The characteristics of postprandial glycemic response patterns to white rice and glucose in healthy adults: Identifying subgroups by clustering analysis.

Objectives: Large interpersonal variability in postprandial glycemic response (PGR) to white rice has been reported, and differences in the PGR patterns during the oral glucose tolerance test (OGTT) have been documented. However, there is scant study on the PGR patterns of white rice. We examined the typical PGR patterns of white rice and glucose and the association between them.

Materials And Methods: We analyzed the data of 3-h PGRs to white rice (WR) and glucose (G) of 114 normoglycemic female subjects of similar age, weight status, and same ethnic group. Diverse glycemic parameters, based on the discrete blood glucose values, were calculated over 120 and 180 min. K-means clustering based on glycemic parameters calculated over 180 min was applied to identify subgroups and representative PGR patterns. Principal factor analysis based on the parameters used in the cluster analysis was applied to characterize PGR patterns. Simple correspondence analysis was performed on the clustering categories of WR and G.

Results: More distinct differences were found in glycemic parameters calculated over 180 min compared with that calculated over 120 min, especially in the negative area under the curve and Nadir. We identified four distinct PGR patterns to WR (WR1, WR2, WR3, and WR4) and G (G1, G2, G3, and G4), respectively. There were significant differences among the patterns regard to postprandial hyperglycemia, hypoglycemic, and glycemic variability. The WR1 clusters had significantly lower glycemic index (59 ± 19), while no difference was found among the glycemic index based on the other three clusters. Each given G subgroup presented multiple patterns of PGR to WR, especially in the largest G subgroup (G1), and in subgroup with the greatest glycemic variability (G3).

Conclusion: Multiple subgroups could be classified based on the PGR patterns to white rice and glucose even in seemingly homogeneous subjects. Extending the monitoring time to 180 min was conducive to more effective discrimination of PGR patterns. It may not be reliable to extrapolate the patterns of PGR to rice from that to glucose, suggesting a need of combining OGTT and meal tolerance test for individualized glycemic management.