Cognition, Cytokines, Blood–Brain Barrier, and Beyond in COVID-19: A Narrative Review

The summarized information on cognitive deficits reported in the literature during the acute phase is shown in Table 1 and described below.

Several studies agree that a significant proportion of patients hospitalized with COVID-19 experienced cognitive deficits during the acute phase of the disease, with prevalence rates ranging between 60% and 80% [37,38]. These impairments primarily affect executive functions, memory, and attention [37,38,39,40,41].

The cognitive impairment observed during the acute phase of COVID-19, according to various studies, does not manifest as a global dysfunction nor is it consistent with a typical pattern of dementia-related decline, but it primarily affects specific, mainly higher-order, cognitive functions. In this regard, Hosp et al. [40] found that although a large proportion of patients exhibited deficits in memory and executive functions, domains such as orientation and language remained relatively preserved, which rules out the possibility that the observed impairment is solely attributable to a generally weakened state or fatigue. This distribution suggests a focal pattern of impairment rather than a generalized cognitive dysfunction. Similarly, Kanberg et al. [41] demonstrated that despite the presence of cognitive symptoms, biomarkers indicative of active neurodegeneration (such as neurofilament light chain (NfL) or glial fibrillary acidic protein (GFAP)) were not observed, reinforcing the hypothesis that the cognitive deficits are not due to neurodegenerative processes but likely to transient functional or inflammatory mechanisms related to the infection.

The more severely ill patients who underwent invasive ventilation and sedation showed better cognitive performance in the subacute phase compared to those who received less aggressive respiratory support. Alemanno et al. [37] suggest that this difference could be explained by a higher oxygen supply during hospitalization or by a protective effect of sedation against the emotional and physiological stress associated with critical illness in conscious patients. In fact, younger patients, who more frequently received mechanical ventilation, also exhibited better cognitive status after the acute phase. On the other hand, Kanberg et al. [41] observed that cognitive symptoms such as memory and concentration difficulties were more common in patients with severe disease, although no biomarkers indicative of persistent neurodegenerative damage were found, suggesting that the cognitive deficits may be more related to transient inflammatory mechanisms rather than structural neurodegeneration.

In the acute phase of the disease, several studies used cognitive screening tools to quickly assess the neuropsychological status of patients, and these instruments appear to be useful for the initial detection of deficits. In particular, the Montreal Cognitive Assessment (MoCA) has shown higher sensitivity than the Mini-Mental State Examination (MMSE) in identifying subtle impairments in domains such as memory, attention, executive functions, and visuoconstructive abilities [37,38,40]. However, these authors also note that the MoCA provides limited information and does not replace a comprehensive neuropsychological evaluation. Therefore, in cases where screening results are positive, a more extensive and detailed assessment is recommended, ideally involving neuropsychology specialists, in order to establish a more accurate diagnosis and guide for treatment.

Some studies have identified specific cognitive syndromes, such as dysexecutive syndrome, present in one-third of discharged patients, characterized by disorientation, inattention, and disorganized motor responses to verbal commands [39]. This pattern, assessed using tools such as the Confusion Assessment Method for the Intensive Care Unit (CAM-ICU), highlights the complexity of the neurocognitive deficits associated with COVID-19, particularly in patients who have been admitted to intensive care units.

The specificity of the deficits observed during the acute phase of COVID-19, particularly in executive functions and memory, together with the absence of structural neurodegenerative damage [41], points to the need for establishing clinical protocols for cognitive follow-up and rehabilitation. Such an approach could contribute not only to improving the quality of life of patients but also to facilitating a more complete functional recovery after COVID-19 hospitalization.

The summarized information on cognitive deficits reported in the literature during the post-acute phase I is shown in Table 2 and described below.

Several studies have reported a high prevalence of cognitive impairment in patients who have recovered from the acute phase of COVID-19 infection, even weeks after infection (4–12 weeks). The prevalence estimates vary depending on the methodology used and the time of assessment but generally range between 58% and 81% in hospitalized samples, reflecting a significant impact on cognitive function [44,45]. In addition, a substantial number of patients, approximately 34%, report subjective complaints, mainly related to attention and naming, although these do not always align with the objective findings from neuropsychological assessments [42].

Regarding the frequency and severity of deficits, impairments have been documented ranging from mild to severe. Objectively, moderate impairment has been observed in immediate memory (38%) and severe impairment in a smaller proportion (11%). These findings were accompanied by deficits in semantic verbal fluency (34.6% moderate, 8.4% severe) and working memory (6.1% moderate, 1.1% severe) [45].

