B-077 Statistics of the ratio of 0-sialo transferrin to 2-sialo transferrin among plasma and cerebrospinal fluid (CSF) specimens, using the Sebia beta-2 transferrin electrophoresis/immunofixation assay

Background Assays for beta-2 transferrin (asialo transferrin, or 0-sialo transferrin) are conducted to assess for presence of CSF in body fluids. Clinical contexts are those of potential cranial leakage (surgery, trauma) or leakage of intrathecal pumps. We evaluated the Sebia beta-2 transferrin electrophoresis/immunofixation assay as a laboratory-developed test (LDT) for detection of presence of CSF. Results assessment relies on the ratio (R) of integrals of chromatographic peaks for 0-sialo transferrin (0-ST) to 2-sialo transferrin (2-ST). According to the Sebia assay instructions, R >1 is a sufficient condition for indication of the presence of CSF. As a matter of assay validation, we examined the statistics of variation in R among plasma and CSF control specimens, in order to determine sensitivity and specificity of the assay to distinguish CSF from plasma as a function of R. Methods The beta-2 transferrin assay was conducted according to manufacturer*s instructions. Pooled plasma (A) and pooled CSF (B) specimens were prepared de-novo on each day of testing for use as controls. Specimen A was created from a combination of equal-volume aliquots from 3 pre-existing, to-be-discarded, de-identified K-EDTA plasma samples. Specimen B was created from a combination of equal-volume aliquots from 3 pre-existing, to-be-discarded, de-identified CSF samples. R data for A and B specimens were collected retrospectively for tests run over a 16-month interval (July 2023-October 2024; n = 91). Peak integration used a flat baseline method, which is used in example data analyses in the Sebia kit insert, and which is the default method in the Sebia chromatographic data analysis software. Results After removal of results for R deemed to be outliers (deviation of R from mean > ±2.5 SD), results for R for both plasma (n = 90) and CSF (n = 88) were, expectedly, normally distributed (r > 0.98): R(plasma) = 0.79 ± 0.252; R(CSF) = 2.52 ± 0.411. These distributions were widely separated, such that there was very high sensitivity (>99%) and specificity (>99%) in ability to distinguish pure CSF (R > R(cutoff)) from plasma (R < R(cutoff)), using R(cutoff) = 1.47. Unexpectedly, R(plasma) had a significant fraction of specimens for which R was >1 (20.3%); correspondingly, use of R(cutoff) = 1 would reduce specificity relative to plasma to 79.7%. Determination of specificity and sensitivity to detect CSF in samples that may be admixtures of CSF and plasma is complicated by the fact that densitometry of peaks in the assay is not a linear function of dilution. Per Sebia kit instructions, the best interpretation is to declare the test result as “CSF positive” when R(specimen) > R(plasma). In cases where a plasma sample is not available, R(plasma) for the control plasma specimen should be used. Conclusions: Distributions in R = (0-ST)/(2-ST) for plasma and CSF control specimens were widely separated in the Sebia beta-2 transferrin assay, leading to high sensitivity and specificity for distinguishing pure CSF from plasma, using R(cutoff) = 1.47. For potential CSF/plasma admixtures with R(specimen) < R(cutoff), R(specimen) > R(plasma) might be used for interpretations as “CSF positive".