Verbal and nonverbal memory impairment in aphasia

Transcription

1 J Neurol (2012) 259: DOI /s ORIGINAL COMMUNICATION Verbal and nonverbal memory impairment in aphasia Christoph J. G. Lang Andrea Quitz Received: 7 October 2011 / Revised: 21 December 2011 / Accepted: 21 December 2011 / Published online: 19 January 2012 Ó Springer-Verlag 2012 Abstract Repetition is frequently impaired in aphasia, most strikingly in conduction aphasia. The still not fully answered question is whether this relates to a linguistic deficit or to a general impairment of working memory extending to other modalities as well. To contribute to this problem, we assessed 49 aphasic and 50 non-aphasic stroke patients using an aphasia test plus three memory tests in forward and backward fashion, taxing verbal, numerical, spatial, and facial retention. The results show that in aphasics there is a memory gradient declining gradually from verbal to nonverbal content reflecting aphasia severity and that aphasics generally perform worse than non-aphasics, even if they present with similar cerebral lesions. Keywords Language Repetition Auditory Visual Verbal Nonverbal Introduction The question of whether at least some aphasic deficits are due to dysfunctional memory is not new. In the dawn of aphasiology, researchers asked themselves if disorders of repetition for phonemes, words, and sentences could possibly be the result of a breakdown in linguistic memory. It was hypothesized that the disruption of a direct route (fasciculus arcuatus) between an acoustic receptive language center, C. J. G. Lang (&) A. Quitz Neurological Unit, University of Erlangen-Nuremberg at Erlangen, Neurologische Universitätsklinik, Schwabachanlage 6, Erlangen, Germany christoph.lang@uk-erlangen.de A. Quitz aaquitz@web.de located in the temporal lobe, and a motor expressive language center, located in the frontal lobe, was responsible for a pervasive deficit in repetition, as seen prominently with conduction aphasia [1], while it was assumed that the same pathway was spared in the transcortical aphasias [2]. The specific repetition deficit was thought to be due to an impaired memory for sequences [3, 4]. Susan Kohn [1] saw a role for short-term memory in sentence comprehension and suggested that conduction aphasia is sometimes associated with impaired auditory-verbal short-term memory, although the relationship between this and the phonemic output deficit seemed unclear. What was not known and only insufficiently assessed was the question, whether aphasia was regularly or even consistently associated with disorders to repeat or manipulate less or not overtly verbally coded material, such as numbers, geometric, visual-spatial arrays, or human faces [5]. It is also not well known whether there is a marked difference between simple repetition or imitation tasks and tasks requiring the manipulation of material, namely reproducing it in reverse or backward fashion. All of those are considered to tax working memory. In elaborating on his articulatory loop hypothesis, Baddeley [6] suggested that phonological similarity and word length effects were instrumental for working memory performance. He later added a visuo-spatial sketchpad and an episodic buffer to explain the retention of pure nonverbal material. These three components were thought to feed separate inputs into a central executive, a flexible system responsible for the control and regulation of cognitive processes. He acknowledged, however, that the aphasias are often associated with defective short-term memory performance [7, 8]. In a later publication [9] it was contended that Broca s aphasics had disordered phonological memory skills while their immediate visual memory abilities were

2 1656 J Neurol (2012) 259: unimpaired. The authors cited evidence that in this type of aphasia recurring spoken or printed words were not well recognized while memory for shapes that could not be verbalized was not compromised. Other types of aphasia were not mentioned. The question that formed the outline of our research was if aphasias per se or deficits of verbal repetition in particular were more or less regularly accompanied by repetition or (short-term) memory deficits in other domains or modalities, which did not prima facie depend on verbal abilities. It is interesting that Martin and Ayala [10] found a correlation between measures of phonological abilities and a nonverbal span task, suggesting either a general cognitive deficit affecting verbal and nonverbal short-term memory or possibly the (covert) use of a verbal strategy