Relationship Between Cytologically Determined Early Lactation Hepatic Lipid Content and Energy Balance, Health, and Milk Production in Grazing Dairy Cows

Simple Summary

A significant proportion of dairy cows suffer from fat accumulation in liver cells during calving. This condition affects milk composition and predisposes cows to diseases such as ketosis, hypocalcemia, and metritis. This study aims to improve the understanding of how liver fat accumulation relates to the health and production of grazing dairy cows. By examining liver cells using fine needle biopsy at ten days postpartum as well as measuring blood metabolites and recording health and milk production outcomes from a grazing herd in Chile, we observed that moderate or severe fat accumulation in liver cells was associated with an increased risk of ketosis, while severe accumulation was associated with high milk production but lower protein content in the milk. These findings are consistent with previous studies on housing and grazing. In addition, we suggest that fine needle biopsy is a practical, minimally invasive, and cost-effective method for assessing liver fat accumulation, with potential applications in herd monitoring.

Abstract

The aim of this study was to evaluate the relationship between cows’ hepatic lipid content (HLC) at 10 days in milk (DIM) and their metabolic status, health, and production during transition and early lactation periods. HLC was determined in 103 cows from a grazing Chilean dairy herd via cytologic examination of the liver through fine needle biopsies, categorized as mild, moderate, or severe. Blood metabolites were evaluated in the transition period, together with diseases in the postpartum period and milk production during the first 90 DIM. In pre-partum and postpartum periods, primiparous cows with severe HLC showed higher plasma cholesterol than multiparous cows with mild HLC. Postpartum, cows with severe HLC had higher serum non-esterified fatty acids (NEFAs) and NEFA/cholesterol ratios than those with mild HLC. Similarly, cows with moderate and severe HLC presented higher plasma β-hydroxybutyrate and greater risk of subclinical ketosis than cows with mild HLC. Additionally, cows with severe HLC had higher milk production and lower milk protein content than those with mild HLC. These results indicate that moderate to severe HLC at 10 DIM was associated with negative energy balance and subclinical ketosis, whereas severe HLC was associated with increased milk production and decreased milk protein content.

1. Introduction

During the transition period, from three weeks before to three weeks after calving, various metabolic, endocrine, and immunological pathways are involved, including homeorhetic mechanisms, all aimed at adapting from gestation to lactation [1]. After calving, energy demands increase with the onset of lactation, triggering a negative energy balance (NEB) and adipose tissue mobilization [2]. Consequently, increased amounts of non-esterified fatty acids (NEFAs) are released and metabolized by the liver [2]. The hepatocyte’s ability to completely oxidize NEFAs is limited, leading to increased ketone body formation or re-esterification to triacylglycerols (TAG; [3]). Excessive hepatic accumulation of TAG results in hepatic lipidosis (HL), a syndrome ranging from subclinical to severe [4]. Some authors indicate that HL has become increasingly common and often more severe in dairy cows during the transition period, and that the criteria used to classify hepatic lipidosis severity may need to be reconsidered [5].

According to available data, the prevalence of HL in pasture-based systems has only been reported in New Zealand [6]. In a study by Spaans et al. [6], median liver TAG (% wet weight) at the sampling times closest to what was used in our study were 2.5% at 7 DIM (minimum = 0.3%, maximum = 16.9%) and 2.2% at 14 DIM (minimum = 1.1%, maximum = 11.7%); at no point between −30 and 28 DIM did median liver TAG exceed 2.5%. However, the data in the postpartum period were skewed (mean values were consistently higher than median values), and a subset of cows developed liver TAG of 5–10% (maximum prevalence of 13% at 28 DIM) or >10% (maximum prevalence of 5%, also at 28 DIM).

