Association of dry matter intake, milk yield, and days to first ovulation with cytological endometritis in Holstein cows

Dry matter intake (DMI, kg/d) is closely related to the magnitude of negative energy and protein balance during the transition period, and the metabolic adaptations to support lactation in dairy cows. Thus, DMI might affect the development of cytological endo-metritis in the early postpartum period. Difficulty to adapt to these metabolic changes is related to impaired immune function and increased occurrence of reproductive disorders. We aimed to examine the association of pre-and postpartum DMI, body weight (BW), body condition score, milk yield and milk composition, and days to first ovulation with cytological endometritis at 15 (CYT15) and 30 DIM (CYT30). A second ob-jective was to understand the association of vaginal discharge with CYT15 and CYT30 and performance. We conducted a pooled statistical analysis of 5 studies, including data from 280 multiparous Holstein cows. Based on the cutoffs for the percentage of uterine polymorphonuclear cells (PMN), determined by taking the median value of the data set for 15 and 30 DIM, cows were categorized as follows: LOW15 (PMN % at 15 DIM ≤24%; n = 125), HIGH15 (PMN % at 15 DIM >24%; n = 125), LOW30 (PMN % at 30DIM ≤7%; n = 141); and HIGH30 (PMN % at 30DIM >7%; n = 139). Cows in HIGH15 consumed an average of 1.97 ± 0.5 kg/d less DM than cows in LOW15 during prepartum, and 3.01 ± 0.5 kg/d less DM during postpartum. Dry matter intake (as a percentage of BW) was higher for cows in LOW15 during pre-and postpartum than for cows in HIGH15. Moreover, cows in HIGH15 tended to have lower milk yield than cows in LOW15 from the third until the fifth week postpartum. Although DMI was not associated with CYT30, DMI (as a percentage of BW) was lower for cows in LOW30 pre-and postpartum than for cows in HIGH30. There was no association between CYT30 and milk yield. Cows in LOW15 had greater days to first ovulation than cows in HIGH15, while cows in LOW30 also had greater days to first ovulation than cows in HIGH30. Simple regression analyses demonstrated linear associations of increased DMI, particularly postpartum, with decreased uterine PMN percentage and lower vaginal discharge score. Additionally, increased units of vaginal discharge score and increased percentage units of uterine PMN were linearly associated with decreased milk yield. Corroborating with the notion of the ovarian function being associated with uterine inflammatory status, cows in HIGH15 and HIGH30 ovulated on average 3 d before than cows in LOW15 and LOW30, respectively. Cytological endometritis at 15 DIM was associated with lower DMI from 4 wk before calving until 4 wk postpartum and was associated with lower milk yield. The association of vaginal discharge with cytological endometritis was variable and dependent on the day of evaluation.


INTRODUCTION
The transition period is characterized by the progression from gestation into lactation and it is a unique period in a dairy cow's productive life.In this period, decreased DMI and the increased milk production costs result in a state of negative energy balance (NEB; Drackley, 1999).Thus, to support milk production in such a physiological state, metabolic adaptations take place.Briefly summarizing, faced with decreased DMI and the mammary gland's increased glucose requirement to support milk production, almost all available glucose is taken up by the mammary gland in early lactation (Bell, 1995).As a result, circulating nonesterified fatty acids (NEFA) increase because lipolysis increases in response to reduced systemic insulin sensitivity along with decreased circulating insulin (Rhoads et al., 2004).The occurrence of health problems in the early postpartum period is related, in part, with difficulty in responding to these metabolic challenges and meeting the increased demand for nutrients (Ingvartsen and Moyes, 2013).Thus, DMI can affect metabolic adaptations because it is directly related to nutrient availability, and is one of the main determinants of the magnitude of the NEB (Drackley, 1999).
A prompt and robust inflammatory response with a substantial influx of PMN in the uterus early postpartum is associated with a reduced incidence of reproductive disorders (Hammon et al., 2006;Galvão et al., 2011b;Gilbert and Santos, 2016).However, when excessive, this inflammatory response can lead to cytological endometritis (Pascottini and LeBlanc, 2020), which is characterized by increased proportion of PMN in uterine cytology (Kasimanickam et al., 2004;Gilbert et al., 2005;Dubuc et al., 2010).The specific cause for cytological endometritis has not been described, but reduction of immune function and impairment of inflammation regulatory mechanisms are fundamental players in the development of uterine diseases (Sheldon, 2019), and were reported to precede diagnosis of diseases (Hammon et al., 2006).Dairy cows face impairment of PMN function during the transition period, which is attributed to increased exposure to circulating NEFA and BHB (Ster et al., 2012;Ingvartsen and Moyes, 2013); and decreased availability of glucose (Garcia et al., 2015;LeBlanc, 2020) and calcium (Ducusin et al., 2003;Martinez et al., 2012).Therefore, DMI might be an important factor in the development of cytological endometritis because it is closely related to the metabolic adaptations leading to increased NEFA concentrations and decreased glucose availability.
Uterine health is crucial for the reproductive success of postpartum dairy cows (Fonseca et al., 1983;Sheldon, 2004).Uterine inflammatory status can affect ovarian resumption and cyclicity after parturition through inhibition of hypothalamic GnRH release and pituitary LH secretion by inflammatory mediators (Williams et al., 2001).Furthermore, the occurrence of metritis is associated with lower conception rates and prolonged calving-to-conception intervals (LeBlanc et al., 2011).Cytological endometritis between 34 and 60 DIM is also associated with lower conception rates and higher reproductive failure rates (Hammon et al., 2001;Kasimanickam et al., 2004;Gilbert et al., 2005).Few studies have evaluated the association of uterine cytology at 15 DIM with measurements of fertility.For instance, a study with 406 Holstein cows from 5 herds in the United States reported that cows with ≥8.5% of PMN at 21 DIM had a similar interval from calving to pregnancy as cows with <8.5% of uterine PMN (Galvão et al., 2009).Nevertheless, to the best of our knowledge, associations of individual DMI during the transition period, uterine cytology in the early postpartum, and ovulation of the first dominant follicle postpartum were not yet explored.Therefore, we aimed to determine the association of DMI, lactation performance, days to first ovulation, and vaginal discharge with cytological endometritis at 15 (CYT15) and 30 DIM (CYT30).Our objectives were 2-fold.First, to evaluate the association of prepartum and postpartum DMI, BW, BCS, milk yield, and milk composition, and days to first ovulation with CYT15 and CYT30.We hypothesized that these performance parameters and days to first ovulation would be associated with CYT15 as well as with CYT30, and whenever a cow would develop both, this would incur an additive negative effect.A second objective was to explore the association of vaginal discharge with CYT15 and CYT30; and performance.We hypothesized that vaginal discharge is not necessarily associated with cytological endometritis, but it is negatively associated with performance.

