Subarachnoid hemorrhage (SAH) is a neurologic emergency with high rates of mortality and morbidity.1 Non-traumatic SAH comprises 3% of all stroke types, most commonly associated with the rupture of an aneurysm (80%).1,2 The annual incidence of aneurysmal SAH is 9/100 000 people in the United States and with approximately 600 000 cases worldwide.1 Mortality approaches 50%, and it has a reduction in health-related quality of life, which has been reported to occur in 35% of patients 1 year after SAH, with only one-third of patients demonstrating full recovery after treatment.1,2

Patients can deteriorate days after SAH because of delayed cerebral ischemia (DCI), one of the most feared complications of aneurysmal SAH (aSAH).3 DCI occurs in approximately 30% of patients with aSAH and accounts for 30%–40% of their deaths and permanent severe neurologic deficits.4,5 Although vasospasm is a significant contributor to DCI, there is a discrepancy between angiographic findings and clinical signs of stroke.6 Clinical, laboratory values, ultrasound parameters, and radiological findings have been proposed as risk predictors of this complication.7

Early identification of patients at high risk of developing DCI allows better clinical decision-making for the length of hospital stay and management to prevent irreversible brain damage.8 For that reason, many studies have developed several grading scales.4,9 These scales include mostly clinical predictors such as clot thickness, poor clinical admission status, aneurysm location, and aSAH management.4,9 Other studies have tested cerebral vascular reactivity and serum biomarkers, such as kallikrein.10,11 On the other hand, anatomical variants and other radiological findings could play a major role in the pathophysiology of DCI not described in the literature before. Therefore, this study aimed to establish these features as potential risk factors of DCI in this retrospective cohort.

The study comprised 60 female patients (73.2%), and the mean age of the sample was 62.15 (±13.11). Most aneurysms were located in the anterior circulation (95.1%). Neurological status at presentation was overall good, with more than half of patients having low WFNS grades (I–II; 60.97%), HH scores (I–II; 53.65%), and GCS 14 (IQR 11–15). Most aneurysms underwent surgical clipping (62.2%), and only 3 patients (3.6%) received only medical management because these patients died before being taken to surgical management, these data were collected until they died.

Vasospasm was diagnosed in patients by either CTA, DSA, or TCD ultrasonography, with 37.8%, in this cohort. DCI occurred in 30 patients (36.6%), and the mean day of presentation was 6.89 (±4.59). There was a good correlation between vasospasm and DCI, as this coexisting phenomenon occurred in 16 of 30 patients (53.3%). Twenty-five patients (30.5%) had hydrocephalus, 20 patients (24.4%) had an intracerebral hemorrhage, and 14 patients (17.1%) had global cerebral edema on admission CT. In TCD, the highest peak systolic blood flow velocity and mean blood flow velocity were 296 and 199.92 cm/sg, respectively, while the lowest values reported in the sample were 25.48 and 14.76 cm/sg for the middle cerebral artery in 62 patients. Twenty patients had an inadequate window for the middle cerebral artery.

Clinical characteristics and ultrasound parameters of patients with and without DCI are compared in Table 1. Smoking history, high values of systolic blood pressure, GCS, global cerebral edema, intracerebral hemorrhage, hydrocephalus, angiographic vasospasm, type of intervention, diameter>10 mm of the aneurysm, the presence of anatomic variants, peak systolic velocity, and mean flow velocity were all associated with DCI (Table 1). In particular, patients with DCI had higher values peak systolic flow (median, 89.95 [IQR 61.75–138.45] vs median, 62.74 [IQR 41.59–90.79]), and mean flow velocity (median, 52.46 [IQR 32.43–79.37] vs median, 36.35 [IQR 22.74–60.68]) (Fig. 1).

Fig. 1.

Box-plot of peak systolic velocity and mean flow velocity corresponding to delayed cerebral ischemia. In the box-plot, the central rectangle spans the first quartile to the third quartile (interquartile range). A bold segment inside the rectangle shows the median and “whiskers” above and below the box showing the minimum and maximum; outside box-plot points are outliers.