Several studies in housed systems have established associations between hepatic lipid content (HLC), measured as the TAG concentration, and biochemical indicators of energy balance. It has been reported that as the percentage of hepatic TAG increases, blood concentrations of NEFAs and β-hydroxybutyrate (BHB) also increase, and blood cholesterol concentrations decrease [7,8,9]. Although dairy cows often experience a period of NEB during early lactation, the severity of this condition may be greater in pasture-based systems, where dry matter intake is usually restricted [10]. Therefore, significant variations in the relationship of NEB indicators and the presence of HL in grazing dairy cows during the transition period could be expected.

The HLC has been reported to be correlated with a higher incidence of diseases such as ketosis, hypocalcemia, or clinical metritis [8,11,12,13]. Additionally, grazing systems are often associated with a high incidence of metabolic diseases in transition cows, highlighting the importance of this issue [14]. Thus, it is worth noting that research on HL in pasture-based dairy cows remains limited.

The relationship between HLC and milk composition parameters, such as milk fat and protein, has not been fully described, except for two studies conducted on housing dairy systems [8,15]. While Jorritsma et al. [15] found that hepatic TAG accumulation does not negatively affect productive performance during early lactation, more recently, Arshad and Santos [8] reported that milk production rose when hepatic TAG content increased from 2.5 to 7.5% and decreased in severe HL (≥7.5%). Thus, the impact of HL on the milk production performance of grazing dairy cows remains unknown; it is conceivable that productive losses in milk production or composition occur, thereby reducing the farm’s profitability.

The present study is a logical extension of previous work showing that HLC can be estimated cytologically—a faster and less expensive, invasive, and technically demanding method than the “gold standard” method of measuring HLC by performing chemical extraction from samples obtained via conventional biopsy—and can easily be applied to herd management [16]. The purpose of the present study is to improve understanding of how cytologically determined HLC in grazing dairy cattle during early lactation relates to clinically and economically relevant outcomes and to potentially help develop guidelines for using liver cytology to inform herd management decisions. The specific objective is to evaluate the relationship between cows’ HLC at 10 DIM and their measured metabolic status, health, parity, and milk production during the transition and early lactation period. The null hypothesis is that HLC would eventually have no significant relationship with any of the measured outcomes.

2. Materials and Methods

This longitudinal study was conducted during two consecutive spring calving seasons at the Austral Agricultural Research Station, at Universidad Austral de Chile, in the city of Valdivia (39°47046″ S, 73°13013″ W). The experiment was approved by the Animal Care and Use Committee of the Universidad Austral de Chile (Protocol N° 459, 23 May 2022), and all procedures were performed in accordance with the farm’s management. The dairy herd was managed under standard grazing conditions in Southern Chile over two calving seasons: autumn (February to April) and spring (July to September).

2.1. Animal Handling and Study Design

This study was an extension of our previous work supporting fine needle biopsy (FNB) cytology as a practical, minimally invasive addition to the diagnostic toolkit for herd management [16]. Because the study objective was to evaluate the relationship between HLC at 10 DIM and a spectrum of clinicoeconomically relevant outcomes (metabolic status, health, production), we enrolled the maximum possible number of cows from our grazing herd, instead of calculating the number of samples required to demonstrate a statistically significant difference for a particular metric. The assumptions considering the hypotheses α (two-tailed) = 0.05, β = 0.20, r = 0.27 were used.

A total of 105 Holstein Friesian primiparous (PP) and multiparous (MP) cows were included in this study, corresponding to two cohorts from 2022 (12 PP and 48 MP; year 1) and 2023 (17 PP and 28 MP; year 2), which calved during the spring calving season. Parity was 3 [1,2,3,4,5], and body condition score was 3.5 [3.25–3.5] (median [interquartile range; IQR]) at calving. Body weight was 497 ± 71.6 kg (mean ± standard deviation; SD) at 10 days in milk (DIM). A clinical examination of each cow was conducted at the time of enrollment, three weeks prior to the expected calving date. Only clinically healthy cows were included.

Pre-partum cows were housed in a paddock without pasture (i.e., a bare soil surface paddock) with 8.2 cows/ha as the stocking rate. The study subject received grass silage, commercial concentrate with anionic salts (

Table 1. Nutritional composition of diets supplied to dairy cows in transition during the spring calving season in years 1 and 2.

