Skip to main content
Download PDF

Causal influence of immune factors on the risk of diabetic retinopathy: a mendelian randomization study

Diabetology & Metabolic Syndrome volume 16, Article number: 194 (2024) Cite this article

Abstract

Objectives

Diabetic retinopathy (DR) is a prevalent microvascular complication in diabetic patients. Various mechanisms have been implicated in the pathogenesis of DR. Previous studies have observed the relationship between immune factors and DR, but the causal relationship has not been determined.

Methods

We conducted a two-sample Mendelian randomization (MR) analysis of 731 immune cells and DR, using publicly available genome-wide association study (GWAS) summary statistics, to evaluate potential causal relationships between them. Four types of immune traits were included in the analysis through flow cytometry. GWAS statistics for DR were obtained from the Finngen database, which performed GWAS on 190,594 European individuals (Ncase = 14,584, Ncontrol = 176,010) to assess genetically predicted DR. The primary method used to perform causality analysis was inverse variance weighting (IVW).

Results

Following false discovery rate (FDR) correction, 11MFI-DR, 5AC-DR, 5RC-DR, and 1MP-DR reached a significant causal association level (PFDR < 0.05). Notably, all AC traits exhibited potential associations with a decreased risk of DR(OR < 1), while a majority of MFI traits, along with the singular MP trait, exhibited potential associations with an increased risk of DR (OR > 1). The highest proportion of T-cell subsets in the final results.

Conclusion

This study elucidates that the progression of DR is intricately influenced by immune responses, thereby confirming the immunological susceptibility of DR. Our findings may offer new targets for diagnosing and treating DR, as well as aid in developing therapeutic strategies from an immunological standpoint.

Introduction

Diabetic retinopathy (DR) is a condition resulting from prolonged damage to the retina caused by chronic hyperglycemia. The initial sign is the breakdown of the blood-retinal barrier (BRB), primarily due to vascular endothelial growth factor (VEGF) [1]. Current research indicates that the intricate mechanisms underlying DR are closely associated with oxidative stress, activation of protein kinase C, elevated levels of advanced glycation end-products [2], and inflammatory injury [3], among various factors. Pertinent risk factors encompass prolonged diabetic duration, pregnancy, suboptimal glycemic and blood pressure control [4]. Studies propose that the global burden of DR is projected to remain substantial until 2045 [5]. Nevertheless, to date, satisfactory therapeutic strategies for DR have yet to be established.

In a healthy state, the retina is immunoprivileged to prevent damage from immune stimuli. The BRB is important to limit the entry of immune cells into the eye. However, chronic inflammation-induced BRB destruction can threaten the immune privilege state, leading to retinal destruction [6]. Immunodysregulation is acknowledged as a pivotal factor in the pathophysiology of DR [7]. A growing body of evidence suggests that the activation of the immune system plays a crucial role in the progression of DR. Microglial cells, resident immune cells in the retina, assume a significant role in the development of diabetic retinal lesions. When exposed to elevated glucose levels, these microglial cells undergo activation and can mediate inflammatory responses leading to visual deterioration [8]. It has been suggested that T cells may promote inflammatory damage in the DR vasculature by releasing cytokines and cytotoxic factors, resulting in retinal inflammation, angiogenesis, and vascular leakage [9]. A study discovered increased levels of B-cell-generated antibodies in the vitreous fluid of type 2 diabetic patients with DR, indicating that a B-cell-mediated immune response could be a factor in the advancement of the disease [10]. Overall, the involvement of the immune system in the inflammatory and angiogenic processes of DR has been postulated, yet a definitive causal impact of immune cells on the risk of DR remains inconclusive [11]. Activation of these immune cells and their role in DR pathology will open up new possibilities for clinical applications and treatment strategies.

Mendelian Randomization (MR) studies serve as effective tools in mitigating biases arising from confounding factors and reverse causation [12]. Leveraging naturally occurring genetic variations as Instrumental Variables (IVs), MR enables a robust assessment of causal relationships between exposures and outcomes [13]. In this study, immune cells were chosen as the exposure variable, and DR as the outcome variable, to explore the causal relationship between immune cells and the occurrence and progression of DR. Not only did the study yield results similar to existing research analyses, but it also uncovered some previously unexplored immune factors that may be causally related to the onset of DR. This contributes novel perspectives for the diagnosis and treatment of DR in the future. It is important to note that this analysis is based on genetic data, which can provide evidence for causal inference. It is designed to generate causal hypotheses rather than confirm causal relationships.

Materials and methods

Study design

Based on aggregated summary-level data from a large-scale genome-wide association study (GWAS), we employed a two-sample Mendelian randomization approach to assess the causal relationship between a myriad of immune cells (731 immune cells within 7 immune panels) and the risk of DR. To enhance the precision of our results, single nucleotide polymorphisms (SNPs) were selected as IVs. Simultaneously, adherence to three fundamental assumptions (Fig. 1) was imperative: (1) genetic variations are directly associated with the exposure (the relevance assumption); (2) genetic variations are unrelated to potential confounders between the exposure and the outcome, ensuring that the results are not influenced by confounding factors (the independence assumption); (3) genetic variations do not affect the outcome through pathways other than the exposure (the exclusivity assumption). All studies included in the utilized dataset have obtained approval from the relevant institutional review boards.

