Sodium Amobarbital Memory Tests: What Do They Predict?

BRAIN AND COGNITION 33, 189–209 (1997)ARTICLE NO. BR970892

Sodium Amobarbital Memory Tests: What Do They Predict?

L. A. DADE AND M. JONES-GOTMAN

Montreal Neurological Institute and Hospital, McGill University, Montreal, Canada

Severe transient postoperative memory deficits among epilepsy patients with re-section from one temporal lobe may be indicative of increased risk for amnesiahad more extensive removal of mesial structures occurred. Immediate postoperativetesting may provide some validation for risk of amnesia as predicted by the intraca-rotid sodium amobarbital memory test (IAP-M). Thirty patients (24 not consideredat risk for amnesia and 5 who failed the IAP-M) were tested on the first, second,and third days following resection from the right or left temporal lobe. Resultssuggest that the IAP-M paradigm used does not necessarily predict postoperativememory performance. 1997 Academic Press

INTRODUCTION

Since the original reports of unexpected amnesia following unilateral tem-poral-lobe resection for the treatment of medically intractable epilepsy (Mil-ner, 1958; Milner & Penfield, 1955), the intracarotid sodium amobarbitalprocedure (IAP) has become widely used to assess risk of postoperative am-nesia. The IAP memory test (IAP-M) is thought to test the patency of brainstructures important for memory. Patients who are unable to remember testitems introduced after drug injection (anaesthetisation) into the proposed sur-gical hemisphere are believed to have additional damage in the contralateralhemisphere and could be at risk for postoperative amnesia. Therefore, thesepatients may not be offered surgery, or they may receive a limited resectionwith sparing of the mesial structures.

In recent years controversy has arisen surrounding the validity of theIAP-M procedure (Jones-Gotman, Barr, Dodrill, Gotman, Meador, Rausch,

We thank the subjects for graciously consenting to participate in these tests during the earlypostoperative period, and we are grateful to Drs. A. Olivier, J.-G. Villemure, and R. Leblancfor the opportunity to study their patients. We also thank Rhonda Amsel for statistical advice.This work was supported by Grant MT-10314 awarded to M. Jones-Gotman and R. J. Zatorreby the Medical Research Council of Canada; it was presented in part at the annual meetingof the American Epilepsy Society, December 1994. Address reprint requests to L. A. Dade,at the Montreal Neurological Institute, 3801 University Street, Montreal, Quebec, Canada H3A2B4.

1890278-2626/97 $25.00

Copyright 1997 by Academic PressAll rights of reproduction in any form reserved.

190 DADE AND JONES-GOTMAN

Sass, Sharbrough, Silfvenius, & Wieser, 1993; Rausch, Silfvenius, Wieser,Dodrill, Meador, & Jones-Gotman, 1993). While there have been reports ofpatients who failed IAP-M tests from injection on the side of proposed sur-gery and who went on to have surgery and became amnesic (Loring, Her-mann, Meador, Lee, Gallagher, King, Murro, Smith, & Wyler, 1994a;Rausch, Babb, & Brown, 1985; Rausch et al., 1993), there have also beenreports of patients who passed the tests and suffered from amnesia postopera-tively (Barr, Schaul, Decker, & Lantos, 1992; Rausch et al., 1993). As well,several studies have examined patients who had failed the IAP-M tests butwho underwent temporal lobe surgery including the hippocampus withoutsubsequent amnesia (Girvin, McGlone, Mclaughlin, & Blume, 1987; Jones-Gotman, 1992; Jones-Gotman et al., 1993; Loring, Lee, Meador, Flanigin,Smith, Figueroa, & Martin, 1990; Novelly, 1987).

At the Montreal Neurological Institute (MNI), patients who fail the IAP-Mtests after injection on the side of intended surgery usually undergo a resec-tion that spares the hippocampus and the parahippocampal gyrus. However,even with this type of limited resection, severe transient postoperative mem-ory deficits have been observed. Jones-Gotman (1987) reported the case ofpatient WT, who had epileptiform disturbance in the left temporal lobe. Inhis preoperative amobarbital procedure he failed the memory test after left-sided injection, and following left anterior temporal-lobe resection with hip-pocampal sparing he suffered a transitory global amnesia during the immedi-ate postoperative period.

WT’s postoperative memory deficit suggested that he was indeed at riskfor amnesia as predicted by the IAP-M test, and one could infer that a perma-nent amnesia might have resulted if the surgical resection had extended fur-ther into hippocampal structures. This case raised an interesting question:would it be possible to verify the accuracy of a preoperative IAP-M progno-sis of risk for amnesia through early postoperative memory testing?

Because the cognitive status of patients in the first few postoperative dayshas not been studied systematically, it is possible that cases with transientearly memory deficits have gone unnoticed. If patients, determined to be atrisk for amnesia, show transient deficits postsurgically in a fashion similar toWT, this would provide some confirmation of the validity of the amobarbitalmemory procedure.

During the days following brain injury, swelling of tissue as well as adecrease in glucose metabolism occurs (Langfitt, Obrist, Alavi, Grossman,Zimmerman, Jaggi, Uzzell, Reivich, & Dreux, 1986). This metabolic de-crease has been correlated with impairments in behavioral measures in ratsduring the first 3 days following small lesions to the parietal cortex (Colle,Holmes, & Pappius, 1986). Similar metabolic decreases have been docu-mented in humans following head trauma, with a more notable decline ofcerebral glucose metabolism in the hemisphere of greater injury (Langfitt et

SODIUM AMOBARBITAL MEMORY 191

al., 1986). This may also occur in the postsurgical epilepsy patient, leadingto early transient impairments in cognitive functions.

Swelling and decreased brain metabolism could create particular difficul-ties for the patient who is at risk for postoperative amnesia. In these patientsthe mesial structures on the side of surgery are thought to be crucial forsustaining a patent memory system; therefore, interference with the function-ing of these areas could induce memory impairment. Thus, despite a limitedtissue removal, the transient postsurgical effects might be sufficient to inter-fere significantly with memory function in the early postoperative period,resulting in a greater deficit than that related to the excision alone.

The present study was designed to examine systematically the early post-operative performance of patients with surgical removal from one temporallobe. Some of the memory tests used consisted of highly polarized (verbal vs.nonverbal) material in order to investigate material-specific deficits; duallyencodable material was also used to test for more global memory loss.

