Identification of the key targets for sepsis-associated acute kidney injury by RNA sequencing combined with bioinformatics methods

Abstract

Purpose

This research probes into genes related to the risk of concurrent kidney injury in septic patients to provide reliable targets for early identification of sepsis-associated kidney injury and prognosis research.

Methods

Peripheral blood samples were isolated from 10 healthy individuals and 22 septic patients for RNA sequencing and differential analyses. Meanwhile, the top 1000 kidney-associated genes were chosen from the GTEx website. Subsequently, DEGs in sepsis were intersected with kidney-specific genes, followed by GO and KEGG analyses on these intersection genes. The predictive ability of hub genes for prognosis was evaluated using survival analysis. A meta-analysis was carried out to determine the differential expression profiles of hub genes between the sepsis surviving and dead groups. ROC curves were plotted to screen hub genes and clarify their diagnostic value. Cell line localization of hub genes was further clarified through single-cell RNA sequencing.

Results

There were 40 targets in the intersection between 1328 DEGs in sepsis and 1000 kidney-associated genes. These intersection genes were mainly engaged in functions and signaling pathways .Survival curves linked the higher levels of CD74 and IL32 to raised survival rates of patients, indicating positive correlations of CD74 and IL32 with patient prognosis. Meta-analysis revealed that CD74 and IL32 were highly expressed in the sepsis surviving group but poorly expressed in the sepsis dead group, showing statistically significant differences between these two groups. In the ROC analysis of hub genes, AUC values of CD74 (0.983) and IL32 (0.980) suggested their high diagnostic value. Lastly, CD74 was principally expressed in macrophages, while IL32 was mainly presented in T cells.

Conclusion

CD74 and IL32, as biomarkers for early diagnosis and prognostic evaluation of sepsis complicated with kidney injury, are highly expressed in macrophages and T cells, respectively, providing new diagnostic and prognostic targets for sepsis complicated with acute kidney injury.

Introduction

Sepsis represents an abnormal host response to infection, accompanied by a considerable incidence of complications. Patients experiencing sepsis-induced shock exhibit circulatory, cellular, and metabolic aberrations, resulting in an elevated mortality rate [1]. Currently, sepsis causes 11 million deaths annually, with a particularly severe impact on infants, the elderly, pregnant women, and populations in low-income countries [2]. Despite significant progress in the comprehension of pathophysiology and supportive therapeutic schemes, sepsis and septic shock are still associated with high mortality. It is estimated that one-fifth of septic patients will die [3]. Sepsis is defined as impaired organ functions elicited by the host’s harmful response to infection. Kidney is one of the most frequently-affected organs in sepsis, which causes sepsis-associated acute kidney injury (SA-AKI), leading to significant increases in the incidence and mortality of sepsis [4]. AKI exists in 40–50% of septic patients, with a 6 - to 8-fold increase in mortality [5]. SA-AKI has a high incidence and is difficult to prevent. Additionally, for patients without AKI at the time of diagnosis, several septic patients complicated with AKI have no significant changes in creatinine and urine volume at the early stage. Thus, the diagnosis merely relying on creatinine and urine volume as criteria frequently results in delayed diagnosis, influencing the therapeutic effect. Regarding the pathophysiological mechanism of SA-AKI, it was previously believed that SA-AKI was caused by circulatory disorders, among which low blood pressure and reduced renal blood flow lead to acute tubular necrosis [6]. However, recent animal models have illustrated that although tubular cell damage and the expression of markers such as KIM-1 are common, inflammation and cell apoptosis also have an effect [4]. It can be seen that inflammation and the disrupted immune system have a profound impact on sepsis-elicited organ dysfunction. Early diagnosis and timely intervention of SA-AKI can remarkably lower the incidence of AKI [7]. Therefore, early diagnosis of SA-AKI can improve patient prognosis. However, studies on the key genes and proteins associated with SA-AKI in humans are limited. Recently, biomarkers have been introduced in many medical fields, including sepsis or AKI, which may aid in the timely diagnosis of diseases. With advancements in biotechnology, a lot of information on genes, transcription factors, and signaling pathways has become available, which has enabled the screening and identification of key biomarkers and potential targets of SA-AKI. In light of this situation, increasing studies focus on novel biomarkers for early AKI [8].

This study adopted an RNA sequencing technique in conjunction with bioinformatics methods to screen relevant targets for early identification of SA-AKI, providing a basis and direction for early diagnosis and treatment of this disease. The flowchart of this experiment is depicted in Fig. 1.