Experimental Design and Sample Size
The experimental design was a retrospective longitudinal study using performance and health data for 404 primiparous and multiparous Holstein cows assembled from 5 studies conducted at the University of Illinois at Urbana-Champaign (Table 1).Therefore, sample size calculations were not performed a priori.For continuous variables, a minimum of 45 cows would be required in each group to provide sufficient power (α = 0.05; β = 0.20) to detect significant associations between cytological endometritis and DMI (difference between groups of at least 1 kg/d when SD is 1.8 kg/d of DMI), milk yield (difference between groups of at least 1 kg/d when SD is 2 kg/d of milk yield), and days to first ovulation (difference between groups of 1 ± 3 d at first ovulation).Of the 404 cows originally assembled, 124 cows were removed because they had incomplete data for DMI postpartum, endometrial cytology, or primiparous.The final data set then consisted of data for 280 multiparous cows (Table 2).

Data Collection and Data Set Construction
In all studies, dry cows were housed in an enclosed ventilated barn with access to sand-bedded freestalls, where they were fed once a day using an individual feeding system (American Calan Inc.).After calving, cows were moved into a tiestall barn where they continued to be fed once a day.In all studies, cows were provided a TMR (Supplemental Table S1, https: / / uofi .box.com/s/ zklebr5asxu5300rcc4xspp9jjzzfi70), had free access to water, and were milked on a set schedule (2 or 3 times a day).Daily DMI was determined for Guadagnin and Cardoso: DRY MATTER INTAKE AND UTERINE HEALTH each cow by weighing total amount of feed and refusals and determining the difference on a DM basis.To avoid bias due to inaccuracy in records of DMI of d 0 due to a change from a prepartum diet to a postpartum diet, DMI of d 0 was not included.Weekly composited milk samples were analyzed for contents of fat, true protein, and MUN, using mid-infrared procedures (AOAC International, 1995) by a commercial laboratory (Dairy One Cooperative Inc., Ithaca, NY).Body condition score was assigned independently by 3 individuals weekly using a 5-point scale (Ferguson et al., 1994) and the average score was used for each cow.Body weight was individually measured (Ohaus digital scale, model CW-11) weekly.
Cytology of the endometrium was performed using a cytology brush (Andwin Scientific) at 15 ± 2 DIM and at 30 ± 2 DIM.Cytology slides were prepped immediately following the procedure described in Guadagnin et al. (2022).A minimum of 100 cells were counted within the individual areas (ImageJ, National Insti-tutes of Health) to determine the percentage of PMN to epithelial cells.Cytological endometritis at 15 DIM (CYT15) and at 30 DIM (CYT30) were defined based on the percentage of PMN from endometrial cytology samples.Cut-off values for samples being classified as having cytological endometritis were obtained from the median values of the data set for each specific day and used to categorize all cows as follows: LOW15 (PMN percentage at 15 DIM ≤24%; n = 125), HIGH15 (PMN percentage at 15 DIM >24%; n = 125), LOW30 (PMN percentage at 30 DIM ≤7%; n = 141), and HIGH30 (PMN percentage at 30 DIM >7%; n = 139).
Evaluations of vaginal discharge were performed at 4 ± 1, 7 ± 1, 10 ± 1, 13 ± 1, 15 ± 1, and 17 ± 1 DIM via the Metricheck device (Simcro; Metricheck score = MS) following guidelines reported by Stella et al. (2018).The vaginal discharge was scored according to smell (0 = no odor, 3 = fetid odor) and appearance: score 0 = clear or translucent mucus; score 1 = mucus containing flecks of white or off-white pus; score 2 = discharge containing ≤50% white or off-white mucopurulent material; and score 3 = discharge containing ≥50% purulent material, may be white, yellow, or sanguineous (Sheldon et al., 2006;LeBlanc and Bicalho 2017).The first postpartum follicular cycle was monitored via transrectal ultrasonography (7.5-MHz linear array probe, E.I. 141 Medical Imaging) every other day from 7 until 28 DIM.Ovulation was classified as the disappearance of the previously identified dominant follicle (DF) and the appearance of a corpus luteum in the subsequent examinations.Ultrasonography examinations were performed until the putative DF (≥10 mm in diameter) disappeared or until 28 DIM.Vaginal discharge examined using a Metricheck device (Simcro) and scored on a scale of 0 to 3: score 0 = clear or translucent mucus; score 1 = mucus containing flecks of white or off-white pus; score 2 = discharge containing ≤50% white or off-white mucopurulent material; and score 3 = discharge containing ≥50% purulent material, which may be white, yellow, or sanguineous (Sheldon et al., 2006).
Health disorders included retained placenta, hypocalcemia, displaced abomasum, and clinical mastitis.Retained placenta was defined as a placenta that failed to deliver completely more than 12 h after calf delivery; hypocalcemia, or milk fever, was defined as serum total Ca concentration ≤1.374 mM with physical signs of parturient paresis (Mulligan et al., 2006); and displaced abomasum was defined as the presence of an acute ping sound at auscultation and percussion on the left or right side of the abdomen were diagnosed by a veterinarian from the College of Veterinary Medicine at University of Illinois at Urbana-Champaign, and clinical mastitis was diagnosed by altered milk composition confirmed by positive microbiological culture.

