• Users Online: 633
  • Home
  • Print this page
  • Email this page
Home About us Editorial board Ahead of print Current issue Search Archives Submit article Instructions Subscribe Contacts Login 


 
 Table of Contents  
ORIGINAL ARTICLE
Year : 2019  |  Volume : 5  |  Issue : 2  |  Page : 80-86

A gas chromatography–Mass spectrometry-based metabonomic study on estimation of toxicant in rats


Department of Forensic Toxicological Analysis, The West China School of Basic Medical Sciences and Forensic Medicine, Sichuan University, Chengdu, Sichuan, China

Date of Web Publication26-Jun-2019

Correspondence Address:
Linchuan Liao
The West China School of Basic Medical Sciences and Forensic Medicine, Sichuan University, Chengdu, Sichuan
China
Login to access the Email id

Source of Support: None, Conflict of Interest: None


DOI: 10.4103/jfsm.jfsm_4_19

Rights and Permissions
  Abstract 


The aim of this study was to develop a gas chromatography-mass spectrometry (GC-MS)-based metabolomics method to distinguish different kinds of poisons in the blood. We examined the changes in blood metabolites using GC-MS following administration of four different poisons (paraquat, dichlorvos, aconitine, and sodium nitrite). The data were analyzed with orthogonal partial least squares. Then, total and single differential metabolite profiles were evaluated with support vector machine (SVM) models. The results showed that various metabolites (5-ketone proline, 1,5-anhydrohexitol, lactic acid, glycine 2,2-furoic acid, and 3-hydroxybutyric acid) were differential between the experimental groups and the control groups. The accuracy rates of the SVM models established using total and single differential metabolites were 80% and 100%, respectively. In conclusion, we successfully developed a poison screening method. The established SVM models can distinguish four kinds of poisons and could be used to establish a complete poison metabonomic information database. Furthermore, some of the metabolites could be biomarkers of these poisons. Finally, both the models and potential biomarkers may reduce the time required for poison detection and provide direction for solving cases and auxiliary diagnosis.

Keywords: Aconitine, dichlorvos, gas chromatography-mass spectrometry, metabonomics, paraquat, sodium nitrite


How to cite this article:
Fu Y, Dai X, Chen F, Zheng Y, Fang S, Lin Y, Ye Y, Liao L. A gas chromatography–Mass spectrometry-based metabonomic study on estimation of toxicant in rats. J Forensic Sci Med 2019;5:80-6

How to cite this URL:
Fu Y, Dai X, Chen F, Zheng Y, Fang S, Lin Y, Ye Y, Liao L. A gas chromatography–Mass spectrometry-based metabonomic study on estimation of toxicant in rats. J Forensic Sci Med [serial online] 2019 [cited 2019 Jul 16];5:80-6. Available from: http://www.jfsmonline.com/text.asp?2019/5/2/80/261529

Yingqiang Fu, Xinhua Dai. These authors have contributed equally to this work





  Introduction Top


As a crucial laboratory component of forensic toxicology, poison screening currently uses modern instrumental analysis for qualitative or quantitative analyses of poisonous elements in vivo and in vitro and provides a scientific basis for criminal investigations and trials.[1],[2] Because the types of poisons are unknown in many cases, the nature of many important factors that could affect the accuracy and application of poison screening, such as the properties of different poisons, sample pretreatment methods, and instrumental analysis methods, are unclear. Moreover, some cases lack characteristic clinical manifestation, and some poisons produce the same symptoms, which makes it more difficult to identify the correct testing method, especially when cases lack details. In addition, some poisons are metabolized over time, and if sampling is untimely, the content of the sample may be below the detection limit and unable to be quantified.[3],[4] The poison content may be assessed qualitatively through metabolite changes or potential biomarkers. Under these circumstances, determination of the poison content may require a great deal of time and effort, causing errors and lowering the efficiency.

