{"title":"Estimation Following Surrogate Based Selection in Multi-Arm Drop-The-Losers Designs.","authors":"Mojammel Haque Sarkar, Mohd Arshad, Masihuddin","doi":"10.1002/pst.70140","DOIUrl":"https://doi.org/10.1002/pst.70140","url":null,"abstract":"<p><p>This paper studies inference for primary endpoints following surrogate-based selection in a multi-arm, two-stage drop-the-losers design, assuming bivariate normal responses with a common but unknown covariance matrix. In this framework, treatment selection at the first stage is based on the largest sample mean of the surrogate endpoint (the first component), after which additional observations are collected from the selected treatment arm to estimate the primary endpoint (the second component of the mean vector). We consider several estimators for the primary endpoint, including the natural (maximum likelihood) estimator, two plug-in estimators that account for covariance estimation, and the uniformly minimum variance conditionally unbiased estimator (UMVCUE) derived using Rao-Blackwellization. The natural estimator is shown to be minimax under the scaled squared error loss function, while the UMVCUE provides unbiased inference with only a modest loss in efficiency. Simulation studies demonstrate that for a fixed total sample size, the bias of the natural estimator increases with stronger correlation and larger first-stage sample sizes, whereas the UMVCUE exhibits stable performance across a wide range of configurations. An illustrative real-data example is also presented, highlighting the practical relevance of the proposed methodology for adaptive and sequential clinical trial designs.</p>","PeriodicalId":19934,"journal":{"name":"Pharmaceutical Statistics","volume":"25 6","pages":""},"PeriodicalIF":1.5,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149299779","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Estimands for Randomized Discontinuation Designs in Oncology.","authors":"Ayon Mukherjee, Oleksandr Sverdlov, Yu Deng","doi":"10.1002/pst.70141","DOIUrl":"https://doi.org/10.1002/pst.70141","url":null,"abstract":"<p><p>The Randomized Discontinuation Design (RDD) is a two-stage adaptive enrichment design that is uniquely suited to answer causal questions about the benefit of continuing treatment in patients who have already demonstrated response to an induction therapy. The estimand framework from the International Council for Harmonisation (ICH) E9(R1) addendum provides the necessary language to precisely define this scientific question. A major challenge in formulating estimands for RDD oncology trials is that the target population for the primary scientific question is itself defined by a post-treatment biological outcome in Stage 1, which makes it difficult to separate the process of selecting eligible participants from the causal inference problem in Stage 2. In this paper, we address this challenge using the Phase III JAVELIN Gastric 100 trial as an example. We argue that the \"conventional\" estimand can be formulated as an estimand targeting the response-defined subgroup of participants with evidence of response to the induction therapy in Stage 1, and we clarify the precise sense in which this subgroup is, and is not, related to the concept of a principal stratum. We also formulate a \"full-path\" estimand-capturing the patients' entire two-stage treatment journey-and connect it to the dynamic treatment regime literature. We also provide Directed Acyclic Graphs (DAGs) for the two estimands to clarify their respective causal structures. This paper aims to strengthen the dialogue among clinical research stakeholders tasked with the design and analysis of RDD trials and support more efficient regulatory decision-making. Trial Registration: ClinicalTrials.gov identifier: NCT02625610.</p>","PeriodicalId":19934,"journal":{"name":"Pharmaceutical Statistics","volume":"25 6","pages":""},"PeriodicalIF":1.5,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149323912","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Stefan Englert, Zoe Craig, Nicolas Sauvageot, Valentine Jehl, Anirban Mitra, Jane Gilbert, Lilla Di Scala, Leen Slaets
{"title":"Improving Safety Reporting: Putting the Question First and Choosing the Right Estimator for the Incidence of Treatment-Emergent Adverse Events.","authors":"Stefan Englert, Zoe Craig, Nicolas Sauvageot, Valentine Jehl, Anirban Mitra, Jane Gilbert, Lilla Di Scala, Leen Slaets","doi":"10.1002/pst.70146","DOIUrl":"https://doi.org/10.1002/pst.70146","url":null,"abstract":"<p><p>The assessment of safety is critical throughout the development of drugs, across all therapeutical areas and indications. For marketing authorization, the common practice is to summarize adverse events (AEs) using incidence proportion (IP): the proportion of participants who experience the AE at least once while being exposed to the treatment of interest. However, IP can be biased, particularly when follow-up is limited and data are immature or when comparing treatment groups with different efficacy profiles. We introduce two distinct research questions regarding AE incidence. The first, the \"while-on-treatment\" approach quantifies AE incidence observed in routine clinical practice where intercurrent events (IE) such as