Objective: To analyse serum concentrations of brain specific S100 protein in patients with Creutzfeldt-Jakob disease and in controls.
Design: Prospective case-control study.
Setting: National Creutzfeldt-Jakob disease surveillance unit.
Subjects: 224 patients referred to the surveillance unit with suspected Creutzfeldt-Jakob disease and 35 control patients without dementia.
Main outcome measure: Serum concentration of S100 protein in patients with Creutzfeldt-Jakob disease, in patients with other diseases causing dementia, and in the control group.
Results: Of the 224 patients with suspected Creutzfeldt-Jakob disease, 65 were classed as definitely having the disease after neuropathological verification, an additional 6 were classed as definitely having the disease as a result of a genetic mutation, 43 as probably having the disease, 36 as possibly having the disease, and 74 patients were classed as having other disease. In the 108 patients classed as definitely or probably having Creutzfeldt-Jakob disease the median serum concentration of S100 was 395 pg/ml (SD 387 pg/ml). This was significantly higher than concentrations found in the 74 patients classed as having other diseases (median 109 pg/ml; SD 177 pg/ml; P = 0.0001). At a cut off point of 213 pg/ml sensitivity for the diagnosis of the disease was 77.8% (95% confidence interval 68.8% to 85.2%) and specificity was 81.1% (70.3% to 89.3%). There was a significant difference in survival at different concentrations of S100 in Kaplan-Meier curves (P = 0.023).
Conclusion: Measurement of serum concentrations of S100 is a valuable tool which can be used more easily than tests on cerebrospinal fluid in the differential diagnosis of Creutzfeldt-Jakob disease. More studies are needed to determine whether serial testing of serum S100 improves diagnostic accuracy.