The genetic associations of lipidome on bladder cancer: a Mendelian randomization study

Abstract

Purpose

Multiple epidemiological studies have demonstrated a correlation between the lipidome and bladder cancer (BC). Nonetheless, the current literature lacks consensus on the causal link between the lipidome and BC. We utilise a two-sample Mendelian randomisation approach to meticulously evaluate the causal link between the two variables.

Methods

A two-sample Mendelian randomisation (MR) analysis was performed utilising publically accessible genome-wide association research data. The principal technique utilised for the inquiry was the inverse variance-weighted (IVW) meta-analysis. Furthermore, Bayesian weighted Mendelian randomisation (BWMR) was employed to corroborate the findings. Supplementary validation was conducted utilising Cochran’s Q test and methodologies including MR-Egger.

Results

The two-sample MR analysis, in conjunction with BWMR analysis, revealed a correlation between the genetic prediction of the lipidome and the risk of BC. Sterol ester (SE) 27:1/16:0 (OR: 1.148, 95% CI: 1.020–1.293, p = 0.022), Sterol ester 27:1/20:3 (OR: 1.132, 95% CI: 1.006–1.274, p = 0.040), Phosphatidylcholine (PC) 18:0_20:3 (OR: 1.257, 95% CI: 1.101–1.436, p = 0.001), Sphingomyelin (SM) d38:1 (OR: 1.120, 95% CI: 1.016–1.235, p = 0.023), Sphingomyelin d40:2 (OR: 1.156, 95% CI: 1.033–1.295, p = 0.012), Triacylglycerol (TAG) 46:1 (OR: 1.178, 95% CI: 1.013–1.369, p = 0.034), Triacylglycerol 49:2 (OR: 1.219, 95% CI: 1.031–1.442, p = 0.021), Triacylglycerol 50:5 (OR: 1.173, 95% CI: 1.038–1.326, p = 0.011), and Triacylglycerol 52:6 (OR: 1.161, 95% CI: 1.007–1.339, p = 0.040) exhibited a positive correlation with an elevated risk of BC.

Conclusion

Genetic projections suggest a reciprocal causal link between the lipidome and BC, offering theoretical support and a basis for future clinical research.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-02389-3.

Introduction

Bladder cancer (BC) is the tenth most prevalent malignancy worldwide and the thirteenth largest cause of cancer-related mortality internationally [1]. Research predicts that by 2040, the yearly incidence of breast cancer cases and fatalities worldwide is expected to rise by 73 and 87%, respectively [2]. This rising trend highlights the critical necessity for comprehensive research into the aetiology of BC, associated risk factors, and early detection techniques. Prior research has identified advanced age, smoking, certain occupational exposures, exposure to hazardous substances, and pelvic radiation therapy as established risk factors linked to BC [3]. Recent studies have emphasised the potential of diverse biomarkers, including miRNAs, protein markers, and lipid profiles, as early diagnostic and prognostic indications for BC [4]. The causal relationship between liposomes and bladder cancer remains unclear, and little is known whether liposomes can become targets for the diagnosis and treatment of BC.

The lipidome denotes the aggregate of thousands of lipid molecules present within an organism. Lipids consist of a varied category of biological substances, encompassing fatty acids, triglycerides, phospholipids, sphingolipids, and sterols. They serve essential biochemical and structural functions inside the organism [5]. The lipidome is essential in the mechanisms of carcinogenesis and tumour growth. Sodium palmitate C16:0 has demonstrated the ability to impede the onset and advancement of colorectal cancer by diminishing cell membrane fluidity [6], whereas phosphatidylcholine (34:1) has been identified as having anti-thyroid papillary carcinoma properties [7]. Linda Ottensmann conducted a comprehensive analysis of the lipidome’s whole genome, revealing 14 gene coding variants that may possess causal relationships [8]. These findings offer significant insights into diseases associated with human lipid metabolism and suggest potential therapeutic targets for tumor-targeted therapy. Nonetheless, disagreement endures concerning the association between the lipidome and BC, as well as the precise lipidome components linked to BC. While several studies have shown correlations between specific lipidome profiles and BC, the quality of evidence is frequently low and vulnerable to confounding variables or reverse causation. Consequently, there is a necessity for novel, dependable techniques to clarify the genetic causal link between the lipidome and BC.

