Dataset Information

Elucidation of the Application of Blood Test Biomarkers to Predict Immune-Related Adverse Events in Atezolizumab-Treated NSCLC Patients Using Machine Learning Methods

ABSTRACT:

Background

Development of severe immune-related adverse events (irAEs) is a major predicament to stop treatment with immune checkpoint inhibitors, even though tumor progression is suppressed. However, no effective early phase biomarker has been established to predict irAE until now.

Method

This study retrospectively used the data of four international, multi-center clinical trials to investigate the application of blood test biomarkers to predict irAEs in atezolizumab-treated advanced non-small cell lung cancer (NSCLC) patients. Seven machine learning methods were exploited to dissect the importance score of 21 blood test biomarkers after 1,000 simulations by the training cohort consisting of 80%, 70%, and 60% of the combined cohort with 1,320 eligible patients.

Results

XGBoost and LASSO exhibited the best performance in this study with relatively higher consistency between the training and test cohorts. The best area under the curve (AUC) was obtained by a 10-biomarker panel using the XGBoost method for the 8:2 training:test cohort ratio (training cohort AUC = 0.692, test cohort AUC = 0.681). This panel could be further narrowed down to a three-biomarker panel consisting of C-reactive protein (CRP), platelet-to-lymphocyte ratio (PLR), and thyroid-stimulating hormone (TSH) with a small median AUC difference using the XGBoost method [for the 8:2 training:test cohort ratio, training cohort AUC difference = −0.035 (p < 0.0001), and test cohort AUC difference = 0.001 (p=0.965)].

Conclusion

Blood test biomarkers currently do not have sufficient predictive power to predict irAE development in atezolizumab-treated advanced NSCLC patients. Nevertheless, biomarkers related to adaptive immunity and liver or thyroid dysfunction warrant further investigation.

SUBMITTER: Zhou J

PROVIDER: S-EPMC9284319 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:BackgroundIn the context of patients with hepatocellular carcinoma (HCC) treated with systemic therapy, the correlation between the appearance of adverse events (AEs) and reported efficacy outcomes is well-known and widely investigated. From other pathological settings, we are aware of the prognostic and predictive value of the occurrence of immune-related AEs in patients treated with immune-checkpoint inhibitors.ObjectiveThis retrospective multicenter real-world study aims to investigate the potential prognostic value of AEs in patients with HCC treated with atezolizumab plus bevacizumab in the first-line setting.Patients and methodsThe study population consisted of 823 patients from five countries (Italy, Germany, Portugal, Japan, and the Republic of Korea).ResultsOf the patients, 73.3% presented at least one AE during the study period. The most common AEs were proteinuria (29.6%), arterial hypertension (27.2%), and fatigue (26.0%). In all, 17.3% of the AEs were grade (G) 3. One death due to bleeding was reported. The multivariate analysis confirmed the appearance of decreased appetite G < 2 [versus G ≥ 2; hazard ratio (HR) 0.60; 95% confidence interval (CI) 0.13-0.90; p < 0.01] and immunotoxicity G < 2 (versus G ≥ 2; HR: 0.70; 95% CI 0.24-0.99; p = 0.04) as independent prognostic factors for overall survival, and the appearance of decreased appetite G < 2 (versus G ≥ 2; HR: 0.73; 95% CI 0.43-0.95; p = 0.01), diarrhea (yes versus no; HR: 0.57, 95% CI 0.38-0.85; p = 0.01), fatigue (yes versus no; HR: 0.82, 95% CI 0.65-0.95; p < 0.01), arterial hypertension G < 2 (versus G ≥ 2; HR: 0.68, 95% CI 0.52-0.87; p < 0.01), and proteinuria (yes versus no; HR: 0.79, 95% CI 0.64-0.98; p = 0.03) as independent prognostic factors for progression-free survival.ConclusionsAs demonstrated for other therapies, there is also a correlation between the occurrence of AEs and outcomes for patients with HCC for the combination of atezolizumab plus bevacizumab.

