中国P站

Sepsis; Inflammatory proteins; TRAIL; Susceptibility

AKI: Acute Kidney Injury; CCL19: C-C motif Chemokine Ligand 19; CCL28: C-C motif Chemokine Ligand 28; CCLs: C-C Chemokine Ligands; CCR7: C-C Chemokine Receptor type 7; GWAS: GenomeWide Association Studies; IVW: Inverse-Variance Weighted; IVs: Instrumental Variables; LDA: Linkage Disequilibrium Analysis; MR: Mendelian Randomization; MR-PRESSO: MR Pleiotropy Residual Sum and Outlier; STROBE-MR: Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomization; TRAIL: TNF-Related Apoptosis-Inducing Ligand; VEGF-A: Vascular Endothelial Growth Factor A

Sepsis is a fatal condition involving organ dysfunction that develops following an imbalanced response of the host to various infections. Globally, approximately 19 million sepsis cases occur annually [1].

Most affected patients show long-term consequences of sepsis and approximately 6 million patients die due to sepsis. Additionally, around 3 million individuals who survive after sepsis encounter various complications, including cognitive dysfunction and immunosuppression. Early identification and treatment are essential for improved outcomes of sepsis patients. However, the current diagnostic methods, such as blood culture, are time-consuming and may yield poor results. Presently, specific drugs for sepsis are unavailable, and the treatment includes antibiotic therapy, organ protection and fluid resuscitation. The recognition of reliable early detection indicators and specific therapeutic targets is therefore critical.

Sepsis has a high prevalence among the elderly population; moreover, because of the aging global population, a significant economic burden is imposed on healthcare systems and society.

Hence, a critical requirement is to develop effective, simple and affordable approaches to prevent and manage sepsis. Additionally, a downweighing analysis was conducted to address data complexity. Sepsis is a complex pathological process involving the interaction of anti-inflammatory and pro-inflammatory mechanisms. Therefore, the modulation of different anti-inflammatory and pro-inflammatory proteins is particularly important for treating sepsis [2].

Mendelian Randomization (MR) is a reliable and convenient tool for determining causal associations of exposures and outcomes through data based on Genome-Wide Association Studies (GWAS). MR effectively controls for confounding factors and has a reasonable causal time sequence. It can effectively provide supplementary causal relationships for observational research data. Additionally, the use of open accessible and anonymous GWAS data eliminates ethical concerns.

In this study, we applied a two-sample MR approach to analyze 91 inflammatory proteins in batches for identifying protective and risk proteins for treating sepsis. Previous MR studies have focused on the causal implications of biomarkers related to inflammation in susceptibility to infectious diseases [3].

However, a comprehensive evaluation is required to determine whether circulating inflammatory proteins levels causally contribute to predisposition to sepsis. This study aimed to offer new perspectives regarding the vital role of circulating inflammatory protein mediators in sepsis pathogenesis.

Study design

Figure 1 presents the overall study design of the MR analysis. The GWAS data of 91 inflammatory proteins were derived from a pooled dataset of 14,824 individuals. Subsequently, we extracted sepsisrelated GWAS data from the aggregated data of European descent populations. The primary analytical approach was the InverseVariance Weighted (IVW) method, while MR-Egger, weighted median and weighted mode were utilized as supplementary tools. Sensitivity analyses were carried out to confirm whether the study results were robust. Additionally, reverse causation was prevented by the Steiger test. Table 1 lists the aggregated summary statistics data for sepsis:

• Sepsis patients.
• Sepsis patients in critical care.
• Sepsis patients under 75 years of age.
• Sepsis patients who died within 28 days.
• Sepsis patients who died within 28 days in critical care [4].

Figure 1: Information on the research flowchart for this study.

Table 1: Information of sepsis and related outcomes.

This MR analysis was guided by three main assumptions:

• Strong association between the genetic instruments and exposures.
• Absence of relationship between the genetic instruments and confounders.
• No direct association of the genetic instruments with outcomes (Figure 2) [5].

Figure 2: Information on the three major assumptions of mendelian randomization illustrated in a diagram.

GWAS data related to inflammatory proteins

The data on cytokines were obtained from a genetic investigation. The latest GWAS data on circulating inflammatory proteins related to inflammation included 91 cytokines. This extensive dataset provides further opportunities to investigate the relationship between sepsis and the inflammatory factors. To obtain a broad cytokine range, we utilized the GWAS data on 91 inflammatory cytokines from 11 cohorts containing 14,824 European descent individuals. The original study provides more information regarding aggregation of the GWAS data. For the main analysis, p<5 × 10-8 was considered the threshold for separating circulating inflammatory proteins with independent SNPs. We then conducted a Linkage Disequilibrium Analysis (LDA) to ensure that the SNPs were independent according to the following criteria: r2<0.001 and 10,000 base pairs (kb) [6].