Fig. 1
figure 1

Three basic assumptions of Mendelian randomization analysis. ①the relevance assumption; ②the independence assumption; ③the exclusivity assumption. IVS, instr-umental variables

Full size image

Data sources for exposure and outcome

The summary statistics of 731 immune traits are publicly available in the GWAS Catalog (https://www.ebi.ac.uk/gwas/), accession numbers from GCST90001391 to GCST90002121 [14], encompassing a total of 3,757 Sardinian samples (57% female). All included data in the analysis were derived through flow cytometry, involving the examination of approximately 22 million SNPs. These data are categorized based on trait type, distinguishing absolute cell (AC) counts (n = 118), median fluorescence intensity (MFI) reflecting surface antigen levels (n = 389), morphological parameters (MP) (n = 32), and relative cell (RC) counts (n = 192). The GWAS summary statistics pertaining to diabetic retinopathy (GWAS ID: finngen_R9_DM_RETINOPATHY_EXMORE) were extracted from the Finngen research project (https://r9.finngen.fi/) [15], a large-scale case-control study. This GWAS incorporated 14,584 cases of DR and 176,010 controls of European ancestry, involving a total of 16,380,347 SNPs. Rigorous quality checks were conducted on SNPs to ensure data robustness and result accuracy, meeting the requirements for IVs.

Selection of IVs

In this study, immune cells were utilized as the exposure variable, and DR was employed as the outcome variable to assess the role of immune cells in DR. To satisfy the three assumptions of the two-sample MR analysis, several quality control steps were implemented to select IVs closely associated with immune cells and meeting the criteria. (1) The paucity of significant IVs below the stringent threshold of P < 5 × 10− 8 compromises the statistical power of subsequent analyses. Upon careful investigation, we noted that a threshold of P < 1 × 10− 5 is commonly used in studies utilizing this dataset with 731 immune cells for MR analyses [16,17,18,19]. After careful consideration, and in line with recent research trends, we adopted a more relaxed significance threshold of P < 1 × 10− 5 for our IVs selection. While this threshold is more lenient than traditional standards, it represents a pragmatic compromise given the current limitations of available genetic variations within the study domain. (2) Additionally, the Clump function was employed to conduct linkage disequilibrium testing (setting criteria as r2 < 0.001, kb = 10000). (3) The F-statistic was employed to gauge IVs strength [20]. We computed the F-statistic and retained IVs with F > 10 to prevent bias introduced by weak IVs. (4) To meet the independence assumption, potential confounding factors correlated with the outcome were identified and removed using PhenoScanner (http://www.phenoscanner.medschl.cam.ac.uk/) [21] for comprehensive cross-validation. (5) To ensure that IVs corresponded to the same allelic genes in the same order for exposure and outcome, we adjusted for non-palindromic IVs chains and removed IVs containing palindromic sequences, thereby harmonizing the exposure and outcome datasets. In summary, subsequent MR analysis was conducted using the IVs that remained after a rigorous multistep selection process.

Statistical analysis

Recent MR analyses have confirmed that risk factors like Body mass index [22], obesity [23], waist circumference [23], Waist hip ratio [24], and smoking [24] are causally linked to DR. Through a comprehensive search on the Phenoscanner website, we identified 96 SNPs (Supplementary Table S1) that displayed suggestive associations (P < 1 × 10− 5) with these risk factors and subsequently excluded them to effectively minimize potential confounding influences. To prevent outliers from causing horizontal pleiotropy in subsequent analyses, we utilized MR-PRESSO to detect and remove outliers while performing MR analyses [25]. 13 outliers (Supplementary Table S2) with Distortion test P-values < 0.05 (indicating a difference in results before and after correction) were removed from the IVs that were found to potentially cause horizontal pleiotropy legacy. Subsequently, the filtered set of 18,468 IVs (Supplementary Table S3) underwent a renewed MR analysis (MR workflow in Fig. 2). This study used inverse variance-weighted (IVW), weighted median, simple mode, Weighted mode, and MR-Egger methods to assess potential causal effects (Supplementary Table S4), and we gave priority to positive results from IVW methods. We opted for the random-effects IVW method due to acknowledging the presence of heterogeneity between studies. Random-effects method enhances the consideration of heterogeneity. However, we are also aware that even with the use of random-effects model, there remains some uncertainty and the results may be influenced by heterogeneity. We conduct the sensitivity analysis to confirm the reliability of our results.

