COVID-19, inflammatory cytokines, and digestive system cancers: a bidirectional Mendelian study and mediation analysis

Abstract

Background

Gastrointestinal cancers pose a significant global health burden, contributing to 8% of cancer cases and a substantial number of deaths each year. Meanwhile, the Corona Virus Disease 2019(COVID-19) pandemic has potentially elevated the risk of digestive system cancers (DSCs), and this work aims to investigate the causal relationship between COVID-19 infection andDSCs, and the role of inflammatory cytokines (IC).

Methods

This study employs a two-sample mendelian randomization (MR) approach, using COVID-19, ICs, and DSCs single-nucleotide polymorphisms (SNPs) as instrumental variables to investigate their causal relationship. The primary outcome is assessed using the inverse variance-weighted method. Moreover, mediation analysis was conducted using Two-step method.

Results

This work revealed that COVID-19 infection increases the susceptibility of malignant stomach neoplasm and benign small intestine neoplasm, while malignant colorectal cancer and biliary tract neoplasm increase COVID-19 susceptibility. Additionally, 32 ICs were linked to 12 DSCs and 13 ICs associated with all COVID-19 outcomes.

Conclusions

There is a bidirectional causal relationship between COVID-19 and DSCs. ICs are closely associated with both COVID-19 and DSCs, but their mediating role has been overstated in previous studies.

Supplementary Information

The online version contains supplementary material available at 10.1007/s12672-025-03740-4.

Introduction

Gastrointestinal cancer accounts for 8% of global cancer cases, resulting in approximately 40,000 new incidences and sadly contributing to the premature demise of around 260,000 individuals, constituting 3% of cancer-related fatalities [1]. This burden on global health is substantial. Previous investigations have identified various risk factors for gastrointestinal cancer, including obesity, smoking, alcohol consumption, and hepatitis B virus infection [2]. Simultaneously, the COVID-19 continues its global proliferation, resulting in numerous fatalities. As of 7 September 2025, data provided by the World Health Organization (WHO) indicates a staggering 778,612,822 COVID-19 cases and 7,101,631 confirmed deaths. Despite significant reductions in incidence attributed to advancements in prevention, therapeutics, and empirically targeted treatments, morbidity and mortality rates associated with COVID-19 remain alarmingly high [3].

Gastrointestinal manifestations of COVID-19 encompass an array of symptoms, including anorexia, nausea, vomiting, diarrhea, abdominal pain, and hepatic impairment. Notably, the triad of symptoms most commonly encountered in COVID-19 patients consists of nausea, anorexia (loss of appetite), diarrhea, and vomiting [4, 5]. Additionally, several reports have documented a subset of patients presenting solely with diarrhea and vomiting, devoid of fever or cough [6].

Patients with digestive system damage face an elevated likelihood of progressing to severe or critical illness [7]. Moreover, COVID-19 has emerged as a contributing factor to hepatic deterioration in individuals previously diagnosed with chronic liver conditions, thereby augmenting the risk of severe illness [8].

Furthermore, inflammatory cytokines (ICs) play a critical role in both the progression of COVID-19 and tumor development. Studies have shown that following COVID-19 infection, IC are released [5, 9], with their expression levels remaining elevated above baseline even up to 8 months after the resolution of physical symptoms [9]. On the other hand, the prolonged release of IC triggered by inflammatory responses may contribute to tumor progression and has even been implicated as a potential origin of tumors [10]. Additionally, research indicates that tumor-induced changes in the microenvironment can lead to chronic IC release [11]. However, whether chronic aberrations in ICs increase susceptibility to or exacerbate the severity of COVID-19 or DSCs remains unclear. Thus, further investigation is urgently needed to elucidate the causal relationship between COVID-19, IC, and DSCs.

Mendelian Randomization (MR) is an epidemiological method that enhances causal inference by utilizing genetic variations as instrumental variables (IVs) for exposure. This approach confers two notable advantages: it minimizes confounding effects, as genetic variations are randomly allocated during conception and are unrelated to environmental or self-selection factors, and it mitigates reverse causality since the onset and progression of diseases cannot influence one’s genetic lineage. However, no investigations have explored the association between COVID-19 and DSCs. In this study, we conducted an MR investigation to explore the causal associations (Fig. 1).

Fig. 1.