to perform this task. This question therefore gives rise to two different hypotheses regarding short-term memory: the first one assuming a pure verbal deficit being confined to language only, the second one assuming a supramodal memory buffer not exclusively confined to language material. If the first hypothesis should be correct, then (working) memory tasks other than verbal tasks should pose no problem to aphasic patients, while the second hypothesis would predict that in aphasia there should be nonverbal memory problems as well, regardless of the material presented. Methods Patients were recruited from four hospitals in northern Bavaria. One of us (AQ) assessed altogether 99 persons in the acute phase, 49 aphasics (A?) and 50 brain-damaged patients who were not aphasic (A-), matched for sex, age, and schooling, by means of an aphasia test developed by the other author (Kurze Aphasieprüfung KAP Short Assessment for Aphasia [11]) including verbal repetition, and three additional short-term memory, respectively, repetition tests for numbers (digit span subtest of the Wechsler Memory Scale Revised [WMS-R], German version, [12]), visual-spatial arrays (block tapping subtest of the WMS-R), and human faces (Recognition Memory Test [RMT] [13]). Since patients who evidently were not aphasic according to clinical expertise and an extended conversation were not tested for aphasia, the number of aphasia tests in A- was only eight. The memory tests, however, were administered to every patient. Patients (A?/ A-) presented with cerebral infarcts or ischemic lesions (82/74%), cerebral hemorrhage (16/10%), and unspecific white matter changes or no demonstrable lesion (2/16%). Lesion location was left hemispheric (82/34%), right hemispheric (4/34%), or bihemispheric (12/16%). The hippocampus was spared in all patients. Handedness was right (94%/90%), left (4/8%), and ambidextrous (2/2%). The two left-handers and one ambidextrous patient in the A? group suffered from left-hemisphere lesions, the four left-handers and one ambidextrous patient in the A- group from right-hemisphere lesions. All human data included in this manuscript were obtained in compliance with the Helsinki Declaration, the patients or their custodians giving informed consent. More details are given in Table 1. All memory tests except repetition being part of the aphasia examination were given in forward and backward fashion. The memory and aphasia raw sum scores were calculated adding the raw scores of each subtest, the maximum being 70 for aphasia and 76 for memory. The results were then correlated with the type of aphasia, aphasia severity, and the degree of repetition disorder as measured by the KAP. The correlation coefficients were tested for significance (p B 0.05) and compared with each other. Statistical analysis was performed using non-parametric tests (Kolmogorov Smirnov, v 2, median test, regression analysis, Spearman rank correlation) on a commercially available computer program (SPSS Statistics for Windows, version 18). Results The global memory performance of the A? (raw sum score) was distributed normally, that of the A- was not, being skewed. A? scored markedly inferior to A- (Fig. 1a); the difference being highly significant (p = 0.000). This finding held true for each of the memory subtests, regardless if the answer mode was forward or backward, even if a Bonferroni correction for multiple comparisons was applied. There was a significant positive correlation (q = 0.852, p = 0.000) between the aphasia and memory raw sum scores (Fig. 1b). Thus, within certain margins, the severity or type of aphasia was predictive of the results of memory testing (Table 2). Each memory task was correlated with the KAP global score. All results were statistically significant. A regression analysis showed that the slope of the linear regression line declined in the order repetition [ digit span [ block tapping span [ span for faces (Fig. 2a g). A correlation matrix within the A- yielded significant results for all memory tests except repetition (q C.316; p B 0.05), but the correlation coefficients within the A? were higher (q C 0.415; p B 0.01). Remarkably, the correlation between each of the memory tests and the aphasia test sum score was considerable (q C 0.593, p = 0.000). The correlations between digits forward/backward (q = vs ) and block span forward/backward (q = vs ) were higher for A?, the face span forward/backward (q = vs ) higher for A-. There was a general tendency for more severe aphasias (e.g., global) to display more severe memory impairments.

