Journal of Biopharmaceutical Statistics最新文献

Traditional phase I dose finding cancer clinical trial designs aim to determine the maximum tolerated dose (MTD) of the investigational cytotoxic agent based on a single toxicity outcome, assuming a monotone dose-response relationship. However, this assumption might not always hold for newly emerging therapies such as immuno-oncology therapies and molecularly targeted therapies, making conventional dose finding trial designs based on toxicity no longer appropriate. To tackle this issue, numerous early-phase dose finding clinical trial designs have been developed to identify the optimal biological dose (OBD), which takes both toxicity and efficacy outcomes into account. In this article, we review the current model-assisted dose finding designs, BOIN-ET, BOIN12, UBI, TEPI-2, PRINTE, STEIN, and uTPI to identify the OBD and compare their operating characteristics. Extensive simulation studies and a case study using a CAR T-cell therapy phase I trial have been conducted to compare the performance of the aforementioned designs under different possible dose-response relationship scenarios. The simulation results demonstrate that the performance of different designs varies depending on the particular dose-response relationship and the specific metric considered. Based on our simulation results and practical considerations, STEIN, PRINTE, and BOIN12 outperform the other designs from different perspectives.

Randomized controlled trials (RCTs) are the gold standard for clinical research but may not accurately reflect the impact of medicines in real-world settings. Supplementing RCTs with insights from real-world data (RWD) can address known limitations by including more diverse patient populations, additional types of sites-of-care, and practices more representative of the care most people receive. One current challenge in using RWD is the lack of an algorithmic approach to identifying outcomes. To address this, machine learning models for identifying a frequently used outcome, Major Adverse Cardiovascular Events (MACE), were developed in Clinical Trial Data (CTD). Anonymized CTD sourced from the Medidata Enterprise Data Store were used to develop model features on the condition that they would be useful for labelling MACE events and that they could also be found in RWD. These features were used to train three random forest models to identify each component of 3-point MACE in a patient's clinical trial journey. Performance metrics for the models are presented (recall = 0.72 [0.07], precision = 0.68 [0.12] - mean, [SD]) along with the top contributing features. We show that the models can be tuned specifically to replicate the adjudication panels' results and present a cost-benefit analysis for deploying such models in clinical trial settings. We demonstrate the viability of using advanced algorithms for identifying clinical outcomes in prospective clinical trials. Deployment of such models could reduce the resources required to conduct RCTs. Extending such models to RWD would facilitate approval of pragmatic clinical trials for regulatory submissions.

In this paper, we propose a novel method for the analysis of cure rate data with competing risks using defective distributions. We develop two defective regression models for the analysis of competing risk data subjected to random right censoring. The proposed models enable us to estimate the cure fraction directly from the model. Simultaneously, we also estimate the regression parameters corresponding to each cause of failure using the method of maximum likelihood. We conduct a simulation study to evaluate the finite sample performance of the proposed estimators. The practical usefulness of the procedures is illustrated using two real-life data sets.

Dose-response relationships are important in assessing the efficacy and potency of compounds, which can usually be characterized by a 4-parameter logistic (4-PL) model estimating EC50, slope factor, lower asymptote, and upper asymptote. EC50, the concentration of a compound that induces a response halfway between the baseline and maximum, is a key quantity to evaluate compound potency. For multi-donor dose-response data, it is often of interest to estimate the overall EC50 (i.e. the average EC50 of the population of donors) and its 95% confidence interval (CI). A few multi-donor EC50 estimation methods have been proposed in the literature. Jiang and Kopp-Schneider (2014) systematically compared the meta-analysis approach and the nonlinear mixed-effects approach and concluded that the meta-analysis approach is simple and robust to summarize EC50 estimates from multiple experiments, especially suited in the case of a small number of experiments, while the nonlinear mixed-effects approach has the issue of convergence failures probably due to overparameterization. In this article, we propose a modification of the nonlinear mixed-effects approach by using the stochastic approximation expectation-maximization (SAEM) algorithm to estimate model parameters and using multiple starting points to search for globally optimal values, which can substantially alleviate the issue of convergence failures even for small number of donors (e.g. n = 3), and achieve a smaller absolute median bias and better coverage probability of 95% confidence interval than the meta-analysis approach when the number of donors is not too small (e.g. n ≥ 7).

