OUP user menu

Language outcome following multiple subpial transection for Landau–Kleffner syndrome

Christopher L. Grote, Patricia Van Slyke, Jo-Ann B. Hoeppner
DOI: http://dx.doi.org/10.1093/brain/122.3.561 561-566 First published online: 1 March 1999


Landau–Kleffner syndrome is an acquired epileptic aphasia occurring in normal children who lose previously acquired speech and language abilities. Although some children recover some of these abilities, many children with Landau–Kleffner syndrome have significant language impairments that persist. Multiple subpial transection is a surgical technique that has been proposed as an appropriate treatment for Landau–Kleffner syndrome in that it is designed to eliminate the capacity of cortical tissue to generate seizures or subclinical epileptiform activity, while preserving the cortical functions subserved by that tissue. We report on the speech and language outcome of 14 children who underwent multiple subpial transection for treatment of Landau–Kleffner syndrome. Eleven children demonstrated significant postoperative improvement on measures of receptive or expressive vocabulary. Results indicate that early diagnosis and treatment optimize outcome, and that gains in language function are most likely to be seen years, rather than months, after surgery. Since an appropriate control group was not available, and that the best predictor of postoperative improvements in language function was that of length of time since surgery, these data might best be used as a benchmark against other Landau–Kleffner syndrome outcome studies. We conclude that multiple subpial transection may be useful in allowing for a restoration of speech and language abilities in children diagnosed with Landau–Kleffner syndrome.

  • Landau–Kleffner syndrome
  • language
  • multiple subpial transection
  • EOWPVT-R = Expressive One Word Picture Vocabulary Test—revised
  • PPVT-R = Peabody Picture Vocabulary Test—revised
  • WRAML = wide range assessment of memory and learning


Landau–Kleffner syndrome (Landau and Kleffner, 1957) is characterized by acquired aphasia and behavioural changes, a highly abnormal EEG and seizures that regress over time. Nearly 200 cases have been reported since the initial description in 1957 (Beaumanoir, 1992).

Although this disorder has attracted more attention in recent years, the aetiology, prognosis and optimal method of treatment are unknown. Most interventions have utilized anticonvulsant or steroid therapies. However, outcome of language and behaviour functioning following pharmacological treatment is variable and unpredictable. Marescaux et al. (1990) reported that use of anticonvulsants typically did not lead to an improvement of language ability. High doses of corticosteroids resulted in some improvement in three children, although significant deficits in speech and intellectual ability often remained. Stefanatos et al. (1995) described one child whose language improved after treatment with corticosteroids. Deonna et al. (1989) reported that most patients had severe language deficits when studied several years after the diagnosis of Landau–Kleffner syndrome was made.

A significant limitation noted throughout much of the literature is the absence of psychometric data from either the time of diagnosis or after treatment. Many authors have instead relied on clinical impressions when reporting language and behaviour outcome.

Morrell et al. (1995) reported recently on the outcome of 14 children who underwent multiple subpial transection surgery for treatment of Landau–Kleffner syndrome. This surgical technique is described in detail elsewhere (Morrell et al., 1989), but briefly, the surgery severs tangential, or horizontal, intracortical fibres while preserving the vertical fibre connections of both incoming and outgoing nerve pathways. This procedure is based on experimental evidence (Mountcastle, 1957; Hubel and Wiesel, 1962; Reichenthal and Hocherman, 1977) showing that the basic functional unit of cortical physiology is the vertically oriented column, while the anatomical substrate necessary to generate epileptic activity includes neurons having side-to-side or horizontal linkages. The rationale for using multiple subpial transection for treatment of Landau–Kleffner syndrome is that it is capable of selectively disrupting intracortical fibres with minimal injury to the main vertically aligned elements which might eliminate the capacity of the treated cortex to produce seizures or epileptiform activity, while not significantly compromising its physiological action (Morrell et al., 1995).

In this study, we analysed the results of language assessment performed before and after multiple subpial transection in children diagnosed with Landau–Kleffner syndrome.


