ORIGINAL ARTICLES

The Clinical Course of Perimesencephahc Nonaneurysmal Subarachnoid Hemorrhage Gabriel J. E. Rmkel, MD, Eelco F. M. Wijdicks, MD, Marinus Vermeulen, MD, Djo Hasan, MD, Paul J. A. M. Brouwers, MD, and Jan van Gijn, M D

We studied the early clinical course of 65 patients with perimesencephalic (nonaneurysmal) subarachnoid hemorrhage. None of the patients rebled; none had delayed cerebral ischemia; and only 3 patients (5%) developed clinical signs of acute hydrocephalus, 2 requiring ventricular shunting. Hyponatremia and electrocardiographic changes were found in the same proportions as in patients with aneurysmal rupture. All patients had a good outcome after 3 months. TO control for the influence of the relatively small amount of cisternal blood in perimesencephalic hemorrhage on the absence of delayed cerebral ischemia, we compared these 65 patients with 49 patients who had aneurysmal subarachnoid hemorrhage. This control group with aneurysms was selected according to a similar amount of cisternal blood, a comparable level of consciousness on admission, and absence of other factors known to increase the risk for delayed cerebral ischemia. We found that 4 of the 49 patients with aneurysmal hemorrhage developed delayed cerebral ischemia (Fisher's exact test, p = 0.03); therefore the relatively small amount of blood does not account for the absence of delayed cerebral ischemia in perimesencephalic hemorrhage. Patients with a perimesencephalic pattern of hemorrhage and a normal angiogram should be considered to have a distinct subset of subarachnoid hemorrhage and should be excluded from future treatment trials of patients with subarachnoid hemorrhage. Rinkel GJE, Wijdicks EFM, Vermeulen M, Hasan D, Brouwers PJAM, van Gijn J. The clinical course of perimesencephalic nonaneurysmal subarachnoid hemorrhage. Ann Neurol 1991;29:463-468

Patients with perimesencephalic subarachnoid hemorrhage are radiologically distinguished from other patients with subarachnoid hemorrhage by extravasation of small amounts of blood, mainly in the midbrain cisterns (Fig l), and a normal angiogram 111. In our first report, on 13 patients with perimesencephalic hemorrhage, none of these rebled and none developed delayed cerebral ischemia (DCI) ll]. To validate these findings in a much larger series, we studied the clinical features of 65 patients with perimesencephalic hemorrhage. As thick collections of cisternal blood on early computed tomograms (CT) after aneurysmal hemorrhage are associated with DCI 12-41, the absence of DCI in patients with perimesencephalic hemorrhage might be explained by a lesser degree of hemorrhage from the outset. In addition, the invariably normal level of consciousness in these patients may favor outcome as well. Therefore, we also compared the outcome events in patients with perimesencephalic hemorrhage with those in a subgroup of patients with a ruptured aneurysm, that is, with a normal level of consciousness on admission, similar amounts of subarachnoid blood on

From the Department of Neurology, University Hospitals, Utrecht and Rotterd;un, The Netherlands.

Received Jul 13, 1990, and in revised form Oct 8 and Nov 8. Accepted for publication Nov 11, 1990.

CT, and no other known risk factors for the development of DCI. Patients and Methods Patients for this study were selected from consecutive series of patients with subarachnoid hemorrhage who were admitted betweenJanuary 1982 and January 1989to the University Hospitals in Utrecht, Rotterdam, and Amsterdam and to the Westeinde Hospital in The Hague. Criteria for the diagnosis of perimesencephalichemorrhage were (1) headache coming on within seconds or at most a few minutes, ( 2 ) CT scans showing blood mainly or only in the cisterns around the midbrain, and (3) no detectable aneurysm on four-vesselangiography {I). Only patients who underwent CT scanning within 72 hours of the first clinical symptoms were included. The control group was drawn from a prospectively collected, consecutive series of patients with symptoms and signs of subarachnoid hemorrhage, confirmed by C?' scanning within 72 hours, and with a proven aneurysm on a.ngiography or autopsy. The amount of subarachnoid blood on CT was graded as previously described {5, 61. Each of the ten basal cisterns and fissures (frontal interhemispheric fissure, quadrigeminal cistern, and the paired suprasellar cisterns, ambient cisterns, basal sylvian fissures, and lateral sylvian fissures) was graded separately on a semiquantitative scale, ac-

