Introduction

Transition dairy cows enter a state of negative energy balance (NEB) due to increased fatty acids, glucose, and amino acids for colostrum and milk synthesis at parturition (^{Grummer et al., 2004}). This deficit of nutrients is met by mobilizing adipose and muscle tissue in the form of non-esterified fatty acids (NEFA) and amino acids. Therefore, physiological changes occur to support the flux of nutrients to the mammary gland at the expense of other body tissues (^{Roche et al., 2017}). The restricted availability of nutrients, in association with increased energy demand for milk synthesis, leads to modifications in many metabolic pathways and impairment of the finely tuned inter-relation between energy metabolism and reproductive function. Although these changes are normal adaptive processes in high-yielding cows, when cows fail to adjust to this metabolic challenge the resulting NEB adversely affects the reproductive potential (^{Butler, 2003}; ^{Wathes et al., 2007a}) even months later (^{Leroy et al., 2008}), as well as milk yield (^{Ospina et al., 2010}). These metabolic and endocrinological changes associated with reduced reproductive performance are attributed, in part, to a suboptimal follicular environment and reduced developmental competence of the oocyte and subsequent embryo (^{Ashworth et al., 2009}; ^{Van Hoeck et al., 2014}).

Additionally, mobilization of muscle and fat during early lactation results in deteriorating body energy reserves that is reflected in the cows' metabolic and hormone profiles, such as reduced blood glucose concentrations, increased levels of β-hydroxybutyrate (BHBA) and non-esterified fatty acids (NEFAs) (^{Pires et al., 2013}). Body condition score (BCS) is also readily affected by the mobilization of body energy reserves, and its assessment is a physical indicator of negative energy balance. These reduced body energy reserves at calving in high-yielding dairy cows may interfere with conception and maintenance of pregnancy in cattle (^{Santos et al., 2009}; ^{Gomez et al., 2018}), as well as milk yield (^{Roche et al., 2009}). Given that, in dairy cows, reproductive function and metabolism are reciprocally regulated and tightly connected, blood metabolites are interlinked and provide a tool to assess the physiological state and to assist in the identification of possible biomarkers for fertility research (^{Oikonomou et al., 2008}; ^{Garverick et al., 2013}). Blood metabolites can be used objectively, reliably and routinely to assess the nutritional status of cows.

The objectives of this study were to determine the association between serum metabolites (indicative of body energy reserves) and BCS during the first week of lactation on various reproductive variables and milk yield of contemporary high-yielding dairy cows in a hot environment.

Materials and Methods

Results

There was a significant association between blood BHBA concentration within the first week after calving and first-service pregnancy rate. Cows with serum BHBA levels ≤ 0.8 mmol/L about one-week post-calving were three times more likely to become pregnant at first service than were cows with higher levels of this metabolite (Table 1).

Likewise, cows with serum creatinine levels ≥2.0 mg/dL had 2.7 times greater chance of becoming pregnant at first service than cows with serum creatinine levels ≤2.0 mg/dL at the beginning of lactation. Cows experiencing a lower reduction in BCS (≤0.41 units) from 30 to 60 d postpartum were 2.5 times more likely to become pregnant at first breeding than cows losing more BCS postpartum.

The critical thresholds of serum BHBA for predicting pregnancy at first service was 0.80 mmol/L (area under the curve; AUC, 0.69) within the first week postpartum, 1.8 mg/dL for creatinine (AUC, 0.61) and 3.0 (AUC, 0.61) for BCS at 30 and 60 d postpartum.

As shown in Table 2, both BCS at 30 and 60 d postpartum were the only variables significantly associated with conception rate to all services. BCS at 30 and 60 d postpartum were equally good predictors of pregnancy potential to all services (AUCs 0.64 and 0.66, respectively; p<0.01). The percentage of cows pregnant with more than two services, grouped by serum metabolite concentrations one-week postpartum are shown in Table 3.

Table 1 Associations of postpartum serum β-hydroxybutyrate and creatinine concentrations within one week after calving and body condition score at 30 and 60 d post-calving with the risk of first-service pregnancy rate (FSCR) in high-yielding Holstein cows (n= 165) in a hot environment. Only significant variables are presented. 

AUC= area under the ROC (receiver operating characteristic) curve; BCS= Body condition score.

¹Cutpoint = herd alarm level where a reduced probability of first-service conception rate exists based on serum metabolite concentration and body condition score in transition cows.

²Sen = Epidemiologic sensitivity from a receiver operating characteristic curve.

