Using frequency domain analysis, spontaneous EEG signals can be decomposed into several frequency bands: delta (1–4 Hz), theta (4–8 Hz), alpha (8–12 Hz), beta (12–30 Hz), and gamma (30–80 Hz) (Babiloni et al., 2020). The power of spontaneous neural oscillation in different frequency bands has served as the potential neural biomarker correlated with the injury and recovery from stroke. Cross-sectional studies found that the ipsilesional hemisphere is associated with an increase in low-frequency power in delta and theta bands, and a decrease in higher-frequency power in alpha and beta bands, compared with the contralesional hemisphere (Molnár et al., 2006). The alteration of cortical rhythmic activity after stroke is also functionally relevant. For example, the increase of delta power over the bilateral sensorimotor cortex (SMC) was associated with more severe neurological deficits (measured by the NIHSS), while a higher level of beta power over the ipsilesional SMC was related to a lower degree of neurological deficits (Wu et al., 2016). The redistribution of the power spectrum between lower and higher frequency oscillations can also be quantified by using ratio indices, such as delta/alpha ratio (Fanciullacci et al., 2017), theta/beta ratio (Molnár et al., 2006), and delta + theta/alpha + beta ratio (Sheorajpanday et al., 2011). The increase in lower-frequency waves and the decrease in higher-frequency waves also contributed to hemispheric asymmetry in stroke patients, using the brain symmetry index (BSI; van Putten and Tavy, 2004). The ratio index and/or the BSI have also been found to be functionally relevant cross-sectionally, as assessed by the NIHSS (van Putten and Tavy, 2004; Sheorajpanday et al., 2009; Fanciullacci et al., 2017) and the FMA upper extremity scores (FMA-UE; Saes et al., 2020).

Longitudinal studies have further confirmed that some of the cross-sectional markers are indicative of individual recovery potential after acute stroke. The BSI in theta band measured within 3 weeks after stroke was predictive for upper extremity recovery outcomes (assessed by the FMA-UE) at 26 weeks poststroke (Saes et al., 2021). In another study, the delta/alpha ratio over the ipsilesional hemisphere, as well as the BSI in delta band measured within 3 weeks after stroke, showed a positive correlation with the level of neurological deficits (assessed by the NIHSS) at 26 weeks poststroke (Saes et al., 2020).

In summary, an asymmetric hemispheric rhythmic activity can be observed with an increase in lower-frequency activities and a decrease in higher-frequency activities over the ipsilesional hemisphere rather than over the contralesional hemisphere. The redistribution in the power spectrum of spontaneous EEG observed at the acute stroke phase is also an indication of recovery three to 6 months after stroke (Saes et al., 2020; Saes et al., 2021). An increase in lower-frequency activities (e.g., delta and theta bands) and a decrease in higher-frequency activities (e.g., alpha and beta bands) after acute stroke indicate an unfavorable recovery outcome three to 6 months after stroke.

EEG demonstrates excellent temporal resolution at millisecond level; however, the frequency domain analysis of EEG alone cannot utilize any temporal information as the continuous data are merged across the time domain (Gasser and Molinari, 1996). Event-related potential (ERP) is a time- and phase-locked evoked response to a variety of stimuli or events, thus it gives more context-specific information about how the brain processes afferent signals and makes behavioral responses (Müller-Putz, 2020). Therefore, ERP has the potential to be used as a biomarker in the prediction of poststroke functional recovery.

Somatosensory evoked potential (SSEP) is a form of EEG evoked potential elicited by somatosensory stimuli, such as electrical stimulation to a peripheral nerve. SSEP is believed to be associated with the activation of the somatosensory cortex along with somatosensory ascending volley and is therefore a proxy for functional integrity of the somatosensory pathway (Guerreiro and Ehrenberg, 1982). SSEP has demonstrated prognostic values of functional recovery after stroke. Stroke patients with or without SSEP response at baseline showed a significant difference in the amount of recovery of the hemiparetic arm at 6 and 12 months poststroke (Feys et al., 2000), and the pathological change in the morphology of SSEP was associated with a higher level of disability (mRS scores ≥5) at 6 months poststroke (Su et al., 2010).

Time-frequency analysis decomposes EEG signals into both time and frequency domains, therefore reserving the temporal evolution of neural dynamics across different frequency bands (Pfurtscheller, 2001). Event-related desynchronization (ERD) is typically elicited by neuronal activation and can therefore be regarded as a marker of cortical activation and excitability, while event-related synchronization (ERS) can be interpreted as a deactivated cortical region (Pfurtscheller, 2001). Movement-related ERD/ERS has been investigated in stroke research. An ERD in both mu and beta bands can be observed during movement preparation and execution, while an ERS in beta band can be found after the termination of movement (Pfurtscheller and Lopes Da Silva, 1999). A reduction of movement-related beta ERD over the ipsilesional sensorimotor cortex (SMC) during hemiparetic hand movement was found in acute (Tangwiriyasakul et al., 2014), subacute (Bartur et al., 2019; Tang et al., 2020), and chronic stroke patients (Rossiter et al., 2014). In addition, the level of movement-related ERD was associated with the level of upper limb motor impairment (Rossiter et al., 2014; Bartur et al., 2019; Tang et al., 2020), functional activity (Tang et al., 2020), and the lesion volume of structural damage caused by stroke (Bartur et al., 2019). Although the findings were based on cross-sectional experiments, they also indicate the potential predictive value of ERD/ERS.