Inclusionary criteria for a diagnosis of Landau–Kleffner syndrome at our centre currently include: (i) a history of acquired aphasia, including auditory verbal agnosia, in the context of having previously normal language abilities; (ii) severe epileptiform EEG abnormality characterized by bilateral spike-wave discharges in slow wave stages of sleep; (iii) neuropsychological test findings indicating relative preservation of non-verbal skills; and (iv) electrophysiological evidence of a unilateral origin of the bilateral epileptiform discharge.

Details about the way in which these criteria are determined are presented elsewhere (Morrell et al., 1995), including the use of the methohexital suppression test to demonstrate that despite the apparent bilaterality and widespread distribution, the actual source of primary epileptogenic activity is unilateral and well localized.

In all cases, the experimental nature of the surgical procedure as applied to Landau–Kleffner syndrome was fully explained to the parents and informed consent obtained. The Institutional Review Board (IRB) of Rush-Presbyterian-St Luke's Medical Center gave its support to this specific study, formally including it under the previous approval for the application of subpial transection to the treatment of focal epilepsy.

Over 80 children have presented at our centre with a presumptive diagnosis of Landau–Kleffner syndrome. Of these, 18 met our criteria for Landau–Kleffner syndrome at the time of our evaluation, and subsequently underwent multiple subpial transection surgery. The perisylvian area was transected, to varying degrees, based on the electrocorticogram. The other children referred for evaluation were ultimately diagnosed with some other disorder, typically pervasive developmental disorder, on the basis of their failing to meet one or more of the above-described criteria.

This analysis presents language data on 14 of the 18 children who were diagnosed with Landau–Kleffner syndrome and then underwent surgery. Of the other four children not reported in this series, three are now thought to have a progressive disorder other than Landau–Kleffner syndrome. One child was later found to have a chronic enchephalitis, and the second and third were later shown to have completely autonomous epileptogenic foci in each hemisphere. This is now considered a contra-indication to surgery, but was not at the time of their surgery. These children are described more completely in the paper by Morrell et al. (1995). The fourth child was not available for postoperative language assessment. Ten of the children in this study were included in the previously cited work by Morrell et al. (1995). The additional four children in this study were operated upon more recently.

The number of language evaluations completed for each child varied and were conducted by different examiners. Most language evaluations were conducted at our centre, but some follow-up evaluations took place in the child's state of residence. Where multiple pre- and/or postoperative evaluations had been conducted, data from the last preoperative and from the most recent postoperative evaluations were chosen for analysis.

Two tests were given to nearly all children both before and after surgery. The Peabody Picture Vocabulary Test—revised (PPVT-R) (Dunn and Dunn, 1981) is a measure of receptive vocabulary. The test requires the child to point to one of four pictures displayed on a card as depicting the word spoken by the examiner. Normative data for ages 2.5–40 years are available. The Expressive One Word Picture Vocabulary Test—revised (EOWPVT-R) (Gardner, 1990) is a test of expressive vocabulary. The child is shown a picture and asked to name it. Norms are available from age 2 years 0 months up to 11 years 11 months.

Twelve of the 14 childrens' parents were recently contacted by phone to determine that we had the latest available language test results. Parents were also questioned about their children's school placement and performance, and whether the children were enrolled in any type of language remediation programs. The other two children have moved, without leaving a forwarding address or phone number.

Each of the 14 children was reported to have had a history of normal language and cognitive development until a sudden onset of impairments in receptive and expressive language. The mean age (standard deviation) at which deficits were noted was 4.0 years (1.1) with a range of 3.0–6.5 years (Table 1). In two cases (9 and 10 of Table 1) the children had recurring acquisitions and losses of language ability. Both patients had lost all, or nearly all, of their expressive language at some point before surgery. Although their language function at time of surgery was not typical of the other children in this sample, they were operated upon because they each had an active epileptic focus, had been experiencing losses and re-acquisitions of language ability, and had not regained all of their premorbid language abilities.

