Following ICH, the mass effect during hematoma formation can physically distend, distort and finally disrupt loose WM fibers. Those nerve fibers within the hematoma cannot recover, while the fibers surrounding the peri-hematoma show damage of various degrees of severity depending on the bleeding speed and hematoma size. Alternatively, to date, the biomechanical mechanism of barotrauma was considered similar in ICH and TBI because many similar pathophysiological changes between TBI and ICH have been observed, such as bilateral diffuse brain injury and brain atrophy due to an unilateral brain lesion (Powers, 2010; Kummer et al., 2015). In contrast to TBI, in which the barotrauma arises externally from a blow to the head, the barotrauma in ICH originates from the fluid percussion waves due to the sudden expansion of the hematoma under arterial pressure in the cranium and immediately propagates through the intracranial contents (Powers, 2010), resulting in distal axotomy and demyelination. (Figure 2). The mechanism of the barotrauma may partially explain the global WMI observed during the acute stage after regional ICH.

Inflammation is a major contributor to secondary WMI. Inflammation progresses in response to various stimuli produced after hemorrhage via the activation of microglia, peripheral inflammatory infiltration and the production of cytokines and chemokines. In addition, local microglial activation and peripheral leucocyte infiltration mutually interact to propagate inflammatory injury. The inflammatory reaction to ICH within WM has been demonstrated; for instance, nuclear factor kappa B (NF-κB) is rapidly activated, the pro-inflammatory cytokine genes are upregulated in the perihematomal WM (Wagner et al., 2006) and CD47, which is a transmembrane protein that regulates the entry of leukocytes into the brain, is elevated in the oligodendrocytes and microglia in the WM after experimental ICH (Zhou et al., 2014). Inhibiting inflammation can reduce WMI, providing further evidence of inflammation-induced WMI. For instance, inhibiting one of the pro-inflammatory factors, i.e., EP1R, which is expressed primarily in axons, could attenuate WMI, reduce brain atrophy and improve functional outcomes in an ICH animal model (Zhao et al., 2015). Furthermore, neutrophil depletion before ICH induction in the rat striatum reduced the infiltration of activated microglia in the peri-hematoma WM tracts and decreased myelin fragmentation and axon damage (Moxon-Emre and Schlichter, 2011).

Although the exact mechanisms of the inflammation-mediated WMI remain elusive, matrix metallopeptidase 9 (MMP-9) is considered to play a key role during this process. MMP-9 is produced by leukocytes, astrocytes and oligodendrocytes, and the MMP-9 promoter region contains activator protein-1 and NF-κB, both of which respond to inflammatory stimuli and specifically attack the basal lamina through the fibronectin binding domain (Seo et al., 2013; Vafadari et al., 2016). MMP-9 has been reported to be activated after ICH with consequential WMI, and many inflammatory mechanisms underlying WMI are also associated with the activation of MMP-9. For instance, EP1R was found to have a toxic effect on the WM surrounding the hematoma via the Src kinases–MMP-9 signaling pathway following ICH (Zhao et al., 2015); neutrophil depletion leads to attenuated myelin fragmentation and axon damage with reduced MMP-9, indicating that MMP-9 produced by infiltrated neutrophils may contribute to WMI (Moxon-Emre and Schlichter, 2011). Another inflammatory signaling axis recently identified using an ICH model, i.e., iron—the receptor for advanced glycation end-products (RAGE)—NF-κB, is associated with the BBB and WMI via the activation of MMP-9 (Yang F. et al., 2015). However, in contrast to its deleterious role during the early period, MMP-9 secreted by oligodendrocytes can promote WM remodeling by accelerating angiogenesis that is mediated by an analogous oligovascular niche during the recovery stage after lysophosphatidylcholine-induced WMI in mice (Pham et al., 2012). It is unclear whether the late beneficial effect of MMP-9 exists and contributes to WM repair following ICH. If this hypothesis is confirmed in the future, treatments targeting MMP-9 should avoid the complete inhibition of MMP-9 during all stages.

WM is vulnerable to oxidative stress induced by ischemic or hemorrhagic events (Hall et al., 2000; Wang Y. et al., 2016). Wagner et al. (2002, 2005), evaluated the oxidative injury in WM by measuring the formation of protein carbonyl and gene expression of HO-1 after infusing either plasma or whole blood into the porcine frontal WM and concluded that not only whole blood but also its plasma components are capable of rapidly inducing oxidative stress in WM.