(0.04MB).

Further analysis of neurological outcome at discharge revealed that 12 characteristics were involved with this outcome: age, neurological status at admission (GCS, HH scale, WFNS, and MFS), hydrocephalus, DCI, the presence of an anatomical variant, peak systolic velocity, mFV, and pulsatility index (Table 2). On univariate logistic regression, systolic blood pressure, the presence of anatomic variants, global cerebral edema, intracerebral hemorrhage, and hydrocephalus had a statistically significant association with the development of DCI (Table 3). In multivariate analysis, none of these variables remained significant.

In this cohort, the anatomical variants were associated with the development of DCI and poor neurological outcome. The distribution of anatomical variations in this cohort, was as follows: 1 patient (1.21%) had a persistent trigeminal artery (PTA), 6 patients (7.3%) had either agenesis or hypoplasia of the A1 segment of anterior cerebral artery, 9 patients (10.9%) had either a complete or partial fetal posterior cerebral artery, and 3 patients had a hypoplastic vertebral artery (3.7%) (Fig. 2). Table 4 shows the patients affected by DCI, with anatomical variants, the respective arterial territory involved, and mRS at discharge.

Fig. 2.

Anatomic variations in our sample, 3D-CTA (A) and (B) A. Absence of an A1 segment of the right anterior cerebral artery (white arrow). B. Right posterior fetal cerebral artery (black arrow). The P1 segment is absent. DSA image of the right internal carotid and right vertebral artery (C) and (D). C. Trigeminal persistent artery arising from the cavernous segment of the internal carotid (Black arrow), posterior communicating artery, (orange arrow). D. Hypoplasia of the basilar artery (black arrow).

(0.36MB).

What allowed us to confirm that it were indeed anatomical variants and not vasospasm secondary to SAH, were the imaging and TCD indirect findings, particularly bony structures, vessel's diameters, and Soustiel index (“modified Lindegaard” index) (for basilar artery). In case of hypoplastic vertebral arteries (HVA) (2 cases), we compared the size transverse foramen of C6 and C2 with the contralateral one, and also estimated the diameter difference of the artery (which was 2.05 mm). Regarding the only case of hypoplastic basilar artery (HBA), we measured the artery's diameter (1.92 mm), and we obtained a normal Soustiel index by TCD means, confirming that it corresponds to anatomical variants.

DCI is one of the main complications in patients with SAH and carries a poor prognosis if not detected on time.20 DCI was mainly thought to be caused by cerebral vasospasm; however, studies in humans and animal models have since supported the notion of a multifactorial development of DCI (including cerebral vascular dysregulation, multifocal microthrombosis, cortical spreading depolarizations, delayed cell apoptosis, and neuroinflammation).21 As a result, there is a need to find other screening and managing factors, to offer the best possible individualized treatment and prevention strategies tailored to each patient.

In this study, we performed a retrospective analysis of a 4-year retrospective cohort, which analyzed clinical features, radiological, and anatomical factors involved in the risk of DCI. We found that smoking history and high values of systolic blood pressure on admission were significantly associated with DCI. Only smoking history has been proved to have a strong level of evidence as a risk factor for DCI, according to the systematic review performed by Rooji et al.22 As a unique finding of our study, 4 neuroradiological factors were statistically significant associated with DCI, namely, hydrocephalus, intracerebral hemorrhage (ICH), global brain edema, and anatomical cerebral arterial variants. To the best of our knowledge, anatomical cerebral arterial variants have never been described in the literature as risk factors of DCI.

