This article is mentioned in:

Introduction

Idiopathic pulmonary fibrosis (IPF), a progressive fibrotic lung ailment (1), exhibits notably unfavorable prognosis, with a median survival rate of ~3-4 years subsequent to diagnosis (2). This condition frequently culminates in fatality due to respiratory failure, primarily caused by the extensive restructuring of the pulmonary tissue (3). Antifibrotic treatment has shown effectiveness in reducing the advancement of the disease and lessening the deterioration of lung function in individuals with IPF. Furthermore, other studies demonstrated that the survival rates of patients with IPF treated with antifibrotic medications were enhanced compared with untreated ones (4,5). However, several patients stop treatment due to the onset of adverse effects (6). Genetic studies suggested that the susceptibility to IPF could be significantly affected by genetics. Therefore, the investigation of sporadic and familial cases promoted the successful identification of multiple IPF-related variants (7). The establishment and verification of evidence-supported recommendations for gene-oriented screening for IPF could enable the redefinition and reclassification of this particular illness based on its molecular characteristics, thus promoting the adoption of precision medicine strategies (8). Therefore, the timely detection of possible genetic risk factors could prevent the onset of IPF. Currently, there are limited methods for the early detection of IPF. However, the early detection and intervention are crucial for slowing down the progression of the disease. Due to their powerful monitoring capabilities, wide applicability and accessible analysis, serum metabolites have become promising biomarkers for IPF.

The understanding of disease progression has been improved with the introduction of transcriptomics, proteomics and metabolomics, thus providing novel insights into the cellular mechanisms involved. It has been reported that regulating metabolic balance could affect IPF (9). Pulmonary fibrosis can occur due to irregular collagen production and disrupted airway remodeling caused by metabolic abnormalities in alveolar epithelial cells, myofibroblasts and macrophages. Several metabolic pathways are involved in lipid, protein and carbohydrate metabolism, and more particularly in IPF (10). Metabolomics, via identifying altered metabolites or metabolic pathways, offer a fresh perspective for investigating the underlying mechanisms of different diseases, thus providing valuable insights into the biological processes associated with these diseases (11). Metabolic changes are not just a characteristic of fibrosis development, but they can also significantly affect the development of fibrosis, particularly in IPF (12). It has been also suggested that dysfunctions in the metabolic processes of glycine, glutamine and arginine, along with irregularities in glycolysis, can activate a pro-fibrotic feature via a TGF-β-dependent mechanism (10,13,14). Increased content of extended and moderate chain fatty acids has been documented in the lungs of patients with IPF. In addition, previous studies reported an enhanced reprogramming of macrophages, resulting in IPF improvement via fatty acid β-oxidation (15,16). Another study demonstrated that metabolites derived from arachidonic acid could hinder pulmonary fibrosis via altering the TGF-β1-induced pro-fibrotic signaling pathway (17). The aforementioned finding indicated that targeting these metabolites could be a promising approach for treating IPF (17). However, the effect and relevance of serum metabolomic profiles on IPF remains understudied.

Investigating metabolites associated with the pathogenesis and progression of IPF is not just significant for early detection and prevention, but also serve a key role in comprehending the biological mechanisms of IPF therapy and uncovering possible therapeutic objectives. Several metabolites associated with IPF have been identified, such as Krebs von den Lungen 6, matrix metalloproteinases and periostin (18). Nevertheless, the uncertain causal association between metabolites and IPF persists, since there is a dearth of prospective studies investigating metabolites and IPF. Randomized controlled trials are widely considered as the benchmark for establishing causal effects. However, the complexities and obstacles in this particular scenario pose challenges in reaching conclusive findings regarding the causal association between metabolites and IPF. Mendelian randomization (MR) is an extensively employed and novel epidemiological approach. In the absence of randomized controlled trials, MR stands out as a highly effective method for investigating causal associations between exposures of interest and outcomes. This can be achieved via utilizing genetic variants as instrumental variables (IVs) to estimate the causal associations between exposures and outcomes. Furthermore, MR can help in reducing confounding factors and the bias of reverse causality, which is inherent in observational studies. Due to the limited comprehension of the cause-and-effect association between blood metabolites and IPF, more studies are urgently needed. Therefore, the current study employed MR analysis to thoroughly investigate the potential causality of 486 blood metabolites in the progression of IPF via using pooled data from genome-wide association studies (GWAS). Exploring these particular metabolites and their association with IPF could promote the improved understanding of the pathogenesis of IPF and lay a solid foundation for exploring the potential mechanisms and therapeutic targets of IPF in the future.

