Hippocampal sclerosis in epilepsy and childhood febrile seizures

6 Bone marrow transplantation in thalassemia: the Pesaro experience. Bone Marrow Transplant 1993; 12 (suppl 1): S56-73. Muretto P, Del Fiasco S, Angelucci E, De Rosa F, Lucarelli G. Bone marrow transplantation in thalassemia: modifications of hepatic iron overload and associated lesions after long time engrafting. Liver (in

7

press). 8

International Committee for Standardization in Haematology. The measurement of total and unsaturated iron-binding capacity in serum. Br J Haematol 1978; 38: 281-87.

9

Borgna-Pignatti C, Castriota-Scanderberg A. Methods for evaluating iron stores and efficacy of chelation in transfusional haemosiderosis. Haematologica 1991; 76: 409-13.

10 Brittenham GM, Cohen AR, McLaren CE, et al. Hepatic iron stores and plasma ferritin concentration in patients with sickle cell anemia and thalassemia. Am J Hematol 1993; 42: 81-85. 11 Overmoyer BA, McLaren CE, Brittenham GM. Uniformity of liver density and nonheme (storage) iron distribution. Arch Pathol Lab Med 1987; 111: 549-54. 12 De Virgiliis S, Sanna G, Cornacchia G, et al. Serum ferritin, liver iron stores and liver histology in children with thalassemia. Arch Dis Child 1980; 55: 43-45. 13 Bronspiegel-Weintrob N, Olivieri NF, Tyler BJ, Andrews DF, Freedman MH, Holland FJ. Effect of age at the start of iron chelation therapy in gonadal function in beta-thalassemia major. N Engl J Med 1990; 323: 713-19.

Hippocampal sclerosis in epilepsy and childhood febrile seizures

Summary

Introduction

The connection between hippocampal sclerosis and childhood febrile seizures (CFS) is a contentious issue in the study of epilepsy. We investigated 107 patients with drug-resistant

Childhood febrile seizures (CFS) occur in 2-5% of the population, and are associated with an increase in the risk of subsequent nonfebrile seizures, especially of temporal lobe origin.l-6 An association between CFS and hippocampal sclerosis has been noted/,8but the reason is unknown. Previous studies have been restricted to histological examination of operative or necropsy material. Operative specimens only provide information from the site of surgery, while necropsy evidence may represent only

epilepsy by high-resolution volumetric magnetic resonance imaging (MRI). 20 had a history of CFS, 45 had focal (26) or diffuse (19) hippocampal volume loss (HVL). The frequency of CFS was significantly (p<0·001) higher in the patients with HVL, especially of the diffuse pattern, compared to other epileptic patients without HVL and to the general population. Furthermore, the severity of HVL was greatest in those with a history of CFS. No other clinical or demographic features were associated with either a history of CFS or HVL. The frequency of CFS in patients with other structural congenital causes of epilepsy did not differ from that in a general population. Although these findings show that hippocampal sclerosis is strongly associated with a history of CFS, they do not indicate whether this is a causal relationship. If CFS do cause some cases of hippocampal sclerosis, this can not be the only mechanism, as 64% of those with HVL gave no history of CFS. As diffuse HVL is more strongly associated with a history of CFS than focal HVL, it is also possible that CFS convert pre-existing congenital focal abnormalities into diffuse hippocampal sclerosis. Given the possibility that CFS may cause hippocampal damage and epilepsy, they require urgent medical intervention.

Lancet 1993; 342: 1391-94

chronic,

severe,

or

complex

cases.

Furthermore,

sclerosis may also be related

to disturbed hippocampal or embryogenesis,9 perinatal postnatal trauma,10,11repeated

seizures,12,13 or other intracranial lesions." volumetric High-resolution magnetic-resonance the imaging (MRI) provides opportunity to study in ViVO.15,16 Asymmetric hippocampal morphology volume loss (HVL) is correlated with hippocampal histologically-proven sclerosis.16,17 The distribution of volume loss within the hippocampus can be imaged and divided into focal or diffuse patterns .15, I’ Diffuse hippocampal atrophy involves relatively uniform loss of cross sectional area along the entire length of the hippocampus. In focal hippocampal atrophy the bulk of the volume loss is clearly restricted to one region of the hippocampus, usually the anterior half (figure 1). We used this technique: (a) to determine the frequency of a history of CFS in patients with HVL, and to what extent this differed in patients with diffuse and focal volume loss; (b) to identify other factors which may be associated with the MRI finding of HVL or with a history of CFS; (c) to compare the frequency of CFS in other epileptic patients without HVL; and (d) to study the relationship between the extent of volume loss and the occurrence of CFS. These issues are important to one of the most contentious propositions in the study of epilepsy: that childhood febrile seizures damage the brain and cause hippocampal sclerosis, resulting in the subsequent development of chronic temporal lobe

epilepsy.7 National Hospital for Neurology and Neurosurgery, London (M J Cook FRACP, D R Fish MRCP, S D Shorvon FRCP); St Mary’s Hospital, London, UK (J M Stevens FRACR); and Department of Neurology, State