Address correspondence to Dr Rinkel, University Department of Neurology, Heidelberglaan 100, 3584 CX Utrecht, The Netherlands'

Copyright 0 1991 by the American Neurological Association 463

tive complications were not included in the comparison for this study. The relative size of the lateral ventricles (the actual bicaudate index divided by the upper limit for age) was computed for all patients to assess the presence of ventricular enlargement as described previously [ 5 ] . All patients in both series were under continuous observation and a n y clinical deterioration was carefully documented and evaluated by C T scanning. DCI was diagnosed if (1)new focal signs gradually developed and (2) CT excluded other causes. Rebleeding was diagnosed if (1) there was a sudden aggravation of headache or a decrease in level of consciousness, with or without new focal signs; and (2) CT showed an increase of subarachnoid, intracerebral, or intraventricular blood, in comparison with the previous scan.

Fig I . Perimesencephulic hemorrhagee,with thick clots of blood in the interpedznculur and ldt ambient cistern. without extension to the frontul interhemispheric or sylvian fi.lsures.

cording to the amount of extravasated blood: 0, no blood; 1, small amount of blood; 2, moderately filled with blood; 3, completely filled with blood. The density of the clot was not measured. Clots that had expanded the original size of a cistern or fissure were still graded as 3. The total amount of subarachnoid blood (total sum score) was calculated by adding the ten scores and could range from 0 to 30. The grading scale for the amount of blood in the four ventricles was constructed in a comparable fashion, as follows: 0, no blood; 1, sedimentation of blood in the posterior part; 2, partly filled with blood; or 3, completely filled with blood. The total amount of intraventricular blood (total sum score) was the total of the four scores and could range from 0 to 12. We selected those patients with aneurysmal rupture in whom the amount of cisternal blood was comparable to that of the patients with perimesencephalic hemorrhage, that is, with a sum score of cisternal blood of not more than 16, which was the highest sum score observed for perimesencephalic hemorrhage. In addition, we excluded all patients with aneurysmal hemorrhage who had entry characteristics or subsequent treatments or events associated with an increased risk of DCI: (1) a decreased level of consciousness on admission f2}; (2) intraventricular hemorrhage on CT, exceeding sedimentation of blood in the posterior parts of the ventricles { 2 ) ; (3) treatment with antifibrinolytic or antihypertensive drugs or with fluid restriction 17-11}; ( 4 ) patients with a rebleed within 14 days; and ( 5 ) early surgery. Clipping of the aneurysm, if feasible at all, was carried out on the twelfth day after the hemorrhage presented in most patients. Postopera-

464 Annals of Neurology Vol 29 No 5 May 1991

Results Sixty-five patients fulfilled the criteria for perimesencephalic hemorrhage. Twenty-five were women and 40 men, with ages ranging from 22 to 70 years (mean, 53 years). Two patients had a history of hypertension. Only 4 patients briefly lost consciousness at the time of the ictus. Neurological findings on admission were unremarkable, except for neck stiffness. One patient had a transient third nerve palsy, which resolved within the first week.

Clinical Course in Perimesencepbulic Hemowbage Only 4 patients deteriorated after the hemorrhage, but in none could a rebleed or DCI be incriminated. Two patients developed a gradual decrease in the level of consciousness, associated with enlarged ventricles on a repeated CT scan. Both made a complete recovery after ventricular shunting. The clinical course of 1 of these 2 patients, who developed small nonreactive pupils and an upgaze palsy, is described more fully in a separate report [12). Another patient developed an upgaze palsy and gait apraxia in the first week, concomitantly with enlargement of the ventricles on a repeat CT scan, but in this patient the symptoms resolved spontaneously. A fourth patient, a right-handed 54year-old man, being treated for hypertension and diabetes, suddenly experienced weakness of the right arm and a right facial droop in the second week after the hemorrhage, which disappeared within 9 hours. His level of consciousness remained unaltered during this episode and his language was not affected. A repeat CT scan 1 week after the episode was normal. We diagnosed a transient ischemic attack in the territory of a small penetrating artery, probably unrelated to the previous hemorrhage. One patient deteriorated during angiography. He became less alert and on examination he was disoriented and had slight weakness of the left arm. The symptoms resolved within 4 hours. A repeated CT scan 1 week after the episode was normal, but a magnetic resonance image (MRI) 4 weeks later showed a

right occipital infarct. On neurological examination after 3 months, again no hemianopia could be found. Of the remaining patients, one noticed a drift of her left arm on awakening on the tenth day after the bleeding. Her symptoms disappeared within 30 minutes, and weakness could not be confirmed on examination. The clinical course in the other patients was uneventful. All 65 patients were seen as outpatients after 3 months; all had a good outcome, as measured on the Glasgow Outcome Scale { 131.