The average age at which children underwent multiple subpial transection was 7.4 years (SD = 2.2); the youngest at time of surgery was 5.1 years, the oldest 13.1 years. In nine cases, the left hemisphere was transected; the right was operated upon in the other five cases. The average time elapsed between the age of language decline to the age at surgery was 3.4 years (2.0); the shortest period of time between decline and surgery was 0.7 years, the longest was 7.1 years. Three of the children (2, 8 and 14 from Table 1) were operated on twice, as the first surgery did not completely eliminate the epileptiform activity.

Seven of the cases were male, seven female. There were no significant gender differences in the age at time of speech decline, age at time of surgery, amount of elapsed time between age of decline and surgery, or side of brain operated upon (all P > 0.10).

The standard scores of the vocabulary tests are listed in Table 1 as they provide better measures of age-related language functioning than do raw scores.


Table 2 lists the pre- and postoperative raw and standard scores for the PPVT-R and EOWPVT-R. Application of paired t tests show statistically significant improvements for both raw and standard scores of both tests.

Although group statistics clearly point to a significant improvement in test performance across time, an attempt was made to determine what percentage of the children showed a significant improvement on each of these tests. For the purposes of this study, a change of 5 or more points of the standard score was considered to be evidence of a significant change. This criterion was based on the finding that the median increase in PPVT-R standard scores over a period of 6 years was only 4.5 points in the standardization sample (Dunn and Dunn, 1981).

Thirteen children completed the PPVT-R before and after surgery. Data analysis indicates that the PPVT-R scores of seven children improved significantly, for five children the scores did not change significantly and for one child the scores worsened significantly, following surgery. All 14 children completed the EOWPVT-R before and after surgery. Eight children improved significantly, five did not change significantly, and one worsened significantly following surgery. Summarizing data across the two tests indicates that four children showed a significant improvement on both tests, six children improved on one test but not the other, one child improved on one test but worsened on the other, two children did not show any significant changes on either test and one child worsened on one test but did not change on the other test.

Two patients experienced some morbidity following surgery. Patient 8 contracted meningitis following surgery, which likely accounts for her diminished ability on the PPVT-R. Patient 10 suffered a stroke, from which she has largely or completely recovered. We then examined which factors might account for changes in test scores following surgery. We first calculated changes in raw test score performance, by subtracting the preoperative language score from the postoperative language score. A resulting positive difference score represents the number of additional items passed on postoperative examination.

Table 3 indicates that the extent of improvement demonstrated for both PPVT-R and EOWPVT-R raw scores was significantly correlated (P < 0.001) with the length of time elapsed between surgery and the date of the most recent language evaluation, i.e. those children who showed the greatest improvements in language function were those last assessed at a time that was years, instead of months, following surgery. As an illustration, patients 7 and 11 (Table 1) showed the greatest postoperative gains on language testing in this sample. They were last assessed 6.3 and 3.9 years, respectively, after surgery. In contrast, children 4 and 12 had shown no postoperative improvement in test scores, and were last tested only 0.6 and 0.8 years, respectively, after surgery.

Table 3 indicates that changes in PPVT-R scores also correlated significantly (P < 0.05), but inversely, with the number of months elapsed between age at which language declined, and age at which surgery was performed, i.e. children who had longer periods of time between age of onset and age of surgery were less likely to demonstrate improvements in their receptive vocabulary following surgery. Age at time of language decline, and age at time of surgery, did not significantly correlate with postoperative changes in test scores.

Dichotomous variables (gender; number of surgeries; left versus right hemisphere surgery; existence of preoperative seizures) were also examined. None of these variables had a significant relationship (all P > 0.10) with language change scores.