The findings suggest a domain-specific pattern of impairment, with a characteristic profile. Executive functions (such as cognitive control, response inhibition, cognitive flexibility, and set-shifting), attention (particularly divided and sustained), immediate memory, and working memory are the most frequently affected domains [42,44,47]. To a lesser extent, deficits have also been identified in verbal fluency, processing speed, and aspects of language, while recognition memory tends to remain more preserved [44]. This neuropsychological profile is consistent with a dysexecutive syndrome, similar to that observed in neuroinflammatory or early-stage neurodegenerative disorders [47]. Although neuroimaging was not routinely performed, these findings suggest the involvement of brain regions relevant to executive control processes, including the prefrontal cortex, parietal cortex, cingulate cortex, and striatum [44].

The relationship between the clinical complications associated with the course of COVID-19 and cognitive impairment has been widely documented. Patients who experienced acute neurological symptoms such as headache, anosmia, or dysgeusia were more likely to show deficits in domains such as attention, memory, and executive functions [42]. Headache was the most strongly associated symptom with poor performance on cognitive assessments. Prolonged hypoxia, systemic inflammation, intubation, and coagulopathy are clinical factors associated with higher cognitive compromise [44]. In fact, there is a linear association between the ICU length of stay and the severity of cognitive impairment, as measured by tests such as the MMSE [46].

It has also been noted that cognitive impairment can coexist with symptoms of mental fatigue, anxiety, and insomnia, which may amplify subjective complaints without necessarily reflecting objective neuropsychological deficits [42,47].

The summarized information on cognitive deficits reported in the literature during the post-acute phase II is shown in Table 3 and described below.

The persistence of cognitive deficits in the post-acute phase of COVID-19 (between approximately 13 and 24 weeks after infection) has been widely documented in various studies, with prevalence rates ranging from 38% to 69%, depending on the methodology used, the severity of the initial clinical presentation, and the timing of the post-acute evaluation [48,50]. Our review confirms that neurocognitive sequelae, such as executive dysfunction, attentional alterations, slowed processing speed, and memory problems, are among the most frequent and persistent symptoms of post-COVID syndrome within this timeframe, even in patients who were not hospitalized. Specifically, Chaumont et al. [48] objectively identified neurocognitive deficits in 68.9% of patients evaluated six months after the acute neurological episode. Similarly, Ferrando et al. [50] found that 38% of participants seeking care for post-COVID cognitive complaints had extremely low scores on neuropsychological assessments, compared to 14% in the group without post-COVID complaints. Krishnan et al. [51] also reported cognitive impairments exceeding 35% prevalence in specific domains such as executive functions and sustained attention. Complementary self-reported data indicate that 85.1% of respondents experienced brain fog and cognitive dysfunction, including problems with attention, executive functioning, problem-solving, and decision-making [49].

Regarding the progression of cognitive symptoms, the evidence suggests these tend to be most severe and prevalent in the first few months post-infection, peaking around the third month when up to 66.7% report brain fog or cognitive difficulties, followed by a slight decline but remaining present in over 50% of patients at seven months [49].

A notable observation is the partial discrepancy between the self-reported cognitive complaints and the neuropsychological test performance. Although many patients report symptoms such as brain fog, forgetfulness, or concentration difficulties, these do not always correspond to significant objective deficits. While previous studies have questioned this association, current data suggest that the perception of cognitive difficulties, even months after acute COVID-19, may be a reliable indicator of actual impairment. Ferrando et al. [50] noted that although only 38% of the group reporting complaints exhibited extremely low performance, significant differences found in neuropsychological tests suggest these individuals might be detecting subtle but real cognitive decline relative to expectations for their age and premorbid functioning. In this regard, including a measure of subjective cognitive complaints helps explore whether the perceived impairment correlates with real neuropsychological deficits. However, the absence of objective deficits does not imply that complaints are irrelevant, as they are associated with genuine functional impairment and reduced quality of life, justifying specific objective cognitive assessments and personalized clinical interventions.

Key risk factors for developing long-term cognitive deficits include the severity of symptoms during the acute COVID-19 phase, the presence of prior medical comorbidities, current depressive symptoms, the subjective perception of cognitive difficulties, and older age [50,52]. Additionally, factors associated with critical illness, such as prolonged delirium, sepsis, hypoxemia, or administration of high-dose sedatives, may significantly contribute to post-COVID cognitive impairment. Nevertheless, this deterioration appears to be mediated by multiple underlying mechanisms, including prolonged systemic inflammation, encephalopathy linked to critical disease phases, and more specific processes such as retrograde neuroinvasion via olfactory pathways, blood–brain barrier disruption, and infiltration of the central nervous system by peripheral immune cells [48]. Longitudinal studies have also shown that patients with higher depressive symptom levels perform worse on neuropsychological tests, although the causal relationship between cognitive and affective dysfunction remains unclear [50].

