9 79
Amnesia after
Femorocerebral
Angiography
J. H. W. Pexman 1
R. K. Coates 1
The charts and radiographs of 12 of 1 ,520 patients who developed amnesia after
transfemoral cerebral angiography performed with local anesthesia and minimal sedation were retrospectively reviewed . In three younger patients , exacerbation of organic or idiopathic temporal lobe epilepsy was thought to be the cause of the amnesia,
while the amnesia of the nine patients over 40 years old had the characteristics of
transient global amnesia . Eight of the patients had had similar episodes before angiography , and three others had other evidence of vertebrobasilar disease . Emboli from
atheromatous debris or catheter clot due either to catheter manipulation problems , an
inexperienced operator, or the use of a large catheter were considered likely causes in
11 cases, while in three patients contrast emboli or toxicity may have been the etiologic
factor.
Amn esia after angiography is se ld o m mentioned in reviews of th e compli ca tion s
of ce rebral angiogra phy performed by transfemoral cath eterizati on [1]. Wi shart
[2] bri efly mentioned six cases in 231 ve rtebral angi ograms. Wales and Nov [3 ]
described two cases of transient g loba l amn es ia compli cating cerebral ang iography . Memory d isturbance was described in nin e patients after ve rt eb ral angiography by de Tribo let et al. [4] , while two cases have been reco rd ed during
coro nary angiog raphy w ith the catheter positi oned in the aorta [5]. Our expe rie nce is th at amn es ia is more co mmo n th an th e literature sugges ts; th is pape r
presents 12 cases see n in our department over th e past 4 years.
MClterials and Methods
Received November 20, 198 1; accepted aft er
revision December 1 , 1982.
' Departm ent o f Diag nostic Rad io logy, Vi c tori a
Hospita l, London, Ontario, Canad a N6A 4G5 . Address reprint req uests to J . H . W. Pex man.
AJNR 4 :979-983, July / AU9ust 1983
0 195-6108 / 83 / 0404-0979 $00 .00
© Ame ri can Roentgen Ray Society
B etween Ap ril 1977 and Ap ril 1981 , 1,520 pati ents had ce reb ra l ang iog rap hy pe rfo rm ed
by transfemoral cat hete ri zati o n under loca l a nesth esia w ith pre med ication using eith er 10
mg Valium o r 30 mg Serax given o rall y about 1 hr before th e angiog ram. Th e po ste ri o r
c ircu lati o n was stu d ied in 1,3 2 1 of th e pati e nts. Fo rty pe rce nt o f th e pati ents we re examin ed
fo r suspected carotid ste nosis, 11 % fo r poss ibl e vertebrobasi lar isc hemi a, 3 0 % fo r possible
intrac ra ni al b leed in g, 10% fo r suspec ted brain tum o rs, and 9 % for misce ll aneo us reasons.
One ne uroradiol og ist use d a red Kifa 6 Frenc h cath eter fo r 750 angiog ra ms. Th e oth e r
neurorad io log ist ob tain ed 3 13 angiograms w ith a 7 .3 Frenc h Cook Head Hunte r cath ete r
o n cases presentin g wi th tra nsient isc he mi c attacks and vertebral basi lar di sease , and used
a M an i 5 Fre nch ca th ete r for th e o th er 4 57 a ng iog rams. Fo rt y-e ig ht ang iog rams we re
obtain ed by a resid ent under supe rvisio n, and th e two ex perienced neurorad io log ists
obta in ed th e oth er 1 ,472.
Conray-260 was used in all cases ; the cath eters were frequ entl y flu shed with heparini zed
sa li ne (4 ,000 U / L). Subc lavian o r inn om inate inj ec ti ons we re made in 519 pati ent s;
se lec ti ve ve rt ebral injec ti o ns w ere made in 802. Fo r eac h subc lavian o r innom in ate a rtery
se ri es, 10 ml o f co ntrast materi a l was used pe r injec ti on , whil e 2 '12 -6 ml o f co ntrast materi a l
was inj ec ted eac h tim e a ve rt ebra l artery was se lec ti ve ly stu d ied . Th e to tal amo unt of
co ntrast materi a l injec ted into th e ve rte bro basil ar c ircu lati o n by th e neuro rad io log ist using
th e M ani and Cook ca th ete rs was 20 ml , and th e o th e r neuro rad io logist never injected
more th a n 3 5 ml. If both ve rteb ral a rt eri es need ed to be ca th eteri zed, bot h in vest igators
PEXMAN AND COATES
980
studi ed one before and th e oth er after th e anterior c irc ulation had
been examin ed so th at th ere was a '/2-1 hr interva l betwee n th e
two se ri es of vertebrobasil ar injections.