Materials and methods

Data of 486 blood metabolites and IPF from GWAS

Data for 486 metabolites were obtained from the Metabolomics GWAS summary database (https://metabolomics.helmholtz-muenchen.de/gwas/), which includes a total of 2,163,597 single nucleotide polymorphisms (SNPs) associated with them (19). The names of the 486 metabolites are listed in detail in Table SI. X indicates an unidentified chemical composition. The FinnGen biospecimen repository (https://www.finngen.fi/en) was utilized as an outcome variable for conducting GWAS analysis on IPF. In the current study the analysis included genotypic data from 1,028 patients with IPF and 196,986 controls (20).

Selection of IVs

To evaluate the possible cause-and-effect association between circulating metabolites and IPF, a two-sample MR analysis was conducted (21). The present study assessed blood metabolites and IPF as factors of exposure and outcome. The three essential assumptions (22) that should be met for IVs to be valid are depicted in Fig. 1. Therefore, it is crucial that the genetic variants employed as IVs exert a robust association with blood metabolites. Furthermore, genetic mutations should not be associated with any potential confounding variables. Thirdly, the genetic variations should exclusively affect IPF via blood metabolites rather than any alternative pathways. Therefore, to meet the aforementioned three assumptions, a significance threshold of P<1x10-5 was set for autonomy (23). A lower P-value threshold, such as that of P<1x10-5, could effectively control the false positive rate. However, it could also increase the risk of false negative results. Choosing this value could balance the two to a particular extent, ensuring that the results would have a certain level of reliability when discovering potential related SNPs. In large sample studies, lower P-value thresholds, such as P<5x10-8, are commonly used in GWAS, since they can effectively reduce false positive results caused by multiple comparisons. However, if the sample size is small, using overly strict thresholds can miss out several truly relevant SNPs. Therefore, choosing a P<1x10-5 could be more suitable for the current research scenario. Additionally, to guarantee the autonomy of SNPs and eliminate linkage disequilibrium (LD) (24), a LD threshold of r2<0.001 and a distance of 10,000 kb were implemented. r² is a statistical measure of LD between two loci. Therefore, the closer the value is to 1, the stronger the LD between these two loci. r²<0.001 indicates that there is almost no linkage imbalance between these two loci and they can be therefore considered independent from each other. Setting r2<0.001 could help identify markers that are independent of the phenotype and avoid LD-caused bias. This strict selection could improve the accuracy of the results, ensuring that the discovered effects are independent and not caused by the effects of other loci. Furthermore, the PhenoScanner (25) (http://www.phenoscanner.medschl.cam.ac.uk/) website was utilized to evaluate if these SNPs were associated with confounding variables in IPF. Currently, several risk factors for IPF have been identified, such as smoking, exposure to dust and reflux esophagitis. Therefore, in the present study, when any associations were identified, individuals with the aforementioned variations were excluded.

Figure 1 Overview of the analysis. MR, Mendelian randomization; IPF, idiopathic pulmonary fibrosis.

Statistical analysis

Various strong statistical methods were employed to guarantee the dependability and authenticity of the results. However, owing to its exceptional resilience, the inverse variance weighted (IVW) method was used as the main analytical approach (26). The IVW method is commonly applied in fixed- and random-effect models, which help mitigate the heterogeneity-related bias (27). Furthermore, the robustness of the findings was assessed via employing the MR-Egger regression and weighted median techniques. Several studieshave argued that the MR-Egger regression technique lacks statistical validity for estimating causality (27,28). Therefore, it was suggested that this method should only serve as a sensitivity analysis to assess the violation of core assumptions of IVs, rather than being used as a substitute for the IVW method (27). It is generally recognized that SNPs with F-statistics <10 are weakly instrumented variables, while those of >10 indicate no significant weakly instrumented variable bias. In the present study, SNPs with F-statistics of <10 in the MR analysis were excluded. To calculate the F value, the following formula was used: F=R2 x (n-k-1)/(1-R2) x k. The sample size, denoted by n, indicates the genetic variance denoted by R2, while k indicates the number of SNPs present in the sample (29).

