Silent Celiac Disease in Patients with Childhood Localization-Related Epilepsies


Address correspondence and reprint requests to Dr. A. Gambardella at Cattedra ed U.O. di Neurologia, Università degli studi “Magna Graecia,” Policlinico Mater Domini, Via Tommaso Campanella, 88100 Catanzaro, Italy. E-mail: Dr. R.L. Oliveri is deceased.


Summary:  Purpose: To evaluate how many patients with a clinical picture of idiopathic childhood localization-related epilepsies may also have silent celiac disease (CD). This will help determine whether investigation for CD should be restricted to those patients with childhood partial epilepsy with occipital paroxysms (CPEO) or should be extended to all patients with childhood partial epilepsy (CPE) regardless of seizure type and electroencephalographic (EEG) paroxysms.

Methods: The study group consisted of 72 patients (31 girls and 41 boys; mean age, 12.6 ± 4.28 years; age at onset, 6.4 ± 3.7 years) who were observed consecutively over a 5-year period and who received an initial diagnosis of idiopathic CPE. A diagnosis of CD was confirmed by using enzyme-linked immunosorbent assay (ELISA) to assess the presence of antigliadin antibodies and the immunofluorescent undirected test to assess the presence of antiendomysium antibodies.

Results: Twenty-five patients had CPEO, whereas the remaining 47 had CPE with centrotemporal spikes (CPEC). None of the patients with CPEC had positive antibody tests. Of the 25 patients with CPEO, two (8%) had antiendomysium immunoglobulin (Ig) A antibodies. In both of these patients, the jejunal biopsy showed atrophy of the villi and hyperplasia of the crypts, consistent with a diagnosis of CD. Brain computed tomography (CT) was normal in one of these patients and revealed occipital corticosubcortical calcifications in the other.

Conclusions: Our study indicates that CD screening should be performed routinely only in patients with CPEO.

Celiac disease (CD) is a common enteropathy characterized by malabsorption, abnormal small-bowel mucosa, and permanent intolerance to gluten protein (1). Presenting symptoms vary considerably and depend greatly on the patient's age at onset. Gastrointestinal symptoms are the most common manifestations of the classic form of CD, which begins within the first 2 years of life (1). Conversely, in children older than 2 years or in adults, the clinical picture is often dominated by nonbowel manifestations (2), particularly neurologic disorders, which occur in ∼10% of patients (3). These diverse clinical manifestations can delay diagnosis and lead to complications that can be prevented by adherence to a gluten-free diet (4).

Epilepsy is one of the most common neurologic manifestations associated with CD (5). In the last decade there was an explosion of interest in this condition, which resulted in the delineation of the syndromes of childhood localization-related epilepsies, occipital calcifications, and CD (6). It has been demonstrated, however, that cerebral calcifications may not be a feature of localization-related epilepsies associated with CD (7). Affected patients may have other types of partial seizures, particularly in frontal regions (6), and show extraoccipital, mainly temporal, electroencephalographic (EEG) epileptiform discharges as well (8). Overall, especially in the early years of the disease, patients with CD may be misdiagnosed as having idiopathic childhood partial epilepsy (CPE). Testing these patients for serum immunoglobulin (Ig)A endomysial antibody, with or without performing a brain computed tomography (CT) scan, is the only way to assure a correct diagnosis. Nonetheless, questions remain as to whether this deeper investigation should be restricted to those patients with CPE and occipital paroxysms (CPEO) or extended to all CPE patients regardless of their seizure type and EEG paroxysms.

To answer this question, we examined 72 patients with an initial diagnosis of idiopathic CPE and tested them for CD.


Patient selection

The study group consisted of 72 consecutive newly diagnosed patients (31 girls and 41 boys; mean age, 12.6 ± 4.28 years; age at onset, 6.4 ± 3.7 years) observed over a 5-year period. All of the patients fulfilled the criteria proposed by the International League Against Epilepsy (ILAE) for idiopathic CPE (9) based on a comprehensive clinical, neuropsychological, and EEG investigations. Awake and sleep-deprived video-EEG recordings were obtained in all subjects. Neurologic examination and intellectual level were normal in all patients. In 25 patients, the electroclinical features were consistent with a diagnosis of CPEO, whereas the remaining 47 received a diagnosis of CPE with centrotemporal spikes (CPEC).

Once the diagnosis of idiopathic CPE was confirmed, each patient was enrolled to be tested for antigliadin (AGA) and antiendomysium (AEA) IgA and IgG antibodies. At the time of the study, none of these 72 CPE patients had previously been tested for CD. As part of routine study at our institute, 60 (88%) patients also underwent brain magnetic resonance imaging (MRI). All 25 patients with CPEO also underwent a brain CT scan. Overall, all 72 patients had at least one neuroradiologic investigation (either brain CT or MRI).