This study compared the performance of patients who had passed theIAP-M test after injection on the side of proposed surgery with that of pa-tients who had failed that critical test. Effect of side of temporal-lobe resec-tion was also examined. It was expected that patients with right-sided re-moval would perform poorly on the nonverbal tasks and that patients witha left-sided removal would show a decrement on the verbal ones. Addition-ally, the most important prediction was that patients who had failed the criti-cal amobarbital memory test would show a more severe postoperative mem-ory deficit than those who had passed it.

METHOD

Subjects

Thirty patients were studied; all had undergone unilateral resection from a temporal lobe(18 left resection (LR), 12 right resection (RR)) for the relief of intractable epilepsy and hadnormal speech representation in the left cerebral hemisphere. All had Full-Scale WAIS-R IQ(Wechsler, 1981) ratings above 75, no evidence of gross cerebral damage, and were undergoingtheir first surgical procedure for the treatment of epilepsy. Eleven normal control (NC) subjectswho had no known history of brain injury or neurological illness were also tested on thecognitivetasks. Control subjects were matched to the patient groups for age and level of education.

Eighteen of the 30 patients had undergone a preoperative IAP (13 of these received the testspecifically for evaluation of memory). The remaining 12 had demonstrated adequate memoryon other preoperative neuropsychological measures and were not required to undergo an amo-barbital procedure (Jones-Gotman, 1991); they were considered equivalent to the patients whohad passed the IAP-M tests and were included in the ‘‘pass’’ groups. These Pass subjectsacted as patient controls for subjects who failed the IAP-M.

Patients who had failed IAP memory tests after the critical injection (into the proposed sideof surgery) were the focus of this study. As most patients pass IAP-M tests ipsilateral to theirseizure focus, the memory failure group is unavoidably small. In our sample, four patientsfrom the LR group (Cases 1–4, Table 1) and one from the RR group (Case 5) failed the


TABLE 1Surgeon’s Estimation of Resection

HippocampusSubjects (in cm) Amygdalaa

LR IAP-Pass Group Mean, 2.1 5 MostRange, 2–3 7 Complete

RR IAP-Pass Group Mean, 2.1 6 MostRange, 0–3.5 5 Complete

Case 1 0.3 CompleteCase 2 2.5 CompleteCase 3 1.0 CompleteCase 4 0.5 MostCase 5 1.5 CompleteCase 6 2.0 Most

a Amygdala removals dichotomized into ‘‘most’’ or‘‘complete.’’

relevant test. In addition, one patient (Case 6) who had passed the IAP-M had unexpectedpostoperative memory deficits and will be discussed separately.

The patients were grouped by side of eventual temporal-lobe resection. Surgeons’ estimatesof resection from amygdala and hippocampus are shown for pass groups and individual casesin Table 1.

Pre- and postoperative results from four of the neuropsychological tests given for clinicalpurposes, at 2 weeks following surgery, are shown in Table 2. The protocol for postoperativeneuropsychological assessment at the MNI was changed while this study was in progress;therefore, 8 of the 30 subjects did not receive a full Wechsler IQ test or the Wechsler memory(Wechsler, 1945) subtests postoperatively.1

Amobarbital Procedure

Patients were selected to undergo the amobarbital procedure based either on suspicion ofatypical hemispheric dominance for language or on evidence of deficient memory from nonin-vasive memory tests (Jones-Gotman, 1991). Patients whose psychology results were in dis-agreement with EEG or neuroimaging lateralization, or who had bitemporal abnormality onEEG recordings, were also selected for an IAP. The procedure used in our study was as follows:injection of sodium amobarbital (mean dose, 140 ml; range, 125–175 ml) was made transfemo-rally via a catheter into the internal carotid artery. Simultaneous EEG recording was performed.The memory items were one object, two pictures, a word, and a sentence, which were presentedwhile the drug was active, interspersed with speech testing; this was the traditional MNI test(Jones-Gotman, 1987). In most patients, tests were considered valid if memory items werepresented before EEG slow waves had dissipated; in rare cases without EEG coverage, recov-

1 Of the 15 IAP-M Pass subjects (8 LR, 7 RR) who had the full battery of pre- and post-operative clinical testing no significant differences were found on the WAIS-R Full Scale IQ,Wechsler Memory Scale designs, or the complex figure. There was a difference on the Wechs-ler Memory Scale stories: the Group 3 Time of Testing interaction was not significant, butthe main effects of group and time of testing were. The group effect was attributable to poorerperformance of the LR group (F(1, 13) 5 5.04 p , .05), and both groups performed worsepostoperatively than preoperatively.


TA

BL

E2

Pre-

and

Post

oper

ativ

ePe

rfor

man

ceon

Neu

rops

ycho

logi

cal

Tes

ts

WM

Sst

orie

sW

MS

desi

gns

Com

plex

Figu

reFu

llsc

ale

IQde

laye

dre

call

dela

yed

reca

llde

laye

dre

call

Pre-

2W

eeks

Pre-

2W

eeks

Pre-

2W

eeks

Pre-

2W

eeks

Subj

ects

oper

ativ

epo

stop

erat

ive

oper

ativ

epo

stop

erat

ive

oper

ativ

epo

stop

erat

ive

oper

ativ

epo

stop

erat

ive

LR

Pass

Gro

upN

58

90.6

89.5

5.8

3.3

7.7

8.75

13.2

16.1

(N5

13)a

(89.

0)(5

.9)

(7.7

)(1

4.6)

RR

Pass

Gro

upN

57

93.7

96.7

8.8

7.2

5.7

8.6

12.5

16.3

(N5

11)a

(94.