	Year	DM (%)	CP (%)	NDF (%)	ADF (%)	ME (Mcal/kg)
Pre-partum
Grass silage 1	1	41.9	5.6	68.6	45.4	2.0
Grass silage 1	2	49.9	5.2	72.5	47.2	1.4
Concentrate 2	Both 3	87	23	23	—	3.0
Postpartum
Grass silage 1	1	35.8	9.5	61.3	39.0	2.2
Fresh pasture	1	15.9	21.9	51.3	33.4	2.9
Grass silage 1	2	59.4	10.1	60.5	40.5	1.7
Fresh pasture	2	20.8	17.8	48.6	25.7	2.5
Concentrate	Both 3	87	14	22	11	—

¹ The grass silage is mainly composed of the following species: Dactylis glomerata, Bromus spp., Lolium perenne, Festuca arundinacea, and Holcus lanatus L. ² Inclusion of anionic salts at 3% per kg of concentrate, with a DCAD of −675 mEq/kg of concentrate. Mineral composition per 1000 g of mineral salt: 170 g of Cl; 100 g of S; 60 g of Mg; 3 g of Na; 1000 mg of Fe max; 1900 mg of Cu; 2500 mg of Mn; 4500 mg of Zn; 200 mg of I; 15 mg of Se; 3 mg of organic Se; 1500 IU of vitamin E. ³ During the two spring seasons, the same commercial concentrates were used before and after calving. All components were analyzed using near-infrared spectroscopy (NIRS). DM: dry matter, CP: crude protein, NDF: neutral detergent fiber, ADF: acid detergent fiber, ME: metabolizable energy. Cells with “—” indicate not applicable or no data available.

), and fresh water supplied ad libitum from a water trough. Cows were calved in the same place where they were managed and, subsequently, transferred to the lactation group. Cows were milked twice a day, at 05:00 and 14:00 h, in a milking parlor, at which time they were fed 5 kg/cow/day of commercial concentrate. After milking, cows were moved to a feedyard and fed 16 kg DM/cow/day of grass silage (Table 1) when the pasture was not available. Afterward, cows were moved to a pasture in a rotational grazing system and remained there until the next milking. In months when the pasture was abundant (September), cows grazed over an area of 2 ha/day of pasture with a fresh grass intake composed of mixed species of grasses (mainly Lolium perenne, Bromus sp., and Dactylis glomerata L.) and legumes (Trifolium repens and Trifolium pratense) at a rate of approximately 12.5 kg DM/day.

2.2. Liver Fine Needle Biopsy Procedure, Sample Processing, and Cytological Analysis

Each cow was positioned in a cattle handling chute and maintained in a partially immobilized state through a head-locking mechanism. A liver FNB was performed in the right eleventh intercostal space [17,18]. First, a trichotomy was carried out within a radius of three to five cm from the estimated puncture site, followed by an ultrasound confirmation using a 2.5 MHz micro-linear transducer to pinpoint the exact location of the puncture. The image with the fewest blood vessels was considered adequate. Second, a surgical antisepsis was performed, involving alternating three applications of 70° alcohol and three applications of povidone–iodine. Then, local anesthesia was induced via subcutaneous injection of 1 mL of Bupivacaine hydrochloride (Richmond Vet Pharma, Buenos Aires, Aregentina) administered above the FNB site. Later, a small incision of less than 1 cm was carefully made through the skin with a scalpel to reach the subcutaneous space. An 18G × 3.5″ spinal needle (NIPRO^®, Osaka, Japan) was inserted toward the left scapulohumeral joint, passing through intercostal muscles and peritoneum until reaching the liver (2.5″ of the needle). A stylet was subsequently removed to allow for complete needle insertion, which was then partially withdrawn and slightly redirected to obtain tissue from a different area, and the needle was fully reinserted a maximum of 3 times to obtain the sample. After the needle was removed, the incision was treated with a topical antiseptic. Two trained veterinarians carried out all procedures, which took a mean of 3′54″(±0.02) min per cow.