Tangwiriyasakul et al. (2014) investigated the longitudinal change of ERD patterns in the bilateral hemisphere accompanying stroke recovery. An increase in movement-related ERD over the ipsilesional hemisphere and a decrease in movement-related ERD over the contralesional hemisphere were observed across time from one to 4 weeks after stroke onset (Tangwiriyasakul et al., 2014). In another longitudinal study, postmovement beta ERS (16–30 Hz) peak amplitude over the ipsilesional SMC at the acute phase, predicted the upper limb recovery outcomes in FMA-UE 12 weeks after stroke (Tang et al., 2020). ERD/ERS can also be induced by observation of movement. Zhang et al. (2022) found that mirror visual feedback-induced high beta ERD (21–30 Hz) over the ipsilesional SMC was associated with motor training-induced improvement assessed by FMA-UE and ARAT in chronic stroke patients (Zhang et al., 2022). However, ERD/ERS induced by movement observation has not been implicated in the prediction of functional recovery after acute stroke, and awaits further investigation (Zhang et al., 2022).

To sum up, the findings from time-frequency EEG analyses suggest a predictive role of sensorimotor beta oscillation in poststroke functional recovery, specifically in the motor domain. However, in most studies, ERD/ERS is triggered with the voluntary movement of the hemiparetic hand. The potential predictive role of ERD/ERS induced by other forms of stimuli or events needs to be further investigated in stroke patients.

The connectivity measured by EEG refers to the relationship between two or more scalp EEG sensors (Guevara and Corsi-Cabrera, 1996). There are plenty of EEG connectivity parameters that have been used in clinical studies, among which, coherence, which measures the similarity between two signals in the frequency domain, has been applied in stroke research with prognostic values (Cassidy et al., 2020). The EEG connectivity provides unique insights into the dynamic intra-/interhemispheric interactions; therefore, it can further facilitate the conceptualization of recovery models poststroke from a brain network perspective.

According to two cross-sectional studies, high beta (20–30 Hz) coherence between the ipsilesional M1 and the premotor cortex positively correlated with the arm motor recovery as measured by the FMA-UE (Wu et al., 2015), while delta (1–3 Hz) coherence between the ipsilesional and contralesional M1 inversely correlated with the arm function as measured by the FMA-UE (Cassidy et al., 2020). Longitudinally, a decrease in delta coherence between the bilateral M1 over the first 3 months poststroke correlated with arm motor recovery in the FMA-UE (Cassidy et al., 2020). In another study with chronic stroke patients, a greater degree of high beta coherence between the M1 and premotor cortex over the ipsilesional hemisphere was positively correlated with motor gains in the FMA-UE after 1 month of intensive motor training, while a greater degree of high beta coherence between the M1 and parietal cortex over the ipsilesional hemisphere negatively correlated with motor gains in response to training (Wu et al., 2015). Taken together, the results indicate that the prediction of stroke recovery using EEG connectivity outcomes is likely influenced by frequency bands, brain regions, and time after stroke onset.

Transcranial magnetic stimulation (TMS) studies revealed the potential values of using MEP for functional prognosis. MEP is a marker to measure the functional integrity and excitability of the corticospinal tract (CST; Hallett, 2007). According to a seminal study, the presence of MEP evoked by the ipsilesional M1 stimulation in the acute stroke stage was associated with good recovery at 6 months after stroke, defined as a BI ≥ 60 and/or strength index of the paretic limb muscles in the medical research council (MRC) scale ≥4, while the absence of MEP evoked by the ipsilesional M1 stimulation was related to poor recovery (measured by the BI or the MRC scale), with sensitivity from 77.1 to 87.1% and specificity from 81.0% to 85.7% (Escudero et al., 1998). MEP studies mainly focused on upper extremity motor recovery after stroke and their results suggest that the presence of MEP evoked by the ipsilesional M1 stimulation at the acute phase indicate a sign of good recovery (Escudero et al., 1998); however, the absence of the MEP evoked by the ipsilesional M1 stimulation at the acute phase does not imply poor recovery (Powell et al., 2019).

Paired-pulse TMS paradigms have been used to investigate the intracortical or interhemispheric facilitation/inhibition after stroke. In the paired-pulse paradigms, two consecutive pulses, i.e., a conditioning stimulus (CS) and a testing stimulus (TS), are delivered to one or over two different cortical regions. Depending on the intensity of the CS as well as the interval between the CS and TS, paired-pulse TMS can result in either facilitatory or inhibitory influence on the induced MEP. In a longitudinal study by Manganotti et al. (2008), the change of the short-interval intracortical inhibition (SICI), i.e., subthreshold CS and suprathreshold TS to the unaffected M1, with an interstimulus interval of 2–4 ms, was correlated with upper extremity motor improvement (assessed by the Lindmark scale) from the acute phase (day 5–8) to 3-month poststroke, suggesting that the increase of motor inhibition over the contralesional M1 mirrored the motor improvement poststroke. Subgroup analysis revealed that the longitudinal correlation only existed in patients with a lower level of disability at baseline (BI ≥ 60), but not in patients with a higher level of disability. However, when using an intracortical facilitation (ICF) paradigm (subthreshold CS and suprathreshold TS to each M1, with an interstimulus interval of 7–15 ms), no significant correlation was observed (Manganotti et al., 2008). In another longitudinal study (Xu et al., 2019) using an interhemispheric inhibition (IHI) paradigm (subthreshold CS to the contralesional M1 and suprathreshold TS to the ipsilesional M1, with an interval of 10 ms), stroke patients at the acute and subacute phases demonstrated a normal level of IHI prior to movement onset, in comparison to their healthy counterparts. However, stroke patients at the chronic phase showed abnormal IHI prior to movement onset—the contralesional M1 remained inhibitory to the ipsilesional M1 before paretic hand movement. The longitudinal emergence of abnormal premovement IHI from the acute to chronic phases was negatively correlated with the improvement of hand functions in a finger individualization task. Therefore, this study concluded that the existence of an inhibitory flow from the contralesional to the ipsilesional M1 may not be the cause of poor recovery poststroke (Xu et al., 2019).

