Introduction
The abilities to regulate alertness and sustain attention are essential for efficient information processing and behavior (Posner and Petersen
1990). Such cognitive processes are traditionally measured with reaction time variability (RTV), capturing the consistency and short-term fluctuations in response speed during attentional performance in cognitive tasks (Kuntsi et al.
2013; Ode et al.
2011). Increases in RTV are characteristic of several psychiatric disorders (Kaiser et al.
2008), including attention-deficit/hyperactivity disorder (ADHD) (Kofler et al.
2013; Wood et al.
2010) and bipolar disorder (BD) (Brotman et al.
2009; Moss et al.
2016). ADHD and BD are common psychiatric conditions in adults (Merikangas et al.
2011; Willcutt
2012), which severely impact many aspects of individuals’ lives (Asherson et al.
2014; Skirrow et al.
2012). Although ADHD and BD represent distinct disorders, they present with common symptoms of distractibility and difficulty concentrating, which can lead to uncertainty regarding the boundaries of the two disorders (Asherson et al.
2014; Kitsune et al.
2016). These overlapping symptoms may reflect, at the cognitive level, the common fluctuations in attentional performance and increased RTV displayed by individuals with ADHD and BD (Albaugh et al.
2017; Kuntsi et al.
2014). Increased RTV is also observed in unaffected first-degree relatives of individuals with either disorder, compared to individuals without family risk, representing a candidate marker of genetic/familial risk for both disorders (Adleman et al.
2014; Andreou et al.
2007). Direct comparisons of impairments in attentional performance between ADHD and BD may lead to new insights into the pathways to overlapping symptoms and cognitive dysfunction in both disorders. Yet, cross-disorder comparisons in ADHD and BD are limited to date (Michelini et al.
2016; Rommel et al.
2016; Torralva et al.
2011).
Previous research on RTV in psychiatric disorders has addressed the question of whether dysfunctions in alertness and attentional performance, rather than being stable, could be malleable and sensitive to context changes, such as task manipulations. RTV impairments in children and adolescents with ADHD are maximal in slow and unrewarded conditions, but with the introduction of faster event rate and incentives may improve significantly more than in neurotypical individuals (Andreou et al.
2007; Cheung et al.
2017; Kuntsi et al.
2013; Slusarek et al.
2001; Uebel et al.
2010). It remains unknown, however, whether RTV also improves in adults with ADHD. Initial evidence also indicates potential malleability of RTV in BD, as suggested by one study showing increased RTV in individuals with BD in a continuous performance task (CPT) with low target frequency, but not with high target frequency (Moss et al.
2016). The evidence of malleability in RTV is clinically relevant, as it may point to room for improvement in the observed cognitive impairment, which could be targeted in new interventions for the disorders, aimed at reaching and maintaining an optimal state of alertness (Cheung et al.
2017; Kuntsi and Klein
2012). Understanding whether the same or different mechanisms underlie attentional fluctuations and their potential reduction in individuals with ADHD and BD may thus potentially inform the development of interventions for ADHD and BD. No study to date, however, has compared adults with ADHD and adults with BD on the malleability of attentional fluctuation indexed by RTV.
The investigation of brain responses using the millisecond temporal precision of electroencephalography (EEG) can help elucidate the neural correlates of a suboptimal attentional performance. Most EEG studies on attentional impairments in ADHD or BD samples have employed event-related potentials (ERPs), measuring transient enhancements in brain activity following an event (Luck
2014). ERP studies in adults with ADHD have shown attenuated contingent negative variation (CNV) components over central regions (reflecting atypical response anticipation and preparation) (McLoughlin et al.
2010; Michelini et al.
2016; Valko et al.
2009) and reduced attentional P3 components over parietal regions (reflecting impaired attentional resource allocation) (Cheung et al.
2017,
2016; McLoughlin et al.
2010; Szuromi et al.
2011). Similarly, impairments in P3 and CNV in BD during attentional tasks have also been found (Fridberg et al.
2009; Li et al.
2015; Maekawa et al.
2013). Yet, only a few direct comparisons have examined whether cognitive and ERP indices are affected to a similar extent in ADHD and BD. In a recent investigation using a cued CPT paradigm, we showed that increased RTV and reduced CNV may represent shared attentional impairments in ADHD and BD (Michelini et al.
