An IAP to assess cerebral dominance is performed when there is reason to suspect an atypical cerebral organization for language, because this might interfere with the planned surgical intervention. Left-handed individuals, those with a strong family history of left-handedness, and people with evidence of early damage in or near speech areas of the left hemisphere fall into this category. Patients whose anatomical and functional (cognitive) lateralization is discordant are also candidates for an IAP because in such cases the mismatch can reflect right-hemisphere speech dominance. Dichotic-listening tasks are sometimes used as a screening test for atypical speech representation. In those tasks, competing verbal stimuli are delivered simultaneously to the two ears, with the expectation that the stimuli will be perceived best by the ear opposite the speaking hemisphere.^113,129 If there is little or no difference between the ears, or if best performance is observed from the left ear (right hemisphere) in the presence of normal hearing, then an IAP to determine cerebral speech dominance may be performed.

Patients are selected to undergo an IAP for memory evaluation based either on the presence of significant deficits on verbal and nonverbal memory tests uncovered during a basic neuropsychological assessment or on electroencephalogram (EEG), anatomic magnetic resonance imaging (MRI), or other evidence suggesting bitemporal damage. An IAP might also be performed in cases of mismatch between EEG and anatomic MRI findings such that an EEG focus is observed in one temporal lobe and a significantly small hippocampus is found on the opposite side.

Before injection, basic speech and memory tests are performed to establish a baseline. On injection, more speech and memory tests are administered while only one hemisphere is functional. The same speech tests as used in baseline testing are performed after injection, but new memory material is shown. Memory for the new material is tested later, after the drug is no longer active and the injected hemisphere has returned to baseline functioning.

IAP speech tests sample different aspects of language, including serial or automatic speech (counting, reciting days of the week), naming, repetition, spelling, reading, and auditory comprehension (see, e.g., Fedio et al.²¹ and Ravdin et al.¹⁰²). Cerebral dominance should not be assessed with a single task because dissociations among types of speech functions are sometimes observed.^11,103,110 Tasks should be rotated throughout the crucial period of hemianesthesia so that all can be sampled.

Superimposed on the basic methodology of memory testing during the IAP are the many variations used by different institutions. The number of items introduced after injection to test the formation of new memories varies from only a few, such as five,⁴⁶ to continuous presentation for as long as the hemianesthesia is present.^67,100,101 The timing of memory item presentation can also vary.⁷² Some centers^15,16 test memory using a mixture of materials that includes words, pictures, phrases, and commands, whereas others use all words or all pictures or all real objects. Real objects have the advantage of being perceived more easily by either hemisphere.⁵² In most but not all centers the final interpretation of memory results is based on the patient’s ability to recognize later the new material that had been shown during hemianesthesia, but free recall is taken into account in some institutions.¹⁰⁴ Despite this lack of uniformity in details of IAP memory-test methods, the essence of the procedure does not differ, and results from one institution can usually be interpreted by another.

Although most centers carry out a “standard” intracarotid amobarbital test, some perform special procedures for selective temporary inactivation of temporal and extratemporal structures. The development of selective temporal lobe (TL) amobarbital tests was motivated by skepticism about the ability of “standard” IAP to predict postoperative memory performance in patients who would receive a TL resection,⁶⁹ especially a restricted resection such as selective amygdalohippocampectomy.¹²⁵ The main reasons for the skepticism were that the hippocampal formation may not be sufficiently anesthetized with the IAP⁴² (but see refs. 26, 79, and 80) and that inactivation of large parts of one hemisphere does not allow examination of the functional role of the specific structures of interest. Furthermore, the sudden anesthesia of one hemisphere can produce an initial period of confusion, inattention, and disorientation, as well as aphasia if the speech-dominant hemisphere is involved. These side effects interfere with memory testing.

For anesthesia of medial TL structures, at least three selective procedures were developed—two anterior procedures and a posterior one. One anterior procedure^{120,121,122,123,124} consists of a temporary balloon occlusion of the internal carotid artery distal to the origin of the anterior choroidal artery (acha), with subsequent injection of amobarbital into the territories of the acha, the posterior communicating artery, and the opthalmic artery. The second anterior procedure is a selective catheterization of, and injection of amobarbital into, the acha. Whether the selective inactivation of medial temporal-lobe structures by injection into the anterior choroidal artery has the advantages that are claimed for it should be verified by further studies.

The posterior TL amobarbital test^{42,43,94,120,123} consists of selective catheterization and subsequent injection of amobarbital into the P2 segment of the posterior cerebral artery.

In the hands of an expert interventional neuroradiologist with appropriate sophisticated catheter techniques, extratemporal brain regions can be selectively approached and temporarily inactivated by amobarbital. Because of the inherent risks of such procedures, they are rarely performed. The need for extratemporal amobarbital testing arises if the function of the brain region under consideration involves potential essential cortex, such as classic speech or motor areas, or if the alternative approach of intraoperative functional mapping with the patient awake is not possible. Between October 1986 and December 2005 the Zürich group carried out selective amobarbital tests in 106 patients, only 6 of whom received an extratemporal procedure. One of these was a selective inactivation of Broca’s region in a patient slated for resection of a tumor invading the left frontal operculum, 1 was a regional left frontal inactivation in a patient with a left anterior frontal epileptogenic lesion, 1 was a sequential Broca and Wernicke inactivation, 1 was a left posterior insula and Wernicke inactivation, and the remaining 2 underwent inactivation of selected branches of the middle cerebral artery.