After completing the FNB, the needle was connected to a syringe containing at least 5 mL of air to gently eject the sample onto a glass slide. The sample was then smoothly spread between two slides, dried at room temperature, and identified using the last four digits of each cow’s official individual identification device. Later, it was stained using Hemacolor (Merck Chemicals, Darmstadt, Germany) and evaluated via cytological analysis by a trained observer to determine the HLC. The HLC was scored from 0 to 4 according to an ordinal scale proposed by Fry et al. [16]. This scoring system considers the estimated percentage of hepatocytes with cytoplasmic vacuolation and the mean percentage of the cytoplasm occupied by the lipid vacuoles. For the present study, HLC was categorized as mild (score ≤ 1), moderate (score = 2), or severe (score ≥ 3).

2.3. Blood Sampling and Analysis

The pre-partum sampling scheme was established according to the estimated calving date. During the pre-partum period (3 weeks before calving), blood samples were obtained from each cow once a week, on days −18 (±2), −11 (±2), and −4 (±2) relative to calving. Blood samples were collected within 24 h after calving and then at 6 or 7, 10 (±2), and 19 (±1) days after calving. Samples were obtained via coccygeal venipuncture between 08:00 and 09:00 h using sterile vacuum tubes with heparin and without additives. Serum and plasma were extracted within the next hours and later stored in microtubes at −20 °C for subsequent analysis. Commercial reagents were used for metabolite analysis. The serum NEFA concentration (NEFA, Randox Laboratories, Crumlin, Ireland), plasma BHB concentration (RANBUT, Randox Laboratories), cholesterol (Cholesterol liquicolor, Human Laboratories, Wiesbaden, Germany), and total calcium concentration (Ca-Color Arsenazo III AA, Wiener Lab, Rosario, Argentina) were measured using a Metrolab 2300 autoanalyzer (Metrolab S.A., Rosario, Argentina). The magnesium concentration was determined at 285.2 nm using an atomic absorption spectrophotometer (Thermo^®, Series AA Solaar, Whaltam, MA, USA). All analyses were performed at the Veterinary Clinical Pathology Laboratory of Universidad Austral de Chile.

2.4. Health Status Assessment

On the day of calving and the day after, each cow was monitored for retained fetal membranes and clinical hypocalcemia. Then, from day 3 to day 14 after calving, with an interval of 3 to 4 days between visits, each animal was clinically examined for the diagnosis of metritis and lameness. In addition, blood concentrations of total calcium, magnesium, and BHB, obtained from the four blood samples per animal in the postpartum period, were used for the diagnosis of hypocalcemia, hypomagnesemia, and subclinical ketosis. All health checks were performed by two trained veterinarians. To monitor clinical mastitis cases, the milker assessed each cow twice daily at the beginning of each milking from calving to 21 DIM.

Clinical and puerperal metritis, defined according to Sheldon et al. [19], were determined via vaginal examination with the metricheck^® (Hamilton, New Zealand) device and rectal temperature evaluation according to Huzzey et al. [20]. After cleaning and disinfection of the vulva with 10% iodine solution, the device was introduced into the vaginal cavity up to the cervix to collect vaginal discharge (VD), which was evaluated and classified according to appearance and odor into 5 categories: VD= 0 with clear mucus or no discharge; VD = 1 with bloody mucus or pus foci; VD = 2 with <50% pus, plus foul odor with or without fever; VD = 3 with >50% pus, plus foul odor with or without fever; and VD = 4 with a brownish/reddish discharge, plus putrid foul odor. The absence of metritis was considered when a cow presented a VD ≤ 1; clinical metritis when a cow presented a VD = 2 or 3 at least once but never a VD of 4, and puerperal metritis when a cow presented a VD = 4 at least once.

Lameness was evaluated according to the 0 to 3 ordinal scoring system for mobility described by Reader et al. [21]. Mobility scores 0, 1, 2 and 3 were considered sound, imperfect, impaired, and severely impaired, respectively. For data analysis, each cow was assigned to the highest mobility score it received during the evaluation period.