Statistical Analyses
Mixed multivariable linear regression models were built using the MIXED procedure of SAS (version 9.4, SAS Institute Inc.).The effect of time (day or week), endometrial inflammatory status (LOW15, HIGH15, LOW30, and HIGH30), and their interaction were forced into each model to test associations with DMI, BW, BCS, milk yield, and milk composition.Independent experiment was treated as a random effect.Cow nested within experiment was the random error term.Models account for repeated measures of time [DIM was used for the variables DMI, DMI (as a percentage of BW), milk yield (k/d), and week was used for the variables BW, BCS, and milk composition].The following model was used: where Y jklm = the observations for dependent variables; μ = the overall mean; A j = the fixed effect of CYT15; B k = the fixed effect of CYT30; T l = the repeated effect of time (T); (AT) jl = the interaction of CYT15 and T; (BT) kl = the interaction of CYT30 and T; (AB) jk = the interaction of CYT15 and CYT30; (ABT) jkl = the interaction of CYT15 and CYT30 and T; C m = the random effect of experiment; and ε jklm = the random residual error.Cow was used as the experimental unit.Parity, BCS at enrollment (≤3.5 and >3.5), and retained placenta were offered as covariates and retained when P ≤ 0.05.Numerator degrees of freedom = 1.Denominator degrees of freedom was estimated using the Kenward-Rogers method (Littell, 2002).Model residuals were assessed using a scatterplot of the studentized residuals for homoscedasticity, linear predictor for linearity, and a Shapiro-Wilk test for normality.The covariance structure used was compound symmetry, due to yielding the smallest Akaike information criterion and Schwarz Bayesian criterion (Littell, 2002).A Q-Q (quantile-quantile) plot and box plot were used to determine outliers (calculated by multiplying the interquartile range by 3 and subtracting or adding from the corresponding quartile).
Simple regression models were built using the MIXED procedure of SAS.Vaginal discharge (average of evaluations performed at 4, 7, 10, 13, 15, and 17 DIM), PMN percentage at 15 DIM, PMN percentage at 30 DIM, and days to first ovulation were considered as outcomes.Dry matter intake and DMI (as a percentage of BW) by week relative to calving were considered as explanatory variables.Univariate associations for each explanatory variable with each outcome of interest were tested.Additional simple regression models considering milk yield as the outcome of interest and vaginal discharge (average of evaluations performed at 4, 7, 10, 13, 15, and 17 DIM), PMN percentage at 15 DIM, PMN percentage at 30 DIM as the explanatory variables were built.For all univariate linear regression models, cow was considered as the experimental unit and cow nested within experiment was considered as a random effect.Statistical parameters including the coefficient of determination and root mean square error were calculated to evaluate model fit.Assumptions of normality and homoscedasticity of variance were evaluated with normal probability plots and plots of residuals versus predicted values.
Development of cytological endometritis at 15 DIM, cytological endometritis at 30 DIM, and vaginal discharge (by day) were modeled by logistic regression using a binomial distribution in the GLIMMIX procedure of SAS fitting a binary distribution and a logit link function.The following model was used: where P(Y ij ) is the probability of the outcome of interest of cow i in herd j and is a function of the predictor variable through the logit function; β 0ij is the intercept; β 1 is the regression coefficient for the predictor A ij (DMI or milk yield); and μ j is the random herd effect.Odds ratio (OR) was used to compare the likelihood of high or low intake pre-and postpartum, and RP to incur in the event.Dichotomization of DMI values was performed using the median DMI prepartum (11 kg/d) and postpartum (13 kg/d).The Kenward-Roger method was used to calculate the approximate denominator degrees of freedom for the F tests in the statistical models.Goodness of fit was evaluated to test for overdispersion.
The interval from calving to ovulation of the first dominant follicle was analyzed using PHREG procedure.Cox proportional-hazards regression models included the fixed effects of endometrial cytology (LOW15, HIGH15, LOW30, and HIGH30).Cows that did not ovulate and cows that developed a follicular cyst by 30 DIM were censored.The Cox's hazard regression analysis used the following model: where logarithm of the ratio of the hazard of time [h(t)] over the basal hazard [ho(t)] was a function of the set of predictors representing the fixed effect of endometrial cytology (β 1 ) and the random effect of experiment (β 2 ).Time interval (median ± 95% CI) for days to first ovulation was obtained using the LIFETEST procedure of SAS.The assumption of the proportionality of hazard of the model was assessed graphically by plotting the logarithm of the hazard function by the logarithm of time.Residuals were evaluated for homogeneous distribution.
Data are presented as least squares means ± standard errors of the mean, unless otherwise stated.Solid diamonds (♦) within graphs represent the results of the mean separation procedure PDIFF within MIXED procedure at P < 0.05.Open diamonds (◊) within graphs represent the results of the mean separation procedure PDIFF incorporating the Tukey adjustment within MIXED procedure at 0.05 < P ≤ 0.10.Significance was declared at P ≤ 0.05, and tendencies were reported if 0.05 < P ≤ 0.10.Interactions were considered when P ≤ 0.05.

Primary Objective
To evaluate the association of prepartum and postpartum DMI (and as a percentage of BW), BW, and BCS, early postpartum milk yield and milk composition, and days to first ovulation with cytological endometritis at 15 and at 30 DIM.