death or treatment discontinuation may occur. The second, the \"hypothetical\" approach quantifies AE incidence under a counterfactual scenario where all participants remain exposed to treatment. These research questions differ in how intercurrent events are handled and therefore require different analytic approaches. Using a simulation study that mimics a typical early-development oncology setting, we compared four estimators under heterogeneous follow-up, varying frequencies and timing of intercurrent events, and both constant and time-varying AE hazards: IP, Aalen-Johansen (AJ), one minus Kaplan-Meier (1-KM), and probability transformed incidence density (PTID). Our results indicate that AJ performs best for the while-on-treatment research question. For a hypothetical research question, PTID is recommended due to its higher precision but can be biased when hazards vary over time. The 1-KM estimator is unbiased for the hypothetical research question but tends to be less precise in small samples. Putting the research question first and selecting the adequate estimator is essential for valid AE reporting to improve interpretability and to reduce bias when characterizing AE incidence.</p>","PeriodicalId":19934,"journal":{"name":"Pharmaceutical Statistics","volume":"25 6","pages":""},"PeriodicalIF":1.5,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149409452","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Yan Liu, Seung Won Hyun, Chithra Sangli, Kyle Beauchamp, Farahnaz Islam, Dongho Shin
{"title":"An Inverse-Selection-Probability Weighted Approach for Real-World Progression-Free Survival With Left-Truncated Clinico-Genomic Data.","authors":"Yan Liu, Seung Won Hyun, Chithra Sangli, Kyle Beauchamp, Farahnaz Islam, Dongho Shin","doi":"10.1002/pst.70145","DOIUrl":"10.1002/pst.70145","url":null,"abstract":"<p><p>Next-generation sequencing (NGS) has become a vital tool in precision medicine, revolutionizing cancer diagnosis and treatment. However, in real-world studies where NGS defines study entry, data are often left-truncated because patients who do not survive until sequencing are excluded. Consequently, standard survival analysis techniques can result in biased, overestimated survival probabilities. While the commonly used risk set adjustment method addresses this bias, it excludes patients experiencing progression prior to sequencing, reducing statistical efficiency and power. We introduce a novel weighted approach utilizing the inverse of sample selection probability to include all observed patients. We also developed a computationally efficient, user-friendly method for calculating weights and estimating variance. The performance of the proposed method is evaluated through comprehensive simulation studies and further demonstrated using real-world data of metastatic breast cancer patients. The proposed weighted approach, utilizing inverse sample selection probabilities, ensures accurate parameter estimates and reliable inference.</p>","PeriodicalId":19934,"journal":{"name":"Pharmaceutical Statistics","volume":"25 6","pages":""},"PeriodicalIF":1.5,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13716531/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149288116","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Inverse Probability of Censoring Weights for Single and Competing Events Survival Analyses With Dependent Censoring.","authors":"Emily C Zabor, Anne A Eaton","doi":"10.1002/pst.70138","DOIUrl":"10.1002/pst.70138","url":null,"abstract":"<p><p>Survival analysis is a common statistical method in biomedical research that can be applied to censored time-to-event endpoints such as overall survival or progression-free survival. Standard survival analysis methods to analyze censored time-to-event data, in the context of either a single event or of competing events, rely on the assumption that censoring is independent of the event of interest. Dependent censoring occurs when participants who are censored at any point in time are more or less likely to have an event than the participants who remain in the study. Many clinical trials and real-world data sources may be subject to dependent censoring, which requires specialized statistical methods. This review paper demonstrates how to estimate inverse probability of censoring weights (IPCW) in the context of both single event and competing events survival analyses. The examples show how to account for dependent censoring using IPCW Kaplan-Meier curves for single events and using the IPCW Aalen-Johansen estimator to estimate cumulative incidence for competing events. The examples further demonstrate how to incorporate IPCW into a Cox regression model and a Fine and Gray regression model. Simulation studies show that naive methods can lead to substantial bias when estimating the survival curve, the hazard ratio in a Cox model, and the cumulative incidence. However, failing to account for censoring that depends on a variable that is included in the regression model led to minimal bias in Fine and Gray regression estimators.</p>","PeriodicalId":19934,"journal":{"name":"Pharmaceutical Statistics","volume":"25 6","pages":""},"PeriodicalIF":1.5,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13748472/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149380038","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Telba Irony, Tobias Mielke, Fredrik Ohrn, Vladimir Dragalin
