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: A Refined Continuous Risk Index for Accurately Predicting Outcomes of Patients with CLL after Limited-Duration Therapy
Scientific Abstract:
Background: In chronic lymphocytic leukemia (CLL), measurable residual disease (MRD) after fixed-duration therapy predicts progression-free and overall survival. However, most prognostic tools rely on baseline characteristics and do not incorporate longitudinal treatment response.
Objective: To validate a refined Continuous Individualized Risk Index (CIRI2-CLL) that integrates clinical risk factors with repeated MRD assessments to improve individualized outcome prediction after fixed-duration targeted therapy.
Study Design: Retrospective external validation study using individual patient-level data from a completed randomized trial.
Participants: Patients with previously untreated CLL enrolled in the GLOW trial, with available baseline data, MRD assessments and survival outcomes.
Primary and Secondary Outcome Measures: The primary outcome is progression-free survival. Secondary outcomes include overall survival, model calibration, discrimination and risk stratification.
Statistical Analysis: CIRI2-CLL uses a Bayesian Cox proportional hazards framework integrating baseline clinical factors and MRD as time-varying covariates at interim, end-of-treatment, and post-treatment landmarks. Prior information from previously analyzed trials is incorporated through Bayesian priors to support stable risk estimation and uncertainty quantification. A piecewise proportional hazards approach addresses non-proportional hazards. Model parameters remain fixed, and performance in GLOW will be evaluated using calibration metrics, C-statistics and comparison with other prognostic indices.
Brief Project Background and Statement of Project Significance:
Chronic lymphocytic leukemia (CLL) has undergone a major therapeutic shift toward fixed-duration, chemotherapy-free regimens using targeted agents. While these approaches achieve high response rates and deep remissions, a substantial proportion of patients relapse after treatment completion. Accurately identifying patients at increased risk of relapse remains a key unmet need. Existing prognostic tools, such as the CLL International Prognostic Index (CLL-IPI), were developed primarily in the context of chemoimmunotherapy and rely on baseline characteristics alone. As a result, they do not adequately capture the dynamic treatment response patterns observed with modern targeted therapies.
Measurable residual disease (MRD) has emerged as a robust prognostic marker in CLL across treatment modalities and timepoints, including after fixed-duration therapy. However, MRD is typically evaluated at single landmarks and is not routinely integrated with clinical risk factors into a unified prognostic framework. To address this gap, we previously developed the Continuous Individualized Risk Index (CIRI), a Bayesian survival modeling approach that integrates baseline risk factors with longitudinal response data to generate individualized, continuously updated survival predictions. A refined version of this model (CIRI2-CLL) incorporates repeated MRD assessments during and after therapy and has demonstrated superior prognostic performance compared with conventional indices in multiple completed clinical trials.
The proposed project seeks to externally validate CIRI2-CLL using individual patient-level data from the GLOW trial, which evaluates fixed-duration ibrutinib plus venetoclax. This regimen represents a widely adopted treatment strategy that differs mechanistically from prior venetoclax--CD20 antibody combinations, providing a critical test of model generalizability across modern targeted therapies. Importantly, the GLOW dataset offers high-quality longitudinal MRD and long-term outcome data necessary for independent validation without model refitting.
By confirming the robustness and transportability of a dynamic, MRD-integrated risk model, this work will materially enhance prognostic assessment in CLL and support more personalized disease monitoring, follow-up strategies, and clinical trial design. More broadly, the project advances generalizable methodology for integrating longitudinal biomarkers into survival prediction, with relevance beyond CLL to other cancers treated with time-limited targeted therapies. The results are expected to inform future MRD-guided clinical strategies and contribute to improved long-term outcomes and public health.
Specific Aims of the Project:
The overall objective of this project is to externally validate a refined Continuous Individualized Risk Index (CIRI2-CLL) for patients with chronic lymphocytic leukemia (CLL) treated with fixed-duration targeted therapy. CIRI2-CLL integrates baseline clinical risk factors with longitudinal measurements of measurable residual disease (MRD) to provide individualized predictions of disease progression.
Aim 1: To evaluate the discrimination and calibration of CIRI2-CLL for predicting progression-free survival in patients enrolled in the GLOW trial of fixed-duration ibrutinib plus venetoclax.