6)(7

.2)

(5.9

)(1

1.7)

Cas

e1

7663

0.25

0.0

5.0

1.0

6.0

2.0

Cas

e2b

8990

5.25

4.0

2.0

4.0

9.0

9.0

Cas

e3

7779

1.25

1.25

3.5

8.0

6.5

17.5

Cas

e4

113

101

00

3.0

2.5

5.5

6.0

Cas

e5

9894

3.0

6.0

6.0

4.5

6.0

6.5

Cas

e6

8794

6.0

0.0

5.5

0.0

7.0

4.0

Not

e.W

MS,

Wec

hsle

rM

emor

ySc

ale.

aN

ine

ofth

e24

Pass

subj

ects

did

not

rece

ive

this

post

oper

ativ

ete

stin

gba

ttery

due

topr

otoc

olch

ange

sdu

ring

the

time

ofst

udy.

bT

hree

wee

kspo

stop

erat

ive.


ery from the injection-induced hemiplegia, as determined by grip strength and fine fingermovements, was the criterion. After full recovery, memory was tested by a yes–no recognitionprocedure. Two or more recognition errors (2/5 or 40% errors) constituted a failure of theIAP-M test.

Early Postoperative Test: Materials and Procedure

Test Materials

Five comparable sets of test materials were prepared, for use on 5 different days of testing.Each set consisted of five categories of stimuli: objects, pictures selected from the PeabodyPicture Vocabulary Test—Revised (Dunn & Dunn, 1981), photographs of faces, writtenwords, and orally presented sentences. For each stimulus type there were eight items: threetargets and five recognition foils.

The objects and pictures were selected to promote easy verbal and pictorial encoding ofthese relatively nonverbal stimuli. To encourage the same kind of dual encoding with thewords, subjects were requested to form a visual image of what the word represented. Allobjects and pictures were familiar items. The words were taken from among 925 nouns ratedfor concreteness, imagery, and meaningfulness (Paivio, Yuille, & Madigan, 1968); all wererated high on the visual imagery scale (range 6.0–6.83, scale of 7) and were thus deemed tobe easily pictured.

Two types of stimuli that were more polarized into the verbal or nonverbal domain, sentencesand photographs of faces, were introduced to test for more material-specific deficits. The sen-tences were composed of abstract words that expressed concepts rather than concrete actionsor objects (e.g., Cathy couldn’t concentrate), making them difficult to visualize. They wereprerecorded on a cassette tape and played on a cassette recorder during testing. The picturesof faces were black and white 4.5 3 5.0-cm photographs mounted on white cards. It wasexpected that they could not be verbalized and that recognition would be dependent on visualimagery.

The distractor tasks were verbal and consisted primarily of simple speech tasks, to allowinvestigation of any dissociations between language and memory deficits.

Test Procedure

Order of presentation of the different test sets across days was determined randomly. Sub-jects were tested once preoperatively to establish baseline performance. Testing was then car-ried out at bedside on the first, second, and third days postoperatively. The fifth set of testingmaterials was used on the seventh postoperative day for patients who had continued to displaysevere deficits on the third day of postoperative testing.

The three target items from each of the five stimulus categories were presented in the follow-ing order: objects, words, pictures, faces, sentences. Presentation of the three target items fromeach category was followed by an immediate recall test. If subjects failed to recall all threeitems at that time, the items were shown again. Delayed recall and recognition of the materialswere tested following a distractor interval of approximately 5 min. Faces were tested forrecognition only. A yes–no recognition procedure was used in all cases, with stimuli pre-sented one at a time and target items interspersed amongst five foil items, for each category(Fig. 1).

FIG. 1. Examples of early postoperative test stimuli: Within each category the top threeitems are target stimuli, the bottom five items are recognition foils.



Scoring

Free recall. One point was allotted for each item recalled correctly (maximum recall score,3 per category). Intrusions occurred so rarely that they were not calculated in recall scores.Because sentences varied slightly in length across the different stimulus sets, a percentagescore for the number of words recalled was used. This procedure provided credit for partiallyrecalled sentences.

Recognition. To take into account the number of false positive responses that occurred inrecognition, a subtraction procedure (correct recognitions minus false positive responses) wasemployed. Thus the maximum possible score was 3 and the minimum score was 25 (0 correctand 5 false positive responses) per type of material.

RESULTS

Recall results were analyzed using a three-way analysis of variance(ANOVA) comparing LR and RR IAP-M Pass and NC groups, day oftesting, and stimulus category. Five separate two-way ANOVAs (Group 3Day) were carried out on the recognition data, one per stimulus category.Tukey HSD tests were used for post hoc analyses. Results of the IAP-Mfailure cases are presented and discussed separately, as are the results of onesubject (Case 6) who had passed the IAP-M test but had severe postoperativememory deficits.

Group Analyses: IAP-M Pass Groups vs. NC Group

Recall

The main effects of group [F(2, 32) 5 29.59, p , .000], day of testing[F(3, 96) 5 16.43, p , .000], and stimulus category [F(3, 96) 5 6.36, p ,.000] were all significant, as were the group by stimulus [F(6, 96) 5 4.33,p , .001] and group by day interactions [F(6, 96) 5 6.08, p , .000] (Figs.2 and 3). Group by stimulus by day did not yield a significant interaction(F 5 0.76).

Post hoc comparisons of the group by stimulus interaction revealed thatboth LR and RR groups performed significantly worse than the NC groupon word (LR q 5 14.28, p , .01; RR q 5 6.95, p , .01), object (LR q 58.52, p , .01; RR q 5 5.81, p , .01), and picture (LR q 5 10.21, p , .01;RR q 5 6.95, p , .01) recall. In addition, the LR group performed signifi-cantly worse than the RR group on recall of the verbal material: words (q5 7.33, p , .01) and sentences (q 5 10.55, p , .01) (Fig. 2).

Group by day interaction showed no differences among the three groupsat baseline, but postoperatively the LR group performed significantly worsethan both the NC and the RR groups across all three postoperative days (Day1 LR vs. NC q 5 10.36, p , .01; LR vs. RR q 5 5.81, p , .01; Day 2 LRvs. NC q 5 11.61, p , .01; LR vs. RR q 5 7.86, p , .01; Day 3 LR vs.RR q 5 13.73, p , .01; LR vs. RR q 5 7.44, p , .01). Only on Day 3 didthe performance of the RR group differ significantly from the NC group (q5 6.29, p , .01) (Fig. 3).


FIG. 2. Mean recall according to type of stimulus and side of surgical removal (averagedacross days). LR, Left Resection Pass Group; RR, Right Resection Pass Group; NC, NormalControl Group.

Recognition

There were no significant findings for recognition of objects or pictures,but there were for words, sentences, and faces (Table 3).

