Multiple pathways to metabolic syndrome identified by hierarchical clustering of Japanese longitudinal health checkup data

Abstract To identify pre-onset pathways of metabolic syndrome (MetS), we retrospectively analyzed Japanese health checkup data from 296 male MetS cases and 296 non-MetS controls matched by initial status (aged 49–64 years). Hierarchical clustering was applied to four MetS components—abdominal obesity (OB), hypertension (HT), dyslipidemia (DLP), and hyperglycemia (HG)—using longitudinal status from the 3 years (years −3 to −1) before onset (year 0). Six clusters were identified, showing shifts from early (year −3 to −1) to late (year −1 to 0) patterns. Cluster 1 (n = 97, 32.8%): early OB and HT; followed by DLP. Cluster 2 (n = 36, 12.2%): early OB and DLP; followed by HT. Cluster 3 (n = 48, 16.2%): early OB; followed by HT and HG. Cluster 4 (n = 44, 14.9%): early HT and DLP; followed by OB. Cluster 5 (n = 46, 15.5%): early HT; followed by OB, DLP, and HG. Cluster 6 (n = 25, 8.4%): early HT, DLP, and HG; followed by OB. Controls showed no clear changes. The waist circumference and triglyceride-glucose related indices were significantly higher at year 0 in the MetS group. These findings provide insight into the heterogeneous pathways leading to MetS. This clustering approach offers a flexible framework for capturing population-specific patterns in MetS development and may allow more targeted early intervention strategies.