Introduction

Sepsis is a life-threatening condition characterized by organ dysfunction due to a dysregulated immune response to an infection1. Current research efforts are directed towards identifying key proteins that decisively influence clinical outcomes, potentially guiding therapeutic interventions towards recovery or mitigating mortality2,3,4. One important system that modulates inflammation in sepsis is the glucocorticoid system. Within this system, polymorphisms in key proteins appear to influence the response to glucocorticoid therapy. In this context, we have previously investigated single nucleotide polymorphism (SNPs) in glucocorticoid-induced leucine zipper (GILZ)6,7 and NF-κB5, both of which were associated with significant effects on treatment response. However, the central protein of this system, the glucocorticoid receptor (GR), has received relatively little attention5. Given its pivotal role in the immune response to sepsis8, a deeper understanding of genetic variants affecting GR function is warranted while it regulates a wide array of pro- and anti-inflammatory signaling pathways6. The receptor exists in two primary isoforms: the alpha isoform (GRα) and the beta isoform (GRβ)7. GRα generally exerts anti-inflammatory effects by inducing transcription of genes such as GILZ and monocyte chemotactic protein (MCP-1), and by inhibiting the activation of pro-inflammatory mediators such as the p65 subunit of NFκB8,9. In contrast, GRβ tends to enhance inflammatory responses due to its antagonistic interaction with GRα10. Considerable attention has been paid to genetic variations within the NR3C1 gene, in particular the SNP rs6198 11,12,13,14. This SNP has been associated with adverse outcomes in immune-mediated diseases such as rheumatoid arthritis and post-traumatic stress disorder, mainly through its influence on mRNA stability, which favors the pro-inflammatory GRβ isoform15,16. However, the impact of rs6198 SNP within NR3C1 in sepsis, specifically its effects on the differential expression of GRα versus GRβ, and consequently on sepsis outcomes, has not been investigated. Therefore, we tested the hypothesis that the presence of a particular genotype of the rs6198 SNP is associated with an increased 30-day mortality rate in patients with sepsis.

Materials and methods

Study design and conceptual overview

204 patients from the prospective, multicenter Sepsis.Data.Net NRW cohort were included. Patients were recruited consecutively between 2018 and 2020. Systematic screening ensured that all eligible ICU patients were considered for inclusion (see supplementary file 1). Written consent was obtained from all patients or their legal guardians. This study was approved by the Ethics Committee of the Medical Faculty of the Ruhr-University of Bochum (Registration no. 19–6606 3-BR). All research involving human participants was conducted in full accordance with the Declaration of Helsinki, as well as institutional and national guidelines and regulations. The study strictly adhered to the protocols outlined in the approved ethics vote. Intensive care patients aged 18 and older were eligible for recruitment if they met the current Sepsis-3 criteria for sepsis diagnosis1. To enhance the generalizability of our findings and account for the heterogeneity of sepsis progression, the study protocol allowed for patient inclusion within 48 h of sepsis diagnosis. This ensured that patients initially treated on the general ward before ICU transfer were not systematically excluded. For patients diagnosed with sepsis upon ICU admission, enrollment and sample collection were performed immediately to capture the earliest possible disease stage. Treatment of patients was carried out according to the current national and international guidelines and was not influenced by participation in the study17. Hydrocortisone therapy was defined as 200 mg/24 h via infusion pump for at least 24 h, according to guidelines17. Blood samples for DNA, RNA, and serum analysis were collected within the first 48 h after diagnosis and stored at − 80 °C after initial processing.

DNA genotyping

DNA was isolated from EDTA-blood samples using the my-Budget Blood DNA Midi Kit (Bio-Budget Technologies GmbH, Krefeld, Germany) according to the manufacturer’s instructions as previously described18. Genotyping was performed using the Thermo Fisher Scientific TaqMan® SNP Genotyping Assay (Thermo Fisher Scientific, Wilmington, USA) and Bio-Rad CFX Connect Cycler Systems (Bio-Rad Laboratories, Inc., Hercules, USA) using a protocol of 95 °C for 10 min and 40 cycles of 95 °C for 15 s followed by 60 °C for 60 s.