I

Systemic Disorders in Perimesencephalic Hemorrhage Thirty-five patients were enrolled in a prospective hyponatremia study, in which two of the four hospitals participated C7). Serum sodium levels were measured at least three times a week. Ten patients (29%) developed hyponatremia. It was mild (sodium level of 130 to 134 mmol/liter) in 5 , moderate (125 to 129 mmoli liter) in 4, and severe (120 to 124 mmol/liter) in 1. In all patients the hyponatremia developed between the first and seventh day after the hemorrhage. The mean duration of hyponatremia was 7 days (range, 3 to 11 days). In 32 patients an electrocardiogram (ECG) was made on admission. Fourteen of these were normal. In 7 patients (22%) ischemic changes were found (ischemic ST segment, ischemic T wave, or pathological Q wave), in 6 patients (19%) there were signs of left ventricular hypertrophy, and in 6 patients there were nonspecific changes. In 7 patients in whom ECGs were recorded daily for the first 12 days after the hemorrhage [14], transient ECG changes were found in all but 1. Ischemic changes were found in 5, signs of left ventricular hypertrophy in 2, supraventricular extrasystoles in 3, bundle-branch block in 2, and prolongation of the QT interval in 2 patients. Sinus tachycardia, long PR interval, and isoelectric T wave were found in 1 patient each. Comparison with Aneurysmal Hemorrhage Of the 66 patients with aneurysmal hemorrhage, 17 had a rebleed (in 1 patient 3 days after ventricular drainage for acute hydrocephalus). This left 49 patients with aneurysmal hemorrhage who fulfilled all criteria for inclusion in the control group. The aneurysm was located at the posterior communicating artery in 25 of these patients, at the anterior communicating artery in 11, at the middle cerebral artery in 7, at the posterior circulation in 4 , and at the pericallosal artery in 2 patients.

C T Scan and Angiographic Featares CT scanning was performed within 24 hours of the ictus in 27 patients with perimesencephalic hemorrhage (52%) and in 26 of the patients with aneurysmal hemorrhage (53%), between 24 and 48 hours of the

of patients

5-8 9 -12 13-16 1-4 sum scores of cisternal blood Fig 2. Sum scores of cisternal blood in 65 patients with perimesencephalic hemowhage (open bars) and in 49 patients with aneurysmal subarachnoid hemowhage (dotted barsj. (See text for explanation of grades.)

ictus in 18 patients with perimesencephalic hemorrhage (35%) and in 11 patients with aneurysmal hemorrhage (23%), and between 48 and 72 hours in 7 patients with perimesencephalic hemorrhage (14%) and in 12 patients with aneurysmal hemorrhage (25%). Four-vessel angiograms were repeated in 18 patients, in 2 patients because of slight narrowing of the basilar artery. In none of the patients was an aneurysm or severe vasospasm of the posterior circulation found. In perimesencephalic hemorrhage, the sum scores of cisternal blood ranged between 1 and 16. The sum scores of cisternal blood in rhe patients with perimesencephahc hemorrhage and in the control series with aneurysmal hemorrhage were similarly distributed (Fig 2). All cisterns were adequately visualized. The proportion of patients with thick clots of blood (grade 3) in one or more basal cisterns and fissures was higher in the group with perimesencephalic hemorrhage (309%) than in those with aneurysmal rupture (996). In patients with perimesencephalic hemorrhage the thick clots were invariably located in the suprasellar and ambient cisterns, in 2 patients with extension into the basal sylvian fissure (Fig 3). In the parients with aneurysmal rupture, the clots were equally common in the central cisterns and rhe more peripheral fissures. In 10 patients with perimesencephalic hemorrhage (15%) the relative size of the lateral ventricles was above the upper limit for age, the highest value exceeding this limit by a factor of 1.37. In 6 patients with aneurysmal rupture (1295) the relative size of the

