Multivariate analysis of the serum-cerebrospinal fluid-protein-relation for the diagnosis of neurological diseases of the central nervous systems.
In a population comprising 197 patients, serum and CSF proteins were assayed using the radial immunodiffusion technique devised by Mancini. Multiple discriminants analysis was applied to investigate whether the measured CSF/serum protein relations and their ratios could be regarded as an indicator of specific neurological diseases. One significant finding was that the slope angle alpha of the regression line between the serum/CSF relation and molecular weight may represent an important indicative parameter. A small angle is suggestive of enhanced permeability of the BBB, a large angle of a correspondingly lowered permeability. Further, the analyses demonstrated that the combined use of several predictors can markedly improve differential diagnosis. The study also demonstrates the potential of a statistical analytic technique that is still rarely applied in medicine.