After stroke, a cascade of inflammatory and degenerative mechanisms is triggered that ultimately determine the extent and severity of damage, the development of late secondary neurodegeneration, and the degree of disability or recovery (17). In the first phase, injury of the central nervous system (CNS) results from the dysfunction and death of neurons and of multiple cell types, including astrocytes, pericytes, smooth muscle cells, endothelial cells, oligodendrocytes, microglia and neural and glial precursor cells, which together constitute the so-called neurovascular unit (NVU) (18, 19). The NVU has been hypothesized to regulate the balance of the neural–glial–vascular signaling in a biphasic mode, that is mainly oriented to the damage during the acute injury phase and to the regeneration in the chronic recovery (20–22). Thus, post-stroke recovery results from reversal of tissue dysfunction, by promoting neurogenesis, angiogenesis and vascular remodeling (23–27) (Figure 1).

The imbalance between persistence of chronic inflammatory degenerative status and neurogenesis is considered responsible for persistent neurological dysfunction and impairment of functional recovery following damage of the CNS (28–30). Indeed, it is known that the occurrence of brain lesions, due to trauma or stroke injury, increases the incidence of late impairments and chronic neurodegenerative conditions, such as Alzheimer’s disease (AD), Parkinson’s disease (PD) and chronic traumatic encephalopathy (31, 32).

SNDG refers to neurophysiological and histopathologic changes occurring in non-ischemic remote brain regions that have anterograde or retrograde synaptic connections with the primary lesion site (6); thus, SNDG has been hypothesized to be a potential modulator of post-stroke functional disorders (33).

Most of the knowledge on neurodegenerative processes associated with stroke is based on studies with neuroimaging and in particular brain MRI, providing an excellent anatomical detail and gray/white matter contrast. For this reason, structural MRI, using conventional sequences, especially T1, T2-weighted, in addition to contrast enhancement, has become the accepted standard for routine examination of the brain, offering high sensitivity to anatomical location and morphological characteristics of pathological processes.

Studying the SNDG with conventional MRI, disruption to global functional connectivity has been revealed in the ipsi-lesional corticospinal tract and in the inter-hemispheric connections (corpus callosum), in the bilateral inferior fronto-occipital fasciculus and in the bilateral superior longitudinal fasciculus (34–36), in the ipsilateral thalamus, in the substantia nigra, hippocampus and in amygdala (37–39).

The involvement of these areas distal to the primary lesion defines the onset of cognitive and behavioral symptoms different from those primarily related to stroke area (36, 40). For example, involvement of the thalamus can lead to hyperalgesia; damage in the substantia nigra provokes Parkinson-like symptoms (slow movements, tremor, stiffness and difficulty with walking and balance); involvement of amygdala results in difficulty with memory processing and emotional reactions, whereas an involvement of hippocampus to memory impairment (Table 1).

In addition, other pathogenic mechanisms occurring after stroke have been reported as potential factors leading to cognitive decline. Interestingly, some studies described alterations of functional connectivity measured with functional MRI in different parts of cortex that appear structurally normal, not contiguous to the stroke area (49, 50). This suggests that uninvolved areas of the brain, even if structurally normal, may have functional impairment. It was supposed that the stroke lesion can disrupt the mutual balanced inhibition between hemispheres by decreasing the inhibition of the contralateral non lesioned hemisphere and by reducing the number of cortical-spinal tract projections to the other brain areas, with the effect of a remote functional depression (51).

Following ischemic stroke, microglia and astrocytes are activated within hours, inducing the production of cytokines and chemokines and the infiltration of leukocytes (52, 53). Oxidative stress, a disturbance in the balance between the production of reactive oxygen species and antioxidant defenses, is induced in cerebral ischemia especially through inflammation and reperfusion, increasing the production of free radicals (54). After the activation of the peripheral immune response, macrophages and neutrophils are released by the spleen into the bloodstream, which can easily reach the brain due to the alteration of the blood–brain barrier (55). This immune response, associated with the activation of resident inflammatory cells such as microglia and astrocytes, has been demonstrated to contribute to the development of SNDG (56, 57).

On a cellular level, stroke lesions cause the disruption of the structure of the neuron’s cytoskeleton (58). Those changes in cytoskeletal structure and subsequent neuron instability and final neuronal death are associated with the formation of protein aggregates and the release of cytoplasmic proteins in the extracellular space, representing a valuable surrogate index of acute necrosis or slow

neurodegeneration (59, 60).

