Research Of Metformin Accelerates Nrf2 Degrading To Improve Chemotherapeutic Effects In Lung Cancer And Constructing The Prediction Model Of Adverse Events

BackgroundNon-small cell lung cancer(NSCLC)is responsible for more deaths worldwide than any other malignancy,the prognosis of locally advanced patients is poor.Platinum-based systemic chemotherapy is still the cornerstone in the treatment of resectable patients who undergo neoadjuvant or adjuvant chemotherapy even in the era of immunotherapy.Our study find that the efficacy of chemotheray and prognosis of of locally advanced patients are limited by the cisplatin-resistance of NSCLC and various adverse events(AEs).Searching for low toxic drugs to overcome chemoresistance and accurately predicting AEs can greatly enhance the efficacy of chemotherapy.In NSCLC,hyperactivation of nuclear factor erythroid 2-related factor 2(Nrf2)counteracts oxidative stress and scavenges ROS to promote chemoresistance.Nowadays,recent studies tend to combine metformin with other treatment strategies in view of its anti-cancer effects.Our research teams found that metformin-mediated downregulation of Nrf2 plays a pivotal role in overcoming drug resistance in NSCLC cells.Therefore,a deeper understanding of the molecular mechanisms of combination therapy and the role of Nrf2 in chemotherapeutic response is critical to clinical translation.Additionally,as disciplines cross over,machine learning models have been applied to predict the risk of disease,which means that the tools that use high-dimensional data of chemotherapy to predict the adverse drug reactions of next chemo-cycle will become a reality.Aims1.To explore the adverse prognostic factors in lung cancer patients with neoadjuvant chemotherapy,and to clarify the Nrf2 fluctuating pattern and the influence of multi-AEs on efficacy of chemotherapy.2.To explore the effectiveness of metformin in inhibiting NSCLC Nrf2 fluctuation as a synergistic target and its specific molecular mechanism.3.To build machine learning-based AEs predition model and evaluate its performance.Methods1.In order to determine the relationship between objective response rate(ORR)of chemotherapy and AEs,and the impact on the prognosis of patients,tumor samples and clinical characteristics of 50 neoadjuvant lung cancer patients were collected to evaluate the objective response by RECIST 1.1 standard,and adverse reaction events were determined by CTCAE 5.0.Immunohistochemical(IHC)staining was performed to detect Nrf2 expression in matched tumor samples before and after neoadjuvant chemotherapy.The correlation between Nrf2 variation,objective response rate and prognosis were further analyzed.2.The synergistic effects of combination therapy with metformin and cisplatin on NSCLC and Nrf2 fluctuating in vitro and in vivo were evaluated by cell viability,cell apoptosis assay,combination index(CI)calculating and cisplatin-resistance xenograft tumor formation.The expression of Nrf2/HO-1 were up-regulated and down-regulated by gene overexpression or knockdown to observe the proliferation,apoptosis and changes in mitochondrial ROS or other phenotypes of NSCLC cells,so as to clarify the anti-tumor effect mediated by Nrf2 axis.Western blot and co-immunoprecipitation(co-IP)were used to investigate the changes of Nrf2 ubiquitin and the effect of Ras-Raf-ERK on Keapl/Nrf2/HO-1 pathway under the treatment of metformin.GPS 5.0 phosphorylation site prediction software and Phos-tag assays were used to identify potential binding sites of phosphorylation on Nrf2 and its regualting model.3.A total of 1659 chemotherapeutic information data of 403 lung cancer patients were extracted from Electronic Health Record(EHR)system,which included baseline features,lung cancer features,features of chemotherapeutic agents,blood markers features,and interventions for adverse reactions.Then we used machine learning algorithm to build a prediction model of adverse reactions.The experiment was carried out with 10 five-fold cross-validation,and the ROC curve,AUC value and calibration curve were used to evaluate the performance of the model.Results1.Patients with multi-AEs or SD objective response had poor prognosis(p<0.05,log-rank test).Patients with multi-AEs showed a difference in the efficacy of chemotherapy compared with those who did not(p<0.001,Fisher’s exact test).The decreased Nrf2 expression in patients after neoadjuvant chemotherapy had better ORR(p<0.05)and better overall survival(p<0.05).2.Metformin inhibited cisplatin-induced Nrf2/HO-1 axis activation both in vitro and in vivo.Metformin and cisplatin synergistically suppressed cell proliferation,enhanced the induction of apoptosis,inhibited cisplatin-resistant NSCLC growth in an xenograft tumor model and markedly increased the ROS accumulation in NSCLC cells which promoted the cisplatin-induced oxidative burst.The synergistic antitumor effect and ROS accumulation of combination therapy is blocked by treatment with the ROS scavenger N-acetyl cysteine(NAC)as well as overexpression of Nrf2/HO-1 axis.The sensitivity and cytotoxicity of the NSCLC cells to cisplatin was remarkably enhanced upon Nrf2 knockdown or suppression of HO-1 activity in NSCLC cells.3.U0126,a highly MEK1/2 inhibitor,strongly suppressed the expression of Nrf2.Metformin inactivated the Ras/Raf/ERK pathway,weakened the interaction between ERKI/2 and Nrf2,increased Nrf2 ubiquitination level,promoted the transition of phosphorylated Nrf2 to the non-phosphorylated state.Nrf2 ubiquitin modification was increased after extensive dephosphorylation.ERK/2 is a potential phosphorylated kinase of Nrf2,and mutations at key amino acid sites may have effect on Nrf2 ubiquitination modification.4.We used algorithm of random forest,multi-layer perceptron,adaboost and logistics regression to establish four prediction models of adverse drug reactions.The random forest model had achieved the best performance in the task of predicting the side effects of bone marrow suppression,cachexia,and liver function damage.Its AUC values are 0.75,0.74 and 0.76,respectively,and there was an optimal calibration curve.Conclusions1.Chemoresistance and chemo-associated AEs limit the efficacy of patients with lung cancer,in which the Nrf2 fluctuating pattern is related to the ORR and prognosis of neoadjuvant patients.The occurrence of multi-AEs also limit the chemo-efficacy and prognosis of lung cancer patients.2.Metformin reduces the Nrf2 fluctuating level and synergically enhances the cytotoxicity of cisplatin in NSCLC.The synergistic effect is related to metformin weakens the detoxification ability of Nrf2/HO-1 axis on cisplatin and enhances ROS-mediated apoptosis.Metformin inhibits the interaction of Nrf2 by the kinase ERK1/2,thereby restoring the ubiquitination modification of Nrf2 and ultimately promoting the degradation of Nrf2.3.The prediction model constructed by machine learning algorithm has good prediction performance.It is recommended to develop tools based on random forest algorithm to predicte AEs of chemotherapy,thus improving the efficacy of chemotherapy.