Investigation On Breath Screening Method For Diseases Based On Rapid Mass Spectrometry

Breath analysis can reflect the pathological or physiological state of the human body by analyzing Volatile Organic Compounds(VOCs)of exhaled breath.The operation processes are effective,fast,and without any damage to the human body,which can effectively determine the chemical components of exhaled breath.In clinical research,it is expected to become a new method for screening diagnosing method the early disease.In this study,an online mass spectrometry system that combined with Thermal Desorption Instrument(TD-1000)and Single Photon Ionization Mass Spectrometry(SPI-MS 3000)was established for exhaled breath analysis.The system parameters,including desorption temperature,carrier gas flow rate,system stability,etc.,were optimized to ensure the best operate conditions of this system.In this way,the chemical composition of more than hundreds of VOCs in exhaled breath could be measured online,while the diseases could be intelligently,effectively and high throughput screening without damage.A set of analytical processes and discriminant models suitable for early screening of disease was established based on the classification modeling and differential marker screening methods by using mass spectral data.Firstly,the feasibility of the model was verified by using the exhalation spectral data of human edible candies.Subsequently,it was used to detect the VOCs components of the patient’s exhalation with gastric cancer.The results showed that the self-constructed method of breath analysis for disease screening could effectively screen out differential markers,which could rapidly identify patients and healthy populations.1.Breath analysis of human edible candies: PCA,OPLS-DA and HCA models were used to analyze the breath spectral data of 20 volunteers after consuming 3 kinds of candies.The results obtained by the three models were consistent,indicating that those kinds of candies could be distinguished clearly.In addition,the accuracy of models could reach to 98.9%.Based on the VIP value,11 characteristic mass-to-charge ratios(m/z)of 136,138,172,121,108,120,153,168,107,124,115,190 were screened out.2.Breath analysis of patients with gastric cancer: PCA,OPLS-DA and ANN classification methods were used to analyze the exhalation spectra data of 88 gastric cancer patients and 65 healthy subjects,which provided reliable results of classification and model accuracy.The precision of OPLS-DA model was up to 83.3%.The precision,specificity and sensitivity of ANN model were 98.0%,96.0% and 100%,respectively.Combined with the variable screening methods of t-test and VIP value,10 common differential markers were obtained,presented by the m/z of 124、110、120、70、126、82、56、166、98、100.They were identified as 6-methylhept-5-en-5-one,Acrolein,(S)-3,4-dihydroxybutyric acid,3-aminopropionitrile,a-Angelica lactone,Ethyl salicylate,2-Acethl-5-methylfuran,2-Propylfuran,hexanal,2-Methylfuran,respectively.In this study,the established diagnostic model based on the exhalation mass spectral datasets could not only improve the identity and classification efficiency of disease discrimination due to its good accuracy,but also determine the differential markers.The establishment of this method has a guiding significance for clinical research on the application of breath detection analysis in disease screening.