Hypothesis: CIRI2-CLL will demonstrate accurate calibration and superior discrimination compared with established prognostic tools, including CLL-IPI and single timepoint MRD assessments.
Aim 2: To assess the generalizability of CIRI2-CLL across a chemotherapy-free regimen distinct from those used in model development.
Hypothesis: Model performance will be maintained without refitting, supporting robustness across modern targeted treatment strategies.
Aim 3: To evaluate risk stratification for overall survival and clinically relevant subgroups.
Hypothesis: CIRI2-CLL will identify distinct low-, intermediate-, and high-risk groups with significantly different outcomes.
Study Design: Individual trial analysis
What is the purpose of the analysis being proposed? Please select all that apply.: New research question to examine treatment effectiveness on secondary endpoints and/or within subgroup populations Confirm or validate previously conducted research on treatment effectiveness Participant-level data meta-analysis Meta-analysis using data from the YODA Project and other data sources Develop or refine statistical methods Research on clinical prediction or risk prediction
Software Used: R, RStudio
Data Source and Inclusion/Exclusion Criteria to be used to define the patient sample for your study:
The YODA Project data requested for this study are individual patient-level data from the phase III GLOW trial in previously untreated patients with chronic lymphocytic leukemia (CLL). These data will be used exclusively for independent external validation of a previously developed prognostic model.
Inclusion criteria: All patients randomized in the GLOW trial will be included according to the intention-to-treat (ITT) principle.
Exclusion criteria: None beyond those defined by the original GLOW trial protocol. No additional demographic or clinical exclusions will be applied.
Data from other clinical trials (including CLL8, CLL10, CLL11, MURANO, and CLL14) were previously used for model development and internal validation and are held in-house under institutional sponsorship. These datasets will not be pooled with the YODA-provided GLOW data. All analyses using GLOW data will be conducted separately for external validation purposes only, and no combined individual patient-level analyses across YODA and non-YODA datasets will be performed.
Primary and Secondary Outcome Measure(s) and how they will be categorized/defined for your study:
The primary outcome measure for this study is progression-free survival (PFS), defined in accordance with the GLOW trial protocol as the time from randomization to the first occurrence of disease progression, relapse or death from any cause, whichever occurs first. Patients without an event will be censored at the date of last disease assessment. PFS will be evaluated as a time-to-event outcome and used to assess discrimination, calibration and risk stratification performance of the prognostic model.
The secondary outcome measure is overall survival (OS), defined as the time from randomization to death from any cause. Patients alive at last follow-up will be censored at the date of last contact. OS will be analyzed as a time-to-event outcome to assess model performance for longer-term survival prediction.
Additional secondary performance measures include model calibration (agreement between predicted and observed event probabilities), discrimination (C-statistics) and predefined risk group stratification based on predicted survival probabilities. These are analytic performance metrics rather than clinical endpoints.
All outcomes will be analyzed using the intention-to-treat population and according to the original GLOW trial endpoint definitions. No changes to the primary or secondary outcome measures are planned between the current study and any subsequent publication. No exploratory or post-hoc clinical endpoints beyond those specified above will be introduced.
Main Predictor/Independent Variable and how it will be categorized/defined for your study:
The primary independent variable for this study is the Continuous Individualized Risk Index for CLL (CIRI2-CLL), a pre-specified composite prognostic score that provides individualized predicted survival probabilities following fixed-duration targeted therapy. CIRI2-CLL is calculated using a Bayesian Cox proportional hazards framework that integrates baseline clinical risk factors, treatment regimen and longitudinal measurable residual disease (MRD) assessments.
Baseline components include established clinical and biological risk factors summarized by the CLL International Prognostic Index (CLL-IPI), encompassing age, disease stage, IGHV mutation status, TP53 aberrations and serum β2-microglobulin. Longitudinal components include MRD status measured at protocol-defined timepoints (interim/on-treatment, end-of-treatment and post-treatment follow-up), incorporated as time-varying covariates. MRD is defined according to trial standards, with undetectable MRD typically defined as <10⁻⁴ leukemic cells.
CIRI2-CLL will be analyzed both as a continuous variable (predicted probability of progression-free or overall survival at predefined time horizons) and as a categorical variable, stratified into pre-specified low-, intermediate- and high-risk groups based on predicted survival probabilities. Cut points for risk categories are fixed based on prior model development and will not be recalibrated using GLOW data.
