Non-Small Cell Lung Cancer Detection and Subtyping by UPLC-HRMS-Based Tissue Metabolomics

Abstract

Non-small cell lung cancer (NSCLC) is the prevalent histological subtype of lung cancer. In this study, we performed ultraperformance liquid chromatography-high-resolution mass spectrometry (UPLC-HRMS)-based metabolic profiling of 227 tissue samples from 79 lung cancer patients with adenocarcinoma (AC) or squamous cell carcinoma (SCC). Orthogonal partial least squares-discriminant analysis (oPLS-DA) analyses showed that AC, SCC, and NSCLC tumors were discriminated from adjacent noncancerous tissue (ANT) and distant noncancerous tissue (DNT) samples with good accuracies (91.3, 100, and 88.3%), sensitivities (85.7, 100, and 83.9%), and specificities (94.3, 100, and 90.7%), using 12, 4, and 7 discriminant metabolites, respectively. The discriminant panel for AC detection included valine, sphingosine, glutamic acid γ-methyl ester, and lysophosphatidylcholine (LPC) (16:0), levels of which in tumor tissues were significantly altered. Valine, sphingosine, LPC (18:1), and leucine derivatives were used for SCC detection. The discrimination between AC and SCC had 96.8% accuracy, 98.2% sensitivity, and 85.7% specificity using a five-metabolite panel, of which valine and creatine had significant differences. The classification models were further verified with external validation sets, showing a promising prospect for NSCLC tissue detection and subtyping.

Multi-omic profiling of squamous cell lung cancer identifies metabolites and related genes associated with squamous cell carcinoma

“…Deregulated creatine (in particular) and phosphocholine metabolite levels between AC and SqCC have been reported previously, 12,13,29,30 with similar patterns as observed in this study, but without the link to SLC6A8 or CHKA expression through metabolic models. When comparing bulk tumor tissue to matched normal tissue, the increased mRNA expression of the creatine transporter SLC6A8 (solute carrier family 6, member 8) gene and decreased CHKA (phosphocholine) expression appeared tumor-associated.…”

Ultrahigh‐performance liquid chromatography–high‐resolution mass spectrometry‐based plasma metabolomics study of thymoma and thymic hyperplasia

“…After initial pre‐processing including retention time alignment and peak detection using Progenesis QI, a total of 2477 and 1136 features were obtained in the positive and negative ionization modes, respectively. Data post‐processing was then applied, as used in our previous study 21 . Briefly, the post‐processing steps included annotation of adducts, in‐source fragment ions and chloride cluster ions, signal blank correction, deletion of artefacts and contaminants, QC‐based signal correction, and feature abundance normalization and scaling.…”

“…A data matrix table of metabolic feature abundances with samples in columns and feature identities (retention time ( R t ) and m / z pair) in rows was exported. Next, a developed Fortran program was applied to post‐process the output of Progenesis QI, similar to our previous study, 21 which included analyses of adducts, insource fragments and chloride isotopic ions, contaminant ion removal, QC‐based scaling and normalization.…”

“…Data postprocessing was then applied, as used in our previous study. 21 1). In the youth group, all LPCs and LPEs had elevated abundances in the five thymoma samples and one TC sample.…”

Metabolomics Analysis and Diagnosis of Lung Cancer: Insights from Diverse Sample Types

“…Due to the differences in diet, living habits, and other aspects of various individuals, blood, urine, and other specimens are affected by many non-experimental factors. Tissue specimens can be collected from the same cancer tissue and adjacent tissue (or normal tissue), significantly reducing the influence of interference factors 21 , 115 .…”