Clustering patterns of metabolic syndrome: A cross-sectional study in children and adolescents in Kyiv

Objective: The aim: to identify subgroups by cluster analysis according parameters: original homeostatic model of insulin resistance (HOMA-1 IR), updated computer model of insulin resistance (HOMA-2 IR), beta-cell function (%B) and insulin sensitivity (%S) for the prognosis of different variants of metabolic syndrome in children for more individualized treatment selection. Patients and methods: The observational cross-sectional study on 75 children aged from 10 to 17 with metabolic syndrome according to the International Diabetes Federation criteria was conducted at the Cardiology Department of Children's Clinical Hospital No.6 in Kyiv. HOMA-1 IR was calculated as follows: fasting insulin (mu IU/ml) x fasting glucose (mmol/L)/22.5. HOMA-2 IR with %B and %S were calculated according to the computer model in []. All biochemical analysis were carried out using Cobas 6000 analyzer and Roche Diagnostics (Switzerland). The statistical analysis was performed using STATISTICA 7.0 and Easy R. The hierarchical method Ward was used for cluster analysis according the parameters: HOMA-1 IR, HOMA-2 IR, %B and %S. Results: Four clusters were identified from the dendrogram, which could predict four variants in the course of metabolic syndrome such that children in cluster 1 would have the worst values of the studied parameters and those in cluster 4 - the best. It was found that HOMA-1 IR was much higher in cluster 1 (6.32 +/- 0.66) than in cluster 4 (2.19 +/- 0.13). HOMA-2 IR was also much higher in cluster 1 (3.80 & PLUSMN; 0.34) than in cluster 4 (1.31 +/- 0.06). By the analysis of variance using Scheffe's multiple comparison method, a statistically significant difference was obtained between the laboratory parameters among the subgroups: HOMA-1 IR (p < 0,001), glucose (p < 0.001), insulin (p < 0,001), HOMA-2 IR (p < 0.001), %B (p < 0.001), %S (p < 0.001), TG (p = 0.005) and VLDL-C (p = 0.002). Conclusions: A cluster analysis revealed that the first two subgroups of children had the worst insulin resistance and lipid profile parameters. It was found positive correlation between HOMA-1 IR, HOMA-2 IR, %B and %S with lipid metabolism parameters TG and VLDL-C and negative correlation between %B and HDL-C in children with metabolic syndrome (MetS).The risk of getting a high TG result in the blood analysis in children with MetS was significantly dependent with the HOMA-2 IR > 2.26.

Automated summary

Cluster analysis was used to identify subgroups of children with metabolic syndrome based on insulin resistance and lipid metabolism parameters. Significant differences were found among the subgroups, providing evidence for individualized treatment selection.