Neurology & Neurosurgery resident University of Illinois Urbana, Illinois
Elevations of total nucleated cell count (TNCC) or total protein (TP) concentration in the cerebrospinal fluid (CSF) of dogs with neurologic disease can help categorize lesions in the central nervous system. Iatrogenic blood contamination at the time of CSF collection is a common occurrence, which can make diagnostic interpretation difficult. The purpose of this study was to identify the best method of correction of TNCC in CSF, based on the amount of hemodilution present. Paired sequential samples of CSF were prospectively obtained from a single collection site and patient in which one sample had visibly more hemodilution than the other. The TP concentration, red blood cell (RBC) count, and TNCC were analyzed within 30 minutes of collection. A complete blood count (CBC) was collected within 24 hours of the CSF collection. Two linear regression models were fitted, and the results were compared with 3 previously suggested correction formulas. Pearson’s correlations coefficients were calculated for each correction formula with the difference between the TNCC in the hemodiluted and clear sample. Sum squared errors were compared amongst the 5 correction formulas. Nineteen paired samples were collected. The differences between the hemodiluted and clear samples were significant for TNCC (P < .00042), RBC count (P < .0001) and TP (P = .0013). Using a ratio of 500:1 and 1000:1, RBC:WBC yielded a correlation of r = 0.78. A previously suggested correction formula that included the WBC and RBC from the CBC yielded a correlation of r = 0.98. Two linear regression models fitted using the CSF TNCC, RBC, and the peripheral blood RBC, with one additionally including the peripheral WBC, both had a correlation of r > 0.99. The linear regression model including both the peripheral RBC and WBC had the lowest sum squared errors.
The results of the study indicate that hemodilution has an effect on the TNCC, RBC, and TP in CSF. All correction methods including CBC parameters were more accurate than evaluating the hemodiluted CSF parameters in isolation. A linear regression model including the RBC and WBC from the CBC provided the most accurate correction of TNCC in hemodiluted CSF.