Functional Neuromarkers for Psychiatry, 1st Edition
April, 2016
Applications for Diagnosis and Treatment
This cutting edge book explores, in detail, the brain pattern biomarkers for psychiatric and neurologic disorders and how they can be used in diagnosis, determining personalized neurotherapy, and monitoring treatment results. Focuses on the analysis of EEG and ERP, and covers the functional neuromarkers for ADHD, schizophrenia, and obsessive-compulsive disorder.
Here is an exclusive summary of the book provided by
Prof. Yury Kropotov to our members:
Objective measures of human brain functioning
Human brain functioning can be directly and objectively measured by neuroscience methods such as fMRI (functional Magnetic Resonance Imaging), PET (Positron Emission Tomography), EEG (electroencephalogram) and ERPs (Event-Related Potentials). Indirectly, human brain functioning is measured by behavioral indexes of speed and quality of performance such as reaction time, reaction time variability, omission and commission errors.
fMRI and PET provide information about slow metabolic responses of the brain during behavioral tasks. Because of low temporal resolution, indirect relation to neuronal activity, and low intra-subject reliability the application of fMRI and most of PET indexes in psychiatric practice is quite limited. In contrast, EEG and ERP provide high temporal resolution indexes of cortical neuronal activity, high test-retest reliability and have been in used in clinical practice since 1970s.
EEG rhythms of the healthy brain
The EEG is electric currents measured from scalp electrodes. The main feature of EEG is oscillatory nature of the currents. The oscillations are seen at different frequency bands: infra-low (0.01 -0.1), low (0.1-1), delta (1-4 Hz), theta (4-8 Hz), alpha (8-12 Hz), and beta (>13 Hz) bands. In the infra-low frequency band there are at least two types of spontaneous oscillations: 1) periodic oscillations with a peak frequency around 0.1 Hz and 2) arrhythmic fluctuations with no clear peak at EEG spectrograms. Experimental evidence suggests that the 0.1 Hz oscillations are associated with local hemodynamic oscillations of the human brain.
During relaxed wakefulness the human brain exhibits several types of rhythmic electrical activity of the alpha frequency band (8–13 Hz) in occipital, parietal and central areas (Fig. 1 top). These rhythms differ in topography, frequency, and sensitivity to tasks. Despite these differences alpha oscillations share a general function –inhibition of the irrelevant sensory pathways. If we think of the thalamus as the gate to the cortex, this inhibitory function can be considered as closing the sensory gate. The absence of alpha rhythms is found in 10% of population with prevalence in anxiety disorders.
FIGURE 1. Rhythms of cortical self-regulation: alpha (top), frontal midline theta (middle) and beta (bottom) rhythms. (a) Hypothetical neuronal network for rhythm generation. Excitatory neurons and pathways are marked by red. Inhibitory neurons and pathways are marked by blue. (b) An example of rhythm recorded from the human scalp. (c) Locations of maximums of the rhythm power. (d) Functional meaning.
Beta rhythms (Fig. 1 bottom) come in ranges: beta 1 (13-20 Hz), beta 2 (21-30 Hz), and gamma (30-60 Hz). The basal ganglia beta rhythms are reflected in the scalp recorded Rolandic beta rhythms. The frontal beta rhythm synchronizes in response to activation of the frontal lobes. The vertex beta rhythm is induced in response in unexpected situations. Existence of several beta rhythms with different frequencies, topographies and functional properties presumes no single neuronal mechanism of their generation. The scheme in Fig. 1, bottom reflects locality of beta rhythms rather than a single mechanism.
In healthy brain during wakefulness there is only one rhythm in the theta band – the frontal midline theta rhythm (Fig. 1 middle). Because this rhythm appears in short bursts (of few seconds) with long and varied inter-bust intervals in a small (10-40%) group of healthy subjects, and is enhanced by task load, it can be reliably measured by spectral analysis in highly demanding cognitive tasks. Appearance of the frontal midline theta rhythm is more likely in less neurotic and less anxious subjects. The frontal midline theta rhythm in humans is often associated with the hippocampal theta rhythms in animal research. The scalp topography of the rhythm power is frontal with a maximum at Fz and frequency from 5 to 7.5 Hz. The rhythm is associated with working memory, episodic encoding and retrieval.
Information flow in the healthy brain.
