Establishment and validation of an LC-MS/MS method for detecting lipid metabolites in serum as biomarkers in colorectal cancer

To the Editor: The incidence and mortality of colorectal cancer (CRC) in China have increased significantly in recent years. Surpassing gastric cancer, CRC has become one of the main cancers threatening the life and health of Chinese people and causing a serious social burden.^[1] Early detection of CRC is critical for improving outcomes. Although colonoscopy remains the primary diagnostic and screening method, its invasiveness and low patient compliance hinder broad application.^[2] Non-invasive methods including fecal occult blood tests and carcinoembryonic antigen (CEA) lack sensitivity (20–40%) and specificity for early detection.^[3] Therefore, non-invasive biomarkers with high sensitivity and specificity for detecting CRC and precancerous adenomas are urgently needed. Altered gut microbiota is closely associated with colorectal carcinogenesis but exhibits high interindividual variability, limiting its clinical utility. Microbiota-derived metabolites such as bile acids and short-chain fatty acids are critical in CRC progression.^[4] Serum metabolite levels are strongly correlated with gut microbial abundance, providing a new direction for noninvasive diagnosis of CRC.^[5]

This study included a total of 376 subjects, comprising 129 healthy controls, 50 patients with adenoma, and 197 patients with CRC [Supplementary Table 1, http://links.lww.com/CM9/C679]. Subjects were enrolled from the Cancer Institute, Chinese Academy of Medical Sciences (CICAMS), with samples collected between 2019 and 2024. All participants with colorectal abnormalities were first-time diagnosed and treatment-naïve, with pathology-confirmed adenoma or CRC. Patients with secondary CRC or a history of other malignancies were excluded. This study was approved by the Independent Ethics Committee of the National Cancer Center/Cancer Hospital, CICAMS (No. 22/139-3340). The committee waived the requirement for informed consent because of the retrospective and anonymized nature of the data used in this study.

Pathway enrichment analysis based on untargeted metabolomics data from serum samples of 31 healthy individuals and 61 patients with colorectal abnormality revealed significant enrichment of linoleic acid metabolism, fatty acid β-oxidation, and bile acid metabolism pathways during colorectal carcinogenesis [Supplementary Figure 1, http://links.lww.com/CM9/C679]. Subsequently, integrated correlation analysis of untargeted metabolomic data from serum samples and matched metagenomic data from fecal samples of 11 healthy controls and 33 patients with colorectal abnormality identified 22 CRC-related gut microbiota metabolites, which we termed CRC-related gut microbiota metabolites (CGMMs) [Supplementary Figure 1, http://links.lww.com/CM9/C679].

The complexity of metabolic models for detecting colorectal abnormalities limits their clinical applicability. Therefore, we developed a quantitative liquid chromatography-tandem mass spectrometry (LC-MS/MS) method for simultaneous detection of 22 CGMMs in serum [Supplementary Figures 2–4 and Supplementary Tables 2–5, http://links.lww.com/CM9/C679], including octadecenoic acid, carnitine, and bile acids, within 6.0 min. This analytical method was designed based on previously reported conditions for bile acids,^[6] carnitines,^[7] and octadecenoic acids.^[8] Our approach exhibits high-throughput and is inexpensive and convenient for clinical use.

The performance of this method for 22 CGMMs was also validated according to Clinical and Laboratory Standards Institute guidelines.^[9] The linearity, lower limit of quantification, and upper limit of quantification of 22 CGMMs are shown in Supplementary Table 6, http://links.lww.com/CM9/C679. The concentration ranges were 20–2000 nmol/L for 15 bile acids; 0.31–31.25 ng/mL for 12,13-DiHOME (BN02), trans-2-octenoyl-L-carnitine (BP02), and trans-2-hexadecenoy-1-L-carnitine (BP03); 0.63–62.50 ng/mL for myristoyl-l-carnitine (BP01); 0.94–93.75 ng/mL for (±)15-HETE (BN01); and 1.3–125.0 ng/mL for 9,12,13-TriHOME (BN03) and (±)-hexanoylcamitine chloride (BP04) [Supplementary Table 6, http://links.lww.com/CM9/C679]. Our method provided a broader linear dynamic range of 20–2000 nmol/L for bile acids compared with other methods. At high, medium, and low concentrations, the intra-assay precision coefficient of variation and inter-assay precision coefficient of variation for the 22 CGMMs were less than 15% [Supplementary Table 7, http://links.lww.com/CM9/C679]. For all metabolites, extraction recovery rates after internal standard correction were 80–120% [Supplementary Table 8, http://links.lww.com/CM9/C679]. In addition, except for BN01, BP03, BP04, and GDCA, the serum matrix effects of the remaining compounds were ±15% [Supplementary Table 8, http://links.lww.com/CM9/C679].

