Table 2.
Details of EEG studies investigating candidate predictive/prognostic biomarkers for treatment response.
| Authors, year | Country | N ADHD | N Control | Age | % Male | % White | Design | Candidate biomarker(s) | Key findings |
|---|---|---|---|---|---|---|---|---|---|
| Arns et al. [38] | USA, Australia, Netherlands | 336 (all medication free for >7 days) | 158 | M = 12 | 72% | Not reported | 6-week MPH, open label | iAPF during rest | Lower pre-treatment frontal iAPF in male adolescent non-responders relative to responders. No difference in pre-treatment TBR, age, medication dosage, ADHD severity. |
| Chiarenza et al. [44] | Italy | 61 (all medication free for >5 times half-lives) | Not reported (reference database) | M = 10.4, SD = 2.9 | 85% | Not reported | 12-month ATX, open label | Absolute power across frequency bands during rest | Higher pre-treatment frontal alpha and fronto-temporal delta and theta power in responders relative to controls. Higher pre-treatment absolute power in all frequency bands (especially frontal and central) in non-responders relative to controls. |
| Griffiths et al. [45] | Australia | 52 (all medication free) | 52 | M = 11.9, SD = 2.5 | 83% | Not reported | 6-week ATX vs. placebo, cross-over RCT | N2 amplitude during an auditory oddball task | Lower pre-treatment N2 amplitudes (especially right fronto-central) in responders relative to non-responders and controls. N2 predicted responders vs. non-responders with specificity = 80.8% and sensitivity = 47.1% in a leave-one-out cross validation analyses. |
| Krepel et al.[50] | Netherlands, Germany, Australia | 136 (43 medication free, 93 medicated) | 0 | M = 24.9, SD = 14.9 | 89% | Not reported | QEEG-informed NF (NF protocols based on individual EEG), open label | P3 amplitude (females), iAPF (males) during rest | Shorter pre-treatment P3 latencies in girls/women who remitted; lower pre-treatment iAPFs in boys/men who remitted. |
| Leuchter et al. [47] | USA | 44 (medication free for >10 days) | 0 | Range = 18–30 | Not reported | Not reported | 12-week ATX vs. placebo RCT | Change in theta cordance during rest at 1-week post-treatment | Lower left temporoparietal theta cordance at 1-week post-treatment in ATX responders relative to non-responders. No difference between placebo responders and non-responders. Lower theta cordance predicted improvement in ADHD symptoms and quality of life. No association between absolute and relative power measures and clinical outcomes. |
| Loo et al. [49] | USA | 51 (all medication free for >1 month) | 0 | M = 10.3, SD = 1.4 | 67% | Not reported | 4-week active vs. sham TNS, RCT | Theta and alpha power during rest | Lower pre-treatment right-frontal theta and alpha power in responders relative to non-responders. Treatment-related change right-frontal theta predicted response AUC = 00.81. |
| Luo et al. [56] | China | 121 | 0 | M = 8.9, range= 7.1–12.3 | 83% | Not reported | 3-month remote computerized cognitive, NF, and combined training, RCT | Relative alpha power during rest | Pre-training relative alpha power correlated positively with ADHD improvements. |
| Michelini et al. [57] | USA | 207 (all medication naïve or medication free for >5 times half-lives) | 0 | M = 10.1, SD = 2.1 | 68% | 83% | 8-week MPH, GUAN, MPH + GUAN, RCT | Event-related midfrontal beta power localized in the ACC during the Sternberg spatial working memory task with encoding, maintenance and retrieval phases | Weaker mid-frontal beta power modulations across task phases predicted greater ADHD improvements with MPH + GUAN. Stronger mid-frontal beta power modulations predicted clinical improvements with GUAN (during retrieval) and binary response with MPH (during encoding). Mid-frontal beta & clinical measures at pre-treatment explained R2 = 0.41 in MPH + GUAN and R2 = 0.34 in GUAN groups; clinical measures alone explained R2 = 0.21 in MPH + GUAN and R2 = 0.14 in GUAN groups. |
| Ogrim et al. [36] | Norway | 98 | 90 | Range = 7–17 | 67% | Not reported | 4-week MPH or DEX, open label | Theta power, contingent negative variation, cue P3 and no-go-P3 during a cued go/no-go task |