Rinkel et al: Clinical Course of Perimesencephalic Hemorrhage 465

% filled with clots of blood

total

central cisterns

f issu res

Fig 3. Percentage of basal cisterns with thick clots of blood (grade 3 ) in 65 patients with perimesencephalichemorrhage (open bars) and in 49 patients with aneugwnal hemorrhage (dotted bars). Ten cisterns were distinguished in each patient. The total percentages were firrther subdivided into five central cisterns isuprasellar right and left, ambient right and left, and quadrigeminall and the five more peripherally locatedj.fi.r.rures iinterhemispheric, basal sylvian, and lateral sylvian).

lateral ventricles was abnormal, the highest value being 1.25.

Clinical Course in Aneurysmal Hemorrhage Three patients deteriorated due to CT-proved acute hydrocephalus. Two patients developed focal signs during angiography; one of them was left with a residual hemiparesis. Two other patients developed epilepsy and 1 patient developed an oculomotor palsy associated with a posterior communicating aneurysm. Four patients had secondary deterioration judged to be caused by DCI. The total sum scores of cisternal blood in these patients ranged from 4 to 15. (CT scanning had been performed within 24 hours in 3 of these patients and between 24 and 48 hours in 1.) All these patients progressively deteriorated to drowsiness or stupor together with focal motor signs. In 3 patients repeated CT scanning revealed multiple hypodense lesions. Two patients remained incapacitated, 1 patient died of a fatal rebleed, and 1 patient recovered after successful clipping of a middle cerebral artery aneurysm. Of the remaining 45 patients, 36 had a good outcome after 3 months. Three patients died after surgery, 2 were dependent on others, and 4 were independent but disabled. The difference in incidence of DCI in both series (0165 versus 4/49)is statistically significant (Fisher’s exact test,$ = 0.03). 466 Annals of Neurology Vol 29 No 5 May 1991

Discussion The most striking findings in this large series of patients with perimesencephalic hemorrhage were the complete absence of rebleeding and DCI, the low incidence (55%) of symptoms due to acute hydrocephalus, and the excellent outcome in all patients after 3 months. The absence of DCI in patients with perimesencephalic hemorrhage, however, might be simply explained by a lesser degree of cisternal blood on initial CT scan. The increased risk of DCI in patients with large amounts of blood on CT has been confirmed in several studies [2, 15, 161. In particular, localized, thick clots of blood have been incriminated {3, 41. Therefore, we compared the 65 patients with perimesencephalic hemorrhage with a subgroup of patients with aneurysmal hemorrhage, selected on the basis of a similar amount of subarachnoid blood and absence of other well-established predictors of DCI. We found that in this selected series of patients with aneurysmal hemorrhage, DCI did occur, despite a larger proportion of thick clots of blood on the baseline CT scan in the group with perimesencephalic hemorrhage. The absence of DCI could not be explained by absence of acute hydrocephalus in patients with perimesencephalic hemorrhage, as the incidence and degree of acute hydrocephalus were similar in both series of patients. One patient with perimesencephahc hemorrhage had transient focal deficits, but the mode of onset strongly argues against DCI. The hemiparesis started abruptly and without a concomitant decrease in the level of consciousness, whereas in DCI the onset of the focal signs tends to be gradual, and in the vast majority of patients this i s accompanied or preceded by a decreased level of consciousness {17). In fact all our 4 patients with DCI after aneurysmal hemorrhage demonstrated this stereotyped clinical pattern. The ischemic symptoms of this single patient should probably be attributed to atherosclerotic small-vessel disease, which is further supported by the presence of hypertension and diabetes. However, we cannot definitively exclude the possibility of a sudden and brief manifestation of vasospasm. The location of the blood remained the only apparent difference between the two series, apart from the presence of a demonstrable aneurysm. Perimesencephalic hemorrhages have little or no extension to the chiasmatic, interhemispheric, or sylvian cisterns, which are typically filled in aneurysmal hemorrhage. This difference, however, is not a likely explanation for the absence of DCI in patients with perimesencephalic hemorrhage; in the series of Kistler and colleagues, focal clots in the interpeduncular or ambient cisterns were associated with severe vasospasm on angiography and corresponding clinical deterioration [4).More-