Sensitivity analysis

The heterogeneity of IVW and MR Egger regression methods was evaluated using Cochran's Q test and funnel plots. The results of Cochran's Q test (30) with P>0.05 indicated the absence of significant heterogeneity. The intercept in the MR-Egger regression represents the magnitude of horizontal pleiotropy. Therefore, the smaller the horizontal pleiotropy, the more the intercept tends to 0. In addition, to evaluate whether the findings were significantly affected by a solitary SNP, a leave-one-out examination was performed (31). This involved excluding each SNP at a time and subsequently conducting an MR analysis.

Results

Analysis of the associations between 486 blood metabolites and risk of IPF

After rigorous quality evaluation of the IVs, a total of 486 blood metabolites were obtained for MR analysis. LD analysis and removal of palindromic sequences to enhance accuracy and reliability of the selected IVs were also performed. Among the 486 blood metabolites identified, five appeared in two forms, thus resulting in a total of 491 metabolites. All SNPs demonstrated adequate validity as indicated by the F-statistics, which exceeded the empirical threshold of 10. The results of the analysis of the 486 metabolites are displayed in Table SII. Additionally, using the IVW methodology, a total of 21 circulating metabolites were identified, that could exert a causal role in IPF risk. Among the aforementioned circulating metabolites, 10 were known compounds and 11 were unknown (Fig. 2). Namely, the circulating metabolites identified were the following: Cotinine (OR=1.206; 95% CI, 1.002-1.452; P=0.047), hypoxanthine (OR=0.225; 95% CI, 0.056-0.899; P=0.034), aspartyl phenylalanine (OR=4.309; 95% CI, 1.084-17.131; P=0.038), acetyl-carnitine (OR=5.767; 95% CI, 1.398-23.789; P=0.015), 2-aminobutyrate (OR=0.155; 95% CI, 0.033-0.713; P=0.016), docosapentaenoate (OR=0.214; 95% CI, 0.055-0.833; P=0.026), octanoyl-carnitine (OR=3.398; 95% CI, 1.179-9.794; P=0.023), alpha-hydroxy-isovalerate (OR=0.324; 95% CI, 0.112-0.931; P=0.036), 1,7-dimethylurate (OR=0.401; 95% CI, 0.172-0.931; P=0.033), 1-linoleoylglycerophosphocholine (OR=6.559; 95% CI, 1.060-40.557; P=0.043), X-10395 (OR=3.592; 95% CI, 1.263-10.210; P=0.016), X-10810 (OR=3.42; 95% CI, 1.263-10.210; P=0.016), X-11261 (OR=2.321; 95% CI, 1.009-5.335; P=0.047), X-11795 (OR=5.13; 95% CI, 1.150-22.885; P=0.032), X-12189 (OR=0.803; 95% CI; 0.669-0.964; P=0.019), X-12428 (OR=0.503; 95% CI, 0.259-0.973; P=0.041), X-12443 (OR=1.806; 95% CI, 1.051-3.104; P=0.032), X-12786 (OR=3.371; 95% CI, 1.206-9.423, P=0.020), X-13671 (OR=23.162; 95% CI; 3.091-173.553; P=0.002), X-14374 (OR=0.102; 95% CI; 0.011-0.936; P=0.043) and X-14745 (OR=4.947; 95% CI; 1.024-23.895; P=0.046). The analysis also revealed that the high levels of cotinine, aspartyl-phenylalanine, acetyl-carnitine, octanoyl-carnitine and 1-lipoyllglycophorophosphacholine were potential risk metabolites for IPF. By contrast, the high levels of hypoxanthine, 2-aminobutyrate, docosapentaenoate alpha hydroxy-isovalerate and 1,7-dimethylurate could reduce the risk of IPF. These findings could be helpful for detecting the incidence and predicting the prognosis of IPF in clinical practice. The results of the MR analysis are visualized in Table SIII, while those of sensitivity analysis are listed in Table I. Cochran's Q-test yielded insignificant heterogeneity, while MR-Egger's intercept analysis verified the lack of horizontal pleiotropy. The reliability of the MR analysis was also verified by the leave-one-out graphing (Fig. 3). The scatter plot is displayed in Fig. 4.