Project description:ImportanceImmune-related adverse events (irAEs) secondary to immune checkpoint inhibitor (ICI) therapy reportedly improve overall survival (OS) in patients with non-small cell lung cancer (NSCLC). However, studies have been small and the association between irAE severity and OS remains poorly defined.ObjectiveTo examine the association between irAEs and their severity with OS in patients with locally advanced or metastatic NSCLC receiving ICIs.Design, setting, and participantsThis retrospective observational cohort study included patients with NSCLC receiving ICIs between March 1, 2014, and November 30, 2021, with follow-up until March 31, 2023. Data analysis was completed April 26, 2023. The Alberta Immunotherapy Database, a provincial, multicenter cohort, was used to capture data from patients receiving ICIs in Alberta, Canada. Participants included 803 patients 18 years or older who received at least 1 cycle of ICI (alone or with chemotherapy), agnostic to treatment line.ExposureDeveloping an irAE mandating delay or discontinuation of ICI therapy and/or systematic corticosteroids for management of toxic effects (hereinafter referred to as clinically meaningful irAEs).Main outcomes and measuresThe primary outcome was association between irAEs and OS according to Kaplan-Meier analysis. Clinically meaningful irAEs were identified. Patients with poor prognosis (survival <3 months) who may have died prior to irAE development were excluded from OS analysis, mitigating immortal time bias. Adjusted Cox proportional hazards regression analyses ascertained variables associated with OS.ResultsAmong the 803 patients included in the analysis, the median age of patients with irAEs was 69.7 (IQR, 63.1-75.2) years and the median age of those without irAEs was 67.5 (IQR, 60.4-73.3) years, with comparable sex distribution (139 of 295 men [47.1%] and 156 of 295 women [52.9%] with irAEs vs 254 of 505 men [50.3%] and 251 of 505 women [49.7%] without irAEs). Mitigating immortal time bias (n = 611), irAEs were associated with OS (median OS with irAEs, 23.7 [95% CI, 19.3-29.1] months; median OS without irAEs, 9.8 [95% CI, 8.7-11.4] months; P < .001). No OS difference was associated with treatment in hospital vs as outpatients for an irAE (median OS, 20.8 [95% CI, 11.7-30.6] vs 25.6 [95% CI, 20.1-29.8] months; P = .33). Developing irAEs remained associated with OS in the total cohort after Cox proportional hazards regression with known prognostic characteristics (hazard ratio, 0.53 [95% CI, 0.40-0.70]; P < .001).Conclusions and relevanceIn this cohort study of 803 patients with locally advanced or metastatic NSCLC receiving ICIs, developing a clinically meaningful irAE was associated with improved OS. This association was not compromised by hospitalization for severe toxic effects. Whether and how ICI therapy resumption after an irAE is associated with OS warrants further study.

Project description:BackgroundAt present, there is a lack of studies focusing on the survival prediction of patients with non-small cell lung cancer (NSCLC) receiving atezolizumab in light of gene mutation characteristic.MethodsPatients with NSCLC receiving atezolizumab from the OAK study were defined as the training group. LASSO Cox regressions were applied to establish the gene mutation signature model to predict the overall survival (OS) rate of the training group. NSCLC patients receiving atezolizumab from the POPLAR study were defined as the testing group to validate the gene mutation signature model. In addition, we compared the OS rate between patients receiving atezolizumab and docetaxel classified according to their risk score based on our gene mutation signature model.ResultsWe successfully established a 5-genomic mutation signature that included CREBBP, KEAP1, RAF1, STK11 and TP53 mutations. We found it was superior to the blood tumor mutation burden (bTMB) score and programmed death ligand 1 (PDL1) expression in the prediction of the OS rate for patients receiving atezolizumab. High-risk patients receiving atezolizumab had a worse OS rate compared with low-risk patients in the training (P = 0.0004) and testing (P = 0.0001) groups. In addition, low-risk patients using atezolizumab had a better OS rate compared with those in use of docetaxel for the training (P <0.0001) and testing groups (P = 0.0095). High-risk patients of the training group (P = 0.0265) using atezolizumab had a better OS rate compared with those using docetaxel. However, the OS difference between atezolizumab and docetaxel was not found in high-risk patients from the testing group (P = 0.6403). Multivariate Cox regression analysis showed that the risk model in light of 5-genomic mutation signature was an independent prognostic factor on OS for patients receiving atezolizumab (P <0.0001). In addition, significant OS benefit could only be found in low-risk patients receiving atezolizumab compared with docetaxel (P <0.0001).ConclusionsThe 5-genomic mutation signature could predict OS benefit for patients with NSCLC receiving atezolizumab. Therefore, the establishment of the 5-genomic mutation panel will guide clinicians to identify optimal patients who could benefit from atezolizumab treatment.