Overall design and assumptions

Methods

Statistical analysis

The two-sample MR method was used to evaluate the relationship, following the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines, and Inverse-variance weighted (IVW) was chosen as the primary analysis method, using the Wald ratio when SNP < 2, and significance was determined by a p-value < 0.05. Single Nucleotide Polymorphisms (SNPs) are determined as IVs if they satisfy the following three assumptions: (A) They exhibited significant associations with the exposure. (B) They did not show any associations with known confounding factors. (C) The SNPs exerted their effects on outcome exclusively through the IVs of exposure, as is shown in Fig. 1. Finally, two-step and MR analyses were used to identify the mediation effect of ICs.

Data acquisition

We obtained Genome-Wide Association Study (GWAS) data related to COVID-19 from the COVID-19 Host Genetics Initiative, which included a sample size of 1,683,768 for COVID-19 infection, 1,887,658 for COVID-19 hospitalization, and 1,388,342 for COVID-19 severe disease.

GWAS data for DSCs were obtained from the Finn Gen database, with sample sizes as follows: 174,238 for esophageal cancer, 174,639 for stomach cancer, 181,173 for benign stomach conditions, 174,310 for liver intrahepatic bile ducts, 180,839 for benign liver bile, 180,828 for benign liver conditions, 174,115 for biliary tract cancer, 180,829 for benign pancreas conditions, 174,611 for pancreatic cancer, 180,959 for benign small intestine conditions, 174,258 for small intestine cancer, and 177,028 for colorectal cancer.

The ICs GWAS was obtained from the Cytokines GWAS results published by university of Bristol [3, 12]. This GWAS report includes meta-analysis summary statistics for 41 IC and did not adjust for BMI as a covariate in the model, with a total of 8,293 participants. All relevant data used in this study are publicly available. Ethical approval was granted by the respective ethics review boards. Participants in these databases provided informed consent, obviating the need for further ethical scrutiny in this study.

Selection of genetic instruments variables

SNPs meeting the following two criteria: (A) GWAS-associated P value < 1*10^{− 5} [13]. (B) linkage disequilibrium (LD) r² < 0.001, and clumping distance equal to 10,000 kb [14], were considered can minimize the risk of bias from high LD values. Besides, the F statistics were calculated to evaluate the strength of the associations, and SNPs with F statistic values >10 were considered to be independently associated with exposure [15, 16].

Additionally, we focused solely on SNPs found within the GWAS results for the MR analysis, as the impact of SNPs not encompassed by the outcome on the results was deemed negligible [14].

MR analysis

TwoSampleMR package were used for MR using five complementary methodologies: IVW, MR Egger, weighted median, simple mode, and weighted mode. These diverse approaches enabled us to explore distinct assumptions regarding horizontal pleiotropic effects, all within the framework of a random effects model. The primary outcome was guided by IVW, and a significance level of P < 0.05 indicated a causal relationship between the exposure and the outcome [17]. MR Egger, Weighted median, Simple mode, and Weighted mode served as valuable complements to the IVW method [18] .

Besides, MR-Egger method allowed for the assessment of SNP pleiotropy, where P > 0.05 suggested the absence of pleiotropy, thus affirming the reliability of the MR analysis results. Conversely, P-values below this threshold indicated the presence of SNP pleiotropy, thereby rendering the MR analysis results less reliable [19] .

Furthermore, we employed Cochran’s Q statistic (MR-IVW) and Rucker’s Q statistic (MR Egger) to discern the presence of heterogeneity in our MR analysis. A significance level of p > 0.05 served as an indicator of the absence of heterogeneity [20]. To gauge the sensitivity of each SNP with respect to the outcome, we adopted the leave-one-out method. Moreover, we harnessed both the TwosampleMR and MR-PRESSO packages to detect potential multi-effectivity among the SNPs [21] .

Mediation analysis

Mediation analysis aims to assess the mediating effects through which exposure influences the outcome, thereby helping to explore the underlying mechanisms of the exposure’s impact on the result. In this study, the mediation analysis focuses on examining the mediating role of 41 ICs in the bidirectional causal relationship between COVID-19 and digestive system cancers (DSCs). First, the two-sample MR approach is used to evaluate the causal relationship between COVID-19 and DSCs to obtain beta0. Then, the MR analysis is conducted to assess the causal relationships between COVID-19, DSCs, and the 41 ICs, calculating beta1. Next, MR is applied to identify the independent causal associations between the 41 ICs and COVID-19 and DSCs, yielding beta2. The mediation effect is calculated using a two-step MR approach: mediation effect = beta1 × beta2. The direct effect is computed as (beta0 - mediation effect). The mediation proportion is calculated using the formula: mediation proportion = (mediation effect / beta0) × 100%. Based on the results, identified mediators are categorized into different levels of evidence. A triangular relationship alone indicates causal links between exposure and outcome, between mediator and outcome, and between exposure and mediator. Identified ICs are considered potential mediators in the causal relationship between COVID-19 and DSCs. If identified metabolites, peripheral cells, or cytokines not only exist within a triangular relationship but also show significant mediation effects (beta > 0), they are regarded as strong mediators with compelling evidence.

