https://orcid.org/0000-0003-0997-6272
Figures
Abstract
Background
Kidney dysfunction is a common complication in septic patients. Studies have identified numerous risk factors for sepsis-associated acute kidney injury (S-AKI), yet there is wide variability in the incidence even among patients with similar risk factors, suggesting the presence of additional uncharacterized risk factors, including genetic differences. The expansion of biobanks, advances in genotyping, and standardized diagnostic criteria have enabled large-scale, hypothesis-generating studies into the genetic mechanisms underlying S-AKI. We hypothesize that the genetic pathway behind S-AKI has overlapping mechanisms with key differences based upon the specific subtype of acute kidney injury (AKI).
Methods
To test this hypothesis, we performed a genome-wide association study (GWAS) of S-AKI in three logistic regression models. Model 1, controlled for 1) age, 2) sex, 3) genotyping chip, and 4) the first five principal components. In Model 2, pre-sepsis baseline serum creatinine was added to the variables in Model 1. Finally, in Model 3, we controlled for the full range of patient, clinical, and ICU-related risk factors. Each of the 3-models were repeated in a pre-specified sensitivity analysis of higher severity S-AKI, defined as KDIGO Stage 2 or 3. We then compare associated variants and genes from our GWAS with previously published AKI sub-types and model other factors associated with S-AKI in our dataset.
Findings
3,348 qualifying Sepsis-3 patients have been genotyped in our dataset. Of these patients, 383 (11.4%) developed Stage 1, 2, or 3 AKI (primary outcome) and 181 (5.4%) developed Stage 2 or 3 AKI (sensitivity analysis). The median age was 61 years (interquartile range (IQR): 51,69), 42% were female, and the increase in SOFA score (between 48-hours before to 24-hours after the onset of suspected infection) was 2 (2–3). No variants exceeded our threshold for genome-wide significance (P<5x10-8), however, a total of 13 variants exceeded the suggestive (P<1x10-6) threshold. Notably, rs184516290 (chr1:199814965:G:A), near the NR5A2 gene, chr1:199805801:T:TA, also near the NR5A2 gene, and rs117313146 (chr15:31999784:G:C), near the CHRNA7 gene, were associated with S-AKI at the suggestive level in all three models presented. Variants in the suppressor of fused homolog (SUFU) gene, previously shown to be correlated with renal function in bacteremic patients, consistently exceeded the P<0.05 threshold in our models.
Conclusions
While failing to identify any novel association for S-AKI at the level of genome-wide significance, our study did suggest multiple variants in previously characterized pathways for S-AKI including CHRNA7, NR5A2, and SUFU. We failed to replicate associations from multiple prior studies which may result from differences in how the phenotype was defined or, alternatively, limited genetic contribution and low heritability.
Citation: Douville NJ, Bastarache L, Bertucci-Richter E, Patil S, Jewell ES, Freundlich RE, et al. (2024) Genetic variants associated with sepsis-associated acute kidney injury. PLoS ONE 19(12): e0311318. https://doi.org/10.1371/journal.pone.0311318
Editor: Tarek Samy Abdelaziz, Cairo University Kasr Alainy Faculty of Medicine, EGYPT
Received: May 17, 2024; Accepted: September 17, 2024; Published: December 5, 2024
Copyright: © 2024 Douville et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: GWAS Summary Statistics have been made publicly available (uploaded to the NHGRI-EBI Catalog of Human Genome-Wide Association Studies). In accordance with the PLOS ONE open data sharing policy, GWAS Summary Statistics are available for download from the NHGRI-EBI Catalog (https://www.ebi.ac.uk/gwas/). Model 1 data (GCST90451676) is available here (https://www.ebi.ac.uk/gwas/studies/GCST90451676). Model 2 data (GCST90451677) is available here (https://www.ebi.ac.uk/gwas/studies/GCST90451677). Model 3 data (GCST90451678) is available here (https://www.ebi.ac.uk/gwas/studies/GCST90451678). Individual patient level data cannot be shared publicly because of institutional rules, however, de-identified (non-genetic) data on the sepsis cohort will be available for researchers who meet the criteria for access to confidential data following activation of necessary data use agreement. Please contact University of Michigan, Precision Health (phdatahelp@umich.edu) with requests to share de-identified data and additional summary statistics (beyond those uploaded to the NHGRI-EBI Catalog).
Funding: All work and partial funding was attributed to the Department of Anesthesiology, University of Michigan Medical School (Ann Arbor, Michigan, USA) and the Department of Anesthesiology, Vanderbilt University Medical Center (Nashville, Tennessee, USA). NJD received support from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) (1 K08 DK131346-02) and a Foundation for Anesthesia Education and Research (FAER) - Mentored Research Training Grant (MRTG). LB received funding from the National Library of Medicine (R01- LM010685-11). REF received funding from the National Heart Lung and Blood Institute (K23 HL148640). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have declared that no competing interests exist.
Abbreviations: AKI, Acute Kidney Injury; GWAS, Genome-wide association study; MGI, Michigan Genomics Initiative; MPOG, Multicenter Perioperative Outcomes Group; S-AKI, Sepsis-associated AKI
Background
Sepsis is a life-threatening organ dysfunction resulting from dysregulated response to infection, [1] and despite decades of research remains a leading causes of morbidity and death in hospitalized patients [2]. Kidney dysfunction occurs in 30–50% of septic patients [3–5] leading to sepsis-associated AKI (S-AKI) [6]. S-AKI is associated with increased hospital stay, greater resource consumption, and increased mortality, [7–10] even in less severe cases of S-AKI not requiring dialysis [5].
