Application of machine learning algorithms to screen potential biomarkers under cadmium exposure based on human urine metabolic profiles

Exposure to environmental cadmium increases the health risk of residents. Early urine metabolic detection using high-resolution mass spectrometry and machine learning algorithms would be advantageous to predict the adverse health effects. Here, we conducted machine learning approaches to screen potential biomarkers under cadmium exposure in 403 urine samples. In positive and negative ionization mode, 4207 and 3558 features were extracted, respectively. We compared seven machine learning algorithms and found that the extreme gradient boosting (XGBoost) and random forest (RF) classifiers showed better accuracy and predictive performance than others. Following 5-fold cross-validation, the value of area under curve (AUC) was both 0.93 for positive and negative ionization modes in XGBoost classifier. In the RF classifier, AUC were 0.80 and 0.84 for positive and negative ionization modes, respectively. We then identified a biomarker panel based on XGBoost and RF classifiers. The incorporation of machine learning models into urine analysis using high-resolution mass spectrometry could allow a convenient assessment of cadmium exposure.     

Urine profiling using capillary electrophoresis-mass spectrometry and multivariate data analysis

This work presents the development of a general and fast method for metabolic profiling of urine, using capillary electrophoresis-electrospray ionisation mass spectrometry (CE-ESIMS) and multivariate data analysis (DA). Human urine samples collected before and after ingestion of paracetamol were analysed at acidic and basic CE conditions, using both positive and negative ESI-MS detection. Analysis of the entire resulting data set, with no prior knowledge of the target compounds, using pair-wise 'fuzzy' correlation and eigenvalue analysis enabled the samples to be discriminated between on the basis of blank urine and urine collected after drug intake. By generating two-dimensional loadings plots, it was also possible to identify the m/z values of the substances responsible for the differentiation between control and dosed samples.     

Paper spray mass spectrometry and PLS-DA improved by variable selection for the forensic discrimination of beers

Paper spray mass spectrometry (PS-MS) combined with partial least squares discriminant analysis (PLS-DA) was applied for the first time in a forensic context to a fast and effective differentiation of beers. Eight different brands of American standard lager beers produced by four different breweries (141 samples from 55 batches) were studied with the aim at performing a differentiation according to their market prices. The three leader brands in the Brazilian beer market, which have been subject to fraud, were modeled as the higher-price class, while the five brands most used for counterfeiting were modeled as the lower-price class. Parameters affecting the paper spray ionization were examined and optimized. The best MS signal stability and intensity was obtained while using the positive ion mode, with PS(+) mass spectra characterized by intense pairs of signals corresponding to sodium and potassium adducts of malto-oligosaccharides. Discrimination was not apparent neither by using visual inspection nor principal component analysis (PCA). However, supervised classification models provided high rates of sensitivity and specificity. A PLS-DA model using full scan mass spectra were improved by variable selection with ordered predictors selection (OPS), providing 100% of reliability rate and reducing the number of variables from 1701 to 60. This model was interpreted by detecting fifteen variables as the most significant VIP (variable importance in projection) scores, which were therefore considered diagnostic ions for this type of beer counterfeit.     

超高效液相色谱/飞行时间质谱法分析尿液中的代谢物用于区分人类性别的研究

尿中的代谢产物可以反映生命个体的生理状态。为了考察在非严格控制条件下(即对志愿者的饮食、生活方式以及样品采集时间等诸多条件不加以控制)基于尿中代谢物的指纹图谱对男女性别进行区分的可行性,采用超高效液相色谱/飞行时间质谱(UPLC/TOF-MS)联用技术分析了31个随机尿样,并用主成分分析法（PCA）和偏最小二乘法判别分析（PLS-DA）两种数据处理方法对数据进行处理,与PCA法比较,PLS-DA法能提高分类效果,并筛选出4种可能的与性别相关的生物标记物。研究结果表明,UPLC/MS联用技术通量高,数据量丰富;模式识别数据处理方法适合于从大量数据中提取信息,两者的结合有利于代谢组学的研究。     

The use of high performance liquid chromatography-quadrupole time-of-flight mass spectrometry coupled to advanced data mining and chemometric tools for discrimination and classification of red wines according to their variety