Fig. 2
figure 2

Flowchart of MR design. MR, Mendelian randomization; IVW, inverse variance weighted

Full size image

To mitigate potential over-bias, sensitivity analyses were conducted. Heterogeneity was evaluated using Cochran’s Q test [26]. Horizontal pleiotropy tests were conducted for MR-PRESSO with MR-Egger regression (commonly denoted by the intercept term). If the intercept term of the MR-Egger regression differed significantly from 0 (P-value < 0.05), it indicated the presence of horizontal pleiotropy [27]. Furthermore, the results were re-calculated using the Leave-one-out(LOO) method after stepwise removal of individual SNPs. The optimal outcome would be minimal changes in the results after sequentially removing each SNP. Causal effects were presented in terms of odds ratios (OR) and their 95% confidence intervals (CI). The direction and magnitude of the OR values are crucial indicators for evaluating causal relationships. Furthermore, to guard against Type I errors in multiple hypothesis testing, the false discovery rate (FDR) was applied to correct P-values obtained from the IVW method. In the field of MR analysis, it is common and advisable to use FDR as a means of controlling false positive rates [28]. Discoveries that yield adjusted P-values below the conventional 0.05 threshold are considered statistically significant [29, 30]. The aforementioned statistical analyses were primarily executed using the R software (version 4.3.2) and the “TwoSampleMR” package (version 0.5.8), along with the “MR-PRESSO” package (version 1.0) [31].

Results

Overview

After rigorous quality checks, a two-sample Mendelian randomization analysis excluded 96 confounder-associated IVs and 13 outliers, resulting in 33 trait pairs. To account for potential false positives, we applied the FDR method to adjust the P values obtained from the IVW method. Following this correction, 11 trait pairs (5AC-DR, 5RC-DR, and 1MFI-DR) were identified with PFDR >0.05. As the goal of the study was to investigate immunological factors causally linked to DR using Mendelian randomization, these 11 pairs were excluded from the final analysis. The sensitivity analyses results for the remaining 22 trait pairs (11 MFI-DR, 5 AC-DR, 5 RC-DR, and 1 MP-DR) indicated no horizontal pleiotropy, and four trait pairs showed potential heterogeneity. As previously mentioned, we utilized a random-effects IVW method that effectively addresses heterogeneity [32], so the results are acceptable [33]. In all results, the F-statistic was > 19, ensuring the strength of the IVs. When the 22 traits were classified into seven sets of panel types, most were found to be linked to T-cell subsets (Supplementary Table 6). Furthermore, upon classifying based on OR values, there were 9 trait pairs with OR < 1 (5 AC-DR, 2 RC-DR, 2 MFI-DR), and 13 trait pairs with OR > 1 (9 MFI-DR, 3 RC-DR, 1 MP-DR). We found that all AC traits among these 22 traits exhibited a potential association with a decreased risk of DR, while most MFI traits and the sole MP trait may be associated with an increased risk of DR.

Immune traits with OR values < 1

Figure 3 clearly shows 9 immune traits with ORs < 1, and none of their 95% CIs crossed 1. This indicates that our results are statistically significant. After FDR correction, the results belonging to the AC trait all presented the possibility of reducing the risk of DR [CD11c + monocyte AC(OR:0.962, 95% CI: 0.929–0.998, PFDR=0.0475), CD14 + CD16- monocyte AC(OR: 0.974, 95% CI. 0.954–0.994, PFDR=0.0396), CD4 + CD8dim AC(OR:0.958, 95% CI: 0.922–0.996, PFDR=0.0443), CD45RA- CD4 + AC(OR:0.978, 95% CI: 0.959–0.996, PFDR=0.0410), EM DN(CD4-CD8-) AC (OR:0.946, 95% CI: 0.904–0.991, PFDR=0.0400)]. We further established the causal link between these five AC traits and the risk of DR through the scatter plot in Fig. 4. It’s worth noting that CD25 on lgD- CD24- within the B-cell group showed a more substantial causal connection with the risk of developing DR (OR: 0.943, 95% CI: 0.901–0.987, PFDR=0.0369). This suggests that this particular subpopulation of immune cells might have a significant impact on reducing the risk of DR. Furthermore, CD14 + CD16 + monocyte %monocyte, CD39 + secreting Treg %CD4 Treg and HVEM on TD CD4 + may all act as protective factors for DR. Up to four of these nine immune traits are linked to membership in the T-cell group.

Fig. 3
figure 3

The forest plot illustrating the causal impact of immune cells on the risk of DR, derived using the Inverse Variance Weighted method. OR, odds ratios; CI, confidence intervals; PIVW, P-value of the IVW method; PFDR, FDR-corrected P-value

Full size image
Fig. 4
figure 4

The scatter plot depicting the risk of DR associated with 5 AC traits with OR < 1 (primarily focusing on the IVW method). DR, diabetic retinopathy; AC, absolute cell; OR, odds ratios; IVW, inverse variance-weighted

Full size image

Immune traits with OR values > 1

The LOO plots for the 9 MFI traits with OR values > 1 are depicted in Fig. 5, illustrating the robustness of the results. This suggests a causal relationship between the MFI features of these cells and the risk of DR. Among them, the Myeloid cell CD45 on CD33- HLA DR- (OR: 1.050, 95% CI: 1.010–1.092, PFDR=0.0377) exhibits a relatively significant positive causal relationship with the risk of DR. FSC-A on CD14 + monocyte (OR: 1.033, 95% CI: 1.004–1.063, PFDR=0.0432) is the only MP trait in the analysis results. Additionally, it is noteworthy that from Fig. 3, Activated Treg %CD4 (OR: 1.040, 95% CI: 1.000-1.081, PFDR=0.0495), and CD3 on NKT (OR: 1.059, 95% CI: 1.000-1.122, PFDR=0.0495) exhibit distinct characteristics. The 95% CI for both traits covered 1.000, indicating that we cannot dismiss the possibility of a null effect, and thus this may necessitate further in-depth study and understanding. Consistent with earlier findings, immune traits with OR > 1 showed the most significant associations for T cell subset characteristics compared to other groups. This underscores the significance of T cells in regulating the immune response and offers clues for further exploration of the potential role of these T cell subsets in the mechanisms of DR occurrence.

