EEG Biofeedback to Improve Memory in Adults With Dementia

Objectives and Significance

1. What is (are) the primary goal(s) of the study The purpose of this study is to evaluate
the efficacy of biofeedback training to increase regional cerebral blood flow (CBF) and
peak alpha frequency (PAF) on the cognitive and behavioral symptoms of frontotemporal
dementia (FTD). We will also measure the effects of EEG driven photostimulation during
PAF neurofeedback on learning of higher PAF states and related clinical outcomes.

2. What is (are) the secondary goal(s) of the study A diversity of clinical,
neuropsychological, and physiological variables will be collected in this study. For
instance, a full 19-channel quantitative EEG (QEEG) assessment will be performed at
intake. The relationship between EEG power and coherence and the symptoms of this
disorder could potentially provide an important contribution to the literature on the
usefulness of QEEG for differential diagnosis.

An additional goal of the study is to contribute to the understanding of how neurofeedback
works. For instance, we will take measurements to determine if learning to control the PAF
involves awareness of some internal subjective state related to the PAF. Also, the effect of
the reward signal on a brainwave called the P300 may show the importance of having the reward
tone sounding the majority of the time during neurofeedback sessions. Finally, we will
measure whether blood flow changes during EEG biofeedback and whether the EEG is affected by
blood flow biofeedback.

We propose several measurements that will contribute to an understanding of the mechanism of
action of neurofeedback.

Frontotemporal Dementia and Cerebral Hyperperfusion Single photon emission computed
tomography (SPECT) studies have shown cerebral blood flow to be significantly reduced in the
frontal and temporal regions in FTD patients (Miller et al., 1997; Read et al., 1995). The
anatomical distribution of reduced CBF corresponds to the pattern of neuropsychological
deficits (McMurtray et al., 2006).

Not surprisingly, magnetic resonance imaging (MRI) and computed tomography (CT) in FTD
patients shows atrophy in the frontal and temporal regions (Mendez et al., 1996; Neary and
Snowden, 1996). However, Spilt et al. (2005) hypothesized that neurodegeneration and dementia
are largely secondary to pathologies of cerebral blood flow. When compared to elderly
controls with optimal cognitive function, patients with DSM-IV dementia did not differ
significantly from elderly controls with respect to the number of cerebral infarctions.
Demented patients showed significantly more white matter lesions (p=.028) and cerebrospinal
fluid (CSF; p=.016), but a reduction in cerebral blood flow had the largest effect size
(p<.001). An attempt to build a logistic regression model showed that no significant residual
variance could be explained after cerebral blood flow was included in the model.

Another argument for the central role of blood flow in dementia is that Alzheimer's patients
with brain damage (regions of MRI signal hyperintensity) have increased oxygen extraction per
mL/min. That is, blood supply rather than demand seems to be the problem. Oxygen extraction
would be expected to be unaltered if reduced blood flow were secondary to tissue damage
(Spilt et al., 2005; Yamaji et al, 1997).

Positron emission tomography (PET) imaging in FTD patients reveals reduced glucose metabolism
in the frontal and anterior temporal lobes, but also in the cingulate gyrus, insula, uncus,
and subcortical structures (Jeong et al., 2005; Garraux et al., 1999; Ishi et al., 1998).
Grimmer et al. (2003) performed a longitudinal study on ten patients diagnosed with FTD. At
the initial assessment, FTD patients had reduced metabolic activity compared to controls in
frontal cortical areas, the caudate nuclei, and the thalami. On a 1-2 year follow-up,
significant progression of the original deficits was observed in the orbitofrontal cortex and
the subcortical structures.

Given the substantial evidence linking dementia and FTD in particular to reduced cerebral
blood flow, we hypothesize that training FTD patients to increase cerebral blood flow will
alleviate FTD symptoms and slow the progress of the disease.

Recent studies have suggested that individuals can learn to increase CBF through biofeedback.
Yoo et al. (2006) showed that participants given feedback of fMRI activity of the auditory
cortex while listening to music were able to significantly increase the mean blood
oxygenation as well as the number of significant voxels. Another study (deCharms et al.,
2005) trained participants to change fMRI activity in the rostral anterior cingulate gyrus
(RACG), a region implicated in pain perception. Control conditions included sham feedback or
feedback from a different brain region. When a noxious thermal stimulus was applied,
participants had decreased pain sensation when trained to decrease RACG activity and
increased pain sensation when trained to increase RACG activity. In another phase of the
study, eight chronic pain patients reported decreased pain after down-training fMRI in the
same region.

fMRI costs more than $1000 per session, which places this form of therapy beyond the reach of
most patients. However, it is possible to provide CBF neurofeedback for the outermost 1.5 cm
of cerebral cortex with a relatively inexpensive device that uses the refractive properties
of oxygenated hemogoblin to red and infrared light (Toomim et al., 2004). A light source is
attached to the scalp (typically on the forehead) with a headband, 3 cm away from an infrared
sensor, which detects the relative absorption by oxygenated blood. This procedure is known as
hemoencephalography or HEG. Toomim et al. (2004) showed that ten sessions improved
impulsivity scores on the Test Of Variables of Attention (TOVA) in 28 patients of diverse
psychopathology. Carmen (2004) provided frontal HEG to 100 migraine patients, and found that
90% of those who completed at least six sessions reported significant improvement in migraine
symptoms. In a single case study, Mize (2004) reported that a child with ADHD showed
significant improvement on the IVA, which improvement persisting into the 18-month follow-up.

