Transcript Related File

An overview of EEG signal processing algorithms for the
detection of EEG biomarkers of treatment response in
depression
Dr. Reza Rostami
Psychiatrist, Associate professor
University of Tehran
Major depressive disorder (MDD(
• A major depressive disorder occurs without a history of a manic, mixed, or
hypomanic episode.
• A major depressive episode must last at least 2 weeks, and typically a person
with a diagnosis of a major depressive episode also experiences at least Five
(or more) symptoms from a list that includes:
• Depressed mood most of the day, nearly every day
• Markedly diminished interest or pleasure in all, or almost all, activities most of
the day
• Significant weight loss when not dieting or weight gain or increase in appetite
• Insomnia or hypersomnia
• Psychomotor agitation and retardation
• Fatigue or loss of energy
• Feelings of worthlessness or excessive or inappropriate guilt
• Diminished ability to think or concentrate, indecisiveness
• Recurring thoughts of death or suicide
Epidemiology
Lifetime Prevalence Rates of Depressive Disorders
Type
Lifetime
Major depressive episode
Range
Average
5-17
12
Dysthymic disorder
Range
Average
3-6
5
Minor depressive disorder
Range
Average
10
----
Recurrent brief depressive disorder
Range
16
Burden disease
• The World Health Organization predicts depression as the second burden of
disease in 2020 (Muray & Lopez, 1996).
• Currently depression is the fourth leading cause of disability worldwide
(Muray & Lopez, 1996).
• The annual cost of depression is 118 million Euros (Sobocki et al., 2006) in
Europe and 400 million Dollars (Greenberg et al., 2015) in the United
States.
• Less than 50% of MDD patients respond to their first consumed
antidepressant and only 30% experience full recovery (Trivedi et al., 2006).
• Most of the patients experience a long period of trial and error before
finding the right antidepressant (Bauer et al., 2007)
Types of treatment in depression
•
•
•
•
•
Medication
Psychotherapy
Combination both
Brain stimulation method
Alternative medicine
The algorithm of treatment selection for Major depressive disorder
The percentage reduction of depressive symptoms in varies
treatments
Main Brain Stimulation Techniques
• ECT - Electroconvulsive Therapy
▫ FEAST
• rTMS - repeated Transcranial Magnetic Stimulation
▫ Neuronetics, Magstim, Brainsway, Neuralieve, Neostim, Neosync
• DBS - Deep Brain Stimulation
▫ RST - Responsive Stimulation Therapy
▫ Epidural Cortical Stimulation
• VNS - Vagus Nerve Stimulation
• MST - Magnetic Seizure Therapy
• tDCS - transcranial Direct Current Stimulation
• TENS - transcutaneous Electrical Nerve Stimulation
▫ Cranial Electrical Stimulation (CES) Alpha-stim
• EPI-fMRI - echoplanar fMRI
• Transcranial pulsed ultrasound
What is Vagus Nerve Stimulation (VNS) Therapy?
• Mild pulses applied to the
left vagus nerve in the neck
send signals to the brain
• Automatic intermittent
stimulation
• Simple in-office dose
adjustment
• Assured treatment
adherence
• FDA Approved for Epilepsy,
Class I evidence
From Higgins and George, Brain Stimulation Therapies for Clinicians, 2008, APPI Press.
FEAST – Clinical Status
• 17 Depressed Patients treated to date
▫ Columbia University (NY), MUSC
• Remarkable time to recovery after FEAST (17 min average)
• 7 Antidepressant remitters, 5 of last 8
• Continued development underway
What is TMS?
• Electrical current flowing through a coil induces a magnetic field
• Scalp and skull are transparent to magnetic fields
• Alternating magnetic fields induce electrical current in
underlying brain tissue
Overview of TMS
1) Electrical energy in
insulated coil on the
scalp induces
2) Pulsed magnetic
field of about 1.5 Tesla
in strength
3) Passes unimpeded
through the
cranium for 2-3 cm
4) In turn induces a
focal electrical current
in the brain
5) Get desired local
and distal effects on
the target neural
circuitry
6) Delivered as single
pulses or repeated
trains (rTMS)
How TMS Works
George MS. Sci Am. 2003;289:66-73.