Fig. 5
figure 5

The LOO forest plot illustrating the OR > 1 for 9 MFI traits in relation to DR. LOO, leave-one-out; OR, odds ratios; MFI, median fluorescence intensity; DR, diabetic retinopathy

Full size image

Discussion

Earlier studies have hinted at the role of immune factors in DR development, but this study is the first to comprehensively examine the causal link between multiple immune traits and DR using MR analysis. To ensure the reliability of causality, we minimized the impact of potential confounders and outliers. The number of causal relationships between MFI traits and DR was the highest compared to the other three. Our results show that CD25 on lgD- CD24- (belonging to the B cell) with MFI traits demonstrated a significant negative causal association with the risk of DR. CD25 has long been known to have a negative regulatory role in the immune system, primarily by modulating the immune response through inhibition of free Interleukin-2 (IL-2), which helps prevent excessive immune responses [34]. However, past literature has reported that CD25 is typically associated with regulatory T cells (Treg) [35]. Research on the specific role of CD25 in B cells and its effects on immune responses [36] is relatively limited and requires further exploration and investigation.

The AC and RC traits serve as valuable parameters for the analysis of specific immune cell subsets, allowing for the examination of both the quantity and relative proportions. Notably, both the percentage of CD14 + CD16 + monocytes and the AC of CD14 + CD16- monocytes demonstrate potential negative correlations with the risk of DR. The CD14 + CD16 + monocytes, recognized as intermediate monocytes, constitute the third major subpopulation of monocytes alongside classical (CD14 + CD16-) and nonclassical (CD14 CD16-) monocytes. This unique subpopulation exhibits a distinct gene expression profile that has been linked to various diseases [37]. CD14 + CD16- monocytes make up about 80% of all monocytes and play a role in phagocytosis, defense against pathogens, homeostasis, and tissue repair [38]. Previous research has linked CD11c + monocytes with atherosclerosis [39], intestinal inflammation [40], and the promotion of chronic inflammatory responses [41], making them a focal point in studies of immune system function and disease pathogenesis. Contrary to expectations, our findings suggest that CD11c + monocytes with AC features may be negatively linked to DR risk, warranting further experiments and studies for verification.

Additionally, it is of significance to highlight that, mirroring previous research trends, a majority of the immune cells identified to have a causal association with the risk of DR in this study belong to the T-cell category. As early as 1995, Tang et al. [42]. made a seminal observation that T lymphocytes, in conjunction with interleukin-2 (IL-2), a growth-promoting factor for various immune cells, infiltrated the retinal tissue of individuals with DR. Furthermore, in the fibrovascular membranes (FVM) of proliferative diabetic retinopathy (PDR) patients, researchers noted a significant elevation in the densities of both CD4 + T lymphocytes and CD8 + T lymphocytes [43]. Several T cell subsets associated with CD4 or CD8 surface markers were also identified in our findings, and their heightened risk of DR demonstrated a positive and causal correlation. Notably, both CD28 on CD28 + CD45RA + CD8br and CD28 on CD45RA + CD4 + encompass a subpopulation of T cells expressing CD28, with an additional association with CD45RA expression. Previous literature has hinted at a potential link between CD28 expression in T cells and the pathogenesis of DR [44]. Regarding CD8 on CD39 + CD8br, upon thorough examination, we found that CD39 serves as an extracellular nucleotidase that, in conjunction with CD73, catalyzes the conversion of extracellular ATP to adenosine, contributing to the immunosuppressive adenosine pathway. Research has indicated that the CD39 + T cell population exhibits robust anti-tumor cytotoxicity across various cancer types [45]. In line with our current investigation, an increased level of HVEM on EM CD8br, belonging to the Maturation stages of T cells, is identified as a potential contributor to an elevated risk of DR. The HVEM signaling pathway is postulated to play a regulatory role in influencing the function of the CD8 T cell subpopulation [46]. Furthermore, our analysis has brought attention to the involvement of double-negative T (DNT) cells. DNT cells, inclusive of naive subpopulations, have been a focal point of research [47] in the realms of autoimmune diseases, inflammation, and cancer [48]. However, additional investigations may be warranted to comprehensively understand the impact of these distinctly expressed T cell subsets on DR within diverse physiological and pathological contexts.