Studies have identified numerous risk factors for S-AKI, including male sex, older age, comorbidities including hypertension and diabetes, source and type of infection, and angiotensin converting enzyme inhibitors or angiotensin receptor antagonists use [11–13]. More severe sepsis, as marked by shock, receipt of vasopressors, or mechanical ventilation, is also a risk factor for S-AKI [12, 13]. Yet there is wide variability in the incidence of AKI, even among patients with similar risk factors, suggesting the presence of additional uncharacterized risk factors, such as genetic differences [14]. The genetics of AKI had been traditionally studied in a hypothesis-driven manner, testing a limited number of candidate polymorphisms [15]. The expansion of biobanks and recent advances in genotyping have enabled large-scale, hypothesis-generating studies into the genetic mechanisms underlying AKI phenotypes, including surgical [16, 17] and intensive care populations [18]. Furthermore, standardized definitions for both sepsis (Sepsis-3) [1] and AKI (Kidney Disease: Improving Global Outcomes) [19] improve the breadth and generalizability of phenotypes that can be curated and studied. A recent study, characterizing genetic differences resulting from transition between the Sepsis-2 to Sepsis-3 diagnostic criteria, showed 16% of variants remained whether either definition was employed, while 49% and 35% were isolated in the Sepis-2 and Sepsis-3 cohorts, respectively [20]. Large-scale, unbiased investigations on S-AKI using standardized definitions for sepsis (Sepsis-3) and AKI (KDIGO) are lacking [15].
We hypothesize that the genetic pathway behind S-AKI has overlapping mechanisms with key differences based upon the specific subtype of AKI. To test this hypothesis, we performed a genome-wide association study (GWAS) of S-AKI in a large, single-center biorepository of septic patients [13, 20] with linked genotype and electronic health records [16, 17]. We then compare associated variants and genes from our GWAS with previously published AKI sub-types and model other factors associated with S-AKI in our dataset.
Methods
Study design and setting
This study was approved by the Institutional Review Board (HUM00126162). We used the STREGA checklist when writing our report [21]. All participants provided written informed consent. Our study cohort was created from the intersection between two previously described sources: (i) Michigan Medicine S-AKI dataset [13, 22] (N = 35,020 patients) and (ii) MGI biobank (Data Freeze 6, N = 80,529 patients). General inclusion criteria were adult patients treated at Michigan Medicine between July 10, 2009 and September 7, 2019 with genotype available. Data were accessed for research purposes September 12, 2023. Patients were excluded if they required renal replacement therapy prior to sepsis, were missing baseline creatinine, or had no follow-up creatinine (Fig 1).
Abbreviations: AKI = Acute Kidney Injury; AKI 1 = Kidney Disease: Improving Global Outcome (KDIGO) Stage 1 AKI; AKI 2 = KDIGO Stage 2 AKI; AKI 3 = KDIGO Stage 3 AKI; MGI = Michigan Genomics Initiative; RRT = Renal Replacement Therapy; S-AKI = Sepsis-associated Acute Kidney Injury; SCr = Serum Creatinine; Sepsis-3 = The Third International Consensus Definitions for Sepsis and Septic Shock.
Outcome measure and data collection
The outcome of interest was Stage 1, 2, or 3 AKI, defined per KDIGO guidelines, [19] within 7-days of sepsis onset, defined by Sepsis-3 criteria [1]. Baseline creatinine was defined as the most recent value at or before sepsis onset. The control population was patients with no AKI, within 7-days of developing sepsis. Sepsis-3 criteria were determined from the S-AKI dataset, which included vital signs, laboratory results, microbiology results, antibiotic use, and receipt of dialysis, based upon a previously described methodology [13]. A pre-specified sensitivity analysis was performed on higher severity (KDIGO, Stage 2 or 3) cases of S-AKI. In this sensitivity analysis, Stage 1 AKI was omitted from the analysis (ie: considered neither “case,” nor “control’).
Selection criteria for sepsis database
The sepsis database was created by querying (DataDirect, University of Michigan, Ann Arbor, MI) the data warehouse for all patients meeting Sepsis-3 criteria between July 10, 2009 and September 7, 2019, as previously described [13]. In brief, the data warehouse contains demographics, vital signs, laboratory results, microbiology results, antibiotic use, operation, and processes of care, including receipt of vasopressors, mechanical ventilation, kidney replacement therapy, and blood transfusion. Patients who (i) received acute kidney replacement therapy in the 14 days prior to sepsis, (ii) required chronic dialysis, (iii) were missing baseline creatinine, or (iv) had no follow-up creatinine were excluded. Patients met Sepsis-3 criteria if they had suspected infection as documented by blood cultures, received antibiotics within the time window from 24 hours before to 72 hours after blood cultures were obtained, and had ≥ 2 points increase in Sequential Organ Failure Assessment (SOFA) score between 48-hours prior to 24-hours after onset of suspected infection. The timing of suspected infection was defined by antibiotic administration or obtaining cultures (whichever occurred first). The cardiovascular component of the SOFA score was calculated including vasopressin (0.4 U/min vasopressin = 0.1 mcg/kg/min norepinephrine) [23]. Patients who had Kidney Disease Improving Global Outcomes (KDIGO) Stage 1, 2 or 3 acute kidney injury (AKI) were compared to patients without any AKI.
