Elderly patients (>65 years old) diagnosed with sepsis were enrolled as the research subjects. For enrollment, patients had to meet the international diagnostic criteria for sepsis. The diagnosis of AKI was based on the 2012 Kidney Disease Improving Global Outcomes diagnostic criteria (58). These criteria involve a sharp decline in renal function within 48 h, manifested by an increase in serum creatinine >0.3 mg/dl (26.5 µmol/l) or an increase >50% (According to KDIGO, AKI is defined as an increase in serum creatinine levels by at least 0.3 mg/dl within 48 h with 1.5-fold being the baseline), patient age >65 years, expected survival time >3 days, and pathogen culture or laboratory test results showing gram-negative bacteria. The diagnostic criteria for sepsis were based on the Third International Consensus on the Management of Sepsis and Septic Shock (Sepsis-3) in 2016, which entails a joint diagnosis by >2 attending physicians. The exclusion criteria were as follows: i) Patients with tumor, acute stroke, rheumatic immune system disease and mental illness; ii) patients with viral myocarditis; iii) patients with severe hepatitis and cirrhosis; iv) patients who received anti-infective treatment before enrollment; v) patients with end-stage renal diseases; vi) patients who died or were discharged within 48 h after admission; and vii) patients who do not have complete clinical records or do not cooperate with urine sample collection.

The control group included healthy elderly individuals [elderly people with no previous history of chronic disease, age (75.29±5.46), female (58.82%)] who underwent a physical examination during the same period. All subjects were recruited between August 2020 and December 2021. The current study was approved by the Ethics Committee of Huadong Hospital Affiliated to Fudan University (Shanghai, China). All patients or their family members (some older people lose the ability to write) signed informed consent forms before enrollment.

Urine samples were collected within 24 h after the onset of sepsis in elderly patients admitted to Huadong Hospital Affiliated to Fudan University (Shanghai, China). Urine samples were collected in the morning of the physical examination day in healthy subjects. All urine samples were centrifuged at 845 x g for 10 min at 4˚C, and the supernatant was aliquoted into 1.8-ml Eppendorf tubes and frozen within 4 h of collection at -80˚C.

Total RNA was extracted using the mirVana^™ miRNA Isolation kit (cat. no. AM1561; Thermo Fisher Scientific, Inc.) and the samples were extracted for total RNA according to the standard procedure provided by the manufacturer, and the extracted total RNA was electrophoresed by an Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.) for quality control. The samples were then prepared for use by Agilent 2100 Bioanalyzer (Agilent Technologies, Inc.). The purified total RNA was subjected to 3' end-joining, 5' end-joining, reverse transcription, amplification, cDNA library size selection and purification according to the experimental instructions to complete the library construction of the sequenced samples. Total RNA was separated using 17% denaturing polyacrylamide gels and small RNAs between 10 and 60 nucleotides (nt) were collected. Then, 5'- and 3'-RNA adaptors were ligated to the small RNAs, followed by reverse transcription to produce cDNAs. These cDNAs were subsequently amplified by PCR and subjected to Solexa/Illumina sequencing by Shanghai Biotechnology Corporation. The libraries were created using the Qubit 2.0 Fluorometer (; Thermo Fisher Scientific, Inc.) for concentration and the Agilent 2100 for library size. Cluster generation and first-way sequencing primer hybridization were performed on the Illumina HiSeq sequencer's cBot (Illumina, Inc.) according to the appropriate procedure shown in the cBot User Guide. Sequencing reagents are prepared according to the Illumina User Guide and flow cells with clusters are loaded onto the machine. Single-ended sequencing was performed using the single-read program. The sequencing process was controlled by Illumina's data collection software (Illumina, Inc.) and real-time data analysis was performed. FastX software (https://anaconda.org/biobuilds/fastx-toolkit,fastx-toolkit 0.0.14) was used to preprocess the original reads for sequencing, remove linker sequences and low-quality sequences (including ambiguous base N sequences with a base quality <10 nt and length <18 nt), and provide (statistical analysis based on the processed results table and length distribution diagram. The sequences obtained through the Sanger miRBase database (https://www.mirbase.org; such as those of known ribosomal RNA, transfer RNA and repeat regions), RefSeq database (https://www.ncbi.nlm.nih.gov/refseq/) and other noncoding RNA databases, including the non-coding RNA, PIWI-interacting RNA (https://www.smallrnagroup.uni-mainz.de/piRNAclusterDB) and Rfam databases (https://rfam.xfam.org/), were compared, and the known miRNAs were annotated. The sequence obtained by sequencing was compared with the genome database corresponding to the species, the annotated reads were classified and counted, and the known miRNAs and various other types of small RNA molecules were identified and counted. The DEGseq R language package combined with Perl script was used to group samples according to the current requirements (such as the control and experimental groups) for comparative analysis of miRNA expression. In the differential analysis, the transcripts per million (TPM) formula (single miRNA reads x 10⁶/total reads) was used to present the data.

