A high-performance metabolomic diagnostic panel for early-stage non-small cell lung cancer detection based on UPLC‒MS/MS

doi:10.21203/rs.3.rs-4293656/v1

Download PDF

Article

A high-performance metabolomic diagnostic panel for early-stage non-small cell lung cancer detection based on UPLC‒MS/MS

https://doi.org/10.21203/rs.3.rs-4293656/v1

This work is licensed under a CC BY 4.0 License

Version 1

posted

You are reading this latest preprint version

Lung cancer is the most common cancer in the world and has a consistently high mortality rate, with the majority of patients being diagnosed at an advanced stage. This study aimed to identify potential biomarkers through metabolomics to provide clues for the diagnosis and treatment of early-stage non-small cell lung cancer (NSCLC). We enrolled two prospective cohorts with a total of 180 patients (115 patients with I-II a NSCLC and 65 healthy controls) and tested serum samples for tumour markers, cytokines, and 306 metabolites by ultrahigh-performance liquid chromatography coupled with tandem mass spectrometry (UPLC‒MS/MS). In both the discovery and validation cohorts, there were 57 differentially abundant metabolites in the serum between patients with early-stage NSCLC and healthy controls, which were concentrated in the fatty acid metabolic pathway and amino acid metabolic pathway. Finally, three metabolites with significant differences were screened as isoleucine, 5Z-dodecenoic acid and 9E-tetradecenoic acid. The AUC of centralized combined diagnosis reached 0.95. This study provides new evidence that abnormalities in valine, leucine, and isoleucine metabolism and dysregulation of fatty acid synthesis may play important roles in the development of NSCLC.

Biological sciences/Cancer/Lung cancer

Biological sciences/Biochemistry/Metabolomics

early-stage NSCLC

metabolomics

fatty acid metabolism

amino acid metabolism

cytokines

Lung cancer is one of the major causes of cancer-related deaths worldwide, and according to statistics, in 2020, there were 816,000 newly diagnosed cases of lung cancer and 715,000 lung cancer-related deaths in China^1,2. One of the important reasons for the high mortality rate of lung cancer is that it occurs mostly in the form of lung nodules in the early stage and lacks specific clinical symptoms; therefore, approximately 70% of patients are already in the advanced stage at the time of diagnosis and lose the best time for treatment. Moreover, because NSCLC is the most common tissue type, and studies have shown that the 5-year survival rate of patients with stage Ia disease can reach 80% after surgery, early diagnosis of lung cancer is highly important for improving patient prognosis³.

Histopathological examination is the gold standard for the diagnosis of NSCLC. However, due to the invasive nature of histopathological examination, detecting lung cancer tumour markers is an important way to achieve early diagnosis of lung cancer, and biomarker testing using body fluids such as blood, urine, and saliva has become an important method for cancer diagnosis^4,5. In recent years, markers such as DNA methylation^6,7, circulating tumour DNA^8,9, circulating tumour cells^10,11, and microRNAs¹² have provided certain molecular biological information at an early stage to assist in the screening of early-stage lung cancer. Although the available markers show great potential, these high specimen requirements make clinical application difficult. There are no standardized biomarkers for the early diagnosis of lung cancer that are routinely used for clinical testing⁵.

Metabolomics has become an important complement to genomics and proteomics. Tumour cells undergo specific metabolic processes during development¹³. In recent years, metabolomics has been used in the development of biomarkers, and numerous scholars have conducted studies on the diagnosis^14,15 and efficacy prediction¹⁶ of lung cancer using a variety of assays and obtained excellent research results. However, we know that lung cancer is a heterogeneous disease and that metabolite levels vary among different biological samples, and different techniques are used to detect metabolites. However, the use of metabolomics in the early diagnosis of NSCLC has not yet entered clinical application, and additional clinical studies are needed to prove the accuracy and reliability of these metabolomics methods.

In this study, a total of 180 subjectspatients (I-IIa NSCLC patients and healthy controls) were enrolled in the study using UPLC‒MS/MS-targeted metabolomics technology to discover potential biomarkers and further validate them through discovery and independent validation cohorts. An attempt was made to establish a new early diagnostic model of NSCLC from a serologic perspective.

The clinical characteristics of the participants are shown in Table 1. A total of 180 volunteers from the Fourth Hospital of Hebei Medical University were enrolled in this study; 45 of these patients were included in the discovery cohort, and 70 were included in the validation cohort. Tumour markers (CEA, CYFRA21-1, NSE, ProGRP, SCC) and cytokines (IL-2, IL-4, IL-6, IL-10, IL-17A, TNF-α, INF-γ) were detected preoperatively in all patients.

