A Review of Epidemiology, Risk Factors, Presumed Cause, Clinical Course, and Outcome
Background and Purpose—We systematically reviewed the literature on epidemiology, risk factors, presumed cause, clinical course, and outcome of perimesencephalic hemorrhage.
Methods—PubMed, Embase, and the Cochrane Library were searched until March 2016. Quality assessment was done by 2 authors independently. Pooled prevalence ratios and pooled odds ratios with 95% confidence intervals were calculated for data extracted from case–control studies.
Results—We included 208 papers. The incidence of perimesencephalic hemorrhage is ≈0.5 per 100.000 person-years, men are more often affected, and no risk factors were confirmed. Two decision analyses both found that a single, high-quality computed tomography angiography is the preferred diagnostic approach. Short-term complications, such as hydrocephalus or cranial nerve palsies, are rare, and usually transient, with the exception of acute symptomatic hydrocephalus necessitating treatment in 3% of patients. Lacunar infarcts in the brain stem were convincingly described in 4 patients only. Fatal rebleeding after installment of anticoagulation in the initial days after the hemorrhage was described in 1 patient. At long-term follow-up, death related to the hemorrhage has not been reported, disability is found in 0% to 6%, and neuropsychological sequelae are suggested.
Conclusions—A single, high-quality computed tomography angiography is the preferred diagnostic strategy. Short-term complications are rare and often transient. Long-term outcome is excellent with respect to disability and death, but high-quality studies focused at neuropsychological sequelae are needed.
In 1985, a subset of patients with subarachnoid hemorrhage (SAH) was identified with a pattern of hemorrhage centered immediately anterior to the midbrain without intraparenchymal or intraventricular extension, no source of bleeding on 4-vessel angiography, and a benign clinical course.1–3 We will summarize the epidemiology, risk factors, presumed causes, and associations for this so-called perimesencephalic hemorrhage (PMH) and describe the diagnostic approach, clinical course, and long-term outcomes. Throughout the review, we will make distinction between a perimesencephalic pattern of hemorrhage and a diagnosis PMH, which requires both a perimesencephalic pattern of hemorrhage and exclusion of an aneurysm.
Search Strategy and Selection Criteria
We conducted a systematic review and meta-analysis in accordance with Preferred-Reporting-Items-for-Systematic-Reviews-and-Meta-analyses guidelines.4 We searched PubMed, Embase, and the Cochrane Library with search terms perimesencephalic hemorrhage and pretruncal subarachnoid hemorrhage until March 2016. We included case-series as well as case-reports. Papers written in English, Dutch, German, French, Spanish, Portuguese, Italian, or one of the Scandinavian languages were reviewed. Papers in other languages were translated if the title or English abstract suggested additional value. We excluded articles of which full-text was not available, animal studies, and other types of articles than original contributions. We searched for full-texts in the libraries of 2 Dutch academic institutions and through online searches. If full-texts were unavailable, authors were contacted. Cross-checking of references was performed until no new articles were identified. We also cross-checked the search with the personal database of 1 author (G.J.E. Rinkel), which has been built prospectively by daily searches of Pubmed for the past 20 years with the terms subarachnoid hemorrhage [AllFields] OR aneurysm [AllFields] OR arteriovenous malformation [AllFields] OR perimesencephalic [AllFields] OR subarachnoid hemorrhage [AllFields] OR aneurysm*. The results of the search are given in Figure 1. A quality assessment was performed of the included case-series with ≥5 patients with PMH reporting on risk factors, diagnostic approach, short-term complications, and long-term outcomes (Table I in the online-only Data Supplement), using an adapted version of a recently used scoring system.5 Quality assessment was done independently by 2 authors (L.A. Mensing/K.G. Laban), and in case of conflicting results, the study was assessed by a third author (M.D.I. Vergouwen) and discussed afterward in a consensus meeting. A score of 11 to 15 was considered a high-quality study. To increase validity of the paragraphs on risk factors, diagnostic approach, short-term complications, and long-term outcomes, we only report proportions using the high-quality case-series. For the paragraphs on short-term complications and long-term outcome, we also describe the findings of case-reports separately. Local ethical committees approved individual studies. No additional ethical approval was required for this review and meta-analysis. Data supporting the findings of this study are available from the corresponding author on request.
