Plasma VGF as a Predictive Biomarker for Early Acute Kidney Injury in Critically Ill Elderly Patients: A Combined Clinical and Mendelian Randomization Study

Summary

Objectives:

To assess whether VGF nerve growth factor inducible (VGF) can serve as a reliable predictor of acute kidney injury (AKI) among patients requiring intensive care.

Methods:

We first conducted a Mendelian randomization (MR) analysis to explore the potential causal link between plasma VGF levels and AKI risk. Subsequently, 128 critically ill individuals admitted to our intensive care unit were prospectively enrolled. Within 24 hours of either hospital admission or clinical decline, both blood and urine specimens were obtained. We measured VGF concentrations in plasma and urine, along with plasma cystatin C (pCysC), plasma and urinary neutrophil gelatinase-associated lipocalin (NGAL), and urinary N-acetyl-β-D-glucosaminidase (NAG). Kidney function was assessed using serum creatinine and hourly urine output. Receiver operating characteristic (ROC) curve analysis was employed to evaluate each biomarker ‘s predictive capacity for AKI.

Results:

The inverse-variance weighted (IVW) method revealed a statistically significant positive association between plasma VGF levels and AKI development. The area under the ROC curve (AUC) values for individual markers were as follows: pVGF (0.651), uVGF (0.483), pCysC (0.750), pNGAL (0.804), uNGAL (0.803), and uNAG (0.632). When pVGF, pCysC, and pNGAL were combined, the AUC increased to 0.861, indicating substantially better predictive performance than any single biomarker.

Conclusions:

Plasma VGF shows promise as a predictive biomarker for AKI in critically ill populations. Its diagnostic utility is considerably strengthened when measured alongside pCysC and pNGAL.

Introduction

Acute kidney injury (AKI) represents a frequently encountered and severe clinical condition marked by an abrupt deterioration of kidney function. According to pooled data from multiple meta-analyses, the occurrence rate of AKI within intensive care units (ICUs) substantially exceeds that observed in general hospital wards, with estimates ranging from 30% to over 50% [1,2,3]. This condition correlates closely with elevated death rates, extended hospital stays, and substantial healthcare expenditures [4,5,6]. Timely identification and therapeutic intervention have the potential to attenuate renal impairment.

Serum creatinine and urine output remain the standard diagnostic markers for renal dysfunction. However, both are influenced by numerous non-renal factors and tend to rise only after significant renal impairment has occurred, limiting their value for early detection [7,8,9]. Consequently, there is a pressing need for highly sensitive and specific biomarkers that can identify AKI at an earlier stage, thereby facilitating timely clinical intervention. Several biomarkers, including plasma cystatin C (pCysC), neutrophil gelatinase-associated lipocalin (NGAL), and N-acetyl-β-D-glucosaminidase (NAG), have shown considerable potential for early AKI prediction [10,11,12,13].

Studies indicate that VGF nerve growth factor inducible (VGF) protein levels, particularly the VGF-derived peptide TLQP-21, significantly increase during early AKI, suggesting that VGF/TLQP-21 may serve as valuable biomarkers for renal function [14]. Our Mendelian randomization (MR) analysis further confirmed a causal association between plasma VGF concentration and AKI. Therefore, the present study measured both plasma and urinary VGF levels to assess their correlation with AKI and compared these findings with established biomarkers (CysC, NGAL, and NAG) to assess the value of VGF as a diagnostic tool for the early identification of AKI in critically ill patients.

Methods

Study design

Plasma VGF was designated as the exposure variable, with significantly associated single nucleotide polymorphisms (SNPs) serving as instrumental variables. Acute kidney injury was defined as the outcome. The analysis met the 3 core MR criteria: relevance, independence, and the exclusion restriction.

Data sources

Exposure data were obtained from the GWAS catalog (https://www.ebi.ac.uk/gwas/), and outcome data were obtained from the IEU). OpenGWAS database (https://gwas.mrcieu.ac.uk/). Both genetic datasets originated from populations of European descent.