Diagnostic assay for celiac disease

Enzyme-linked immunosorbent assay (ELISA) (Alfa-gliatest; Eurospital, Trieste, Italy) was used to assess AGA antibodies, whereas indirect immunofluorescence (Endomisium; Eurospital) measured AEA antibodies. Previous studies showed ELISA to have a sensitivity of ≥96% and specificity of 97% for combined IgA-AGA and IgG-AGA antibodies (10). The AEA test is more specific but less sensitive than the AGA test , being negative in those patients with CD who also carry a selective IgA deficiency (1).


None of the enrolled patients carried a selective IgA deficiency. None of the 47 patients with CPEC had positive screening tests for CD. Neuroradiologic investigation also was normal in these patients. Of the 25 patients with CPEO, two (8%; one girl and one boy, age at onset, 9 and 6 years, respectively) were positive for AEA IgA antibodies. Both of these patients underwent intestinal biopsy taken at the Treitz ligament. In both, the jejunal biopsy showed an atrophy of the villi and hyperplasia of the crypts, which are consistent with a diagnosis of active CD. Importantly, brain CT was normal in one patient, but revealed occipital corticosubcortical calcifications in the other. Brain MRI was normal in both patients. Both of these patients were indistinguishable clinically and electroencephalographically from the other patients with idiopathic CPEO. Moreover, neither of these two patients had earlier symptoms that were in accordance with malabsorption or other gastrointestinal symptoms, even though we stressed this aspect after the diagnosis of CD was made. Hematologic and biochemical investigations—including plasma levels of vitamin B12, vitamin E, and folate—were normal. After the diagnosis, both patients started a gluten-free dietary regimen. Follow-up biopsy samples 6 months later showed mucosal recovery, which allowed us to confirm further the diagnosis of CD (11). The patients have continued to follow a gluten-free diet.

After the gluten-free diet was introduced, treatment for seizures with carbamazepine (CBZ) or lamotrigine (LTG) was started. The patient receiving CBZ has been seizure free for 52 months; the patient receiving LTG has been seizure free for 9 months. Importantly, the patient whose CT was initially normal had a repeated scan 3 years later, which was still normal.


It may be difficult for physicians to establish which patients with epilepsy should be evaluated for CD and to know how extensive an evaluation to perform. This is especially true when, as in our patient population, symptoms of malabsorption are not present. The results of our study indicate that CPEO associated with CD is a definite entity and should be considered in the differential diagnosis of patients with infantile occipital seizures (5,6). Indeed, the percentage of CPEO patients with silent CD in our study [two (8%) of 25] is significantly greater (Fisher's exact test, p < 0.00001) than the estimated prevalence (3.3 per 1,000) reported in the general population of comparable age and sex (10). Our results also indicate that the existence of silent CD associated with extraoccipital infantile partial epilepsy is unlikely, as CD screenings were negative in all our patients with CPEC.

Of note, the two patients in our study with CD and CPEO had a remission in their seizures after gluten-free diets were started; cerebral calcifications have not occurred after a 3-year follow-up in the patient with normal CT scan. These findings reinforce the notion that early identification and treatment of CD may reverse the tendency to epilepsy and probably the development of calcifications (12). Moreover, the absence of cortical calcifications in one patient is in keeping with the current view that calcifications are not necessarily caused by CPE associated with CD, as seizures often develop earlier than their appearance (13). Although the underlying mechanism of calcifications in CD is still a matter of discussion, it is possible that a genetic predisposition plays a part in their pathogenesis (7).

There is also a great deal of speculation about the pathogenic mechanism of epilepsy in CD. Some authors claimed that metabolic disturbances—particularly vitamin B12 deficiency or low serum levels of folate or other vitamins—may play an important role (14). This hypothesis seems unlikely, however, because these are common biochemical abnormalities in uncomplicated CD, and often, as in our patients, are not present in CD patients with epilepsy (6). Other authors have speculated that the cause of epilepsy has to do with the altered immune function (6), which may produce both the gastrointestinal abnormalities and epilepsy. It has been shown that the autoimmune process causes an isolated vasculitis (15), which in turn may lead to the pathologic changes seen in CD, characterized by cortical vascular abnormality with patchy pial angiomatosis, fibrosed veins, and large jagged microcalcifications (16). In this way, it is reasonable to hypothesize that early treatment might preempt the development of epilepsy in predisposed individuals. Nonetheless, such an altered immune function affects selectively the occipital regions of the brain for poorly understood reasons.

Finally, an innovative diagnostic assay for CD using human recombinant tissue transglutaminase (TAGA) has been introduced recently as a novel cost-effective strategy for identifying both symptomatic and silent forms of CD (17). This technique is cheaper and easier to perform than the epithelial membrane antigen (EMA) markers traditionally used; it also appears to be more sensitive. Nonetheless, the improved sensitivity of these techniques seems not to be significantly higher than that of the established methods. Therefore we believe that performing serum measurements of TAGA would not have allowed us to identify additional silent-CD patients.

In conclusion, the results of our study further indicate that serologic screening for CD should be routinely performed in patients with infantile partial epilepsy and occipital paroxysms, because clinical or even radiologic abnormalities are often absent. Conversely, there seems to be no indication to extend CD screening to those patients with infantile extraoccipital seizures.