Among the T cell subsets associated with CD4 or CD8 surface markers, CD39 + secreting Treg %CD4 Treg showed a negative causal association with DR risk in this analysis. This observation implies a pivotal role for this specific T cell subset in the regulation of immune responses and the maintenance of immune homeostasis. Additionally, three Maturation stages of T cells, namely CD45RA- CD4+, EM DN(CD4-CD8-) AC, and HVEM on TD CD4+, were also found to exhibit a negative association, potentially contributing to the regulation of T cell function [49, 50]. However, their relevance to the mechanism underlying the occurrence of DR remains a subject that warrants further, in-depth investigation.

In summary, it is evident that there is a dearth of studies focusing on the segmentation and elucidation of clear causality within these cell subpopulations. Our analysis, however, has produced more nuanced results, offering suggestive evidence for subsequent investigations. Nevertheless, our study is not without limitations. Firstly, the reliance on a European database restricts the applicability of our conclusions to other racial groups, thus limiting the generalizability of the results in terms of demographic aspects. Secondly, despite the execution of multiple sensitivity analyses, a comprehensive evaluation of horizontal pleiotropy remains challenging. Thirdly, all conclusions are derived from exploratory analyses based on existing data and require further validation. Even though our results are statistically significant with FDR < 0.05, they exceed 0.036, which may indicate weaker associations. Further analysis is needed once stronger and more suitable data become accessible. Fourthly, relaxing the selection criterion from 5 × 10− 8 to 1 × 10− 5 represents a trade-off aimed at improving statistical power, but also carries the risk of increased false-positive rates. A stringent threshold (p < 5 × 10− 8) is generally favorable for obtaining robust MR results. However, MR analysis requires a trade-off between statistical power and risk of bias, based on study objectives and sensitivity analyses [51]. In exceptional cases, such as when the exposure lacks sufficiently potent IVs resulting in inadequate statistical power, a moderate relaxation of the standards may be warranted. Indeed, Wootton et al. [52]. adopted the relaxed threshold of 1 × 10− 5 in their 2018 MR study. The complexity of immune system regulation requires us to not only focus on highly significant genetic signals but also to explore genetic variants that exhibit multiple coincident associations with various immune phenotypes and disease states. These variants may be crucial for understanding disease mechanisms, but their effects may appear statistically weaker due to multifactorial influences. Therefore, we choose to moderately relax the threshold to avoid overlooking these potentially key genetic markers, while rigorously validating their authenticity and effect size through subsequent analyses. While relaxing the significance threshold, we are fully aware of the necessity to control the risk of false positives. Our strategies include but are not limited to implementing sensitivity analyses, using P-value correction methods to ensure the biological plausibility of the selected IVs. These supplementary measures aim to seek a reasonable balance between statistical power and biological significance. Bowden et al. [27]. suggested using the MR Egger method to provide unbiased estimates of the overall causal effect and to quantify potential systematic biases when IVs may not meet all necessary assumptions, thereby assisting researchers in assessing IVs efficacy and providing more reliable causal inferences. Our application of the MR Egger method revealed no evidence of horizontal pleiotropy, suggesting that the relaxed criteria still yield effective IVs. We recognize that relaxing the threshold is applicable to the exploratory research phase, aimed at discovering new genetic clues and hypotheses. Like other studies that have also relaxed thresholds, our goal is to open up new research directions rather than reaching final conclusions. Future analyses should be conducted when larger genetic studies become available to bolster the selection of IVs. Furthermore, we call upon the scientific community to collaboratively explore more systematic approaches to threshold selection, particularly in devising reasonable compromises when confronted with data limitations. Future studies should strive to uncover additional genetic variants closely tied to exposure variables, thereby enhancing the precision of MR analyses. This concerted effort will not only strengthen the validity of our findings but also pave the way for a more nuanced understanding of complex disease etiologies and the intricate interplay between genetics and phenotype. In conclusion, utilizing MR as an exploratory tool, this study is a significant advancement in understanding the potential causal link between immune cells and DR. However, limitations in IVs emphasize the importance of interpreting findings cautiously and call for future studies to enhance genetic marker data and sample size. These enhancements are crucial for creating stronger and more widely applicable scientific insights.

Data availability

The GWAS data on 731 immune traits originated from the GWAS Catalog (GCST0001391 to GCST0002121), https://www.ebi.ac.uk/gwas/. The GWAS summary statistics pertaining to diabetic retinopathy (GWAS ID: finngen_R9_DM_RETINOPATHY_EXMORE) were extracted from the Finngen research project, https://r9.finngen.fi/. Use PhenoScanner to remove confounders, http://www.phenoscanner.medschl.cam.ac.uk/.

References

  1. Qaum T, Xu Q, Joussen AM, Clemens MW, Qin W, Miyamoto K, et al. VEGF-initiated blood-retinal barrier breakdown in early diabetes. Invest Ophthalmol Vis Sci. 2001;42(10):2408–13.

    CAS  PubMed  Google Scholar 

  2. Ahsan H. Diabetic retinopathy – biomolecules and multiple pathophysiology. Diabetes Metabolic Syndrome: Clin Res Reviews. 2015;9(1):51–4.

