• Rasmussen syndrome;
  • Hemiparesis;
  • Cortical damage;
  • Burden of pathology


  1. Top of page
  2. Abstract
  7. Acknowledgments

Summary: Purpose: Rasmussen syndrome (RS) is a rare form of epilepsy characterized by progressive destruction of a single hemisphere. To characterize the profile of cortical involvement in RS, we studied the pathological changes in the cerebral cortex of 45 hemispherectomies performed at Johns Hopkins Hospital between 1985 and 2002.

Methods: The patterns of pathologic changes and stages of cortical abnormalities were studied by histology and immunocy to chemistry methods. The burden of pathology (BP) was quantified in all brain regions of each of the 45 hemispheres.

Results: Our study demonstrated significant heterogeneity in the stages of cortical pathology and the multifocal nature of the disease. These stages varied from early inflammation defined by infiltration of T lymphocytes and neuroglial reactions, to more severe stages with extensive neuronal cell death and cavitation of the cerebral cortex. A greater BP was significantly associated with an early age at onset (p = 0.01) and longer duration of disease (p ≤ 0.001). The BP was similar in all brain regions except the occipital lobe, where the BP was significantly lower (p = 0.032).

Conclusions: The multifocal distribution of pathologic changes, as well as the heterogeneity in the stages of cortical damage in each patient, is consistent with an ongoing and progressive immune-mediated process of neuronal damage that involves neuroglial and lymphocytic responses, resembling other autoimmune CNS disorders such as multiple sclerosis.

Rasmussen syndrome (RS) is a rare, progressive, unilateral neurologic disorder of previously normal children who usually are first seen with intractable focal seizures (1–4). As the illness progresses, the child slowly but inexorably loses function of the hemisphere, develops a hemiparesis, and, when the dominant hemisphere is involved, loses language. Progressive intellectual and functional deterioration ranges in duration from months to years. The ultimate outcome is almost always an individual with a dense hemiparesis, substantial intellectual impairment, continued seizures, and a severely impaired quality of life (4,5). Cortical inflammation, neuronal loss, and gliosis confined to one hemisphere are the characteristic pathologic hallmarks of the condition (6–8).

The purpose of this study was to examine the profile, topography, patterns of cortical involvement, and the burden of pathology in the brain obtained after hemispherectomy in patients with RS. This is a cross-sectional study, with tissue obtained at a single point in time for each patient. Through this process, we developed a conceptualization of the nature of the disease process and understanding of the evolution and pathogenic mechanisms of the disease.


  1. Top of page
  2. Abstract
  7. Acknowledgments

Patient population

Forty-five patients underwent hemispherectomy for treatment of RS at Johns Hopkins Hospital between January 1985 and December 2002. All patients demonstrated the typical clinical profile described for RS (1–4,9), as they were normal children who experienced intractable partial seizures that occasionally generalized, progressive hemiparesis, and progressive deterioration in intellectual function. Early specimens were handled as part of routine pathology studies, whereas recent specimens were obtained as part of an ongoing research protocol (1993–2002) to correlate pathology and brain function, and consent was obtained under a protocol approved by the Johns Hopkins Committee for Protection of Human Subjects. The demographic information of the patients included in the study is described in Table 1. Clinical information about some of these patients has been included in previous publications (4,10–12). The duration of illness was defined as the period between onset of seizure activity and the hemispherectomy.

Table 1. Descriptive statistics of patient population
Mean age at onset (yr) 6.8 ± 3.4 6.9 ± 2.9 7.0 ± 3.8
Range of age at onset (yr) 1.5–16.5 1.9–11.1 1.5–16.5
Mean age at surgery (yr) 9.6 ± 4.3 9.4 ± 3.4 9.8 ± 4.8
Range of age at surgery (yr) 3.8–20.6 4.5–16.7 3.8–20.6
Mean duration of symptoms (mo)33.7 ± 31.033.7 ± 26.633.7 ± 34.0
Range of duration (mo)   5–163  10–110   5–163

Surgical procedure and tissue handling

The patients included in this study underwent a hemidecortication, as described previously (11). All tissue obtained from the surgical procedure was processed for neuropathological analysis. Representative cortical areas from each lobar region of the brain were sampled for histopathological analysis. The tissue obtained for histopathological studies was fixed in 10% buffered formalin or 4% paraformaldehyde in 0.1 M phosphate buffer for 48 h.

Histologic and immunocytochemical techniques

The 1,155 samples representative of all regions of the brain from the 45 RS hemispherectomies were paraffin processed (a mean of 25.7 cortical areas per hemispherectomy). All blocks of tissue were cut and 10-μm sections stained with hematoxylin and eosin (H&E) and Nissl stains. These sections were surveyed for an initial evaluation to define cytoarchitectural and pathologic features. After the initial assessment, selected areas with different types of pathologic changes were stained with Luxol fast blue/HE (LFB/HE) to study myelin changes and immunocytochemistry techniques to define the characteristics of astroglial and microglial reactions as well as lymphocyte infiltration and axonal injury (Table 2). The immunocytochemical procedures followed the recommendations given by the antibody manufacturers and included antigen-unmasking techniques. Double immunocytochemistry was used to evaluate CD3 versus CD4 and CD8 cell populations. The immunocytochemical procedures for single and double immunocytochemistry have been described previously (13), and the immune reaction was disclosed with Vector SG and/or diaminobenzidine tetrahydrochloride (DAB) substrates as chromogens (Vector Laboratories, Inc., Burlingame, CA, U.S.A.).