Selective procedures are carried out far less frequently than the “standard” IAP; in a survey of epilepsy centers it was determined (among 68 respondents) that only 4% of all amobarbital procedures performed annually were selective.¹⁰¹ Although no newer survey has been conducted, a PubMed search yielded only 13 articles about selective amobarbital procedures in the period from 1996 to 2006. This represents 5% of the approximately 250 articles that were published about the IAP in that same time period. Furthermore, judging by those publications, most selective procedures are being performed in Europe. The largest series of selective tests has been performed by the Zürich group; for details of their results, including angiographic, clinical, electroencephalographic, positron emission tomography (PET), and single photon emission computed tomography (SPECT) findings and memory performance, see Wieser et al.¹²³

Owing to repeated shortages of amobarbital, within but especially outside of the United States, some centers have switched to other drugs. The most frequent alternative has been methohexital,^3,13,41 but propofol has also been used.^6,108,115 Clinicians who use methohexital are pleased with it as an alternative to amobarbital. It has an important limitation, however, in that it is so short-acting that it usually has to be reinjected within a test, leading to waxing and waning of the effect at least twice within critical testing time. This lack of control over the level of anesthesia is a problem for presentation of new memory items and consequently for interpretation of memory results. Propofol, which is used in the same way as amobarbital with a single injection and subsequent waning of the effect, is also problematic because it must be injected in a lipid carrier. It is a newer alternative, and has been used in only a few patients except for one series of 12¹¹⁵; the articles about propofol have also reported overall satisfaction with it. A third alternative, etomidate, has been used successfully in >42 patients (84 tests), and in this case the alternative also includes an important change in the procedure: In the etomidate Speech and Memory test (eSAM), the level of anesthesia is maintained with an infusion following the initial bolus injection.⁵² This allows all critical speech and memory tests to be administered
during full hemianesthesia, eliminating the time pressure during test presentation, and consequently eliminating the incidence of uninterpretable memory tests that occur in the IAP owing to presentation of items when the drug is no longer active. The limitation for eSAM is that in about 45% of the first 26 tests performed, a side effect was observed that consisted of a shiver-like tremor. This was very mild or almost imperceptible in more than half of those incidents and was seldom noticed or remembered by the patients.

There were no serious adverse effects reported for any of these three alternatives to amobarbital.

Another agent is also used for a different but related purpose: Lidocaine has been used together with amobarbital for patients with cerebral arteriovenous malformations (AVM) who undergo superselective intraarterial injections and behavioral testing (language, memory, motor) before receiving embolization to treat the AVM.^23,98 According to Fitzsimmons et al.,²³ the reason for including lidocaine in those procedures is that amobarbital selectively inhibits gray matter structures, whereas lidocaine inhibits white matter tracts as well. They found that using both agents provided a fuller picture in predicting possible deficits following embolization of AVMs.

One reason for carrying out an IAP is to screen for a risk of potential severe memory loss after resection from a temporal lobe, but it is very difficult to prove whether this test actually does predict amnesia. If patients who have failed amobarbital memory tests are denied surgery, or if they do undergo surgery but the hippocampus is spared, one cannot know whether they would have become amnesic if an extensive removal of hippocampus had been performed.

Some evidence does exist, however. There are two known cases of postoperative amnestic syndrome in patients who had been designated by IAP to be a risk and who still underwent resection from a temporal lobe.^68,99 Furthermore, in a survey of 71 epilepsy centers concerning IAP issues, an additional four such amnesic cases were reported,¹⁰¹ and a later publication reported an additional patient who became globally amnesic for an extended period of time but who gradually recovered to a state of deficient memory but not amnesia.²⁸ Many other patients who have undergone temporal lobectomy after “failing” the memory test have not developed amnesia.^17,19,71

Evidence that the IAP memory test does address hippocampal function can also be found. Sass et al.¹⁰⁶ counted hippocampal cells in resected temporal-lobe tissue and showed that patients with severe cell loss had shown deficient memory during the relevant amobarbital test when injection had been made opposite that damaged temporal lobe. Similarly, studies examining amobarbital memory performance as a function of hippocampal atrophy, as measured on magnetic resonance imaging (MRI), show significant memory failures after injection opposite the hemisphere harboring hippocampal atrophy and good memory after injection opposite a normal hippocampus.^51,75

The foregoing results suggest that we are indeed testing hippocampal function with amobarbital memory tests. We infer further that doing so indeed allows us to predict postoperative memory loss, although this remains difficult or impossible to prove. Studies of later memory performance in patients who have passed versus those who have failed IAP memory tests suggest that the IAP is predictive.^19,48 Those studies show postoperative losses and long-term memory deficits in patients who had been determined to be at risk for serious global memory impairments. However, the patients in those studies were not amnesic. As we have just seen, postsurgical amnesia does occur, but rarely.