Machine learning identifies the role of purine metabolism disorders in the pathogenesis of gout

New machine learning analysis of intestinal microbiomes can identify purine metabolic pathways as the main contributors to distinguish gout from other groups, revealing important intestinal microbiome biomarkers, and supporting new diagnostic strategies for hyperuricemia (HUA) and gout.¹

“In healthy people, about two-thirds of UA are excreted from the renal system, and the remaining third is excreted through the intestine. When renal function is reduced, the proportion of UA excreted through the intestine can increase by two-thirds.² Therefore, investigating differences in gut microbiota between patients provides a valuable approach to understanding the pathogenesis of gout and HUA.¹

Tang and colleagues collected 16S rRNA amplicon sequence data from 233 fecal samples from five population studies conducted in Japan, Korea, and China, and a total dataset of 100 healthy controls (HC), 93 from the HUA group and 40 from the gout group. The mean ages for these groups were 40.87, 54.88, and 50.27 years, respectively, with mean BMI of 24.34, 26.22, and 24.08. The group consisted of 75 (75%), 62% (67%), and 38 (93%), respectively.¹

Using the Shannon Diversity Index and the Inverse Simpson Index, researchers found that gout groups had the lowest diversity and that of the HC groups had the highest diversity.p <.05).

One-way anosim further confirmed significant separation between the three groups (HC vs. HUA: r = 0.167; p = .001; HC vs. gout: R = 0.651, p = .001; Hua vs. Gout: r = 0.452, p = .001) (Figure 2c). Further evaluation of the interactions between genera revealed that the HC group has the most complex microbial community network compared to the HUA and gout groups.¹

Machine Learning (ML) and Shapley Additive Description (SHAP) interpretability algorithms were found to be five genera. G-Curisten Neraches, G-Streptococcus, G-Prevotella, G-Coprococcus, and G-Elysiperotricha family, It was shared between the HC and HUA groups and seven genera. G-Subdoligranulum, g-agathobacter, g-collinsella, g-dorea, g-alistipes, g-lachnospira, and G-Bacteroides, It was shared between the HC group and the gout group. Among these genera, Christensenellaceae High contribution to SHAP analysis Sub-Dried Granurum Both the LEFSE and SHAP analysis showed strong correlation and significance with the HC group. Alicepe It was identified as the most contributing genus by SHAP.

Between Hua and the gout group, 8 genera, subdoligranulum, g-agathobacter, g-blautia, g-kkermansia, g-lachnospira, g-fusobacterium, g-phascolarctobacterium, and G-Bacteroides, Shared. Among these, Levs discovered Sub-Dried Granurum As HUA Group and SHAP analyses have become richer, Rachnospira As the most important contributor to model classification. Among the unique and important genus, Halomonas It turns out to be the most important genus of model classification. Rhodococcus Ranked 2nd in model classification contribution.¹

A random forest (RF)-based SHAP approach was chosen to assess the performance of the top 20 genera identified by Lefse and SHAP based on best diagnostic performance, with a predictive accuracy of 82–96%. use Tax4fun2, Tang and colleagues also found that compared to HC, the HUA group showed reduced activity in thiamine, fructose, mannose, and propanoic acid metabolism, suggesting metabolic suppression that may contribute to hyperuricemia. Compared to the HC group, purine metabolism and fructose and mannose metabolism showed the most significant differences in the gout group. This was further strengthened when comparing Hua and gout, where significant differences in purine metabolism were also observed.¹

In summary, the ML-based SHAP approach to identifying core taxa developed in this study outperforms traditional Lefse methods in terms of classification accuracy. Functional predictions of identified core taxa reveal significant enrichment of metabolic pathways associated with HUA and gout, further supporting the reliability of the proposed method. Hua and Gout's strategy,” Tang and colleagues concluded.²

reference

1. Tang, J.W., Tay, Acy, Wang, L. Prediction of interpretive predictions of hyperuricemia and gout patients via machine learning analysis of the human gut microbiota. BMC Microbiol twenty five429 (2025). doi:10.1186/s12866-025-04125-x

2. Ichinose K, Matsuguchi H, Takayama T, Nakayama A, Murakami K, Shimamura T, and others Reduced uric acid excretion outside the kidneys is a common cause of hyperuricemia. Nut commune.2012; 3(1): 764.