Postpartum endometritis in cattle is a multifactorial disease with high economic impact. Inflammation of the bovine uterus has been demonstrated to decrease reproductive performance. Both, clinical (CE) and subclinical endometritis (SCE) were associated with increased days to first service as well as decreased conception and pregnancy rates resulting in an increased risk of culling [1–4].

Bacteriological contaminations of the uterus after parturition and metabolic changes in the transition period are important etiological factors. Negative energy balance, is known to influence the number and functional properties of polymorphonuclear cells (PMN) [5, 6]. There is evidence that periparturient depression of functional properties and number of PMN is of great importance for the pathogenesis of endometritis, as recently reviewed [7–9].

The identification of risk factors for endometritis will contribute to a better understanding of the underlying mechanisms in the pathogenesis of endometritis. An early identification of cows at risk for endometritis could provide new targets for intervention and the basis for changes in management practices to prevent this disease. Decreased dry matter intake (DMI) prior to parturition is associated with mobilization of lipids, which are released as non-esterified fatty acids (NEFA) from adipose tissue [10]. Decreased DMI and increased NEFA levels are associated with periparturient suppression of the immune function, resulting in a greater susceptibility of cows to infection [11]. Cows with clinical or subclinical ketosis have shown impaired phagocytic activity of PMN isolated from milk and blood [12–14]. Elevated serum BHBA concentrations in the first two weeks postpartum were indicative of an increased risk (OR = 3.35) of metritis [15], a common pre-stage for CE and SCE, and decreased pregnancy risk (OR = 0.48)[16]. A reduction of probability of pregnancy of 20 to 50% was found, depending on the magnitude and duration of elevated serum BHBA (≥1,000 μmol in wk +1 postpartum, P = 0.04 or ≥1,400 μmol in wk +2 postpartum, P = 0.01)[16]. A relationship between the plasma concentration of bilirubin and the respiratory burst activity of bovine PMN in vitro was found [17]. It was hypothesized by the authors that bilirubin has potential as a diagnostic marker of impaired neutrophil function and consequently for identification of cows at risk around parturition. Furthermore, elevated urea concentrations had an effect of decreasing uterine pH during the luteal phase [18–20]. Bovine endometrial cells in culture responded directly to increasing urea concentrations with alteration in pH gradient [21]. This effect might facilitate subsequent bacterial growth and lower local immune defense.

Few publications are available describing relationships between metabolic disorders and the prevalence of CE and SCE under field conditions. Hammon et al. [5] reported significantly elevated concentrations of NEFA and BHBA around parturition for cows with CE and SCE compared with healthy cows. Decreased PMN function and elevated plasma levels of NEFA prior to parturition and elevated plasma levels of BHBA postpartum were associated with uterine disorders later in lactation. However, predictor cut points have not been determined yet.

The objective of this study was to investigate the relationship between elevated serum concentrations of NEFA, BHBA, bilirubin and urea in the periparturient period and the prevalence of CE and SCE. Specifically, we set out to determine sensitivity and specificity of these metabolic parameters for the detection of CE and SCE, establishing optimal predictor cut points for daily practical use.

Previous studies demonstrated negative impacts of elevated concentrations of NEFA, BHBA, bilirubin, and urea on PMN function [5, 12, 13, 17], uterine environment [18–20, 22] and the prevalence of metritis [5, 15, 16, 23, 24]. These findings led to our hypothesis that elevated concentrations of these metabolic parameters might serve as indicators for the presence of endometritis. Differences between PP and MP cows in the regulation of fat tissue mobilization [25] and the DMI around parturition [23] have been described. Therefore, in our study metabolic profiles were analyzed separately for MP and PP cows. In MP cows, NEFA concentrations measured at wk -1 were a significant predictor for cows at higher risk of CE or SCE (CE: P = 0.05, SCE: P = 0.04) with lower concentrations in healthy cows. Sensitivities were low, but specificities were fairly high (CE: sensitivity 38%, specificity 87%; SCE: sensitivity 35%, specificity 89%). Yet, our data generated in the field support evidence from earlier in vitro studies [6, 26, 27] showing that high concentrations of NEFA can affect functions of bovine blood PMN. Also this study confirms previous findings [5] on an association between energy status prior to calving and uterine health in the postpartum dairy cow utilizing a different diagnostic technique (i.e. cytobrush) for SCE that has been described as more reliable than the uterine lavage [28]. To our knowledge this is the first study to describe an exact predictor cut point for NEFA to discriminate between healthy and diseased cows.

