The main findings of this study are as follows: (1) SAH induced a significant increase of oxidative stress and evident brain edema, resulting in obvious brain damage in the rabbit SAH model; (2) Intraperitoneal administration of hydrogen-rich saline abated the level of oxidative stress and brain edema following SAH and, in the meanwhile, alleviated EBI following SAH. These findings suggest that hydrogen-rich saline could alleviate brain injury via decreasing oxidative stress and brain edema after SAH. Hence, we conclude that hydrogen-rich saline could be a potential therapeutic agent for EBI after SAH.
Oxidative stress arises from the strong cellular oxidizing potential of excess reactive oxygen species (ROS), or free radicals [28, 29]. Plenty of ROS are generated after SAH and considerable evidences have confirmed that oxidative stress, which is the focus of this study, is an important mediator of brain injury after SAH . Triggered by clot-derived hemoglobin(Hb), free radicals, including ·O-
2·, H2O2 and ·OH, explode subsequently. ·O-
2· is produced by hemoglobin auto-oxidation and consequent dismutation of two ·O-
2· forms H2O2, and the latter is the source of highly reactive ·OH in the reaction catalyzed by ferric ion . ·OH is the strongest of the oxidant species and reacts indiscriminately with nucleic acids, lipids and proteins to produce strong cytotoxic effect [20, 31]. The production of ·OH from extravasated hemoglobin , mitochondria disruption  and the disruption of the intrinsic antioxidant system  have all been proved in experimental or human SAH. It has been demonstrated that explosion of free radical species after SAH caused oxidative brain injury , and antioxidative treatments alleviated brain damage and improved the neurological deficits in a rat model .
Lipid peroxidation is mediated by superoxide, such as ·OH, ONOO-, as well as H2O2, resulting in structural alterations of membranes and functional impairment of cellular components. Malondialdehyde (MDA), the end product of lipid peroxidation of polyunsaturated fatty acids in cellular membranes, was identified as a reliable marker of oxidative stress, which reflects the damages of the cell indirectly . In agreement with previous reports [25, 38], we confirmed in the present study that experimental SAH elevated MDA significantly in the brain cortex after SAH. And the administration of hydrogen-rich saline decreased MDA dramatically in comparison with that in the SAH or SAH + saline group.
Caspases, or cysteine-aspartic proteases or cysteine-dependent aspartate-directed proteases are a family of cysteine proteases that play essential roles in apoptosis (programmed cell death), necrosis, and inflammation . Caspase-3 is one member of the caspase family and it plays a central role in the execution-phase of cell apoptosis. It has been reported that the expression of caspase-3 was intensified in the cortical neurons after SAH  and its inhibition could reduce the neuron loss. Caspase-12 is recognized to be essential for the endoplastic reticulum (ER) stress-induced apoptosis. The ER is sensitive to alterations in homeostasis from a variety of stimuli, such as glucose deprivation, perturbation of calcium homeostasis, and exposure to free radicals. Excessive stress to the ER results in apoptosis. It has been demonstrated that during apoptosis induced by ER stress, caspase-12 is localized to the ER and gets activated . Caspase-12 is considered as an upstream caspase that might activate caspase-3 , and calcium–calpain–caspase-12–caspase-3 cascade was also suggested to play a role in ER-mediated apoptosis . In the present study, we analyzed the level of caspase-3 and caspase-12 in the brain cortex following SAH. Our data showed that caspase-3 and caspase-12 increased significantly in the SAH and SAH + saline group but decreased dramatically in the SAH + hydrogen-rich saline group. These result are in line with the previous study in brain ischemia  and spinal cord injury . Therefore, we can postulate that ER stress plays a pivotal role in the neuron death after SAH, and it might be a promising filed to studying the concrete protective mechanism of hydrogen in vivo.
