Cognition decline after stroke is not rare. The clinical course of post-stroke cognitive impairment (PSCI) is not a unitary syndrome but varies from individual to individual (Patel et al., 2002). Cognitive impairment assessment is often performed at 3–6 months after acute stroke to provide sufficient time for delirium resolution and neurological stability. Notably, stroke patients free of early-onset PSCI (3–6 months after stroke) are still at risk of delayed-onset PSCI (>6 months after stroke), which suggests an underlying pathological process beyond 3 months after stroke (Wagle et al., 2011). Many risk factors for PSCI have been proposed based on observational studies, including age, sex, educational attainment, stroke severity, stroke histories, and cardiovascular risk factors—particularly diabetes mellitus and hypertension (Sun et al., 2014). Brain image risk factors include white matter hyperintensities and gray matter and hippocampal volumes loss (Dichgans and Leys, 2017; Mok et al., 2017; Ding et al., 2019). However, there is still a lack of precise blood biomarkers for predicting PSCI.

Accumulation of amyloid-beta (Aβ) peptides and phosphorylated tau (p-tau) in the brain are both key pathological features of the Alzheimer’s disease (Glenner and Wong, 1984; Grundke-Iqbal et al., 1986; Jack et al., 2018; Drummond et al., 2020). Several previous studies implicated plasma Aβ40 and Aβ42 levels as associated with cognitive decline among older adults and stroke patients (Tang et al., 2018; Giudici et al., 2020). Our previous work also showed plasma Aβ42 and tau levels at 3 months were lower in the patients with PSCI at 1 year than in those without PSCI (Chi et al., 2019).

Some cohorts found that plasma tau phosphorylated at threonine 181 (p-tau181) was associated with cognitive decline (Karikari et al., 2020). In addition, brain-derived neurotrophic factor (BDNF) is an important neurotrophin in the adult brain, which can help the brain to repair (Liu et al., 2020). A previous study further found that serum BDNF levels are decreased in the acute phase of stroke, and lower circulating concentrations of BDNF protein are associated with poor long-term functional outcomes (Stanne et al., 2016).

Since these key peptides and proteins play important roles in cognitive performance, and there are few studies to examine the plasma levels of Aβ42, tau, BDNF, and p-tau181 in patients after stroke through longitudinal follow-up, the purpose of this study was to investigate whether these blood with imaging markers can improve the prediction for PSCI. In addition, the association between blood biomarkers with the trajectories of PSCI was also examined.

All 173 participants completed initial clinical and neuropsychological examinations as well as plasma biomarker assays and a brain MRI scan within 7 days after stroke. Each individual was given a standardized treatment scheme during hospitalization. No patient died during the follow-up time of 1 year. Attrition was due to loss to follow-up in 37 patients at 3 months. No significant difference between the baseline profiles of participants involved in this study and those lost in follow-up was found. Data from 136 patients were processed in the final analysis. At 3 months after stroke, 40 patients met the criteria for PSCI according to CDR-SB > 0, whereas at 12 months after stroke, 50 patients had PSCI (Figure 1).

This study investigated 136 patients with acute ischemic stroke and followed them up for consequential cognitive function trajectory for 1 year. The results demonstrated that a high plasma p-tau181 level in the acute post-stroke stage is related to a low PSCI risk at 3 (OR = 0.62, 95% CI = 0.40–0.94, p = 0.0243) and 12 (OR = 0.69, 95% CI = 0.47–0.99, p = 0.0443) months. Integrating p-tau181 to the model containing conventional risk factors (age, education, hypertension, diabetes mellitus, and NIHSS score at ≤7 days) and image biomarker significantly improves prediction.

Cognitive improvement after stroke often occurs within 6 months, and early recognition of PSCI is necessary for assessing the need for rehabilitation (Turunen et al., 2018). In the case of cerebral ischemia, hyperphosphorylated tau protein may play a protective role by promoting β-catenin and other proteins to inhibit cell apoptosis, suggesting that neurons survive apoptotic attacks and achieve self-repair (Wang et al., 2010). Phosphorylation of tau results in reduced binding to microtubules and potentially enhances plasticity, as observed during brain development, which is also important in neural repair after stroke (Alonso et al., 2001, 2006; Wang et al., 2007; Morris et al., 2011). In Alzheimer’s disease (AD), plasma p-tau181 is a prognostic and confirmatory biomarker (Bateman et al., 2020). Hyperphosphorylated tau protein forms paired helical filaments and progressively aggregates to form the main component of neurofibrillary tangles in AD pathology (Pevalova et al., 2006). However, unlike in AD pathology, neurofibrillary tangles are rarely seen in the brain after minor stroke. Exceptional case series of NFTs in the ipsilateral basal nucleus of Meynert (BNM) associated with a massive cerebral infarct in the MCA territory or a putaminal hemorrhage, speculated to form within the 5–10 years after stroke onset (Hatsuta et al., 2019). Studies of cerebrospinal fluid (CSF) samples from patients with acute stroke showed that CSF p-tau did not increase, whereas CSF total tau increased and returned to normal levels at 3–5 months after stroke, suggesting different pathogenic processes from AD (Hesse et al., 2001; Hjalmarsson et al., 2014; Hagberg et al., 2020). Therefore, the diagnostic criteria for vascular cognitive disorders from the International Society of Vascular Behavioral and Cognitive Disorders statement suggest excluding CSF p-tau when diagnosing vascular cognitive impairment in research (Sachdev et al., 2014), whereas a persistent increase in CSF p-tau is considered a defining biomarker of AD in the National Institute on Aging—Alzheimer’s Association Research Framework (Albert et al., 2011). In addition, plasma p-tau181 predicted PSCI in a dose-dependent manner in our longitudinal cohort study, and the p-tau181 level was the highest in the persistent non-PSCI group and the lowest in the persistent PSCI group. A protective role for tau phosphorylation at threonine 181 could facilitate its binding to exosomes and the release of excess tau (Avila et al., 2012). Further studies are warranted to understand the mechanism underlying this finding.

