Abstract
We recently demonstrated statistically significant correlations between presurgical memory impairment and hippocampal volumetric cell densities (in CA3 and the hilar area only) for patients with idiopathic left temporal lobe epilepsy who exhibited marked hippocampal neuron loss.In the present research we determine whether the same relationship exists for patients with structural lesions, in whom hippocampal neuron loss was minimal. Rank-order correlations of verbal memory test results (ie, Long Term Retrieval score of the verbal Selective Reminding Test, Percent Retention index of the Logical Memory subtest of the Wechsler Memory Scale) and hippocampal volumetric cell densities (subfields CA1, CA2, CA3, the hilar area, and the granule layer of area dentata) were computed for 22 patients with structural lesions and medically refractory epilepsy of temporal lobe onset (11 left, 11 right). There were statistically significant correlations between Long Term Retrieval and the volumetric cell density of CA1 (r equals 0.62, p less than 0.05) and between percent retention and the volumetric cell density of CA2 (r equals 0.60, p less than 0.05) for patients with left hemisphere lesions. No other correlations were found for patients with left or right temporal lobe lesions.
NEUROLOGY 1995;45: 2154-2158
Sass et al [1] recently reported that anteromedial temporal lobectomy for the treatment of seizures of hippocampal onset can cause or exacerbate a verbal memory impairment when performed for the speech-dominant hemisphere of patients who manifest no more than moderate hippocampal neuron loss. Prior research indicates that the hippocampus is often spared severe neuron loss when seizures originate from a temporal lobe tumor or other structural lesion (eg, vascular malformations, cortical dysplasia) as opposed to when they originate from within the hippocampus. [2-6] As a result, many patients with temporal lobe lesions within the speech-dominant hemisphere may be at risk of increased verbal memory impairment if the hippocampus is surgically excised.
Sass et al [7-9] demonstrated a correspondence between the degree of hippocampal pathology and certain impairments of verbal learning and retention that allowed the estimation of hippocampal integrity prior to surgery. Preoperative verbal memory test results could be employed to identify patients with temporal lobe structural lesions who are at risk for increased memory impairment after surgery, if the correspondence between the degree of hippocampal pathology and memory impairment was the same as that demonstrated for patients without structural lesions who suffer temporal lobe epilepsy. This correspondence may not exist, however. The temporal lobe mass may impair verbal learning or retention independently of the dysfunction caused by hippocampal pathology. The impact of the temporal lobe structural lesion upon verbal memory may vary with its location or type. Such variability would attenuate the statistical relationship between hippocampal pathology and verbal memory impairment. The purpose of the present research was to examine the relationship of verbal memory impairment and hippocampal pathology for patients with structural lesions who suffer temporal lobe epilepsy.
Methods.
Subjects.
Twenty-two patients with structural lesions and medically refractory epilepsy of temporal lobe onset (11 left, 11 right) were examined neuropsychologically prior to surgery. Preoperative evaluation also included CT, MRI, angiography, and the intracarotid amobarbital procedure (IAP). The seizure focus was established via scalp EEG with simultaneous audiovisual recording during at least three clinical seizures, when noninvasive diagnostic tests converged. For five patients, invasive EEG evaluation (depth electrodes, subdural electrodes, or both) was necessary to determine the precise area of seizure onset.
The sample was confined to English-speaking patients, 16 years or older, with a Wechsler Adult Intelligence Scale-Revised (WAIS-R) Full Scale IQ (FSIQ) of 70 or more. All patients were shown to be left-hemisphere speech dominant via the IAP. Inclusion in this study required the presence of a structural lesion. The sample contained patients with a variety of structural lesions. Fifteen patients had tumors (six low-grade gliomas, one oligodendroglioma, two gangliogliomas, five anaplastic astrocytomas, one cellular astrocytoma). Two patients had vascular malformations, and one had an arachnoid cyst. One patient had cortical dysplasia, and another had a hamartoma. The remaining two patients, one in each group, had abnormal sulci that probably represented dysgenesis.
Descriptive information for the two patient groups (left- versus right-sided lesions) is summarized in Table 1. The two groups did not differ significantly with regard to history of febrile convulsions, FSIQ, age at first seizure, age at seizure intractability, or age at surgery, although no attempt was made to equate the groups with respect to demographic variables.
Table 1. Demographic information
Examination techniques.
Examiners who were blind with respect to the location of the seizure focus administered the verbal Selective Reminding Test (vSRT) [10] and the Logical Memory subtest of Russell's adaptation of the Wechsler Memory Scale (WMS) [11] during audiovisual monitoring of seizures. All patients were treated with anticonvulsant medication, and serum levels were maintained within the therapeutic range during evaluation.
The vSRT was employed to evaluate verbal learning because Sass et al [7] demonstrated that this measure is sensitive to the degree of hippocampal pathology found for patients without structural lesions. The vSRT is a list-learning task. Subjects are read 12 words, which they attempt to repeat immediately following the examiner's recitation. This test is unique in that all items are presented on trial 1, but stimulus exposure is limited thereafter to items that the patient failed to recall on the previous trial. Stimulus exposure continues until the patient can recite the entire list on two consecutive trials or a total of 12 trials have been performed. Standard scoring yields measures of total recall, long-term storage, long-term memory retrieval, consistent long-term memory retrieval, and short-term memory retrieval. Perrine et al [12] demonstrated that these five measures are highly intercorrelated. We employed only Long Term Retrieval in this study, as we did in our prior research, because Long Term Retrieval had the highest correlation with the other vSRT measures. This reduced the number of statistical relationships that were evaluated, thereby maintaining an acceptable probability of Type I error.
The WMS Logical Memory subtest was also employed, because prior research indicated that scores derived from this test were differentially sensitive to cortical versus subcortical left temporal lobe dysfunction. [8,9] Immediate Recall and Delayed Recall correlated with the results of measures of visual confrontation naming and verbal analytic reasoning, but not with hippocampal neuron density. In contrast, Percent Retention was correlated with hippocampal neuron density, but not with visual confrontation naming or verbal analytic reasoning. On this basis, the measures of Immediate Recall and Delayed Recall were considered to be indices of cortical pathology, whereas Percent Retention was considered to be an index of hippocampal integrity.
Surgical procedure and analysis.
All patients underwent a cortical resection that was tailored to the location of the lesion. Total hippocampectomy was performed for 14 patients and partial hippocampectomy was performed for the remaining eight patients. Total hippocampectomy was most often performed for lesions that were in close proximity to the hippocampus. Fried et al [2] demonstrated that hippocampal pathology was common when the structural lesion was mesial in location. The hippocampus was exposed by dissection of the occipitotemporal fasciculus. Medial structures, including the amygdala, parahippocampus, and hippocampal formation, were removed en bloc using the operating microscope. The hippocampus was cut posteriorly at its curve toward the superior colliculus when total hippocampectomy was performed. The cortical incisions spared the superior temporal gyrus in 14 of the cases; for all others the inferior, middle, and superior temporal gyri were incised. The anterior-posterior margins ranged from 3 to 6.5 cm and were all at a depth of 2.3 to 3 cm. Intraoperative cortical stimulation or electrocorticography was performed in five of the cases.
A coronal section, 2 mm in thickness, was obtained from the midsection of the hippocampus. This was placed in 10% buffered formalin in the operating room and then transferred to the laboratory for further fixation, dehydration, and paraffin embedding. Paraffin sections, 6 micro meter in thickness, were stained with hematoxylin-eosin.
Separate cell counts were obtained for CA1, CA2, CA3, the hilar area, and the granule layer of area dentata. Neuronal nuclei were counted in areas measuring 200 micro meter by 400 micro meter for the pyramidal cell fields and 50 micro meter by 100 micro meter for the granule cell area. The neuronal counts were adjusted using Abercrombie's formula (8 micro meter was used as the nuclear diameter of pyramidal neurons and 6 micro meter for granular neurons). Five separate areas were counted for each subfield, and the average for each subfield was computed. Separate counts were obtained from two observers who were blind with respect to presurgical examination findings. Tissue samples occasionally did not permit analysis of all subfields. When this was a consideration during statistical analysis, the number of available samples is presented in the relevant Table 1. Neuronal densities are expressed as measures of volume (mean cell count/mm3).
Normative cell densities were established using tissue collected during autopsy of 22 adults (mean age, 31.3 years) who died of nonneurologic causes. The tissue was obtained within 8 hours postmortem.
Statistical analyses.
The method for establishing the relationship between hippocampal volumetric neuron density and verbal memory ability was determined by computing the Spearman rank-order correlations of FSIQ and the measures of memory. Partial correlations controlling for FSIQ were planned if either Long Term Retrieval or Logical Memory Percent Retention was significantly correlated with the FSIQ of patients with left temporal lobe lesions. Spearman rank-order correlations were planned if neither Long Term Retrieval nor Percent Retention were correlated with FSIQ, because rank-order correlations minimize the impact of outliers upon the statistical results. Rank-order correlations were computed because the sample size was small.
Results.
The hippocampal neuron densities of our patients varied markedly, suggesting that their seizure disorders affected their hippocampal formations differently (see Table 2). Contrasted with autopsy controls, one-way analyses of variance indicated that temporal lobe epilepsy patients with right temporal lobe lesions had significantly reduced neuron densities in all five subfields (see Table 3). Patients with left temporal lobe lesions had significantly reduced neuron densities only in subfields CA1, CA2, and CA3. The mean degree of neuron loss observed for the patient sample did not vary by subfield. The mean reduction of neuron density ranged from 22.7% (CA3) to 17.3% (granule layer). The neuron densities of patients with left temporal lobe lesions did not differ significantly in any subfield from those with right temporal lobe lesions.
Table 2. Ranges for dependent and independent variables
Table 3. Hippocampal volumetric cell densities
The mean Long Term Retrieval of patients with left temporal lobe lesions was significantly less than that of patients with right temporal lobe lesions (see Table 4). Patients with left temporal lobe lesions also performed less effectively on the measures of Logical Memory Immediate Recall, Delayed Recall, and Percent Retention, but the differences were not statistically significant. The mean FSIQ of patients with left temporal lobe masses did not differ significantly from that of patients with right temporal lobe masses.
Table 4. Group differences on measures of verbal learning and memory
The pattern of correlations among measures of cognition differed with the side of the lesion (see Table 5). For patients with left-sided lesions, Logical Memory Immediate Recall was significantly correlated with Logical Memory Delayed Recall but not Percent Retention. Percent Retention was significantly correlated with Long Term Retrieval and Logical Memory Delayed Recall. FSIQ was significantly correlated only with Logical Memory Immediate Recall.
Table 5. Correlations of measures of cognition
For patients with right-sided lesions, Logical Memory Immediate Recall was significantly correlated with Delayed Recall and Long Term Retrieval. Delayed Recall was significantly correlated with Percent Retention and Long Term Retrieval. FSIQ was significantly correlated with both Immediate Memory and Long Term Retrieval.
The FSIQ of patients with left temporal lobe masses was not significantly correlated with Long Term Retrieval or Percent Retention (see Table 5). Therefore, Spearman rank-order correlations were computed to investigate the relationship between hippocampal neuron density and these measures of memory. Neuron density in CA1 was significantly correlated with Long Term Retrieval for patients with left temporal lesions only (see Table 6). Neuron density in CA2 was significantly correlated with Percent Retention for patients with left temporal lesions only. WMS Immediate Recall and Delayed Recall scores did not correlate with neuron density in any subfield for patients with left temporal lobe lesions. No statistically significant correlations were found between neuron density and any measure of memory for patients with right temporal lobe lesions.
Table 6. Spearman rank-order correlations of volumetric neuron density and memory test results
Discussion.
We found statistically significant correlations between neuronal density in subfields CA1 and CA2 and measures of verbal memory for patients with left-sided lesions. There were no significant correlations between neuronal density and verbal memory for patients with right-sided lesions. This is consistent with our prior studies of patients with medically refractory epilepsy who did not have structural lesions. Those studies demonstrated that only left hippocampal neuron loss was significantly correlated with verbal memory impairment. [7,8]
The results of the present study also support our prior conclusions regarding important differences that exist among measures of verbal memory with respect to their sensitivity to cortical versus subcortical left temporal lobe dysfunction. We previously reported that the Long Term Retrieval score of the verbal Selective Reminding Test, as well as the three measures of verbal memory that were derived from the Wechsler Memory Scale (ie, Immediate Recall, Delayed Recall, Percent Retention) distinguished patients with left temporal lobe epilepsy from those whose seizures involved the right temporal lobe. However, only Long Term Retrieval and Percent Retention were correlated with hippocampal neuron loss. During the present study, there were significant correlations with neuron density again only for Long Term Retrieval and Percent Retention. We previously concluded that Immediate Recall could be interpreted as an index of cortical dysfunction within the dominant hemisphere [8,9] and that Percent Retention and Long Term Retrieval could be interpreted as indices of hippocampal dysfunction. [7,9] The present findings indicate that such interpretations can be extended to include patients with structural lesions. Therefore, these test results may be useful in predicting the outcome of surgery for patients with lesions, as they have been for patients without such lesions. [1]
Excision of the hippocampus in addition to a temporal lobe lesion is sometimes performed to obtain adequate margins, particularly when the tumor encroaches upon the hippocampus or to improve seizure control when there is evidence that the seizure focus is within the hippocampus. If the hippocampus is merely a pathway for seizure discharge, rather than an active epileptogenic participant, a functional hippocampus may be removed, causing a predictable decline in memory function. No decline in memory would be expected if the tumor had compromised hippocampal integrity by exerting mass effect, altering local metabolism, or incorporating the hippocampus in the propagation of excitatory impulses. Our analyses suggest that patients at risk for a significant decline in memory after excision of the hippocampus can be identified by their average Percent Retention and Long Term Retrieval scores. [7,9]
Although the present findings are similar to those that we previously reported for patients without structural lesions, there is a potentially important difference. During the present study, there were significant correlations between the results of memory tests and hippocampal neuron density for subfields CA1 and CA2. Prior studies of epileptic patients who had no structural lesions identified statistically significant correlations between the results of verbal memory tests and hippocampal neuron density in subfields CA3 and CA4. [7,8] A contrast of the patterns of hippocampal neuron loss observed for patients with lesions versus those without lesions identified no compelling differences that would account for these observations. We report herein that the degree of neuron reduction was relatively consistent among the five subfields examined (range, 17.3 to 22.7% reduction). We reported previously that neuron loss was greatest in CA1 (70% reduction relative to the autopsy controls reported herein) and least in CA2 (42% reduction) for patients without structural lesions. Therefore, differences in the patterns of neuron loss do not appear to account for the differences in the results of the present study versus those of past studies with respect to the patterns of correlations that were found between hippocampal neuron densities and the memory test results of patients with lesions versus those with epilepsy only. These differences were not attributable to variation in memory testing. The same instruments were employed to assess patients with lesions and those without them. Reassessment of this issue is warranted when more patients are available for study, since there is no compelling explanation for the observed differences.
- Copyright 1995 by Advanstar Communications Inc.
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