Table 2. Antibodies used in immunocytochemical studies
Antibody/typeCell population/epitopeSourceDilution
  1. p, polyclonal antibody; m, monoclonal antibody; MHC, major histocompatibility complex.

GFAP (p)Astrocytes, glial fibrillary acidic proteinDAKO, Inc.1:500
CD-68 (m)Monocytes, macrophages, microgliaDAKO, Inc.1:500
HLA-DR (m)Activated microglia, MHC class IIDAKO, Inc.1:500
CD3 (p)T lymphocytesDAKO, Inc.1:500
CD19 and CD20 (m)B lymphocytesDAKO, Inc.1:200
CD4 (m)CD4+ lymphocytesNovocastra, Inc.1:50
CD8 (m)CD8+ lymphocytesNovocastra, Inc.1:50
SMI 312 (m)Axons, phosphorylated neurofilamentsSternberger Monoclonals, Inc.1:1,000

Staging of pathologic changes in RS

To define pathologic features of the disease, we adapted a conceptualization of disease stages for RS patients established by Robitaille (6). The assessment of histopathologic changes was based on the profile of the destructive process throughout the cerebral cortex of the different regions of the hemisphere. A summary of the staging is found in Table 3 and Fig. 1. The normal cerebral cortex in which no evidence of neuronal loss and absence of reactive astrogliosis, microgliosis, or lymphocyte infiltration was found was defined as stage 0.

Table 3. Staging of cortical pathology in Rasmussen syndrome
StagesStage 0Stage 1Stage 2Stage 3Stage 4
DefinitionNormal cortexEarly stageIntermediate stageLate stageEnd stage
Cerebral cortexNormalMild focal inflammation and gliosisPanlaminar cortical inflammation and gliosisPanlaminar cortical degeneration and gliosisPanlaminar cortical cavitation and/or gliosis
Neuronal lossAbsentMinimal, focalModerate to severe, multifocalSevere, panlaminarSevere, rare neurons
AstrogliosisAbsentMild to moderate, focalMarked, frequently panlaminar, gemistocytesMarked, panlaminar, gemistocytesVariable
Microglial activationAbsentMild to moderate, focalMarked, panlaminarVariableVariable
T-cell infiltrationAbsentMild to moderate, few T-lymphocyte clusters and perivascular cuffsMarked, panlaminar or multifocal, frequent perivascular cuffsMinimalRare

Figure 1. Stages of cortical pathology in Rasmussen syndrome (RS). A–D: The profile of histopathologic changes observed in the different stages of cortical damage in RS as observed in HE/LFB stain. A: In stage 1, minimal or focal inflammatory changes appear (arrow). Inserts show images of perineuronal lymphocytes and lymphocyte clusters. In stage 1, the magnitude of astroglial reaction (glial fibrillary acidic protein immunostain) (E), activation of microglia (HLA-DR immunostain) (G) and T-lymphocyte infiltration (CD3 immunostain) (I) are restricted to focal areas. The inserts in E and G show the relation of reactive astroglia and microglial nodules with lymphocyte clusters. In stage 2 (B), the inflammatory reaction has a multifocal panlaminar distribution (arrows). Frequent figures of neuronophagia, lymphocyte clusters, and marked perivascular astroglial reaction are observed (inserts in B). The astroglial reaction (F), microglial activation (H), and T-lymphocyte infiltration (J) have a panlaminar distribution. Gemistocytic astrocytes are widespread, distributed in cortex and perivascular regions (inserts in B and F). Perivascular macrophages also are prominent (CD68 immunostain, insert in H). In stage 3 (C), neuronal loss, focal spongiosis, panlaminar astroglial reaction, and few foci of T lymphocytes are observed. In this stage, decrease in the cortical thickness (*) and atrophy are prominent. Gemistocytic astrocytes are prominent as well as foci of cortical spongiosis (inserts). In stage 4 (D), extensive cortical vacuolation and panlaminar degeneration are the most important features. Large cavitary lesions and septal astrogliosis are seen (inserts). The lymphocyte infiltration is composed of T cells (CD3 immunostain) (K) with a large proportion of CD8+ T lymphocytes (M), many of them closely attached to neurons (L). Few B lymphocytes are observed, mostly in the perivascular compartment (N). (Bars A–J, 200 μm; inserts A–H and K, M, and N: 25 μm; L: 10 μm).

Download figure to PowerPoint

Early stage of cortical pathology (stage 1)

Stage 1 was defined as the earliest stage of cortical involvement with discrete foci of inflammatory cells without significant evidence of cortical or neuronal injury. The most consistent histopathologic change observed, and considered the earliest pathologic abnormality, was the presence of subtle infiltration by lymphocytes in superficial and deep neuronal layers of the cerebral cortex (Fig. 1A and I). Additional immunocytochemical studies, using astroglial and microglial markers, consistently demonstrated the concomitant presence of focal astroglial (Fig. 1E) and microglial reactions (Fig. 1G). No significant neuronal loss was noted in areas with early pathologic changes, but occasional examples of injury were found with perineuronal satellitosis due to accumulation of lymphocytes and microglial cells (Fig. 1A, E, and G inserts). The lymphocytic infiltrates were characteristically located in perineuronal (Fig. 1A, inserts K and L) and perivascular regions. The perineuronal accumulation of lymphocytes gave the appearance of lymphocytic nodules or clusters (Fig. 1A insert and 1I).