Table 1

Patients characteristics for exploratory, identification and validation studies
Characteristic	Discovery cohort			Validation cohort
Characteristic	NSCLC (N = 45)	Control (N = 35)	p-Value	NSCLC (N = 70)	Control (N = 30)	p-Value
Age
< 60 years	25(55.56%)	18(51.43%)	p = 0.713	36(51.43%)	29(64.44%)	p = 0.169
≥ 60years	20(44.44%)	17(48.57%)	p = 0.713	34(48.57%)	16(35.56%)	p = 0.169
Gender
Male	23(51.11%)	17(48.57%)	p = 0.822	37(52.86%)	24(53.33%)	p = 0.960
Female	22(48.89%)	18(51.43%)	p = 0.822	33(47.14%)	21(46.67%)	p = 0.960
Smoking
No	30(66.67%)	27(77.14%)	p = 0.304	43(61.43%)	24(53.33%)	p = 0.390
Yes	15(33.33%)	8(22.86%)	p = 0.304	27(38.57%)	21(46.67%)	p = 0.390
Stage
I	28(62.22%)	26(74.29%)	p = 0.253	48(68.57%)	31(68.89%)	p = 0.971
IIa	17(37.78%)	9(25.71%)	p = 0.253	22(31.43%)	14(31.11%)	p = 0.971
Tumor size
< 1.5cm	39(86.67%)	/		61(87.14%)	/
≥ 1.5cm	6(13.33%)	/		9(12.86%)	/

In the discovery cohort, the CEA, CYFRA21-2 and NSE levels in the early-stage NSCLC cohort were significantly greater than those in the control cohort. The median levels of CEA, CYFRA211 and NSE in the early-stage NSCLC group and control group were 3.62 and 2.16 (p < 0.001), 2.99 and 2.11 (p = 0.008), and 13.26 and 10.36 (p = 0.008), respectively. For ProGRP and SCC, there was no significant difference between the early-stage NSCLC group and the control group. Similar results were obtained in the independent validation cohort (Table 2).

Table 2

Comparison of tumor biomarkers between different groups
Tumor biomarkers	Discovery cohort			Validation cohort
Tumor biomarkers	NSCLC	Control		NSCLC	Control
CEA	3.62 (2.31 ~ 4.93)	2.16 (1.64 ~ 2.95)	Z = 3.303 p < 0.001	3.54 (2.19 ~ 4.46)	2.44 (1.94 ~ 2.68)	Z = 3.946 p < 0.001
CYFRA211	2.99 (2.02 ~ 3.71)	2.11 (1.60 ~ 2.70)	Z = 2.653 p = 0.008	2.77 (1.96 ~ 3.71)	2.12 (1.57 ~ 2.97)	Z = 2.994 p = 0.003
NSE	13.26 (11.75 ~ 15.28)	10.36 (9.25 ~ 12.55)	Z = 4.452 p < 0.001	13.52 (11.38 ~ 14.64)	10.54 (9.32 ~ 11.93)	Z = 5.017 p < 0.001
ProGRP	46.54 (36.08 ~ 54.31)	45.11 (37.89 ~ 54.74)	Z = 0.373 p = 0.709	45.51 (36.76 ~ 56.39)	42.87 (34.28 ~ 51.68)	Z = 1.252 p = 0.211
SCC	0.80 (0.60 ~ 1.06)	0.64 (0.48 ~ 1.30)	Z = 0.898 p = 0.369	1.04 (0.70 ~ 1.43)	1.03 (0.71 ~ 1.37)	Z = 0.063 p = 0.950

Comparison of cytokines between the Ⅰ-Ⅱa NSCLC patients and the healthy volunteers in the discovery cohort. The results showed that the IL-2 concentration was lower in the early-stage NSCLC group than in the control group, with median values of 0.82 and 1.89, respectively (Z = 3.182, p = 0.001). IL-4, IL-6, IL-10, IL-17A and TNF-α were greater in the early-stage NSCLC group than in the control group, with median levels of 1.56 vs. 1.21 (p = 0.012), 2.41 vs. 1.63 (p = 0.012), 1.51 vs. 1.09 (p = 0.007), 7.42 vs. 3.80 (p < 0.001), and 1.98 vs. 1.54 (p < 0.001), respectively. Moreover, there was no significant difference in INF-γ between the early-stage NSCLC group and the control group (p = 0.554). Similar results were obtained in the independent validation cohort (Table 3).

