The biomarker is a kind of characteristic biochemical index that can be evaluated objectively, and the biological process of the organism can be known by its measurement. Examining specific biomarkers for a disease can play a key role in the diagnosis and prevention of the disease.
In the field of medical research, the research ideas of biomarkers are generally divided into three stages: Discovery, Verification, Validation. The screening of biomarkers usually requires the use of high-throughput omics methods to conduct metabonomics or proteomics tests on large-scale clinical samples to screen out statistically significant differential metabolites or proteins, and then screen out target biomarkers after a series of complex bioinformatics analyses. In the following validation stage, it is necessary to conduct large sample size validation of targeted proteomics or targeted metabolomics for a smaller range of biomarkers, statistical analysis, and calculation of specificity and sensitivity of the target markers. If you want to make your own research results more complete, you can also use clinical samples and combine them with clinical data for supplementary verification, such as ELISA and WB.
In 2017, a study titled “Metabolic biomarker signature to differentiate pancreatic ductal adenocarcinoma from chronic pancreatitis” published in the journal Gut (IF=17.016) by Greifswald University in Germany was the use of metabolomics technology to determine biomarkers A typical example.
Clinically, pancreatic cancer is known as the “king of cancer” and is one of the worst-prognosis malignant tumors. Chronic pancreatitis is a risk factor for pancreatic cancer, and clinically it is difficult to distinguish between the two, which can easily lead to the misdiagnosis of early pancreatic cancer and the delay of treatment. Due to the poor effect of the original markers, this series of facts prompted researchers Efforts to find alternative biomarkers.
In this study, a total of 914 subjects were recruited, including pancreatic ductal adenocarcinoma (PDAC, 271), chronic pancreatitis (CP, 282), liver cirrhosis (LC, 100), and healthy blood donors ( BDs) and 261 control samples of preoperative patients with non-pancreatic diseases, using LC-MS and GC-MS multiple metabolomics platforms including lipidomics (non-targeted analysis/steroids/lipids) to compare 914 cases Samples were tested. A three-stage biomarker development strategy (exploration set/training set/test set) was used to identify a total of 477 metabolites.
Finally, nine potential biomarkers were found based on the results of metabolomics data. These nine metabolites were used in combination with the existing pancreatic cancer diagnostic blood index CA19-9, and the combined marker group could even detect 98% of the biomarkers. Resection of pancreatic cancer has an accuracy rate of 90.4%. The AUC of the combined markers is significantly higher than the AUC of CA19-9 (0.94 vs. 0.85, p <0.001), sensitivity (89.9% vs. 74.7%, p <0.01) and specificity ( 91.3% compared with 77.5%, p <0.05) also significantly improved.
Not only metabolomics, in the research of some disease biomarkers, but the application of proteomics is also becoming more and more extensive. And the application of multi-omics technology is in general trend. Next, let us introduce to you the indispensable role of multi-omics technology in the process of screening biomarkers through another research example.
The results of a 2017 study by Greifswald University in Germany were published in the journal BMC Medicine (IF=8.097). Titled “Plasma proteome and metabolome characterization of an experimental human thyrotoxicosis model”. The researchers aimed to screen biomarkers that characterize the characteristics of human plasma thyroid-stimulating hormone (TSH) and free thyroxine (FT4). Using the thyrotoxicosis model for research, and through a two-stage crossover procedure through a random forest, verify whether the screened biomarkers can distinguish abnormal thyroid function.
According to the statistics of metabolome and proteome data, a total of 380 metabolites and 497 human proteins were identified. To ensure the availability of data, through filtering analysis, only metabolites and proteins with missing values less than 40% are selected for subsequent analysis. That is, 349 metabolites and 437 proteins are analyzed in the next step.
In order to find a new biomarker to classify the TH state, the researchers established a random forest classifier through a two-stage cross-validation procedure to comprehensively analyze the differential metabolites and differential proteins. In the end, 15 substances including metabolites and proteins were obtained. The results of 30 verifications all showed stable and good classification ability, which can be used as a potential Biomarker.
The above two high-level papers screened out different metabolites and proteins based on metabolomics and metabolism + proteomics methods. Combine the corresponding verification analysis to screen out the target biomarkers.