Additional, but varied, measures of postoperative language and cognitive function were administered to these 14 children. For example, patient 11, whose preoperative language scores were below the 1st percentile of ability, demonstrated a postoperative Wechsler Intelligence Scale for Children—third revision (WISC-III) Verbal IQ score of 114 and a Performance IQ score of 121. This child scored above the 95th percentile on a measure of general verbal knowledge (Information), and in the above to above-average range on most other measures of verbal ability. However, this child still performed relatively poorly on tests which require attending to auditorily presented information and then repeating it exactly as it was said and in the same order. He scored at the 5th percentile on the wide range assessment of memory and learning (WRAML) (Sheslow and Adams, 1990) Number/Letter subtest (which requires the child to repeat a string of numbers and letters) and at the 9th percentile on WRAML Sentence Repetition (which requires the child to repeat a sentence exactly as it was read to him). In contrast, he did well both on tests of visual attention (WRAML Finger Windows = 91st percentile) and on tests which allow the child to paraphrase verbal material (WRAML Story Memory = 50th percentile) or to learn by rote single words, but repeat them in any order (WRAML Verbal Learning = 95th percentile). Similar patterns of strengths and weaknesses were noted in children 5 and 7. In each case the child did well on most measures of verbal ability, but displayed relative weaknesses on measures which required the child to recite verbal material exactly as he or she heard it with no provision for paraphrasing or re-sequencing.

Finally, 12 parents were interviewed recently by phone to determine whether their children required any type of special or remedial education in school. Four parents reported that their children were in mainstream classrooms and received no type of remedial or special services. Another two parents reported that their children were `mainstreamed', but had `resource specialists' available to assist them when needed. Six parents reported that their children were in `special' setting classrooms. Two of these children (2 and 12) were in rooms for hearing-impaired children; the other four children were in rooms for children with learning problems.


In this study, we administered tests of receptive and expressive vocabulary to 14 children before and after they underwent multiple subpial transection surgery for treatment of Landau–Kleffner syndrome. Paired t-tests indicate significant improvements in raw and standard scores on both tests following surgery. Analysis of individual test–retest scores indicates that four children improved on both tests, six children improved on one test but not the other, one child improved on one test but worsened on the other, two children did not show change on either test, and one child worsened on one test.

The most obvious difference between children who did and did not show significant postoperative score gains was the length of time between surgery and the most recent evaluation; those children tested the longest time after surgery were those showing the greatest postoperative gains, while those tested only a short time after surgery typically did not show significant postoperative improvements on tests of language function. Also, the amount of postoperative gain on the PPVT-R was significantly inversely correlated with the length of time between age of language decline and age at surgery. That is, children with longer periods of language impairment prior to surgery were less likely to show significant postoperative improvements on this measure of receptive vocabulary.

It is difficult to compare the postoperative gains in language ability demonstrated in this sample against those of Landau–Kleffner syndrome children in other investigations who were either untreated or who received pharmacological therapy. No other investigations have consistently utilized psychometric data to monitor outcome over time in a group of children with Landau–Kleffner syndrome. However, three reports have used at least some psychometric data to report outcome. Deonna et al. (1989) presented follow-up psychometric data on seven adults who had previously been diagnosed with Landau–Kleffner syndrome, but scores from the time at which decline was first noticed were not available. Also, information about pharmacological treatment (if any) is not described in detail. Stefanatos et al. (1995) reported language and IQ scores in a case study of a child whose abilities improved after treatment with corticosteroids. However, it is not clear whether this child had Landau–Kleffner syndrome as no EEG disturbance was evident. Finally, Mantovani and Landau (1980) evaluated nine children 10–28 years after the onset of aphasia. Four children were reported to have recovered fully, one had mild language dysfunction, and four had moderate language disability. Follow-up measures of language (Token Test, Porch Index of Communicative Ability), intelligence (Wechsler Adult Intelligence Scale) and visuoperceptual ability (Benton Visual Retention Test) are reported for seven of the children, but scores from an earlier point in their illness were not available.

Other studies describe outcome, but apparently did not utilize psychometrics to measure change in language or cognitive abilities. Paquier et al. (1992) reported follow-ups on six children with Landau–Kleffner syndrome, the length of follow-up ranging from 3 to19 years. Outcome among the children varied. One child, treated with sodium valproate, became `one of the brightest pupils in her class'. However, another child, treated with ethosuximide and valproate, apparently could speak only a few words.

