Genome-wide Mendelian randomization and single-cell RNA sequencing analyses identify the causal effects of glucocorticoids on inflammatory disorders of breast

doi:10.21203/rs.3.rs-4772475/v1

Download PDF

Article

Genome-wide Mendelian randomization and single-cell RNA sequencing analyses identify the causal effects of glucocorticoids on inflammatory disorders of breast

https://doi.org/10.21203/rs.3.rs-4772475/v1

This work is licensed under a CC BY 4.0 License

You are reading this latest preprint version

Background

Inflammatory disorders of breast(IDB) is a common reason for glucocorticoids(GCs) treatment in women. However, there are no studies elucidating the relationship between GCs and the risk of IDB in general population. Therefore, we employed Mendelian randomization (MR) to explore the causal associations between GCs and IDB.

Methods

A two-sample MR analysis was performed using the summary statistics sourced from the largest genome-wide association studies(GWAS) conducted in GCs, glucocorticoid receptor-related mRNAs(GRs) and IDB. Inverse-variance weighting (IVW), MR‒Egger, and weighted median (WM) were further supported by several sensitivity analyses. Metascape and Single-cell RNA sequencing(scRNA-Seq) were used to analyze the functions and distribution of GRs.

Results

We detected causal genetic associations between GCS and IDB (OR, 1.22 (95% CI, 1.019– 1.462), P = 0.021). Further WM measure (OR, 1.294 (95% CI, 1.002– 1.671), P = 0.048) also showed similar results. No causal association was found between GCs and Childbirth-Associated Breast Infections. ScRNA-Seq confirmed that GRs were expressed in almost all immune cells, but more highly expressed in macrophages. The expression quantitative trait locus (eQTL) data suggest that NR1I3 is a high-risk factor for IDB.

Conclusions

We are the first to use MR analysis to explore the causal relationships between GCs and IDB, revealing an increased risk of IDB with GCs. These may caused by the highly expressed GRs on macrophages in breast tissue.

Biological sciences/Computational biology and bioinformatics

Biological sciences/Immunology

Health sciences/Medical research

Mendelian randomization

glucocorticoids

Inflammatory disorders of breast

macrophages

IDB is a commonly benign breast conditions with an increasing incidence rate^[1]. Idiopathic Granulomatous Mastitis (IGM) and Plasma Cell Mastitis (PCM) and tuberculous mastitis were three different types of IDB^{[1; 2]}. Currently, the cause of IDB is still unclear, but: autoimmune process, changes of pregnancy, hyperprolactinemia, and inflammation from extravasation of lactation products may contribute to IDB risk^{[1; 2; 3]}. IDB has a long disease course (6–12 months), which usually brings a huge financial and psychological burden to patients^{[1; 2]}. Looking for the risk factor of IDB is crucial for developing targeted preventative strategies and personalized management for those affected.

First-line treatment for moderate or severe IDB cases is typically oral steroids^[1]. GCs mediate their effects through the GR. GR is transported to the nucleus, where it binds to the glucocorticoid-responsive element (GRE) to upregulate the transcription of its target genes. Although GCs are widely mentioned for their anti-inflammatory activity, the pro-inflammatory effects of GCs are also used documented. The anti-inflammatory effects of GCs was related to dosage. High dosage of GCs could inhibit the transcription of inflammatory genes in LPS-activated and IFNγ-activated macrophages^[4]. In addition, LPS enhanced the expression and secretion of pro-inflammatory cytokines from macrophages. The pro-inflammatory cytokines was not suppressed but rather augmented following the administration of GCs to rats before LPS treatment^[5]. These studies suggest that the use of glucocorticoids may promote the occurrence and development of inflammation throughout the body in some situations. However, it is common that patients receive GCs but not suffering from any infections or inflammatory in clinical practice. For example, GCs were an important component of standard therapy for several lymphoid malignancies, including multiple myeloma, and acute lymphoblastic leukemia^{[6; 7; 8]}. Therefore, exploring the relationship between GCs and IDB has important theoretical and clinical application value.

