Screening of glycan biomarkers for early detection of colorectal cancer based on novel isotope labeling relative quantitative method

Abstract

Colorectal cancer (CRC) is a prevalent gastrointestinal malignancy. However, the lack of diagnostic accuracy of traditional clinical serum biomarkers carcinoembryonic antigen (CEA) and cancer antigen 19-9 (CA19-9) results in patients being diagnosed at an advanced stage. Herein, we developed a novel method of ultrahigh-performance liquid chromatography coupled to quadrupole-Orbitrap high-resolution mass spectrometry (UHPLC-Q-Orbitrap HRMS) for relative quantification based on the non-specific enzyme pronase E and an isotope mass spectrometry (MS) probe 3-benzoyl/(benzoyl-2,3,4,5,6-d5)-2-oxothiazolidine-4-carboxylic acid d0/d5-BOTC to screen novel glycan biomarkers. We applied the method in a cohort of 102 serum samples (including 51 healthy volunteers (HV), 26 stage II CRC, and 25 stage III CRC) and 90 tissue samples (including 45 paracancerous tissue and 45 cancerous tissue). Results revealed that the serum levels of H5N4F, H5N4F3SA, H4N5F1SA, and H5N4SA2 in CRC patients were significantly different from those in HV (P < 0.01). The area under the curve values of H5N4F, H5N4F3SA, and H4N5F1SA in serum samples were 0.77, 0.71, and 0.91, respectively. The clinical diagnostic accuracies of these glycans ranged from 65% to 91%, which were significantly higher than those of CEA. Additionally, differential glycan profiles in tissues were further examined using the same method and compared with serum levels. H5N4F was found to be significantly down-regulated in all CRC groups (P < 0.0001), indicating strong specificity for CRC diagnosis. The glycans identified in this study are expected to serve as potential biomarkers for the early diagnosis of CRC, offering valuable reference points for clinical diagnosis and treatment.

Graphical abstract

Highlights

• •A novel method for differential glycans in CRC patients' serum and tissue was developed.
• •Seven glycans had a significant difference in the serum and tissue of CRC patients.
• •The sensitivity of H4N5F1SA, H5N4F, H5N4F3SA, and H5N4SA2 exceeds CEA for CRC diagnosis.
• •A novel strategy for early warning of CRC stages II and III was provided.

1. Introduction

Colorectal cancer (CRC) is a prevalent malignant tumor of the digestive system and accounts for 10% of global cancer-related deaths [1]. It is the second most common cancer diagnosed in women and the third most common cancer diagnosed in men [2,3]. Notably, the majority of CRC cases are diagnosed in the late or metastatic stages, at which point the 5-year survival rate of patients is only 10% [4]. However, early diagnosis and treatment of CRC can significantly reduce the mortality rate and increase the 5-year survival rate of patients to 90% [5]. Therefore, it is crucial to detect and treat CRC in the early stages. Currently, the primary screening methods for CRC include barium enema, colonoscopy, and fecal occult blood test, which although accurate, are highly invasive and can lead to poor compliance and other complication [[6], [7], [8]]. The use of blood samples for early CRC screening can significantly improve patient compliance. Recently, Gold et al. [9] demonstrated the presence of small molecule biomarkers, such as amino acids, in the serum of CRC patients; however, the use of small molecule compounds for CRC diagnosis has a few drawbacks, including limited sensitivity and poor specificity.

Glycosylation is a post-translational modification of proteins that primarily takes place in the Golgi apparatus and endoplasmic reticulum [10]. Since glycosylation affects protein function, variations in glycoprotein biomarkers between healthy individuals and patients may indirectly indicate the presence of a disease [11]. Sethi et al. [12] reviewed the differentiation of glycans in biological samples of CRC patients and concluded that variations in glycan content are associated with the type of glycan. However, these aforementioned glycans cannot be used for the early detection of CRC since they only exhibit significant alterations in the advanced stages of CRC [13,14]. Therefore, it is crucial to identify glycans that undergo significant changes in the early stages of CRC. In protein glycosylation, glycans are enzymatically attached to proteins or peptide chains. Therefore, if the glycans are to be analyzed, they need to be separated from the protein or peptide chain for qualitative or quantitative analysis. The enzyme release method is the most effective and common methodology for glycan release [15]. However, different enzymes have varying specificity for glycans, which limits their cleavage [16,17]. Currently, PNGase F is the broadest range N-glycan-releasing enzyme that is used for glycan release. However, PNGase F is relatively expensive and thus unsuitable for large-scale use [18]. In recent years, Xu et al., Lu et al., and Li et al. [[19], [20], [21]] have discovered that pronase E can hydrolyze peptide bonds in glycoproteins, releasing glycan in the form of glycosaminoglycans. Pronase E effectively circumvents the limitations posed by PNGase F, and pronase E can theoretically act on both N-glycans and O-glycans. In addition, pronase E also preserves the structural integrity of glycans during glycan release, which enables further determination of the structure of sugar chains through mass spectrometry (MS) information.

