Comments submitted in response to the FDA Oncology Center of Excellence’s (OCE) recent dose optimization guidance demonstrate stark differences in stakeholder perspectives, ranging between solid support and fervent opposition to the operational approaches proposed by FDA’s guidance. In this extensive analysis, AgencyIQ unpacks the contrasting opinions regarding what dose optimization truly means and what is most important to establish during the early stages of oncology clinical trials.

Background on dose-finding in oncology

• In the early phases of clinical development of oncology products, sponsors conduct dose-finding studies to determine the maximum tolerated dose (MTD) of an investigational product. MTD seeks to find the tolerability and safety of treatment at the highest dose that can both be tolerated and treat the cancer.
• There are several limitations to the MTD approach. This approach, which was initially developed for systemic chemotherapies, is generally no longer applicable to new treatments that are more targeted in their effects, such as immunotherapies and new molecular targeted agents (MTAs). In addition, the MTD approach ignores target interactions and off-target toxicities, evaluates too few patients at each dose, offers only a short observation period to observe dose-limiting toxicities (DLTs), and offers limited consideration of safety information beyond DLTs. These challenges may result in approved treatments that may be poorly tolerated by patients, eventually affecting their quality of life and potentially leading to drug discontinuation.
• There have been several workshops on oncology dose finding hosted by the FDA in the last few years. In 2015, the FDA, along with the American Association for Cancer Research (AACR), held a workshop on dose optimization for small molecules. A year later, a  similar workshop was conducted that included biologics and nonclinical models for efficacy, as well as modeling and simulation for dose-finding and dose optimization trials design. A  third workshop was held in 2017 and focused on combination products, use of biomarkers and novel endpoints, and modeling and simulation approaches. Friends of Cancer Research (FOCR) has also been working to  improve tolerability and dose optimization programs in collaboration with various stakeholders. [Read our analysis of the meeting held by FOCR  here.]
• The FDA’s Oncology Center of Excellence (OCE) recently established a new program, Project Optimus, focused on dose optimization. The Project was first unveiled in an  interview in 2021 when Richard Pazdur, the Director of the FDA’s OCE, explained that the field of oncology has long suffered from a cultural bias that “more is better” when it comes to dosing oncology products, under the idea that the highest dose that a patient can withstand is preferable to a lower dose. “We’re not asking anything more than [we are for] any other therapeutic areas,” Pazdur explained, adding that OCE is “going to start making it more of a requirement” that companies conduct dose-finding to facilitate optimization and that it is in companies’ “best interest” to follow this approach. [Read our brief here.]
• Last year, the FDA dedicated a two-day workshop to Project Optimus, partnering with the American Society of Clinical Oncology (ASCO) to focus on dose optimization strategies in oncology drug development. At the workshop, FDA and external presenters highlighted the importance of conducting randomized dose-finding trials early in the development program and of capturing patient-reported outcomes (PROs) when measuring toxicity; presenters also provided strategies for the use of non-clinical pharmacology and model-informed drug development (MIDD) to bolster dose optimization work during drug development. [ Read AgencyIQ’s analysis of the workshop here.]
• In recent years some stakeholders have advocated about the importance of treatment tolerability when determining dose(s) and regimen(s). Due to the severe side effects of some anticancer treatments, patients are either discontinuing treatment before reaching the targeted maximum benefit or are receiving the treatments at a lower dose. These issues are especially apparent in treatments that are intended for chronic use. As a result, the products receiving approval due to their demonstrated safety and efficacy appear to lack real-world effectiveness. As a result, patients are failing to see material benefit despite dealing with persistent and severe side effects.
• Some research has indicated that the tolerability issue is being further exacerbated by accelerated approval pathway (AAP) because the designation is granted using limited data, because sponsors are either failing to enroll/conduct confirmatory trials, and/or because the confirmatory trials report underwhelming results that are not as strong as the initial readouts due to issues with high toxicity, differences in patient populations, dosage/regimen adherence issues, and common product availability via expanded access. One example of this issue was demonstrated during a recent Oncologic Drugs Advisory Committee (ODAC) meeting to discuss Pepaxto, where the committee faulted the sponsors’ analytical methods and sharply criticized the sponsor’s dose  optimization process. In particular, the sponsor did not explore lower doses or alternate dosing regimens and used the maximum tolerated dose for the product, though FDA reviewers noted there was no pharmacokinetic or exposure-response analysis conducted using the maximum tolerated dose. “An optimal dose remains unestablished,” said Alexandria Schwarsin, clinical reviewer at the Division of Hematologic Malignancies II, despite post-marketing requirements for dosing, because of inadequate study design, conduct and analysis. [Read AgencyIQ’s full report of the ODAC meeting here]

