Neurodegeneration and vascular pathology are the two most common causes of cognitive impairment in the elderly. Post-stroke cognitive impairment (PSCI) is a special condition of vascular cognitive impairment (VCI) and refers to the presence of cognitive impairment after stroke. PSCI is noted in about one-third of stroke survivors, which usually manifest 3–6 months after stroke occurrence (Pendlebury and Rothwell, 2009; Gottesman and Hillis, 2010). The occurrence of PSCI can be attributed to vascular injury, pre-existing neurodegenerative substrates, and a combination of both (Mijajlović et al., 2017). Besides, stroke patients with co-existing beta-amyloid (Aβ) plaque tend to present steeper cognitive declines in longitudinal follow-up (Liu et al., 2015).

Regarding the in vivo detection of neurodegenerative pathology in stroke patients, positron emission tomography (PET) imaging, cerebrospinal fluid (CSF) and blood examination have been developed in the past decades. PET imaging can yield an excellent visual resolution of the topographical distribution of Aβ pathology and tau protein, but it is expensive with limited availability and requires multiple scanning sessions if different pathological substrates are to be detected. The concentrations of Aβ protein and tau protein in CSF have been shown highly correlated with brain pathological changes and clinical presentations, but the invasive nature of lumbar puncture limits its clinical application. Compared to PET and CSF studies, blood biomarkers of neurodegeneration have the advantage of more convenient accessibility with no risk for radiation exposure and invasive procedures (Blennow, 2017). Detection of plasma neurodegeneration markers, such as Aβ40, Aβ42, and tau protein, has been shown feasible and reliable by the immunomagnetic reduction (IMR) technique (Yang et al., 2011). Patients with Alzheimer’s disease (AD) had lower plasma Aβ40 and higher plasma Aβ42 and tau protein levels than patients with mild cognitive impairment (MCI) and healthy controls (Chiu et al., 2013, 2014). Regarding the reliability of plasma neurodegenerative biomarkers, the CSF and plasma Aβ42 were shown correlated in a recent AD study (Teunissen et al., 2018). Moreover, plasma Aβ40 and Aβ42 levels were significantly correlated with amyloid accumulation on Pittsburgh compound B (PiB) PET (Tzen et al., 2014).

Detection of fluid biomarkers of amyloid plaque and tau protein has been applied in stroke patients, and their levels would vary according to sampling time point, vascular lesion characteristics, and coexisting neurodegeneration states (Hesse et al., 2000; Zhang et al., 2010; Bielewicz et al., 2011; Skillback et al., 2015). Previous studies have shown that there is no significant change in amyloid plaque accumulation after acute stroke (Hesse et al., 2000; Sahathevan et al., 2016). On the other hand, tau protein level has an abrupt increase within 5–10 days after acute stroke, and then gradually decreases to a stable level in the following 3 months after stroke (Hesse et al., 2000; Kaerst et al., 2013). Therefore, plasma amyloid peptide and tau protein levels measured 3 months after stroke would be within a relatively stable condition and may represent the overall neurodegenerative condition in stroke patients.

Although plasma Aβ42 and tau protein levels have been recently reported to have significant associations with VCI presentations (Tang et al., 2018; Chi et al., 2019), there is limited literature on the relationship between the imaging and blood biomarkers of neurodegeneration in the context of PSCI. In this study, we aimed to explore: (1) the associations of plasma Aβ40, Aβ42, and tau protein levels with cognitive presentations around 3 months after first-ever ischemic stroke; and (2) the relationships of plasma Aβ40, Aβ42, and tau protein with the relevant imaging markers, such as cortical atrophy on MRI and amyloid burden on PET imaging.

The study recruited 24 patients with first-ever ischemic stroke and 13 age- and education-matched normal controls. Stroke patients had higher proportions of hypertension and dyslipidemia than those of the controls. There were no differences between the two groups in mean cortical thickness, plasma levels of Aβ40, Aβ42, tau, or the composite scores for Aβ42/Aβ40, tau/Aβ40, tau*Aβ42, and tau*Aβ42/Aβ40 (Table 1). The median days of blood sampling, neurocognition assessment, and neuroimaging evaluation were 96 (86–105, interquartile range), 98 (88–110), and 100 (87–118) days after stroke occurrence for stroke patients, and these intervals were highly correlated (r = 0.48–0.68, p values < 0.03). These evaluations were performed within 15 days for the controls.