3 J Neurol (2012) 259: Table 1 Characteristic of aphasics and their control group SD standard deviation, F female, M male, n. a. not applicable, TSA transcortical sensory aphasia Aphasics Controls Number 49 (23 F, 26 M) 50 (22 F, 28 M) Mean age (SD) (±15.89) (±12.14) Schooling (years) B9: 79.7%, [9: 20.3% B9:68.0%, [9:32.0% Disease duration (days) 4.7 ± ± 8.1 Aphasia type Global 34.7% n. a. Wernicke and TSA 12.2% n. a. Broca 12.2% n. a. Amnestic 24.5% n. a. Residual 16.3% n. a. Memory raw sum score (SD) (±11.33) (±9.86) Lesion localization Left Right Left Right Anterior cerebral artery Middle cerebral artery Posterior cerebral artery Thalamus Basal ganglia White matter Other (mesencephalon, pons, cerebellum) Another correlational analysis was dedicated to memory testing and subtests of the KAP. All memory subtests correlated significantly with all aphasia subtests (token test, repetition, writing to dictation, reading aloud, naming, auditory comprehension, visual comprehension, B q B 0.842, p B 0.003). The correlations between memory tests and repetition did not stand out among other subtests, although the highest correlation was seen between digit span forward and repetition (Table 3). Digit span forward and face span forward correlated least among the memory measures. The lowest correlation among memory and language parameters was between block span forward and naming. A test for differences between the correlation coefficients among A? and A- yielded significant results (p B 0.05) only for each of the three forward/backward test modes. Discussion The main result of this experimental prospective controlled study is that any memory variable, being overtly verbal or not, forward or backward, was significantly correlated with language variables not only repetition and with the severity of aphasia. It seems as if a certain degree of memory impairment would be intrinsically interwoven with aphasia, even if material is used that is almost impossible to verbalize (faces). This favors the notion of a global (working) memory buffer that is impaired in aphasia. The more severe the aphasia was, the more severe was the memory impairment, although a gradient was seen declining from overtly verbal to overtly nonverbal material. However, each memory task was better solved by A- than by A? patients, although both groups suffered from brain damage. It was Warrington and Shallice [14] who first linked deficits in repetition to an auditory verbal short-term memory deficit in a patient suffering from a left parietal traumatic lesion and subsequent epilepsy; however, when tested for the recognition of incomplete pictures, his performance was normal. De Renzi and Nichelli [15] investigated verbal and nonverbal short-term memory in unilaterally brain-damaged patients and controls. Not surprisingly, they found that left brain-damaged patients were impaired on verbal tests, while right brain-damaged patients were not. A? had a significantly shorter verbal span than A-, whereas spatial span was significantly affected by posterior lesions regardless of their side and independent of aphasia. One single patient suffering from conduction aphasia presented with an exceedingly poor verbal memory span. Interestingly, in the left brain-damaged group there was a high and significant correlation (r = 0.57, p \ 0.001) between the token test indicating aphasia severity, and figure pointing, which requires the patient to point to an array of pictures in the same order as the examiner similar to the block span test. This may be regarded as a faint hint to an impairment in spatial memory, too. Other than Ostergaard and Meudell [9], we found a dependency of memory performance on the severity rather than the type of aphasia. Our data do not lend support to the suggestion that only Broca s aphasics show deficits in

4 1658 J Neurol (2012) 259: Fig. 1 a Memory raw sum scores (mean and standard deviation) for aphasics and controls. b Correlation between memory and aphasia raw sum scores visual recognition performance while Wernicke s aphasics do not. When Wold and Reinvang [4] concentrated on token test (TT) performance in aphasics, they found out that it surprisingly correlated highly (r C-0.60) not only with language comprehension but also with picture pointing span (PPS), which coincides with our results. Other findings, however, were inconclusive or even contradictory: while there was no correlation with other language and memory measures such as repetition, word fluency, average length of utterances, digit span backwards, and block span forwards, the correlation with