2016). Using quantitative EEG (QEEG), we further reported that both ADHD and BD groups showed higher spontaneous EEG theta power during rest and a lack of a task-related increase in theta from rest to CPT task compared to controls (Rommel et al.
2016). These results indicate potentially shared impairments in attentional processes in both disorders. Yet, in ERP analyses, the attentional P3 components in response to cue and target stimuli were intact in both groups, consistent with other studies that also failed to report P3 reductions in adults with ADHD (Dhar et al.
2010; Grane et al.
2016; Michelini et al.
2016) or BD (Bestelmeyer
2012; Chun et al.
2013; Michelini et al.
2016). One possible reason for inconsistencies between studies using different attentional paradigms is that the attentional P3, similar to RTV, may reflect a context-dependent and potentially malleable, rather than stable, impairment (Cheung et al.
2017). We recently reported that a reduced parietal P3 in a slow and unrewarded condition in adolescents and young adults with ADHD improved with faster event rate and rewards significantly more than in neurotypical controls (Cheung et al.
2017). In contrast, for CNV, the ADHD group showed reduced amplitude compared to controls only in the fast and rewarded condition. No study has examined the malleability of these ERPs with faster rate and incentives in BD. Further direct comparisons between ADHD and BD are needed to clarify what neurophysiological impairments overlap between the two disorders, and whether ADHD and BD may show similar malleability with a changed context.
Advances in EEG methods called time–frequency analyses, combining the strengths of ERP and QEEG methods, further allow to capture event-related brain oscillatory dynamics, which reflect sub-second modulations of power and phase in response to an event across the full EEG spectrum (Klimesch
1999; Loo et al.
2015; Makeig et al.
2004; Pfurtscheller and Lopes da Silva
1999). Processing and focusing attention on a relevant stimulus have been associated with various event-related brain oscillatory phenomena in the time–frequency domain not captured by ERP or QEEG approaches: (1) an event-related synchronization (ERS) or increase in theta (3–7 Hz) power over fronto-central (Bickel et al.
2012; Lenartowicz et al.
2014; Mazaheri et al.
2014) or parietal (Babiloni et al.
2004; Jacobs et al.
2006) regions, reflecting the initial processing of the stimulus; (2) an event-related desynchronization (ERD) or suppression of power in posterior alpha (8–13 Hz), reflecting attentional selection and cortical activation (Klimesch
2012; Mazaheri and Picton
2005); and (3) an ERD in central beta (14–30 Hz) when a motor response is required (Guntekin et al.
2013; Pfurtscheller
1981). Additionally, indices of consistency of the phase (i.e., the “timing”) of brain oscillations over trials can reveal whether the processing of a stimulus repeated over time reflects stable or variable neural mechanisms (Klimesch
2012; Makeig et al.
2004; Papenberg et al.
2013). Greater alpha and beta ERD and theta phase consistency have further been associated with better task performance (Bickel et al.
2012; Klimesch
2012; McLoughlin et al.
2014). Multiple brain-oscillatory correlates of attentional processes may be affected in ADHD and BD. Individuals with ADHD have been reported to show reductions in event-related phase consistency in the theta band (Groom et al.
2010; McLoughlin et al.
2014), alpha ERD (Lenartowicz et al.
2014; ter Huurne et al.
2013), and beta ERD (Hasler et al.
2016). Emerging evidence also suggests that individuals with BD show attenuations in event-related theta (Atagun et al.
2013; Ethridge et al.
2012) and alpha power (Basar et al.
2012; Ethridge et al.
2012) and increases in beta power (Ozerdema et al.
2013; Tan et al.
2016). These studies in BD, however, applied time–frequency analyses on averaged ERP responses, thus not allowing the characterization of both ERD and ERS dynamics (Bickel et al.
2012). The investigation of fine-grained brain-oscillatory indices underlying attentional processes with time–frequency analyses may allow a deeper investigation into the neural correlates of attentional performance, and help identify distinct or comparable impairments in neural processes between the two disorders (Loo et al.
2015). However, no study to date has compared ADHD and BD on time–frequency indices of brain oscillations, or whether these indices, like RTV, show adjustments under context changes, such as fast and rewarded conditions.
The present study aims to investigate and compare cognitive-performance, ERP and detailed event-related power modulations of theta, alpha and beta oscillations and of phase variability in theta oscillations, previously linked to attentional processes, in adults with ADHD and adults with BD. We used an all-female sample, to match the groups on sex but also because little is known on these processes in females, especially in relation to ADHD (McLoughlin et al.