Erythrocyte lysis followed by the release of hemoglobin and its metabolites occurs several days after ICH (Cao et al., 2016). Heme, which is the major product of hemoglobin, can be degraded into iron, carbon monoxide and biliverdin by heme oxygenase (HO; Hu et al., 2016). Iron overload in the brain subsequently generates abundant reactive oxygen species via the Fenton reaction (Hu et al., 2016), resulting in WMI. Iron oxidation has been shown to cause greater oxidative stress in the WM than GM in vitro (Hall et al., 2000). An intrastriatal injection of FeCl₃ in rats leads to remarkable axonal damage and demyelination (Zou et al., 2017), while iron chelators, such as minocycline (Zou et al., 2017) and deferoxamine (Gu et al., 2009; Xie et al., 2014; Ni et al., 2015), decrease iron accumulation, attenuate WMI, and improve neurological function in rodents and piglets. Furthermore, another metabolic by-product of hemoglobin, i.e., bilirubin, has been reported to cause structural and functional damage in the WM in vitro and in vivo through its oxidation products (Lakovic et al., 2014).

Iron-induced oxidative stress and inflammation injury can overlap to further aggravate WMI through the iron dependent c-Jun N-terminal kinase (JNK) pathway. JNK is a member of the mitogen-activated protein kinase (MAPK) superfamily, which has already been widely investigated due to its active actions in response to various stimuli, such as the exposure to inflammatory cytokines (Kumar et al., 2015). JNK is activated in ICH animal models primarily by iron, and iron chelator can reduce the free iron contents in the CSF, suppress JNK activation, reduce WMI and improve neurological deficits (Yatsushige et al., 2007; Wan et al., 2009; Ni et al., 2015; Chen et al., 2016; Zou et al., 2017). Furthermore, the inhibition of JNK could deactivate inflammation, attenuate brain edema and improve functional outcome after ICH (Ohnishi et al., 2007; Michel-Monigadon et al., 2010). Interestingly, similarly to MMP-9, the JNK pathways have been found to promote axonal regeneration after cytoskeletal disruption (Quintanilla et al., 2013; Valakh et al., 2015). Thus, iron is an ideal potential candidate for the treatment of WMI post-ICH due to its dual toxic roles in oxidative stress and inflammation.

Preclinical studies have revealed that the level of glutamate is up to eight times higher than the baseline value during the early period of ICH due to the influx of glutamate from the hematoma or through a disrupted BBB from circulation (Qureshi et al., 2003), and these values correlate with the brain edema and BBB disruption (Wu et al., 2013b). Moreover, the perihematomal glutamate level is associated with the postoperative outcome in ICH patients (Wu et al., 2013a). For hemiplegic patients with basal ganglia hematoma affecting the internal capsule, earlier minimally invasive surgery is accompanied by a lower level of peri-hematoma glutamate and a better prognosis, indicating that glutamate-related WMI may play a role in the functional outcome, and the early attenuation of glutamate-induced WMI may enhance recovery (Wu et al., 2013a). The increased glutamate activates glutamate receptors in oligodendrocytes and axons with intracellular Ca²⁺ overload, leading to oligodendrocyte death, demyelination and axon injury in MS, TBI, spinal cord injury and cerebral ischemia (Park et al., 2004; Matute et al., 2007; Matute, 2011; Zhang et al., 2013). Similarly, glutamate-mediated excitotoxicity is believed to play a detrimental role in WMI following ICH, although pertinent studies are lacking.

The molecular mechanisms of WMI after ICH are currently unclear. More thorough research regarding studies exploring the potential mechanisms underlying WMI are required to identify novel treatment targets. Oligodendrocyte differentiation and remyelination are vital for WM repair. The expression and function of either their negative regulators, including the Wnt/β-catenin signaling pathway (Dai et al., 2014; Lee et al., 2015a,b) and myelin-associated inhibitory factors, or positive regulators, such as the morphogen Sonic hedgehog (MSH; Ferent et al., 2013) and trophic factors such as Netrin-1 (He et al., 2013; Tepavčević et al., 2014), should be investigated in the context of ICH in future studies.

Accelerating axon regeneration is another option for restoring the integrity of WM. Myelin-associated inhibitors, such as myelin-associated glycoprotein (MAG), Nogo-A, and oligodendrocyte myelin glycoprotein (OMGP), can lead to growth cone collapse and inhibit axon regeneration (Muramatsu and Yamashita, 2014). However, while a wealth of new data regarding activating axonal repair after IS and other neurological diseases are available, this knowledge has not yet been applied to ICH. For example, the induction of growth and differentiation factor 10 (GDF10) in peri-infarct neurons can produce axonal sprouting and enhance functional recovery (Li et al., 2015), and cell-type-specific G protein-coupled receptor (GPCR) signaling essential for axonal guidance and targeting during development was recently found to promote axon regeneration after CNS injury (Li et al., 2016). By targeting modulators that are directly related to the physiological processes of axonal regeneration, new treatment can be developed.