There are different sensory modalities that give us sensations of images, sounds, and body movements. The information flow in the sensory systems is modulated by attention. Acanonical ERP in a sensory modality includes P1 and N1 waves. When stimuli are repeatedly presented the brain forms the neuronal model of sensory stimulation. When an unexpected stimulus violates the model hypothetical “change” detectors are activated. The activation of these change detectors is reflected in visual and auditory mismatch negativity (MMN) ERP waves.
If the deviancy is large it produces orienting response which is reflected in the Novelty P3 wave. Readout from a personal memory is attributed to the visual N170 wave generated in the fusiform gyrus. This area is also involved in generation of N250 wave which appears in response to stimulus repetition. When the stimulus is mismatched with the template in working memory the temporaly generated P2 (or P250) wave emerges.
The operations of sensory processing are separated by methods of blind source separation. Fig. 2b represents grand average ERPs in response to GO, NOGO and Ignore stimuli in the cued GO/NOGO task schematically shown in Fig. 2.a. Two sensory-related components are shown at Fig. 2c at the bottom. The time course of these components is modulated by operations of categorization and comparison to working memory as seen at the differences on Fig, 2d. sLORETA localizes the first component in the prestriate cortex and the second component in the fusiform gyrus.
FIGURE 2. ERPs and ERP components in healthy brain. (a) Stimuli in the cued GO/NOGO task (examples). (b) Left: grand average ERPs for a group of 114 healthy subjects of 18-23years old for GO (green), NOGO (red) and Ignore (black) conditions. Right: maps of ERPs for the three conditions at maximums around 300 ms. (c) Latent components extracted from the collection of ERPs for GO (green) and NOGO (red) conditions. (d) Difference NOGO-GO for the corresponding components. Note a progressive delay of the P3 peak latency from occipital to frontal locations. (e) sLORETA images with arrows indicating hypothetical information flow.
One of the basic goals in human behavior is selection from a large repertoire of possible behavioral options those actions that are more likely to promote human well-being. The hypothetical operations of cognitive control include: action preparation and action selection, working memory, shift from one action to another, suppression of prepared action, inhibition of ongoing activity, detection of conflict, adjusting behavior in order to avoid conflicts.
Two modes of cognitive control are separated: proactive mode reflecting the sustained and anticipatory maintenance of internal goals, and reactive control reflecting transient mechanism in response to conflict detection. The ERP correlates of proactive cognitive control include readiness potential, the contingent negative variation (CNV), and the stimulus preceding negativity (SPN).
Concept of prepotent response (or prepotent model of behavior) is used as a theoretical background in cognitive control studies. The prepotent automaticity makes our life efficient because it frees limited cognitive resources from the numerous routine requirements. When conflict is detected the automatic response is shifted to the reactive cognitive control mode. This is reflected in N2 and P3 NOGO waves (Fig. 2b). However the N2/P3 dichotomy does not fit experimental data and is substituted by decomposition of ERPs into functional components on the basis of the blind source separation approach. Three components of cognitive are presented in Fig. 2c at the top. Time course of these components is differently modulated by operations of conflict detection, inhibition of ongoing activity and suppression/overriding of the prepared action.
The affective system maps sensory stimuli into rewards and punishments, expresses emotions and is responsible for feelings of those emotional reactions. One of the hypotheses suggests asymmetric involvement of prefrontal cortical regions in positive affect (dominance of the left hemisphere) and negative affect (dominance of the right hemisphere). The left-right frontal asymmetry in human subjects is assessed in the anterior alpha asymmetry index of QEEG. The key structure of the affective system is the amygdala which extracts affective memories and sends the results to the prefrontal cortex. The amygdala reacts with increased activation to fearful stimuli and via feedback connections enhances early visual ERPs components. As a result, anxiety is often associated with enhancement of P1/N1 waves. The affective and cognitive control systems are mutually interconnected so that the loss of cognitive control during stress exposure leads to a number of maladaptive behaviors, such as drug addiction, smoking, drinking alcohol and overeating. Prolonged stress is a major risk factor for depression, and exposure to traumatic stress can cause post-traumatic stress disorder.
There are different types of memory. The contralateral delayed activity in the posterior brain areas represents an ERP correlate of short-term (working) memory. In relation to content the long term memory is divided into declarative and procedural memories. The hippocampus serves as a key element in declarative memory by mapping representations of the current episode into an intermediate form. The encoding of episode is accompanied by a burst of hippocampal theta rhythm. Procedural memory is associated with learning motor and cognitive skills. The representations of actions are stored in the frontal-parietal networks and are mapped into striatum. Procedural memory, unlike episodic memory, does not need a separate system for encoding and consolidating events.
Current treatment options in psychiatry.
