Targeted Metabolomic Profiling: A Diagnostic Tool for Non-Small-Cell Lung Cancer Patients
A recent study published in Scientific Reports has unveiled the potential of targeted metabolomic profiling as a diagnostic tool for patients with non-small-cell lung cancer (NSCLC). The primary objective of the study was to gain a deeper understanding of lung cancer development and design the most effective machine learning (ML) model for the classification of NSCLC and noncancer (NC) patients based on the results of targeted metabolomic profiling.
The study involved two groups: 100 patients diagnosed with NSCLC at different TNM stages and 100 participants without any malignancies (NC group). Prior to their participation, all individuals provided informed consent. The NSCLC patients were not administered antibiotics, pre- and/or probiotic preparations for three months prior to the study. On the other hand, the exclusion criteria for the noncancer participants included age below eighteen, the presence of severe somatic diseases, diabetes mellitus, gastrointestinal tract diseases, acute respiratory viral diseases, psychosis, alcoholism, drug addiction, pregnancy, and lactation. Blood samples, approximately 2.5 mL each, were collected from all the study participants after overnight fasting and stored at -80 °C.
The study was conducted in accordance with the ethical principles of medical research involving humans as stated in the Declaration of Helsinki and was approved by the ethical committee of Sechenov University.
The researchers used standard solutions for amino acid, tryptophan metabolism intermediates, acylcarnitine profiling, and other compounds obtained from reputable sources. They employed various analytical methods for the metabolic profiling, ensuring accuracy, precision, and stability. The validation process included the assessment of selectivity, linearity, accuracy, matrix effect, and stability. Quality control samples were used to evaluate analysis reproducibility.
Using statistical analysis, the researchers identified the most discriminative metabolites between the NSCLC and NC groups. They employed the debiased sparse partial correlation (DSPC) network analysis to detect connections between a large number of metabolites. Based on the metabolite combinations, machine learning-based classification models were developed to assess the diagnostic ability.
The study’s findings showed promising results in utilizing targeted metabolomic profiling as a tool for identifying non-small-cell lung cancer. The ML models proved effective in classifying NSCLC and NC patients based on their metabolomic profiles. The accuracy, precision, and area under the curve of the receiver operator characteristics (AUCROC) were evaluated to determine the diagnostic capabilities of the models.
In conclusion, this study highlights the potential of targeted metabolomic profiling as a diagnostic tool for non-small-cell lung cancer. By analyzing the metabolites in the blood samples of patients, researchers were able to develop machine learning models that accurately classified individuals with NSCLC. Further research and validation are required to fully establish the effectiveness of this approach in clinical settings. Nonetheless, this study provides valuable insights into the development of innovative methods for lung cancer diagnosis.
For more information on this study, you can visit the GitHub repository of the FimaLab research group, where the open-source python scripts used in the analysis are available.
Disclaimer: The content and views expressed in this article are of the researchers mentioned in the study. The article is for informational purposes only and should not replace professional medical advice or diagnosis. Always consult a qualified healthcare provider for individualized guidance on medical conditions and treatments.