    Article  Google Scholar 

  3. Yue T, Shi Y, Luo S, Weng J, Wu Y, Zheng X. The role of inflammation in immune system of diabetic retinopathy: molecular mechanisms, pathogenetic role and therapeutic implications. Front Immunol. 2022;13:1055087.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  4. Yau JWY, Rogers SL, Kawasaki R, Lamoureux EL, Kowalski JW, Bek T, et al. Global prevalence and major risk factors of Diabetic Retinopathy. Diabetes Care. 2012;35(3):556–64.

    Article  PubMed  PubMed Central  Google Scholar 

  5. Teo ZL, Tham Y-C, Yu M, Chee ML, Rim TH, Cheung N, et al. Global prevalence of Diabetic Retinopathy and Projection of Burden through 2045. Ophthalmology. 2021;128(11):1580–91.

    Article  PubMed  Google Scholar 

  6. Wang X, Wang T, Lam E, Alvarez D, Sun Y. Ocular vascular diseases: from Retinal Immune privilege to inflammation. IJMS. 2023;24(15):12090.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  7. Pan WW, Lin F, Fort PE. The innate immune system in diabetic retinopathy. Prog Retin Eye Res. 2021;84:100940.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  8. Kinuthia UM, Wolf A, Langmann T. Microglia and inflammatory responses in Diabetic Retinopathy. Front Immunol. 2020;11:564077.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  9. Liu Y, Yang Z, Lai P, Huang Z, Sun X, Zhou T, et al. Bcl-6-directed follicular helper T cells promote vascular inflammatory injury in diabetic retinopathy. Theranostics. 2020;10(9):4250–64.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  10. Liu B, Hu Y, Wu Q, Zeng Y, Xiao Y, Zeng X et al. Qualitative and Quantitative Analysis of B-Cell-Produced Antibodies in Vitreous Humor of Type 2 Diabetic Patients with Diabetic Retinopathy. Journal of Diabetes Research. 2020;2020:1–7.

  11. DeMaio A, Mehrotra S, Sambamurti K, Husain S. The role of the adaptive immune system and T cell dysfunction in neurodegenerative diseases. J Neuroinflammation. 2022;19(1):251.

    Article  PubMed  PubMed Central  Google Scholar 

  12. Davey Smith G, Ebrahim S. Mendelian randomization’: can genetic epidemiology contribute to understanding environmental determinants of disease?*. Int J Epidemiol. 2003;32(1):1–22.

    Article  Google Scholar 

  13. Ference BA, Holmes MV, Smith GD. Using mendelian randomization to improve the design of Randomized trials. Cold Spring Harb Perspect Med. 2021;11(7):a040980.

    Article  PubMed  PubMed Central  Google Scholar 

  14. Orrù V, Steri M, Sidore C, Marongiu M, Serra V, Olla S, et al. Complex genetic signatures in immune cells underlie autoimmunity and inform therapy. Nat Genet. 2020;52(10):1036–45.

    Article  PubMed  PubMed Central  Google Scholar 

  15. Kurki MI, Karjalainen J, Palta P, Sipilä TP, Kristiansson K, Donner KM, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023;613(7944):508–18.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  16. Wang C, Zhu D, Zhang D, Zuo X, Yao L, Liu T, et al. Causal role of immune cells in schizophrenia: mendelian randomization (MR) study. BMC Psychiatry. 2023;23(1):590.

    Article  PubMed  PubMed Central  Google Scholar 

  17. Han S, Gao J, Wang Z, Xiao Y, Ge Y, Liang Y, et al. Genetically supported causality between gut microbiota, immune cells and morphine tolerance: a two-sample mendelian randomization study. Front Microbiol. 2024;15:1343763.

    Article  PubMed  PubMed Central  Google Scholar 

  18. Song Z, Li W, Han Y, Xu Y, Ding H, Wang Y. Association of immune cell traits with Parkinson’s disease: a mendelian randomization study. Front Aging Neurosci. 2024;16:1340110.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  19. Cao R-R, Yu X-H, Xiong M-F, Li X-T, Deng F-Y, Lei S-F. The immune factors have complex causal regulation effects on bone mineral density. Front Immunol. 2022;13:959417.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  20. Burgess S, Thompson SG, CRP CHD Genetics Collaboration. Avoiding bias from weak instruments in mendelian randomization studies. Int J Epidemiol. 2011;40(3):755–64.

    Article  PubMed  Google Scholar 

  21. Kamat MA, Blackshaw JA, Young R, Surendran P, Burgess S, Danesh J, et al. PhenoScanner V2: an expanded tool for searching human genotype-phenotype associations. Bioinformatics. 2019;35(22):4851–3.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  22. Shu Y, Zhou Q, Shao Y, Lin H, Qu S, Han W, et al. BMI and plasma lipid levels with risk of proliferative diabetic retinopathy: a univariable and multivariable mendelian randomization study. Front Nutr. 2023;10:1099807.

    Article  PubMed  PubMed Central  Google Scholar 

  23. Zheng C, Wei X, Cao X. The causal effect of obesity on diabetic retinopathy: a two-sample mendelian randomization study. Front Endocrinol. 2023;14:1108731.

    Article  Google Scholar 

  24. Su Z, Wu Z, Liang X, Xie M, Xie J, Li H, et al. Diabetic retinopathy risk in patients with unhealthy lifestyle: a mendelian randomization study. Front Endocrinol. 2023;13:1087965.

