- Top of page
Summary: Purpose: The aim of this study was to compare the utility of baseline neuropsychological measures and scores from the intracarotid amobarbital procedure (IAP) in the prediction of postoperative memory decline in temporal lobe epilepsy surgery patients.
Methods: Logistic regression analyses were used to determine the relation between demographic variables, baseline neuropsychological scores, and scores from the IAP (using mixed verbal and nonverbal stimuli) and postoperative deterioration in verbal learning and verbal recall in 91 patients (48 right, RTL; 43 left, LTL) who had undergone a standard anterior temporal lobe resection for the relief of medically intractable epilepsy and who had been followed up 1 year postoperatively.
Results: In the RTL group, the IAP scores were not significant predictors of a postoperative decline in verbal learning or recall. In the LTL group, postoperative decline in verbal learning was associated with good preoperative baseline scores, an older age at the time of surgery, and an unexpected asymmetry on the IAP. Baseline neuropsychological scores and scores from the IAP were associated with a significant postoperative decline in verbal recall in the LTL group.
Conclusions: Scores from the IAP using mixed stimuli were not helpful in the prediction of postoperative verbal memory decline in RTL patients. The significance of IAP scores in predicting verbal memory deficits in LTL patients may be task specific.
An ongoing debate exists regarding the future of the intracarotid amobarbital procedure (IAP) (van Emde, 1999; Medina et al., 2004; Duncan et al., 2005; Helmstaedter, 2005; Kirsch et al., 2005; Kloppel and Buchel, 2005). Traditionally, the IAP played an important primary role in the assessment of amnesic risk and is the gold standard against which newer language-lateralizing paradigms are measured (Baxendale, 2002). Data from the IAP have also been used to confirm the lateralization of the seizure focus (Lancman et al., 1998; Lee et al., 2002; Cohen-Gadol et al., 2004). More recently, IAP scores have been used in the statistical modelling of postoperative memory decline (Jokeit et al., 1997; Stroup et al., 2003; Kirsch et al., 2005).
However, these primary and secondary roles have been challenged by new, noninvasive technologies. Patients who “fail” an IAP may not always be at risk of a postoperative amnesic syndrome (Lacruz et al., 2004). Loring et al. (1990) reported 10 such cases who proceeded to surgery, none of whom became amnesic postoperatively. Structural and functional MRI scans can now ensure the integrity of the contralateral structures preoperatively and provide valuable data regarding amnesic risk (Lencz et al., 1992; Baxendale et al., 1998; Kapur and Prevett, 2003; Woermann et al., 2003; Richardson et al., 2004; Janszky et al., 2005; Koepp and Woermann, 2005). In addition, functional imaging paradigms are also proving useful in both the lateralization and localization of specific language functions (Matthews et al., 2003; Sullivan and Detre, 2005). As these techniques begin to supersede the IAP in its traditional roles within the presurgical evaluation, clinical and research emphases have shifted toward the importance of the IAP in the prediction of postoperative memory change (Stroup et al., 2003; Kirsch et al., 2005).
A number of studies have investigated the relation between preoperative IAP scores and postoperative memory function (Loring et al., 1990; Kneebone et al., 1995; Bell et al., 2000; Chiaravalloti and Glosser, 2001; Stroup et al., 2003; Kirsch et al., 2005; Lee et al., 2005). Collectively these studies suggest that the IAP provides valuable prognostic data for memory decline in both adults and children, particularly with respect to verbal memory decline after a left temporal lobe resection (Kneebone et al., 1995; Bell et al., 2000; Chiaravalloti and Glosser, 2001). Stroup et al. (2003) found that IAP scores, side of surgery, preoperative memory function, and underlying pathology were all significantly and independently associated with memory outcome after a temporal lobe resection for epilepsy. Further, the IAP scores were more potent predictors than baseline neuropsychological measures.
As an invasive and expensive procedure, the IAP must be shown to contribute uniquely valuable data to the presurgical evaluation of epilepsy patients if it is to continue to play an important role in epilepsy surgery. Although IAP scores may predict postoperative memory decline, the “added value” of IAP scores over and above other readily available noninvasive indices of risk has yet to be established.
The aim of this study was to compare the contributions of baseline neuropsychological measures with mixed-stimuli IAP scores in the prediction of verbal memory decline in patients who underwent an anterior temporal lobe resection. The focus of this study is solely on the role of the IAP in the prediction of postoperative memory decline, rather than language lateralization or the determination of amnesic risk.
- Top of page
Our results suggest that an IAP using mixed stimuli provides different data for RTL and LTL groups. The confounding effects of language laterality mean that LTL patients have significantly higher ipsilateral hemisphere scores and lower contralateral hemisphere scores than the RTL patients. As a result, LTL patients are far more likely to have an unexpected asymmetry score than are RTL candidates. These findings suggest that it is inappropriate to equate the mixed-stimuli IAP ipsilateral/contralateral hemisphere memory scores of RTL and LTL groups. These findings support Jokeit's (2004) contention than the IAP data from RTL and LTL groups must be analyzed separately in regression analyses using IAP data.
