KEY TAKEAWAYS
- The observational trial aimed to investigate the applicability of the Weibull parametric model for estimating ETR in patients with EC survival analysis.
- Researchers noticed improved insights into treatment effects and potential survival extensions.
Semiparametric survival analysis, such as the Cox proportional hazards (CPH) regression model, is commonly employed in endometrial cancer (EC) studies due to its flexibility and robustness. However, while CPH regression does not require knowledge of the baseline hazard function, it lacks the capability to estimate the event time ratio (ETR), which quantifies the relative increase or decrease in survival time.
Xingfeng Li and the team aimed to assess the Weibull parametric model, which offers a method to estimate ETR and thereby enhances the understanding of survival dynamics in patients with EC.
They performed an inclusive analysis utilizing retrospective datasets comprising (n = 411) training and (n = 80) testing samples from patients with EC. They applied a bi-level model selection approach with a minimax concave penalty on the training dataset to identify optimal CPH models, incorporating clinical and radiomic features extracted from T2-weighted MRI images. Following model selection, diagnostic assessments were conducted to verify the proportional hazard assumption using the Schoenfeld test.
Subsequently, survival data were fitted to a Weibull model to calculate hazard ratios HR and ETR. Comparison metrics, including the Brier score and time-dependent area under the receiver operating characteristic curve (AUC), were employed to evaluate the performance of both CPH and Weibull models. The goodness of fit was assessed using the Kolmogorov-Smirnov (KS) statistic.
It was found that the proportional hazard assumption holds for fitting EC survival data. However, suspicion arose regarding the linearity of the model assumption due to observable trends in predictors such as age and cancer grade. Furthermore, a significant relationship between EC survival data and the Weibull distribution was observed.
Notably, the Weibull model demonstrated a higher AUC value than the CPH model, indicating superior predictive accuracy for EC survival in training and testing datasets. Additionally, the Weibull model exhibited a smaller Brier score, further supporting its enhanced performance in EC survival prediction.
The study concluded that the Weibull parametric model provides a comprehensive framework for EC survival analysis, offering simultaneous characterization of treatment effects through HR and ETR. This method’s potential extension to investigate progression-free survival (PFS) and disease-specific survival suggests its broader applicability in clinical research and treatment planning for patients with EC.
The trial was sponsored by Imperial College London.
Source: https://pubmed.ncbi.nlm.nih.gov/38724889/
Clinical Trial: https://clinicaltrials.gov/study/NCT03543215
Li X, Marcus D, Russell J, et al. (2024). “Weibull parametric model for survival analysis in women with endometrial cancer using clinical and T2-weighted MRI radiomic features.” BMC Med Res Methodol. 2024 May 9;24(1):107. doi: 10.1186/s12874-024-02234-1. PMID: 38724889; PMCID: PMC11080307.