Words. Analysis of the recognition results for words showed significantmain effects of group and day and a significant interaction (Table 3). Posthoc tests showed that the groups did not differ preoperatively, but the LR

FIG. 3. Mean recall on each day of testing, averaged across stimulus types. LR, Left Resec-tion Pass Group; RR, Right Resection Pass Group; NC, Normal Control Group.


TABLE 3Anova Results for Recognition Performance of Pass and NC Groups

Group Day Group 3 daymain effect main effect interaction

F(2, 32) F(3, 96) F(6, 96)

Words 10.33** 4.87** 4.70**Sentences 25.97** 3.87* 3.31**Faces 8.00** 4.77** 1.14

* p , .01** p , .001

group was impaired compared to both NC and RR groups on the second (LRvs. NC q 5 4.83, p , .05; LR vs. RR q 5 5.37, p , .05) and third (LR vs.NC q 5 9.39, p , .01; LR vs. RR q 5 8.33, p , .01) postoperative days.The RR and NC groups did not differ.

Sentences. The main effects of group and day, as well as the interactionof group by day were significant (Table 3). Post hoc analysis again showedno difference between groups preoperatively but an impairment in the LRgroup compared to NC across all postoperative days (Day 1 q 5 5.35, p ,.05; Day 2 q 5 8.58, p , .01; Day 3 q 5 8.47, p , .01). In addition, LRperformed significantly worse than RR on Day 2 (q 5 7.50, p , .01) andDay 3 (q 5 6.66, p , .01). There were again no differences between theRR and the NC groups.

Faces. Analysis of the face recognition task revealed a significant maineffect of group and day but no interaction of group by day (Table 3). Posthoc analyses indicated that on this nonverbal task, RR as well as LR wereimpaired (NC vs. LR, q 5 3.89, p , .05; NC vs. RR, q 5 5.68, p , .01).The LR and RR groups did not differ significantly. However, given our smallsample size, the power to detect a large effect (effect size 5 .4) is only 50%(Cohen, 1988). Therefore, with a larger sample these nonsignificant groupdifferences could potentially be significant. The main effect of day reflecteda decline in performance from baseline to postoperative Days 1 and 2 (Day0 vs. Day 1, q 5 5.08, p , .01; Day 0 vs. Day 2, q 5 3.90, p , .05).

Results of Failure Cases

Recall

Among the four LR failure cases, two (Cases 1 and 2) showed profoundpostoperative recall deficits. Case 1 was unable to recall any target item,while Case 2 recalled only 1/36 items across all 3 postoperative days. Onlyone of the left IAP-M Pass subjects performed this poorly. The recall perfor-mance of the other two left resection failure subjects (Cases 3 and 4) andthe right failure subject (Case 5) was equal to, and sometimes surpassed, theperformance of their respective patient control groups.


FIG. 4. Postoperative recognition performance (correct minus false positive responses) ofleft IAP-M Failure Cases (1,2,3,4) vs. LR IAP-M Pass Group (represented by the horizontalline). Case 2 scores 5 0. LR, left resection.

Recognition

Recognition performance averaged across postoperative days showed thatCases 1 and 2 also showed deficient recognition compared to the Pass sub-jects (Fig. 4; Table 4). Poorest performance was on the verbal material: Case1 averaged 22.67 on words and 20.67 on sentences across the 3 postopera-tive days. This subject’s negative scores reflect a large number of false posi-tive responses as well as a low number of correct responses (e.g., words: 5hits, 13 false positive responses across 3 days). Case 2 did not recognizeany of the words or any of the sentences postoperatively compared to hisperfect performance on these two categories preoperatively. In addition,these two patients made errors across all 3 days on the easier tasks of objectand picture recognition, where both LR and RR groups had performed virtu-ally without error. Their face recognition was also well below the averagefor the LR Pass cases.

Conversely, Cases 3 and 4 performed above the LR group average forrecognition of objects, words and sentences. Case 3 also performed abovethe LR average for picture recognition, while Case 4 was only 0.25 belowthe group average. Case 4 performed well on face recognition, scoring 0.93above the LR group average (1.74), while Case 3 exhibited a deficit in thisarea, obtaining an average score of zero across all 3 postoperative days.

Right-resection failure Case 5 performed above the average of the RRIAP-M Pass control group on recognition of objects, words, sentences, andpictures. However, performance on face recognition was below the RR groupaverage, with an average postoperative score of zero.


TA

BL

E4

Pre-

and

Post

oper

ativ

eR

ecog

nitio

nPe

rfor

man

ceon

Ear

lyPo

stop

erat

ive

Tes

t:Pa

ssG

roup

san

dFa

ilSu

bjec

ts

Obj

ects

Pict

ures

Face

sW

ords

Sent

ence

s

Pre-

Post

-Pr

e-Po

st-

Pre-

Post

-Pr

e-Po

st-

Pre-

Post

-Su

bjec

tsop

erat

ive

oper

ativ

eaop

erat

ive

oper

ativ

eaop

erat

ive

oper

ativ

eaop

erat

ive

oper

ativ

eaop

erat

ive

oper

ativ

ea

LR

Pass

Gro

up(N

513

)b3.

02.

952.

92.

92.

41.

72.

91.

952.

41.

0R

RPa

ssG

roup

(N5

11)b

3.0

2.9

2.9

2.9

2.4

1.3

2.9

2.8

2.8

2.55

Cas

e1

31.

33

2.0

30.

01

22.

71

20.

7C

ase

23

2.0

21.

32

0.7

30.

03

0.0

Cas

e3

33.

03

3.0

30.

03

2.0

21.

3C

ase

43

3.0

22.

71

2.7

22.

01

1.7

Cas

e5

33.

03

3.0

20.

03

3.0

32.

0

aR

ecog

nitio

nsc

ores

aver

aged

acro

ssal

l3

post

oper

ativ

eda

ys.

bA

vera

gegr

oup

scor

es.