The SNP selection

We selected instrumental variables by applying a genome-wide significance level (P < 5 × 10⁸) and a linkage disequilibrium threshold (r² = 0.001, kb = 10,000). The SNPs with statistics >10 were retained to avoid weak instrument bias and ensure independence. The SNPs associated with known confounding factors were excluded.

Mendelian randomization analysis

The primary analytical method was inverse-variance weighting, with supplementary sensitivity analyses conducted via WME, MR-Egger regression, simple mode (SM), and weighted mode (WM). Sensitivity analyses assessed robustness evaluation of horizontal pleiotropy and heterogeneity. Horizontal pleiotropy was tested using MR-Egger and MR-PRESSO (p > 0.05 indicating absence). Heterogeneity was assessed using the Cochrane's Q test (p > 0.05 indicating no heterogeneity), permitting the use of fixed-effect IVW results. The leave-one-out approach was used to evaluate SNP stability.

Clinical data analysis

Between June 2022 and October 2023, a total of 128 patients (85 male, 43 female) admitted to the Department of Critical Care Medicine at the Second Affiliated Hospital of Wenzhou Medical University were included in this study. Acute kidney injury developed in 38 patients, whereas AKI did not develop in 90 patients. The study received approval from the hospital's medical ethics committee.

Sample collection and processing

Within 24 hours after admission or clinical deterioration, anticoagulated blood (3 mL) and urine (3 mL) were collected. The samples were centrifuged at 3000 rpm for 15 minutes, and the resulting supernatants (plasma and urine) were aliquoted into EP tubes and stored at -80 °C for subsequent analysis of plasma VGF and urinary VGF concentrations, as well as pCysC, plasma neutrophil gelatinase-associated lipocalin (pNGAL), urinary neutrophil gelatinase-associated lipocalin (uNGAL), and unique green fluorescent protein (uNAG) levels.

Biomarker measurement

Plasma and urinary concentrations of CysC, NGAL, and NAG were quantified using immunoturbidimetric assays, whereas plasma (pVGF) and urinary (uVGF) were determined using enzyme-linked immunosorbent assay. Routine hematologic, hepatic, and renal function tests, as well as electrolyte and blood lactate assessments, were also performed. Urine output was monitored every 6 hours.

Renal function assessment

Acute kidney injury was diagnosed according to the Kidney Disease: Improving Global Outcomes criteria: i) serum creatinine (Scr) increase ≥26.5 μmol/L within 48 hours; ii) Scr increase ≥1.5× baseline within 7 days; iii) urine output <0.5 mL/kg/h for more than 6 hours.

Monitoring indicators

Clinical data collected included the presence of hypertension, diabetes, and smoking history, as well as the use of renal replacement therapy and vasopressors, and ICU discharge outcomes (improvement or deterioration). Laboratory parameters recorded included white blood cell count, neutrophil count, lymphocyte count, hemoglobin level, platelet count, blood lactate level, serum albumin level, alkaline phosphatase level, cholinesterase level, and electrolyte levels.

Statistical analysis

The MR analysis was performed using R v4.4.2 (TwoSampleMR package v0.6.14). All other analyses were conducted using SPSS v17.0. Comparisons for continuous variables were performed with t-tests, while categorical variables were analyzed using chi-square tests, with a significance threshold of p < 0.05. Binary logistic regression was applied to determine independent predictors of AKI. The ROC curves were generated for individual and combined predictors derived from the logistic regression model. The Youden index was calculated to determine the optimal cutoff values based on the ROC-derived sensitivity and specificity.

Results

Mendelian randomization analysis

A total of 12,716,125 plasma VGF-associated SNPs were extracted from the GWAS Catalog. After quality control and selection, 115 independent SNPs were retained for MR analysis.

Mendelian randomization results

Inverse-variance weighted (IVW) showed a positive causal association between plasma VGF concentration and acute kidney failure [odds ratio = 1.09, 95% CI (1.01, 1.17), p = 0.023].

Sensitivity analysis

Cochrane's Q test revealed no evidence of significant heterogeneity across the instrumental variables (IVW Q = 94.88, p = 0.750; MR-Egger Q = 96.39, p = 0.737), supporting the robustness of the IVW model as the primary analytical method. The MR-Egger regression indicated no horizontal pleiotropy (p > 0.05), confirming the validity and stability of the causal estimates. Overall, these results support a causal association between plasma VGF levels and the risk of AKI (Fig. 1,Fig. 2,Fig. 3,Fig. 4).

