Unveiling the atlas of associations between 1,400 plasma metabolites and 24 tumors: Mendelian randomization analyses
Original Article

Unveiling the atlas of associations between 1,400 plasma metabolites and 24 tumors: Mendelian randomization analyses

Jili Zhang1,2#, Zhibin Hao3#, Zewei Chen4#, Xingxing Su5, Wentao Xu6, Xin Jiang2, Xinwen Nian1

1Department of Urology, Changhai Hospital, Naval Medical University (Second Military Medical University), Shanghai, China; 2Department of Urology, The First Navy Hospital of Southern Theater Command, Zhanjiang, China; 3Department of Oncology, Second Mobile Corps Hospital of Chinese People’s Armed Police Force, Wuxi, China; 4Department of Nephrology, The First Navy Hospital of Southern Theater Command, Zhanjiang, China; 5Department of Oncology, China Coast Guard Hospital of the People’s Armed Police Force, Jiaxing, China; 6General Surgery Department, The First Navy Hospital of Southern Theater Command, Zhanjiang, China

Contributions: (I) Conception and design: J Zhang, Z Hao, Z Chen, X Nian; (II) Administrative support: X Jiang; (III) Provision of study materials or patients: X Su, W Xu; (IV) Collection and assembly of data: Z Hao, Z Chen, X Su; (V) Data analysis and interpretation: J Zhang, Z Hao; (VI) Manuscript writing: All authors; (VII) Final approval of manuscript: All authors.

#These authors contributed equally to this work.

Correspondence to: Wentao Xu, PhD. General Surgery Department, The First Navy Hospital of Southern Theater Command, No. 40 Haibin 3rd Road, Zhanjiang 524000, China. Email: axuwentao@qq.com; Xin Jiang, PhD. Department of Urology, The First Navy Hospital of Southern Theater Command, No. 40 Haibin 3rd Road, Zhanjiang 524000, China. Email: jiangx_1314666@163.com; Xinwen Nian, PhD. Department of Urology, Changhai Hospital, Naval Medical University (Second Military Medical University), 168 Changhai Road, Shanghai 200433, China. Email: 18801765626@163.com.

Background: Association between plasma metabolites and pan-cancer remains controversial. Herein, we performed a two-sample Mendelian randomization (MR) analysis to verify whether there is a causal relationship between the two and to point the way for cancer metabolism research.

Methods: In our research, we downloaded 1,400 plasma metabolites from a large genome-wide association study (GWAS). We also obtained GWAS summary statistics for 24 types of cancers from the publicly available GWAS database, totaling 5,003,410 European individuals. We mainly used the fixed/random-effects inverse variance-weighted (IVW) method for two-sample MR analysis.

Results: In a combined sample of 291,202 cancer cases and 4,712,208 controls, a total of 55 plasma metabolites were identified as causally associated with nine types of cancer as a result of our MR analysis [P<0.05, false discovery rate (FDR) <0.2], including methionine sulfone, gamma-glutamylcitrulline, alliin, tetradecanedioate, hexadecanedioate, glutarate, ceramide, linolenoylcarnitine, hydroxypalmitoyl sphingomyelin, 1-palmitoyl-2-linoleoyl-glycerylphosphorylcholine (1-palmitoyl-2-linoleoyl-GPC), 3-acetylphenol sulfate, retinol (vitamin a) to linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) ratio, etc. Reverse MR analysis revealed a causal relationship between lung cancer and the only plasma metabolite, 1-palmitoyl-2-linoleoyl-GPC (P<0.05, FDR <0.2).

Conclusions: Our study provides a comprehensive atlas of cancer-related plasma metabolites, offering possible targets for cancer detection, as well as a reference for future research on tumorigenesis mechanisms and therapeutic targets.

Keywords: Mendelian randomization (MR); plasma metabolites; metabolism; pan-cancer; cancer detection


Submitted Mar 06, 2024. Accepted for publication Jul 19, 2024. Published online Sep 09, 2024.

doi: 10.21037/tcr-24-359


Highlight box

Key findings

• This study revealed the causal relationship between some plasma metabolites and several cancers.

What is known and what is new?

• Abnormal changes of different tumor markers in plasma can predict the occurrence, development and treatment prognosis of the corresponding tumor.

• Our results show that different plasma metabolite changes predict different cancers, and we provide a comprehensive tumor-associated plasma metabolite mapping.

What is the implication, and what should change now?

• The tumor-associated plasma metabolite profiles in this study provide targets for plasma cancer detection and suggest basic oncology research approaches.


Introduction

Cancer has been a threat to human health since its discovery (1). In 2023, it was expected to have 1,958,310 new cancer cases and 609,820 cancer deaths in the United States (2), and malignant tumors are the major cause of death among people aged 45 to 64 years (3). In the current times, although malignant tumors have gone through several iterations of treatment modalities, they remain one of the major risk factors for patient mortality (1-3). As humans continue to fight against tumors, our understanding of the complex pathogenesis of tumors has improved and plasma tumor detection has become possible (4,5). Plasma tumor testing is used to find tumor-associated plasma substances in a patient’s blood to provide a basis for tumor diagnosis (4). However, plasma cancer detection also faces a number of problems, including how to differentiate between the genetic mutations and metabolic changes that result from normal physical aging and those that result from tumors (6-9). This requires specific markers that are causally related to the onset of malignant tumors. Specific markers of malignant tumors can be divided into tissue and serological sources, depending on how they are detected (10). It is worth highlighting that liquid phase markers (serologic markers/plasma markers) have the advantage of being more sensitive, safer, and less damaging to patients compared to histologic markers (11).

Some of the existing cancer plasma markers are malignant cell metabolites, such as neuron-specific enolase (NSE) (12), which marks small cell lung cancer (SCLC), polyamine (13) and alpha-fetoprotein (AFP) (14,15), which are associated with hepatocellular carcinoma (HCC), suggesting that we can look for new plasma markers from tumor metabolism. Classically, aberrant phospholipid metabolism is one of the features of tumor metabolism (16,17), and even an important hallmark of tumors (18). Phospholipid metabolism regulates the composition of cell membranes, modulates tumor cell energy accumulation, influences cell signaling, governs tumor immunity, and plays an important role in the growth, metastasis, and resistance of tumor cells to drugs (19). Next, disorders of glucose metabolism occur in various types of malignant tumors (20). Inadequate blood supply within the tumor leads to a hypoxic environment that induces hypoxia-inducible factors 1 and 2 (expression of HIF-1 and HIF-2), which in turn leads to the expression of glucose transporter proteins, GLUT1 and GLUT3, which promotes glucose uptake and allows glucose to enter into the glycolytic pathway. This enhanced glycolysis promotes tumor cell migration, induces angiogenesis, and mediates chemoresistance (21,22). With the reprogramming of glycolysis, in the absence of glucose, amino acids can be involved in metabolism, such as elevated levels of glutamine metabolism (23,24). Precursors of glutamine and its metabolites can regulate signaling pathways, proliferation, and metastasis of tumor cells by modulating the tricarboxylic acid cycle (TCA cycle) in tumor cells (25,26). In addition, the role of altered lipid metabolism in tumor growth has attracted attention (27). Unlike other metabolites that are directly involved in energy metabolism, lipid metabolism mainly provides biomolecules for metabolites (28,29). In conclusion, with the development of metabolomics, more and more new metabolites have been defined, and it is practical and effective to explore the causality of tumorigenesis through a metabolic perspective (30-32).

Observational studies are one of the commonly used methods when exploring causal relationships in research subjects (33), but often vary from the results of randomized controlled trials (RCTs) when confronted with real-world issues due to confounding factors and reverse causation (33). However, RCTs require long-term follow-up with a large clinical cohort and can be expensive (34-37). Mendelian randomization (MR) analysis is an emerging analytical method that further assesses the reliability of causality by adding genetic variation as instrumental variables (IVs) between exposure and outcome to overcome the limitations of RCTs and observational studies (38). Although some MR analyses suggest a causal relationship between metabolites and cancer, to our knowledge there have been no comprehensive MR analyses exploring the causality between plasma metabolites and pan-cancers (39). In this study, we used the most current and comprehensive human plasma metabolomics data to explore causal relationships with 24 common human tumors, with the aim of obtaining plasma metabolites that could be used for cancer detection and providing a reference for research on tumorigenesis mechanisms and therapeutic targets. We present this article in accordance with the STROBE-MR reporting checklist (available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-359/rc).


Methods

Study design

In this study, a two-sample MR analysis was used to identify causal relationships between 1,400 plasma metabolites (including 1,091 blood metabolites and 309 metabolite ratios) and 24 types of cancers (including 20 solid tumors: bladder cancer, brain cancer, breast cancer, cervical cancer, colorectal cancer, corpus uteri cancer, endometrial cancer, esophageal adenocarcinoma, gastric cancer, liver cancer, lung cancer, malignant melanoma, malignant neoplasm of head and neck, oral cavity cancer, oropharynx cancer, ovarian cancer, pancreatic cancer, prostate cancer, renal cell carcinoma, thyroid cancer; and 4 non-solid tumors: chronic lymphocytic leukemia, diffuse large B-cell lymphoma, Hodgkin lymphoma, multiple myeloma). From substantial genome-wide association studies (GWASs), 5,003,410 European individuals, including 291,202 cancer cases and 4,712,208 controls across 24 types of cancer were included. Considering the fact that the study used open-access summary data, no further participant consent or ethical approval was required. In order to minimize the potential bias in results, IVs must satisfy three core assumptions: (I) IVs exhibit a significant association with the exposure; (II) genetic IVs are independent of potential confounders; and (III) IVs influence outcome only through exposure. The directionality of the causal linkage was further determined with a bidirectional MR design, which ruled out a potential reverse causal linkage. A comprehensive study design was presented in Figure 1. This study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Figure 1 The study design. Flowchart of the study. GWAS, genome-wide association study; IVs, instrumental variables; MR, Mendelian randomization; LD, linkage disequilibrium.