digit span forwards was low, but significant (r =-0.25), and the correlation with block span backwards considerable (r =-0.42). Repetition in turn correlated highly with digit and picture pointing span, but not with block span. Maeshima et al. [16] supposed that there were connections between short-term memory and aphasia, especially when the temporal or parietal lobe was involved. They also suggested that nonverbal memory may be impaired with left hemisphere lesions. They made the important observation that in traumatic brain injury there may be remitting aphasia with persistent verbal and nonverbal amnesia. Because of the different etiology, these results cannot be readily compared with ours. The assumption of Gathercole and Baddeley [9] that in Broca s aphasia nonverbal memory is intact was in contrast to our findings. Our results clearly show that they were inferior to non-aphasics not only with regard to verbal repetition and digit span but also to memory span for spatial arrays (block span) and faces. Martin and Ayala [10] found a discrepancy between memory span for digits and words, the latter being inferior; they also examined the relation between lexicalsemantic and phonological abilities as well as nonverbal memory span. It correlated with phonological abilities yielding a hint as to an intertwining of verbal and nonverbal faculties. The differentiation between semantics and phonology therefore does not seem very helpful in explaining why visual-spatial relations or unfamiliar faces can be less well memorized than expected. Others thought that an impairment of phonemics or articulation would reduce the capacity of the working memory since this depends largely on subverbalization or inner articulation. Although this assumption would explain a Table 2 Short-term memory performance of aphasics and controls No aphasia (n = 50) Global (n = 17) Broca (n = 6) Wernicke a (n = 6) Amnestic (n = 12) Residual (n = 8) Repetition 9.6 ± 0.7 b 2.3 ± ± ± ± ± 0.9 Digit span 8.5 ± ± ± ± ± ± 2.8 Block span 6.6 ± ± ± ± ± ± 3.0 Face span 3.5 ± ± ± ± ± ± 1.8 Sum score 32.3 ± ± ± ± ± ± 8.7 Short-term memory performance (raw scores, forward fashion) for different modalities broken down according to aphasia type. Given are means ± standard deviations a Wernicke and transcortical sensory aphasia b Only data for eight patients available

5 J Neurol (2012) 259: Fig. 2 a g Correlations between aphasia raw sum scores and verbal and nonverbal memory spans reduction of word or digit span, it can hardly come up for a reduction of overtly nonverbal material. Since verbal contents may be encoded along semantic as well as phonological routes, this distinction does not explain how material that cannot easily be named or encoded verbally can be stored this way at all. Leśniak et al. [17], who concentrated on cognitive impairments as sequelae of brain damage, found out that in

6 1660 J Neurol (2012) 259: Fig. 2 continued Table 3 Correlation between repetition and short-term memory tests Digit span forward Digit span backward Block span forward Block span backward Face span forward Face span backward Aphasics Repetition 0.847** 0.621** 0.490** 0.460** 0.504** 0.494** Digit span forward 0.714** 0.444** 0.453** 0.415** 0.440** Digit span backward 0.548** 0.650** 0.553** 0.569** Block span forward 0.848** 0.707** 0.587** Block span backward 0.786** 0.709** Face span forward 0.704** Controls Repetition (n = 8) Digit span forward 0.413** 0.390** 0.394** 0.316* 0.395** Digit span backward 0.455** 0.523** 0.474** 0.513** Block span forward 0.590** 0.669** 0.673** Block span backward 0.614** 0.541** Face span forward 0.836** Rank correlations between repetition and short-term memory tests in A? (n = 49) and A- (n = 50, except repetition where n = 8). All correlations marked ** are significant at the 0.01 error level; * signifies the 0.05 error level. Correlation coefficients that differ significantly (p B 0.05) between A? and A- are printed in bold type the acute phase, 27% showed some degree of language impairment and 24.5% some degree of memory impairment. To their impression, it could also be the case that the covariation of language and memory impairments was only coincidental proving that a lesion that is large enough or located in critical areas may damage both. A recent study by Seniów et al. [18] dealt with the relation between nonlinguistic cognitive deficits and language rehabilitation in aphasics. Their results were somewhat heterogeneous. Visual-spatial working memory corresponded with naming and comprehension performance, and there was no correlation