Additional Testing of Case 1 and Case 2

Due to the poor memory performance of Cases 1 and 2, and to Case 1’sdysphasia, both were given an additional day of testing 1 week followingsurgery, using the extra set of testing materials. Their difficulty in recallcontinued, with neither subject able to recall any of the stimuli. This occurreddespite Case 1’s language recovery, showing that her memory deficit wasindependent from the speech deficit seen during the first 3 postoperativedays. Case 1 performed well on the easy object and picture recognition(3/3 each), but continued to have difficulty with word (21), face (0), andsentence (21) recognition. Case 2 scored zero on recognition of the words,faces, and sentences. He recognized all three objects but only two picturesand one face. On 2-week postoperative clinical neuropsychological testsCase 2 showed a level of recall similar to his preoperative performance,while Case 1 showed extremely poor recall that was somewhat worse thanher very low preoperative scores (Table 2). Therefore, the early postoperativetesting was able to detect severe memory deficits that were not obvious later.

Case 6: IAP-M Pass/Amnesic?

Unfortunately Case 6 was not given a baseline test because he spoke nei-ther French nor English and, having passed his IAP-M procedure, no memorydeficits beyond those typically associated with the surgical procedure wereanticipated. Due to the unexpectedness of this patient’s memory difficulties,initial testing did not occur until the 3rd postoperative day, and because ofthe severity of his deficits, the subsequent tests were performed on the 6thand 10th days following surgery. An equivalent form of the sentences inhis native language was not available, and therefore the sentences were notincluded. This patient was tested again at 1 year follow-up. Both sets of datawill be presented.

Postoperative

Case 6 had a striking recall deficit, with zero recall for all stimuli acrossall 3 postoperative days. Recognition was better than recall but still worsethan the LR Pass group (Fig. 5). Across all 3 days he recognized only 3/9target words and 3/9 faces (subtracting two false positive responses), withsomewhat better performance for objects and no difference from the compar-ison group for pictures. One should bear in mind that these deficits werepresent even though Case 6 was tested several days later than the other sub-jects. Thus he was able to recognize some items, but his recall deficit wasprofound, as demonstrated in his daily interactions in which he showed theclassic amnesic pattern of repeatedly asking the same questions. Severememory deficits were still apparent at 2 weeks following surgery, as shownby his clinical neuropsychological test results (Table 2).


FIG. 5. Postoperative recognition performance of Case 6 vs. LR Pass Group for each stimu-lus type (correct minus false positive responses). Performance of Case 6 is averaged acrossthe third, sixth, and ninth postoperative days vs. the average of the first, second, and thirdpostoperative days of the LR Pass Group. LR, left resection.

One-Year Followup

One year following surgery Case 6 was retested over 4 days with the sametests. There was no change in his severe recall deficit. Over all four sessionsand among all stimuli, only one word from one of the target sentences wasrecalled (1/69). Recognition was somewhat better than in the immediatepostoperative period, with recognition of nonverbal visual stimuli showingmore improvement than recognition of verbal information. Across 4 testingdays Case 6 scored 10/12 each for face and picture recognition and 11/12for object recognition. In contrast, word recognition (3/12) did not improvecompared to the early postoperative time period. Similarly, in his clinicalneuropsychological testing, he remained unable to recall anything of theWechsler Memory Scale stories or designs following a delay interval.

DISCUSSION

In this study of people whose sodium amobarbital memory failuresshowed them to be at risk for severe postoperative memory deficits, possiblyamnesia, we have shown two of five patients who failed the IAP-M to havesignificant memory deficits and three to be no different from the patientswho were not considered to be at risk. In addition one patient who passedthe IAP-M experienced postoperative amnesia.

IAP-M Pass Cases

In support of our notion that very early postoperative testing would elicit agreater memory impairment than is seen at the usual, later, neuropsychology


testing, we found that our control patients who had passed the IAP-M testshowed postoperative decrements on these very easy memory items. Thiswas in comparison both to their own preoperative baselines and to the healthycontrol subjects. Their impairments were most notable for the polarized ver-bal versus nonverbal material, which is suggestive of the greater interferenceoccurring in the operated hemisphere.

IAP-M Failure Cases

There was little consistency in the performance of the five failure casesin relation to their respective patient control groups. Two subjects (Cases 1and 2) appeared to be truly at risk, while Cases 3, 4, and 5 tended to performabove the average of the IAP-M Pass subjects, suggesting that they were nodifferent from patients who were not deemed at risk for amnesia.

Case 1 experienced a greater postoperative speech deficit than had the LRPass patients, but testing on the seventh postoperative day, after speech defi-cits had subsided, showed continued memory deficits that were thus indepen-dent of speech impairment. Recognition for the highly verbal material (sen-tences and words) was dramatically impaired, with this patient performingworse than all other subjects. Even on the easier tasks of object and picturerecognition, where the LR IAP-M Pass subjects performed virtually withouterror, Case 1 showed an impairment. Furthermore, she was unable to recall asingle item across all postoperative days, thus showing a very severe memorydeficit. Postoperative clinical neuropsychological testing performed 2 weeksafter her surgery revealed memory performance below that of her preopera-tive results but still within the same very low range (Table 2). However,these tests did not reflect the degree of deficit demonstrated with the simpletasks used in the early postoperative tests. Therefore, the early postoperativetest did add information about the frailty of her memory system. Also notethat her deficits cannot be attributed to greater extent of hippocampal exci-sion, as her resection was the smallest in the sample (Table 1).

Case 2’s language abilities following surgery were not different from thoseof the other LR cases, but his memory performance was deficient. Like Case1, Case 2 displayed a severe recall deficit: he was able to recall only oneitem across all 3 postoperative days (i.e., 1/36). This deficit was still presenton the additional day of testing 7 days following surgery. Postoperative rec-ognition performance remained at the low range of the LR Pass subjects andoccasionally fell below their lowest scores. As with Case 1, Case 2 alsoshowed impairment on the easy object and picture recognition, and this wasstill evident 7 days after surgery. As well, in striking similarity to patientWT (Jones-Gotman, 1987), Case 2 did not recognize his surgeon the firstday after surgery, had difficulty recalling his daughter’s name, and displayedthe amnesic pattern of repeatedly asking the same question. Thus this patientwas apparently truly at risk for a postoperative memory deficit as predictedby his failure on the IAP-M test.