Plasma metabolites data

GWAS summary statistics for plasma metabolites were obtained from the GWAS Catalog (https://www.ebi.ac.uk/gwas/, accession numbers from GCST90199621 to GCST90201020; table available at https://cdn.amegroups.cn/static/public/tcr-24-359-1.xlsx) (40). A cohort of 8,299 unrelated European individuals participating in the metabolomics study underwent genome-wide genotyping and had their circulating plasma metabolites measured in the Canadian Longitudinal Study of Aging (CLSA). Batch-normalized metabolites levels were used in the study, and metabolites with missing measurements were detected in less than 50% of samples (n=1,091). According to the Human Metabolome Database (HMDB), 309 metabolite pairs share enzymes or transporters. By dividing one metabolite’s batch-normalized measurement value by the other metabolite’s measurement in the same individual, the metabolite ratio was calculated for each pair of metabolites.

Resource of 24 types of cancers

Data from publicly available databases were used to generate summary statistics on GWAS for cancer outcomes. The cancer outcomes included a total of 5,003,410 European individuals (291,202 cancer cases and 4,712,208 controls, Table 1). FinnGen studies, R9 release (https://www.finngen.fi), a research project involving European descent participants, supplied summary statistics for bladder cancer (2,053 cases, 287,137 controls), brain cancer (764 cases, 287,137 controls), chronic lymphocytic leukemia (624 cases, 287,133 controls), corpus uteri cancer (1,967 cases, 167,189 controls), diffuse large B-cell lymphoma (1,010 cases, 287,137 controls), gastric cancer (1,307 cases, 287,137 controls), Hodgkin lymphoma (780 cases, 299,952 controls), malignant melanoma (3,960 cases, 286,874 controls), malignant neoplasm of head and neck (2,131 cases, 287,137 controls), pancreatic cancer (1,416 cases, 287,137 controls), and thyroid cancer (1,783 cases, 287,137 controls). IEU Open GWAS project (https://gwas.mrcieu.ac.uk/) provided breast cancer (122,977 cases, 105,974 controls), cervical cancer (1,889 cases, 461,044 controls), colorectal cancer (6,581 cases, 463,421 controls), endometrial cancer (12,906 cases, 108,979 controls), esophageal adenocarcinoma (4,112 cases, 17,159 controls), liver cancer (775 cases, 1,332 controls), lung cancer (11,348 cases, 15,861 controls), multiple myeloma (601 cases, 372,016 controls), oral cavity cancer (1,223 cases, 2,928 controls), oropharynx cancer (1,119 cases, 2,329 controls), ovarian cancer (25,509 cases, 40941 controls), and prostate cancer (79,148 cases, 61,106 controls). The renal cell carcinoma dataset (male: 3,227 cases, 4,916 controls; female: 1,992 cases, 3,095 controls) was downloaded from GWAS Catalog (https://www.ebi.ac.uk/gwas/). Table 1 provided detailed information.

Table 1

Information of cancer-specific GWAS datasets

Cancer type Resource Population Sample size (n=5,003,410)
Cases (n=291,202) Controls (n=4,712,208)
Bladder cancer FINNGEN R9 European 2,053 287,137
Brain cancer FINNGEN R9 European 764 287,137
Breast cancer IEU OpenGWAS project: ieu-a-1126 European 122,977 105,974
Cervical cancer IEU OpenGWAS project: ukb-b-8777 European 1,889 461,044
Chronic lymphocytic leukemia FINNGEN R9 European 624 287,133
Colorectal cancer IEU OpenGWAS project: ebi-a-GCST90018808 European 6,581 463,421
Corpus uteri cancer FINNGEN R9 European 1,967 167,189
Diffuse large B-cell lymphoma FINNGEN R9 European 1,010 287,137
Endometrial cancer IEU OpenGWAS project: ebi-a-GCST006464 European 12,906 108,979
Esophageal adenocarcinoma IEU OpenGWAS project: ebi-a-GCST003739 European 4,112 17,159
Gastric cancer FINNGEN R9 European 1,307 287,137
Hodgkin lymphoma FINNGEN R9 European 780 299,952
Liver cancer IEU OpenGWAS project: ebi-a-GCST90092003 European 775 1,332
Lung cancer IEU OpenGWAS project: ieu-a-966 European 11,348 15,861
Malignant melanoma FINNGEN R9 European 3,960 286,874
Malignant neoplasm of head and neck FINNGEN R9 European 2,131 287,137
Multiple myeloma IEU OpenGWAS project: ieu-b-4957 European 601 372,016
Oral cavity cancer IEU OpenGWAS project: ieu-b-94 European 1,223 2,928
Oropharynx cancer IEU OpenGWAS project: ebi-a-GCST012242 European 1,119 2,329
Ovarian cancer IEU OpenGWAS project: ieu-a-1120 European 25,509 40,941
Pancreatic cancer FINNGEN R9 European 1,416 287,137
Prostate cancer IEU OpenGWAS project: ebi-a-GCST006085 European 79,148 61,106
Renal cell carcinoma (female) GWAS Catalog: GCST008225 European 1,992 3,095
Renal cell carcinoma (male) GWAS Catalog: GCST008226 European 3,227 4,916
Thyroid cancer FINNGEN R9 European 1,783 287,137

GWAS, genome-wide association study.

Selection of IVs

For forward MR analysis, the significance level of IVs associated with plasma metabolites was set to 1×10−5. For obtaining IVs with independent loci, we set the linkage disequilibrium (LD) threshold of r2<0.001 and clumping distance =10,000 kb in 1,000 Genomes European data using the “TwoSampleMR” package. We calculated F-statistics to verify IV strength and mitigate weak instrumental bias, F-statistics greater than 10 were considered to be sufficiently strong for the correlation between IVs and exposure. F-statistics were calculated through the following formula: F = R2*(n − k − 1)/k*(1 − R2), where “n” signifies the sample size, “k” denotes the number of IVs, and “R2” indicates the portion of the exposure variance elucidated by the IVs (41). We adopted the following formula to calculate R2 for more than 10 single nucleotide polymorphisms (SNPs): R2=2*(beta2)*EAF*(1 − EAF)/[2*(beta2)*EAF*(1 − EAF) + 2*(SE2)*N*EAF*(1 − EAF)); whereas for less than 10 SNPs we used: R2=2*(beta2)*EAF*(1 − EAF), where EAF identifies the effect allele frequency, beta denotes the effect size of SNP on exposure, Ν is the sample size, and SE is the standard error of the beta (42). Last but not least, we used “PhenoScanner” (http://www.phenoscanner.medschl.cam.ac.uk/) to eliminate SNPs that definitely correlated with cancer risk factors.

In the reverse MR analysis, we used plasma metabolites with positive results in forward MR as the outcome and corresponding cancers as the exposure to explore the causality of each. We screened for IVs related to cancers at threshold (P<5×10−8 for breast cancer, colorectal cancer, endometrial cancer, lung cancer, and ovarian cancer; P<1×10−6 for bladder cancer and pancreatic cancer; P<1×10−5 for brain cancer, cervical cancer, and oropharynx cancer). The remaining steps of the IVs screening in reverse MR analysis were similar to that described above for forward MR analysis.

Statistical analysis

All statistical analyses were conducted with R software (version 4.3.1) using the package “TwoSampleMR” (version 0.5.7). To evaluate the causal relationship between 1,400 plasma metabolite and 24 types of cancers, the fixed/random-effects inverse variance-weighted (IVW) method, weighted median method, simple mode, weighted mode, and MR-Egger regression were mainly performed. The IVW method, which provides the most precise effect estimates, was performed as the primary analysis in this study and for almost all MR-analysis (43-45). Multiple correction of the IVW P value was performed by the false discovery rate (FDR) method. The Cochran’s Q test was conducted to scrutinize the heterogeneity of the selected SNPs. In cases where significant heterogeneity emerged (P<0.05), a random-effects IVW method was chosen; if not, a fixed-effects IVW method was applied (46). Moreover, we implemented the MR-Egger intercept test to detect horizontal pleiotropy (statistically significant if P<0.05) (47). Leave-one-out analysis was performed to assess whether there was a pleiotropic effect on the results after the removal of individual SNP (48). In addition, the MR pleiotropy residual sum and outlier (MR-PRESSO) test which conducted a global test of heterogeneity was used to identify potential outliers in each SNP and obtain a corrected association result after removing the potential outliers (49). In order to increase the reliability and stability of the results, we used five criteria for screening positive results: (I) P value of IVW was less than 0.05; (II) P value of FDR-corrected IVW was less than 0.2; (III) the trend of OR values of the five MR analytical methods was consistent; (IV) MR-Egger intercept test P value was greater than 0.05; (V) MR-PRESSO global test P value was greater than 0.05. Lastly, reverse MR analysis was conducted to investigate any reverse causal relationship between 10 types of cancers and plasma metabolites.


Results

Causality of plasma metabolites on tumorigenesis

Following a series of rigorous quality control procedures as described above, 34,843 SNP associated with 1,400 plasma metabolites were selected as IVs for MR analysis (table available at https://cdn.amegroups.cn/static/public/tcr-24-359-2.xlsx). Based on a set of rigorous screening criteria, we identified suggestive evidence that 2 plasma metabolites were associated with bladder cancer risk, 1 plasma metabolite with Brain cancer risk, 6 plasma metabolites with Breast cancer risk, 12 plasma metabolites with cervical cancer risk, 7 plasma metabolites with colorectal cancer risk, 10 plasma metabolites with endometrial cancer risk, 3 plasma metabolites with lung cancer risk, 1 plasma metabolite with oropharynx cancer risk, 10 plasma metabolites with ovarian cancer risk, and 3 plasma metabolites with pancreatic cancer risk (Figure 2). Summary information for IVs used in these significant results was listed in the online table (available at https://cdn.amegroups.cn/static/public/tcr-24-359-3.xlsx). An analysis of five MR analysis methods and sensitivity analysis further confirmed the robustness of the causal associations (table available at https://cdn.amegroups.cn/static/public/tcr-24-359-4.xlsx). Specifically, both the MR-Egger intercept test and the MR-PRESSO global test ruled out the possibility of horizontal pleiotropy (Table 2). Forest plots, funnel plots, leave-one-out plots, and scatter plots also indicated the stability of the results (Figures S1-S4).