As urine output was not reliably obtained in all patients, only the creatinine criteria were used. Stage 2 AKI was defined as serum creatinine 2–2.9 times baseline value; Stage 3 as serum creatinine ≥ 3 times baseline value, a rise to ≥ 4.0 mg/dL, or initiation of kidney replacement therapy. No AKI was defined as any rise in serum creatinine <0.3 mg/dL within 48 hours and absence of ≥ 50% within 7 days.
Genotype
The Michigan Genomics Initiative (MGI) is a biobank that collects blood or saliva samples from adult patients receiving care at Michigan Medicine. In brief, MGI patients were genotyped on customized Illumina Human-CoreExome (v1.0, v1.1, v1.3) or Illumina Infinium Global Screening Arrays (v1.3) (Illumina, Inc. San Diego, CA) and subsequently imputed to the Trans-Omics for Precision Medicine (TOPMed) reference panel using the TOPMed Imputation Server [24]. Variants with R2 ≥ 0.3 and minor allele frequency (MAF) ≥ 0.01% were selected, resulting in more than 52-million imputed variants after quality control and filtering. The European population was defined using the majority ADMIXTURE global ancestry fraction (Q-value). There were not enough patients of non-European ancestry (enrolled in Michigan Genomics Initiative) to perform a more comprehensive analysis in cohorts of non-European ancestry, therefore, our study population was restricted to patients of European descent. Additional genotype characterization and quality control from MGI is described in S1 Appendix.
Covariates
Demographic, comorbidity, laboratory, and exposure data were derived from the S-AKI database. Specific pre-sepsis comorbidities were obtained from International Classification of Diseases-9 and -10 based upon grouping schema from the Elixhauser comorbidity index [25].
Data analysis
Perioperative and intraoperative characteristics were summarized using medians and interquartile ranges for continuous variables and counts and percentages for categorical covariates. Comparisons of continuous data were made using Mann-Whitney U Tests and categorical data were compared using Pearson chi-square tests. P-value < 0.05 was selected a priori to denote statistical significance.
GWAS was performed using the Scalable and Accurate Implementation of Generalized mixed model (SAIGE) logistic regression model with kinship adjustment on genetic data from a European cohort in MGI Freeze 6 [26]. The European population was defined using the majority ADMIXTURE global ancestry fraction (Q-value). For association tests we used TOPMed Imputed genotypes and excluded variants with low imputation quality (Rsq < 0.8) or very rare minor alleles (minor allele frequency < 0.1%). Genome-wide significance threshold (P < 5 × 10−8) was selected a priori for identification of novel loci and a lower threshold was selected as a suggestive threshold (and for replication of previously reported loci) (P < 1 × 10−6) [27]. Genome-wide association was assessed in three logistic regression models, as previously described [16, 17]. Model 1, controlled for 1) age, 2) sex, 3) genotyping chip, and 4) the first five principal components (PCs). In Model 2, pre-sepsis baseline serum creatinine was added to the variables in Model 1. Finally, in Model 3, we controlled for the full range of patient, clinical, and ICU-related risk factors. Each of the 3-models were repeated in a pre-specified sensitivity analysis of higher severity S-AKI, defined as KDIGO Stage 2 or 3.
Results of the GWAS were visualized and top hits annotated using the gwasTools repository (https://github.com/ilarsf/gwasTools) with the option to set a suggestive significance threshold (—sigthreshold) of 1x10-6. The Manhattan plot was annotated with quasi-independent top hits by creating 1Mb regions centered around variants with p-values less than 1x10-6 and combining overlapping intervals.
Differential tissue enrichment
We used FUMA to test for enrichment of differential expression across tissues for the eight candidate genes suggested by our study [28]. We compared expression levels of these with 57,241 background genes. Our analysis used the GENE2FUNC function with GTEx version 8 and tested 30 general tissue types. P-values for the enrichment were corrected using a false discovery rate.
Results
3,348 qualifying Sepsis-3 patients from the S-AKI database have been genotyped in MGI (Fig 1). Of these patients, 383 (11.4%) developed AKI (primary outcome) and 181 (5.4%) developed Stage 2 or 3 AKI. The median age was 61 years (IQR: 51,69), 42% were female, and median SOFA score at time of Sepsis-3 diagnosis was 2 (2,4). Additional details on the cohort can be found in Table 1.
On univariate analysis, patients developing S-AKI tended to be male (66.1% in the AKI group compared to 57.0 in the No AKI group, P = 0.0008) (Table 1). Additionally, patients with comorbidities, including diabetes, peripheral vascular disease, and cardiac arrhythmias had a greater risk of developing S-AKI.
Genome-wide association studies
Primary outcome.
Q-Q plots confirmed that the analyses were well-controlled, without an elevated Type I error (S1 Fig). Median genomic control λ values calculated across all tested variants were 1.051, 1.044, and 1.043 for Models 1, 2, and 3, respectively. Median genomic control λ values calculated across only variants with MAF≥ 1% were 1.004, 0.997, and 0.996 for Models 1, 2, and 3 [29]. No variants exceeded our threshold for genome-wide significance (P<5x10-8), however, a total of 13 variants exceeded the suggestive (P<1x10-6) threshold in at least one of our Models (Table 2). Manhattan plots for each of the 3 Models can be found in Fig 2. Variants exceeding the suggestive (P<1x10-6) threshold for each of our 3-models are described in Table 2. Notably, rs184516290 (chr1:199814965:G:A), chr1:199805801:T:TA, and rs117313146 (chr15:31999784:G:C) exceeded the suggestive significance threshold in all 3-models. rs113501598 (chr5:142181517:G:A), rs17526864 (chr5:142170708:C:T), rs113887864 (chr5:142179290:G:A), rs61394874 (chr5:142166870:G:A) exceeded the suggestive threshold in Models 2 and 3 (Table 2). The overlap and intersection of suggestive genetic variants (and accompanying genes) between each of the three genome-wide association studies can be visualized in Fig 3.
