Lead Investigator: Benoit You, Université Claude Bernard Lyon
Title of Proposal Research: Prognostic And Predictive Value Of The Modeled PSA Kinetics Parameters In Patients With Metastatic Castration-Resistant Prostate Cancer Treated With Docetaxel Or Cabazitaxel (Phase III Trial — FIRSTANA)
Vivli Data Request: 7662
Funding Source: None
Potential Conflicts of Interest: None
Summary of the Proposed Research:
Prostate cancer (PCa) is the second most frequent cancer in men worldwide in 2018 representing 7.1% of total diagnosed cancer, after lung cancer (11.8% of all cancers). Generally, PCa is diagnosed early by high serum Prostate-Specific Antigen (PSA) levels. PSA is mainly released into the seminal fluid and is an organ-specific marker of the prostate, expressed in both normal and tumor tissues. Indeed, an elevation of PSA levels (currently defined at 4 ng/mL) usually reflects prostatic diseases and most patients with prostate cancers have high value of PSA at diagnosis. PSA is an interesting tool to characterize PCa because PSA blood test is currently a cheap, simple and early test available to diagnose PCa and to follow it evolution.
It was shown that PSA decline during treatment was related to treatment efficacy, leading to the official definition of PSA response by The PSA working group: PSA response corresponds to a 50% decrease of PSA levels in a period of four or more weeks after treatment beginning. However, this definition is not optimal, as it relies on a two time-point algorithm (PSA reference level, measured within 2 weeks before therapy starting, and PSA levels measured at least four weeks after treatment beginning), potentially influenced by the variability of PSA assays. Indeed, it does not take into account PSA level changes between these 2 measures, and PSA levels are likely to increase or decrease before observing a potential PSA response.
In that context, EA 3738 CICLY (Centre pour l’Innovation en Cancérologie de LYon) team of the “Faculté de Médecine et de Maïeutique Lyon-Sud Charles Mérieux” (Lyon University), under the direction of Prof. B. You, uses mathematical modeling and simulation analyses to assess the longitudinal kinetics of serum tumor markers such as PSA concentrations during the first 100 days of treatment. To carry out this type of analyses, the team uses mathematical equations to describe PSA kinetics of the population and to specify inter-individual variability, in other words, to explain differences observed in each individual and the global population kinetics. This method allows describing the early longitudinal dynamics of PSA during the first 100 days of therapy using all PSA time points measured during this period of time, and secondarily extract kinetic parameters characterizing the whole decline curve that describe PSA concentration. Indeed, with that approach, PSA kinetics is described by several parameters such as the PSA ELIMiniation rate constant K (KELIM), that reflects the PSA elimination rate during the treatment, or the PSA PRODuction rate K (KPROD), which represents the PSA production rate over the time.
The approach was shown to be effective for other serum markers such as CA-125 and hCG in other cancers. The modeled kinetic parameter KELIM of CA-125 is now recognized as an indicator of the tumor primary chemosensitivity, and may become a new standard biomarker in first line setting.
We assume that modeled parameters of PSA early longitudinal kinetics during treatment may provide a new early prognostic marker, and an indicator of the therapy efficacy.
An initial study that assessed prognostic value of PSA kinetics in patients with prostate cancers from a phase III trial testing the efficacy of adding a chemotherapy (docetaxel) to an endocrine therapy (Androgen-Deprivation Therapy (ADT)). This method showed promising results. We found that: 1) the analysis of PSA longitudinal kinetics during the first 100 days of systemic treatment was feasible; 2) the two kinetic parameters KELIM and KPROD exhibited independent prognostic value regarding patients’ survival.
To simplify, KELIM and KPROD represent rates of PSA elimination and production respectively. The higher KELIM, the more the chemotherapy is effective, the faster the CA-125 is eliminated.
In other words, we assume that PSA kinetics during the first 100 days of treatment might enable to characterize the patient prognosis, and to identify the patients who will respond to the treatment early.
The objective of this new project is to confirm this hypothesis by assessing the prognostic value of PSA KELIM and KPROD in prostate cancer patients treated with docetaxel or cabazitaxel, in FIRSTANA clinical trial. The mathematical equations driving the decline of PSA in these patients will be determined, in order to extract kinetic parameters KELIM and KPROD as indicators of the quality of PSA decline. Then, the aim will be to investigate the value of these indicators of PSA kinetics as prognostic markers regarding patient survival, and response to treatment, and the predictive value regarding either study drugs.
In other words, we want to see if these new kinetic parameters could separate patients according to their prognosis, so treatment can be densified in those with poor prognosis, or de-escalated in those with good prognosis. Moreover, we want to see if these kinetic parameters could identify early the patients who will benefit more from docetaxel or cabazitaxel.
Statistical Analysis Plan:
1. Estimation of modeled Prostate-Specific Antigen (PSA) PRODuction rate K (KPROD) and ELIMination constant rate K (KELIM) values for each patients.
To eliminate right-skewness in the PSA distribution and to normalize the titer distribution, PSA concentrations will be log-transformed. The model previously developed for assessing the PSA kinetics during the Androgen-Deprivation Therapy (ADT) +/- docetaxel regimens will be adjusted to patients treated with cabazitaxel and docetaxel.
The mathematical modeling of early PSA kinetics with a non-linear mixed effect model was previously described (Aurore Carrot et al., 2022; A. Carrot et al., 2021 ESMO, 2021 PAGE).
2. Prognostic and predictive value of standardized (std) KELIM and KPROD calculated during the 100 days of treatment regarding the overall response rate, Progression-Free Survival (PFS) and Overall Survival (OS) of patients.
The discriminatory predictive ability, along with the prognostic value of categorical KPROD and KELIM, regarding PFS and OS will be assessed using univariate analyses with Box plots, Kaplan-Meier method and Log-rank test, and multivariate analyses with Concordance-index (C-index) method and Cox regression tests. The other tested prognostic factors in the multivariate analyses will be: disease stage, Gleason score, PSA response criterion, prior treatment.
The final multivariate C-index and Cox survival models will be obtained using backward selections.
KELIM and KPROD will be assessed as a continuous covariates, and a discrete covariates separated by the median as favorable (≥ median) or unfavorable (< median).
All survival analyses will be implemented with a landmark time point set at 100 days after the start of treatment. Indeed, PSA is modeled from day 0 to 100, and exclusion of the early progressions observed during the first 100 days will avoid the biases related to the links between early progressions and PSA kinetics, or radiological tumor response.
3. Statistics and Computing Process.
All tests will be implemented using a two-sided 0.05 alpha risk. Monolix software will be used to fit the semi-mechanistic model to PSA kinetic data. Logistic analyses; Survival analyses; concordance probability (C-index) will be obtained in R software version 4.1.1.
Requested Studies:
Cabazitaxel Versus Docetaxel Both With Prednisone in Patients With Metastatic Castration Resistant Prostate Cancer (FIRSTANA)
Data Contributor: Sanofi
Study ID: NCT01308567
Public Disclosures:
A. Carrot, A., Oudard, S., Fizazi, K., Maillet, D., Sartor, O., Freyer, G., You, B. Prognostic Value of the Modeled Prostate-Specific Antigen KELIM Confirmation in Metastatic Castration-Resistant Prostate Cancer Treated With Taxanes in FIRSTANA. JCO Clin Cancer Inform 8, e2300208(2024). DOI:10.1200/CCI.23.00208