Lerman et al. (1991) reported that four children treated with corticosteroids were doing well after following medication. Marescaux et al. (1990) and Hirsch et al. (1990) reported that, in a sample of five patients with Landau–Kleffner syndrome, antiepileptic medication was either ineffective or aggravative. However, the three patients treated with corticosteroids showed improvements both clinically and electroencephalographically.

Although the efficacy of treatment with multiple subpial transection cannot easily be compared with other methods of treatment, differences in other aspects of outcome between our samples and others deserve comment. The first concerns the relationship between age at onset and outcome. Bishop (1985) and Dulac et al. (1983) each found that a younger age of onset was associated with poorer outcome of language function. This was not true in our sample, as age of onset did not significantly correlate with either language scores following surgery or the amount of gain made on language tests following surgery (all P > 0.10). The reasons for this discrepancy are not clear. It might be that surgery allowed for a recovery of language function not demonstrated in these previous studies. It is also possible that the lack of empirical follow-up data available to Bishop (1985) and Dulac et al. (1983) which forced them to make qualitative judgments about outcome, led to conclusions that will not be supported by studies that do have access to quantitative assessments of language function across time. Of course, the relatively small number of subjects in our study and others' studies might also be the cause of this discrepancy. Secondly, other investigations apparently have not commented upon the relationship between outcome and the time at which outcome was measured. Examination of the available data listed in these other reports, however, does not clearly indicate that superior outcomes were more likely to be found when the patients were examined at longer intervals after either treatment or the onset of symptoms. In contrast, patients in our sample were more likely to show significant gains in test performance if tested 2 or more years after surgery than if tested a few months after surgery. Examination of serial testing performed in the postoperative phase of our sample generally indicates that children did not show significant gains on language testing until ≥6 months after surgery, and that they generally then made steady improvements from one evaluation to the next. There were no instances of a child making gains and then losing them. We interpret these findings as evidence that surgery allows for a reacquisition of language, perhaps by halting the epileptiform discharges that interfere with optimal pruning following synaptogenesis, and allowing for a reacquisition of language while the child is still capable of such relearning (Morrell et al., 1995). The delay between surgery and reacquistion of language function demonstrated in our patients, also suggests that other variables, such as schooling, language therapy or time itself, may also be instrumental in the reacquistion of these abilities.

Given that children generally show improvements some time after surgery, the hypothesis that spontaneous remission, and not surgery, accounts for these improvements must be considered. However, as previously reviewed, we are not aware of evidence from any other investigation that children with Landau–Kleffner syndrome will spontaneously improve to the degree seen in at least some children here.

Other aspects of our patients' outcome are now commented upon. First, in this paper we have focused on outcome of two standardized tests of language function, the PPVT-R and the EOWPVT-R, as these were the most frequently administered tests. Because these children were referred from all over the country, we have sometimes had to rely on follow-up evaluations done by professionals closer to the child's home, thus leading to a variety of tests administered in the postoperative phase. Review of these other tests generally indicates that patient's language abilities improved in other ways than recognition or expression of single vocabulary words. For instance, patient 14 was shown the Cookie Theft Picture card from the Boston Diagnostic Aphasia Examination before and after surgery. Preoperatively, his description of the card was `oops', `down the water' `and eat'. Postoperatively, he stated that the `the boy is falling and the water is falling down', `the mom is washing the dishes'. Similarly, this child showed significant improvements following surgery on each of the subtests of the Porch Index of Communicative Ability for Children. Similar types of improvements in language ability were seen in several other children, but systematic comparison of pre- and postoperative test scores is not possible because of the variety of tests employed. A second feature of these data is the finding that patients continue to have difficulty in repeating words or numbers exactly as they were presented to them. Deficits are seen on tests of digit span or sentence repetition, but not on those tests that allow the child to paraphrase what they heard (e.g. WRAML Story Memory) or to repeat single words, but in any order that they desire (e.g. WRAML Verbal Learning). Perhaps this reduced auditory attention span is a residual deficit of the auditory verbal agnosia. A third feature of our data set is the finding of age-appropriate (defined as performance at or above the 5th percentile of same-aged peers) non-verbal ability. We have previously reported upon the concordance of age-appropriate non-verbal skills and EEG abnormalities that are consistent with a diagnosis of Landau–Kleffner syndrome (Grote et al., 1996). In that study, children with age-appropriate non-verbal skills were very likely to have EEG abnormalities consistent with Landau–Kleffner syndrome, while those children who had impaired non-verbal skills were very unlikely to have these EEG findings.