MR, the use of genetic variants as instrumental variables for assessing the effect of modifiable exposures, is a powerful tool for inferring causality between risk factors and various health outcomes^{[9; 10]}. The principle behind MR is analogous to the random segregation of alleles during gamete formation, as described by Mendel's laws of inheritance. This approach was often described as a "natural randomized controlled trial" as it could mitigate the impact of confounding factors significantly^{[11; 12; 13]}. In this study, a two-sample MR analysis was employed to investigate the causal associations between GCs and vIDB. Furthermore, we utilized single-cell sequencing to provide complementary insights into the underlying mechanisms, aiming to guide the clinical use of GCs.

Study design

The MR design is shown in Fig. 1. In this study, we performed a two-sample MR analysis to explore the causal associations between GRs and IDB based on GWAS database. This MR study was conducted according to the guidelines of the STROBE-MR Statement^[14].

Genetic IV selection

We use the following criteria to screen instrumental variables(IVs) for GCs and GRs: 1) extract_instruments was used to select single nucleotide polymorphisms (SNPs) strongly related to exposure. Only SNPs that reached genome-wide signiffcance (P < 5 × 10^− 8) and displayed independence as determined through linkage disequilibrium (r2 < 0.001, kb > 10000) were considered^[15]; 2) Only SNPs with F > 10 were considered^[16]; 3) harmonise_data was used to remove the SNPs for palindromic or incompatible alleles; 4) ieu open gwas project(https://gwas.mrcieu.ac.uk/) was used to remove the SNPs that were associated with confounding factors. These strictly selected SNPs were used as instrumental variables in the subsequent MR analysis of the two samples.

GWAS data for GCs, IDB, and Childbirth-Associated Breast Infections

The details of the GWASs for GCs, IDB and Childbirth-Associated Breast Infections were shown in Table 1. Medication use (glucocorticoids) was come from a Meta-analyses with the UK Biobank and FinnGen^[17]. This cohort included 17,352 patients and 188,348 participants from European. The GWAS dates of IDB was come from FinnGen (https://www.finngen.fi/en). This cohort included 2,495 patients and 392,423 participants from European. The eQTL instruments for GRs were obtained from the IEU.

Table 1

Details of the GWAS included in the Mendelian randomization.
Trait	Population	Cases	Controls	Source
Medication use (glucocorticoids)	European	17,352	188,348	ebi-a-GCsST90019000
Inflammatory disorders of breast		2167	233,585	r11.finngen.fi
Childbirth-Associated Breast Infections		1233	244,020	r11.finngen.fi

Statistical analysis

To detect the causal effect of GCs and GRs on the risk of IDB, the IVW method was applied for the main MR estimates. The causal effect estimates were averaged from multiple genetic variants, utilizing the inverse of their variances as weights^[18]. MR- Egger and weighted median methods were meanwhile used as a complemental role for IVW estimates. When there are more than 3 genetic variant instrumental variables, a random effects model is used for analysis, otherwise a fixed effects model is adopted. MR‒Egger regression allows for the intercept to differ from zero, providing a test for directional pleiotropy^[19]. The heterogeneity of each SNP was evaluated employing Cochran’s Q statistic, with a P value < 0.05 indicating significant heterogeneity among the estimates of the included SNPs. The TwoSampleMR^[20] (https://github.com/MRCIEU/TwoSampleMR) was used to perform MR-egger_intercept by the function “mr_pleiotropy_test”. Among the included SNPs, SNPs with a P value < 0.05 indicating signiffcant pleiotropy. The Calculation of the OR and 95%CI: OR = e^b, LowerCI = e^{b−1.96* se}, UpperCI = e^{b+1.96* se}. The TwoSampleMR was used to perform Leave-one-out plot, scatter_plot, forest_plot and funnel_plot by the function “mr_leaveoneout”, “mr_scatter_plot”, “mr_forest_plot” and “mr_funnel_plot”. In addition, the leave-one-out sensitivity analysis to identify the effect of individual SNPs^[21]. The detected outliers mentioned above were excluded from the IVs to ensure the more precise corrected estimate.