Results

Relationship between COVID-19 and DSCs

Impact of COVID-19 on DSCs

As shown in Table S7, all SNPs exhibited F-values greater than 10, indicating a strong association between the IVs and COVID-19. A total of 45 SNPs were strongly associated with COVID-19 infection, 51 SNPs with severe COVID-19, and 41 SNPs with COVID-19-related hospitalization. MR results revealed a clear causal relationship between COVID-19 and malignant stomach neoplasm, with estimates as follows: COVID-19 (beta = 0.646, p = 0.002), severe COVID-19 (beta = 0.130, p = 0.017), and COVID-19 hospitalization (beta = 0.130, p = 0.017). Additionally, very severe COVID-19 infection was found to have a causal relationship with benign small intestine neoplasm (beta = 0.148, p = 0.012). No clear causal relationship was observed between any severity of COVID-19 and other DSCs (Table 1, other results are provided in Table S1).

Table 1.

The significant results from MR analysis of COVID-19 and DSCs

Exposure	Outcome	Nsnp	Beta	Se	P-value
COVID-19	Malignant stomach neoplasm	42	0.6458	0.2139	0.0025
severe COVID-19	Malignant stomach neoplasm	50	0.1298	0.0544	0.0171
COVID-19 hospitalization	Malignant stomach neoplasm	40	0.2334	0.1019	0.0220
severe COVID-19	Benign Small intestine neoplasm	50	0.1478	0.0589	0.0121
Colorectal cancer	COVID-19 hospitalization	20	0.0958	0.0286	0.0008
Colorectal cancer	COVID-19	22	0.0374	0.0123	0.0023
Malignant stomach neoplasm	severe COVID-19	18	-0.0439	0.0152	0.0039
Colorectal cancer	severe COVID-19	22	0.1198	0.0456	0.0086
Malignant biliary tract neoplasm	severe COVID-19	13	0.0228	0.0113	0.0436

Note [1]. The beta value represents the strength of the exposure’s effect on the outcome [2]. Standard Error (SE) indicates the precision of the beta estimate, a smaller SE indicates greater precision in the beta estimate [3]. P-value less than 0.05 denotes statistical significance

Impact of DSCs on COVID-19

DSC-related SNPs also demonstrated F-values exceeding 10, with the following associations: 13 SNPs were strongly associated with malignant biliary tract neoplasm, 23 SNPs with colorectal cancer, 11 SNPs with malignant liver and intrahepatic bile duct neoplasms, 18 SNPs with malignant esophageal neoplasm, 16 SNPs with malignant pancreatic neoplasm, 21 SNPs with malignant small intestine neoplasm, 19 SNPs with malignant stomach neoplasm, 13 SNPs with benign liver/bile duct neoplasm, 11 SNPs with benign liver neoplasm, 11 SNPs with benign pancreatic neoplasm, 23 SNPs with benign small intestine neoplasm, and 22 SNPs with benign stomach neoplasm. MR results demonstrated that malignant colorectal cancer increased the risk of COVID-19 infection (beta = 0.037, p = 0.002), severe COVID-19 (beta = 0.120, p = 0.008), and COVID-19-related hospitalization (beta = 0.096, p = 0.001). Malignant biliary tract neoplasm increased the risk of severe COVID-19 (beta = 0.022, p = 0.044). Furthermore, a causal relationship was found between malignant stomach neoplasm and severe COVID-19 (beta = -0.044, p = 0.004) (Table 1, other results are provided in Table S2).

Impact of ICs on DSCs and COVID-19

MR of ICs with COVID and DSCs was performed using two-sample IVW analysis as the primary method to determine the causal relationship between ICs and DSCs and COVID-19. (The complete results are presented in Table S3-6)

Impact of ICs on DSCs

In the 492 MR analyses of 41 serum ICs and DSCs, 69 IVW results showed significant causal associations, including 32 ICs levels associated with 12 DSCs in GWAS (p < 0.05). ICs with significant results were selected for further mediation analysis (Table S5).