Intermediate cortical pathology (stage 2)

Stage 2 was defined as intermediate cortical pathology. An increase in the magnitude of lymphocytic infiltration occurred (Fig. 1B) as well as the progression of astroglial and microglial reactions from focal areas to panlaminar distribution (Fig. 1F and H). At this stage, evidence was found of neuronal injury, demonstrated by the presence of cytoplasmic and nuclear changes and increased amounts of perineuronal satellitosis (Fig. 1B insert) and neuronal degeneration. Evidence of patchy neuronal dropout also was noted. In some cortical regions, evidence of cytoarchitectural change was characterized by ballooning and distortion of the neuronal shape, reminiscent of cortical dysgenesis. Both microglia and astroglia showed an increase in cytoplasmic volume and extension of multipolar processes as indication of their stage of activation (14–16) (Fig. 1B, F and H inserts). The lymphocyte population was predominantly a T-lymphocyte infiltrate, as shown by immunocytochemical studies with anti-CD3 antibodies (Fig. 1J) and was composed of a mixture of CD4+ and CD8+ lymphocytes, although CD8+ cells predominated (Fig. 1M). Infrequent B lymphocytes were seen exclusively in perivascular but not in perineuronal regions (Fig. 1N). Two patterns of cellular reaction were seen at this stage: (a) an inflammatory pattern in which an extensive panlaminar infiltration of lymphocytes was seen (Fig. 1B and J), and (b) a prominent gemistocytic astroglial pattern (Fig. 1F insert). In many cases, both patterns coexisted and also were associated with marked microglial activation (Fig. 1H).

Late stage of cortical pathology (stage 3)

Stage 3 was defined as showing significant decrease in the neuronal population, either in large focal areas or panlaminar distribution (Fig. 1C). Astrocytes still continued to show gemistocytic features (Fig. 1C insert), and foci of activated microglia and, in some instances, large phagocytic microglia were seen. The brain at this stage showed cortical atrophy due to marked neuronal loss. Focal cortical spongiosis was seen in some cortical areas (Fig. 1C) without panlaminar involvement. Reactive astrocytosis, but not necessarily an overwhelming inflammatory lymphocytic reaction occurred; sometimes only a few clusters of lymphocytes were observed (Fig. 1C).

End-stage cortical disease (stage 4)

Stage 4 was defined as extensive destruction of the cerebral cortex. Invariably, cortical vacuolation (Fig. 1D) or evidence of complete panlaminar neuronal dropout and degeneration was found. Residual astrogliosis was prominent in such regions, whereas minimal or absent inflammatory changes remained.

Topographic analysis of lesions

The hemispherectomy tissue was sampled exhaustively in a subset of six cases to obtain a detailed topographic distribution of the pathologic changes in RS. Slices of tissue were obtained and processed as described in the histological analysis section. The pathologic changes were localized by using the photographs taken at the time of surgery and projected to brain diagrams to disclose the topography of the lesions and the variability of pathologic stages throughout the hemisphere.

Clinical–pathologic correlations

To define pathologic features of the disease, we used the conceptualization of disease stages for RS patients, as it was described earlier. We, however, used 1,155 samples from the cerebral cortex to provide a more detailed analysis of the variability and magnitude of the cortical injury. This initial analysis was based on the evaluation on H&E and Nissl-stained slides, in which we assessed neuroglial reactions, lymphocyte infiltration, and neuronal loss. For each child, we collected n1, n2, n3 and n4 slides for the frontal, temporal, parietal, and occipital lobes, respectively. For the jth slide (1 ≤ j ≤ ni) of the ith lobe (1 ≤ i ≤4), we assessed its pathology with a score Sij as either: (0) normal, (1) early cortical changes, (2) intermediate-stage cortical changes, (3) late-stage cortical changes, or (4) end-stage cortical pathology.

To establish a quantitative assessment of the burden of pathology (BP) in each of the four cortical regions, we divided the total score in a region (sum of Sij for 1 ≤jni) by the total number of slides in that region (ni) and standardized it by dividing by 4 (the maximum score) to arrive at a measure of BP to be between 0 and 1. That is the BP in the ith lobe. BPi= (∑jSij/4ni) corresponding to the frontal, temporal, parietal, and occipital lobes for i= 1, 2, 3, and 4, respectively. A BP equal to 0 will correspond to all the slides in a region being classified as normal. Conversely, a value of 1 as BP will result from all slides in a region being determined as end-stage pathology. The overall pathology in the brain will be characterized by BP1, BP2, BP3, and BP4, and an overall BP can be derived by (n1BP1+n2BP2+n3BP3+n4BP4)/(n1+n2+n3+n4).