Figure 2 Significant MR results of causal effects between plasma metabolites and cancer risk. N.SNP, the number of SNPs used as IVs; SNP, single-nucleotide polymorphism; IVs, instrumental variables; IVW, inverse variance-weighted; FDR, false discovery rate; OR, odds ratio; CI, confidence interval; MR, mendelian randomization; GPC, glycerylphosphorylcholine; GPG, glycerophosphoethanolamine; GPE, glycero-3-phosphorylglycerol.

Table 2

The horizontal pleiotropy test results of causal effects between plasma metabolites and cancer risk

Cancer type (outcome) Plasma metabolites/metabolite ratios (exposure) MR-Egger intercept test P value MR-PRESSO global test P value
Bladder cancer 5-acetylamino-6-formylamino-3-methyluracil levels 0.791 0.492
Paraxanthine to 5-acetylamino-6-formylamino-3-methyluracil ratio 0.566 0.327
Brain cancer Adenosine 5'-monophosphate (AMP) to flavin adenine dinucleotide (FAD) ratio 0.901 0.855
Breast cancer Docosadienoate (22:2n6) levels 0.740 0.302
Myristoleate (14:1n5) levels 0.226 0.137
Cysteinylglycine disulfide levels 0.167 0.526
3-bromo-5-chloro-2,6-dihydroxybenzoic acid levels 0.501 0.332
Histidine to pyruvate ratio 0.261 0.872
Caffeine to paraxanthine ratio 0.930 0.068
Cervical cancer N-alpha-acetylornithine levels 0.220 0.499
Alliin levels 0.195 0.406
Methionine sulfone levels 0.770 0.943
Gamma-glutamylcitrulline levels 0.963 0.429
N-acetylleucine levels 0.567 0.837
X-24518 levels 0.537 0.891
N-acetyltyrosine levels 0.912 0.729
N-acetyl-L-glutamine levels 0.554 0.647
N-acetylarginine levels 0.936 0.968
N-acetylcitrulline levels 0.845 0.811
N-acetyl-1-methylhistidine levels 0.433 0.595
N-delta-acetylornithine levels 0.744 0.846
Colorectal cancer 1,2-dipalmitoyl-GPC (16:0/16:0) levels 0.814 0.783
1-palmitoyl-2-stearoyl-GPC (16:0/18:0) levels 0.927 0.570
1-stearoyl-2-arachidonoyl-GPE (18:0/20:4) levels 0.585 0.105
1-oleoyl-2-linoleoyl-GPE (18:1/18:2) levels 0.054 0.074
1-palmitoyl-2-arachidonoyl-GPC (16:0/20:4n6) levels 0.176 0.470
Retinol (vitamin A) to linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) [1] ratio 0.674 0.585
Arachidonate (20:4n6) to linoleate (18:2n6) ratio 0.206 0.183
Endometrial cancer Tetradecanedioate (C14-DC) levels 0.146 0.554
Hexadecanedioate (C16-DC) levels 0.608 0.119
5alpha-androstan-3alpha,17beta-diol monosulfate (1) levels 0.638 0.429
1-oleoyl-GPG (18:1) levels 0.588 0.334
Glycodeoxycholate 3-sulfate levels 0.183 0.313
1-linoleoyl-GPG (18:2) levels 0.109 0.598
Octadecenedioate (C18:1-DC) levels 0.053 0.236
Glutarate (C5-DC) levels 0.211 0.854
X-22509 levels 0.886 0.966
Adenosine 5'-diphosphate (ADP) to pantothenate ratio 0.833 0.794
Lung cancer 5alpha-androstan-3beta,17alpha-diol disulfate levels 0.361 0.662
1-palmitoyl-2-linoleoyl-GPC (16:0/18:2) levels 0.584 0.602
1-stearoyl-2-arachidonoyl-GPC (18:0/20:4) levels 0.569 0.579
Oropharynx cancer 3-acetylphenol sulfate levels 0.189 0.561
Ovarian cancer 5-acetylamino-6-amino-3-methyluracil levels 0.555 0.431
2R,3R-dihydroxybutyrate levels 0.997 0.913
Ceramide (d18:1/14:0, d16:1/16:0) levels 0.144 0.740
Linolenoylcarnitine (C18:3) levels 0.680 0.462
N-lactoyl phenylalanine levels 0.596 0.889
Hydroxypalmitoyl sphingomyelin [d18:1/16:0(OH)] levels 0.736 0.761
X-12221 levels 0.286 0.787
X-12410 levels 0.385 0.375
5-acetylamino-6-formylamino-3-methyluracil levels 0.224 0.940
Spermidine to N-acetylputrescine ratio 0.129 0.734
Pancreatic cancer X-18888 levels 0.602 0.980
X-24494 levels 0.269 0.401
Retinol (vitamin A) to linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) [1] ratio 0.303 0.924

MR, Mendelian randomization; PRESSO, pleiotropy residual sum and outlier; GPC, glycerylphosphorylcholine; GPG, glycerophosphoethanolamine; GPE, glycero-3-phosphorylglycerol.

Genitourinary systems

In the urinary system, a decrease in 5-acetylamino-6-formylamino-3-methyluracil levels [odds ratio (OR) =0.837, 95% confidence interval (CI): 0.765–0.916, P=1.07E−04, FDR =0.075) and an increase in paraxanthine to 5-acetylamino-6-formylamino-3-methyluracil ratio (OR =1.215, 95% CI: 1.113–1.327, P=1.36E−05, FDR =0.019) were positively associated with bladder carcinogenesis (Figure 2). In the reproductive system, 32 metabolites were causally associated with cervical, endometrial, and ovarian cancers. As presented in Figure 2, we found that N-alpha-acetylornithine levels (OR =1.002, 95% CI: 1.001–1.003, P=1.02E−04, FDR =0.020), N-acetylleucine levels (OR =1.001, 95% CI: 1.000–1.001, P=1.51E−04, FDR =0.023), N-acetyltyrosine levels (OR =1.001, 95% CI: 1.000–1.001, P=5.25E−05, FDR =0.018), N-acetyl-L-glutamine levels (OR =1.001, 95% CI: 1.000–1.001, P=8.13E−05, FDR =0.019), N-acetylarginine levels (OR =1.001, 95% CI: 1.000–1.001, P=2.55E−05, FDR =0.018), N-acetylcitrulline levels (OR =1.001, 95% CI: 1.000–1.001, P=3.94E−05, FDR =0.018), and N-acetyl-1-methylhistidine levels (OR =1.001, 95% CI: 1.000–1.001, P=1.36E−04, FDR =0.023) increased the risk of cervical cancer, while alliin levels (OR =0.999, 95% CI: 0.998–1.000, P=1.64E−03, FDR =0.191), methionine sulfone levels (OR =0.999, 95% CI: 0.998–0.999, P=1.45E−05, FDR =0.018), gamma-glutamylcitrulline levels (OR =0.998, 95% CI: 0.997–0.999, P=5.04E−04, FDR =0.070), X-24518 levels (OR =0.999, 95% CI: 0.998–1.000, P=1.33E−03, FDR =0.168), and N-delta-acetylornithine levels (OR =0.999, 95% CI: 0.999–1.000, P=7.65E−05, FDR =0.019) were negatively associated with the risk of cervical cancer. Meanwhile, we identified a total of ten plasma metabolites that were positively associated with the risk of endometrial cancer, including tetradecanedioate (C14-DC) levels, hexadecanedioate (C16-DC) levels, 5alpha-androstan-3alpha,17beta-diol monosulfate (1) levels, 1-oleoyl-glycero-3-phosphorylglycerol (GPG) (18:1) levels, glycodeoxycholate 3-sulfate levels, 1-linoleoyl-GPG (18:2) levels, octadecenedioate (C18:1-DC) levels, glutarate (C5-DC) levels, X-22509 levels, and adenosine 5’-diphosphate (ADP) to pantothenate ratio. As for ovarian cancer, 10 plasma metabolites were statistically associated with the risk of ovarian cancer, including 4 protective risk factors [2R,3R-dihydroxybutyrate levels, linolenoylcarnitine (C18:3) levels, N-lactoyl phenylalanine levels, and spermidine to N-acetylputrescine ratio] and 6 risk factors [5-acetylamino-6-amino-3-methyluracil levels, ceramide (d18:1/14:0, d16:1/16:0) levels, hydroxypalmitoyl sphingomyelin [d18:1/16:0(OH)] levels, X-12221 levels, X-12410 levels, and 5-acetylamino-6-formylamino-3-methyluracil levels]. In summary, our results revealed specific plasma metabolites associated with the risk of genitourinary tumors and provided new perspectives for human understanding of the pathogenesis of genitourinary tumors.