Coil Type
Figure 8
Double Cone
Round
TMS as a novel treatment for depression
Non-invasive, office-based procedure
No anesthetic required, awake, alert, no recovery room or
nurses, immediate resumption of activities
Flexible targeting of treatment possible, device can be
moved over surface of the brain and so different circuits and
disorders can be targeted
Possible advantages of rTMS
No cognitive side effects
High tolerability generally, better than medication, more
precise delivery
High patient acceptability, with lack of stigma, although labor
intensive
Which treatment is better?
• Currently a combination of antidepressant medications and psychotherapy
is the most effective treatment for depression, while their individual
effectiveness equals any Alternative medicine treatment or control group.
• Less than 50% of MDD patients respond to their first consumed
antidepressant and only 30% experience full recovery.
• Most of the patients experience a long period of trial and error before
finding the right antidepressant.
Biomarkers
• A new research scope that has attracted a lot of attention recently is the use
of biomarkers to predict the response to treatment.
• Different genetic, metabolic, neuroendocrinology, neuroimaging and
neurophysiology parameters are proposed as potential biomarkers of
response to antidepressant treatment but none has been approved.
• One neurophysiologic biomarker of response to treatment which has had
promising results is EEG
EEG biomarkers
•
•
•
•
•
•
•
EEG alpha band activity
Alpha asymmetry
Alpha Peak Frequency
EEG theta band activity
Antidepressant treatment response index (ATR)
Theta cordance
Event-related potentials (ERPs)
EEG alpha band activity
• Pre-treatment changes in the alpha band, that is increased alpha power
has been shown to differentiate responders from non-responders.
• A possible mechanism involving the right temporoparietal and
subcortical regions was to account for the increased alpha power seen in
depressed patients who respond to SSRI treatment.
• Depression may be related to dysfunction of temporoparietal mechanisms,
which may mediate emotional arousal
• This biological mechanism has been proposed to play a role in both
increased alpha power and alpha asymmetry found in SSRI responders.
• Asymmetry in the alpha band characterized by left lateralization has been
demonstrated in treatment response with TCAs and the SSRI, fluoxetine.
Alpha asymmetry
• Inverse relation between alpha power and cortical activity, increased
alpha is reflective of reduced cortical activity whereas, decreased alpha
reflects increased activity.
• While greater left than right hemispheric activation at occipital sites was
associated with treatment response, non-responders tended to show the
opposite pattern of hemispheric activation (greater right than left) in
frontal and posterior regions.
Alpha asymmetry in depressed patient
• Findings of left lateralizaton are in conflict with reports of reduced left
frontal activity (increased alpha activity) in depressed patients.
• Explanation 1: Left frontal hypoactivation can be interpreted as a deficit
in approach mechanisms, while right frontal hypoactivation can be
interpreted as a deficit in withdrawal mechanisms.
• Explanation 2: Decreased left prefrontal activation in depression may
disinhibit left temporoparietal regions, resulting in enhanced left
hemisphere advantage in fluoxetine responders.
• Alpha asymmetry in SSRI responders may be reflective of low arousal
associated with low serotonergic activity
Alpha Peak Frequency
• Alpha peak and rCBF: A direct relationship between regional CBF
and iAPF has been established, where increased iAPF was associated with
increased rCBF, most specifically in the bilateral inferior frontal gyrus (BA
45) and right insular cortex (BA 13).
• Alpha peak and cerebral perfusion: A direct relationship
between iAPF and cerebral perfusion on one hand and their
relationship to the modulation of attention and arousal on the other
hand, which are also impaired in both ADHD and depression.
• Alpha peak and medication : Given that benzodiazepines have
been shown to decrease the iAPF (specifically Carbamazepine, and
oxcarbazepine), these studies suggest interplay between the GABAergic system and cerebral blood-flow.
Alpha peak and medication
•
The power spectral plots of non-responders (left) and responders (right) to antidepressant medication at
posterior sites. Note the decreased alpha power and lower iAPF (8 Hz) in non-responders. The red line
indicates the iAPF pre-treatment and the blue line the iAPF post-treatment. Only responders to medication
exhibited an increase in iAPF of approx. 0.5 Hz.
