《FDA Optimus项目要点》

We asked Parexel regulatory experts to share their insights about FDA’s Project Optimus. 1. It is clear that small molecules will fall under Project Optimus. Likewise, large molecules (nivolumab) andcytotoxics (capecitabine) have been used as ‘poster children.’ Do you believe that vaccines and cell-based Generally, the type of drug does not exempt from assessing a proper dose-(R)response analysis. D-R orE(xposure)-R analyses applies to all drugs and uses both non-clinical and clinical data. 2. Under Project Optimus, should a sponsor plan to request meeting requests with US FDA at least twice?The first one to discuss dose optimization study design, and the second one to get FDA agreement on the Discussing the selected dose for a pivotal study with the FDA would be beneficial. You could discuss your initialplan during the pre-IND meeting and then plan at least one more additional meeting to discuss pivotal study 3. Do you think the US FDA will also apply this standard to TIL and TCR therapies which have a one-time Overall, the tools/approaches discussed are not typically used for individually optimized treatments and precisiondosing. However, we are seeing increasing importance of assessing clinical pharmacology aspects for CAR-T cells(e.g., https://ascpt.onlinelibrary.wiley.com/doi/epdf/10.1111/cts.13349) and the principles of quantitative clinical 4. You state that Bayesian Optimal Interval (BOIN) designs are better than 3+3. However, it is not great fordose escalation (due to the low sample size). 3+3 is typically only used during dose escalation, so how do youjustify recommending BOIN instead of 3+3 when BOIN is not ideal during a dose escalation? It seems not to The BOIN design is also algorithm-based but has better-operating characteristics than the 3+3 design. Itallows for more flexibility, e.g., setting the DLT rate and the cohort size. It allows for decision-making at any timeduring the trial by comparing the observed DLT rate at the current dose with the escalation and de-escalation 5. RECIST ORR is a very blunt endpoint. What about the use of biomarkers, e.g., ctDNA? We agree that ORR is a blunt measure of efficacy. Dynamic endpoints such as ctDNA, tumor kinetics, andother biomarkers should be considered during the exposure-response analysis to identify the efficacy gainswith increasing doses. It would be beneficial to discuss all planned dynamic endpoints with the FDA. The most PRO-CTCAE should be implemented at least in dose expansion. 7. Please elaborate on randomization in dose-finding studies: why (what to randomize for), how (small N),when (Ph1 expansion or Ph2/3), and why not to randomize/alternate strategies. The population should be randomly assigned to treatments to avoid bias in selecting patients, which can falsifythe results by a biased pre-selection. There can be seamless designs with study sub-parts as a continuum from 8. The reason most companies go up to MTD/MAD might be due to hesitancy in leaving efficacy on the table,considering that tumor microenvironment availability of the drug is based on assumptions. Is that correct? With targeted therapy, you may not reach MTD. If D-R or E-R response relationship is flat, a lower dose may workas well. Safety as a response needs to be factored in and appropriate analyses provided to the agency. It may still 9. Should you be looking for a sub-optimal dose in the randomized phase 2 with multiple doses? It would be best if you did not look for a sub-optimal dose. The goal is to find the optimal dose that does not 10. While not typically done - in light of Project Optimus, do you see any role of initial dose finding in healthyvolunteers in Oncology, such as for molecularly targeted agents? Is there a stepping stone to obtaining data in Healthy volunteer studies typically are not performed with oncology therapies because of the potential short-and long-term adverse reactions associated with such therapies. Some therapies may have a toxicology profile 11. For oncology studies that have already started, what are the options? E.g., what should we do regardinglooking back at your expansion cohorts and opening up additional dose cohorts to support dose justification? Many sponsors will find themselves in late-stage drug development using the MTD approach to identify theRP2D. We would suggest reviewing all the data and material to assess the quality of the dose justificationarguments. Modeling and simulations may play a role in supporting a registrational dose; however, this will be 12. What mechanisms are available for early and often FDA interactions? There are pre-IND meetings that FDA highly encourages to use. Afterward, sponsors may engage with theappropriate FDA division during type B or C meetings when more data are available to be discussed. 13. Are there mechanisms (besides a pre-IND meeting) to discuss dosing considerations with the Project Sponsors may engage with the appropriate FDA division during a type B or C me