Total RNA was extracted using TransZol Up reagent (Beijing Transgen Biotech Co., Ltd.). Briefly, the process was as follows: A total of 1 ml TransZol UP reagent was added into 500 µl of urine sample, followed by mixing with 200 µl chloroform and centrifugation at 4˚C for 15 min at 10,000 x g. The aqueous phase containing the RNA was transferred to a new Eppendorf tube and the same volume (~500 µl) isopropyl alcohol was added. A total of 1 ml pre-cooled (4˚C) 75% ethanol was added, and centrifugation at 4˚C for 5 min at 7,500 x g. The RNA precipitate was air-dried, followed by dissolution in RNA solution buffer. cDNA synthesis was conducted with TransScript miRNA First-Strand cDNA Synthesis SuperMix (TransGen Biotech Co., Ltd.). The RT kit was used according to the manufacturer's protocol. The reactions were performed in a PCR instrument and the reaction program was set to 37˚C for 1 h and 85˚C for 5 sec. The Hieff qPCR SYBR Green Master Mix kit (Shanghai Yeasen Biotechnology Co., Ltd.) was used to perform RT-qPCR assays. The qPCR cycling conditions were 95˚C for 10 sec, 55˚C for 30 sec and 72˚C for 30 sec for 40 cycles. Relative quantification of hsa-miR-31-5p, hsa-miR-151a-3p, hsa-miR-142-5p and hsa-miR-16-5p was performed by normalization to U6 small nuclear (sn)RNA expression levels. The 2^-ΔΔCq method was used to analyze miRNA levels (59). The primer sequences used are presented in Table SI.

Bioinformatics analysis was performed to preprocess sequencing data and analyze the results. Bioinformatics analyses included miRNA expression quantitative analysis, expression correlation analysis, miRNA differential expression analysis, differential miRNA target gene prediction, and Gene Ontology (GO; http://www.geneontology.org) and Kyoto Encyclopedia of Genes and Genomes (KEGG; http://www.kegg.jp/kegg/pathway.html) enrichment analysis of different miRNA target genes. The results of the GO enrichment analysis were displayed in plots, where Rich Factor=(the number of miRNA target genes in a GO term/the number of all target genes that can correspond to the GO database)/(the number of genes contained in a GO term/the total number of genes that can correspond to the GO database). The greater the Rich Factor is, the greater the degree of enrichment, while the smaller the Q-value is, the more significant the enrichment (60).