The physicochemical properties of different kinds of poisons are obviously different, which means the pretreatment methods and instrument detection methods used for biological samples also vary. Therefore, traditional screening methods cannot detect different poisons simultaneously.[5],[6] When poisons enter organisms, they disturb the internal homeostasis of the normal physiological state and change the metabolites within the body. Consequently, metabonomic profiling is becoming an impactful tool in this area as it provides a distinct perspective on characterization and understanding of the mechanisms of responses to poisons.[7],[8] In recent years, there have been many applications of metabonomic technology to observe metabolite changes in the body after poisoning. Wang et al. used a gas chromatography-mass spectrometry (GC-MS) metabonomic method to show that the levels of octadecanoic acid, l-serine, l-threonine, l-valine, and glycerol in an acute paraquat poisoning group increased in serum samples.[1] Yang et al. established a model of chronic low-dose dichlorvos poisoning in rats and used ultra performance liquid chromatography–MS to detect changes in plasma sphingosine, sphinganine, C16 sphinganine, C17 sphinganine, and arachidonic acid.[9] With the aid of GC-quadrupole-time-of-flight MS and nuclear magnetic resonance spectroscopy metabonomic technology, Sun et al. studied changes in the concentrations of glucose, acetate, dimethylglycine, succinate, and alanine in the plasma and urine of aconitine-poisoned rats.[10]

Previous studies have shown that high-throughput metabonomic detection technology combined with multivariate statistical analysis can help to clarify toxicity mechanisms by detecting in vivo metabolite changes induced by paraquat, dichlorvos, and aconitine poisoning in rats.[9],[10],[11] However, the overwhelming majority of research on screening of characteristic metabolites of poisons has concentrated on acute or chronic poisoning models established using a single poison, and different methods have been used to develop single poisoning models.

To increase the efficiency of toxicological screening, we aimed to develop a model using GC-MS metabonomic technology that could be used to infer what poisons were present in a sample. Considering the high occurrences of poisoning cases caused by paraquat, dichlorvos, aconitine, and sodium nitrite, we selected these four toxins to establish the model. In this research, rather than analyzing the plasma concentrations of different poisons, we focused on qualitative evaluation of blood metabolite variations induced by toxic reactions. These metabolites could be used to identify the poisons and reduce the number of instrumental analyses required. To a large extent, this method can reduce the time and cost of analysis.


  Materials and Methods Top


Animals

The Experimental Animal Center of Sichuan University (Chengdu, China) provided male Sprague Dawley (SD) rats (10 weeks old). The rats were divided into five experimental groups with 10 rats per group and one group with six rats for double-blind testing of the developed model. Animal care and the experimental procedures were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals.[12] The rats weighed 200.2 ± 7.7 g on arrival at the laboratory and were habituated for 7 days before the experiments. A constant temperature (23°C ± 3°C) and a 12-h light/dark cycle were maintained throughout the study. Food and water were available ad libitum.

Experimental reagents

An aqueous solution of paraquat (20% purity) was provided by ShunYi (Beijing, China). A dichlorvos emulsion (77.5% purity) was provided by RunDe (China). Aconitine (98.11% purity) was provided by RuiFenSi (Chengdu, China). Sodium nitrite (99% purity), methyl stearate (99.5% purity), acetonitrile, pyridine, and n-heptane were purchased from KNOWLES (Chengdu, China). Bis (trimethylsilyl) trifluoroacetamide + 1% trimethylchlorosilane was obtained from Regis (Beijing, China). Methoxylamine hydrochloride (98% purity) was purchased from Sigma-Aldrich (Chengdu, China).

A gas chromatograph–mass spectrometer equipped with a GC-MS-5975C triple-axis detector (Agilent Technologies, Santa Clara, CA) was used during this study. A DB-5 capillary column (30 m × 0.25 mm i.d., 0.25 μm film thickness, Agilent Technologies) was used for the chromatographic separation. The column oven temperature was maintained at 60°C for 1 min, increased by 10°C/min to 325°C, and held at 325°C for 10 min. The injection temperature and carrier gas (He) flow rate were set at 250°C and 1.0 mL/min, respectively. The split ratio for the injection was 5:1. For electron ionization, the ion source and quadrupole temperatures were 230°C and 150°C, respectively, and the electron energy was 70 eV. Mass spectra were collected from m/z 50–600 at scan rate of two spectra per se cond. GC/MS was conducted in scan mode.