TMS-evoked EEG potential (TEP), by its nature, is a form of ERP, reflecting the brain’s responsivity to an external stimulus using a single TMS pulse. Besides, TMS-EEG outcomes can also be analyzed in the time-frequency domain to reflect the ongoing neural oscillation perturbated by TMS (Tremblay et al., 2019). Cross-sectionally, the ipsilesional hemisphere showed a simplified focal response to TMS pulses, with a smaller number of deflections and a higher amplitude in the TEP curve, compared with the contralesional hemisphere and healthy controls (Tscherpel et al., 2020). After considering the initial deficits and the ipsilesional MEP in a multivariate regression analysis, TMS-EEG outcomes are significant variables in the prediction model with a composite recovery score of the NIHSS, ARAT, and grip strength measured by a vigorimeter as the dependent variable (Tscherpel et al., 2020). The results indicate a complementary role of using TMS-EEG outcomes in predicting poststroke motor recovery, in addition to the ipsilesional MEP and the initial motor deficits.

Succinctly, the redistribution of rhythmic powers (spontaneous EEG powers at resting-state and ERD/ERS), pathological alterations of cortical evoked potentials (MEP, SSEP, and TEP), as well as change of the intra-/interhemispheric coherent activities represent distinct forms of functional disturbance of hemispheric activities after acute stroke. The evidence from neurophysiological studies largely supports that a greater level of initial interhemispheric imbalance poststroke predicts poorer motor and functional outcomes. Besides, paired-pulse TMS studies highlight the role of contralesional M1, especially for its intracortical inhibition, in monitoring poststroke motor recovery over time.

Structural MRI biomarkers provide straightforward information about the nature of stroke injury that may be relevant to functional prognosis (Jiang et al., 2010). Routine MRI techniques for acute stroke include T1- and T2-weighted imaging, diffusion-weighted imaging (DWI), etc. (Nanjundaswamy et al., 2011). In addition, diffusion tension imaging (DTI) is often used for measuring microstructural alteration of structural connections via white matter (Puig et al., 2017). No consistent correlation was found between global lesion volume and recovery outcomes (Schiemanck et al., 2006; Rehme et al., 2015); however, lesion to certain functionally relevant anatomical structures could determine, at least partially, the amount of functional recovery after stroke (Puig et al., 2017). For example, damage to the posterior limb of the internal capsule, which contains the corticospinal descending fibers, could best predict the severity of neurological deficits (assessed by the NIHSS) 3 months after stroke (Puig et al., 2011). Quantitative measures of CST show prognostic values as well: a weighted CST lesion load in DWI, measured at the acute phase, in addition to the initial motor impairment, predicted motor recovery (measured by the FMA-UE) at 3 months after stroke (Feng et al., 2015; Doughty et al., 2016). Similarly, a greater extent of CST degeneration, as revealed by its apparent diffusion coefficient (ADC) at the acute phase, was associated with a more severe neurological deficit at 3 months after stroke (DeVetten et al., 2010). The laterality index of the fractional anisotropy (FA) obtained from DTI is also commonly used in stroke recovery prediction—the level of FA asymmetry at the pons measured 1-month poststroke was found to be predictive of motor recovery 2 years after stroke, as assessed by the motricity index (Puig et al., 2013).

Apart from the CST, the intra- and interhemispheric corticocortical connection of the motor system (the M1, premotor, and supplementary motor area [SMA]) also shows a prognostic value (Salvalaggio et al., 2020). For example, using diffusion spectrum imaging MRI, generalized fractional anisotropy of both intra- and interhemispheric motor connections at the contralesional hemisphere measured at the acute phase, combined with age and the initial severity of stroke, could explain 96% of the variance of the severity of neurological deficits (measured by the NIHSS) at 6 months after stroke (Granziera et al., 2012). This evidence indicates the compensatory role of structural motor connections over the contralesional hemisphere in motor recovery poststroke.

Functional MRI (fMRI) provides an indirect measure of neuronal activity by detecting brain oxygenation and hemodynamics (Crofts et al., 2020). The level of the ipsilesional M1 activation during paretic hand movements, measured by fMRI, is strongly correlated with the level of upper extremity motor impairment, according to two meta-analyses (Rehme et al., 2012; Favre et al., 2014). In addition, better motor performance in patients with stroke is found to be related to a higher level of activation over the ipsilesional premotor cortex, pre-SMA, as well as the contralesional premotor cortex and cerebellum (Rehme et al., 2012). Therefore, functional activation of the motor-related system has been assessed as a form of imaging biomarker for prognostic prediction of poststroke recovery (Ward et al., 2003). For instance, neural activation during handgrip of the paretic side measured at the acute phase could predict motor outcomes at four to 6 months after stroke (Rehme et al., 2015). Patients with a good outcome (assessed by grip strength and the ARAT) showed a higher level of activation over the ipsilesional M1, SMA, and dorsal premotor cortex, and the contralesional cerebellum at the acute stroke phase (6 ± 3 days), compared with those with a poor outcome. However, results from a subsequent fMRI measure at 12 ± 4 days were not predictive (Rehme et al., 2015). Interestingly, other longitudinal fMRI studies demonstrated different forms of neuroplastic changes in patients after stroke (Ward et al., 2003; Hamzei et al., 2006). In acute stroke patients, overactivation in the primary and secondary motor areas was found initially. The movement-related brain activation started to reduce gradually over time, and a negative correlation between the change of brain activation levels in the cortical motor areas and the amount of functional recovery was reported in ischemic stroke patients without M1 lesion involvement (Ward et al., 2003). Research also indicates the neuroplastic change after stroke differs in patients with different levels of structural lesions. For example, a study with chronic stroke patients showed that patients with evocable ipsilesional MEP and intact hand area of M1 showed a reduction of the ipsilesional SMC activation along with recovery, while patients with a lesion involving the M1 or absent ipsilesional MEP showed an increase in the ipsilesional SMC activity accompanying recovery (Hamzei et al., 2006). These findings suggest that the structural and functional integrity of the CST may mediate the recovery patterns of brain activations after stroke.