Figure 2 Forest plot of the MR-analyzed blood metabolites on idiopathic pulmonary causality. MR, Mendelian randomization; CI, confidence interval; OR, odds ratio; SNPs, single nucleotide polymorphisms.

Figure 3 Leave-one-out results are shown.

Figure 4 Scatter plots of the Mendelian randomization analysis are presented.

Table I Sensitivity analysis.

Table I

Sensitivity analysis.

Metabolites	Q_pval (IVW)	MRegger_intercept	MRegger_interpreter_pval
Cotinine	0.634204819	-0.016253799	0.594349888
Hypoxanthine	0.64539929	-0.014255373	0.658951597
Aspartylphenylalanine	0.507548038	-0.0100362	0.8937339
X-10395	0.421558055	0.005325463	0.697704685
X-10810	0.942302112	0.016006743	0.482089942
Acetylcarnitine	0.854090843	0.007536244	0.833125497
2-aminobutyrate	0.94823233	-0.004991164	0.84015343
Docosapentaenoate (n3 DPA; 22:5n3)	0.857218151	-0.020539902	0.729874938
X-11261	0.752776285	-0.029700565	0.386080027
X-11795	0.812512936	-0.032772067	0.537999752
X-12189	0.19702192	-0.02414237	0.332548791
X-12428	0.49662327	-0.009779736	0.805574022
X-12443	0.748525852	0.004634389	0.921132003
Octanoylcarnitine	0.198556909	0.003739538	0.894565868
Alpha-hydroxyisovalerate	0.480656564	-0.057310933	0.218243317
1,7-dimethylurate	0.86047744	0.026635303	0.450327514
1-linoleoylglycerophosphocholine	0.806629731	-0.040877387	0.380697155
X-12786	0.824668697	0.009070153	0.770926683
X-13671	0.667042516	0.037205512	0.501665959
X-14374	0.124767596	-0.026570989	0.822873085
X-14745	0.384835525	-0.057805481	0.606030326

[i] MR, Mendelian randomization; IVW, inverse variance weighted.

Discussion

The present study aimed to investigate the possible causal effect of 486 blood metabolites on IPF progression. Therefore, via combining two extensive GWAS datasets and implementing a rigorous MR design, a total of 13 genetic metabolites, namely cotinine, aspartyl phenylalanine, acetyl-carnitine, octanoyl-carnitine, 1-linoleoylglycerophosphocholine, X-10395, X-10810, X-11261, X-11795, X-12443, X-12786, X-13671 and X-14745, were identified, which were associated with enhanced susceptibility to IPF. By contrast, the analysis predicted eight genetic metabolites, namely hypoxanthine, 2-aminobutyrate, docosapentaenoate, alpha-hydroxy-isovalerate, 1,7-dimethylurate, X-12189, X-12428 and X-14374, that could reduce the risk of IPF.

The rising morbidity and significant mortality of IPF in recent years have imposed a substantial burden on individuals globally, emphasizing the need for prompt screening and prevention as a critical strategy. The emergence of metabolomics technologies has sparked growing curiosity in investigating the significance of metabolites associated with IPF. Due to the chemical nature of exposures originating from both internal and external sources, blood samples can serve as a means to identify exposure groups (32). A previous study supported the reliability of S100 calcium-binding protein A12 as a serum biomarker for assessing the severity and prognosis of IPF (33). 5-methoxytryptophan (5-MTP) has been recently identified as a tryptophan metabolite. Therefore, a study indicated that 5-MTP could combat inflammation, inhibit tumor growth, protect blood vessels and prevent fibrosis in kidney disorders. Additionally, the aforementioned study showed that 5-MTP could be also significantly involved in IPF and it was therefore anticipated to serve as a treatment strategy for pulmonary fibrosis (34). Glutaminolysis, a crucial metabolic pathway in IPF, involves the conversion of glutamine to glutamate by glutaminase and subsequently to α-ketoglutarate, a circulating metabolite of the tricarboxylic acid cycle (35). Emerging evidence has suggested that the immune system also exerts a significant role in the development of IPF. Therefore, it has been reported that IL-1β, IL-6, IL-23 and IL-17A, involved in type 17 immunity, play a significant role in both pulmonary fibrosis and acute exacerbation of pulmonary fibrosis. Targeting type 17 immunity could emerge as a novel therapeutic approach to hinder the advancement or onset of pulmonary fibrosis (36). Although the existing literature has provided conflicting results for particular metabolites, significant differences in the expression of serum metabolites have been identified between patients with IPF and healthy controls. However, further studies are needed to establish a definitive cause-and-effect association and gain a deeper comprehension of the underlying mechanisms. In the present study, the c MR method was used to elucidate the cause-and-effect association between blood metabolites and IPF, along with the metabolic pathways implicated. The present study aimed to offer guidance for the screening and management of IPF.

