Predicting the efficacy of classic ketogenic diet therapy in children with refractory epilepsy using dynamic changes in the glucose-ketone index (GKI).

OBJECTIVE

To investigate the predictive value of dynamic changes in blood glucose, blood ketones and glucose-ketone index (GKI) for the efficacy of the classic ketogenic diet (CKD)in children with refractory epilepsy.

METHODS

Data from 189 children with refractory epilepsy who received classic ketogenic diet (CKD) between January 2018 and December 2024 were retrospectively analyzed. Blood glucose, blood ketones and the glucose-ketone index (GKI) which defined as the ratio of blood glucose to blood ketone were monitored at multiple time points, from the 2-day period of halving carbohydrate intake to 90 days after CKD initiation. Patients were divided into a responder group (≥50% seizure reduction) and a non-responder group based on Engel's classification after three months of treatment. Group differences in metabolic indices were assessed, and receiver operating characteristic (ROC) analysis identified optimal GKI thresholds for predicting response.

RESULTS

On the first day of CKD therapy, GKI (AUC = 0.638) was the best single predictor of efficacy among all time points, with a cut-off value of 9.63 (sensitivity and specificity both 0.63). Combining blood glucose (<4.95 mmol/L) and blood ketone (>0.75 mmol/L) levels improved predictive performance (AUC = 0.661).

CONCLUSION

A GKI > 9.63 on the first day of CKD therapy may serve as an early-warning indicator of poor treatment response. The early treatment phase-from the second day of halving carbohydrate intake to the first day of CKD-represents a critical window for metabolic transition. Monitoring dynamic changes in GKI and blood ketone may facilitate timely adjustment of intervention strategies.