In this study, the potential of high performance liquid chromatography coupled to quadrupole time-of-flight mass spectrometry (HPLC–QTOFMS) for metabolomic profiling of red wine samples was examined. Fifty one wines representing three varieties (Cabernet Sauvignon, Merlot, and Pinot Noir) of various geographical origins were sourced from the European and US retail market. To find compounds detected in analyzed samples, an automated compound (feature) extraction algorithm was employed for processing background subtracted single MS data. Stepwise reduction of the data dimensionality was followed by principal component analysis (PCA) and partial least square-discriminant analysis (PLS-DA) which were employed to explore the structure of the data and construct classification models. The validated PLS-DA model based on data recorded in positive ionization mode enabled correct classification of 96% of samples. Determination of molecular formula and tentative identification of marker compound was carried out using accurate mass measurement of full single MS spectra. Additional information was obtained by correlating the fragments obtained by MS/MS accurate mass spectra using the QTOF with collision induced dissociation (CID) of precursor ions.     

Metabolite profiling in rat urine by liquid chromatography/electrospray ion trap mass spectrometry. Application to the study of heavy metal toxicity

This work reports the use of reverse-phase liquid chromatography coupled to electrospray ion trap (QIT) mass spectrometry for the analysis of the metabolome in rat urine. An injection of 20 microL of urine into the chromatographic system is followed by a slow gradient elution and mass spectrometric detection in the scanning mode from m/z 100-1000 in both positive and negative modes. Over a time scale of 90 min, 30 and 20 resolved peaks were observed in the positive and the negative modes, respectively, corresponding to the presence of a few hundred m/z ratios. By using a QIT analyzer, data-dependent tandem mass spectrometry of selected m/z ratios identified several compounds in rat urine and characterized various chemical families, including carboxylic acids, amines, sulfated compounds, glucuronides and glycosides, by the observation of characteristic fragment ions or neutral losses. The method has been applied to the investigation of the chronic toxicity of heavy metals in rat urine. A few tens of m/z ratios, differing in intensity more than threefold from control values, were observed in both positive and negative modes. The time variations for some selected ions suggest that LC/ESI-MS could allow selective characterization of biomarkers in response to specific toxic compounds.     

Tissue-spray ionization mass spectrometry for raw herb analysis

Tissue-spray ionization mass spectrometry is developed for the in situ chemical analysis of raw herbs under ambient conditions. We demonstrated that analyte molecules could be directly sprayed and ionized from solvent-wetted ginseng tissues upon the application of high electrical voltage to the tissue sample. Abundant phytochemicals/ metabolites, including ginsenosides, amino acids and oligosaccharides, could be detected from ginseng tissues when the tissue-spray experiments were conducted in positive ion mode. Thermally labile and easily hydrolyzed malonyl-ginsenosides were also detected in negative ion mode. The tissue-spray ionization method enables the direct detection of analytes from raw herb samples and preserves the sample integrity for subsequent morphological and/ or microscopic examination. In addition, this method is simple and fast for chemical profiling of wild-type and cultivated-type American ginsengs with differentiation.     

高效液相色谱-离子阱质谱法测定尿液中β2-受体激动剂及β-受体阻断剂

建立了尿液中23种β2-受体激动剂及5种β-受体阻断剂的高效液相色谱-离子阱质谱(HPLC-IT-MS)测定方法。尿液样品采用冷冻高速离心沉淀蛋白,上清液过ExtrelutTM硅藻土柱,用乙酸乙酯洗脱后,洗脱液经旋转蒸发仪浓缩并复溶待测。HPLC分离采用AtlantisT3-150 mm色谱柱,以甲醇和含0.1%甲酸的水溶液为流动相梯度洗脱,IT-MS采用电喷雾离子源在多反应离子监测模式下测定。定量分析选择9种经过氘代同位素标记的β2-受体激动剂为内标。各化合物的线性范围为0.005～0.16 mg/L,尿液中的检出限均能达到0.2 μg/L。空白尿液样品中不同加标水平的回收率为57.1%～127.1%,相对标准偏差为1.1%～31.1%。该方法简便快速,灵敏度高,适用于人或动物尿液中23种β2-受体激动剂及5种β-受体阻断剂的定性和定量分析。     