Drug treatment

Experimental group

Half lethal doses for paraquat (150 mg/kg), dichlorvos (80 mg/kg), aconitine (1.8 mg/kg), and sodium nitrite (85 mg/kg) were administered orally in a volume of 1 mL to separate groups of rats (n = 10). An acute poisoning model for the four poisons was established by intragastric administration. The symptoms of poisoning and the time of death were observed and recorded after intragastric administration.

If the rats died after the gastric lavage, they were dissected immediately and blood samples were collected (0.5 mL per tube, heparin sodium anticoagulant). If the rats did not die, we sacrificed them 6 h after the gavage, as performed in previous studies,[13] and collected blood samples. All blood samples were frozen and stored at −80°C.

Blank group

The rats in the blank group (n = 10) were treated via intragastric administration with saline (1 mL) and then sacrificed 6 h later. The rats were dissected, and blood samples were collected and stored at −80°C.

Sample pretreatment

Blood samples

The blood samples were placed in a 4°C refrigerator for thawing. A 50-μL aliquot was pipetted into a 1.5-mL blood EP tube, and then, 150 μL of acetonitrile was added at − 20°C. After vortexing for 2 min, the mixture was centrifuged at 4°C and 12,000 rpm for 10 min. The supernatant (100 μL) was transferred to a 300-μL Du canaliculus glass tube, which was then placed in a rotary evaporator at 30°C to concentrate the residual liquid in preparation for derivatization. A 15 mg/mL solution (30 μL) pyridine solution of methoxyamine hydrochloride was added to the tube. Each tube was wrapped with aluminum foil and sealed with oxime at 16°C for 16 h. Then, 20 μL of the silane reagent bis (trimethylsilyl) trifluoroacetamide + 1% trimethylchlorosilane was added rapidly. The sealed tube was placed in an oven at 70°C for 2 min and then removed and cooled in the dark. After cooling, 100 μL of the internal standard working solution (methyl stearate heptane solution, 10 g/mL) was added. After vortexing for 2 min and static incubation for 10 min, the supernatant was analyzed by GC-MS.

Data preprocessing

The spectra obtained after injection were deconvoluted with AMDIS 6.51 software (Agilent Technologies, CA, USA), and the chromatographic peaks were qualitatively analyzed using the NIST 2014 (Agilent Technologies) MS library. The matching ratio with the MS library was >80%. All peak areas >10,000 were taken as the standard metabolite chromatographic peaks, and the metabolite data (i.e., metabolite name, retention time, and peak area) were obtained. The peak area of each metabolite in each sample was divided by the internal standard peak area for normalization.[14],[15]

Data analysis

Metabonomics and multivariate statistical analysis

The pretreated data were imported from SIMCA-P 14 software (Umetrics, Umea, Sweden) and analyzed using multivariate statistical analysis. The supervised pattern recognition method and partial least squares (PLS)-discriminant analysis were used to analyze the data, and classification effects for each experimental group and the blank group were observed. Then, multivariate statistical analysis of all experimental and blank group samples was performed, and classification of all samples was observed by the orthogonal PLS (OPLS) pattern recognition method.

After analysis of the supervised pattern recognition method, the variable importance in projection (VIP) scores of all variables was calculated. We used the VIP scores and Kruskal–Wallis test to select different metabolites associated with the different poisons (VIP >1 and P < 0.01).

For researching the relationship of the differential metabolites among each experimental group, four OPLS models were established (DDVP group and blank group; paraquat group and blank group; aconitine group and blank group; sodium nitrite group and blank group), and differential metabolites were selected by VIP value.