Fig. 1

Research flowchart. RNA sequencing and differential analyses were conducted on peripheral blood samples of 10 healthy participants and 22 septic patients. Meanwhile, the top 1000 kidney-related genes were screened from the online GTEx tool. Subsequently, the intersection of differentially expressed genes (DEGs) and renal damage-associated genes was identified, followed by GO and KEGG analyses on these intersection genes. The prognostic potential of the hub genes was assessed through survival analysis. Through a meta-analysis, we determined the differential expression patterns of hub genes between the sepsis surviving group and the dead group. Then, receiver operator characteristic (ROC) curves were generated to further screen hub genes and clarify their diagnostic value. A single-cell RNA sequencing analysis was the final to elucidate the localization of hub genes in cell lines

Methods

Subject recruitment

In this study, 22 septic patients (sepsis group) were chosen from the Emergency Intensive Care Unit (EICU) of the Affiliated Hospital Of Southwest Medical University from January 2019 to December 2020. Additionally, 10 healthy participants were assigned as the control group. The experimental protocol was reviewed and approved by the Ethics Committee of the Affiliated Hospital Of Southwest Medical University (Ethics No. ky2018029; clinical trial registration No. ChiCTR1900021261). The participants met the following inclusion criteria: (1) Participants with sepsis conformed to the SCCM and ESICM consensus diagnostic criteria for sepsis 3.0 in 2016: infection plus organ dysfunction (SOFA score ≥ 2) [3]; (2) Patient age ≥ 18 years; (3) Patients voluntarily participated in the experiment, and patients or their legal representative signed an informed consent form. The subjects with one of the following conditions were eliminated from the study: (1) Pregnant or lactating women; (2) Patients with a mental illness; (3) Hypoimmunity, immunodeficiency, or HIV-positive.

Blood collection and processing

The PAXgene blood collection tube system can collect whole blood, stabilize nucleic acids, minimize the risk of RNA degradation, and greatly improve the accuracy of transcriptomic analysis of cellular RNA data. Before use, it is necessary to correctly label the patient’s ID and ensure that the temperature of the PAXgene tube is maintained between 18℃ and 25℃. From each patient, 2 tubes of PAXgene blood samples were collected. To ensure that the volume inside the blood collection device can be pre-perfused with blood, the first tube of blood from each patient should be collected into a regular blood collection tube and discarded. Next, the PAXgene tube was connected for blood collection. For blood collection using the PAXgene tube, 2.5 mL of blood was extracted into the tube under negative pressure for about 10 s. Thereafter, the PAXgene tube was gently shaken 8–10 times immediately. Then, the PAXgene tube was vertically placed in a metal wire tube rack at room temperature and transferred to a -20℃ freezer. The tube was frozen at -20℃ for 24 h before preservation at -80℃.

RNA extraction and sequencing

Tissue total RNA extraction was realized as per the manual of a Trizol kit (Invitrogen, Carlsbad, CA, USA). The RNA mixture was subjected to centrifugation at 12,000 xg, 4 °C, and the collected supernatant was subsequently placed into a new EP tube that contained 0.3 mL chloroform/isoamylol (24:1). Following the addition of chloroform, the sample was subjected to centrifugation. Then, the sample was separated into an aqueous-phase layer and an organic-phase layer, with RNA mainly existing in the aqueous phase. The aqueous phase liquid was loaded into a new centrifuge tube and centrifuged with the addition of isopropanol to obtain RNA precipitate. The RNA precipitate was rinsed with 75% ethanol to remove residual isopropanol and other impurities. The washed RNA precipitate was dissolved in an appropriate amount of RNase-free water. Subsequently, the total RNA was qualitatively and quantitatively analyzed using Thermo Fisher Scientific NanoDrop and Agilent 2100 Bioanalyzer. The extracted RNA was purified with the application of Oligo (dT)-coated magnetic beads. Using random hexamer primers, mRNA was reverse-transcribed to generate the first-strand cDNA and then synthesized into the second-strand cDNA, followed by amplification of the cDNA fragment using PCR; A DNA nanosphere with a molecular weight over 300 copies was prepared using phi29 amplification, and a single-ended sequence (50 bases) was generated on the BGIseq500 platform. The filtering of sequencing data was achieved with the SOAPNuke (v1.5.2) tool, and the filtered fragments were stored in FASTQ format and next mapped to the reference genome with the assistance of the HISAT2 software. Lastly, the Bowtie2 software was utilized to align the fragments to the reference genome and acquire the genomic coordinates of each fragment, and then the RSEM software was adopted to calculate the gene expression levels.