Case 2 was seen 6 months following his surgery and, although he was nolonger suffering from seizures, his memory problems persisted. He com-plained that he could not remember what he had read in the newspaper afterreading it, and he complained of being unable to recognize some people hehad known before surgery. Formal testing with the neuropsychological clini-cal battery revealed very poor learning and recall for verbal and visual mate-rial, but as with Case 1 his scores were not markedly different from his verylow preoperative performance. Thus the scores on the clinical tests do notseem to corroborate the severity of his subjective memory difficulties in dayto day interactions. This may be because the tests used, with recall intervalsranging between 45 and 90 min, do not approach closely enough the demandsof everyday life.

IAP-M False Negative Case

Particularly important was Case 6, who was apparently at risk for postop-erative amnesia but was not predicted to be so by his IAP-M test. He showeda severe deficit after surgery in recall memory that still persisted when hewas last seen at 1 year follow-up. Why did he pass his IAP memory test?

Reexamination of his test showed nothing unusual: it had been interpretedbased on five critical items according to the traditional MNI protocol. How-ever, recent analysis of the IAP-M and EEG suggests that items shown whilecontralateral slow waves are present, during approximately the first 5 min(Bouwer, Jones-Gotman, & Gotman, 1993), are the most reliable indicatorsof memory impairment (Jones-Gotman, Bouwer, & Gotman, 1994). Loring,Meador, Lee, King, Gallagher, Murro, and Smith (1994b) have also foundthat objects presented early in the IAP-M test are superior to later item pre-sentation in predicting lateralized brain dysfunction. We reexamined theEEG measures obtained during Case 6’s amobarbital procedure. Contralat-eral slow waves were present for the first 3 min of the test, and only two‘‘experimental’’ items, which were not part of the traditional protocol usedat the time of his IAP tests, had been presented. Neither item was recognizedlater, which in retrospect could have signaled that he was potentially at risk.Nonetheless, the findings with regard to contralateral slow waves and mem-ory impairment still require further investigation and therefore provide onlya hypothesis rather than an explanation for this patient’s incorrect IAP-Mprediction. Consequently the general conclusion remains: the traditionalMNI IAP-M test, using five target memory stimuli and interpreted usingmeasures of hand strength and visual field recovery to determine return tobaseline functioning, failed to identify this patient as being at risk for amne-sia. A present modification of the MNI IAP-M includes the presentation ofmore items within the initial 3 min of testing.

There is understandably little in the literature addressing the question ofthe validity of the IAP-M as a predictor of amnesia, because the results of


the test itself influence the patient’s subsequent surgical treatment. However,a few studies have attempted to examine the issue in patients who receivedtemporal-lobe resections despite having failed the critical IAP-M test. Jones-Gotman reported significantly lower postoperative memory scores in groupsof patients who had failed the test than those who had passed (Jones-Gotman,1992; Jones-Gotman et al., 1993), but also before surgery, the group whohad failed performed worse than the group who passed, and no patient wasamnesic. Wyllie, Naugle, Awad, Chelune, Luders, Dinner, Skibinski, andAhl (1991) also looked at group data, but they found no difference in postop-erative memory score between the group that had passed versus the one thathad failed the IAP-M. Loring et al. (1990) compared pre- to postoperativememory scores within individual patients of a group that had failed IAPmemory tests, and they found no systematic loss after surgery. Finally, Do-drill and Ojemann (this issue) report differing concordance between memoryoutcome and IAP-M pass–fail results, depending on the approach used totest memory during the IAP (Dodrill & Ojemann, 1997).

Single case reports in the literature include patients who failed the proce-dure from the proposed side of surgery and became amnesic (Loring et al.,1994a; Rausch, Babb, & Brown, 1985; Rausch et al., 1993), those whopassed but were amnesic postoperatively (Barr et al., 1992; Rausch et al.,1993), and still others who failed the IAP-M and went on to have surgerywithout subsequent amnesia (e.g., Loring et al., 1992). The small numberof reports of postoperative amnesia suggests that it is a rare phenomenon,and therefore one has to question whether the IAP-M has the sensitivityrequired to detect risk for something with such a low rate of occurrence.

Dasheiff, Shelton, and Ryan (1993) found that patients with frontal lobeepilepsy had the same IAP-M failure rate as temporal lobe epilepsy patients,while patients with generalized seizures had nearly twice the failure rate.This suggests that IAP-M failure is not specific to dysfunction of the tempo-ral lobes and, by extension, perhaps not specific to memory interference. Theissue is further complicated by our limited understanding of the possibleeffects of disease processes on the brain. It is not known what impact damagein one brain region has on reorganization of function in other regions, owingto brain plasticity, and what effect such reorganization can have on IAP-Mresults. For example, if brain regions that will be removed during surgeryare anaesthetized and memory performance is maintained, it is typically as-sumed that a nonanaesthetized contralateral structure is sustaining this func-tion and that resection from the anaesthetized area will not cause a memorydeficit. However, if an alternate structure within the same hemisphere hassubsumed some memory function, and it has also been anaesthetized, a pa-tient could fail the IAP-M yet maintain adequate memory performance fol-lowing focal resection of the damaged structure.

In addition to unknown changes within the brain due to disease processes,there is the relatively unknown effect of the sodium amobarbital on the brain.


It is only with recent technological advances that we are able to examinemore closely the physiological effects of amobarbital injection and to esti-mate the extent of inhibition that occurs within the brain. Regional effectshave recently been investigated using EEG recorded from implanted elec-trodes (Gotman, Bouwer, & Jones-Gotman, 1992) and using single photonemission tomography (SPECT) (Johannson, Silfvenius, Bjornebrink, Hie-tala, Olivecrona, Saisa, & Christianson, 1993; Kurthen, Reichmann, Linke,Biersack, Reuter, Durwen, & Grunwald, 1990; McMackin, Dubeau, Jones-Gotman, Gotman, Lukban, Dean, Evans, Tampieri, & Lisbona, 1995;Wieser, Muller, Scheiss, Khan, Regard, Landis, Bjeljac, Buck, Vosavanis,Yasargil, & Yonekawa, 1995). These studies have shown changes in regionsoutside those directly irrigated by intracarotid injection, indicating a morewidespread effect than previously suspected. However, despite our growingunderstanding of these effects in the procedure as a whole, it is difficult topredict the individual effects on any specific patient. Physiological effectsmay vary according to differences in vascularization, metabolic function, orextent of damaged tissue. Thus, determining the validity of the IAP-M as apredictor of postoperative memory function remains a complex issue. Thisis further complicated by the use of many different testing methods and dif-ferent ways of judging memory performance at the various epilepsy centers(Rausch et al., 1993).