Among the 22 CGMMs, BN03, BP01, taurolithocholic acid (TLCA), and lithocholic acid (LCA) concentrations were statistically significantly increased in patients with CRC compared to normal control (P <0.05) [Supplementary Table 9 and Supplementary Figure 5, http://links.lww.com/CM9/C679]. TLCA is a conjugated bile acid synthesized by the conjugation of LCA with taurine, both of which are secondary bile acids produced through gut microbial metabolism of primary bile acids. These results suggest an abnormality in the LCA-related secondary bile acid metabolic pathway in patients with CRC.

Among the four metabolites (BN03, BP01, TLCA, and LCA), BN03 had an area under the receiver operating characteristic curve (AUC) of 0.930 (95% confidence interval [CI]: 0.899–0.962) for distinguishing between colorectal abnormality and healthy individuals, which was higher than the AUC of CEA (0.778, 95% CI: 0.718–0.838) [Supplementary Table 10 and Supplementary Figure 6A, http://links.lww.com/CM9/C679]. In addition, BN03 showed an AUC of 0.825 (95% CI: 0.759–0.869) in distinguishing early CRC (stages I/II) from healthy individuals [Supplementary Table 10 and Supplementary Figure 6B, http://links.lww.com/CM9/C679]. As a trihydroxy-octadecenoic acid derived from linoleic acid metabolism, the mechanism of BN03 in CRC remains unclear. Previous studies suggested that gut microbiota-related linoleic acid metabolites promote carcinogenesis by inducing p21 expression, thereby inhibiting cyclin-dependent kinases and cyclin complex activity. Moreover, various isomers of linoleic acid may function as competitive ligands in multiple signaling pathways, potentially contributing to CRC development.^[10] We identified the clinical potential of BN03 in the early detection of CRC. The significant upregulation of BN03 in the serum of patients with CRC supports its potential as an early diagnostic biomarker and provides new avenues for investigating the mechanisms of colorectal carcinogenesis, particularly the role of oxidized lipid metabolism.

Adenomas can be categorized into high- and low-grade adenomas; patients with high-grade adenomas have a greater likelihood of progressing to CRC and are more likely to warrant clinical intervention. The sensitivity of BN03 to distinguish CRC and high- from low-risk adenomas and healthy people was 82.61% and specificity was 85.29% (AUC = 0.882), suggesting the potential of BN03 as a noninvasive serological marker for clinical intervention [Supplementary Table 11 and Supplementary Figure 6C, http://links.lww.com/CM9/C679].

We also analyzed the ability of BN03 and CEA at their respective cutoff values to distinguish between colorectal abnormality and healthy individuals [Supplementary Table 12, http://links.lww.com/CM9/C679]. The cutoff value for BN03 was selected at 3.865 ng/mL, which yielded the maximum sum of sensitivity (85.62%) and specificity (91.95%). To facilitate a direct comparison with the clinical standard, we also evaluated BN03 at a higher cutoff of 5.410 ng/mL, which matched the specificity of CEA (98.85%) while demonstrating a substantially higher sensitivity (62.75% for BN03 vs. 33.33% for CEA). For CEA, the widely accepted clinical cutoff value of 5 ng/mL was used [Supplementary Table 12, http://links.lww.com/CM9/C679]. When a combined criterion of BN03 >3.865 ng/mL or CEA >5 ng/mL was applied, the sensitivity reached 89.31% with a specificity of 90.80%, highlighting the complementary value of BN03 [Supplementary Table 12, http://links.lww.com/CM9/C679]. When BN03 is used in combination with CEA (cutoff values: BN03 >5.410 ng/mL or CEA >5 ng/mL), the sensitivity for detecting colorectal abnormalities improved significantly compared with using CEA alone (33.33% vs. 72.87%) while retaining a specificity of 97.7% [Supplementary Table 12, http://links.lww.com/CM9/C679]. Among 86 CEA-negative patients with colorectal abnormality, 51 showed elevated BN03 levels (cutoff at 5.410 ng/mL) [Supplementary Figure 6D, http://links.lww.com/CM9/C679]. Therefore, BN03 improved the sensitivity of CEA in distinguishing colorectal abnormalities and healthy individuals.

In summary, this study established and validated an LC-MS/MS method for the highly efficient quantification of 22 CRC-related microbiota metabolites. Most importantly, we successfully developed an absolute quantification method for the oxidized lipid BN03, achieving a critical transition from biomarker discovery to clinical quantitative detection. Compared with existing noninvasive screening techniques, such as whole-blood cfDNA methylation assays, the LC-MS/MS method developed in this study demonstrates significant advantages in operational convenience, detection cost, and standardization. This study not only initially established the value of BN03 as a new serum biomarker in the early detection of CRC, but also provided a new research direction for further exploring the role mechanism of oxidative lipid metabolism in the occurrence and development of CRC.

Acknowledgments

We gratefully acknowledge Bingxin Xiao for her crucial technical insight into the development and optimization of the LC-MS/MS method employed in this study.

Funding

This work was supported by a grant from the Beijing Hope Run Special Fund of Cancer Foundation of China (No. LC2021L02).

Conflicts of interest

None.

Supplementary Material

cm9-138-3479-s001.docx ^{(1.7MB, docx)}