thediagnosis ofsubarachnoid haemorrhagenoid haemorrhage (sah). at present, com-puted tomography (ct)...

8
Journal of Neurology, Neurosurgery, and Psychiatry 1990;53:365-372 OCCASIONAL REVIEW The diagnosis of subarachnoid haemorrhage M Vermeulen, J van Gijn Lumbar puncture (LP) has for a long time been the mainstay of diagnosis in patients who presented with symptoms or signs of subarach- noid haemorrhage (SAH). At present, com- puted tomography (CT) has replaced LP for this indication. In this review we shall outline the reasons for this change in diagnostic approach. In the first place, there are draw- backs in starting with an LP. One of these is that patients with SAH may harbour an intracerebral haematoma, even if they are fully conscious, and that withdrawal of cerebro- spinal fluid (CSF) may occasionally precipitate brain shift and herniation. Another disadvan- tage of LP is the difficulty in distinguishing between a traumatic tap and true subarachnoid haemorrhage. Secondly, the use of CT within the first two or three days offers a wealth of information about the origin and extent of the haemorrhage, and about the presence of early complications requiring urgent treatment. An early brain scan also serves as a baseline against which future changes can be measured. The information about the location of the haemor- rhage is especially valuable in patients with SAH and a negative angiogram. This is a heterogeneous group of patients which will be discussed separately. The review continues with guidelines about the interpretation of the CSF in patients with a negative CT scan, which is the only remaining indication for LP in the diagnosis of SAH. The final paragraph points out the many pitfalls in the diagnosis of rebleeding by analysis of CSF samples, which makes serial CT scanning by far the preferred method. Department of Neurology, University Hospital Dijkzigt, Rotterdam M Vermeulen University Department of Neurology, Utrecht, The Netherlands J van Gijn The dangers of lumbar puncture The possibility of a haematoma being present in patients who present with the clinical features of SAH should be considered. In a series of 100 consecutive patients who were suspected of aneurysmal rupture, fifteen had a non-aneurysmal haematoma and eight of these were located in the cerebellum.' It is obvious that an LP could have done harm in these patients. In addition, a third of patients with a ruptured aneurysm harbour an intracerebral haematoma,`'' and even in patients without neurological signs other than neck stifEfess 10 per cent may have a haematoma of at least 30 mm.4 Deterioration after LP is not only a theor- etical possibility as shown by Duffy, who described severe clinical deterioration in seven of 55 patients with SAH who had LP, before CT scanning and within 12 hours of the bleed. Intracranial haematomas with brain shift was proven by operation or subsequent CT scan- ning in six of the seven patients, and it was suspected in the remaining patient who stop- ped breathing at the end of the procedure.5 Rebleeding may have occurred in some of these patients. We therefore agree with Hillman that it is advisable to perform a CT scan first in all patients who present within 72 hours of a suspected SAH, even if this requires referral to another centre.4 It could be argued that by first performing CT the diagnosis may be delayed and that this may be dangerous if the patient has bacterial meningitis. However, the likelihood of menin- gitis is extremely small in patients with an adequate history of headache or unconscious- ness coming on within seconds and only rarely is the distinction more difficult, for instance, in patients who have been found in a- confused state, with marked neck stiffness and moderate fever. Ideally, patients with suspected SAH should not be admitted to a hospital without CT scanning facilities (and patients with confirmed SAH not to hospitals without a neurosurgical unit). Traumatic tap or intracranial bleeding? Neck stiffness following SAH takes three to 12 hours to develop-it is an unreliable test to perform on domiciliary calls. Patients arriving in casualty with a history of "thunderclap headache" a few hours earlier and with equi- vocal neck stiffness may either have a true SAH or a more usual type of headache with an unusual onset.' If an LP is performed and blood-stained CSF is found it is difficult to distinguish between a traumatic tap and intra- cranial bleeding. It is a widely held belief that this distinction can be made by collecting the CSF in three test tubes and by counting the red cells in the consecutive tubes. A decrease in red cells would indicate a traumatic tap. Indeed, a decrease in red cells occurs more often in traumatic punctures than in intracranial bleed- ing, but it also occurs in patients with a previous bleed.9 Conversely, a constant num- ber of cells can be seen with traumatic taps. In the individual case therefore no firm diagnosis can be made with this "three tube method". The opinion of the doctor who carried out 365 on February 10, 2020 by guest. Protected by copyright. http://jnnp.bmj.com/ J Neurol Neurosurg Psychiatry: first published as 10.1136/jnnp.53.5.365 on 1 May 1990. Downloaded from

Upload: others

Post on 25-Jan-2020

10 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Thediagnosis ofsubarachnoid haemorrhagenoid haemorrhage (SAH). At present, com-puted tomography (CT) has replaced LPfor this indication. Inthis reviewweshall outline the reasons for

Journal of Neurology, Neurosurgery, and Psychiatry 1990;53:365-372

OCCASIONAL REVIEW

The diagnosis of subarachnoid haemorrhage

M Vermeulen, J van Gijn

Lumbar puncture (LP) has for a long time beenthe mainstay of diagnosis in patients whopresented with symptoms or signs of subarach-noid haemorrhage (SAH). At present, com-puted tomography (CT) has replaced LP forthis indication. In this review we shall outlinethe reasons for this change in diagnosticapproach. In the first place, there are draw-backs in starting with an LP. One of these isthat patients with SAH may harbour anintracerebral haematoma, even if they are fullyconscious, and that withdrawal of cerebro-spinal fluid (CSF) may occasionally precipitatebrain shift and herniation. Another disadvan-tage of LP is the difficulty in distinguishingbetween a traumatic tap and true subarachnoidhaemorrhage. Secondly, the use of CT withinthe first two or three days offers a wealth ofinformation about the origin and extent of thehaemorrhage, and about the presence of earlycomplications requiring urgent treatment. Anearly brain scan also serves as a baseline againstwhich future changes can be measured. Theinformation about the location of the haemor-rhage is especially valuable in patients withSAH and a negative angiogram. This is aheterogeneous group of patients which will bediscussed separately. The review continueswith guidelines about the interpretation of theCSF in patients with a negative CT scan, whichis the only remaining indication for LP in thediagnosis of SAH. The final paragraph pointsout the many pitfalls in the diagnosis ofrebleeding by analysis of CSF samples, whichmakes serial CT scanning by far the preferredmethod.