Digestive systems

In the digestive system, 1,2-dipalmitoyl-glycerylphosphorylcholine (1,2-dipalmitoyl-GPC) (16:0/16:0) levels (OR =1.133, 95% CI: 1.073–1.197, P=7.37E−06, FDR =0.005), 1-palmitoyl-2-stearoyl-GPC (16:0/18:0) levels (OR =1.119, 95% CI: 1.065–1.177, P=1.05E−05, FDR =0.005), 1-stearoyl-2-arachidonoyl-glycerophosphoethanolamine (GPE) (18:0/20:4) levels (OR =1.076, 95% CI: 1.032–1.121, P=6.24E−04, FDR =0.109), 1-palmitoyl-2-arachidonoyl-GPC (16:0/20:4n6) levels (OR =1.083, 95% CI: 1.047–1.120, P=4.04E−06, FDR =0.005), and arachidonate (20:4n6) to linoleate (18:2n6) ratio (OR =1.112, 95% CI: 1.053–1.175, P=1.58E−04, FDR =0.033) were positively correlated with the risk of colorectal cancer, while 1-oleoyl-2-linoleoyl-GPE (18:1/18:2) levels (OR =0.929, 95% CI: 0.889–0.970, P=9.02E−04, FDR =0.140) and retinol (vitamin A) to linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) ratio (OR =0.903, 95% CI: 0.857–0.951, P=1.19E−04, FDR =0.033) were negatively associated with colorectal cancer (Figure 2, Table 2). In addition, high X-18888 levels (OR =1.334, 95% CI: 1.140–1.560, P=3.15E−04, FDR =0.183) and high retinol (vitamin A) to linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) ratio (OR =1.365, 95% CI: 1.168–1.596, P=1.19E−04, FDR =0.125) could promote the onset of pancreatic cancer, while X-24494 levels (OR =0.722, 95% CI: 0.602–0.866, P=4.43E−04, FDR =0.183) could be a protective factor against pancreatic cancer (Figure 2, Table 2). Notably, the retinol (vitamin A) to linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) ratio plays diametrically opposed roles in colorectal and pancreatic cancers, which may suggest differences in metabolic levels pathway between the two cancers. Our results illustrate that in digestive system tumors, different tissue origins have different metabolic backgrounds, which may suggest new solutions for screening cancers of the digestive tract.

Other cancers

We also found that high 5alpha-androstan-3beta,17alpha-diol disulfate levels (OR =1.193, 95% CI: 1.096–1.298, P=4.56E−05, FDR =0.022) and high 1-stearoyl-2-arachidonoyl-GPC (18:0/20:4) levels (OR =1.094, 95% CI: 1.045–1.145, P=1.34E−04, FDR =0.047) were causally associated with lung cancer development, whereas 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2) levels (OR =0.822, 95% CI: 0.763–0.886, P=2.66E−07, FDR <0.001) was a protective factor for lung cancer. As for breast cancer, 6 plasma metabolites were statistically associated with the risk of breast cancer, including 2 protective factors (3-bromo-5-chloro-2,6-dihydroxybenzoic acid levels and Caffeine to paraxanthine ratio) and 4 risk factors [docosadienoate (22:2n6) levels, myristoleate (14:1n5) levels, cysteinylglycine disulfide levels, and histidine to pyruvate ratio]. Meanwhile, 3-acetylphenol sulfate levels (OR =0.448, 95% CI: 0.305–0.657, P=4.04E−05, FDR =0.057) were negatively associated with the risk of oropharynx cancer (Figure 2, Table 2). In brain cancer, adenosine 5’-monophosphate (AMP) to flavin adenine dinucleotide (FAD) ratio (OR =1.894, 95% CI: 1.427–2.514, P=9.69E−06, FDR =0.014) is a risk factor for tumorigenesis (Figure 2, Table 2). This suggests that high energy conversion rate metabolism is selected for brain cancer. Therapeutic agents targeting the correction of excessive metabolism may yield favorable results in brain cancer.

In conclusion, our results indicated that tumors from different tissue sources have different metabolic profiles and different metabolites. At the same time, different plasm metabolites also respond to different lifestyle habits of the population, and poor lifestyle habits are responsible for metabolic derangement and causing tumorigenesis. Our study provides a basis for the health management of the population. Finally, our results demonstrate a causal relationship between different plasm metabolites and different tumors, which makes non-invasive cancer screening possible.

Causality of tumorigenesis on plasma metabolites

To explore the causal effects of cancers on plasma metabolites, reverse MR analyses were conducted and 1,390 SNP associated with 10 tumors were selected as IVs for reverse MR analysis (table available at https://cdn.amegroups.cn/static/public/tcr-24-359-5.xlsx). Remarkably, we screened the only plasma metabolite 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2) levels (OR =0.902, 95% CI: 0.836–0.973, P=0.008, FDR =0.024) with causality by IVW method in lung cancer patients with FDR value less than 0.2 (Figure 3A). Meanwhile, both the MR-Egger intercept test and the MR-PRESSO global test ruled out the possibility of horizontal pleiotropy (table available at https://cdn.amegroups.cn/static/public/tcr-24-359-6.xlsx). In addition, the OR values calculated by the five methods were in the same direction (Figure 3A,3B). Methodological scatter plots, forest plots of effect sizes, leave-one-out plots, and funnel plots all indicated reliable results (Figure 3B-3E). Interestingly, the plasm metabolite 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2) levels had a negative causal association with lung cancer in forward MR above, which is consistent with the results of reverse MR analyses. This suggests that lecithin metabolism may play a pivotal role in both lung cancer development and progression. In summary, our results may point to a novel way of understanding the specific mechanisms of lung cancer development.

Figure 3 Significant results of reverse MR. (A) MR results of causal effects between lung cancer risk and 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2) levels. (B) Scatter plots of causality, the slope of each line corresponding to the estimated MR effect in different models. (C) Forest plot with the estimated MR effect of each IV in IVW and MR-Egger models. (D) Leave one out of sensitivity tests. Calculate the MR results of the remaining IVs after removing the IVs one by one. (E) Funnel plot from genetically predicted 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2) levels on lung cancer. N.NSP, the number of SNPs used as IVs; SNP, single-nucleotide polymorphism; IVs, instrumental variables; FDR, false discovery rate; OR, odds ratio; CI, confidence interval; MR, Mendelian randomization; SE, standard error; GPC, glycerylphosphorylcholine.

Discussion

Based on a large amount of publicly available genome-wide data, we explored causal associations between 1,400 plasma metabolites and 24 cancers. To the best of our knowledge, this is the first MR analysis to explore the causal relationship between plasma metabolites and pan-cancer. It was found that 53 metabolites had substantial causal associations with ten malignancies of different tissue origins in this study (FDR <0.2, IVW method). These plasma metabolites are not only star molecules in glucose metabolism, fatty acid metabolism, and amino acid metabolism but are also involved in phospholipid metabolism, caffeine metabolism, and amino acid acetylation. In conclusion, our results demonstrate the diversity of metabolic profiles of malignant tumors and provide a molecular basis for cancer detection and drug targeting.

In cervical cancer, 11 plasma metabolites were associated with amino acid metabolism (X-24518 role is currently unknown) and eight plasma metabolites carried N-terminal amino acid acetylation, suggesting a specific relationship between cervical cancer and protein acetylation modifications. Protein acetylation is a widespread and reversible post-translational protein modification that regulates biological behaviors such as cellular metabolism, division, and signaling by acetylating thousands of histones and non-histone proteins in cells through lysine deacetylases (KDACs) and lysine acetyltransferases (KATs) (50). In our results, eight plasma metabolites with N-terminal acetylation were causally associated with cervical cancer, but in inverse MR analysis, cervical cancer was not causally associated with all 11 plasma metabolites, suggesting that amino acid acetylation may be associated with cervical carcinogenesis only. Human papilloma virus (HPV), the main cause of cervical cancer, can induce H3K9 acetylation and affect chromatin modification through its early genes, E6 and E7, which in turn directly or indirectly affects apoptosis, proliferation, and growth, leading to tumorigenesis (51-54), which is consistent with our results. Three other causally related plasma metabolites, gamma-glutamylcitrulline, methionine sulfone, and alliin, have also been shown to have inhibitory effects on other malignant tumors (55-59), which provides a sidebar to the potential of these three metabolites to inhibit cervical cancer, consistent with our results.

In endometrial cancer, the scenario changes. 1-linoleoyl-GPG (18:2) is a regulator of fatty acid metabolism, which can promote tumorigenesis by regulating key enzymes of lipid metabolism (60,61). Zhao et al. demonstrated that abnormal fatty acid metabolism can directly promote endometrial cancer formation and invasive migration of endometrial cancer cells through mTOR signaling by observing the intestinal flora and spatial visualization of fatty acids (62). Mozihim et al. and Wang et al. also illustrated the critical role of uncontrolled excess fatty acid metabolism in the development of endometrial cancers from the perspectives of sexual hormone metabolism and fatty acid metabolism regulatory enzymes, respectively (63-67). In our study, both 1-linoleoyl-GPG (18:2) and tetradecanedioate were risk factors for endometrial cancer, which may be due to the fact that the transporter protein tetradecanedioate can promote tumorigenesis by participating in fatty acid metabolism (63-66). Glycodeoxycholate 3-sulfate is a bile acid metabolite, and its elevation in plasma is also predictive of endometrial cancer, in line with established studies (67,68). In addition, it has been shown that endometrium under the environment of uncontrolled glutarate metabolism develops abnormalities of decidualization, and this abnormality not only leads to miscarriage but also the development of endometrial cancer (69). This is also illustrated by Our results that there is a causal relationship between high plasma glutarate levels and the development of endometrial cancer. Moreover, our results suggest that the metabolite hexadecanedioate (C16-DC), which also leads to abnormal endometrial decidualization, is also causally associated with the development of endometrial cancer (FDR <0.2, IVW method), which is the first report of this kind to the best of our knowledge.

