Article Text


Determinants of the copper concentration in cerebrospinal fluid
  1. Neurological Department, University Hospital Hamburg-Eppendorf, Germany
  1. Dr Hans Joerg Stuerenburg, Neurological Department, University Hospital Hamburg-Eppendorf, Martinistrasse 52, 20246 Hamburg, Germany. Telephone 0049 40 4717 4832; fax 0049 40 4717 5086.

Statistics from

The measurement of CSF copper concentration can serve as an indicator of brain copper concentration.1 2 However, the complex mechanisms by which copper crosses into the CSF, and the factors determining the CSF copper concentration in humans are largely obscure. Copper can pass into and out of the CSF by various mechanisms. For example, active transport through the blood-brain barrier or the blood-CSF barrier, or passive diffusion of the free or the bound fraction (bound to albumin or coeruloplasmin) through the blood-CSF barrier. We studied the factors influencing CSF copper concentration using a stepwise multiple linear regression model. The independent variables were age, plasma coeruloplasmin, CSF/serum albumin ratio, total serum copper concentration, and calculated serum free copper concentration (based on serum coeruloplasmin and total serum copper concentration). The CSF copper concentration was treated as a dependent variable of continuous type. We investigated lumbar CSF samples from 113 patients. These patients had dementia, extrapyramidal, or tremor symptoms; lumbar puncture was performed to exclude Wilson’s disease, and none of the patients had the disease. Copper was measured by flameless atomic absorption (Perkin Elmer, HGA 500, Ueberlingen, Germany). Coeruloplasmin was determined nephelometrically (Beckman Array: Beckman Instruments, Brea, CA, USA). The age of the patients was 50.0 (SD15.5) years; 50 were women and 63 were men. Mean serum coeruloplasmin concentrations were 394.3 (SD 11.7) mg/l. Mean serum copper concentrations were 1194 (SD 335) μg/l. Mean calculated free copper concentrations in serum were 78.5 (SD 1285) μg/l. Mean CSF copper concentrations were 14.16 (SD 6.0) μg/l. The mean albumin ratio (AR) was 6.63×10−3. The mean ratio of calculated serum free copper concentration to total serum copper was 6.6%, the ratio of CSF copper to serum copper was 1.2%, and the ratio of free serum copper to CSF copper was 18%. In the stepwise linear regression model (F to enter 4.0,F to remove: 3.996), significant positive predictors of the CSF copper concentration were found to be AR (p=0.0001) and serum coeruloplasmin (p=0.0057). The other independent variables mentioned above showed no statistically significant relation with CSF copper concentration. The figure shows the simple linear regression between CSF/serum albumin ratio and CSF copper concentration (on logarithmic axes;R=0.46, p=0.0001). The formula for the CSF copper concentration, derived from the multiple linear regression model, is: copper CSF (μg/l)=5.32 μg/l±0.653 × CSF/serum albumin ratio (×10−3)+0.012×serum coeruloplasmin (mg/l). According to this analysis,CSF/serum albumin ratio and serum coeruloplasmin together determine 25.3% of the variation in CSF copper concentration (adjustedR 2=0.253), implying that other (unknown) factors determine the remaining 74.7% of the variation. We have been able to demonstrate here that the CSF copper concentration is determined in a highly significant manner by disturbances in the blood-CSF barrier and by the serum coeruloplasmin concentration. It can be assumed from this that in the case of normal blood-CSF barrier function and a normal serum coeruloplasmin concentration, 29.7% of the measured CSF copper entered the CSF by passive diffusion bound to coeruloplasmin, and only around 0.09% by passive diffusion bound to albumin. In the case of a markedly raised CSF/serum albumin ratio of 20×10−3, this would mean that 60.6% of the measured CSF copper originated from the blood (bound to coeruloplasmin). A variable fraction of the CSF copper concentration, depending on the degree of damage to the blood-CSF barrier, therefore crosses from the blood into the CSF and can be measured there. Our formula would therefore predict, in patients with Wilson’s disease with anintact blood-CSF barrier (assuming a CSF/serum albumin ratio of 6.5×10−3), that the CSF copper concentration is actually reduced by 27.4%, when the serum coeruloplasmin concentration falls from its normal value of 394 mg/l to 68 mg/l. In consequence, CSF copper in patients with Wilson’s disease is evidently substantially free, implying that a larger fraction than previously assumed of the raised CSF copper in patients with untreated Wilson’s disease originates from the brain; the fraction entering the CSF by passive diffusion (bound to coeruloplasmin) tends towards zero. It can be concluded from this that, when the aim of therapy is considered in terms of the total CSF copper concentration, a region around 30% lower than the upper limit of the normal range should be aimed for. This is supported by the clinical finding that patients report feeling better when the CSF copper concentration is below this value. This analysis also shows that the raised copper concentration in the CSF can only originate from the brain. In particular, it is not associated with free serum copper, but evidently only via storage in the brain. The investigation here also shows that, after determining the CSF copper concentration, the coeruloplasmin-bound fraction originating from the plasma should be subtracted according to the formula we have given, or better, all measured copper concentrations in the CSF should be adjusted using the CSF/serum albumin ratio and serum coeruloplasmin concentration. A statistical relation with a low correlation (p<0.05) between CSF protein content and CSF copper was already shown in 1989 in various neurological diseases3; our study shows a much higher significance and, in addition, the effect of serum coeruloplasmin (therefore of bound serum copper). Furthermore, we have been able to determine quantitatively the fraction of CSF copper which enters the CSF across the blood-CSF barrier.

Correlation of blood-CSF barrier (albumin ratio, (AR)) with total CSF copper concentration (on logarithmic axes). R=0.46, p=0.0001; 95% confidence bands for the true mean of the total CSF copper concentration are shown.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.