The modern psychiatry has in its arsenal different treatment options including pharmacotherapy, neurotherapy and neuromodulation techniques. The basic idea of the pharmacotherapy is that the core of the brain functioning is chemical so that mental illness is the result of imbalances among neurotransmitters. During the last years the psycho-pharmacology faces a certain crisis and shrinks neuroscience research projects. Neurotherapy and neuromodulation options emerge as alternative approaches.
Neurofeedback (NF) is a technique of a self-regulation in which current parameters of EEG are presented to a subject through visual, auditory or tactile modality while the subject is supposed to alter these parameters to reach a more efficient mode of brain functioning. According to the two types of electrical brain phenomena there are two main types of neurofeedback: conventional EEG biofeedback and infra-low frequency (ILF) neurofeedback. The conventional NF uses the spectral characteristics of EEG in 0.5-50 Hz frequency band whereas the ILF NF uses either amplitude itself or phase of the voltage fluctuations below 0.1 Hz. The protocols of the conventional NF can be divided into activation and relaxation protocols. ILF NF is done in discrete and continuous forms. In contrast to the “one size fits all” discrete protocol the continuous ILF protocols are different for different symptoms. The neurofeedback of at the beginning (middle of 20 century) was driven by the theory of operant conditioning but recently a bulldozer principle has been suggested with the aim to normalize a pathologically abnormal EEG pattern. NF shouldn’t be applied without relevant diagnostic procedures of QEEG and ERPs.
In transcranial Direct Current Stimulation (tDCS) a small amount of direct electric current (1-2 mA) is applied to the skin of the head by two relatively large electrodes. The electric current flows according to the Ohm’s law and depolarizes/hyperpolarizes pyramidal cells at their basal membrane depending on direction of the current. tDCS long-term after-effect is a function of its intensity and duration and occur through NMDA-dependent mechanisms similar to long-term synaptic potentiation and depression.
Transcranial Magnetic Stimulation (TMS) is based on the law of electromagnetic induction. TMS in clinical practice is applied in form of continuous trains and is named repetitive TMS (rTMS). rTMS induces changes in neuronal excitability that persist beyond the time of stimulation. rTMS at a low frequency (about 1 Hz) induces decrease of cortical excitability while higher frequency rTMS (usually between 5 and 20 Hz) increases cortical excitability.
At the beginning of 20th century psychosurgery for severe psychiatric conditions aimed in destruction of large portions of the brain. In the mid of 20th century it was replaced by stereotactic local lesions. Currently deep brain stimulation (DBS) substitutes these ablation techniques.
Functional neuromarkers in diseased brain.
During the last 20 years intensive research on functional neuromarkers in different psychiatric conditions like ADHD, schizophrenia, OCD provided a vast amount of empirical knowledge with clear indication that some of those neuromarkers are reliable and powerful tools in discriminating patients from healthy controls. Nowadays we are facing the decade of translation that focuses on application of neuromarkers for providing an early detection of brain dysfunction as well as a personalized care for patients with disease.
Attention-Deficit/Hyperactivity Disorder (ADHD) is a highly prevalent neuropsychiatric condition with onset in childhood and characterized by a persistent and age-inappropriate pattern of descriptive symptoms of inattention, hyperactivity and impulsivity. ADHD puts children at risk for other psychiatric and substance abuse disorders. Although symptoms decline with age in some cases ADHD persist into adulthood. Methylphenidate, by blocking dopamine reuptake in the striatum, has a positive therapeutic effect in approximately 65-70% of patients. Based on EEG several subgroups of ADHD are separated including those characterized by presence of Rolandic spikes, excessive theta/beta ratio (TBR), excessive frontal beta and persistent excessive alpha in eyes open condition. Based on ERPs in GO/NOGO paradigm at least two subgroups of ADHD with decrease of parietal and frontal ERP components respectively are separated. Patients with the ERP frontal deficit respond to psychostimulants. The pattern of ERP change in response to a single dose stimulant medication predicts a positive response to stimulant medication. There were numerous studies on application of neurofeedback in treatment symptoms of ADHD.