Establishing models for inferring the identities of the poisons

All metabolic data and differential metabolite data were used as input vectors with a support vector machine (SVM) to establish models for inferring the identities of the poisons.

Application of the models

To test the prediction abilities of the established models, a double-blind method was used. Six male SD rats dosed with nitrite were labeled as an unknown/prediction group and analyzed with the same experimental method and procedures by a scientist who did not know the identity of the poison. Then, the predictive success rates of the different models were calculated.


  Results and Discussion Top


Section of poisons

In this study, we selected paraquat, dichlorvos, aconitine, and sodium nitrite to establish the models because they are frequently encountered in poisoning cases. Paraquat, one of the most common herbicides, is widely used to control weed growth and has a strong toxic effect on humans and animals. Chen et al. analyzed patient epidemiological data from a hospital's 5-year paraquat poisoning database and found that paraquat poisoning accounted for 31.40% of all cases of acute poisoning.[16] Dichlorvos is a common organophosphorus pesticide. Worldwide, up to 200,000 deaths per year are caused by pesticide poisoning, and two-thirds of these are related to organophosphorus pesticide poisoning, among which the vast majority is dichlorvos poisoning.[17],[18] Aconitum plants have been used in Chinese medicine for hundreds of years. Their main component, aconitine, has limited drug safety. Therefore, poisoning and death cases are often caused by improper use.[19],[20] Nitrite is a crystalline white powder that looks similar to salt and can preserve color and prevent corrosion. Consequently, illegal traders often use this compound as a substitute for salt. Cases of nitrite poisoning are common in China.[21] To increase the efficiency of screening, we established a model for inferring the identities of these four poisons using GC-MS metabonomic technology.

Metabolic profiling and metabolite identification

Blood samples were pretreated, analyzed by GC-MS, and the total ion chromatograms were obtained. By comparing the chromatographic peaks of the quality control samples with the NIST 14 library, we identified 83 metabolites with high degrees of matching (>80%). These metabolites were mainly organic acids, amino acids, carbohydrates, and lipids.

Multivariate statistical analysis

[Figure 1] was the result of OPLS analysis. It showed that the blank and experimental groups could be approximately separated. Cross-validation showed R2Y = 0.772 and Q2Y = 0.607 in the model. To evaluate the validity of the PLS model, we used permutation tests, and the 20 validation results showed that R2 on the y-axis intercept was 0.324 and Q2 on the y-axis intercept was −0.256 [Figure 2]. According to the literature, in the cross-validation of the model, the R2 value reflects the degree of fit of the model and a value closer to one suggests that the model fits better. The Q2 value reflects the predictive ability of the model and a value closer to one means that the model has stronger prediction ability.[22] In the model permutation test, the intercept of the R2 and Q2 on the y-axis reflects the degree of overfitting of the model. The following criteria can be used to evaluate the effectiveness of a model: in cross-validation, Q2Y should be >0.5 for an effective model and >0.9 for an outstanding model, and the R2YQ2Y value should be <0 and >0.2–0.3; and in the permutation test, R2 on the y-axis intercept should be <0.4 and Q2 on the y-axis intercept should be <0.05.[23] According to the above criteria, our model is valid and reliable. According to literature[24],[25] and our results, we believed that the combination of GC-MS metabonomics and OPLS analysis can classify poisoning caused by different chemicals.
Figure 1: Orthogonal partial least squares-discriminant analysis of the blank and experimental groups

Click here to view
Figure 2: The partial least squares-discriminant analysis permutation plot of the blank group and all experimental groups

Click here to view


Screening of differential metabolites

[Figure 3] was the loading plot of OPLS analysis. Differential metabolites could be observed from it. Then, we used VIP value to select the differential metabolites. According to the literature, variables with a VIP value >1 have variables, so we used the Kruskal–Wallis test to observe whether the differences were statistically significant. We screened for differential metabolites for the different poisons (VIP >1 and P < 0.01) and obtained 18 metabolites [Table 1]. The relationship of the differential metabolites among each experimental group was showed in [Table 2].
Figure 3: The orthogonal partial least squares-discriminant analysis loading plot of the blank group and all experimental groups