Matrix-assisted laser desorption ionization-time of flight (MALDI-TOF) and electrospray ionization (ESI) MS are highly sensitive and accurate methods for quantitative screening and structural identification of macromolecular components, including proteins and glycans [14,22,23]. MALDI-TOF is suitable for analyzing protein sequence and glycan structure and is commonly used for qualitative analysis of glycans. However, its low sensitivity in detecting non-methylated glycans makes it unsuitable for quantitative analysis of glycans [[24], [25], [26]]. In contrast to MALDI-TOF, ESI can produce a variety of charged fragment ions by directly spraying the analytes from the solution into a MS, which enables direct detection of the molecular weight of reactants, intermediates, and products [27]. Therefore, ESI-MS is more appropriate for quantitative glycomic analysis and screening of glycan biomarkers. Moreover, owing to the lack of standards and strong hydrophilicity of glycans, it is important to develop a derivatization reagent with isotopic structure for the relative quantification of glycans [28]. Therefore, our team developed a novel isotope MS probe 3-benzoyl/(benzoyl-2,3,4,5,6-d5)-2-oxothiazolidine-4-carboxylic acid (d0/d5-BOTC), which has carboxyl groups that can specifically recognize glycosylamine treated with pronase E [29] (Figs. S1A–D). Furthermore, the isotopic structure of d0/d5-BOTC makes it suitable for the relative quantification of glycans in complex biological samples. In a previous investigation, we effectively analyzed the glycans in cetuximab and identified glycan biomarkers in the serum of patients with acute myocardial infarction through the use of d0/d5-BOTC.

In this study, we obtained the glycan from the serum and tissue samples of CRC patients and age- and sex-matched healthy volunteers (HV) using non-specific pronase E. These glycans were then labeled with d0/d5-BOTC and examined using the ultrahigh-performance liquid chromatography coupled to quadrupole-Orbitrap high-resolution mass spectrometry (UHPLC-Q-Orbitrap HRMS) and TraceFinder database (Waltham, MA, USA) (Fig. 1). The relative quantitative method was used to qualitatively screen and quantitatively analyze the glycan derivatives, and the potential glycan biomarkers for early diagnosis of CRC were screened. In addition, the specificity of the screened glycan biomarkers was further compared with that of carcinoembryonic antigen (CEA), which is currently used clinical diagnosis of CRC.

Fig. 1.

d0/d5-BOTC-labeled serum and tissue glycan derivatives examined using ultrahigh-performance liquid chromatography coupled to quadrupole-Orbitrap high-resolution mass spectrometry (UHPLC-Q-Orbitrap HRMS).

2. Materials and methods

2.1. Chemicals and reagents

LC-MS grade acetonitrile (ACN) was bought from Fisher Scientific (Pittsburgh, PA, USA). Hydroxybenzotriazole (HOBT) was obtained from Tokyo Chemical Industry Co. (Tokyo, Chiba, Japan). Trihydroxymethyl aminomethane (Tris), anhydrous ethanol (C₂H₅OH), and 1-(3-dimethylaminopropyl)-3-ethylcarbondiimine (EDC) were purchased from Aladdin Biochemical Technology Co. (Shanghai, China). Pronase E was obtained from Merck (Darmstadt, Hesse, Germany). Hematoxylin-Eosin (H&E) stain kit was purchased from Solarbio (Beijing, China). d0/d5-BOTC were synthesized in our laboratory (Yanbian, Jilin, China). Ammonia water (NH₃·H₂O) were purchased from Sigma-Aldrich (St. Louis, MO, USA). Amicon Ultra-0.5 3kDa tube was bought from Millipore (Billerica, MA, USA). Deionized and ultrapure water was obtained from a Milli-Q purification system (Unique-R20, Research Scientific Instruments Co. (Xiamen, Fujian, China). Monospin Amide columns were obtained from GL Science (Tokyo, Tokyo Metropolis, Japan).