In contrast to ADHD schizophrenia (SZ) is a less common psychiatric condition characterized by a diverse set of symptoms including specific distortions in sensory, motor, executive and affective systems. Positive symptoms (delusions and hallucinations) start between ages 16 and 30, are associated with dopaminergic hyperactivity and are suppressed by antipsychotic drugs via blocking dopaminergic receptors. Negative symptoms include impairments in the affective domain such abulia, anhedonia and apathy as well as impairments in cognitive domain with no medication known to reduce them. SZ is developing during prodromal phase so that defining neuromarkers of this state is of big importance for SZ prevention. No consistent changes in QEEG have been so far reported in SZ, however ERP research shows reliable decrease of many ERP components with large effect sizes. The ERP neuromarkers reflect sensory-related deficits such decrease of auditory/visual N1 waves, MMN, and novelty P3, as well as cognitive related deficits such as reduction of the target P3, N2 and P3 NOGO waves, CNV and SPN. Some of these waves predict conversion to psychosis while the others predict response to antipsychotic medication. Attempts of applying tDCS for schizophrenia report reduction of hallucinations and improvement of cognitive functions.
Obsessions include socially unacceptable thoughts or impulses, chronic doubting, fears of contamination etc. Compulsions include excessive hand washing, repeatedly checking, placing objects symmetrically… The nature of the obsessions and compulsions varies greatly between patients with obsessive-compulsive disorder (OCD). In contrast to ADHD and schizophrenia OCD patients consistently show hyper-activation of the medial prefrontal cortex including the anterior cingulate area. The error-related negativity and conflict-related N2 ERP waves generated in the anterior cingulum are elevated in OCD and correlate of the feeling that things are “not just right”. Until recently reports on successful neurofeedback treatment in OCD were rare but individually tailored protocols could be beneficial. Inhibition of the pre-supplementary motor area by TMS or tDCS might be a promising protocol.
Implementation of functional neuromarkers in clinical practice
A single diagnostic category represents a heterogeneous condition in which similar symptoms may be caused by different sources. For example, ADHD patient may have a focus in the Rolandic fissure, maturation lag, hypo-arousal, limbic dysfunction, or cortical-striatal-thalamic-cortical dysfunction with either frontal or parietal nodes. Knowledge about the cause of the particular behavioral pattern helps not only in making the correct diagnosis but also in providing the prognosis of how to treat the cause. For example, depending on the biological source of the behavioral pattern of ADHD the psychiatrist may use dopamine reuptake or noradrenalin reuptake inhibitors, GABA agonists, neurofeedback, tDCS, TMS, or even DBS.
The working space of the 21 century psychiatrist includes the EEG/ERP system for recording spontaneous EEG and event-related potentials (Fig. 3a). A possible content of the report made by an expert depends on the goals of the psychiatrist and includes recommendations for the individual treatment. There are many EEG/ERP systems in the world but there are few QEEG/ERP databases including BRC and HBI databases.
FIGIRE 3. Implementing functional neuromarkers into psychiatric practice. (a) EEG recording in resting and task conditions. (b) Top: automatic spike detection and averaging, mapping, and constructing sLORETA image. Middle: computing EEG spectra, comparing with the normative data, extracting the corresponding independent component, and constructing sLORETA image. Bottom: computing ERPs and ERP components by spatial filtration, comparing with the normative data, and constructing sLORETA image for the deviant component. (c) Selection of appropriate treatment on the basis of the results of assessment as well as on the basis of the obtained research knowledge.
Since normal mechanisms of cortical self-regulations are reflected in brain oscillations in several frequency bands different kinds of EEG dysrhythmia are expected in brain disorders. For example, ADHD is characterized by a specific pattern of cortical dysregulation including Rolandic spikes (Fig. 3b top), excess of theta beta ratio and/or central theta (Fig. 3b middle), increase of frontal beta oscillations and frontal midline theta rhythm activity, persistent alpha rhythms with eyes open. Individual independent components for the QEEG deviations from the reference might be used for constructing neurofeedback protocols.
In substantial number of psychiatric patients the ERP abnormalities occur without any significant deviations from the reference in spectral EEG and even behavioral characteristics. The ERP analysis includes comparing raw ERPs and ERP components with the parameters obtained in healthy controls of the same age. An example at Fig. 3b bottom shows ERP in NOGO condition of an ADHD patient without any impairment in EEG spectra and behavioral characteristics. One can see statistically significant decrease of P3 NOGO wave and selective decrease of the P3 early component associated with operation of response inhibition. The topography and sLORETA image of the component clarify the source of dysfunction and possible neuromodulation techniques for activating this part of the prefrontal cortex (Fig. 3c).
Recently ERPs in the cued GO/NOGO task were successfully applied for predicting response to stimulant medication of the basis on a single-dose trial. ERPs provide a good tool for monitoring changes in brain functioning induced by neurotherapeutic and neuromodulation techniques.