Elevated concentrations of BHBA and other ketone bodies have been shown to impair the proliferation of bone marrow cells (> 1.0 mmol/L) [14], the proliferation of lymphocytes (6.25 mmol/L) in vitro [29], the in vitro chemotactic differentials of leukocytes (> 1.6 mmol/L) [30] and the respiratory burst activity of PMN (2.5 mmol/L) [12] in cattle. Field studies found elevated concentrations of BHBA during early lactation in cows with CE and SCE compared to healthy cows [5] and elevated BHBA concentrations (> 1.2 mmol/L) in the first week postpartum indicative of an increased risk of subsequent metritis [15]. Surprisingly, our data do not confirm the diagnostic value of BHBA concentrations to predict endometritis. Differences in experimental designs, methodologies and disease definitions might have contributed to this discrepancy. In the field studies cited metritis or endometritis was diagnosed earlier in lactation (21 to 28 days postpartum [5] and before 15 days postpartum [15]) than in the current study (28 to 35 days postpartum). Furthermore, SCE was diagnosed by uterine lavage [5] and a higher threshold of PMN (> 25% vs. > 18% in our study) was used.

Alterations of uterine pH caused by elevated concentrations of urea [19, 20] have been demonstrated to affect the viability of embryos [31–33]. This study investigated the relationship between serum concentrations of urea and uterine health. Multiparous cows with CE had lower (healthy vs. CE: P = 0.01, SCE vs. CE: P = 0.03) plasma concentrations of urea at wk +5. This could be explained by a possible lower feed intake of cows with CE. A recent study [23] showed that cows with uterine disease occurring in the 3 weeks postpartum consumed less dry matter during the transition period compared to healthy cows. Huzzey et al. [23], however, diagnosed metritis, not endometritis, so criteria for uterine disease were not similar to the present study. Also, it is not clear whether the reduction in DMI is the reason for uterine disease or a consequence. Information concerning the effect of a lower urea concentration, as found in cows with CE in this study, on the uterine environment or on PMN function is to our knowledge not available. On the other side, PP cows with CE had higher serum concentrations of urea at wk +1. We can only speculate about the reason for the opposing results concerning urea concentrations in PP and MP cows with CE in this study. In cows with a ruminal flora not adapted to lactational rations, the dietary protein supply exceeds the energy availability in the bovine rumen, which results in higher urea concentrations [34]. It is possible that in our study PP cows faced a more abrupt diet transition than MP cows due to the change in ration. But our study does not provide sufficient data to prove this hypothesis.

The body condition score (BCS) is used to appraise body fat in dairy cattle. The change in BCS is an indirect measurement of the fat metabolism and thereby related to the tested metabolic parameters, particularly NEFA and BHBA [35]. Measurement of the chosen metabolites, however, is more objective, repeatable, and subtle than the BCS. Furthermore, the parameters offer information about the actual differences between cows [34]. Therefore, BSC was not included in the study. We are aware, however, that the inclusion of the BCS could have added another semi-quantifiable factor of potential influence on CE and SCE.

For the determination of the optimal predictor cut points the statistical software MedCalc utilizes the Youden's index [36]. The index balances sensitivity and specificity equally. It is important to note that depending on the goals of diagnosing cows with CE or SCE one could use a different approach such as choosing a threshold based on high sensitivity or high specificity.

The combinations of sensitivity and specificity for the predictors cut points determined in this study (in PP cows: urea at wk +1, in MP cows: NEFA at wk -1 and urea at wk +5) were low (35 to 61% and 70 to 90%, respectively) and unsatisfactory for practical use. Therefore, the serum concentrations of NEFA, BHBA, bilirubin or urea at wk -1, wk +1 or wk+5 relative to calving are unsatisfactory for disease prediction. This study was designed as a first approach to establish predictor cut points for daily practical use. Further research, with tighter sampling intervals around parturition, measurement of body weight, and monitoring of actual DMI, is required to better describe factors that contribute to the risk of postpartum uterine disease.