Introduction

Ischemic stroke is a major cause of morbidity and mortality worldwide [12, 33, 53]. Carotid artery disease contributes to 18–29% of all cases [59, 66]. Carotid invasive interventions such as carotid endarterectomy (CEA) and carotid angioplasty with stenting (CAS) are widely used to treat carotid artery disease [6, 26, 62]. One potentially severe complication is intracerebral hemorrhage (ICH) in the context of cerebral hyperperfusion syndrome (CHS) [37]. Clinical manifestations of CHS are diverse and include symptoms such as throbbing headaches, confusion, focal neurological deficits, partial or generalized seizures, among others [25, 27, 60]. Diagnostic examinations such computerized tomography scans (CT), magnetic resonance imaging (MRI), transcranial Doppler (TCD), single-photon emission computerized tomography (SPECT) and positron emission tomography (PET) can be used to confirm or exclude this diagnosis [3, 11, 24, 60]. The pathophysiology behind CHS is still poorly understood [41]. Three mechanisms have been associated with the syndrome: failure of brain vessels’ autoregulatory mechanisms to adapt to the sudden and deregulated increase in cerebral blood flow after carotid revascularization in long-standing hypoperfused brains due to severe stenosis/obstruction [2, 36, 48, 54, 60]; baroreflex disturbances secondary to carotid revascularization [58]; disturbances in the trigeminovascular reflex [35, 60]. Despite its severity, the knowledge about the frequency, risk factors and prognosis of ICH in the context of CHS is scarce. Previous reviews did not specifically address the occurrence of ICH in the context of CHS [32, 41]. Therefore, we performed a systematic review of the existent studies and meta-analytically estimated the frequency of ICH after CHS and its case fatality.

Materials and methods

Protocol and registration

This systematic review was registered at the PROSPERO database (CRD42016033190) and written in accordance with the PRISMA guidelines [31].

Eligibility criteria

Primary studies involving patients submitted to carotid revascularization (CAS or CEA) due to carotid occlusive disease were included. Studies with CEA/CAS performed for other specific conditions, case reports and animal studies, and studies in which the definition and frequency of ICH related CHS were not described were excluded.

Information sources

The search process was performed using the search engines PubMed and EBSCOHost (1986 to January 2016). Databases accessed via EBSCOhost include MEDLINE, sciencedirect, academic one file, J-stage, general one file, OAlster, expanded academic ASAP, China/Asia on demand, SciELO, Scitech conect, MedicLatina and Korean studies information study system. Only full-text English-written publications were included.

Search strategy and study selection

The Mesh terms “cerebral hyperperfusion syndrome,” “complications,” “carotid revascularization,” “endarterectomy” and “carotid angioplasty” were used to retrieve relevant literature. Studies were selected by two independent investigators. A consensus between the authors was used to resolve any disagreements about the inclusion of specific studies.

Data collection process

Studies were analyzed by two independent investigators. A consensus between the authors was used to resolve any disagreements about the inclusion of specific studies.

Data items

The following items were extracted: type of surgical procedure, frequency of CHS with ICH, associated risk factors and outcome (case fatality, morbidity rates).

Studies' risk of bias

The National Institutes of Health (NIH) tools were used [15] for quality assessment (supplement). Two reviewers performed the assessment independently. Discrepancies in the classifications were discussed and agreement achieved.

Planned methods of analysis

Qualitative analysis with quantitative description including all selected studies was performed whenever applicable. To address the risk factors, and for the meta-analysis of frequency and case fatality of ICH after CHS, only large studies (≥ 100 patients) were considered. This arbitrary threshold was selected to minimize the effects of substantial variability in the diagnostic criteria, time of evaluation and populations included in our study. We used Stata/SE 14.0 software for conducting the analysis and to derive forest plots. Random-effects meta-analysis weighted by the inverse-variance method was performed to estimate pooled frequency and 95% confidence intervals (CI). Heterogeneity was assessed with the I2 test. We used a random-effects model as substantial heterogeneity between studies results was expected. The limit for statistical significance was established at 0.05.