Enrichment analysis and single cell RNA-sequencing (scRNA-Seq) analysis

We used Gene Set Enrichment Analysis (GSEA) database to obtain GR-related mRNAs (WP_GLUCOCORTICOID_RECEPTOR_PATHWAY)^[22]. The Metascape (version 3.5) is a database that provides more than forty bioinformatics databases. To investigate the biological functions mediated by the GR-related mRNAs, we conducted the enrichment analysis based on Metascape^[23]. Cellxgene (https://cellxgene.cziscience.com/) is a publicly available resource covering 87.2 million unique cells in 881 cell types. It provides a platform for researchers to search, download, and analyze scRNA-Seq datasets generated from cellxgene data sets. To explore effects of GCs on immune cell function in breast tissue, we used the cellxgene to analyze the distribution of GR-related mRNAs in different immune subtypes. All scRNA-Seq datasets in this paper were from European.

Genetic instruments for GCs

Initially, we identified 25 SNPs that were significantly (5 × 10^− 8) and independently (LDr2 < 0.001) associated with GCs. Subsequently, we calculated F-statistics, which ranged from 30.63 to 163.89 and thereby indicated the absence of weak instruments. The SNPs rs1391371 and rs1689510 were excluded due to their associations with body mass index by ieu open gwas project. The SNPs rs7717955 was excluded due to its great heterogeneity by mr_scatter_plot function. Following this, we harmonised GCs data with those of IDB, and Childbirth-Associated Breast Infections. The harmonised data of GCs data and IDB are presented in Table S1.

MR estimates and sensitivity analyses

The MR results are shown in Fig. 2. No significant causal relationship was found for GCs and the risk of Childbirth-Associated Breast Infections (odds ratio (OR) = 0.97, 95% confidence interval (CI) = 0.745–1.264, P = 0.823), which was supported by the MR‒Egger (OR = 0.768, 95% CI = 0.333–1.771, P = 0.542) and WM (OR = 0.969, 95% CI = 0.687–1.368, P = 0.859) analyses (Table S2). No significant heterogeneity (P for Cochran’s Q = 0.903) (Table S3) or horizontal pleiotropy (P for intercept = 0.845) (Table S4) was detected in this MR analysis (Fig. 2).

For IDB, by the IVW method, our results discovered that GCs could exert a significantly potential causal and risk effect on the IDB (OR, 1.2 (95% CI, 1.014– 1.421), P = 0.033)(Fig. 2). Scatter_plot and forest_plot showed the GCs is positively correlated with IDB (Fig S1A and S1C). Leave-one-out analysis and funnel_plot showed the potential outlying SNPs (Fig S1B、D). The association between SLE and PBC remained significant after we removed the outliers and repeated the MR analyses. By the IVW method, our results discovered that GCs could exert a significantly potential causal and risk effect on the IDB (OR, 1.22 (95% CI, 1.019– 1.462), P = 0.021). Further weighted median measure (OR, 1.294 (95% CI, 1.002– 1.671), P = 0.048) also showed similar results (Fig. 2). Scatter_plot and forest_plot reinforce previously obtained outcome (Fig S2A and S2C). Besides, funnel_plot (Fig S2D) and leave-one-out analysis (Fig S2B) indicated no heterogeneity among the instrumental SNPs.

The functions and distribution of GRs in breast tissue

We obtained 70 GRs from the GSEA database (Table S5)。The enrichment analysis results show that GRs were related to the activation macrophages (Fig S3)。UMAP plots of single cells from breast tissue showed the distribution of 14 immune cell subtypes based on 70 GRs (Fig. 3C)。Besides innate lymphoid cell, macrophages led to the highest expression of 70 GRs, followed by dendritic cell, and then Tc1 cells. Based on the eQTL data from 70 GRs, we identified 203 SNPs to investigate the impact of GCs on IDM (Table S6). Genes with p-values < 0.05 using either the "Wald ratio" or "Inverse variance weighted" test methods were MFGE8 and NR1I3. These results were visualized using a volcano plot (Fig. 3A). The forest plot indicates that MFGE8 acts as a protective factor for IDB (Figure S4), whereas NR1I3 represents a high-risk factor for IDB (Fig. 3B).