DEG screening

Integrated Differential Expression and Pathway Analysis (iDEP) online platform is a commonly used tool for genome and transcriptome Big data analysis [9]. RNA sequencing data were uploaded onto the iDEP 0.96 platform (http://149.165.154.220/idepg/) for logarithmic processing, with “Homo Sapiens” chosen as the species and “Read counts data” as the data type. The overall RNA levels of the two groups were determined through box plots. PCA analysis was conducted to screen sample clusters with high similarity and identify outlier samples. The genes obeying the thresholds: false discovery rate (FDR) < 0.05, |Fold Change| (FC) ≥ 4.0, were regarded as DEGs between healthy individuals and septic patients.

Screening of kidney-associated genes

The GTEx website establishes a resource database and related tissue libraries that allows the scientific community to investigate the relationship between genetic variants and gene expression in human tissues [10]. RNA sequencing data were selected in TPM format, and the tissue source was chosen as the kidney. The top 1000 genes with high enrichment in the kidney were screened. The Venny 2.1 tool was utilized to visualize the intersection targets of DEGs in sepsis and kidney-specific genes were obtained, which were potential research targets for SA-AKI.

Gene ontology (GO) analysis

GO, a community-based bioinformatics resource, utilizes ontologies to signify biological knowledge and supplies data illustrating the function of gene products [11]. GO enrichment analysis is a typical method for gene Big data analysis that aims to determine which specific functions or pathways the gene set is primarily enriched in, so as to infer whether these functions or pathways are significantly enriched under experimental conditions. At present, over 45,000 terms are displayed in the ontology, connected by nearly 134,000 relationships. Ontology covers three different aspects of gene function: molecular function (MF; a gene product’s activity at the molecular level), cellular component (CC; the position of gene product’s activity relative to biological structures), and biological process (BP; a larger biological program that utilizes a gene’s molecular function) [12]. GO analysis was conducted here with the Omicshare tool to globally view the functional enrichment of the aforementioned intersection genes, and the top 20 gene sets of the aforementioned aspects were enriched individually and visualized in interactive graphs.

Kyoto Encyclopedia of genes and genomes (KEGG) analysis

KEGG comprehensively comprises 15 manually-planned databases and one computationally-produced database. The databases in the system information category are PATHWAY, BRITE, and MODULE, which form a reference repository that aids in the comprehension of higher-level cell and organism system functions, involving metabolism, other cellular events, biological activities, and human diseases [13]. The KEGG enrichment analysis of the intersection genes was conducted on the Omicshare website to identify the top 20 pathways significantly enriched in the whole genome background, with p < 0.05 deemed as statistically significant.

Survival curve analysis

Survival curve analysis was mainly adopted to analyze the clinical significance of key targets. To further explore the relations between key genes and septic patients’ prognosis, this study performed a survival analysis of the GEO public GSE65682 dataset [10]. GSE65682 contained RNA sequencing results in peripheral blood from 479 septic patients as well as gene expression profiles and clinical prognostic data of each patient. Graphpad Prism7.0 software was adopted for survival analysis; p < 0.05 was regarded as statistically significant through a log-rank test.

Meta-analysis

Meta-analysis is a statistical tool that provides pooled estimates of effect from the data extracted from individual studies in the systematic review. To enlarge the sample scale for validating the accuracy of the differences in hub genes between the sepsis surviving and dead groups and to reduce the probability of false positives, this research downloaded three sepsis gene sequencing datasets, namely GSE54514, GSE63042, and GSE95233, from the GEO website. After homogenization, the above-mentioned dataset was assigned into sepsis surviving and dead groups. To validate data reliability, a comprehensive meta-analysis based on the R language package was made on a single gene of the same group in various datasets, and hub genes were further screened from the aforementioned intersection genes.

ROC analysis

ROC, which can describe the discriminant accuracy of diagnostic tests or predictive models, is a curve generated based on the true positive rate and false positive rate that reflect the relevance of the sensitivity of specific indicators to their specificity [14]. ROC curves were generated to further screen and validate hub genes with a diagnostic accuracy assessment. The area under the curve (AUC) can reflect the value of the diagnostic test, with a higher value, closer to 1.0 reflecting a higher diagnostic accuracy. When the AUC is equal to 0.5, it is regarded as no diagnostic value. Our research utilized the GSE95233 dataset from the GEO database [15] containing genetic sequencing data of blood samples from 51 septic patients and 21 healthy individuals. ROC curves were obtained by GraphPad Prism software (version 6.05).