Other Variables of Interest that will be used in your analysis and how they will be categorized/defined for your study:
Other Variables of Interest
In addition to the primary predictor (CIRI2-CLL), several variables will be used to characterize the study population and for descriptive and comparative analyses. These variables are derived from the GLOW trial dataset and defined according to the original trial protocol.
Baseline demographic and clinical variables include age at randomization (continuous and categorical, as defined in CLL-IPI), sex, disease stage (Rai or Binet, as applicable), Eastern Cooperative Oncology Group (ECOG) performance status, and serum β2-microglobulin level (continuous and categorized per CLL-IPI thresholds).
Biological risk factors include IGHV mutation status (mutated vs unmutated) and TP53 aberrations (presence vs absence of del(17p) and/or TP53 mutation), defined using trial-standard assays.
Treatment-related variables include randomized treatment assignment (ibrutinib plus venetoclax vs comparator arm), treatment start and end dates, and duration of therapy. Treatment assignment will be used for descriptive purposes and benchmarking only, not as an independent predictor.
Measurable residual disease (MRD) variables include MRD status assessed in peripheral blood at protocol-defined timepoints (interim/on-treatment, end-of-treatment, and post-treatment follow-up). MRD will be analyzed as detectable vs undetectable using the trial-defined threshold (typically <10⁻⁴). Individual MRD timepoints may be reported descriptively but will not replace the composite CIRI2-CLL predictor in primary analyses.
Outcome-related variables include dates of disease progression, relapse, death, and last follow-up, used to derive progression-free survival and overall survival endpoints.
Statistical Analysis Plan:
All analyses will be conducted using individual patient-level data from the GLOW trial under the intention-to-treat principle. The primary objective is external validation of a previously developed prognostic model; therefore, no model refitting or data-driven optimization will be performed using the GLOW dataset.
Descriptive Analyses
Baseline demographic, clinical, and biological characteristics will be summarized for the overall study population and by randomized treatment arm using appropriate descriptive statistics. Continuous variables will be summarized using means with standard deviations or medians with interquartile ranges, as appropriate. Categorical variables will be summarized using frequencies and percentages. Baseline characteristics will be compared descriptively between treatment arms to confirm balance achieved by randomization; no formal hypothesis testing of baseline differences is planned.
Primary Outcome Analysis
The primary endpoint is progression-free survival (PFS), defined according to the GLOW trial protocol. Kaplan--Meier methods will be used to estimate PFS distributions, with median PFS and landmark PFS rates reported with 95% confidence intervals. The Continuous Individualized Risk Index (CIRI2-CLL) will be applied to each patient to generate individualized predicted PFS probabilities at predefined time horizons without recalibration.
Model discrimination will be assessed using time-dependent concordance statistics (C-statistics). Model calibration will be evaluated by comparing predicted versus observed PFS probabilities using calibration plots and absolute differences between predicted and observed event rates at clinically relevant timepoints. Risk stratification performance will be assessed by categorizing patients into pre-specified low-, intermediate-, and high-risk groups based on CIRI2-CLL predictions, with Kaplan--Meier curves used to visualize separation between risk groups.
Secondary Outcome Analysis
Overall survival (OS) will be analyzed as a secondary endpoint using Kaplan--Meier methods. Model discrimination and calibration for OS will be evaluated using the same approaches applied for PFS. The ability of CIRI2-CLL to stratify patients into distinct OS risk groups will be assessed descriptively.
Comparative Analyses
The performance of CIRI2-CLL will be compared with established prognostic tools, including the CLL International Prognostic Index (CLL-IPI) and single timepoint MRD assessments. Comparative performance will be evaluated using differences in C-statistics and calibration metrics. Treatment assignment and individual MRD timepoints will be analyzed for benchmarking purposes only and will not be used as alternative primary predictors.
Multivariable Modeling
No new multivariable Cox models will be fitted to the GLOW data for outcome prediction. The analysis is limited to application and evaluation of the pre-specified CIRI2-CLL model parameters derived from prior studies. Sensitivity analyses may include stratified evaluation by treatment arm and predefined clinical subgroups to assess consistency of model performance.
