Sage Journals: Discover world-class research

Abstract

Purpose of the Review:

Diagnostic and treatment accuracy of depression can lead to a better and possibly earlier response and remission in patients. The literature, though scanty, seems to suggest that quantitative electroencephalography (QEEG) can predict the outcome of antidepressant effects.

Collection and Analysis of Data:

Articles published between January 1990 and July 2019, including those dealing with QEEG recordings before and after the initiation of antidepressant medication, were included. The pooled effect size and subgroup analysis of waveforms were calculated to predict response to antidepressants. In all, 572 results were retrieved from the searches, of which 20 studies were included. Pooled data using a random-effects model (REM) calculated an effect size of 0.80 (95% CI [0.64–0.97]). Heterogeneity of the sample was low with Tau² = 0.02; df = 18 (P = .30); I² = 12%. Moreover, subgroup analysis showed that theta band frequencies were better at predicting response than alpha band frequencies (the standard mean difference [SMD] for theta was 0.91 compared to 0.68 for alpha waves).

Conclusions:

QEEG is a valuable predictor of the antidepressant response. Among the EEG frequencies, the theta band showed the most significant change with treatment.

Keywords

QEEG depression treatment response antidepressants

Key Messages:

Question: What is the evidence for using resting-state QEEG in predicting antidepressant treatment response (ATR)?

Findings: Response to antidepressants correlates with decreased prefrontal cordance, changes in the ATR index, and frontal spectral powers. Frontal alpha and theta waves also predict SSRI treatment response, with significant effects observed after two weeks. Theta waves show better predictive ability for ATR in subgroup analysis.

Meaning: Resting-state QEEG shows promise in predicting antidepressant treatment response.

The remission rate of first-line antidepressants is approximately 30% in depression,^1,2 and it can take as much as six to eight weeks of an adequate dose of an antidepressant before treatment failure is considered.³ Studies addressing an early prediction of treatment response have included predictions based on depression rating scales,⁴ and/or neuroimaging scans,⁵ but the results are equivocal.

Quantitative electroencephalography (QEEG) might be time-consuming considering the pre-placement of electrodes, but as per a recent meta-analysis by Watts et al. in 2022, EEG represents a cost-effective and scalable potential measure to predict treatment response with promising predictive models.⁶ QEEG is also non-invasive and easy to administer, though it seems less cost-effective in the Indian scenario or other low to middle-income countries. Various factors derived from QEEG recordings are mentioned in the literature, such as power at each electrode, Cordance (which is also shown to be a reliable indicator of low perfusion, low metabolism, and association with undercutting lesions in the brain)^;7–10 increased theta and delta power, before starting medication, is related to poor treatment response;¹¹ and the antidepressant treatment response (ATR) index, which is calculated from the EEG at the baseline and after one week of antidepressant therapy.^9,12 Detailed definitions and details for calculations of these parameters are given.^9,13

A recent meta-analysis¹⁴ used multiple QEEG markers (spectral power density, Cordance, Event-Related Potentials [ERP]), Loudness Dependence of Auditory Evoked Potentials [LDAEP], and ATR) to predict the treatment response from different treatment modalities (medications, somatic therapy, psychotherapy and combinations of these) in depressed patients. The results indicated insignificant overall predictive power. The present review aims to quantitatively summarize the strength of the studies dealing with the resting-state QEEG and prediction of the ATR.

Methods

Search Strategy

The research studies were searched following a fourfold process. Databases searched were PubMed Central, PsycINFO, Embase, ClinicalTrials.gov, Cochrane Central Register of Controlled Trials, WHO International Clinical Trials Registry Platform (ICTRP), Cochrane Database of Systematic Reviews, Database of Abstracts of Reviews of Effects (DARE) and Open Gray database (research reports, doctoral dissertations and conference papers), using the keywords ((“Major Depression” OR “Anaclitic Depression” OR “Dysthymic Disorder” OR “Endogenous Depression” OR “Late Life Depression” OR “Postpartum Depression” OR “Reactive Depression” OR “Recurrent Depression” OR “Treatment Resistant Depression” OR depressi*) AND (“Quantitative Electroencephalogram” OR QEEG)), for articles published between January 1990 and June 2019. A hand search of selected journals for relevant articles from 2010 onwards was then performed (SJ). Finally, authors were contacted via email for study protocols, conference proceedings, unpublished data, and unavailable full-text articles. The preliminary screening of the articles, based on title and abstract, was performed independently by researchers AS and SJS, using the software RAYYAN.¹⁵ Any dispute(s) was resolved consensually by the authors (SS, SJS, AS, and SJ).