2.2. CRC and HV samples and ethical statement

Biological samples were collected from departments of Gastrointestinal Surgery and Physical Examination of the Affiliated Hospital of Yanbian University (Jilin, China), and stored at −80 °C. The experimental group samples were consisted of 51 individuals CRC (17 females aged 48–89 years and 34 males aged 50–85 years) (Table S1). And control group serum samples were obtained from 51 age- and sex-matched HV. Both the experimental group and the control group underwent colonoscopy screening. In addition, the tissues of patients in the experimental group underwent further histopathological screening (H&E staining) to determine staging. A total of 26 stage II and 25 stage III CRC patients were ultimately selected (Figs. S2A and B). This study was conducted in accordance with the Declaration of Helsinki. Written informed consent forms were obtained from the participants. This study was approved by the Medical Ethics Committee of Yanbian University and the Affiliated Hospital of Yanbian University (Approval No.: 2023211, Approval date: February 11, 2023).

2.3. Biological sample pretreatment and glycan release

Glycans were obtained from the biological samples using a previously mentioned protocol. Briefly, 50 μL serum sample was mixed with 450 μL Tris-HCl solution (pH 7.5) containing 10 mM CaCl₂ solution and 5 μL 1.0 M DTT and vortexed. Thereafter, the mixture was heated at 100 °C for 5 min to denature the proteins. Subsequently, the mixture was cooled, and the protein samples were purified using a 3K ultrafiltration tube to eliminate <3 kDa salt molecules. The white residue in the 3K tube was collected and transferred into a 1.5 mL Eppendorf tube. Thereafter, 4 mg pronase E was added to the tube and the samples were incubated in a water bath at 37 °C for 48 h to induce glycan release. After glycan release, the samples were heated at high temperatures and subjected to ethanol precipitation to remove excessive enzyme, peptides, and proteins. The glycans were purified using the solid-phase extraction (SPE) column and vacuum dried by the vacuum concentration system at 60 °C. Lastly, the samples were dissolved in 200 μL ultrapure water and stored at −20 °C. 50 μL aliquots were added in 1.5 mL Eppendorf tubes for later derivatization.

Tissue samples (50 mg) were sectioned and stored in 450 μL 75% ethanol solution at −20 °C for 30 min to remove soluble contaminants. The tissue samples were then homogenized on ice and centrifuged at 13,000 rpm for 5 min. The supernatant was discarded and the precipitate was mixed with 450 μL Tris-HCl solution (pH 7.5) containing 10 mM CaCl₂ solution and 5 μL 1.0 M DTT and vortexed. Thereafter, the mixture was heated at 100 °C for 5 min to denature the proteins. The subsequent glycan release and purification process was consistent with the treatment of serum samples.

2.4. Glycan labeling and relative quantification strategies

In general, the asparagine residue is retained at the reducing end of the glycans after treatment with pronase E. In this study, we labeled purified glycan samples from the experimental (CRC serum and cancerous tissue) and control (HV serum and paracancerous tissue) groups using d5-BOTC and d0-BOTC, respectively, to obtain glycan derivatives. Briefly, a mixture of 50 μL of 200 mM ethylene dichloride, 50 μL of 100 mM HOBT, 50 μL of 150 mM d0/d5-BOTC, and 50 μL of glycan solution was added to a 1.5 mL Eppendorf tube and mixed at 1,500 rpm and 40 ^oC for 2 h to glycan derivatization. The experimental and control derivatives were filtered through a 0.22-μm nylon membrane and combined in a 1:1 M ratio. Lastly, 6 μL of this solution was injected into the UHPLC-Q-Orbitrap HRMS system for analysis.

2.5. Establishment of a human serum and tissue glycan database labeled with d0/d5-BOTC

Kronewitter et al. and Aldredge et al. [30,31] successfully constructed a biological sample glycan library containing 331 types of glycans using established glycobiology principles and reverse synthesis techniques. Building on this foundation, this study further combined these glycans with d0/d5-BOTC derivatization and manually determined the exact molecular formulas of the glycan derivatives, thereby creating a database of 331 d0/d5-BOTC-labeled glycan derivatives from human biological samples (Table S2). Subsequently, the molecular formulas, molecular weights, mass errors, added ions, charge states, and other relevant information for these derivatives were imported into TraceFinder v4.2 software.

