Clinical risk factors for CIE

We identified demographic similarities between patients in our series (Table 1) and prior reports. ^{Reference Quintas-Neves, Araújo and Xavier5} These include risk factors for preexisting microvascular disease, such as advanced age, uncontrolled preoperative arterial HTN, DM and CKD. ^{Reference Quintas-Neves, Araújo and Xavier5} Five patients in our series had prior stroke and 2 patients had connective tissue disorders which may predispose patients to pathological neurovascular manifestations. ^{Reference Kim, Brinjikji, Lanzino and Kallmes33} Previously reported risk factors for CIE include CKD, prior stroke, anticoagulant use and heart failure. The association between microvascular disease, ischemic stroke and BBBd is established. ^{Reference Teng, Pannell and Rennert26,Reference Kassner and Merali27,Reference Serlin, Ofer and Ben-Arie34–Reference Amantea, Russo and Gliozzi38} As such, we hypothesize that the aforementioned variables are also risk factors for CIE.

Clinical and radiographic findings associated with CIE

Clinical manifestations of CIE may include encephalopathy (39.6%), cortical blindness (39.6%), motor weakness (37.5%)and decreased LOC (20.8%). ^{Reference Quintas-Neves, Araújo and Xavier5} Aphasia, headache, seizures and neglect have also been reported. We identified a similar spectrum of symptoms (Table 3). Decreased LOC was observed in all cases. Symptom onset time was similar to prior reports. ^{Reference Quintas-Neves, Araújo and Xavier5}

CIE-related radiological findings typically include effacement of cortical sulci (37.5%), parenchymal edema (29.2%), cortical (54.8%) or subarachnoid (22.9%) contrast enhancement. ^{Reference Quintas-Neves, Araújo and Xavier5} These findings (Table 3) were also similar to those previously reported. ^{Reference Quintas-Neves, Araújo and Xavier5,Reference Chu, Lee and Chen7}

Prognosis following CIE

Clinical outcomes following CIE vary. ^{Reference Guimaraens, Vivas and Fonnegra3,Reference Quintas-Neves, Araújo and Xavier5,Reference Niimi, Kupersmith, Ahmad, Song and Berenstein9–Reference Leong and Fanning14,Reference Lantos39,Reference Nakao, Joshi and Hirose40} Following diagnostic procedures, most patients recover clinically and radiographically within days. ^{Reference Quintas-Neves, Araújo and Xavier5} While data is sparse, outcomes following intervention for AIS may be less favorable; preoperative neurological deficits were associated with postoperative progression and incomplete clinical recovery. Similarly, Chu et al. reported unfavorable clinical outcomes in 6 of 7 CIE patients treated with EVT for AIS. ^{Reference Chu, Lee and Chen7} No long term data is available for patients with CIE.

BBBd and CIE

The BBB is composed of microvascular endothelial cells, pericytes and astrocytic foot processes. ^{Reference MacLean, Kamintsky, Leck and Friedman24} Tight junctions bridging endothelial cells and the basement membrane are comprised of transmembrane (e.g. Claudin and Occludin) and cytoplasmic proteins (e.g. ZO-1); both play an integral role in regulating integrity and permeability of the BBB. ^{Reference Wang, Li and Zhao20} Several animal and human studies have implicated a mechanistic role for BBBd in CIE. BBBd may underly brain dysfunction and injury, either by direct disruption of brain homeostasis due to serum extravasation into brain neuropil, or indirectly via astroglia activation, endothelial dysfunction and pathological neuroinflammation (e.g. cytokine release). Below, we discuss the association between BBBd, microvascular disease, iodinated contrast medium, intraoperative HTN, neuroinflammation and iatrogenic endothelial injury as risk factors for CIE.

BBBd and iodinated contrast medium

Iodinated contrast agents do not typically cross the BBB. However, both ionic and nonionic iodine contrast media have been associated with CIE. Uchiyama et al. provided direct evidence of high iodinated contrast concentration in the cerebrospinal fluid of a patient with CIE, compared to controls. ^{Reference Uchiyama, Abe and Hirohata13} The mechanism of contrast extravasation is likely multifactorial. Iodinated contrast agents have been shown to reduce expression of endothelial tight junction proteins. ^{Reference Wang, Li and Zhao20} BBBd may also be related to a chemo-toxic effect of contrast on cerebral microvasculature, ^{Reference Salvesen, Nilsen and Holtermann41–Reference Chase and Kricheff43} particularly in the setting of slow CBF with prolonged contact of contrast on the endothelium. ^{Reference Uchiyama, Abe and Hirohata13,Reference Bassett, Rogers, Cherry and Cruzhit29,Reference Horwitz and Wener42} Alternatively, BBBd has been associated with iodinated contrast osmolality and volume. ^{Reference Bassett, Rogers, Cherry and Cruzhit29} Contrast hyperosmolality may induce BBBd by drawing water out of endothelial cells, with resultant cell shrinkage and separation of tight junctions. ^{Reference Uchiyama, Abe and Hirohata13} To this end, Wilson et al. compared mannitol to several nonionic, iodinated contrast agents varying in osmolality. They found that iodinated contrast media delivery resulted in BBBd in rabbits, whereas mannitol did not. ^{Reference Wilson, Evill and Sage18} These findings suggest that BBBd is not related to contrast osmolality alone and other physical or chemical effects of iodinated contrast, such as viscosity, may alter BBB function. ^{Reference Parker, Aboghazleh and Mumby52} Lastly, BBBd may also be related to contrast volume. ^{Reference Bassett, Rogers, Cherry and Cruzhit29} The median lethal dose LD₅₀ for commonly used iodinated contrast agents is 10–20 ml/kg. ^{Reference Junck and Marshall44} In our study, high volumes of media delivered, ranging from 210 to 580 mL (i.e. up to 8 mL/kg; approaching the LD₅₀), may have contributed to CIE.