Eligibility Criteria

Studies included were (a) written in English; (b) had a study population age range from 20 to 80 years; (c) studies designed as randomized controlled trials—double-blind, open-label, and cross-over designs; (d) and studies with comparator group(s) as either an active comparator or placebo. Only those studies were further included that (e) used ICD or DSM criteria for diagnosis, (f) standardized interviews (SCID, CIDI) as well as standard rating scales (Beck Depression Inventory (BDI),¹⁶ Hamilton Depression Rating Scale (Ham-D),¹⁷ Montgomery-Asberg Depression Rating Scale (MADRS),¹⁸ Quick Inventory of Depressive Symptomatology Self Report (QIDS-SR)¹⁹ for recording symptoms severity of (g) patients with depression, bipolar depression, treatment-resistant depression and comorbid medical and psychiatric conditions, and (h) received treatment from the following classes of antidepressants (SSRI, SNRI, TCA, NaSSa, NARI, SARI), with an optimum treatment dose for at least two weeks. Their (a) treatment outcomes were recorded by researchers as either a reduction in depression scores or categorization as responders/non-responders by clinicians (>50% reduction in scores on the standardized rating scales). Studies were excluded in the case of (a) Factorial RCT study design with a combination of treatment modalities and (b) psychological/somatic therapy as comparator groups.

Quality of the Studies

The methodological quality of the selected studies was assessed independently by authors SJS and AS using Cochrane Collaboration assessment tools—RoB-2 for randomized studies^20,21 for non-randomized studies. The studies were then classed as either having a high risk of bias (one or more criteria not applied/met), a moderate risk of bias (one or more criteria unclear), or a low risk of bias (all criteria applied/met).

Data Extraction and Statistical Analysis

The relevant data regarding sample size, depression subtype, depression assessment tool, antidepressant drug along with the dosage, duration of treatment, and QEEG markers from all studies were extracted by the authors, SJS and AS (Table 1). OpenMeta[Analyst]²² software was used to calculate the effect size. The publication bias was assessed using funnel plot characteristics. Heterogeneity between studies was calculated using Cochran’s Q and Thompson—Higgins I² statistic. A P value of <.10 for the Chi² statistic or an I² > 50% was considered for significant heterogeneity. Due to significant heterogeneity in the included studies, a random-effects model (REM) was used. The standardized mean difference (SMD) with 95% CI was calculated from the change in spectral power among responders and non-responders. The SMD was calculated using various statistical derivations in the studies that did not report the mean spectral powers (Supplemental Table S1).

Table 1.

Characteristics of Included Studies.

