Longitudinal Pharmacometabonomics for Predicting Malignant Tumor Patient Responses to Anlotinib Therapy: Phenotype, Efficacy, and Toxicity.

Abstract

Anlotinib is an oral small molecule inhibitor of multiple receptor tyrosine kinases (RTKs), which was approved by the National Medical Products Administration (NMPA) of China in 2018 for the third-line treatment of non-small cell lung cancer (NSCLC). Here, for the first time, the longitudinal pharmacometabonomics was explored for predicting malignant tumor patient responses to anlotinib, including the metabolic phenotype variation, drug efficacy, and toxicity. A total of 393 plasma samples from 16 subjects collected from a phase I additional study of anlotinib (NCT02752516) were submitted to targeted metabolomics analysis. The orthogonal partial least-squares discriminant analysis (OPLS-DA) models were constructed for the predication of anlotinib efficacy and toxicity based on the longitudinal pharmacometabonomics data. Statistical results showed that 38 metabolites, mainly involved in aminoacyl-tRNA biosynthesis, alanine, aspartate, and glutamate metabolism, and steroid hormone biosynthesis, were all significantly upregulated attributing to anlotinib treatment. The anti-tumor efficacy and occurrence of proteinuria after anlotinib administration can be predicted with 100% accuracy using the established OPLS-DA models. Glycodeoxycholic acid and glycocholic acid possessed the most excellent sensitivity and specificity in predicting the efficacy of anlotinib, with area under the receiver operating characteristic curve (AUC of ROC curve) 0.847 and 0.828, respectively. NG, NG-dimethylarginine was the most promising biomarker for the prediction of proteinuria occurrence after anlotinib administration, with AUC of ROC curve 0.814. In conclusion, this work developed efficient and convenient discriminant models that can accurately predict the efficacy and toxicity of anlotinib based on longitudinal pharmacometabonomics study.