2A. Model 1. 2B. Model 2. 2C. Model 3. (•) Genome Wide Significance Threshold (P<5x10-8) red dashed line. (•) Suggestive/Replicative Significance Threshold (P<1x10-6) blue dashed line.
Gene Names: ARL4C = ADP Ribosylation Factor like GTPase 4C; CHRNA7 = Cholinergic Receptor Nicotinic Alpha 7 Subunit; IQCJ = IQ Domain-Containing Protein J; KCNMB2 = Potassium Calcium-Activated Channel Subfamily M Regulatory Beta Subunit 2; NDFIP1 = Nedd4 Family Interacting Protein 1; SCHIP1 = Schwannomin Interacting Protein 1; SLAMF6 = SLAM Family Member 6; TRIB2 = Tribbles Pseudokinase 2.
GWAS results for variants previously reported to be associated with AKI (but not exceeding the suggestive threshold) are also reported to facilitate comparison and replication (S1 Table).
Sensitivity analysis: Severe AKI
Results of the analysis repeated for severe AKI (defined as KDIGO, Stage 2 or 3) are shown in S2 Table. As with the primary outcome, no variants exceeded the genome-wide significance threshold (P<10−8), but numerous variants exceeded the suggestive (P<1x10-6) threshold and appeared across multiple models. Notably, rs11140550 (chr9:84369833:G:A) and rs11140551 (chr9:834369848:C:T) near SLC28A3 and rs2671704 (chr10:48929632:C:T), rs2377695 (chr10:4829132:G:A), rs2671703 (chr10:48927949:C:G), rs2663023 (chr10:48928027:T:C), rs2663025 (chr10:48928128:G:A), rs1877803 (chr10:48929468:A:G), and rs2671706 (chr10:48930286:T:C) near WDFY4 exceeded the suggestive significance threshold in Models 2 and 3.
Gene expression patterns
We explored the gene expression patterns of the candidate genes suggested by genome-wide association studies using the Genotype-Tissue Expression (GTEx) portal [30]. ARL4C (kidney-cortex: 7.956 median transcripts per million; kidney-medulla:10.210), CHRNA7 (0.037; 0.016), IQCJ-SCHIP1 (1.111; 1.431), NDFIP1 (46.780; 67.020), KCNMB2 (0.872, 1.861), TRIB2 (29.800; 89.660), and SLAMF6 (0.419; 0.270) were all expressed by the kidneys. Lung was the only tissue that was significantly differentially expressed among our candidate genes on the FUMA test [28] (FDR corrected p-value = 0.0306), driven by candidate genes ARL4C, TRIB2, SCHIP1, and SLAMF6. Kidney, however, was not found to be differentially expressed (FDR corrected p-value = 1).
Discussion
In a large, unbiased GWAS on S-AKI, we failed to identify any variants reaching genome-wide significance (P<5x10-8). Numerous loci reached the “suggestive” threshold (P<1x10-6), [27] however, these should not be viewed as significant novel associations, as adoption of less stringent thresholds leads to increased rate of false discovery [31]. rs117313146 (chr15:31999784:G:C), near the CHRNA7 gene, was associated with S-AKI at the suggestive threshold in all 3-models presented. While our study is the first to suggest an association between a specific acetylcholine receptor subunit and S-AKI in a human GWAS; nicotinic acetylcholine receptors have been linked to both septic [32] and ischemic [33] AKI in mouse models. Furthermore, the CHRNA7 variant (chr15:32413847:G:A) was associated with the largest odds for developing Sepsis-3 in a recent publication [20].
CHRNA7 is the gene for the human α7 subunit of the neuronal nicotinic acetylcholine receptor gene and has previously been shown to regulate inflammation by attenuating the production of multiple inflammatory cytokines from macrophages [34, 35]. The α7 controls systemic inflammation in sepsis by inhibiting TNF production and the NF-kB immunomodulatory response [34]. Inhibition of this pathway has been shown to improve survival in experimental models of sepsis [36]. While our study was likely underpowered to discover new variants at the level of genome-wide significance (P<5x10-8); the strongly concordant results within our 3-models, confirmation of findings from an overlapping phenotype [20] at replication threshold (P<1x10-6) [20], and plausible mechanistic link should be viewed as hypothesis-generating. Additional studies are necessary to confirm the association between CHRNA7 and S-AKI as well as identify potential therapeutic interventions that block the inflammatory signaling pathways involving nicotinic acetylcholine receptor α7 subunit.
Additionally, rs184516290 (chr1:199814965:G:A) and chr1:199805801:T:TA (no rsID) both near the NR5A2 gene exceeded the suggestive threshold in all 3-models. The NR5A2 gene is a pioneer transcription factor which initiates transcription from closed chromatin and regulates metabolism in adult tissues [37]. While not specifically implicated in S-AKI, NR5A2 regulates the calreticulin gene during renal fibrosis [38, 39] and plays a role in multiple immune pathways including inflammation [40] and tumor-necrosis factor- (TNF) induced cell death [41].