Open Label Studies	Sample Size	Age (in Years) Mean ± SD or Mdn, IQR	Depression Subtype	Depression Scale and Severity	Depression Scores Mean ± SD								Antidepressant Drug Used and Dose	Duration of Treatment	QEEG Analysis Using
					Baseline		1 Week		4 Weeks		>6 Weeks
					Responders	Non Responders	Responders	Non Responders	Responders	Non Responders	Responders	Non Responders
Knott et al., 2000	51	NA	MDD	HAM-D > 18	22.9 ± 3.2	NA	NA	NA	NA	NA	8.9 ± 13.9	NA	Paroxetine	>4 weeks	Absolute and relative power
Bares et al., 2007	17	Mdn = 45.75, IQR = 33.5–57	MDD	MADRS >25	28	32.5	24	25.5	11	21.5	NA	NA	NA	2–4 weeks	Cordance
Bares et al., 2008	25	Mdn = 44.5, IQR = 29–54	MDD	MADRS > 20	27.5	25	22.5	26	10.5	24	NA	NA	Venlafaxine 150–234 mg	2–4 weeks	Cordance
Spronk et al., 2011	31	42.8 ± 14.2	MDD	HAM-D	NA	NA	NA	NA	NA	NA	NA	NA	NA	NA	Spectral power
Knott et al., 1996	29	39.9 ± 11.8	MDD	HAM-D > 18	24.5 ± 5	23.8 ± 2.3	NA	NA	6.4 ± 3.6	19.3 ± 3.8	NA	NA	Imipramine 1 mg per kg	2–4 weeks	Relative power
Caudill et al., 2014	25	42.9 ± 11.1	MDD	HAM-D > 16	22.8 ± 3.07	NA	NA	NA	NA	NA	NA	NA	Reboxetine	>4 weeks	ATR
Arns et al., 2016	655	38.55	MDD	HAM-D > 16	NA	NA	NA	NA	NA	NA	NA	NA	Escitalopram, sertraline, venlafaxine	>4 weeks	Frontal alpha asymmetry
Bares et al., 2014	87	46.15 ± 10.7	MDD	MADRS > 20	NA	NA	NA	NA	NA	NA	NA	NA	NA	<4 weeks	Cordance
Tenke et al., 2011	41	35.65 ± 10.3	MDD	BDI, HAM-D	14.8 ± 4.1	17.4 ± 5.4	NA	NA	NA	NA	4 ± 2.6	14.8 ± 5.7	Escitalopram 7.5–40 mg; Fluoxetine 60 mg; Sertraline 50 mg; Bupropion 150–450 mg; Duloxetine 30–120 mg; Venlafaxine 125–375 mg	>4 weeks	Alpha wave amplitude
Cook et al., 2009	37	42.7 ± 12.3	MDD	HAM-D > 16	NA	NA	NA	NA	NA	NA	NA	NA	Fluoxetine 20 mg; Venlafaxine 150 mg	2–4 weeks	Frontal theta cordance
Bruder et al., 2008	18	35.85 ± 9.55	MDD	HAM-D	19 ± 2.1	19.7 ± 6.6	NA	NA	NA	NA	5.4 ± 2.5	15.4 ± 8.6	Fluoxetine	>4 weeks	Alpha asymmetry
Schmidt et al., 2017	65	36.28 ± 12.13	MDD	HAM-D	21.10 ± 6.24	23.03 ± 5.92	NA	NA	11.62 ± 7.93	19.26 ± 5.49	6.34 ± 3.32	17.59 ± 4.56	Escitalopram, mirtazapine, sertraline, duloxetine, agomelatine	>4 weeks	EEG vigilance in alpha & theta waveforms
Bruder et al., 2001	53	39.6 ± 13.5	MDD, dysthymia	BDI	NA	NA	NA	NA	NA	NA	NA	NA	Fluoxetine 10–60 mg	>4 weeks	Alpha power
Bares et al., 2012	20	Mdn = 42, IQR = 32–59	Bipolar depression	MADRS > 20	24	26.5	19.5	24	6	20	NA	NA	Bupropion, carbamazepine, escitalopram, lithium, olanzapine, quetiapine, sertraline, valproate, venlafaxine	2–4 weeks	Frontal theta cordance
Bares et al., 2010	18	43	TRD	MADRS > 20	28	29	24	31	12	24	NA	NA	Bupropion 150 mg	2–4 weeks	Cordance
Double-Blind RCT
Leuchter et al., 2009 (Esc)	73	42.7 ± 12.7	MDD	QIDS-SR > 12	20.7 ± 4.6	20.4 ± 4.2	12.6 ± 5.4	16.3 ± 6.1	NA	NA	5.8 ± 3.8	15.4 ± 4.2	Escitalopram 10 mg	>4 weeks	ATR
Leuchter et al., 2009 (Bup)	73	41.6 ± 13.4	MDD	QIDS-SR > 12	21.6 ± 3.1	21.7 ± 4.6	14.9 ± 4.0	17.7 ± 5.5	NA	NA	5.9 ± 3.2	16.4 ± 4.9	Bupropion 300 mg	>4 weeks	ATR
Leuchter et al., 2004	51	41.45 ± 12	MDD	HAM-D > 16	21.5 ± 3.0	23.3 ± 4.3	NA	NA	NA	NA	6.0 ± 3.1	18.4 ± 4.6	Fluoxetine 20 mg; Venlafaxine 150 mg	>4 weeks	Cordance
Cook et al., 1999	13	40.29 ± 11.51	MDD	HAM-D > 17	NA	NA	NA	NA	NA	NA	NA	NA	Fluoxetine 20 mg	2–4 weeks	Cordance
Korb et al., 2009	37	41.6 ± 12.4	MDD	HAM-D > 16	21.0 ± 2.7	23.0 ± 4.1	NA	NA	NA	NA	5.4 ± 3.4	17.9 ± 4.5	Fluoxetine 20 mg; Venlafaxine 150 mg	>4 weeks	Theta density

RCT: Randomized Controlled Trial, MDD: Major Depressive Disorder, TRD: Treatment Resistant Depression, HAM-D Hamilton Depression Rating Scale, BDI: Beck Depression Inventory, MADRS: Montgomery-Asberg Depression Rating Scale, QIDS-SR: Quick Inventory of Depressive Symptomatology Self Report, ATR: Antidepressant Treatment Response, SD: Standard Deviation, Mdn: Median score, IQR: Interquartile range, NA: not mentioned in the studies.

Results

Search Results

The search strategy identified 572 articles (363 from Ovid, 55 from Embase, 61 from PubMed, 64 from PsychInfo, and another 29 studies through hand search). After removing duplicates, the studies were screened based on titles and abstracts. Forty-three studies were reviewed as full text, and 19 were included in the analysis. The reasons for exclusion are detailed in Figure 1. The selected studies were then reported using PRISMA²³ guidelines.

Figure 1.

PRISMA Flow Chart of the Studies Included in the Meta-analysis.