GWAS data related to sepsis

The UK Biobank is a major health resource and a large and detailed biomedical database for genetic, lifestyle and health-related information and contains biological samples from more than 500,000 UK participants. It is the most widely used dataset with in-depth genetic data. In the present study, we selected four sets of GWAS data for sepsis, sepsis (under 75 years of age), sepsis (critical care), sepsis (28-day death) and sepsis (28-day death in critical care). We utilized the databases from the IEU Open GWAS (https://gwas.mrcieu.ac.uk/), which sourced summary-level data from the UK Biobank.

Statistical analysis

A two-sample MR approach was used by leveraging Instrumental Variables (IVs) derived from two separate GWAS summary results to improve the study’s statistical power. Furthermore, our study adhered to the STROBE-MR (Strengthening the Reporting of Observational Studies in Epidemiology using Mendelian Randomization) guidelines to ensure high-quality reporting.

Genetic variants associated with Parkinson’s disease were selected based on rigorous criteria for reliability (p<5 × 10-8) and independence (Linkage Disequilibrium (LD): r2<0.001 within a 10000-kb window). We then harmonized the data for exposures and outcomes. Additionally, any palindromic alleles were excluded, as they can lead to ambiguous interpretations. For identifying weak instruments, the variance explained by each instrument was computed and the Fstatistic was estimated as F=(beta2 /se2 ).

A comprehensive set of four analytical methods was utilized to elucidate the causal relationship between exposures and outcomes, including IVW, MR-Egger, weighted median, and weighted mode. By utilizing multiple approaches, we derived robust and reliable estimates of causal effects while ensuring that our findings are valid. Among these methods, the IVW method was adopted as the primary approach. The remaining methods, which differed from the IVW method and relied on distinct assumptions, served as alternative approaches. These methods exhibited relative robustness against horizontal pleiotropy and complemented the IVW method to enhance the reliability of MR estimates [7].

In the IVW method, the causal relationship of exposures with outcomes was analyzed using the fixed-effect model or the randomeffect model. If the results exhibited heterogeneity, the random-effects model was primarily used, with particular emphasis on the outcomes of the weighted median method.

Heterogeneity was evaluated with Cochran’s Q test. Directional pleiotropy was determined with the MR-Egger intercept test. Furthermore, we utilized the MR Pleiotropy Residual Sum and Outlier (MR-PRESSO) test for recognizing potential outliers that could introduce horizontal pleiotropy. After removing these outliers, corrected MR estimates were obtained, which improved the robustness and reliability of the results. Furthermore, the leave-one-out analysis was carried out to clarify whether individual SNPs influenced the MR estimates. The Steiger test was also utilized for mitigating the influence of reverse causality. For significant results (p<0.05) obtained with the IVW method, we utilized the LDtrait online tool (https:// ldlink.nih.gov/?tab=ldtrait) to investigate the associated SNPs. The MRPRESSO (version 1.0) package and the TwoSampleMR (version 0.5.6) package in R software (version 4.2.1) were utilized for analysis [8].

After a stringent threshold of 5 × 10⁸ was implemented during SNP selection, all IVs utilized in our study constituted strong IVs. This indicated that our MR analysis outcomes were not prone to weakinstrument bias effects. Based on our IVW analysis, we observed that 10 inflammatory proteins were causally related to sepsis and related outcomes, with all associations showing statistical significance at p<0.05. Figure 3 depicts the corresponding forest plot. Our sensitivity analyses indicated that both heterogeneity and pleiotropy were not detected. Table 2 shows more details regarding these analyses. The stability and reliability of our findings were also supported by the leave-one-out analysis. There were no outliers in our analysis based on the MR-PRESSO test. The MR-PRESSO test identified abnormal data in Vascular endothelial growth factor A levels (P=0.0495) in the sepsis group, C-C motif chemokine 19 levels (P=0.0360) and C-C motif chemokine 28 levels (P=0.0245) in the sepsis (28 day death in critical care) group, C-C motif chemokine 19 levels (P=0.0137) and cystatin D levels (P=0.0317) in the sepsis (28 day death) group, TNF-related apoptosis-inducing ligand levels (P=0.0475) in the sepsis (critical care) group, and TNF-related apoptosis-inducing ligand levels (P=0.0107) in the sepsis (under 75) group [9].

Figure 3: Information on the details of a forest plot.