If possible, data were pooled using Review Manager software (version 5.2), calculating pooled prevalence ratios with 95% confidence interval (95% CI) and pooled odds ratios (ORs) with 95% CI for data extracted from case–control studies. Heterogeneity was assessed with χ2 statistics, and results were considered heterogeneous at P<0.10.
The pooled proportion of patients with PMH within series of patients with spontaneous SAH was 6.8% (95% CI, 5.7%–8.0%; n=1637; Figure I in the online-only Data Supplement). Assuming an incidence of SAH of 9 per 100.000 person-years in population-based studies6 and taking into account that ≈12% of patients with aneurysmal SAH (aSAH) dies before reaching the hospital,7 the incidence of PMH can be estimated to be ≈0.5 per 100.000 person-years. A study in the Greater Cincinnati area indeed reported an incidence for PMH of 0.5 (95% CI, 0.3–0.7; n=431) per 100.000 person-years.8 This calculated and observed incidence is probably an underestimation because a substantial proportion of patients with PMH may not seek medical attention.9
Age and Sex Distribution
Estimates of age at time of PMH have been fairly consistent, with a pooled mean age of 53 years (range, 3–90; References in the online-only Data Supplement), which is not different from patients presenting with aSAH (pooled mean difference, −0.7 years [95% CI, −4.0 to 2.6; n=1539]; Figure II in the online-only Data Supplement). Three children with PMH have been described.10–12 Forty-two percent (95% CI, 36%–49%; n=223) of patients with PMH is female compared with 67% (95% CI, 65%–70%; n=1636) of patients with aSAH (Figure II in the online-only Data Supplement).
PMH Versus General Population
Three case–control studies, totaling 142 patients with PMH, investigated the prevalence of vascular risk factors in PMH, with the general population or patients admitted for acute radicular pain as a comparison group.13–15 Data were obtained by interview in 1 prospective study13 through chart review in a retrospective study14 and by questionnaire in another retrospective study.15 Compared with the controls, pooled ORs was 1.6 (95% CI, 0.8–3.2) for hypertension and 1.1 (95% CI, 0.4–3.7) for smoking (Figure III in the online-only Data Supplement).13–15 There was substantial heterogeneity in these analyses (range, 58%–79%), which may be caused by differences between the control groups and the different methods of data collection. Other potential risk factors were studied only in single studies, showing no difference in the occurrence of excessive alcohol consumption (OR, 0.8 [95% CI, 0.4–1.6]), diabetes mellitus (OR, 2.1 [95% CI, 0.9–5.2]), hypercholesterolemia (OR, 1.2 [95% CI, 0.8–1.7]), high triglycerides (OR, 1.3 [95% CI, 0.5–3.0]), and use of oral contraceptives (OR, 1.9 [95% CI, 0.4–8.0]).13–15 The tendency toward an increased prevalence of diabetes mellitus is a noteworthy finding because both a previous meta-analysis and a large cohort study found diabetes mellitus to be associated with a reduced risk of aSAH (References in the online-only Data Supplement).
PMH Versus aSAH
Five studies compared risk factors between PMH and aSAH in univariable analysis. Data were obtained by interview in 3 prospective studies8,13,16 and by chart review in 2 retrospective studies.17,18 Hypertension and smoking are established risk factors for aSAH.19 Compared with patients with aSAH, the pooled ORs for hypertension in patients with PMH was 0.5 (95% CI, 0.3–0.9) and for smoking 0.4 (95% CI, 0.2–0.8; Figure IV in the online-only Data Supplement).8,13,16–18 In both analyses, there was no heterogeneity, which supports the robustness of these findings. For other potential risk factors, only data from single studies were available, showing an OR for obesity of 0.7 (95% CI, 0.3–1.4)13 and for exertion at time of onset of 0.8 (95% CI, 0.4–1.4).20
In conclusion, no clear risk factor profile for PMH has been identified, and although definitive evidence is lacking, these data suggest that risk factors involved in aSAH do not play a role in PMH.
