Metabolomics analysis in acute paraquat poisoning patients based on

Subscriber access provided by WEBSTER UNIV

Article

Metabolomics analysis in acute paraquat poisoning patients based on UPLC-Q-TOF-MS and machine learning approach Congcong Wen, Feiyan Lin, Binge Huang, Zhiguang Zhang, Xianqin Wang, Jianshe Ma, Guanyang Lin, Huiling Chen, and Lufeng Hu Chem. Res. Toxicol., Just Accepted Manuscript • DOI: 10.1021/acs.chemrestox.8b00328 • Publication Date (Web): 26 Feb 2019 Downloaded from http://pubs.acs.org on February 27, 2019

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Chemical Research in Toxicology

Metabolomics analysis in acute paraquat poisoning patients based on UPLC-Q-TOF-MS and machine learning approach

Congcong Wen1#, Feiyan Lin2#, Binge Huang1, Zhiguang Zhang1, Xianqin Wang1, Jianshe Ma1, Guanyang Lin3, Huiling Chen4∗, Lufeng Hu3*

1 Analytical and Testing Center of Wenzhou Medical University, Wenzhou 325035, China, Wenzhou 325035, China 2 Central laboratory, The First Affiliated Hospital of Wenzhou Medical University, Wenzhou, Zhejiang, China 3. Department of pharmacy, The First Affliated Hospital of Wenzhou Medical University, Wenzhou 325000, China 4 College of Physics and Electronic Information Engineering, Wenzhou University, Wenzhou 325035, China

# These authors contributed equal to this work.  Corresponding Author: Lufeng Hu and Huiling Chen Huiling Chen: College of Physics and Electronic Information, Wenzhou University, Wenzhou, Zhejiang, 325035, China. E-mail: [email protected] Lufeng Hu: Department of pharmacy, The First Affliated Hospital of Wenzhou Medical University, Wenzhou 325000, China 1

ACS Paragon Plus Environment

Chemical Research in Toxicology 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Abstract: Most of paraquat (PQ) poisoned patients died for acute multiple organ failure (MOF) such as lung, kidney, and heart. However, the exact mechanism of intoxication is still unclear. In order to find out the initial toxic mechanism of PQ poisoning, a blood metabolomics study based on ultra-performance liquid chromatography coupled to quadrupole time-of-flight mass spectrometry (UPLC-Q-TOF-MS) and efficient machine learning approach was performed on 23 PQ poisoned patients and 29 healthy subjects. The initial PQ plasma concentrations of PQ poisoned patients were > 1000 ng/mL, and the blood samples were collected at before first hemoperfusion (HP), after first HP, and after last HP. The results showed that PQ poisoned patients all differed from healthy subjects, whatever they were before or after first HP or after last HP. The efficient machine learning approaches selected key metabolites from three UPLC/Q-TOF-MS datasets which had highest classification performance in terms of classification accuracy, Matthews Correlation Coefficients, sensitivity and specificity, respectively. The mass identification revealed that the most important metabolite was adenosine, which sustained in low level, regardless of whether PQ poisoned patients received HP treatment. In conclusion, decreased adenosine was the most important metabolite in PQ poisoned patients. The metabolic disturbance caused by PQ poisoning can't be improved by HP treatment even the PQ was cleared from the blood.

Keywords: Paraquat; Patient; Metabonomics; Machine learning; Adenosine

2


Page 2 of 23

Page 3 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


1. Introduction Paraquat (PQ; 1,1-dimethyl-4,4-bipyridinium dichloride) is a widely used herbicidewith highly toxic to human. Acute ingestion of PQ and serum concentration over 5000 ng/mL, the mortality is essentially 100% within 1-7 days 1. In order to reduce mortality, various kinds of studies have been carried out for PQ poisoning. Only in 2018, there over 100 articles of PQ have been published around the world. For example, Gawarammana et al reported a randomized controlled trial about high-dose immunosuppression to prevent death after PQ self-poisoning 2. He et al reported mesenchymal stem cell showed anti-fibrosis therapeutic effects in animal models of lung injury caused by PQ poisoning 3. So far, it has been acknowledged that the injury of PQ poisoning is nonselective. Ingestion of PQ can lead to inflammation and ulceration of the mouth, throat and gastrointestinal tract. After absorbed, PQ will result in multiple organ function failure, especially lung, kidney, and liver

4-6.

As for the mechanism, oxidative stress plays an

important role in toxic pathogenesis of PQ poisoning 7. The increased level of reactive oxygen species (ROS), excessive consumption of NADPH will result in cellular injury such as lipid peroxidation and mitochondrial dysfunction, by the generation of superoxide anions and other free radicals

8-10.