All candidate genes were expressed, to varying extent, in the kidneys indicating a potential role in sepsis-associated AKI; however, differential expression within the lung (and not kidney) indicates that the relationship is likely more complex and requires further exploration. Gene expression patterns have been quantified in S-AKI models [42–44]. Notably, the ARL4C gene (rs558059451; chr2:234602989:G:A, in our study) was downregulated in S-AKI in a previously published Gene Ontology enrichment analysis [42]. Additionally, ARL4C, TRIB2, and SLAMF6 were found to be downregulated within 0.5, 24, and 48-hours after septic shock [44]. Full details on differential expression of suggested genes from our study can be found in S3 Table. Future studies are necessary to assess gene expression patterns of the candidate genes and compare those to published models of sepsis and kidney injury.
The sensitivity analysis restricted to more severe, KDIGO Stage 2 or 3 AKI, cases suggested (P<1x10-6) a different set of genes (SLC28A3 and WDFY4) compared to the primary analysis. This could be due to a different genetic effect between more and less severe AKI, but more likely is because most cases were Stage 1 AKI, leaving the sensitivity analysis of Stage 2 and 3 severely underpowered to detect novel associations.
Comparison to previously published results
We failed to replicate any of the variants previously reported to be associated with S-AKI at P<1x10-6 threshold (S1 Table). Variants in the suppressor of fused homolog (SUFU) gene previously shown to be correlated with renal function in adult patients hospitalized with Enterobacteriaceae bacteremia (at threshold P<0.05) [45] also consistently demonstrated P<0.05 in our Models (rs10786691, P = 0.040–0.053; rs12414407, P = 0.007–0.026; rs10748825, P = 0.007–0.019; rs2296590, P = 0.008–0.016) [45]. Although significance-level was consistent between the two studies across multiple SUFU variants and models, since they did not exceed the a priori significance threshold for replication (P<1x10-6), these findings should be viewed as hypothesis-generating and additional studies are necessary to elucidate the role of the sonic hedgehog signaling pathway in S-AKI.
Inability to replicate findings from other prior studies could be due to differences in how phenotypes were defined. Specifically, we assessed KDIGO Stage 1, 2, or 3 AKI [19] following Sepsis (as defined by the Sepsis-3 diagnostic criteria) [1]. This contrasts with other studies using alternative definitions or severity classifications: (i) RIFLE (AKI) after American College of Chest Physicians/Society of Critical Care Medicine (Sepsis) [46] or AKIN Stage 3 in the setting of severe sepsis or septic shock (based upon SOFA score) [47]. Although our study demonstrated similar significance-levels to prior work from Hanao-Martinez, their outcome was renal function as defined by serum creatinine (continuous),precluding direct effect-size comparisons with our KDIGO AKI outcome (binary) [45]. Alternatively, issues replicating prior studies could be because S-AKI has limited genetic contribution and low heritability (as found in other AKI subtypes) [16, 17]. In this case, a greater understanding of risk for AKI could be gained from characterizing non-genetic clinical risk factors (some of which may be modifiable) rather than from further genetic studies.
In addition to comparing our work to studies on the narrow S-AKI phenotype, we also compare our results to other AKI sub-phenotypes, including: (i) critically ill [18] and (ii) post-surgical populations to facilitate the broadest understanding of overlapping mechanisms [16, 17]. The lack of concordance with these studies could be due to statistical variation in significance between studies or, alternatively, due to unique genetic pathways responsible for S-AKI.
Study strengths and limitations
Unlike prior studies which have largely employed a candidate-gene approach, [14, 45, 46, 48] our study provides a large, unbiased analysis on the genetics underlying S-AKI. We adjust for multiple clinical covariates (including baseline serum creatinine, lactate prior to diagnosis of sepsis, and SOFA-score). As studies demonstrated considerable variation based upon phenotype definition, [20] our study employs standardized diagnostic criteria for Sepsis (Sepsis-3) [1] and AKI (KDIGO), [19] in a previously validated dataset [13]. We assess for differential genetic influence at greater severity phenotypes in a pre-specified sensitivity analysis. The major limitations of this study are lack of external validation (ie: single-center design) and lack of statistical power to detect novel genetic associations in low-frequency variants or variants with moderate effect sizes.
While our study selected cases and controls using the entire population of the Sepsis Database with available genetic data in Michigan Genomics Initiative, there are potentially unaccounted biases in the initial curation of this dataset. For example, more critically ill patients are likely to have more frequent laboratory draws and vital sign documentation, leading to higher probability of reaching the Sepsis-3 diagnostic criteria. Furthermore, as the data within the data warehouse was generated during the process of providing clinical care (and not specifically for research purposes), additional inter-provider variation and other sources of bias are possible. While we were unable to include stages of AKI based on urine creatinine, this is similar to other studies [13, 49, 50]. Patients with AKI by urine criteria only have better outcomes than patients who meet serum creatinine criteria for AKI [51]. However, our exclusion of urine criteria may bias the results in unknown ways.
While the data within Michigan Genomics Initiative are reflective of the patient population served by Michigan Medicine and lack of ancestral diversity is a commonly encountered problem across many biobanks, [52] lack of genetic diversity (92% European Ancestry) is an inherent limitation of our dataset. Furthermore, analysis of patients at a tertiary academic center in the United States may not reflect findings from international or community centers.
Conclusions
While failing to identify any novel association for S-AKI at the level of genome-wide significance, our study did suggest multiple variants in previously characterized pathways for S-AKI including CHRNA7, NR5A2, and SUFU. We failed to replicate associations from multiple prior studies which may result from differences in how the phenotype was defined or, alternatively, limited genetic contribution and low heritability.