Criteria for a perimesencephalic pattern of hemorrhage on computed tomography (CT) are described in Figures 2 and 3. Using these criteria, the characterization of a perimesencephalic bleeding pattern on unenhanced head CT has a good interobserver (κ=0.79) and intraobserver (κ=0.80) agreement.22 Computed tomography angiography (CTA) or digital subtraction angiography (DSA) is often used to rule out an intracranial aneurysm. An aneurysm in the posterior circulation as a causative factor is found in 4.1% (95% CI, 2.0%–8.2%; n=171) of patients with a perimesencephalic bleeding pattern (Figure 4; References in the online-only Data Supplement).1,23 Because CTA has been reported to be false-negative in rare instances, some studies advocate ≥1 follow-up DSAs in all patients with a perimesencephalic bleeding pattern despite the many negative results (References in the online-only Data Supplement).23 We found 2 decision analyses in patients with PMH, both showing that a single CTA is the preferred diagnostic approach (References in the online-only Data Supplement). In case of new neurological deficits, a repeated imaging study could be considered to exclude a dissecting pseudo- or perforator aneurysm (References in the online-only Data Supplement). Magnetic resonance imaging (MRI) of the brain and craniocervical junction has no additional benefit for the detection of a bleeding source and the costs outweigh the benefits in the evaluation of PMH (References in the online-only Data Supplement).23,24
Clinical Features on Admission
Nine articles described clinical features on admission in 220 patients with PMH, of which 8 were retrospective chart reviews (References in the online-only Data Supplement). Onset of headache is instantaneous in 86% (95% CI, 79%–91%; n=117) but can also be more gradual, starting in minutes rather than seconds in the remaining 14% (95% CI, 9%–21%; n=117; References in the online-only Data Supplement) In 75% (95% CI, 62%–85%; n=52), headache is accompanied by nausea and vomiting.19,20,25 The intensity of the headache does not differ between patients with aSAH and PMH.20 No data are available comparing the duration of headache between PMH and aSAH. It is rare for a patient with PMH to lose consciousness at time of ictus. This is reported by ≈4% (95% CI, 1%–11%; n=75) of patients and lasts only a few seconds to minutes.1,20 On admission, 96% (95% CI, 91%–98%; n=141) has a normal level of consciousness (Glasgow Coma Scale=15 or Hunt&Hess I/II; References in the online-only Data Supplement). Transient focal symptoms, such as sensory symptoms, weakness, difficulties with walking, and speech arrest, are reported by 9% (95% CI, 2%–27%; n=23).20 An epileptic seizure was described in 5% (95% CI, 1%–15%; n=43), but the observer, symptoms, and moment of onset were not described sufficiently to conclude that these were genuine epileptic seizures.26,27
Cause of PMH
Support for a venous origin comes from radiological studies that assessed the venous drainage pattern in patients with PMH. We identified 9 such studies, totaling 224 patients with PMH (References in the online-only Data Supplement). A primitive variant of venous drainage directly into a dural sinus instead of via the vein of Galen is found in 1 or both hemispheres in half of PMH patients in comparison with 1 of 5 patients with aSAH.3,28 In 20 of 28 PMH patients with primitive drainage, the hemorrhage was on the same side as the primitive drainage.3,28 Further circumstantial evidence for a nonarterial source of hemorrhage comes from a CT perfusion study that found less compromised cerebral blood flow on admission in patients with PMH than in patients with aSAH.29 Because a CT perfusion change was found in all patients with aSAH, this contrast suggests a nonarterial source of PMH.29
Several case-reports suggest that venous hypertension may lead to PMH, either directly or indirectly (References in the online-only Data Supplement). Direct mechanisms suggested include a stenosis of the vein of Galen at its junction with the straight sinus, cavernous sinus thrombosis, venulitis causing thrombosis, and a spinal dural arteriovenous fistula (References in the online-only Data Supplement). Two case-series described the occurrence of PMH after physical exertion, in 9 of 23 patients and in 7 of 9 patients.20,30 The suggested pathophysiological substrate was that physical exertion results in venous hypertension, which increases intrathoracic pressure and hereby blocks venous return through the internal jugular vein. The subsequent increase in intracranial venous pressure may result in mechanical swelling of intracranial veins and venous or capillary breakdown.30 An argument against the pathophysiological reasoning that venous hypertension might lead to PMH is that PMH has been described during pregnancy in 2 patients but never during childbirth while childbirth should be considered as an event involving components of the Valsalva manoeuvre.31
Arterial causes associated with a perimesencephalic bleeding pattern other than an aneurysm have only been described in case-reports, such as rupture of a perforating artery, basilar artery dissection, a cerebral cavernous malformation, and a transient microaneurysm of the basilar artery (References in the online-only Data Supplement). Except for an intracranial aneurysm, none of these lesions have been reported in more than a few patients.