BBBd in AIS: molecular mechanisms

An association between BBBd and cerebral microvascular disease has been demonstrated in animals, ^{Reference Schoknecht, Prager and Vazana25,Reference Mazumder, Giri, Kumar and Borah45} and humans. ^{Reference Bowman, Kaye and Quinn21,Reference Kassner and Merali27,Reference Serlin, Ofer and Ben-Arie34,Reference Lakhan, Kirchgessner, Tepper and Leonard36,Reference Sweeney, Sagare and Zlokovic46,Reference Jabbari and Vaziri47} In our series, and prior reports of CIE following EVT for AIS, ^{Reference Quintas-Neves, Araújo and Xavier5,Reference Aguirre, Trillo, Ximénez-Carrillo and Vivancos17} multiple preexisting patient risk factors for microvascular disease were identified. As such, we hypothesize that preoperative microvascular disease is a risk factor for both BBBd and CIE during intervention.

Disruption of pulsatile flow across endothelial cells in the setting of ischemic stroke has numerous pathological implications, including: (a) increased BBBd via decreased expression of tight junction proteins; (b) thrombosis-induced endothelial cell release of proinflammatory cytokines (e.g. TNF-alpha, IL-1-beta); (c) glial cell activation; (d) leukocyte recruitment with local release of cytotoxic molecules; and (e) decreased nitric oxide production by injured endothelial cells, which may compromise autoregulation of the microcirculation, particularly in sensitive penumbra. ^{Reference Teng, Pannell and Rennert26}

In the context of ischemic stroke, BBBd has also been associated with cortical spreading depolarization (CSD), ^{Reference Gursoy-Ozdemir, Qiu and Matsuoka48} seizures, ^{Reference Vazana, Veksler and Pell49} elevated levels of glutamate, ^{Reference Vazana, Veksler and Pell49} proinflammatory cytokines, ^{Reference Kassner and Merali27,Reference Amantea, Russo and Gliozzi38} increased levels of matrix metalloproteinases (MMPs) and neuronal injury. ^{Reference Wang, Li and Zhao20,Reference MacLean, Kamintsky, Leck and Friedman24,Reference Lakhan, Kirchgessner, Tepper and Leonard36,Reference Rempe, Hartz and Bauer37,Reference Gursoy-Ozdemir, Qiu and Matsuoka48} MMPs, including MMP-9, have an established role in modulating BBB function and structure; they degrade tight junction proteins and may induce paracellular leakage. ^{Reference Rempe, Hartz and Bauer37} MMPs are upregulated in response to acquired brain injury and play a role in the neuroinflammatory response. ^{Reference Cunningham, Wetzel and Rosenberg50} CSDs are common following AIS, ^{Reference Dreier51} trigger the release of MMPs, ^{Reference Parker, Aboghazleh and Mumby52} result in BBB breakdown, ^{Reference Takizawa, Qin and Lopes de Morais53} upregulate proinflammatory cytokines, ^{Reference Nakamura, Strong and Dohmen54} and contribute to ischemic lesion progression. ^{Reference Park, Hwang, Song, Suh and Lee55} While these factors may contribute to CIE following AIS, the extent of subarachnoid contrast extravasation among our cases suggest other mechanisms may also be involved.

Blood-brain barrier dysfunction in acute ischemic stoke: cerebral blood flow dynamics and iatrogenic endothelial injury

Endothelial cell damage may result from iatrogenic mechanical trauma, altered CBF dynamics or reperfusion injury. Mechanical endovascular stenting or clot extraction may induce device-specific iatrogenic endothelial trauma following treatment of AIS in animals. ^{Reference Gounis, Wakhloo and Chueh56,Reference Whisson, Wilson, Evill and Sage57} The use of such devices is increasing, particularly among the aging population, at longer time intervals since the onset of symptoms, with the goal of rapid and complete revascularization. ^{Reference Zaidat, Lazzaro and McGinley30,Reference Riina31} These factors may potentially increase the likelihood of vessel perforation, contrast extravasation and subarchnoid hemorrhage. In vitro and in vivo studies provide valuable insight regarding the pattern and degree of endothelial injury inflicted, based on device size and shape. ^{Reference Teng, Pannell and Rennert26} Furthermore, endothelial cell injury may vary based on vessel diameter, with greater endothelial cell injury in smaller vessels, emphasizing the important of appropriately matching device size based on vessel dimensions. ^{Reference Teng, Pannell and Rennert26} These findings highlight the importance of laboratory studies in an effort to design novel technologies with the goal of minimizing endothelial trauma, thereby preventing BBBd, neuroinflammation and secondary neuronal injury. ^{Reference Teng, Pannell and Rennert26}

