VICTORIA, B.C. – The results of a cluster analysis may help streamline the diagnostic work-up in children with inflammatory brain diseases, sparing some of them invasive tests such as lumbar puncture or brain biopsy.
Investigators led by Dr. Tania Cellucci, a pediatric rheumatologist at the Hospital for Sick Children, University of Toronto, retrospectively analyzed presenting clinical, laboratory, and magnetic resonance imaging features in 147 pediatric patients having one of these diseases in an as-yet unpublished study.
Dr. Tania Cellucci
The results, reported in a poster session at the annual meeting of the Canadian Rheumatology Association, identified three clusters of children who had different groups of diseases and who differed significantly with respect to the prevalence of various presenting features.
The investigators studied consecutive children seen at the hospital for inflammatory brain disease between December 1998 and June 2010, reviewing medical records to identify presenting clinical, laboratory, and imaging findings, and biopsy histology. The median age was about 9 years, and half the children were girls.
The most common disease was primary CNS vasculitis (seen in 71% of the children), Dr. Cellucci reported. That was followed by CNS vasculitis secondary to a systemic condition, such as lupus (7%), the neuronal antibody syndromes (6%), and postinfectious inflammatory brain disease (4%). The rest fell into a collective category of other diseases, such as neurosarcoidosis and inflammatory channelopathy (12%).
A k-means cluster analysis identified three distinct clusters:
• Patients in cluster 1 had angiography-negative primary CNS vasculitis and neuronal antibody syndromes that were due to antibodies against NMDA (N-methyl-d-aspartate).
• Patients in cluster 2 had angiography-negative primary CNS vasculitis and neuronal antibody syndromes that were due to antibodies against some other antigen.
• Patients in cluster 3 had angiography-positive primary CNS vasculitis and secondary CNS vasculitis.
The clusters differed significantly with respect to 14 presenting features. In terms of clinical features, patients in cluster 1 were more likely than their counterparts in the other clusters to have behavioral, cognitive, and mood changes, and seizures. Patients in cluster 2 were more likely to have ataxia, fatigue, and vision abnormalities. And patients in cluster 3 were more likely to be male and to have paresis and speech deficits.
Generally, laboratory features were not helpful in differentiating among clusters, Dr. Cellucci noted. For example, the proportions of patients having an elevated erythrocyte sedimentation rate and C-reactive protein level were essentially the same across clusters. The exception was that patients in cluster 3 were less likely to have an elevated cerebrospinal fluid white blood cell count.
In contrast, presenting MRI features were helpful (but not sufficient to confirm a diagnosis). Although the majority of patients in each cluster had some abnormalities on this imaging, the specific findings differed. Patients in cluster 3 were more likely than those in the other clusters to have unilateral lesions and ischemic lesions.
"The spectrum of childhood inflammatory brain diseases has really rapidly expanded over the past decade. A lot of new diseases have been described in the literature," Dr. Cellucci explained in an interview. Furthermore, these diseases have overlapping presenting features, leading to diagnostic uncertainty.
"So this is a new field with which a lot of people feel uncomfortable," she said. "This study is the first step in which we compare the different diseases and figure out whether can we use these findings to come up with a diagnostic algorithm."
If validated in an independent cohort, the findings may help clinicians "figure out which patient needs one test, which patient needs another test, and whether we can cut down on invasive tests, like a brain biopsy or a lumbar puncture, in patients who don’t actually need it," she said.
The study’s findings, while requiring validation, could help tailor the work-up to a child’s presenting features, according to Dr. Cellucci. "All patients will need an MRI, which is reasonable since this is not invasive. But patients in cluster 1 and 2 are those who may need a lumbar puncture, testing for neuronal antibodies, and possibly a biopsy. Patients who present with the findings typical for cluster 3 may require only an angiogram in addition to the MRI," she explained.
In addition, the findings may go a long way toward identifying children in the community whose inflammatory brain disease is misdiagnosed as some other condition, such as depression, psychosis, or seizure disorder.
"It would be a really important study to find out how many of these kids we are missing in the community," Dr. Cellucci commented. "One of the critical things is that we have treatment that works well for these patients: You treat them, and they get back to normal or almost normal, most of them. So that makes it even more critical to identify them."