Finally, the functional and social impact of these cognitive deficits is considerable. Over 86% of patients with persistent symptoms report moderate to severe work limitations, the need for reduced hours or medical leave, and relapses triggered by mental exertion [49]. These difficulties occur across all ages, affecting the quality of life, autonomy in instrumental activities of daily living, and overall mental health. This underscores the importance of ensuring adequate recovery time, access to benefits, and workplace accommodations, especially for lower-income workers who face greater challenges.

The summarized information on cognitive deficits reported in the literature during the post-acute phase III is shown in Table 4 and described below.

Hospitalized COVID-19 patients have been shown to continue presenting cognitive deficits one year after discharge, with high prevalence rates ranging from 39% [57] to 46.8% [54] or 48% and 56% when compared to a matched healthy control group and demographically adjusted norms, respectively [55]. Furthermore, follow-up data indicate that up to three years after hospitalization, many still experience a substantial cognitive and psychiatric burden, likely due to both the worsening of existing symptoms and the emergence of new symptoms [58]. These findings underscore the need for prolonged follow-up and the planning of long-term rehabilitation strategies, especially in patients with persistent symptoms.

In terms of cognitive domains affected after 1 year of COVID-19 infection, the most frequent deficits are observed in attention, impaired in up to 61.9% of patients, followed by executive functions, affected in 43% of cases [56,58]. Attentional deficits not only manifest in concentration tasks but also impact the performance of other cognitive functions, such as memory and verbal fluency, suggesting that attention acts as a “central system” whose dysfunction can compromise overall performance. In this context, cognitive impairment often involves multiple domains simultaneously, constituting a characteristic multidomain impairment pattern: 60.3% of patients showed alterations in two or more cognitive domains, compared to 39.7% with impairment limited to a single domain [53]. In fact, attention deficits are present in many combinations of multidomain impairment, including alongside executive function deficits. This interdependence may be explained by shared neural networks in fronto-subcortical structures. In contrast, memory impairment appears to result from hippocampal dysfunction and occurs relatively independently of other deficits [53]. The presence of executive deficits alongside attentional impairment suggests the need for interventions that not only target specific skills but also enhance the coordination and regulation of multiple cognitive domains.

Several studies have identified neurobiological correlates that could explain post-COVID cognitive alterations. Wood et al. [59] reported reductions in the anterior cingulate cortex volume and elevated serum biomarkers of neuronal damage, such as NfL and glial fibrillary acidic protein (GFAP), which were associated with cognitive deficits. These findings suggest underlying neuronal and glial damage that may contribute to cognitive dysfunction. Additionally, it has been suggested that a biocognitive profile marked by elevated D-dimer relative to c-reactive protein at 6 months may explain part of the risk for depression, fatigue, and subjective cognitive decline at 2–3 years, indicating a potential long-lasting biological process such as cerebral microthrombi, which reflects a neurobiological correlate [58]. These biological correlates provide evidence of potential mechanisms for the persistence of deficits and support the need for interventions combining cognitive rehabilitation with strategies aimed at neurobiological recovery and neuroplasticity. However, studies such as Ruzicka et al. [56] report the absence of structural changes specifically associated with the post-COVID condition. These findings indicate that routine magnetic resonance imaging or computed tomography scans in patients with post-COVID conditions who present cognitive symptoms often fail to identify neurostructural correlates, due to the lack of clinically defined structural or functional abnormalities. Therefore, further research is needed to better understand the neurobiological underpinnings of post-COVID cognitive deficits.