All the experimental results in this study were verified by three biological repetitions to ensure the accuracy of the experimental results. Data analysis was performed using SPSS 23 software (IBM Corp.). Data are presented as the mean ± SEM. Each experiment, controlling a single variable and setting up two experimental groups (the AKI and non-AKI group) and a control group, had at least three biological repetitions. EdgeR (http://www.R-project.org/) was used to analyze the difference in miRNAs between samples. After obtaining the P-value, multiple hypothesis test correction was performed, and the P-value threshold was determined by controlling the false discovery rate, thereby providing the Q-value. Fold-change was calculated as the differential expression based on the TPM value. The screening conditions for differential genes were as follows: Q-value ≤0.05; fold-change ≥2. One-way ANOVA followed by Dunnett's multiple comparisons test was used to compare the groups. Receiver operating characteristic (ROC) curves were plotted to analyze the predictive value of miR-31-5p, miR-151a-3p, miR-142-5p and miR-16-5p for the prognosis and 28-day mortality of elderly patients with sepsis. The ROC curve analysis and the derived area under the curve (AUC) statistic provide a global and standardized appreciation of the accuracy of a marker or a composite score for predicting an event (61). ROC curves were generated by plotting sensitivity against 1-specificity. P<0.05 was considered to indicate a statistically significant difference.

Patients were diagnosed with sepsis, and the main infection sites were the lung, urinary system and gastrointestinal tract. According to the general clinical data, the 74 study subjects included 17 healthy elderly patients, 29 septic patients with AKI and 28 septic patients without AKI. The mean age was 81 years, with a range of 65-97 years. After 28 days of follow-up in the observation group, 18 patients of the 57 patients with sepsis succumbed, accounting for 31.58% (Table SII).

Details of the patients are provided in Table I. Heatmap and cluster analysis demonstrated that there were differentially expressed miRNAs between the sepsis AKI and non-AKI groups. The sepsis AKI and non-AKI groups were compared with the normal group. Among the differentially expressed miRNAs in the sepsis AKI group, six miRNAs were upregulated compared with the normal group (Fig. 1A). Among the differentially expressed miRNAs in the sepsis non-AKI group, 28 miRNAs were upregulated compared with the normal group (Fig. 1C). The volcano plots show the differentially expressed miRNAs under the two different conditions (AKI vs. non-AKI groups) (Fig. 1B and D).

Firstly, the number of target genes corresponding to the three GO elements, biological process, cellular component and molecular function, was counted. Only the top 30 GO entries are shown in Fig. 2A. Using the same principle as for GO enrichment analysis, KEGG pathway enrichment analysis was also performed for target genes of differentially expressed miRNAs, and the results are shown in Fig. 2B.

There were six upregulated miRNAs in the AKI group (Table II) and 27 in the non-AKI group (Table SIII) compared with the control. Analysis of the data revealed that several miRNAs in the sepsis AKI and non-AKI groups were differentially expressed and upregulated compared with the control. A high expression trend was found in the sepsis AKI and non-AKI groups.

To verify the expression levels of these four miRNAs in sepsis, 17 samples from healthy controls, 29 samples from patients with sepsis and AKI and 28 samples from patients with sepsis without AKI were collected. Details of the patients are provided in Table SII. Compared with those in the control group, the expression levels of miR-31-5p, miR-151a-3p, miR-142-5p and miR-16-5p were significantly increased in the sepsis AKI and sepsis non-AKI groups (Fig. 3A-D).

ROC curve analysis of miR-31-5p, miR-151a-3p, miR-142-5p and miR-16-5p was performed to assess their predictive value in the diagnosis of AKI in elderly patients with sepsis. Fig. 4 shows that the AUC for miR-142-5p and miR-16-5p expression was 0.746 and 0.820, respectively, indicating a good predictive value of miR-142-5p and miR-16-5p for patients with sepsis-induced AKI. The AUC for miR-31-5p and miR-151-3p expression was 0.416 (P=0.274) and 0.450 (P=0.513) respectively, with no statistical significance.

ROC curves were generated to evaluate the predictive value of miR-31-5p, miR-151a-3p, miR-142-5p and miR-16-5p for the 28-day mortality in patients with sepsis (Fig. 5). The AUC for each miRNA was 0.668, 0.747,0.714 and 0.838, respectively. These results indicated a good predictive value of miR-31-5p, miR-151a-3p, miR-142-5p and miR-16-5p in the prognosis of 57 patients with sepsis.