Among these, tau proteins, a microtubule associated protein, and the corresponding hyperphosphorylated forms, p-tau, are well-established biomarkers of neurodegeneration, and represent predictors of functional outcome (61) or development of cognitive dysfunction after cerebral ischemia (62–64). Further, increase of microtubule-associated protein (MAP2) reflects the fragmentation of neuronal dendrites, not only in the areas of stroke but even in other brain regions resulting in a widespread loss of synaptic plasticity (65, 66). Finally, stroke can also lead to accumulation of amyloid-beta (AB), a peptide that is the main component of amyloid plaques in the brains of subjects with Alzheimer’s disease (67).

The deposition of the neurotoxic aggregates of protein amyloid-beta could be further stimulated by the glial activation and pro-inflammatory cytokine release after cerebral ischemia, which persists for a long time and in some cases could determine the latter onset of Alzheimer’s type dementia (17).

Neurofilament proteins are components of the cytoskeleton of neurons, classified on their molecular weight in light (NF-L), middle (NF-M) and heavy (NF-H). As a subunit of neurofilament, serum NF-L has emerged in the last few years as the most promising biomarker of axonal injury and neurodegeneration, mainly in multiple sclerosis (MS), showing potential applications for both patient monitoring and for observational and interventional trials (78, 108).

Recently, numerous studies have been carried out on patients with stroke showing significant correlation with severity, according to National Institutes of Health Stroke Scale (NIHSS) upon admission (77, 78, 80, 109, 110) and after clinical outcome (111).

A meta-analysis on ischemic stroke or transient ischemic attack patients demonstrated that patients with higher serum NF-L, had increased risk of poor functional outcome during follow-up, compared with those with lower NF-L, strengthening the value of NF-L as predictive biomarker for ischemic stroke outcome (75). Furthermore, a study on a cohort of 1,694 patients with first-ever acute ischemic stroke investigated the correlation among NF-L and post-stroke cognitive impairment (PSCI). Among the cohort, 60.74% of patients developed cognitive impairment. Interestingly, authors found that the PSCI group exhibited significantly higher levels of NF-L within 48 h of stroke onset; levels of NF-L were negatively correlated with cognitive impairment defined by Montreal Cognitive Assessment (MOCA) (MOCA <26) at 90 days after stroke onset, and directly correlated with age, cerebral infarction volumes and NIHSS score (76).

Importantly, results from different studies showed a wide heterogeneity regarding the sampling time, reporting best association among clinical outcome and serum NF-L measured at 7 days or within 48 h after symptom onset (77, 78). Pending multicentric studies and large-scale validation, it is however suggested to perform the analysis of NF-L before the 7-day time point.

Converging evidence suggests that NF-L level has an added predictive value apart from stroke severity. This may be explained by the presence of SDNG outside the infarct area, such as white matter tracts connected to the infarct, which could contribute to the development of cognitive impairment after the stroke (76). A study in the acute and post-acute phase after stroke confirmed that high levels of NF-L are associated with poor clinical outcome, and later on, the concentration of NF-L positively correlates with the occurrence of SDNG, as assessed by MRI.

Surprisingly, increase of NF-L has shown to predict functional improvement in the late phase after stroke in patients who survived for more than 10 months. This could suggest that the kinetic of NF-L may reflect two concurrent but very distinct processes of NF-L release, namely neuroaxonal injury and synaptic damage, that are features of secondary neurodegeneration, and oppositely of late adaptive neural plasticity. Indeed, while in the acute phase high blood NF-L seem to reflect the extent of neuronal injury, in the late phase increase of NF-L may serve as a biomarker of adaptive neural plasticity and a positive predictor of functional improvement and, therefore, effectiveness of neurorehabilitation (79).

Thus, results from different studies suggest that at different time points, NF-L levels may be considered a measure of structural brain lesion, complementary to MRI (80) but in the late phase after stroke, the interpretation of elevated blood levels of NF-L should not be therefore limited to the extent of injury or neurodegeneration (79).

Tau, a microtubule-associated protein that regulates stability and dynamics of axons, is considered a well-recognized marker of neuronal degeneration by a huge amount of both clinical and preclinical studies. In the research framework of AD, CSF tau is one of the core markers of the so called “ATN system,” where “A” refers to the value of an amyloid β biomarker (amyloid PET or CSF Aβ42); “T,” the value of a tau pathology biomarker (CSF p-tau or tau PET); and “N,” a quantitative or topographic biomarker of neurodegeneration or neuronal injury (112, 113).