Coupling ultra high-pressure liquid chromatography with single quadrupole mass spectrometry for the analysis of a complex drug mixture

The coupling of ultra high-pressure liquid chromatography with a single quadrupole mass spectrometer was investigated for the analysis of several cytochromes P450 (CYP450) substrates and respective metabolites. The effect of numerous operating parameters (e.g. mobile phase pH, flow rate, gradient length, MS acquisition mode, dwell time, polarity switching, etc.) on selectivity, sensitivity and acquisition rate was studied. It was demonstrated that basic pH conditions provided the best compromise in terms of sensitivity and chromatographic selectivity with both acidic and basic compounds. The optimal mobile phase flow rate for UHPLC-MS experiments should be comprised between 300 and 600 μL/min for 2.1 mm ID columns, while a higher flow rate generated up to 3-fold loss in sensitivity. It was also demonstrated that the fast polarity switching mode represented a valuable tool to improve throughput, maintaining acceptable performance. Finally, limits of detection were included in the range [1-50 ng/mL] in positive ionization mode and [50-250 ng/mL] in negative ionization mode, for investigated compounds.     

Liquid chromatography/time-of-flight mass spectrometry analysis of postmortem blood samples for targeted toxicological screening

A liquid chromatography/time-of-flight mass spectrometry (LC-TOF-MS) method for targeted toxicological screening in human postmortem blood samples from forensic autopsy cases has been developed, validated and compared with a previously used method using gas chromatography with nitrogen–phosphorus detection (GC-NPD). Separation was achieved within 12 min by high-resolution gradient chromatography. Ions were generated in positive and negative electrospray ionization mode and were detected in 2-GHz single mass spectrometry mode, m/z range 50–1,000. Before injection, 0.25 g blood was prepared by protein precipitation with 500 μL of a mixture of acetonitrile and ethanol containing deuterated internal standards. An in-house database comprising 240 drugs and metabolites was built by analysing solutions from certified standards or other documented reference material available. Identification was based on scoring of retention time, accurate mass measurement and isotopic pattern. Validation was performed on spiked blood samples and authentic postmortem blood samples. The thresholds defined as minimum required performance levels were for most compounds in the range from 0.01 to 0.10 μg/g. Typically, a mass error of less than 2 ppm and a precision of area measurements of less than 5 % coefficient of variation were achieved. Positive identification was confirmed at concentrations up to 500 μg/g. Most compounds were determined in positive ionization mode, but for a limited number of compounds (fewer than 4 %) negative ionization was needed and a few early-eluted compounds could not be identified owing to substantial influence of interferences from the matrix and were thus not included in the screening. A robust and valid toxicological screening by LC-TOF-MS for postmortem blood samples, covering 50 % more compounds, and with higher precision and sensitivity than the previously used screening by GC-NPD was achieved.     

Sensitive quantification of clozapine and its main metabolites norclozapine and clozapine-<Emphasis Type="Italic">N</Emphasis>-oxide in serum and urine using LC-MS/MS after simple liquid–liquid extraction work-up

An LC-MS/MS method for the determination of the atypic neuroleptic clozapine and its two main metabolites norclozapine and clozapine-N-oxide has been developed and validated for serum and urine. After addition of d4-clozapine as deuterated internal standard a fast single-step liquid–liquid extraction under alkaline conditions and with ethyl acetate as organic solvent followed. The analytes were chromatographically separated on a Synergi Polar RP column using gradient elution with 1 mM ammonium formate and methanol. Data acquisition was performed on a QTrap 2000 tandem mass spectrometer in multiple reaction monitoring mode with positive electrospray ionization. Two transitions were monitored for each analyte in order to fulfill the established identification criteria. The validation included the determination of the limits of quantification (1.0 ng/mL for all analytes in serum and 2.0 ng/mL for all analytes in urine), assessment of matrix effects (77% to 92% in serum, 21 to 78% in urine) and the determination of extraction efficiencies (52% to 85% for serum, 59% to 88% for urine) and accuracy data. Imprecision was <10%, only the quantification of norclozapine in urine yielded higher relative standard deviations (11.2% and 15.7%). Bias values were below ±10%. Dilution of samples had no impact on the correctness for clozapine and norclozapine in both matrices and for clozapine-N-oxide in serum. For quantification of clozapine-N-oxide in urine a calibration with diluted calibrators has to be used. Calibration curves were measured from the LOQ up to 2,000 ng/mL and proved to be linear over the whole range with regression coefficients higher than 0.98. The method was finally applied to several clinical serum and urine samples and a cerebro-spinal fluid sample of an intoxicated 13-month-old girl.     

