Prediabetes Subtypes with Divergent Iron Metabolism Profiles: A Cluster Analysis from Two Cohorts in the United States and China.

We aim to investigate the relationship between iron metabolism and prediabetes using data from the National Health and Nutrition Examination Survey (NHANES) and the China Health and Nutrition Survey (CHNS) databases, and to identify prediabetes subgroups with distinct distinct iron-related biomarker profiles through cluster analysis, while examining both universal patterns and population-specific differences across U.S. and Chinese populations. Data from NHANES 2003-2010 and 2015-2020 were used for primary analysis; these cycles were selected based on the availability of complete iron metabolism biomarker data. Participants were classified into prediabetes group and normoglycemic group. Linear regression analysis was conducted to evaluate the association between clinical indicators and iron metabolism markers. K-means cluster analysis was performed in the NHANES prediabetes subsample to identify phenotypically distinct subgroups. The identified cluster structure was subsequently validated by independently applying the same algorithm to the CHNS 2009 prediabetes subsample. Between-group comparisons were performed using the t-test or Mann-Whitney U test for continuous variables and the χ² test for categorical variables, as appropriate. A two-sided P-value of < 0.05 was considered statistically significant. Among 7110 participants from NHANES, 1528 were classified into the prediabetes group. Compared with the normoglycemic group, the prediabetes group exhibited significantly higher ferritin [82.0 (IQR 40.1-158.0) vs. 47.00 (IQR 26.1-89.3) µg/L, P < 0.001] and lower serum iron (14.9 ± 6.3 vs. 16.0 ± 7.2 µmol/L, P = 0.031), reflecting a pattern of sub-clinical iron redistribution rather than clinically defined iron overload. Three distinct prediabetes phenotypes were identified: Cluster 1 (obesity-related insulin resistant-younger phenotype, N = 424), cluster 2 (mild insulin secretion impairment-younger phenotype, N = 456), and cluster 3 (high iron store-older phenotype, N = 648). Cluster 3 demonstrated older phenotype with higher ferritin concentrations within the normal reference range and showed the highest ferritin and serum iron concentrations among the identified clusters, while Cluster 1 has intermediate ferritin levels and Cluster 2 has the lowest levels (P < 0.001). A comparable three-cluster structure was replicated in the CHNS 2009 cohort, with Cluster 3 representing a higher proportion (48.89% vs. 42.41%) and retaining the highest ferritin levels [92.6 (IQR 52.0-155.3) µg/L, P < 0.001]. Variation in iron-related markers is present in the prediabetes stage, with distinct patterns across subgroups. We identified three distinct prediabetes phenotypes characterized by different ferritin profiles within the normal reference range, providing novel insights for individualized clinical practice and risk stratification.