Study Characteristics

The total number of patients included in the 19 studies^{8–11,13,24–36} was 859 (Table 1). The mean age of this sample was 40.62 ± 3.04, except for Bares 2010³⁷ (treatment-resistant depression) and 2012³⁸ (bipolar depression), all the other studies included patients with major depression. ROBINS-1 and ROB-2 were used to assess the risk of bias for observational (n = 12) and RCT (n = 7) studies, respectively (Suppl Table 2). One observational study¹⁴ was assessed at moderate risk of bias. Two RCTs^27,32 had some concerns of bias regarding the intended intervention and measurement of outcome, respectively, as per ROB-2. The details are included in Supplemental Table S2. Since none of the studies reported a high risk of bias, all of them were included in the analyses.

Efficacy of QEEG in the Prediction of Antidepressant Response

Table 2 shows the pooled SMD of all the studies included in the analysis using a REM. The overall weighted effect size was calculated to be d = 0.80 (95% CI 0.64–0.97) with significantly low heterogeneity, Tau² = 0.02; χ² = 20.54, df = 18 (P = .30); I² = 12%. These results indicate a significant change in the QEEG waveforms with antidepressant therapy.

Table 2.

Meta-analysis of QEEG Predicting Treatment Response in Depression.

Study or Sub-group	Std. Mean Difference	SE	Responders Total	Non-responders Total	Weight	Std. Mean Difference (IV)	Std. Mean Difference (Random Model) EFFECT SIZE	Std. Mean Difference (95% CI)	Y Values	Left Label × Position	Right Labels × Position	Right Labels Text	Threshold 1 × Axis	Threshold 2 × Axis
Spronk et al., 2011	0.449	1.022	25	1	0.7%	0.45	–1.55	2.45	19	–6	5	–1.55 ± 2.45	–1	1
Cook et al., 2009	0.519	0.3653	11	26	4.8%	0.52	–0.2	1.23	18	–6	5	–0.2 ± 1.23	–1	1
Leuchter et al., 2009 bup.	0.27	0.239	30	43	9.8%	0.27	–0.2	0.74	17	–6	5	–0.2 ± 0.74	–1	1
Leuchter et al., 2004	0.4121	0.2846	23	28	7.4%	0.41	–0.15	0.97	16	–6	5	–0.15 ± 0.97	–1	1
Bares et al., 2012	0.83	0.4794	8	12	2.9%	0.83	–0.11	1.77	15	–6	5	–0.11 ± 1.77	–1	1
Bruder et al., 2008	1.006	0.5194	11	7	2.5%	1.01	–0.01	2.02	14	–6	5	–0.01 ± 2.02	–1	1
Bares et al., 2007	1.19	0.581	5	12	2.0%	1.19	0.05	2.33	13	–6	5	0.05 ± 2.33	–1	1
Korb et al., 2009	0.727	0.3466	22	15	5.3%	0.73	0.05	1.41	12	–6	5	0.05 ± 1.41	–1	1
Caudil et al., 2014	0.95	0.453	17	8	3.2%	0.95	0.06	1.84	11	–6	5	0.06 ± 1.84	–1	1
Tenke et al., 2011	0.747	0.3466	28	13	5.3%	0.75	0.07	1.43	10	–6	5	0.07 ± 1.43	–1	1
Bares et al., 2010	1.16	0.5307	11	7	2.4%	1.16	0.12	2.2	9	–6	5	0.12 ± 2.2	–1	1
Bruder et al., 2001	1.072	0.463	21	7	3.1%	1.07	0.16	1.98	8	–6	5	0.16 ± 1.98	–1	1
Arns et al., 2016	0.55	0.187	57	2	14.0%	0.55	0.18	0.92	7	–6	5	0.18 ± 0.92	–1	1
Bares et al., 2008	1.027	0.4305	12	13	3.6%	1.03	0.18	1.87	6	–6	5	0.18 ± 1.87	–1	1
Leuchter et al., 2009 esc.	1.013	0.3479	11	48	5.2%	1.01	0.33	1.69	5	–6	5	0.33 ± 1.69	–1	1
Bares et al., 2014	0.85	0.2254	48	39	10.7%	0.85	0.41	1.29	4	–6	5	0.41 ± 1.29	–1	1
Schmidt et al., 2017	0.93	0.2633	29	36	8.4%	0.93	0.41	1.45	3	–6	5	0.41 ± 1.45	–1	1
Knott et al., 2000	1.5	0.3391	16	35	5.5%	1.5	0.84	2.16	2	–6	5	0.84 ± 2.16	–1	1
Knott et al., 1996	1.751	0.4479	13	16	3.3%	1.75	0.87	2.63	1	–6	5	0.87 ± 2.63	–1	1

Pooled standard mean difference using a random-effects model.