With the advancement of programs such as the Human Genome Project, a growing number of diseases are linked to epigenetics. MR is an epidemiological study design that employs single nucleotide polymorphisms (SNPs) as exposure variables and genetic variations as instrumental variables (IVs) to mitigate the risks of confounding factors and reverse causality typically encountered in observational studies [9]. This method allows for a more reliable assessment of the impact of exposure factors on health outcomes, yielding estimates that are closer to actual causal correlations. BWMR analysis, an advancement of conventional MR analysis, provides a more accurate and dependable approach for examining causal relationships between exposure factors and outcomes by allocating greater weights to genetic variants, thereby enhancing their validity [10]. This work used MR and BWMR techniques to investigate the genetic causal link between the lipidome and BC. This establishes a theoretical foundation for future clinical research initiatives.

Methods

This study used a two-sample MR analysis to evaluate the influence of the lipidome on BC. MR analysis is predicated on three fundamental assumptions: Genetic variants influence outcomes only through exposure, without involvement of alternative routes; genetic variation correlates with exposure; genetic variation is unassociated with confounding factors. Figure 1 depicts the schematic diagram of this design.

Fig. 1.

The overview of the present MR for the association between lipidome and BC

The data source

Acquire SNP-phenotype association data from the Genome-wide association studies (GWAS) catalogue, which includes 179 lipidome-associated symptoms. The genotypic data for BC utilised in this study is exclusively sourced from the Finngen database (R9 version), which comprises 314,193 persons of European ancestry. Detailed information concerning the genetic datasets utilised in this design is included in Table 1.

Table 1.

Details of studies and datasets used for analyses

Exposures	Consortium	Ethnicity	Participants	Sex
Lipidosome
Sterol ester (27:1/16:0)	Gwas catalog	European	7174	Males and females
Sterol ester (27:1/20:3)	Gwas catalog	European	7171	Males and females
Phosphatidylcholine (18:0_20:3)	Gwas catalog	European	7169	Males and females
Phosphatidylcholine (18:2_20:4)	Gwas catalog	European	7049	Males and females
Sphingomyelin (d38:1)	Gwas catalog	European	7174	Males and females
Sphingomyelin (d40:2)	Gwas catalog	European	7174	Males and females
Triacylglycerol (46:1)	Gwas catalog	European	6070	Males and females
Triacylglycerol (49:2)	Gwas catalog	European	6076	Males and females
Triacylglycerol (50:5)	Gwas catalog	European	5491	Males and females
Triacylglycerol (52:6)	Gwas catalog	European	6554	Males and females
Outcomes
Bladder cancer	Finngen	European	314,193	Males and females

The selection of genetic instrumental variables (IVs)

This study identifies genome-wide SNPs for the lipidome as IVs (p < 5 × 10⁻⁸) to establish a persistent causal connection with BC [11]. To ascertain the presence of SNPs exhibiting linkage disequilibrium, we evaluate linkage disequilibrium by filtering SNPs within a 10,000 kb window using a threshold of R² less than 0.001 to guarantee independence among SNPs [12]. Subsequently, SNPs linked to possible confounders were eliminated. The variables examined comprised age, sex, body mass index (BMI), smoking status, and comorbidities recognised to influence lipid metabolism and the risk of bladder cancer. SNPs significantly associated with these parameters were systematically excluded from the analysis to guarantee impartial results. Furthermore, SNPs harmonisation is performed to rectify the orientation of the alleles.

The MR combined with BWMR analysis

Through a two-sample MR analysis, we established initial causal correlations. The preliminary analysis employed the IVW method, often regarded as the most reliable approach in the absence of evidence for directional pleiotropy [13]. Furthermore, we performed supplementary analyses utilising MR Egger, weighted median, simple mode, and weighted mode. Among these five methodologies, IVW and MR-Egger are regarded as the most significant for ascertaining outcomes. A significant score of p < 0.05 signifies a genetic causal link between the lipidome and BC. Additionally, to more precisely evaluate the causal influence of the lipidome on BC, we conducted further validation via BWMR analysis. BWMR is a statistical technique for causal inference that accommodates uncertainty, facilitating a more adaptable approach to intricate causal inference challenges. Taking into account uncertainty and bias in the conclusion can more precisely represent the actual causal link within the data [10]. To address potential directional pleiotropy, the MR-Egger regression method was utilised, which may identify and rectify pleiotropic effects by estimating an intercept term. The MR-Egger method employs weighted linear regression analysis, utilising genetic exposure coefficients as the independent variable and genetic outcome coefficients as the dependent variable [14]. In addition, MR-pleiotropy residual sum and outliers (MR-PRESSO) was employed to identify horizontal pleiotropy [15].