2.6. Data acquisition and statistical analysis

The TraceFinder v4.2 software was used to collect and analyze MS data. Additionally, the qualitative and quantitative data of the glycan derivatives were verified using the Xcalibur v4.0 software. Unpaired t-test, paired t-test, nonparametric test, receiver operating characteristic (ROC) curve, and box plot were employed for statistical analysis using the GraphPad Prism v8.02 software for the screening of glycan biomarkers. The P value < 0.05 was considered statistically significant.

3. Results

3.1. Optimization of enzyme dosage and tissue pretreatment

In recent years, pronase E enzyme has been used for the release and analysis of glycan [20]. Nevertheless, there have been no reports on optimizing the dose of pronase E. The amount of enzyme used in MS detection is correlated with the intensity of glycan response. Insufficient enzyme may lead to incomplete glycan release, while excessive enzyme could reduce the sensitivity of glycan detection. Therefore, in this study, we used four, eight, and 12 mg pronase E to process the serum samples of six HVs (three males and three females). Then the quantity and the peak areas of the glycans were compared by MS to determine the optimal enzyme usage. The results shown most glycans were detected at high levels in the samples treated with 4 mg pronase E (Fig. S3A). Although some glycans with larger peak areas were detected by MS at an enzyme dose of 8 mg (Figs. S3B and C), more attention should be paid to the number of detectable glycans in the screening of biomarkers, and the detection limit of low abundance polysaccharides should be maximized. Therefore, this study ultimately chose 4 mg enzyme as the enzyme dosage for glycan release. Furthermore, to further elucidate and compare the distinctive glycan profiles in serum and tissue, we have applied pronase E for the first time in the release of tissue glycan. Compared to the serum samples, tissue samples had fewer glycans and had a more complex pretreatment process. For quantifying glycans in tissue samples, it is necessary to first obtain the total protein from the tissues. Therefore, we compared tissue lysis solution, 75% ethanol precipitation, and Tris-HCl for effective tissue pretreatment by calculating the content and peak area of five high mannose-type glycans after sample treatment [32,33]. As shown in Fig. S4, a total of five high mannose-type glycans were detected in the tissue samples treated with tissue lysis solution and 75% ethanol precipitation, while only two glycans were detected in the tissue samples treated with Tris-HCl. Moreover, the peak areas of the glycans treated with 75% ethanol were approximately 1.49–2.81 times larger than those treated with tissue lysis solution. This may be because some soluble impurities, such as inorganic salts and cell walls, are removed during 75% ethanol precipitation, which improves the sensitivity of glycan detection by MS. Therefore, 75% ethanol precipitation was chosen for total protein extraction from the tissue samples.

3.2. Screening of potential glycan biomarkers in the serum of CRC patients

We used the relative quantitative analysis approach to examine the glycan content in the serum of the CRC group and the control group. 17 N-glycans (high mannose-type glycans H6N2, H8N2, H9N2, and complex glycans with varying degrees of sialylation and fucosylation H4N5F1SA, H5N4F, H5N4F3SA, H5N4SA2, H3N4F, H4N4F, H5N4, H6N5F2SA2, H5N4SA, H5N4F2, H5N5F2, H5N4F2SA, H5N5F1SA, and H4N7F2) (H: hexose; N: N-acetylglucosamine; F: fucose; SA: N-acetylneuraminic acid) were screened out using the established human serum 331 glycan database and the Xcalibur v4.0 software (Table 1). To ensure the accuracy of the screened glycans and reduce false positives, the glycans screened in this study were either [M−2H]²⁻ or [M−3H]³⁻. We subsequently used exact m/z values to get extract ion chromatograms (XIC) and MS spectra of each glycan (Figs. S5, 2A and B). The blue part represents the glycans in the serum of HV labeled with d0-BOTC, while the red part represents the glycans in the serum of CRC patients labeled with d5-BOTC. The peak areas of three replicates were manually counted and summarized for subsequent statistical analysis (Fig. S6Figs. S6 and 2C–F).

Table 1.

Relative quantitative and statistical analysis of d0/d5-BOTC-labeled N-glycan derivatives in the serum samples of the colorectal cancer (CRC) patients (d5) and healthy volunteers (HV) (d0) (n = 51).