Department ofNeurology, UniversityHospital Dijkzigt,RotterdamM VermeulenUniversityDepartment ofNeurology, Utrecht,The NetherlandsJ van Gijn

The dangers of lumbar punctureThe possibility of a haematoma being presentin patients who present with the clinicalfeatures of SAH should be considered. In aseries of 100 consecutive patients who weresuspected of aneurysmal rupture, fifteen had anon-aneurysmal haematoma and eight of thesewere located in the cerebellum.' It is obviousthat an LP could have done harm in thesepatients. In addition, a third of patients with aruptured aneurysm harbour an intracerebralhaematoma,`'' and even in patients withoutneurological signs other than neck stifEfess 10per cent may have a haematoma of at least30 mm.4

Deterioration after LP is not only a theor-etical possibility as shown by Duffy, whodescribed severe clinical deterioration in seven

of 55 patients with SAH who had LP, beforeCT scanning and within 12 hours of the bleed.Intracranial haematomas with brain shift wasproven by operation or subsequent CT scan-ning in six of the seven patients, and it wassuspected in the remaining patient who stop-ped breathing at the end of the procedure.5Rebleeding may have occurred in some ofthesepatients.We therefore agree with Hillman that it is

advisable to perform a CT scan first in allpatients who present within 72 hours of asuspected SAH, even if this requires referral toanother centre.4

It could be argued that by first performingCT the diagnosis may be delayed and that thismay be dangerous if the patient has bacterialmeningitis. However, the likelihood of menin-gitis is extremely small in patients with anadequate history of headache or unconscious-ness coming on within seconds and only rarelyis the distinction more difficult, for instance, inpatients who have been found in a- confusedstate, with marked neck stiffness and moderatefever.

Ideally, patients with suspected SAH shouldnot be admitted to a hospital without CTscanning facilities (and patients with confirmedSAH not to hospitals without a neurosurgicalunit).

Traumatic tap or intracranial bleeding?Neck stiffness following SAH takes three to 12hours to develop-it is an unreliable test toperform on domiciliary calls. Patients arrivingin casualty with a history of "thunderclapheadache" a few hours earlier and with equi-vocal neck stiffness may either have a true SAHor a more usual type of headache with anunusual onset.' If an LP is performed andblood-stained CSF is found it is difficult todistinguish between a traumatic tap and intra-cranial bleeding. It is a widely held belief thatthis distinction can be made by collecting theCSF in three test tubes and by counting the redcells in the consecutive tubes. A decrease in redcells would indicate a traumatic tap. Indeed, adecrease in red cells occurs more often intraumatic punctures than in intracranial bleed-ing, but it also occurs in patients with aprevious bleed.9 Conversely, a constant num-ber of cells can be seen with traumatic taps. Inthe individual case therefore no firm diagnosiscan be made with this "three tube method".The opinion of the doctor who carried out

365

on February 10, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.53.5.365 on 1 M

ay 1990. Dow

nloaded from

Page 2: Thediagnosis ofsubarachnoid haemorrhagenoid haemorrhage (SAH). At present, com-puted tomography (CT) has replaced LPfor this indication. Inthis reviewweshall outline the reasons for

Vermeulen, van Giin

the LP also appears unreliable. In a study byBuruma et al doctors were often unable todecide for themselves whether the blood-stained CSF had resulted from the procedureor not.9 The form of the red cells changes in theCSF, and this so-called crenation of red cellshas been advocated as a means of differentiat-ing a traumatic tap from a true haemorrhage.However, crenation of red cells occurs verysoon after red cells have entered the CSF andmay therefore be seen in many traumatic taps.'0Cytology of the CSF is also of limited value.The presence of erythrophages indicatesintracranial haemorrhage if an LP was notcarried out earlier, but it takes some time beforesiderophages are demonstrable. Moreover,negative cytology does not exclude a haemor-rhage.9The most reliable method for distinguishing

a traumatic tap from a genuine bleed is to spindown the CSF and to examine the supernatantfluid for the presence of xanthochromia.Examination of the supernatant should be dohewith spectrophotometry and not with thenaked eye, since direct vision is a rather in-sensitive method. In a series of32 CSF samplesin which spectrophotometry had detected xan-thochromia, this was visible with the naked eyein only half the samples." The pigments thatare responsible for the yellow colour ofthe CSFafter a haemorrhage result from lysis of redcells. The average survival of red cells in theCSF is much shorter than in the circulation.This rapid haemolysis has not been fully ex-plained. Recently, it has been suggested thatred cells in the subarachnoid space lose theiridentity as autologous cells and this results inimmune mediated haemolysis.'2

