Improvement in latent variable indirect response modeling of multiple categorical clinical endpoints: application to modeling of guselkumab treatment effects in psoriatic patients

Employing zero-inflated beta distribution in an exposure-response analysis of TYK2/JAK1 inhibitor brepocitinib in patients with plaque psoriasis

Brepocitinib is an oral selective dual TYK2/JAK1 inhibitor and based on its cytokine inhibition profile is expected to provide therapeutic benefit in the treatment of plaque psoriasis. Efficacy data from a completed Phase 2a study in patients with moderate-to-severe plaque psoriasis were utilized to develop a population exposure-response model that can be employed to inform dose selection decisions for further clinical development. A modeling approach that employs the zero-inflated beta distribution was used to account for the bounded nature and distributional characteristics of the Psoriasis Area and Severity Index (PASI) score data. The developed exposure-response model provided an adequate description of the observed PASI scores across all the treatment arms tested and across both the induction and maintenance dosing periods of the study. In addition, the developed model exhibited a good predictive capacity with regard to the derived responder metrics (e.g., 75%/90%/100% improvement in PASI score [PASI75/90/100]). Clinical trial simulations indicated that the induction/maintenance dosing paradigm explored in this study does not offer any advantages from an efficacy perspective and that doses of 10, 30, and 60 mg once-daily may be suitable candidates for clinical evaluation in subsequent Phase 2b studies.

Latent variable indirect response modeling of clinical efficacy endpoints with combination therapy: application to guselkumab and golimumab in patients with ulcerative colitis

Accurate characterization of longitudinal exposure-response of clinical trial endpoints is important in optimizing dose and dosing regimens in drug development. Clinical endpoints are often categorical, for which much progress has been made recently in latent variable indirect response (IDR) modeling with single drugs. However, such applications have not yet been used for trials employing multiple drugs administered concurrently. This study aims to demonstrate that the latent variable IDR approach provides a convenient longitudinal exposure-response modeling framework to assess potential interaction effects of combination therapies. This is illustrated by an application to the exposure-response modeling of guselkumab, a monoclonal antibody in clinical development that blocks the interleukin-23p19 subunit, and golimumab, a monoclonal antibody that binds with high affinity to tumor necrosis factor-alpha. A Phase 2a study was conducted in 214 patients with moderate-to severe active ulcerative colitis for which longitudinal assessments of disease severity based on patient-reported measures of rectal bleeding, stool frequency, and symptomatic remission were evaluated as categorical endpoints, and fecal calprotectin as a continuous endpoint. The modeling results suggested independent pharmacodynamic guselkumab and golimumab effects on fecal calprotectin as a continuous endpoint, as well as interaction effects on the categorical endpoints that may be explained by an additional pathway of competitive interaction.

Role of Disease Progression Models in Drug Development

The use of Disease progression models (DPMs) in Drug Development has been widely adopted across therapeutic areas as a method for integrating previously obtained disease knowledge to elucidate the impact of novel therapeutics or vaccines on disease course, thus quantifying the potential clinical benefit at different stages of drug development programs. This paper provides a brief overview of DPMs and the evolution in data types, analytic methods, and applications that have occurred in their use by Quantitive Clinical Pharmacologists. It also provides examples of how these models have informed decisions and clinical trial design across several therapeutic areas and at various stages of development. It briefly describes potential new applications of DPMs utilizing emerging data sources, and utilizing new analytic techniques, and discuss new challenges faced such as requiring description of multiple endpoints, rapid model development, application of machine learning-based analytics, and use of high dimensional and real-world data. Considerations for the continued evolution future of DPMs to serve as community-maintained expert systems are also provided.

Improving categorical endpoint longitudinal exposure-response modeling through the joint modeling with a related endpoint

Exposure-response modeling is important to optimize dose and dosing regimens in clinical drug development. While primary clinical trial endpoints often have few categories and thus provide only limited information, sometimes there may be additional, more informative endpoints. Benefits of fully incorporating relevant information in longitudinal exposure-response modeling through joint modeling have recently been shown. This manuscript aims to further investigate the benefit of joint modeling of an ordered categorical primary endpoint with a related near-continuous endpoint, through the sharing of model parameters in the latent variable indirect response (IDR) modeling framework. This is illustrated by analyzing the data collected through up to 116 weeks from a phase 3b response-adaptive trial of ustekinumab in patients with psoriasis. The primary endpoint was based on the 6-point physician's global assessment (PGA) score. The Psoriasis area and severity Index (PASI) data, ranging from 0 to 72 with 0.1 increments, were also available. Separate and joint latent variable Type I IDR models of PGA and PASI scores were developed and compared. The results showed that the separate PGA model had a substantial structural bias, which was corrected by the joint modeling of PGA and PASI scores.