Age was negatively correlated with mean cortical thickness in all participants (r = −0.35, p value = 0.04), and plasma tau protein level and tau-related composite scores were correlated with mean cortical thickness after age adjustment. Subgroup analyses were then performed to evaluate associations between plasma biomarkers and mean cortical thickness in stroke patients and normal controls, respectively. In stroke patients, plasma tau, tau*Aβ42, and tau/Aβ42 levels were correlated with mean cortical thickness, but not with stroke volume (Figure 1). However, such correlations were not observed in normal controls (Table 2).

In all participants, plasma levels of Aβ40, Aβ42/Aβ40, tau/Aβ40, and tau*Aβ42/Aβ40 were all significantly correlated with the GDS and DRS-2 total scores, respectively, and remained significant after controlling for age and education (Table 3). Similarly, in stroke patients alone, the plasma Aβ42/Aβ40, tau, tau/Aβ42, tau/Aβ40, and tau*Aβ42/Aβ40 levels were significantly correlated with the GDS score, DRS-2 total score, and DRS-2 I/P subtest score after age and education adjustment (Figure 2). The correlations of stroke volume with the NIHSS, MMSE, GDS and DRS-2 total scores were not significant.

An ¹⁸F-florbetapir PET scanning was done in 19 of 24 stroke patients and 11 of 13 controls. There were no differences in age, education, or cognitive test scores between participants with and without ¹⁸F-florbetapir PET scanning. The ¹⁸F-florbetapir PET imaging was visually negative for amyloid plaque in all participants. The ¹⁸F-florbetapir SUVR was positively and moderately correlated with Aβ42/Aβ40 (r = 0.42, p value = 0.07) in stroke patients (Figure 3), but was not significantly correlated with any cognitive test scores in either group.

PSCI is not a single disease entity; rather, it describes an unspecified cognitive decline that usually occurs 3–6 months after stroke onset (Tuladhar and de Leeuw, 2010). In this study, we checked plasma Aβ42, Aβ40, and tau protein levels as well as structural MRI and amyloid PET scanning around 3 months after the occurrence of first-ever ischemic stroke. We found that plasma tau protein level and tau-related composite scores were correlated with mean cortical thickness. In addition, the Aβ42/Aβ40 ratio was moderately correlated with ¹⁸F-florbetapir PET SUVR. Furthermore, plasma Aβ42, Aβ40, tau protein and their composite scores were correlated with cognitive performance 3 months after stroke. All these findings suggest plasma biomarkers of Aβ42, Aβ40, and tau protein may be associated with the development of PSCI.

Amyloid plaque is a pathognomonic marker of Alzheimer’s disease and could be present for decades before the manifestations of cognitive impairment (Jack et al., 2013). Although vascular lesions and Aβ pathology frequently coexist in stroke patients, previous CSF or amyloid PET studies have shown that stroke itself would not induce amyloid plaque accumulation (Hesse et al., 2000; Sahathevan et al., 2016). Nonetheless, the presence of co-existing amyloid plaque has been implicated as a risk factor for PSCI development (Thiel et al., 2014; Skillback et al., 2015). Therefore, detection of both vascular injury and neurodegeneration pathology would be helpful to disentangle the complex etiologies of PSCI. Our study found that plasma Aβ proteins were correlated with PSCI performance. Of note, the Aβ42/Aβ40 ratio was more sensitive to PSCI than Aβ42 or Aβ40 alone in our study. It has been proposed that the Aβ42/Aβ40 ratio could compensate for general inter-individual variations than Aβ42 or Aβ40 alone (Hansson et al., 2019), and the Aβ42/Aβ40 ratio in either CSF and blood studies had adequate accuracy and reliability to differentiate AD patients from controls (Chiu et al., 2012; Hansson et al., 2019).