However, the MR-Egger intercept test did not find evidence of pleiotropy. Additionally, because of the insufficient number of SNPs associated with some of the exposures, we concluded that the positive results for those exposures were irrelevant; hence, they were excluded from the analysis.

Table 2: Information of heterogeneity and pleiotropy analyses.

After conducting a series of sensitivity analyses, we ultimately identified 5 inflammatory proteins that exhibited causal relationships with sepsis and related outcomes.

The two-sample MR analysis with summary data from a European population provided evidence that supported the causal effects of the levels of various circulating inflammatory proteins, including TNFRelated Apoptosis-Inducing Ligand (TRAIL), Vascular Endothelial Growth Factor A (VEGF-A), C-C motif Chemokine Ligand 19 (CCL19), C-C motif Chemokine Ligand 28 (CCL28) and cystatin D, on sepsis risk and related outcomes. The study design employed genetic instruments to infer causality, thereby bypassing confounding factors in the observational data. Apart from cystatin D levels, all other inflammatory proteins were consistently identified as risk factors.

The two-sample MR analysis with summary data from a European population provided evidence that supported the causal effects of the levels of various circulating inflammatory proteins, including TNFRelated Apoptosis-Inducing Ligand (TRAIL), Vascular Endothelial Growth Factor A (VEGF-A), C-C motif Chemokine Ligand 19 (CCL19), C-C motif Chemokine Ligand 28 (CCL28) and cystatin D, on sepsis risk and related outcomes. The study design employed genetic instruments to infer causality, thereby bypassing confounding factors in the observational data. Apart from cystatin D levels, all other inflammatory proteins were consistently identified as risk factors.

TRAIL

TRAIL, a TNF ligand superfamily member, is a type II transmembrane protein and contains an extracellular carboxy-terminal domain. A prospective observational study revealed an association of plasma TRAIL levels with the poor prognosis of sepsis patients; furthermore, increased plasma TRAIL levels were also observed during the recovery process. The association between plasma TRAIL levels and necroptosis in sepsis was also investigated in a multicenter study; however, an inconsistent correlation with mortality was observed. Additionally, in a separate observational study, sepsis patients exhibited higher TRAIL levels than the control group. Nonetheless, the study revealed that plasma soluble TRAIL levels were markedly higher in individuals surviving for 28 days than in dead patients. However, in a retrospective study, the MeMed BV index comprising TRAIL did not demonstrate good diagnostic and monitoring value for infection in ICU patients. Unlike clinical research, studies in mouse models revealed that neutrophils are sensitive to apoptosis induced by TRAIL stimulation during the early stages of abdominal sepsis. Another animal experiment study yielded similar results, demonstrating that TRAIL can alleviate multiorgan damage caused by sepsis by inducing apoptosis of infiltrating neutrophils within the tissues. In our MR analysis, TRAIL was recognized as a risk factor for sepsis development as well as sepsisrelated outcomes, and a causal relationship was observed with sepsis (28-day death), increasing the mortality risk within 28 days for sepsis patients. We should therefore correctly recognize the distinction between randomized controlled trials and MR studies and carefully interpret their results. The complete explanation of TRAIL’s impact on prognosis in sepsis necessitates further investigation, as current evidence may not provide a complete picture.

VEGF

VEGF is a crucial protein involved in angiogenesis, a process in which new blood vessels are formed from the pre-existing ones. Regarding sepsis, VEGF plays a significant role in vascular permeability, i.e., the ease with which substances can cross the walls of blood vessels. An elevated level of vascular permeability is a key pathological mechanism in sepsis and VEGF can enhance this process. Furthermore, in animal model experiments, VEGF-A blockade reduced the mortality rate of mice with sepsis. In a prospective study, Shu-Min Lin et al., found a significant association of the sustained decrease in angiotensin-1 levels with the development of multi-organ dysfunction syndrome; subsequently, a direct correlation was observed with the increased mortality rate of sepsis patients. A subsequent prospective study reported that patients with sepsis exhibited elevated VEGF levels; however, VEGF levels did not differ significantly between patients with severe sepsis and those with non-facultative sepsis without organ dysfunction. Additionally, several observational studies exhibited a correlation of VEGF levels with sepsis prognosis, thus suggesting that VEGF could function as a potential biomarker for predicting disease severity and outcome. A meta-analysis reinforced the predictive value of VEGF, thus indicating a strong correlation with a higher mortality risk in sepsis. The non-survival sepsis group exhibited significantly higher VEGF levels than survivors and patients in the severe sepsis subset had even elevated VEGF concentrations as compared to those in the less severe subset. These findings further underscore the association between VEGF and adverse outcomes in sepsis. In this MR study, a causal relationship of VEGF with sepsis was detected. However, for certain outcomes, such as the 28-day mortality group, a significant effect was not observed. Despite the ability of the MR analysis to minimize confounding effects, it cannot fully replace a randomized controlled trial. Consequently, our interpretation of the findings should be cautious, and additional experiments or observations may be necessary to confirm the clinical significance of this association.