³Spec = Epidemiologic specificity from a receiver operating characteristic curve.

Table 2 Associations of body condition score (BCS) at 30 and 60 days post-calving with the risk of pregnancy rate (PR) to all services in high-yielding Holstein cows (n= 165) in a hot environment. Only significant variables are presented. 

¹Cutpoint = herd alarm level where an increased probability of non-pregnancy to all services exists based on body condition score in transition cows.

²Sen = Epidemiologic sensitivity from a receiver operating characteristic curve.

³Spec = Epidemiologic specificity from a receiver operating characteristic curve.

Prediction threshold for less than two services per pregnancy was ≤0.80 mmol/L BHBH one week postpartum. Cows with serum glucose ≤70 mg/dL were 1.9 times more likely to require more than two services to become pregnant, relative to cows with higher blood glucose one week postpartum.

Table 3 Associations of postpartum serum metabolite concentrations one week after calving with the risk of services per conception greater than two (S/C >2) in high-yielding Holstein cows (n= 165) in a hot environment. Only significant variables are presented. 

AUC= area under the ROC (receiver operating characteristic) curve.

¹Cutpoint = herd alarm level where an increased probability of more than two services per pregnancy exists based on serum metabolite concentration in transition cows.

²Sen = Epidemiologic sensitivity from a receiver operating characteristic curve.

³Spec = Epidemiologic specificity from a receiver operating characteristic curve.

Cows with serum levels ≥5.0 mg/dL TP had greater odds of requiring more than two services to become pregnant than cows with lower serum levels of this metabolite. Both serum BHBA and creatinine were better predictors of pregnancy with 2 or fewer services (AUCs 0.64 for both serum metabolites; p<0.01) than other serum metabolites measured early in lactation.

Odds ratios and corresponding confidence intervals for variables associated with pregnancy before 80 d postpartum are presented in Table 4. The only factor associated with pregnancy before 80 d postpartum was serum levels of BHBA (p<0.05). Data showed that a cutoff point of 0.81 mmol/L serum BHBA one-week post-calving could differentiate between cows conceiving by 80 days postpartum.

Table 4 Associations of postpartum serum β-hydroxybutyrate one week after calving with the risk of pregnancy before 80 days (d) post-partum (pp) in high-yielding Holstein cows (n = 165) in a hot environment. Only significant variables are presented. 

AUC= area under the ROC (receiver operating characteristic) curve.

¹Cutpoint = herd alarm level where a reduced probability of pregnancy before 80 days post-calving exists based on serum metabolite concentration in transition cows.

²Sen = Epidemiologic sensitivity from a receiver operating characteristic curve.

³Spec = Epidemiologic specificity from a receiver operating characteristic curve.

Cows with serum BUN levels ≥15 mg/ dL one week postpartum were 2.0 times more likely to produce >10,500 kg in 305 d lactations compared to herd mates with lower levels of this metabolite (Table 5). The threshold value had a significant effect (p<0.01) in discriminating cows with less or more than 10,500 kg of milk yield in 305 d (AUC= 0.60). The AUC for the groups with serum creatinine higher or lower than 2.0 mg/dL creatinine was 0.66 (p<0.01) with an optimal prediction threshold for 305- d lactations >10,500 kg of 1.8 mmol/L of this metabolite. Higher BCS change at 30-d postpartum had higher odds of 305-d milk yield greater than 10,500 kg. Compared to cows with serum total proteins ≥5.0 mg/dL, cows with serum total proteins ≤5.0 mg/dL were 3.1 times more likely to have a 305-d lactation >10,500 kg. Serum alkaline phosphatase concentrations did not evidence the presence of fatty liver, and consequently, this blood metabolite did not alter the reproductive performance of cows.

Discussion

Serum BHBA one-week postpartum was successful at discriminating between cows becoming pregnant at first service, cows requiring less than two services per pregnancy, and cows becoming pregnant before 80 d postpartum. Although circulating ketone bodies can be used to a certain extent as energy source by mammary tissue and other organs, excessive blood ketone bodies negatively affect the reproduction function. In line with the results of the present study, ^{Rutherford et al. (2016)} and ^{Mellado et al. (2018b)} also found that low blood BHBA levels shortly after calving had a positive effect on the establishment of pregnancy at first service. However, in the present study this metabolite was not related to conception rate to all services, which is not in line with ^{Walsh et al. (2007)} and ^{Ospina et al. 2010}) who observed higher overall conception rates in nonketotic than in ketotic cows.