Missing Data
Missing baseline covariates or MRD values will be handled according to predefined rules established during model development. No data-driven imputation strategies will be developed using the GLOW dataset. The extent and pattern of missing data will be reported descriptively.
Statistical Software
All analyses will be performed using R/RStudio. Statistical significance testing will be descriptive in nature, with emphasis placed on effect estimates, confidence intervals,and model performance metrics rather than formal hypothesis testing.
Narrative Summary: Patients with chronic lymphocytic leukemia (CLL) increasingly receive fixed-duration therapies, but predicting who will remain in remission is challenging. Current risk scores rely only on information collected before treatment and do not account for treatment response over time. This study aims to develop and validate a refined, easy-to-use risk index that combines standard clinical factors with repeated measurements of minimal residual disease (MRD) during and after therapy. Using data from completed clinical trials, we will build and test a model that more accurately estimates an individual patient's risk of relapse. Data from the GLOW trial of ibrutinib plus venetoclax are essential to evaluate performance across modern chemotherapy-free treatment strategies. Improved risk prediction will support personalized disease management.
Project Timeline:
Anticipated project start date: Month 0 (upon execution of the YODA Data Use Agreement and receipt of the GLOW dataset).
Data preparation and quality checks: Months 0--0.5. Dataset familiarization, verification of variable definitions, and confirmation of data completeness required for application of the pre-specified CIRI2-CLL model.
Primary and secondary analyses: Months 0.5--3. External validation analyses will be conducted, including calculation of CIRI2-CLL scores, assessment of model discrimination and calibration for progression-free and overall survival, risk stratification, and predefined subgroup analyses.
Manuscript drafting: Months 3--4. Preparation of the manuscript describing methods, results, and interpretation.
Manuscript submission: Months 4--5. First submission of the manuscript to a peer-reviewed journal.
Revision and reporting: Months 5--6. Initial response to peer-review feedback (if applicable) and reporting of results to the YODA Project in accordance with Data Use Agreement requirements.
Dissemination Plan:
The primary product of this project will be a peer-reviewed scientific manuscript reporting the external validation of the CIRI2-CLL prognostic model using data from the GLOW trial. The manuscript will describe the study rationale, methods, results, and clinical implications, with full transparency regarding data sources and analytic approach. Based on an editorial review of an initial manuscript draft describing model development and validation in earlier trials, editors at Blood have indicated interest in the work and specifically encouraged inclusion of data from patients treated with venetoclax plus ibrutinib, underscoring the relevance of the proposed analysis. Accordingly, Blood will be a primary target journal, with additional suitable journals including Journal of Clinical Oncology, Leukemia, and Haematologica.
In addition to journal publication, results are expected to be presented at major international scientific meetings such as the American Society of Hematology (ASH) Annual Meeting or the European Hematology Association (EHA) Congress, subject to abstract acceptance.
All dissemination activities will comply with the YODA Project Data Use Agreement, including acknowledgment of data access through YODA. No patient-identifiable information will be disclosed.
Bibliography:
Kurtz DM, Esfahani MS, Scherer F, et al. Dynamic Risk Profiling Using Serial Tumor Biomarkers for Personalized Outcome Prediction. Cell. 2019 Jul 25;178(3):699-713.e19.
Al-Sawaf O, Zhang C, Lu T, et al. Minimal Residual Disease Dynamics after Venetoclax-Obinutuzumab Treatment: Extended Off-Treatment Follow-up From the Randomized CLL14 Study. J Clin Oncol. 2021 Dec 20;39(36):4049-4060.
Kater AP, Owen C, Moreno C, et al. Fixed-Duration Ibrutinib--Venetoclax in Patients with Chronic Lymphocytic Leukemia and Comorbidities. NEJM Evidence. 2022 Jul;1(7):EVIDoa2200006
Niemann CU, Munir T, Moreno C, et al. Fixed-duration ibrutinib--venetoclax versus chlorambucil--obinutuzumab in previously untreated chronic lymphocytic leukaemia (GLOW): 4-year follow-up from a multicentre, open-label, randomised, phase 3 trial. Lancet Oncology. 2023 Dec;24(12):1423--1433
Supplementary Material: iwCLL-2025_CIRI_poster_v0.5-3.pdf