Longitudinal Circulating Tumor DNA Modeling to Predict Disease Progression in First-Line Mutant Epidermal Growth Factor Receptor Non-Small Cell Lung Cancer.

This exploratory, post hoc analysis aimed to model circulating tumor DNA (ctDNA) dynamics and predict disease progression in patients with treatment-naïve locally advanced/metastatic epidermal growth factor receptor mutation (EGFRm)-positive non-small cell lung cancer, from the FLAURA trial (NCT02296125). Patients were randomized 1:1 and received osimertinib 80 mg once daily (q.d.) or comparator EGFR-TKIs (gefitinib 250 mg q.d. or erlotinib 150 mg q.d.). Plasma was collected at baseline and multiple timepoints until treatment discontinuation. Patients with Response Evaluation Criteria in Solid Tumors (RECIST) imaging data and detectable EGFR mutations (Ex19del/L858R) at baseline and ≥ 3 additional timepoints were evaluable. Joint modeling was conducted to characterize the relationship between longitudinal changes in ctDNA and probability of progression-free survival (PFS). A Bayesian joint model of ctDNA and PFS was developed solving differential equations with the ctDNA dynamics and the PFS time-to-event probability. Of 556 patients, 353 had detectable ctDNA at baseline. Evaluable patients (with available imaging and ≥ 3 additional timepoints, n = 320; ctDNA set) were divided into training (n = 259) and validation (n = 61) sets. In the validation set, the model predicted a median PFS of 17.7 months (95% confidence interval (CI): 11.9-28.3) for osimertinib (n = 23) and 9.1 months (95% CI: 6.3-14.8) for comparator (n = 38), consistent with observed RECIST PFS (16.4 months and 9.7, respectively). The model demonstrates that EGFRm ctDNA dynamics can predict the risk of disease progression in this patient population and could be used to predict RECIST-defined disease progression.

Imputing Biomarker Status from RWE Datasets-A Comparative Study.

Missing data is a universal problem in analysing Real-World Evidence (RWE) datasets. In RWE datasets, there is a need to understand which features best correlate with clinical outcomes. In this context, the missing status of several biomarkers may appear as gaps in the dataset that hide meaningful values for analysis. Imputation methods are general strategies that replace missing values with plausible values. Using the Flatiron NSCLC dataset, including more than 35,000 subjects, we compare the imputation performance of six such methods on missing data: predictive mean matching, expectation-maximisation, factorial analysis, random forest, generative adversarial networks and multivariate imputations with tabular networks. We also conduct extensive synthetic data experiments with structural causal models. Statistical learning from incomplete datasets should select an appropriate imputation algorithm accounting for the nature of missingness, the impact of missing data, and the distribution shift induced by the imputation algorithm. For our synthetic data experiments, tabular networks had the best overall performance. Methods using neural networks are promising for complex datasets with non-linearities. However, conventional methods such as predictive mean matching work well for the Flatiron NSCLC biomarker dataset.