(240) Bridging clinical trial life expectancy (LE) inclusion criterion in the real-world (RW) setting: targeted literature review results for modeling feasibility in relapsed and refractory multiple myeloma (RRMM) patients

Publication Number: 1231

Background: Many clinical trials require that patients at enrollment have a minimum predicted LE (e.g., >6 months) to ensure patients are fit to receive newer therapies. However, LE assessment is typically based on a subjective evaluation by an investigator and is rarely recorded in clinical practice as reflected in real-world data (RWD). It is important to explore the applicability of such a key trial criterion in RWD, so that RW populations can more closely mimic those from trials for contextualization of outcomes in the absence of an active trial comparator.

Objectives: This study aimed to assess the feasibility of modeling LE in RWD for RRMM using published models.

Methods: A targeted literature review (TLR) was conducted. Searches were restricted to studies modeling LE or overall survival (OS) in MM patients published between 1/1/2016 and 3/15/2022 in the English language. Identified studies were assessed according to population, model design, outcome definitions, and study quality to determine the feasibility of application to RW RRMM patients.

Results: The search identified 1,391 citations, of which 1,171 were excluded at the title/abstract level. The remaining 219 studies were assessed at the full-text level and 6 studies (from 8 publications) were selected for inclusion. One study was conducted among RRMM patients, one among mixed newly diagnosed and RRMM patients, with the rest unclear about relapse/refractory status. Studies were heterogeneous in terms of patient populations/characteristics and data sources. All studies predicted relative risk or risk categories of mortality and no study predicted actual LE or OS rate at 6 months. All studies used multivariate Cox regression models. Predictors included in the models also differed by study, with most including patient and disease characteristics while others also included cytogenetic risk factors. Only three studies performed model validation including one qualitative validation through an expert panel.

Conclusions: This review identified a limited number of published LE models in patients with RRMM, with no study predicting actual LE or OS rate at 6 months. Therefore, LE prediction is considered not feasible at present for application to RW RRMM patients due to the lack of available models. Future studies are warranted to build de novo LE prediction models using RWD, incorporating known predictors for LE and clinical expert opinions.