Th e smallest artery cath eterized in eac h examin ation of the
posterior c irc ulat ion is recorded in tabl e 1.
Results
Th e ce ntral nervous system (CNS) co mpli cati ons are
li sted in table 2 and are div ided into four types as described
TABLE 1: Relation of Catheter Size and Artery Injected to
Amnesic Complications
Artery Inj ected: Catheter Type and
French Size
Subclavian:
Head Hunter 7.3
Kifa 6
Mani 5
Vert ebral:
Head Hunter 7.3
Kifa 6
M ani 5
N o . Patie nt s
Tolals
284
225
10
13
4 20
369
With Amnesic Complica ti ons (Ages)
5(57,57,59,64,72)
o
o
o
5(22,36,43,43,62)
1 (65)
Note.-A 12 th pati ent (ag ed 24) developed amn esia aft er a right ca rot id angiogram
was obtain ed .
TABLE 2: Central Nervous System Complications in 1 ,520
Patients after Femorocerebral Angiography
Type
Transient minor
Transient major
Perman ent
Deaths
Total
No . (% )
14 (0. 9)
17 (1.1)
3 (0.2)
Notes
Fi ve < 5 min; all < 20 min
12 amnesias < 24 hr; three
strokes < 10 days; one cortica l blin dness < 24 hr; one
hall ucination < 48 hr
Two strokes; one internu c lear
op hthalm op le9 ia
0
34 (2.2)
AJNR:4, Jul. / Aug. 1983
by Mani et al. [1]. A transient minor co mplication lasted less
th an 10 days and did not significantly affect the hea lth of
th e patient, while a transient major compli cation was of th e
same duration, but th e activity and hea lth of the patient were
affected . A perman ent complication persisted for more th an
10 days . Technical problems during angiography were reco rded on our radiologic reports . The data concern ing amnes ia and other co mplications were recorded at the time of
th e angiogram and from information supplied during our
daily neuroscience case conferences. Details of the course
of amnesia were derived retrospectively from the patients '
charts.
Twelve patients developed sudden amnesia for recent
events and could not remember where they were , why they
were there, or what had happened to them (table 3). Frequently, they asked a question that they had asked just a
few minutes before and to which they had been g iven an
answer. Some felt vaguely th at something was wrong and
others felt shivery .
In three younger pati ents (cases 1-3), the amnesic episodes were thought to be due to a temporal lobe epilepsy
because th e patients were more ag itated , restl ess, or belligerent, and because th ey had previously had a similar attack
labe led as temporal lobe epilepsy. These patients, two men
and one woman aged 22-36 years, had no evidence of
ce rebrovascular disease. The temporal lobe epi lepsy was
idiopathi c in one patient, the result of a viral encephaliti s in
another, and produced by a th alamic g liom a in th e third .
The amnes ia started during or at the end of the angiograms
and lasted 3 , 6 , and 24 hr, respectively. The posterior
c irc ulation was studied angiographically in two patients, but
in the patient with the thalami c glioma on ly a right carotid
angiogram was obtained . In this patient the right carotid
artery supplied the right posterior cerebral artery.
In nine older patients , six men and three women 43-72
years old (mean , 58 years), transient g lobal amnes ia developed after angiography . This was recognized during or at
the end of angiography in eight patients, while in one patient
(case 12), the amnesia was recognized only 8 hr after
TAB LE 3: Etiologic Factors in Amnesia after Femorocerebral Angiography
Case No .
Temporal lobe epil epsy
Transient globa l amn esia
Cli nica l (C) or rad iologic (R) evidence of di sease in th e territory of th e posterior
ce rebral basil ar and ve rtebral art eri es:
Previous amn esic episode
Other evid ence of ve rt ebrobasil ar disease
Spasm
Possible tec hnic al prob lems:
Catheter manipul ati on
Inexperien ced resident
Bi g cath eter-7 .3 Frenc h in subc lavi an (S) or 6 French in ve rtebral (V)
Poss ible contrast problems:
" Bad batc h"
..............