Regression Analyses
Linear regression results for the association of DMI by week relative to calving with vaginal discharge score, PMN percentage at 15 DIM, PMN percentage at 30 DIM, and days to first ovulation are in Table 5.For every 1 kg increase in weekly DMI on the third week before calving, PMN percentage at 15 DIM decreased by 13 percentage units (P = 0.01; R 2 = 0.06).For every 1 kg increase in weekly DMI on the first week postpartum, vaginal discharge score decreased 0.06 units (P < 0.01; R 2 = 0.11), and PMN percentage at 30 DIM decreased by 0.72 percentage units (P = 0.05; R 2 = 0.02).For every 1 kg increase in DMI during wk 2 postpartum, the vaginal discharge score decreased by 0.03 units (P < 0.01; R 2 = 0.09).For every 1 kg increase in DMI on wk 3 postpartum, the PMN percentage at 15 DIM decreased by 0.90 percentage units (P = 0.04; R 2 = 0.15).Finally, for every 1 kg increase in DMI on wk 4 postpartum, the vaginal discharge was lower by 0.04 units (P = 0.02; R 2 = 0.09).
Linear regression results for the association of DMI (as a percentage of BW) by week relative to calving with vaginal discharge score, PMN percentage at 15 DIM, PMN percentage at 30 DIM, and days to first ovulation are in Table 6.For every unit increase in DMI (as a percentage of BW) on the fourth week before calving, the vaginal discharge score decreased by 0.45 units (P = 0.01; R 2 = 0.12).For every 1 unit increase in DMI (as a percentage of BW) on the first week before calving, PMN percentage at 30 DIM increased by 6.49 percentage units (P = 0.04; R 2 = 0.04).For every 1 unit increase in DMI as a percentage of BW) on the wk 1 postpartum, vaginal discharge score decreased by 0.43 units (P < 0.01; R 2 = 0.14).For every 1 unit increase in DMI (as a percentage of BW) on wk 3 postpartum, the PMN percentage at 15 DIM decreased by 4.45 percentage units (P = 0.03; R 2 = 0.07).For every 1 unit increase in the DMI (as a percentage of BW) on wk 4 postpartum, the vaginal discharge score was lower by 0.27 units (P = 0.02; R 2 = 0.07).
Linear regression results for the association of vaginal discharge score, PMN % at 15 DIM, and PMN % at 30 DIM with milk yield by week postpartum are in Figure 7.For every 1 unit of increase in the vaginal discharge score, the milk yield decreased by 1.84 kg/d during the first week postpartum (P < 0.01; R 2 = 0.54), 2.26 kg/d in the second week postpartum (P < 0.01; R 2 = 0.50), 2.18 kg/d for the third week postpartum (P < 0.01; R 2 = 0.49), and 1.73 kg/d in the fourth week postpartum (P < 0.01; R 2 = 0.47).For every 1 percentage unit increase in the PMN percentage at 30 DIM, milk yield decreased by 0.06 kg/d during the first week postpartum (P < 0.01; R 2 = 0.95), 0.04 kg/d for the second week postpartum (P < 0.01; R 2 = 0.95), 0.11 kg/d in the third week postpartum (P < 0.01; R 2 = 0.95), and 0.10 kg/d for the fourth week postpartum (P < 0.01; R 2 = 0.43).

Second Objective
To determine the association of vaginal discharge with endometrial cytology (CYT15 and CYT30) and performance (prepartum DMI, DMI as a percentage of BW, BW, and BCS, and postpartum DMI, DMI as a percentage of BW, BW, BCS, and milk yield).

DMI, BW, BCS, and Milk Yield
Performance variables associated with vaginal discharge are in Table 4 and Supplemental Table S2.Prepartum DMI and DMI as a percentage of BW was greater (P ≤ 0.05) for cows with a MS ≤3 at 15 DIM (12.3 ± 1.53 kg/d and 1.53 ± 0.24%, respectively) than for cows with a MS >3 at 15 DIM (9.91 ± 1.53 kg/d and 1.19 ± 0.24%, respectively).Prepartum BW was lower (P < 0.01) for cows with a MS ≤3 at 15 DIM (790 ± 11.0 kg) than for cows with MS >3 at 15 DIM (812 ± 11.0 kg).Prepartum BCS was lower (P = 0.02) for cows with a MS ≤3 at 10 and 13 DIM (3.77 ± 0.18 and 3.76 ± 0.16, respectively) than for cows with a MS >3 at 10 and 13 DIM (4.06 ± 0.18 and 4.01 ± 0.16, respectively).Postpartum DMI as a percentage of BW was lower (P = 0.04) for cows with a MS ≤3 at 10 DIM (2.14 ± 0.04%) than for cows with a MS >3 at 10 DIM (2.24 ± 0.04%); and tended to be higher (P = 0.10) for cows with a MS ≤3 at 15 DIM (2.21 ± 0.03%) than for cows with a MS >3 at 15 DIM (2.16 ± 0.03%).Postpartum BW was greater (P < 0.01) for cows with a MS ≤3 at 10 and 15 DIM (680 ± 1.88 kg and 683 ± 1.68 kg, respectively) than for cows with a MS >  Vaginal discharge was evaluated using the Metricheck device (Simcro) following guidelines reported by LeBlanc and Bicalho (2017).Discharge appearance was scored based on a scale of 0 to 3: score 0 = clear or translucent mucus; score 1 = mucus containing flecks of white or off-white pus; score 2 = discharge containing ≤50% white or off-white mucopurulent material; and score 3 = discharge containing ≥50% purulent material, which may be white, yellow, or sanguineous (Sheldon et al., 2006).Discharge smell was scored based on the absence (smell = 0) or presence (smell = 3) of a fetid odor.Sums of MS score plus smell were used.Endometrial samples collected at 15 DIM for the evaluation of PMN percentage through endometrial cytology analysis.

5
Endometrial samples collected at 30 DIM for the evaluation of PMN percentage through endometrial cytology analysis. 6 Ovulation of the first dominant follicle (DF) postpartum was monitored via transrectal ultrasonography, and ovulation was classified as the disappearance of the previously identified DF and the appearance of a corpus luteum in the subsequent examinations.

DISCUSSION
The primary objective of this study was to evaluate the association of prepartum and postpartum DMI (and as a percentage of BW), BW, and BCS, early postpartum milk yield and milk composition, and days to first ovulation with cytological endometritis at 15 and at 30 DIM.We hypothesized that performance and days to first ovulation would be associated with CYT15 as well  as with CYT30; and that whenever a cow had both CYT15 and CYT30 it would have an additive negative effect.A second objective was to associate vaginal discharge with CYT15 and CYT30 and performance.We hypothesized that vaginal discharge was not associated with uterine cytological endometritis, but it would be associated with performance.

