions.

	The high cost and competitive landscape for oncology drugs have resulted in increased pressure from healthcare payers to justify adding the therapy to the formulary-another area for M&S, albeit related to health economics.

Exposure-response and pharmacokinetic/pharmacodynamic modeling

 incorporate PK and PD data gathered during early- and late-stage clinical trials. 

	Exposure-response modeling is increasingly used to support optimal dose selection; provide proof-of-concept for the drug; elucidate and validate the treatment’s mechanism of action; improve characterization of relationships between drug exposure, efficacy, and toxicity; and inform interpretation of risk-benefit profiles. Such insight is essential not only for dose justification (which relies heavily on establishing dose-response relations) but for regulatory labels where dose modifications may be required for specific patient populations to avoid adverse events.

	Population PK modeling is used to leverage sparse concentration data and evaluate the variability of drug exposure across individuals in a population over time. Such M&S approaches can facilitate not just development of new treatments for adults but also streamline pediatric drug development programs.

 helps to predict the drug’s PK activity both in the body and the tumor site. PBPK models are built using preclinical and clinical trial data and can assess the therapy in simulated patient subpopulations to inform further clinical trial design and product labeling.

	PBPK modeling is often use to predict DDIs but it can also extrapolate drug-function findings between patient cohorts, based on age and disease demographics and physiological differences (e.g., adult versus pediatric populations or cancer patients versus healthy volunteers). PBPK has gained significant traction with global regulators and PBPK-modeled results (in lieu of clinical studies) have been accepted by FDA to support more than 150 label claims.

combines computational modeling and experimental methods to examine the mechanistic relationships between an investigational therapy, the biological system, and the disease process. QSP models can help to elucidate how target exposure, binding, and expression occur in biological pathways, impacting disease determinants, drug efficacy, and disease progression. This allows for optimal combinations and dosing regimens to be explored within a virtual population. QSP modeling is also used to reduce Phase II attrition by enabling a wide range of “what-if” scenarios to be investigated-ahead of the actual clinical investigation-to optimize trial design.

	QSP models are also being used to simulate biomarker responses for a drug or multiple-drug regimen across virtual trials, providing added insight that can help to inform ongoing trial design. Meanwhile, many researchers consider QSP modeling an essential element for successful immuno-oncology drug discovery, because the number of possible combinations (in terms of drugs and dosing schedules) is simply too numerous to explore experimentally.

is used to compare the investigational therapy with other drugs being tested in clinical trials or on the market. MBMA-demonstrated superiority for the drug candidate provides ongoing encouragement for investigators. In contrast, demonstration of non-superiority allows the drug developer to either leverage the findings and fine-tune the design of ongoing trials, or revise corporate priorities (perhaps focusing limited resources on more-promising candidates). MBMA can help to understand expected response in control arms (e.g., against the standard of care), and to identify a dose that is expected to be associated with competitive efficacy and safety outcomes in clinical trials.

	MBMA is also used to provide “virtual comparator” data, to put into context benefits seen in single-arm trials.

aims to better characterize tumor-size responses to therapy to establish the optimal dose regimen and therapeutic window and allow use of that relationship as an early marker for survival. Several promising oncology products have received their initial approval on the basis of tumor size in response to therapy (objective response data)-rather than survival outcomes. This places further emphasis on developing a thorough understanding of tumor-size dynamics and the effects of investigational drug candidates on tumor growth or shrinkage. As a result, novel approaches to tumor-size modeling are being developed and applied to support both drug development and regulatory decision-making.

	The strongest development program makes maximal use of all available data at the outset, and then applies M&S to answer essential questions, explore “what if?” scenarios, fill in critical gaps, and assess how the investigational therapy works and is tolerated in expanded patient subpopulations and different scenarios to those evaluated in the clinical trial dataset.

	There is a growing body of evidence in academia, industry, regulatory agencies, and health authorities that M&S can facilitate development, approval, and cost justification of oncology drugs. This trend will continue.

		PBPK Modeling and Simulation: Yesterday’s Scientific Endeavor is Today’s Regulatory Necessity. 

https://www.aapsnewsmagazine.org/articles/2017/dec17

		The Model-informed Drug Development Imperative in Oncology R&D. 

https://www.certara.com/wp-content/uploads/Resources/Booklets/BK_MIDDImperativeOncology.pdf

, MD, FCP, are vice presidents at Certara. Dr. Pfister is also Professor of Pharmacology and Pharmacometrics at the University of Basel.

How the use of M&S in cancer trials from the outset can help address those critical “what if?” scenarios and accelerate oncology drug development. 

Julie Bullock, PharmD

Articles

FDA’s Project Optimus Guidance: Highlights and Tips for Success for Oncology Drug Developers

Leveraging Modeling & Simulation in Oncology