General Information
Conflict of Interest
Request Clinical Trials
Associated Trial(s): What type of data are you looking for?: Individual Participant-Level Data, which includes Full CSR and all supporting documentationRequest Clinical Trials
Data Request Status
Status: OngoingResearch Proposal
Project Title: Population Exposure--Patient-Reported Outcome (PRO) Modeling to Evaluate PRO-Informed Dose Selection in Early-Phase Clinical Trials
Scientific Abstract:
Background: Early-phase dose selection is typically guided by safety, pharmacokinetics (PK), and clinician-reported toxicity. Patient-reported outcomes (PROs) may provide additional quantitative insight into tolerability and efficacy. Establishing exposure--PRO relationships could support more patient-centered dose optimization strategies.
Objective: To evaluate whether population exposure--response modeling can characterize relationships between drug exposure and PRO measures and assess the feasibility of PRO-informed dose selection in early-phase trials.
Study Design: Secondary analysis of de-identified participant-level data from NCT01381874 using population PK and exposure--response modeling.
Participants: Adults enrolled in NCT01381874 with available PK data and PRO assessments (EORTC QLQ-C30 and EQ-5D-5L).
Primary and Secondary Outcome Measure(s): The primary outcome is the quantitative relationship between systemic drug exposure and longitudinal PRO toxicity-related items. Secondary outcomes include exposure--response relationships for PRO efficacy-related items and simulation-based evaluation of alternative dosing regimens to assess potential benefit--risk improvements.
Statistical Analysis: Exposure--PRO relationships will be developed using population modeling. Model performance will be evaluated using standard diagnostics and internal validation. Simulations will assess the impact of alternative dosing regimens on predicted PRO toxicity and efficacy.
Brief Project Background and Statement of Project Significance:
Dose selection in early-phase clinical trials is traditionally guided by safety, pharmacokinetics (PK), and clinician-reported adverse events. [1] The maximum tolerated dose paradigm remains central in oncology and many other therapeutic areas. [2] However, clinician-reported toxicity may underestimate the frequency and severity of symptomatic adverse events experienced by patients. Patient-reported outcomes (PROs), including validated instruments such as EORTC QLQ-C30 and EQ-5D-5L, capture patients' direct experience of symptoms, functioning, and quality of life. Despite increasing regulatory and scientific interest in PROs, they are rarely incorporated quantitatively into dose-finding decisions in early-phase development. [3,4]
To our knowledge, no prior study has systematically evaluated the feasibility of using quantitative exposure--PRO relationships to inform dose selection in early-phase trials. [1] A key reason is the analytical complexity of PRO data. PRO measurements are inherently noisy, highly variable, longitudinal, and often subject to substantial missingness. Traditional exposure--response methods are not well suited to handle these features, and therefore PRO data have typically been analyzed descriptively rather than integrated into quantitative dose-optimization frameworks. Our lab has developed a population modeling approach specifically designed to characterize longitudinal PRO trajectories while accounting for variability and missing data. Using this framework, we have successfully demonstrated that PRO data can predict important clinical outcomes, including overall survival, supporting their validity as quantitative endpoints. [5-7]
In this project, we will conduct a secondary analysis of participant-level data from NCT01381874 to develop population PK models and characterize exposure--PRO relationships for both toxicity-related and efficacy-related PRO domains. By linking drug exposure over time with longitudinal PRO data, we will assess whether robust and predictive exposure--PRO relationships can be established. If feasible, we will simulate alternative dosing regimens to evaluate whether PRO-informed dose optimization could improve the balance between symptom burden and therapeutic benefit.
This will be the first study to formally evaluate the feasibility of PRO-driven dose selection in early-phase clinical trials using quantitative population modeling. The knowledge generated will advance methodological understanding of how PRO data can be integrated into exposure--response frameworks and early drug development. If successful, this work could inform more patient-centered dose optimization strategies, improve benefit--risk assessment, and contribute to the modernization of early-phase trial design. The modeling framework developed here will be generalizable to other therapeutic areas and clinical development programs, thereby enhancing scientific and public health decision-making.