over, in the present control series of patients with aneurysmal hemorrhage, thick clots were equally common in the central and the peripheral cisterns and fissures. Some recent studies reported DCI in patients with subarachnoid hemorrhage and a normal fourvessel angiogram. In Barlow’s series, 3 of 50 patients developed DCI r18). However, the site of the blood on CT scan in these 3 patients was different from that in patients with perimesencephalic hemorrhage: only intraventricular blood, or clots in the frontal interhemispheric and left sylvian cisterns, or no blood at all. In other series of nonaneurysmal subarachnoid hemorrhage, with DCI reported in up to 3 1% 1191,the distribution of subarachnoid blood was not mentioned 119-21) or CT scanning was not performed [22). Therefore, the origin of the hemorrhage in these series is uncertain. Another explanation for the absence of DCI in perimesencephalic hemorrhage could be that other, recently identified predictors of DCI and poor outcome such as hyponatremia 15, 23) and ECG changes 114) are less frequent in patients with perimesencephalic hemorrhage. However, we found that both the incidence and the degree of hyponatremia in the present series of patients with perimesencephalic hemorrhage were similar to those in two recent studies on patients with aneurysmal rupture 17, 231, in which the hyponatremia was associated with DCI. Also, the ECG changes found in patients with perimesencephalic hemorrhage are comparable with the changes found in a recent prospective study on ECGs in patients with aneurysmal rupture [14}. The most important difference berween the two subsets of patients therefore seems to be the origin of the subarachnoid hemorrhage. We speculated earlier, on the basis of the radiological features, that a venous or capillary source might be responsible in patients with perimesencephalic hemorrhage 111. In that case the cisternal blood in patients with perimesencephalic hemorrhage may lack a spasmogenic factor that is associated with a ruptured arterial aneurysm. This factor might be a component of arterial blood that is absent in venous blood, or it might be directly derived from the damaged aneurysm wall. Despite the arguments in favor of a venous or capillary origin, the cause of perimesencephalic hemorrhage has not been elucidated. Thanks to the excellent outcome in these patients [24}, postmortem studies are not available, and the possibility of finding the offending vascular abnormality seems remote thus far. Although pathological proof is not available, we conclude that perimesencephalic subarachnoid hemorrhage probably is a distinct clinical entity, not associated with rebleeding or DCI. The implication is that patients with perimesencephalic hemorrhage should be excluded from treatment trials in patients with sub-

arachnoid hemorrhage. Also, treatment with calcium entry blockers or other measures to prevent ischemia seems unnecessary. Another practical issue is the need for a potentially risky angiogram in patients with a perimesencephalic hemorrhage demonstrated on CT. At present we still recommend a baseline four-vessel angiogram. A recent study conducted to test the ability of experienced neuroradiologists to discriminate this perimesencephalic pattern from aneurysmal patterns on CT failed to show perfect agreement or a maximal predictive value (0.95) of a perimesencephalic pattern of hemorrhage for a normal angiogram (Rinkel GJE et al, unpublished data). We thank J. Th J. Tans and L. Hageman for providing us with the CT scans and clinical details of their patients with penmesencephahc hemorrhage, and H P Adams, Jr, for his frlutful discussion.

References 1. van Gijn J, van Dongen KJ, Vermeulen M, Hijdra A. Pee-

2.

3.

4.

5.

mesencephalic hemorrhage: a nonaneurysmal and benign form of subarachnoid hemorrhage. Neurology 1985;35:493-497 Hijdra A, van Gijn J, Nagelkerke N, et al. Prediction of delayed cerebral ischemia, rebleeding and outcome after aneurysmal subarachnoid hemorrhage. Stroke 1988;19:1250- 125 6 Adams H P Jr, Kassel NF, Torner JC, Haley EC Jr. Predicting cerebral ischemia after aneurysmal subarachnoid hemorrhage: influences of clinical condition, CT results, and antlfibrinolytic therapy. Neurology 1987 ;37 :15 86- 159 1 KistlerJP, Crowell RM,Davis KR, et al. The relation of cerebral vasospasm to the extent and location of subarachnoid blood visualized by CT scan: a prospective study. Neurology 1983; 33:424-436 van Gijn J, Hijdra A, Wijdicks EFM, et al. Acute hydrocephalus after aneurysmal subarachnoid hemorrhage. J Neurosurg 1985;