" Overdose "
. . . . . . . . . . . . . . . . . . .
. Right carotid ang iog raph y onl y: thi s supplied the ri ght posterior ce re bra l arte ry .
5 French Mani used .
t
X
C
2
3
X
X
C
C
5
6
X
X
X
C
C
R
R
C
C
X
X
V S'
X
V
V
V
9
10
11
X
X
X
X
X X
C
R
C
RC
C
C C
X
S
X
X
8
8
8
V
X
X
X
8
t
12
8
TOlals
3
9
8
7
1
4
2
68; 5V
2
AJNR:4. Jul. / Aug . 1983
ANGIOGRAPHIC AMNESIA
angiography. The nine patients had both their anterior and
posterior c irc ulations stud ied. Cases 4-7 and 10-12 had
cl ini cal and/or radiologic evidence of cerebrovascular d isease. Five of the nine cases (cases 4 , 6-8 , and 12) also
had had a previous episode of transient g lobal amnesia. The
precise duration of that amnesia was diffi c ult to evaluate,
but five patients had recovered within 4 hr and all had
recovered by 24 hr.
Seven patients (cases 1, 2, 6, 7, and 9 -1 1), questioned
at various times during their hospital stay, could not recall
the angiogram . Four patients (cases 3, 5, 8, and 12) were
not asked if they cou ld remember the angiogram. Case 4
was amnesic for 3 hr after angiography, but later c laimed to
remember the angiogram . One of the older patients, case 7,
developed transient signs indicative of vertebrobasi lar ischemia during his amnesia. No new signs occurred in any
other patient. Computed tomography (CT) was performed
on all patients before angiography and was normal in all
except case 2, the patient with the thalamic glioma.
Clinical vertebrobasilar d isease and a history of transient
global amnesia was prominent in those patients who developed transient g lobal amnesia after angiography. Seven
(4 %) of 167 patients presenting with vertebrobasilar disease developed postangiographic transient g loba l amnesia.
Excluding all patients who may conceivably have had temporal lobe ep il epsy, only one (0.1 % ) in 1,200 of those
patients without vertebrobasilar disease developed postangiographic transient g lobal amnesia. A patient who developed postangiographic transient g lobal amnesia was at least
10 (and not more than 30) times more li kely to have a history
of transient global amnesia than one who did not develop
this complication.
In case 4 , a patient with spasm of the basilar artery in
whom it was important clinically to prove that a basilar tip
aneurysm was present , four injections of 6 ml of contrast
material were made into the left vertebral artery, and it was
estim ated in this case that 30 ml of the total 50 ml of con trast
agent used for the investigation entered the posterior circulation.
Technical problems with catheter manipulation were recorded in four patients (cases 5, 7 , 10, and 11). All the
amnesic compli cations occurred in the hands of experienced neuroradiologists with the exception of two of the
younger patients (cases 1 and 3), who were catheterized by
a resident under supervision.
In five patients the amnesic episodes followed subclavian
injections using a 7.3 French Head Hunter catheter . There
were no episodes of amnesia when subclavian angiography
was carried out using 6 French Kifa or 5 French Mani
catheters. In five patients the amnesic episodes followed
selective vertebral injections using a 6 French Kifa catheter.
One patient developed amnesia after a selective vertebral
injection with a 5 French Mani catheter. In the patient in
whom the posterior cerebral artery originated from the carotid artery , the amnesic episode followed carotid angiography using a 7.3 Frenc h Head Hunter cathete r. When the
re lation of catheter size and the sma llest artery injected in
age-matched groups is studied (table 1), a soft size 6
catheter is safer than a stiff size 7.3 catheter in a subclavian
981
artery and a size 5 catheter is safer than a size 6 ca theter
in a vertebral artery.
Cases 6 and 9 developed amnesia with the same batch of
co ntrast material (table 3), and at that time six of nin e
consecutive patients developed complications. Five of th ese
had neurologic comp li cations, and one had an all ergic co mplication. Three of the patients had the prese nting neurologi c symptoms reprodu ce d. The probability that two of fiv e
consecutive patients wo uld develop amnesia afte r angiography when the approximate incidence of amnesi a is 1 %
(12 in 1,520) is 9.7 X 10- 4 , and the probability of fiv e of
nine patients developing neurologic comp li cations when our
neurologi c compli cation rate is 2 .2% (table 2) is 6 x 10- 7
using binomi al probability .