Fig. 1.

Scatter plot of SNPs associated with VGF and AKI.

GVF: nerve growth factor inducible, SNP: single nucleotide polymorphisms, AKI: acute kidney injury, MR: mendelian randomization.

Fig. 2.

Leave-one-out analysis of SNPs associated with VGF and the risk of AKI.

VGF: nerve growth factor inducible, SNP: single nucleotide polymorphisms, AKI: acute kidney injury.

Fig. 3.

Funnel plot visualizing overall heterogeneity of MR estimates for the effect of VGF on AKI risk.

VGF: nerve growth factor inducible, AKI: acute kidney injury, MR: mendelian randomization.

Fig. 4.

Forest plot of SNPs associated with VGF and the risk of AKI.

VGF: nerve growth factor inducible, SNP: single nucleotide polymorphisms, AKI: acute kidney injury.

Patient group comparisons

Significant differences were observed between patients with AKI (n = 38) and non-AKI (n = 90) groups in terms of creatinine, blood urea nitrogen, uric acid, white blood cells, neutrophils, lactate, serum potassium, serum phosphate, pVGF, pCysC, pNGAL, and uNGAL (p < 0.05). The proportions of patients receiving renal replacement therapy or norepinephrine also differed significantly between groups (p < 0.05) (Table 1).

Table 1.

Baseline demographic and clinical characteristics of study participants, stratified by AKI status.

Variables	Non AKI (n = 90)	AKI (n = 38)	T/χ2	P - value
Male/Female	57/33	28/10	1.283	0.309
Age (years)	70 ± 10	70 ± 9	0.190	0.850
SCr	78.3 ± 52.8	176.3 ± 94.3	–6.016	0.000
Urine volume	1858 ± 864	1622 ± 1543	0.886	0.380
Urea nitrogen	7.43 ± 5.82	15.6 ± 11.81	–4.062	0.000
Uric acid	258 ± 116	432 ± 220	–4.621	0.000
Urea/SCr	0.10 ± 0.06	0.09 ± 0.05	1.219	0.225
CRRT (Y/N)	6/84	7/31	4.046	0.044
Norepinephrine (Y/N)	44/46	27/11	5.314	0.021
Outcome (Improvement/Deteriorate)	62/28	24/14	0.398	0.528
High blood pressure (Y/N)	51/39	19/19	0.479	0.561
Diabetes (Y/N)	27/63	12/26	0.031	0.859
Smoking (Y/N)	17/73	11/27	1.582	0.209
Hemoglobin	103 ± 21	98 ± 24	1.382	0.169
White blood cells	10.76 ± 5.23	15.24 ± 8.79	–2.93	0.005
Neutrophils	9.14 ± 5.19	13.27 ± 8.16	–2.891	0.006
Neutrophil ratio	0.92 ± 0.79	0.87 ± 0.11	0.431	0.667
Lymphocytes	0.93 ± 0.53	1.18 ± 2.15	–1.057	0.293
Lymphocyte ratio	0.10 ± 0.08	0.08 ± 0.09	1.242	0.216
Platelets	179 ± 96	162 ± 100	0.920	0.359
PLR	270.9 ± 284.7	280.6 ± 346.6	–0.165	0.869
Lactic acid	1.64 ± 0.97	3.27 ± 4.41	–2.257	0.03
Na	141.8 ± 6.8	144.6 ± 8.8	–1.873	0.063
K	3.89 ± 0.45	4.23 ± 0.77	–2.56	0.014
CL	109.2 ± 5.6	111.0 ± 8.4	–1.462	0.146
Ca	1.93 ± 0.45	1.97 ± 0.42	–0.385	0.701
P	0.88 ± 0.36	1.14 ± 0.53	–2.757	0.08
Mg	0.74 ± 0.19	0.77 ± 0.21	–0.756	0.451
Albumin	31.2 ± 6.1	29.9 ± 4.4	1.141	0.256
Alkaline phosphatase	96 ± 60	137 ± 258	–0.975	0.336
Cholinesterase	4365 ± 1756	3929 ± 1559	1.303	0.195
pCysC	1.27 ± 0.76	2.02 ± 1.09	–3.85	0.000
pNGAL	256 ± 241	639 ± 545	–4.159	0.000
uNGAL	358 ± 846	1263 ± 1379	–3.705	0.001
uNAG	29.5 ± 31.9	53.5 ± 87.9	–1.620	0.113
pVGF	540 ± 151	697 ± 287	–3.187	0.003
uVGF	364 ± 182	383 ± 224	0.029	0.977