In addition to oligodendrocytes, other cell types, including astrocytes, microglia and endothelial cells, are contained in the WM, and their interaction is essential for the maintenance of the integrity of the WM. Crosstalk among these cell components is required for oligodendrogenesis and remyelination after WM damage (Pham et al., 2012; Itoh et al., 2015; Miyamoto et al., 2015; Rosenzweig and Carmichael, 2015). Therefore, an understanding of the signaling pathways involved in intracellular crosstalk will add new knowledge that will be helpful for alleviating WMI and accelerating WM restoration. Astrocytes support oligodendrocyte function. Several soluble factors secreted by astrocytes have been implicated in stimulating oligodendrocyte differentiation and enhancing myelination. in vitro and in vivo experiments have demonstrated that astrocyte-derived brain-derived neurotrophic factor (BDNF) supports oligodendrogenesis and regeneration after WM damage (Miyamoto et al., 2015). Furthermore, leukemia inhibitory factor-like protein (LIF), signal transducer and activator of transcription 3 (STAT 3), ciliary neurotrophic factor (CNTF) and tumor necrosis factor receptor 2 (TNFR2) promote oligodendrocyte survival and maturation (Patel et al., 2012; Fischer et al., 2014; Monteiro de Castro et al., 2015; Domingues et al., 2016). Microglia also play a crucial role in remyelination by driving the phagocytosis of myelin debris and oligodendrocyte differentiation during CNS disorders. After ICH, microglia can be polarized into two phenotypes, i.e., the early pro-inflammatory M1 and the late anti-inflammatory M2 (Zhang et al., 2017). The switch from M1 to M2 can enhance oligodendrocyte differentiation, and M2 cell polarization is essential for efficient remyelination via M2 cell–derived activin-A after demyelination (Miron et al., 2013). Additionally, endothelial-oligodendrocyte interactions may be important for WM repair after stroke (Arai and Lo, 2009; Pham et al., 2012). The “oligovascular niche” was recently proposed as a microenvironment between cerebral endothelial cells and the oligodendrocyte lineage in which these cells exchange soluble factors, such MMP-9, BDNF and vascular endothelial growth factor (VEGF), to maintain WM homeostasis (Miyamoto et al., 2014). Under stroke conditions, oligovascular coupling is interrupted, which may contribute to WMI (Arai and Lo, 2009). A deeper understanding of the mechanisms of endothelial-oligodendrocyte trophic coupling may lead to new therapeutic approaches for ICH.

ICH damages both the GM and WM. It is hypothesized that the sole focus on neuronal protection is partially responsible for the failure of ICH treatments (NINDS ICH Workshop Participants, 2005). However, focusing only on WMI may also be fruitless. Therefore, treatments targeting both GM and WM are encouraged in the future. However, a single treatment is not easy to develop because there are significant distinctions in the cellular properties between the GM and WM. Therefore, developing agents with multiple pharmaceutic characteristics; combining therapeutic modalities, such as surgical hematoma evacuation, followed by drug therapy; or combining multiple agents with separate roles may be the management of choice.

Currently, to accelerate the translation from the bench to the clinic, several points should be noted. Regarding animal studies, rodents are the predominant species used in stroke research studies worldwide. However, the rodent WM comprises only 14% of the total brain tissue, whereas this proportion reaches 50% in humans (Zhang and Sejnowski, 2000; Rosenzweig and Carmichael, 2015). Other animal models, such as pigs, with poly-gyrated brains are needed, and nonhuman primates are even more ideal for assessing WMI. In addition, advanced MRI equipped with a multi-parameter measuring system is most advantageous for dynamically observing the integrity and microstructural changes in the WM throughout the whole period in living subjects and should be routinely incorporated into the evaluation of WMI after ICH. More importantly, more in vivo studies of WMI in patients are desperately needed because of the translational difficulty of the neuroprotection benefits observed in animal studies.