Subgroup Analysis of QEEG

Table 3 shows subgroup analysis for different EEG waveforms. Both alpha (0.68, 95% CI [0.39–0.97], Tau² = 0; χ² = 1.63, df = 3 (P = .65); I² = 0) and theta (0.91, 95% CI [0.66–1.16], Tau² = 0.03; χ² = 11.86, df = 10 (P = .29); I² = 16%) waveforms show evidence of significant change in spectral power with antidepressant medications, with higher effect size recorded for theta waves.

Table 3.

Subgroup Analysis of the Different EEG Waveforms Included in the Meta-analysis.

Theta Power			Total	Total		IV	Random Model Effect Size	95% CI	Y Values	Left Labels × Position	Right Labels × Position	Right Labels Text	Threshold 1 × Axis Position	Threshold 2 × Axis Position
Knott et al., 1996	1.751	0.4479	13	16	3.3%	1.75	0.87	2.63	11	–5	4	0.87 ± 2.63	–0.5	0.5
Knott et al., 2000	1.5	0.3391	16	35	5.5%	1.5	0.84	2.16	10	–5	4	0.84 ± 2.16	–0.5	0.5
Bares et al., 2014	0.85	0.2254	48	39	10.7%	0.85	0.41	1.29	9	–5	4	0.41 ± 1.29	–0.5	0.5
Bares et al., 2008	1.027	0.4305	12	13	3.6%	1.03	0.18	1.87	8	–5	4	0.18 ± 1.87	–0.5	0.5
Bares et al., 2010	1.16	0.5307	11	7	2.4%	1.16	0.12	2.2	7	–5	4	0.12 ± 2.2	–0.5	0.5
Bares et al., 2007	1.19	0.581	5	12	2.0%	1.19	0.05	2.33	6	–5	4	0.05 ± 2.33	–0.5	0.5
Korb et al., 2009	0.727	0.3466	22	15	5.3%	0.73	0.05	1.41	5	–5	4	0.05 ± 1.41	–0.5	0.5
Bares et al., 2012	0.83	0.4794	8	12	2.9%	0.83	–0.11	1.77	4	–5	4	–0.11 ± 1.77	–0.5	0.5
Leuchter et al., 2004	0.4121	0.2846	23	28	7.4%	0.41	–0.15	0.97	3	–5	4	–0.15 ± 0.97	–0.5	0.5
Cook et al., 2009	0.519	0.3653	11	26	4.8%	0.52	–0.2	1.23	2	–5	4	–0.2 ± 1.23	–0.5	0.5
Spronk et al., 2011	0.449	1.022	25	1	0.7%	0.45	–1.55	2.45	1	–5	4	–1.55 ± 2.45	–0.5	0.5
Subtotal (95% CI)			194	204	48.50%	0.91	0.66	0.16
Heterogeneity: Tau² = 0.03; χ² = 11.86, df = 10 (P = .29); I² = 16%.
Test of overall effect: Z = 7.19 (P <.00001)
1.1.2 Alpha power
Arns et al., 2016	0.55	0.187	57	2	14.0%	0.55	0.18	0.92	4	–2.5	2.25	0.18 ± 0.92	–0.25	0.25
Bruder et al., 2001	1.072	0.463	21	7	3.1%	1.07	0.16	1.98	3	–2.5	2.25	0.16 ± 1.98	–0.25	0.25
Tenke et al., 2011	0.747	0.3466	28	13	5.3%	0.75	0.07	1.43	2	–2.5	2.25	0.07 ± 1.43	–0.25	0.25
Bruder et al., 2008	1.006	0.5194	11	7	2.5%	1.01	–0.01	2.02	1	–2.5	2.25	–0.01 ± 2.02	–0.25	0.25
Subtotal (95% CI)			117	89	24.90%	0.68	0.39	0.97				0.39 ± 0.97
Heterogeneity: Tau² = 0.00; χ² = 1.63, df = 3 (P = .65); I² = 0
Test of overall effect: Z = 4.56 (P <.00001)
1.1.3 Alpha and Theta Power
Schmidt et al., 2017	0.93	0.2633	29	36	8.4%	0.93	0.41	1.45	4	–3.5	2.5	0.41 ± 1.45	–0.5	0.5
Leuchter et al., 2009 esc.	1.013	0.3479	11	48	5.2%	1.01	0.33	1.69	3	–3.5	2.5	0.33 ± 1.69	–0.5	0.5
Caudil et al., 2014	0.95	0.453	17	8	3.2%	0.95	0.06	1.84	2	–3.5	2.5	0.06 ± 1.84	–0.5	0.5
Leuchter et al., 2009 bup.	0.27	0.239	30	43	9.8%	0.27	–0.2	0.74	1	–3.5	2.5	–0.2 ± 0.74	–0.5	0.5
Subtotal (95% CI)			87	135	26.60%	0.73	0.34	1.13
Heterogeneity: Tau² = 0.07; χ² = 5.11, df = 3 (P = .16); I² = 41%
Test of overall effect: Z = 3.62 (P <.00001)
Total (95% CI)			398	428	100.00%	0.8	0.64	0.97
Heterogeneity: Tau² = 0.02; χ² = 20.54, df = 18 (P = .30); I² = 12%
Test of overall effect: Z = 9.48 (P <.00001)
Test for subgroup differences: χ² = 1.58, df = 2 (P = .45); I² = 0