Results

Study selection

The initial search yielded a total of 545 manuscripts (423 publications at EBSCOhost and 122 publications at PubMed) (Fig. 1). After extraction removal of duplicates and studies not fulfilling our eligibility criteria, 41 studies were included in the final analysis (Fig. 1). Reasons for study exclusions are documented in supplement 1.

Fig. 1
figure 1

Flowchart of the article inclusion process in the systematic review

Study characteristics

A total of 28,956 participants were included, with study sample sizes ranging from 26 to 4494 participants (Table 1). Eighteen studies (44%) defined both CHS and ICH in the Methods section [1, 2, 7,8,9, 16, 19, 20, 22, 24, 29, 34, 42, 44, 46, 47, 61, 65]. The frequency of ICH after carotid intervention in studies with fewer than 100 participants ranged from 0% to 4.44% [4, 8, 14, 17, 18, 21, 22, 24, 40, 46, 48, 64]. In studies with 101 to 1000 participants, the range was from 0% to 2.21% [1, 2, 5, 7, 9, 10, 13, 19, 20, 23, 29, 30, 34, 39, 42, 43, 51, 55, 57, 65]. In studies that included 1001 or more participants, a range from 0.09% to 0.6% was found [16, 44, 45, 47, 49, 50, 52, 61, 63]. With regard to the quality evaluation, the majority were rated as “good” [1, 2, 5, 7, 8, 18, 22, 29, 30, 34, 50, 65] or fair [4, 9, 10, 13, 16, 17, 19,20,21, 23, 24, 39, 40, 42,43,44,45,46,47,48,49, 51, 55,56,57, 63, 64], and three were “poor” [14, 52, 61]. The risk of bias was considerable in most studies (Supplement 2) [4, 9, 10, 13, 16, 17, 19,20,21, 23, 24, 39, 42,43,44,45, 47, 49, 55, 64]. The pooled frequency of ICH in the context of CHS was 38% (95% CI: 26% to 51%, I2 = 84%) (Fig. 2).

Table 1 Frequency and case fatality of intracerebral hemorrhage in large and small studies in the context of hyperperfusion syndrome
Fig. 2
figure 2

Forest plot showing the frequency of intracerebral hemorrhage in the context of cerebral hyperperfusion syndrome in studies including at least 100 patients (point estimates and 95% CIs are shown along with pooled estimates)

Risk factors

Table 2 summarizes the risk factors found in larger studies (≥ 100 patients). Table 1 shows that when comparing CEA and CAS, the frequency of ICH in the context of CHS in large studies was higher after CAS, ranging from 0.28% to 4.05% [1, 7, 9, 19, 30, 39, 42, 47, 49, 55, 57, 65]. In CEA, it varied from 0% to 2.15% [2, 5, 9, 10, 13, 16, 18, 20, 23, 29, 34, 43,44,45, 47, 61]. Also, 73% of the studies involving CAS had a frequency of ICH in the context of CHS above 0.5% [1, 7, 9, 19, 30, 39, 42, 47, 49, 55, 57, 65]. In CEA, these frequencies occurred in only 26% of the cases [2, 5, 9, 10, 13, 16, 18, 20, 23, 29, 34, 43,44,45, 47, 61]. The “ICH to CHS proportion” was higher after CAS in comparison to CES: 7 out of 11 CAS studies (63.6%) had 50% or more hemorrhagic CHS (range 0–100%) [2, 5, 9, 10, 13, 16, 18, 20, 23, 29, 34, 43,44,45, 47, 61]. In CEA, 5 out of 12 studies (41.6%) had 50% or more cases of hemorrhagic CHS (range 0–80%) [2, 5, 9, 10, 13, 16, 18, 20, 23, 29, 34, 43,44,45, 47, 61]. Post-procedure ICH in asymptomatic patients was addressed in nine large studies [1, 7, 10, 13, 47, 49, 55, 57, 61] and occurred in three [47, 57, 61]. Overall, periprocedural hypertension is the most frequent risk factor, being documented in four studies [16, 20, 47, 50]. Three studies mention severe ipsilateral stenosis as a risk factor [50, 51, 63]. Younger age was considered a risk factor in two studies but denied as a risk factor in another two studies [16, 47, 51, 63].