Single-cell RNA sequencing

To further elucidate the localization of hub genes in diverse cell lines, peripheral blood specimens were harvested from 2 healthy participants, 1 SIRS patient, and 2 septic patients for 10 × single-cell RNA sequencing. The construction of sepsis single-cell RNA library, RNA sequencing, and data processing and analyses were completed under the guidance of Shanghai Oebiotech Co., Ltd. Raw data from high-throughput sequencing were presented in FASTQ format. The quality statistics of the samples and alignment to the reference genome were performed using the official 10× genomics Cell Ranger software. In light of the preliminary quality control results of Cellranger, we conducted in-depth quality control and data processing with the use of the Seurat package. Subsequently, low-quality cells such as double cells, multiple cells, or dead cells were eliminated. Finally, gene expression was subjected to PCA linear dimensionality reduction analysis and quality control. The tSNE nonlinear dimensionality reduction method was employed to visualize the PCA results from a two-dimensional spatial perspective.

Cell-type annotation

Genes that were significantly up-regulated in each cell type in contrast to other cell types were defined as marker genes. Marker genes in each cell population were screened using the FindAllMarkers function in the Seurat package, which were subsequently visualized with the use of VlnPlot and FeaturePlot functions. The SingleR package based on the HPCA reference dataset was utilized for cell-type annotation of sequencing data. Next, the SingleR package was adopted for a correlation analysis between the reference dataset and the expression profile of the cells to be identified. For sepsis single-cell bank construction, cell types showing the most significant correlation in the reference dataset were designated as the cells to be identified. The single-cell sequencing results were stored on the Oebiotech single-cell transcriptome platform (https://cloud.oebiotech.cn/task/category/scrna/).

Results

DEGs between sepsis and healthy controls

Quality control was performed on peripheral blood mRNA sequencing results from 22 septic patients and 10 healthy subjects. The box plots exhibited good sample homogeneity and comparability between these two groups (Fig. 2A). The PCA plot showed good discriminability between the two sample groups without outlier samples (Fig. 2B). The data of these two groups were subjected to differential expression analysis with |FC|≥4.0 and FDR < 0.05 serving as thresholds, yielding 1328 DEGs. There were 221 up-regulated genes (red dots) and 1107 down-regulated genes (blue dots) in the septic patients, while non-differentially expressed genes were displayed as gray dots (Fig. 2C-D).

Fig. 2

Screening of DEGs. A: The box plot shows good sample homogeneity and comparability between the control group and the sepsis group; B: The PCA graph shows good discrimination between the two groups without outlier samples. C-D: Differential expression analysis is performed on two sets of data with |FC|≥4.0 and FDR < 0.05 as thresholds, resulting in 1328 DEGs. The red dots represent 221 up-regulated genes in the septic participants, and the blue dots represent 1107 down-regulated genes in the septic participants

Kidney-associated genes in sepsis

The top 1000 genes with high enrichment in the kidney were selected from the GTEx database and downloaded for subsequent analysis. Then, 1328 DEGs in sepsis were intersected with 1000 kidney-related genes, yielding 40 kidney-specific genes that were linked to sepsis (Fig. 3A).

Fig. 3

Screening of sepsis-associated kidney-specific genes. A: 40 kidney-specific genes that are linked to sepsis in the intersection between 1328 DEGs in sepsis and 1000 kidney-related genes. B: Gene expression heatmap in RNA sequencing data

Functional enrichment analyses

By enriching top 20 gene sets from each aspect (BP, CC, and MF), it was illustrated that the intersection genes were mainly engaged in BPs such as receptor metabolic process, vacuolar transport, cellular catabolic process, lysosomal transport (Fig. 4A), principally localized in CCs like lytic vacuole, lysosome, intracellular vesicle, and lysosomal membrane (Fig. 4B), primarily involved in MFs including clathrin adaptor activity, cargo adaptor activity, molecular adaptor activity, and calcitonin receptor activity (Fig. 4C). The intersection genes were mainly enriched in pathways such as Endocrine and other factor-regulated calcium reabsorption, Synaptic vesicle cycle, Ubiquitin mediated proteolysis, and Protein processing in endoplasmic reticulum (Fig. 4D).