In this study, ovarian cancer was causally linked to a variety of plasma metabolites, including amino acid metabolism, lipid metabolism, and caffeine metabolism (the role of X-12221 and X-12410 is unknown at this time). In our results, both 5-acetylamino-6-formylamino-3-methyluracil and 5-acetylamino-6-formylamino-3-methyluracil are products of caffeine metabolism (70) and our results that the more of these metabolites, the higher the probability of developing ovarian cancer. In previous prospective studies, caffeine intake predicted a better prognosis in ovarian cancer (71), and the same results were obtained in in vitro experiments (72). However, whether caffeine intake has a causal relationship with the occurrence of ovarian cancer is still controversial in current studies (73,74). We speculate that caffeine and its metabolites may play different roles in ovarian carcinogenesis and progression, which needs to be explored in further studies. Meanwhile, the relationship between caffeine and metabolites also needs more explicit evidence. Sphingomyelin is composed of ceramide with either choline phosphate or ethanolamine phosphate attached to the hydroxyl group (75). Our results show that both hydroxypalmitoyl sphingomyelin [d18:1/16:0(OH)] levels and ceramide (d18:1/14:0, d16:1/16:0) levels can promote ovarian carcinogenesis. Existing studies illustrate that sphingomyelin can contribute to poor prognosis by activating AKT/mTOR/4EBP1 (75), mediate cisplatin resistance in ovarian cancer by altering ceramide 1-phosphate (C1P) activity (76), and enhance the viability of tumor cells in harsh metabolic environments (77), they proved the reliability of our results on the other hand.

Our results indicate that breast cancer is similarly associated with multiple metabolic pathways, including caffeine metabolism, histidine-pyruvate-mediated glucose metabolism, and lipid metabolism. In breast cancer, we used the caffeine-to-paraxanthine ratio to evaluate the direction of the caffeine metabolic pathway. Paraxanthine is an intermediate in the caffeine metabolic pathway, which serves as a substrate for the synthesis of caffeine or continues to be metabolized into 5-acetylamino-6-amino-3-methyluracil and 5-acetylamino-6-formylamino-3-methyluracil (70). Our results showed that the caffeine to paraxanthine ratio was inversely related to the occurrence of breast cancer. Jiang et al. and Zheng et al. conducted a meta-analysis of 37 published studies and found that caffeine intake was negatively associated with the risk of breast cancer in postmenopausal women (78,79). Bellerba et al. used metformin to lower breast cancer recurrence rates and found that paraxanthine was lower in a population with low breast cancer recurrence rates (80). These studies partially explain our results from both caffeine and paraxanthine perspectives. Histidine is one of the essential amino acids, which can be partially synthesized in adults through the gluconeogenic pathway (pyruvate as a raw material), can protect tumor cells to a certain extent from harsh environments, and has some potential as a plasma marker (81). In this study, we used histidine to pyruvate ratio to show the possible direction of histidine metabolism and found a positive correlation with breast cancer development, which is consistent with the results of existing studies (81), which may serve as a new prognostic marker for tumor patients.

In bladder cancer, our results could not circumvent the caffeine metabolic pathway. As mentioned above, 5-acetylamino-6-formylamino-3-methyluracil is one of the end products of the caffeine metabolic pathway, while paraxanthine is an intermediate product of caffeine metabolism (70). Cui et al. found that low levels of 5-acetylamino-6-formylamino-3-methyluracil levels in urine were associated with the development of bladder cancer by first giving caffeine to subjects and then testing their urine (82). Landi et al. explained the effect of caffeine on bladder carcinogenesis by detecting changes in the levels of 5-acetylamino-6-formylamino-3-methyluracil levels in the blood of subjects after caffeine consumption by affecting the activity of cytochrome P4501A2 (CYP1A2), indicating that acetylamino-6-formylamino-3-methyluracil is a protective factor against bladder cancer (83). In our results, there was a causal relationship between low levels of 5-acetylamino-6-formylamino-3-methyluracil levels and bladder carcinogenesis, and a causal relationship between high paraxanthine to 5-acetylamino-6-formylamino-3-methyluracil ratio and bladder cancer, and these results are consistent with the above studies (84). However, we note that all causal relationships between caffeine metabolic pathways and tumors avoid the role of caffeine itself. Further studies are needed to clarify the specific role of caffeine and its metabolites in the body and in vitro.

Retinol (vitamin A) to linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) ratio plays different roles in colorectal and pancreatic cancer. Our results indicate that the retinol (vitamin A) to linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) ratio is inversely associated with colorectal carcinogenesis but promotes pancreatic cancer. A meta-analysis has demonstrated that retinol is a protective factor against colorectal carcinogenesis (85), as well as in vitro experiments demonstrating that retinol can reduce chronic inflammation of the intestinal mucosa and decrease the incidence of colorectal cancer by inducing foxp3+ regulatory T cells to regulate T cells into the Th1/Th2/Th17/Tregs pathway (86), and in a mouse model, a retinol deficiency leads to dysregulation of T-cell responses, leading to colitis and colorectal cancer (87), in partial agreement with our findings. On the other side, retinol promotes pancreatic development and maintains the normal secretory function of the pancreas (88), and loss of retinol leads to pancreatic astrocyte apoptosis and pancreatic fibrosis (89). The question arises then as to what role the plasma metabolite linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) plays in pancreatic cancer. It has been shown that linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) may be a product formed by esterification of retinol, an intracellular storage form of retinol (90). Linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) may have roles in regulating the ability of the pancreas to metabolize lipids (91) and regulating neutrophil chemotaxis in some tumors (92). However, to the best of our knowledge, there is no evidence to indicate a specific role for linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) in pancreatic and colorectal cancers. Our results suggest that the direct use of retinol (vitamin A) to linoleoyl-arachidonoyl-glycerol (18:2 to 20:4) ratio to predict tumorigenesis may be more reliable than the use of one of the plasma metabolites alone.

Disorders of lipid metabolism are the main protagonists of lung cancer. Available evidence indicates that in lung cancer, 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2) plays a critical role in regulating tumor cell metabolism (93). Unsaturated short-chain fatty acids are reduced in tumor cell membranes, whereas the increase in saturated long-chain fatty acids is to protect cells from oxidative stress that arises during the development of various cancers (93,94), consistent with our results that 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2), which is a saturated long-chain fatty acid, is a protective factor in lung cancer. Fatty acid metabolism has a precise predictive effect for specific malignant tumors, and our inverse MR results also obtained a causal relationship between lung cancer and 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2). Our results suggest that lung carcinogenesis may lead to decreased 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2) levels, implying that 1-palmitoyl-2-linoleoyl-GPC (16:0/18:2) may be involved in both lung carcinogenesis and progression, and is a novel and valuable target throughout the whole process of lung cancer prediction, diagnosis, and treatment. Loss of 1-palmitoyl-2-linoleoyl-GPC to an L-α-GPC (also known as choline glycerophosphate) and attachment of a 3-acetyl-rac-glycerol to form 1-palmitoyl-2-linoleoyl-3-acetyl-rac-glycerol (PLAG), an inhibitory factor in lung cancer (95-97). PLAG inhibits tumor progression in lung cancer animal models by promoting the degradation of adenosine 2B receptors (A2BRs) (96), and PLAG also modulates PAR2/EGFR transactivation by accelerating protease-activated receptor 2 (PAR2) degradation to attenuate metastatic activity in lung cancer (97). In our results, 1-palmitoyl-2-linoleoyl-GPC, which shares the same motif as PLAG, has similar inhibitory effects on lung cancer as PLAG in both forward MR and reverse MR. Unfortunately, there is no definite evidence for the relationship between 1-palmitoyl-2-linoleoyl-GPC and PLAG in metabolism and its direct effect on lung cancer. In conclusion, our results on 1-palmitoyl-2-linoleoyl-GPC are very rare and give a new perspective on plasma cancer detection in lung cancer. Furthermore, our results also suggest that 1-stearoyl-2-arachidonoyl-GPC (18:0/20:4) levels are a high-risk factor for lung carcinogenesis. In existing studies, 1-stearoyl-2-arachidonoyl-GPC (18:0/20:4) can regulate cell growth via exosomes, which in turn promotes lung carcinogenesis (98), is consistent with findings in a mouse model (99), validating the reliability of our results. Our results also revealed a positive correlation between 5alpha-androstan-3beta,17alpha-diol disulfate levels and lung carcinogenesis, which is the first report to our knowledge.

In oropharynx cancer, our results showed that 3-acetylphenol sulfate levels were the only protective factor, which was similarly demonstrated in a recent RCT study (100). 3-acetylphenol sulfate levels are a promising marker for oropharynx cancer with potential as markers for plasma cancer detection in cancer. The major difference between tumor cells and normal tissue cells resides in their higher energy demand and higher metabolic efficiency, and it is no exception in brain cancer (101-103). AMP and FAD are both important energy metabolism intermediates, which can directly participate in electron transfer, form electron chains, and directly generate energy (104-107). The difference is that the same amount of AMP provides more energy and is more efficient than FAD (108,109). Maher et al. used [U-13 C]-labeled glucose to trace the metabolic processes of brain cancer and obtained the result that brain cancer has a high metabolic rate (101); Martínez-Reyes et al. revisited the high metabolic profile of brain cancer again from the perspective of the mitochondrial electron transport chain (ETC) (102). Based on our findings, AMP to FAD ratio is a risk factor for brain cancer, reflecting the high energy demand of brain cancer, and is a strong corroboration of the results of existing studies (101,102). Notably, in addition to energy metabolism, FAD is also associated with the activity of several biological enzymes, such as spermine oxidase (SMO), which can lead to tumor progression by regulating the biological behavior of tumor cells (110). On the other hand, AMP can directly participate in the activation of adenosine monophosphate-activated protein kinase (AMPK), and then regulate the PI3K signaling pathway, mTOR signaling pathway, Wnt signaling pathway, etc., which can directly affect the survival of tumor cells and play a two-way regulatory role of “inhibiting cancer when cancer is absent and promoting cancer cell growth when cancer is present” (111). This demonstrates the superiority of using ratio rather than AMP or FAD alone to predict brain cancer. Rather than taking into account individual metabolites, we only consider the overall effect. It could be a source of new markers.