So far, the assessment of total tau level in the CSF or blood is thought to reflect the extent and intensity of neuronal damage of any etiology. Oppositely, the presence of post translationally modified tau species, i.e., truncated or phosphorylated, is considered a typical feature of some tauopathies. Blood and CSF levels of tau, but not p-tau, increases after ischemic stroke and mild head trauma, being highly increased in case of prion disease as Creutzfeldt-Jakob disease (114–116). Conversely, in tauopathies such as AD or other dementia, the production of p-tau, C- or N terminal truncated tau species are specifically increased, reflecting underlying neurofibrillary pathology and the formation of post translationally modified species which accumulate in brain and can be also detected in biological fluid (117–119).

Tau proteins have been proposed as candidate markers of stroke and SDNG for evaluating the extent of neurodegeneration, and potentially discriminating neuronal damage with the presence of degenerative processes of other etiology, such as Alzheimer’s pathology. Studies combining CSF analysis and neuroimaging demonstrate a correlation among the magnitude of t-tau increase and the volume of tissue damage following ischemic stroke (81). A recent study compared levels of plasma tau and NfL as cross-sectional and longitudinal markers of cognitive decline and neuroimaging changes, on a cohort of 995 subjects from the community-based Mayo Clinic Study of Aging (MCSA) (82). Both plasma NfL and T-tau have been found to be elevated in stroke patients, and among those with other cardiovascular conditions; the combination of having both higher NfL and T-tau, compared to either alone, was more strongly associated with lower memory, global cognitive decline, brain atrophy and a higher number of infarcts cross-sectionally (82).

Interestingly, in stroke patients, concentration of plasma tau progressively increases from admission reaching a maximum peak 7 days after event, showing a correlation with severity, long-term outcome, as well as infarct volume (83, 84). Analysis of kinetics of tau after global brain ischemia revealed a bimodal elevation after day 2 and 4, probably due to the occurrence of early first necrosis and secondary delayed neuronal death (85, 120). This is also confirmed by a recent retrospective study on head trauma and remote injury, involving a total of 164 subjects, 94 PD patients and 70 healthy controls, in which the levels of CSF tau were found to be increased in a subgroup of PD patients who reported lifetime head trauma preceding diagnosis, probably due to extended degeneration or occurrence of tauopathy (121).

Evidence from experimental models also suggest that tau has a key role in regulating neuronal damage and SDNG after stroke, up to the development of Alzheimer’s-type dementia (64). In animal models, tau protein dysfunction following ischemia may trigger neurofibrillary tangle-like tauopathy and neurofibrillary tangles (86, 87). In humans, history of ischemic stroke increases likelihood of developing AD-type dementia (122) and further, patients suffering of AD with previous brain ischemic injuries show more severe dementia phenotype (123). Certainly, tau in combination with other biomarkers is useful for clinical and research purposes in order to reveal the development of future or clinical subtle neurodegeneration.

S100b is a calcium-binding protein mainly expressed in astroglial and Schwann cells, myeloid-derived cells, and a few other cell types. Under physiological conditions, S100b is released in the extracellular space in response to hormonal or inflammatory stimuli and exerts both paracrine and autocrine effects on neurons and glia, but increases are observed in neuronal pathological conditions, such as brain trauma, ischemia and neurodegenerative, inflammatory and psychiatric diseases.

S100b is widely used in emergency medicine due to its high positive predictive value in cases of brain injury, showing a correlation with the severity of post-traumatic neurological deficit as well as with the infarct volume after stroke (88–90, 124).

A study on 32 consecutive patients evaluated changes of S100b and the neuron specific enolase (NSE) levels between the first 6 h, and in the 5 days after stroke. Serum concentrations of S100b from 6 h on and of NSE from 24 h on were associated with the outcome at 3 months with a maximum of correlation obtained for protein S100b at 48 h (91).

Moreover, the changes in S100b concentration reflect the extent of astroglial damage, the enlargement of the ischemic core, (92) as well as the formation of malignant edema and of blood–brain barrier damage (93). Conversely, in transient ischemia not associated with substantial tissue injury, S100b levels in serum are generally normal (94, 125). Recently, S100b has been proposed as a surrogate marker to monitor the stroke response after endovascular treatment. Indeed, levels of S100b are correlated with the recanalization following mechanical thrombectomy, being low in case of successful recanalization, and oppositely being increased in case of ineffective recanalization and in patients who developed infarcts despite recanalization (95).

So far, studies in patients with neurodegenerative diseases showed conflicting results. Some authors found that serum S100b concentrations were similar (126) or lower in CSF of AD patients compared with elderly controls, not correlating with brain atrophy (127), whereas others reported correlations between CSF S100b levels and AD brain atrophy or cognitive status as measured by the Mini Mental State Exam score (128, 129).

