Memory Outcome after Selective Amygdalohippocampectomy in Patients with Temporal Lobe Epilepsy: One-year Follow-up


Address correspondence and reprint requests to Dr. U. Gleissner at University Hospital of Epileptology, Sigmund-Freud Str. 25, 53105 Bonn, Germany. E-mail:


Summary: Purpose: In a previous study we reported clinically significant memory declines 3 months after selective amygdalohippocampectomy (SAH) in 140 patients with mesial temporal lobe epilepsy, particularly if the resection was left-sided. We supposed that the observed postoperative impairments might have reflected acute effects of surgery. Therefore we evaluated in the present study whether a recovery can be found 1 year after surgery.

Methods: Verbal and nonverbal memory functions were assessed in 115 patients before and 3 and 12 months after unilateral SAH.

Results: No recovery of postoperative verbal memory declines was found in the left-SAH group. Clinically meaningful losses were still evident in 33 to 50% of patients. In right-SAH patients, a recovery of verbal memory was indicated, and effects of surgical complications were no longer evident. One year after surgery, the corresponding preoperative performance was the only significant predictor of a postoperative change in the left-SAH group.

Conclusions: Verbal memory decline observed 3 months after left SAH is persistent 1 year after surgery. Declines in verbal memory, which were observed in some right-SAH patients at the short-term follow-up, seem to be temporary.

Cognitive effects of selective amygdalohippocampectomy (SAH) are not unequivocal (1–7). Although several studies indicated a relatively better outcome than after standard anterior temporal lobectomy (2–4,7), it is not clear to what extent SAH also can lead to a significant memory decline, because some studies reported significant verbal memory decreases after left SAH (3–6), but others did not (1,7). Most of these studies referred to relatively small samples. Therefore we recently investigated the effects of lateralized SAH on memory and determinants of memory outcome in a very large sample of 140 patients (8). Our results indicated relatively broad and strong declines in verbal memory 3 months after left SAH, and we supposed that the observed impairments might reflect immediate effects of surgery that could recover in the later course. This was supported by the observation that memory decline was particularly reported in studies with short retest intervals of 1 to 4 months (3,5,6), whereas studies that found no declines had longer retest intervals of 1 to 4 years after surgery (1,7). We now evaluated for a subgroup of 115 patients whether a recovery is indicated 1 year after surgery and whether the observed relations between clinical and neuropsychological variables remain valid.


Characteristics of the 115 right-handed patients are given in Table 1. Preoperative diagnostics followed the Bonn Algorithm (9). The exact surgical SAH technique was reported in detail in our first study (8). Seizure outcome was slightly less favorable now than in the first study (65 vs. 76%). Verbal and visual memory functions were assessed with parallel test versions of the VLMT [Verbal Learning and Memory Test (10)] and the DCS-R [Diagnosticum für Cerebralschädigungen–Revision (11)]. Preoperative IQ was estimated by a vocabulary test (12). Psychomotor speed was assessed pre- and postoperatively by a letter-cancellation test (13). As in the first study, standardized scores (m, 100; SD, 10) were used for verbal and nonverbal learning capacity, verbal delayed recall, and verbal recognition. Decisions about the significance of individual postoperative changes were based on 90% confidence intervals derived from test–retest data of 85 nonsurgical patients, who had been tested twice within a mean retest interval of 11.6 months (SD, 6.2), by following the proposals of Hermann et al. (14). Twelve of those patients later underwent SAH and also were included in the surgical group.

Table 1. Demographic and clinical data
 Right SAH
(n = 63)
Left SAH
(n = 52)
  1. The cells contain the number of patients (for gender, pathology, and postoperative seizure status) or group means (with standard deviations in parentheses).

  2. SAH, selective amygdalohippocampectomy; yr, years; NS, not significant (p > 0.1) in ANOVA or χ2 test.

  3. aEstimated by a vocabulary test (10).

  4. bAssessed with a letter-cancellation test (11).

  5. cEngel outcome class I.

Males/females29/3425/27χ2= 0.8, NS
Age at surgery (yr) 31.5 (10.2) 31.4 (10.9)F= 0.005, NS
Age at onset of epilepsy (yr)10.9 (8.7)11.0 (9.9)F= 0.05, NS
Preoperative IQa  103 (16.8) 98.7 (11.2)F= 1.8, NS
Psychomotor speedb
 Preoperatively103.1 (13.4)100.9 (12.1)F= 0.8, NS
 3 mo after surgery109.1 (12.3)105.6 (14.0)F= 2.0, NS
 1 yr after surgery110.5 (13.7)109.0 (13.5)F= 0.3, NS
 Ammon's horn sclerosis5843 
 Other lesion 5 9 
Seizure-free patientsc
 3 mo after surgery   49 (78%)   38 (73%) χ2= 0.34, NS
 1 yr after surgery   43 (68%)   32 (62%) χ2= 0.57, NS


Pre- and postoperative memory performance is reported in Table 2. Separate repeated-measurement multivariate analyses of variance (MANOVAs) were performed for intervals (a) preoperative, 3 months; (b) preoperative, 12 months; and (c) 3 to 12 months with the four memory scores as dependent variables, test repetition as within-subjects factor, and side of surgery, gender, and seizure outcome (seizure free vs. not seizure free) as between-subjects factors. As in our previous study, MANOVAs 1 and 2 revealed significant interaction effects between test repetition and side of surgery, and post hoc analyses indicated left-resected patients to show significantly worse overall verbal memory performance and significant postoperative declines in all verbal memory parameters at both follow-ups (a): Hotellings T = 0.17; all F > 4.2 with p < 0.024; and (b): T = 0.28; all F > 6.9 with p < 0.001 for the overall effect and all verbal memory parameters). No significant effects were found in MANOVA c, confirming that 1 year after surgery, no recovery of the postoperative memory declines had occurred.