Two clinical scales have been validated for the prediction of DCI.4,9 Both models took into account some clinical characteristics such as the WFNS, the amount of cisternal and intraventricular blood on CT, HH scale, mFS, and age.4,9 Radiological features, different from mFS like the Barrow neurological institute scale (BNI) (OR, 3.4; 95% CI 2.1–5.3) and the subarachnoid hemorrhage early brain edema score (SEBES) (OR, 2.24; 95% CI 1.58–3.17) have also been tested in their respective cohorts as significant predictors of DCI.23,24

De Rooij et al. performed the most recent systematic review for DCI predictors.22 In this systematic review, acute hydrocephalus had strong evidence for an association with DCI (OR, 1.3; 1.1–1.5 and OR, 2.6; 1.2–5.5).22 Some authors like Black et al. hypothesize that hydrocephalus is correlated with vasospasm by diminishing the clearance of vasoactive substances from the blood.25 Others, like Van Asch et al., demonstrated that hydrocephalus induces a reduction in cerebral blood flow of the deep gray matter and periventricular white matter, compromising cerebral perfusion at the expense of increased intracranial pressure (ICP).25 Fugate et al. reported that aggressive CSF drainage at low levels of ICP (e.g., 5 mmHg) improves blood flow in the microcirculation and tissue perfusion, making plausible a relationship between CSF and development of DCI.26

ICH is an uncommon complication of aSAH, but it might be related to DCI because it compromises cerebral blood flow.27 Qureshi et al. review identified several clinical and experimental studies that showed mild hypoperfusion of the ipsilateral hemisphere with ICH during the first 48 h, making brain tissue vulnerable to ischemia.27 Mechanical compression of surrounding vasculature by the hematoma has also been suggested as a mechanism for ischemia.27 In a similar manner to hydrocephalus, ICH provokes an increase of ICP, affecting cerebral blood perfusion and promoting vasospasm by subarachnoid extension of the bleeding.28

Although many cerebral arterial variants may be asymptomatic, their recognition is important, as some of these might be pathological and play an important role in the planning of neurosurgical procedures.15 The anatomy of the brain arterial system can influence the development of vascular diseases like aneurysms, dolichoectasia, atherosclerosis, stroke, and, as we propose in this paper, DCI.29 Developmental variants of the cerebral circulation can be fenestrations and duplications, variants of the circle of Willis, persistent carotid-basilar anastomoses, and other vascular anomalies identified in the skull base.30 The most frequently reported persistent variant fetal arteries in adults include the persistent trigeminal artery (PTA) (0.1%–0.5%), hypoglossal artery (0.1%), proatlantal artery (0.020%), and otic artery (0.001%).31 There are some few but important caveats that are worth-mentioning regarding some of these vessels and how they are related to cerebrovascular disease. Fetal posterior cerebral artery (FPCA), PTA, and hypoglossal arteries increased the likelihood of aneurysm formation in-situ or distant and rupture due to turbulent flow and increased wall shear stress.29 They have also been associated with ischemic cerebrovascular disease (either transient ischemic attack or ischemic stroke), and also with the presence of other important vascular arterial malformations (FPCA, A1 segment hypoplasia, or azygos anterior cerebral artery).30,32–35 PTA arises from the cavernous ICA in the region where it leaves the carotid canal and joins the basilar artery through the sella turcica, normally regressing.36 Both PTA and fetal posterior cerebral artery supply the posterior cerebral territory via the anterior cerebral circulation.37

Patients with these variants and concomitant carotid disease cannot develop leptomeningeal anastomoses between cerebral arteries, making them vulnerable to ischemia due to hemodynamic failure.38 In the case of anomalies of the anterior circulation, there is an interhemispheric collateral circulation failure and a deficient supply of the distal brain.32,39 Finally, another interesting point may arise here, although it must be clarified that this is only a pure speculative theory from the authors. We believe that, in certain way, the sole presence of fetal arteries or other anatomical arterial variations may have a basal and inherent risk of aneurysm formation, and, possibly, a secondary susceptibility to DCI in SAH patients, due to a diminished “vascular compensatory capacity”, which is inherent to a limited-vessel number and a “high monoarterial-dependency” in the circle of Willis, and that it might be depicted as a “double effect” or “double hit” phenomenon. As a matter of fact, this is just a supposition from the author's personal view, and we believe that it may be difficult to prove it in a clinical trial.