Cognitive impairments are associated with reduced work capacity, self-reported cognitive difficulties in daily life, poorer quality of life, and elevated affective symptoms one year after hospital discharge [55,57]. Many people who changed occupation in the months and years following acute COVID-19 did so because they could no longer meet the cognitive demands of their job, rather than due to a lack of energy, interest, or confidence. Some authors [55,60] have suggested that returning to work after COVID-19, even with persistent symptoms, may promote recovery and help maintain cognitive function through environmental stimulation and neuroplasticity. However, the job demands should not exceed the cognitive capacity, making cognitive screening and providing support to those with cognitive impairment clinically valuable. Additionally, in a mutually reinforcing cycle, Miskowiak et al. [55] hypothesize that cognitive impairment could exacerbate depression and anxiety due to difficulties in coping with cognitive challenges in daily life and work, which in turn leads to a poorer quality of life, stress, and more affective symptoms. In turn, affective symptoms could worsen cognitive impairment and functional disability. This underscores the need to assess and address both cognitive alterations and affective symptoms after severe COVID-19. Moreover, the correlation between objectively measured and subjectively reported cognitive impairment suggests that patients have adequate awareness of their cognitive status. Therefore, cognitive screening using self-report questionnaires may be appropriate for patients attending long COVID-19 clinics, defined as individuals experiencing lingering physical, cognitive, neurological, or psychiatric symptoms persisting for 12 or more weeks after recovery from acute illness, collectively referred to as long COVID or post-COVID syndrome [55]. However, there are studies, such as that of Ruzicka et al. [56], in which cognitive performance is similar among groups reporting high, moderate, and mild cognitive complaints, suggesting that the perception of neurocognitive manifestations may be subject to confounding by several factors, including fatigue, insomnia, stress, or emotional impact. This evidence highlights the importance of a comprehensive approach that includes not only patient-reported cognitive impairment but also objective cognitive assessment and evaluation of comorbid symptoms to design personalized interventions.

Finally, the implications for rehabilitation are clear. The high prevalence and persistence of deficits in attention and executive function call for targeted programs of cognitive stimulation, concentration training, and compensation strategies for daily life. Moreover, the frequent coexistence of multiple affected domains justifies a multidimensional approach, combining cognitive rehabilitation, psychological support, and regular medical follow-up.

An integrative model of post-COVID cognitive sequelae across different phases of the disease is presented in Figure 1.

One of the first studies available on the COVID-19 pandemic reported an increased level of different types of cytokines in patients in the intensive care unit during the acute phase of infection [13]. Inflammatory IL-2, IL-7, IL-10, and TNFα were increased in this study. In addition, the chemoattractant proteins MCP1, MIP-1α, and G-SCF were also increased. Later, Chen et al. reported one of the most consistent findings in COVID-19 patients, the elevated increase in IL-6 levels during the infection [63]. Broader screening targeted studies confirmed that not only did IL-6 increase in COVID-19 patients compared to healthy controls, but there was a progressive elevation of this cytokine, CXCL10, and the chemokine GM-CSF in different degrees of COVID-19 severity linked to endothelial damage and thrombosis [61]. Thus, IL-6 and GM-CSF have a central role in COVID-19, with GM-CSF more highly dysregulated than in influenza virus infection [61]. In 2022, a study compared different phenotypic patients, comparing patients in the intensive care unit and COVID-19 patients with neurological deficits showing increased IL-10 levels, while the opposite changes were observed, marking BBB disruption [64]. Continuing with this clinical phenotypic segregation, a recent study in long-COVID-19 patients with cognitive impairment performed a broader transcriptomic screening analysis in peripheral blood mononuclear cells with different degrees of severity [65]. In the acute phase, CXCL10, IFNγ, IL-1β, and IL-8 were increased in patients with moderate severity while IL-1β and GM-CSF were also elevated in severely ill patients. In addition, a recent study reported that the levels of IL-1β and IL-6 during the acute infection were associated with verbal and visuospatial memory and are proposed as predictors of these long-term cognitive difficulties [66]. A summary of those findings is described in Table 5 and Figure 2. Thus, a specific profile of antiviral and inflammatory cytokines and chemoattractant proteins has been reported in the acute phase, with some of them related to neurological clinical phenotype and severely ill COVID-19 patients.

A few studies have also investigated the cytokine profile after the acute phase, which are summarized in Table 5 and Figure 2 as well. Surprisingly, some antiviral cytokines are elevated between week 5 and 12 post-infection including CXCL-1 and CXCL-10 [67]. IL-6 and TNFα were also mainly elevated in this phase [47,64], as well as the chemoattractant protein MCP1 [67]. In particular, IL-6, TNFα, and MCP1 were increased in patients with neurological impairments compared to those without this clinical phenotype [67]. Thus, these studies demonstrate that antiviral and inflammatory cytokines are still upregulated in the post-acute phase, with some inflammatory cytokines related to neurological symptoms.

The effects of COVID-19 disease have also been investigated beyond 12 weeks. Several studies have reported the maintenance of elevated IL-6 in COVID-19 patients and in those with neurological symptoms [68]. Other inflammatory cytokines such as IL-1β, IL-4, IL-2, IL-10, and TNFα have also been found to increase [69], with IL-6 and TNFα still increased in patients with a neurological symptom profile, and IL-1β was upregulated in patients with persistent COVID-19 symptoms [69]. Regarding antiviral cytokines and chemoattractant proteins, IFNγ was detected as elevated in one study, and GM-CSF was downregulated in this time window of 13 and 24 weeks. Therefore, in this medium-term post-acute phase, inflammatory cytokines are increased in patients with a neurological clinical phenotype.