Impact of ICs on COVID-19

In the 123 MR analyses of 41 serum ICs and COVID-19, 16 IVW results showed significant causal associations, including 13 ICs levels associated with 3 COVID-19 GWAS results (p < 0.05) (Table S6).

Mediation analysis

Mediation effects of ICs in COVID-19 on DSCs

As all COVID-19 GWAS demonstrated causal associations with malignant stomach neoplasm, and severe COVID-19 infection was causally associated with benign small intestine neoplasm, the relevant analysis results were further extracted for mediation analysis.

As shown in Table S5, ICs causally associated with malignant stomach neoplasm include FGFB (beta = -0.402, p = 0.0004), IL6 (beta = -0.618, p = 0.001), VEGF (beta = 0.096, p = 0.003), SCGFb (beta = 0.226, p = 0.007), bNGF (beta = -0.140, p = 0.031), IP10 (beta = 0.241, p = 0.036), and IL1ra (beta = -0.281, p = 0.039).

The results show that in the MR analysis between COVID-19 and ICs, COVID-19 infection is negatively causally associated with VEGF (beta = -0.138, p = 0.005). No causal association was observed between other COVID-19 GWAS and ICs related to DSCs. We used a two-step method to calculate the mediation effect. The total effect of COVID-19 infection on malignant stomach neoplasm was 0.646, with a mediation effect of VEGF of -0.013. The direct effect of COVID-19 infection on malignant stomach neoplasm was 0.659.

The ICs causally associated with benign small intestine neoplasm include IL1ra (beta = -0.632, p < 0.001), IL6 (beta = -0.687, p = 0.006), and bNGF (beta = 0.536, p = 0.017). However, the MR results for COVID-19 and IL1ra, IL6, and bNGF did not show a causal relationship, indicating that these ICs are not mediators of the increased risk of benign small intestine neoplasm in severe COVID-19 infection.

Mediation effects of ICs in COVID-19 on DSCs

Malignant colorectal cancer increased the risk of COVID-19 infections, so 13 ICs associated with COVID-19 was extracted for further analysis. However, the results indicated that malignant colorectal cancer was not causally associated with these 13 ICs (Table S4), suggesting that the increased risk of COVID-19 due to malignant colorectal cancer is not mediated by these ICs.

Malignant biliary tract neoplasm increased the risk of severe COVID-19, and mediation analysis revealed a causal association between malignant biliary tract neoplasm and the CTACK (beta = 0.0184, p = 0.024). The total effect of malignant biliary tract neoplasm on severe COVID-19 was 0.023, with a mediation effect through CTACK of -0.001, and the direct effect of malignant biliary tract neoplasm on severe COVID-19 was 0.024.

Additionally, the mediation analysis for the causal relationship between malignant stomach neoplasm and severe COVID-19 showed no involvement of ICs in the progress.

Discussion

This is the first study to analyze the bidirectional causal relationship between COVID-19 and DSCs and to explore the mediating role of serum ICs. The results indicate that all severities of COVID-19 increase the susceptibility of malignant stomach neoplasm. Besides, severe COVID-19 infection is associated with an increased risk of benign small intestine neoplasm. In the reverse analysis, malignant colorectal cancer was found to increase the susceptibility of all severities of COVID-19, malignant biliary tract neoplasm increased the risk of severe COVID-19, and malignant stomach neoplasm reduced the risk of severe COVID-19.

Additionally, 32 ICs were associated with 12 DSCs, and 13 ICs were associated with all COVID-19 results. Despite the strong associations between ICs, DSCs, and COVID-19, mediation analysis revealed that only VEGF and CTACK involved in the causal relationship between DSCs and COVID-19 as mediators. Specifically, the downregulation of VEGF caused by COVID-19 infection slightly reduced susceptibility to malignant stomach neoplasm, while the downregulation of CTACK caused by malignant biliary tract neoplasm slightly reduced susceptibility to severe COVID-19.

Gastrointestinal distress represents a noteworthy symptom in COVID-19 infections that demands close attention. Research has indicated that individuals typically exhibit clinical symptoms within the initial 3–14 days following COVID-19 infection. These symptoms predominantly include cough, fever, sore throat, myalgia, diarrhea, vomiting, and abdominal pain, even in the absence of radiographic evidence of pneumonia [22, 23]. As the disease advances, approximately one week after the onset, radiological signs of pneumonia become apparent, accompanied by symptoms such as exertional dyspnea, acute cardiac injury, sepsis, and, in severe cases, the emergence of ground-glass opacities that can lead to fatalities [22, 23]. Despite the passage of 6 years, COVID-19 and its variants, including the Delta, Omicron, and other variants, persist in their propagation and cause many after-effects.