DMI, DMI as a Percentage of BW, BW, and BCS
Prepartum and postpartum DMI (and as a percentage of BW) was associated with CYT15 from 4 wk prepartum until 4 wk postpartum, where cows in HIGH15 had lower DMI than cows in LOW15.Until now, the literature does not support a defined trigger for cytological endometritis in dairy cows.However, prepartum energy intake is associated with decreased postpartum blood neutrophil function (Graugnard et al., 2012).Graugnard et al. (2012)  Impaired PMN function characterizes immunosuppression, which is observed to some degree in peripartum dairy cows (Hammon et al., 2006;Goff, 2008).Although a physiological event, immunosuppression might be aggravated by nutritional aspects, such as the previously mentioned excess of energy intake in the prepartum period (Graugnard et al., 2012) or increased negative energy balance metabolites (Ingvartsen and Moyes, 2013).In the present study, cows were consuming diets averaging 1.45 ± 0.05 Mcal of NE L /kg of DMI during prepartum, which could be considered a moderateenergy diet (Cardoso et al., 2020).
Overconsuming energy prepartum may predispose dairy cows to health problems when DMI is limited (Cardoso et al., 2013).Thus, it is possible that the lower prepartum intake of cows in HIGH15 lead to greater mobilization of adipose tissue fatty acids, elevating blood concentration of NEFA, potentially influencing immune cells in the circulation and inflammation indices (LeBlanc, 2020).In fact, Yasui et al. (2014) reported an association of lower energy status in the first 3 wk postpartum, denoted by higher area under the curve for NEFA and BHB, with subsequent development of cytological endometritis.Migration of PMN cells is activated upon stimulation, for example by IL-8 or tumor necrosis factor α (TNFα) released from macrophages (Sheldon, 2019).Studies evaluating the effects of systemic delivering of different concentrations of bovine recombinant TNFα to dairy cows reported decreased DMI (Kushibiki et al., 2003;Bradford et al., 2009;Yuan et al., 2013), elevated plasma NEFA concentrations, increased markers of oxidative stress (Kushibiki et al., 2003), and increased liver triglyceride content (Bradford et al., 2009).Negative energy balance metabolites, such as NEFA, can also activate toll-like receptor 4, leading to secretion of TNFα and IL-8 (Ingvartsen and Moyes, 2013).At the same time, negative energy balance metabolites are associated with decreased function of PMN cells.Thus, we hypothesize that the association of lower intake pre-and postpartum and HIGH15 is possibly explained by an increased migration of PMN cells in the uterus to compensate a potential lower PMN function.
Dry matter intake was not associated with CYT30 in the prepartum or postpartum period.However, we report greater DMI (as percentage of BW) for cows in HIGH30 in comparison with cows in LOW30.As it is a relative expression of the function of both DMI and BW, it could be driven by a difference in one of the variables or in both.Similarly, Osorio et al. (2013) reported greater postpartum DMI as percentage of BW (P = 0.04) for cows receiving rumen-protected methionine (Met; 2.24 ± 0.18% of BW) in comparison with cows not receiving (Con; 1.88 ± 0.18% of BW), driven by a tendency for difference in DMI (P = 0.06; Met = 15.2 ± 1.01 kg/d and Con = 13.3 ± 1.01 kg/d) even though BW was not different (P = 0.37; Met = 660.3± 20.5 kg and Con = 670.8± 20.5 kg).Thus, is possible that the greater DMI (as a percentage of BW) for cows in HIGH30 was driven by the difference in BW for these cows, because there was no difference in DMI.Wittrock et al. (2011) used vaginal discharge associated with fever to classify cows as having metritis or not in the first 3 wk postpartum and reported that multiparous cows with metritis ate less during 3 wk after calving (12.2 ± 1.2 vs.14.0 ± 0.8 kg/d).Directly comparing our results with these of Wittrock et al. (2011), is equivocal because vaginal discharge was evaluated using a different scoring technique; however, we report no differences in postpartum DMI.Vaginal discharge evaluated at 10 and 15 DIM were associated with differences in the DMI as a percentage of BW.At 10 DIM, cows with a MS ≤3 ate less than the cows with MS >3, but when vaginal discharge is evaluated at 15 DIM, cows with MS ≤3 ate more than cows with a MS > 3.These could be indicating that at 10 DIM the vaginal discharge could be more associated with the natural reproductive tract cleansing process, which physiologically happens after calving.Conversely, when evaluated at 15 DIM, the vaginal discharge could be more associated with a uterine inflammatory process, which would concomitantly have an increased influx of PMN cells in the uterus.Additionally, we highlight that the timing of vaginal discharge evaluation is important because it affects the result.