Three patients had a history of temporal lobe epilepsy
and five patients had a hi sto ry of transient g loba l amn esia.
Seven patients had either c lin ical or radiologic evidence of
vertebrobasi lar disease (table 3). Overall , therefore , 11 of
12 patie nts had ev id ence of disease in th e territories of th e
posterior ce rebral arteries; case 9 was the on ly exce ption .
These 11 patients were regard ed as having " susceptibl e
brains. " Situations likely to produce atheromatous debri s or
catheter c lot emboli occurred in 11 cases (table 3). In thre e
patients contrast emboli or to xicity may have contribu ted to
the deve lopment of th e amnesia , and in case 9 it was most
strong ly suspected. Thi s 60-year-old man presenting with
amaurosis fug ax had no radiologic or c lini cal evid ence of
vertebrobasilar disease , and his technically uneventfu l angiogram was obtained by an experienced operator using a
6 French Kifa catheter in the vertebral artery. Thi s pati ent
was one of the six of nine consecutive patients who developed complications during ang iography using the same
batch of contrast material.
Discussion
Transient global amn esia is a syndrome desc ribed by
Fisher and Adams [6] in 1964 in which th ere is a sudden
loss of memory for the present and recent past in individu als
in or past middle age, and the patients cannot retain new
facts although the remote memory is intact . Charac teristicall y, the patient keeps asking the same question. There is
no loss of consc iousn ess, no physical signs , and th e pati ent
recovers in several hours but has amnesia for th e period of
th e attack. Single occ urren ces are the rul e but multipl e
attacks have been reported [7 -9]. Th e authors speculated
that th e cause of the transient global amnesia might be a
localized ep ilepti c discharge in the temporal lobe, a th eo ry
subsequently favored by few [10 , 11], or an ep isode of
ischemia involving the limbic system of the medial temporal
lobe. The generally accepted theory is that it is du e to
transient vascular insufficiency of the arteries supplying th e
medial temporal lobe [12]. Our series has two distinct groups
of patients-a younger group of three labeled as tem poral
lobe epi lepsy and an o ld er group of nin e classified as
transient global amnesia .
Of our 1 ,520 pati ents, 12 (0 .8% or one in 125) developed
postangiographi c amn esia , yet thi s co mpli ca ti on is rarely
recorded by others [1 -5]. Th e com pli cati o n rate for patients
982
PE XM AN AND CO ATE S
with cereb rovascul ar disease and subarachn oid hemorrh age in the 2,316 cases of M ani and Ei senberg [1 3] was
1.7 % . They did not say wheth er am nesia occurred . Ei se nberg et al. [1 4] reported a co mplicati o n rate of 1.3% in 301
cases of cerebrovasc ul ar d isease, but they d id not attempt
either ve rtebral or subc lavian catheteri zati o n. Th ey repo rted
no am nes ic com pli cati ons. Alth o ug h our CN S complicati on
rate was 2 .2% , we had a hi g h num ber of mino r transient
CNS compli cations (0.9% ) compared w ith Mani et al. [1] ,
possibly due to o ur lig ht sedati on. Because o ur overa ll
comp licati on rates are si mil ar to th ose of other investigators,
we spec ul ate th at oth ers may not have recog nized amn esia
d ue to heavie r sedation. We wo ul d hav e been un aware of
these com pl icati o ns if th e pati ents had been premedicated
with neuro lepta nalgesia [ 15] or Demerol or had th e in vesti gati on been pe rformed unde r general anesth es ia. The type
of sedati o n used by M ani et al. [1 ] and by Ei se nberg et al.
[ 14] is not recorded. Palmer et al. [1 6] , w ho injected th e
innominate and subc lavian arteri es, may not have noticed
amnesic compli cations because of th e neuroleptanalges ia.
Sil ent radio log ic emboli from cereb rofe moral ang iog raphy
were desc ri bed by Cro nqvi st [17].
