Altered temporal correlations in parietal alpha and prefrontal theta oscillations in early-stage Alzheimer disease

Abstract

Encoding and retention of information in memory are associated with a sustained increase in the amplitude of neuronal oscillations for up to several seconds. We reasoned that coordination of oscillatory activity over time might be important for memory and, therefore, that the amplitude modulation of oscillations may be abnormal in Alzheimer disease (AD). To test this hypothesis, we measured magnetoencephalography (MEG) during eyes-closed rest in 19 patients diagnosed with early-stage AD and 16 age-matched control subjects and characterized the autocorrelation structure of ongoing oscillations using detrended fluctuation analysis and an analysis of the life- and waiting-time statistics of oscillation bursts. We found that Alzheimer's patients had a strongly reduced incidence of alpha-band oscillation bursts with long life- or waiting-times (< 1 s) over temporo-parietal regions and markedly weaker autocorrelations on long time scales (1-25 seconds). Interestingly, the life- and waiting-times of theta oscillations over medial prefrontal regions were greatly increased. Whereas both temporo-parietal alpha and medial prefrontal theta oscillations are associated with retrieval and retention of information, metabolic and structural deficits in early-stage AD are observed primarily in temporo-parietal areas, suggesting that the enhanced oscillations in medial prefrontal cortex reflect a compensatory mechanism. Together, our results suggest that amplitude modulation of neuronal oscillations is important for cognition and that indices of amplitude dynamics of oscillations may prove useful as neuroimaging biomarkers of early-stage AD.

Fig. 1.

The study of spatial and temporal dimensions of neuronal processing requires different correlation analyses. Coordination of anatomically distributed activity (parallel processing) may be studied by computing correlations between neuronal signals from different brain areas (Cross-correlations). Many different algorithms have been used to detect and quantify the nature of correlations in the spatial domain, e.g., coherence, phase-locking factors, or synchronization likelihood. In contrast, coordination of brain activity over time (serial processing) may be studied by computing autocorrelations in neuronal signals within a single brain area (Autocorrelations). Serial processing requires a sequence of causally related neuronal activities, which is likely to give rise to correlations over time (temporal correlations), e.g., persistent oscillatory activity as reflected in a slow amplitude modulation as it is studied here. Thus, by studying autocorrelation properties we may learn more about the mechanisms of attention and memory.

Fig. 2.

Three biomarkers for characterizing the amplitude fluctuations of neuronal oscillations. To characterize the amplitude dynamics of ongoing alpha oscillations, the MEG signals were band-pass filtered from 6–13 Hz (thin green line) and the amplitude envelope of the oscillations (thick blue line) extracted with the Hilbert transform (A, B). Non-random fluctuations are qualitatively identified as a tendency for oscillations to exhibit amplitude modulations on multiple time scales, as seen in the control subject (C), as opposed to rapidly changing amplitude levels even on short time scales, as seen in the AD patient (D) and the MEG recording without a subject in the device (E). To quantify differences in amplitude dynamics of oscillations on short to intermediate time scales (< 1 s), we introduced a threshold at the median amplitude (horizontal dashed line in A) and defined the beginning and the end of an oscillation burst as the time points of crossing this threshold. An inherent persistence in the amplitude dynamics on short time scale was reflected in probability distributions of oscillation-burst “life-times” and “waiting-times” that exhibited power-law–like decays as indicated by the least-squares fit in the double logarithmic plot (F, G, straight blue line), compared with the exponential fit (F, G, dashed line). The probability distributions, however, did not have a sufficiently wide dynamic range to make statistical conclusions about which mathematical function best described the distributions. Instead, we used the 95^th percentile of the cumulative probability distributions as an index that captures the fat tail in the distribution of “life-times” (H) and “waiting-times” (I) for the AD patient (thick red line) and control (thin blue line). The DFA exponent provides a quantitative index of the persistence of autocorrelations on longer time scales (1–25 s) and is the slope of the least-square fitted lines in (J). The stronger autocorrelations in the control subject (J, blue circles) compared with the AD patient (J, red diamonds) is reflected in a DFA exponent closer to 1 (0.81 vs. 0.58). The lack of temporal correlations in (E) is reflected in the DFA exponent having the value of ≈0.5, which is characteristic of an uncorrelated random process (J, black dots). All data were taken from a parietal channel.

Fig. 3.

Aberrant temporal structure of temporo-parietal alpha and medial prefrontal theta oscillations in AD. Grand-average topographies of biomarkers are shown for AD patients (small left column), controls (small middle column), and controls minus patients (right column). Alpha oscillations (6–13 Hz) are shown in the left panel (A, C, E, and G) and theta oscillations (4–5 Hz) on the right panel (B, D, F, and H). Individual-subject values for the patients (red diamonds) and the controls (blue circles), and mean ± SEM of the two groups were computed as the average over the channels marked with white circles in the difference topography of a given biomarker. In (F) the channels marked in (B) were used, and in (G) the average was made across the 12 channels with the largest group difference. White circles denote channels with P < 0.05 (open), and P < 0.01 (filled). Each row displays a biomarker: life-time (A, B), waiting-time (C, D), DFA exponent (E, F), and mean amplitude (G, H).**, group difference at P < 0.005; ns (nonsignificant), P > 0.05 .