C-C chemokine ligands

C-C Chemokine Ligands (CCLs), also known as β-chemokines, comprise 28 chemokines with the N-terminal CC domain and designated CCL1-28. The symbol’s digits depend on the order in which they are discovered. However, the actual number of C-C chemokines is 27, because CCL9 and CCL10 are considered equivalent. These chemokines play critical roles in the immune system, particularly in eosinophils, CD4+ and CD8+ T cells, dendritic cells, monocytes, macrophages and natural killer cells. Our investigation revealed a causal relationship of CCL19 with sepsis (28-day mortality) and a causal relationship of CCL19 and CCL28 with sepsis (28-day mortality in critical care). Therefore, we believe that CCL19 and CCL28 may be potential biomarkers for predicting sepsis-related mortality within 28 days. Further studies are required to confirm this. CCL19, also known as B7-H3 or lymphotactin, is a chemokine (C-X-C motif ligand) that belongs to the CCL family. It is primarily produced by immune cells, particularly dendritic cells and it is critically involved in immune cell trafficking. Functionally, CCL19 is a T cell chemoattractant, particularly for memory T cells and CD8+ T cells. It serves as a “homing signal” for T cells during immune responses, guiding them toward lymph nodes and other lymphoid tissues. The binding of CCL19 to its receptor, C-C Chemokine Receptor type 7 (CCR7), is essential for T cell accumulation in these sites. This process is a critical part of the immune system’s surveillance and response to infections and tissue injury. In sepsis, CCL19 has a crucial function in the complex immune response to infections. Initially, immune cells, for example, monocytes and neutrophils, release CCL19 locally to recruit CD8+ T cells and memory T cells to the infection site. During the chronic phase of sepsis, increased CCL19 expression can result from persistent inflammation and immune activation. This increased production of CCL19 attracts more immune cells, potentially exacerbating the “immunological storm” where excessive immune activation can induce tissue damage and organ failure. CCL19 also influences the homing and differentiation of T cells, thereby affecting the formation of immune memory and the subsequent immune tolerance. According to observational studies, sepsis patients show remarkably higher CCL19 levels than those with non-sepsis infections. In our MR analysis, CCL19 emerged as a crucial factor contributing to both sepsis (28-day mortality) and severe sepsis (28-day mortality in critical care) and its presence in the bloodstream indicated a causal association with increased risk of death within 28 days in sepsis patients. Therefore, targeting CCL19, particularly through its receptor CCR7, has emerged as a potential therapeutic strategy in sepsis. By modulating immune cell recruitment and function, these interventions aim to reduce inflammation and prevent further organ damage. However, more research is required to fully understand the complex interplay between CCL19 and sepsis and to develop effective clinical treatments. CCL28, a mucosae-associated epithelial chemokine, is an essential chemokine for the normal functioning of the mucosal immune pathway. It activates the CCR3 and CCR10 receptors. CCL28 shows constitutive expression in multiple tissues and the expression of CCL28 is induced by inflammation and infection. CCL28 expression is observed in intestinal columnar epithelial cells, mammary glands, lungs and salivary glands. This chemokine facilitates the homing of CCR10-expressing T and B lymphocytes to mucosal tissues; it also promotes the migration of CCR3-expressing eosinophils. Although CCL28 exhibits constitutive expression in the colon, its levels are increased by certain bacterial products and pro-inflammatory cytokines, thus indicating its involvement in recruiting effector cells at the epithelial injury site. Apart from its constitutive expression, CCL28 is involved in IgA-expressing cell migration to the salivary glands, intestine, mammary glands, and other mucosal tissues. To date, there has been no direct studies on the relationship between CCL28 and sepsis. CCL28 functions as both an antimicrobial agent and a modulator of the immune system. Its elevated presence in epithelial and mucosal surfaces can provide a non-specific, innate defense against multiple bacterial pathogens by enhancing their antimicrobial activity. CCL28 is significantly instrumental in driving the maturation, trafficking and activation of immune cells, particularly T helper cells, which form a vital component of the immune response. Previous research has suggested that under immune stress conditions, CCR3 tends to interact more frequently with CCL28. These findings provide indirect support for our results, suggesting that the elevated levels of CCL28 predicted by the gene may increase the risk of sepsis.

