PS1-11. Resolution Outcomes and Predictive Capacity of Pregnancy-Associated Glycoprotein (PAG) Concentrations in Lactating Dairy Cows with an Initial.

Abstract

Pregnancy-associated glycoproteins (PAG) are widely used for pregnancy diagnosis in cattle, but cows classified as “recheck” present a challenge. This study aimed to characterize the outcomes of recheck diagnoses and evaluate the predictive value of initial concentrations of PAG for final pregnancy status post-artificial insemination (AI). We hypothesized that recheck cows would have a greater risk of pregnancy loss and longer interval to resolution than cows diagnosed as pregnant or nonpregnant, and that initial concentrations of PAG would predict final outcomes. The dataset included lactating dairy cows with milk samples collected at 25 to 45 days post-AI (n = 26,263). PAG were determined with the Alertys milk pregnancy test (IDEXX Laboratories), and cows were classified as nonpregnant (< 0.100), pregnant (≥ 0.250), or recheck (≥ 0.100 and < 0.250). Only pregnancies where the first sample was a recheck (n = 1,130) were analyzed. Recheck status was followed until resolved as nonpregnant, pregnant, or pregnant followed by pregnancy loss. A reference population of cows initially diagnosed as pregnant (n = 18,598) was used to estimate pregnancy loss prevalence. Analyses were performed using SAS 9.4. Among recheck diagnoses, 74.0% (836/1,130; P < 0.01) were later determined to be nonpregnant, whereas 26.0% (294/1,130) were determined to be pregnant. Subsequently pregnancy loss occurred in 25.5% (75/294) of recheck cows initially diagnosed as pregnant, which was greater (P < 0.01) than the (11.7%; 2,167/18,598) loss in the reference population. Time to resolution differed by outcome (P ≤ 0.03), with the pregnancy loss interval being longer (45.0 ± 3.3 d) than pregnant (34.5 ± 1.0 days) and nonpregnant (16.6 ± 0.7 days) categories. The number of samples required also varied (P < 0.01), with nonpregnant requiring the fewest (1.9 ± 0.03) and pregnant or pregnancy loss requiring more (3.7 ± 0.1 and 3.0 ± 0.1, respectively). Initial PAG differed among groups (P < 0.01): nonpregnant (0.14; S-N), pregnancy loss (0.17), and pregnant outcomes (0.21). Initial PAG was a predictor of final pregnancy status (AUC = 0.72) and pregnancy loss (AUC = 0.67; P < 0.01). PAG was not correlated with days post-AI at the time of sampling (ρ = 0.07, P = 0.03). In conclusion, a majority of recheck diagnoses result in a non-pregnant diagnosis, and those resolving as pregnant have an elevated risk of loss. Additionally, PAG provide moderate predictability, supporting their utility in refining management decisions.