Developmental atlas of phase-amplitude coupling between physiologic high-frequency oscillations and slow waves

Eishi Asano, Kazuki Sakakura, Naoto Kuroda, Masaki Sonoda, Takumi Mitsuhashi, Ethan Firestone, Aimée F. Luat, Neena I. Marupudi, Sandeep Sood

Abstract

The cortex generates high-frequency oscillations (HFO) nested in slow waves during sleep, and these signals are especially elevated in the seizure onset zone. Thus, HFO occurrence rate and Modulation Index (MI), which quantifies the strength of coupling between HFO amplitude and slow-wave phase, represent promising epilepsy biomarkers. However, their diagnostic utility may be suboptimal because the endogenous developmental distributions are unknown. To improve age-appropriate localization of the epileptogenic zone, we hence constructed normative brain atlases demonstrating the developmental changes in MI and HFO rate. Our study investigated extraoperative, intracranial EEG data from 114 patients with focal epilepsy (ages 1.0 to 41.5 years) who achieved International League Against Epilepsy class I outcomes following resective surgery. We analyzed 20-minute slow-wave sleep epochs at 8,251 nonepileptic electrode sites. Each electrode was transposed onto a standard brain template, and we then calculated its MI and HFO occurrence rate using four different detector toolboxes. Linear and nonlinear regression models determined the developmental slope of MI and HFO rate at each cortical mesh point. Mixed model analysis established the significance of MI and HFO rate developmental changes in each region of interest, while accounting for the independent effects of patient and epilepsy profiles. Finally, we created a dynamic tractography movie visualizing white matter pathways connecting cortical regions showing developmental co-growth in MI. We found that the occipital lobe exhibited enhanced MI compared to other lobes in both children and adults. Increased age, square root of age, and log base 10 of age were independently associated with elevated MI exclusively in the occipital lobe. The cortical regions showing developmental co-growth in MI were connected via the vertical occipital fasciculi and posterior callosal fibers. In contrast, we did not observe any significant association between age measures and HFO rate in the occipital lobe, but rather noted an inverse relationship between age and HFO rate in the temporal, frontal, and parietal lobes. Our study suggests that phase-amplitude coupling between physiologic HFO and delta waves, as rated by MI, is strengthened during development, in the occipital lobe particularly during toddlerhood and preschool. Given that occipital delta-nested HFO are believed to support visual memory consolidation, our observations imply that process may be significantly strengthened during early childhood. The data is publicly available to provide investigators with a crucial reference for MI and HFO-based presurgical evaluation of the epileptogenic zone.

Introduction

When surgeons plan epilepsy surgery, they use intracranial EEG (iEEG) to identify the "seizure onset zone." Two primary biomarkers for these zones are High-Frequency Oscillations (HFOs) and Phase-Amplitude Coupling (PAC)—specifically, how the amplitude of high-frequency bursts is "nested" within the phase of slower brain waves.

The challenge is that these signals aren't always pathological; they occur naturally in healthy brain tissue as well. Until now, clinicians didn't have a clear "map" of what normal PAC or HFO rates look like at different ages. Without this normative data, a surgeon might mistake a healthy, developing brain region (like a 3-year-old’s occipital lobe) for a seizure site, potentially leading to the unnecessary removal of functional brain tissue. This study establishes a "normative atlas" of brain activity to address this clinical gap.

Methods

The study is a retrospective analysis of iEEG data from 114 patients (ages 1 to 41.5 years) who achieved seizure freedom after surgery. To ensure they were only looking at "healthy" activity, the researchers analyzed 8,251 electrode sites located outside the seizure zones and MRI lesions.

Patient demographics and clinical characteristics of the study cohort.

Key technical aspects include:

  • Signal Processing: The team analyzed 20-minute epochs of slow-wave sleep.
  • Modulation Index (MI): They used the EEGLAB winPACT toolbox to quantify the strength of PAC.
  • Robust HFO Detection: To avoid tool-specific bias, they employed four different automated detector toolboxes (STE, SLL, HIL, and MNI).
  • Statistical Modeling: They used linear and nonlinear regression (incorporating $\sqrt{age}$ and $\log_{10} age$) alongside mixed-model analysis to account for individual patient variables.
  • Dynamic Tractography: By combining iEEG data with diffusion MRI templates from over 1,000 healthy subjects, they visualized the white matter "highways" (tracts) that support the development of these signals.