63:355-362 6. Hijdra A: Brouwers PJAM, Vermeulen M, van Gijn J. Grading the m o u n t of blood on computed tomograms after subarachnoid hemorrhage. Stroke 1990;21:1156-1161 7. Wijdicks EFM, Vermeulen M, Hijdra A, van Gijn J. Hyponatremia and cerebral infarction in patients with ruptured intracranial aneurysms: is fluid restriction harmful? Ann Neurol 1985;18:137-140 8. Wijdicks EFM, Vermeulen M, Ten Haaf JA, et al. Volume depletion and natriuresis in patients with a ruptured intracranial aneurysm. Ann Neurol 1985;18:211-216 9. Vermeulen M, Lindsay KW, Murray GD, et al. Antifibrinolytic treatment in subarachnoid hemorrhage. N Engl J Med 1984; 311:432-437 10. Hasan D, Vermeulen M, Wijdicks EFM, et al. Effect of fluid intake and antihypertensive treatment on cerebral ischemia after subarachnoid hemorrhage. Stroke 1989;20:15 11-1515 1I. Wijdicks EFM, Vermeulen M, Murray GD, et al. The effects of treating hypertension following aneurysmal subarachnoid hemorrhage. Clin Neurol Neurosurg 1990;92:111-1 17 12. Rinkel GJE, Wijdicks EFM, Ramos LMP, van Gijn J. Progression of acute hydrocephalus in subarachnoid hemorrhage: a case report documented by serial CT scanning.J Neurol Neurosurg Psychiatry 1990;53:354-3 5 5 13. Jennett B, Bond M. Assessment of outcome after severe brain damage: a practical scale. Lancet 1975;1:480-484

Rinkel et al: Clinical Course of Perirnesencephalic Hemorrhage 467

14. Brouwers PJAM, Wijdicks EFM, Hasan D, et al. Serial electrocardiographic recording in aneurysmal subarachnoid hemorrhage. Stroke 1989;20:1162-1167 15. Mohsen F, Pomonis S , Illingworth R. Prediction of delayed cerebral ischemia after subarachnoid haemorrhage by computed tomography. J Neurol Neurosurg Psychiatry 1984;47: 1179- 1202 16. Pasqualin A, Rosta L, Da Pian R, et al. Role of computed tomography in the management of vasospasm after subarachnoid haemorrhage. Neurosurgery 198415 :344-3 53 17. Hijdra A, van Gijn J, Stefanko S, et al. Delayed cerebral ischemia after aneurysmal subarachnoid hemorrhage: clinicoanacomic correlations. Neurology 1986;36:329-333 18. Barlow P. Incidence of delayed cerebral ischemia following subarachnoid haemorrhage of unknown cause. J Neurol Neurosurg Psychiatry 1985;48:132-1 36 19. Juul R, Tjorb0rn A, Frederiksen TA, Rinkj~jbR. Prognosis in

468 Annals of Neurology Vol 29 No 5 May 1991

20.

21. 22.

23.

24.

subarachnoid hemorrhage of unknown etiology. J Neurosurg 1986;64:359-362 Alexander MSM, Dias PS, Uttley D. Spontaneous subarachnoid hemorrhage and negative cerebral panangiography.J Neurosurg 1986;64:537-542 Giombini S, Bruzzone MG, Pluchino F. Subarachnoid hemorrhage of unexplained cause. Neurosurgery 1988;22:3 13-3 16 Rrismar J, Sundbarg G. Subarachnoid hemorrhage of unknown origin: prognosis and prognostic factors. J Neurosurg 1985;63: 349-354 Hasan D, Wijdicks EFM, Vermeulen M. Hyponatrernia is associated with cerebral ischemia in patients with aneurysmal subarachnoid hemorrhage. Ann Neurol 1990;27:106-108 Rinkel GJE, Wijdicks EFM, Vermeulen M, et al. Outcome in perimesencephalic (nonaneurysmal) subarachnoid hemorrhage: a follow up study in 37 patients. Neurology 1990;40:11301132