between the outcome of therapy and abstract thinking. Our study yields two arguments in favor of an intrinsic interrelation between language and memory that is not due to chance. First, all memory measures correlated with all language measures, regardless of their degree of nameability or verbalizability, the aphasia severity being a significant determinant of the degree of memory impairment, although the correlation was all the higher, the closer the memory task was to language abilities. Second, aphasics were definitely more impaired than a similar group of brain-damaged but non-aphasic patients. In theoretical terms, this may either signify that there exists a central short-term or working memory store operating supramodally and being impaired in aphasia

7 J Neurol (2012) 259: producing lesions, or that subverbalization however improbable is also and always active in apparently nonverbal tasks. Further information could possibly be gained by simultaneously analyzing morphological and functional correlates of memory and language performance which, however, was outside the scope of this study. One shortcoming of our work was that lesion type (infarction, hemorrhage) was heterogeneous, that patients were assessed in the acute phase and that not all A- patients were formally tested using an aphasia test. It would also be interesting to examine a larger subgroup of conduction and transcortical aphasics. However, for clinical and rehabilitation purposes, it should be kept in mind that aphasia independent of its type is likely to involve more than mere linguistic deficits, that nonverbal memory may be in need to be assessed and treated too, and that most aphasics are likely to show some impairment in a memory domain. The gradient seen between our different memory tasks indicates that tasks that are close to overt language abilities are more likely to be involved than pure visual(-spatial) tasks, but that the neuropsychological correlates of aphasia-producing lesions are more wide-spread than may be described in pure linguistic terms. Acknowledgments We are grateful to doctors Nückel at Nuremberg, Bößenecker at Amberg (deceased on July 17, 2011), and Rieckmann at Bamberg for help with the data acquisition. Conflicts of interest References None. 1. Kohn SE (ed) (1992) Conduction aphasia. Lawrence Erlbaum Publishers, Hillsdale 2. Berthier ML (1999) Transcortical aphasias. Psychology Press, Hove 3. Tzortzis C, Albert ML (1974) Impairment of memory for sequences in conduction aphasia. Neuropsychologia 12: Wold AH, Reinvang I (1990) The relation between integration, sequence of information, short-term memory, and token test performance of aphasic subjects. J Commun Disord 23: Lang CJ (1989) Continuous figure recognition in dementia and unilateral cerebral damage. Neuropsychologia 27: Baddeley A (1986) Working memory. Clarendon Press, Oxford 7. Allport DA (1984) Auditory-verbal short-term memory and conduction aphasia. In: Bouma H, Bouwhuis DG (eds) Attention and performance X. Lawrence Erlbaum Publishers, London 8. Ostergaard AL, Meudell PR (1984) Immediate memory span: recognition memory for subspan series of words, and serial position effects in recognition memory for supraspan series of verbal and nonverbal items in Broca s and Wernicke s aphasia. Brain Lang 22: Gathercole SE, Baddeley AD (2001) Working memory and language. Psychology Press, Hove 10. Martin N, Ayala J (2004) Measurements of auditory-verbal STM span in aphasia: effects of item, task, and lexical impairment. Brain Lang 89: Lang C, Dehm A, Dehm B, Leuschner T. (1999) Kurze Aphasieprüfung KAP. Manual. Frankfurt/Main, Germany: Swets 12. Härting C, Markowitsch HJ, Neufeld H, Calabrese P, Deisinger K, Kessler J.(2000) WMS-R. Wechsler Gedächtnis test Revidierte Fassung. Bern, Switzerland: Huber 13. Warrington EK (1984) Recognition memory test. NFER-Nelson, Berkshire 14. Warrington EK, Shallice T (1969) The selective impairment of auditory verbal short-term memory. Brain 92: De Renzi E, Nichelli P (1975) Verbal and nonverbal short-term memory impairment following hemispheric damage. Cortex 11: Maeshima S, Uematsu Y, Ozaki F, Fujita K, Nakai K, Itakura T, Komai N (1997) Impairment of short-term memory in left hemispheric traumatic brain injuries. Brain Inj 11: Leśniak M, Bak T, Czepiel W, Seniów J, Członkowska A (2008) Frequency and prognostic value of cognitive disorders in stroke patients. Dem Geriatr Cogn Disord 26: Seniów J, Litwin M, Leśniak M (2009) The relationship between non-linguistic cognitive deficits and language recovery in patients with aphasia. J Neurol Sci 283:91 94