Our MR analysis results also underscore the causal relationship between 32 ICs and 12 DSCs, which aligns with findings from numerous studies. Specifically, alterations in the levels of IL-8, IL-10, and TNF-α have been shown to play pivotal roles in the development of stomach cancer [24]. Furthermore, the TNF-α/TNFR1 signaling pathway has been shown to promote the occurrence of stomach tumors [25], while elevated levels of IL-6 are closely associated with an increased risk of colorectal cancer [26]. A literature-based retrospective study has indicated that COVID-19 infection may lead to immune dysfunction, toxic exposure, and an elevated risk of gastrointestinal cancer [27]. Although our study did not observe evidence of COVID-19 increasing ICs at the genetic level, previous research has indicated that COVID-19 infection can induce alterations in the expression levels of ICs in patients’ serum, including IL-1, IL-1β, IL-6, IL-8, IL-19, interferons, and TNF-α [28–30]. In addition, a review has proposed that COVID infection may influence the expression of signaling pathways, including IL-6/ Janus tyrosine Kinase (JAK)/Signal Transducer and Activator of Transcription (STAT), Interferon-I (IFN-I), and Androgen Receptor signaling. These alterations in signaling pathways have the potential to play a regulatory role in the progression of DSCs [31].

Due to the high incidence of DSCs and their tendency to present with mild early symptoms that are challenging to detect, DSCs patients are often diagnosed at an advanced stage. Therefore, it is crucial to perform MR analysis to investigate the causal relationship. Undoubtedly, these efforts could effectively guide individuals at high risk for DSCs to precise diagnostic procedures, including endoscopy, colonoscopy, abdominal ultrasound, and relevant biomarker assessments, and help the early intervention and diagnosis, ultimately leading to improved patient outcomes. The findings of this study indicate that, in comparison to other DSCs, colorectal cancer exhibits a heightened susceptibility to all severities of COVID-19. This suggests that patients with colorectal cancer should prioritize measures to protect themselves from viral infections.,

This study is with its limitations. First, like most MR analyses, it relies on data from the IEU Open GWAS Project. Consequently, data not included in this dataset but published elsewhere were not considered, which may pose potential challenges. Additionally, MR analyses are sometimes affected by weak instrument effects, which can lead to false-positive results. In our study, we mitigated this risk by applying stringent criteria for SNP selection and by removing potential confounders using the PhenoScanner database, aiming to minimize the impact of weak instruments.Furthermore, MR does not account for interactions between genetic variations and environmental factors, which could lead to incomplete causal estimates. Factors such as education, healthcare, and socioeconomic status might influence the observed relationships but were not considered in our analysis. Additionally, the GWAS data used in this analysis were primarily sourced from European populations. Although data on DSCs from Asian populations are available, there is a lack of relevant GWAS data on COVID-19 for Asian populations, which led to their exclusion from this study. Therefore, the generalizability of the results remains uncertain. To confirm the robustness of our conclusions, large-scale, multi-center, and long-term follow-up studies are necessary to validate the findings of this MR analysis.

Conclusion

This work revealed that COVID-19 infection increases the susceptibility of malignant stomach neoplasm and benign small intestine neoplasm, while malignant colorectal cancer and biliary tract neoplasm increase COVID-19 susceptibility. Additionally, 32 ICs were linked to 12 DSCs and 13 ICs associated with all COVID-19 outcomes.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Liang Y and Deng B conceived, designed, and planned the study. Liang Y and Deng B acquired and analyzed the data. Xiao W and Deng B interpreted the results. Liang Y and Chen Q drafted the manuscript，Chen Q and Deng B contributed to the critical revision of the manuscript. All authors read and approved the final manuscript.

Funding

This work was supported by the National Natural Science Foundation of China (Grant No. 82574805), the Noncommunicable Chronic Diseases - National Science and Technology Major Project (No. 2023ZD0509400), and the Major Special Project of Scientific Research of Sichuan Provincial Administration of Traditional Chinese Medicine (Grant No. 25ZDAZX005).

Data availability

All the data generated or analyzed during this study are included in this published article and its supplementary information files.

Declarations

Ethics approval and consent to participate

Review and/or approval by an ethics committee was not needed for this study because our research is based on public databases.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.