Taken together, structural damage to CST as well as neuroimaging biomarkers of motor connections could potentially predict motor recovery after stroke. The level of activation of the motor system over the ipsilesional hemisphere during motor execution may also be indicative of the potential for recovery (Rehme et al., 2012; Favre et al., 2014). However, the structural lesion is likely to play a mediating role in brain reorganization, leading to different recovery and brain activation patterns (Ward et al., 2003; Hamzei et al., 2006; Rehme et al., 2015).

The S100B is a calcium-binding peptide abundantly found within glial cells and is used as a marker of glial activation and/or death in several neurological disorders (Yardan et al., 2011; Kamtchum-Tatuene and Jickling, 2019; Vuuren et al., 2022). High levels of S100B in cerebrospinal fluids (CSF) and the peripheral circulation have been associated with pathological injury and neuronal damage, and reliably predict injury severity (Yardan et al., 2011; Vuuren et al., 2022). In the course of metabolic injury associated with oxygen, serum, and glucose deprivation, glial cells respond by secreting S100B, and high concentrations of S100B lead to neuronal death via nitric oxygen release from astrocytes (Yardan et al., 2011). The prolonged presence of S100B in serum is an indication of its continuous release from affected tissue (Yardan et al., 2011). In ischemic stroke, high levels of S100B have been associated with the degree of systemic inflammation disregarding the size of the infarction (Michetti et al., 2012), therefore S100B has been proposed as a potential biomarker to predict recovery outcomes after stroke (Rahmati et al., 2020).

High levels of S100B in serum could significantly predict the level of neurological disability at 3 months after ischemic stroke, measured by the BI, mRS, and Lindley score (Abraha et al., 1997; Abd-Elhamid et al., 2016). In another study, S100B was measured in blood 3 months after ischemic stroke (Rahmati et al., 2020) and a significant correlation between high levels of S100B and worse prognosis (measured by the mRS and NIHSS) was observed (Rahmati et al., 2020). Similarly, high concentrations of S100B (blood samples taken 24 h after symptom onset) detected in serum by ELISA, were associated with more severe neurological deficits (NIHSS>10) after ischemic stroke (Altintas Kadirhan et al., 2021). These findings suggest that S100B could be a powerful prognostic biomarker in ischemic stroke.

Progranulin, a 68.5 kDa cysteine-rich secreted glycoprotein, is a multipotent growth factor also known as proetithelin 1, acrogranin 2, PC cell-derived growth factor 3, and granulin-epithelin precursor 4 (Xie et al., 2016; Abella et al., 2017). Progranulin is involved in several biological and pathological processes, including cell growth, tumorigenesis, wound healing, inflammation, and insulin resistance (Abella et al., 2017). Progranulin deficiency and overexpression have been associated with neurodegeneration and the onset of different types of cancer, respectively (Abella et al., 2017). This growth factor is primarily found in mature neurons and microglia, in which its expression is upregulated in response to injury and inflammation (Alexander et al., 2019).

The relationship between progranulin levels in serum and adverse functional outcomes after stroke has been studied. The levels of progranulin in fasting blood samples taken on the first morning after admission were analyzed by ELISA. A higher level of progranulin was correlated with a higher level of physical disability (measured by mRS) 6 months after stroke (Xie et al., 2016).

NfL is a cytoskeletal protein whose expression is limited to neurons and has emerged as a neuronal injury biomarker (Barro et al., 2020). The neuronal cytoskeleton is composed of neurofilaments consisting of three subunits, namely neurofilament light, neurofilament medium, and neurofilament heavy chains (Yuan et al., 2012). The NfL, together with the medium and heavy subunits, is one of the scaffolding proteins implicated in axonal injury that gets released into the extracellular matrix following axonal damage (Wang P. et al., 2020).

In patients with small vessel disease, NfL levels remained elevated for at least 3 months post-stroke (Wang P. et al., 2020). Measured in the first 24 h after admission, high concentrations of NfL can predict poor outcomes, as observed by the positive correlation between mRS, NIHSS scores, and serum levels of NfL (Wang and Wei, 2009). The advantage of NfL as a biomarker of neuronal damage is that it does not only predict prognosis if measured 24 h after admission (Wang P. et al., 2020), but it can also be used to quantify neuronal damage in stroke to predict recovery progress (Barro et al., 2020). NfL levels are not specific to stroke as they can also be influenced by physiological processes such as the body mass index and disease states (e.g., diabetes and hypertension; (Barro et al., 2020). Therefore, studies to validate NfL as a prognostic biomarker should consider the factors that can also affect NfL levels.

GFAP is a monomeric filament protein found as the main component of the cytoskeleton of mature astrocytes in the central nervous system (Amalia, 2021). Astrocytes react in response to injury or ischemic processes to maintain cell homeostasis through complex response mechanisms known as astrocytosis (Liu and Geng, 2018; Cole and Einstein, 2019; Amalia, 2021). In ischemic stroke, high levels of GFAP detected in CSF and serum in samples taken within 24 h after symptom onset were correlated to higher NIHSS scores (Amalia, 2021). GFAP levels peak after 48–74 h after symptom onset with a positive relationship between elevated GFAP levels, the severity of neurological deficits, and the extent of the brain damage (Amalia, 2021). Higher GFAP levels in serum predicted poorer outcomes according to the NIHSS at 1-year follow-up poststroke (Liu and Geng, 2018).

