A pregnancy detection assay using milk samples: Evaluation and considerations

Two experiments were conducted to evaluate a pregnancy-detection assay based on the measurement of pregnancy-associated glycoproteins (PAG) in milk samples. In experiment 1, milk samples were collected on the day of first pregnancy check (33-52 d postinsemination; n = 119) or second check (60-74 d postinsemination; n = 60). The accuracy in identification of pregnant and nonpregnant cows was 99% at first check. Only 6% of samples were found to be within an intermediate range of PAG concentrations and classified as requiring recheck by the assay. At second check, the accuracy of the assay was 98%. Fifteen percent of these samples were classified as requiring recheck. In experiments 2a (n = 17 cows) and 2b (n = 16 cows), milk and plasma samples were collected from cows at weekly intervals beginning 2 (experiment 2a) or 4 d (experiment 2b) after insemination. The earliest time point at which pregnant cows were accurately classified as pregnant by the assay was on d 30 postinsemination. A transient decline in PAG levels into the intermediate range was observed on d 46 to 72 postinsemination. This coincides with the time of recheck in experiment 1. Results obtained with the plasma samples were essentially the same. The accuracy of pregnancy identification based on milk samples from nonpregnant and pregnant cows was 99%. Levels of PAG in milk were useful in identifying 6 incidences of embryonic mortality. No consistent relationship was noted between the timing of the decline in PAG levels and the timing of luteal regression in this small number of cows.