Specific Aims of the Project:
The overall objective of this project is to evaluate the feasibility of incorporating patient-reported outcomes (PROs) into dose selection decisions in early-phase clinical trials using quantitative exposure--response modeling.
Aim 1: Develop a population pharmacokinetics model to characterize individual drug exposure profiles and establish quantitative exposure--PRO relationships for toxicity-related PRO domains.
Hypothesis: Systemic drug exposure is significantly associated with longitudinal changes in PRO toxicity-related items.
Aim 2: Characterize exposure--PRO relationships for efficacy-related PRO domains and assess their predictive performance.
Hypothesis: Drug exposure is associated with improvements in PRO efficacy-related measures.
Aim 3: Conduct simulation analyses to evaluate whether alternative dosing regimens informed by exposure--PRO relationships could optimize the balance between symptom burden and therapeutic benefit.
Hypothesis: PRO-informed dosing strategies may identify dose regimens that improve patient-reported tolerability without compromising efficacy.
This study will determine whether robust exposure--PRO relationships can support PRO-driven dose selection in early-phase development.
Study Design: Individual trial analysis
What is the purpose of the analysis being proposed? Please select all that apply.: Develop or refine statistical methods Research on clinical trial methods
Software Used: R, RStudio
Data Source and Inclusion/Exclusion Criteria to be used to define the patient sample for your study:
Data Source: De-identified individual participant-level data from NCT01381874 obtained through the YODA Project secure analysis environment, including pharmacokinetic concentration data (PC), dosing/exposure information (e.g., EX), demographics/baseline characteristics (DM/Baseline), safety/labs as available, and patient-reported outcomes questionnaires (QS) including EORTC QLQ-C30 and EQ-5D-5L. Supporting documents (CSR/protocol) will be used to verify visit schedules, dose levels, and endpoint definitions. No external IPD will be pooled for the primary analyses.
Inclusion Criteria (analysis sample): Participants randomized/treated in NCT01381874 who received at least one dose of study drug and have sufficient data to contribute to modeling, defined as having (1) dosing information and at least one post-dose PK concentration measurement for population PK estimation, and (2) at least one post-baseline PRO assessment (EORTC QLQ-C30 and/or EQ-5D-5L) for longitudinal exposure--PRO analyses.
Exclusion Criteria: Participants with no dosing records, no post-dose PK concentrations, or no post-baseline PRO data will be excluded from the corresponding analyses. For PK modeling, participants with implausible or non-interpretable PK records that cannot be reconciled with dosing/visit timing based on protocol/CSR (e.g., missing sample times) will be excluded from PK estimation but may remain eligible for PRO-only descriptive summaries. For exposure--PRO analyses, PRO records collected after documented discontinuation due to non-treatment-related reasons or after major protocol deviations affecting PRO interpretability (as defined in the CSR) will be excluded in sensitivity analyses. No additional demographic restrictions (age/sex/race) will be applied beyond the trial's original eligibility criteria.
Primary and Secondary Outcome Measure(s) and how they will be categorized/defined for your study:
The primary outcome of this study is the quantitative exposure--PRO relationship between systemic drug exposure and longitudinal PRO toxicity-related domains. Drug exposure will be defined using individual-level exposure metrics derived from population pharmacokinetic modeling (e.g., AUC, Cmax, or time-varying predicted plasma concentrations). PRO toxicity-related outcomes will be defined using selected symptom scales/items from EORTC QLQ-C30 (e.g., fatigue, nausea/vomiting, pain) and EQ-5D-5L health status measures. These outcomes will be analyzed longitudinally as continuous scores according to instrument scoring manuals, with baseline-adjusted trajectories modeled over time.
Secondary outcomes include: (1) exposure--PRO relationships for efficacy-related domains (e.g., global health status/QoL from EORTC QLQ-C30), and (2) simulation-based evaluation of alternative dosing regimens to assess predicted changes in PRO toxicity and efficacy profiles. For secondary analyses, summary exposure metrics (e.g., steady-state AUC) and time-varying exposure predictions will be explored.