No.	Glycans	d0-glycan	d5-glycan	IVS	AR	Changes	P value	AUC
1	H4N5F1SA	1231.9075	1234.4232	[M=−2H]2−	0.07	Down	<0.0001	0.9109
2	H5N4F	1065.8466	1068.3623	[M−2H]2−	0.44	Down	<0.0001	0.7702
3	H5N4F3SA	904.6326	906.3108	[M−3H]3−	2.23	Up	<0.001	0.7105
4	H5N4SA2	1283.913	1286.4287	[M−2H]2−	1.58	Up	<0.01	0.6784
5	H6N2	870.7646	873.2803	[M−2H]2−	1.02	No	ns	0.505
6	H3N4F	903.7937	906.3094	[M−2H]2−	1.37	Up	ns	0.5525
7	H4N4F	984.8202	987.3358	[M−2H]2−	0.81	No	ns	0.5965
8	H5N4	992.8176	995.3333	[M−2H]2−	0.67	Down	ns	0.6087
9	H8N2	1032.8175	1035.3331	[M−2H]2−	1.10	No	ns	0.5527
10	H6N5F2SA2	1074.6892	1076.3673	[M−3H]3−	1.08	No	ns	0.5407
11	H9N2	1113.8439	1116.359	[M−2H]2−	1.15	No	ns	0.5582
12	H5N4SA	1138.3653	1140.881	[M−2H]2−	1.15	No	ns	0.5821
13	H5N4F2	1138.8755	1141.3912	[M−2H]2−	0.85	No	ns	0.5887
14	H5N5F2	1240.4152	1242.9309	[M−2H]2−	1.07	No	ns	0.5037
15	H5N4F2SA	1284.4232	1286.9389	[M−2H]2−	1.45	Up	ns	0.569
16	H5N5F1SA	1312.934	1315.4496	[M−2H]2−	1.06	No	ns	0.5011
17	H4N7F2	1362.4682	1364.9839	[M−2H]2−	0.98	No	ns	0.5673

H: hexose; N: N-acetylglucosamine; F: fucose; SA: N-acetylneuraminic acid; IVS: ionic valence state; AR: average ration; AUC: area under the curve; ns: no significance.

Fig. 2.

Differential glycans in the serum samples of the healthy volunteers (HV) and colorectal cancer (CRC) patients. (A) Extract ion chromatograms of the glycans in the serum samples. (B) Mass spectra of the glycans in the serum samples. (C) Relative abundances of the four significantly differentially present glycans in the serum samples of the HV and CRC patients (^∗∗P < 0.01, ^∗∗∗P < 0.001, ^∗∗∗∗P < 0.0001). (D) Box plot of the peak area ratio of the four significantly differentially present glycans in the serum samples of the HV and CRC patients. (E) Receiver operating characteristic (ROC) curves of the four significantly differentially present glycans in the serum samples of the HV and CRC patients. (F) Sensitivity of the glycan biomarkers in CRC diagnosis. H: hexose; N: N-acetylglucosamine; F: fucose; SA: N-acetylneuraminic acid; NL: ionic stregth.

3.3. Screening of potential glycan biomarkers in the tissue of CRC patients

We also performed a glycan screening in the cancerous and paracancerous tissues of CRC patients using the same technique. A total of nine glycans were identified in the cancerous and paracancerous tissues of CRC patients using the relative quantitative method, among which five were high mannose-type glycans, H5N2, H6N2, H7N2, H8N2, and H9N2, while four were complex glycans, H5N4F, H5N4F1SA, H4N4F, and H5N4 (Table 2). XIC chromatograms and mass spectra of the four differential glycans and five non-differential glycans were shown in Figs. S7A, 3A and B, respectively. Statistical analysis revealed that the four differential glycans, H5N4 (P < 0.01), H4N4F (P < 0.001), H5N4F1SA (P < 0.001), and H5N4F (P < 0.0001) showed a significantly decreasing trend in the cancerous tissues compared with the paracancerous tissues (Figs. S7B, 3C and D). Besides, the area under the curve (AUC) of H5N4F, H4N4F, and H5N4F1SA were greater than 0.7 (Table 2), indicating three glycans have certain accuracy in diagnosing CRC diseases (Fig. 3E).

Table 2.

Relative quantitative and statistical analysis of d0/d5-BOTC-labeled N-glycan derivatives in the tissue samples of the cancerous (d5) and paracancerous (d0) (n = 45).