NSE, also known as the γγ-isoenzyme of the glycolytic enzyme enolase, gets released into the extracellular matrix as a result of neuronal injury and death, and can be detected in blood shortly thereafter (Kurakina et al., 2021; Shash et al., 2021). NSE has been used as a biomarker to assess the extent of structural and functional impairment in the central nervous system (Kurakina et al., 2021). Although no consensus has been reached on the correlation between NSE level and the severity of neurological symptoms in stroke (Kurakina et al., 2021), recent evidence supports its usefulness as a prognosis biomarker in stroke. NSE serum levels measured in samples taken shortly after symptom onset were high in patients with higher NIHSS scores measured 28 days after stroke (Shash et al., 2021). In a study following up patients for 90 days, high NSE levels were positively correlated with higher scores in the NIHSS and mRS (Kurakina et al., 2021). A similar trend was observed in ischemic stroke patients in a study assessing functional recovery 60 days after stroke using the NIHSS (González-García et al., 2012). NSE levels were measured in blood samples taken between 12 and 48 h after symptom onset, and NSE has positively correlated to the neurological outcome in stroke patients (González-García et al., 2012). In CSF and blood samples taken from both ischemic and hemorrhagic stroke patients, NSE was identified between four and 8 hours after symptom onset (Nasution and Bangun, 2022). NSE levels positively correlated with functional outcomes assessed using the NIHSS and mRS at 14 days post-stroke (Nasution and Bangun, 2022). In blood samples taken within the first 48 h after ischemic stroke, high NSE levels were correlated with the severity of neurological symptoms, assessed by the NIHSS (Kurakina et al., 2021). On the other hand, low levels of NSE were associated with good motor function outcomes 2 weeks after symptom onset, suggesting that NSE is a good predictor of motor recovery, measured by the Rivermead mobility index (Kurakina et al., 2021).

Copeptin, the C-terminal portion of pro-vasopressin, is a neurohypophysis hormone released in an equimolar ratio to vasopressin mirroring the levels of vasopressin (Zhang et al., 2013; Xu et al., 2017; Kamtchum-Tatuene and Jickling, 2019). In stroke, vasopressin release is driven by cerebral edema, volume overload, and hyponatremia, and copeptin increases in bacterial infections and febrile conditions (Zhang et al., 2013). Hypoxemia can lead to abnormal renal water excretion, which is most likely mediated via vasopressin release (Zhang et al., 2013; Wang C.-W. et al., 2014).

Several studies place copeptin as one of the most promising strong and independent biomarkers in ischemic stroke that predicts functional outcomes (Katan et al., 2009; De Marchis et al., 2013; Dong et al., 2013; Tu et al., 2013; Zhang et al., 2013; Wang C.-W. et al., 2014; Xu et al., 2017). A meta-analysis that evaluated the role of copeptin in prognostic prediction after ischemic stroke revealed that high levels of copeptin (tenfold increase) were associated with poor functional outcomes within 3 months or 1 year after stroke (Xu et al., 2017). Blood samples of the studies included in the meta-analysis were taken on the first day of admission (Xu et al., 2017). Furthermore, copeptin improved the prognostic value of the NIHSS compared to other prognostic biomarkers (Zhang et al., 2013).

Leptin, a 16 kDa protein secreted by the adipose tissue, is responsible for the regulation of food intake, non-shivering thermogenesis, reproduction, hemostasis, angiogenesis, regulation of arterial pressure, and neuroendocrine and immune functions (Zhao et al., 2021). Adiponectin, a 30 kDa glycoprotein expressed in adipose tissue, regulates the body’s sensitization to insulin (Wang Z. et al., 2019; Zhao et al., 2021). Low concentrations of adiponectin have been linked to disorders such as metabolic syndrome, atherosclerosis, hypertension, and cardiovascular disease, while high concentrations have been linked to increased risk for ischemic stroke (Wang Z. et al., 2019). Leptin functions as a neuroprotective agent against ischemic neuronal injury (Zou et al., 2020), while it has been suggested that adiponectin increases to counteract the systemic inflammation conditions observed in ischemic stroke (Wang Z. et al., 2019). A peak in serum levels of leptin and leptin/adiponectin ratio >1.16 has been observed 1 day after ischemic stroke and predicted poor functional outcomes measured using mRS after 3 months (Carbone et al., 2015; Kamtchum-Tatuene and Jickling, 2019). Also, high levels of adiponectin have been independently associated with poor prognosis (mRS scores 3–6) in ischemic stroke patients in a 3-month follow-up study (Wang Z. et al., 2019).

Fibrinogen is an inflammation-related protein that participates in thrombokinesis, platelet aggregation, blood viscosity, thrombus formation, and fibrin formation (Kim and Lee, 2014; Zhai et al., 2022). Fibrinogen is secreted in the liver and plays a role in cardiovascular risk through the regulation of coagulation and inflammation pathways (Hou et al., 2021). High concentrations of this hemostatic biomarker are directly associated with a higher incidence of ischemic stroke (Kim and Lee, 2014). In blood samples collected shortly after the onset of ischemic stroke, the levels of fibrinogen were found to be high (Kim and Lee, 2014; Sale et al., 2016; Peycheva et al., 2021). A negative correlation between high fibrinogen and improved outcomes measured by the NIHSS in ischemic stroke patients was reported, suggesting that high fibrinogen levels were an indicative biomarker of poor outcomes (Hou et al., 2021).

hFABP is a 15 kDa cytoplasmic protein that participates in fatty acid metabolism in the myocardium and neurons (Park et al., 2013). hFABP concentration in plasma was shown to be independently associated with stroke prognosis, as assessed 3 months after ischemic stroke (Park et al., 2013). Increased plasma levels of hFABP indicate acute cardiovascular dysfunction, which in turn is associated with neuronal injury (Park et al., 2013).

