Table 1.
Details of MRI studies investigating candidate predictive/prognostic biomarkers for treatment response.
| Authors, year | Country | N ADHD | N Controls | Age | % Male | % White | Design | Candidate biomarker(s) | Key findings |
|---|---|---|---|---|---|---|---|---|---|
| Griffiths et al. [20] | Australia | 37 (19 had taken psychostimulants in the previous 6-months, and underwent a washout period) | 26 | M = 13.29 SD = 2.64 | 70% | Not reported | 6-week open label study of MPH, pre-treatment neuroimaging | White matter local efficiency (graph theoretical measure of how efficiently information can be distributed between a brain region and its neighbors) | Support vector machine learning algorithm applied to multivariate measures of local efficiency predict treatment response assessed at 6-weeks using the ADHD-RS-IV. The most predictive features were higher local efficiency of the thalamus, precentral gyrus and DLPFC. |
| Hong et al. [24] | South Korea | 83 (medication-free for >4 weeks, and with no history of long-term treatment for ADHD, defined as medicated >6 months) | 22 | M = 9.63 SD = 2.61 | 25% | Not reported | 8-week RCT of MPH, pre-treatment neuroimaging | Resting-state connectivity assessed using striatal seeds | Treatment responders (n = 48), defined according to the CGI-I at 8-weeks, showed greater pre-treatment connectivity between striatal seeds and orbitofrontal, cingulo-opercular and middle and medial temporal regions than did non-responders. |
| Kim et al. [18] | South Korea | 67 (medication-free for >4 weeks, and with no history of long-term treatment for ADHD, defined as medicated >6 months) | 25 | M = 9.83 SD = 2.5 | 25% | Not reported | 8-week RCT of MPH, pre-treatment neuroimaging | Subcortical volume | Responders had smaller volumes in bilateral amygdala and hippocampal subregions and right thalamus than non-responders. |
| Lam et al. [29] | UK | 31 (24 subjects were receiving stable medication) | 0 | M = 13.90 SD = 1.58 | 100% | Not reported | 2-week real-time fMRI neurofeedback of the right IFG vs. neurofeedback of a control para-hippocampal region. Single-blind RCT | Brain activation during inhibitory control task as a predictor of neurofeedback learning | Better neurofeedback learning was associated with pre-treatment activation in left IFG/insula and striatum during the fMRI stop task. |
| Mizuno et al. [83] | Japan | 27 (all medication-free for >5 times half-lives) | 49 | M = 10.96, SD = 2.14 | 100% | Not reported | Double-blind, placebo-controlled, crossover design comparing single-dose MPH and placebo | Dynamic resting-state functional connectivity | Dynamic network interactions under placebo predicted individual differences in sustained attention improvements under MPH. |
| Moreno et al. [19] | Spain | 27 (all treatment naïve) | 0 | M = 9.33 SD = 2.49 | 70% | Not reported | 4-week open-label trial of MPH, pre-treatment neuroimaging | Subcortical gray matter concentration | Treatment responders, as defined via clinical interview and administration of CGI and CGAS, showed greater gray matter concentration within the nucleus accumbens and caudate compared with non-responders. |
| Norman et al. [22] | USA | 110 (medicated) | 142 | age range, 6–17 years | 65% | Not reported |
Naturalistic longitudinal study of chronically medicated subjects, including up to 5 assessments. Scanning was performed during washout period. |
Resting-state connectivity within and between cingulo-opercular, default mode and subcortical networks was assessed while subjects were off medication. | ADHD symptoms were rated on and off medication using the DICA-IV interview for parents. Non-responders showed developmentally atypical increases in cingulo-opercular connectivity with age, while responders showed a developmental trajectory that tracked that of the controls. |
| Peterson et al. [21] | USA | 16 (psychostimulant responders) | 20 | M = 13.71 SD = 2.85 | 69% | 94% | Non-blinded study of chronically medicated psychostimulant responders with ADHD. Subjects were scanned on and off medication | Brain activation was assessed during a stop task both on and off medication | Off-medication left lateral prefrontal cortex activation correlated with differences in total ADHD symptoms, which were assessed using on and off medication versions of the CPRS in youth with ADHD. |
| Schulz et al. [27] | USA | 36 (medication-naïve) | 0 | M = 11.0 SD = 2.4 | 83% | Not reported | 8-week MPH/ATX randomized cross-over design, pre-treatment | Brain activation during a go/no-go task | Greater pre-treatment caudate activation was associated with a better treatment response to MPH, but a worse response to ATX, as assessed using ADHD-RS-IV. |
ADHD attention deficit hyperactivity disorder, ADHD-RS-IV ADHD Rating Scale-IV, ATX atomoxetine, CGAS Children’s Global Assessment Scale, CGI Clinical Global Impressions scale, CPRS Conners’ Parent Rating Scale, DICA-IV Diagnostic Interview for Children and Adolescents – IV; fMRI functional magnetic resonance imaging; IFG inferior prefrontal gyrus, MPH methylphenidate, MRI magnetic resonance imaging, RCT randomized controlled trial.