No.	Glycans	d0-glycan	d5-glycan	IVS	AR	Changes	P value	AUC
1	H5N4F	1065.8466	1068.3623	[M−2H]2−	0.59	Down	<0.0001	0.7833
2	H5N4F1SA	1211.3943	1213.9099	[M−2H]2−	0.53	Down	<0.001	0.7323
3	H4N4F	984.8202	987.3358	[M−2H]2−	0.71	Down	<0.001	0.7067
4	H5N4	992.8176	995.3333	[M−2H]2−	0.61	Down	<0.01	0.6691
5	H5N2	789.7382	792.2539	[M−2H]2−	0.88	No	ns	0.5951
6	H6N2	870.7646	873.2803	[M−2H]2−	0.88	No	ns	0.5828
7	H7N2	951.7911	954.3068	[M−2H]2−	0.87	No	ns	0.5052
8	H8N2	1032.8175	1035.3332	[M−2H]2−	0.99	No	ns	0.6025
9	H9N2	1113.8439	1116.3596	[M−2H]2−	1.14	No	ns	0.5644

H: hexose; N: N-acetylglucosamine; F: fucose; SA: N-acetylneuraminic acid; IVS: ionic valence state; AR: average ration; AUC: area under the curve; ns: no significance.

Fig. 3.

Differential glycans in the cancerous and paracancerous tissue samples of the colorectal cancer patients. (A) Extract ion chromatograms (XIC) of the glycans in the tissue samples. (B) Mass spectra of the glycans in the tissue samples. (C) Relative abundances of the four significantly differentially present glycans in the cancerous and paracancerous tissue samples (^∗∗P < 0.01, ^∗∗∗P < 0.001, ^∗∗∗∗P < 0.0001). (D) Box plot of the peak area ratio of the four significantly differentially present glycans in the cancerous and paracancerous tissue samples. (E) Receiver operating characteristic (ROC) curves of the four significantly differentially present glycans in the cancerous and paracancerous tissue sample. H: hexose; N: N-acetylglucosamine; F: fucose; SA: N-acetylneuraminic acid; NL: ionic strength.

3.4. Differential changes in glycans in patients with different stages of CRC

According to the level of cancer cell infiltration in the tissues, CRC can be divided into different pathological stages. To explore the correlation between variations in the serum and tissue glycans in CRC patients at different stages, we conducted an in-depth analysis of the glycans from stage II and stage III CRC patients (Figs. 4A and B). Among the four glycans, the levels of H4N5F1SA, H5N4F, and H5N4F3SA in serum samples of the stage II and stage III patients were significantly different from those of the HV. The levels of H5N4F and H5N4F3SA exhibited a significant upward trend in stage III patients. H5N4SA2 was significantly differentially present in the sera of stage III patients but not of stage II patients. Compared to the serum samples, the difference in the levels of the four differential glycans was reduced in the tissue samples.

Fig. 4.

Relative abundance of the four significantly differentially present glycans in stage II and stage III colorectal cancer (CRC) patients and healthy volunteers (HV). (A) Differential glycans in the serum samples. (B) Differential glycans in the tissue samples. ^∗P < 0.01,^∗∗P < 0.01, ^∗∗∗P < 0.001, ^∗∗∗∗P < 0.0001. H: hexose; N: N-acetylglucosamine; F: fucose; SA: N-acetylneuraminic acid; ns: no significance.

4. Discussion

CRC is a major public health concern worldwide; however, early diagnosis and treatment can effectively improve the survival rate of CRC patients [4]. To date, colonoscopy is the gold standard for diagnosis of CRC [34]. In contrast to the intestines of healthy people, CRC patients' intestines contain lumps with uneven surfaces and edges, a rough texture, easy bleeding, and have a different color compared to the surrounding tissue (Fig. S2A). However, the staging of CRC patients cannot be reliably determined by colonoscopy alone. Therefore, further pathological testing is required to establish. In this study, we employed hematoxylin eosin for pathological staining of cancer tissue sections and measured the extent of inflammatory infiltration at magnifications of 40x, 100x, and 200x in order to precisely ascertain the staging of CRC patients (Fig. S2B). In comparison to HV, a pronounced accumulation of inflammatory mediators was evident within the tumor tissue of patients. Furthermore, neoplastic cells from individuals with stage II CRC were found to have breached the muscularis propria, while those from stage III patients had advanced to penetrate the serosal layer. In this study, we screened 51 stage II and III CRC patients for potential glycan biomarkers for subsequent CRC through colonoscopy and histopathology. The stage of CRC may be reliably diagnosed by combining electron microscopy and pathological investigations; however, these approaches are quite intrusive and have low patient compliance. As a result, it is crucial to screening blood biomarkers with high sensitivity and specificity.