Lactation Performance
Cytological endometritis at an earlier stage of the postpartum period (15 DIM) is associated with decreased milk yield.We attributed decreased milk yield for cows in HIGH15 to be a consequence of a decrease in DMI.A decline in milk yield was previously reported to be driven by decreased DMI (Bell and Roberts, 2007;Huzzey et al., 2007).In contrast, CYT30 was not associated with milk yield, agreeing with Dubuc et al. (2010).Dubuc et al. (2010) suggested that this lack of association occurred because the condition was localized in the reproductive tract and did not have systemic effects.However, the present study only used milk yield records until 30 DIM, thus, prospective effects of cytological endometritis on milk yield are not discarded.The lack of association of CYT30 and milk yield could be explained by the lack of association of CYT30 and DMI because, among dietary factors, milk yield is mostly driven by intake (Hristov et al., 2004;Bell and Roberts, 2007;Gohary et al., 2014).Conversely, Akbar et al. (2014) reported that cows with uterine PMN >18% between 22 and 25 DIM had lesser milk yield through the first 3 wk postpartum than cows with uterine PMN <18%.Akbar et al. (2014) suggested the lower milk yield to be, at least in part, a consequence of increased utilization of glucose by the immune system, thus limiting the glucose available for milk production.It is possible that differences in the PMN percentage cutoff values and sampling date contributed to different results.For instance, Gobikrushanth et al. (2016) reported no association of DMI and milk yield with different categories of endometritis (cytological, clinical, or both) assessed at 25 ± 1 DIM.It is plausible that DMI is contributing to explain the milk yield differences reported in Akbar et al. (2014), although DMI data were not reported by Akbar et al. (2014).
Additionally, every unit increase in MS is associated with up to 2.26 kg/d decreased milk yield in the early postpartum.Because the evaluation of vaginal discharge using MS is a tool for the on-farm diagnosis of metritis (Sheldon et al., 2006), this result agrees with previous studies reporting reduction in milk yield due to metritis (Wittrock et al., 2011;Giuliodori et al., 2013;Stangaferro et al., 2016).Although beyond the scope of the present study, it is important to consider that vaginal discharge could be originating from the uterus or cervix (LeBlanc et al., 2011;LeBlanc, 2014).Moreover, every unit increase in uterine PMN percentage at 30 DIM was associated with up to 0.11 kg/d decreased milk yield.In a study evaluating the association of cytological endometritis (classified based on PMN ≥10% evaluated at 42 DIM) with systemic energy status, Galvão et al. (2010) reported no difference in milk yield between cows with cytological endometritis and healthy primiparous cows.Healthy multiparous cows had greater milk yield than multiparous cows with cytological endometritis on the fifth and sixth weeks of lactation (Galvão et al., 2010), in corroboration with the presented association.
Differences in milk composition were associated with CYT15 and CYT30.The greater milk protein yield for cows in LOW15 at wk 3 and 4 postpartum are likely a result of the greater DMI (and as percentage of BW) observed for these cows since DMI and milk protein yield are interrelated (Roseler et al., 1997a,b;NRC, 2001).A meta-analysis reported by Hristov et al. (2004) identified DMI as the major variable influencing milk yield and milk protein yield.A greater DMI leads to greater outflow of microbial protein from the rumen, leading to greater absorption of indispensable AA and a greater supply of glucose or propionate (Rode et al., 1985;Raggio et al., 2006).Thus, milk protein yield is stimulated by both indispensable AA and insulin activation of mechanistic target of rapamycin complex 1 (mTORC1) in mammary epithelial cells (Appuhamy et al., 2014).Milk fat content was greater during wk 1 postpartum for cows in HIGH30 than for cows in LOW30, which could indicate a greater mobilization of adipose tissue.During a negative energy balance, the greater mobilization of adipose tissue for energy would result in a greater concentration of preformed fatty acids (FA) than de novo and mixed FA (Palmquist et al., 1993;Barbano et al., 2019).However, milk FA were not evaluated.Milk fat yield tended to be greater for cows in LOW15 than cows in HIGH15 on wk 4 and 5 postpartum, which is likely an effect of more available energy, since LOW15 cows had greater DMI in the postpartum period and, at 4 wk postpartum the NEB is likely resolved (Drackley, 1999).In the same way, had FA composition of milk fat being available and this hypothesis was correct, de novo and mixed FA would likely be in greater concentration due to increased acetate production in the rumen and decreased plasma NEFA (Palmquist et al., 1993;Barbano et al., 2019).Somatic cell count was greater for cows in HIGH30 at wk 4, but lower at wk 5 than cows in LOW30.Because SCC is frequently more related to events occurring in the mammary gland, we suggest there is a minimal biological relationship of these results to uterine cytology.Besides diurnal variation, mammary gland insult leading to modulation of inflammatory mediators is the main factor significantly affected SCC (Harmon, 1994, Sharma et al., 2011).

Vaginal Discharge and Days to First Ovulation
Little is known about the association of vaginal discharge in the first 2 wk postpartum and uterine PMN influx.Even less is known about whether having a persistent uterine PMN influx at 2 wk postpartum is detrimental to their reproductive performance or not.The association of MS at the first 2 wk postpartum and uterine PMN percentage at 15 DIM is variable and depends on the day of the vaginal discharge evaluation, whereas there was no association of MS and PMN percentage at 30 DIM.Most studies evaluating vaginal discharge do so to diagnose metritis in the first 2 wk postpartum because most cows develop metritis 4 to 8 DIM (Galvão, 2011a), or to diagnose purulent vaginal discharge in wk 5 postpartum (Pascottini and LeBlanc, 2020;Van Schyndel et al., 2021).For instance, a fetid watery red-brownish uterine discharge within 21 DIM, associated with signs of systemic illness and accompanied or not by rectal temperature >103°F characterizes puerperal metritis (Sheldon et al. 2006;Galvão, 2011a).In the present data set, 67 cows had fetid vaginal discharge at least once from 4 to 17 DIM, but only 20 of these cows had a rectal temperature >103°F.Therefore, it is possible that the association of MS and uterine PMN percentage is due to a delay in the uterine involution process or a persistent inflammatory response rather than puerperal metritis.
The cutoff value of 3 for the MS is based on the characteristic and odor of the vaginal mucus being associated with the uterine bacterial load, because the presence of mucopurulent, purulent, or fetid vaginal discharge is associated with presence of Arcanobacterium pyogenes, Fusobacterium necrophorum, Escherichia coli, and Mannheimia haemolytica (Williams et al., 2005).An MS >3 would mean that a fetid odor is necessarily present and, thus, the characteristic of the vaginal discharge would be more related to a bacterial contaminated uterus.However, a MS ≤3 at 10 DIM was associated with greater PMN influx in the uterus at 15 DIM, which could be reflecting physiological mechanisms of uterine involution.However, cows with a MS ≤3 also had lower DMI (as a percentage of BW) than cows with a MS >3, which indicates that the lower intake could be also affecting the uterine PMN influx as for cows in HIGH15.Still, it is notable that the evaluation of vaginal discharge at several DIM is differentially associated with uterine PMN influx.Finally, although the evaluation of vaginal discharge might not be a direct function of uterine cytology at both 15 and at 30 DIM, its association with milk yield suggests it should be a practice performed at the farm.
Cows in HIGH15 ovulated on average 3 d earlier than cows in LOW15, which was also observed regarding cows in HIGH30 and LOW30.It has been previously reported that cows with uterine infections in the first week postpartum are susceptible to have perturbed ovarian follicular growth, specifically the growth of dominant follicle into a preovulatory size follicle, which leads to delayed ovulation (Sheldon, 2004;Williams et al., 2001).The mechanisms behind uterine infections causing ovarian dysfunction appear to be related to pathogenic organisms disrupting the uterine mucosal layer and releasing LPS, which can enter the ovarian follicular fluid through the circulation and disturb ovarian cyclic events (Peter et al., 1989;Huszenicza et al., 1999;Dahiya et al., 2018).Moreover, inflammatory mediators can also disturb cyclic events through effect on the hypothalamic-pituitary-gonadal-axis (Sheldon et al., 2009;Williams et al., 2001).However, these events are usually related to delayed ovulation and development of follicular cysts, while cows in HIGH15 and HIGH30 had earlier ovulation.In a study comparing the follicular fluid in cystic ovarian follicles from cows in early lactation (≤35 DIM) and from cows in mid lactation (≥118 DIM), Lima et al. (2019) reported that cows in early lactation had disturbances in steroidogenic and metabolic markers in the follicular fluid, while cows in mid lactation had disturbances in immunological markers as well, besides steroidogenic and metabolic.Thus, it is possible that the earlier days to first ovulation in HIGH15 and HIGH30 cows is not related to their inflammatory status as much as to their metabolic status.Thus, the association of uterine inflammation and early postpartum ovarian cyclic events requires further investigation, with the inclusion of other explanatory variables related to the cows' nutritional and metabolic status.Moreover, whether these events would positively or negatively impact subsequent fertility remains to be determined.