Table 3

Comparison of cytokines levels in different groups
Cytokines	Discovery cohort			Validation cohort
Cytokines	NSCLC	Control		NSCLC	Control
IL_2	0.82 (0.58 ~ 1.23)	1.89 (0.95 ~ 2.94)	Z = 3.182 p = 0.001	0.84 (0.57 ~ 1.24)	1.66 (0.82 ~ 2.51)	Z = 3.407 p < 0.001
IL_4	1.56 (1.18 ~ 2.70)	1.21 (0.86 ~ 1.75)	Z = 2.522 p = 0.012	1.63 (1.19 ~ 2.37)	1.09 (0.97 ~ 1.79)	Z = 2.559 p = 0.011
IL_6	2.41 (1.87 ~ 3.23)	1.63 (0.82 ~ 2.27)	Z = 2.973 p = 0.003	2.61 (1.97 ~ 3.31)	1.75 (1.37 ~ 2.87)	Z = 3.032 p = 0.002
IL_10	1.51 (1.11 ~ 2.07)	1.09 (0.76 ~ 1.73)	Z = 2.721 p = 0.007	2.02 (1.33 ~ 2.89)	1.28 (0.96 ~ 1.96)	Z = 3.141 p = 0.002
INF_γ	3.27 (2.44 ~ 3.83)	3.19 (2.76 ~ 4.19)	Z = 0.592 p = 0.554	3.35 (2.91 ~ 4.02)	3.50 (3.00 ~ 3.95)	Z = 0.097 p = 0.922
TNF_α	1.98 (1.72 ~ 2.38)	1.54 (1.19 ~ 1.99)	Z = 3.696 p < 0.001	1.96 (1.73 ~ 2.32)	1.62 (1.33 ~ 2.01)	Z = 3.398 p < 0.001
IL_17A	7.42 (6.74 ~ 8.11)	3.80 (1.51 ~ 6.70)	Z = 5.126 p < 0.001	7.54 (6.65 ~ 8.39)	3.77 (2.70 ~ 6.49)	Z = 5.206 p < 0.001

Metabolic profiling

UPLC‒MS/MS was used to detect targeted metabolites in the two groups of samples in the discovery cohort. A total of 194 metabolites were quantitatively detected and classified as HMDB compounds, with a total of 11 classes of metabolites identified. Among these, 44 amino acids were detected, with the highest percentage being 22.68%. Fatty acids (FAs) and their derivatives, organic acids, carnitines and sugars accounted for 18.56%, 14.43%, 10.31% and 8.25%, respectively (Fig. 1a).

Principal component analysis (PCA) revealed that most of the samples were clustered within the 95% confidence interval (the oval region of the score chart). Moreover, the difference between the control group and the early-stage NSCLC group was not significant according to the PCA score. To further distinguish the differences between the two groups, OPLS-DA was performed (Fig. 1b). As shown in Fig. 1B, the control group and the early-stage NSCLC group could be distinguished, and the clustering trend of the two groups was obvious. The results of the random permutation test with 1000 randomly disrupted sample groupings showed that the R2X and Q2Y values of the OPLS-DA model were 0.801 and 0.409 (> 0.2), respectively, and the intercept of the Q2Y fitted curve on the Y-axis in the permutation test was < 0, which indicated that the model was valid (Fig. 1c).

Marker discovery

To screen potential biomarkers with biological significance, we further performed unidimensional and multidimensional statistical analysis (unidimensional analysis P < 0.05, |log2FC| >= 0 and multidimensional analysis VIP > 1). A total of 57 differentially abundant metabolites were screened, with 18 metabolites upregulated and 39 downregulated in the early-stage NSCLC group (Fig. 2a-c and Table S1). The screened differentially abundant metabolites were enriched and analysed in the KEGG hsa library, which revealed that the differentially abundant metabolites were enriched mainly in phenylalanine metabolism; phenylalanine, tyrosine and tryptophan biosynthesis; linoleic acid metabolism; FA biosynthesis; tyrosine metabolism; valine, leucine and isoleucine degradation; and propanoate metabolism (Fig. 2d).

Tumour development is closely related to fatty acid metabolism and amino acid metabolism. In this study, we focused on the fatty acid anabolic pathway, and the differentially abundant metabolites involved were myristic acid, dodecanoic acid, palmitoleic acid, octanoic acid, and oleic acid. The amino acid metabolic pathways associated with the differentially abundant metabolites were phenylalanine, tyrosine, tryptophan metabolism combined with valine, leucine, isoleucine, the beta-alanine metabolic pathway, phenylpyruvic acid, phenylalanine, tyrosine, 4-hydroxyphenylpyruvic acid combined with isoleucine, kinesin, methylmalonic acid, and beta-alanine.

To identify potential biomarkers from these candidate differentially abundant metabolites, we performed random forest (RF), support vector machine (SVM) and Boruta analyses for each differentially abundant metabolite. Using Boruta to summarize the differentially abundant metabolites obtained from RF and SVM, a total of 3 tentative and 21 confirmed significant differentially abundant metabolites were identified as potential markers (Fig. 3a). The diagnostic efficacy of the markers was evaluated using receiver operating curves, and the combination of the three differentially abundant metabolites was shown to have the best diagnostic effect for Ⅰ-Ⅱa NSCLC. The three most common SCFAs were isoleucine (AUC = 0.80), 5Z-dodecenoic acid (AUC = 0.92) and 9E-tetradecenoic acid (AUC = 0.90). The AUC of the combined detection of the three markers was 0.95 (Fig. 3b), the relative levels of isoleucine, 5Z-dodecenoic acid and 9E-tetradecenoic acid in the NSCLC group and the healthy control group are shown in Fig. 3c-e.