AKI: acute kidney injury, SCr: serum creatinine, CRRT: continuous renal replacement therapy, PLR: platelet-lymphocyte ratio, Na: sodium, K: potassium, Cl: chloride, Ca: Calcium, P: phosphorus, MG: magnesium, pCysC: plasma cystatin C, uNGAL: urinary neutrophil gelatinase-associated lipocalin, uNAG: urinary N-acetyl-β-D-glucosaminidase, pVGF: plasma VGF nerve growth factor inducible, uVGF: urinary VGF nerve growth factor inducible.

Logistic regression analysis for AKI

Univariate regression (with AKI as the dependent variable and biomarkers as the independent variable) identified variables with p<0.1 for multivariate analysis. Multivariate analysis identified pVGF (p = 0.001), pCysC (p = 0.001), and pNGAL (p = 0.023) as significant risk factors for AKI.

ROC curve analysis for AKI prediction

Based on multivariate results, the combined predictive probability of pVGF, pCysC and pNGAL were calculated. The ROC analysis yielded the following area under curve (AUCs): pVGF = 0.651 (p = 0.007), uVGF = 0.483 (p = 0.764), pCysC = 0.750 (p < 0.001), pNGAL = 0.804 (p < 0.001), uNGAL = 0.803 (p < 0.001), and uNAG = 0.632 (p = 0.020). The combined AUC for pVGF, pCysC, and pNGAL was 0.861(p < .001) (Fig. 5).

Fig. 5.

ROC curves of biomarkers in predicting AKI occurrence.

ROC: receiver operating characteristic, AKI: acute kidney injury, pCysC: plasma cystatin C, uNGAL: urinary neutrophil gelatinase-associated lipocalin, uNAG: urinary N-acetyl-β-D-glucosaminidase, pVGF: plasma VGF nerve growth factor inducible, uVGF: urinary VGF nerve growth factor inducible.

Discussion

In humans, VGF is composed of 615 amino acids, whereas in rats it consists of 617 amino acids [16], is processed into multiple biologically active fragments by endopeptidase cleavage at specific sites [17]. A VGF shows extensive expression throughout the central and peripheral nervous systems, as well as in the pituitary gland, adrenal medulla, gastrointestinal endocrine cells, and pancreatic β-cells [18,19]. The VGF-derived peptides play key roles in energy homeostasis, metabolic regulation, and reproductive function [20,21]. However, the detailed role of VGF in both renal physiology and pathophysiology has yet to be fully elucidated. Kim et al. [14] observed a 200–2000-fold induction of the VGF gene in AKI mouse models (ischemia-reperfusion, cisplatin nephrotoxicity, rhabdomyolysis). Immunoblotting revealed a marked increase in VGF protein expression during early AKI, along with elevated renal TLQP-21, a VGF-derived peptide. Systemic TLQP-21 administration partially reversed the exacerbated renal Injury in VGF-deficient mice, suggesting a protective role for TLQP-21 in AKI. Moreover, in renal tubular epithelial cells, the increase in VGF expression following injury is transcriptionally governed by Sox9, which identifies VGF as a stress-inducible gene with reno-protective properties.

We employed MR to confirm the relationship between VGF and AKI, thereby enhancing the credibility of our results. The MR serves as a technique for causal inference that employs genetic variants as instrumental variables to emulate randomized controlled trials, thereby mitigating confounding and reverse causation [22,23,24]. Leveraging the work of Loya H et al. [25], we obtained plasma VGF-associated SNPs from the GWAS Catalog and ARF-related SNPs from the IEU OpenGWAS database. Our MR study confirms a positive causal relationship between plasma VGF and ARF, supporting the diagnostic potential of VGF in AKI. Moreover, we found that the AUC of pVGF for predicting the onset of AKI was 0.651, indicating moderate predictive ability. Although the predictive performance of pVGF was lower than that of pCysC and pNGAL, its combination with these biomarkers significantly enhanced overall predictive accuracy.