In summary, a majority of children who underwent multiple subpial transection for treatment of Landau–Kleffner syndrome experienced a significant gain in one or more areas of language functioning following surgery. This suggests that multiple subpial transection surgery can allow for substantial and long-lasting improvements in language functioning for children with Landau–Kleffner syndrome. However, as appropriate non-operated controls were not available to us for study, and as the best predictor of outcome is the amount of time elapsed since surgery, it is difficult to disentangle the effects of surgery from those of the natural history of the disease. Therefore, these data might also be viewed as a benchmark against which data of other Landau–Kleffner syndrome outcome studies might be compared.

View this table:
Table 1

Clinical characteristics and test scores of children operated on for Landau–Kleffner syndrome

PatientAge at decline (years)Preop. szs?Age at surgery (years)Area of MSTPre-PPVT (SS)Post-PPVT (SS)Pre-EOW-PVT (SS)Post-EOW-PVT (SS)Length of follow-up (years)Special services at school?
Calculations refer to time of second surgery. Preop. szs? = preoperative seizures?; for test scores: SS = Standard Score. *Those patients who underwent MST on two separate occasions. L = left; R = right; F = frontal; T = temporal; P = parietal.
13.0No 6.0LFTP<4073<55 1074.9No
24.8Yes 7.4*LFTP<4059<55<554.9Yes
33.1Yes10.2RFTP565383 1192.2Yes
43.0Yes 7.6LFT<40<40<55<550.6Yes
55.5Yes 7.1LTP<4097<55 1096.6?
66.5Yes13.1RFTP<4042<55 673.9Yes
73.5Yes 5.5LFTP<40 103<55 1246.3No
83.9No 6.8*RFTP6949<55 832.8Yes
93.0Yes 5.1RFTP797590 810.6?
103.5Yes 8.2LTP7976 830.5Yes
113.5Yes 5.2RTP<4095<55 1103.9No
124.5No 6.9LFT<40<40<55<550.8Yes
134.1Yes 8.8LFTP588086 860.6Yes
144.8No 5.5*LT<407067 671.4No
View this table:
Table 2

Mean (SD) raw and standard scores (SS) before and after multiple subpial transection

PPVT-R (n = 13)
Raw20.9 (24.0)70.0 (32.2)0.001
SS47.8 (13.4)67.4 (22.2)0.022
EOWPVT-R (n = 14)
Raw24.1 (24.7)57.9 (26.3)0.002
SS64.1 (13.6)85.8 (24.4)0.009
View this table:
Table 3

Pearson r correlations between changes in raw test scores and other variables

Change in PPVT-RChange in EOWPVT-R
Changes in PPVT-R and EOWPVT-R were calculated as postoperative score minus preoperative score, so that resulting positive difference scores reflect improvement after surgery, and negative scores reflect a decline. Length of illness was calculated as the number of months between age at which language decline was noticed and age at which surgery was performed. *P < 05; ** P < 0.001.
Age at time of language decline 0.21 0.13
Age at time of surgery–0.27–0.07
Length of illness prior to surgery–0.51*–0.29
Length of time between surgery and most recent evaluation 0.88** 0.95**


We are deeply indebted to the late Frank Morrell, MD, for sharing his patients with us.


View Abstract