Impact of Pharmacokinetic and Pharmacodynamic Properties of Monoclonal Antibodies in the Management of Psoriasis

The treatment of psoriasis has been revolutionized by the emergence of biological therapies. Monoclonal antibodies (mAb) generally have complex pharmacokinetic (PK) properties with nonlinear distribution and elimination. In recent years, several population pharmacokinetic/pharmacodynamic (PK/PD) models capable of describing different types of mAb have been published. This study aims to summarize the findings of a literature search about population PK/PD modeling and therapeutic drug monitoring (TDM) of mAb in psoriasis. A total of 22 articles corresponding to population PK/PD models of tumor necrosis factor (TNF)-α inhibitors (adalimumab and golimumab), interleukin (IL)-23 inhibitors (guselkumab, tildrakizumab, and risankizumab), IL-23/IL-12 inhibitor (ustekinumab), and IL-17 inhibitors (secukinumab, ixekizumab, and brodalumab) were collected. A summary of the clinical trials conducted so far in psoriasis was included, together with the current structural population PK and PD models. The most significant and clinical covariates were body weight (BW) and the presence of immunogenicity on clearance (CL). The lack of consensus on PK/PD relationships has prevented establishing an adequate dosage and, therefore, accentuates the need for TDM in psoriasis.

Approaches to handling missing or "problematic" pharmacology data: Pharmacokinetics

Missing or erroneous information is a common problem in the analysis of pharmacokinetic (PK) data. This may present as missing or inaccurate dose level or dose time, drug concentrations below the analytical limit of quantification, missing sample times, or missing or incorrect covariate information. Several methods to handle problematic data have been evaluated, although no single, broad set of recommendations for commonly occurring errors has been published. In this tutorial, we review the existing literature and present the results of our simulation studies that evaluated common methods to handle known data errors to bridge the remaining gaps and expand on the existing knowledge. This tutorial is intended for any scientist analyzing a PK data set with missing or apparently erroneous data. The approaches described herein may also be useful for the analysis of nonclinical PK data.

Bounded Integer Modeling of Symptom Scales Specific to Lower Urinary Tract Symptoms Secondary to Benign Prostatic Hyperplasia

The International Prostate Symptom Score (IPSS), the quality of life (QoL) score, and the benign prostatic hyperplasia impact index (BII) are three different scales commonly used to assess the severity of lower urinary tract symptoms associated with benign prostatic hyperplasia (BPH-LUTS). Based on a phase II clinical trial including 403 patients with moderate to severe BPH-LUTS, the objectives of this study were to (i) develop traditional pharmacometric and bounded integer (BI) models for the IPSS, QoL score, and BII endpoints, respectively; (ii) compare the power and type I error in detecting drug effects of BI modeling with traditional methods through simulation; and (iii) obtain quantitative translation between scores on the three abovementioned scales using a BI modeling framework. All developed models described the data adequately. Pharmacometric modeling using a continuous variable (CV) approach was overall found to be the most robust in terms of type I error and power to detect a drug effect. In most cases, BI modeling showed similar performance to the CV approach, yet severely inflated type I error was generally observed when inter-individual variability (IIV) was incorporated in the BI variance function (g()). BI modeling without IIV in g() showed greater type I error control compared to the ordered categorical approach. Lastly, a multiple-scale BI model was developed and estimated the relationship between scores on the three BPH-LUTS scales with overall low uncertainty. The current study yields greater understanding of the operating characteristics of the novel BI modeling approach and highlights areas potentially requiring further improvement.

Comparison of Precision and Accuracy of Five Methods to Analyse Total Score Data

Total score (TS) data is generated from composite scales consisting of several questions/items, such as the Movement Disorder Society-Unified Parkinson's Disease Rating Scale (MDS-UPDRS). The analysis method that most fully uses the information gathered is item response theory (IRT) models, but these are complex and require item-level data which may not be available. Therefore, the TS is commonly analysed with standard continuous variable (CV) models, which do not respect the bounded nature of data. Bounded integer (BI) models do respect the data nature but are not as extensively researched. Mixed models for repeated measures (MMRM) are an alternative that requires few assumptions and handles dropout without bias. If an IRT model exists, the expected mean and standard deviation of TS can be computed through IRT-informed functions-which allows CV and BI models to estimate parameters on the IRT scale. The fit, performance on external data and parameter precision (when applicable) of CV, BI and MMRM to analyse simulated TS data from the MDS-UPDRS motor subscale are investigated in this work. All models provided accurate predictions and residuals without trends, but the fit of CV and BI models was improved by IRT-informed functions. The IRT-informed BI model had more precise parameter estimates than the IRT-informed CV model. The IRT-informed models also had the best performance on external data, while the MMRM model was worst. In conclusion, (1) IRT-informed functions improve TS analyses and (2) IRT-informed BI models had more precise IRT parameter estimates than IRT-informed CV models.