In 2022, more studies started to address longer-term post-acute analysis of cytokines [50]. IL-6 remained elevated after 24–72 weeks of infection [50]. In these studies, increased IL-6 levels were found in patients with cognitive deficits or post-acute sequelae. Moreover, the inflammatory cytokines IL-1β and TNFα were also elevated in patients with long COVID-19 and post-acute sequelae of COVID-19, defined as persistent symptoms beyond four weeks after acute infection, including fatigue, exercise intolerance, cognitive impairment, and respiratory and gastrointestinal symptoms, which may last for months and range from mild to debilitating [65,70]. Cytokines, such as IL-4 and IL-8, were found at reduced levels at 72 weeks compared to 24 weeks [65]. The inflammatory cytokine IL-8 and the chemoattractant MCP1 were increased in patients with brain fog at this phase [65]. This study also reported an increase in IL-13 in COVID-19 patients with brain fog. Thus, inflammatory cytokines IL-6, IL-1β, IL-13, and TNFα are elevated at this longer phase and are related to cognitive deficits and/or post-acute sequelae.

In 2021, Kanberg et al. [41] identified increased levels of the glial protein GFAP and NfL in the blood of moderate COVID-19 patients in the acute phase. A later report extended this analysis to other proteins of the neurovascular unit [64]. This study found that proteins of the extracellular matrix (PPIA and MMP-9), astrocytes (S100β and GFAP), and neurons (NfL) were increased in both COVID-19 patients with neurological deficits and those in the intensive care unit (ICU), compared to healthy controls, with the levels of these proteins higher in ICU patients, with the exception of MMP-9, which showed the opposite change. In agreement with these findings, a study performed in a cohort of 500 COVID-19 patients observed higher levels of MMP-9 and also MMP-3 in COVID-19 patients with neurological symptoms compared to healthy controls [76]. In the same year, another report with a modest sample size found higher levels of GFAP and NfL in the plasma of COVID-19 patients with encephalopathy compared to neuro-COVID-19 patients, defined as those with neurological manifestations associated with COVID-19, including cognitive, sensory, and motor deficits, who never required hospitalization. The authors argued that different mechanisms are involved in COVID-19 patients with neurological post-acute sequelae at 3 weeks from the onset of symptoms [77]. Indeed, these patients also showed higher anxiety and depression levels and a neuroglial GFAP/NfL score, suggesting an astroglial activation in this particular clinical phenotype. Overall, proteins of different elements of the neurovascular unit have been reported to be elevated in blood during the active infection phase in both COVID-19 patients with neurological symptoms and with a severe degree of the disease.

Subsequent studies identified in the literature on the post-acute phase also report changes in some of these proteins of the BBB (Table 6). Increased levels of MMP-9 were found in severe and critical patients compared to moderate patients or healthy subjects, with higher levels in critical patients compared to severe patients [64,78]. GFAP levels were increased in neuro-COVID-19 and long COVID-19 patients compared to non-neuro-COVID19 or non-long-COVID-19 patients, respectively [67,79]. Moreover, higher levels of NfL were also still detected in this initial post-acute phase in patients with neuropathic pain and in severe cases, compared to recovered patients or moderate patients [64,80]. Therefore, biomarkers of BBB disruption are still present in this initial post-acute phase.

After 12 weeks post-infection, Telser et al. [79] found no changes in the NfL and GFAP in long-COVID-19 patients, while Wallensten et al. [81] reported increased levels in GFAP in extracellular vesicles and S100β that peaked at 4 months and returned to normal levels at 12 months post-infection (Table 6). Thus, these studies suggest that the leakage of BBB started to be restored.

Several follow-up studies covering a time frame from 24 to 48 weeks in COVID-19 patients also reported normalized levels of S100β, GFAP, and NfL [41,79,81] (Table 6). However, Greene et al. [65] showed increased levels of glial markers in plasma in patients with brain fog, suggesting a persistence of a non-restored BBB in post-acute patients. In addition, they found that peripheral mononuclear cells of those patients showed increased adhesion to human brain endothelial cells in vitro, and the exposure of endothelial cells to serum with long-COVID-19 activated the expression of inflammatory biomarkers. This study also found that TGFβ correlates with the degree of BBB disruption in brain fog patients. Thus, BBB disruption is still altered between 24 and 72 weeks post-infection, restricted to a particular clinical phenotype linked with cognitive impairment and related to systemic inflammation.