Click here to view
Table 1: Entire differential metabolites in the rat blood samples of the blank and experimental groups

Click here to view
Table 2: Differential metabolites in the rat blood samples in the blank and experimental groups

Click here to view


Even though they have different mechanisms of toxicity, the paraquat and sodium nitrite groups had the same differential metabolites (i.e., lactic acid, glycine 2,2-furoic acid, 3-hydroxybutyric acid, urea, and seminose). The reason for this needs to be investigated. We also identified 5-ketone proline and 1,5-anhydrohexitol as potential biomarkers of dichlorvos and aconitine, but additional experiments are needed to confirm this.[25]

Support vector machine models

Although we were able to visualize classification of the blank and experimental groups using OPLS analysis of the rat blood samples, we failed to provide quantitative information to establish a poison classification model and were unable to identify unknown samples. SVM could be adopted to solve this problem using this data mining method.[19],[20],[21]

The differential metabolite data from our research on the application of SVM to metabonomics were classified as input variables. The eight groups samples (1 – blank group, 2 – paraquat group, 3 – dichlorvos poisoning survival group, 4 – dichlorvos poisoning death group, 5 – aconitine poisoning survival group, 6 – aconitine poisoning death group, 7 – sodium nitrite poisoning survival Group, 8 – sodium nitrite poisoning death group), and the data were classified as output variables using the radial basis function algorithm to establish a SVM classification model for the poisons.

Cross-validation was used to evaluate the accuracy of the model and prevent overfitting. We established a model for inferring the identities of the poisons by SVM classification of all metabolites using single poison differential metabolite data [Table 3]. In the training set, all samples except the W9 sample could be accurately identified, and the identification accuracy was 97.22%. The test set prediction accuracy was 100%. The accuracy of the training set identification was 80.56%, and the prediction accuracy of the test set was 85.71% [Table 4].
Table 3: The identities of the poisons by SVM classification using one kind poison differential metabolite data, training set, and test set

Click here to view
Table 4: The identities of the poisons by SVM classification using entire poison differential metabolite data, training set, and test set

Click here to view


Application of the model for inferring the identities of poisons

To test the prediction ability of the model, we used a double-blind method with a group of six male SD rats poisoned with an unknown substance. All metabolic data of unknown group were processing by OPLS, the results was [Figure 4]. It showed that the unknown group and experimental group could not be approximately separated. So we used SVM to distinguish it. After processing the blood samples using the established method, the deviation sample (U6) was very different to the other sample, so we deleted it. The single differential metabolite predictive accuracy (100%) was significantly higher than the total differential metabolite predictive accuracy (80%) [Table 5].
Figure 4: Orthogonal partial least squares-discriminant analysis of the unknown group and the blank group

Click here to view
Table 5: Toxic substances prediction results of support vector machine models using different type data

Click here to view


A comparison of the data [Table 3] and [Table 4] showed that the prediction accuracy of the total metabolite information model was higher than that of the differential metabolite information model. This difference may occur because OPLS analysis was used for screening to select statistically significant metabolites, but metabolites without statistically significant differences may also affect the sample classification. Not using this metabolite information for modeling may decrease the accuracy of the model forecast.

In summary, total metabolite and differential metabolite information with the SVM model can be used to construct models for identification of poisons. The prediction accuracy of the total metabolite model is better than that of the differential metabolite model. Furthermore, the single differential metabolite model is better than the total differential metabolites model.


  Conclusions Top


We established a poison prediction model in SD rats using a GC-MS metabonomic method. In contrast to methods that require sequential quantification of plasma concentrations, this model can quickly and accurately forecast toxic responses to paraquat, dichlorvos, aconitine, and nitrite simultaneously. In addition, we identified potential biomarkers for the four poisons. Considering the similar metabolic profiles of these poisons, in practical toxicology screening work, other models will be required to perfect the poison metabonomic information database, and this should be pursued in follow-up studies. This work will increase efficiency and provide new direction for auxiliary diagnoses to solve criminal cases.