Then, in January 2023, FDA released a long-awaited guidance on dose optimization for oncology products

• The draft guidance document is intended to help sponsors of both prescription drugs and biologics identify optimal doses for their products during the development process, before regulatory submission for approval. The document specifically excludes dose optimization for radiopharmaceutical or cell and gene therapy products; cancer vaccines and micriobiota-related products are also excluded. The guidance also is not meant for “selection of the starting dosage for first-in-human trials,” the FDA clarified.
• In the guidance, the FDA outlines why the MTD strategy is particularly inappropriate for targeted therapies, which “demonstrate different dose-response relationships compared to cytotoxic chemotherapy, such that doses below the MTD may have similar efficacy to the MTD but with fewer toxicities,” according to the guidance. “Additionally, the MTD may never be reached in certain situations.” The FDA makes the additional point that many modern therapies are meant to be taken for long periods of time, even chronically in some cases, shifting considerations for evaluating toxicity.
• The FDA advocates for an approach that can take advantage of the “seamless” design of many newer oncology trials, “characterized by rapid transitions between initial dose-finding trials and registrations trial(s) to expedite development.” With good planning on the sponsor’s part, dose-finding strategies “can be merged into a seamless development program.” In any case, dosage optimization should happen before approval.

What the guidance recommends:

• When selecting doses for a clinical trial, sponsors should look to “relevant nonclinical and clinical data,” as well as what’s known about both dose-response and exposure-response for the candidate drug or biologic in terms of both safety and efficacy. The FDA will take a dim view of approaches that select doses “without adequate justification or consideration of relevant data,” both because of the potential risk to patients and because the trial design might not be able to meet required objectives for approval.
• These considerations also apply to drugs being developed under expedited pathways, such as breakthrough designation. The agency makes this clear in no uncertain terms: “Sponsors should note that development of a drug under an FDA expedited program (e.g., breakthrough therapy designation) is not a sufficient justification to avoid identifying an optimal dosage(s) prior to submitting a marketing application.”
• Trials involving dose-finding should have a robust clinical pharmacokinetic (PK) component that includes PK characteristics such as linearity, absorption, and elimination of the candidate drug, after multiple doses. The sampling and analysis plan should also support population PK and the required analyses of dose-response and exposure-response relationships.
• When these data are available, they should be evaluated “along with the anti-tumor activity, safety, and tolerability data” to select which doses should be evaluated further. Additional PK and safety data should be gathered, as appropriate, for oral drugs.
• The draft guidance recommends enrolling “an appropriately broad population” to evaluate how the dose will perform in “relevant subpopulations.” Here, the FDA inserts a series of five footnotes referring sponsors to its guidance in related areas, including  enhancing diversity of clinical trial populations, making trials available in  non-curative settings and for people with  organ dysfunction or previous cancers, and more. When doses might differ for subpopulations, these alternative doses can be calculated by “simulated exposure metrics” and included in registration trials.
• Pharmacodynamic (PD) and pharmacogenomic data may also be appropriate for inclusion in the clinical trial; in any case, the FDA should review the proposed sampling and analysis plans before the sponsor proceeds.
• In terms of how dosages should be compared, the FDA recommends a randomized, parallel dose-response trial, when feasible. Though randomization is preferred, the trial does not have to be powered to demonstrate superiority or even non-inferiority of one dose over another. An adaptive design with stopping rules following interim safety and efficacy assessments could be an acceptable strategy, according to the draft guidance.
• Multiple doses can be studied in a registration trial by adding an additional dosage arm or arms. However, a trial that is meant to show that one of the dosages is superior to a control arm should “provide strong control of Type I [statistical] error,” accounting for a design that tests multiple doses against a single control.
• If a sponsor is looking to use safety and efficacy data from multiple doses to support its application for marketing authorization, this design should be discussed with the FDA “early in clinical development.”
• Considerations for evaluating safety and tolerability of doses include capturing a host of variables, according to the draft guidance document. These include duration of exposure, how many patients were able to take all planned doses, and how many patients had dose interruptions, reduction, or discontinuations because of adverse reactions to the candidate therapy. All of these data, as well as the total percentage of patients in each arm with serious or fatal adverse reactions, “should be compared across the multiple dosages,” according to the FDA.
• When a trial will include dosages “associated with a high percentage of dosage modifications or serious adverse reactions,” sponsors should pre-specify safety monitoring rules, including what the sponsor will do if serious adverse reactions or dose modifications cross a given threshold. In such cases, sponsors may pause the trial, reduce the starting dose for future trial enrollees, or even end the trial.
• In a nod to discussions had at the 2022 two-day meeting on dose optimization, the draft guidance notes that some adverse reactions that are reported as “less severe,” such as grade 1 or 2 diarrhea, “may still significantly affect a patient’s ability to remain on the drug for extended periods.” For drug reactions like these, sponsors should track both the frequency and the impact of the reactions on dose interruptions or discontinuations.
• Conversely, for some oncology drugs, the most serious toxicities happen with the first dose, or few doses. Toxicity “may lessen in severity of not occur with subsequent administration” for these drugs; the FDA recommends considering titration or other stepwise dosing “to improve tolerability.”
• Another topic discussed extensively at the dose optimization meeting made its way into today’s draft guidance: Patient Reported Outcomes (PROs). The FDA recommends including PROs “to enhance the assessment of tolerability in early phase dosage finding trials,” since these measures “can provide a systematic and quantitative assessment of expected symptomatic adverse events and their impact on function.”
• The draft guidance document provides considerations for dose formulations, including the admonition that “Perceived difficulty in manufacturing multiple dose strengths is an insufficient rationale for not comparing multiple dosages in clinical trials.” Considerations for both oral and parenteral use are included in the draft.
• Generally, when subsequent indications are being considered, sponsors should recognize that dosages may differ because of differences in the tumor biology, patient population, and concurrent treatments, among other reasons. Accordingly, nonclinical and clinical data will need to be gathered to support a subsequent indication, the guidance document clarifies, adding that “Strong rationale for choice of dosage should be provided” before a registration trial for a new dosage is launched. Sponsors should be prepared for the possibility of additional dose-finding studies being required.