Alpha peak and rTMS
• Studies have demonstrated that non-responders to rTMS were
characterized by a slow iAPF, suggesting regular rTMS is unlikely to be a
likely candidate for this sub-group.
• One study demonstrated in schizophrenia that rTMS at the iAPF
demonstrated better effects on negative symptoms than LF or HF
rTMS.
Pre treatment
Pre treatment (BDI)
Pre treatment (CANTAB)
Post treatment (After the 10th session)
Post treatment (After the 10th session)
Post treatment (After the 10th session)
Post treatment (After the 20th session)
The summary of studies in the field of response to treatment biomarkers(Alpha
frequency) in antidepressant
Pharmacolo
gy
Amitriptyline
study
Ulrich et
al ,1984
Pirlindol
Clomipramin
e
Number
of
Subjects
Status of response
Changs in EEG Resp . vs Non- Resp
9
5 Resp. vs 4 Non-Resp
11
4 Resp. vs 7 Non-Resp
Power decrease in the alpha range (7.5-13 cps)/ power increase of the
slower frequencies (2-5 cps)/ higher peak frequency in the alpha
range/increase in the peak frequency in the alpha range of about 0.5
cps/ tendency towards a left lateralisation of the occipital alpha power
23
12 Resp. vs 11 Non-Resp
Ulrich et
al ,1988
22
Maprotiline
7 Resp. vs 15 Non-Resp
Paroxetine
Knott et al
, 2000
70
51 Resp. vs 19 Non-Resp
Fluoxetine
Bruder et
al , 2001
53
34 Resp. vs 19 Non-Resp
18
11 Resp. vs 7 Non-Resp
Fluoxetine
Early changes in alpha band EEG after the first TCA dose were
associated with treatment response at three weeks
Bruder et
al, 2008
Baseline increases in alpha power and decreases in theta power were
detected in responders compared to non-responders to a six-week
treatment with paroxetine.
EEG alpha asymmetry between brain hemispheres was shown to
differentiate responder and non-responder/
Responders had great baseline EEG alpha power at occipital lobe/
responders showed greater alpha over right/ alpha power and
asymmetry at occipital lobe showed good predictive ability
EEG theta band activity
• Studies investigating pre-treatment and early changes in the theta band
have reported conflicting results.
• Elevated, pre-treatment theta current density, localized by LORETA to the
rACC has been associated with response to nortriptyline, citalopram,
reboxetine, fluoxetine, or venlafaxine, in depressed patients.
• The rACC is a main hub within the default network (DN) of the brain,
which is involved in self-focused processing.
• Elevated resting state activity in these brain regions is associated with
focusing on reflective thought or task-independent introspection such as
rumination, remembering and planning.
• Pizzagalli (2011) proposes that elevated resting rACC activity may lead
to treatment response through adaptive self-referential functions such as
mindfulness and non-evaluative self-focus.
Default Mode Network
Blue :
regions that negatively correlate
with the default network
Red:
regions that positively correlate
with the default network.
rACC as a biomarker response to treatment
• Pizzagalli et al (2001): Increased pre-treatment resting rACC theta activity
in responders relative to nonresponders (red).
• Pizzagalli (2011) further proposes that increased rACC activity associated
with treatment response to antidepressants may play a key role in
reestablishing the functional connections between the DN and TPN.
• several domains of research including neuropsychology, neurophysiology
and neuroimaging have demonstrated significant frontocingulate
dysfunction in depression.
Antidepressant treatment response index (ATR)
• Antidepressant Treatment Response Index (ATR) as a composite, threeparameter QEEG index that combines prefrontal EEG theta and alpha
power from baseline and week 1.
• ATR is a nonlinear combination of three features: relative combined theta
and alpha power (3–12 Hz), plus alpha power in two different alpha bands
(8.5–12 Hz and 9–11.5 Hz). ATR also includes the change in alpha power
from baseline (8.5–12 Hz) and week 1 (9–11.5 Hz).