Table 3 summarizes the different EEG waveforms into three major subtypes used in the studies. The theta band frequencies had the highest change with antidepressant treatment. Heterogeneity among the subgroups was not significant. Details of the individual studies are represented in Table 1.

Publication Bias

The funnel plot analysis for publication bias appeared symmetrical upon visual inspection, indicating no bias. Low-power studies were relatively absent from the funnel plot, and high-power studies presented lesser variations and asymmetry (Supplemental figure S1). These results indicate that the current analysis is unlikely to be influenced by publication bias, and therefore, significant confidence can be held in the conclusion of the current analyses.

Discussion

The study was undertaken with the research question of whether QEEG changes predict the treatment response to antidepressants among patients with depression. The included studies measured various QEEG markers – Cordance (n = 8), ATR (n = 2), and relative and absolute spectral power (n = 10). Response to the prescribed antidepressant can be predicted based on decreased prefrontal cordance, change in ATR Index,³¹ and relative and absolute spectral powers for channels majorly in the frontal region.^11,29,30 The frontal alpha and theta waves also significantly predicted treatment response to SSRIs.^11,27 A robust effect size (d = 0.80), with low heterogeneity, for at least two weeks of antidepressant therapy, and the subgroup analysis suggesting theta waves as a better predictor of the ATR (d = 0.91) were the main conclusions of this meta-analysis.

These analyses reflect the predictive power of QEEG for ATR and a strong unidirectional funnel plot symmetry, indicating the overall evidence for predicting medication response with a high effect size.³⁹ The meta-analysis has some limitations, however. First, considering the study’s methodological aspects, some artifacts may arise from the nature of the studies selected/included. We considered individual studies’ operational definitions for “marker” and “responders,” which left us with a diverse nature of variables/data in quality but not enough quantity to consider each variable for all QEEG measures reliably. Though narrowing these variables further in our selection criteria could have generated a data set uniformly, it could have also left us with an even lesser number of studies to gather the evidence in any direction. The literature and clinical acumen resonate toward six to eight weeks of lapse before we gauge treatment response. Still, we considered a one to two-week decline in measures to predict the treatment response and could not streamline the studies for longer durations of active medications. Second, a lack of patient-level data and small sample sizes in studies could have inflated the overall effect size, indicating plausible false impressions on QEEG efficacy as a marker.³⁹ Third, the diverse nature of technological options with which QEEG data was analyzed could lead to heterogeneous interpretations and redundant multiple-hypothesis testing likely to overgeneralize the findings. The risk of bias tools was incorporated to guard off such misinterpretations, but like any meta-analysis, the quality of studies included directly impacts the quality of meta-analyzed results.⁴⁰ Lastly, considering the factors related to depression, including heterogeneity in terms of diagnosis, sub-typing, treatment regime, treatment response, and duration of treatment response, could have limited the generalizability of our results.⁴¹

To quantitatively evaluate the antidepressant response using EEG, further studies can be designed using a baseline EEG and EEG a few weeks into treatment. The data extracted from the two resting awake EEGs may be used to calculate the different QEEG parameters (Cordance, ATR, and spectral power). A homogenous patient group selection and a randomized placebo-controlled study design would help further increase the results’ generalizability. The above QEEG parameters have been shown to have a large effect size in predicting antidepressant response and can be clinically used to predict treatment response in depression. Further, QEEG needs to be studied or validated rigorously with marker-specific study designs for different drug trials to ascertain its efficacy in clinical practice. Prioritizing activity-recorded EEG over resting-state markers can also bridge our understanding of functional aspects of depression since literature seems to hint toward an association between physical activity and risk for depression.⁴²

Supplemental Material

Supplemental material for this article is available online.

Supplemental Material

Supplemental material for this article available online.

Supplemental Material

Supplemental material for this article available online.

Footnotes

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.

Declaration Regarding the Use of Generative AI

None of the content in the manuscript is generated by AI, and the authors assume full responsibility for the entire content. However, Grammarly was used to ensure the language for any grammatical errors, spelling mistakes, and punctuation inaccuracies.