What industry thinks of the guidance

• Earlier in March, the public comment period on the guidance document ended with over fifty comments submitted to the docket. In addition to pharmaceutical companies and industry trades, there were more than a few noteworthy organizations included in the list of public comments submitted including the Association for Clinical Oncology (ASCO), Friends of Cancer Research (FOCR), National Cancer Institute (NIH), Society for Immunotherapy of Cancer (SITC), LUNGevity Foundation to name a few. Interestingly, one contract research organization, Certara, who has been closely involved in this space previously, also submitted comments. The breadth of important stakeholders that responded and the depth of comments submitted, led to an interesting comparison of perspectives.
• Across the board, there were a few topics that respondents largely agreed on. These included the need for additional detail regarding how the guidance applied to combination products where the investigational product has minimal activity when used as a monotherapy; requests for more specific terminology (especially regarding the use of the word “broad” to describe patient populations since there are also broad tumor types); requests for more information on the use of mechanistic modeling/simulation to inform dose optimization; and appeals for FDA to add in more detail on the statistical considerations covered in the guidance.
• There is a need for additional guidance, and flexibilities to accommodate, novel oncology therapeutics: The Society for Immunotherapy of Cancer (SITC) noted that the word “biologics” appears in the title of the guidance, yet the guidance states that it does not cover cell or gene therapies. Numerous comments suggested that additional flexibilities and recommendations should be added into the guidance on antibody-drug conjugates (ADCs).
• Furthermore, respondents largely agreed on the need for additional detail and process clarification regarding early and effective communication between sponsors and FDA. Friends of Cancer Research (FOCR) suggested that a template for the exchange of key information on dosing would be especially useful for sponsors. Multiple comments endorsed FOCR’s white paper, Optimizing Dosing in Oncology Drug Development, for its discussion of this exact topic and inclusion of a tool in the appendix that could be leveraged to prompt earlier discussions on dose-finding topics between sponsors and FDA. The Society for Immunotherapy of Cancer (SITC) also requested further detail on the type of meeting FDA suggests sponsors to utilize to discuss dose-finding information and cited concerns that such a meeting would not occur as rapidly as necessary.

However, comments also highlighted some fundamental differences in stakeholder priorities when it comes to dose optimization in early oncology trials.