2010; Saville et al.
2015). Participants completed the same four-choice reaction time task used in our previous studies of ADHD (Andreou et al.
2007; Cheung et al.
2017; Kuntsi et al.
2006), which compares a slow-unrewarded baseline condition with a fast-incentive condition designed to specifically reward reduction of RTV. A further aim is to examine whether differences in adjustments in the investigated cognitive-performance, ERP and brain-oscillatory indices with a faster event rate and incentives emerge between groups, which could inform the development of cognitive/brain training programs for ADHD and BD.
Discussion
In this comparison between ADHD and BD on cognitive, ERP and brain-oscillatory markers of attentional processes, women with ADHD and women with BD showed overlapping impairments in fluctuations in attentional performance (RTV), neural variability (theta ITC) and neural response preparation (CNV). Individuals with either disorder further displayed a similar inability to adjust neural attention allocation (P3) and activation (alpha suppression) from a baseline to a fast-paced and rewarded condition, suggesting no adaptation to a changed context in these processes. Additional disorder-specific alterations in alpha and beta suppression were displayed by women with ADHD only, but impairments in most processes were shared between the two disorders. By examining both ERP and fine-grained brain-oscillatory indices of brain activity, these findings reveal novel neural mechanisms of shared attentional dysfunction in ADHD and BD, which potentially underlie some of the common symptoms in both disorders.
At the cognitive level, both ADHD and BD groups showed increased RTV in both task conditions, indicating more frequent fluctuations in response speed and impairments in the ability to sustain attention during the task. Increased RTV in both disorders is consistent with our results with this sample using a cued CPT task (Michelini et al.
2016), and previous studies on ADHD (Cheung et al.
2016; Kofler et al.
2013; Kuntsi et al.
2010) and BD (Bora et al.
2006; Brotman et al.
2009; Moss et al.
2016). We further show novel evidence of intra-individual variability also at the neural level in the phase of theta oscillations in both women with ADHD and women with BD. Low phase variability over trials is thought to reflect an adaptive mechanism to maintain stable neural processing of a stimulus (Makeig et al.
2004; Papenberg et al.
2013). The increased variability in theta oscillations, previously reported in adolescents with ADHD (Groom et al.
2010; McLoughlin et al.
2014), thus points to increased variability in the timing of evoked theta responses to targets over trials in adults with ADHD and BD (Cavanagh et al.
2009; McLoughlin et al.
2014). Although these differences emerged as significant only in the fast-incentive condition, the group-by-condition interaction was not significant, suggesting that there may be subtle differences also in the baseline condition, non-significant in this sample. Further analyses in the pre-stimulus window indicated that, compared to individuals with ADHD or BD, control women displayed greater phase consistency upon target presentation, but lower consistency before targets. As such, with presentations of targets across trials, the controls displayed a consistent alignment and increase in consistency in the phase of theta (called phase resetting) (Lakatos et al.
2009; Palaniyappan et al.
2012) from the low consistency observed in the pre-stimulus window. This mechanism may be lacking in women with either ADHD or BD as indicated by the more frequent fluctuations in this neural mechanism across trials. Overall, our findings of increased variability in cognitive and neural processes in women with ADHD or BD indicate an overlap in the neural underpinnings of impaired attentional fluctuations in both disorders, which may point to common neurobiological dysfunctions.
By further examining pre-stimulus response preparation in ADHD and BD, we found shared preparatory impairments, as indicated by reduced CNV, in both clinical groups in the fast-incentive condition. This finding is consistent with our previous results in this sample using a CPT task (Michelini et al.
2016), and in adolescents and young adults with ADHD using the same task employed in this study (Cheung et al.
2017). Suggestive (trend-level) differences between ADHD and BD in this measure may also indicate more pronounced CNV impairment in ADHD, although this awaits replication in future studies. The pattern for P3 amplitude in response to targets, which was not different from controls in either ADHD or BD groups, indicates that women with either disorder may not be impaired in this ERP of attentional allocation. This result is consistent with our previous study with this sample (Michelini et al.
2016), showing intact P3s following cue and target stimuli, and other previous studies reporting normal attentional P3 amplitudes in adults with ADHD (Barry et al.