University Hospital, Groningen, Netherlands (J B M Kuks MD) Correspondence to: Dr D R Fish, National Hospital for Neurology and Neurosurgery, Queen Square, London WC1N 3BG, UK

Patients and methods We investigated 107 patients with medically-intractable epilepsy who had normal computerised tomograms with a specific volumetric MRI scanning protocolls between 1990 and 1992. The protocol included a volumetric sequence in the coronal plane which

1391

childhood were recorded (questioning of the patient and/or family about CFS is part of our routine evaluation). A CFS was defined as a seizure occurring before the age of 5 years during a non-cerebral febrile illness in patients without previous seizures. We excluded from analysis a further 34 patients: 14 because of insufficient clinical information, 6 because of an acquired neocortical abnormality (3 gliomas, 3 traumatic lesions), and 14 who had congenital vascular lesions (eg, cavernomas or large arterio-venous malformations which might have bled). Volume measures showed only one with asymmetric HVL (a patient with a large vascular lesion which directly involved the mesial temporal structures), and none of the 34 had a known history of CFS.

Statistics and the Kruskal-Wallis one-way analysis of variance were used to compare categorical and continuous data between different groups, and the differences in categorical data were then tested with a X2 and Fischer’s exact test where appropriate. The relationship between hippocampal volume ratios in different patient groups were analysed with Spearman’s nonparametric rank correlation. The incidence of CFS in the different patients groups were also compared to the normal population (with the incidence of febrile seizures estimated to be 4%) with the binomial

Global X2

test.

Results

Slice numoer

-

-

Right hippocampus

-B-

Left hippocampus

Figure 1: Examples of hippocampal surface-area plots used to calculate total hippocampal volumes Vertical axis: cross sectional area of the left and right hippocampus at successive 1 5 mm intervals along its length. The MRI-slice numbers are shown on the horizontal axis. MRI-slice number 1 is at the posterior border of the hippocampus, and each subsequent slice is 1 5 mm anterior to its predecessor. (a) normal pattern; (b) focal anterior HVL; (c) diffuse HVL. The overall ratio of HVL (lesser:greater) for all MRI-slices were (a) 0 97, (b)0-54, (c)0 72.

allows the reconstruction of 1-5 mm contiguous coronal slices by a spoiled-gradient echo technique with a 35/5/1 (TR/TE/NEX) pulse sequence, flip angle 35 degrees, matrix size 256 x 128. The images were magnified 3 times, the grey-white interface enhanced by computerised thresholding, the interface outlined manually with a tracker-ball, and the area of the hippocampus in each of the 20 or so slices through the hippocampus calculated automatically by pixel counting, and the area plotted against the slice location (figure 1). The techniques have been previously described in detail. 15,18 Hippocampal volume measurements were used to define a hippocampal volume ratio (lesser:greater) for each patient. We have shown that these are between 0-96 and 1-00 in normal Patients with controls.8 HVL was defined as a ratio of <0-95. bilateral HVL may be difficult to detect but will usually have an abnormal ratio as even bilateral hippocampal sclerosis is usually asymmetric.7,10,16,21 Patients were divided into three groups on the basis of MRI scanning (table): 1: Patients with no MRI abnormalities (25). 2: Patients with small focal lesions which were presumed to have been present from birth and which do not overlay or encroach on the hippocampus (37: 20 cortical dysplasia, 10 developmental tumours, 1 tuberous sclerosis, 2 porencephalic cysts, 3 diffuse white matter lesions, 1 small arterio-venous malformation). 3: Patients with HVL (45). This group was subdivided into: 3a patients with focal HVL (26; 23 anterior, 3 posterior) and 3b patients with diffuse HVL (19). Patients’ records were analysed to see if febrile seizures in

1392

patients had HVL;

focal and 19 with a diffuse pattern, and 20 had a history of CFS. The most important finding of the study was of a clear association between a history of CFS and HVL, especially diffuse HVL: 16 of the 20 with a history of CFS had HVL. The frequency of CFS in patients with HVL (16/45; 36%) was significantly different (p < 0-0001) from the 4% frequency in a normal population. A history of CFS was found in 3/26 (12%) of those with focal HVL and 13/19 (68%) of those with diffuse HVL (proportions different from the general population at significance levels of p < 008 and p < 0-001, respectively). In addition, 3 of the other 6 with diffuse HVL had