The CEA and cancer antigen 19-9 (CA19-9), which are used for clinical diagnosis of cancer, have several drawbacks, including low sensitivity and poor specificity, making them unsuitable for CRC diagnosis. Recent research has identified that specific glycosyltransferases in the human body may induce structural alterations and aberrant expression of glycans on serum and tissue surfaces during the progression of various diseases. In the present study, we identified four differential glycans with distinct fucosylation and sialylation profiles from serum samples: H5N4F, H4N5F1SA, H5N4F3SA, and H5N4SA2. Among these, H5N4F3SA and H5N4SA2 were found to be upregulated in the sera of CRC. This upregulation is likely associated with the overexpression of ST6GAL1 and ST3GAL4 enzymes driven by the epidermal growth factor receptor (EGFR) or transforming growth factor beta (TGF-β) signaling pathways, resulting in excessive addition of sialic acid to the terminal positions of glycans, thereby increasing the levels of sialylated glycans [33,35]. However, glycan expression is not solely regulated by a single enzyme, as fucosyltransferases also play a crucial role in modulating glycan expression. Studies have demonstrated that the concentrations of FUT3 and FUT6 enzymes are abnormally expressed in the sera of CRC patients under the influence of the same signaling pathways. While the expression of core fucosylation is primarily regulated by the FUT8 transferase, abnormalities in FUT3 and FUT6 may indirectly affect the concentration of FUT8, leading to a reduction in core-fucosylated glycans. Consequently, the glycan H4N5F1SA, characterized by both sialylation and core fucosylation, exhibited a downregulated expression trend. Notably, the upregulated glycan H5N4F3SA not only contains sialylation and core fucosylation but also involves branched fucosylation, which is frequently influenced by the concentrations of FUT3 and FUT6 enzyme [[36], [37], [38]]. While insights into the causes of glycan differentiation are important for understanding disease pathogenesis, there is still a need to accurately determine the differential glycan structure of sex-specific alterations to be used as biomarkers if they are to be convenient for clinical screening of diseases. In this study, we employed pronase E digestion and relative quantitative analysis to identify glycosaminoglycans as biomarkers in the serum of CRC patients for the first time (Table 1 and Figs. 2A–D). While the expression trends of certain glycans, such as H4N5F1SA and H5N4F, were consistent with previous studies, differences in pretreatment methods resulted in variations in the specificity and accuracy of the selected biomarkers [37,39]. In this study, the AUC value for H4N5F1SA was determined to be 0.91, indicating a strong correlation with CRC. To the best of our knowledge, no previous study has reported glyco-biomarkers with individual diagnostic accuracies exceeding 0.9 for CRC screening [40,41]. Besides, we further evaluated the accuracy of four potential glycan biomarkers and commonly used CEA antigens in clinical practice in diagnosing CRC, in order to further test the practical application value of detecting possible glycan biomarkers in blood (Figs. 2E and F). Firstly, we compared the AUC values of four differential glycans and CEA in the serum of 51 patients. The results showed that, except for H5N4SA2, the AUC values of H4N5F1SA, H5N4F, and H5N4F3SA were all greater than CEA. In addition, we further compared the positive detection rates of differential glycans and CEA in 51 samples. Clinically, content > 5.0 ng/mL CEA is considered to be associated with cancer risk, but only 53% of the 51 diagnosed CRC patients met these conditions, which is consistent with previous research reports [42]. The diagnostic effectiveness of the four differential glycans for CRC was determined according to their relative quantitative ratios (d5/d0). For CRC diagnosis, the ratios < 0.8 was considered effective for the downregulated glycans and the ratios > 1.2 was considered effective for the upregulated glycans. These results indicate that the sensitivity of the four differential glycans is higher than that of CEA for CRC diagnosis, with H4N5F1SA showing > 90% sensitivity.