RNA analysis

RNA was extracted from whole blood collected with Tempus™ Blood RNA Tubes (Applied Biosystems, Waltham, USA) using Tempus™ Spin RNA Isolation Reagent Kits (Applied Biosystems, Waltham, USA), followed by complementary DNA (cDNA) synthesis using the High-Capacity cDNA Reverse Transcription Kit by Applied Biosystems (Applied Biosystems, Waltham, USA). Then, quantitative polymerase chain reaction was performed using our primers listed in supplementary file 2 for expression analysis of total GR, GRα and GRβ in relation to ACTB, as proposed19. The protocol used with GoTaq® qPCR MasterMix (Promega, Madison, USA) involved 2 min of 95 °C followed by 40 cycles of 95 °C for 15 s and 60 °C for 60 s.

Cortisol and GILZ, MCP-1 concentrations

Serum samples were analyzed using enzyme-linked immunosorbent assay (ELISA) kits to determine cortisol, MCP-1, and GILZ levels (Enzo Life Sciences, Inc., Farmingdale, USA; Antibodies-online GmbH, Aachen, Germany). Samples were diluted as necessary to fall within the standard detection range of each kit. Blood samples were consistently drawn at 6 AM to account for potential circadian variations in cortisol concentrations.

Cytokine measurements

The concentrations of various cytokines were quantified on day one using a customized human LegendPlex assay (BioLegend, San Diego, CA). The cytokines measured included Interleukin-1 beta (IL-1β), Interleukin-6 (IL-6), Interleukin-10 (IL-10), Interleukin-12 (IL-12), Interleukin-18 (IL-18), Interleukin-23 (IL-23), tumor necrosis factor alpha (TNF-α), Interferon alpha2 (INF-α2), and Interferon gamma (INF-γ),

Cytospin and immunofluorescence

Peripheral blood mononuclear cells (PBMCs) from septic patients were collected on day one of sepsis and centrifuged onto glass slides using a Cellspin device (Tharmac, Limburg, Germany), fixed with 4% formaldehyde (F1635, Sigma Aldrich, Taufkirchen, Germany) and permeabilised with 0.1% (v/v) Triton-X100 (T8787, Sigma Aldrich) and 0.1% (m/v) sodium dodecyl sulfate (SDS) (0183.1, Carl Roth, Karlsruhe, Germany). Non-specific binding sites were blocked using Duolink® blocking solution (DUP82007, Sigma Aldrich). The slides were then incubated overnight at 4 °C with a mouse anti-human GR antibody (Glucocorticoid Receptor (D4X9S) Mouse mAb #47,411). After washing, the slides were incubated for 1 h at 37 °C with goat anti-mouse IgG AlexaFluor® 488 (1:400 in PBS, ab150077, Abcam, Cambridge, UK). Counterstaining was performed using SlowFade™ Gold Antifade Mountant with DAPI (S36939, Invitrogen) and phalloidin solution. Fluorescence microscopy was carried out using an Olympus IX51 microscope (Olympus, Hamburg, Germany). All samples were imaged and analyzed under consistent settings using Fiji ImageJ software.

Statistical analysis

Continuous variables are presented as means ± standard deviation (SD) when normally distributed and as medians with interquartile ranges (IQR; 25th to 75th percentile) for distributions that are not normal. Differences between groups for continuous data were determined using the t-test or the Wilcoxon rank-sum test, depending on the distribution of the data. For categorical variables, differences between groups were evaluated using either the Chi-square test or Fisher’s exact test, as appropriate. The distribution of genotypes were tested for Hardy–Weinberg equilibrium to ensure that genetic variation was consistent with expected frequencies using a chi-squared test. Survival curves were constructed using the Kaplan–Meier method to estimate the time-dependent probability of survival. The log-rank test was applied to statistically compare these curves between the different groups and to identify significant differences in survival time. For the multivariable regression analysis, we applied a structured feature selection approach to ensure robustness and prevent overfitting. Univariate analysis was first conducted on the rs6198 SNP, key clinical variables (SOFA/SAPS II scores), and laboratory markers. Statistically significant variables were considered for the final model, with collinearity addressed by selecting the strongest predictor among correlated variables. To adhere to standard guidelines, we included a maximum of six variables for our 204-patient cohort. The full univariate analysis is provided in the supplementary material (see supplementary file 3). Multivariate Cox regression analysis was used to evaluate the influence of various predictors on survival, adjusting for potential confounders and quantifying hazard ratios. We assessed interaction effects within a generalized linear model by including interaction terms between the rs6198 TT genotype and both SOFA Score ≥ 9 and hydrocortisone therapy to evaluate whether the association with 30-day mortality varied across these subgroups. To assess outcomes in patients with more severe illness, we defined a high-severity subgroup based on a SOFA score ≥ 9. This threshold was chosen as it exceeds the median SOFA score of 8 in our cohort, allowing us to focus on patients with more pronounced organ dysfunction. Unlike septic shock alone, which is already partly reflected in the cardiovascular SOFA subcomponent, this approach provides a broader measure of disease severity, incorporating dysfunction across multiple organ systems. A p-value of less than 0.05 was deemed to indicate statistical significance. Confidence intervals (CIs) were set at 95% coverage. All statistical analyses were conducted using R software (version 3.5.3; The R Foundation for Statistical Computing; http://www.R-project.org).