Notwithstanding, studies are still conducted on limited cohorts, therefore results need to be consolidated by larger cohorts and longitudinal studies, evaluating the association between S100b changes, volume of tissue damage and the functional or cognitive outcome over the post-stroke phases. Although S100B can be considered a useful biomarker in the acute phase after stroke to evaluate the damage of the NVU and astroglial cells, further investigations to understand its role in SDNG following stroke are needed.

Neuron specific enolase (NSE) is an enzyme involved in glycolytic energy metabolism in the brain, released from neurons during injury as a nonspecific marker of neuronal damage. Studies have investigated the potential of NSE as a predictor of outcome in patients in the early phase after stroke. In 150 cases of patients with stroke and 101 controls, NSE level showed a positive correlation with stroke severity at the time of admission and with the degree of disability, categorized into mild, moderate and severe according to NIHSS; further, NSE was associated with neurological worsening after 7 days of event (96).

Data obtained from 12 studies including 597 patients, found that serum NSE levels were higher in stroke patients compared with controls and correlated with volume of infarct; however, results do not support a correlation among NSE and functional outcome and further the relationship to stroke severity is unclear (97). Increase of NSE as well as of tau was highly correlated with severity of neurological deficits and infarct volume, and with the functional outcome at 3 months. Interestingly, NSE release was associated with the neurovascular status on admission (84).

Findings on CSF in patients with primary neurodegenerative diseases, such as AD, PD, LBD, suggest that NSE may be used to evaluate the presence and extent of axonal and glial degeneration (130). NSE in combination with tau may predict secondary damage after stroke, with specific windows that reflect different release mechanisms (84). In fact, the NSE concentration in serum increase 2–3 h after onset of first stroke, afterwards, NSE decreased, followed by a secondary increase until day 5, that is the last measurement in the observation period; the secondary increase indicates further release of NSE, which probably reflects a secondary mechanism of brain damage, ongoing neuronal cell death (98, 99), or persistent disturbance of the blood–brain barrier (131). A recent study found a correlation between NSE and both stroke severity and prognosis after 1 year of follow up documented by NIHSS score and modified Rankin Scale (mRS) score, respectively, on 1,086 patients grouped as hypertension and non-hypertension (100). However, studies mainly focus on the role of NSE in the acute and subacute phase after stroke and its potential predictive or prognostic value. Longitudinal studies on large cohorts are needed to evaluate the association between NSE levels and the development of brain morphological changes in secondary neurodegeneration.

The brain derived neurotrophic factor (BDNF) is the most abundant neurotrophin in the adult brain, and has a remarkable capability to repair brain damage and maintain synaptic plasticity by inducing neuronal proliferation, survival and differentiation (132). However, unlike synaptic plasticity involved in normal cognitive function, post-stroke and rehabilitation neuroplasticity primarily refers to the brain’s ability to recover from injury to restore its normal structure and function. In this context, the significant role of BDNF in the regulation and maintenance of synaptic plasticity after stroke has been extensively investigated in both clinical and experimental studies, including the potential use of BDNF as a direct therapeutic agent for the stroke treatment (133).

After a stroke event, reactive astrocytes upregulate the expression of BDNF and other neurotrophic factors, and the resulting levels have been demonstrated to be associated with the clinical and functional outcome. For instance, BDNF decrease after stroke is associated with poor long term functional outcome (101) and development of post stroke depression (102); further, low level of BDNF is associated with increased risk of stroke (103).

A recent meta-analysis of data from 62 studies showed that subjects with stroke (n = 1856) had lower BDNF levels compared to healthy controls (n = 1,191), but there was no significant difference in the level of BDNF through time points post stroke. Furthermore, BDNF levels were lower in the patients with depression compared to non-depressed subjects, and positively affected by performing physical training in the early but not in the long term (104).

Preclinical studies demonstrated that BDNF induces anti apoptotic mechanisms, reducing the size of the lesion and secondary death (105), promoting synaptogenesis, neuronal plasticity and recovery post stroke (105, 106); moreover, increase of BDNF concentration in the cortex is related to motor learning after-stroke (107). Experimental studies demonstrated clinically positive outcomes reached by administration of stroke treatments that modulate BDNF expression, leading to consider BDNF as potential therapeutic target (133). Together, these studies suggest that BDNF exerts favorable effects in post-stroke recovery due to its attenuation of cell death and promotion of neurogenesis. However, longitudinal studies investigating the correlation between BDNF levels and other biomarkers of neurodegeneration, as well as morphological and clinical changes, are needed to better understand the development of SDNG and recovery in the chronic phase after stroke.