A total of 1824 articles were recovered for screening, of which 803 duplicates were removed and 687 were excluded as they did not deal with BD, while 4 articles were included as a result of the Snowballing effect (Wohlin 2014). Subsequent to review of titles and abstracts, 687 records were discarded and the full manuscripts of 336 studies were examined in detail. Of the articles included, only 16 explored an association between biomarkers and cognition in different affective states, most of which demonstrated a correlation between the cognitive functions evaluated and the different alterations during the mood phases of the disorder.
↑, increased; ↓, decreased; >, higher; <, lower; N/A, not applicable; HAMD, Hamilton Depression Rating Scale; HDRS, Hamilton Depression Scale; MAS, Bech and Rafaelsen Mania Scale; YMRS, Young Mania Rating Scale; MADRS, Montgomery-Asberg Depression Rating Scale; BDM, bipolar mania; BDD, bipolar depression; BDE, bipolar euthymic; BDH, bipolar hypomania; BDX, bipolar mixed state; MDD, major depressive disorder; HC, healthy control; FCr, right frontal cortex; PaC, Parietal cortex; PCr, parietal cortex right; DLPFC, dorsolateral prefrontal cortex; OFC, orbitofrontal cortex; SACC, supragenual anterior cingulate cortex; TPJ L, temporal parietal junction left; TP, temporal polar cortex; IPL, inferior parietal lobe; PACC, perigenual anterior cingulate cortex; PCC, posterior cingulate cortex; vlPFC, ventrolateral prefrontal cortex; MD, mean diffusivity; RD, radial diffusivity; FA, DTI-derived fractional anisotropy; FCo, functional connectivity; SPECT, brain perfusion single-photon emission computed tomography; SC, structural connectivity; WM, White matter; NAA/Cr, N-acetylaspartate/creatine; GDNF, Glial cell line-derived neurotrophic factor; Tchol, Total cholesterol; HDL-C, high-density lipoprotein cholesterol; hs-CRP, high sensitivity C reactive protein; LDL-C, low-density lipoprotein-cholesterol; OCH-24/27, cholesterol 24/27 hydroxycholesterol; oxy-Hb, Relative concentration changes of oxygenated; deoxy-Hb, concentration changes of deoxygenated; rs-fMRI, resting-state functional magnetic resonance imaging; MRI, Magnetic Resonance Imaging; NIRS, near infrared spectroscopy; fNIRS, functional near-infrared spectroscopy; EEG, electroencephalography; DTI, probabilistic tractographic diffusion tensor imaging; MoCA, Montreal Cognitive Assessment; TAP, Word Accentuation Test; VFT, Verbal Fluency Test; WAIS, Wechsler Adult Intelligence Scale; IQ, intelligence quotient; JART, Japanese Adult Reading Test; CPT, Continuous performance test; SCIP-S, Screen for Cognitive Impairment in Psychiatry Scale; ToL, Tower of London; WCST, Wisconsin card sorting test; TMT-B, Trail making test part B; PDQ, Perceived Deficits Questionnaire.
3.1. Cognitive and biomarkers findings across affective state
Studies included markers from serum or plasma and neuroimaging. The studies were grouped into the following cognitive domains according to the tasks used or the fMRI paradigms used: "attention", "executive functions", "memory (working memory and verbal memory)", "IQ" “Self-reported cognitive” and "Cognitive Screening Test ".
Fourteen studies used a combination of neuroimaging and neurocognitive assessments to investigate the affective states in BD (Alonso-Lana et al. 2019; Estudillo-Guerra et al. 2020; Gao et al. 2023; Kopf et al. 2023; Lai et al. 2018; Magioncalda et al. 2016; Magioncalda et al. 2015; Martino et al., 2016; Mikawa et al. 2015; Nishimura et al. 2015; Pomarol-Clotet et al. 2015; Rive et al. 2016; Velasques et al. 2013; Yang et al. 2020). We find different neuroimaging modalities (e.g., structural magnetic resonance imaging (MRI), fMRI, diffusion tensor imaging, resting-state, brain perfusion, proton magnetic resonance spectroscopy). Two studies used peripheral markers (Guidara et al. 2021; Idemoto et al. 2021).