In our series, all patients experienced intraoperative arterial HTN (Table 2). We propose that severe intraoperative HTN, particularly in the setting of preexisting cerebrovascular disease, is a risk factor for BBBd and CIE. Aguirre et al., reported a case of CIE with severe intraoperative arterial HTN during EVT, in a patient with multiple risk factors for microvascular disease. ^{Reference Aguirre, Trillo, Ximénez-Carrillo and Vivancos17} Whisson et. al., induced arterial HTN (190 mmHg peak systolic pressure) during carotid angiography in rodents and reported an association with increased BBBd. ^{Reference Ivens, Gabriel, Greenberg, Friedman and Shelef58} While it is standard to address intra-procedural blood pressure optimization and be mindful of contrast volume administered, our cases highlight the occurrence of these variables among patients with CIE.

All our patients with AIS had restoration of CBF to the affected vascular territory. A mechanism analogous to cerebral hyperperfusion syndrome (CHS) may have facilitated CIE, as both ischemia and reperfusion injury are known to induce BBBd. BBBd has been documented in animal models of cerebral hyperperfusion, ^{Reference Sakaki, Tsujimoto, Nishitani, Ishida and Morimoto22,Reference Amantea, Russo and Gliozzi38,Reference Chassidim, Veksler, Lublinsky, Pell, Friedman and Shelef59} as well as a clinical case. ^{Reference Chassidim, Veksler, Lublinsky, Pell, Friedman and Shelef59} The perfusion pressure breakthrough threshold in chronically ischemic brains is decreased in the setting of new ischemic insults and HTN, both resulting in BBBd. ^{Reference Sakaki, Tsujimoto, Nishitani, Ishida and Morimoto22} Reperfusion of ischemic brain regions may contribute to BBBd and CIE via mechanisms related to impaired autoregulation such as those observed with CHS.

Long procedures could potentially result in downstream ischemia, leading to BBBd and secondary CIE, particularly if a support platform is in place for a prolonged period and occlusive in a vessel. In our long cases, however, no contrast stasis was noticed during the injections, indicating a nonocclusive guiding catheter position in the neck vessels. Operative durations in the EVT cases were all under 200 minutes.

Future avenues for research: BBB imaging and CIE-directed therapies

A future cohort study would provide the opportunity for more comprehensive multi-variate analysis. The inclusion of a control group in future studies may allow for analyses assessing variables predictive of CIE. Dynamic-contrast enhanced MRI (DCE-MRI) has been used to describe the extent and localization of BBBd in patients with neurological conditions, including TIA and stroke. ^{Reference Serlin, Ofer and Ben-Arie34,Reference Yamasaki, Matsuura, Shozuhara, Onodera, Itoyama and Kogure60} DCE-MRI may be useful for the diagnostic work-up of patients with suspected CIE and serve as a biomarker for measuring CIE-directed treatment efficacy.

Previously described CIE-directed therapies include corticosteroids, vigorous hydration and blood pressure optimization. ^{Reference Quintas-Neves, Araújo and Xavier5} In the absence of standardized guidelines for the treatment of CIE, these options are reasonable supportive adjuncts. Airway protection, anticonvulsants and hyperosmolar therapy may be considered on an as needed basis. The optimal duration and intensity of such therapies warrant further study. The utility of such therapies likely depends on timely diagnosis and education around CIE, which may improve diagnostic recognition and facilitate early treatment.

Future studies may explore therapies for CIE that target pathological mechanisms underlying BBBd. These may include inhibition of proinflammatory cytokines, CSDs or MMP release. Inhibition using IL-1-receptor antagonists has been shown to attenuate brain damage induced by stroke, ^{Reference MacLean, Muradov and Greene61} whereas inhibition of MMP-9 activity prevents cytokine maturation and reduces ischemic brain volume. ^{Reference Amantea, Russo and Gliozzi38} Lastly, CSDs are associated with massive release of glutamate and increased BBBd. ^{Reference Takizawa, Qin and Lopes de Morais53} NMDA-receptor antagonism has been shown to inhibit CSD, ^{Reference Sakowitz, Kiening and Krajewski62,Reference Reinhart, Morton, Brennan, Carlson and Shuttleworth63} reduce BBBd, ^{Reference Vazana, Veksler and Pell49} and improve neuronal recovery following CSD in peri-infarct brain tissue.⁶⁴

Limitations

Given the small sample size, the analysis is limited to univariate, nonparametric comparisons. As described above, our future work includes conducting a cohort study with a control group to elicit pre- and peri-operative variables predictive of developing CIE. The time course for obtaining postoperative radiographic imaging was nonstandardized and driven by findings on neurological assessment in this retrospective series. This reflects lack of wide-spread consensus on the definition of CIE and absence of guidelines pertaining to the work-up and management.