Beta‐trace protein concentration in nasal secretion: discrepancies and flaws in recent publications

Dear Sirs, I would like to reply to Bachmann and Petereit: Beta-trace protein as sensitive marker for liquorrhea. Acta Neurologica Scandinavica 2004; 110: 337–338. A number of recent publications (1–3) deal with the beta-trace protein concentration in nasal secretions as a sensitive marker for cerebrospinal fluid (CSF) rhinorhea and CSF otorhea. In their letter to the editor (4) Bachmann and Petereit discussed our report (1) on the reference range of normal beta trace values and in particular the cut-off value reported for discrimination between a normal nasal secretion and a secretion contaminated with CSF. We appreciate and share their concerns to find a clinically relevant cut-off value, which is high enough to avoid false-positive interpretations. We also agree that a determination of the clinically relevant cut-off value must be based on values of clinically defined cases. But the data of a control group with normal beta trace values in the nasal secretion (normal reference range) is also a mandatory information to get a reliable cut-off value. Unfortunately none of the reports, Bachmann et al. (4) refer to, i.e. their own data (2) and the data of Arrer et al. (3, 5) (Table 1) present reliable data for a normal reference range. This reply comments therefore primarily on the accurate determination of the reference range for beta trace protein of normal controls and the discrepancies between different reports (Table 1), which are obviously due to analytical flaws. Our recent publication (1) reports reference values of nasal secretions from normal controls, (range: 0.003–0.12 mg/l, median: 0.016 mg/l), which are significantly different from the results reported by Arrer et al. (3, 5) with a 100-fold higher reference range 0.219–1.69 mg/l (mean: 0.39 mg/l). Both groups used the same commercial assay (N-Latex b-trace protein; BNA Nephelometer or BN Pro Spec; Dade Behring, Germany) with a sensitivity of 0.002 mg/l (according to the assay supplier and confirmed by (1) and (2)). The main difference between our application of the assay (1) and that of Arrer et al. (3) originate from the use of undiluted (1) versus 1:100 prediluted (3, 5) secretion samples. Petereit et al. (2) applied undiluted samples but analysed the samples with a 1:100 default dilution. Our routine analysis (1) of CSF and serum samples was performed according to the proposal of the supplier (Dade Behring). Undiluted CSF samples were analysed with a default dilution (automated dilution in the machine) of 1:400 and undiluted serum samples with an automated default dilution of 1:100. Nasal secretion samples of patients were analysed undiluted, first using the automated default dilution of 1:100. This step helps to avoid possible errors in the antigen excess range as observed for values >50 mg/l. But usually a reliable value for nasal secretions (in particular with values in the reference range) is only obtained for undiluted samples and a default dilution of 1:1 (modified automated measuring process using 1:1 instead of 1:100 default