Table 5 Associations of postpartum serum metabolite concentrations one week after calving with the risk of more than 10,500 kg of milk in 305-day lactations in high-yielding Holstein cows (n= 165) in a hot environment. Only significant variables are presented. 

BUN= blood urea nitrogen; BCS= body condition score; AUC= area under the ROC (receiver operating characteristic) curve.

¹Cutpoint = herd alarm level where an increased probability of 305-d milk yield <10,500 kg exists based on serum metabolite concentration in transition cows.

²Sen = Epidemiologic sensitivity from a receiver operating characteristic curve.

³Spec = Epidemiologic specificity from a receiver operating characteristic curve.

This lack of difference in overall conception rate compared with previous studies seems to be because non-pregnant cows in our study were recurrently re-bred (cows were served on average 2.7 times per pregnancy), and it is likely that by the time cows received their second or third service, high serum BHBA had subsided. Most reports for postpartum blood BHBA concentration cut-points that maximize accuracy of disease prediction and production measures range from 1.2 to 1.46 mmol/L (McArt et al., 2013). It was clear in the present study that much lower levels of serum BHBA (0.80 mmol/L) one week after calving affect the reproduction function of high-yielding cows, based on increased negative effects on the reproductive performance of transition Holstein cows.

Surprisingly, serum NEFA levels did not affect the probability of pregnancy at first AI, which is not in line with other studies in lactating dairy cows (^{Ospina et al., 2010}; ^{Garverick et al., 2013}) where the likelihood of pregnancy is reduced when blood NEFA concentrations increase postpartum. Thus, our data suggest that serum NEFA is a less efficient indicator of subsequent fertility than serum BHBA in high-yielding dairy cows during the early lactation period. This is in agreement with ^{Duffield (2000)} who stated that the use of BHBA is a better indicator of energy imbalance than NEFA in postpartum animals.

Our results also indicate that serum creatinine levels one-week postpartum can be used to distinguish cows more likely to become pregnant at first service. Creatinine formation occurs almost exclusively in the muscle and this metabolite is highly related to total body muscle mass (^{Costa-Silva et al., 2012}; ^{Kim et al., 2016}); therefore, serum creatinine could be considered a specific index to define muscle mass. Thus, individuals with higher muscle mass present higher blood creatinine (^{Baxmann et al., 2008}), and larger muscle mass is a reflection of a higher body condition score, which is highly related to improved reproduction function of dairy cows (^{López-Gatius et al., 2003}; ^{Carvalho et al., 2014}).

Low BCS at 30 and 60 days postpartum had a detrimental effect on both first-service conception rate and conception rate to all services. These results are in line with other studies where BCS at 60 to 100 d of lactation was positively associated with pregnancy rate (^{Buckley et al., 2003}; ^{Gomez et al., 2018}). This response may be explained because high milk yielding dairy cows use a great amount of energy to cope with the marked increase in milk yield at the beginning of lactation. This increased energy requirement is partially met by increased feed intake but at a slower rate than milk production; the remainder being met by retrieval of body energy reserves which results in cows with a NEB (^{Grummer, 2007}). The timing of BCS loss (NEB) coincides with involution and remodeling of the uterine and oviductal structures, ovarian activity reinitiation, and, for some cows, fertilization and blastocyst and early conceptus growth. This NEB leads to decrease BCS and cows in low BCS at calving, or those suffering excess BCS loss early post-calving, are more likely to present delayed first ovulation (^{Shrestha et al., 2005}), are less likely to ovulate, have reduced conception rate to first service and also have increased likelihood for pregnancy loss and increased calving to conception interval (^{Roche et al., 2009}; ^{Shin et al., 2015}; ^{Barletta et al., 2017}).

In the present study, high serum BHBA concentrations were associated with higher number of services per conception, which is in line with other studies with dairy cows experiencing ketosis shortly after calving (^{Gillund et al., 2001}; ^{Mellado et al., 2018b}). Considering AUC (0.64), blood BHBA concentrations one-week post-calving presented a moderately accurate discriminating ability for cows requiring more than two services to become pregnant. It was also found that blood glucose at the beginning of lactation is useful for defining whether a cow requires more than two services for conception. These results are in agreement with findings of ^{Guzel and Tanriverdi (2014)}, who observed that repeat breeder cows had lower blood glucose levels compared to fertile ones. It has been considered that blood glucose has a moderate diagnostic value in the assessment of the nutritional status of cattle as it varies moderately in blood, however, in the present study, the AUC of 0.61 showed that this blood metabolite was a fair classifier to separate cows needing more than two services per pregnancy.