Results

Baseline characteristics

In all 204 patients, genotyping of the rs6198 NR3C1 was performed. A total of 137 patients exhibited the TT-genotype. The genotype distribution was in Hardy–Weinberg equilibrium (p = 0.892) and the less frequent CC-genotype (n = 7) was pooled with the heterozygous CT-genotype (n = 60) for further analyses. There was no significant difference between the TT and CC/CT genotype groups in terms of gender, age and SOFA score at enrollment, as shown in Table 1. Similarly, no significant differences were found in terms of infection focus, length of stay in the ICU. Regarding laboratory variables at admission, the CC/CT-genotype group exhibited higher bilirubin concentrations (0.77 mg/dL, IQR: 0.41–1.64) compared to the TT-genotype group (0.52 mg/dL, IQR: 0.32–0.91, p = 0.043).

Table 1 Cohort description, classification according to rs6198 SNP TT and CC/CT-genotype.
Full size table

The rs6198 SNP association with 30-day survival

Carriers of the TT-genotype demonstrated a 30-day survival rate of 65% (89 out of 137 patients). In contrast, individuals with CC/CT genotype had a significantly higher 30-day survival rate of 82% (55 out of 67 patients, p = 0.006), as illustrated in Fig. 1.

Fig. 1
figure 1

Kaplan–Meier Survival Analysis of the entire Cohort (n = 204 patients) based on rs6198 SNP genotypes. This figure displays the survival curves for patients stratified by TT and CC/CT genotypes of the rs6198 SNP. Patients with the TT-genotype exhibit significantly higher 30-day mortality compared to those with the CC/CT-genotype.

Full size image

In the multivariate Cox regression analysis, we adjusted the rs6198 genotype for additional potential confounding variables such as age and the SOFA score on day one. Further adjustments were made for the administration of hydrocortisone during the ICU stay, serum bilirubin concentrations, which differed significantly between the groups (see above) and serum lactate concentrations. The TT-genotype remained a significant, independent risk factor, with a hazard ratio of 3.56 (95% CI 1.22–10.38, p = 0.02), as demonstrated in Table 2.

Table 2 Cox regression analysis for 30d mortality of septic patients (n = 204).
Full size table

The interaction analysis revealed a significant modification of the association between the rs6198 TT genotype and 30-day mortality by SOFA Score ≥ 9 (HR: 1.43, p = 0.003) and hydrocortisone therapy (HR: 1.19, p = 0.021), as detailed in supplementary file 4. These effects were further explored within the following subgroup analysis.

The influence of SNP rs6198 depending on disease severity

To analyze the extent to which the described effects impact 30-day survival in patients with higher disease severity (i.e., presence of septic shock within 24 h of study enrollment or a SOFA score of ≥ 9 at study entry), we conducted a subgroup analysis. Of the 204 patients, 94 fell into this subgroup (46%) and their baseline characteristics are available in supplementary file 5. The association of TT-genotype with a higher mortality rate was even more pronounced in this subgroup with greater disease severity, as shown in Fig. 2. Performing the previously described multivariate Cox regression in this subgroup, the presence of a TT-genotype resulted in a hazard ratio of 6.16 (95% CI 1.66–22.80, p = 0.007; see supplementary File 6 for the complete analysis). No significant effects were observed in the corresponding subgroup (i.e. SOFA score < 9 and no septic shock) (see supplementary File 7).