Sensitivity analysis

To further enhance the confidence in the study results, sensitivity analysis was conducted to validate positive results from the MR analysis. For positive results, heterogeneity was first examined using the Cochran’s Q test method. A p-value less than 0.05 indicates heterogeneity among instrumental variables (IVs), while a p-value greater than 0.05 indicates no heterogeneity. Lastly, since the MR-Egger method considers an intercept term in the regression analysis, we compared the MR-Egger method with the IVW method using the horizontal pleiotropy test to verify whether there is horizontal pleiotropy among IVs [16]. If the p-value for the intercept is less than 0.05, it indicates significant horizontal pleiotropy. Additionally, sensitivity analysis was performed using scatter plots and funnel plots. If the funnel plot is symmetrical, the results are stable.

Statistical analysis

The effects of exposure and outcome on allele genes were matched using the "TwoSampleMR" and "MRPRESSO" packages in R (version 4.3.3). A p-value less than 0.05 indicates statistical significance for the observed differences.

Ethical approval

This study is a Mendelian randomization analysis based on publicly available data from the GWAS database. As it does not involve direct human participation, it poses no ethical risks and therefore does not require approval from an ethics committee.

Results

MR analysis and BWMR analysis

The IVW analysis results for 179 lipidome and BC indicate that there are a total of 10 lipidome showing causal relationships with BC. The IVW analysis reveals that an increase in Phosphatidylcholine (18:2_20:4) (OR: 0.855, 95% CI: 0.754–0.970, p = 0.015) is negatively correlated with the risk of BC incidence. Conversely, the risk of BC incidence is positively associated with Sterol ester 27:1/16:0 (OR: 1.148, 95% CI: 1.020–1.293, p = 0.022), Sterol ester 27:1/20:3 (OR: 1.132, 95% CI: 1.006–1.274, p = 0.040), Phosphatidylcholine 18:0_20:3 (OR: 1.257, 95% CI: 1.101–1.436, p = 0.001), Sphingomyelin d38:1 (OR: 1.120, 95% CI: 1.016–1.235, p = 0.023), Sphingomyelin d40:2 (OR: 1.156, 95% CI: 1.033–1.295, p = 0.012), Triacylglycerol 46:1 (OR: 1.178, 95% CI: 1.013–1.369, p = 0.034), Triacylglycerol 49:2 (OR: 1.219, 95% CI: 1.031–1.442, p = 0.021), Triacylglycerol 50:5 (OR: 1.173, 95% CI: 1.038–1.326, p = 0.011), and Triacylglycerol 52:6 (OR: 1.161, 95% CI: 1.007–1.339, p = 0.040) (Fig. 2). To augment the dependability of the analytical outcomes, we employed BWMR analysis to corroborate the IVW analysis results. The results of BWMR analysis are roughly consistent with those of IVW Analysis. The ten lipidome components positively correlated with an elevated risk of breast cancer include Sterol ester 27:1/16:0 (OR: 1.151, 95% CI: 1.011–1.312, p = 0.034), Sterol ester 27:1/20:3 (OR: 1.178, 95% CI: 1.027–1.352, p = 0.019), Phosphatidylcholine 18:0_20:3 (OR: 1.125, 95% CI: 1.077–1.416, p = 0.003), Phosphatidylcholine (18:2_20:4) (OR: 0.848, 95% CI:0.743–0.968, p = 0.015), Sphingomyelin d38:1 (OR: 1.124, 95% CI: 1.016–1.242, p = 0.023), Sphingomyelin d40:2 (OR: 1.158, 95% CI: 1.031–1.301, p = 0.014), Triacylglycerol 46:1 (OR: 1.158, 95% CI: 1.007–1.387, p = 0.041), Triacylglycerol 49:2 (OR: 1.233, 95% CI: 1.035–1.469, p = 0.019), Triacylglycerol 50:5 (OR: 1.184, 95% CI: 1.041–1.348, p = 0.010), and Triacylglycerol 52:6 (OR: 1.172, 95% CI: 1.016–1.353, p = 0.030). (Fig. 3). These findings further illustrate the reliability and consistency of the present MR study.