In the classic study by Barrows et al thedevelopment of xanthochromia was studied bymeans of spectrophotometry.'3 This techniquemeasures the light intensity in different regionsof the visible spectrum (400-700 nm), aftertransmission of light through the colouredCSF. Three different pigments can be distin-guished by their characteristic absorptionbands in the spectrum: oxyhaemoglobin,methaemoglobin, and bilirubin. The wave-length where absorption occurs permits quali-tative determination, whereas the height of thepeak makes quantification possible.The rate at which the pigments appear in the

CSF is of great practical importance, as theirpresence excludes a traumatic puncture. Oxy-haemoglobin has been detected as early as twohours after the bleeding,'3 but usually it takes afew hours or more for red cells to lyse and forxanthochromia to develop. In two large seriesofpatients with SAH it was shown that xantho-chromia could be detected in all patients inwhom the CSF was examined at least 12 hoursafter the haemorrhage. 4 15

Extra information by CT scanningComputed tomography not only obviates theneed for a potentially hazardous LP, it offersimportant information other than the merepresence or absence of intracranial blood. Thisapplies particularly if the scan is made as early

as possible. On the day of SAH, intracranialblood is detected in about 95% of patients.This proportion declines to 90% after one day,80% after five days, and 50% after one week.316

In the first place, if intracranial blood isdetected, the site of the haemorrhage oftenindicates the likely source. In case of anintracerebral haematoma, it is often possible todistinguish primary intraparenchymal bleed-ing from SAH extending into the brain tissue(fig 1).1' It is not always easy to determinewhether a haematoma associated withaneurysm rupture is in fact intraparenchymalor whether it has only distended the subarach-noid space, usually the frontal interhemis-pheric or sylvian fissure.'8 However, this dis-tinction is of limited importance for practicalmanagement, and follow up scans will usuallyresolve the issue. In case the extravasated bloodis confined to the basal cisterns, a rupturedaneurysm can be diagnosed if the subarachnoidclots are most dense at one of the classicalaneurysm sites (fig 2): the frontal interhemis-pheric fissure (aneurysm of the anterior com-municating artery), the chiasmatic cistern onone side (aneurysm of the internal carotidartery, usually at the origin of the posteriorcommunicating artery), or the most lateral partof the sylvian fissure (aneurysm of the middlecerebral artery).' '9

Bleeding from aneurysms at more unusualsites may be more difficult to recognise;aneurysmal bleeding from a posterior inferiorcerebellar artery, for instance, may be easilymissed by CT.202' Predicting the site of aruptured aneurysm from the CT scan is morethan an intellectual game. This information ishelpful in planning angiography, in particularthe order and the extent of selective catheter-isation, and it is vital ifmore than one aneurysmis found.' 21-23 Even the uncommon event oftwoaneurysms rupturing at the same time can becorrectly documented by CT.24 Visualisation ofthe aneurysm itself, after injection of contrast,used to be the exception with CT, indicatingthat the aneurysm is very large, with or withoutcalcification.' High-resolution CT scanning,with 1-5 mm slices, is more sensitive but notinfallible.25 Magnetic resonance imaging(MRI) is not well suited for imaging subarach-noid haemorrhage in the acute stage.26 Afterseveral days or weeks, however, when the CTscan has become normal, MRI may detectsubpial deposition of haemosiderin near thesource of the haemorrhage.2'26 In addition,MRI allows visualisation of the aneurysmitself. In two studies, a T2-weighted 2000/80spin echo sequence, identified an aneurysm inabout half the cases, by means of a signal-voidarea.27 28

Not all haemorrhage in the basal cisterns isof aneurysmal origin. A recently recognisedform is the perimesencephalic haemorrhage,29with extravasated blood mainly or only in theinterpeduncular, ambient, or quadrigeminalcisterns (fig 3a). Not only is the pattern ofhaemorrhage on CT different from that withaneurysmal haemorrhage, the clinical featuresmay also differ; loss of consciousness occursless often and a gradual onset of headache

366

on February 10, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.53.5.365 on 1 M

ay 1990. Dow

nloaded from

Page 3: Thediagnosis ofsubarachnoid haemorrhagenoid haemorrhage (SAH). At present, com-puted tomography (CT) has replaced LPfor this indication. Inthis reviewweshall outline the reasons for

The diagnosis of subarachnoid haemorrhage

Figure I (a) Bifrontalhaematoma,from rupturedaneurysm of the anteriorcommunicating artery; (b)haematoma in the medialpart of the temporal lobe,from ruptured aneurysm ofthe internal carotid artery,at thejunction with theposterior communicatingartery; (c) haematoma inthe lateral part of thetemporal lobe,fromruptured aneurysm of themiddle cerebral artery (theaneurysm itself is visible asa less dense area within thehaematoma).

i ~~~~~~~~~ ~~~~~~~~ ~~~~~~~~~~~...y t ':

& ~ ~~~~~~~~~ ~ ~~~~~~~~:.4.