Fig. 2
figure 2

Kaplan–Meier Survival Analysis for the Subgroup with Septic Shock or SOFA Score ≥ 9 at Enrollment (n = 94) Based on rs6198 SNP Genotypes. This figure presents survival curves for patients with higher disease severity, stratified by TT and CC/CT genotypes of the rs6198 SNP. The increased 30-day mortality associated with the TT genotype observed in the entire cohort (see Fig. 1) is more pronounced in this subgroup analysis.

Full size image

The influence of SNP rs6198 in relation to hydrocortisone therapy

Within the entire cohort, 53 patients (26%; TT-genotype 40 out of 53 patients vs. CC/CT-genotype 13 out of 53 patients) received hydrocortisone therapy during their sepsis management in the intensive care unit. The mortality rate among all patients undergoing hydrocortisone therapy was 43% (23 out of 53 patients), and thus significantly higher than in patients who did not receive hydrocortisone therapy (25%, 38 out of 151 patients, p = 0.006). In patients treated with hydrocortisone, no statistically significant association was observed between SNP rs6198 and 30-day mortality (TT-genotype 48%, 19 out of 40 patients vs. CC/CT-genotype 31%, 4 out of 13 patients, p = 0.34). In contrast, among patients not receiving hydrocortisone therapy, the TT-genotype was associated with a significantly higher mortality rate (TT-genotype 31%, 30 out of 97 patients vs. CC/CT-genotype 15%, 8 out of 54 patients, p = 0.017).

The rs6198 SNP association with GR expression

We conducted a qualitative assessment of GR protein expression In the immunofluorescence staining of PBMCs from septic patients. Notably, individuals with TT genotypes showed an enhanced GR expression signal, as illustrated in Fig. 3A. However, subsequent RNA analysis revealed no significant differences in overall GR expression between the TT and CC/CT genotypes (p = 0.220), as shown in Fig. 3B. Further analyses were performed separately for the alpha and beta isoforms of GR. The TT-genotype was significantly associated with increased expression of GRα (p = 0.023), while GRβ expression showed no significant differences (p = 0.330). We also investigated the impact of GR expression on 30-day mortality using COX- regression. Increased GRα expression was significantly associated with higher 30-day mortality (HR 2.38, 95% CI 1.48–3.82, p < 0.001). In contrast, for GRβ (HR 1.00, 95% CI 1.00–1.00, p = 0.211) and overall GR expression (HR 1.21, 95% CI 0.81–1.83, p = 0.352), no association with 30-day mortality was observed.

Fig. 3
figure 3

Qualitative and Quantitative Expression Analysis of GR and Its Isoforms Alpha and Beta in PBMCs of Septic Patients. A Immunofluorescence depiction of overall GR protein (green dots) alongside actin filaments, stained with phalloidin (red). B Quantitative mRNA expression of GR and its isoforms, alpha and beta. While no significant differences in overall GR expression are observed between patients possessing TT and CC/CT genotypes of the rs6198 SNP, there is a significant increase in GRα expression associated with the TT genotype.

Full size image

Analysis of downstream proteins GILZ and MCP-1, serum cortisol levels and cytokines

As indicated in Table 3, there were no significant differences in the expression levels of the downstream proteins MCP-1 (p = 0.449) and GILZ (p = 0.115) between the TT-genotype and the CC/CT genotype, when considering the entire cohort. Additionally, the overall serum cortisol concentrations did not demostrate significant differences between these subgroups (p = 0.733). Similarly, cytokine measurements showed no significant differences between the TT genotype and the CC/CT genotypes (all p-values exceeding 0.05). However, again when examining the subgroup defined by higher disease severity—in particular, those with septic shock or a SOFA score ≥ 9 —the TT genotype was associated with significantly higher concentrations of the cytokines IL-12, IL-23, and IFN-γ (each p-values below 0.05).

Table 3 Expression analysis of downstream proteins, cortisol and cytokines.
Full size table

Discussion

Our study demonstrates that the presence of the NR3C1 rs6198 SNP influences survival in sepsis. In particular, the presence of the TT-genotype is associated with significantly higher mortality, especially in patients with initially severe disease states.

Upon genotyping our cohort, we observed a TT to CC/CT ratio of 2:1, which is consistent with previous literature12,15. At baseline, there were no other significant baseline characteristic differences or disease severity factors at study entry that could have influenced the outcomes. Nevertheless, both the Kaplan–Meier analysis and the Cox regression, adjusted for potential confounders, confirmed the TT-genotype as an independent risk factor for increased mortality. This raises questions about the underlying mechanisms involved.