For example, if there are n1= 7 slides from the frontal lobe of a patient, two samples might be at pathologic stage 0 (a score of 0 of a possible 8 = 4 × 2), three samples at stage 1 (a score of 3 of a possible 12 = 4 × 3), and two samples at stage 3 (a score of 6 of a possible 8 = 4 × 2). The frontal lobe BP score would be 9 (= ∑jSij) of a possible 28 or a BP score of 9/28 = 0.32. The minimum score would be zero. The maximum score would be 1.

Statistical modeling and analysis

The measures of the BP in the four regions for the 45 children were used as the outcome to determine the effect and significance of differences by region (using the frontal region as the reference category), duration of disease (measured in 6-month intervals), and age at onset (in years). Because the four regions within individuals are expected to be correlated with each other, we used mixed (i.e., fixed and random) effects linear models (PROC MIXED of SAS). This model allowed us to test for the heterogeneity of BP according to the region, duration of disease, and age at onset (fixed effects) and to determine the magnitude and statistical significance of the within-individual correlations of the BP in the four regions. Specifically, the BP of the ith region in the kth child is modeled as BPik=β011(i= 2) +β21(i= 3) +β31(i= 4) +β4 duration of disease +β5 age at onset + bk+ cik where variance (bk) =σB2 and variance (cik) =σω2 are the between- and within-individual variances of the BP. The regression coefficients β1, β2, and β3 represent the differences of the BP in the temporal (T), parietal (P), and occipital (O) relative to frontal (F), which is used as the reference category; β4 and β5 represent the effect of 6-months and 1-year increase in the duration of disease and age at onset, respectively. The variance components σB2 and σω2 can be used to quantify the within-individual correlation as σB2/(σB2ω2). To calculate the 95% confidence interval for the correlation, we estimated the standard error based on the equivalence of the Wald test and the likelihood ratio test. Additionally, covariates related to the subjects (gender) and to the sampling procedure (number of slides) were tested in a similar fashion to determine their influence on BP.


  1. Top of page
  2. Abstract
  7. Acknowledgments

Distribution and multifocality of RS lesions

The ability to examine multiple areas of each lobe of the hemisphere allowed us to recognize the heterogeneity and multifocality of pathology within a single individual. Areas with normal architecture and devoid of pathologic changes (stage 0) were adjacent to cortical areas that demonstrated considerable foci of cortical inflammation and destruction (Fig. 2A, B, D, F, and H). Within a given lobe, a wide spectrum of cortical damage was noted, ranging from a normal cortex to degeneration and vacuolation of the cerebral cortex (Fig. 2I and J). In some areas, a gradient of inflammation was found (Fig. 2B; stages 1, 2, and 3 in contiguity), whereas areas in which contiguous areas of the cortex were at distinctly different stages of pathologic involvement also were seen (Fig. 2D, F, H). For instance, an area at stage 0 is contiguous with areas at stages 3 or 4. The distribution of these changes was either discrete in some specimens, and within the laminar framework of the cortex (gyral pattern, Fig. 2A and G), or translaminar and penetrating the cortical layers, producing a wedge-shaped region of injury or “punch out” lesion (Fig. 2C and E).


Figure 2. Patterns of cortical damage and multifocality of cortical lesions in Rasmussen syndrome. A: Panlaminar gyral pattern confined to the top of the gyrus. The transition from stage 1 to stage 2 and 3 is shown in (B). “Punch out” cortical and subcortical lesions involve focal areas of the cerebral cortex (C and E). A sharp transition is seen from stages 0 or 1 to complete degeneration of the cerebral cortex (D and F). Complete degeneration and vacuolation of the cerebral cortex is seen occasionally (G). This pattern is frequently well delimited and surrounded by normal-looking cortex (H). (Bars, 500 μm). I: Representation of large regions of cortical resection in a single patient. Red lines represent slices taken, and representative photomicrographs with scores are at the right, showing the variation of pathology scores within individual gyral areas of the brain. J: Each slide (on red bar) is given the highest score seen in the section, and this is projected onto the brain region involved. K: Histogram constructed from scoring done as in J illustrates the distribution of cortical pathology in the different brain regions (cortical samples) of 45 hemispheres.

Download figure to PowerPoint

Figure 2I and J illustrates the heterogeneity of pathology within each section examined and show the method described earlier in which the final score for each section is denoted by the highest grading. Figure 2I and J, in which each of the major brain regions is shown, demonstrates the complete topographic scoring for the hemisphere. Thus we saw pathology ranging from mild to moderately severe in the temporal and parietal lobes (stages 1–3), from absent to moderately severe (stages 0–3) in the frontal lobe and from absent to mild (stages 0–1) in the occipital lobe. Figure 2K shows the distribution of BP for all 45 hemispheres in which 1,155 cortical samples were analyzed. Variability was found in the sampling of the different brain regions. Fewer samples were obtained from the occipital lobe (mean, 4.31; SD, 2.47) when compared with frontal (mean, 7.36; SD, 4.48), temporal (mean, 6.36; SD, 2.79), and parietal (mean, 7.62; SD, 3.97) lobes. This difference is roughly proportional to volume differences among the different lobes.

Patterns of cortical damage

Two patterns of cortical damage were identified: a gyral pattern characterized by involvement of large areas of the top or sulcal regions of a gyrus (Fig. 2A and G), or a less frequent wedge-shaped or “punch out” lesion (Fig. 2C, E, and F). These patterns of cortical involvement showed an area of marked cortical damage surrounded by normal cerebral cortex or milder stages of inflammation (Fig. 2B, D, F, and H). In a few cases with very prolonged disease, a more uniform and widespread pattern of lesions gave the appearance of a more generalized and diffuse pattern of cortical destruction.