(minutes rather than seconds) occurs more

often than with aneurysms. Together with theexcellent prognosis this suggests a venous or

capillary source of bleeding.29 This type ofhaemorrhage is not rare, occurring in about1000 of all subarachnoid haemorrhages, and inhalf the patients with a negative angiogram(table). It is also rather specific. Basilaraneurysms may also produce an isolated clotin the interpeduncular cistern, but theseaneurysms constitute only one out of 10 rup-

tured aneurysms, and in turn only one in eighthaemorrhages from basilar aneurysms do notextend into the chiasmatic and other basalcisterns (fig 3b).29 For every 100 SAHs 80 are

aneurysmal. After the exclusion by CT ofintracerebral haemorrhages and arteriovenousmalformations the remaining 10 are of non-

aneurysmal perimesencephalic type and abouthalf of these occur mainly in the interpedun-cular cistern. Thus, the chance of a rupturedbasilar aneurysm with such a type of haemorr-hage is about one in six (fig 4). This is of coursereason enough to perform a vertebral angio-gram if the patient is fit for operation, but if thisis negative repeated angiography does not seem

warranted.A similar calculation applies to patients

without blood in the basal cisterns or even

without any evidence of extravasated blood on

CT scanning, despite an interval of less thanthree days from the onset of symptoms, and yetwith true SAH demonstrated by xantho-chromia of the CSF. With third or fourthgeneration CT scanners, and with adequateexperience in the detection of subarachnoidblood, evidence of blood in the basal cisterns islacking on the first day in about 5 %// of rupturedaneurysms,3 16 29 and in about a quarter ofpatients with SAH and a normal angiogram.29Given that 800/ of patients with SAH have an

aneurysm, the chance of an aneurysm in

Table Patterns of bleeding on CT and results ofangiography in 120 consecutive patients withsubarachnoid haemorrhage. Patients with primaryintracerebral haemorrhage have been excluded.24 CT wasperformed within five days in all patients, and withinthree days in most

Angiography

NoAneurysm aneurysm(n = 92) (n = 28)

Pattern of haemorrhage on CT:Basal cisterns, compatible withaneurysm 85 7

Predominantly around midbrain 1 13Aspecific or no evidence ofhaemorrhage 6 8

367

on February 10, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.53.5.365 on 1 M

ay 1990. Dow

nloaded from

Page 4: Thediagnosis ofsubarachnoid haemorrhagenoid haemorrhage (SAH). At present, com-puted tomography (CT) has replaced LPfor this indication. Inthis reviewweshall outline the reasons for

Figure 2 (a)Subarachnoidhaemorrhage,predominantly in thefrontal part of theinterhemispheric fissure,from a ruptured aneurysmof the anteriorcommunicating aneurysm;(b) subarachnoidhaemorrhage, with centrein the right suprasellarcisterns (to reader's left),from a ruptured aneurysmof the internal carotidartery, at thejunction withthe posteriorcommunicating artery;(c) subarachnoidhaemorrhage,predominantly in thesylvianfissure,from aruptured aneurysm of themiddle cerebral artery. |~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~

, r ... ,g F_~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~...

lFo ~~~~~~~~~~~~~~~~~~~~~~~~~~.6.''...

!......'

.;k~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~... ...

patients with true SAH but without cisternalblood on an early CT scan is about one in two.

Early CT scanning not only gives importantdiagnostic information on the source ofhaemorrhage, it also demonstrates complica-tions that may require urgent treatment,especially acute hydrocephalus"0 or a life-threatening haematoma." Finally, the initialCT scan provides an important baseline for thediagnosis of future complications, particularlyrebleeding, infarction or hydrocephalus.

"Angiogram-negative SAH": a mixedbagThe fate of patients with the diagnosis ofsubarachnoid haemorrhage but a negativeangiogram has been the subject of a spate ofpublications that have reported follow up

studies. We shall abstain from listing thesestudies, as it is abundantly clear from thepreceding paragraphs that these patients are

not all of one kind. The distinction betweenthese groups is difficult from the perspective ofthe neurosurgeon whose patients are referredafter five days or more, without a CT scan, thediagnosis being based on a lumbar puncture ofwhich there are no details. The relative propor-tions of the different categories of patients with

. a negative angiogram may differ betweencentres, which makes it even more pointless tolump all these patients under one heading. Itmay be useful therefore to remember that thereare, at present, at least four subgroups ofpatients with "angio-negative SAH".The first and most unfortunate category is

that of patients who are brought to the emer-gency room with what in fact is nothing morethan "crash migraine" or a sudden exacerba-tion of tension headache, 8 and who undergoan ill-timed and traumatic lumbar puncture byyoung and eager physicians. They are thenconfined to bed with a diagnosis of SAH,subsequently transferred to a specialised unitfor angiography, and they are finally told thatthe weak spot in one of their brain vessels hasprobably healed but that they would be wise totake things a bit easier in the future.The remaining groups of patients all have

a true SAH, but without a demonstrableaneurysm. Again, it is assumed that on the basisof the CT scan primary intracerebral haemor-rhages consistent with either the "hyper-tensive" type or with arteriovenous malforma-tions, cryptic or overt, have been ruled outbefore. About half of these patients show bloodon CT around the midbrain, unlikely to becompatible with aneurysmal rupture (fig 3a).The localised nature of the extravasated blood,the less explosive onset of the symptoms insome cases, and the excellent prognosis allmake it improbable that an occult arterialaneurysm underlies these haemorrhages."9 Thethird group of patients, making up about one

quarter ofpatients with true SAH and a normalangiogramj shows no evidence of blood in the

368 Vermeulen, van Gijn

on February 10, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.53.5.365 on 1 M

ay 1990. Dow

nloaded from

Page 5: Thediagnosis ofsubarachnoid haemorrhagenoid haemorrhage (SAH). At present, com-puted tomography (CT) has replaced LPfor this indication. Inthis reviewweshall outline the reasons for

The diagnosis of subarachnoid haemorrhage

Figure 3 (a) Non-aneurysmalperimesencephalichaemorrhage, withextravasated bloodpredominantly in theinterpeduncularfossa, andto a lesser extent in theambient cisterns; (b)subarachnoid haemorrhagefrom ruptured aneurysm ofthe basilar artery, withextravasated bloodpredominantly in theinterpeduncularfossa, butalso more anteriorly, in thechiasmatic cistern, theSylvian fissures, and thefrontal part of theinterhemispheric fissure.

r

I*. .X.:~...