Ou r statistics for vertebral and subc lavian ang iog raphy
suggest th at a small er, softer cath ete r will produce th e
fewest complicati ons. Ei se nberg et al. [14] stated th at a 5
Frenc h cath ete r has a surface area onl y half th at of a 7
Frenc h cath eter , w hi c h would correspo ndi ngly dec rease th e
potential for emboli , and th at the stiffer cath eters, wh ose
torque facilitates cath eterizati on, may also damage th e intima by flailin g of th e cath eter tip . Wi shart [ 2] , wh o reco rd ed
six cases of amnesia in 447 cases, used an 8 Frenc h Hinc k
Head Hunter cath eter [1 8]. De Tribolet et al. [4 ], wh o desc ribed Ko rsakoff syndrome in nin e of 832 vertebra l angi ograms, used a 6.5 Fre nc h BD red cath eter, and we used a
7 .3 Frenc h Head Hunter in six of o ur cases. Lin [19] recomm ended onl y usi ng bigger Head Hunters fo r diffi c ult
cases, preferring, as do M ani et al. [ 1] and Eisenberg et al.
[1 4 ], the soft 5 Frenc h cath eter.
Wishart [2 ] also stated th at most of hi s cases we re done
by res idents, and M ani et al. [ 1] fo und in 5, 0 0 0 cases th at
res idents prod uced 4 .5 tim es as many co mpli cati o ns as
mo re experienced operato rs. Alth oug h Olivec ro na [20], w ho
revi ewed a simil ar num ber of cases, d isag rees, comparisons
suggest th at reside nts in Californ ia and Stock holm have
sim il ar compli catio n rates. Two of our cases occ urred in th e
hands of a resident .
In Europe, fi ve commo n brands of contrast medi um have
been fo un d to contain intrin sic parti c les [ 2 1]. In rub be rtipped via ls th ere are betwee n 6 1 and 369 partic les measurin g 5- 10 fLm, 45 to 183 measuring 10 - 3 0 fL m , and three
to f ive measuring greater th an 30 fLm in va ri o us brand s in 1
ml of contrast mate ri al. Th ese small pa rtic les coul d embo li ze
into th e territori es of both posterior cerebral arteri es, th e
reg ions thoug ht by Horel [22] to be responsibl e fo r amnesia,
and cou ld acco unt fo r our cases of am nesia in w hich verteb robasil ar disease or tempo ral lobe epil epsy was present
in 1 1 of o ur 12 patients. Two cases described by Wales and
Nov [3] occurred wh ile usi ng the same batc h of contrast
medium (Con ray-60), and it was theo rized that th e compli-
AJNR:4 , Ju l. / Aug . 1983
cati ons we re du e to contrast emboli . We had two simil ar
cases associated wi th one batc h of co ntrast medium .
On e of our cases of amn esia occ urred w hen about 30 ml
of co ntrast materi al was inj ected into a ve rt e bra l artery to
di ag nose a bas il ar aneurysm in th e presence of so me
spasm. Th is amnesia may be d ue to contrast tox ic ity in a
brain made susceptibl e by ath erosc lerosis and arteri al
spasm. Tox ic effects of co ntrast medium ca n occ ur wi th
v e ry hig h doses of mod ern age nts [23].
In 14 patients with tra nsient gl obal amnesia reported by
Matth ew and Meye r [7] , 1 1 had c lini cal and 1 2 had rad iolog ic ev ide nce of vertebrobas il ar vasc ul ar di sease; and in
our nin e cases of transient glo bal amnesia, seve n had
c linical and three had rad iol og ic evid e nc e of vertebro basil ar
disease. Five of our nine cases of transient global
amnesi a had a previ ous amnesic epi sode, and all of our
cases di ag nosed as temporal lobe epil epsy had had a simil ar
epi sode. All th ese factors suggest th at th e re are areas in
th e hi ppocampi made susceptibl e by disease processes that
can be tri ggered either by an o perato r o r co ntrast-induce d
factor. Ov e rd oses with di azepam may produce transi ent
gl obal amnes ia [24] , but in our cases 10 mg oral Val ium or
30 mg Serax was c hose n to prod uce mild tranq uil izatio n
with minimal drowsin ess.
REFEREN CES
1. Man i RL , Ei senberg RL , Mc Donald EJ, Pollock JA, Man i JR .
Complicat ions of cath eter ce rebral arteriog raph y: analysis of
5000 procedures. 1. Criteri a and inc idence. AJR 1978 ;
13 1 :86 1-865
2 . Wi shart DL. Comp licati ons of ve rtebral angiography as co mpared to non-vertebral cereb ral angiog raph y in 447 stud ies.