In pathological terms, IDB manifests as an inflammatory response within the mammary gland connective tissue. Subsequently, macrophages and lymphocytes migrate to this region, resulting in the development of mammary gland inflammation. Consequently, IDB is considered an autoimmune disease. At present, the etiology of IDB remains unclear. Currently, there is only one study demonstrated a causal relationship between Body Mass Index and the occurrence of IDB^[24]. Notably, Childbirth-Associated Breast Infections usually caused by bacterial infection, unlike IDB. The main treatment modalities for IDB was GCs, while anti-infective was the main therapeutic modalities for the treatment of Childbirth-Associated Breast Infections^[25]. Utilizing Mendelian randomization, we have established a causal relationship between GCs and IDB, while excluding a correlation with childbirth-associated breast infections. This suggests that GCs' contribution to IDB is more likely associated with immune system abnormalities rather than infections.

IGM and PCM were two types of IDB, both exhibiting infiltration of various immune cells, including macrophages^[26]. Our study found that GRs are associated with the activation of macrophages. GRs, as crucial nuclear transcription factors within cells, regulated the expression of genes related to growth, development, metabolism, and immune response, thereby influencing a wide range of biological activities^{[27; 28]}. ScRNA-Seq showed that macrophages in breast tissue exhibit a higher enrichment of GRs compared to other immune cells. This suggests that macrophages within breast tissue may be more sensitive to the application of GCs. Consistent with our hypothesis, MR confirms a causal relationship between GCs and the occurrence of IDB. These findings indicate that GCs may promote the development of IDB by acting on the increased GRs present on macrophages within breast tissue.

Glucocorticoids have been extensively utilized in clinical practice as anti-inflammatory drugs^[29]. Research indicated that GCs facilitate substantial metabolic reorganization in lipopolysaccharide (LPS)-activated macrophages, modulating the metabolic pathways of these LPS-activated macrophages. This further alters the metabolic state of the macrophages, thereby inhibiting the onset of inflammation^[30]. However, this study has found that GCs have a pro-inflammatory effect, which appears to contradict previous conclusions. In fact, this is not the case, as existing research has suggested that GCs may play a role in promoting the development of inflammation. Macrophages have pro-inflammatory (M1) and anti-inflammatory (M2) effects, which are involved in the different pathological outcome^[31]. M1 macrophages exert their pro-inflammatory effects by activating their own oxidase system, leading to the production of reactive oxygen species^[32]. Previous research had found that the activation of GRs could promote the polarization of M1 macrophages in vitro. The deletion of GRs in macrophages and the pharmacological inhibition of GRs can ameliorate skin inflammation^[33]. Inflammation is a key factor in the pathogenesis of glucocorticoid-associated osteonecrosis of the femoral head, with a significant increase observed in the number of M1 macrophages in the femoral head of GA-ONFH patients^[34]. Therefore, the promotion of IDB by GCs may be due to their reshaping of the normal immune microenvironment. The pharmacological inhibition of GRs may represent a promising approach to prevent the occurrence of IDB

To our knowledge, this is the first MR study to explore the positive causal relationships between GCs and IDB. Our study has demonstrated a causal relationship between GCs and the occurrence of IDB, suggesting that GCs may promote the development of IDB by acting on the increased GRs in macrophages within breast tissue. Overall, our research selected instrumental variables based on strict criteria and ensured that the F-statistics for all instrumental variables exceeded the threshold of 10, thereby eliminating potential biases from weak instruments and meeting the relevance assumptions required for Mendelian randomization studies. Secondly, we excluded SNPs unrelated to any known or unknown confounding factors, thus satisfying the independence assumption required for Mendelian studies. Lastly, we ruled out the pleiotropy assumption through heterogeneity tests and the MR-Egger test. These measures ensured the rigor and reliability of our study. However, there are some limitations to our research. Firstly, due to the lack of GWAS data from non-European populations, our study only discusses the role of GCs in IDB among European populations. However, given the varying prevalence of GCs and IDB across different ethnic groups, the results may differ. Additionally, the sample size included in our study is relatively small, which could lead to estimation biases. Finally, although our research provides insights into the etiology of GCs in IDB, due to the lack of detailed pathological and clinical data on IDB, our analysis is limited to statistical data from large datasets, and the specific biological mechanisms await further exploration by future researchers.

In summary, our study provided genetic evidence indicating a significant association between GCs and IDB in European. This finding underscored the clinical importance of controlling the use of GCs. Furthermore, the occurrence of IDB due to GCs is associated with the high expression of GRs in macrophages, which represents a promising therapeutic target for IDB. Additionally, further basic and clinical experiments are required to validate the relationship between GCs and IDB.