Marker validation

To further validate the feasibility of the screened metabolic markers as diagnostic markers for early-stage NSCLC, we re-enrolled a new cohort (70 Ⅰ-Ⅱa NSCLC patients and 30 healthy volunteers) as an independent validation cohort for the target validation of the above 57 differentially abundant metabolites. Using unidimensional and multidimensional statistical analyses (unidimensional analysis P < 0.05, |log2FC| >= 0 and multidimensional analysis VIP > 1) for differentially abundant metabolite screening, a total of 22 differentially abundant metabolites were identified, and the metabolic pathways to which they belonged were consistent with the above findings (Fig. 4a). Moreover, RF, SVM and Boruta analyses were conducted for these differentially abundant metabolites, and the three identified differentially abundant metabolites were stably and repeatedly detected in the verification cohort (Fig. 4b-e). The receiver operating curve was used to assess diagnostic efficacy. The AUC of isoleucine, 5Z-dodecenoic acid and 9E-tetradecenoic acid combined reached 0.88, and the area under the ROC curve of isoleucine, 5Z-dodecenoic acid and 9e-tetradecenoic acid was 0.78, 0.86 and 0.89, respectively. These results are similar to those obtained for the discovery cohort (Fig. 5).

Tumorigenesis and development lead to metabolic remodelling to adapt to nutritional/metabolic stress in vivo, promoting malignant cell transformation and tumour evolution. Monitoring metabolite levels is important for tumour diagnosis and treatment. In this study, we identified 57 differential serum metabolites between the early-stage NSCLC group and the control group. We screened the significant differentially abundant metabolites according to the Brouta model, performed ROC analysis, and ultimately confirmed three differentially abundant metabolites, isoleucine, 5Z-dodecenoic acid and 9E-tetradecenoic acid. The ROC-AUCs of this panel in the discovery and validation cohorts were 0.95 and 0.88, respectively.

Fifty-seven different metabolites were involved in the FA and amino acid metabolic pathways and were especially enriched in the fatty acid synthesis pathway and branched-chain amino acid (BCAA) metabolism pathway. FAs are the major components of several lipids, including phospholipids, sphingolipids, and triglycerides, which play important roles in energy storage, membrane biosynthesis, signalling, and regulation of gene transcription^17–19. In tumour cells, a high rate of de novo synthesis leading to the production of large amounts of monounsaturated fatty acids (MUFAs) is considered to be the 3rd most common typical tumour feature after alterations in glucose and glutamine metabolism¹⁹. Through scRNA-seq, Yin et al. reconfirmed that aberrant lipid metabolism is common in nontumorigenic cells, including fatty acid biosynthesis and unsaturated fatty acid biosynthesis, and further identified nine lipids as the most important features for early cancer detection through plasma untargeted lipidomics²⁰. In the present study, 5Z-dodecenoic acid and 9E-tedecenoic acid were shown to be associated with the development of NSCLC, which, to our knowledge, has not been reported previously.

Amino acids, especially BCAAs (valine, leucine and isoleucine), play important roles in tumour development. Recent evidence suggests that altered gene expression related to BCAA metabolism is observed in various cancer types and is correlated with tumour aggressiveness and treatment resistance^21,22. The TCGA database analysis showed that BCAA catabolic enzymes have prognostic value in NSCLC²³. Studies have shown that changes in BCAA metabolism specifically affect the status of tumour cells and the systemic metabolism of individuals with malignant tumours. BCAAs cannot be synthesized in the human body; they can be catabolized and metabolized by the highly reversible branched-chain amino acid aminotransferase 1/2 (BCAT1/2) to provide energy for tumour cells, or they can be reversed and converted to leucine by BCKA to activate the mTOR signalling pathway, which results in metabolic abnormalities between different cells or tissues and organs in the body and promotes tumour growth²⁴.

A study from foreign scholars included five types of cancer patients, namely, lung cancer, gastric cancer, colorectal cancer, breast cancer and prostate cancer patients, to determine the plasma free amino acid profiles of the patients and its application in early diagnosis. By comparing the changes in amino acid profiles in the blood of patients with the above five types of cancer, it was found that the AUCs for the discrimination of patients based on amino acid concentrations and ratios was 0.802 and 0.802 for lung cancer, respectively. The levels of valine and leucine change significantly in tumour patients, and the use of amino acid profiles can effectively assist in improving cancer screening and diagnosis²⁵. Another study enrolled 22 patients with COPD and 77 lung cancer patients (stage I-IV), and a collection of NMR metabolic fingerprints was modelled using OPLS-DA. The model successfully distinguished between COPD and NSCLC (AUC = 0.993), suggesting that isoleucine may be able to distinguish between COPD and lung cancer patients²⁶. Decreased concentrations of arginine, alanine, isoleucine, tyrosine and tryptophan are helpful for diagnosing breast cancer and predicting tumour progression²⁷. Elevated plasma levels of circulating metabolites of BCAAs can more than triple the risk of pancreatic cancer development²⁸.

