A survival mediation model with Bayesian model averaging

Abstract

Determining the extent to which a patient is benefiting from cancer therapy is challenging. Criteria for quantifying the extent of “tumor response” observed within a few cycles of treatment have been established for various types of solids as well as hematologic malignancies. These measures comprise the primary endpoints of phase II trials. Regulatory approvals of new cancer therapies, however, are usually contingent upon the demonstration of superior overall survival with randomized evidence acquired with a phase III trial comparing the novel therapy to an appropriate standard of care treatment. With nearly two-thirds of phase III oncology trials failing to achieve statistically significant results, researchers continue to refine and propose new surrogate endpoints. This article presents a Bayesian framework for studying relationships among treatment, patient subgroups, tumor response, and survival. Combining classical components of a mediation analysis with Bayesian model averaging, the methodology is robust to model misspecification among various possible relationships among the observable entities. A posterior inference is demonstrated via an application to a randomized controlled phase III trial in metastatic colorectal cancer. Moreover, the article details posterior predictive distributions of survival and statistical metrics for quantifying the extent of direct and indirect, or tumor response mediated treatment effects.

Predicting outcomes of phase III oncology trials with Bayesian mediation modeling of tumor response

“…Let 𝜽 = (𝜷, 𝜸, 𝜈, 𝜆) denote the vector of parameters of interest and assume a Weibull baseline hazard model. After fitting the MED discussed in Zhou et al 25 using data O 𝜏 , we will obtain K posterior samples…”

“…Under the Weibull assumption for survival baseline distribution, the parameters for the model are denoted as . Estimation of mediation models and risk ratios have been discussed in Vandenberghe et al 24 using frequentist approaches and in Zhou et al 25 under the Bayesian framework. Here we adopt the Bayesian estimation procedures introduced by Zhou et al 25 and develop methodology for clinical trial prediction.…”

“…Estimation of mediation models and risk ratios have been discussed in Vandenberghe et al 24 using frequentist approaches and in Zhou et al 25 under the Bayesian framework. Here we adopt the Bayesian estimation procedures introduced by Zhou et al 25 and develop methodology for clinical trial prediction. Two advantages of the Bayesian modeling framework include: (1) we can estimate the posterior distribution of the parameters and risk ratio measures via Markov chain Monte Carlo (MCMC) and (2) we can predict the response and survival outcomes based on the posterior samples for .…”

“…Let denote the vector of parameters of interest and assume a Weibull baseline hazard model. After fitting the MED discussed in Zhou et al 25 using data , we will obtain K posterior samples . The goal is to predict whether the trial will end successfully, as defined by statistically significant evidence of improved survival for the treatment arm.…”

Radiomics, deep learning and early diagnosis in oncology

“…Patients with TNBC usually undergo neoadjuvant systemic therapy (NAST) for downstaging of disease to facilitate less invasive surgery. The extent of downstaging is used as a surrogate prognostic marker [ 17 ]. Pathological complete response (pCR) to NAST is only seen in about half of the TNBC patients undergoing the treatment.…”