Large hemispheric infarction (10%) is defined as a massive middle cerebral infarction (MCI), with or without the involvement of the anterior and posterior cerebral artery.40 It is associated with substantial morbidity and mortality due to the high chance of developing malignant cerebral edema (50%).40 A lesion 82 ml size or greater in diffusion-weighted imaging (DWI) or an Alberta Stroke Program Early CT Score (ASPECTS) <6 are indicative of LHI.40 Patients with anatomic variations and DCI had large territory infarctions that frequently involved 2 cerebral artery territories, such as the case of the fetal posterior cerebral arteries. Although patients did not undergo diffusion studies to measure the lesions, most patients with ischemia had borderline ASPECTS for LHI (Table 4), and 1 patient had an LHI (Fig. 3) in our sample. For this reason and due to the involvement of multiple arterial territories, patients with anatomic variations had a worse neurological outcome than those without.

Fig. 3.

Hemispheric cerebral infarction in our sample, Axial CT scan (A), (B), and (C). Images show hypoattenuation of the total right MCA and ACA territory, with effacement of the adjacent sulci and loss of the gray-white matter distinction.

(0.12MB).

The most recent and updated meta-analysis of TCD for the diagnosis of vasospasm in the middle cerebral artery shows that both conventional and color Doppler are likely to detect vasospasm, but neither is useful to exclude one.42 In the case of DCI, TCD was also proved to have a high sensitivity (90%) (95% confidence interval [CI] 77%–96%), and predictive-negative value 92% (95% CI 83%–96%).41 However, TCD is not a mandated standard of care in aSAH due to the paucity of evidence on clinically relevant outcomes, and there is a need for high-quality for randomized trials.41 In our study, none of the parameters evaluated by TCD were statistically significant in the logistic regression; however, the mean values for systolic peak flow and mean flow velocity in the group with DCI compared to those who did not develop it was higher as depicted in the boxplots (Fig. 1).

Some of the characteristics associated with the development of DCI was also associated with an adverse neurological outcome, namely GCS at admission, hydrocephalus on admission, the presence of an anatomic variation, PSV, and MFV. This accounts for the validity of our data as DCI is also associated with a bad neurological outcome.

The limitations of the present study included its single-center retrospective nature, with inherent bias. Only 62 patients had a temporal window on TCD for assessment of the middle cerebral artery, limiting the statistical analysis and results of the TCD parameters. The small sample size limits the ability to find differences between groups on the multivariate analysis compared to other larger prospective cohorts. Additionally, most of the patients had a low WFNS score, which may correspond to information bias. Studies with higher evidence (prospective cohorts and randomized control trials) need to address all of the information presented here. Regarding the role of hydrocephalus and DCI, we did not make any Evans index adjustments for gender and age, as reported previously in the literature.43 Although our study had some limitations, it introduces the association of novel neuroradiological features (anatomical cerebral arterial variants), and the assessment of DCI in a Latinamerican sample.

Intracerebral hemorrhage, hydrocephalus, global cerebral edema on admission, and anatomic variations of the circle of Willis were significantly associated with DCI development in this sample of patients with aSAH. Patients with anatomic variations were more likely to present worse neurological outcomes because of a more severe compromise of the infarcted territories. Further larger prospective randomized studies are needed to validate the importance of these radiologic features and anatomic variations as risk predictors of DCI.

Not applicable.

In the present study we don't require informed consent, the authors don't include case details, other personal information or images of patients.

According to what is established in resolution 008430 of 1993, it is established that this investigation is Minimum Risk, taking into account that in the development of this, retrospective documentary investigation techniques and methods will be used and no intentional intervention or modification of the biological variables will be carried out. Physiological, psychological, or social conditions of the individuals participating in the study.

Orrego-González analyzed and interpreted the data, collected study data, and drafted and revised the manuscript for important intellectual content. Vasquez revised the manuscript for important intellectual content. Soto-Moreno revised the manuscript for important intellectual content. Roa-Wandurraga originated the idea for the study, designed and conceptualized the study, analyzed and interpreted the data, collected study data, and drafted and revised the manuscript for important intellectual content.