Simple Summary: This paper presents a paradigm-shifted approach for adverse event (AE) anal. from classic descriptive summary into modern informative statistics to fulfill precision medicine.It defines early AE and derives a series of innovative AE biomarkers to assess grade, treatment relatedness, occurrence, frequency, and duration.Comprehensive data anal. generates opportunities for global discovery of predictive early AE-derived biomarkers.A proof of concept in two lung cancer studies demonstrates the potential clin. utility of early AE-derived biomarkers in patients with advanced non-small cell lung cancer treated with immunotherapy.The methodol. also provides a useful tool to help discover predictive AE biomarkers for treatment response and survival outcomes.Abstract: Rationale: Adverse events (AEs) have been shown to have clin. associations, in addition to patient safety assessments of drugs of interest.However, due to their complex content and associated data structure, AE evaluation has been restricted to descriptive statistics and small AE subset for efficacy anal., limiting the opportunity for global discovery.This study takes a unique approach to utilize AE-associated parameters to derive a set of innovative AE metrics.Comprehensive anal. of the AE-derived biomarkers enhances the chance of discovering new predictive AE biomarkers of clin. outcomes.Methods.We utilized a set of AE-associated parameters (grade, treatment relatedness, occurrence, frequency, and duration) to derive 24 AE biomarkers.We further innovatively defined early AE biomarkers by landmark anal. at an early time point to assess the predictive value.Statistical methods included the Cox proportional hazards model for progression-free survival (PFS) and overall survival (OS), two-sample t-test for mean difference of AE frequency and duration between disease control (DC: complete response (CR) + partial response (PR) + stable disease (SD)) vs. progressive disease (PD), and Pearson correlation anal. for relationship of AE frequency and duration vs. treatment duration.Two study cohorts (Cohort A: vorinostat + pembrolizumab, and B: Taminadenant) from two immunotherapy trials in late-stage non-small cell lung cancer were used to test the potential predictiveness of AE-derived biomarkers.Data from over 800 AEs were collected per standard operating procedure in a clin. trial using the Common Terminol. Criteria for Adverse Events v5 (CTCAE).Clin. outcomes for statistical anal. included PFS, OS, and DC.Results: An early AE was defined as event occurrence at or prior to day 30 from initial treatment date.The early AEs were then used to calculate the 24 early AE biomarkers to assess overall AE, each toxicity category, and each individual AE.These early AE-derived biomarkers were evaluated for global discovery of clin. associationBoth cohorts showed that early AE biomarkers were associated with clin. outcomes.Patients previously experienced with low-grade AEs (including treatment related AEs (TrAE)) had improved PFS, OS, and were associated with DC.The significant early AEs included low-grade TrAE in overall AE, endocrine disorders, hypothyroidism (pembrolizumab's immune-related adverse event (irAE)), and platelet count decreased (vorinostat related TrAE) for Cohort A and low-grade AE in overall AE, gastrointestinal disorders, and nausea for Cohort B.In contrast, patients with early development of high-grade AEs tended to have poorer PFS, OS, and correlated with PD.The associated early AEs included high-grade TrAE in overall AE, gastrointestinal disorders with two members, diarrhea and vomiting, for Cohort A and high-grade AE in overall AE, three toxicity categories, and five related individual AEs for Cohort B.One low-grade TrAE, alanine aminotransferase increased (vorinostat + pembrolizumab related), was an irAE and correlated with worse OS in Cohort A.Conclusions: The study demonstrated the potential clin. utility of early AE-derived biomarkers in predicting pos. and neg. clin. outcomes.It could be TrAEs or combination of TrAEs and nonTrAEs from overall AEs, toxicity category AEs, to individual AEs with low-grade event leaning to encouraging effect and high-grade event to undesirable impact.Moreover, the methodol. of the AE-derived biomarkers could change current AE anal. practice from a descriptive summary into modern informative statistics.It modernizes AE data anal. by helping clinicians discover novel AE biomarkers to predict clin. outcomes and facilitate the generation of vast clin. meaningful research hypotheses in a new AE content to fulfill the demands of precision medicine.