Fig. 4

Enrichment analyses. A: biological process (BP); B: cellular component (CC); C: molecular function (MF); D: KEGG signaling pathways

The potential clinical significance of CD74 and IL32 in sepsis.

The survival analysis based on the GSE65682 dataset (GEO) revealed that patients with more abundant CD74 and IL32 had significantly higher survival rates than those with a deficiency (p < 0.05), suggesting that CD74 and IL32 were positively relevant to septic patients’ prognosis, and abundant CD74 and IL32 contributed to better prognosis. CD74 and IL32 may become new research targets for SA-AKI, providing new directions for disease diagnosis and treatment (Fig. 5A-B).

Fig. 5

Survival curves. A-B: the patients with high CD74 and IL32 levels have markedly higher survival rates than those with low CD74 and IL32 levels (p < 0.05)

Meta-analysis of candidate genes

Three publicly available GEO sepsis datasets (GSE54514, GSE63042, and GSE95233) were submitted to a meta-analysis. For the multi-data heterogeneity test, the random effects model was chosen if I² > 50%, and the fixed effects model was chosen if I² ≤ 50%. Compared with the sepsis dead group, the sepsis surviving group showed statistically significant increases in CD74 and IL32 expression (Fig. 6A-B).

Fig. 6

Results of the meta-analysis. A: CD74; B: IL32

The diagnostic values of CD74 and IL32 shown by ROC curves

ROC curves for CD74-based and IL32-based prediction had AUC values of 0.983 (CD74) and 0.98 (IL32), respectively, showing relatively high diagnostic value (Fig. 7A-B). These two hub genes had high sensitivity and specificity for SA-AKI, which, therefore, provides an early diagnostic and therapeutic target for SA-AKI.

Fig. 7

ROC curves. A-B: The AUC values of ROC curves are 0.983 (CD74) and 0.98 (IL32)

Identification of marker genes

The number of high-quality cells for quantitative quality control of single-cell sequencing samples ranged from 4050 to 10,191. After quality control (double cells, multiple cells, and apoptotic cells were eliminated), the final cell quantity ranged from 3108 to 8509. Genes in each cell had an average value of 343 to 2337. PTPRC is a indicator for immune cells, CD8A is a biomarker for CD8⁺T cells, CD14 is a parameter for macrophages, CD3E is a marker for CD4⁺T cells, CD79A refers to a biomarker for B cells, FCGR3A is a natural killer cell marker, FLT3 is a dendritic cell indicator, PPBP is a platelet biomarker, and PRF1 is an initial T cell parameter (Fig. 8A).

Cell clustering based on marker genes

After clustering of marker genes by dimensionality reduction, the cells were assigned into 9 cell populations, with 3 and 5 representing macrophage population, 4 representing natural killer cell population, 1, 2, 6, and 8 representing T cell sets, 7 representing B cell population, and 9 representing platelets (Fig. 8B). Single-cell RNA sequencing results in Fig. 9A-D illustrated that CD74 was principally localized in cell populations 3 and 5 (macrophages), while IL32 was dominantly localized in cell sets 1, 2, 6, and 8 (T cells).

Fig. 8

Identification of marker genes and positioning analysis on cell lines. A: Identification of marker genes; B: Positioning analysis on cell lines

Fig. 9

Single-cell sequencing. A-D: Single-cell RNA sequencing shows that CD74 is dominantly presented in cell populations 3 and 5, namely macrophages, while IL32 is principally localized in cell sets 1, 2, 6, and 8, namely T cells

Discussion

AKI is a fundamental yet intricate syndrome that is induced by heterogeneous mechanisms and features a high incidence and mortality [16]. Approximately 10 -15% of in-patients develop AKI, with an incidence of AKI up to 50% in ICU patients [17]. Despite advancements in AKI therapy (including KRT), there has been limited improvement in the prognosis of AKI patients over the past few decades [18]. In this respect, it is imperative to seek novel diagnostic and therapeutic approaches that can prevent and curb the initiation and progression of AKI at the early stages. To this end, it is critical to more deeply and accurately identify the associations between AKI and other clinical diseases or factors causing AKI [19].

It is well known that sepsis increases the risk of AKI [20]. A multicenter prospective cohort study involving critically ill patients from 24 European countries revealed that the proportion of AKI patients among septic patients was 54% [21]. AKI is a common and life-threatening complication of sepsis, the global incidence of SA-AKI is approximately 11 million every year [22]. The pathogenesis of SA-AKI is intimately associated with renal hemodynamic abnormalities, inflammatory damage, and adaptive mechanisms.The previous criteria based on urine volume and creatinine are insufficient for early diagnosis of AKI. The emergence of new biomarkers may compensate for the lack of early diagnosis [23].