    Article  Google Scholar 

  25. Verbanck M, Chen C-Y, Neale B, Do R. Detection of widespread horizontal pleiotropy in causal relationships inferred from mendelian randomization between complex traits and diseases. Nat Genet. 2018;50(5):693–8.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  26. Greco MFD, Minelli C, Sheehan NA, Thompson JR. Detecting pleiotropy in mendelian randomisation studies with summary data and a continuous outcome. Stat Med. 2015;34(21):2926–40.

    Article  Google Scholar 

  27. Bowden J, Davey Smith G, Burgess S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression. Int J Epidemiol. 2015;44(2):512–25.

    Article  PubMed  PubMed Central  Google Scholar 

  28. Jafari M, Ansari-Pour N, Why. When and how to adjust your P values? Cell J. 2018; 20(4).

  29. Cao Y, Lu H, Xu W, Zhong M. Gut microbiota and Sjögren’s syndrome: a two-sample mendelian randomization study. Front Immunol. 2023;14:1187906.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  30. Mavromatis LA, Rosoff DB, Cupertino RB, Garavan H, Mackey S, Lohoff FW. Association between Brain structure and Alcohol Use behaviors in adults: a mendelian randomization and Multiomics Study. JAMA Psychiatry. 2022;79(9):869.

    Article  PubMed  PubMed Central  Google Scholar 

  31. Hemani G, Zheng J, Elsworth B, Wade KH, Haberland V, Baird D, et al. The MR-Base platform supports systematic causal inference across the human phenome. eLife. 2018;7:e34408.

    Article  PubMed  PubMed Central  Google Scholar 

  32. Papadimitriou N, Dimou N, Tsilidis KK, Banbury B, Martin RM, Lewis SJ, et al. Physical activity and risks of breast and colorectal cancer: a mendelian randomisation analysis. Nat Commun. 2020;11(1):597.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  33. Burgess S, Davey Smith G, Davies NM, Dudbridge F, Gill D, Glymour MM, et al. Guidelines for performing mendelian randomization investigations: update for summer 2023. Wellcome Open Res. 2023;4:186.

    Article  PubMed  PubMed Central  Google Scholar 

  34. Lindqvist CA, Christiansson LH, Simonsson B, Enblad G, Olsson-Strömberg U, Loskog ASI. T regulatory cells control T‐cell proliferation partly by the release of soluble CD25 in patients with B‐cell malignancies. Immunology. 2010;131(3):371–6.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  35. Brusko TM, Wasserfall CH, Hulme MA, Cabrera R, Schatz D, Atkinson MA. Influence of membrane CD25 Stability on T lymphocyte activity: implications for Immunoregulation. Unutmaz D, directeur. PLoS ONE. 2009;4(11):e7980.

    Article  PubMed  PubMed Central  Google Scholar 

  36. Amu S, Strömberg K, Bokarewa M, Tarkowski A, Brisslert M. CD25-expressing B‐lymphocytes in Rheumatic diseases. Scand J Immunol. 2007;65(2):182–91.

    Article  CAS  PubMed  Google Scholar 

  37. Zawada AM, Rogacev KS, Rotter B, Winter P, Marell R-R, Fliser D, et al. SuperSAGE evidence for CD14 + + CD16 + monocytes as a third monocyte subset. Blood. 2011;118(12):50–61.

    Article  Google Scholar 

  38. Wu H, Gower RM, Wang H, Perrard X-YD, Ma R, Bullard DC, et al. Functional role of CD11c + monocytes in Atherogenesis Associated with Hypercholesterolemia. Circulation. 2009;119(20):2708–17.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  39. Schneider L, Marcondes NA, Hax V, Da Silva Moreira IF, Ueda CY, Piovesan RR, et al. Flow cytometry evaluation of CD14/CD16 monocyte subpopulations in systemic sclerosis patients: a cross sectional controlled study. Adv Rheumatol. 2021;61(1):27.

    Article  PubMed  Google Scholar 

  40. Arnold IC, Mathisen S, Schulthess J, Danne C, Hegazy AN, Powrie F. CD11c + monocyte/macrophages promote chronic Helicobacter hepaticus-induced intestinal inflammation through the production of IL-23. Mucosal Immunol. 2016;9(2):352–63.

    Article  CAS  PubMed  Google Scholar 

  41. Corbin AL, Gomez-Vazquez M, Berthold DL, Attar M, Arnold IC, Powrie FM, et al. IRF5 guides monocytes toward an inflammatory CD11c + macrophage phenotype and promotes intestinal inflammation. Sci Immunol. 2020;5(47):eaax6085.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  42. Tang S, Le-Ruppert KC. Activated T lymphocytes in epiretinal membranes from eyes of patients with proliferative diabetic retinopathy. Graefes Arch Clin Exp Ophthalmol. 1995;233(1):21–5.

    Article  CAS  PubMed  Google Scholar 

  43. Urbančič M, Štunf Š, Milutinović Živin A, Petrovič D, GlobočnikPetrovič M. Epiretinal membrane inflammatory cell density might reflect the activity of proliferative diabetic retinopathy. Invest Ophthalmol Vis Sci. 2014;55(12):8576–82.