Table 2. Memory performance before and after SAH
 Right SAHLeft SAH
Preop.3 MP1 YPPreop.3 MP1 YP
  1. Cells provide the mean standardized scores with the standard deviations in parentheses. Standardized scores have a mean of 100 (SD, 10).

  2. SAH, selective amygdalohippocampectomy; MP, months postoperative; YP, year postoperative.

Verbal memory
 Learning91.2 (12.5)90.4 (12.1)91.1 (12.5)83.4 (10.4)74.8 (10.9)75.3 (11.5)
 Delayed recall91.1 (13.1)87.8 (13.6)91.4 (12.9)78.3 (10.3)68.1 (10.1)68.6 (11.0)
 Recognition91.8 (18.7)86.2 (18.4)89.6 (21.6)76.6 (22.9)49.8 (27.2)54.2 (27)  
Nonverbal memory
 Learning84.7 (15)  83.4 (16.1)84.6 (15.9)83.5 (14.7)84.3 (15.6)81.9 (13)  

Table 3 reports the frequency of significant individual postoperative changes. In the left-resected group, still ≤50% of the patients experienced a significant decline in verbal memory. This is comparable with the outcome 3 months after surgery. Table 3 indicates that slightly fewer right-resected patients have significant verbal memory declines 1 year after surgery (≤16%) than at the 3-month follow-up (≤27%). No significant differences were found between right-resected patients with or without complications (n = 7 and 56; all F values <3.1, with p > 0.08). Thus effects of surgical complications, which had been observed 3 months after surgery, were in remission 1 year after surgery. To identify predictors of memory outcome, separate multiple-regression analyses (forward-selection, inclusion criterion F > 3.83, with p ≤ 0.05) were computed for the left- and right-SAH groups, with the difference scores in memory as the dependent variables (preoperative minus postoperative score), and the corresponding preoperative memory score, age at the onset of epilepsy, age at surgery, gender, evaluation with intrahippocampal depth electrodes, Ammon's horn sclerosis (yes/no), and seizure outcome as the independent variables for both follow-ups. The proportions of explained variance were between 0.23 and 0.45 in the left-SAH group and between 0.1 and 0.28 in the right-SAH group. Independent of the side of surgery and the memory parameter, a higher preoperative performance was associated with a postoperative decline at both follow-ups (left SAH: all t < –3.5, with p < 0.001; right SAH: all t < –2.6, with p < 0.012). For the left-SAH group, further predictors of a postoperative decline were a later onset of epilepsy (for learning, t=–2.5; p = 0.014), and an older age at surgery (for delayed recall and recognition: all t < –2.5, with p < 0.016) at the short-term follow-up. One year after surgery, only the corresponding preoperative performance remained as a significant predictor for all three parameters of verbal memory in the left-SAH group. For the right-SAH group, an older age at surgery was a predictor for delayed recall at both follow-ups and for recognition 3 months after surgery (all t < –2.1, with p < 0.037).

Table 3. Individual changes in memory 3 and 12 months after selective amygdalohippocampectomy
 3 mo postoperative
1 yr postoperative
Side of surgeryLeftRight LeftRight 
  1. Cells provide the percentage of patients in the left- or right-SAH group showing a loss or a gain in postoperative performance compared with the preoperative score.

  2. ap < 0.01, bp < 0.05, NS not significant in a χ2 test (including patients with unchanged performance).

Verbal memory
Delayed recall50/227/8a50/416/11b
Nonverbal memory26/1432/14NS29/1227/16NS


The technique of SAH was originally developed in epilepsy surgery to spare unaffected brain tissue from surgery and thus to minimize the negative cognitive consequences of temporal lobe surgery. Our data clearly show that this does not mean no or only very mild memory decline after SAH. Our previous study (8) showed clinically significant verbal memory declines after SAH in patients with mesial TLE, particularly if the resection is on the left side. We had expected that a longer follow-up interval might reveal some recovery if the declines reflected acute effects of surgery. However, neither on the level of group means nor on the level of individual data was a recovery evident for the left-resected group 1 year after surgery. The analyses of the 3-months' follow-up, which were repeated to control for the possibility that the present findings reflect only an effect of the smaller sample, confirmed the results of our previous study (8). For the right-SAH group, the 3-months' follow-up had shown a verbal memory decline, particularly in patients with surgical complications. Our data indicate a recovery in this respect. The short-term data might have reflected a temporary functional disturbance, because no significant differences were found between right-resected patients with and without complications 1 year after surgery. At the short-term follow-up, a higher preoperative performance, a later onset of epilepsy, and an older chronologic age at surgery were predictors for a postoperative verbal memory decline similar to the predictors reported for patients after anterior temporal lobectomy (15,16). One year after surgery, the declines in left-SAH patients were associated solely with a higher preoperative performance in the respective score, revealing this as the strongest predictor. A higher preoperative performance indicates a relatively intact preoperative functioning of the cerebral tissue and hence implies a higher risk for cognitive declines in case of a resection. The strength of the relation probably indicates also effects of regression toward the mean. In the present study, all aspects of verbal memory were affected by the left SAH. The current data thus do not support the findings of a previous study (4), in which we had observed in a smaller SAH group (n = 15) declines only in delayed recall and recognition. The assumption that temporomesial structures are specifically involved in long-term consolidation and retrieval processes, whereas learning or data acquisition is mediated by neocortical temporal structures is therefore not supported by the present results, and this must be more closely investigated in the future.