Funding

The authors received no financial support for the research, authorship and/or publication of this article.

Registration and Primary funding source

The study is registered on the PROSPERO network (https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=147997, https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=147997) and received a grant from the parent institute.

References

Tritschler

, Gaillard

, Gardier

, . Conséquences de l’inter-relation des systèmes monoaminergiques dans l’activité antidépressive. L’Encéphale, 2018; 44: 264–273.

Sinyor

, Schaffer

, Levitt

. The sequenced treatment alternatives to relieve depression (STAR*D) trial: a review. Can J Psychiatry, 2010; 55: 126–135.

Bennabi

, Charpeaud

, Yrondi

, . Clinical guidelines for the management of treatment-resistant depression: French recommendations from experts, the French Association for Biological Psychiatry and Neuropsychopharmacology and the fondation FondaMental. BMC Psychiatry, 2019; 19: 262.

Kingslake

, Dias

, Dawson

, . The effects of using the PReDicT test to guide the antidepressant treatment of depressed patients: Study protocol for a randomized controlled trial. Trials, 2017; 18: 558.

Godlewska

, Browning

, Norbury

, . Predicting treatment response in depression: The role of anterior cingulate cortex. Int J Neuropsychopharmacol, 2018; 21: 988–996.

Watts

, Pulice

, Reilly

, . Predicting treatment response using EEG in major depressive disorder: A machine-learning meta-analysis. Transl Psychiatry, 2022; 12: 332.

Cook

. Early changes in prefrontal activity characterize clinical responders to antidepressants. Neuropsychopharmacology, 2002; 27: 120–131.

Cook

, Hunter

, Abrams

, . Midline and right frontal brain function as a physiologic biomarker of remission in major depression. Psychiatry Res Neuroimaging, 2009; 174: 152–157.

Leuchter

, Cook

, Gilmer

, . Effectiveness of a quantitative electroencephalographic biomarker for predicting differential response or remission with escitalopram and bupropion in major depressive disorder. Psychiatry Res, 2009; 169: 132–138.

10.

Bares

, Brunovsky

, Kopecek

, . Early reduction in prefrontal theta QEEG cordance value predicts response to venlafaxine treatment in patients with resistant depressive disorder. Eur Psychiatry, 2008; 23: 350–355.

11.

Knott

, Mahoney

, Kennedy

, . Pre-treatment EEG and it’s relationship to depression severity and paroxetine treatment outcome. Pharmacopsychiatry, 2000; 33: 201–205.

12.

Iosifescu

, Greenwald

, Devlin

, . Frontal EEG predictors of treatment outcome in major depressive disorder. Eur Neuropsychopharmacol, 2009; 19: 772–777.

13.

Caudill

, Hunter

, Cook

, . The antidepressant treatment response index as a predictor of reboxetine treatment outcome in major depressive disorder. Clin EEG Neurosci, 2015; 46: 277–284.

14.

Widge

, Rodriguez

, Carpenter

, . EEG biomarkers for treatment response prediction in major depressive illness. Am J Psychiatry, 2019; 176: 82–82.

15.

Ouzzani

, Hammady

, Fedorowicz

, . Rayyan—a web and mobile app for systematic reviews. Syst Rev, 2016; 5: 210.

16.

Beck

. An inventory for measuring depression. Arch Gen Psychiatry, 1961; 4: 561.

17.

Hamilton

. A rating scale for depression. J Neurol Neurosurg Psychiatry 1960; 23(1):56-62. doi: 10.1136/jnnp.23.1.56.

18.

Montgomery

, Smeyatsky

, De Ruiter

, . Profiles of antidepressant activity with the Montgomery-Asberg depression rating scale. Acta Psychiatr Scand, 1985; 72: 38–42.

19.

Rush

, Trivedi

, Ibrahim

, . The 16-item quick inventory of depressive symptomatology (QIDS), clinician rating (QIDS-C), and self-report (QIDS-SR): A psychometric evaluation in patients with chronic major depression. Biol Psychiatry, 2003; 54: 573–583.

20.

Sterne

JAC

, Savović

, Page

, . RoB 2: A revised tool for assessing risk of bias in randomised trials. BMJ, 2019; 14898.

21.

Sterne

, Hernán

, Reeves

, . ROBINS-I: A tool for assessing risk of bias in non-randomised studies of interventions. BMJ, 2016; i4919.

22.

Wallace

, Dahabreh

, Trikalinos

, . Closing the gap between methodologists and end-users: R as a computational back-end. J Stat Softw; 49. Epub ahead of print 2012. DOI: 10.18637/jss.v049.i05.

23.

Moher

, Liberati

, Tetzlaff

, . Preferred reporting items for systematic reviews and meta-analyses: The PRISMA statement. Review.

24.