Short-Term Complications and Sequelae
Acute hydrocephalus, defined as enlarged ventricles on head CT, can be caused by a block in the circulation of cerebrospinal fluid at the tentorial hiatus caused by filling of all perimesencephalic cisterns with blood (Figure 3).27,32 Acute hydrocephalus is seen in 14% (95% CI, 10%–18%; n=277) of patients with PMH25,26 (References in the online-only Data Supplement) while symptomatic hydrocephalus, defined as hydrocephalus on head CT in combination with decreased level of consciousness necessitating drainage, occurs in only 3% (95% CI, 2%–6%; n=330; References in the online-only Data Supplement).11,23 In our experience, hydrocephalus usually becomes symptomatic within 24 hours after PMH onset.
Although it was initially described that delayed cerebral ischemia (DCI) does not occur in patients with PMH, several reports in the last 2 decades suggest that DCI can also occur after PMH (References in the online-only Data Supplement).23 However, in most of the reported patients the symptoms had a close temporal relationship with DSA and resolved within a few hours, which suggests that the symptoms were procedure-related ischemia instead of spontaneous DCI (References in the online-only Data Supplement). Nevertheless, we also identified 9 patients in the literature with symptoms suggestive of DCI where a temporal relationship with angiography was not evident (References in the online-only Data Supplement).23 In 4 of these patients, the articles left no doubt that the CT scans showed a perimesencephalic pattern of hemorrhage and that an underlying aneurysm was ruled out (References in the online-only Data Supplement).23 In the 5 other patients, the diagnosis could not be reconfirmed based on the data provided (References in the online-only Data Supplement). Two of the 4 patients with reconfirmed PMH developed a transient lacunar ischemic syndrome in the anterior circulation, with a lacunar infarction confirmed on CT in one of them (References in the online-only Data Supplement). In the other 2 patients, the type of ischemia could not be determined because of lack of detail on the clinical course or symptoms (References in the online-only Data Supplement).23 Three of the 5 patients in whom the diagnosis of PMH could not be reassessed developed a lacunar ischemic brain stem syndrome with infarction in the mesencephalon, pons, or medulla (References in the online-only Data Supplement). In a fourth patient, an asymptomatic small acute left paramedian pontine infarct was noticed on diffusion-weighted MRI 2 days after ictus (References in the online-only Data Supplement).
In conclusion, a convincing clinical picture of DCI supported by imaging has not been reported in any patient with PMH. A few patients with confirmed PMH developed a lacunar ischemic syndrome, but this is not congruent with the presentation of DCI in patients with aSAH (References in the online-only Data Supplement). On the basis of the current evidence, it is uncertain whether the lacunar ischemic syndrome is a coincidence, caused by a missed basilar artery dissection, or is directly related to PMH. Angiographic vasospasm has been noticed in 9% of patients with PMH (95% CI, 6%–14%; n=198), but in none of them, this was accompanied by the clinical syndrome of DCI (References in the online-only Data Supplement).
In-hospital rebleeding has been reported in 3 patients23,26,33 of the >1220 patients with PMH described in the literature to date. The rebleeding was convincingly documented in only 1 patient.33 This 57-year-old woman experienced an increase of headache 5 days after the hemorrhage and 3 days after starting treatment with aspirin and intravenous heparin for an acute coronary syndrome, with a head CT showing an increased amount of extravasated blood compatible with recurrent PMH.33 She died the same day after developing ventricular fibrillation. Autopsy was not permitted. The second patient turned out to have an undiscovered vertebrobasilar aneurysm, making the rebleeding not a consequence of PMH, but of an undetected aneurysm.26 For the third patient, we cannot confirm rebleeding because the head CT on admission and after suspected rebleeding were not presented.23 In conclusion, rebleeding hardly ever occurs after PMH. If possible, the use of anticoagulant drugs should be avoided in the acute phase.