• One initial topic of disagreement was whether sponsors can and should enroll broader patient populations in early dose-finding studies. The draft guidance recommends sponsors identify and account for specific subpopulations (race, gender, age, organ impairment etc.) that demonstrate clinically meaningful differences in drug exposure. Certara, the only contract research organization (CRO) to submit comments stated “It is unlikely that early clinical trials will be able to (or allowed to) enroll adequately across broad subpopulations to facilitate any meaningful analyses. 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 to include subpopulations as drug development progresses.” In stark contrast, the Association for Clinical Oncology (ASCO) confidently asserted that strategies to optimize dosage can be merged into a seamless dosage-optimization development program. However, these strategies necessitate the enrollment of appropriately broad populations to identify differences across relevant subpopulations. Furthermore, it recommended the Agency go a “step further to reference all FDA guidance documents on broad eligibility criteria and provide examples of these broad populations to include older adults, sexual and gender minorities, race and ethnicity, and other patients who have historically been underrepresented in cancer clinical trials.”
• Use of Patient-Reported Outcomes (PRO) Data: Patient advocates endorse the idea, but industry isn’t convinced. As the Pharmaceutical Research and Manufacturers of America (PhRMA) wrote, “We encourage the Agency to provide flexibility and acknowledge in the final guidance that interpreting PRO data may be challenging due to potential limitations including small patient numbers or situations where it may be difficult to attribute patient experience to either disease symptoms or tolerability.” Furthermore, the ‘Statisticians in Pharmaceutical Industry (PSI) & European Federation of Statisticians in Pharmaceutical Industry (EFSPI)’ stated that PROs may be appropriate in later studies but are not relevant during dose-escalation studies. Furthermore, it interestingly cited that, “Considering the high number of assessments already done in dose escalation trials, it may be too much burden for these late line patients.” In either case, other organizations noted that blinding status is an important factor to consider when incorporating PRO data. Additional guidance from FDA covering approaches to mitigating bias or publication of a standardized instrument sponsors could use to collect PROs in early clinical trials for the purpose of evaluating tolerability and side effects would be especially useful, they said.
• In contrast, patient advocacy organizations stated strong support for the use of PROs in early phase dose-finding trials as one factor for assessing tolerability. In particular, the LUNGevity Foundation pointed out that “Patient insights can provide a depth of understanding about side effect burden and drug tolerability that cannot be replicated with clinician-reported outcomes, as has been shown in the literature. […] If PRO data are thoughtfully captured during the dose finding process, they could help to manage side effects and mitigate risk in later phases of drug development.”
• The next controversial topic was timing of dose optimization studies. Perhaps the best summary of the comments received is the statement submitted by the Statisticians in Pharmaceutical Industry (PSI) & European Federation of Statisticians in Pharmaceutical Industry (EFSPI): “It is not clear where the dose optimization study best fits in a drug’s development pathway.” In agreement, Certara noted that some of the language in the guidance seems to be “confusing early First-in-Human (FIH) study and trials directly comparing multiple dosages,” and this imprecise language could lead sponsors to consider their first expansion after dose escalation as the definitive randomized dose optimization study. In AgencyIQ’s opinion, it’s possible that some of this mixed feedback regarding timing of dose optimization studies is a result of widespread seamless trial design adoption and the increasingly blurred line between phase 1 and 2 oncology clinical trials.
• Some respondents further questioned the ethics of enrolling additional patients on multiple dose levels to assess a product’s efficacy at different levels before the clinical effectiveness of the therapy has been sufficiently established. The NIH’s National Cancer Institute (NCI) had a particularly strong stance on this topic stating, “Subjecting substantial number of additional patients to dose optimization studies of ineffective/non-viable agents will expose large number of patients to unnecessary toxicity and would be a suboptimal use of resources.” It argued that “Dose optimization studies should be undertaken in the late phase 2 setting only after there is clear evidence of efficacy and clear, pre-defined levels of unacceptable tolerability.”
• Interestingly, patients and clinicians disagreed, according to the Association for Clinical Oncology (ASCO), a leading voice on topics in oncology. The organization offered strong support for the FDA’s decision to put this recommendation in guidance. “There may be some concern that more patients will be needed to evaluate agents in the early phase of drug development if doses are expected to be optimized, which may lead to more patients being exposed to ineffective agents. However, better dosing studies will increase the likelihood of successful later phase trials for promising agents, and avoidance of poorly tolerated doses and schedules in later phase trials and ultimately in practice.”
• Next: to randomize, or not to randomize… that is the question. PhRMA requested more flexibility on randomization of trials, stating, “We believe that dosage selection should be based on the totality of information available for a specific oncologic product. Additionally, principles outlined in the guidance may not be feasible for rare diseases due to challenges in recruiting a sufficient number of patients.” On the opposite end of the spectrum, ASCO not only stated that “randomized, parallel dose-response trials and trials designed to assess activity, safety, and tolerability comparing multiple dosages are appropriate and key,” but went even further and called for FDA to put stronger (binding) measures in place beyond just issuing guidances to “ensure sponsors adhere to these recommendations of trial designs to compare multiple dosages.”
• What does “optimal dose” actually mean in oncology and how should it be measured? Industry stated that products are often not effective at doses below the maximum tolerated dose (MTD) and noted concern for unethically “underdosing” sick patients in early clinical trials. PhRMA wrote that FDA acknowledges only the negative consequences of unnecessary toxicities using the maximum tolerated dose in the draft guidance and fails to recognize the potential risk of exposing patients with life threatening disease to a sub‐therapeutic dosage and lower efficacy in instances where “the maximum tolerated dose (MTD) or maximum administered dose (MAD) may be the only biologically active dose or in instances where observed non‐clinical pharmacodynamic data is the main basis for requests to evaluate lower dosage(s).” The National Cancer Institute agreed with PhRMA on this point noting, “Most investigational drugs that are developed into marketable agents are tolerable at the dose used in phase 3 studies. Only a minority of drugs are dosed too high. The FDA guidance focuses on a small minority of approved oncologic agents.”
• At least three organizations representing patients and clinicians directly discredited these statements, asserting that maximum tolerated dose (MTD) is no longer relevant and is not the only dose effective in treating cancer. According to Friends of Cancer Research (FOCR), “Historically, dose-finding trials in oncology were designed to identify the maximum tolerated dose (MTD) that does not cause unacceptable side effects. […] With the advent of precision medicines designed to interact with specific oncologic pathways, the maximum tolerated dose (MTD) may not only be unachievable but lower doses may achieve similar efficacy as higher doses while also having better tolerability.” The Association for Clinical Oncology (ASCO) affirmed this point stating that “use of the classic MTD is not relevant” for many treatments and an active dose can be effective in treatment at lower doses than the MTD. Other patient organizations agreed and stated that more resources should actually be put into understanding which doses are tolerable before therapies are pushed forward in development and patients are ultimately prescribed treatment plans that are not actually feasible.
• Where exactly do tolerability and feasible regimen adherence factor into evaluating a product’s efficacy/effectiveness? The Association for Clinical Oncology (ASCO) wrote that it “agrees with the FDA that delaying dose optimization until after approval could result in patients experiencing excess toxicities due to a poorly tolerated dosage and potentially discontinuing therapy without maximal clinical benefit.” The LUNGevity Foundation further affirmed that OCE’s efforts to modernize the “dose selection process and weigh the impact of persistent, low-grade adverse events on patients’ quality of life, are valuable and necessary.” Yet comments submitted by the National Cancer Institute fired back about role tolerability plays in drug development, especially in early phase trials. “The FDA guidance does not contain clear criteria for defining tolerability and does not describe how precision of its measure can be determined. If tolerability is to be a primary endpoint of a study, then it should be defined as rigorously as efficacy.”
• An interesting point raised by the LUNGevity Foundation: Finally, [we] are concerned that terminology used in this guidance and for Project Optimus, namely, dose optimization and optimal dose overpromises as the word “optimal” denotes a singular best, just as “optimization” is the process of arriving at the best. This terminology is misleading for patients, who may believe that with dose optimization in place, drug toxicities and dose reductions will be things of the past. However, in reality, no dose will be “best” for everyone regardless of how carefully and with how much information it was selected. We suggest OCE adopt more practical language, such as “dose selection,” and be candid about what the new expectations for arriving at the recommended dose will—and will not—mean for patients.