Clinical–pathological correlation

The statistical modeling of differences in the BP by cortical region and the effect of duration of illness and age at onset on BP are shown in Table 4. The analysis of the distribution of cortical abnormalities demonstrated that no statistically significant differences occurred in the distribution of BP in the different brain regions, except the occipital lobe, where the magnitude of pathology was significantly lower (p = 0.032; Table 4). An additional analysis of the other three differences showed no significant differences between temporal and parietal (p = 0.077) or temporal and occipital (p = 0.15) regions. However, the occipital region was significantly lower (p = 0.002) than the parietal region. In summary, the occipital region was significantly lower than frontal and parietal regions. The BPs in the four regions of the brain were significantly correlated, but at the same time, a significant heterogeneity was seen between them (0.37; 95% CI, 0.24–0.51; Table 4,

Table 4. Statistical modeling of burden of pathology (BP) by brain region, duration of disease, and age at onset
Mean level of BP in frontal0.37
 Temporal–frontal−0.027 (p = 0.473)
 Parietal–frontal 0.040 (p = 0.290)
 Occipital–frontal−0.082(p = 0.032)
Per 6-mo duration0.021 (p < 0.001)
Per year of age at onset−0.019(p = 0.011)
Between-individual variance [σB2] 0.019
Within-individual variance [σω2] 0.032
Within-individual correlation between regions [σB2/(σB2ω2)] 0.37 (95% CI: 0.24–0.51)

Fig. 3). The age at onset of symptoms was significantly correlated with the overall BP in the sense that the earlier the onset of illness, the higher the BP (p = 0.011; Table 4). The effect of earlier age at onset on the magnitude of pathology was particularly noted in the occipital lobe, where an earlier age at onset of disease significantly correlated with a higher BP (p = 0.01; Fig. 5). A similar but not statistically significant trend was seen in the parietal lobe (p = 0.05). The effect of disease duration appeared to influence the BP, as a significant association appeared with duration of disease (p ≤ 0.001; Table 4, Fig. 6).


Figure 3. Distribution of the burden of pathology (BP) in the four different lobar regions of 45 hemispheres. The BPs in all regions correlated with each other, but at the same time, heterogeneity was found between them. A significantly lower BP in the occipital lobe (p = 0.032) was noted as compared with the frontal region that was used as reference category.

Download figure to PowerPoint


Figure 5. Distribution of the burden of pathology (BP) as a function of age at onset. A significant statistical correlation was noted between a higher magnitude of pathology (BP) in the occipital lobe and early onset of disease (p = 0.01).

Download figure to PowerPoint


Figure 6. Effect of disease duration on burden of pathology (BP). BP as a function of duration of disease (log10). The BP was significantly associated with longer duration of disease (p < 0.001).

Download figure to PowerPoint

Subcortical and white matter involvement

Although typically the focus of attention has been on the cortical pathology of RS, we also identified a pattern of subcortical and white matter involvement. This pattern of changes occurred in more advanced areas of cortical injury (stages 2 and 3) and where the underlying white matter demonstrated pallor consistent with loss of both myelin and axons (Fig. 4A). Changes involved foci of lymphocytic infiltration (Fig. 4B–D), astrogliosis (Fig. 4B), and microglial reaction (Fig. 4C) at the junction of cerebral cortex and white matter and in deep white matter. Axonal injury in the white matter affected by inflammatory infiltrates was demonstrated immunocytochemically by the presence of axonal swellings and


Figure 4. White matter and subcortical involvement in Rasmussen syndrome. A: Patchy pallor of subcortical and deep white matter was identified in myelin stain (LFB/HE); bar, 400 μm. Perivascular lymphocyte infiltration also is seen in the white matter (insert). The arrow in the insert shows the delimitation between an area of myelin pallor and other normal-looking area. Areas of white matter pallor were associated with foci of lymphocyte accumulation and reactive astrogliosis (GFAP immunostain) (B), microglial activation (HLA-DR immunostain) (C), and T-lymphocyte infiltration composed mostly of CD8+ lymphocytes (D). All of the inflammatory changes in the white matter resembled the histopathologic reactions observed in the cerebral cortex. E: White matter axonal loss and increased frequency of axonal swellings and axonal retraction bulbs were seen in immunocytochemical preparations with antibodies against phosphorylated neurofilaments (SMI 312 antibody; bar in B, C, and D, 25 μm; E, 10 (m). F: The hippocampus showed similar foci of lymphocyte infiltration by T lymphocytes (CD3 immunostain) in the corpus ammonis and (G) dentate gyrus (arrow) (bar in G, 100 μm; H, 25 μm).

Download figure to PowerPoint

axonal “retraction bulbs” (Fig. 4E). In addition, fragments of basal ganglia, mostly putamen, were studied in eight cases in which neuroradiologic evidence of involvement was found. In all these cases, basal ganglia showed marked inflammatory changes and neuronal loss similar to those seen in the cerebral cortex. The hippocampal formation studied in eight cases consistently demonstrated inflammatory changes (Fig. 4F and G).