There have been several MR articles published investigating the relationship between plasma metabolites and tumors. Despite this, they use fewer types of plasma metabolites than we do, and our data sources are more current and comprehensive. Additionally, they tend to focus on a specific tumor or group of tumors, which can lead to one-sided results due to individual differences. Finally, we used the ratio between plasma metabolites to illustrate their relationship with tumorigenesis, which can avoid the disadvantage that some plasma metabolites play multiple roles in the organism and cannot be judged alone. The overall effect of plasma metabolites on malignant tumors can be more accurately described by our approach, which considers the overall effect. In conclusion, our work provides the most comprehensive atlas to date of the causal relationships between 1,400 plasma metabolites and 24 cancers with the expectation of providing novel ideas for specific tumor plasma markers, plasma cancer detection, and therapeutic targets.

We used published large genome-wide cohorts for our two-sample MR analyses, which is statistically efficient and statistically reliable. Genetic IVs and causal inference were used in this study to derive conclusions. There were no confounding factors such as horizontal polytomies or other confounding factors that affected the results. There are, however, still limitations to our study. First, horizontal polyvalence cannot be completely ruled out even when multiple methods of quality control were conducted. Second, the lack of individual information on participants prevented us from further stratifying the population. Third, due to the study’s European database, the conclusions cannot be generalized to other races, which limits our results’ generalizability. Finally, we adapted more flexible thresholds for assessing the results, which may result in more false positives, but this simultaneously enabled us to assess plasma metabolites’ association with tumors in a more comprehensive manner.


Conclusions

In summary, our MR study found that there is a causal relationship between plasma metabolites and cancer and that they may play different roles at different times of tumor growth. In addition, our study provides a comprehensive map of plasma metabolites associated with pan-cancer and suggests an important role of biometabolism in tumorigenesis and development, showing the direction of basic cancer research and providing possible targets for plasma cancer detection.


Acknowledgments

We appreciate all the researchers and consortia for providing the high-grade GWAS resources. All data and material are available from the corresponding consortia.

Funding: This research was gratefully supported by the Ministry of Education double first-class construction disciplines: basic medicine [Naval Medical University (Second Military Medical University)] and the Shanghai Municipal Health Commission Health Industry Clinical Research Special (No. 20234Y0257). The funding agencies had no role in the study design; collection, analysis, or interpretation of data; writing of the report; or the decision to submit the article for publication.


Footnote

Reporting Checklist: The authors have completed the STROBE-MR reporting checklist. Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-359/rc

Peer Review File: Available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-359/prf

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://tcr.amegroups.com/article/view/10.21037/tcr-24-359/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The study was conducted in accordance with the Declaration of Helsinki (as revised in 2013).

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.