The tumour microenvironment (TME) is a highly structured ecosystem that contains cancer cells and different types of nonmalignant cells, such as rich and diverse immune cells, cancer-associated fibroblasts (CAFs), endothelial cells (ECs), peripheral cells, various cell types and a vascularized extracellular matrix ²⁹. The various types of cells in the TME, as well as their secretion of cytokines, play key roles in the pathogenesis of cancer. IL-17 is a universal cytokine in the TME that can induce the production of multiple factors, play a multifactorial role in tumour immunity, exert antitumour effects and promote tumour progression^30,31. Abnormally elevated levels of IL-17 promote tumour progression by increasing cell proliferation and mutation rates and increasing immune tolerance in transformed cells³². Reppert et al. performed qPCR on tissue samples from 25 NSCLC patients and reported that the expression of IL-17A was 10-fold greater in these patients than in healthy controls³³. Xiao-Tang Yang et al. also reported that the level of IL-17A was greater in NSCLC patients than in healthy controls and was also significantly greater in stage III and IV patients than in stage I or II patients³⁴. IL-2 can induce the depletion of CD8 + T cells, thereby inhibiting the antitumour immune response³⁵. Our results demonstrated a significantly high level of IL-17A expression and downregulation of IL-2 in the peripheral serum of early-stage NSCLC patients, which was similar to the results of previous studies^35,36.

In recent years, there have been several studies on cfDNA methylation and fragmentation omics for early detection and localization of multiple cancers^37–39. However, due to the low content of cfDNA, there is a need for high sequencing depth, library construction and whole-genome bisulfite sequencing in combination with machine learning. Integrative profiling of the cfDNA methylome and fragmentome has been technologically challenging, and a unified standard has not been established, which limits clinical applications. In this study, we established an early-stage NSCLC diagnostic model through metabolomics detection of serum, which is simple, noninvasive and easily reproducible. It is easier to realize clinical translation, and this approach is conducive to clinical practice.

This study has several limitations. First, the sample in this study was small, but because all the patients had early-stage NSCLC, the sample was representative. Second, these markers found in the case‒control study are relatively innovative but need to be further confirmed in prospective cohort studies. Finally, at the same time, we need to explore the relevant mechanisms in animal experiments. However, whether cytokine and metabolomics detection can be combined for the early diagnosis of NSCLC and the relationship between the two need further study.

In this study, dysregulated amino acid metabolism and fatty acid metabolism were identified in early-stage NSCLC by targeted metabolomics. Isoleucine, 5Z-dodecenoic acid and 9E-tetradecenoic acid were stably detected and exhibited good diagnostic performance. In the present study, 5Z-dodecenoic acid and 9E-tedecenoic acid were shown to be associated with the development of NSCLC, which, to our knowledge, has not been reported previously. With respect to the validation cohort, the ROC-AUC of the combination of the panel was 0.88. Overall, we established an early-stage NSCLC diagnostic model through metabolomics detection of serum, which is simple, noninvasive and easily reproducible. It is easier to realize clinical translation, and this approach is conducive to clinical practice.

Study design and population

This was a case‒control study in which patients who underwent surgical resection of Ⅰ-Ⅱa NSCLC for clinical diagnosis at the Fourth Hospital of Hebei Medical University were enrolled. A total of 180 participants were enrolled, including a discovery cohort (45 patients with Ⅰ-Ⅱa NSCLC and 35 healthy volunteers) and an independent validation cohort (70 patients with confirmed Ⅰ-Ⅱa NSCLC and 30 healthy volunteers), from June-July 2020 to March-April 2022. All the subjects signed an informed consent form. The study was approved by the Ethics Committee of the Fourth Hospital of Hebei Medical University and was conducted in accordance with the Declaration of Helsinki.

Detailed clinical information was available for all subjects (Table 1). In the discovery cohort, among the patients in the early-stage NSCLC cohort, 39 had stage I NSCLC (86.67%), and 6 had stage IIa NSCLC (13.33%). Twenty-five patients had a tumour < 1.5 cm in length (55.56%), and 20 had a tumour ≥ 1.5 cm in length (44.44%). Comparisons revealed that the differences in age, sex, smoking history, and family history between patients in the early-stage NSCLC group and those in the control group were not significant (p > 0.05). There were 70 patients with stage Ⅰ-Ⅱa NSCLC in the validation cohort; 61 patients with stage I NSCLC, accounting for 87.14%; and 9 patients with stage IIa NSCLC, accounting for 12.86%. There were 37 patients with a tumour size < 1.5 cm, accounting for 52.86%, and 33 patients with a tumour size ≥ 1.5 cm, accounting for 47.14%. There were no significant differences in age, sex, smoking history, or family history between patients in the early-stage NSCLC group and those in the control group (p > 0.05).