Supporting information
S1 Appendix. Characterization and quality control within Michigan genomics initiative.
https://doi.org/10.1371/journal.pone.0311318.s001
(DOCX)
S1 Fig. Quantile-Quantile (Q-Q) plots for each of the three genome-wide association studies.
https://doi.org/10.1371/journal.pone.0311318.s002
(PDF)
S1 Table. Attempted replication of prior GWAS findings.
* SNP not found in MGI dataset. ** Study by Henao-Martinez et al., 2013 used Serum Creatinine (mg/dL) as a marker of renal function (ie: continuous variable), therefore, direct comparison of effect size to our logistic regression, which used KDIGO-staged AKI (ie: binary, categorical variable) is not possible. Studies replicated at P < 0.05 threshold are denoted in bold. Abbreviations: CHF = congestive heart failure; Chr = Chromosome; COPD = chronic obstructive pulmonary disease; GWAS = genome-wide association study; KDIGO = Kidney Disease: Improving Global Outcomes; MAF = Minor Allele; MGI = Michigan Genomics Initiative OR = Odds Ratio; Pos = Position; Frequency; S-AKI = Sepsis-associated Acute Kidney Injury; SOFA = Sequential Organ Failure Assessment; SE = Standard Error; SNP = single nucleotide polymorphism.
https://doi.org/10.1371/journal.pone.0311318.s003
(DOCX)
S2 Table. Genome Wide Association Study (GWAS) results: P < 1 x 10−6, minor allele frequency > 1%.
A. Model 1: GWAS Performed on Primary Outcome (KDIGO Stage 2 or 3)—Controlling for age, sex, chip, and first 5 principal components. B. Model 2: GWAS Performed on Primary Outcome (KDIGO Stage 2 or 3)—Controlling for age, sex, chip, baseline serum creatinine, and first 5 principal components. C. Model 3: GWAS Performed on Primary Outcome (KDIGO Stage 2 or 3)—Controlling for age, sex, chip, baseline serum creatinine, lactate, total increase in SOFA score, comorbidities: diabetes complicated/uncomplicated, CHF, liver disease, peripheral vascular disease, COPD, and cardiac arrhythmias, and first 5 principal components. Abbreviations: CHF = congestive heart failure; Chr = Chromosome; COPD = chronic obstructive pulmonary disease; GWAS = genome-wide association study; KDIGO = Kidney Disease: Improving Global Outcomes; MAF = Minor Allele; OR = Odds Ratio; Pos = Position; Frequency; SOFA = Sequential Organ Failure Assessment; SE = Standard Error; SNP = single nucleotide polymorphism.
https://doi.org/10.1371/journal.pone.0311318.s004
(DOCX)
S3 Table. Differential gene expression of suggested genes in published S-AKI models.
Abbreviations: logFC: The log fold change, or the change in gene expression for each unit increase; AveExpr: The average expression across all samples; t: The logFC divided by its standard error; P.Value: The raw p-value based on t from the test that logFC differs from 0.
https://doi.org/10.1371/journal.pone.0311318.s005
(DOCX)
Acknowledgments
The authors acknowledge the Michigan Genomics Initiative participants, Precision Health at the University of Michigan, the University of Michigan Medical School Central Biorepository, and the University of Michigan Advanced Genomics Core for providing data and specimen storage, management, processing, and distribution services, and the Center for Statistical Genetics in the Department of Biostatistics at the School of Public Health for genotype data curation, imputation, and management in support of the research reported in this publication. We thank the clinicians, staff, and study participants from the Michigan Genomics Initiative.
References
- 1. Singer M, Deutschman CS, Seymour CW, Shankar-Hari M, Annane D, Bauer M, et al. The Third International Consensus Definitions for Sepsis and Septic Shock (Sepsis-3). JAMA. 2016;315: 801–810. pmid:26903338
- 2. Mayr FB, Yende S, Angus DC. Epidemiology of severe sepsis. Virulence. 2014;5: 4–11. pmid:24335434
- 3. Poukkanen M, Vaara ST, Pettilä V, Kaukonen K-M, Korhonen A-M, Hovilehto S, et al. Acute kidney injury in patients with severe sepsis in Finnish Intensive Care Units. Acta Anaesthesiol Scand. 2013;57: 863–872. pmid:23713703
- 4. Lopes JA, Jorge S, Resina C, Santos C, Pereira A, Neves J, et al. Acute kidney injury in patients with sepsis: a contemporary analysis. Int J Infect Dis. 2009;13: 176–181. pmid:18771942
- 5. Bagshaw SM, George C, Bellomo R, ANZICS Database Management Committee. Changes in the incidence and outcome for early acute kidney injury in a cohort of Australian intensive care units. Crit Care. 2007;11: R68. pmid:17588270
- 6. Poston JT, Koyner JL. Sepsis associated acute kidney injury. BMJ. 2019;364: k4891. pmid:30626586
- 7. Kerr M, Bedford M, Matthews B, O’Donoghue D. The economic impact of acute kidney injury in England. Nephrol Dial Transplant. 2014;29: 1362–1368. pmid:24753459
- 8. Zeng X, McMahon GM, Brunelli SM, Bates DW, Waikar SS. Incidence, outcomes, and comparisons across definitions of AKI in hospitalized individuals. Clin J Am Soc Nephrol. 2014;9: 12–20. pmid:24178971