References

  1. Rahib L, Wehner MR, Matrisian LM, et al. Estimated Projection of US Cancer Incidence and Death to 2040. JAMA Netw Open 2021;4:e214708. [Crossref] [PubMed]
  2. Siegel RL, Miller KD, Wagle NS, et al. Cancer statistics, 2023. CA Cancer J Clin 2023;73:17-48. [Crossref] [PubMed]
  3. Heron M. Deaths: Leading Causes for 2017. Natl Vital Stat Rep 2019;68:1-77. [PubMed]
  4. Fitzgerald RC, Antoniou AC, Fruk L, et al. The future of early cancer detection. Nat Med 2022;28:666-77. [Crossref] [PubMed]
  5. Singh AK, McGuirk JP. CAR T cells: continuation in a revolution of immunotherapy. Lancet Oncol 2020;21:e168-78. [Crossref] [PubMed]
  6. Martincorena I, Fowler JC, Wabik A, et al. Somatic mutant clones colonize the human esophagus with age. Science 2018;362:911-7. [Crossref] [PubMed]
  7. Tomasetti C, Vogelstein B. Cancer etiology. Variation in cancer risk among tissues can be explained by the number of stem cell divisions. Science 2015;347:78-81. [Crossref] [PubMed]
  8. Yokoyama A, Kakiuchi N, Yoshizato T, et al. Age-related remodelling of oesophageal epithelia by mutated cancer drivers. Nature 2019;565:312-7. [Crossref] [PubMed]
  9. Krimmel JD, Schmitt MW, Harrell MI, et al. Ultra-deep sequencing detects ovarian cancer cells in peritoneal fluid and reveals somatic TP53 mutations in noncancerous tissues. Proc Natl Acad Sci U S A 2016;113:6005-10. [Crossref] [PubMed]
  10. Ortega MA, Navarro F, Pekarek L, et al. Exploring histopathological and serum biomarkers in lung adenocarcinoma: Clinical applications and translational opportunities Int J Oncol 2022;61:154. (Review). [Crossref] [PubMed]
  11. Ma S, Zhou M, Xu Y, et al. Clinical application and detection techniques of liquid biopsy in gastric cancer. Mol Cancer 2023;22:7. [Crossref] [PubMed]
  12. Tian Z, Liang C, Zhang Z, et al. Prognostic value of neuron-specific enolase for small cell lung cancer: a systematic review and meta-analysis. World J Surg Oncol 2020;18:116. [Crossref] [PubMed]
  13. Wang X, Zhang Q, Zhou J, et al. T cell-mediated targeted delivery of tadalafil regulates immunosuppression and polyamine metabolism to overcome immune checkpoint blockade resistance in hepatocellular carcinoma. J Immunother Cancer 2023;11:e006493. [Crossref] [PubMed]
  14. Norman JS, Li PJ, Kotwani P, et al. AFP-L3 and DCP strongly predict early hepatocellular carcinoma recurrence after liver transplantation. J Hepatol 2023;79:1469-77. [Crossref] [PubMed]
  15. Głowska-Ciemny J, Szymanski M, Kuszerska A, et al. Role of Alpha-Fetoprotein (AFP) in Diagnosing Childhood Cancers and Genetic-Related Chronic Diseases. Cancers (Basel) 2023;15:4302. [Crossref] [PubMed]
  16. Feng J, Zhang Q, Zhou Y, et al. Integration of Proteomics and Metabolomics Revealed Metabolite-Protein Networks in ACTH-Secreting Pituitary Adenoma. Front Endocrinol (Lausanne) 2018;9:678. [Crossref] [PubMed]
  17. Finley LWS. What is cancer metabolism? Cell 2023;186:1670-88. [Crossref] [PubMed]
  18. Ma Q, Jiang H, Ma L, et al. The moonlighting function of glycolytic enzyme enolase-1 promotes choline phospholipid metabolism and tumor cell proliferation. Proc Natl Acad Sci U S A 2023;120:e2209435120. [Crossref] [PubMed]
  19. Szlasa W, Zendran I, Zalesińska A, et al. Lipid composition of the cancer cell membrane. J Bioenerg Biomembr 2020;52:321-42. [Crossref] [PubMed]
  20. Gyamfi J, Kim J, Choi J. Cancer as a Metabolic Disorder. Int J Mol Sci 2022;23:1155. [Crossref] [PubMed]
  21. Schito L, Semenza GL. Hypoxia-Inducible Factors: Master Regulators of Cancer Progression. Trends Cancer 2016;2:758-70. [Crossref] [PubMed]
  22. Vaupel P. The role of hypoxia-induced factors in tumor progression. Oncologist 2004;9:10-7. [Crossref] [PubMed]
  23. Martinez-Outschoorn UE, Peiris-Pagés M, Pestell RG, et al. Cancer metabolism: a therapeutic perspective. Nat Rev Clin Oncol 2017;14:113. [Crossref] [PubMed]
  24. Altman BJ, Stine ZE, Dang CV. From Krebs to clinic: glutamine metabolism to cancer therapy. Nat Rev Cancer 2016;16:749. [Crossref] [PubMed]
  25. Lu W, Pelicano H, Huang P. Cancer metabolism: is glutamine sweeter than glucose? Cancer Cell 2010;18:199-200. [Crossref] [PubMed]
  26. Shanware NP, Mullen AR, DeBerardinis RJ, et al. Glutamine: pleiotropic roles in tumor growth and stress resistance. J Mol Med (Berl) 2011;89:229-36. [Crossref] [PubMed]
  27. Peck B, Schulze A. Lipid desaturation - the next step in targeting lipogenesis in cancer? FEBS J 2016;283:2767-78. [Crossref] [PubMed]
  28. Santos CR, Schulze A. Lipid metabolism in cancer. FEBS J 2012;279:2610-23. [Crossref] [PubMed]
  29. Cheng C, Geng F, Cheng X, et al. Lipid metabolism reprogramming and its potential targets in cancer. Cancer Commun (Lond) 2018;38:27. [Crossref] [PubMed]
  30. Muñoz-Pinedo C, El Mjiyad N, Ricci JE. Cancer metabolism: current perspectives and future directions. Cell Death Dis 2012;3:e248. [Crossref] [PubMed]
  31. Deng F, Zhou R, Lin C, et al. Tumor-secreted dickkopf2 accelerates aerobic glycolysis and promotes angiogenesis in colorectal cancer. Theranostics 2019;9:1001-14. [Crossref] [PubMed]
  32. Hanahan D. Hallmarks of Cancer: New Dimensions. Cancer Discov 2022;12:31-46. [Crossref] [PubMed]
  33. Kim JY, Song M, Kim MS, et al. An atlas of associations between 14 micronutrients and 22 cancer outcomes: Mendelian randomization analyses. BMC Med 2023;21:316. [Crossref] [PubMed]
  34. Veronese N, Demurtas J, Pesolillo G, et al. Magnesium and health outcomes: an umbrella review of systematic reviews and meta-analyses of observational and intervention studies. Eur J Nutr 2020;59:263-72. [Crossref] [PubMed]
  35. Chen Z, Huang Y, Cao D, et al. Vitamin C Intake and Cancers: An Umbrella Review. Front Nutr 2022;8:812394. [Crossref] [PubMed]
  36. Liu D, Meng X, Tian Q, et al. Vitamin D and Multiple Health Outcomes: An Umbrella Review of Observational Studies, Randomized Controlled Trials, and Mendelian Randomization Studies. Adv Nutr 2022;13:1044-62. [Crossref] [PubMed]
  37. Manson JE, Cook NR, Lee IM, et al. Vitamin D Supplements and Prevention of Cancer and Cardiovascular Disease. N Engl J Med 2019;380:33-44. [Crossref] [PubMed]
  38. Smith GD, Ebrahim S. 'Mendelian randomization': can genetic epidemiology contribute to understanding environmental determinants of disease? Int J Epidemiol 2003;32:1-22. [Crossref] [PubMed]
  39. Wang Y, Liu F, Sun L, et al. Association between human blood metabolome and the risk of breast cancer. Breast Cancer Res 2023;25:9. [Crossref] [PubMed]
  40. Chen Y, Lu T, Pettersson-Kymmer U, et al. Genomic atlas of the plasma metabolome prioritizes metabolites implicated in human diseases. Nat Genet 2023;55:44-53. [Crossref] [PubMed]
  41. Pierce BL, Ahsan H, Vanderweele TJ. Power and instrument strength requirements for Mendelian randomization studies using multiple genetic variants. Int J Epidemiol 2011;40:740-52. [Crossref] [PubMed]
  42. Papadimitriou N, Dimou N, Tsilidis KK, et al. Physical activity and risks of breast and colorectal cancer: a Mendelian randomisation analysis. Nat Commun 2020;11:597. [Crossref] [PubMed]
  43. Larsson SC, Traylor M, Malik R, et al. Modifiable pathways in Alzheimer's disease: Mendelian randomisation analysis. BMJ 2017;359:j5375. [Crossref] [PubMed]
  44. Larsson SC, Burgess S. Appraising the causal role of smoking in multiple diseases: A systematic review and meta-analysis of Mendelian randomization studies. EBioMedicine 2022;82:104154. [Crossref] [PubMed]
  45. Mingdong W, Xiang G, Yongjun Q, et al. Causal associations between gut microbiota and urological tumors: a two-sample mendelian randomization study. BMC Cancer 2023;23:854. [Crossref] [PubMed]
  46. Greco M FD, Minelli C, Sheehan NA, et al. Detecting pleiotropy in Mendelian randomisation studies with summary data and a continuous outcome. Stat Med 2015;34:2926-40. [Crossref] [PubMed]
  47. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol 2015;44:512-25. [Crossref] [PubMed]
  48. Li Y, Liu H, Ye S, et al. The effects of coagulation factors on the risk of endometriosis: a Mendelian randomization study. BMC Med 2023;21:195. [Crossref] [PubMed]
  49. Verbanck M, Chen CY, Neale B, et al. Detection of widespread horizontal pleiotropy in causal relationships inferred from Mendelian randomization between complex traits and diseases. Nat Genet 2018;50:693-8. [Crossref] [PubMed]
  50. Shvedunova M, Akhtar A. Modulation of cellular processes by histone and non-histone protein acetylation. Nat Rev Mol Cell Biol 2022;23:329-49. [Crossref] [PubMed]
  51. Liu Y, Tergaonkar V, Krishna S, et al. Human papillomavirus type 16 E6-enhanced susceptibility of L929 cells to tumor necrosis factor alpha correlates with increased accumulation of reactive oxygen species. J Biol Chem 1999;274:24819-27. [Crossref] [PubMed]
  52. Chakrabarti O, Veeraraghavalu K, Tergaonkar V, et al. Human papillomavirus type 16 E6 amino acid 83 variants enhance E6-mediated MAPK signaling and differentially regulate tumorigenesis by notch signaling and oncogenic Ras. J Virol 2004;78:5934-45. [Crossref] [PubMed]
  53. Feng D, Yan K, Zhou Y, et al. Piwil2 is reactivated by HPV oncoproteins and initiates cell reprogramming via epigenetic regulation during cervical cancer tumorigenesis. Oncotarget 2016;7:64575-88. [Crossref] [PubMed]
  54. Evans MR, James CD, Bristol ML, et al. Human Papillomavirus 16 E2 Regulates Keratinocyte Gene Expression Relevant to Cancer and the Viral Life Cycle. J Virol 2019;93:e01941-18. [Crossref] [PubMed]
  55. Fentiman IS. Gamma-glutamyl transferase: risk and prognosis of cancer. Br J Cancer 2012;106:1467-8. [Crossref] [PubMed]
  56. Chen YJ, Guo X, Liu ML, et al. Interaction between glycolysis-cholesterol synthesis axis and tumor microenvironment reveal that gamma-glutamyl hydrolase suppresses glycolysis in colon cancer. Front Immunol 2022;13:979521. [Crossref] [PubMed]
  57. López-López Á, Ciborowski M, Niklinski J, et al. Optimization of capillary electrophoresis coupled to negative mode electrospray ionization-mass spectrometry using polyvinyl alcohol coated capillaries. Application to a study on non-small cell lung cancer. Anal Chim Acta 2022;1226:340259. [Crossref] [PubMed]
  58. Good S, Walter MA, Waser B, et al. Macrocyclic chelator-coupled gastrin-based radiopharmaceuticals for targeting of gastrin receptor-expressing tumours. Eur J Nucl Med Mol Imaging 2008;35:1868-77. [Crossref] [PubMed]
  59. Livingstone TL, Saha S, Bernuzzi F, et al. Accumulation of Sulforaphane and Alliin in Human Prostate Tissue. Nutrients 2022;14:3263. [Crossref] [PubMed]
  60. Liu XW, Sok DE, Yook HS, et al. Inhibition of lysophospholipase D activity by unsaturated lysophosphatidic acids or seed extracts containing 1-linoleoyl and 1-oleoyl lysophosphatidic acid. J Agric Food Chem 2007;55:8717-22. [Crossref] [PubMed]