The study revealed that elevated levels of cotinine, aspartyl phenylalanine, acetyl-carnitine, octanoyl-carnitine and 1-linoleoylglycerophosphocholine were associated with a heightened susceptibility to IPF. A previous study demonstrated that cotinine, which serves as a biomarker for smoking exposure, was associated with an increased risk of kidney stones, coronary heart disease and chronic obstructive pulmonary disease (37). In addition, high levels of serum cotinine could also reduce muscle mass, hepatic steatosis and liver fibrosis (38-40). The present study verified that cotinine could be involved in elevated risk of IPF, thus suggesting that the smoking cessation management should be strengthened in clinical practice. Aspartyl phenylalanine, which is formed when cholecystokinin-8 breaks down, but not when angiotensin breaks down by angiotensin-converting enzyme, has been associated with an increased risk of acute coronary syndromes (41). Additionally, another study revealed a cause-and-effect association between aspartyl phenylalanine and hypertension and hypertriglyceridemia (41). The deficiency of alpha aspartyl phenylalanine hydrolase activity is considered to be the reason for the possible adverse reactions of aspartame intake (42,43). Therefore, reducing the intake of aspartame could be beneficial for preventing IPF. However, further research is needed to confirm this finding. Fatty acid metabolites, such as acetyl-carnitine and octanoyl-carnitine, serve significant roles in several cellular energy metabolic pathways and are commonly utilized as nutritional supplements. The identification of acyl-carnitines has become increasingly crucial for the study of metabolism in various diseases, such as metabolic disorders, heart disease, diabetes, depression, neurological disorders and particular types of cancer (44). Therefore, it was hypothesized that its effect on IPF could be associated to its involvement in histone acetylation. However, further studies are needed to verify this finding. 1-linoleoylglycerophosphocholine is a phospholipid, which is formed from linoleic acid and glycerol. It is involved in several processes in the human body, such as building and upkeeping cell membranes and acting as a signaling molecule. It is also used as an ingredient in nutritional supplements and medications. A previous study suggested that 1-oleoylglycerophosphate choline could notably reduce the risk of type 2 diabetes (45). In the present study, elevated 1-lipoloylglycophorophosphacholine levels were associated with increased susceptibility to IPF. During the pathological process of IPF, the cell membrane structure can be damaged, while the cellular function can be disrupted. When the levels of 1-lipoloylglycophorophosphacholine are abnormally enhanced, it can disrupt the lipid balance of the cell membrane, and affect membrane protein function and the exchange and signal transmission of substances inside and outside the cell. For example, it can interfere with the binding of cell surface receptors and ligands, thus affecting the response of cells to signaling molecules, such as growth factors and cytokines, thereby promoting abnormal activation and proliferation of fibroblasts, excessive deposition of extracellular matrix, and advancing the process of pulmonary fibrosis (46,47).