A

I

basal cisterns on CT, or no blood at all.29 Inthese patients the cause of bleeding is evenmore elusive than in the perimesencephalicgroup, but it may be the same. The fate of thesepatients has not been separately investigated,but the impression is that they do well.The fourth and last category are patients

with a pattern of bleeding on CT, that is typicalof ruptured aneurysm but in whom angio-graphy fails to demonstrate the expectedaneurysm. This group constitutes the lastquarter of patients with SAH and normalangiography, and about 5%/ of all patients withcisternal haemorrhage (table). It may be thatoccasional observations of local thrombosis,32vasospasm33 or rebleeding34 apply only to thesepatients, and that these harbour "occult"aneurysms, despite a normal angiogram. It isthis category of patients in which repeatedangiography seems most indicated.

Lumbar puncture after a negative CTscanExamination ofthe CSF is still important ifCTscanning provides no evidence of subarachnoid

100patients withisubarachnoid

haemorrhage

Aneurysm Other(80) aneurysm (72) Blood also in

anterior cisternsL Basilar (7)

aneurysm (8)-(1)

Blood mainily in 1 in 6

interpedu ncu larcistern

(5)Peri -

mesencephalic -

(10) _I-Blood elsewhere

-No aneurysm Other (10) around midbrain

(20) (5)

Figure 4 The chance of a ruptured aneurysm in patients with an isolated clot in theinterpeduncular cistern on CT is 1 in 6.

or other intracranial haemorrhage. Yet, if thepatient is seen soon after the headache, we wishto emphasise again that it is vital to defer lumbarpuncture to at least 12 hours after the onset ofsymptoms and that xanthochromia should beexamined not with the naked eye but withspectrophotometry. If LP is carried out tooearly, xanthochromia may not have had enoughtime to develop and it is then impossible todistinguish between a traumatic lumbar punc-ture and a genuine SAH. Blood artificiallyintroduced into the CSF will later lead to theappearance of pigments, and repeated CSFstudies can never solve the problem that hasbeen created by an LP that has been performedtoo soon.CSF examination in the diagnosis of SAH is

also important when the patient presents lateafter a suspected SAH. The 90% probability3 16ofdetecting blood in the basal cisterns on CT inthe first two days after an SAH decreasesrapidly thereafter. In contrast, all patients witha true SAH have CSF xanthochromia between12 hours and two weeks after the bleeding.'4 15Xanthochromia is still present in over 700,, ofpatients after three weeks, and even after fourweeks this proportion is still over 40%." Thesefindings contrast with another study35 wherexanthochromia was investigated with the nakedeye, which explains the high proportion ofpatients with SAH who had no xanthochromia.Could an SAH have occurred in patients

with symptoms suspected of SAH who haveblood-stained CSF but no xanthochromia?Although one study suggested that this mayoccur,35 this is extremely unlikely. In a centrewith approximately 80 patients with a provenSAH admitted each year, only 12 patients in asix year period had bloodstained CSF but noxanthochromia. In three of these 12 patientsangiography was performed; all three werenegative. All 12 patients survived withoutdisability and were not admitted again for an

369

I

on February 10, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.53.5.365 on 1 M

ay 1990. Dow

nloaded from

Page 6: Thediagnosis ofsubarachnoid haemorrhagenoid haemorrhage (SAH). At present, com-puted tomography (CT) has replaced LPfor this indication. Inthis reviewweshall outline the reasons for

Vermeulen, van Gijn

SAH during a mean follow up period of fouryears.'5 Again, the most probable explanationfor bloodstained CSF without xanthochromiain patients suspected of SAH is that thesepatients had a non-haemorrhagic "thunderclapheadache" and a traumatic LP.

It has been suggested that angiographyshould be carried out in patients with a suspec-

ted SAH but with bloodless CT or LP. Thissuggestion is based on two reports of singlepatients suspected of SAH, who had a normalCT and normal CSF but in whom subsequentangiography showed an aneurysm with vaso-

spasm.36 37

In contrast, a study of 71 patients withsudden, severe and unusual headache but withnormal CT and CSF showed that none of thesepatients had complications related to SAH andnone later had a documented SAH, during a

mean follow up period of 3-3 years.6 It is likelythat these patients have had benign "thunder-clap headaches", without bleeding. In bothreported cases with aneurysms, vasospasm dn

the angiogram was the feature believed to havea causal relation with the headache. We thinkthat migraine is a more probable explanationfor both the headache and the vasospasm.38 Asmall proportion of adults harbour asympto-matic aneurysms,39 and indiscriminate angio-graphy is bound to uncover some of these. Weagree that angiography should be done inpatients without blood on CT and even withoutCSF xanthochromia if these patients presentafter more than two weeks of the suspectedbleeding, since not only the blood on CT butalso xanthochromia of the CSF may havedisappeared by that time.'5