2009; McLoughlin et al.
2010) or BD (Bestelmeyer
2012). Yet, this P3 finding does not align with our previous larger-scale investigation using this task in ADHD, where our predominantly-male group of adolescents and young adults with ADHD (mean age: 18 years) showed a reduced P3 in the baseline condition (with a small effect size) compared to controls (Cheung et al.
2017). In the current study, the intact target P3 in ADHD may be due to gender or age, the present study being the first on this task using an all-female and all-adult sample (mean age: 37 years). In addition, the ADHD group had lower IQ than the control group in our previous study, and the ADHD-control difference on the P3 was non-significant when IQ was controlled for (Cheung et al.
2017). The lack of IQ differences between groups in the current sample may have contributed to the lack of group differences in the P3. Taken together, these findings indicate that both ADHD and BD are associated with reduced ERP activity of attentional preparation and anticipation of motor responses.
With faster target presentation and incentives, further shared impairments between ADHD and BD emerged in adjustments between conditions. These task manipulations, originally designed in ADHD studies to reward more consistent response times, produced comparable reductions in RTV in clinical and control groups. At the neural level, women with ADHD, and potentially (at trend-level) with BD, displayed significantly reduced increases in CNV amplitude compared to controls, and no improvements in allocation of attentional resources (P3) (Polich
2007) or attentional selection (alpha suppression) (Klimesch
2012). The novel finding of a reduced ability to increase alpha suppression with task demands in both disorders points to a common inability in individuals with ADHD and BD to regulate brain activity implicated in attentional selection processes (Klimesch
2012; Klimesch et al.
2007). A reduced adjustment in the response preparation CNV in women with ADHD replicates our previous findings in adolescents and young adults with the disorder (Cheung et al.
2017). Yet, in the P3, neither of the clinical groups showed the improvement between conditions displayed by controls. This pattern for the P3 contrasts with our previous findings using this task in a sample of adolescents and young adults with ADHD, where the ADHD group showed improvements between conditions in the P3, which were greater than those observed in the control group, suggesting malleability in this attentional ERP component in ADHD (Cheung et al.
2017). Similarly, these results in ADHD do not align with studies in children, adolescents and young adults indicating greater RTV malleability and improvements in ADHD than in neurotypical samples (Andreou et al.
2007; Cheung et al.
2017; Kuntsi et al.
2013,
2009). A possible explanation for the inconsistencies in P3 and RTV adjustments is the age difference between the samples of current and previous studies: it could be hypothesized that adults with ADHD, compared to younger individuals, may be less sensitive to task manipulations in these processes. Gender effects represent another possible reason for inconsistencies with previous studies based on predominantly-male samples (Andreou et al.
2007; Cheung et al.
2017). Longitudinal studies and replications in larger samples, including individuals of both sexes, are needed to examine potential developmental and gender effects on the malleability of markers of attentional processes in ADHD.
While most impairments were shared between ADHD and BD, we further found impairments specific to ADHD relative to controls, which were not displayed by women with BD. Women with ADHD displayed a dysfunction in attentional selection, as indexed by lower alpha power suppression in response to targets in the fast-incentive condition (Klimesch
2012; Klimesch et al.
2007). These results are consistent with previous studies reporting attenuated event-related alpha suppression in ADHD (Hasler et al.
2016; Lenartowicz et al.
2014; Mazaheri et al.
2014; Missonnier et al.
2013). In addition, in the change from the baseline to the fast-incentive condition, individuals with ADHD were specifically associated with lower adjustments in the suppression of beta power than in individuals with BD and controls, indicating reduced improvements in neural mechanisms associated with response execution (Bickel et al.
2012; Mazaheri et al.
2014). The lack of a difference between women with BD and controls on measures of alpha and beta power suppression has not been reported in previous studies of BD, which measured event-related power extracted from averaged ERP responses and thus could not capture event-related power suppressions. Future investigations using the finer-grained time–frequency methods employed in this study are needed to confirm typical profiles in alpha and beta power suppression measures in individuals with BD. While the ADHD-specific impairment in alpha suppression did not distinguish women with ADHD from women with BD, the reduction in the adjustment in beta power suppression with task demands significantly differentiated the two clinical groups. The latter brain-oscillatory process may thus represent neurobiological dysfunctions specific to ADHD, which may potentially help delineate ADHD from BD in adults.