Taking into account the local dependence structure in large-scale multiple testing is expected to improve both the efficiency of the testing procedure and the interpretability of scientific findings. The hidden Markov model (HMM), as an effective model to describe the sequential dependence, has been successfully applied to large-scale multiple testing with local correlations. However, in many applications, the first-order Markov chain is not flexible enough to capture the complexity of local correlations. To address this issue, this paper proposes a novel multiple testing procedure that uses a higher-order Markov chain to better characterize local correlations among tests. The proposed procedure is validated by theoretical results and simulation studies, which show that it outperforms its competitors in terms of power. Finally, a real data analysis is presented to demonstrate the favorable performance of the proposed procedure.

In the realm of medical diagnostic testing, diagnostic tests can assume either binary forms, distinguishing between diseased and non-diseased states, or ordinal forms, categorizing states from non-diseased to various stages (1 to K). Another significant classification scheme for multi-class scenarios is tree or umbrella ordering, which entails several unordered sub-classes (subtypes) within a biomarker. Within tree or umbrella ordering, the classifier assesses whether the marker measurement for one class surpasses or falls below those for the other classes. Although Receiver Operating Characteristic (ROC) curves and summary measures have been adapted to accommodate tree and umbrella ordering, these approaches often yield cut-off points that generate highly sensitive tests for certain disease subtypes while compromising specificity for others. This may not be ideal for all diseases. Hence, in this investigation, we explore diverse measures of diagnostic test accuracy and optimal cut-off point selection procedures under tree or umbrella ordering to foster more specific tests. We present numerical examples and simulation studies and demonstrate the approach using real data on lung cancer.

The mixed model for repeated measures (MMRM) analysis is sometimes used as a primary statistical analysis for a longitudinal randomized clinical trial. When the MMRM analysis is implemented in ordinary statistical software, the standard error of the treatment effect is estimated by assuming orthogonality between the fixed effects and covariance parameters, based on the characteristics of the normal distribution. However, orthogonality does not hold unless the normality assumption of the error distribution holds, and/or the missing data are derived from the missing completely at random structure. Therefore, assuming orthogonality in the MMRM analysis is not preferable. However, without the assumption of orthogonality, the small-sample bias in the standard error of the treatment effect is significant. Nonetheless, there is no method to improve small-sample performance. Furthermore, there is no software that can easily implement inferences on treatment effects without assuming orthogonality. Hence, we propose two small-sample adjustment methods inflating standard errors that are reasonable in ideal situations and achieve empirical conservatism even in general situations. We also provide an R package to implement these inference processes. The simulation results show that one of the proposed small-sample adjustment methods performs particularly well in terms of underestimation bias of standard errors; consequently, the proposed method is recommended. When using the MMRM analysis, our proposed method is recommended if the sample size is not large and between-group heteroscedasticity is expected.

Conducting randomized controlled trials for medications targeting rare diseases presents significant challenges, due to the scarcity of participants and ethical considerations. Under such circumstances, leveraging real-world data (RWD) to generate supporting evidence may be accepted by the regulatory agency. Constructing an external control arm (ECA) from RWD for a single-arm trial has been conducted occasionally. A complication in this design is that patients from RWD may be eligible at multiple time points. Most studies approach this by selecting one time point as the index date for ECA patients. Here, we propose a novel design for externally controlled trials that permits the inclusion of ECA patients at various entry points. Accompanying this design, we make recommendations for statistical methods to account for measured confounders, limited sample size, within-subject correlation, and potential overdispersion inherent in count data. Furthermore, we present an idea for the blinding process for this type of study. We have conducted a series of simulations to assess the performance of the design and statistical methods in terms of bias, type I error, and efficiency, as compared to the approach of selecting only one entry per ECA patient. The study and parameter setup were based on a hypothetical case inspired by a rare disease study. The results indicate that allowing multiple entries for ECA patients can lead to enhanced performance in many aspects. It provides a controlled type I error, robustness against certain model misspecifications, and a moderate power improvement compared with selecting a single entry per ECA patient.

Dong et al. (2023) showed that the win statistics (win ratio, win odds, and net benefit) can complement each another to demonstrate the strength of treatment effects in randomized trials with prioritized multiple outcomes. This result was built on the connections among the point and variance estimates of the three statistics, and the approximate equality of Z-values in their statistical tests. However, the impact of this approximation was not clear. This Discussion refines this approach and shows that the approximate equality of Z-values for the win statistics holds more generally. Thus, the three win statistics consistently yield closely similar p-values. In addition, our simulations show an example that the naive approach without adjustment for censoring bias may produce a completely opposite conclusion from the true results, whereas the IPCW (inverse-probability-of-censoring weighting) approach can effectively adjust the win statistics to the corresponding true values (i.e. IPCW-adjusted win statistics are unbiased estimators of treatment effect).