45

26 with

a

Kruskal-Wallis one-way analysis of variance to compare continuous variables, and X’ noncontinuous vanables. *=p<0 0 001 (value compared to the normal population [binomial] test) p < 0 005 (value compared between groups by global X’-square test). No other comparisons reached significance levels greater than p < 0 01. t = median values with inter-quartile range in brackets. HVL = hippocampal volume loss. CFS = childhood febrile seizure.

Table : Clinical and

findings

demographic features related to MRI

Discussion

Figure 2: Hippocampal volume ratios (lesser:greater) in 3 groups Ratio of 96% or greater is taken as normal, and the number above the dotted line is the number of patients inside this normal range. For patients outside the normal range, the degree of volume loss is given on the vertical axis, with each dot representing one patient. A= 20 patients with a history of a childhood febrile seizure (CFS). B= 18 patients with a history of non-febrile seizures. C 69 patients without a history of childhood seizures. =

childhood seizures (before the age of 5 years) but a relationship with a febrile illness was uncertain and these were therefore considered for the purposes of this study as "non-febrile" fits; one other patient had a severe bout of malaria in childhood but without recognised seizures. The frequencies of CFS in the patients without HVL (1 /25 [4%]] and 3/37 [8%]) in groups 1 and 2 respectively) were not significantly different from that in a normal population, nor from each other. To study the effect of a history of CFS on the severity of HVL, a comparison was made of the degree of HVL in the patients with a history of CFS (n = 20), with a history of non-febrile seizures in childhood (n= 18), and those with no history of childhood seizures (n = 69) (figure 2). The most severe loss occurred in those with a history of CFS We used a narrow normal range of (p<0-0001). hippocampal volume ratios ( > 096), but if the range is extended (eg, to > 0-90) an even greater difference between those with and without a history of CFS would be shown. 38 of 45 in group 3a and 3b had partial seizures of temporal lobe type; however, 2 had predominantly generalised convulsions, and 5 had seizures suggestive of extratemporal involvement. Only 3 gave a clear history of unilateral febrile seizures (although the proportion might have been higher in a prospective study), and all 3 had smaller hippocampal volumes on the side contralateral to the spasms. Analysis of subgroups 3a and 3b separately showed no relationship between the degree of HVL (expressed either as absolute hippocampal volume or hippocampal volume asymmetry) and age of CFS (although the first CFS occurred before the age of 18 months in 72%), number of CFS (60% of the patients with CFS had only a single CFS), history of status epilepticus, duration of epilepsy, or estimated life-time number of generalised seizures. No notable differences between groups were found in other clinical factors, although there was a slight excess of patients with a history of complicated pregnancy or birth injury in group 2, the age at onset of epilepsy (non-febrile seizures) was somewhat lower in group 3b than in the other groups, and patients with focal HVL had somewhat fewer secondary generalised seizures (p < 0-025) than other groups

(table).

Volumetric MRI provides, for the first time in vivo, an accurate method for measuring the volume of the hippocampus and thus detecting hippocampal sclerosis which is the most common underlying pathology of complex partial seizures. Our study set out to determine the relationship between childhood febrile seizures (CFS) and hippocampal sclerosis (as indicated by HVL). The main finding was that in patients with chronic partial epilepsy a history of preceding CFS is often associated with characteristic loss of hippocampal volume. This study differs from previous work in that a non-invasive in-vivo method was used instead of the pathological examination of surgical or necropsy tissue to detect hippocampal atrophy. This method also allows the investigation of hippocampal volumes in normal individuals and patients with other cerebral abnormalities. In the majority of patients with CFS, the pattern was of diffuse (as opposed to focal) HVL. 13 out of 19 patients with diffuse HVL had a clear history of CFS and a further 4 a history that may have accounted for these changes. Thus, in only 2 out of 19 were these changes totally unexplained. Furthermore, a history of CFS was clearly associated with the most severe HVL.