Sample collection and processing

A total of 115 NSCLC patients underwent fasting blood collection before surgery. Five millilitres of whole blood was collected in sterile clotting BD vacuum blood collection tubes, immediately mixed 5–8 times, and incubated at room temperature (20°C ~ 25°C) for 30–120 min. After the blood had completely coagulated and the serum had sufficiently precipitated, centrifugation was performed at 1300 × g for 10 min, after which 0.2-1 mL of serum was transferred to 1.5 mL EP tubes and stored at -80°C. Moreover, 65 healthy volunteers were subjected to fasting blood collection as a control group, and the serum preparation was the same as that used for the test group.

Serum (20 µL) thawed in an ice bath was added to a 96-well plate and transferred to an Eppendorf Epomotion workstation (Eppendorf, Inc., Humburg, Germany), after which 120 µL of precooled methanol solution (with an internal standard) was added, after which the mixture was vortexed vigorously for 5 min. The 96-well plate was centrifuged at 4000 × g for 30 min at 4°C, after which the plate was placed into the workstation, 20 µL of freshly prepared derivatization reagent was added to each well, and the plate was sealed and placed at 30°C for 60 min of derivatization. Then, 330 µL of 50% methanol solution in an ice bath was added to dilute the samples, which was followed by centrifugation at 4000 × g for 30 min at 4°C. Afterwards, 135 µL of the supernatant was aspirated and transferred to a new 96-well plate, after which 10 µL of internal standard was added to each well. A gradient dilution of the derivatized standard stock solution was added to the left well, after which the plate was sealed and used for LC–MS analysis.

Peripheral serum tumour markers, such as carcinoembryonic antigen (CEA), cytokeratin 19 fragment (CYFRA21), neuron-specific enolase (NSE), progastrin releasing peptide (Pro-GRP), and squamous cell carcinoma antigen (SCC), were detected in NSCLC patients and controls by Roche's Cobas e 801 automated chemiluminescent immunoassay analyser.

IL-2, IL-4, IL-6, IL-10, TNF-a, INF-g, and IL-17A levels in the peripheral serum of NSCLC patients and controls were measured using a Beckman NAVIOS flow cytometry microsphere array.

UPLC‒MS/MS targeted metabolomics

Targeted LC‒MS/MS analyses were performed using a UPLC‒MS/MS system (ACQUITY UPLC-Xevo TQ-S, Waters, MA, USA). A total of 306 standards were obtained from Sigma‒Aldrich (St. Louis, MO, USA), Steraloids, Inc. (Newport, RI, USA) and TRC Chemicals (Toronto, ON, Canada). All the standards were accurately weighed and prepared in appropriate solutions configured as individual stock solutions at a concentration of 5.0 mg/mL. An appropriate amount of each stock solution was mixed to form a stock calibration solution.

Mobile phase A was water (containing 0.1% formic acid), mobile phase B was acetonitrile/methanol (70/30), the flow rate was 0.40 mL/min, and the column temperature was 40°C. As part of the system tuning and quality control process, composite quality control (QC) samples were prepared by mixing all the samples in equal volumes to monitor the analytical stability. QC samples were treated and tested in the same way as the analytical samples were. QC samples were handled and tested in the same manner as the analytical samples. The mass spectrometer was operated in negative mode with a 2.0 kV capillary voltage and positive mode with a 1.5 kV capillary voltage. Capillary voltage. The source and desorption temperatures were 150°C and 550°C, respectively.

Data analysis

The target raw data were processed through a calibration curve of the standards. Calibration curves were analysed by instrumental response and standard concentration methods. Target metabolites were subsequently analysed via iMAP software (v1.0; Metabo-profile; Shanghai, China). Orthogonal partial least squares discriminant analysis (OPLS-DA) was used to maximize the identification of differences in metabolic profiles between groups. In addition to multivariate statistical methods, Student's t test or Wilcoxon's rank sum test was used to estimate the significance of each metabolite. In each comparison, P values for all metabolites were adjusted for the false discovery rate. P values for comparisons were adjusted by the false discovery rate method. Finally, significantly different metabolites were required to meet the following criteria: (i) VIP > 1, (ii) log2FC > 0.25 or <-0.25, and (iii) FDR < 0.05.

Statistical analysis

All the data were analysed in R (v3.5.0), and differential metabolomics analysis was performed with the MSstats R software package, which includes logarithmic transformation, normalization, and p value calculations for Spectronaut and Skyline quantitative data. Clustering algorithms were performed using hierarchical clustering (HCA), and Euclidean distance calculations were performed with the heatmap package in R software to reveal relationships between samples and metabolites. Differentially abundant metabolites were analysed for KEGG pathway enrichment using Fisher's exact test and the KEGG database (v89.1) and compared with all identified metabolites. Correlation analysis was performed using Pearson correlation coefficient analysis with the corrplot R package.

Author contributions

H.Y.T and S.B.E. conceived of the idea, designed the study, performed research and writing – review & editing. W.L.L conducted research and data analysis, and wrote articles. L.D. and G.Y.D. conducted literature research and data collation. R.J.X. and L.G.J collected the clinical data and pathological data of all the patients. R.J.X. and G.X.X assisted with literature search and analysis. All authors reviewed the manuscript.

