γ-band abnormalities as markers of autism spectrum disorders

Abstract

Autism is a behaviorally diagnosed neurodevelopmental disorder with no current biomarkers with high specificity and sensitivity. γ-band abnormalities have been reported in many studies of autism spectrum disorders. γ-band activity is associated with perceptual and cognitive functions that are compromised in autism. Some γ-band deficits have also been seen in unaffected first-degree relatives, suggesting heritability of these findings. This review covers the published literature on γ abnormalities in autism, the proposed mechanisms underlying the deficits and the potential for translation into new treatments. Although the utility of γ-band metrics as diagnostic biomarkers is currently limited, such changes in autism are also useful as endophenotypes, for evaluating potential neural mechanisms, and for use as surrogate markers of treatment response to interventions.

Figure 1

Phase-locking and the transient auditory gamma-band response. A) Single trial amplitude plot of responses to 1000 Hz pure tone stimulus in passive task. Warmer colors indicate positive amplitude and cooler colors indicate negative amplitudes. Note variability of response latency from trial to trial. B) Time frequency spectrograms produced by Morlet wavelet decomposition, from left to right: Induced, or total power, consisting of both phase-locked and non-phase-locked components; Evoked power, from the spectral analysis of the time-domain average; Phase-locking factor (PLF). Note the strong phase-locked response around 40 Hz. C) Mean phase from trials at 45 ms post stimulus at peak of gamma-band response (blue) compared with mean phase from trials at -45 ms post-stimulus period (red). Note that the mean phase of the pre-stimulus region is near zero, while there is a clear non-zero phase for the post-stimulus region. Conceptually, this is what the PLF plots for each time-frequency bin, normalized from 0 (non-phase-locked) to 1 (phase-locked).

Figure 2

Gamma-band responses to two different types of stimuli. The top two panels are spectrograms from a visual phonological processing task, from a left occipito-temporal channel in an MEG sensor array: A) Induced power and B) PLF. Note the high gamma-band responses beginning approximately 700 ms post stimulus. In the PLF, these are not seen, indicating that the gamma-band power in plot A is primarily non-phase-locked, or induced power. In contrast, there are low frequency components that are highly phase-locked, or evoked. Bottom two panels: C) Induced power of auditory steady-state response to 40 Hz amplitude-modulated stimulus lasting 500 ms, from channel T5 of an EEG dataset. Note the gamma-band response at approximately 40 Hz beginning around 100 ms. D) PLF of response in C. The PLF for this type of gamma is high, in contrast to plot B, indicating that the power of this response is highly phase-locked to the stimulus, or evoked. All time-frequency spectrograms were produced by Morlet wavelet decomposition.