Frontotemporal Dementia and Peak Alpha Frequency

The PAF in health adults has an average of 10-11 Hz. Higher PAF is associated with higher
memory performance (Klimesch, 1997), reading ability (Suldo, 2000), vocabulary, and response
control (Angelakis et al., 2004a). After a series of cognitive tasks, PAF was reduced in
traumatic brain injury patients compared to normal controls, but only weakly or
nonsignificantly reduced compared to controls during the task or the baseline conditions.
Thus, Angelakis et al. (2004b) argued that PAF is both a trait and a state marker of
cognitive preparedness. Passant et al (2005), Chan et al. (2004) and Yenner et al. (1996) all
observed a reduction in PAF in FTD patients.

We hypothesize that EEG biofeedback rewarding higher PAF will result in an improvement in
symptoms in FTD patients. In EEG biofeedback or neurofeedback, an individual's real-time EEG
is presented continuously as a visual or auditory signal, and desired variations are
rewarded. A recent double-blind controlled study (Angelakis et al., 2007) showed that
neurofeedback rewarding increased PAF improved cognitive processing speed and executive
function in normal elderly adults.

The efficacy of neurofeedback as a therapy has been demonstrated for attention deficit
hyperactivity disorder (ADHD), epilepsy, anxiety, and addictive disorders. Other disorders
such as schizophrenia, depression, learning disabilities (LD), and traumatic brain injury are
under investigation as candidates for neurofeedback therapy (Monastra, 2003).

Frontotemporal Dementia and EEG-Driven AVS

Like EEG neurofeedback, EEG-dependent auditory and visual stimulation (AVS), has showed
promise for improving cognitive function by modifying the PAF. A substantial body of research
has demonstrated that rhythmic AVS can induce EEG rhythms corresponding to the frequency of
stimulation (Frederick et al., 2004). Russell (1997) reported on a study in which the
continuously varying PAF of LD and ADHD children was used as a signal to produce AVS
alternately at 5% above and 5% below the PAF for 30 second intervals, for 20 minute sessions.
While the theoretical aim of this study was to improve the flexibility of the PAF (not to
change the mean frequency), it showed that treating the PAF can effectively treat cognitive
dysfunction. These children showed significant gains in cognitive and behavioral measures
that persisted to the 16-month follow-up.

We hypothesize concurrent EEG-Driven photostimulation during PAF enhancement neurofeedback
(where rewards are presented as auditory tones, with eyes closed) will increase the rate of
learning of PAF enhancement, and have increased therapeutic efficacy compared to PAF
neurofeedback alone. In addition to helping induce higher PAF, rhythmic photostimulation has
the benefit of increasing CBF by inducing repetitive waves of activation throughout the
brain. It may therefore also enhance the effects of the HEG training described above in part
I.

Understanding the mechanism of action of neurofeedback could potentially lead to more refined
methods of treatment with improved efficacy. We propose three measurements that would
contribute to an improved understanding of how neurofeedback works.

1. A common practice in EEG neurofeedback is for the reward signal to be adjusted so that
the client spends the majority of the time with the reward tone on. This is believed to
improve motivation and compliance, but there might be another reason. When the reward
tone goes off, it amounts to a rare event requiring a behavioral change. This situation
is similar to the design of many studies which have shown that the absence of an
expected stimulus evokes a positive deflection in the EEG called the P300 wave. The P300
has been associated with the orientation reflex, and with a phase-resetting of the
background EEG rhythms. If a greater P300 during reward tone offset is observed than
during reward tone onset, this would suggest that the brain responds to the loss of
reward with an electrical signal that disrupts and resets ongoing EEG activity. This
would provide independent physiological support to the conventional clinical wisdom of
providing frequent rewards.

2. Some theorists argue that the efficacy of neurofeedback depends on the normalization of
the abnormal physiology (e.g., PAF or CBF) correlating with the disorder (Duff, 2004).
Others (Othmer, Othmer and Kaiser, 1999) argue that the efficacy of neurofeedback
results from the brain's adaptive responses to the altered brain states induced by
training. This model is similar to that explaining the efficacy of antidepressant drugs,
and could explain the importance of multiple sessions over weeks and months of training.
Thus, we will measure whether changes in clinical and neuropsychological test variables
relate with the changes of PAF and CBF (average and variance per session). The
relationships (or lack thereof) between physiological and psychological variables from
this study will contribute to the field's understanding of how neurofeedback works.

3. A criticism raised against HEG training in favor of EEG biofeedback is that blood flow
changes are typically secondary to metabolic demands created by neuronal activity. Thus,
it is unclear whether HEG feedback is training a nonspecific EEG activation or if some
other mechanism is involved. We propose to measure the EEG adjacent to the HEG training
site to determine if HEG changes are correlated with EEG activation (enhanced 14-21 Hz
and reduced 4-7 Hz amplitude, or increased PAF). It is of similar interest to see if the
HEG blood flow intensity measure increases during PAF training.

Inclusion Criteria:

FTD Symptoms reported by self or caregiver Significantly abnormal scores on Delis-Kaplan
Executive Function System and Behavior Rating Inventory of Executive Function- Adult
Version

Exclusion Criteria:

- 45 years of age

- no one with Axis I dx of Bipolar, Psychosis or active Substance Abuse

- no severe Axis II disorders

- symptoms manifesting within the last 5 years and person is still somewhat independent
regarding their ADLs.