Considering these limitations, how well does the IAP-M manage to predictrisk of amnesia among those who fail the test, and with what confidence canone say that there is no risk among patients who pass? Among the 30 patientsincluded in our study, 13 had undergone the test specifically for memoryevaluation. Seven of those patients passed and underwent surgery withoutresulting memory deficits: if we adopt the nomenclature of signal detectiontheory, these would be called Correct Rejections. Five patients failed; how-ever, of these five, only two appeared to be at risk (Correct Hits), and threehad no memory deficit (False Alarms). In addition, there was one Miss bythe test: one patient passed the critical test and was amnesic following sur-gery. Thus we found four incorrect predictions (three False Alarms and oneMiss). If the IAP-M were not in use and all these patients simply underwenta ‘‘standard’’ hippocampal resection (2 cm), the seven patients previouslydeemed ‘‘Correct Rejections’’ would have had the same nonamnesic out-come because they were never at risk in the first place. More interestingly,the resulting cost and benefits of surgery would have been different. Thetwo cases that were considered ‘‘hits’’ according to the IAP-M results wouldprobably be amnesic in this scenario, as would the patient who was missedby the test. However, there would be no instances of patients incorrectlydenied treatment. The type of error that one believes to be most detrimentalwill influence whether the IAP-M will continue to be used to predict riskfor amnesia. If an outcome of amnesia is considered more detrimental thana false prediction of high risk (resulting in a more limited surgical interven-


tion or being refused surgery), and the goal is to avoid or reduce incidence ofamnesia, then continued use of the IAP-M in this way is indicated. However,improvements in sensitivity and specificity are needed.

An approach that is sometimes used to deal with ‘‘surprise failures’’ orsuspected false positive cases is to repeat the test. In some instances, externalfactors such as panic attacks or a patient missing eyeglasses warrant a repeti-tion of the IAP-M test (McGlone & MacDonald, 1989). However, repeattesting has also been carried out when a patient has failed the critical testwithout any obvious influencing external factors. In situations of a failedmemory test with an objective confounding factor (such as prolonged unre-sponsiveness), it is reasonable to repeat it with a lower drug dosage(McGlone & MacDonald, 1989; Novelly, 1987). However, if an apparentlyvalid test with an unexpected memory failure is repeated at a lower dose,and if the repeat test results in a changed outcome, it will be impossible tojudge which is the ‘‘correct’’ result. The effectiveness of this approach invalidating IAP-M outcomes remains unclear.

Finally, one can question whether it is fruitful to continue to interpretIAP-M results in a dichotomous, pass versus fail, fashion. The use of this testhas evolved, and its contribution toward confirming or questioning seizurelateralization by comparing the memory performance of the two hemispheres(Engel, Rausch, Lieb, Kuhl, & Crandall, 1981; Loring, Murro, Meador, Lee,Gratton, Nichols, Gallagher, King, & Smith, 1993; Perrine, Westerveld, Sass,Dovinsky, Dogali, Spencer, Luciano, & Nelson, 1995; Wyllie et al., 1991)has become at least as important a function as attempting to predict amnesia.

CONCLUSION

Many issues of reliability and validity of the IAP-M test continue to beunresolved; in addition, further knowledge about individual differences indrug reaction and brain plasticity are needed to better understand the limita-tions of the IAP-M. At present the majority of patients who undergo thisprocedure are not at risk for postoperative amnesia; hence the task becomesone of correctly predicting the small number of individuals truly at risk whileavoiding false positive predictions in others. The findings in this study sug-gest that the IAP memory results in the paradigm used do not necessarilypredict postoperative memory performance and that further refinements toimprove the sensitivity and specificity of IAP-M protocols must continue.

REFERENCES

Barr, W., Schaul, N., Decker, R., & Lantos, G. 1992. Post-operative amnesia after ‘‘passing’’memory testing during the intracarotid sodium amytal procedure. Epilepsia, 33, 138. [Ab-stract]

Bouwer, S., Jones-Gotman, M., & Gotman, J. 1993. Duration of sodium amytal effect: Behav-ioral and EEG measures. Epilepsia, 34, 61–68.


Cohen, J. 1988. Statistical power analysis for the behavioral sciences 2nd ed. Hillsdale, NJ:Erlbaum.

Colle, L. M., Holmes, L. J., & Pappius, H. M. 1986. Correlation between behavioral statusand cerebral glucose utilization in rats following freezing lesion. Brain Research, 397,27–36.

Dasheiff, R. M., Shelton, J., & Ryan, C. 1993. Memory performance during the amytal testin patients with non-temporal lobe epilepsy. Archives of Neurology, 50, 701–705.

Dodrill, C., & Ojemann, G. A. 1997. A comparison of three methods of memory assessmentwith the intracarotid amytal procedure. Brain and Cognition, 57, 210–223.

Dunn, L., & Dunn, L. M. 1981. Peabody picture vocabulary test—Revised. Circle Pines, MN:American Guidance Service.

Engel, J. J., Rausch, R., Lieb, J. P., Kuhl, D. E., & Crandall, P. H. 1981. Correlation of criteriaused for localizing epileptic foci in patients considered for surgical therapy of epilepsy.Annals of Neurology, 9, 215–224.

Girvin, J. P., McGlone, J., McLaughlin, R. S., & Blume, W. T. 1987. Validity of the sodiumamobarbital test for memory in selected patients. Epilepsia, 28, 636. [Abstract]

Gotman, J., Bouwer, S., & Jones-Gotman, M. 1992. Intracarotid EEG study of brain structuresaffected by internal carotid injection of amobarbital. Neurology, 42, 2136–2143.

Johannson, B., Silfvenius, H., Bjornebrink, J., Hietala, S., Olivecrona, M., Saisa, J., & Chris-tianson, S. 1993. A SPECT-[99mTc] HMPAO study in connection with intracarotid amy-tal testing. Epilepsia, 34(Suppl 2), 33. [Abstract].

Jones-Gotman, M. 1987. Commentary: Psychological evaluation—Testing hippocampal func-tion. In J. J. Engel (Ed.), Surgical treatment of the epilepsies. New York: Raven Press.Pp. 203–211.