The mixed-stimuli IAP scores were not significant predictors of a postoperative decline in verbal learning or recall in our RTL group. Consistent with previous findings, we found that higher baseline neuropsychological scores and an older age at seizure onset were significant predictors of decline in verbal recall and learning in this group (Chelune et al., 1991; Hermann et al., 1995; Jokeit et al., 1997; Davies et al., 1998). However, the mixed-stimuli IAP scores were significant predictors of postoperative decline in both verbal learning and recall in the LTL group. An unexpected IAP asymmetry score, together with an older age at the time of surgery and good preoperative functioning were significant predictors of postoperative decline in verbal learning. All of the IAP indices were significant predictors of a postoperative decline in verbal recall, together with higher baseline scores. Although the sensitivity of the model increased when the IAP asymmetry score was excluded in the prediction of a postoperative decline in verbal learning, this was at the expense of specificity. In our data set, the IAP asymmetry score increased the specificity of the model in predicting significant postoperative decline in verbal learning in LTL patients, over models including simple demographic data and baseline neuropsychological scores. Similarly, the addition of our mixed-stimuli IAP indices increased both the sensitivity and specificity of the model used to predict postoperative decline in verbal recall, over the model based simply on baseline neuropsychological scores in LTL patients.
It is possible that the significance of the IAP variables in predicting postoperative verbal memory decline may be due to an artifact. The LTL patients are more likely to decline on verbal memory measures postoperatively and are also more likely to have unexpected IAP asymmetry scores because of the confounds of language dominance. Thus the verbal stimuli used in our IAP may be significant factors in the results. Further work is under way to investigate the relation between IAP scores derived from a different IAP protocol using real objects as memory stimuli and postoperative decline in verbal memory.
This study has a number of limitations that constrain the conclusions that can be drawn from the data. Our sample included only patients who were left hemisphere dominant for language. Intracarotid amobarbital procedures vary greatly in methods and materials. This protocol uses a mixture of both verbal and nonverbal stimuli, and we always injected the side of the suspected seizure onset first, introducing the possibility of a small, but systematic, “second injection” effect on memory performance. The measures of verbal memory include a measure of learning and immediate recall but not delayed recall. These limitations should be borne in mind when considering the clinical implications of our findings outlined later.
After the introduction of fMRI paradigms to the clinical investigation of epilepsy surgery patients, the role of the IAP has increazingly come into the spotlight. The traditional roles of the IAP in lateralizing language and in screening for amnesic risk are being superseded in the presurgical evaluation of epilepsy patients by structural and functional MRI scans. Although no doubt exists that the IAP can provide lateralizing and prognostic data, the added value of the data must be carefully assessed to justify using an invasive and expensive procedure. Our findings suggest that the prognostic value of our mixed-stimuli IAP data in predicting postoperative memory change is very limited in left hemisphere language-dominant RTL patients. As such, it is difficult to justify the application of such an IAP in these patients solely to gain prognostic data to predict postoperative deterioration in verbal learning and recall.
Statistically, our mixed-stimuli IAP indices provided superior data to baseline neuropsychological scores in the prediction of postoperative decline in our LTL patients. However, in practical terms, the models including IAP indices enabled the identification in just one additional patient who was at risk of a postoperative decline in verbal recall, and just four LTL patients were correctly reclassified at low risk in our sample, of a denominator of 43, 21 of whom had a decline in verbal memory.
In our view, this refinement of prediction does not justify the use of the mixed-stimuli IAP in all LTL patients solely to gain prognostic data to predict postoperative deterioration in verbal learning and recall. This raises ethical and economic issues about the “added value” of mixed-stimuli IAP data in the prediction of postoperative change in LTL patients. These can be addressed partly by examining the use to which postoperative prediction are put. It is important that every patient be given as much information as possible regarding the risks of a temporal lobe resection to their memory function, to enable them to make an informed decision before surgery. However, whether a statistical definition of deterioration is clinically meaningful for a given patient will depend on many factors, including the nature and extent of his environmental demands and, to some extent, his or her psychosocial resources and support systems (Baxendale et al., 2006). It is extremely difficult for patients to weigh precise statistical likelihoods in the surgical decision-making process. The difference between a 60% chance and a 65% chance of an event occurring is unlikely to alter the decision-making process. Thus the added value of these IAP scores in improving the prediction of postoperative memory decline may have negligible clinical relevance for the patient. In addition, because generally a poor correlation is found between objective measures of memory function and patients' complaints (Baxendale and Thompson, 2005), even if our IAP data did improve the accuracy of a predictive model, the clinical relevance of the information to the patient might well be outweighed by the invasive nature of the procedure.
In addition, now other noninvasive indices of structure and function can be used to predict postoperative memory decline in both RTL and LTL groups (Hermann et al., 1995; Jokeit et al., 1997; Davies et al., 1998; Baxendale et al., 2006). Because of the historical nature of the sample, it was not possible to include quantitative or qualitative indices from structural or functional MRI studies in this study that have been shown to be significant predictors of postoperative memory change.
The IAP has often been cited as the “gold standard” in the assessment of language lateralization and in the screening for amnesic risk of prospective temporal lobe surgery patients. However, it has not been established as the gold standard for the prediction of postoperative memory decline. A number of other demographic and clinical factors have also been shown to be associated with postoperative memory decline, the majority of which can be quantified by using noninvasive methods. Whereas it makes sense to use IAP measures where they are already available to predict postoperative memory decline, our results suggest that it is not appropriate to conduct an invasive mixed-stimuli IAP solely to gain prognostic data regarding postoperative memory decline.