The diagnosis of rebleedingIf a patient with a recently ruptured aneurysmcomplains of a sudden increase ofheadache andthen abruptly becomes deeply unconscious,with wide, non-reacting pupils and apnoea,there is no diagnostic problem. However, othercomplications may mimick rebleeding.YDeterioration of consciousness and focalneurological signs may have various causes,particularly delayed cerebral ischaemia.Furthermore, the clinical picture is oftenclouded by the previous haemorrhage. Coma-tose patients do not report an increase ofheadache, and "increased neck stiffness" isoften equivocal. Finally, close observation ofthe patient around the clock for several weeksis not always practicable, with the result thatthe onset of any deterioration is not alwaysdocumented. Even if a sudden loss of con-sciousness (within minutes) is witnessed in a

patient with SAH, only two out of three turnout to have a rebleed on CT. If the fatal casesare excluded an even larger proportion ofabrupt episodes, nearly 50%, was not causedby rebleeding.4l Other causes of acute

deterioration were epilepsy, ventricularfibrillation (in only one of 62 episodes ofdeterioration) and, surprisingly, cerebralischaemia. Cerebral ischaemia after SAHusually presents with a gradual onset ofdeterioration. In contrast three patients in this

series of 62 episodes mimicking a rebleed had asudden impairment of consciousness, followedby a period of a fe hours in which there was nochange, and then had further gradual deter-ioration while evidence of infarction developedon CT.4' Therefore, rebleeding cannot be diag-nosed on purely clinical grounds. Theseproblems used to be often dismissed with thestatement that rebleeding "was confirmed bylumbar puncture or autopsy".The diagnosis of rebleeding by analysis of

CSF samples is, in fact, extremely difficult.Just as the diagnosis of a first haemorrhage byCSF analysis requires the demonstration ofxanthochromia, the diagnosis of rebleedingrequires the demonstration of "fresh" pig-ments. This raises the question of whetherblood pigments appear in and disappear fromthe CSF in a predictable fashion. Oxyhaemo-globin may be detected in the CSF a few hoursafter the bleed.'3 In the early study by Barrowset al it remained the predominant pigment onlyfor the first few days, after which it wasconverted to bilirubin, the iron-free derivativeof haemoglobin.'3 This conversion does nottake place in test tubes, as it is dependent uponthe enzyme haem oxygenase, which is presentin macrophages, the arachnoid membrane, andthe choroid plexus.42 In a later study, however,the relative concentration of oxyhaemoglobinwas found to increase spontaneously after aninitial decrease, and oxyhaemoglobin couldstill be the predominant pigment as late asthree weeks after the initial haemorrhage.43 Thesame fluctuating pattern was observed whenthe haemoglobin concentration was seriallymeasured.43 Therefore, measurement ofneither the proportion of oxyhaemoglobin northe haemoglobin concentration can help in thediagnosis of rebleeding. It has also been sug-gested that the total amount of pigments fromred cell lysis, the xanthochromic index,decreases after the first few days following thehaemorrhage.44 Rebleeding would becomeapparent by the interruption of this normalresolution pattern. However, the xantho-chromic index may continue to rise up to thesixteenth day after SAH.43 The marked dif-ferences in the velocity of pigment clearancefrom the CSF have been attributed to advancedage, diabetes, and atherosclerosis.45 Since mostrebleeds occur within this period' a suspectedrebleed cannot be confirmed merely by demon-strating an increase in xanthochromia.These findings make the often reported

statement that rebleeding was confirmed by LPbarely tenable. If rebleeding is suspected andLP is repeated, the demonstration of blood-stained CSF does not prove rebleeding, as theCSF may remain blood-stained for as long asfour weeks after the initial bleed.'2 The blood-staining may also have resulted from a trau-matic tap, and the distinction between atraumatic tap and recurrent bleeding is impos-sible to make since the CSF supernatant wasalready xanthochromic before the suspectedrebleed. In summary, examination of the CSFcan only be of help in the diagnosis of rebleed-ing if two earlier LPs had demonstrated adecrease of the xanthochromic index, which is

370

on February 10, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.53.5.365 on 1 M

ay 1990. Dow

nloaded from

Page 7: Thediagnosis ofsubarachnoid haemorrhagenoid haemorrhage (SAH). At present, com-puted tomography (CT) has replaced LPfor this indication. Inthis reviewweshall outline the reasons for

The diagnosis of subarachnoid haemorrhage 371

then followed by an increase.43 It is clear that inpractice this method can seldom be applied.Apart from the unpredictable clearance of

pigments the diagnosis of rebleeding by CSFanalysis is also difficult because LP cannotalways be performed after a rebleed. Since"fresh" xanthochromia needs time to develop,the LP has to be postponed to at least 12 hoursafter the rebleed. Meanwhile the patient mayhave died. In addition, the risk of LP inpatients with an unsuspected haematoma maybe even greater than after a first bleed. In aconsecutive series of 17 patients with CTevidence of rebleeding, it was possible in onlysix patients to carry out an LP and compare theresults ofspectrophotometry with two previoussamples (the diagnosis was confirmed in five).43In short, the diagnosis of rebleeding by LP ishardly possible.

Necropsy, if feasible, does not settle thematter either. The pathologist can confidentlydiagnose rebleeding only if the interval be-tween the first haemorrhage and rebleedingwas not too short, and that between rebleedingand death was not too long.A' The presentaccepted standard in the diagnosis of rebleed-ing consists of the demonstration of a freshhaemorrhage on CT scanning or at necropsy,not previously demonstrated on CT scan. Suchcomparisons are quite sensitive, as even threeor more successive haemorrhages can be relia-bly distinguished.4' Nevertheless, earlyrebleeds cannot be diagnosed if these occurbefore the admission CT scan.'Conclusions1 Lumbar puncture after SAH may pre-

cipitate brain shift and herniation from anunsuspected haematoma, even in patientswho are fully conscious.

2 Subarachnoid haemorrhage can be reliablydiagnosed by analysis of the CSF only afterexamination of the supernatant fluid with aspectrophotometer. This implies that lum-bar puncture should be postponed until atleast 12 hours after the onset of symptoms.If lumbar puncture is performed earlier, atraumatic tap may unnecessarily stigmatisepatients with benign forms of "thunderclapheadache".