Data availability statement

The GC datasets generated and/or analysed during the current study are available in the https://r11.finngen.fi/. The GRs datasets generated and/or analysed during the current study are available in the https://gwas.mrcieu.ac.uk/.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Ethics statement

This MR research utilized only published or publicly available GWAS data. Each participant received ethical approval and informed consent for the respective study, as detailed in the original publication and consortium.

Authorship contribution statement

Shuran Chen , Lei Xue and Wencan Yang: Writing – original draft, Conceptualization, Formal analysis, Methodology, Software.

Wencan Yang, Lixiang Li and Zhenglang Yin: Investigation, Validation, Visualization.

Declaration of competing interest

None declared.

Acknowledgements

We express our gratitude to all the genetics consortiums for making the GWAS summary data publicly available. We express our gratitude to Cellxgene platform and Anhui Huaxiao Technology Co. for their technical support.

Scott, D. M. Inflammatory diseases of the breast[J]. Best Pract. Res. Clin. Obstet. Gynaecol. 83, 72–87 (2022).
Omranipour, R., Vasigh, M., Mastitis & Breast Abscess, and Granulomatous Mastitis[J]. Adv. Exp. Med. Biol. 1252, 53–61 (2020).
Yin, Y. et al. Idiopathic Granulomatous Mastitis: Etiology, Clinical Manifestation, Diagnosis and Treatment[J]. J. Invest. Surg. 35, 709–720 (2022).
Lim, H. Y. et al. Glucocorticoids exert opposing effects on macrophage function dependent on their concentration[J]. Immunology. 122, 47–53 (2007).
Frank, M. G. et al. Prior exposure to glucocorticoids sensitizes the neuroinflammatory and peripheral inflammatory responses to E. coli lipopolysaccharide[J]. Brain Behav. Immun. 24, 19–30 (2010).
Olivas-Aguirre, M. et al. Dexamethasone-Induced Fatty Acid Oxidation and Autophagy/Mitophagy Are Essential for T-ALL Glucocorticoid Resistance[J]. (2023). Cancers (Basel), 15 .
Burwick, N. & Sharma, S. Glucocorticoids in multiple myeloma: past, present, and future[J]. Ann. Hematol. 98, 19–28 (2019).
Clarisse, D. et al. Crosstalk between glucocorticoid and mineralocorticoid receptors boosts glucocorticoid-induced killing of multiple myeloma cells[J]. Cell. Mol. Life Sci. 80, 249 (2023).
Davey Smith, G. & Ebrahim, S. What can mendelian randomisation tell us about modifiable behavioural and environmental exposures?[J]. Bmj. 330, 1076–1079 (2005).
Nitsch, D. et al. Limits to causal inference based on Mendelian randomization: a comparison with randomized controlled trials[J]. Am. J. Epidemiol. 163, 397–403 (2006).
Skrivankova, V. W. et al. Strengthening the Reporting of Observational Studies in Epidemiology Using Mendelian Randomization: The STROBE-MR Statement[J]. Jama. 326, 1614–1621 (2021).
Davey Smith, G. & Hemani, G. Mendelian randomization: genetic anchors for causal inference in epidemiological studies[J]. Hum. Mol. Genet. 23, R89–98 (2014).
Zheng, J. et al. Recent Developments in Mendelian Randomization Studies[J]. Curr. Epidemiol. Rep., 4, 330–345. (2017).
Spiga, F. et al. Tools for assessing quality and risk of bias in Mendelian randomization studies: a systematic review[J]. Int. J. Epidemiol. 52, 227–249 (2023).
Birney, E. Mendelian Randomization[J]. Cold Spring Harb Perspect. Med., 12. (2022).
Sanderson, E. Multivariable Mendelian Randomization and Mediation[J]. Cold Spring Harb Perspect. Med., 11. (2021).
Sakaue, S. et al. A cross-population atlas of genetic associations for 220 human phenotypes[J]. Nat. Genet. 53, 1415–1424 (2021).
Burgess, S., Butterworth, A. & Thompson, S. G. Mendelian randomization analysis with multiple genetic variants using summarized data[J]. Genet. Epidemiol., 37, 658–665. (2013).
Bowden, J., Davey Smith, G. & Burgess, S. Mendelian randomization with invalid instruments: effect estimation and bias detection through Egger regression[J]. Int. J. Epidemiol. 44, 512–525 (2015).
Hemani, G. et al. The MR-Base platform supports systematic causal inference across the human phenome[J]. Elife, 7. (2018).
Burgess, S. et al. Guidelines for performing Mendelian randomization investigations: update for summer 2023[J]. Wellcome Open. Res. 4, 186 (2019).
Subramanian, A. et al. Gene set enrichment analysis: a knowledge-based approach for interpreting genome-wide expression profiles[J]. Proc. Natl. Acad. Sci. U S A. 102, 15545–15550 (2005).
Zhou, Y. et al. Metascape provides a biologist-oriented resource for the analysis of systems-level datasets[J]. Nat. Commun. 10, 1523 (2019).
Wei, C. et al. Body mass index and risk of inflammatory breast disease: a Mendelian randomization study[J]. Nutr. Hosp. 41, 96–111 (2024).
Koziol, K. J. et al. Predictive Risk Factors for Childbirth-Associated Breast Infections in the United States: A 10-Year Perspective[J]. Int. J. Environ. Res. Public. Health, 20. (2023).
Dilaveri, C. et al. Idiopathic Granulomatous Mastitis[J]. Breast J, 2024, 6693720. (2024).
McGowan, P. O. et al. Epigenetic regulation of the glucocorticoid receptor in human brain associates with childhood abuse[J]. Nat. Neurosci. 12, 342–348 (2009).
Weikum, E. R. et al. Glucocorticoid receptor control of transcription: precision and plasticity via allostery[J]. Nat. Rev. Mol. Cell. Biol. 18, 159–174 (2017).
Scherholz, M. L., Schlesinger, N. & Androulakis, I. P. Chronopharmacology of glucocorticoids[J]. Adv. Drug Deliv Rev., 151–152, 245–261. (2019).
Auger, J. P. et al. Metabolic rewiring promotes anti-inflammatory effects of glucocorticoids[J]. Nature. 629, 184–192 (2024).
Gauthier, T. & Chen, W. Modulation of Macrophage Immunometabolism: A New Approach to Fight Infections[J]. Front. Immunol. 13, 780839 (2022).
Wang, C. et al. Macrophage Polarization and Its Role in Liver Disease[J]. Front. Immunol. 12, 803037 (2021).
Ge, H. et al. Stress aggravates imiquimod-induced psoriasiform inflammation by promoting M1 macrophage polarization[J]. Int. Immunopharmacol. 124, 110899 (2023).
Cheng, Y. et al. Early depletion of M1 macrophages retards the progression of glucocorticoid-associated osteonecrosis of the femoral head[J]. Int. Immunopharmacol. 122, 110639 (2023).