  1. Top of page
  2. Abstract
  7. Acknowledgments

We performed a comprehensive pathological evaluation of tissues obtained between 1985 and 2002 from 45 hemispherectomies done at Johns Hopkins Hospital as treatment for RS. This study demonstrated that cerebral cortex pathology in RS is multifocal within the hemisphere and progressive, with evidence of an ongoing pathologic process. The nature of the progression appears to be that of an immune-mediated disease that involves both neuroglial and T-lymphocyte responses. The heterogeneity and variability in lesion location, progression of disease, and magnitude of pathologic changes was remarkable within and between patients. The patchy nature of the pathology means that biopsy for diagnosis of RS may be misleading, and partial cortical resections will be not a satisfactory treatment for this devastating disorder.

Stages of cortical pathology in RS

Our ability to examine the complete hemisphere in a large number of patients with RS provides the first clear picture that the destructive process involved is not homogeneous. It is not homogeneous within a given gyrus; it is not homogeneous within a given region of the brain, and it is certainly not homogeneous from region to region. Robitaille and others (6–8,17) recognized that the pathology of RS could be quite variable from patient to patient, but no study has previously been able to examine comprehensively the variability within each patient. This heterogeneity within a given patient suggests that the disease process is ongoing, affecting different parts of the brain at different times.

Our observations support the view that in RS, T lymphocytes initiate an inflammatory process that progresses with the activation of astroglial and microglial cell populations and ultimately produces neuronal and cortical injury. The early morphologic findings observed in the cerebral cortex of RS patients (stage 1), infiltration by T lymphocytes and activation of astrocytes and microglia with little evidence of neuronal damage, suggests the critical role of immunologic mechanisms in the pathogenesis of this disorder. The overall neuronal cell population in early stages of RS appears morphologically healthy, except for foci of satellitosis, in which they are surrounded by T lymphocytes and activated microglial and astroglial cells. The observations of different stages of cortical involvement indicate that in RS, a continuum of cortical injury ends with cavitation and vacuolation of the cerebral cortex (stage 4); i.e., that neuronal loss progresses because of an ongoing injury mediated by T-lymphocyte and neuroglial activation. These morphologic observations expose the role of immune cellular and neuroglial pathogenic mechanisms in RS. In addition, the presence of morphologic changes such as focal cortical spongiosis observed in late stages of cortical pathology, a change that resembles the morphologic observations described for excitotoxic cortical injury (18), may suggest that in addition to immunologic injury, other neurotoxic factors may play roles in neuronal and cortical damage. Focal spongiosis was not observed in early cortical pathology, a finding that supports the view that its presence is associated with secondary effects that occur in the cortex after the initial lymphocyte and neuroglial reactions.

Distribution and multifocality of RS lesions

The topographic analysis of the distribution of lesions demonstrates that in RS, at a single point in time, as is well demonstrated in this cross-sectional sampling, we are able to discern all of these stages of injury in a single region of the hemisphere, sometimes even within a single gyrus. We are able to demonstrate that all brain regions of the hemisphere are affected and the BP of each region is significantly correlated, albeit in a heterogeneous fashion. The involvement of all lobar regions of the brain and the patchy distribution of lesions demonstrate that cortical pathology in RS represents a multifocal process rather than a centrifugal pattern, as was suggested earlier (3,5,6). The involvement of the white matter, with focal or diffuse infiltration by T lymphocytes, microgliosis, and astrogliosis similar to those changes seen in the cerebral cortex, suggests that the immunologic reactions are not strictly restricted to the cerebral cortex. We would infer from this that an ongoing and either episodic or continuous process is present, as one would expect from the clinical progression of the disease. An inciting factor causes the immune mechanism to attack regions of the brain over a variable period. The multifocal pattern and profile of pathologic changes in RS is more compatible with an immunologically mediated process that produces damage at multiple times and in various sites rather than a monophasic disorder such as a viral encephalitis. This multifocality and heterogeneity of lesions resembles those seen in other immunologically mediated disorders such as multiple sclerosis and postinfectious encephalomyelitis (19–23), in which patchy and multifocal lesions also are found.

Pathogenic implications of cortical pathology

The factors that trigger the immunopathogenic process in RS remain unknown. Specifically, we do not know the precipitating cause for T-lymphocyte infiltration and neuroglial activation that leads to neuronal injury and death. No consistent evidence is seen of viral infection of the neuron or neuroglial cell populations, as shown by multiple studies (24–27). No evidence exists of cytopathic neuronal changes similar to those induced by seizures (28–30) in areas undergoing early stages of inflammation; thus it is unlikely that cellular damage produced by seizure activity is the primary event that triggers the inflammatory cascade. Although focal cortical spongiosis, a change suggestive of excitotoxicity, is seen in some cases, this change is seen mostly in advanced stages of cortical damage (stage 3), an observation that suggest that this abnormality is the result of secondary effects after an immunologically mediated injury is already established.

Our morphologic observations support the hypothesis that cellular-mediated mechanisms are the primary factors that contribute to the neuronal injury in RS. The possible role of T-lymphocyte cytotoxicity has been entertained by different studies (7,8,31) and supported by our preliminary work (32), as well recently published studies (33). Previous studies of T-cell receptor (TCR) expression in RS tissues demonstrated the presence of restricted T-lymphocyte populations that likely expanded locally from few precursors in response to specific antigenic epitopes, a finding that suggests that a specific immune response is triggered in RS (31). The possibility that local factors affecting one specific hemisphere trigger these specific immune responses is supported by the evidence that patients do not deteriorate once the “inciting” areas are removed (10,11,34).

