Cisplatin has been one of the most widely used anticancer agents, but its nephrotoxicity remains a dose-limiting complication. Here, we evaluated the idiopathic nature and the predose prediction of cisplatin-induced nephrotoxicity using a nuclear magnetic resonance (NMR)-based pharmacometabonomic approach. Cisplatin produced serious toxic responses in some animals (toxic group), but had little effect in others (nontoxic group), as judged by hematological and histological results. The individual metabolic profiles, assessed by urine NMR spectra, showed large differences between the post-administration profiles of the two groups, indicating the relevance of the NMR approach. Importantly, multivariate analysis of the NMR data showed that the toxic and nontoxic groups can be differentiated based on the pretreatment metabolite profiles. Leave-one-out analysis, performed to evaluate the practical performance of our approach, gave a 66% accuracy rate in predicting toxic responses based on the pretreatment metabolite profiles. Hence, we provide a working model that can explain the idiopathic toxicity mechanism based on marker metabolites found by NMR analysis consistent with tissue NADH measurements. Thus, a pharmacometabonomic approach using pretreatment metabolite profiles may help expedite personalized chemotherapy of anticancer drugs.
Bibliographical noteFunding Information:
This work was supported by grants from the Korea Healthcare technology R&D Project, Ministry for Health, Welfare and Family Affairs, Republic of Korea (A084338) and (A092006).
All one-dimensional NMR spectra of the urine samples were measured with an NMR spectrometer (Bruker Biospin, Rheinstetten, Germany, Avance 500) operating at a proton NMR frequency of 500.13 MHz. The acquisition parameters were essentially the same as previously reported. 37,55,56 For structural analysis of the metabolites, ultra high field, two-dimensional NMR spectra were acquired using a 900 MHz Bruker Avance spectrometer equipped with a cryogenic probe. For proton correlations, double quantum filtered correlation spectroscopy data were obtained in phase sensitive mode. The data set comprises 1024 × 192 complex points for the direct and indirect dimensions, respectively. For proton and carbon long-range correlation, a heteronuclear multiple-bond correlation spectrum was measured in magnitude mode. The dataset comprises 1024 complex points for direct and 512 real points for indirect dimension. The one-bond proton carbon correlation was measured with a heteronuclear single-quantum coherence spectrum. The dataset comprises 1024 × 192 complex points for the direct and indirect dimensions, respectively. The spectra were referenced against the trimethylsilylpropionic acid sodium salt-d 4 signal. All datasets were processed with nmrPipe software (NMR Science, North Potomac, MD) and analyzed with nmrView software (One Moon Scientific, Westfield, NJ). We also used Chenomx (Spectral database; Edmonton, Canada) for identification of the metabolites. This study used the NMR facility at Korea Basic Science Institute, which is supported by Bio-MR Research Program of the Korean Ministry of Science and Technology (E29070).
ASJC Scopus subject areas