Cystatin D (CST5)

Cystatin D is a member of the cystatin family in humans, distinct from others, with highly limited tissue distribution. It is primarily found in exocrine secretions, specifically in saliva and tear glands. The primary physiological function of CCL19 is to facilitate T cell homing to lymph nodes. This process relies on the presence of CCR7 receptors on these chemokines. Cystatin, a type of immune-regulating protein involved in inflammation, shows a close association with sepsis, particularly in patients with Acute Kidney Injury (AKI) where robust immune responses occur. Elevated levels of cystatin C in sepsis patients are often attributed to increased production by the kidneys due to inflammation or its role as an inflammatory mediator. Cystatin, a protein involved in inflammation regulation, is closely related to sepsis development, particularly in patients with AKI who exhibit intense immune responses. The elevated cystatin C levels in these patients are commonly attributed to increased kidney production due to inflammation or its function as an inflammatory mediator. However, because of limited animal research and clinical observational data, further investigations on cystatin D are warranted. In the present MR analysis, we revealed a protective role for cystatin D. Suppression of the circulating levels of cystatin D directly decreased the 28-day mortality rate in sepsis patients, thus showing a causal relationship of cystatin D with sepsis development with a 28-day mortality endpoint. More research is warranted to investigate the mechanisms related to the protective role of cystatin D in sepsis.

Strengths and Limitations

This study had the following major strengths: 1) Large sample size, which enhanced statistical robustness using multiple MR techniques and 2) The optimization of interpretation by leveraging the GWAS data. We also conducted sensitivity analyses that reinforced the credibility of the study findings. Moreover, MR studies can effectively control for confounding factors, thereby allowing identification of a causal relationship of exposures with outcomes.

Our study has certain limitations. Because of heterogeneity of sepsis, driven by varying sources of infection, host genetics and comorbidities, we could not investigate subtype-specific effects. Furthermore, genetic variations in circulating inflammatory proteins and sepsis GWAS datasets, if present, could introduce bias. To mitigate this, we conducted our analysis by excluding non-European populations. However, because of this limitation, our findings could not be generalized to other racial groups. Additionally, because of limitations related to genomics, we could not directly observe the dynamic changes in circulating inflammatory proteins in sepsis. Continuous monitoring of these proteins could provide deeper insights into their kinetic profile and response to treatment with sepsis progression. This will enhance our understanding of their role in the disease process. Moreover, although genetic research can strongly infer causality, it also carries inherent limitations. We utilized various analytical approaches to minimize the impact of pleiotropy. Lastly, our study did not deeply analyze the potential downstream mechanisms to elucidate the connections between the identified cytokines and the pathophysiology of sepsis.

This MR study indicated a genetic predictive causality between the circulating levels of TRAIL, VEGFA, CCL19, CCL28 and cystatin D and the changes in sepsis risk. These discoveries clarified the potential involvement of these inflammatory proteins in the pathogenesis of sepsis. To validate the observed associations and reveal the underlying biological mechanisms, further experimental research is required. This would involve examining how these markers interact with sepsis development and assessing their specific roles in the process. Additionally, clinical trials in diverse patient populations are needed for confirming the predictive value of these inflammation markers for sepsis outcomes. Despite the requirement for further validation, these preliminary results offer new insights into the inflammatory milieu in sepsis and provide a roadmap for future research.

1. Cecconi M, Evans L, Levy M, Rhodes A (2018) . Lancet 392: 75-87.

   [] [] []

2. Perner A, Cecconi M, Cronhjort M, Darmon M, Jakob SM, et al. (2018) . Intensive Care Med 44: 791-798.

   [] [] []

3. Reinhart K, Daniels R, Kissoon N, Machado FR, Schachter RD, et al. (2017) . N Engl J Med 377: 414-417.

   [] [] []

4. Prescott HC, Angus DC (2018) . JAMA 319: 91.

   [] [] []

5. Zhao Z, Wang Y, Kang Y, Wu G, He J, et al. (2024) . Heliyon 10: e27523.

   [] [] []

6. Saito N, Tsuchiya J, Itoga M, Okamura Y, Tsuyama H, et al. (2024) . Infect Drug Resist 17: 207-219.

   [] [] []

7. Cheng MP, Stenstrom R, Paquette K, Stabler SN, Akhter M, et al. (2019) . Ann Intern Med 171: 547-554.

   [] [] []

8. Perner A, Gordon AC, de Backer D, Dimopoulos G, Russell JA, et al. (2016) . Intensive Care Med 42: 1958-1969.

   [] [] []

9. Gotts JE, Matthay MA (2016) . BMJ 353: 1585.

   [] [] []