3 Computed tomography (CT) scanning isa safe and sensitive method for demon-strating subarachnoid blood, but shouldpreferably be performed within threedays. Important additional informationprovided by CT is: firstly, the demonstra-tion of non-aneurysmal sources of haemor-rhage ("hypertensive" haemorrhages,arteriovenous malformations, cryptichaemorrhages around the midbrain);secondly, the demonstration of early com-plications requiring urgent treatment;thirdly, evidence about the probable site ofruptured aneurysms, and fourthly, estab-lishment of a baseline for future changes.

4 Patients with negative angiograms are nota single category. This group consists ofpatients without SAH but with a traumaticlumbar puncture, patients with varioustypes of non-aneurysmal subarachnoid

haemorrhage, and finally a few patientswith "occult" aneurysms.

5 Lumbar puncture is indicated in SAH onlyif CT scanning is negative. If LP is carriedout between 12 hours and two weeks afterthe haemorrhage, xanthochromia of theCSF can be demonstrated in all patientswith SAH, provided xanthochromia isinvestigated by spectrophotometry. Evenafter three weeks xanthochromia is stillpresent in over 70%0 of patients, and afterfour weeks in over 40%.

6 Rebleeding is the cause of acute deteriora-tion in only two out of three cases withsudden loss of consciousness; the diagnosiscannot be confirmed by CSF analysis,except under exceptional circumstances.Necropsy is also fallible in this respect.Serial CT scanning is mandatory for aconfident diagnosis of rebleeding.

We received helpful comments from Dr E FM Wijdicksand D Hasan and we are grateful to Mrs Betty Mast andMrs Ellen Budelman for secretarial assistance.

1 Van Gijn J, Van Dongen KJ. Computed tomography in thediagnosis of subarachnoid haemorrhage and rupturedaneurysms. Clin Neurol Neurosurg 1980;82:11-24.2 Pasqualin A, Bazzan A, Cavazzani P, Scienza R, Licata C,Da Pian R. Intracranial hematomas following aneurysmalrupture: experience with 309 cases. Surgical Neurology1986;25:6-17.

3 Van Gijn J, Van Dongen KJ. The time course ofaneurysmalhaemorrhage on computed tomograms. Neuroradiol 1982;23:153-6.

4 Hillman J. Should computed tomography scanning replacelumbar puncture in the diagnostic process in suspectedsubarachnoid haemorrhage? Surg Neurol 1986;26:547-50.5 Duffy GP. Lumbar puncture in spontaneous subarachnoidhaemorrhage. Br Med J 1982;285:1 163-4.

6 Wijdicks EFM, Kerkhoff H, Van Gijn J. Long-term followup of 71 patients with thunderclap headache mimickingsubarachnoid haemorrhage. Lancet 1988;ii:68-9.

7 Salloum A, Lebel M, Reiher J. Acces cephalalgiquessimulant une hemorrhagic meningee. Rev Neurol 1977;133:131-8.8 Pearce JMS, Pearce SHS. Benign paroxysmal cranial

neuralgia or cephalgia fugax. Br Med J 1986;292:1015.9 Buruma OJS, Janson HLF, Den Bergh FAJTM, BotsGTHAM. Blood-stained cerebrospinal fluid: traumaticpuncture or haemorrhage? J Neurol Neurosurg Psychiatry1981;44:144-7.

10 Matthews WF, Frommeyer WB. The in vitro behaviourof erythrocytes in human cerebrospinal fluid. J Lab ClinMed 1955;45:508-15.

11 Soderstrom CE. Diagnostic significance of CSF spectro-photometry and computer tomography in cerebrovasculardisease. A comparative study in 231 cases. Stroke 1977;5:606-12.

12 Peterson JW, Kwun BD, Teramura A, et al. Immunologicalreaction against the aging human subarachnoid eryth-rocyte. A model for the onset of cerebral vasospasm aftersubarachnoid hemorrhage. J Neurosurg 1989;71:718-26.13 Barrows LJ, Hunter FT, Banker BQ. The nature and clinicalsignificance of pigments in the cerebrospinal fluid. Brain1955;78:59-80.

14 Walton JN. Subarachnoid haemorrhage. Edinburgh: Living-stone, 1956:124.

15 Vermeulen M, Hasan D, Blijenberg BG, Hijdra A, Van GijnJ. Xanthochromia after subarachnoid haemorrhage needsno revisitation. J Neurol Neurosurg Psychiatry 1989;52:826-8.

16 Adams HP, Kassell NF, Torner JC, Sahs AL. CT andclinical correlations in recent aneurysmal subarachnoidhemorrhage: a preliminary report of the CooperativeAneurysm Study. Neurology 1983;33:981-8.

17 Hayward RD, O'Reilly GVA. Intracerebral haemorrhageaccuracy of computerized transverse axial scanning inpredicting the underlying aetiology. Lancet 1976;i:1-4.

18 Schulder M, Hirano A, Elkin C. "Caval-septal" hematoma:does it exist? Neurosurgery 1987;21:239-41.

19 Scotti G, Ethier R, Melancon D, Terbrugge K, Tchang S.Computed tomography in the evaluation of intracranialaneurysms and subarachnoid haemorrhage. Radiology1977;123:85-90.

20 Ruelle A, Cavazzani P, Andrioli G. Extracranial posteriorinferior cerebellar artery aneurysm causing isolatedintraventricular hemorrhage: a case report. Neurosurgery1988;23:774-7.

