Having models like these that produce more plausible results are crucial for decision-making in an era of targeted treatments that often come with immature data, argued researchers of their findings for minimal residual disease (MRD) in newly diagnosed multiple myeloma (NDMM).
Accounting for minimal residual disease (MRD) in long-term survival models for newly diagnosed multiple myeloma (NDMM) may offer more clinical plausible results than traditional approaches used for determining cost-effectiveness of treatments, say new study findings from the United Kingdom published in Journal of Health Economics and Outcomes Research.
Having models like these that produce more plausible results are crucial for decision-making in an era of targeted treatments that often come with immature data, argued the researchers. Determining the cost-effectiveness of a treatment in diseases like NDMM often requires long-term estimates of overall survival (OS) data, although with OS data often being immature in NDMM, uncertainty often arises.
“In disease areas with a relatively good prognosis, such as NDMM, estimating long-term survival for novel therapies is often challenging, as immature data can lead to considerable uncertainty in long-term survival estimates. Systematic literature reviews and meta-analyses have established MRD status as the most sensitive outcome in NDMM and supported its use as a surrogate for progression-free survival and OS,” explained the researchers, noting that the marker was used as the primary end point in the ongoing CEPHEUS study of patients with the disease.
Data from the current study were incorporated into 2 models to predict outcomes for patients with NDMM who received induction therapy with bortezomib, thalidomide, and dexamethasone, with or without daratumumab (DBTd and BTd, respectively), and consolidation therapy.
Compared with the standard partitioned survival model (PSM), which relied on extrapolations of individual patient data, the response-based PSM that accounted for MRD status reduced uncertainty in long-term survival outcomes. The updated approach demonstrated a lower divergence in OS estimates and mean life-years (LYs) predicted across various parametric extrapolations. The range across extrapolations was 3.4 and 7.7 LYs for DBTd and BTd, respectively, vs 14.8 and 11.8 LYs for the standard PSM.
The outcomes predicted by both PSMs were compared against observed data from the CASSIOPEIA trial and the phase 3 randomized controlled GIMEMA trial, and real-world data from the National Health Service Digital datasets. According to the researchers, the lower divergence remained when clinically implausible extrapolations were excluded.
“Despite the high level of divergence in long-term estimates across different parametric extrapolations for both DBTd and BTd, both the standard and response-based PSMs could generate plausible estimates of LYs compared with long-term survival estimates from GIMEMA and observed data from CASSIOPEIA,” described the group. “However, the reduced divergence of extrapolations in the response-based PSM provided a greater number of extrapolations that were clinically plausible. This demonstrates that, regardless of the approach taken, there is a need to explore multiple extrapolations and assess the clinical plausibility of extrapolations using external sources of validation and clinical expert opinion.”
Hest N, Morten P, Stubbs K, Trevor N. Use of minimal residual disease status to reduce uncertainty in estimating long-term survival outcomes for newly diagnosed multiple myeloma patients. JHEOR. 2023;10(1):1-9. doi:10.36469/001c.56072