The correlations of VCI with plasma Aβ42, Aβ40, and tau protein have been recently reported (Tang et al., 2018; Chi et al., 2019). Tang et al. (2018) found elevated plasma Aβ42 was associated with worse cognitive performance in stroke patients. However, the intervals from stroke occurrence to cognitive evaluation ranged from acute stroke stage to more than 10 years after stroke in their study, and disparity in evaluation intervals may limit the ability to elucidate the temporal relationship between vascular and neurodegenerative contributions to the PSCI development. In our study, the plasma biomarker collection, cognitive evaluation and neuroimaging studies were performed around 3 months after stroke, and we found the plasma Aβ42/Aβ40 ratio and tau-related composite scores were correlated with PSCI presentations. This finding implies the potential role of these markers in signifying the development of PSCI. On the other hand, we found that stroke volume was not correlated with PSCI severity in this study. This could be partly related to the relatively minor stroke severity in our cases, whose NIHSS scores were 2.6 ± 1.4 points at stroke onset and 1.7 ± 1.1 points 3 months after stroke, respectively.

Both plasma and CSF Aβ42/Aβ40 ratios have been reported as surrogate biomarkers of cortical amyloid plaque deposition on PET (Fandos et al., 2017; Alcolea et al., 2019). Although the ¹⁸F-florbetapir PET results were visually rated negative for Aβ pathology in our stroke patients, the plasma Aβ42/Aβ40 ratio had a moderate correlation with global ¹⁸F-florbetapir SUVR. In agreement with the previous studies, our results showed that the plasma Aβ42/Aβ40 ratio was correlated with cognitive performance as well as the ¹⁸F-florbetapir SUVR even under the condition of low Aβ burden on PET imaging.

Tau is a microtubule-associated protein involved in stabilizing the axonal cytoskeleton and is deemed as a potential marker of axonal injury (Seco et al., 2012). Previous CSF and serum studies have shown that tau protein level has an abrupt elevation in the acute ischemic stroke stage, with a peak increase within 5–10 days after stroke onset, followed by a gradual normalization after 3 months. Therefore, the interval from stroke onset to blood collection may have an influence on the plasma tau protein level. In our study, plasma tau protein was collected around 3 months after stroke, and intervals from stroke onset to plasma collection were not correlated with plasma tau level or its composite scores (data not shown). Furthermore, the tau protein level in the acute stage is correlated with stroke volume, implying the role in direct neuronal injury severity (Hesse et al., 2000; Bitsch et al., 2002; Bielewicz et al., 2011; Kaerst et al., 2013). However, the relationship between acute tau protein level and post-stroke cognitive performance has not been well investigated. In a recent study, a single measurement of plasma tau protein which was done within 1 week after stroke onset has not revealed an association with cognitive performance 3 months or 1 year after stroke. Since a steep elevation of tau protein level may occur within days after stroke onset, multiple and intensive sampling of tau protein levels in the acute stroke stage should be more reliable when investigating its influence on long-term neurological and cognitive deficits in future studies.

On the other hand, since the plasma tau protein was detected around 3 months after stroke in our study, its concentration might be more related to either pre-existing or stroke-related neurodegeneration, rather than the direct neuronal injury effect (Chen and Jiang, 2019). Furthermore, we found plasma tau protein level and its composite scores were associated with mean cortical thickness and cognitive performance in stroke patients. Similar findings were also noted in previous studies. Tau protein level is associated with brain atrophy severity in both stroke patients and normal controls (Ihle-Hansen et al., 2017; Harrison et al., 2021). However, stroke patients have a greater brain atrophy rate than normal controls (Brodtmann et al., 2020), and the correlation between CSF tau protein level and brain atrophy severity is still significant 1 year after stroke, suggesting stroke may enhance or trigger tau-linked neurodegeneration with loss of neurons (Ihle-Hansen et al., 2017).