However, the exact pathogenetic

mechanism of multiple organ failure and the effective treatment methods is still unclear. PQ poisoning is still the hot and difficult issues in research. 3



Machine learning is a kind of computer science, which has been widely used in medical area for prediction, classification, and diagnosis. For example, Lei et al. proposed a multimodal algorithm to enhance the prediction performance of protein-protein interactions via multimodal deep polynomial network and extreme learning machine (ELM)

11.

Li et al. applied a support vector machine (SVM) to

develop a predictive model to verify and optimize the exclusive biomarkers 12. ELM and SVM approaches also can achieve high accuracy in diagnosis of PQ poisoned patients by using metabolic data 13, 14. In order to comprehensively investigate the mechanism of PQ poisoning, we carried out a metabolomics study of UPLC-Q-TOF-MS and introduced feature selection approachcombined with PCA, PLS-DA for data analysis. Feature selection, also named variable selection, is the process of developing a subset of relevant variables or features for model construction15. In this study, we applied fisher score, a kind of feature selection method, to rank the weight of each metabolite and develop a subset for ELM model. When the key metabolites were selected by machine learning method and PLS-DA, an animal study was carried out for further investigation.

2. Methods 2.1 Ethics statement The blood samples of PQ poisoning patients admitted to the First Affiliated Hospital of Wenzhou Medical University from January 1, 2015 to December 30, 2017 were collected. Only the patients with serum PQ >1000 ng/mL and received HP treatment were involved in the metabolomics study. The samples of healthy subjects were collected from physical examination center of the First Affiliated Hospital of Wenzhou Medical University. This study was approved by the Medical Ethics Committee of the First Affiliated Hospital of Wenzhou Medical University (Register number: 2016153), China. All clinical examinations and data collection were conducted in accordance with the Declaration of Helsinki.

2. 2. Instruments and chemicals A Waters ACQUITY Ultra Performance LC system equipped a hybrid 4


Page 4 of 23

Page 5 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


quadrupole orthogonal time of flight (Q-TOF) mass spectrometer and a Waters ACQUITY Ultra Performance LC system equipped a Xevo TQ-S Micro triple quadrupole mass spectrometer (Waters Co., Milford, MA, USA) was used for metabolomics study and determination of PQ and adenosine. PQ, adenosine and isoniazide (purity >98%) were purchased from Sigma-Aldrich (St Louis, MO). High performance liquid chromatography (HPLC) grade acetonitrile, methanol and formic acid were purchased from Merck Company (Darmstadt, Germany). Distilled water was purified using a Milli-Q system (Millipore Bedford, MA).

2.3 Study design of UPLC/Q-TOF-MS Sample collection The blood samples of PQ poisoning patients were collected at before first HP, after first HP, and after last HP. The samples of before first HP were collected at the PQ poisoned patients arrived at the emergency room and didn’t received any treatment. The samples after last HP were collected after the PQ concentration lower than < 200 ng/mL. All serum samples were stored under -80oC until for metabonomics analysis. The serum concentrations of PQ were determined by our previous HPLC method, which has been successfully applied in PQ determination 1.The complete blood count (CBC) and biochemical tests of PQ poisoned patient were collected before first HP.

UPLC/Q-TOF-MS assay A total of 100µL serum samples were precipitated by 300µL acetonitrile, then the mixture was vigorously whirled and centrifuged at 14,000×g for 10 min at 4oC. After that, the supernatants were transferred to sample bottles for UPLC/Q-TOF-MS analysis. The control (QC) samples were prepared by withdrawing 10µL of serum from each sample and precipitated with 300 µL acetonitrile and tested ten times to condition the column. An ACQUITY UPLC BEH C18 analytical column (i.d.2.1 mm × 100 mm, Waters, Milford, MA) was used and the column temperature maintained at 40 oC. The 5



analytical condition of UPLC/Q-TOF-MS set at a gradient elution, which was water/formic acid (99.9:0.1, v/v, mobile phase A) and methanol/ formic acid (99.9: 0.1, v/v, mobile phase B). The injection volume was 2 μL. Mass spectrometry was carried out on a Waters Q-TOF Premier mass spectrometer operating in positive ion electrospray mode with electro spray ionization (ESI) source. For more information, please refer to our previous works 16, 17. The raw UPLC/Q-TOF-MS data including retention time, m/z and signal intensity of the peaks, was firstly normalized by MarkerLynx Applications Manager Version 4.1 (Waters, Milford, MA, USA). The principal components analysis (PCA) and projections to latent structures discriminant analysis (PLS-DA) were used to evaluate the difference between PQ poisoned patients and healthy subjects and to select the importance variables by using SIMCA-P+ 12.0 software (Umetrics AB, Sweden).