First, we investigated whether the TT genotype influences GR expression. Initially, immunofluorescence suggested an increased overall expression of GR. Unfortunately, our subsequent quantitative analyses did not achieve statistical significance, possibly due to the small sample size. However, the situation changed when we examined the GR isoforms separately. In this analysis, carriers of the TT genotype exhibited significantly higher expression of the alpha isoform. As previously noted, the alpha isoform is crucial for the anti-inflammatory effects of cortisol5,20. In our cohort, increased expression of GRα was significantly associated with higher mortality, suggesting that the rs6198 SNP may influence sepsis survival through this mechanism. However, the literature shows heterogeneous results regarding GR isoforms, particularly in studies involving critically ill children5,21,22,23. Further confirmatory studies are needed to clarify the impact of increased GRα expression.

Second, we observed enhanced mortality effects of the TT-genotype in a subgroup with particularly high disease severity, such as septic shock. This group is inherently vulnerable and characterized by high mortality rates24. Given that guideline-adherent hydrocortisone therapy may be considered in this subgroup, we intentionally focused our analysis on these patients.17. We specifically examined whether hydrocortisone therapy differentially impacts carriers of the TT versus CC/CT-genotype. In patients receiving hydrocortisone therapy, no significant differences regarding 30-day mortality were observed. Conversely, in the subgroup without hydrocortisone, TT-genotype carriers died significantly more frequently than CC/CT carriers. Given our study’s observational design, it remains an open question whether TT-genotype would have benefited from hydrocortisone therapy in sepsis25. Further research is needed to determine if this could be an avenue for personalized therapy based on patients’ genotype26.

Third, our work examines the rs6198 SNP’s influence on downstream proteins such as GILZ, MCP-1 and various cytokines and examines whether serum cortisol itself could be responsible for the differences. Here, we found no significant group differences concerning TT or CC/CT-genotype, when considering the entire cohort suggesting that the observed effects might be explained by other mechanisms. However, a distinct pattern emerged when analyzing the above described subgroup characterized by high disease severity. In this subgroup, in addition to a more pronounced association between the TT genotype and increased 30-day mortality, significant increases in IL-12, IL-23, and IFN-γ were also observed. IL-12 plays a crucial role in the differentiation of naive T cells into Th1 cells27. It serves as a T cell activator, facilitating the proliferation and activation of T cells. More specifically, IL-12 induces the secretion of INF-γ and TNF-α from T cells and natural killer cells, while concurrently countering the inhibitory impact of IL-4 on IFN-γ production28. Similarly, IL-23 stimulates the production of IFN-γ and enhances the proliferation of activated T cells29. These actions highlight the essential roles of IL-12 and IL-23 in bolstering T-cell mediated immune responses.

Considering these cytokine profiles alongside our genetic findings, it appears that the rs6198 SNP, by influencing higher levels of GRα, especially in critically ill patients, correlates with enhanced cytokine release and augmented T-cell function. This association suggests that GRα may exert a modulatory role on T-cell responses, potentially impacting the efficacy of immune responses. Such findings emphasize the intricate interplay between genetic variations and immune system regulation, offering potential insights into tailored therapeutic approaches based on genetic predispositions.

Limitations

Our study has several limitations. Although our cohort was multicentric and recruited prospectively, the observational study design does not establish causality between the presence of the TT-genotype and increased mortality in sepsis. Moreover, it is beyond the scope of our study to elucidate all potential effects of the rs6198 SNP. Future research investigating GR receptor activity, further downstream proteins in their varying expressions, and longitudinal measurements during sepsis may provide a more comprehensive understanding of the rs6198 SNP’s impact on sepsis survival. Additionally, while the overall cohort size was sufficient for the primary analysis, the sample size within subgroups was smaller, which may have limited statistical power to detect more subtle but potentially clinically relevant associations. This increases the risk of type II errors and necessitates cautious interpretation of the interaction effects observed.

Conclusion

The TT genotype of rs6198 SNP in the NR3C1 gene is associated with higher 30-day mortality in sepsis patients and increased expression of the GRα isoform. Further studies are required to clarify the causal mechanisms and to determine if this represents a potential marker for a personalized therapy in sepsis.