Cows with lower serum creatinine levels required more services to become pregnant than cows with high levels. Given that serum creatinine is a biomarker of skeletal muscle mass (^{Patel et al., 2013}) or body weight (^{Costa- Silva et al., 2012}), cows with more intense mobilization of muscle protein (high blood creatinine concentration indicate proteolysis; ^{McCabe and Boerman, 2020}) presented decreased reproductive performance as cows with lesser muscle mass cannot optimize reproduction function (^{Bewley and Schutz, 2008}). However, it is worth mentioning that urinary creatinine is not related to BCS in cattle (^{Whittet et al., 2019}). Cows with high serum TP levels were also more likely to require more than two services per pregnancy. These results are difficult to explain because blood TP reflects the availability of protein and their level decline with protein deficiency. It could be that the major contribution of TP was from the diet, as the body has a regulatory mechanism to maintain the adequate protein levels in the serum, and high protein intake has a negative influence on services per pregnancy (^{Sinclair et al., 2014}). Increased proteins (mainly serum albumin and immunoglobulins) in the blood of cows indicate activation of the immune response following an infection (^{Tothova et al., 2016}); thus, it is not discarded that high serum TP could also have arisen due to mild periparturient diseases.

Serum BUN during the first week of lactation was identified as a predictive biomarker of 305- d milk yield above 10,500 kg in pluriparous cows. High BUN levels could indicate a high protein intake (energy intake held constant) or the excessive mobilization of muscle. Given that cows in this study received high energy diets, it is believed that the higher milk yield of cows with high BUN was due to higher ingestion of protein. Increased solubility or degradability of dietary protein leads to increased ruminal ammonia concentrations resulting in increased BUN concentrations (^{Hristov et al., 2004}). These results are in agreement with findings of ^{Nozad et al. (2012)}, who found higher BUN levels in high-yielding dairy cows relative to low-yielding cows. ^{Wathes et al. (2007b)} also documented that blood urea is positively related to milk yield. Thus, 14.9 mg/dL serum BUN levels shortly after calving for pluriparous high milk yield is indicative of higher odds of cows to reach 305-d milk yield >10,500 kg.

Serum creatinine levels lower than 1.8 mg/dL resulted in higher proportion of cows presenting >10,500 kg milk in 305-d lactations. Serum creatinine levels correlated significantly with body weight, but the level of correlation with lean mass is even greater (^{Baxmann et al., 2008}). Low serum levels of creatinine in cows producing >10,500 kg of milk evidence that cows were mobilizing high amounts of body proteins during the first weeks of lactation. Thus, these data suggest that cows with greater muscle mass had higher body energy reserves which could be used for greater milk synthesis, which resulted in 305-d milk yields >10,500 kg. This was supported by the fact that cows with higher BCS changes from calving to 30 d post-calving (e.g. higher mobilization of body fat reserves) were more likely to present 305-d milk yield greater than 10,500 kg than cows with less BCS change post-calving. When milk yield is ascending at the beginning of lactation and energy requirements exceed feed intake cows experience NEB, which leads to mobilization of lipid and labile protein reserves, getting thinner, and lose BCS (^{Agenäs et al., 2003}). In line with the results of the present study, other reports have demonstrated that cows with greater body reserve mobilization tend to produce more milk than cows with medium or low levels of body energy mobilization (^{Pedernera et al., 2010}; ^{Gheise et al., 2017}).

In conclusion, this study shows that high serum BHBA and low serum creatinine levels during the first week of lactation in high-yielding Holstein cows in a hot environment are negatively linked to reproductive efficiency. Further, BCS in early lactation is predictive of cows establishing pregnancy, therefore it would be useful to monitor and perhaps modify BCS during the 30-60 d after calving to optimize reproduction in high-producing Holstein cows. Additionally, high serum BUN, low creatinine, and TP, as well as a greater loss of BCS during the first month of lactation are sensitive biochemical markers to detect animals capable of producing >10,500 kg milk in 305 d among high-yielding Holstein cows. For their greater sensitivity and specificity, blood values of BHBA can be used in early lactation to predict reproductive performance, which will not only provide new references for prediction and diagnosis of ketosis but also finds novel directions for research focussed on the connection between BHBA and reproductive variables. Serum alkaline phosphatase was not associated with any of the reproductive variables, which suggests that fatty liver was not a major metabolic disorder in these cows, and consequently did not decrease health status and reproductive performance.