This study aims to identify biomarkers for early diagnosis of SA-AKI through bioinformatics analysis. This study exhibited the results of RNA sequencing and differential analyses on peripheral blood specimens (10 healthy participants and 22 septic patients). Meanwhile, the top 1000 highly enriched genes in the kidneys were selected from the GTEx database for further analysis. Ultimately, 40 targets were identified at the intersection of 1288 DEGs in sepsis and 960 kidney-associated genes. The meta-analysis further identified two potential hub genes, IL32 and CD74, with statistically significant high expression in SA-AKI. As depicted in survival curves, patients with higher CD74 and IL32 had increased survival rates, suggesting that the levels of CD74 and IL32 were positively related to patient prognosis, and their high expression was beneficial for patient prognosis. The AUC values of CD74 (0.983) and IL32 (0.98) displayed the diagnostic value of these two factors. Single-cell RNA sequencing showed that CD74 was principally expressed in macrophages, while IL32 was chiefly expressed in T cells.

Macrophage migration inhibitory factor (MIF), an extensively expressed pleiotropic cytokine, serves as a vital upstream modulator of innate immunity and accelerates diverse pathophysiological processes [24]. MIF can interact with CD74, a cell surface form of class II invariant chain (Ii) [25]. MIF binds to CD74 to induce its intramembrane proteolysis and secretion of its cytosolic intracellular domain (CD74-ICD), thereby mediating cell survival [26]. CD74 performs a crucial role in multiple crucial processes of the immune system, encompassing antigen processing, endocytic maturation, cell migration, and signal transduction [27]. The results of one study suggest that the APP-CD74 pathway between endothelial cells and macrophages may affect kidney injury repair and is a potential therapeutic target for antifibrosis [28].

In one study, the results of the immune infiltration analysis of patients with sepsis showed that lymphocytes, monocytes, macrophages, and NK cells were significantly involved. IL32 was also closely associated with these immune cells. The immune microenvironment is complex and strongly affects the pathogenesis of AKI [29]. IL32 can affect diverse cellular and physiological functions, such as cell death and survival, inflammatory reaction, and response to pathogens [30]. IL32 is a key cytokine involved in a variety of diseases that have extensive roles, including stimulating IL8, IL1, IL6, TNFα, and macrophage inflammatory protein 2 (MIP-2) as well as activating classical pro-inflammatory pathways such as NF-κB and p38 MAPK [30]. A clinical study report says that serum IL32 level increases in individuals with acute or chronic kidney injury. Elevated serum IL32 levels are positively linked to serum creatinine (SCr) levels and 24-hour urinary protein concentration [31]. It follows that the increase in IL32 levels can partly reflect the severity of early AKI.

Currently, the diagnosis of SA-AKI in clinical practice predominantly depends on microbial culture results, procalcitonin, high-sensitivity C-reactive protein, creatinine, and urine volume. The tests for the aforementioned indicators are time-consuming; Particularly, microbial culture often requires waiting for days. In the meantime, the above laboratory test results may be inaccurate due to various factors, such as incorrect sample collection methods, sample contamination, instrumental errors, etc. Based on the aforementioned factors, patients with SA-AKI may not be diagnosed promptly, resulting in delayed early treatment and consequently affecting prognosis. Therefore, the emergence of some novel biomarkers may greatly improve the early diagnostic rate of SA-AKI. This study utilized RNA sequencing technology in conjunction with bioinformatics methods to screen the intersection genes between sepsis and AKI (CD74 and IL32). In combination with ROC curves, survival analysis, and single-cell sequencing, CD74 and IL32 were further confirmed to be early diagnostic biomarkers for SA-AKI.

To conclude, this research provides preliminary evidence for CD74 and IL32 as novel biomarkers for SA-AKI. Relative to traditional laboratory test results, CD74 and IL32 have the advantages of rapid detection and high accuracy, with the potential to supply uncovered targets for early diagnosis and mechanistic investigations of SA-AKI. However, there are still certain shortcomings in current research. The sample size involved in this research is limited, and further prospective studies and experiments are necessary to validate the roles of CD74 and IL32 in the diagnosis and treatment of this disease. Our future research will focus on exploring the biological mechanisms and clinical relevance.