Retained fetal membranes was described as the incapacity to expel fetal membranes within 12 to 24 h after delivery [22]. Furthermore, clinical puerperal hypocalcemia was defined as any cow in recumbency whose serum calcium concentration was less than 2.0 mmol/L after calving [23]. Finally, clinical mastitis was determined as the presence of clinical signs, such as abnormal milk or swelling in 1 or more quarters [24], detected and recorded at milking.

Subclinical hypocalcemia was defined as a serum calcium concentration < 2.15 mmol/L within the first 24 h postpartum [25], while subclinical hypomagnesemia and subclinical ketosis were defined as a magnesium concentration ≤ 0.65 mmol/L [26] and a BHB concentration ≥ 1.2 mmol/L [27], respectively, in at least 1 of the 3 postpartum blood samples. The diagnosis of each of these conditions was based on the absence of clinical signs.

2.5. Milk Production Data

Individual milk production data were recorded daily with a flow sensor (MPC580 DeLaval, Tumba, Sweden) and exported to computer software (DelPro^® Software v.5.3, DeLaval, Tumba, Sweden). The production record was collected using the software from day 7 to day 90 after calving. Cumulative production was established between 7 and 21 DIM, 22 and 60 DIM, 61 and 90 DIM, and for the entire period (7 to 90 DIM). Standardized lactation curves were set using the Wilmink Model projected to 305 DIM. Finally, models were established based on three parameters obtained from each animal’s production record, following the procedure described by Castillo-Umaña et al. [28], considering the following model:

Yt = a + (be)^(−0.05t) + ct

(1)

where a is the parameter defining the curve’s height, b is the degree of slope before peak production, and c is the degree of decline in the postpeak phase in t days. The curve parameters were then compared across HLC categories.

2.6. Milk Sampling and Composition Parameters

Milk samples were collected from each cow in both milkings at (mean ± SD) 31 ± 2.07 DIM, 60 ± 1.96 DIM, and 90 ± 2.34 DIM to determine milk fat and protein contents (%). Samples of 80 to 100 mL were collected using lactometers (Waikato MK V, Waikato, New Zealand) to ensure a representative sample for the entire milking process. The morning milking sample was stored in a sealed flask at 3 to 5 °C until the afternoon milking sample was obtained. A composite sample was created by combining milk from both morning and afternoon milkings in proportions reflecting the relative amounts produced at each milking. The composite sample was stored in flasks with bronopol and labeled using the last four digits of each animal’s official individual identification device. Milk fat and protein were then analyzed using the infrared method outlined in IDF 141C:2000 Milkoscan in the Milk Quality Laboratory from Cooprinsem^® (Osorno, Chile).

2.7. Statistical Analysis

Two cows, both MP, were removed from this study due to insufficient hepatocytes in the FNB sample for cytologic evaluation. Therefore, the final sample amount was 103 animals (29 PP, 74 MP). However, data on blood metabolites during the pre-partum period were incomplete for all animals. At week −3, −2, and −1, 77, 95, and 97 cows were sampled, respectively. In the case of repeated measurements, only those cows for which all the requisite measures were available were included in the analysis for each variable.

Linear mixed models (2) were employed to ascertain the relationship between HLC and the response variables: BHB (postpartum, n = 103), NEFAs (pre-partum, n = 102; postpartum, n = 103), cholesterol (pre-partum, n = 102; postpartum, n = 103), milk protein percentage (n = 95), milk fat percentage (n = 95), and cumulative milk production (n = 96). The full model considered the degree of HLC (mild, moderate, and severe; G), year (1 and 2; Ye), parity (primiparous and multiparous; Par), month of calving (July, August, and September; MC), sampling (Samp), and body condition score at calving (5-point scale, [29]; BCSC). Three interaction terms were also included: HLC × parity (G × Par), HLC × month of calving (G × MC), and HLC × week of sampling (G × Samp). All fixed effects were fitted, with cow nested within HLC group included as a random effect. Additionally, the presence of ketosis during the postpartum transition period was incorporated into the milk fat model.