Four studies used a longitudinal design (Alonso-Lana et al. 2019; Estudillo-Guerra et al. 2020; Kopf et al. 2023; Nishimura et al. 2015) and twelve were cross-sectional studies (Gao et al. 2023; Guidara et al. 2021; Idemoto et al. 2021; Lai et al. 2018; Magioncalda et al. 2016; Magioncalda et al. 2015; Martino et al., 2016; Mikawa et al. 2015; Pomarol-Clotet et al. 2015; Rive et al. 2016; Velasques et al. 2013; Yang et al. 2020).
Attention
Four studies carry out a neurocognitive evaluation of attention (Magioncalda et al. 2015; Magioncalda et al. 2016; Martino et al. 2016; Velasques et al. 2013).
Three of them evaluate sustained attention with the continuous performance test (CPT; Magiocalda et al. 2016; Martino et al. 2016), where we find that BD patients showed lower number of total hits and higher number of total omission errors.
On the one hand, BDM patients showed that structural changes in the cingulum were related to the deficits found at the attentional level. Furthermore, it was found that the perigenual anterior cingulate cortex (PACC) and posterior cingulate cortex (PCC) functional connectivity was decreased in manic patients when compared to both HCs and BDD patients and the SC of the cingulum, especially its anterior part, was decreased in manic patients when compared to HC (Martino et al. 2016).
When microstructural abnormalities in the white matter (WM) were investigated, subgroups of BD patients showed different spatial patterns of WM alterations (Magioncalda et al. 2016). The BDE patients had minor and localized WM alterations in the midline structures, whereas the WM alterations were more diffuse in the BDM patients, affecting both midline and lateral structures, and there were stronger and more widespread WM alterations in BDD patients. In addition, these WM alterations were associated with attention deficits. Similarly, in another study these authors found differences in functional connectivity from the PACC to other regions in the posterior default mode network (DMN) between patients in manic or depressed episode and HCs, but no differences between the BD patient subgroups (Magioncalda et al. 2015).
In Velasques et al. (2013), BDM patients showed lower saccade latency than BDD patients or the HCs. In a prosaccadic attention task the BDM patients showed stronger gamma coherence in the frontal cortex than in the other groups (BDD and HCs).
Processing speed
Only one study evaluates processing speed (Estudillo-Guerra et al. 2020). Six months after an acute episode of mania, patients in euthymic state do not show differences in this cognitive sphere. At follow-up, a decrease in perfusion was observed in the right middle temporal gyrus (MTG) and the right superior temporal gyrus (STG).
Executive functions
Seven studies explored EFs (Estudillo-Guerra et al. 2020; Lai et al. 2018; Magioncalda et al. 2016; Magioncalda et al. 2015; Mikawa et al. 2015; Nishimura et al. 2015; Rive et al. 2016). Three of them found no differences in performance between the groups (BD in different states and HCs) in the cognitive task (Mikawa et al. 2015; Nishimura et al. 2015; Rive et al. 2016).
Estudillo-Guerra et al. 2020 explored cognitive deficits in acute BDM patients and their subsequent evaluation after 6 months (euthymic state). This study evaluates cognitive functions using the Spanish version of the Screen for Cognitive Impairment in Psychiatry Scale (SCIP-S). A subtest contains the Verbal Fluency Test (VFT) to evaluate executive functions. A negative correlation between Brodmann area (BA) 25 and positive with BA 38 and 21 was found during a manic episode. At follow up Cognitive impairment in VFT correlated with changes increased perfusion in the bilateral Anterior cingulate cortex (ACC). Fluency prompted by letter showed a correlation with PACC-SACC (Magioncalda et al., 2015).