PRO measures will primarily be analyzed as continuous longitudinal variables. Sensitivity analyses may categorize clinically meaningful deterioration or improvement using established minimal clinically important difference (MCID) thresholds from published literature. Any categorization thresholds used will be pre-specified and justified based on validated scoring guidelines.
No changes to the primary outcome definition are currently anticipated. If additional exploratory exposure metrics or PRO domains are evaluated, these will be clearly identified as secondary or exploratory analyses in any resulting publication.
Main Predictor/Independent Variable and how it will be categorized/defined for your study:
The primary independent variable is systemic drug exposure derived from population pharmacokinetic (PK) modeling. Individual exposure metrics will be estimated using nonlinear mixed-effects modeling based on observed plasma concentration--time data and dosing history. Exposure will be characterized using both summary metrics (e.g., area under the concentration--time curve [AUC], maximum concentration [Cmax], and average steady-state concentration) and time-varying predicted plasma concentrations.
For the primary exposure--PRO analyses, exposure will be modeled as a continuous variable to preserve quantitative information and support estimation of exposure--response relationships. Time-varying exposure predictions will be linked to longitudinal PRO outcomes to evaluate dynamic exposure--PRO associations. In secondary analyses, exposure may also be categorized (e.g., tertiles or quartiles) for visualization and descriptive comparisons; however, primary inference will rely on continuous modeling.
Dose level assigned in the trial (e.g., multiple-dose cohorts) will be included as a secondary exposure descriptor and may be used in sensitivity analyses to compare model-based exposure metrics with nominal dose groupings.
All exposure variables will be defined a priori and derived consistently with standard pharmacometric practice to ensure comparability with published exposure--response analyses.
Other Variables of Interest that will be used in your analysis and how they will be categorized/defined for your study:
Baseline PRO scores (EORTC QLQ-C30 and EQ-5D-5L domain scores) will be included as continuous covariates to account for pre-treatment symptom burden and health status. Time since treatment initiation (study day or cycle) will be modeled to account for temporal trends independent of exposure.
Demographic variables will include age (continuous), sex (categorical), race/ethnicity (categorical, as available), and body weight or body surface area (continuous), given their potential influence on both PK and PRO outcomes. Disease-related variables such as baseline performance status (e.g., ECOG, categorical), disease stage, and prior lines of therapy (categorical) will be included where available, as these may influence symptom trajectories and treatment tolerability.
Treatment-related variables will include nominal dose level (categorical), cumulative dose (continuous), and treatment discontinuation status. Concomitant medications with potential symptom-modifying effects (e.g., antiemetics, analgesics) will be considered as time-varying covariates if available.
Statistical Analysis Plan:
All analyses will be conducted within the YODA secure data platform using R/RStudio and/or Stata. We also request install of Monolix2024R1 (Lixoft, license will be provided by PI) on Yoda platform to perform population modeling analysis. Analyses will follow a pre-specified plan aligned with the study aims.
Descriptive Analyses: Baseline demographic and clinical characteristics will be summarized using means (SD), medians (IQR), or frequencies (%), as appropriate. Baseline PRO domain scores will be summarized and compared across nominal dose groups using ANOVA or Kruskal--Wallis tests for continuous variables and chi-square tests for categorical variables. PK concentration--time data will be graphically explored to assess distributional properties and data completeness.
Population Pharmacokinetic Modeling: A population PK model will be developed to characterize drug concentration--time profiles and estimate individual exposure metrics (e.g., AUC, Cmax, steady-state concentration). Inter-individual variability will be modeled using random effects, and residual variability will be characterized using appropriate error structures. Covariates (e.g., body weight, age, sex) will be evaluated based on biological plausibility and improvement in model fit.
Exposure--PRO Modeling (Primary Analysis): Longitudinal exposure--PRO relationships will be evaluated using mixed-effects models linking time-varying or summary exposure metrics to repeated PRO domain scores. Baseline PRO scores will be included as covariates to adjust for initial symptom burden. Exposure will be modeled as a continuous predictor to estimate exposure--response slopes. Nonlinear relationships will be explored using Emax or spline functions if suggested by graphical diagnostics. Interaction terms between exposure and key baseline variables (e.g., performance status) will be evaluated to assess effect modification.