Delayed global edema has been shown to be an independent predictor of death . Cerebral edema, which can lead to increased intracranial pressure (ICP), brain herniation, irreversible brain damage or even death, is a significant clinical consequence following SAH. Brain edema is generally classified into cytotoxic or vasogenic edema . Cytotoxic edema is defined as a cellular swelling with fluid accumulating within the cell and with the typical instance of astrocyte swelling . Vasogenic edema is characterized by the breakdown of the BBB, resulting in the increased fluid accumulation around cells . It has been postulated that altered expression of water channels aquaporins (AQPs), development of BBB disruption , active substances derived from blood, as well as secondary events after SAH, such as raised ICP  and hypertension are involved in the pathogenesis of brain edema. Badaut et al.  found a marked increase in AQP1 and AQP2 protein expression 24 hours after SAH in the human neocortex . One of the early indications of brain injury after SAH is the alteration of BBB permeability . In patients with SAH, classic vasogenic edema, a direct result of BBB breakdown, has been proved, which has also been shown in experimental models . In addition, it has been proved that early dysfunction of BBB contributes to brain edema  which expands brain volume and prolongs elevated ICP values after SAH . Consequently, there is a resultant rise in ICP, which further reduces cerebral blood flow, leading to further ischemia . These bring more damage to the BBB. A vicious circle like this never ends. In this study, we found that the brain water content increased obviously after SAH and administration of hydrogen-rich saline abated brain edema significantly. This accords with what was reported in the brain trauma model by Ji et al. .
It is believed that oxidative stress is closely linked with brain edema via the discruption of BBB. Oxidative stress is considered as a major underlying cause of BBB injury. Oxidative stress-mediated disruption of BBB was shown in experimental models in vivo  or in vitro . It has been demonstrated that alcohol-induced loss of BBB integrity is associated with increased production of ROS . In an in vivo study, Katsu,et al. found that hemoglobin-induced oxidative stress contributes to matrix metalloproteinase activation and BBB dysfunction.
Hydrogen is a highly diffusible gas, which can easily penetrate the blood – brain barrier by gaseous diffusion. And hydrogen can also penetrate biomembranes smoothly to diffuse into the cytosol, nucleus and mitochondria. This is particularly important, as mitochondria, the primary site of generation of reactive oxygen species, is notoriously difficult to target. Ohsawa, et al.  reported that hydrogen can target intracellular sources of reactive oxygen species and inhibit reperfusion-induced oxidative damage by selectively scavenging ONOO- and ·OH, the strongest of the oxidant species which reacts indiscriminately with nucleic acids, lipids and proteins. Therefore, H2 could selectively react with the ·OH to produce water, and did not react with other ROS that possess physiological functions. In this aspect, H2 is evidently superior to some other antioxidant with strong reductive reactivity, which can increase mortality, possibly by affecting essential defensive mechanisms . This is immensely advantageous for medical treatment as the use of H2 would not bring serious side effects. Up till now, increasing evidence has proved that H2 can be used as an effective antioxidant therapeutic owing to its ability to decrease cytotoxic ROS [22, 25, 55]. In this study, we find that, the administration of hydrogen-rich saline decreased oxidative stress (MDA) and caspase-3/12 distinctly, which is in accordance with the previous reports. Meanwhile, hydrogen-rich saline abates the brain edema induced by SAH significantly. Correspondingly, the brain injury was alleviated in the SAH + hydrogen-rich saline group, and all the results in the present study were basically in accordance with the previous related reports [25, 61].
With regard to the animal model, till now, a number of SAH models have been applied, and three of them are commonly seen: the intracranial endovascular perforation model, the blood injection into the cisterna magna model, and the blood injection into the prechiasmatic cistern model. However, controversy exists regarding which method of selection is appropriate for different animals and the drawbacks of each of the models . Blood injection into the cisterna magna is the most frequently applied experimental SAH method in rabbits and both the one- and two-hemorrhage models are used to make rabbit SAH model. In our previous study, it was proved that the two-hemorrhage model in rabbits is more appropriate than the one-hemorrhage model for the research on SAH . Consequently, the two-hemorrhage rabbit SAH model was used in the present study. However, like other models, the two-hemorrhage rabbit SAH model could not imitate the clinical SAH ideally as well. None of the rabbits died till 72 hours following SAH and evident neurological deficits could rarely be observed. Admittedly, there are many differences between the two-hemorrhage rabbit SAH model in this study and the clinical SAH. Hence, objectively, the SAH model in this study still needs further improvement, though obvious brain injury was found following the two-time SAH.
In addition, there was no statistically significant difference in clinical evaluation of the rabbits among the groups in this study, while evident neurological improvement was observed in the hydrogen treated hypoxia–ischemia rat model . In effect, there are fewer behavior tests devised for the rabbit SAH model, and the differences in model and the scoring system could bring about inconsistent findings. As a result, we extrapolate that, to some extent, the neurobehavior evaluation system in the present experiment is not sensitive enough to discriminate the neurological difference in different groups. More sensitive neurobehavior tests for rabbit are necessitated in order to render the results more accurate and scientific, with which Hyojin Jeon et al. share the same opinion .