Estimation of circulating bovine leukemia virus levels using conventional blood cell counts

Bovine leukemia virus (BLV) is widely prevalent in Japanese dairy farms. To control BLV infections in Japan, segregating or managing cows according to their proviral load (PVL) is a rational strategy. This study was conducted to establish a quantitative procedure for estimating blood PVL per microliter using a statistical model to offer a cost-effective alternative to the conventional quantitative real-time PCR method. In total, 250 Holstein cows infected with BLV were identified from 10 commercial dairy farms. Information on age was collected and blood samples were tested for white blood cell and lymphocyte counts and PVL using PCR. Generalized linear models with quasi-Poisson errors were used to estimate PVL by selecting age, logarithm of lymphocyte count, and their interaction term as explanatory variables. To evaluate the model, blood samples of 92 BLV-infected Holstein cows from 2 other commercial dairy farms were tested, and measured PVL values were compared with estimated PVL values by the model. The logPVL per microliter was modeled by positive associations with log lymphocyte count and age and a negative association with the interaction term. In the evaluation, measured PVL values had a strong correlation with estimated PVL values (Spearman's rho = 0.87). In conclusion, our model provides a cost-effective and more rapid alternative to the conventional method to facilitate test and segregation or management of BLV-suspected cows.