By contrast, in another study, there was increased activation in the dorsolateral prefrontal cortex (DLPFC) of BDD patients, and in the parietal cortex (PC) compared to the BDE patients (Rive et al., 2016). However, hypoactivation of the left DLPFC and of the left ventrolateral prefrontal cortex (VLPFC) during a verbal fluency task was found in patients with hypomanic symptoms, while this activation was less prominent in the DLPFC of BDD patients (Nishimura et al. 2015). In addition, this study followed hypomanic patients who showed significantly greater concentration changes of oxygenated hemoglobin (oxy-Hb) in the left DLPFC and frontopolar prefrontal cortex (FPPFC) when experiencing hypomanic symptoms compared to when they were absent (8 patients). Similarly, the oxy-Hb levels induced by executive tasks were significantly lower in BDD than BDE patients (Mikawa et al. 2015). Finally, another study failed to find differences between the BD groups (Lai et al. 2018), showing a decrease in the N-acetylaspartate to creatine ratio (NAA/Cr) in the bilateral basal ganglia compared to the HCs. Nevertheless, the decrease in NAA/Cr ratios was negatively correlated with total errors and TMT-B uptake, but there was no correlation between the NAA/Cr and Cho/Cr in the right basal ganglia and the scores of WCST and TMT-B in acute-episode BD patients.
Memory
Working memory
Regarding working memory, four studies used an n-back paradigm (Alonso-Lana et al. 2019; Kopf et al. 2023; Pomarol-Clotet et al. 2015; Yang et al. 2020) and one used the SCIP-S subtest (Estudillo-Guerra et al. 2020). We found worse performance in the manic or depressed state compared to HC and BDE patients.
In a first study, the BDM group obtained worse results in the two versions of the task compared to the BDD patients and HC individuals (Pomarol-Clotet et al. 2015). However, when the cognitive load was increased (2-back version), the BDD patients also differed from the HCs. Surprisingly, the BDE patients did not differ from the HCs. There was reduced activation in the left and right dorsal PC and precuneus in BDM patients, and failure to de-activate the medial frontal cortex was evident in all BD groups.
In a longitudinal study, patients were assessed during a manic episode and later, in a state of euthymia after about 12 months (Alonso-Lana et al. 2019). Similar to previous findings, BDM patients performed worse than HCs and BDE patients. Activation during the cognitive task showed weaker activation in the left DLPFC, PC, and bilateral superior precuneus in BDM patients, while the BDE group continued to exhibit failure in vmPFC deactivation. During the working memory test of SCIP-S, manic episodes were associated with limited perfusion in the right OFC, whereas no significant differences were observed during euthymia (Estudillo-Guerra et al. 2020).
Finally, functional neuroimaging data was used to provide an intuitive method to study fMRI-inferred neural efficiency in the whole brain, allowing interindividual differences related to the task to be predicted (connectome; Yang et al. 2020). An overall increase of the functional connectome was detected and there was a more homogeneous distribution in BDD patients. Interestingly, the maladaptive modulation of the functional connectome was associated with worse performance in working memory.
Verbal memory
Only one study assessed verbal memory (Estudillo-Guerra et al. 2020), with immediate verbal learning correlated to the temporal polar cortex. No significant correlation of manic episodes with delayed verbal learning was detected, although a significant correlation was seen in euthymic states.
Intellectual Quotient (IQ)
Concerning the IQ, HCs had a higher mean current IQ than the BDD and BDM patients but not the BDE patients (Pomarol-Clotet et al. 2015).
When the relationship between neurotrophic factors and cognition was studied in different mood phases of BD (Idemoto et al. 2021), no differences in plasma GDNF levels were evident between the affective states. Furthermore, no correlation was performed to see if there was an association between IQ and serum GDNF levels. However, after controlling for factors such as sex, age, BMI, estimated IQ, and diagnosis, serum GDNF levels in TB patients were lower in remission and depression states than control subjects (this did not occur in patients in a manic or mixed state).
Cognitive Screening Test
Differences in the levels of oxysterols and CRP were analyzed in the distinct groups of bipolar patients, with lower cholesterol levels (Tchol, 24-OCH) reported in BDM patients relative to BDD patients and in patients with severe manic episode compared to those with moderate manic episode for 24-OCH levels (Guidara et al. 2021). By contrast, CRP levels were higher in BDM patients and in patients with severe manic episode compared to those with moderate manic episode. No correlations with the cognitive scale (MoCA) were found.
Self-reported cognitive
A study utilizes a self-report measure to assess cognitive dysfunction with the Perceived Deficits Questionnaire (PDQ) (Gao et al. 2023). Despite finding differences in activation between patients in acute state and their remission state in the follow-up (BDM patients showed reduced network homogeneity compared to BDE), no association with cognition was found.