The reported accuracies for NGAL and CysC show notable variation across studies. Jean-Maxime Côté et al. [26] reported plasma NGAL AUC = 0.66 and a urinary NGAL AUC = 0.80 for discriminating AKI. In a cohort of 187 patients with heart failure–associated AKI, serum NGAL AUC was 0.667, compared with 0.732 for urinary NGAL [27]. A meta-analysis of 28 studies found that urinary, serum, and plasma NGAL AUCs for infection-related AKI were 0.92, 0.87, and 0.84 [28]. Although urinary NGAL often outperforms serum and plasma NGAL, another meta-analysis reported urinary NGAL AUC values of 0.74 and plasma NGAL AUC values of 0.77 [29]. This variability likely reflects differences in sampling time, disease etiology, and patient heterogeneity. The CysC also demonstrates value for early AKI diagnosis. In a study involving 49 patients with cardiac surgery–associated AKI, P-CysC AUCs at 4 hours, 12 hours, and 48 hours were 0.88, 0.93, and 0.74, slightly exceeding those of P-NGAL (0.82, 0.89, 0.74) [30]. Pilarczyk K et al. [31] reported preoperative, 2-hour postoperative, and 6-hour postoperative serum CysC AUCs of 0.822, 0.713, and 0.872 in patients undergoing thoracic aortic surgery, outperforming urinary [TIMP-2]*[IGFBP7]. N-acetyl-β-D-glucosaminidase is another AKI biomarker. In a cohort of 358 septic patients, serum CysC and uNAG achieved AUCs of 0.807 and 0.817, with a combined AUC of 0.831 for AKI prediction [32]. In 605 patients undergoing intracranial tumor resection, serum CysC and uNAG exhibited AUCs of 0.789 and 0.787 [33]. Similarly, a multicenter Chinese study (n = 1162) reported serum CysC and uNAG AUCs of 0.738 and 0.650, for AKI prediction in critically ill patients, with a combined AUC of 0.756, consistent with the present findings [5].

Our findings additionally indicate a link between plasma VGF levels and the development of AKI. However, the ability of any single biomarker to predict outcomes is still constrained. Integrating multiple biomarkers improves diagnostic performance and clinical applicability. Given the potential protective role of the VGF-derived peptide TLQP-21 during AKI, quantifying TLQP-21 may further enhance predictive accuracy, meriting additional investigation.

In conclusion, MR and clinical validation identified plasma VGF as a novel biomarker for AKI, whose predictive accuracy improves significantly when combined with plasma CysC and pNGAL. Although animal studies suggest that the VGF-derived peptide TLQP-21 may be a superior biomarker, its clinical detection remains technically demanding and costly. Therefore, this study focused on VGF as an initial target, with future research planned to evaluate TLQP-21's potential in the early diagnosis of AKI.

Ethical approval

The study was conducted in accordance with the Declaration of Helsinki and Approved by the Institutional Review Board of the Second Affiliated Hospital of Wenzhou Medical University (Approval numbers: LCKY2020-352, approved on 2 November 2020). Written informed consent was obtained from each patient and the copy of the written consent is available upon request.

Funding statement

This research was supported by the Wenzhou Science and Technology Bureau (Grant No. Y2020523 to XianYang Guo, MMed).

Acknowledgments

The authors would like to thank their respective departments for their support in this research. The authors would like to thank Zibo Yimore Translation CO. LTD for the English language editing.

Disclosure

The authors declare no conflicting interests, support or funding from any drug company.

AI disclosure statement

During the preparation of this manuscript, the authors used DeepSeek for English language translation of the original Chinese manuscript, followed by professional editing services. After using this tool, the authors reviewed and edited the content as needed and take full responsibility for the content of the publication. No generative AI was used for data analysis, figure creation, or any other part of the research.