In total, 97 cases were identified for potential review of which 72 (74%) met criteria for inclusion. The most common reasons for exclusion were incomplete data (17, 68%), prior diagnosis of a neurologic disorder (7, 28%), and prior receipt of chemotherapy (2, 8%). The median number of physician evaluations prior to the diagnosis of DSRD was four, including primary care (IQR: 3-6). Clinical data extracted by the authors yielded a Cohen’s kappa coefficient of 0.73 (91% agreement).
Demographic data are presented in Table 1. The median age of onset was 14 years (IQR: 12-17) with the median age at diagnosis was 19 years (IQR: 13-27) (Figure 1). The majority of this cohort was Caucasian (55, 76%) with Hispanic ethnicity (51, 71%), the latter being at a higher rate than our control population (x2(1, N=1289) = 50.5, p<0.001, 95%CI: 1.96-2.76). Individuals with DSRD were more likely to have a history of autoimmune disease (p= 0.02, 95%CI: 1.04-1.75) and thyroid disease (p<0.001, 95%CI: 1.64-3.27) compared to individuals with DS without regression.
Clinical phenotypes are presented in Table 1. A potential preceding trigger was observed in 37 cases (51%) with recent infection reported in 16 cases (43%). The median time to peak symptoms was three months although this was heterogeneous with an IQR of 1-6.
Serum and neurodiagnostic data are presented in Table 2 and overlap of neurodiagnostic study abnormalities is presented in Figure 2. Serum analysis revealed thyroid peroxidase (TPO) and thyroglobulin antibodies were present in 25 (37%) and 20 (30%) individuals, respectively, although only TPO antibodies were significantly greater than controls (p=0.02, 95%CI: 1.06-2.16). Vitamin D 25-OH levels were also significantly lower in individuals with DSRD cohort compared to individuals with DS only (n=384) (p<0.001, 95%CI: 10.57-16.9). Analysis of cytokine profiling revealed abnormalities in 20 individuals (40%) tested with elevations in soluble IL-2 receptor (13, 62%) and IL-10 (5, 24%) most frequently observed. Comparison to individuals with DS without regression was not available for the majority of lab testing reviewed.
Abnormalities in EEG were found in 19 cases (26%). The most frequently reported electrographic feature were epileptiform discharges in the frontal or temporal lobes (n=11, 58%). Neuroimaging was abnormal in 16 cases (22%) with punctate T2 signal abnormalities (n= 13, 81%, Figure 3) and basal ganglia calcification (n=2, 18%, Figure 4) identified. Two individuals were noted to have incidental findings: an anterior temporal arachnoid cyst and Chiari I malformation.
Cerebrospinal fluid was abnormal in nine cases (17%). Pleocytosis was appreciated in five cases (all with >90% lymphocytosis), elevated total protein in nine cases (13%), elevated IgG index in seven cases (10%), and oligoclonal bands in two cases (3%). The Mayo Clinic autoimmune encephalopathy panel was negative in all tested cases in both the serum and CSF (n= 59). Neopterin was elevated in six cases (8%) although pleocytosis and/or elevated total protein were noted in all as well. In cases where time between diagnosis and neurodiagnostic testing was greater than three years, capture of neurodiagnostic abnormalities was very low (8%, 2 of 25 patients) compared to patients receiving assessment prior to two years (40%, 19 of 47 patients) even though clinical presentations were similar (p= 0.01, 95%CI: 1.64-37.06).
Clinical symptoms were predictive of some neurodiagnostic study abnormalities (Appendix 4). Predictors of any neurodiagnostic study abnormality included confusion/disorientation (p=0.01, 95%CI: 1.51-34.67), memory impairment (p=0.04, 95%CI: 1.07-25.46), catatonia (p=0.04, 95%CI: 1.08-16.21), freezing/bradykinesia (p=0.03, 95%CI: 1.04-15.63), urinary retention (p=0.02, 95%CI: 0.02, 1.20-9.13) as well as the cognitive/executive (p=0.01, 95%CI: 1.28-5.72) and motor symptom clusters (p=0.02, 95%CI: 1.09-2.61). Low vitamin D levels were predictive of having any neurodiagnostic abnormalities (r2=0.20, p<0.001, 95%CI: 0.01-0.03).
Therapeutic interventions and responses are reported in Table 3. Amongst all patients, the most effective therapies were IVIg (38/43, 88%), benzodiazepines (46/63, 77%), and electroconvulsive therapy (ECT) (36/49, 74%). Initial immunotherapeutic interventions were either steroids (30 mg/kg/d for 3-5 days, max 1000 mg) or IVIg (2 g/kg divided over 2-3 days followed by monthly infusions of 1g/kg) in all cases. IVIg was markedly effective at improving symptoms with 38 (88%) patients reporting a clinical response as opposed to only 14 (36%) with steroids. Time to therapeutic effect of steroids or IVIg was rapid at a median of 2.5 weeks (IQR: 1-3) in patients who responded. Although less frequently administered, anti-CD20 therapy (750 mg/m2, max 1000 mg), mycophenolate (600 mg/m2, max 2000 mg/d) and azathioprine (2 mg/kg/d, max 200 mg/d) were also effective in individuals who had both neurodiagnostic abnormalities and a prior response to either steroids or IVIg (Table 3).
In individuals with neurodiagnostic abnormalities, use of immunotherapy was nearly four times more likely to have a therapeutic effect compared to individuals without neurodiagnostic abnormalities (OR: 4.11, p=0.001, 95%CI: 1.88-9.02). In those with normal neurodiagnostic studies, IVIg was effective in 14 of 17 individuals (82%) who received it empirically although only one other patient had clinical improvement with other forms of non-steroid immunotherapy (1/14, 7%). Effectiveness of antipsychotics (p=0.12, 95%CI: 0.12-1.26), and benzodiazepines (p=0.42, 95%CI: 0.23-2.59) were not significantly different between groups. Antidepressants and ECT were more likely to be effective in individuals without neurodiagnostic abnormalities compared to those with these findings (OR: 0.20, p=0.02, 95%CI: 0.05-0.79 and OR: 0.09, p=0.04, 95%CI: 0.02-0.39, respectively).