No competing interests reported.

Download PDF

Reviewers agreed at journal
15 Oct, 2024
Reviewers agreed at journal
14 Oct, 2024
Reviewers agreed at journal
12 Oct, 2024
Reviewers invited by journal
10 Oct, 2024
Editor assigned by journal
08 Oct, 2024
Editor invited by journal
22 Aug, 2024
Submission checks completed at journal
20 Aug, 2024
First submitted to journal
20 Jul, 2024

You are reading this latest preprint version

Genome-wide Mendelian randomization and single-cell RNA sequencing analyses identify the causal effects of glucocorticoids on inflammatory disorders of breast

Status:

Version 1

Abstract

Background

Methods

Results

Conclusions

Figures

Introduction

Materials and methods

Study design

Genetic IV selection

GWAS data for GCs, IDB, and Childbirth-Associated Breast Infections

Statistical analysis

Enrichment analysis and single cell RNA-sequencing (scRNA-Seq) analysis

Results

Genetic instruments for GCs

MR estimates and sensitivity analyses

The functions and distribution of GRs in breast tissue

Discussion

Conclusions

Declarations

Data availability statement

Funding

Ethics statement

Authorship contribution statement

Declaration of competing interest

Acknowledgements

References

Additional Declarations

Supplementary Files

Status:

Version 1