Financial support and sponsorship

This work was financially supported by the project of the National Natural Sciences Foundation of China (No. 81373239).

Conflicts of interest

There are no conflicts of interest.



 
  References Top

1.
Wang Z, Ma J, Zhang M, Wen C, Huang X, Sun F, et al. Serum metabolomics in rats after acute paraquat poisoning. Biol Pharm Bull 2015;38:1049-53.  Back to cited text no. 1
    
2.
Mclaughlin P, Pounder D, Maskell P, Osselton D. Real-time near-body drug screening during autopsy I: Use of the randox biochip drugs of abuse DOA I and DOA II immunoassays. Forensic Toxicol 2013;31:113-8.  Back to cited text no. 2
    
3.
Ma C, Bi K, Zhang M, Su D, Fan X, Ji W, et al. Metabonomic study of biochemical changes in the urine of morning glory seed treated rat. J Pharm Biomed Anal 2010;53:559-66.  Back to cited text no. 3
    
4.
Gromski PS, Xu Y, Correa E, Ellis DI, Turner ML, Goodacre R. A comparative investigation of modern feature selection and classification approaches for the analysis of mass spectrometry data. Anal Chim Acta 2014;829:1-8.  Back to cited text no. 4
    
5.
Montenarh D, Hopf M, Maurer HH, Schmidt P, Ewald AH. Development and validation of a multi-analyte LC-MS/MS approach for quantification of neuroleptics in whole blood, plasma, and serum. Drug Test Anal 2016;8:1080-9.  Back to cited text no. 5
    
6.
Di Rago M, Saar E, Rodda LN, Turfus S, Kotsos A, Gerostamoulos D, et al. Fast targeted analysis of 132 acidic and neutral drugs and poisons in whole blood using LC-MS/MS. Forensic Sci Int 2014;243:35-43.  Back to cited text no. 6
    
7.
Wang C, Feng R, Li Y, Zhang Y, Kang Z, Zhang W, et al. The metabolomic profiling of serum in rats exposed to arsenic using UPLC/Q-TOF MS. Toxicol Lett 2014;229:474-81.  Back to cited text no. 7
    
8.
Koek MM, Jellema RH, van der Greef J, Tas AC, Hankemeier T. Quantitative metabolomics based on gas chromatography mass spectrometry: Status and perspectives. Metabolomics 2011;7:307-28.  Back to cited text no. 8
    
9.
Yang J, Wang H, Xu W, Hao D, Du L, Zhao X, et al. Metabolomic analysis of rat plasma following chronic low-dose exposure to dichlorvos. Hum Exp Toxicol 2013;32:196-205.  Back to cited text no. 9
    
10.
Sun H, Wang M, Zhang A, Ni B, Dong H, Wang X. UPLC-Q-TOF-HDMS analysis of constituents in the root of two kinds of aconitum using a metabolomics approach. Phytochem Anal 2013;24:263-76.  Back to cited text no. 10
    
11.
Sun B, Zhang M, Zhang Q, Ma K, Li H, Li F, et al. Metabonomics study of the effects of pretreatment with glycyrrhetinic acid on mesaconitine-induced toxicity in rats. J Ethnopharmacol 2014;154:839-46.  Back to cited text no. 11
    
12.
Greenwod R. Nonhuman Primates; Standards and Guidelines for The Breeding, Care, and Management of Laboratory Animals. Canadian Veterinary Journal 1974;15:231.  Back to cited text no. 12
    
13.
Yu J, Wu H, Lin Z, Su K, Zhang J, Sun F, et al. Metabolic changes in rat serum after administration of suberoylanilide hydroxamic acid and discriminated by SVM. Hum Exp Toxicol 2017;36:1286-94.  Back to cited text no. 13
    