21 Stone JL, Crowell RM, Gandhi YN, Jafar JJ. Multipleintracranial aneurysms: magnetic resonance imaging for

on February 10, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.53.5.365 on 1 M

ay 1990. Dow

nloaded from

Page 8: Thediagnosis ofsubarachnoid haemorrhagenoid haemorrhage (SAH). At present, com-puted tomography (CT) has replaced LPfor this indication. Inthis reviewweshall outline the reasons for

Vermeulen, van Gijn

determination of the site of rupture-report of a case.Neurosurgery 1988;23:97-100.

22 Kendall BE, Lee BCP, Claveria E. Computerized tomogra-phy and angiography in subarachnoid haemorrhage. Br JRadiol 1976;49:483-501.

23 Lim ST, Sage DJ. Detection ofsubarachnoid blood clot andother thin, flat structures by computed tomography.Radiology 1977;123:79-84.

24 Joslyn JN, Williams JP, White JL, White RL. Simultaneousrupture of two intracranial aneurysms: CT diagnosis.Stroke 1985;16:518-21.

25 Schmid UD, Steiger HJ, Huber P. Accuracy of highresolution computed tomography in direct diagnosis ofcerebral aneurysms. Neuroradiology 1987;29:152-9.

26 Barkovich AJ, Atlas SW. Magnetic resonance imaging ofintracranial hemorrhage. Radiol Clin North Am 1988;26:801-20.

27 Jenkins A, Hadly DM, Teasdale GM, Condon B, Macpher-son P, Patterson J. Magnetic resonance imaging of acutesubarachnoid hemorrhage. J Neurosurg 1988;68:731-6.

28 Saton S, Kadoya S. Magnetic resonance imaging of sub-arachnoid hemorrhage. Neuroradiology 1988;30:361-6.

29 Van Gijn J, Van Dongen KJ, Vermeulen M, Hijdra A.Perimesencephalic hemorrhage: A nonaneurysmal andbenign form of subarachnoid hemorrhage. Neurology1985;35:493-7.

30 Van Gijn J, Hijdra A, Wijdicks EFM, Vermeulen M, VanCrevel H. Acute hydrocephalus after aneurysmal sub-arachnoid hemorrhage. J Neurosurg 1985;63:355-62.

31 Brandt L, Sonesson 0, Ljunggren B, Saveland H. Rupturedmiddle cerebral artery aneurysm with intracerebralhemorrhage in younger patients appearing moribund:emergency operation? Neurosurgery 1987;20:925-9.

32 Edner G, Forster DMC, Steiner L, Bergvall V. Spontaneoushealing of intracranial aneurysm after subarachnoidhemorrhage. J Neurosurg 1978;48:540-4.

33 Spetzler RF, Winestock D, Newton HT, Boldrey EG.Disappearance and reappearance of cerebral aneurysm inserial arteriograms. J Neurosurg 1974;48:508-10.

34 Barlow P. Incidence of delayed cerebral ischaemia followingsubarachnoid haemorrhage of unknown cause. J NeurolNeurosurg P-sychiatry 1985;48:132-6.

35 MacDonald A, Mendelow AD. Xanthochromia revisited: are-evaluation of lumbar puncture and CT scanning in thediagnosis of subarachnoid haemorrhage. J Neurol Neuro-surg Psychiatry 1988;51:342-4.

36 Day JW, Raskin NH. Thunderclap headache: symptom ofunruptured cerebral aneurysm. Lancet 1986;ii: 1247-8.

37 Clarke CE, Shepherd DI, Christi L, Victoratos G. Thunder-clap headache. Lancet 1988;ii:625.

38 Call GK, Flemming MC, Sealfon S, Levine H, Kistler JP,Fisher CM. Reversible cerebral segmental vasoconstric-tion. Stroke 1988;19:1159-70.

39 McCormick WF, Nofzinger JD. Saccular intracranialaneurysms. An autopsy study. JNeurosurg 1965;22:155-9.

40 Van Crevel H. Pitfalls in the diagnosis of rebleeding fromintracranial aneurysm. Clin Neurol Neurosurg 1980;82:1-9.

41 Vermeulen M, Van Gijn J, Hijdra A, Van Crevel H. Causesof acute deterioration in patients with a rupturedintracranial aneurysm. J Neurosurg 1984;60:935-9.

42 Roost KT, Pimstone NR, Diamond I, Schmid T. Theformation of cerebrospinal fluid xanthochromia aftersubarachnoid hemorrhage: Enzyme conversion ofhemoglobin to bilirubin by the arachnoid and choroidplexus. Neurology 1972;22:973-7.

43 Vermeulen M, Van Gijn J, Blijenberg BG. Spectro-photometric analysis of CSF after subarachnoid hemor-rhage: Limitations in the diagnosis of rebleeding.Neurology 1983;33:1 12-4.

44 Van der Meulen JP. Cerebrospinal fluid xanthochromia: Anobjective inex. Neurology 1966;16:170-8.

45 Tourtellotte WW, Metz LN, Bryan ER, De Jong RN.Spontaneous subarachnoid hemorrhage. Neurology 1964;14:301-6.

46 Hijdra A, Vermeulen M, Van Gijn J, Van Crevel H.Rerupture of intracranial aneurysms: a clinico anatomicstudy. J Neurosurg 1987;67:29-33.

372

on February 10, 2020 by guest. P

rotected by copyright.http://jnnp.bm

j.com/

J Neurol N

eurosurg Psychiatry: first published as 10.1136/jnnp.53.5.365 on 1 M

ay 1990. Dow

nloaded from