High prevalence of bovine leukemia virus and limited detection of Mycobacterium avium subsp. paratuberculosis, with no detectable association, in dairy cattle from a highland region of Colombia

Enzootic bovine leukosis, paratuberculosis, and bovine tuberculosis are chronic infectious diseases with important implications for dairy productivity and animal health. Evidence integrating the three agents into the same animals is scarce in tropical highland dairy systems and almost absent for herds enrolled in official bovine tuberculosis certification programs. An analytical observational study evaluated 733 cattle from 37 specialized dairy herds in a highland dairy region of Colombia using serological (ELISA) and molecular methods (qPCR, ddPCR) to detect bovine leukemia virus (BLV), Mycobacterium avium subsp. paratuberculosis (MAP), and Mycobacterium bovis . Mixed-effects logistic regression with a random intercept for herds accounted for the clustering of animals within herds, and agreement between diagnostic tests was quantified using Cohen's kappa. At the herd level, BLV was detected in 70.3% of the herds by ddPCR and 67.6% by ELISA, MAP was detected in 27.0% by ELISA and 18.9% by fecal qPCR, and M. bovis was not detected in any herd. At the animal level, BLV proviral DNA was detected in 34.8% of the cattle, and BLV antibodies were detected in 31.9%; MAP fecal DNA was detected in 1.2%, and MAP antibodies were detected in 1.9%, and no animal tested positive for M. bovis by any method. Agreement between the two BLV definitions was substantial (kappa = 0.79), whereas the two MAP definitions identified essentially nonoverlapping animals (kappa = 0.07). The odds of MAP fecal shedding did not differ between BLV-positive and BLV-negative animals, but the estimate was imprecise and compatible with a wide range of effects (OR = 0.26; 95% CI: 0.03–2.43; p = 0.24, from the mixed-effects model), reflecting only nine MAP-positive animals and a single coinfected animal. Age was associated with BLV proviral detection in the mixed-effects model, although the association was less precise and not statistically significant when cluster-robust standard errors were used. BLV was detected in a high proportion of animals and herds, whereas MAP detection was limited, and no evidence of current M. bovis infection was detected using the diagnostic approaches applied in this selected population. The absence of a detectable BLV–MAP association does not imply biological independence, given the small number of MAP-positive animals. The marked clustering of BLV detection within herds (intraclass correlation coefficient of 0.45) indicates substantial unmeasured herd-level heterogeneity and has direct implications for the design of surveillance in these systems. These findings support pathogen-specific surveillance and control strategies.