Previous studies have revealed that plasma Aβ42/40 ratios are surrogate biomarkers of cortical Aβ deposition (Fandos et al., 2017). High-plasma concentrations of Aβ40, especially when combined with low concentrations of Aβ42, indicate an increased dementia risk (van Oijen et al., 2006). A previous publication by our group using the same cohort but few participants showed Aβ42 and tau levels at 3 months were lower in the patients with PSCI at 1 year than in those without PSCI, which may reveal AD pathology one mechanism of PSCI development after 3 months of stroke and decreased levels of plasma tau could be explained by its association with the decreased plasma Aβ42 levels (Chi et al., 2019). In this study, Aβ42, Aβ40, and Aβ42/40 ratios were not different between patients with and without PSCI, which suggests the involvement of additional processes other than amyloid pathology. In addition, BDNF was found to be an indicator of long-term functional outcomes after ischemic stroke, although the additional predictive value of BNDF was modest according to clinical data (Stanne et al., 2016). Our data showed that there was no significant difference in circulating BDNF levels regardless of whether the patient had PSCI or not, which may be due to its limited impact on cognitive outcomes. In this study, a low-education level, hypertension, and pre-existing periventricular white matter disease were also the major risk factors for PSCI at 3 months. This finding is consistent with those of previous studies (Zhou et al., 2005; Sun et al., 2014; Ding et al., 2019).

Clinical Dementia Rating global score, Sum of Boxes was adopted to define PSCI in our study. Given the diverse clinical presentation of PSCI, not only memory but also other cognitive domains should be evaluated (Skrobot et al., 2018). The global CDR is weighted more on memory dysfunction, whereas CDR-SB is weighted equally for all domains (Wyman-Chick and Scott, 2015). In addition, CDR-SB has been considered an effective and reliable assessment method, which combines two sets of questions, one set is for the insider and the other is for the subject. This implies the use of CDR-SB to define the PSCI has a clinically significant impact on the follow-up after stroke. In our study, none of the patients had dementia before the stroke. However, the prevalence of PSCI at 3 and 12 months after stroke was 29.4 and 36.8%, respectively, during the longitudinal follow-up. The reported prevalence varied among previous studies depending on divergent estimates of PSCI according to the population under study and the methods of defining PSCI, suggesting a need for diagnosis criteria consensus (del Ser et al., 2005; Abzhandadze et al., 2019).

Notably, in our cohort, 27 patients without PSCI at 3 months developed PSCI at 12 months after stroke (late PSCI: 19.9%), and 17 patients reversed from PSCI at 3 months to non-PSCI at 12 months (reversal: 42.5%). Our findings further revealed that a dose-dependent trend of plasma p-tau181 level existed among patients with various persistent cognitive impairment statuses, with the highest level in the non-PSCI group followed by the early PSCI with a reversal and delayed-onset PSCI groups; the lowest level was in the persistent PSCI group, which implicated the protective effect on longitudinal cognitive function after a minor ischemic stroke.

This study has several limitations. First, there was no formal test conducted for measuring baseline cognitive function before stroke in our participants. However, patients with known cognitive impairment or neurodegeneration that impaired daily activities before stroke were excluded based on their medical histories at the screening phase. In addition, this study adopted the CDR-SB to define PSCI in order to distinguish functional changes after stroke since we relied on informant-based evidence rather than performance-based tests. Second, the severity of stroke in the patients in this study is relatively small and mainly involves small vessel diseases which might influence the generalizability of the findings to large brain infarction or intracranial hemorrhage. Future studies will be needed to elucidate the natural course of other types of brain insults. Third, in our study, no further AD diagnosis was performed; however, previous findings indicated that after ischemia with reperfusion in the brain, secondary neurodegeneration of AD type may occur (Pluta, 2000). Further longitudinal studies regarding biomarkers are needed to distinguish between PSCI and AD processes.

In conclusion, PSCI is not uncommon in the population with minor stroke. Using plasma p-tau181 as a surrogate biomarker for predicting early- and delayed-onset PSCI is helpful in interventional studies and clinical follow-up.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding authors.

The studies involving human participants were reviewed and approved by the Institutional Review Board of Taipei Medical University. The patients/participants provided their written informed consent to participate in this study.

L-KH wrote the manuscript with support from Y-CH. Y-CH, S-PC, and L-NC verified the analytical methods. C-JH and H-YC conceived the study and were in charge of overall direction and planning. Y-CL, S-PC, and L-NC aided in interpreting the results and worked on the manuscript. All authors discussed the results and contributed to the final manuscript.

The 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.