The following limitations should be considered when interpreting our findings. First, although the groups were matched on gender, age and IQ, there were differences in the prescribed medications that participants were taking. We asked participants with ADHD to stop taking stimulant medication 48 h before assessments, but it was not possible, for ethical reasons, to ask participants to stop mood-stabilizing, anti-psychotic or antidepressant medications. Medication effects are difficult to control for in cross-disorder studies where different groups are prescribed different treatments, resulting in a limited number of participants within medication subgroups. However, previous studies suggest that medication may show positive effects (reducing differences from controls) or no effects on cognitive-EEG measures (Anderer et al.
2002; Degabriele and Lagopoulos
2017; Galletly et al.
2005; Karaaslan et al.
2003). As such, it is unlikely that the significant group differences reported in this study reflect confounding medication effects. Yet, the possibility remains that the lack of differences between clinical groups and the control group on some measures (especially for the BD group, where the majority of individuals were taking medication) may be due to medication effects, which may have attenuated case-control differences. Future studies on samples including non-medicated individuals or a higher number of individuals in each medication sub-group are needed to clarify this issue. Second, while the two task conditions were matched on number of trials, they differed in duration and in length of the fore-period between warning and target stimuli. While we obtained comparable findings in RTV with length-matched segments, ERP/EEG analyses could not be repeated on length-matched segments, as doing so would have produced insufficient number of trials in the baseline condition to obtain reliable ERP/EEG indices. In addition, the different fore-periods and the use of a 0.25 Hz high-pass filter may reduce comparability of preparatory activity between the conditions. Yet, the analysis of the CNV (showing typical topographies at central sites) and the further analyses of EEG activity in the warning-target interval under fast-incentive conditions (Supplementary material) allowed detailed investigation of neural preparatory processes in this latter condition. Future studies could examine stimulus-related and preparatory processes in ADHD and BD using other tasks, as well as examine the influences on slower frequencies on the CNV. Third, although the current study and previous analyses on this sample (Michelini et al.
2016; Rommel et al.
2016) represent the most comprehensive comparisons between ADHD and BD on cognitive, ERP and EEG markers to date, the sample is relatively small. While several significant differences between groups emerged with medium-to-large effects with current sample sizes, larger studies are needed to confirm these results and further investigate subtler impairments in ADHD and BD. Finally, the adult participants in the clinical groups recruited for this study showed higher than expected IQ scores, which did not differ from IQ scores in the control group. Future replication in samples with a wider range of IQs is required in order to generalize these findings to more typical clinical populations.
Taken together, these findings further our understanding of the neural underpinnings of attentional impairments in both disorders, and provide new evidence into the overlap and specificity of impairments in these processes in women with ADHD and BD. The shared markers of attentional dysfunctions may represent biomarkers for both disorders. The shared atypical neural profiles related to attentional processes may underlie similarities in behavioral symptoms (e.g., distractibility) between ADHD and BD, which can lead to difficulty in delineating between ADHD and BD and incorrect treatment decisions. Finally, since ADHD and BD show genetic overlap (Lee et al.
2013; Song et al.
2015; van Hulzen et al.
2016), and increased attentional fluctuations may represent candidate markers of genetic/familial risk for both disorders (Adleman et al.
2014; Andreou et al.
2007), future studies could examine whether shared genetic factors may underlie overlapping attentional dysfunctions in ADHD and BD.
Acknowledgements
We thank all who made this research possible: The National Adult ADHD Clinic at the South London and Maudsley Hospital, Dr Helen Costello, Prof Sophia Frangou, Prof Anne Farmer, Jessica Deadman, Hannah Collyer, Sarah-Jane Gregori, and all participants who contributed their time to the study. Dr Giorgia Michelini was supported by a 1+3 PhD studentship awarded by the MRC Social, Genetic and Developmental Psychiatry Centre, Institute of Psychiatry, Psychology and Neuroscience, King’s College London (G9817803). This project was supported by an Economic and Social Research Council studentship to Dr Viryanaga Kitsune (ES/100971X/1). Dr Giorgia Michelini and Prof Philip Asherson are supported by generous grants from the National Institute for Health Research Biomedical Research Centre for Mental Health at King’s College London, Institute of Psychiatry, Psychology and Neuroscience and South London and Maudsley National Health Service (NHS) Foundation Trust. The funders had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.