Narrative Summary: The objective of this project is to evaluate the feasibility of incorporating PROs into early-phase dose selection using population exposure-response modeling. We will conduct a secondary analysis of de-identified participant-level data from NCT01381874. We will develop a population PK model to estimate individual drug exposure profiles. We will characterize relationships between systemic exposure and longitudinal PRO domains, focusing on both toxicity-related and efficacy-related measures. Finally, we will conduct simulation analyses to evaluate whether alternative dosing regimens informed by exposure-PRO relationships could improve the balance between symptom burden and therapeutic benefit. This study will be the first to formally evaluate whether quantitative exposure-PRO relationships can support PRO dose selection in early-phase CTs.
Project Timeline:
Anticipated project start date: Within 1 month of data access approval and execution of the Data Use Agreement, to ensure that required software such as Monolix installed successfully.
Months 1--2: Data familiarization, review of supporting documentation (protocol, CSR), dataset cleaning, and development of analysis-ready PK and PRO datasets within the secure platform.
Months 3--6: Development of the population pharmacokinetic model and estimation of individual exposure metrics.
Months 6--9: Exposure--PRO longitudinal modeling, sensitivity analyses, and simulation of alternative dosing regimens.
Month 10: Final model validation, interpretation of results, and preparation of tables/figures.
Months 11--12: Manuscript drafting and internal review; first submission for publication by Month 12. Results will be reported back to the YODA Project upon manuscript submission and no later than the end of the 12-month access period.
Dissemination Plan:
The primary anticipated product is a peer-reviewed manuscript describing the development of exposure--PRO models and the feasibility of PRO-informed dose selection in early-phase clinical trials. A secondary methodological manuscript may be developed focusing on the modeling framework and simulation approach for integrating longitudinal PRO data into exposure--response analyses.
Target audiences include clinical pharmacologists, oncologists, biostatisticians, outcomes researchers, regulatory scientists, and early-phase trial investigators. The findings will be relevant to drug development teams, academic researchers, and regulatory agencies interested in patient-centered dose optimization.
Potential journals include Clinical Pharmacology & Therapeutics, CPT: Pharmacometrics & Systems Pharmacology, Journal of Clinical Oncology -- Clinical Cancer Informatics, Statistics in Medicine, and Quality of Life Research. Results may also be presented at scientific conferences such as ASCPT, PAGE, ASCO, or ISOQOL to reach both pharmacometric and PRO research communities.
Bibliography:
[1] Yap, Christina, et al. “Advancing patient-centric care: integrating patient reported outcomes for tolerability assessment in early phase clinical trials--insights from an expert virtual roundtable.” EClinicalMedicine 76 (2024).
[2] Gao, Wei, et al. “Realizing the promise of Project Optimus: Challenges and emerging opportunities for dose optimization in oncology drug development.” CPT: Pharmacometrics & Systems Pharmacology 13.5 (2024): 691-709.
[3] Basch, Ethan. “Beyond the FDA PRO guidance: steps toward integrating meaningful patient-reported outcomes into regulatory trials and US drug labels.” Value in Health 15.3 (2012): 401-403.
[4] Basch, Ethan. “Patient-reported outcomes-harnessing patients’ voices to improve clinical care.” New England Journal of Medicine 376.2 (2017): 105-108.
[5] Zhou, Jiawei, et al. “Revolutionizing patient-reported outcomes analysis for oncology drug development using population models.” Clinical Cancer Research 31.9 (2025): 1580-1586.
[6] Zhou, Jiawei, et al. “Leveraging Longitudinal Patient-Reported Outcome Trajectories to Predict Survival in Non--Small Cell Lung Cancer.” Clinical Cancer Research 31.13 (2025): 2685-2694.
[7] Zhou, Jiawei, et al. “From Symptom to Outcome: Defining Clinically Meaningful Patient‐Reported Appetite Loss in Non‐Small‐Cell Lung Cancer.” Journal of Cachexia, Sarcopenia and Muscle 16.6 (2025): e70150.
Supplementary Material: 2026-0120_Supplementary Material