Application of Beta-Distribution and Combined Uniform and Binomial Methods in Longitudinal Modeling of Bounded Outcome Score Data

Disease status is often measured with bounded outcome scores (BOS) which takes a discrete set of values on a finite range. The distribution of such data is often skewed, rendering the standard analysis methods assuming normal distribution inappropriate. Among the methods used for BOS analyses, two of them have the ability to predict the data within its natural range and accommodate data skewness: (1) a recently proposed beta-distribution based approach and (2) a mixture model known as CUB (combined uniform and binomial). This manuscript compares the two approaches, using an established mechanism-based longitudinal exposure-response model to analyze data from a phase 2 clinical trial in psoriatic patients. The beta-distribution-based approach was confirmed to perform well, and CUB also showed potential. A separate issue of modeling clinical trial data is that the collected baseline disease score range may be more limited than that of post-treatment disease score due to clinical trial inclusion criteria, a fact that is typically ignored in longitudinal modeling. The effect of baseline disease status restriction should in principle be adjusted for in longitudinal modeling.

Facilitating Longitudinal Exposure-Response Modeling of a Composite Endpoint Using the Joint Modeling of Sparsely and Frequently Collected Subcomponents

Longitudinal exposure-response modeling plays an important role in optimizing dose and dosing regimens in clinical drug development. Certain clinical trials contain induction and maintenance phases where the maintenance treatment depends on the subjects' achieving the main endpoint outcome in the induction phase. Due to logistic difficulties and cost considerations, the main endpoint is usually collected more sparsely than a subcomponent (or other related endpoints). The sparse collection of the main endpoint hampers its longitudinal modeling. In principle, the frequent collection of a subcomponent allows its longitudinal modeling. However, the model evaluation via the visual predictive check (VPC) in the maintenance phase is difficult due to the requirement of the main-endpoint model to identify the treatment subgroups. This manuscript proposes a solution to this dilemma via the joint modeling of the main endpoint and the subcomponent. The challenges are illustrated by analyzing the data collected up to 60 weeks from a phase III trial of ustekinumab in patients with moderate-to-severe ulcerative colitis (UC). The main endpoint Mayo score, a commonly used composite endpoint to measure the severity of UC, was collected only at baseline, the end of the induction phase, and the end of the maintenance phase. The partial Mayo score, which is a major subset of the Mayo score, was collected at nearly every 4 weeks. A longitudinal joint exposure-response model, developed under a latent-variable Indirect Response modeling framework, described the Mayo score time course and facilitated the VPC model evaluation under a response-adaptive trial design.

Analysis of Longitudinal-Ordered Categorical Data for Muscle Spasm Adverse Event of Vismodegib: Comparison Between Different Pharmacometric Models

Longitudinal‐ordered categorical data, common in clinical trials, can be effectively analyzed with nonlinear mixed effect models. In this article, we systematically evaluated the performance of three different models in longitudinal muscle spasm adverse event (AE) data obtained from a clinical trial for vismodegib: a proportional odds (PO) model, a discrete‐time Markov model, and a continuous‐time Markov model. All models developed based on weekly spaced data can reasonably capture the proportion of AE grade over time; however, the PO model overpredicted the transition frequency between grades and the cumulative probability of AEs. The influence of data frequency (daily, weekly, or unevenly spaced) was also investigated. The PO model performance reduced with increased data frequency, and the discrete‐time Markov model failed to describe unevenly spaced data, but the continuous‐time Markov model performed consistently well. Clinical trial simulations were conducted to illustrate the muscle spasm resolution time profile during the 8‐week dose interruption period after 12 weeks of continuous treatment.

On the Comparison of Methods in Analyzing Bounded Outcome Score Data

Clinical trial endpoints often take the form of bounded outcome scores (BOS) which report a discrete set of values on a finite range. Conceptually such endpoints are ordered categorical in nature, but in practice they are often analyzed as continuous variables, which may result in data range violations and difficulties to handle data skewness. Analysis methods dedicated for BOS data have been proposed; however, much confusion exists among pharmacometricians on how to compare the possible methods. This commentary reviews the main methods used in pharmacometrics applications and discusses their theoretical and practical comparisons. The expected performance of some conceptually appealing methods in different situations is discussed, and a guideline is provided on selecting analysis methods in practice.

Modeling near-continuous clinical endpoint as categorical: application to longitudinal exposure-response modeling of Mayo scores for golimumab in patients with ulcerative colitis

Accurate characterization of exposure-response relationship of clinical endpoints is important in drug development to identify optimal dose regimens. Endpoints with ≥ 10 ordered categories are typically analyzed as continuous. This manuscript aims to show circumstances where it is advantageous to analyze such data as ordered categorical. The results of continuous and categorical analyses are compared in a latent-variable based Indirect Response modeling framework for the longitudinal modeling of Mayo scores, ranging from 0 to 12, which is commonly used as a composite endpoint to measure the severity of ulcerative colitis (UC). Exposure response modeling of Mayo scores is complicated by the fact that studies typically include induction and maintenance phases with re-randomizations and other response-driven dose adjustments. The challenges are illustrated in this work by analyzing data collected from 3 phase II/III trials of golimumab in patients with moderate-to-severe UC. Visual predictive check was used for model evaluations. The ordered categorical approach is shown to be accurate and robust compared to the continuous approach. In addition, a disease progression model with an empirical bi-phasic rate of onset was found to be superior to the commonly used placebo model with one onset rate. An application of this modeling approach in guiding potential dose-adjustment was illustrated.