In the present study, the following metabolites with protective effects on IPF were identified: Hypoxanthine, 2-aminobutyrate, docosapentaenoate, alpha-hydroxy-isovalerate and 1,7-dimethylurate. Therefore, the particular mechanism of their protective effects on IPF requires further in-depth study. Hypoxanthine is a purine base, which plays a significant role in living organisms. It is a component of DNA and RNA, and is also involved in the process of nucleotide synthesis. In addition, hypoxanthine can be converted to uric acid via metabolic pathways, thus affecting purine metabolism and acting as an indicator of hypoxia. Purine plays a key role in regulating immune response during wound healing and fibrosis and is a key signal hub for scar tissue formation in the fibrotic region. The current study strengthened the study on the improvement of IPF by hypoxanthine, thus providing novel insights into the identification of potential targets for IPF. 2-aminobutyrate, an amino acid, plays multiple functions within the human body. Protein synthesis, metabolism, as well as neurotransmitter synthesis are its areas of involvement. It has been reported that 2-aminobutyrate can also have antioxidant and anti-inflammatory effects, which could account for its protective effects against IPF. Docosapentaenoic acid, a type of omega-3 fatty acid, is specifically classified as a long-chain polyunsaturated fatty acid. It is also known as DPA and is particularly found in fish and marine oils. Docosapentaenoic acid was found to exert similar health benefits with other omega-3 fatty acids, such as reducing inflammation, supporting heart health and potentially improving cognitive function (48). During the degradation of leucine, alpha-hydroxy-isovalerate is produced as a metabolite. Alpha-hydroxy-isovalerate is an organic compound involved in several metabolic pathways in the body. Researchers are currently examining its possible involvement in the identification and treatment of different ailments, such as maple syrup urine disease and other conditions associated with the metabolism of branched-chain amino acids (49). 1,7-Dimethylurate is a metabolite of caffeine, also known as dimethylxanthine. It is produced through the metabolism of caffeine in the body and has antioxidant and anti-inflammatory properties (50). For the unknown serum metabolites found to be associated with IPF risk, further attention should be paid in future studies to explore their possible mechanisms underlying their effect on IPF, thus providing assistance and insights for the clinical prevention and treatment of IPF.

However, the present study has certain limitations. Firstly, a limited number of SNPs, that could be used for genome-wide level exposure, were detected. In order to tackle this problem, slightly more lenient criteria were established for MR analysis, a widely adopted approach. Nevertheless, the F-test statistics for all chosen SNPs surpassed 10, thus suggesting that, in the present study, the instrumental variable was adequately strong. Furthermore, the research sample was presently restricted to individuals of European descent. Therefore, the applicability to other populations requires further investigation and confirmation. Further exploration is required to uncover the precise mechanisms underlying the effects of particular metabolites on the development of IPF. Due to limited data sources, in the current study, only 486 metabolites were covered. Therefore, more studies on additional blood metabolites associated with IPF should be performed. A purely bioinformatic analysis of existing data from one source indeed presents several limitations. Firstly, the data may not be representative of the entire population. For example, if the source database has a bias in terms of geographical location or ethnicity, it could lead to inaccurate or incomplete conclusions about the relationship between metabolites and the studied condition. Secondly, the lack of experimental verification means that the bioinformatic predictions may not accurately reflect the in vivo or in vitro biological processes. There could be confounding factors or interactions that are not accounted for in the data. Thirdly, the static nature of the existing data may not capture the dynamic changes in metabolite levels over time or in response to different stimuli. To address these limitations, future research could involve collecting data from multiple diverse sources to improve representativeness. Experimental studies such as cell culture and animal models could be conducted to validate the bioinformatic findings and explore the underlying mechanisms. Longitudinal studies could also be designed to monitor metabolite fluctuations and their associations with disease progression.

In conclusion, the present magnetic resonance study indicated that numerous blood metabolites could play a key role in IPF, thus offering valuable perspectives on possible approaches for the early detection, avoidance and management of IPF. By combining genomics and metabolomics, this analysis could provide a point of reference for investigating the cause and development of IPF.

Supplementary Material

List of the identification (ID) for each of the 486 blood metabolites.

Analysis results of 486 metabolites and idiopathic pulmonary causality.

Mendelian randomization analysis results of 21 blood metabolites associated with idiopathic pulmonary causality risk.

Acknowledgements

Not applicable.

Funding

Funding: The present study was supported by the Shandong Taishan Scholar Project (grant no. tsqn202306392).

Availability of data and materials

The data generated in the present study are included in the figures and/or tables of this article.

Authors' contributions

FW and XL designed the study and drafted the manuscript. FW, JL, BL, WY, XZ and XL performed the data collection and analysis. All authors read and approved the final version of the manuscript. FW and XL confirm the authenticity of all the raw data.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References