Various combinations of plasma Aβ42, Aβ40, and tau levels have shown better correlations with cognition than single biomarkers in patients with either AD or VCI (Lue et al., 2017; Chen et al., 2019; Chi et al., 2019). During the neurodegenerative process, Aβ tends to elevate during the MCI stage and reaches a plateau towards the demented stage, while tau protein increases as AD progresses. Therefore, the combination of these biomarkers could be synergistically representative of the neurodegeneration profile. For example, the product of plasma Aβ42 and tau has better accuracy for clinically diagnosed AD than either biomarker alone (Lue et al., 2017; Jiao et al., 2020). Moreover, the plasma tau/Aβ42 ratio showed a stronger association with brain tau accumulation than tau alone (Park et al., 2019). Indeed, the characteristic of an inverse relationship between plasma tau and Aβ40 during the neurodegeneration process (Fan et al., 2018) was also epitomized in our study that the tau/Aβ40 ratio had the strongest correlation with cognitive performance among the other plasma biomarkers and their composite scores.

PSCI development is subject to multiple factors, and the underlying mechanisms are not fully understood. In addition to AD-specific biomarkers, there is a considerable number of fluid and imaging biomarkers associated with vascular cognitive impairment. Previous studies have shown elevated CSF/blood albumin ratio, CSF matrix metalloproteinase (MMP) level, CSF neurofilament, and blood inflammatory cytokines and adhesion molecules are associated with worse cognitive performance, which could be attributed to disruption of blood-CSF/brain barriers and breakdown of white matter fibers and extracellular matrix (Wallin et al., 2017). Furthermore, neuroimaging measures, such as total gray matter volume, leukoaraiosis severity, stroke location, CSF volume, and neuroinflammatory presentations, have been implicated in PSCI development (Thiel et al., 2014; Molad et al., 2019; Huang et al., 2020). Therefore, multi-modality studies are required to investigate the relationships among fluid and imaging biomarkers and their composite influence on PSCI prognosis.

There were several limitations of this study that should be considered. First, the sample size was relatively small, and it should be cautious to generalize the study results. A larger sample size is required to validate the influence of plasma Aβ and tau protein on the PSCI development. Second, additional studies may be required to investigate the interactions among stroke location, stroke severity, and plasma AD biomarkers on the PSCI presentations. Since CSF biomarker analysis has been a more direct measure of the CNS condition than blood analysis, the correlations between CSF and blood measurement of neurodegeneration biomarkers, especially for tau protein, in stroke patients may be needed in future studies. Finally, this was a cross-sectional study, and further longitudinal study is deemed necessary to determine the long-term influence of plasma AD biomarkers on PSCI development.

In this study, we found that plasma Aβ42/Aβ40 ratio was correlated with amyloid cortical deposition on ¹⁸F-florbetapir SUVR and the total tau protein value was correlated with mean cortical thickness 3 months after stroke. The plasma Aβ42/Aβ40 ratio, tau protein, and tau-related composite scores were correlated with cognitive performance. The relationship of plasma Aβ40, Aβ42, and tau protein with the long-term post-stroke structural and cognitive changes requires further studies in larger populations.

The datasets presented in this article are not readily available because the used consent does not allow for the public sharing of the data. Requests to access the datasets should be directed to thlee@adm.cgmh.org.tw.

The studies involving human participants were reviewed and approved by The institutional review board at Linkou Chang Gung Memorial Hospital. The patients/participants provided their written informed consent to participate in this study.

K-LH wrote the initial draft, performed neurologic examinations, took part in the data collection and analysis, and scientific interpretation of data. K-JL conceptualized the study design, carried out PET imaging analysis, and wrote a portion of the draft. S-YY conceptualized the study design, took part in scientific interpretation of data, and critical review of the manuscript. T-YC, Y-JC, H-CW, and C-HL performed the data collection, neurologic examination, and scientific interpretation of data. I-TH and Y-MW performed the MRI analysis, took part in data interpretation, and critical review of the manuscript. M-YH conceptualized the study design, wrote a portion of the draft, conducted the cognitive evaluation, took part in data analysis, and critical review of the manuscript. T-HL conceptualized the study design, took part in critical review of the manuscript, and edited the manuscript for content. All authors contributed to the article and approved the submitted version.

S-YY is an employee and a shareholder of MagQu Co., Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.