Mass spectrometry of 3,5- and 4,5-dicaffeoylquinic acids and selected derivatives

The mass spectral properties of 3,5- and 4,5-dicaffeoylquinic acids (DCQAs) and selected derivatives were examined using electron ionization (EI), fast atom bombardment (FAB) and electrospray ionization (ESI). EI analysis of the trimethylsilyl derivatives provides molecular mass (M(r)) information, but the spectrum is dominated by fragment ions of the caffeic acid group; isomers cannot be differentiated using EI. FAB analysis, in both the positive and negative ion detection modes, provides M(r) information on the free compounds, but little fragmentation is observed using normal scan conditions. The FAB mass-analyzed ion kinetic energy spectroscopic analysis of the free compounds does, however, permit differentiation of the isomers, with 3,5-DCQA showing selective loss of water, a process not observed with the 4,5-isomer. Both EI and FAB provide M(r) and some structural information when applied to the peracetate derivatives of the DCQAs. ESI of the DCQAs provides considerably more structural information, especially in the negative ion detection mode, and is the recommended method of analysis of the quinic acid esters. M(r) information, identity of the ester groups and differentiation of isomers are possible using ESI. Copyright 1999 John Wiley & Sons, Ltd.     

Identification of key metabolites of tectorigenin in rat urine by HPLC-MS(n)

This is a report about the identification of key metabolites of tectorigenin in rat urine using high-performance liquid chromatography-electrospray ionization ion trap tandem mass spectrometric method (HPLC-ESI-MS(n)). Six healthy rats were administered a single dose (80 mg/kg) of tectorigenin by oral gavage. Urine was sampled for 0-24 h and centrifuged at 12,000 rpm for 10 min to obtain the supernatants, then the supernatants were purified by solid-phase extraction with a C(18) cartridge. The chromatographic separation was carried out on a reversed-phase C(18) column with a gradient elution program whereas acetonitrile-0.1% formic acid water was used as mobile phase. Mass spectra were acquired in negative ionization mode and a data-dependant scan was used for the identification of the key metabolites of tectorigenin in the urine samples. As a result, four phase II metabolites and the parent drug tectorigenin were found and identified in rat urine for the first time.     

Mass spectrometry molecular fingerprinting of mineral and synthetic lubricant oils

The molecular composition of lubricating oils has a strong impact on how automotive engines function, but the techniques used to monitor the quality parameters of these oils only inspect their gross physical–chemical properties such as viscosity, color, and bulk spectroscopy profiles; hence, bad-quality, adulterated, or counterfeit oils are hard to detect. Herein, we investigated the ability of direct infusion electrospray ionization mass spectrometry (ESI-MS) to provide simple, rapid but characteristic fingerprint profiles for such oils of the mineral and synthetic types. After a simple aqueous extraction, ESI-MS analyses, particularly in the positive ion mode, did indeed show characteristic molecular markers with unique profiles, which were confirmed and more clearly visualized by partial least squares-discriminant analysis (PLS-DA). Nuclear magnetic resonance and Fourier transform infrared-attenuated total reflection spectroscopy were also tested for the bulk samples but showed nearly identical spectra, thus failing to reveal their distinct molecular composition and to differentiate the oil samples. To simulate adulteration, mixtures of mineral and synthetic oils were also analyzed by ESI(+)-MS, and additions as low as 1% of mineral oil to synthetic oil could be detected. The technique therefore offers a simple and fast but powerful tool to monitor the molecular composition of lubricant oils, particularly vias their more polar constituents.     