Clinicopathological observations

Three important observations derived from the clinicopathological assessment in our study can be delineated. The first concerns the variability and multifocality of lesions in RS. The variability of lesions, even in adjacent areas of cortex, would suggest that a biopsy, usually performed in a noneloquent cortical region (frontal or temporal), could be misleading. This recognizes the fact that 60% of the slides from these regions were stage 0 or 1. Having this “false negative” can lead to delays in treatment, especially definitive surgery. The homogeneity in the magnitude of BP in the different region of the brain and its close correlation support the view of a widespread distribution of pathologic changes within the hemisphere of RS patients. This finding also validates the need to remove or disconnect the entire hemisphere. Not a single case was found in which absolutely no pathology was seen (i.e., consistent scores of 0) throughout a given region of brain. Even in the occipital lobe where the pathology scores tended to be lower, a mild degree of inflammation was invariably present, suggesting that over time, this region also would be destroyed.

The second observation and perhaps the most intriguing is the effect of the age at onset on the BP. The age at onset of symptoms is significantly correlated with the BP in the sense that the earlier the onset of illness, the higher the BP. Interestingly, the effect of age at onset significantly influences the magnitude of pathology in the occipital lobe, a finding that may suggest that occipital changes at the beginning of disease in young patients with RS predict a higher BP and aggressive course of the disease. These findings suggest that the factor(s) that trigger inflammatory changes in the cerebral cortex of patients with RS may produce greater inflammatory responses in the brain of younger patients. This effect may be the result of the relative immaturity of the CNS and/or inability of the young brain to modulate inflammatory responses. A similar observation about the effect of age at onset and severity of disease was reported recently in a clinical–neuroradiological study of 13 patients with RS (35).

The third observation concerns the effect of disease duration on the BP. It is very evident that the duration of disease produces a significant impact in the magnitude of pathologic changes in RS. This finding reinforces the view that more aggressive and earlier therapeutic approaches are required in the treatment of this condition and supports the concept of an early removal or disconnection of the affected hemisphere to limit the disabling effects of seizure activity if other treatments fail.


  1. Top of page
  2. Abstract
  7. Acknowledgments

Our neuropathological study of 45 hemispherectomies in RS demonstrates the multifocality and heterogeneity of pathologic changes in this disorder, findings that resemble those present in autoimmune disorders of the CNS such as multiple sclerosis. The presence of T-lymphocyte and neuroglial reactions as the major inflammatory responses in RS suggest important avenues for exploration of potential immunopathogenic mechanisms for RS that may form the basis of future immunomodulatory treatments for this devastating disease of childhood.


  1. Top of page
  2. Abstract
  7. Acknowledgments

Acknowledgment:  Dr. Carlos A. Pardo was supported by a Passano Physician Scientist Award and Clinician Scientist Award from Johns Hopkins University School of Medicine. This study was supported in part by a grant from the Epilepsy Foundation of America (to C.A.P.) and from the Sargent Family Fund. We thank Dr. Alvaro Muñoz for providing guidance on epidemiologic and biostatistical analysis and Drs. David Newman-Toker and Sanjay Keswani for critically reading the manuscript.