Jones-Gotman, M. 1991. Presurgical neuropsychological evaluation for localization and later-alization of seizure focus. In H. Luders (Ed.), Epilepsy surgery. New York: Raven Press.Pp. 469–475.

Jones-Gotman, M. 1992. Neuropsychological techniques in the identification of epileptic foci.In W. H. Theodore (Ed.), Surgical treatment of the epilepsies. Amsterdam: Elsevier Sci-ence. Pp. 87–94.

Jones-Gotman, M., Barr, W. B., Dodrill, C. B., Gotman, J., Meador, K. J., Rausch, R., Sass,J. K., Sharbrough, F. W., Silfvenius, H., & Wieser, H. 1993. Controversies concerningthe use of intraarterial amobarbital procedures. In J. J. Engel (Ed.), Surgical treatmentof the epilepsies 2nd ed. New York: Raven Press. Pp. 445–449.

Jones-Gotman, M., Bouwer, S., & Gotman, J. 1994. EEG slow waves and memory perfor-mance during the intracarotid amobarbital test. Epilepsia, 35, 61–69.

Kurthen, M., Reichmann, K., Linke, D. B., Biersack, H. J., Reuter, B. M., Durwen, H. F., &Grunwald, F. 1990. Crossed cerebellar diaschisis in intracarotid sodium amytal proce-dures: A SPECT study. Acta Neurologica Scandinavia, 81, 416–422.

Langfitt, T. W., Obrist, W. D., Alavi, A., Grossman, R. I., Zimmerman, R., Jaggi, J., Uzzell,B., Reivich, M., & Dreux, R. P. 1986. Computerized tomography, magnetic resonanceimaging, and positron emission tomography in the study of brain trauma. Journal of Neu-rosurgery, 64, 760–767.

Loring, D. W., Hermann, B. P., Meador, K. J., Lee, G. P., Gallagher, B. B., King, D. W.,Murro, A. M., Smith, J. R., & Wyler, A. R. 1994a. Amnesia after unilateral temporallobectomy: A case report. Epilepsia, 35, 757–763.

Loring, D., Meador, K., Lee, G., & King, D. 1992. Amobarbital effect and lateralized brainfunction: The Wada test. New York: Springer-Verlag.

Loring, D. W., Meador, K. J., Lee, G. P., King, D. W., Gallagher, B. B., Murro, A. M., &Smith, J. R. 1994b. Stimulus timing effects on Wada memory testing. Archives of Neurol-ogy, 51, 806–810.

Loring, D. W., Lee, G. P., Meador, K. J., Flanigin, H. F., Smith, J. R., Figueroa, R. E., &Martin, R. C. 1990. The intracarotid amobarbital procedure as a predictor of memoryfailure following unilateral temporal lobectomy. Neurology, 40, 605–610.


Loring, D. W., Murro, A. M., Meador, K. J., Lee, G. P., Gratton, C. A., Nichols, M. E.,Gallagher, B. B., King, D. W., & Smith, J. R. 1993. Wada memory testing and hippocam-pal volume measurements in the evaluation for temporal lobectomy. Neurology, 43, 1789–1793.

McGlone, J., & MacDonald, B. 1989. Reliability of the sodium amobarbital test for memory.Journal of Epilepsy, 2, 31–39.

McMackin, D., Dubeau, F., Jones-Gotman, M., Gotman, J., Lukban, A., Dean, G., Evans, A.,Tampieri, D., & Lisbona, R. 1995. Assessment of the functional effect of the intracarotidsodium amobarbital procedure using co-registered MRI/HMPAO-SPECT and SEEG.Brain and Cognition, 57, 50–70.

Milner, B. 1958. Psychological defects produced by temporal lobe excision. Research Publica-tions–Association for Research in Nervous and Mental Diseases, 36, 244–257.

Milner, B., & Penfield, W. 1955. The effect of hippocampal lesions on recent memory. Trans-actions of the American Neurological Association, 80, 42–48.

Novelly, R. A. 1987. Relationship of intracarotid amytal procedure to clinical and neurosurgi-cal variables in epilepsy surgery. Journal of Clinical and Experimental Neuropsychology,9, 33. [Abstract]

Paivio, A., Yuille, J. C., & Madigan, S. 1968. Concreteness, imagery, and meaningfulnessvalues for 925 nouns. Journal of Experimental Psychology, Monograph Supplement, 76,1–25.

Perrine, K., Westerveld, M., Sass, K. J., Devinsky, O., Dogali, M., Spencer, D. D., Luciano,D. J., & Nelson, P. K. 1995. Wada memory disparities predict postoperative seizure con-trol. Epilepsia, 36, 851–856.

Rausch, R., Babb, T. L., & Brown, W. J. 1985. A case of amnestic syndrome following selec-tive amygdalohippocampectomy. Journal of Clinical and Experimental Neuropsychology,7, 643. [Abstract]

Rausch, R., Silfvenius, H., Wieser, H., Dodrill, C. B., Meador, K. J., & Jones-Gotman, M.1993. Intraarterial amobarbital procedures. In J. J. Engel (Ed.), Surgical treatment of theepilepsies 2nd ed. New York: Raven Press. Pp. 341–357.

Wechsler, D. 1945. A standardized memory scale for clinical use. J Psychol, 19, 87–95.Wechsler, D. 1981. Wechsler Adult Intelligence Scale-Revised. New York: The Psychological

Corporation.Wieser, H.-G., Muller, S., Schiess, N., Khan, N., Regard, M., Landis, T., Bjeljac, M., Buck,

A., Valavanis, A., Yasargil, G., & Yonekawa, Y. 1995. The anterior and posterior selectivetemporal lobe amobarbital tests: Angiographic, clinical, electroencephalographic, PETand SPECT findings and memory performance. Brain and Cognition, 57, 71–97.

Wyllie, E., Naugle, R., Awad, I., Chelune, G., Luders, H., Dinner, D., Skibinski, C., & Ahl,J. 1991. Intracarotid amobarbital procedure. I. Prediction of decreased modality-specificmemory scores after temporal lobectomy. Epilepsia, 32, 857–864.

Documents

Sodium Amobarbital Memory Tests: What Do They Predict?