Diffuse HVL was not associated with other clinical factors which are said to be associated with hippocampal damage: age at the first CFS, duration of epilepsy, estimated lifetime number of generalised seizures, or a history of status epilepticus. We did not find any association between either a history of CFS or of HVL with any other clinical factors, except that the age of onset of non-febrile epilepsy was somewhat lower in diffuse HVL than in other groups, and patients with diffuse HVL were slightly more likely than those with focal HVL to suffer generalised convulsions. It is interesting to note that non-febrile childhood seizures were not associated with hippocampal volume loss; the reason for this is unclear. It is possible that fever in itself has an influence, or that the febrile seizures occurred earlier in life or were more prolonged. It is worth noting, however, that the febrile seizures in this population of epileptic patients occurred at a younger median age than febrile convulsions in the general population. As the incidence of CFS in the normal population is 4% it is not surprising to find some patients (4/62, 6%) with CFS in the other categories. The normal hippocampal volumes in these patients indicate that not all CFS cause, or are due to, HVL detectable with volumetric MRI. There are three possible explanations for the strong relationship between diffuse HVL and CFS. First, it is possible that patients with chronic intractable epilepsy are born with diffuse hippocampal sclerosis and are therefore at risk of having a CFS. Second, that patients are born with focal hippocampal abnormalities, which are then converted into diffuse hippocampal sclerosis by the febrile seizure. Third, that patients are born with normal brains and the CFS causes diffuse HVL. The second possibility is of particular interest, supported by the difference between diffuse HVL in those with a history of CFS compared to the more frequent occurrence of focal HVL in those without. If a causal relationship between CFS and hippocampal sclerosis is proposed, however, other factors must be involved, as a history of CFS was found in only a minority of those with HVL (16/45; 36%). Both the second and third possibilities are important from a practical as well as theoretical standpoint, as there is an obvious potential for prevention if HVL can be 1393

shown to be the result of a CFS. As long as there is a possibility that febrile seizures in childhood cause hippocampal damage, they should be urgently and immediately treated to avoid (further) hippocampal damage and epilepsy in the future.

11

12

We thank The Brain Research Trust and The Sir Jules Thorn Charitable Trust for support for our research, and to Dr W van den Burg for statistical advice.

13

References

14

1 2

Annegers JF, Hauser WA, Elveback LR, Kurland LT. The risk of epilepsy following febrile convulsions. Neurology 1979; 29: 297-303. Nelson KB, Ellenberg JH. Predictors of epilepsy in children who have experienced febrile seizures. N Engl J Med 1976; 295: 1029-33.

Rasmussen T. Relative significance of isolated infantile convulsions as a primary cause of focal epilepsy. Epilepsia 1979; 20: 395-401. 4 Verity CM, Golding J. Risk of epilepsy after febrile convulsions: a national cohort study. BMJ 1991; 303: 1373-76. 5 Verity CM, Ross EM, Golding J. Outcome of childhood status epilepticus and lengthy febrile convulsions: finding of a national cohort study. BMJ 1993; 307: 225-28. 6 Hauser WA. (1981). The natural history of febrile seizures. In: Febrile Seizures, eds. Nelson KB, Ellenberg JH, pp 5-17. New York: Raven Press. 7 Falconer MA. Genetic and related aetiological factors in temporal lobe epilepsy. A review. Epilepsia 1971; 12: 13-31. 8 Bruton CJ. The neuropathology of Temporal Lobe Epilepsy. Oxford University Press. 1988 pp 1-155. 9 Scheibel AB. Are complex partial seizures a sequela of temporal lobe dysgenesis? In: Smith D, Treiman D, Trimble M. Advances in Neurology. Raven Press New York. 1991 pp 59-77. 10 Margerison JH, Corsellis JAN. Epilepsy and the temporal lobes: a clinical, electroencephalographic, and neuropathological study of the 3

toxicity in children with acute promyelocytic leukaemia

Tretinoin

Tretinoin is effective in acute promyelocytic leukaemia in adults. Data about its efficacy and safety in children are limited. We have treated 9 children with tretinoin at 45 mg/m2 per day. Pseudotumour cerebri or hyperleucocytosis occurred in 5 patients. Retinoic acid syndrome was seen in 3 cases. 1 of 2 children who developed hyperleucocytosis, pseudotumour cerebri, and retinoic acid syndrome died despite steroids and mechanical ventilation. Complete remissions with tretinoin alone were achieved in 5 patients. All 8 surviving children received consolidation chemotherapy. Our experience with tretinoin therapy suggests that toxicity is frequent in children. Lancet 1993; 342: 1394-95

Acute promyelocytic leukaemia (APL) is characterised by a reciprocal translocation between chromosome 15 and 17 that results in juxtaposition of the retinoic acid receptor alpha (RAR-a) gene on chromosome 17 and the promyelocytic leukaemia (PML) gene on chromosome 15 plus the synthesis of a unique fusion messenger RNA, P ML/RAR -rx.1 Clinically, this re-arrangement is associated with sensitivity to tretinoin; 50-80% of patients thus treated achieve complete remission.2--4 An advantage of 1394

