Lead Investigator: Tim Brümmendorf, University Hospital Aachen
Title of Proposal Research: Entropy – based Biomarkers for individualized response on Bosutinib treatment in chronic phase CML
Vivli Data Request: 4921
Funding Source: None.
Potential Conflicts of Interest: Consultant (Novartis, Pfizer, Janssen, Merck, Takeda) and research funding (Novartis and Pfizer)
Summary of the Proposed Research:
Despite the progress made in treatment options of chronic myeloid leukemia (CML) with tyrosine kinase inhibitors (TKI), disease progression from the chronic phase (CP) towards an accelerated phase (AP) or blast crisis (BC) still poses a problem, as it can result in lower treatment success and elevated relapse rates. As a result, the prevention of disease progression remains the primary goal in treatment of CML patients and the prediction of progression risk is of high clinical relevance.
To date, the prognostication of CML is focused on clinical scores, such as Sokal, Hasford (“Hasford”) and semi-prospective monitoring of patients based on the kinetics of their clinical response and/or their minimal residual disease (MRD) state defined by the BCR-ABL mutation transcript levels at defined in time [3]. In order to complement this, prognostic or predictive biomarkers, such as the telomere length in leukemic cells or the proportion of BCR-ABL mutated cells in the patients’ immature CD34(+)38(-) stem cell compartment, are currently under investigation.
Despite their limitations, these scores can be used as valid prognostication tools, each fulfilling specific purposes. Still, they do not allow reliable inference of individual progression risk for clinical use yet, which is why we aim to apply a patient stratification method that combines clinical parameters with gene expression-related markers [1] to assess individual disease progression and make predictions about TKI efficacy in those patients.
Statistical Analysis Plan:
Analysis Plan: In detail, we will
1. Use the BCR-ABL mutation ratio of patients’ blood samples before treatment to calculate the corresponding individual disease progression time according to the model by Dingli et al. [5].
Input: BCR-ABL mutation ratio for each patient in BELA study
Method: Model by Dingli et al. [5]
Output: Disease progression time interval for each patient
2. Map the disease progression time of the patients enrolled in BELA study model to the respective time for the individual patients of a prior study by
Radich et al. [6].
Inputs: Disease progression time interval of patients and reference data set by Radich et al. [6]
Method: Mapping of disease progression times from BELA study to respective patient data from GSE4170 using [1]
Output: Expected gene expression profiles of patients
3. Identify the exact patient disease status of patients from comparison of gene expression entropies to sub-stratify chronic phase CML patients into early and late chronic phase patients.
Input: Gene expression profiles of patients
Method: Stratification method based on Shannon entropy of gene expression [1]
Output: Patient subgroups within the chronic phase of CML
4. Find a combination of clinical covariates that can be associated with that substratification.
Input: Clinical parameters assessed in the BELA study
Method: Statistical classification tests; regression analysis (linear-, ridge-, logistic-regression, Lasso, neural networks)
Output: Stratification-associated covariates
5. Estimate the associated gene expression profile of a patient and explore the association to the response to Bosutinib compared to the response to
Imatinib.
Input: Cell line gene expression data from the GDSC data base [8, 9] and expected patient gene expression data when using data set by Radich et al. [6] as a reference
Method: Regression analysis (linear-, ridge-, logistic-regression, Lasso, neural networks)
Output: Patient drug response
6. Characterize specific cellular gene expression patterns that can be associated with the drug responses and relating these to patient gene expression profiles using PhysioSpace [2].
Input: Cell line gene expression data from the GDSC data base [8, 9] and expected patient gene expression data when using data set by Radich et al. [6] as a reference
Method: PhysioSpace [2]
Output: Similarity scores; relationship between cell line drug response (IC50 values) and patient drug response
Requested Studies:
A PHASE 3 RANDOMIZED, OPEN-LABEL STUDY OF BOSUTINIB VERSUS IMATINIB IN SUBJECTS WITH NEWLY DIAGNOSED CHRONIC PHASE PHILADELPHIA CHROMOSOME POSITIVE CHRONIC MYELOGENOUS LEUKEMIA
Sponsor: Pfizer Inc.
Study ID: NCT00574873
Sponsor ID: B1871008
Public Disclosures:
Esfahani, Ali, Susanne Isfort, Tim Brümmendorf, and Andreas Schuppert. Entropy-based Biomarkers for Individualized Response on Bosutinib Treatment in Chronic Phase CML. OSF Preprints. March 1, 2022. doi:10.31219/osf.io/bu2qy.