  61. Araujo M, Beekman JK, Mapa MST, et al. Assessment of intestinal absorption/metabolism of 3-chloro-1,2-propanediol (3-MCPD) and three 3-MCPD monoesters by Caco-2 cells. Toxicol In vitro 2020;67:104887. [Crossref] [PubMed]
  62. Zhao SS, Chen L, Yang J, et al. Altered Gut Microbial Profile Accompanied by Abnormal Fatty Acid Metabolism Activity Exacerbates Endometrial Cancer Progression. Microbiol Spectr 2022;10:e0261222. [Crossref] [PubMed]
  63. Mozihim AK, Chung I, Said NABM, et al. Reprogramming of Fatty Acid Metabolism in Gynaecological Cancers: Is There a Role for Oestradiol? Metabolites 2022;12:350. [Crossref] [PubMed]
  64. Wang B, Ge S, Wang Z, et al. Analysis and experimental validation of fatty acid metabolism-related genes prostacyclin synthase (PTGIS) in endometrial cancer. Aging (Albany NY) 2023;15:10322-46. [Crossref] [PubMed]
  65. Müller F, Sharma A, König J, et al. Biomarkers for In Vivo Assessment of Transporter Function. Pharmacol Rev 2018;70:246-77. [Crossref] [PubMed]
  66. Shen H, Chen W, Drexler DM, et al. Comparative Evaluation of Plasma Bile Acids, Dehydroepiandrosterone Sulfate, Hexadecanedioate, and Tetradecanedioate with Coproporphyrins I and III as Markers of OATP Inhibition in Healthy Subjects. Drug Metab Dispos 2017;45:908-19. [Crossref] [PubMed]
  67. Jin Y, Li Y, Eisenmann ED, et al. Determination of the endogenous OATP1B biomarkers glycochenodeoxycholate-3-sulfate and chenodeoxycholate-24-glucuronide in human and mouse plasma by a validated UHPLC-MS/MS method. J Chromatogr B Analyt Technol Biomed Life Sci 2022;1210:123437. [Crossref] [PubMed]
  68. Orozco CC, Neuvonen M, Bi YA, et al. Characterization of Bile Acid Sulfate Conjugates as Substrates of Human Organic Anion Transporting Polypeptides. Mol Pharm 2023;20:3020-32. [Crossref] [PubMed]
  69. Tang L, Xu XH, Xu S, et al. Dysregulated Gln-Glu-α-ketoglutarate axis impairs maternal decidualization and increases the risk of recurrent spontaneous miscarriage. Cell Rep Med 2023;4:101026. [Crossref] [PubMed]
  70. Xu R, He L, Vatsalya V, et al. Metabolomics analysis of urine from patients with alcohol-associated liver disease reveals dysregulated caffeine metabolism. Am J Physiol Gastrointest Liver Physiol 2023;324:G142-54. [Crossref] [PubMed]
  71. Salari-Moghaddam A, Milajerdi A, Surkan PJ, et al. Caffeine, Type of Coffee, and Risk of Ovarian Cancer: A Dose-Response Meta-Analysis of Prospective Studies. J Clin Endocrinol Metab 2019;104:5349-59. [Crossref] [PubMed]
  72. Lee SJ, Park SY, Bak S, et al. Synergistic Effect of Saccharin and Caffeine on Antiproliferative Activity in Human Ovarian Carcinoma Ovcar-3 Cells. Int J Mol Sci 2023;24:14445. [Crossref] [PubMed]
  73. Grundy A, Sandhu S, Arseneau J, et al. Lifetime caffeine intake and the risk of epithelial ovarian cancer. Cancer Epidemiol 2022;76:102058. [Crossref] [PubMed]
  74. Zhu G, Li Z, Tang L, et al. Associations of Dietary Intakes with Gynecological Cancers: Findings from a Cross-Sectional Study. Nutrients 2022;14:5026. [Crossref] [PubMed]
  75. Shi Y, Zhou C, Lu H, et al. Ceramide synthase 6 predicts poor prognosis and activates the AKT/mTOR/4EBP1 pathway in high-grade serous ovarian cancer. Am J Transl Res 2020;12:5924-39. [PubMed]
  76. Zou J, Jian L. Inhibition of Ceramide Kinase Is Effective against Cisplatin-Resistant Ovarian Cancer Cells by Regulating Ceramide and C1P Levels. Gynecol Obstet Invest 2023;88:61-70. [Crossref] [PubMed]
  77. Tesfay L, Paul BT, Konstorum A, et al. Stearoyl-CoA Desaturase 1 Protects Ovarian Cancer Cells from Ferroptotic Cell Death. Cancer Res 2019;79:5355-66. [Crossref] [PubMed]
  78. Jiang W, Wu Y, Jiang X. Coffee and caffeine intake and breast cancer risk: an updated dose-response meta-analysis of 37 published studies. Gynecol Oncol 2013;129:620-9. [Crossref] [PubMed]
  79. Zheng KH, Zhu K, Wactawski-Wende J, et al. Caffeine intake from coffee and tea and invasive breast cancer incidence among postmenopausal women in the Women's Health Initiative. Int J Cancer 2021;149:2032-44. [Crossref] [PubMed]
  80. Bellerba F, Chatziioannou AC, Jasbi P, et al. Metabolomic profiles of metformin in breast cancer survivors: a pooled analysis of plasmas from two randomized placebo-controlled trials. J Transl Med 2022;20:629. [Crossref] [PubMed]
  81. Matboli M, Eissa S, Said H. Evaluation of histidine-rich glycoprotein tissue RNA and serum protein as novel markers for breast cancer. Med Oncol 2014;31:897. [Crossref] [PubMed]
  82. Cui X, Guo R, Xu Z, et al. Relationship between metabolic phenotype of N-acetylation and bladder cancer. Chin Med J (Engl) 2000;113:303-5. [PubMed]
  83. Landi MT, Zocchetti C, Bernucci I, et al. Cytochrome P4501A2: enzyme induction and genetic control in determining 4-aminobiphenyl-hemoglobin adduct levels. Cancer Epidemiol Biomarkers Prev 1996;5:693-8. [PubMed]
  84. Timson J. Caffeine and bladder cancer. Lancet 1971;2:212-3. [Crossref] [PubMed]
  85. Han X, Zhao R, Zhang G, et al. Association of Retinol and Carotenoids Content in Diet and Serum With Risk for Colorectal Cancer: A Meta-Analysis. Front Nutr 2022;9:918777. [Crossref] [PubMed]
  86. Strober W. Vitamin A rewrites the ABCs of oral tolerance. Mucosal Immunol 2008;1:92-5. [Crossref] [PubMed]
  87. Okayasu I, Hana K, Nemoto N, et al. Vitamin A Inhibits Development of Dextran Sulfate Sodium-Induced Colitis and Colon Cancer in a Mouse Model. Biomed Res Int 2016;2016:4874809. [Crossref] [PubMed]
  88. Brun PJ, Wongsiriroj N, Blaner WS. Retinoids in the pancreas. Hepatobiliary Surg Nutr 2016;5:1-14. [PubMed]
  89. Zhang Y, Yue D, Cheng L, et al. Vitamin A-coupled liposomes carrying TLR4-silencing shRNA induce apoptosis of pancreatic stellate cells and resolution of pancreatic fibrosis. J Mol Med (Berl) 2018;96:445-58. [Crossref] [PubMed]
  90. Ubels JL, Huebler SM. Esterification of retinol in lacrimal gland. Evidence for acyl-CoA:retinol acyltransferase activity. Invest Ophthalmol Vis Sci 1990;31:582-9. [PubMed]
  91. Huang Y, Hui DY. Metabolic fate of pancreas-derived cholesterol esterase in intestine: an in vitro study using Caco-2 cells. J Lipid Res 1990;31:2029-37. [Crossref] [PubMed]
  92. Jeong J, Kim YJ, Lee DY, et al. 1-Palmitoyl-2-linoleoyl-3-acetyl-rac-glycerol (PLAG) attenuates gemcitabine-induced neutrophil extravasation. Cell Biosci 2019;9:4. [Crossref] [PubMed]
  93. Rysman E, Brusselmans K, Scheys K, et al. De novo lipogenesis protects cancer cells from free radicals and chemotherapeutics by promoting membrane lipid saturation. Cancer Res 2010;70:8117-26. [Crossref] [PubMed]
  94. Guo S, Wang Y, Zhou D, et al. Significantly increased monounsaturated lipids relative to polyunsaturated lipids in six types of cancer microenvironment are observed by mass spectrometry imaging. Sci Rep 2014;4:5959. [Crossref] [PubMed]
  95. Kim GT, Kim EY, Shin SH, et al. PLAG alleviates cisplatin-induced cachexia in lung cancer implanted mice. Transl Oncol 2022;20:101398. [Crossref] [PubMed]
  96. Kim GT, Kim EY, Shin SH, et al. Suppression of tumor progression by thioredoxin-interacting protein-dependent adenosine 2B receptor degradation in a PLAG-treated Lewis lung carcinoma-1 model of non-small cell lung cancer. Neoplasia 2022;31:100815. [Crossref] [PubMed]
  97. Kim GT, Hahn KW, Yoon SY, et al. PLAG Exerts Anti-Metastatic Effects by Interfering with Neutrophil Elastase/PAR2/EGFR Signaling in A549 Lung Cancer Orthotopic Model. Cancers (Basel) 2020;12:560. [Crossref] [PubMed]
  98. Zhang X, Xu Y, Ma L, et al. Essential roles of exosome and circRNA_101093 on ferroptosis desensitization in lung adenocarcinoma. Cancer Commun (Lond) 2022;42:287-313. [Crossref] [PubMed]
  99. Kienzl M, Hasenoehrl C, Maitz K, et al. Monoacylglycerol lipase deficiency in the tumor microenvironment slows tumor growth in non-small cell lung cancer. Oncoimmunology 2021;10:1965319. [Crossref] [PubMed]
  100. Rolland V, Meyer F, Guitton MJ, et al. A randomized controlled trial to test the efficacy of trans-tympanic injections of a sodium thiosulfate gel to prevent cisplatin-induced ototoxicity in patients with head and neck cancer. J Otolaryngol Head Neck Surg 2019;48:4. [Crossref] [PubMed]
  101. Maher EA, Marin-Valencia I, Bachoo RM, et al. Metabolism of [U-13 C]glucose in human brain tumors in vivo. NMR Biomed 2012;25:1234-44. [Crossref] [PubMed]
  102. Martínez-Reyes I, Cardona LR, Kong H, et al. Mitochondrial ubiquinol oxidation is necessary for tumour growth. Nature 2020;585:288-92. [Crossref] [PubMed]
  103. Karasová M, Procházková J, Tylichová Z, et al. Inhibition of Aryl Hydrocarbon Receptor (AhR) Expression Disrupts Cell Proliferation and Alters Energy Metabolism and Fatty Acid Synthesis in Colon Cancer Cells. Cancers (Basel) 2022;14:4245. [Crossref] [PubMed]
  104. Sato K, Nishina Y, Shiga K. In vitro assembly of FAD, AMP, and the two subunits of electron-transferring flavoprotein: an important role of AMP related with the conformational change of the apoprotein. J Biochem 1997;121:477-86. [Crossref] [PubMed]
  105. Sato K, Nishina Y, Shiga K. In vitro refolding and unfolding of subunits of electron-transferring flavoprotein: characterization of the folding intermediates and the effects of FAD and AMP on the folding reaction. J Biochem 1996;120:276-85. [Crossref] [PubMed]
  106. Bruno F, Malvaso A, Canterini S, et al. Antimicrobial Peptides (AMPs) in the Pathogenesis of Alzheimer's Disease: Implications for Diagnosis and Treatment. Antibiotics (Basel) 2022;11:726. [Crossref] [PubMed]
  107. Pereira MS, de Araújo SS, Nagem RAP, et al. The role of remote flavin adenine dinucleotide pieces in the oxidative decarboxylation catalyzed by salicylate hydroxylase. Bioorg Chem 2022;119:105561. [Crossref] [PubMed]
  108. Moritzer AC, Niemann HH. Binding of FAD and tryptophan to the tryptophan 6-halogenase Thal is negatively coupled. Protein Sci 2019;28:2112-8. [Crossref] [PubMed]
  109. Deng Y, Zhou Q, Wu Y, et al. Properties and Mechanisms of Flavin-Dependent Monooxygenases and Their Applications in Natural Product Synthesis. Int J Mol Sci 2022;23:2622. [Crossref] [PubMed]
  110. Cervelli M, Amendola R, Polticelli F, et al. Spermine oxidase: ten years after. Amino Acids 2012;42:441-50. [Crossref] [PubMed]
  111. Russell FM, Hardie DG. AMP-Activated Protein Kinase: Do We Need Activators or Inhibitors to Treat or Prevent Cancer? Int J Mol Sci 2020;22:186. [Crossref] [PubMed]
Cite this article as: Zhang J, Hao Z, Chen Z, Su X, Xu W, Jiang X, Nian X. Unveiling the atlas of associations between 1,400 plasma metabolites and 24 tumors: Mendelian randomization analyses. Transl Cancer Res 2024;13(9):4938-4956. doi: 10.21037/tcr-24-359

Download Citation