In a series of 27 patients with PMH, 10 patients (37% [95% CI, 22%–56%]) reported an episode of amnesia a few minutes to a few hours after ictus, lasting between 2 hours and ≈5 days (median, 2 days).34 The occurrence of amnesia is associated with enlargement of the temporal horns on initial CT scan.34
In 5 studies with 196 patients with PMH, we identified 5 patients who died shortly after being diagnosed with a PMH (References in the online-only Data Supplement).8,23,33,35 One of these 5 patients is the patient described above with rebleeding shortly after initiating antithrombotic medication.33 In 2 patients, multiple comorbidities contributed to death, including acute pulmonary embolism and end-stage renal failure in combination with a failed kidney and pancreatic transplant (References in the online-only Data Supplement).8 In 2 other patients, the diagnosis could not be reconfirmed based on the data provided using the head CT criteria specified in Figure 2.23,35 In conclusion, early death after PMH is exceptionally rare and if it occurs is usually not related to PMH.
Cranial nerve palsies after PMH can be caused by direct contact of the nerve with blood in the subarachnoid space. In a series of 148 patients with PMH, 9 (6% [95% CI, 3%–11%]) experienced anosmia: 2 reported a complete recovery after 8 to 12 weeks, whereas in the other 7 patients, smell had not recovered completely after a mean follow-up of 9 years (References in the online-only Data Supplement). Case-reports have documented a transient oculomotor nerve palsy resolving 3 days to 6 months after PMH, a transient trochlear nerve palsy resolving after 2 weeks, and a persistent bilateral sixth nerve palsy 6 years after the hemorrhage (References in the online-only Data Supplement). In 1 patient with a third nerve palsy, diffusion-weighted MRI showed a hyperintensity in the right midbrain and pons, indicating a lacunar infarction (References in the online-only Data Supplement). In 5 other patients with a third or trochlear nerve palsy, MRI did not show an ischemic lesion compatible with the affected cranial nerve (References in the online-only Data Supplement).
Recurrent PMH after discharge has been reported in 2 patients.21,36 One patient had a recurrence 5 months after the index event. However, the initial admission head CT showed more blood in the third and fourth ventricle than only sedimentation which is incompatible with a PMH (Figure 2).36 The patient with recurrent hemorrhage after 12 years had an initial episode compatible with PMH at the age of 48 years.21 The second episode was also compatible with PMH, and extensive diagnostic studies, including CTA, cerebral and spinal DSA, MRI of the brain, and cervical spine with and without gadolinium, did not show a cause of the hemorrhage. The patient had an uneventful neurological clinical course also after the second episode and was discharged home in good condition.21 In a large case-series of 160 patients with PMH and a mean follow-up of 7.5years, none of the patients (95% CI, 0.0–2.3) had a new episode of SAH after discharge (References in the online-only Data Supplement). In conclusion, rebleeding after discharge is exceptional.
At hospital discharge, 99% (95% CI, 94%–100%; n=92) of patients with PMH has a modified Rankin Scale score of 0 to 2.23,24 At follow-up, ranging from 0.8 to 90 months, 94% to 100% had a modified Rankin Scale score of 0 to 2 (References in the online-only Data Supplement). Because of variation in duration of follow-up between and within the included studies, we could not pool data on disability at follow-up. A suboptimal modified Rankin Scale score at long-term follow-up was mostly explained by subjective residual complaints, such as headaches, dizziness, fatigue, irritability, and forgetfulness (References in the online-only Data Supplement). In addition to the sequelae described above, some recent studies suggest that general health perception and vitality may be reduced, with neuropsychological deficits, depression, anxiety, fear of rebleeding, and impaired return to previous work position (Table II in the online-only Data Supplement; References in the online-only Data Supplement). The significance of these findings is not completely clear because much variation between studies is found, screening tools differ between studies so that results cannot be pooled, the number of patients per study is relatively small, and in all but 1 small study,16 no comparisons are made with a representative control group. In addition, only 4 studies used a validated assessment tool (References in the online-only Data Supplement).16 Uncertainty exists if the above-mentioned symptoms are different from those in a representative control group, related to the hemorrhage itself, or result from a post-traumatic stress syndrome related to the experience of sudden illness leading to hospital admission (References in the online-only Data Supplement).