14.
Ye BC, Zhang M, Yin BC. Nanomaterial-enhanced fluorescence polarization and its application. Nano-Bio Probe Design and Its Application for Biochemical Analysis. SpringerBriefs in Molecular Science. Berlin, Heidelberg: Springer; 2012. p. 3-25.  Back to cited text no. 14
    
15.
Werner E, Croixmarie V, Umbdenstock T, Ezan E, Chaminade P, Tabet JC, et al. Mass spectrometry-based metabolomics: Accelerating the characterization of discriminating signals by combining statistical correlations and ultrahigh resolution. Anal Chem 2008;80:4918-32.  Back to cited text no. 15
    
16.
Chen M, Chen F, Zhu R, Wang X. Acute paraquat poisoning epidemiological. Chin J Crit Care Med2014;34:497-501.  Back to cited text no. 16
    
17.
Wang B, Lin B, Shi T, Chen Z, Wu S, He C. Relationship between the heart injury and catecholamine level of patients with severe acute dichlorvos poisoning. Prog Mod Biomed 2014;(31):6139-41.  Back to cited text no. 17
    
18.
Luo R, Zhao C, Zhang J, Zheng J, Feng W, Qiu Z. Epidemiology study on characteristics of acute intoxication in emergency department and analysis on influencing factors of poisoned death. Chin J Public Health 2004;20:1122-4.  Back to cited text no. 18
    
19.
Zhang K, Zhuo L, Liu Q, Zhu S, Lin L. The forensic progress of aconitine poisoning. The Second International Conference on Evidence Law and Forensic Science; 2009.  Back to cited text no. 19
    
20.
Chen F, Zou D, Shan R, Xu Z, Ma L, Mei Z. Literature analysis on causes of aconitine plant poisoning and detoxification methods in recent 10 years. Lishizhen Med Materia Med Res 2012;23:12.  Back to cited text no. 20
    
21.
Li Y, Wang S. Literature analysis of nitrite food poisoning from 2000 to 2009. J Prev Med Inf 2010;26:822-4.  Back to cited text no. 21
    
22.
Deng M, Zhang M, Sun F, Ma J, Hu L, Yang X, et al. A gas chromatography-mass spectrometry based study on urine metabolomics in rats chronically poisoned with hydrogen sulfide. J Forensic Legal Med 2015;32:59-63.  Back to cited text no. 22
    
23.
Eriksson L, Kettaneh-Wold N, Trygg J, Wikström C, Wold S. Multi- and Megavariate Data Analysis: Part I: Basic Principles and Applications. Umeå: Umetrics Inc.; 2006.  Back to cited text no. 23
    
24.
Hirakawa K, Koike K, Uekusa K, Nihira M, Yuta K, Ohno Y. Experimental estimation of postmortem interval using multivariate analysis of proton NMR metabolomic data. Leg Med (Tokyo) 2009;11 Suppl 1:S282-5.  Back to cited text no. 24
    
25.
Sato T, Zaitsu K, Tsuboi K, Nomura M, Kusano M, Shima N, et al. Apreliminary study on postmortem interval estimation of suffocated rats by GC-MS/MS-based plasma metabolic profiling. Anal Bioanal Chem 2015;407:3659-65.  Back to cited text no. 25
    


    Figures

  [Figure 1], [Figure 2], [Figure 3], [Figure 4]
 
 
    Tables

  [Table 1], [Table 2], [Table 3], [Table 4], [Table 5]



 

Top
 
 
  Search
 
Similar in PUBMED
   Search Pubmed for
   Search in Google Scholar for
 Related articles
Access Statistics
Email Alert *
Add to My List *
* Registration required (free)

 
  In this article
Abstract
Introduction
Materials and Me...
Results and Disc...
Conclusions
References
Article Figures
Article Tables

 Article Access Statistics
    Viewed153    
    Printed6    
    Emailed0    
    PDF Downloaded15    
    Comments [Add]    

Recommend this journal


[TAG2]
[TAG3]
[TAG4]