Use of CRM 470/RPPHS has not achieved true consensus for ceruloplasmin measurement.

To the Editor: The use of primary protein reference material CRM 470/RPPHS (1) was intended to lead to reduced method-dependent variation in specific protein analyses. Observations from UK NEQAS for Specific Proteins indicate that this is true for most proteins, but not for ceruloplasmin. Because the measurement of ceruloplasmin is an important part of the initial screening procedure for Wilson disease, we deemed it important to publicize this anomaly to the clinical chemistry community, including both analysts and the diagnostic industry. UK NEQAS distributes specimens for chosen analytes to be analyzed by participating laboratories by their usual laboratory methods as though they were patient specimens. For the Specific Proteins Scheme, monthly distributions are made of material derived from pooled human donor serum stored at 4 °C; the pools reported here contained sodium azide (1 g/L) as preservative. The method used by each laboratory is recorded by UK NEQAS so that results may be studied for method-related differences. Data are analyzed by both method principle (e.g., turbidimetry) and then by the instrument or re-agent used. All participants classified under nephelometric method A used the Beckman Array, whereas all those classified under method B used the Behring Nephelometer II. The turbidimetric group was extremely heterogeneous, with at least nine different manufacturers/suppliers represented according to calibrant used. One-third of this group used a cali-brant from Roche and probably, although not certainly, an antiserum from this supplier. The initial observation was of unexpectedly large differences among method-related mean values for the different methods such that for the January 1999 distribution, the overall mean for ceruloplasmin was 0.ric group; P Ͻ0.01 for nephelometric A v turbidimetric). We have now confirmed that this phenomenon is consistent for 17 pools, each of which was distributed on two occasions, separated by periods of 2– 6 months, in the period January 1999 until October 2000 (Table 1). The overall mean concentrations were relatively constant, and there was no significant difference in concentration between distributions. The mean for the turbidimetric method increased significantly, whereas the mean for nephelometric method A decreased significantly and, on average, to the same degree that the mean value for turbidimetry increased. That this is not purely attributable to the method principle is indicated by the increase of the mean for another nephelometric method (method B; n ϭ 10) between distributions (mean difference, ϩ0.017 g/L; P Ͻ0.001). The degree of this phenomenon appears to be time-dependent …