AJR 1971 ;113:527- 537
3. Wales LR, Nov AA. Transien t global am nesia: complication of
ce rebral ang iography. AJNR 1981 ;2 : 275 - 27 7
4. de Tribolet N, Assai G, Oberson R. Synd rome de Ko rsakoff et
cec ite corti cale transi toires apres ang iog raphi e ve rtebrale.
Schweiz Med Wochenschr 1975; 105: 1 506-1 509
5. Shuttleworth EC, Wi se GR . Transient global amnesia due to
arterial emboli sm. Arch Neuro /1973; 29: 3 40-3 4 2
6. Fi sher CM, Adams RD . Transient global amnesia. Acta Neurol
Scand 1964 ;40[SuppI 9 ];7 - 83
7. Matth ew NT, Meyer JS. Path ogenesis and natural history of
transient global amnesia. Stroke 1974;5: 303-3 1 1
8. Steinmetz ES, Vroom SQ. Transien t g lobal amn esia. Neurology (N Y) 1972 ;22: 1193-1 200
9. Poser CM, Sieg ier OK. Temporary amn esia as a manifestation
of ce rebral vasc ular in suffic iency. Trans Am Neurol Assoc
1960;85:22 1-223
10. Green HH , Bennett DR . Transient global amnesia with previously un re ported EEG abnormality. Electroencepha logr Clin
Neurophysio /1974 ;36: 409 - 413
11 . Lou HOC. Repeated episodes of transient global amnesia. Acta
Neurol Scand 1968;44: 6 12 - 6 18
12 . Shu pin g JR, Rollin son RD , Toole JF. Transient global amnesia.
Ann Neuro /1980 ;7: 28 1-285
13. Mani RL , Eisenberg RL. Complicat ions of cath eter ce rebral
arteriog raphy: analys is of 5000 proced ures. II. Relati on of
co mplicati on rates to clinical and arteri ographi c diag nosis. AJR
1978;13 1 :867 - 869
14. Ei senberg RL , Bank WO, Hedgcock MW . Neurolog ic compli-
AJNR :4, Jul. / Aug. 1983
15 .
1 6.
17 .
18.
ANGIOGRAPHIC AMNESIA
cati ons of angi ograph y for ce rebrovasc ular d isease. Neuro logy
(N Y) 1980;30 : 8 95-8 97
Hoy RJ , Sorby WA . " Minineuro lept. " A note on neuroleptanalgesia in radiologic procedures. Aus tra las Radiol 1968;
1 2:2 68 - 26 9
Palmer FJ, Bri scoe PJ , Sorb y WA, Barry BP , Willi ams RM .
Transfemoral sel ec tive angiograph y in th e in vesti gati on of ce rebral isc haemi c di sease. Review of 400 co nsecutive studi es.
1 . Techniqu e and evalu ati on of meth od . Aus tralas Radiol
1977;2 1 :3 1-38
Cronqvist S, Efsin g HO, Palacios E. Emboli co mpli cati ons in
ce rebral angiograph y with th e cath eter tec hniqu e. Ac ta Radiol
[Diagn] (Stoc kh) 1970; 10 : 9 7 -1 07
Hinck VC, Judkins MP , Paxton HO . Simplified se lecti ve femo-
983
roce rebral angiograp hy . Radiology 1967;89: 1 048-1 052
19. Lin JP . Techniques of ce rebral angiog raphy. Radiol Clin North
Am 1974;1 2:223-2 4 0
20 . Oli vec rona H. Complicati ons of ce rebral angiog raphy. Neuroradiology 1977; 14: 1 75-1 8 1
2 1 . Windin g O. Intrinsic parti c les in angiog raphic con trast med ia.
Radiology 1980;1 34: 317 -320
22. Horel JA . Neuroa natomy of amn esia-a c riti que of hippocampal memory hypothesis. Brain 1978; 101 : 403-445
23. Murph y OJ . Cerebrovasc ular perm eability after meg lum ine iothalamate ad mini strati on. Neurology (NY) 1973;23 : 926-936
24. Gil bert JJ , Benson OF . Tra nsient global am nesia. Repo rt of two
cases with definit e eti olog ies. J Nerv Ment Dis 1972; 154 : 461 464