- 9. Chertow GM, Burdick E, Honour M, Bonventre JV, Bates DW. Acute kidney injury, mortality, length of stay, and costs in hospitalized patients. J Am Soc Nephrol. 2005;16: 3365–3370. pmid:16177006
- 10. Ricci Z, Cruz D, Ronco C. The RIFLE criteria and mortality in acute kidney injury: A systematic review. Kidney Int. 2008;73: 538–546. pmid:18160961
- 11. Suh SH, Kim CS, Choi JS, Bae EH, Ma SK, Kim SW. Acute kidney injury in patients with sepsis and septic shock: risk factors and clinical outcomes. Yonsei Med J. 2013;54: 965–972. pmid:23709433
- 12. Plataki M, Kashani K, Cabello-Garza J, Maldonado F, Kashyap R, Kor DJ, et al. Predictors of acute kidney injury in septic shock patients: an observational cohort study. Clin J Am Soc Nephrol. 2011;6: 1744–1751. pmid:21734090
- 13. Carter AW, Engoren M. Factors associated with occurrence and severity of acute kidney injury in patients with Sepsis—A retrospective database study. J Crit Care. 2022;72: 154150. pmid:36244255
- 14. Frank AJ, Sheu C-C, Zhao Y, Chen F, Su L, Gong MN, et al. BCL2 genetic variants are associated with acute kidney injury in septic shock*. Crit Care Med. 2012;40: 2116–2123. pmid:22710204
- 15. Larach DB, Engoren MC, Schmidt EM, Heung M. Genetic variants and acute kidney injury: A review of the literature. J Crit Care. 2018;44: 203–211. pmid:29161666
- 16. Larach DB, Lewis A, Bastarache L, Pandit A, He J, Sinha A, et al. Limited clinical utility for GWAS or polygenic risk score for postoperative acute kidney injury in non-cardiac surgery in European-ancestry patients. BMC Nephrol. 2022;23: 339. pmid:36271344
- 17. Douville NJ, Larach DB, Lewis A, Bastarache L, Pandit A, He J, et al. Genetic predisposition may not improve prediction of cardiac surgery-associated acute kidney injury. Front Genet. 2023;14: 1094908. pmid:37124606
- 18. Zhao B, Lu Q, Cheng Y, Belcher JM, Siew ED, Leaf DE, et al. A Genome-Wide Association Study to Identify Single-Nucleotide Polymorphisms for Acute Kidney Injury. Am J Respir Crit Care Med. 2017;195: 482–490. pmid:27576016
- 19. Khwaja A. KDIGO clinical practice guidelines for acute kidney injury. Nephron Clin Pract. 2012;120: c179–84. pmid:22890468
- 20. Engoren M, Jewell ES, Douville N, Moser S, Maile MD, Bauer ME. Genetic variants associated with sepsis. PLoS One. 2022;17: e0265052. pmid:35275946
- 21. Little J, Higgins JPT, Ioannidis JPA, Moher D, Gagnon F, von Elm E, et al. STrengthening the REporting of Genetic Association Studies (STREGA)—an extension of the STROBE statement. Genet Epidemiol. 2009;33: 581–598. pmid:19278015
- 22. Freundlich RE, Li G, Leis A, Engoren M. Factors Associated With Initiation of Mechanical Ventilation in Patients With Sepsis: Retrospective Observational Study. Am J Crit Care. 2023;32: 358–367. pmid:37652887
- 23. Khanna A, English SW, Wang XS, Ham K, Tumlin J, Szerlip H, et al. Angiotensin II for the Treatment of Vasodilatory Shock. N Engl J Med. 2017;377: 419–430. pmid:28528561
- 24. Kowalski MH, Qian H, Hou Z, Rosen JD, Tapia AL, Shan Y, et al. Use of >100,000 NHLBI Trans-Omics for Precision Medicine (TOPMed) Consortium whole genome sequences improves imputation quality and detection of rare variant associations in admixed African and Hispanic/Latino populations. PLoS Genet. 2019;15: e1008500.
- 25. Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi J-C, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43: 1130–1139. pmid:16224307
- 26. Zhou W, Nielsen JB, Fritsche LG, Dey R, Gabrielsen ME, Wolford BN, et al. Efficiently controlling for case-control imbalance and sample relatedness in large-scale genetic association studies. Nat Genet. 2018;50: 1335–1341. pmid:30104761
- 27. Duggal P, Gillanders EM, Holmes TN, Bailey-Wilson JE. Establishing an adjusted p-value threshold to control the family-wide type 1 error in genome wide association studies. BMC Genomics. 2008;9: 516. pmid:18976480
- 28. Watanabe K, Taskesen E, van Bochoven A, Posthuma D. Functional mapping and annotation of genetic associations with FUMA. Nat Commun. 2017;8: 1826. pmid:29184056
- 29. Devlin B, Roeder K. Genomic control for association studies. Biometrics. 1999;55: 997–1004. pmid:11315092
- 30. Consortium GTEx. The Genotype-Tissue Expression (GTEx) project. Nat Genet. 2013;45: 580–585. pmid:23715323
- 31. Chen Z, Boehnke M, Wen X, Mukherjee B. Revisiting the genome-wide significance threshold for common variant GWAS. G3. 2021;11. pmid:33585870
- 32. Chatterjee PK, Yeboah MM, Dowling O, Xue X, Powell SR, Al-Abed Y, et al. Nicotinic acetylcholine receptor agonists attenuate septic acute kidney injury in mice by suppressing inflammation and proteasome activity. PLoS One. 2012;7: e35361. pmid:22586448
- 33. Kim H, Kim SR, Je J, Jeong K, Kim S, Kim HJ, et al. The proximal tubular α7 nicotinic acetylcholine receptor attenuates ischemic acute kidney injury through Akt/PKC signaling-mediated HO-1 induction. Exp Mol Med. 2018;50: 1–17.
