BOIN design

• Bayesian optimal interval (BOIN) design provides a novel platform for designing phase I trials with single agent, drug combination, and delayed toxicity in a unified framework. As model-assisted design, the BOIN combines the simplicity of algorithm-based design with the advantages of model-based BOIN design can be implemented in a simple manner similar to a 3+3 design, but yields superior performance comparable to more complex model-based designs, such as the continual reassessment method (CRM).
• Time-to-Event BOIN (TITE-BOIN) allows for real-time dose assignment for new patients while some enrolled patients’ toxicity data are still pending, thereby significantly shortening the trial duration. It is as easy to implement as the rolling 6 design, but yields much better performance.
• Algorithm-based trial design, commonly used in dose escalation trials to find MTD. The method is simple and easy to implement, but the method itself is rough, there are shortcomings such as inaccurate MTD design, too many patients are assigned to the low-dose group for the test, inability to flexibly set the MTD target level (only 33%), and the dose level will need to be set too much.
• Model-based trial design, commonly used in dose-escalation tests, is designed to find MTD. The Continual reassessment method (CRM) is a model-based dose-finding approach that assumes a parametric model for the dose-toxicity curve. As information accumulates during the trial, the dose-toxicity curve will be reassessed by updating the estimates of unknown model parameters and used to determine dose allocation for the next cohort of patients.
• Bayesian Model Average CRM (BMA-CRM) improves the robustness of CRM by utilizing multiple dose toxicity models to reduce the opportunities for misspecification of standard CRM models. Based on accurate data, BMA-CRM automatically adopts a best-fit model to make dose allocation decisions using Bayesian Model Averaging.
• Toxicity probability interval design, based on statistical decision theory, has the theoretical property of a large sample that converges to the true MTD value. Compared with CRM, mTPI needs less related setting work before the test. For example, CRM needs to make pre-trial assumptions on the toxicity - dose model and skeleton probability, while mTPI only needs to set the equivalent interval E1. And the setting of E1 is easy to be understood by the experimental researchers and easy to communicate.
• Keyboard Design provides a novel platform for designing phase I trials with single agent and drug combination. As model-aided design, Keyboard design combines the simplicity of algorithmic design with the superiority of model-based design. Keyboard design can be implemented in a simple manner similar to the 3+3 design, but yield superior performance comparable to more complex model-based designs, such as the Continual reassessment method (CRM).

• U-BOIN is a utility-based seamless Bayes Phase I/II trial designed to find the optimal biological dose (OBD) for targeted and immune therapies. It allows physicians to incorporate the risk-benefit trade-off to more realistically reflect clinical practice.
• The aim of this method is to explore the optimal biological dose, considering immunity, efficacy and toxicity, with good robustness and easy operation.

• it is commonly used in one arm two-stage test of tumor drugs. It has the characteristics of saving sample size, early finding that the test is invalid, timely terminating the test, avoiding subjects from receiving invalid treatment, saving time and cost, and more in line with ethical requirements.
• BOP2 designs provide a Bayesian optimal platform to design phase II clinical trials with various endpoints.
• One practical obstacle to adaptive Phase II trials is the need to observe outcomes as soon as possible in order to apply decision rules to make a go/no-go decision. The common approach of suspending accruals and waiting for data on treated patients to mature is often a lengthy trail. This novel Bayes Phase II design enables clinicians to make a real-time go/no-go decision in the event of delayed results, thus significantly reducing trial time.
• The heterogeneity of each patient was considered to predict drug efficacy based on biomarkers and identify patients who responded well to drugs. The design is easy to implement, and the sample size is small in the case of invalidity, which is ethical.
• Combining the phase IIa single-arm and phase IIb double-arm trial into a protocol, this design can share information from both phases to save time and cost compared to separate phase IIa/IIb trials.

• This protocol design can improve efficiency and speed up drug development. For subtrials with multiple studies that may have different purposes, multiple investigational agents and/or multiple tumor types in adult and pediatric tumors can be evaluated simultaneously within the same general trial framework. Including basket design, umbrella design and platform design.

  Basket design, namely a kind of target clear medicine is a basket, put different diseases with the same target gene in a basket, its essence is a kind of medicine for different diseases or disease subtype, main answer whether a particular targeted drugs in all or some specific mutations in tumor patients is effective. In the basket design, because each type of tumor patients can be regarded as a cohort, researchers can not only analyze the response of a cohort of patients to drugs independently, but also evaluate the drug effect of all the cohort of patients as a large cohort. If patients in one cohort are found to respond well to the drug, other cohorts will be evaluated immediately; If a cohort does not show efficacy or evidence, studies in that cohort can be discontinued after evaluation, while studies in other cohorts continue.

  Umbrella design is aimed at the same disease caused by different gene mutations, namely: one disease, several molecular subtypes, several different therapeutic drugs. It is like holding up a big umbrella, gathering a certain disease with different driving genes under the same umbrella, completing the detection of different targets at the same time, and then distributing different precise target drugs according to different target genes. Its biggest advantage is that it brings together very rare mutation events, turning rare events into "common" events, which has special significance both for speeding up clinical trials for rare diseases and for an individual's chances of receiving precise treatments.

  Platform design is to study multiple targeted therapies in a single disease context in an uninterrupted manner, allowing therapeutic drugs to enter or leave the platform based on decision algorithms (usually Bayesian decision making), and to find the best treatment strategy for the disease through comparative studies of multiple therapeutic measures. Compared with basket or umbrella designs that focus on targeted drugs, platform trials focus more on the disease itself, not only evaluating the initial drug, but also drug combinations, quantifying the efficacy differences between different subgroups, and ensuring that enrolled patients receive the best treatment. Platform test is a dynamic design mode in terms of overall form and operation. It is more like an extension of adaptive design, allowing key factors to be modified during the test based on the information and accumulated data obtained from the previous test.

• Interim analysis was conducted on all subjects who had completed the trial at intervals or after a certain proportion of sample size was completed to determine whether the trial could be concluded in advance with valid or invalid results and to end the test early.
• Compared with the group sequential design, it has stronger flexibility. It not only allows the trial to end in advance due to validity/invalidity in the interim analysis, but also can make adaptive adjustments to the trial according to the results of the interim analysis without compromising its correctness and integrity, such as the re-estimation of sample size, adjustment of sample distribution proportion, adjustment of efficacy parameter, the elimination of inferior group and the addition of superior group.
• Bridge trials are additional trials conducted in the new geographic location to provide information on pharmacokinetics or efficacy, safety, appropriate dose, and dosing regimens relevant to the population in the new geographic location, allowing the extrapolation of clinical data from the original geographic location to the new geographic location, and such studies may include additional pharmacodynamic studies.