microRNA (miRNA) is an important posttranscriptional regulator of gene expression that has been implicated in the pathogenesis of many human diseases and has been suggested as a possible diagnostic and prognostic biomarker. Experimental studies showed a distinct miRNA profile after ICH, which was involved in inflammatory injury and apoptosis (Kim et al., 2014; Yang Z. et al., 2015; Yuan et al., 2015). Analyses of peripheral blood and hematoma samples from ICH patients further reveal that there are some specific miRNAs that are closely correlated to the enlargement of subsequent hematomas, WM edema and functional outcome (Zheng et al., 2012; Zhu et al., 2015; Wang J. et al., 2016; Wang M. D. et al., 2016). Furthermore, certain miRNAs have been regarded as candidate biomarker of WMI in certain neurodegenerative disorders (Gandhi, 2015; Kuswanto et al., 2015). For example, microRNA-137 is associated with the integrity of WM and neurocognitive functioning in patients with schizophrenia (Kuswanto et al., 2015). Therefore, because of their additional advantages of wide distribution, stability and relative ease of detection (Stylli et al., 2017), miRNAs may provide insight into the mechanisms of WMI and prognosis after ICH.

Magnetic resonance spectroscopy (MRS) allows for the noninvasive measurement of in vivo metabolites in a predefined brain region of interest. The metabolic changes in N-acetylaspartate and total creatines in the peri-stroke area are correlated with the WM loss (Yassi et al., 2016), and their ratio is a useful indicator for the early diagnosis of WMI and cognitive impairment in cerebral ischemic patients (Xing et al., 2013). Since the concentrations of N-acetylaspartate, total choline and total creatines are a reflection of axonal health, cellular membrane integrity and energy metabolism, respectively (Acheson et al., 2014), their metabolic activities in the WM after ICH might represent the corresponding axonal damage and demyelination. This hypothesis is supported in patients with TBI (Sinson et al., 2001; Grossman et al., 2015). The MRS technique could be beneficial for detecting WMI early before morphological changes occur.

Although MRS-based techniques can detect changes in most cerebral metabolites, their spatio-temporal resolution is often insufficient for routine examinations of fast-evolving and heterogeneous acute stroke lesions (Sun et al., 2011). Chemical Exchange Saturation Transfer (CEST) has emerged as a novel MRI technique for tissue pH imaging and imaging of various metabolites and is well suited for molecular imaging studies (Liu et al., 2013). Among the techniques of CEST MRI, pH-weighted and Amide Proton Transfer (APT) MRI have been implicated in IS as a surrogate of lactate acidosis and altered tissue metabolism (Sun et al., 2011; Tietze et al., 2014). Although CEST MRI is in its infancy in the field of stroke research, with the development of novel sequence and post-processing methods, it is hopeful that a better understanding of the mechanism of WMI following ICH will be obtained by discovering the subtle metabolic changes.

Cerebral microdialysis (CMD) is a well-established laboratory tool that provides an on-line analysis of the brain biochemistry via a thin fenestrated dialysis catheter that is inserted into the brain parenchyma. In addition to the monitoring of energy metabolism, CMD can dynamically observe excitotoxicity, inflammation and oxidative stress by measuring the levels of glutamate, metalloproteinases, cytokines and 8-iso-PGF₂, respectively (de Lima Oliveira et al., 2014). Moreover, axonal skeleton proteins, such as tau, can also be quantified using CMD (Magnoni et al., 2012). Hence, this technique has great potential in the study of the pathophysiology underlying the ICH-induced WMI.

The importance of WMI in ICH is increasingly being recognized, and abundant evidence has confirmed the intimate relationship between WMI and poor functional outcomes. Although certain pathological factors, including inflammatory injury, oxidative stress and neuro-excitotoxicity, are responsible for ICH-induced WMI, the definite underlying mechanisms need to be further elucidated. However, both inflammatory and oxidative signaling pathways are involved in WMI partially due to the iron overload after ICH. Therefore, treatments targeting iron may be quite promising. From this perspective, the results of the ICH Deferoxamine (iDEF) trial, which will be revealed next year, are highly anticipated. Regarding WM restoration, regulating factors promoting/inhibiting axonal regeneration and remyelination should be defined under the background of ICH. Oligodendrocyte proliferation and differentiation play a central role in remyelination which is involved in the complex crosstalk among oligodendrocytes, astrocytes, microglia and cerebral endothelial cells. An understanding of the cell-cell interaction mechanisms will likely promote oligodendrogenesis. Regarding WMI diagnosis, novel biochemical and neuroimaging markers are expected to detect to assist in the early detection of WMI and the dynamic monitoring of WMI. Finally, for better clinical translation, future experimental treatment strategy designs should provide protection to both GM and WM by targeting multiple pathological factors in poly-gyrated animals rather than using only rodents, and more in vivo studies of WMI in patients are desperately needed.