CONCLUSIONS
We partially confirmed our hypothesis of an association of cytological endometritis and DMI and lactation performance.Cytological endometritis at 15 DIM was associated with lower DMI and a tendency for lower milk yield, while cytological endometritis at 30 DIM was associated with DMI as a percentage of BW; but it was not with milk yield.Additionally, cytological endometritis at both 15 DIM and at 30 DIM was associated with days to first ovulation, and further studies should now investigate this association with subsequent reproductive outcomes (i.e., pregnancy), taking into consideration the cows' intake.Furthermore, we confirmed our second hypothesis because the association of vaginal discharge with cytological endometritis was variable and dependent on the day of vaginal discharge evaluation.However, vaginal discharge was strongly associated with milk yield in the first 4 wk postpartum.
Guadagnin and Cardoso: DRY MATTER INTAKE AND UTERINE HEALTH

Figure 1 .
Figure 1.Association of prepartum and postpartum DMI as a percentage of BW (A) and milk yield (B) with cytological endometritis at 15 DIM.Values are LSM ± SEM.Cytological endometritis at 15 DIM (CYT15) was defined based on the percentage of PMN from endometrial cytology samples.Cut-off values for samples being classified as having cytological endometritis were obtained from the median values of the data set (>24%).Cows were classified as LOW15 (PMN % 15 DIM ≤24%) and HIGH15 (PMN % 15 DIM >24%).Cows in LOW15 had greater DMI (P < 0.01) and tended to have greater milk yield (P = 0.08) than cows in HIGH15.
Figure 2. Association of prepartum and postpartum DMI (as a percentage of BW; A) and milk yield (B) with cytological endometritis at 30 DIM.Values are LSM ± SEM.Cytological endometritis at 30 DIM (CYT30) was defined based on the percentage of PMN from endometrial cytology samples.Cut-off values for samples being classified as having cytological endometritis were obtained from the median values of the data set (>7%).Cows were classified as LOW30 (PMN % 30 DIM ≤7%) and HIGH30 (PMN % 30 DIM >7%).Cows in LOW30 had lesser DMI (P < 0.01) than cows in HIGH30.Milk yield did not differ between cows in LOW30 and in HIGH30 (P = 0.87).

Figure 6 .
Figure 6.Survival curves for cytological endometritis at 15 DIM (A) and cytological endometritis at 30 DIM (B) across days to first ovulation.The y-axis represents the overall probabilities.Cytological endometritis at 15 DIM (CYT15) and at 30 DIM (CYT30) were defined based on the percentage of PMN from endometrial cytology samples.Cut-off values for samples being classified as having cytological endometritis were obtained from the median values of the data set for each specific day and were as follows: 15 DIM, >24%; 30 DIM, >7%.The x-axis represents the number of days post-calving.Cows in LOW15 ovulated on average 3 d later than cows in HIGH15, and cows in LOW30 ovulated on average 3 d later than cows in HIGH30 (log rank P < 0.01).

Figure 7 .
Figure 7. Associations of postpartum vaginal discharge or endometrial cytology with milk yield (kg/d) in the first 4 wk postpartum of Holstein cows.Evaluation of vaginal discharge was performed using Metricheck device (Simcro) at 4, 7, 10, 13, 15, and 17 DIM, following guidelines reported by LeBlanc and Bicalho (2017).Endometrial cytology samples were collected at 30 DIM and evaluated for the percentage of PMN.Scatter diagrams display the specific association of the following: postpartum Metricheck score and smell and milk yield on the first week postpartum (A), milk yield on the second week postpartum (B), milk yield on the third week postpartum (C), and milk yield on the fourth week postpartum (D); and uterine PMN percentage at 30 DIM and milk yield on the first week postpartum (E), milk yield on the second week postpartum (F), milk yield on the third week postpartum (G), and milk yield on the fourth week postpartum (H).Data from 5 different studies were combined.Model accounted for the random effects of cow nested within each experiment.