The ATR components
Relative theta
power
Antidepressant
treatment
response index
(ATR)
Relative
alpha power
alpha power
in two
different
alpha bands
The summary of ATR studies in the field of response to treatment
biomarkers in antidepressant
Pharmacology
SSRI
study
Iosifescu et al,
2009
Escitalopram/bupr
opion
Leutchter et al,
2009
Escitalopram/bupr
opion
Leutchter et al,
2009
Number of
Subjects
82
Status of response
Changes in ATR in Resp . vs Non- Resp
45 Resp. vs 37 Non-Resp
ATR predicted antidepressant response with 70% accuracy
(sensitivity = 82%, specificity = 54%, AUROC = 72%, p =
0.001).
73
38 Resp. vs 35 Non-Resp
375
220 Resp. vs 155 NonResp
ATR was useful for predicting differential response to
either Escitalopram or bupropion monotherapy.
Subjects with high ATR values (above a threshold)
were more than 2.4 times as likely to respond to
escitalopram as those with low ATR values (68% vs.
28%; p = .001)/ Subjects with ATR values below the
threshold who were switched to bupropion treatment
were 1.9 times as likely to respond to bupropion alone
than those who remained on escitalopram treatment
(53% vs. 28%; p = .034).
ATR had 74% accuracy (AUROC = 0.77) in predicting both
response and remission, whereas clinical parameters or
genetic polymorphisms were associated with neither response
nor remission
Theta cordance
• Cordance is a measure that combines EEG absolute and relative power
according to a specific formula.
• Cordance is computed by a normalization and integration of absolute and
relative power values from all electrode sites for a given EEG recording;
cordance values are calculated in three steps.
• First, EEG power values are computed using a re-attributional electrode
montage in which power values from pairs of electrodes that share a
common electrode are averaged together to yield the re-attributed power
These absolute power values are used to derive relative power (percentage
of power in each frequency band) for each electrode.
• Second, these absolute and relative power values for each individual EEG
recording are normalized across electrode sites, using a z-transformation
statistic for each electrode site s in each frequency band f (yielding Anorm
(s,f) and Rnorm (s,f) respectively). This normalization process places
absolute and relative power values into a common unit (standard deviation
or z-score units) which allows them to be combined. These z-scores are
based on the average power values in each band for all electrodes within a
given EEG recording (i.e., these are not z-scores referenced to a normative
population).
• Third, the cordance values are formed by summing the z-scores for
normalized absolute and relative power (Z(s,f)=Anorm(s,f)+Rnorm(s,f), for
each electrode site and in each frequency band). Cordance values have
been shown to have higher correlations with regional cerebral blood
flow than absolute or relative power alone
Cordance Computation
CORDANCE(s,f)=(ANORM(s,f)_0.5)+ (RNORM(s,f) _0.5)
Electrode site
Frequency band
Normalized absolute power
Normalized relative power
• If the site is discordant associated with white-matter lesions,
cordance is negative,
• If the site is concordant, cordance is positive.
The summary of Cordance studies in the field of response to
treatment biomarkers in antidepressant
Pharmacology
study
Venlafaxine
Number of
Subjects
Status of response
Changes in theta cordance Resp . vs Non- Resp
12
6 Resp. vs 6 NonResp
13
7 Resp. vs 6 NonResp
A decrease in prefrontal theta cordance at one week after start of
medication was a significant predictor of antidepressant response/
Change in prefrontal theta cordance at one week significantly
distinguished medication responders from all other groups
25
13 Resp. vs 12 NonResp
Cook et al, 2002
Fluoxetine
Venlafaxine/
Fluoxetine
Leutchter et al,
2002
SSRI
Cook et al, 2005
Prefrontal theta cordance “decrease/no decrease” at one week as a
predictor of clinical response (observed at week 8) led to an accuracy
of 72% (sensitivity = 69%, specificity = 75%).
Of six responders, five showed an early decrease in cordance; only
two of the six non-responders showed an early cordance decrease.