    Article  PubMed  Google Scholar 

  44. Cantón A. CD4-CD8 and CD28 expression in T cells infiltrating the vitreous fluidin patients with proliferative Diabetic Retinopathy: a Flow Cytometric Analysis. Arch Ophthalmol. 2004;122(5):743.

    Article  PubMed  Google Scholar 

  45. Chow A, Fathema Z, Uddin L, Mangarin H, Rizvi A, Dobrin S, Tischfield et al. CD39 identifies tumor-reactive CD8 T cells in patients with Lung Cancer. bioRxiv. 2022;:2022.01.24.477554.

  46. Morel Y, Schiano De Colella J-M, Harrop J, Deen KC, Holmes SD, Wattam TA, et al. Reciprocal expression of the TNF family receptor herpes virus entry mediator and its ligand LIGHT on activated T cells: LIGHT down-regulates its own receptor. J Immunol. 2000;165(8):4397–404.

    Article  CAS  PubMed  Google Scholar 

  47. Li H, Tsokos GC. Double-negative T cells in autoimmune diseases. Curr Opin Rheumatol. 2021;33(2):163–72.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  48. Passos LSA, Koh CC, Magalhães LMD, Nunes MDCP, Gollob KJ, Dutra WO. Distinct CD4 – CD8– (Double-Negative) memory T-Cell subpopulations are Associated with Indeterminate and Cardiac Clinical forms of Chagas Disease. Front Immunol. 2021;12:761795.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  49. Mintz MA, Felce JH, Chou MY, Mayya V, Xu Y, Shui J-W, et al. The HVEM-BTLA Axis restrains T cell help to Germinal Center B cells and functions as a cell-extrinsic suppressor in Lymphomagenesis. Immunity. 2019;51(2):310–23.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  50. Sordo-Bahamonde C, Lorenzo-Herrero S, Granda-Díaz R, Martínez-Pérez A, Aguilar-García C, Rodrigo JP, et al. Beyond the anti-PD-1/PD-L1 era: promising role of the BTLA/HVEM axis as a future target for cancer immunotherapy. Mol Cancer. 2023;22(1):142.

    Article  CAS  PubMed  PubMed Central  Google Scholar 

  51. Emdin CA, Khera AV, Kathiresan S. Mendelian Randomization. 2017;318(19):1925–6.

    Google Scholar 

  52. Wootton RE, Lawn RB, Millard LAC, Davies NM, Taylor AE, Munafò MR, et al. Evaluation of the causal effects between subjective wellbeing and cardiometabolic health: mendelian randomisation study. BMJ. 2018;362:k3788.

    Article  PubMed  PubMed Central  Google Scholar 

Download references

Acknowledgements

We want to acknowledge the participants and investigators of the FinnGen study. Thanks to the Genome-Wide Association Research Alliance for making these summary statistics publicly available. Without them, this study would have been impossible.

Funding

This work was supported by the Enshi Science and Technology Program, D20220052.

Author information

Author notes

  1. Yuanyuan Li and Ying Xiang contributed equally to this work and share first authorship.

Authors and Affiliations

  1. Hubei Minzu University, Enshi, Hubei Province, China

    Yuanyuan Li, Bo Mou & Xiusheng Song

  2. Ophthalmology Center, The Central Hospital Of Enshi Tujia And Miao Autonomous Prefecture, Affiliated Hospital of Hubei University for Nationalities, Enshi, Hubei Province, China

    Yuanyuan Li, Ying Xiang, Bo Mou & Xiusheng Song

  3. Hubei Institute of Selenium and Human Health, Enshi, Hubei Province, China

    Ying Xiang & Xiusheng Song

Authors
  1. Yuanyuan Li
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Ying Xiang
    View author publications

    You can also search for this author in PubMed Google Scholar

  3. Bo Mou
    View author publications

    You can also search for this author in PubMed Google Scholar

  4. Xiusheng Song
    View author publications

    You can also search for this author in PubMed Google Scholar

Contributions

Y.L. and Y.X. collected and analysed the data, and wrote the main manuscript text. X.S. designed the project. B.M. drafted the manuscript. All authors have reviewed and approved the fnal manuscript.

Corresponding author

Correspondence to Xiusheng Song.

Ethics declarations

Ethics approval and consent to participate

Since the data adopted in this MR analysis were all publicly available data from the Finngen database and GWAS Catalog, all data-related studies were approved by their respective ethical review committees and received written informed consent from patients. Therefore, this study does not need additional ethics approval.

Competing interests

The authors declare no competing interests.

Additional information

Publisher’s Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Electronic supplementary material

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Li, Y., Xiang, Y., Mou, B. et al. Causal influence of immune factors on the risk of diabetic retinopathy: a mendelian randomization study. Diabetol Metab Syndr 16, 194 (2024). https://doi.org/10.1186/s13098-024-01441-6

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1186/s13098-024-01441-6

Keywords

Diabetology & Metabolic Syndrome

ISSN: 1758-5996

Contact us