Figure 7 (
Figure 7 (Continued).Associations of postpartum vaginal discharge or endometrial cytology with milk yield (kg/d) in the first 4 wk postpartum of Holstein cows.Evaluation of vaginal discharge was performed using Metricheck device (Simcro) at 4, 7, 10, 13, 15, and 17 DIM, following guidelines reported by LeBlanc and Bicalho (2017).Endometrial cytology samples were collected at 30 DIM and evaluated for the percentage of PMN.Scatter diagrams display the specific association of the following: postpartum Metricheck score and smell and milk yield on the first week postpartum (A), milk yield on the second week postpartum (B), milk yield on the third week postpartum (C), and milk yield on the fourth week postpartum (D); and uterine PMN percentage at 30 DIM and milk yield on the first week postpartum (E), milk yield on the second week postpartum (F), milk yield on the third week postpartum (G), and milk yield on the fourth week postpartum (H).Data from 5 different studies were combined.Model accounted for the random effects of cow nested within each experiment.

Figure 7 (
Figure 7 (Continued).Associations of postpartum vaginal discharge or endometrial cytology with milk yield (kg/d) in the first 4 wk postpartum of Holstein cows.Evaluation of vaginal discharge was performed using Metricheck device (Simcro) at 4, 7, 10, 13, 15, and 17 DIM, following guidelines reported by LeBlanc and Bicalho (2017).Endometrial cytology samples were collected at 30 DIM and evaluated for the percentage of PMN.Scatter diagrams display the specific association of the following: postpartum Metricheck score and smell and milk yield on the first week postpartum (A), milk yield on the second week postpartum (B), milk yield on the third week postpartum (C), and milk yield on the fourth week postpartum (D); and uterine PMN percentage at 30 DIM and milk yield on the first week postpartum (E), milk yield on the second week postpartum (F), milk yield on the third week postpartum (G), and milk yield on the fourth week postpartum (H).Data from 5 different studies were combined.Model accounted for the random effects of cow nested within each experiment.

Figure 7 (
Figure 7 (Continued).Associations of postpartum vaginal discharge or endometrial cytology with milk yield (kg/d) in the first 4 wk postpartum of Holstein cows.Evaluation of vaginal discharge was performed using Metricheck device (Simcro) at 4, 7, 10, 13, 15, and 17 DIM, following guidelines reported by LeBlanc and Bicalho (2017).Endometrial cytology samples were collected at 30 DIM and evaluated for the percentage of PMN.Scatter diagrams display the specific association of the following: postpartum Metricheck score and smell and milk yield on the first week postpartum (A), milk yield on the second week postpartum (B), milk yield on the third week postpartum (C), and milk yield on the fourth week postpartum (D); and uterine PMN percentage at 30 DIM and milk yield on the first week postpartum (E), milk yield on the second week postpartum (F), milk yield on the third week postpartum (G), and milk yield on the fourth week postpartum (H).Data from 5 different studies were combined.Model accounted for the random effects of cow nested within each experiment.
Guadagnin and Cardoso: DRY MATTER INTAKE AND UTERINE HEALTH Guadagnin and Cardoso: DRY MATTER INTAKE AND UTERINE HEALTH

Table 1 .
Guadagnin and Cardoso: DRY MATTER INTAKE AND UTERINE HEALTH Description of treatments, diets, and cows used in experiments to evaluate the association of uterine health and follicular dynamics with milk composition, dietary energy intake, dietary metabolizable protein intake, and supplementation of rumen-protected AA during pre-and postpartum period 1 Number of cows.2Diet characteristics are main ingredients to provide a generalized view of each experimental diet.Full details on all diets can be found in published papers, thesis, or dissertation of each experiment.

Table 2 .
Descriptive statistics of 280 Holstein cows 1 in 5 experiments from 30 d before calving until 30 DIM 2

Table 3 .
Guadagnin and Cardoso: DRY MATTER INTAKE AND UTERINE HEALTH Guadagnin and Cardoso: DRY MATTER INTAKE AND UTERINE HEALTH Least squares means and associated SEM for DMI, BW, BCS, and production parameters of Holstein cows classified according to the endometrial PMN percentage at 2Greatest value of SEM within classification.3

Table 4 .
Least squares means and associated SEM for DMI, BW, BCS, milk yield, and PMN percentage of Holstein cows classified according to their vaginal discharge (Metricheck 1 Rows indicate x-axis variables, and columns indicate y-axis variables.MS at 4 DIM differed only for DMI (% BW), where 2

Table 6 .
Simple regression of DMI as a percentage of BW (DMI, % BW), averaged by week relative to calving, to vaginal discharge, endometrial PMN percentage at 15 DIM, PMN percentage at 30 DIM, and days to first ovulation of multiparous Holstein cows 1 DMI by week (% BW) LeBlanc and Bicalho (2017)ables, and columns indicate y-axis variables.Models accounted for the random effects of experiment and cow.2 Evaluation of vaginal discharge was performed using Metricheck device (Simcro) at 4, 7, 10, 13, 15, and 17 DIM, following guidelines reported byLeBlanc and Bicalho (2017).The average of the sums of Metricheck score plus smell across the indicated days were used.3Root mean square error.

Table 7 .
Logistic regression model, accounting for the random effect of experiment, of variables (by class) associated with subsequent risk of cytological endometritis at 15 DIM in Holstein cows 2Vaginal discharge characterized by the sum of Metricheck score (MS; Simcro) + vaginal discharge smell >3 (LeBlanc and Bicalho, 2017).

Table 8 .
Logistic regression model, accounting for the random effect of experiment, of variables (by class) associated with subsequent risk of cytological endometritis at 30 DIM in Holstein cows reported lower postpartum blood PMN phagocytosis capacity in cows fed a prepartum diet formulated for 1.62 Mcal/kg of NE L / kg of DMI in comparison with cows fed a prepartum diet formulated for 1.34 Mcal/kg of NE L /kg of DMI.