2.4. Machine learning approach Processing flowchart The first stage is mainly to obtain data, mainly using UPLC-Q-TOF-MS for metabolic analysis of blood from PQ poisoned patients and normal controls, and then obtain the blood metabolites. The main purpose of the second phase is to screen for the key substances in the metabolic data obtained during the previous phase. After normalizing the data, the Fisher Score method was used to sort the metabolites, and then the incremental combinations were used to construct an optimal subset of features. The main goal of the third stage is to carry out the model training on the optimal feature subset obtained in the previous stage. In this study, a fast and efficient ELM model is used for modeling. The main task of the final stage is to predict new unknown samples and make a final decision. The whole data processing flowchart is shown in figure 1. Figure 1.

Experimental setup The involved methods including Fisher Score and ELM were implemented using 6


Page 6 of 23

Page 7 of 23 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60


MATLAB. The computational analysis was conducted on a Windows Server2008 R2 operating system with Intel (R) Xeon (R) CPU E5-2660 v3 (2.60 GHz) and 16GB of RAM. Data was scaled into the range [-1, 1] before classification. The 10-fold cross validation scheme was used to evaluate classification performance to guarantee unbiased results. To evaluate the proposed method, the commonly used evaluation criteria such as classification accuracy (ACC), sensitivity, specificity and Matthews Correlation Coefficients (MCC) were analyzed. The formulas were listed in table S1.

2.5. Determination of PQ and adenosine Analytical conditions Chromatographic separation of PQ and adenosine was carried out at hilic C18 column

(2.1 mm × 100 mm, 1.7 μm)

by ACQUITY UPLC and Xevo TQ-S Micro

triple quadrupole mass spectrometer (UPLC-MS,Waters Corporation, USA). The temperature of C18 column was 40°C and the mobile phase consisted of 10 mM ammonium acetate-0.1% formic acid in water (A) and methanol (B). The gradient was eluted as follows: (Tmin/acetonitrile): 0.0-1.0/90% (B), 1.0-2.5/10% (B) and 2.5-3.0/90% (B), the flow rate was set at 0.3 mL/min. The injection volume was 2 μL. MS/MS conditions of PQ, adenosine and IS were 187.13/172.05, 268.15/136.12, and 138/121, the collision energies were 60 eV, 20 eV, and 25 eV, which was detected at the multiple reaction monitoring (MRM) positive mode.

Blood level of adenosine In order to investigate the blood level of adenosine after PQ poisoning, three Sprague Dawley rats (220±20 g) were given PQ by oral administration 50mg/kg, after that 0.3mL blood sample was collected from the caudal vein at 0, 0.5, 1, 2, 4, 6, 8, 10, 12, 24, 48h. The blood samples were centrifuged immediately and 100 µL plasma extracted and stored at -80oC. The blood levels of PQ and adenosine were determined as follows. A total of 100 µL rat plasma was thawed at room temperature and precipitated with 300µL acetonitrile which contained 0.3 μg/mL IS. The mixtures were shocked for 0.5min and centrifuged at 15000 rmp for 10min and 2 μL supernatant was injected 7



into the UPLC-MS system and determined according to the analytical conditions. The calibration curves of PQ and adenosine were constructed over the range of 6.4-6400 ng/mL and 1-1000ng/mL. Quality-control (QC) samples of PQ were prepared at 64, 320, 3200, and adenosine at 10, 50, and 500 ng/mL. The inter-day precision of PQ and adenosine were evaluated at three QC levels with three replicates. The relative recovery was assessed by contrasting the concentration of QC samples with the concentration calculated by calibration curve. The absolute recovery (extraction recovery) was assessed by contrasting the peak areas of extracted QC samples with pure QC samples at same concentration.

3 Results 3.1 Information of subjects A total of 52 subjects were included in this study. There were twenty-three patients in the PQ group, and twenty-nine in the healthy control group. The first CBC test showed there were significant differences between two groups (Table 1). In PQ poisoned patients, the first PQ serum concentrations (before first HP) ranged from 2244 ng/mL to 410055 ng/mL, the second PQ concentrations (after first HP) ranged from 343 ng/mL to 137788 ng/mL, the last PQ concentrations (after last HP) were all

Metabolomics analysis in acute paraquat poisoning patients based on

Recommend Documents