15

16

17

18

19

20 21

brain with particular reference to the temporal lobes. Brain 1966; 89: 499-530. Gastaut H, Toga M, Roger J, Gibson WC. A correlation of clinical, electroencephalographic and anatomical findings in nine autopsied cases of ’temporal lobe epilepsy’. Epilepsia 1959; 1: 56-85. Aicardi J, Chevrie JJ. Consequences of status epilepticus in infants and children. In: Delgado, Escueta AV, Wasterlain CG, Treiman DM. Advances in Neurology. Raven Press New York. 1983 pp 115-28. Sagar HJ, Oxbury JM. Hippocampal neuron loss in temporal lobe epilepsy: Correlation with early childhood convulsions. Ann Neurol 1987; 22: 334-40. Fish DR, Anderman F, Oliver A. Complex partial seizures and small posterior temporal or extratemporal structural lesions: Surgical management. Neurology 1991; 41: 1781-84. Cook MJ, Cendes F, Anderman F, Free SL, Fish DR, Shorvon SD, Stevens JM. Hippocampal volumetric studies in extratemporal epilepsies: 50 cases. Epilepsia 1992; 33: 72. Lencz T, McCarthy G, Bronen RA, et al. Quantitative magnetic resonance imaging in temporal lobe epilepsy: Relationship to neuropathology and neuropsychological function. Ann Neurol 1992; 31: 629-37. Jack CR, Sharbrough FW, Cascino GD, Hirschorn KA, O’Brien PC, Marsh WR. Magnetic resonance image-based hippocampal volumetry: Correlation with outcome after temporal lobectomy. Ann Neurol 1992; 31: 138-46. Cook MJ, Fish DR, Shorvon SD, Straughan K, Stevens JM. Hippocampal volumetric and morphometric studies in frontal and temporal lobe epilepsy. Brain 1992; 115: 1001-05. Babb TL, Brown WJ. Pathological findings in epilepsy. In: Engel J. Surgical treatment of the epilepsies. Raven Press New York. 1987 pp 511-40. Mouritzen-Dam A. Hippocampal neuron loss in epilepsy and after experimental seizures. Acta Neurol Scand 1982; 66: 601-42. Babb TL, Lieb JP, Brown WJ, Pretorius J, Crandall PH. Distribution of Pyramidal cell density and hyperexcitability in the epileptic human hippocampal formation. Epilepsia 1984; 25: 721-28.

tretinoin is avoidance of APL-associated coagulopathy. However, retinoic acid at 45 mg/m2 daily has side-effects.4-6 The retinoic acid syndrome, a rare complication of hypervitaminosis A, has been reported mainly in adult patients with acute leukaemiasand in adults with solid tumours or myelodysplastic syndrome treated with conventional doses of tretinoin.8,9 In children, the maximum tolerated dose of tretinoin has been reported to be 60 mg/m2 per day orally.s This dose in children is lower than that in adults (150 mg/m2 daily).9 We report toxicities that occurred in 8 of 9 children (mean age 12 years, range 1 5-175;6 girls) with APL (de novo, 7; in first relapse and > 3 years off-therapy, 2) treated with tretinoin at 45 mg/m2 per day (as single dose, 7; two daily doses, 2). All had translocation (15; 17)(q22; q21). Mean white cell count was 4-0 x 109/L (0-5-17-0 x 109/L). Platelet count at diagnosis ranged from 8 to 62 x 109/L (mean 35 x 109/L). Coagulation profile revealed normal ranges of prothrombin, partial thromboplastin, and thrombin times. Fibrin split products were slightly increased in 4 and moderately increased in 2 patients. Fibrinogen ranged from 114 to 220 mg/dL (146 mg/dL). None of the 9 children had clinically significant coagulopathy; 2 had a little oozing at venepuncture sites before tretinoin therapy. All 9 required platelet transfusions during therapy. Pseudotumour cerebri (severe headaches, nausea and vomiting, papilloedema, retinal haemorrhages, or visual changes [intermittent loss of vision, 1;ophthalmoplegia, 2]) occurred in 5 patients, 2 of whom had hyperleucocytosis (white cells > 20 x 109/L) (table). 2 other patients had moderate/severe headaches with vomiting, but without ophthalmic findings suggestive of increased intracranial pressure. Retinoic acid syndrome was seen in 3 cases, all of whom had hyperleucocytosis. 2 patients developed