Abstracts

Rationale: Infantile spasms (IS) is a potentially catastrophic form of epilepsy that is characterized by clinical spasms and a chaotic electroencephalographic (EEG) pattern called hypsarrhythmia. The disease frequently progresses into intractable forms of pediatric epilepsy and leaves patients with a bleak developmental outcome, but the prognosis can be improved with early diagnosis and successful treatment. Although classic hypsarrhythmia is easily distinguished from normal activity, the numerous variants of hypsarrhythmia can make diagnosis a challenge. For these reasons, we sought an EEG biomarker to quantify hypsarrhythmia and the patient's response to treatment. Methods: We queried long-range temporal correlations (LRTCs) as a potential biomarker of treatment success in 18 infantile spasms patients and compared results to 24 control subjects. LRTCs are believed to indicate control of neural synchrony, thus we hypothesized that healthy patients would have stronger LRTCs than diseased patients. We measured LRTCs using Detrended Fluctuation Analysis (DFA), an algorithm that measures how correlated a signal is with itself over varying time scales. The output of the algorithm is the DFA exponent, a numerical value between 0.5 and 1.0 that represents the relative strength of the LRTCs in the EEG signal, with higher values indicating stronger correlations. Results: We found that DFA reliably ascribes lower DFA exponents to patients with hypsarrhythmia (0.63 +/- 0.10) than those without (0.82 +/- 0.14) (Wilcoxon rank sum, p < 0.0001). Through simulations with pink noise, we also demonstrate that our measurement of LRTCs is not dependent on the amplitude of the patient's EEG, which is known to be elevated when hypsarrhythmia is present. Lastly, we show that patients that responded to treatment, marked by both a resolution of hypsarrhythmia and a cessation of spasms, had a greater increase in strength of LRTCs after treatment compared to patients who had a resolution of hypsarrhythmia but persistent spasms (right-tailed sign test, p < 0.02). We also find that responding patients had exhibited similar DFA exponent values after treatment (0.85 +/- 0.15) to those of age-matched control subjects (0.91 +/- 0.12). This promotes LRTCs as a potential biomarker of successful treatment of infantile spasms. Conclusions: We demonstrate that LRTCs may have value in measuring treatment outcome in infantile spasms. Assessing LRTCs at several points in time may allow clinicians to promptly initiate a different, more effective drug if the patient does not show early signs of treatment success. This expedited treatment protocol would spare the patient unnecessary side effects, lower the cost of treatment, and maximize the patient's prognosis. Funding: This project was funded in part by an ICTS UCI-CHOC Collaborative Grant.