{"title":"Advancing Drug Development Through Bayesian Approaches: Insights Drawn From Practical Experience in Pharmaceutical R&D.","authors":"Telba Irony, Tobias Mielke, Fredrik Ohrn, Vladimir Dragalin","doi":"10.1002/pst.70147","DOIUrl":"https://doi.org/10.1002/pst.70147","url":null,"abstract":"<p><p>Bayesian approaches are increasingly used in drug development to optimize clinical trial design and analysis by formally integrating prior knowledge, quantifying uncertainty, and supporting decision making throughout the development lifecycle. Nevertheless, broader implementation in confirmatory settings remains constrained by perceived technical complexity and uncertainty regarding regulatory acceptance. This article presents practical applications drawn from our experience, with a focus on how Bayesian thinking can improve drug development. We start with Bayesian methods for borrowing information in clinical trials and explore applications such as external control augmentation and treatment effect extrapolation. We discuss key challenges, including demonstration of exchangeability, over-borrowing, bias, and potential type I error inflation, and outline mitigation strategies such as dynamic borrowing and sensitivity analyses. Regulatory implications of these approaches are illustrated through two pediatric case studies, one unsuccessful and one accepted, highlighting features that influence regulatory acceptability. We then present examples showing how Bayesian models can enhance analytical decision-making during trial conduct even when the primary analyses rely on frequentist methods. Next, we introduce Assurance and Conditional Assurance and demonstrate how these Bayesian metrics can be used to optimize development strategies, including phase progression decisions, number of registrational trials, futility rules, enrichment strategies, and alpha spending. Finally, we illustrate how Bayesian models can support trial operations by forecasting and monitoring the probability of operational success. Overall, the article shows how Bayesian approaches, even when used alongside frequentist methods, can strengthen trial design and analysis, improve efficiency, and enhance transparency across drug development.</p>","PeriodicalId":19934,"journal":{"name":"Pharmaceutical Statistics","volume":"25 6","pages":""},"PeriodicalIF":1.5,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149394994","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Bayesian Approaches in Drug Development: Interpreting the 2026 US Food and Drug Administration Draft Guidance for Optimal Dose Finding.","authors":"Ayon Mukherjee","doi":"10.1002/pst.70149","DOIUrl":"https://doi.org/10.1002/pst.70149","url":null,"abstract":"<p><p>The US Food and Drug Administration (FDA) released a landmark draft guidance in January 2026 entitled 'Use of Bayesian Methodology in Clinical Trials of Drug and Biological Products' (Food and Drug Administration 2026)-the first comprehensive regulatory framework for Bayesian primary inference in drug and biologic trials. This Viewpoint critically appraises the Guidance across four dimensions: (1) The regulatory and philosophical tensions in Section IV.A.1, 'Calibration to Type-I Error Rate', with specific improvement recommendations; (2) the Bayesian borrowing framework articulated in Sections III and V, including gaps in dynamic borrowing and effective sample size (ESS) interpretation; (3) the Guidance's synergy with FDA's Project Optimus and the companion dose-optimisation guidance (August 2024) (Food and Drug Administration 2024), and how Bayesian methods underpin the optimal biological dose (OBD) paradigm; and (4) the limited operational detail in Section III.F for efficacy-toxicity joint models and seamless Phase I/II designs-the design class most capable of OBD identification. It is contended that the Guidance is a historic advance, yet its oncology dose-finding section devotes most of its operational detail to maximum tolerated dose (MTD)-oriented designs. A regulatory taxonomy of Bayesian dose-finding designs and targeted improvements for the final guidance are proposed. The appraisal refers to the nonbinding January 2026 draft text, which may change before finalisation.</p>","PeriodicalId":19934,"journal":{"name":"Pharmaceutical Statistics","volume":"25 6","pages":""},"PeriodicalIF":1.5,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13760598/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149409451","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Marie-Karelle Riviere, Anaïs Andrillon, Pavel Mozgunov, Emma Gerard, Moreno Ursino