D-dimer is a fibrin degradation product that gets released into the circulation by fibrinolysis of cross-linked fibrin; thus, it is used as a biomarker to monitor coagulation and fibrinolysis (Abd-Elhamid et al., 2016; Wang J. et al., 2020). During thrombosis, the plasma levels of D-dimer increase, which makes it a promising biomarker of hemostatic anomalies and thrombosis (Yao et al., 2019). D-dimer is measured in plasma or whole blood to estimate the infraction size, stroke progression, incidence of stroke, and clinical outcomes in ischemic stroke (Abd-Elhamid et al., 2016; Yao et al., 2019). Mixed evidence of the correlation between D-dimer concentrations and the prediction of poor prognosis makes it controversial to conclude with a high degree of confidence that D-dimer can be used as a prognosis biomarker (Yao et al., 2019; Zhang et al., 2019; Wang J. et al., 2020). However, several studies have shown a positive association between D-dimer levels and functional outcomes. For instance, high plasma levels of D-dimer from samples taken within the first 24 h after admission were correlated with unfavorable functional outcomes assessed using the mRS 1 month after admission (Abd-Elhamid et al., 2016). In another study, patients with ischemic stroke were followed up at 3, 6, and 12 months after discharge (Wang J. et al., 2020). A positive and dose-dependent relationship between the plasma levels of D-dimer and increased risk of poor functional outcome was observed. The positive correlation remained significant at 12-month follow-up, suggesting that D-dimer is a promising biomarker for long-term prognosis (Wang J. et al., 2020). Similar results were reported in another long-term study in which D-dimer levels measured by ELISA within 1 week of admission, predicted mortality after 1.2 years of follow-up (Feinberg et al., 1996). However, the predictive ability of D-dimer depends on blood samples taken on admission. Therefore, it is crucial to assess whether D-dimer measured at different time points after stroke could still predict a poor prognosis.

Fibulin-5, found in the extracellular matrix and secreted by vascular cells, is a 66 kDa glycoprotein crucial for the assembly of elastic fibers and vasculogenesis (Hu et al., 2016; Elshony et al., 2021). Endothelial cells increase the expression of fibulin-5 in response to hypoxic conditions (Elshony et al., 2021). Fibulin-5 is considered a neuroprotective factor that improves neurological prognosis by increasing the blood-brain barrier permeability as a compensatory mechanism after ischemic stroke (Hu et al., 2016). Overexpression of fibulin-5 has been associated with improved clinical outcomes following reperfusion and has been regarded as a biomarker to predict severity and prognosis in both ischemic and hemorrhagic stroke (Hu et al., 2016; Elshony et al., 2021). Hypoxic stress, in ischemic stroke, and relative ischemia, in hemorrhagic stroke, induce fibulin-5 expression in endothelial cells via the hypoxia-inducible factor-1, which is an adaptive response to hypoxic conditions (Elshony et al., 2021). A positive correlation between fibulin-5 levels and functional outcomes measured using the mRS in a 3-month follow-up study in both ischemic and hemorrhagic stroke patients, suggested the ability of fibulin-5 to predict poor prognosis (Elshony et al., 2021). Similar results were reported in another 3-month follow-up study with ischemic stroke patients (Hu et al., 2016).

In ischemic stroke, the role of the inflammatory response has been previously established as the levels of several inflammatory biomarkers increased in stroke patients compared to their healthy counterparts (Aref et al., 2020). mHLA-DR is an inflammation biomarker used for injury-associated immunosuppression involved in the cross-talk regulatory mechanism between innate and adaptive immunity (Pfortmueller et al., 2017). Because mHLA-DR is associated with impaired monocyte function resulting in a low rate of antigen presentation, which leads to reduced lymphocyte response, it is considered a biomarker for immunosuppression induced by stroke (Mengel et al., 2019). In patients with ischemic stroke, low concentrations of mHLA-DA predicted poor functional outcomes measured using mRS after 3 months of symptom onset (Mengel et al., 2019).

IL-6 is an inflammatory cytokine that stimulates the release of prostaglandin E2 in the brain and is involved in the pathogenesis of several neurological disorders (Shaafi et al., 2014; Aref et al., 2020; Mosarrezaii et al., 2020). IL-6 acts as a secondary messenger in the communication between leukocytes, vascular endothelium, and parenchymal cells and participates in the response to neuronal damage (Mosarrezaii et al., 2020). Shortly after ischemic events, the levels of IL-6 significantly increase in the CSF and blood; this increase is an indication of early neurological deterioration and poor prognosis (Aref et al., 2020; Mosarrezaii et al., 2020). In a study with ischemic stroke patients, serum IL-6 levels were monitored at different time points (4, 8, 12, 24, 48, and 72 h, 7 and 12 days, 3 months, and 1 year) after stroke to assess the correlation between serum IL-6 with stroke volume, size of the infarction, and long-term functional outcomes (Waje-Andreassen et al., 2005). The study showed a positive relationship between IL-6 levels and these variables, demonstrating the promising use of IL-6 as a biomarker to assess long-term prognosis after ischemic stroke (Waje-Andreassen et al., 2005). Similar findings have been reported in more recent studies. For instance, poor functional outcomes (measured using the European stroke scale and BI) were associated with a high concentration of IL-6 after 3 months of hospital admission in samples taken within 36 h of the stroke event (Mengel et al., 2019). A positive relationship between stroke severity (NIHSS), physical disability (mRS), and IL-6 serum levels after 5 days, 3 months, and 1 year was reported (Shaafi et al., 2014). Furthermore, elevated levels of IL-6 in both CSF and blood have been associated with an increased risk of fever, which in turn causes tissue damage after ischemic stroke (Aref et al., 2020). Poor functional outcomes (NIHSS and mRS) after a 3-month follow-up of ischemic stroke patients were associated with high concentrations of IL-6 in serum measured in samples taken within the first 24 h of symptom onset (Aref et al., 2020).