Fig. 2.

MR analysis between lipidome and BC

Fig. 3.

BWMR analysis between lipidome and BC

The analysis for pleiotropy and heterogeneity

The results of MR-Egger analysis showed that only sterol esters (27:1/20:3) (Egger intercept = −0.582, p = 0.007) had horizontal pleiotropy, while the other nine lipidomes did not show significant horizontal pleiotropy (p > 0.05). Consequently, we utilised a fixed-effects model to illustrate the effect size results in the final analysis (Table 2). The results’ robustness and stability were further validated by leave-one-out analysis and forest plot visualisation (Figures S1–S9).

Table 2.

Heterogeneity and pleiotropy test of this MR analysis

Exposure	Outcome	Heterogeneity		Pleiotropy		MR-PRESSO
		Cochran’s Q	p-value	Egger-intercept	p-value	RSSobs	p-value
Sterol ester (27:1/16:0)	Bladder cancer	38.932	0.186	−0.014	0.396	41.926	0.213
Sterol ester (27:1/20:3)	Bladder cancer	19.763	0.873	−0.582	0.007	29.832	0.519
Phosphatidylcholine (18:0_20:3)	Bladder cancer	15.901	0.918	0.004	0.867	17.427	0.947
Phosphatidylcholine (18:2_20:4)	Bladder cancer	28.855	0.473	−0.012	0.511	31.559	0.49
Sphingomyelin (d38:1)	Bladder cancer	31.001	0.705	−0.013	0.388	33.259	0.767
Sphingomyelin (d40:2)	Bladder cancer	34.651	0.437	−0.019	0.249	37.949	0.444
Triacylglycerol (46:1)	Bladder cancer	16.247	0.701	−0.036	0.141	20.469	0.628
Triacylglycerol (49:2)	Bladder cancer	7.980	0.925	0.004	0.878	9.089	0.946
Triacylglycerol (50:5)	Bladder cancer	21.794	0.592	−0.001	0.970	23.525	0.684
Triacylglycerol (52:6)	Bladder cancer	24.753	0.309	−0.019	0.405	27.554	0.360

Discussion

BC is the most frequent urinary system malignancy and the ninth most common malignant tumour worldwide [17]. BC is treated mostly with surgical resection and intravesical chemotherapy. Clinicians face a major issue because BC patients’ long-term survival rates have remained low for decades despite these therapies. Recent evidence suggests the lipidome is critical to cancer initiation and progression [18]. Cell membranes are made of lipids, which are essential for cell signalling and energy metabolism. Lipids such lysophosphatidylcholines influence pro-tumorigenic signalling pathways in BC cells, encouraging BC development [19]. Others show that BC affects lipid composition [20] and can cause aberrant lipid metabolism. These observational studies often have confounding factors and reverse causality interference. It is unknown if lipidome changes cause BC or respond to it. Thus, more thorough and trustworthy study on BC-lipidome interactions is needed.

Genetic variations are instrumental variables in MR analysis, reducing common confounding factors and reverse causality in observational investigations. BWMR analysis improves research rigour in establishing causal links between exposure factors and outcomes. MR and BWMR studies were used to determine the lipidome-BC causal link. We can infer rather accurate findings from our examination of 179 GWAS catalogue dataset lipidome species. We found that elevated levels of two Sterol esters (SE 27:1/16:0 and SE 27:1/20:3), Phosphatidylcholine 18:0_20:3, two Sphingomyelins (SM d38:1 and SM d40:2), and four Triacylglycerols (TAG 46:1, TAG 49:2, TAG 50:5, and TAG 52:6) increased BC risk.