Schmidt

, Sander

, Dietz

M-E

, . Brain arousal regulation as response predictor for antidepressant therapy in major depression. Sci Rep, 2017; 7: 45187. doi: 10.1038/srep45187

25.

Bruder

, Stewart

, Tenke

, . Electroencephalographic and perceptual asymmetry differences between responders and non-responders to an SSRI antidepressant. Biol Psychiatry.

26.

Cook

, Leuchter

, Witte

, . Neurophysiologic predictors of treatment response to fluoxetine in major depression. Psychiatry Res, 1999; 85(3): 263–273. doi: 10.1016/s0165-1781(99)00010-4

27.

Korb

, Hunter

, Cook

, . Rostral anterior cingulate cortex theta current density and response to antidepressants and placebo in major depression. Clin Neurophysiol, 2009; 120: 1313–1319.

28.

Bares

, Brunovsky

, Kopecek

, . Changes in QEEG prefrontal cordance as a predictor of response to antidepressants in patients with treatment-resistant depressive disorder: A pilot study. J Psychiatr Res, 2007; 41: 319–325.

29.

Spronk

, Arns

, Barnett

, . An investigation of EEG, genetic and cognitive markers of treatment response to antidepressant medication in patients with major depressive disorder: A pilot study. J Affect Disord, 2011; 128: 41–48.

30.

Knott

. Quantitative EEG in the prediction of antidepressant response to imipramine. J Affect Disord, 1996; 39: 175–184.

31.

Kwon

, Youn

, Jung

. Right hemisphere abnormalities in major depression: Quantitative electroencephalographic findings before and after treatment. J Affect Disord, 1996; 40: 169–173.

32.

Arns

, Bruder

, Hegerl

, . EEG alpha asymmetry as a gender-specific predictor of outcome to acute treatment with different antidepressant medications in the randomized iSPOT-D study. Clin Neurophysiol, 2016; 127: 509–519.

33.

Bares

, Brunovsky

, Novak

, . QEEG theta cordance in the prediction of treatment outcome to prefrontal repetitive transcranial magnetic stimulation or venlafaxine ER in patients with major depressive disorder. Clin EEG Neurosci, 2015; 46: 73–80.

34.

Tenke

, Kayser

, Manna

, . Current source density measures of electroencephalographic alpha predict antidepressant treatment response. Biol Psychiatry, 2011; 70: 388–394.

35.

Bruder

, Sedoruk

, Stewart

, . Electroencephalographic alpha measures predict therapeutic response to a selective serotonin reuptake inhibitor antidepressant: Pre- and post-treatment findings. Biol Psychiatry, 2008; 63: 1171–1177.

36.

Leuchter

, Cook

, Hunter

, . A new paradigm for the prediction of antidepressant treatment response. Dialogues Clin Neurosci, 2009; 11: 435–446.

37.

Bares

, Brunovsky

, Novak

, . The change of prefrontal QEEG theta cordance as a predictor of response to bupropion treatment in patients who had failed to respond to previous antidepressant treatments. Eur Neuropsychopharmacol, 2010; 20: 459–466.

38.

Bares

, Novak

, Brunovsky

, . The change of QEEG prefrontal cordance as a response predictor to antidepressive intervention in bipolar depression. A pilot study. J Psychiatr Res, 2012; 46: 219–225.

39.

Mathew

. Systematic reviews and meta-analysis: A guide for beginners. Indian Pediatr, 2022; 59: 320–330.

40.

Muka

, Glisic

, Milic

, . A 24-step guide on how to design, conduct, and successfully publish a systematic review and meta-analysis in medical research. Eur J Epidemiol, 2020; 35: 49–60.

41.

Krishnan

, Nestler

. The molecular neurobiology of depression. Nature, 2008; 455: 894–902.

42.

Pearce

, Garcia

, Abbas

, . Association between physical activity and risk of depression: A systematic review and meta-analysis. JAMA Psychiatry, 2022; 79: 550.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.04 MB

0.05 MB

Meta-analysis on QEEG Changes to Antidepressant Treatment Among Patients with Depression

Abstract

Purpose of the Review:

Collection and Analysis of Data:

Conclusions:

Keywords

Key Messages:

Methods

Search Strategy

Eligibility Criteria

Quality of the Studies

Data Extraction and Statistical Analysis

Results

Search Results

PRISMA Flow Chart of the Studies Included in the Meta-analysis.

Study Characteristics

Efficacy of QEEG in the Prediction of Antidepressant Response

Subgroup Analysis of QEEG

Publication Bias

Discussion

Supplemental Material

Supplemental Material

Supplemental Material

Footnotes

Declaration of Conflicting Interests

Declaration Regarding the Use of Generative AI

Funding

Registration and Primary funding source

References

Supplementary Material