Y_ijklmno = µ + G_i + Cow (G_i)_j + Ye_k + Par_l + MC_m + Samp_n + BCSC_o + (G × Par)_il + (G × MC)_im + (G × Samp)_in + e_ijklmno

(2)

The models were submitted to stepwise backward elimination, with non-significant factors removed. The final model was selected based on the lowest Akaike information criterion. The distribution of the residuals was then assessed through histograms and Q-Q plots. Since the residuals for the NEFAs, NEFA/cholesterol ratio, cholesterol, and BHB variables did not follow a normal distribution, a natural logarithmic transformation was applied. However, this transformation did not alter the significance of the estimated effects or the interpretation of the results. Therefore, to preserve the original scale of the estimates (β coefficients), the results were reported using the untransformed data. Lactation Wilmink curve parameters (n = 82), days at peak (n = 82), and milk yield at peak (n = 82) were analyzed using linear regression models, considering HLC degree, parity, and month of calving as fixed factors.

Regarding the health status-related analysis, the data were transformed into binary responses of “presence = 1” and “absence= 0” for clinical and subclinical disease. Due to the low frequency of clinical diseases, such as retained fetal membranes, mastitis, hypocalcemia, and severe lameness, they were not subjected to statistical evaluation. The presentation of clinical (clinical and puerperal metritis, n = 103 each) and subclinical diseases (hypomagnesemia, n = 103; ketosis, n = 103; and hypocalcemia, n = 102) was analyzed using logistic regression models, in which year, parity, body condition score at calving, month of calving, and HLC were included as explanatory variables. Model selection was performed using the backward elimination method, comparing the models using the likelihood ratio test.

p-values < 0.05 were considered statistically significant. All analyses were performed using the latest version of RStudio (RStudio 4.4.1 ver. R Foundation for Statistical Computing. www.r-project.org. Vienna, Austria).

3. Results

3.1. Cytologic Assessment of HLC

Table 2. The distribution of dairy cows as HLC categories, stratified by parity.

Parity	Mild HLC		Moderate HLC		Severe HLC
	n	%	n	%	n	%
PP	17	58.6	10	34.5	2	6.9
MP	37	50	15	20.3	22	29.7
Total	54	52.4	25	24.3	24	23.3

shows the distribution of dairy cows according to their HLC and parity. More than half of the dairy cows had mild HLC, followed by moderate and severe HLC.

3.2. HLC and Blood Metabolic Biomarkers

Significant relationships between HLC and serum or plasma metabolic biomarkers in the pre-partum period are shown in

Table 3. Variables retained in the final linear mixed models * (mean ± SD and estimate) describing the relationships of parity, month of calving, week of sampling, and HLC with blood metabolites in grazing dairy cows during the pre-partum period.

Blood Metabolite	Variable	Category	Mean ± SD	Estimate	p-Value
NEFAs (µmol/L)	Parity	MP	295 ± 221	Ref. 1
		PP	419 ± 302	108.4	<0.01
	Week of sampling	−3	257 ± 205	Ref.
		−2	316 ± 245	49.5	>0.1
		−1	398 ± 273	126.0	<0.01
	Month of calving	July	388 ± 290	Ref.
		August	252 ± 196	−119.6	<0.01
		September	286 ± 158	−62.6	>0.1
NEFA/cholesterol (ratio)	Parity	MP	0.1 ± 0.1	Ref.
		PP	0.2 ± 0.1	0.03	0.02
	Week of sampling	−3	0.1 ± 0.1	Ref.
		−2	0.1 ± 0.1	0.02	>0.1
		−1	0.2 ± 0.1	0.06	<0.01
	Month of calving	July	0.2 ± 0.1	Ref.
		August	0.1 ± 0.1	−0.04	<0.01
		September	0.1 ± 0.1	−0.03	>0.1
Cholesterol (mmol/L)	Week of sampling	−3	3.0 ± 0.6	Ref.
		−2	2.9 ± 0.6	−0.10	>0.1
		−1	2.6 ± 0.5	−0.40	<0.01
	Parity × HLC	MP × Mild HLC	2.8 ± 0.5	Ref.
		PP × Moderate HLC	3.0 ± 0.7	−0.06	>0.1
		PP × Severe HLC	3.8 ± 0.8	0.97	0.02