Data availability

The datasets used and analysed during the current study available from the corresponding author on reasonable request.

Funding

This work was supported by the Financial Department of Hebei Province [No. [2016] 361006 and No. 2017043367-2] and the Health Commission of Hebei Province [No. 20221721].

Conflicts of Interest

All authors state that they have no conflicts of interest and declare that they have neither financial nor non-financial competing interests.

Zheng, R. et al. Cancer incidence and mortality in China, 2016. Journal of the National Cancer Center 2, 1-9, doi:10.1016/j.jncc.2022.02.002 (2022).
Sung, H. et al. Global cancer statistics 2020: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J. Clin., doi:10.3322/caac.21660 (2021).
Goldstraw, P. et al. The IASLC Lung Cancer Staging Project: Proposals for Revision of the TNM Stage Groupings in the Forthcoming (Eighth) Edition of the TNM Classification for Lung Cancer. J. Thorac. Oncol. 11, 39-51, doi:10.1016/j.jtho.2015.09.009 (2016).
Hofman, P. Liquid Biopsy and Therapeutic Targets: Present and Future Issues in Thoracic Oncology. Cancers (Basel) 9, doi:10.3390/cancers9110154 (2017).
Chang, L., Li, J. & Zhang, R. Liquid biopsy for early diagnosis of non-small cell lung carcinoma: recent research and detection technologies. Biochim Biophys Acta Rev Cancer 1877, 188729, doi:10.1016/j.bbcan.2022.188729 (2022).
Hulbert, A. et al. Early Detection of Lung Cancer Using DNA Promoter Hypermethylation in Plasma and Sputum. Clin. Cancer. Res. 23, 1998-2005, doi:10.1158/1078-0432.CCR-16-1371 (2017).
Goldmann, T. et al. DNA methylation profiles of bronchoscopic biopsies for the diagnosis of lung cancer. Clin. Epigenetics 13, 38, doi:10.1186/s13148-021-01024-6 (2021).
Cohen, J. D. et al. Detection and localization of surgically resectable cancers with a multi-analyte blood test. Science 359, 926-930, doi:10.1126/science.aar3247 (2018).
Campos-Carrillo, A. et al. Circulating tumor DNA as an early cancer detection tool. Pharmacol. Ther. 207, 107458, doi:10.1016/j.pharmthera.2019.107458 (2020).
Wan, L. et al. Circulating Tumor Cell and Metabolites as Novel Biomarkers for Early-Stage Lung Cancer Diagnosis. Frontiers in oncology 11, 630672, doi:10.3389/fonc.2021.630672 (2021).
Maly, V., Maly, O., Kolostova, K. & Bobek, V. Circulating Tumor Cells in Diagnosis and Treatment of Lung Cancer. In Vivo 33, 1027-1037, doi:10.21873/invivo.11571 (2019).
Liang, X., Wu, Q., Wang, Y. & Li, S. MicroRNAs as early diagnostic biomarkers for non‑small cell lung cancer (Review). Oncol. Rep. 49, doi:10.3892/or.2022.8445 (2023).
Wu, R. Q., Zhao, X. F., Wang, Z. Y., Zhou, M. & Chen, Q. M. Novel molecular events in oral carcinogenesis via integrative approaches. J. Dent. Res. 90, 561-572, doi:10.1177/0022034510383691 (2011).
Jelonek, K. & Widłak, P. Metabolome-based biomarkers: their potential role in the early detection of lung cancer. Contemporary oncology (Poznan, Poland) 22, 135-140, doi:10.5114/wo.2018.78942 (2018).
Widłak, P. et al. Serum Metabolite Profiles in Participants of Lung Cancer Screening Study; Comparison of Two Independent Cohorts. Cancers (Basel) 13, doi:10.3390/cancers13112714 (2021).
Botticelli, A. et al. Gut metabolomics profiling of non-small cell lung cancer (NSCLC) patients under immunotherapy treatment. J. Transl. Med. 18, 49, doi:10.1186/s12967-020-02231-0 (2020).
Huang, C. & Freter, C. Lipid metabolism, apoptosis and cancer therapy. Int. J. Mol. Sci. 16, 924-949, doi:10.3390/ijms16010924 (2015).
Koundouros, N. & Poulogiannis, G. Reprogramming of fatty acid metabolism in cancer. Br. J. Cancer 122, 4-22, doi:10.1038/s41416-019-0650-z (2020).
Rohrig, F. & Schulze, A. The multifaceted roles of fatty acid synthesis in cancer. #N/A 16, 732-749, doi:10.1038/nrc.2016.89 (2016).
Wang, G. et al. Lung cancer scRNA-seq and lipidomics reveal aberrant lipid metabolism for early-stage diagnosis. Sci. Transl. Med. 14, eabk2756, doi:10.1126/scitranslmed.abk2756 (2022).