The predictor yielded an accurate classification for 75% of the
subjects
12
6 Resp. vs 6 NonResp
Prefrontal theta cordance decreases after one week predicted
response with an overall accuracy of 88% (sensitivity = 100%,
specificity = 83%)
The decrease of prefrontal cordance after week 1 in responders was
significant ( p =0.03)
Antideppressant
Bares et al, 2007
17
5 Resp. vs 12 NonResp
venlafaxine
Bares et al, 2008
25
12 Resp. vs 13 NonResp
Event-related potentials (ERPs)
• (ERPs) are a measure of change in voltage, which represent brain activity
elicited in response to sensor stimulation (i.e., visual or auditory).
• Recorded ERPs comprise distinctive peaks and troughs reflective of
positive and negative fluctuations in voltage and are referred to as
‘components’. Different components have been identified and named
based on the direction of the waveform deflection (P for Positive and N for
Negative) and on the specific time course of the waveform at which, it
occurs post-stimulus.
• Two ERP components have been the focus of investigation in depression
including the P300 (or P3) and the Loudness Dependent Auditory
Evoked Potential (LDAEP).
The P300 component
• The P300 component is measured at 300 msec after presenting an auditory
stimulus.
• This ERP index is believed to reflect the cognitive processes of attention
and auditory processing.
• In depressed patients, a delay in the latency of the P300 component has
been reported.
• Higher amplitude of the P300 wave at occipital sites was associated with
treatment response with fluoxetine and a TCA.
• Patients who did not remit had longer P300 latency at baseline compared to
those who did.
• The precise neurobiological basis of the P300 in relation to treatment
response is unknown
Grand Grand average ERP waveforms for 20 depressed patients and 20
healthy controls to nontarget, target and novel stimuli at midline sites
Loudness Dependent Auditory Evoked Potential (LDAEP).
• The LDAEP is a measure of the ERP component N1/P2, taken 100-200
msec after presentation of an auditory stimulus.
• The amplitude of the waveform changes with increasing loudness of the
stimulus.
• The LDAEP is thought to relate to the magnitude of 5-HT
neurotransmission in the auditory cortex, particularly in the primary
auditory cortex.
• The majority of studies investigating LDAEP in antidepressant treatment
response have divided their samples into a top 50% range (representative
of higher slopes; ‘strong’ LDAEP) and bottom 50% range (lower slopes;
‘weak’ LDAEP) based on a median split midpoint.
The summary of ERP studies in the field of response to treatment
biomarkers in antidepressant
Pharmacology
Fluoxetine
SSRI
study
Bruder et
al, 1995
Kalayam
&
Alexopoul
os, 1999
Number of
Subjects
86
49
Paige et
al, 1994
17
Fluoxetine
Lee et al,
2005
100
SSRI
Galliant et
al, 2000
29
Reboxetine/citalperam
Juckel et
al, 2007
35
Reboxetine/citalperam
Mulert et
al, 2007
20
Reboxetine/citalperam
Linka et al
, 2009
26
Fluoxetine,buproprion, desipramine
Treatment response was associated with higher amplitude of the P300 wave only at the
occipital electrodes.
Patients who did not remit at six weeks had, at baseline, longer P300 latency than
remitted depressed patients and controls. When combining the P300 result with clinical
predictors (psychomotor retardation and perseveration), 23 of the 24 remitted depressed
patients were correctly identified (sensitivity = 95.8%, specificity = 65.9%).
11 responders to 6 nonresponders after 4 to 8 weeks of treatment with fluoxetine,
buproprion, or desipramine; responders had, at baseline, larger slopes of the P2
amplitude as a function of stimulus intensity.
At week 4, the HDRS score had declined by 44.3% for the group with strong LDAEP
and by 34.4% for the group with weak LDAEP
Patients treated with paroxetine, better four-week outcomes (measured by the HAM-D)
were present in the half of the sample with “stronger” LDAEP
Citalopram responders had a “strong” LDAEP at baseline, whereas responders to
reboxetine had a “weak” LDAEP at baseline. Nonresponders to citalopram or reboxetine
showed the inverse LDAEP characteristics, respectively
Citalopram we found higher LDAEP-values in responders versus non-responders
(p<0.05) and a significant correlation between pre-treatment-LDAEP and improvement in
the Hamilton score after treatment (r=0.71, p<0.05)
High LDAEP being associated with Citalopram response and low LDAEP being
associated with reboxetine response.