超高效液相色谱-串联质谱法测定人尿中毒死蜱的主要代谢产物3,5,6-三氯-2-吡啶醇

建立了人尿液中毒死蜱主要代谢产物3,5,6-三氯-2-吡啶醇(3,5,6-TCP)的超高效液相色谱-串联质谱分析测定方法。采用二氯甲烷-乙酸乙酯(20:80, v/v)溶液中性条件下提取,ACQUITY UPLC BEH C18柱色谱分离,以乙腈-水为流动相梯度洗脱,在电喷雾电离(ESI)负离子模式下,选择离子监测(SIR)方式检测。方法线性范围为0.005~0.4 mg/L,检出限为0.41 μg/L,平均加标回收率为97.9%,日内及日间精密度(RSD)均小于15%。该方法操作简单、灵敏度高、准确性好、重现性强,已成功应用于人尿液实际样品中3,5,6-TCP的测定,为毒死蜱的人体健康风险暴露评估以及其生物负荷的痕量监测提供了有效的方法支撑。     

Comprehensive analysis by liquid chromatography Q‐Orbitrap mass spectrometry: Fast screening of peptides and organic molecules

The number of substances nominally listed in the prohibited list of the World Anti‐Doping Agency increases each year. Moreover, many of these substances do not have a single analytical target and must be monitored through different metabolites, artifacts, degradation products, or biomarkers. A new analytical method was developed and validated for the simultaneous analysis of peptides and organic molecules using a single sample preparation and LC‐Q‐HRMS detection. The simultaneous analysis of 450 target molecules was performed after cleanup on a mixed‐mode solid‐phase extraction cartridge, combined with untreated urine. The cleanup solvent and reconstitution solvent were the most important parameters for achieving a comprehensive sample preparation approach. A fast chromatographic run based on a multistep gradient was optimized under different flows; the detection of all substances without isomeric coelution was achieved in 11 minutes, and the chromatographic resolution was considered a critical parameter, even in high‐resolution mass spectrometry detection. The mass spectrometer was set to operate by switching between positive and negative ionization mode for FULL‐MS, all‐ion fragmentation, and FULL‐MS/MS². The suitable parameters for the curved linear trap (c‐trap) conditions were determined and found to be the most important factors for the development of the method. Only FULL‐MS/MS² enables the detection of steroids and peptides at concentrations lower than the minimum required performance levels set by World Anti‐Doping Agency (1 ng mL⁻¹). The combination of the maximum injection time of the ions into the c‐trap, multiplexing experiments, and loop count under optimized conditions enabled the method to be applied to over 10 000 samples in only 2 months during the 2016 Rio Summer Olympic and Paralympic Games. The procedure details all aspects, from sample preparation to mass spectrometry detection. FULL‐MS data acquisition is performed in positive and negative ion mode simultaneously and can be applied to untargeted approaches.     

高效液相色谱-串联质谱法测定兔尿中溴氰菊酯及其毒性生物标志物

建立了一种应用高效液相色谱-串联质谱（HPLC-MS/MS）测定兔尿中与溴氰菊酯毒性相关的多种生物标志物的检测方法。分析物包括溴氰菊酯及其代谢产物1R,3R-二溴菊酸、3-苯氧基苯甲酸,以及5种生物标志物5-羟色胺、5-羟基吲哚乙酸、3-硝基丙酸、8-羟基脱氧鸟苷和6-甲氧基鸟嘌呤。样品经硅藻土基质固相分散萃取、三氯乙酸沉淀蛋白质和HLB固相萃取小柱净化,使用电喷雾离子源,在多反应监测模式下正负切换采集测定,其中溴氰菊酯、5-羟色胺、5-羟基吲哚乙酸、8-羟基脱氧鸟苷和6-甲氧基鸟嘌呤采用正离子模式,1R,3R-二溴菊酸、3-苯氧基苯甲酸和3-硝基丙酸采用负离子模式。基质校准曲线外标法定量。结果表明,7种生物标志物在各自的浓度范围内线性关系良好（R²不小于0.9914）,5-羟基吲哚乙酸的检出限和定量限分别为20 μg/L和50 μg/L,其余化合物的检出限和定量限分别为0.2~5.0 μg/L和0.5~10 μg/L;在兔尿中3个不同添加水平的平均回收率为74.2%~98.7%,相对标准偏差（RSD）不大于12%,方法简单、快速、准确、灵敏,可作为溴氰菊酯暴露评估的检测方法。     