Esterification of sterols produces sterol esters (SEs) with a C-3 hydroxy sterol core and a C-17 side chain. SEs mediate reciprocal LXR and SREBP2 pathway changes, causing T cell cholesterol deficit, limiting T cell proliferation and activating autophagy-mediated cell death, causing tumour development and progression [21]. There is still no definite research on SE and BC prognosis. Some studies show that BC patients have greater urine SE levels than healthy people [22]. This may be linked to BC patients’ urothelial cell membrane damage.Our investigation found a link between high sphingomyelin (SM) and BC risk. This shows that SM may be a risk factor for BC and contributes to its pathogenesis. SM is a significant cell membrane component, and its metabolism is linked to tumour development. Research shows liver cancer patients’ peripheral blood NK cells can synthesise SM. Further investigations have shown that decreasing SM production in peripheral blood NK cells can impair tumour NK cell membrane topology and cytotoxicity, causing an anti-tumor impact [23]. SM metabolism can be hydrolysed by activating the sphingomyelinase enzyme family to produce physiologically active metabolites like ceramide and S1P. Ceramide levels in normal bladder tissue are lower than in BC tissue, which may be linked to tumour cell growth and migration [24]. BC cells can avoid immune surveillance by migrating Tregs via S1P, SM’s metabolic product. S1P also stimulates BC cell proliferation, migration, and angiogenesis, advancing tumours [25]. Cancer research is increasingly focused on triglyceride (TAG) metabolism. TAG has been thought to store energy, but current research suggest it may affect cancer progression. First, TAG breakdown gives cancer cells an alternative energy source, especially in glucose-limited conditions, encouraging cell survival and multiplication. Fatty acids released from TAG hydrolysis can also signal cancer growth pathways such cell proliferation, apoptosis resistance, inflammation, and angiogenesis. TAG may increase tumour growth and metastasis in some tumours by interacting with tumour microenvironment immune cells [26]. Greater TAG levels are associated with greater BC risk, according to our findings. However, BC TAG processes need further study. SE may boost BC occurrence and progression, while SM may aid BC invasion and immune evasion. The mechanics of BC TAGs need additional study. Recent studies reveal that PC may affect BC progression through many paths. First, biologically active lipid compounds from PC metabolism may regulate cell proliferation, death, and inflammation. Moreover, aberrant PC metabolism causes lipid droplets and cancer invasiveness [27]. Aberrant PC metabolism may regulate epithelial-mesenchymal transition (EMT), boosting tumour spread and invasion [28]. Research shows that bladder cancer tissue has more PC than normal bladder tissue. QCM-D research indicates that unsaturated PC can act as a "recognition motif" for broad-spectrum anticancer compounds like α1-linoleic acid [29]. The mechanisms of PC 18:0_20:3 in BC must be investigated immediately.These findings lay the groundwork for understanding BC pathophysiology and developing effective treatments.

Our research offers many advantages. GWSA data was used to infer causation between the lipidome and BC, eliminating confounding factors and bias. Second, we used sensitivity analysis and other ways to verify our results. Our results were more complete and accurate with BWMR analysis than with traditional MR analysis. Our study has inherent limitations. All of our data came from European populations. Across ethnic groups, genetic background may alter instrumental variable selection and applicability; consequently, the consistency of our findings in various populations warrants further exploration. We also need in vitro research to confirm our genetic regulation of long-term lipidomic functions. Cross-sectional Mendelian randomisation studies may not reliably assess causal links over time. Finally, current studies does not explain how the lipidome and BC interact. Our next investigations will examine the lipidome-BC link to provide new directions and insights for BC therapy and prevention.

Conclusion

In summary, this study utilized the two-sample MR Analysis combined with BWMR analysis to reveal causal relationships between ten lipidomes traits and BC, indicating that increased levels of the nine lipidomes traits are associated with increased BC risk. However, further research is needed to evaluate the potential of these biomarkers as early predictors of BC and as preventive strategies.

Abbreviations

BC

Bladder cancer

MR

Mendelian randomization

BWMR

Bayesian weighted Mendelian randomization

SE

Sterol ester

PC

Phosphatidylcholine

SM

Sphingomyelin

TAG

Triacylglycerol

SNPs

Single nucleotide polymorphisms

IVs

Instrumental variables

S1P

Sphingosine-1-phosphate

EMT

Epithelial-mesenchymal transition

QCM-D

Quartz crystal microbalance with dissipation monitoring

Author contributions

J.J. Data curation (lead). W.W. Data curation (lead). J.J. K.Z. Formal analysis (equal), Funding acquisition (equal). C.P. Formal analysis (equal). J.C. Methodology (equal). Y.L. Visualization (equal). G.X. Data curation (equal), writing–original draft (lead). S.P. Funding acquisition (lead).

Funding

This study was supported by Zhejiang Provincial Medicine, Health, and Science and Technology Project (No. 2022KY1297), (No. 2023KY1253) and (No. 2024KY1720).

Data availability

The datasets generated and/or analysed during the current study are available in the GWSA Catalog, with the primary accession code MONDO_0001187.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.