There is a certain correlation between glycans in serum and tissues in theory, however, some tissues may not fully release glycans into the blood or have lower concentrations in the blood, so the glycans in serum and tissues are not always completely consistent. Although the detection of tissue samples is still a highly invasive method, screening for coexisting differential glycans in serum and tissue is of great significance for accurately screening biomarkers. As reported in the literature, most of the glycans on the tissue surface were high Mannose [33], but unfortunately, high Mannose glycans did not show statistical difference (Table 2). Four complex glycans H5N4, H5N4F, H5N4F1SA, and H4N4F were also screened through relative quantification methods in this study (Figs. 3A–E). Compared to the serum glycan spectrum, a total of 9 glycans were screened in the tissues, of which 6 were the same as the glycans in the serum. Only H5N4F showed a downward trend in both serum and tissue. Therefore, H5N4F has certain clinical reference value in diagnosing CRC.

In the subsequent experiments, we further compared the variations in differential glycan across different stages of CRC. Except for H4N5F1SA, the other three glycans in the serum exhibited more pronounced differences as the condition worsened (Fig. 4A). In contrast, the differences in differential glycans in tissues gradually decreased with disease progression (Fig. 4B). As an example, the shared glycan H5N4F, which was screened in both serum and tissue, showed that with the progression of CRC, the significance of the difference in H5N4F levels between CRC patient serum and HV serum increased from P < 0.01 to P < 0.0001. In contrast, the difference between the two groups in tissue samples was significantly reduced, from P < 0.0001 to P < 0.01. The reasons for the above could be attributed to the fact that, during sample selection, the control group for the serum analysis consisted of HV, whereas the control group for the tissue samples was paracancerous tissue. Since stage III patients are already in stage of metastasis, the paracancerous tissue may have been invaded by cancer cells, reducing the differences in glycan between cancerous and paracancerous tissues. However, the above data further indicates that using the selected differential glycans can effectively distinguish CRC patients from HV, enhancing their potential and reliability as potential biomarkers for diagnosing CRC. Interestingly, we also found that H5N4SA2 showed differences in serum only in stage III patients. This may indicate that H5N4SA2 is a unique potential biomarker for stage III CRC patients. Taken together, we believe that H4N5F1SA, H5N4F, H5N4F3SA, and H5N4SA2 in serum have great potential for diagnosing CRC. However, we are well aware that the screening and clinical application of biomarkers is a complex process. Whether the differential glycans mentioned above can be truly applied to the clinical detection of CRC requires further expansion of the number of biological samples, increase in clinical sample validation, and partial clinical trials for verification.

5. Conclusion

In this study, a novel method of UHPLC-Q-Orbitrap HRMS for relative quantification base on the non-specific enzyme pronase E and an isotope MS probe d0/d5-BOTC was established to screen serum and tissues of CRC patients with stages II and III stages for glycan biomarkers. 17 glycans and nine glycans were detected in the serum and tissue samples, respectively. Among them, statistical analysis was used to screen four potential glycan biomarkers in serum, including H5N4SA2, H5N4F3SA, H5N4F, and H4N5F1SA. With AUC values of 0.68, 0.71, 0.77, and 0.91, respectively, the four serum glycan biomarkers for CRC had a greater diagnostic sensitivity than the CEA antigen that is presently employed in clinical practice. These four novel glycans may serve as potential biomarkers for early diagnosis of CRC. H4N5F1SA is the first potential biomarker for CRC with a sensitivity higher than 0.9 screened based on blood samples. Furthermore, the levels of H5N4F, H4N4F, H5N4F1SA, and H5N4 were significantly different between the cancerous and paracancerous tissues. The AUC values of H4N4F, H5N4F1SA, and H5N4F were 0.71, 0.73, and 0.78, respectively. H5N4F is a differential glycan commonly found in serum and tissues, both of which show a decreasing trend and show significant differences compared to the control group (P < 0.0001). In this study, the glycan screened can serve as potential biomarkers for early diagnosis of stage II and III CRC, providing certain reference value for clinical diagnosis and treatment of CRC, and also provides a novel methodology for identifying glycan biomarkers for the diagnosis of other diseases.

CRediT authorship contribution statement

Yuxuan Li: Writing – original draft, Methodology. Zhenggen Piao: Validation, Methodology. Songze Wang: Visualization, Investigation. Longhai Cui: Investigation. Xinyan Li: Methodology. Jinlong Ma: Investigation. Chengqiang Han: Data curation. Xi-Ling Li: Data curation, Funding acquisition. Toufeng Jin: Writing – review & editing. Jun Zhe Min: Writing – review & editing, Funding acquisition.

Declaration of competing interest

The authors declare that there are no conflicts of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: