- Top of page
- Patients and Methods
Memory skills, or ability to register, encode, and retrieve previously presented information are fundamental neurobiologic processes that may be impaired in TLE. Although widely used as a lateralizing tool to localize seizure focus, the material-specific model of memory impairment in epilepsy is limited by methodologic constraints of the original studies, especially regarding lack of appropriate neuroimaging evaluation (Baxendale & Thompson, 2010). Lack of consistent findings that indicate an association between nonverbal memory impairment and right TLE has further challenged the material-specific memory impairment model (Kennepohl et al., 2007; Baxendale & Thompson, 2010).
Most previous studies that evaluated the lateralizing ability of memory testing studied heterogeneous patient samples, which included cases of nonlesional TLE, epilepsy due to other etiologies (such as tumors and malformations of cortical development), or extratemporal epilepsy cases (Jones-Gotman, 1986; Loring et al., 1989; Hermann et al., 1995; Baxendale et al., 1998; Baxendale & Thompson, 2010). In this study, we included a relatively homogeneous sample of patients with MRI-diagnosed MTS. Dual-pathology and bilateral MTS cases were not included. Our sample consisted of patients with longstanding (approximately 30-year) TLE associated with hippocampal sclerosis, with balanced gender distribution. Both patient groups were comparable in terms of age at disease onset, epilepsy duration, and AED use.
The patient sample in this study consists of a rather homogeneous patient group, harboring a lesion in a pivotal structure for memory function. The patient sample relative homogeneity may allow us to more clearly discern differences in memory performance of patients with hippocampal lesions in the language dominant and nondominant hemispheres. Because we used cutoff scores derived from a demographically matched control group, our findings cannot be extrapolated to different patient populations.
A different approach could be taken, comparing cognitive performance of patients with MTS to that of patients with lateralized nonlesional TLE. This approach would also present limitations, since patients with MTS would have to be matched to non-MTS patients, regarding interictal and ictal electroencephalography (EEG) findings. In addition, nonlesional TLE cases may include both patients with neocortical TLE (without apparent lesions on imaging studies), as well as patients with subtler hippocampal lesions, such as endfolium sclerosis, not easily detected by MRI.
To avoid age-related memory decline as a confounding factor, we did not include patients older than age 55. In addition, we excluded patients with mental retardation. All patients had at least 8 years of education. These inclusion criteria were chosen to minimize impaired global cognition or inadequate schooling on test performance. When compared to controls, we found that patient groups displayed similar performance in tests measuring attention, abstract reasoning, semantic memory, and visual-spatial skills (data not shown). Impairment in selective cognitive domains (verbal and nonverbal memory, confrontation naming, processing speed, inhibitory control, response selection and global cognitive function) were noted between patients and controls, as well as between right and left MTS patients.
Approximately 12% of controls performed on the impaired range on verbal memory testing. This could be interpreted as indicative of cognitive impairment in a subgroup of controls. However, we do not believe this is the case. Although a minimum of eight education years was required as an inclusion criterion, some apparently normal functioning people in urban populations in large Latin American cities may perform in the impaired range on formal testing, as a consequence of inadequate schooling. Because controls were recruited from the same social and educational background as that for the patient population, it is reasonable to consider that the performance of controls represents an adequate comparison for the patient population. This notion is further strengthened by lack of difference on performance in tests measuring reasoning abilities and semantic memory between patients and controls.
The finding of impaired abilities on selective cognitive domains in the patients compared to controls indicates differences in cognitive abilities related to the disease process, and, possibly, at least in part, to medication effect. We also tried to minimize the negative impact on cognition of a recent seizure by not testing patients who reported having had a seizure within the previous 24 hours.
Medication effects can negatively impact on cognitive function. Although medication effects are more likely to negatively affect attention and executive functions, memory and language abilities may also be influenced negatively by antiepileptic drugs. Patients in both study groups (right and left MTS patients), were receiving a comparable number of drugs, with roughly equivalent total and sedative drug loads; therefore, it is unlikely that difference in performance on neuropsychological testing between right and left MTS is due to a drug effect. Measured differences between right and left MTS patients more likely reflect underlying neurobiologic differences related to hemispheric function specialization of memory functions.
Most studies that evaluated performance on verbal and nonverbal memory tests have used only one test to evaluate verbal and nonverbal memory (Naugle et al., 1994; Hermann et al., 1995; Loring et al., 2000; Sawrie et al., 2001; Raspall et al., 2005).
We evaluated verbal and nonverbal memory with two instruments for each type of memory modality, and used two scores for each instrument to assess memory impairment. In addition, we established appropriate cutoff scores to distinguish patients from controls, who were adequately matched to our patient population. Only one other study that evaluated memory function in epilepsy used a control group (Naugle et al., 1994). Other studies relied on standardized data for controls (Loring et al., 1988, 1989; Naugle et al., 1994; Piguet et al., 1994; Hermann et al., 1995; Loring et al., 2000; Sawrie et al., 2001; Raspall et al., 2005; Grammaldo et al., 2006; Keary et al., 2007). These studies merely compared performance of patients with right and left sided lesions or “epileptic focus.”
Rather than comparing group performances, we used individual patient performance in verbal and nonverbal memory tests to compare groups. We also combined performance in verbal and nonverbal memory test to determine memory performance profiles. This approach has allowed us to discern that patients with mesial TLE do not constitute a homogeneous sample in memory performance.
Approximately half of MTS patients display preserved memory function on neuropsychological testing by our criteria. A minority of patients showed global memory impairment, involving verbal and nonverbal memory. These patterns were seen in equal proportions in right and left MTS cases.
Lateralizing profiles of memory impairment were noted in approximately one third of patients. Selective impairment of nonverbal memory was noted in 25% of right MTS patients. This pattern showed high specificity and positive predictive power for right MTS, albeit with low sensitivity. Selective impairment of verbal memory was noted in approximately 25% of the left MTS cases. This pattern was also seen in 11% of the right MTS patients. This pattern showed moderate positive predictive power and high specificity for left MTS.
With testing for robustness of the memory impairment classification system using less and more stringent criteria for selective memory impairment, we found different results for verbal and nonverbal memory. Although for nonverbal memory more or less stringent criteria did not improve sensitivity, specificity, or negative and positive predictive values for right MTS, the use of more stringent criteria to define verbal memory impairment improved sensitivity, specificity, and positive predictive values for left MTS. However, regardless of stringency of criteria for memory impairment, we consistently found low sensitivity, high specificity, and positive predictive values for selective memory impairment and dominant or nondominant hemisphere MTS.
The memory tests used in this study have not been standardized for Brazilian Portuguese or for Brazilian population. We cannot, therefore, compare our cutoff scores to published standardized data.
Our findings indicate that lateralizing profiles of memory impairment occur in approximately one fourth of patients with MTS patients. When encountered, selective verbal and nonverbal memory impairment appear to be very specific, respectively, for left and right MTS. These findings lend further support to the material-specific model of memory processing (Loring et al., 1988; Golby et al., 2001).
The finding of preserved memory in approximately half of our patient sample is intriguing. It may indicate redundancy in memory function between both hippocampi, thereby protecting certain individuals from material-specific memory impairment resulting from a unilateral hippocampal lesion. Alternatively, one could speculate that neuronal plasticity, especially in the setting of an early lesion, may allow preservation of memory function in both modalities with unilateral hippocampal damage. Memory reserve has been traditionally assessed with the intracarotid amytal test, and, more recently, with memory functional MRI (fMRI) paradigms (Bonnici et al., 2013). Greater activation of contralateral mesial temporal lobe structures on fMRI memory paradigms correlates with better postoperative episodic memory performance in patients with TLE, especially left TLE (Köylü et al., 2008). One may also speculate that age at epilepsy onset (or at the precipitating initial insult) may be related to memory impairment. This question should be investigated in future studies.
It is also unclear why a subgroup of individuals with unilateral hippocampal damage presents global impairment of memory functions, involving both memory modalities. It is tempting to speculate that these patients may display bilateral hippocampal dysfunction, not necessarily related to a lesion. One hypothesis, which remains to be proven, is that hippocampal function may be disrupted an effect at-distance of contralateral abnormal electrical activity. In addition, studies with TLE patients have disclosed widespread changes in cortical thickness, white matter pathways, thalamus, and callosal and cerebellar networks, both ipsilateral and contralateral to the epileptogenic focus, that are associated with impaired cognitive profiles (Dabbs et al., 2009; Bell et al., 2011).
The findings of patterns of selective involvement of verbal and nonverbal memory function, with high specificity, respectively, for left and right hippocampal lesions are in agreement with the findings of fMRI studies that indicate differential patterns of hippocampal activation in response to verbal and nonverbal stimuli (Golby et al., 2001). These findings also corroborate in the clinical setting that memory function displays hemispheric subspecialization in memory function, which should not be understood as exclusively lateralized but rather as displaying complex and asymmetric patterns of interaction.
fMRI studies that evaluated preoperative and postoperative word and face encoding paradigms showed that greater asymmetry of activation was the best predictor for postoperative verbal and nonverbal memory decline. Activation asymmetry, language lateralization, and neuropsychological test performance were better predictors of postoperative memory decline for left compared to right TLE. Verbal and nonverbal preoperative memory utilizes both the diseased and the normal hippocampi. In addition, the ipsilateral posterior hippocampus may contribute to memory reserve (Bonelli et al. 2010).
Our findings indicate that neuropsychological testing has low sensitivity for lesion lateralization in patients with MTS. Neuropsychological testing of memory function in patients with MTS, much more than a mere lateralizing tool, should be viewed as an important instrument in assessing hippocampal function in patients with TLE. Neuropsychological testing performance may be a useful tool, in conjunction with fMRI memory paradigms, to evaluate memory reserve and risk of memory decline following epilepsy surgery.
The recognition of different patterns of memory function in patients with hippocampal sclerosis also opens new avenues for the understanding of neuronal networks involved in memory processes. It also poses relevant clinical questions involving the role of neuronal plasticity in early hippocampal lesions, as well as possible differential effects of epilepsy surgery on memory function, in relation to preoperative patterns of memory performance profiles.
We conclude that patients with epilepsy associated with unilateral hippocampal sclerosis present with different profiles of memory impairment. Lateralizing profiles of selective verbal and nonverbal memory deficits are highly specific for left and right MTS, although infrequently encountered in our patients. Nonlateralizing profiles predominated in this population. These findings suggest hemispheric asymmetry memory function, with complex functional interaction of the hippocampi and limbic structures, and possible compensatory mechanisms in the setting of a unilateral lesion.