Higher pre-treatment frontal theta power and cue P3 amplitudes, more negative contingent negative variation amplitude and lower posterior alpha power and no-go P3 amplitudes in responders relative to no-responders. Cue P3, no-go P3, and excess theta predicted response in a multivariate model. No difference in iAPF between responders and non-responders. |
| Ogrim et al. [42] | Norway | 87 | 0 | Range = 7–17 | 69% | Not reported | Single dose and 4-week MPH or DEX, open label | Theta/alpha power, no-go P3, contingent negative variation during cued go/no-go task | Higher pre-treatment Cz theta/alpha ratio, lower pre-treatment no-go P3, higher single-dose change in no-go P3 and lower single-dose change in contingent negative variation in responders than non-responders. An aggregate index of ERP and behavioral predictors yielded AUC = 91%, sensitivity=86%, specificity = 88%. |
| Sangal & Sangal [46] | Not reported (probably USA) | 17 (all medication naïve or medication free for >5 times half-lives) | 0 | M = 10.9, SD = 3.0 | 71% | 82% | 10-week ATX, open label | Auditory P3 amplitude during visual and oddball tasks | Higher pre-treatment P3 amplitude across regions in responders relative to non-responders, yielding PPV = 0.88 and NPV = 0.67. |
| Sangal & Sangal [41] | Not reported (probably USA) | 58 (all medication free for >1 month) | 0 | M = 10.5, SD = 2.1 | 72% | Not reported | 4-week ATX vs. MPH, cross-over RCT | Auditory P3 amplitude during visual and oddball tasks | Greater pre-treatment P3 amplitude across regions in ATX responders relative to non-responders; greater pre-treatment P3 amplitude at right temporal region in MPH responders relative to non-responders. |
| Sari Gokten et al. [35] | Turkey | 51 | 0 | M = 8.57, SD = 1.75 | 82% | Not reported | 13-month MPH, open label | Delta, theta, gamma power, delta/beta, and TBR during rest | Higher pre-treatment delta power at F8, theta power at Fz, F4, C3, Cz, T5, gamma power at T6, lower beta power at F8 and P3, delta/beta ratio at F8 and TBR at F8, F3, Fz, F4, C3, Cz, P3, and T5 predicted greater hyperactivity improvement. Theta power at Cz and T5 and TBR at C3, Cz, and T5 also accurately classified responders vs. non-responders in logistic regressions. |
| Singh et al. [48] | India | 50 (all medication naïve) | 0 | Range = 6–14 | 80% | Not reported | 6-week ATX, open label | Change in theta cordance during rest at 1-week post-treatment | Greater decrease in left temporoparietal theta cordance at 1 week in responders relative to on-responders. No difference between pre-treatment and 1 week in non-responders. |
| Voetterl et al. [52] | Netherlands, Australia, USA | Transfer: 336 MPH & 136 NF; validation: 41 MPH & 71 NF. Exploration: 55 GUAN & 47 ATX | 0 | Range = 7–15 | 100% | Not reported | Various duration, MPH, multimodal NF with with sleep coaching, GUAN, ATX, open label or RCTs | iAPF during rest |
Transfer phase: predicted gain in normalized remission of 17% to 30% after stratifying boys with a higher iAPF to MPH and boys with a lower iAPF to multimodal NF, respectively. Blinded out-of-sample validations: predicted gain in stratified normalized remission of 36% and 29%, respectively. Exploration phase: higher iAPF predicted remission with GUAN and lower iAPF predicted remission with ATX. |
| Young et al. [43] | Not reported (probably Australia) | 35 | 0 | M = 13.3, SD = 2.48 | 58% | Not reported | Single dose and 6-month MPH, open label | P3b during auditory oddball task | Acute P3b amplitude changes accurately predicted treatment outcome in 81% of cases. |
ADHD attention deficit hyperactivity disorder, ATX atomoxetine, AUC area under the curve statistic, EEG electroencephalography, DEX dexamphetamine, ERP event relate potential, GUAN guanfacine, MPH methylphenidate, iAPF individual alpha peak frequency, M mean age, NF neurofeedback, NPV negative predictive value, PPV positive predictive value, SD standard deviation, TNS trigeminal nerve stimulation.