{"title":"Evaluating Early-Stage Oncology Clinical Trial Designs in the Era of Project Optimus: A Scoping Review.","authors":"Marie-Karelle Riviere, Anaïs Andrillon, Pavel Mozgunov, Emma Gerard, Moreno Ursino","doi":"10.1002/pst.70131","DOIUrl":"10.1002/pst.70131","url":null,"abstract":"<p><p>In 2021, the FDA initiated Project Optimus to reform dose optimization and selection in early-phase oncology, highlighting that doses and schedules of new therapies are often inadequately characterized and that traditional dose-finding approaches should be revised. The project also encourages early and comprehensive exploration of multiple doses during clinical development and advocates for a more patient-centered approach incorporating not only toxicity but also the individual's response, biomarkers, pharmacokinetics/pharmacodynamics, or patient-reported outcomes to inform dosage decisions. We conducted a scoping review of the statistical and methodological literature on early-phase oncology trial designs in light of Project Optimus. The review aims to determine which key elements of the Optimus guidelines are incorporated into existing early-phase trial design methodologies and to assess remaining methodological gaps. It also aims to address the following question: What progress has been achieved so far, and what improvements can be made in light of the directives from Project Optimus? The review was organized according to the PRISMA-ScR guidelines. A total of 160 journals were finally retained, with 722 papers considered for the scoping review, published between January 1990 and September 2025. The database containing the reviewed papers is made available to the reader.</p>","PeriodicalId":19934,"journal":{"name":"Pharmaceutical Statistics","volume":"25 6","pages":""},"PeriodicalIF":1.5,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13740141/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149343238","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"A New Optimal Cut-Points Selection Method for Diseases With Multiple Ordinal Stages and Its Confidence Interval Estimation.","authors":"Nan An, Lili Tian","doi":"10.1002/pst.70148","DOIUrl":"https://doi.org/10.1002/pst.70148","url":null,"abstract":"<p><p>Optimal cut-points selection plays a fundamental role in multi-stage classification. The popular Youden index method primarily emphasizes overall diagnostic accuracy while overlooking imbalance among classification rates across disease stages, whereas other methods improve balance at the cost of substantially reduced overall accuracy. Such trade-offs are undesirable in clinical decision-making. This article aims to develop a flexible optimal cut-points selection method that simultaneously preserves overall diagnostic accuracy and balance among stage-specific classification rates. By introducing a tuning parameter, the proposed method integrates the arithmetic mean and the geometric mean of stage-specific classification rates into a unified objective function. Estimation procedures for confidence intervals are developed under both parametric and nonparametric settings. Simulation studies are conducted to evaluate the performance of the proposed cut-points selection method and the confidence interval estimation approaches. The proposed method is applied to a real pancreatic cancer dataset to illustrate its practical utility.</p>","PeriodicalId":19934,"journal":{"name":"Pharmaceutical Statistics","volume":"25 6","pages":""},"PeriodicalIF":1.5,"publicationDate":"2026-11-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"149394993","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"Use of Composite Priors for Assurance of Sequential Therapy Trials: Practical Experiences Using Binary Endpoint Data.","authors":"Jonathan Wright, Li Ye","doi":"10.1002/pst.70125","DOIUrl":"10.1002/pst.70125","url":null,"abstract":"<p><p>Accurately calculating assurance can play an important role in quantifying the confidence in the success of a clinical trial. Formulation of prior distributions and the use of assurance can therefore play a vital part in drug development planning and decision-making. Although approaches in developing priors and calculating probability of success (assurance) have been readily reported, the application for studies incorporating sequential treatment therapy-where multiple medications or therapeutic regimens are administered consecutively to maximise therapeutic efficacy and minimise adverse effects-is less well defined. Challenges relating to eliciting a prior for a single treatment effect across a sequential therapy include reliance on individual experts' subjective beliefs about translating multiple mechanisms of actions into a single overall treatment effect and the inability to capture nuances across and inconsistency in combining individual treatment component effects. In this paper, we outline a comprehensive quantitative framework for developing a composite prior distribution (from individual component priors) and calculating the assurance for study designs involving sequential treatment. We demonstrate how this framework, which employs both prior elicitation and data-driven prior development methods, was applied in two phase 2 study designs with sequential treatment regimens using a binary endpoint. We also discuss the optimal conditions for these methods and compare the benefits and challenges with a conventional holistic prior elicitation and assurance framework.</p>","PeriodicalId":19934,"journal":{"name":"Pharmaceutical Statistics","volume":"25 5","pages":"e70125"},"PeriodicalIF":1.5,"publicationDate":"2026-09-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC13571898/pdf/","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"148963351","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}