- 34. Peña G, Cai B, Liu J, van der Zanden EP, Deitch EA, de Jonge WJ, et al. Unphosphorylated STAT3 modulates alpha 7 nicotinic receptor signaling and cytokine production in sepsis. Eur J Immunol. 2010;40: 2580–2589. pmid:20706987
- 35. Wang H, Yu M, Ochani M, Amella CA, Tanovic M, Susarla S, et al. Nicotinic acetylcholine receptor alpha7 subunit is an essential regulator of inflammation. Nature. 2003;421: 384–388. pmid:12508119
- 36. Wang H, Liao H, Ochani M, Justiniani M, Lin X, Yang L, et al. Cholinergic agonists inhibit HMGB1 release and improve survival in experimental sepsis. Nat Med. 2004;10: 1216–1221. pmid:15502843
- 37. Kobayashi W, Sappler AH, Bollschweiler D, Kümmecke M, Basquin J, Arslantas EN, et al. Nucleosome-bound NR5A2 structure reveals pioneer factor mechanism by DNA minor groove anchor competition. Nat Struct Mol Biol. 2024. pmid:38409506
- 38. Arvaniti E, Vakrakou A, Kaltezioti V, Stergiopoulos A, Prakoura N, Politis PK, et al. Nuclear receptor NR5A2 is involved in the calreticulin gene regulation during renal fibrosis. Biochim Biophys Acta. 2016;1862: 1774–1785. pmid:27345266
- 39. Politis PK, Charonis AS. Calreticulin in renal fibrosis: A short review. J Cell Mol Med. 2022;26: 5949–5954. pmid:36440574
- 40. Sandhu N, Rana S, Meena K. Nuclear receptor subfamily 5 group A member 2 (NR5A2): role in health and diseases. Mol Biol Rep. 2021;48: 8155–8170. pmid:34643922
- 41. Lambrecht R, Delgado ME, Gloe V, Schuetz K, Plazzo AP, Franke B, et al. Liver receptor homolog-1 (NR5A2) orchestrates hepatic inflammation and TNF-induced cell death. Cell Rep. 2023;42: 113513. pmid:38039134
- 42. Yang J-J, Wu B-B, Han F, Chen J-H, Yang Y. Gene expression profiling of sepsis-associated acute kidney injury. Exp Ther Med. 2020;20: 34. pmid:32952625
- 43. Yang S, Guo J, Xiong Y, Han G, Luo T, Peng S, et al. Unraveling the genetic and molecular landscape of sepsis and acute kidney injury: A comprehensive GWAS and machine learning approach. Int Immunopharmacol. 2024;137: 112420. pmid:38851159
- 44. Tang Y, Yang X, Shu H, Yu Y, Pan S, Xu J, et al. Bioinformatic analysis identifies potential biomarkers and therapeutic targets of septic-shock-associated acute kidney injury. Hereditas. 2021;158: 13. pmid:33863396
- 45. Henao-Martínez AF, Agler AH, LaFlamme D, Schwartz DA, Yang IV. Polymorphisms in the SUFU gene are associated with organ injury protection and sepsis severity in patients with Enterobacteriacea bacteremia. Infect Genet Evol. 2013;16: 386–391. pmid:23538333
- 46. Cardinal-Fernández P, Ferruelo A, El-Assar M, Santiago C, Gómez-Gallego F, Martín-Pellicer A, et al. Genetic predisposition to acute kidney injury induced by severe sepsis. J Crit Care. 2013;28: 365–370. pmid:23499421
- 47. Payen D, Lukaszewicz A-C, Legrand M, Gayat E, Faivre V, Megarbane B, et al. A multicentre study of acute kidney injury in severe sepsis and septic shock: association with inflammatory phenotype and HLA genotype. PLoS One. 2012;7: e35838. pmid:22701553
- 48. Vilander LM, Kaunisto MA, Vaara ST, Pettilä V, FINNAKI study group. Genetic variants in SERPINA4 and SERPINA5, but not BCL2 and SIK3 are associated with acute kidney injury in critically ill patients with septic shock. Crit Care. 2017;21: 47.
- 49. Long TE, Helgason D, Helgadottir S, Palsson R, Gudbjartsson T, Sigurdsson GH, et al. Acute Kidney Injury After Abdominal Surgery: Incidence, Risk Factors, and Outcome. Anesth Analg. 2016;122: 1912–1920. pmid:27195635
- 50. Zarkowsky DS, Hicks CW, Bostock IC, Stone DH, Eslami M, Goodney PP. Renal dysfunction and the associated decrease in survival after elective endovascular aneurysm repair. J Vasc Surg. 2016;64: 1278–1285.e1. pmid:27478004
- 51. Bianchi NA, Stavart LL, Altarelli M, Kelevina T, Faouzi M, Schneider AG. Association of Oliguria With Acute Kidney Injury Diagnosis, Severity Assessment, and Mortality Among Patients With Critical Illness. JAMA Netw Open. 2021;4: e2133094. pmid:34735011
- 52.
Zhou W, Neale BM, Daly MJ. Global Biobank Meta-analysis Initiative: Powering genetic discovery across human diseases. 69th Annual Meeting of the American Society of Human Genetics. 2019.