Diagnostic model for predicting hyperuricemia based on alterations of the gut microbiome in individuals with different serum uric acid levels

Abstract

Background: We aimed to assess the differences in the gut microbiome among participants with different uric acid levels (hyperuricemia [HUA] patients, low serum uric acid [LSU] patients, and controls with normal levels) and to develop a model to predict HUA based on microbial biomarkers.

Methods: We sequenced the V3-V4 variable region of the 16S rDNA gene in 168 fecal samples from HUA patients (n=50), LSU patients (n=61), and controls (n=57). We then analyzed the differences in the gut microbiome between these groups. To identify gut microbial biomarkers, the 107 HUA patients and controls were randomly divided (2:1) into development and validation groups and 10-fold cross-validation of a random forest model was performed. We then established three diagnostic models: a clinical model, microbial biomarker model, and combined model.

Results: The gut microbial α diversity, in terms of the Shannon and Simpson indices, was decreased in LSU and HUA patients compared to controls, but only the decreases in the HUA group were significant (P=0.0029 and P=0.013, respectively). The phylum Proteobacteria (P<0.001) and genus Bacteroides (P=0.02) were significantly increased in HUA patients compared to controls, while the genus Ruminococcaceae_Ruminococcus was decreased (P=0.02). Twelve microbial biomarkers were identified. The area under the curve (AUC) for these biomarkers in the development group was 84.9% (P<0.001). Notably, an AUC of 89.1% (P<0.001) was achieved by combining the microbial biomarkers and clinical factors.

Conclusions: The combined model is a reliable tool for predicting HUA and could be used to assist in the clinical evaluation of patients and prevention of HUA.

Keywords: diagnostic model; gut microbiome; hyperuricemia; nomogram; uric acid.