The acute-phase CRP, also known as high-sensitive CRP, is a non-specific biomarker of inflammation rapidly produced in the liver in response to tissue injury or infection (Yu et al., 2017). Cell death, brain injury, and blood-brain barrier disruption have been associated with elevated concentrations of CRP post-stroke (Gu et al., 2022). High levels of CRP have been reported in the first 48 h after symptom onset, and they remained elevated on day seven and even after three to 6 months post-stroke (Mengozzi et al., 2020). Poor functional outcomes assessed with the mRS and stroke recurrence were predicted by high levels of CRP in patients with ischemic stroke in a 3-month follow-up study (Gu et al., 2022).

The complement MBL is an important element of the innate immune response (Osthoff et al., 2011). MBL activates the complement system through MBL-associated serine proteases to initiate complement activation via the MBL pathway (Huang et al., 2016). Complement activation is known to have detrimental effects on the pathogenesis of ischemia/reperfusion injury through the production of free radicals, stimulation of the coagulation cascade, and tissue damage (Osthoff et al., 2011). Complement activation may also result in the stimulation of potent inflammatory reactions that lead to tissue damage (Zhang et al., 2015). MBL is an important role player in brain injury as this molecule regulates and coordinates several pathogenic pathways involved in ischemic brain injury (Neglia et al., 2020). In patients with acute ischemic stroke, low levels of MBL in serum were associated with smaller infarct volume and improved functional prognosis (measured by mRS) in a 3-month follow-up study (Osthoff et al., 2011). Furthermore, higher serum MBL levels in patients with ischemic stroke predicted poor outcomes in the mRS in another study (Zhang et al., 2015).

25-(OH)D is a well-known biomarker to assess the status of vitamin D. Reports indicate that vitamin D deficiency can induce brain damage and cause functional impairment. Macrophage and lymphocyte activity in atherosclerotic plaques is influenced by low vitamin D levels, and this phenomenon promotes inflammation in artery walls (Tu et al., 2014). As vitamin D has anti-inflammatory activity, high levels of vitamin D have been shown to promote recovery after stroke. For instance, a study showed a negative correlation between vitamin D and unfavorable outcomes at discharge (mean hospital stay of 31 days) assessed using the NIHSS (Wang Y. et al., 2014). Another study assessing the level of disability using the mRS after 3 months of the stroke event found that a poor outcome measured by the mRS was associated with low serum levels of 25-(OH)D (Tu et al., 2014).

YKL-40, also known as chitinase 3-like-1 and human chitinase-like protein, is an inflammatory biomarker secreted by monocytes that infiltrate vessel walls upon brain injury (Park et al., 2012; Nadeem et al., 2019). YKL-40 activates modulators of cell proliferation, survival, migration, and adhesion such as mitogen-activated protein kinase (MAPK) and phosphoinositide 3-kinase (PI3K), which increases the chances of a vascular event (Li et al., 2022). In patients with ischemic stroke, a positive correlation between YKL-40 levels and stroke severity (NIHSS) and a poor functional outcome (mRS) was observed in a 3-month follow-up study (Park et al., 2012). YKL-40 long-term prognosis was evaluated after 3 months and one to 5 years using the mRS in patients with ischemic stroke or transient ischemic attack, and an independent association between YKL-40 levels with stroke recurrence, and an unfavorable outcome in the mRS were reported (Li et al., 2022).

SOD is an oxidoreductase that catalyzes dismutation of superoxide radicals into O2 and H2O2 (Fukai and Ushio-Fukai, 2011). Molecular and biochemical studies have characterized three isoforms of superoxide dismutase, namely SOD1 (Cu/ZnSOD) in the cytoplasm, SOD2 (MnSOD) in the mitochondria, and SOD3 (ecSOD) in the extracellular matrix and on the cell surface. SOD1 and SOD2 are present in practically all cells, while SOD3 is found in selected tissues, including blood vessels, lungs, kidneys, and hearts, as well as extracellular fluids including blood (Fukai and Ushio-Fukai, 2011). After ischemic stroke, the SOD activity of the brain decreased dramatically and gradually decreased with infarct size (Spranger et al., 1997). Low serum levels of SOD were found within 24 h of stroke onset and the level of functional impairment, measured by the Scandinavian Stroke Scale (SSS), and the size of the infarction, were negatively correlated with SOD activity.