In a cohort of 160 patients with a follow-up of 1213 patient-years, 11 patients died, but in none it was related to direct consequences of the PMH (References in the online-only Data Supplement). The standardized mortality ratio was 0.6 (95% CI, 0.3–1.1; References in the online-only Data Supplement).
Based on the data described above and our clinical experience, we propose the following treatment strategy. The cornerstone of treatment of patients with PMH is symptomatic care. In the rare patients with symptomatic hydrocephalus, it can be treated by (repeated) lumbar punctures. However, the decline in consciousness usually is only mild, and symptoms resolve spontaneously (References in the online-only Data Supplement). There is no indication for therapy aimed at prevention of DCI (nimodipine) or rebleeding. The use of anticoagulant drugs in the acute phase should be avoided, if possible. The low risk of complications makes forced bed rest or restriction of activities unnecessary. Patients can be discharged 24 hours after ictus because the risk of symptomatic hydrocephalus then is extremely small. However, in our practice, patients with PMH often remain in hospital for 2 to 4 days until the headache has decreased. Because of the normal life expectancy, no restrictions should be imposed on patients with PMH by physicians or health- or life-insurance companies (References in the online-only Data Supplement).
Conclusions and Future Directions
In conclusion, no risk factors were confirmed for PMH, men are affected, and patients typically present with a normal level of consciousness. DSA or repeated CTA can be withheld, provided that the initial unenhanced CT is made within 3 days of onset, the strict criteria of perimesencephalic pattern of hemorrhage are adhered to, and there is a high-quality negative CTA as judged by the neuroradiologist. Short-term complications after PMH are rare, and usually self-limiting, with the exception of rare instances of acute hydrocephalus necessitating lumbar puncture. Whether the reported instances of lacunar infarcts in the brain stem are a chance finding or related to the PMH is uncertain. Anticoagulation should be withheld in the initial days after PMH because it can induce fatal rebleeding. Long-term outcome is excellent with respect to disability and death, especially compared with aSAH.
The most important research questions include the cause of PMH and whether these patients are at risk of long-term cognitive sequelae. For the latter research question, new studies are needed with sufficient numbers of patients (>50), a standardized follow-up, predefined and validated questionnaires, and a control group from the general population. If PMH is associated with long-term sequelae, further studies should sort out whether these are a direct consequence of the hemorrhage or result from failing coping strategies for an acute event.
Guest Editor for this article was Giuseppe Lanzino, MD.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.117.019843/-/DC1.
- Received October 27, 2017.
- Revision received February 7, 2018.
- Accepted March 14, 2018.
- © 2018 American Heart Association, Inc.
- Rinkel GJ,
- Wijdicks EF,
- Vermeulen M,
- Ramos LM,
- Tanghe HL,
- Hasan D,
- et al
- van der Schaaf IC,
- Velthuis BK,
- Gouw A,
- Rinkel GJ
- de Rooij NK,
- Linn FH,
- van der Plas JA,
- Algra A,
- Rinkel GJ
- Linn FH,
- Rinkel GJ,
- Algra A,
- van Gijn J
- Schievink WI,
- Wijdicks EF,
- Piepgras DG,
- Nichols DA,
- Ebersold MJ
- Mensing LA,
- Ruigrok YM,
- Greebe P,
- Vlak MH,
- Algra A,
- Rinkel GJ
- Boerboom W,
- Heijenbrok-Kal MH,
- Khajeh L,
- van Kooten F,
- Ribbers GM
- Lindekleiv H,
- Sandvei MS,
- Romundstad PR,
- Wilsgaard T,
- Njølstad I,
- Ingebrigtsen T,
- et al
- Linn FH,
- Rinkel GJ,
- Algra A,
- van Gijn J
- Rinkel GJ,
- van Gijn J,
- Wijdicks EF
- Brinjikji W,
- Kallmes DF,
- White JB,
- Lanzino G,
- Morris JM,
- Cloft HJ
- Canneti B,
- Mosqueira AJ,
- Nombela F,
- Gilo F,
- Vivancos J
- Hop JW,
- Brilstra EH,
- Rinkel GJ
- Rahme R,
- Vyas NA