How to bridge the gap?

• One set of comments highlighted two opportunities where OCE could work to bridge this gap by including additional recommendations on the use of mathematical and computational methods in dose determination. The Optimal Control Working Group of the Mathematical and Computational Sciences Special Interest Group of the International Society of Pharmacometrics stated that mechanistic mathematical modeling can allow for extrapolation beyond the specific population of individuals included in clinical trials. Secondly, mathematical optimization could be used in cases where well-validated mechanistic models and clear patient outcome objectives exist. Constrained optimization methods could solve for “optimal” dose regimens based on a certain population or even for individuals. The group noted that “This goes beyond the “guess and check” methods that are often used in model simulation studies, which run several simulations of outcomes for selected regimens. If we believe these models for predictions of selected regimens, then we can believe them for regimens that are computed to be optimal. A predicted optimal regimen should then be tested as one arm in a dose-finding study.”
• Incorporation of these methods would involve up-front investment and the collection of additional, more robust data points. The comments did note that additional data collection from study subjects may increase the cost to conduct clinical trials initially. However, in the long-term, incorporation of mechanistic modeling and optimization methods would ultimately reduce the number of studies required and result in (theoretically) better safety and efficacy outcomes.

https://fda.agencyiq.com/article/00000187-30a9-dd77-a1cf-79adab210000