The effects of electron and chemical Ionization Modes on the MS Profiling of Whole Bacteria

Free fatty acid profiling of whole bacteria [Francisella tularensis, Brucella melitensis, Yersinia pestis, Bacillus anthracis (vegetative and sporulated), and Bacillus cereus] was carried out with direct probe mass spectrometry under 70-eV electron ionization (EI) and isobutane chemical ionization in both the positive (CI+) and negative modes (CI-). Electron ionization produced spectra that contained molecular ions and fragment ions from various free fatty acids. Spectra acquired with isobutane chemical ionization in the positive mode yielded molecular ions of free fatty acids as well as ions from other bacterial compounds not observed under EI conditions. Spectra obtained with negative chemical ionization did not contain as much taxonomic information as EI or CI+; however, some taxonomically significant compounds such as dipicolinic acid and poly(3-hydroxybutyrate) did produce negative ions. All ionization modes yielded spectra that could separate the bacteria by Gram-type when observed with principle components analysis (PCA). Chemical ionization in the positive ion mode produced the greatest amount of differentiation between the four genera of bacteria when the spectra where examined by PCA.     

Comparison of atmospheric pressure photoionization and electrospray ionization mass spectrometry for the analysis of UV filters

A comparison was made between the electrospray ionization (ESI) and atmospheric pressure photoionization (APPI) tandem mass spectrometric (MS/MS) responses of eleven ultraviolet (UV) filters. Four of the target compounds were favourably ionized in negative ion mode, and the other seven compounds in positive ion mode. For nine of the compounds APPI generated a similar response to that of ESI, but the APPI signal‐to‐noise (S/N) ratios were 1.3–60 times higher. The two most polar of the UV filter compounds (PBSA and BP‐4) were more efficiently ionized by ESI, offering higher signal intensities and lower detection limits. APPI was, however, less susceptible to ion suppression than ESI when real samples were injected. In order to optimize the APPI conditions different dopant solvents were examined to enhance the efficiency of the photoionization process. Among the evaluated dopants, toluene was selected as the best compromise. At a toluene flow rate of 10% of the solvent flow rates the ionization response increased by a factor of 40–50 over the use of no dopant for the compounds in positive ion mode and by more than 300 for the compounds in negative ion mode. Copyright © 2009 John Wiley & Sons, Ltd.     

Determination of terbuthylazine and desethylterbuthylazine in human urine and hair samples by eletrospray ionization-liquid chromatography/triple quadrupole mass spectrometry

Terbuthylazine (TBA) is a widely applied herbicide and an environmental contaminant. Following its use, humans, such as agricultural workers and rural residents, may be exposed. An isotope-dilution liquid chromatography coupled to electrospray-tandem mass spectrometry method for the determination of TBA, and its metabolite desethylterbuthylazine (DET) in human urine and hair was developed and validated. Under the optimised conditions, analytes were extracted from urine using a solid phase cartridge or from hair by sonication in methanol. Analytes were separated using a C18 reversed-phase chromatographic column and quantified, after positive ionization using a heated electrospray source, by a triple quadrupole mass detector in the selected reaction monitoring mode. Validation showed linear dynamic ranges up to 100 μg/L or 5.00 ng/mg hair, inter- and intra-run precisions <7%, and accuracies within 12% of spiked concentrations. Limits of quantification were 0.25 μg/L in urine and 0.01 ng/mg hair for both TBA and DET. Matrix effect evaluation showed that the isotope dilution approach allowed for the control of bias sources. TBA and DET were determined in specimens of agriculture workers exposed to TBA using the validated method. Hair samples contained TBA levels in the low nanogram per milligram range, and urine samples contained DET levels in the low microgram per liter range. Conversely, TBA levels in urine samples and DET levels in hair samples were always below the limit of quantification.