Investigating potential biomarkers of acute pancreatitis in patients with a BMI>30 using Mendelian randomization and transcriptomic analysis

Abstract

Background: Acute pancreatitis (AP) has become a significant global health concern, and a high body mass index (BMI) has been identified as a key risk factor exacerbating this condition. Within this context, lipid metabolism assumes a critical role. The complex relationship between elevated BMI and AP, mediated by lipid metabolism, markedly increases the risk of complications and mortality. This study aimed to accurately define the correlation between BMI and AP, incorporating a comprehensive analysis of the interactions between individuals with high BMI and AP.

Methods: Mendelian randomization (MR) analysis was first applied to determine the causal relationship between BMI and the risk of AP. Subsequently, three microarray datasets were obtained from the GEO database. This was followed by an analysis of differentially expressed genes and the application of weighted gene coexpression network analysis (WGCNA) to identify key modular genes associated with AP and elevated BMI. Functional enrichment analysis was then performed to shed light on disease pathogenesis. To identify the most informative genes, machine learning algorithms, including Random Forest (RF), Support Vector Machine-Recursive Feature Elimination (SVM-RFE), and Least Absolute Shrinkage and Selection Operator (LASSO), were employed. Subsequent analysis focused on the colocalization of the Quantitative Trait Loci (eQTL) data associated with the selected genes and Genome-Wide Association Studies (GWAS) data related to the disease. Preliminary verification of gene expression trends was conducted using external GEO datasets. Ultimately, the diagnostic potential of these genes was further confirmed through the development of an AP model in mice with a high BMI.

Results: A total of 21 intersecting genes related to BMI>30, AP, and lipid metabolism were identified from the datasets. These genes were primarily enriched in pathways related to cytosolic DNA sensing, cytokine‒cytokine receptor interactions, and various immune and inflammatory responses. Next, three machine learning techniques were utilized to identify HADH as the most prevalent diagnostic gene. Colocalization analysis revealed that HADH significantly influenced the risk factors associated with BMI and AP. Furthermore, the trend in HADH expression within the external validation dataset aligned with the trend in the experimental data, thus providing a preliminary validation of the experimental findings.The changes in its expression were further validated using external datasets and quantitative real-time polymerase chain reaction (qPCR).

Conclusion: This study systematically identified HADH as a potential lipid metabolism-grounded biomarker for AP in patients with a BMI>30.

Keywords: AP; BMI; Lipid metabolism; Machine learning.

Fig. 1

The flow chart of this study

Fig. 2

Modeling diagram. After the mice were anesthetized, the abdomen was disinfected with a cloth, the abdomen was opened layer by layer, the pancreas was exposed, the pancreatic duct was located, and sodium taurocholate or physiological saline was injected

Fig. 3

Scatter plots of causality in AP on 3 BMI datasets. The slope of each line corresponding to the estimated MR effect in different models. A ukb-a-248. B ukb-b-2303. C ukb-b-19953

Fig. 4

Leave-one-out sensitivity tests. The MR results of the remaining IVs were calculated after removing the IVs one by one. A ukb-a-248. B ukb-b-2303. C ukb-b-19953

Fig. 5

Boxplots of gene expression before and after standardization for 2 selected GEO datasets. A Before standardization. B After standardization

Fig. 6

Identification of ORDEGs. A Volcano plot showing DEGs in the BMI>30 and BMI <30 samples. B Soft-thresholding filtering. C Clustering dendrogram of genes. D Correlation heatmap of gene modules and clinical features. E Venn diagram showing the overlap of module genes and DEGs

Fig. 7

Identification of DEGs and AP-related module genes. A Volcano plot showing DEGs in the AP and normal samples. B Soft-thresholding filtering. C Clustering dendrogram of genes. D Correlation heatmap of gene modules and clinical features. E Venn diagram showing the overlap of module genes and DEGs

Fig. 8

Enrichment analysis of the intersecting genes. A A total of 21 overlapping genes were identified among the APRDEGs, ORDEGs and LMRGs. B Gene Ontology (GO) enrichment results of 21 intersecting genes. C Kyoto Encyclopedia of Genes and Genomes (KEGG) enrichment results for 21 intersecting genes

Fig. 9

Box diagram of the proportions of 28 types of immune cells. A dataset with a BMI >30 showed a difference in infiltration between the two groups. B AP dataset showing the difference in infiltration between the two groups

Fig. 10

Selection of potential diagnostic biomarkers with machine learning methods. A LASSO regression analysis was applied to screen diagnostic biomarkers based on the 21 intersecting genes in the AP dataset. The genes with the lowest binominal deviance were identified as the most suitable candidates. B The results of the Gini coefficient method for the random forest classifiers in the AP dataset. The x-axis represents genetic variables, and the y-axis represents importance indices. C The number of CDEGs with the lowest error and highest accuracy were considered the most suitable candidates via the SVM-RFE algorithm in the AP dataset. D Venn diagram visualizing the overlap of selected biomarkers between 3 algorithms, yielding 5 genes selected as candidate biomarkers. E LASSO regression analysis was applied to screen diagnostic biomarkers based on the 21 intersecting genes in the BMI>30 dataset. The genes with the lowest binominal deviance were identified as the most suitable candidates. F The results of the Gini coefficient method for the random forest classifiers in the BMI>30 dataset. The x-axis represents genetic variables, and the y-axis represents importance indices in the BMI>30 dataset. G The number of CDEGs with the lowest error and highest accuracy were considered the most suitable candidates via the SVM-RFE algorithm in the AP dataset. H Venn diagram visualizing the overlap of selected biomarkers between 3 algorithms, yielding 4 genes selected as candidate biomarkers

Fig. 11

Diagnostic values of the candidate biomarkers BMI>30 and AP assessed by expression comparison. A Comparison of HADH gene expression between the AP and normal groups in the GSE109227 test dataset. B Comparison of HADH expression between the BMI >30 and BMI< 30 groups in the GSE166047 test dataset (* P < 0.05, *** P < 0.001)

Fig. 12

The results of q-PCR analysis of mRNA expression levels are shown. The expression levels of HADH in patients with a high BMI were significantly greater than those in patients with AP and a high BMI (** P < 0.01)