Table 2 Risk factors for ICH in the context of CHS

Outcomes

The mortality from ICH related to CHS ranged from 0% to 100% [1, 8, 9, 16, 20, 21, 23, 30, 39, 40, 42, 45, 47, 49, 51, 55, 57, 61, 63,64,65]. In large studies the mortality was ≥ 50% in more than half of the studies (range 0 to 100%) [1, 10, 17, 21, 24, 31, 39, 42, 45, 47, 49, 51, 55, 57, 61, 63, 65]. The pooled case fatality of ICH after CHS was 51% (95% CI: 32% to 71%, I2 = 77%) (Fig. 3).

Fig. 3
figure 3

Forest plot showing the case-fatality rates from intracerebral hemorrhage in the context of cerebral hyperperfusion syndrome in studies including at least 100 patients (point estimates and 95% CIs are shown along with pooled estimates)

Discussion

This is the first systematic review and meta-analysis addressing the frequency, risk factors and outcome of ICH in the context of CHS. Despite being discussed since 1981 [54], no consensual definition exists for CHS, and its pathophysiology is still to be elucidated [28, 60]. We found a high variation in the frequency of ICH in the context of CHS, with a range of 0% to 4.44% [1, 2, 4, 5, 7,8,9,10, 13, 14, 16,17,18,19,20,21,22,23,24, 29, 30, 34, 39, 40, 42,43,44,45,46,47,48,49,50,51,52, 55, 57, 61, 63,64,65]. The overall case fatality associated with ICH in the context of CHS was high [1, 8, 10, 17, 21, 22, 24, 31, 39, 40, 42, 45, 47, 49, 51, 55, 57, 61, 63,64,65]. Of note, the two larger studies reported associated mortality varying from 25.93% to 57.14% [16, 47]. Important variation exists regarding mortality rates when considering all studies' information (0 to 100%). However, this may be explained by the inclusion of different study designs, sample sizes and the classification criteria used for ICH in the context of CHS . In large studies, the frequency of ICH was higher after CAS in comparison to CEA and was rare after asymptomatic carotid disease. The higher “ICH to CHS proportion” post-CAS CHS further supports the notion that patients with CHS after CAS are at increased risk of ICH [38]. The mandatory use of double antiplatelet therapy in CAS could contribute to this finding. Indeed, the use of antithrombotics was associated with the occurrence of post-CAS ICH in two small studies and in one large study based on administrative data [16]. Periprocedural hypertension and ipsilateral severe stenosis were the most common risk factors described [16, 20, 47, 50, 51, 63]. These data are relevant and stress the importance of pre- and post-carotid revascularization blood pressure control, particularly in patients with severe stenosis. One frequent bias found in the included studies was the lack of definition for CHS with ICH. However, the requirement of brain imaging for ICH diagnosis may minimize the impact of this bias for the overall comparison. The use of different methodologies and size discrepancies between the studies can also explain the variation in the frequency of ICH in the context of CHS [1, 2, 4, 5, 7,8,9,10, 13, 14, 16,17,18,19,20,21,22,23,24, 29, 30, 34, 39, 40, 42,43,44,45,46,47,48,49,50,51,52, 55, 57, 61, 63,64,65]. The lack of information regarding associated factors such as use of antithrombotics, time interval from the ischemic event to revascularization procedure and severity of chronic white matter disease represents a limitation when evaluating the occurrence of ICH in the context of CHS.

Conclusion

This systematic review and meta-analysis showed that ICH in the context of CHS is rare in large series, occurs more frequently after CHS secondary to CAS and than post CAS, and is generally associated with high case-fatality rates. The main risk factors are periprocedural hypertension and ipsilateral severe stenosis. Further studies to better describe the contribution of other risk factors are needed.