  1. Top of page
  2. Abstract
  7. Acknowledgments
  • 1
    Rasmussen T, Olszewski J, Lloyd-Smith D. Focal seizures due to chronic localized encephalitis. Neurology 1958;8: 43545.
  • 2
    Rasmussen T, McCann W. Clinical studies of patients with focal epilepsy due to “chronic encephalitis.” Trans Am Neurol Assoc 1968;93: 8994.
  • 3
    Oguni H, Andermann F, Rasmussen T. The natural history of the chronic encephalitis and epilepsy: a study of the MNI series of forty-eight cases. In: AndermannF, ed. Chronic encephalitis and epilepsy: Rasmussen's syndrome. Boston : Butterworth-Heinemann, 1991: 736.
  • 4
    Vining EP, Freeman JM, Brandt J, et al. Progressive unilateral encephalopathy of childhood (Rasmussen's syndrome): a reappraisal. Epilepsia 1993;34: 63950.
  • 5
    Oguni H, Andermann F, Rasmussen TB. The syndrome of chronic encephalitis and epilepsy: a study based on the MNI series of 48 cases. Adv Neurol 1992;57: 41933.
  • 6
    Robitaille Y. Neuropathological aspects of chronic encephalitis. In: AndermannF, ed. Chronic encephalitis and epilepsy: Rasmussen's syndrome. Boston : Butterworth-Heinemann, 1991: 79110.
  • 7
    Honavar M, Janota I, Polkey CE. Rasmussen's encephalitis in surgery for epilepsy. Dev Med Child Neurol 1992;34: 314.
  • 8
    Vining E. Rasmussen's syndrome. In: KotagalP, LudersH, eds. The epilepsies: etiologies and prevention. San Diego : Academic Press, 1999: 2838.
  • 9
    Vining EP, Freeman JM, Pillas DJ, et al. Why would you remove half a brain? The outcome of 58 children after hemispherectomy: the Johns Hopkins experience: 1968 to 1996. Pediatrics 1997;100: 16371.
  • 10
    Farrell MA, Droogan O, Secor DL, et al. Chronic encephalitis associated with epilepsy: immunohistochemical and ultrastructural studies. Acta Neuropathol (Berl) 1995;89: 31321.
  • 11
    Carson BS, Javedan SP, Freeman JM, et al. Hemispherectomy: a hemidecortication approach and review of 52 cases. J Neurosurg 1996;84: 90311.
  • 12
    Boatman D, Freeman J, Vining E, et al. Language recovery after left hemispherectomy in children with late-onset seizures. Ann Neurol 1999;46: 57986.
  • 13
    Furuta A, Price DL, Pardo CA, et al. Localization of superoxide dismutases in Alzheimer's disease and Down's syndrome neocortex and hippocampus. Am J Pathol 1995;146: 35767.
  • 14
    Graeber MB, Streit WJ, Kreutzberg GW. The third glial cell type: the microglia: cellular markers of activation in situ. Acta Histochem Suppl 1990;38: 15760.
  • 15
    Petito CK, Morgello S, Felix JC, et al. The two patterns of reactive astrocytosis in postischemic rat brain. J Cereb Blood Flow Metab 1990;10: 8509.
  • 16
    Gehrmann J, Bonnekoh P, Miyazawa T, et al. Immunocytochemical study of an early microglial activation in ischemia. J Cereb Blood Flow Metab 1992;12: 25769.
  • 17
    Bien CG, Urbach H, Deckert M, et al. Diagnosis and staging of Rasmussen's encephalitis by serial MRI and histopathology. Neurology 2002;58: 2507.
  • 18
    Olney JW. Excitotoxicity: an overview. Can Dis Wkly Rep 1990;16(suppl 1E):4757.
  • 19
    Lassmann H. The pathology of multiple sclerosis and its evolution. Trans R Soc Lond B Biol Sci 1999;354: 163540.
  • 20
    Lucchinetti C, Bruck W, Parisi J, et al. Heterogeneity of multiple sclerosis lesions: implications for the pathogenesis of demyelination. Ann Neurol 2000;47: 70717.
  • 21
    Dale RC, De Sousa C, Chong WK, et al. Acute disseminated encephalomyelitis, multiphasic disseminated encephalomyelitis and multiple sclerosis in children. Brain 2000;123: 240722.
  • 22
    Lassmann H. Classification of demyelinating diseases at the interface between etiology and pathogenesis. Curr Opin Neurol 2001;14: 2538.
  • 23
    Lassmann H, Bruck W, Lucchinetti C. Heterogeneity of multiple sclerosis pathogenesis: implications for diagnosis and therapy. Trends Mol Med 2001;7: 11521.
  • 24
    Asher DM, Gajdusek DC. Virologic studies in chronic encephalitis. In: AndermannF, ed. Chronic encephalitis and epilepsy: Rasmussen's syndrome. Boston : Butterworth-Heinemann, 1991: 14758.
  • 25
    Vinters HV, Wang R, Wiley CA. Herpesviruses in chronic encephalitis associated with intractable childhood epilepsy. Hum Pathol 1993;24: 8719.
  • 26
    Atkins MR, Terrell W, Hulette CM. Rasmussen's syndrome: a study of potential viral etiology. Clin Neuropathol 1995;14: 712.
  • 27
    Park SH, Vinters HV. Ultrastructural study of Rasmussen encephalitis. Ultrastruct Pathol 2002;26: 28792.
  • 28
    Corsellis JA, Bruton CJ. Neuropathology of status epilepticus in humans. Adv Neurol 1983;34: 12939.
  • 29
    Pollard H, Cantagrel S, Charriaut-Marlangue C, et al. Apoptosis associated DNA fragmentation in epileptic brain damage. Neuroreport 1994;5: 10535.
  • 30
    Kubova H, Druga R, Lukasiuk K, et al. Status epilepticus causes necrotic damage in the mediodorsal nucleus of the thalamus in immature rats. J Neurosci 2001;21: 35939.
  • 31
    Li Y, Uccelli A, Laxer KD, et al. Local-clonal expansion of infiltrating T lymphocytes in chronic encephalitis of Rasmussen. J Immunol 1997;158: 142837.
  • 32
    Pardo CA, Arroyo S, Vining E, et al. Neuronal injury in Rasmussen's chronic encephalitis is mediated by cytotoxic T-cells. Epilepsia 1994;35(suppl 8):89.
  • 33
    Bien CG, Bauer J, Deckwerth TL, et al. Destruction of neurons by cytotoxic T cells: a new pathogenic mechanism in Rasmussen's encephalitis. Ann Neurol 2002;51: 3118.
  • 34
    Villemure JG, Andermann F, Rasmussen T. Hemispherectomy for the treatment of epilepsy due to chronic encephalitis. In: AndermannF, ed. Chronic encephalitis and epilepsy: Rasmussen's syndrome. Boston : Butterworth-Heinemann, 1991: 23544.
  • 35
    Bien CG, Widman G, Urbach H, et al. The natural history of Rasmussen's encephalitis. Brain 2002;125: 17519.