Datasets

  • Primary iEEG Dataset: 114 patients from the Children’s Hospital of Michigan and Harper University Hospital.
  • Structural Templates: Electrodes were mapped to the standardized "FSaverage" brain surface.
  • DWI Template: The HCP-842 population-averaged atlas (derived from 1,065 healthy participants) was used for tractography.
  • Public Availability: The researchers have made their iEEG data and analysis codes publicly available via OpenNeuro and GitHub to encourage further clinical validation.

Key Insights and Results

  • The Occipital Powerhouse: The occipital lobe (the brain's visual center) naturally exhibits much higher PAC than any other lobe, regardless of age.
  • The Toddler Surge: PAC strength in the occipital lobe isn't static; it increases rapidly during toddlerhood and preschool years (a significant positive slope of +0.047 per $\sqrt{year}$, $p < 0.001$). This likely reflects the brain's maturation process for visual memory consolidation.
  • HFO Diminution: Outside the occipital lobe, HFO rates showed a significant decrease with age (e.g., a slope of -0.41/$\sqrt{year}$ in the frontal lobe). This suggests that "frequency of activity" and "coordination of activity" follow different developmental trajectories.
  • Structural-Functional Coupling: The regions showing the most significant PAC growth are physically linked by specific white matter tracts, such as the vertical occipital fasciculi. This suggests that the strengthening of these signals is a direct result of large-scale neural network maturation.

Limitations and Future Work

The authors acknowledge several constraints, including clinical sampling bias (iEEG electrodes are only placed where clinically necessary) and the cross-sectional design of the study. Furthermore, because it is unethical to implant electrodes in healthy people, "nonepileptic sites" in epilepsy patients remain the best—though imperfect—proxy for healthy brain tissue.

Future Direction: The next step is a prospective study to see if using this normative atlas to "filter out" healthy developmental signals actually improves surgical outcomes and cognitive preservation in children.

Summary updated

Abstract

Abstract The cortex generates high-frequency oscillations (HFO) nested in slow waves during sleep, and these signals are especially elevated in the seizure onset zone. Thus, HFO occurrence rate and Modulation Index (MI), which quantifies the strength of coupling between HFO amplitude and slow-wave phase, represent promising epilepsy biomarkers. However, their diagnostic utility may be suboptimal because the endogenous developmental distributions are unknown. To improve age-appropriate localization of the epileptogenic zone, we hence constructed normative brain atlases demonstrating the developmental changes in MI and HFO rate. Our study investigated extraoperative, intracranial EEG data from 114 patients with focal epilepsy (ages 1.0 to 41.5 years) who achieved International League Against Epilepsy class I outcomes following resective surgery. We analyzed 20-minute slow-wave sleep epochs at 8,251 nonepileptic electrode sites (those outside the seizure onset zone, interictal spike zone, or MRI-visible lesions). Each electrode was transposed onto a standard brain template, and we then calculated its MI and HFO occurrence rate using four different detector toolboxes. Linear and nonlinear regression models determined the developmental slope of MI and HFO rate at each cortical mesh point. Mixed model analysis established the significance of MI and HFO rate developmental changes in each region of interest, while accounting for the independent effects of patient and epilepsy profiles. Finally, we created a dynamic tractography movie visualizing white matter pathways connecting cortical regions showing developmental co-growth in MI. We found that the occipital lobe exhibited enhanced MI compared to other lobes in both children and adults. Increased age, square root of age, and log base 10 of age were independently associated with elevated MI exclusively in the occipital lobe. The cortical regions showing developmental co-growth in MI were connected via the vertical occipital fasciculi and posterior callosal fibers. In contrast, we did not observe any significant association between age measures and HFO rate in the occipital lobe, but rather noted an inverse relationship between age and HFO rate in the temporal, frontal, and parietal lobes. Our study suggests that phase-amplitude coupling between physiologic HFO and delta waves, as rated by MI, is strengthened during development, in the occipital lobe particularly during toddlerhood and preschool. Given that occipital delta-nested HFO are believed to support visual memory consolidation, our observations imply that process may be significantly strengthened during early childhood. The data is publicly available to provide investigators with a crucial reference for MI and HFO-based presurgical evaluation of the epileptogenic zone.
BibTeX
@inproceedings{asano2023developmental,
  author = {Eishi Asano and Kazuki Sakakura and Naoto Kuroda and Masaki Sonoda and Takumi Mitsuhashi and Ethan Firestone and Aimée F. Luat and Neena I. Marupudi and Sandeep Sood},
  title = {{Developmental atlas of phase-amplitude coupling between physiologic high-frequency oscillations and slow waves}},
  booktitle = {Nature Communications},
  year = {2023},
  doi = {10.1038/s41467-023-42091-y},
}
Figure