Wang, H. et al. High dose isoleucine stabilizes nuclear PTEN to suppress the proliferation of lung cancer. Discov Oncol 14, 25, doi:10.1007/s12672-023-00634-1 (2023).
Wang, Y. et al. AMPK induces degradation of the transcriptional repressor PROX1 impairing branched amino acid metabolism and tumourigenesis. Nat Commun 13, 7215, doi:10.1038/s41467-022-34747-y (2022).
Yao, X., Deng, Y., Zhou, J., Jiang, L. & Song, Y. Expression Pattern and Prognostic Analysis of Branched-Chain Amino Acid Catabolism-Related Genes in Non-Small Cell Lung Cancer. Front Biosci (Landmark Ed) 28, 107, doi:10.31083/j.fbl2806107 (2023).
Qian, L. et al. Enhanced BCAT1 activity and BCAA metabolism promotes RhoC activity in cancer progression. Nat Metab 5, 1159-1173, doi:10.1038/s42255-023-00818-7 (2023).
Miyagi, Y. et al. Plasma free amino acid profiling of five types of cancer patients and its application for early detection. PLoS One 6, e24143, doi:10.1371/journal.pone.0024143 (2011).
Deja, S. et al. Metabolomics provide new insights on lung cancer staging and discrimination from chronic obstructive pulmonary disease. J. Pharm. Biomed. Anal. 100, 369-380, doi:10.1016/j.jpba.2014.08.020 (2014).
Eniu, D. T. et al. The decrease of some serum free amino acids can predict breast cancer diagnosis and progression. Scand J Clin Lab Invest. 79, 17-24, doi:10.1080/00365513.2018.1542541 (2019).
Mayers, J. R. et al. Elevation of circulating branched-chain amino acids is an early event in human pancreatic adenocarcinoma development. Nat. Med. 20, 1193-1198, doi:10.1038/nm.3686 (2014).
de Visser, K. E. & Joyce, J. A. The evolving tumor microenvironment: From cancer initiation to metastatic outgrowth. Cancer Cell 41, 374-403, doi:10.1016/j.ccell.2023.02.016 (2023).
Kuen, D. S., Kim, B. S. & Chung, Y. IL-17-Producing Cells in Tumor Immunity: Friends or Foes? Immune Netw 20, e6, doi:10.4110/in.2020.20.e6 (2020).
Vitiello, G. A. & Miller, G. Targeting the interleukin-17 immune axis for cancer immunotherapy. J. Exp. Med. 217, doi:10.1084/jem.20190456 (2020).
Huangfu, L., Li, R., Huang, Y. & Wang, S. The IL-17 family in diseases: from bench to bedside. Signal Transduct Target Ther 8, 402, doi:10.1038/s41392-023-01620-3 (2023).
Reppert, S. et al. A role for T-bet-mediated tumour immune surveillance in anti-IL-17A treatment of lung cancer. Nat Commun 2, 600, doi:10.1038/ncomms1609 (2011).
Chen, G. et al. Th17 cell frequency and IL-17A production in peripheral blood of patients with non-small-cell lung cancer. J. Int. Med. Res. 48, 300060520925948, doi:10.1177/0300060520925948 (2020).
Liu, Y. et al. IL-2 regulates tumor-reactive CD8(+) T cell exhaustion by activating the aryl hydrocarbon receptor. Nat. Immunol. 22, 358-369, doi:10.1038/s41590-020-00850-9 (2021).
Nicola, S. et al. IL-17 Promotes Nitric Oxide Production in Non-Small-Cell Lung Cancer. J Clin Med 10, doi:10.3390/jcm10194572 (2021).
Bie, F. et al. Multimodal analysis of cell-free DNA whole-methylome sequencing for cancer detection and localization. Nat Commun 14, 6042, doi:10.1038/s41467-023-41774-w (2023).
Nguyen, V. T. C. et al. Multimodal analysis of methylomics and fragmentomics in plasma cell-free DNA for multi-cancer early detection and localization. Elife 12, doi:10.7554/eLife.89083 (2023).
Liu, M. C. et al. Sensitive and specific multi-cancer detection and localization using methylation signatures in cell-free DNA. Ann. Oncol.31, 745-759, doi:10.1016/j.annonc.2020.02.011 (2020).

No competing interests reported.

TableS1.xlsx

Download PDF

Version 1

posted

You are reading this latest preprint version

A high-performance metabolomic diagnostic panel for early-stage non-small cell lung cancer detection based on UPLC‒MS/MS

Status:

Version 1

Abstract

Figures

Introduction

Results

Metabolic profiling

Marker discovery

Marker validation

Discussion

Conclusion

Methods

Study design and population

Sample collection and processing

UPLC‒MS/MS targeted metabolomics

Data analysis

Statistical analysis

Declarations

References

Additional Declarations

Supplementary Files

Status:

Version 1