MDA is a lipid peroxide produced when polyunsaturated fatty acids are peroxidized (Barrera et al., 2018), and it is markedly elevated during the acute phase and persistently maintained during the late phase of stroke (Nakane et al., 2015). When subjects were compared in the acute, subacute, and chronic phases of stroke recovery, a study found that the plasma lipid peroxide level significantly increased over time (Alexandrova et al., 2003). Serum MDA levels of post-stroke patients were significantly higher than their healthy counterparts (Bir et al., 2006; Corrêa Mde et al., 2008). The primary source of MDA production in post-stroke patients is still unclear. Increased lipid peroxide and MDA levels, on the other hand, could be caused by spontaneous reactive oxygen species (ROS) generation by peripheral phagocytes (Alexandrova et al., 2003), extracellular hydrogen peroxide production by paretic muscles during denervation (Jackson and McArdle, 2016), or associated comorbidities such as diabetes, hypertension, and hyperlipidemia (Cheng et al., 2017). MDA, measured in serum, has been proposed as an oxidative stress marker (Ohkawa et al., 1979). In the follow-up evaluation of stroke patients, the assessment of the short-term outcome and disability was measured 3 months after stroke using the mRS (Elsayed et al., 2020). Serum samples were taken within 24 h after admission (Elsayed et al., 2020). An elevated serum MDA level at admission was independently associated with a poor functional outcome, as assessed with the mRS, following an ischemic stroke at a 3-month follow-up (Elsayed et al., 2020).

Among the cytokines produced primarily by adipose tissues, RBP4 is the only protein specifically responsible for delivering retinol (vitamin A) to tissues and blood (Quadro et al., 1999). In the brain, retinol is oxidized to retinoic acid, which has been shown to contribute to neurogenesis and neuroplasticity during both the early stages of growth of the central nervous system and adult functions of the brain (Plane et al., 2008; Shearer et al., 2012). The anti-apoptotic properties of retinoic acid have also been demonstrated to reduce cerebral ischemia by regulating inflammation (Harvey et al., 2004; Choi et al., 2009). RBP4 levels have been examined in the context of ischemic stroke (Sasaki et al., 2010). Patients with cerebral infarction had higher levels of RBP4 compared to healthy controls (Llombart et al., 2016). Elevated levels of RBP4 have been found among women who had a stroke compared to those who did not have a stroke during the discovery phase based on data from the Women’s Health Initiative (Rist et al., 2018). Moreover, a study involving a 3-month evaluation where blood samples were collected within 48 h after the stroke, found that serum RBP4 levels were significantly higher in patients with acute stroke than in those in the healthy controls, and were correlated with stroke severity and poor neurological outcomes as measured using the NIHSS (Liu and Che, 2019).

HbA1c has been recognized as abnormal hemoglobin in diabetes (Rahbar et al., 1969). The discovery of HbA1c led to numerous small studies correlating glucose measurements to HbA1c, resulting in the idea that it could be used objectively to measure glycemic control (Rahbar et al., 1969). Glycated hemoglobin (HbA1c) is a useful biochemical indicator for measuring glycemic control over the past 3 months because the values are not affected by short-term transient fluctuations in blood glucose levels (Bao and Gu, 2021). A positive association was found between blood glucose levels and the levels of HbA1c at the time of admission for stroke (Toni et al., 1994; Mohanty et al., 2000; Kagansky et al., 2001). The triggering of inflammatory pathways may play a role in determining the size or severity of infarcts caused by uncontrolled blood glucose (Toni et al., 1994; Mohanty et al., 2000; Kagansky et al., 2001). Each unit increase in HbA1c (continuous) and HbA1c ≥ 6.5% were found to be associated with an increased risk of poor functional outcome defined by the mRS (Wang H. et al., 2019) after ischemic stroke. Thus, these findings suggest that HbA1c may be a useful biomarker for monitoring or predicting poststroke prognosis, allowing appropriate management to be instituted (Bao and Gu, 2021).

Matrix metalloproteinases (MMPs), a class of proteolytic zinc-dependent enzymes, have been associated with a variety of pathophysiological processes including systemic inflammation, atherosclerosis, and central nervous system diseases (Creemers et al., 2001; Galis et al., 2002). MMP-9 (gelatinase B, 92-kDa collagenase) is a well-studied enzyme in ischemic stroke, and its expression is rapidly upregulated after cerebral ischemia (Romanic et al., 1998; De Marchis et al., 2013). Several clinical studies have found a correlation between circulating MMP-9 levels, disease severity, and infarct volume in the hyperacute phase (Rosell et al., 2005; Demir et al., 2012) and late hemorrhagic infarction between days five and seven after stroke onset (Montaner et al., 2001). A higher serum MMP-9 level in ischemic stroke was associated with an increased risk of physical disability (mRS) in a 3-month follow-up study (Zhong et al., 2017). Therefore, MMP-9 may be a potential prognostic biomarker for ischemic stroke (Ramos-Fernandez et al., 2011).

LBP is an acute-phase protein that is critical for endotoxin signaling (Schumann and Zweigner, 1999; Bas et al., 2004). In endotoxin response, LBP is synthesized by gastrointestinal and liver epithelial cells (Schumann and Zweigner, 1999). As a part of the innate immune response, LBP plays an important role in both Gram-negative and Gram-positive infections (Zweigner et al., 2006). In addition to binding directly to endotoxins, LBP facilitates their transport to CD14 receptors, a receptor that detects patterns, and stimulates the production of proinflammatory cytokines by binding to toll-like receptors promoting the binding of lipopolysaccharides to these receptors. It has been observed that the levels of LBP and sCD14 in blood increase during an ischemic stroke (Klimiec et al., 2016; Hoffmann et al., 2017). In a study measuring plasma LBP levels on days one to four after stroke, it was found that the highest level of plasma LBP after stroke was observed on day four in patients with stroke-associated pneumonia (Hoffmann et al., 2017). A high level of LBP and sCD14 was associated with poor functional prognosis (Mengel et al., 2019). A moderate correlation was found between plasma LBP and acute lesion volume in stroke patients without infections (Hotter et al., 2019). Levels of LBP in stroke patients with infection are increased (Wartenberg et al., 2011). In another study, LBP was predictive of poor prognosis 3 months after stroke, with high levels of LBP associating with poorer outcomes in the mRS (Mengel et al., 2019).