* The R² values for the NEFAs, NEFA/cholesterol ratio, and cholesterol models were 0.14, 0.13, and 0.60, respectively. ¹ Reference.

. Non-esterified fatty acid concentrations and the NEFA/cholesterol ratio were influenced by parity, week of sampling, and month of calving. Cholesterol concentrations decreased as calving approached. Furthermore, PP cows with severe HLC had significantly higher cholesterol concentrations than MP cows with mild HLC.

Table 4. Variables retained in the final linear mixed models * (mean ± SD and estimate) describing the relationships of parity, month of calving, week of sampling, and HLC with blood metabolites in grazing dairy cows during the postpartum period.

Blood Metabolite	Variable	Category	Mean ± SD	Estimate	p-Value
NEFAs (µmol/L)	Parity	MP	400 ± 237	Ref. 1
		PP	458 ± 188	90.4	<0.01
	Week of sampling	1	472 ± 217	Ref.
		2	436 ± 215	−35.6	>0.1
		3	340 ± 227	−131.3	<0.01
	HLC	Mild	368 ± 189	Ref.
		Moderate	422 ± 202	46.9	>0.1
		Severe	518 ± 285	171.3	<0.01
NEFA/cholesterol (ratio)	Parity	MP	0.2 ± 0.1	Ref.
		PP	0.2 ± 0.1	0.02	0.01
	Week of sampling	1	0.2 ± 0.1	Ref.
		2	0.2 ± 0.1	−0.04	<0.01
		3	0.1 ± 0.1	−0.11	<0.01
	HLC	Mild	0.2 ± 0.1	Ref.
		Moderate	0.2 ± 0.1	0.01	>0.1
		Severe	0.2 ± 0.1	0.06	<0.01
Cholesterol (mmol/L)	Week of sampling	1	2.3 ± 0.4	Ref.
		2	2.6 ± 0.5	0.33	<0.01
		3	3.4 ± 0.7	1.1	<0.01
	Month of calving	July	2.7 ± 0.8	Ref.
		August	2.9 ± 0.7	0.24	0.02
		September	2.8 ± 0.6	0.19	0.07
	Parity × HLC	MP × Mild HLC	2.7 ± 0.7	Ref.
		PP × Moderate HLC	2.6 ± 0.6	−0.41	0.07
		PP × Severe HLC	3.8 ± 1.1	0.97	0.03
BHB (mmol/L)	Month of calving × HLC	July × Mild HLC	0.9 ± 0.5	Ref.
		August × Moderate HLC	0.9 ± 0.4	−0.13	>0.1
		September × Moderate HLC	0.5 ± 0.2	−0.43	0.04
		August × Severe HLC	1.1 ± 0.5	−0.45	0.10
		September × Severe HLC	0.8 ± 0.3	−0.70	0.09
	Week of sampling × HLC	1 × Mild HLC	0.8 ± 0.4	Ref.
		2 × Moderate HLC	1.0 ± 0.5	−0.10	>0.1
		2 × Severe HLC	1.3 ± 0.6	−0.07	>0.1
		3 × Moderate HLC	1.0 ± 0.4	0.09	>0.1
		3 × Severe HLC	1.5 ± 1.2	0.40	0.01
	HLC	Mild	0.8 ± 0.4	Ref.
		Moderate	1.0 ± 0.5	0.28	<0.01
		Severe	1.4 ± 0.9	0.87	<0.01

* The R² values for the NEFAs, NEFA/cholesterol ratio, cholesterol, and BHB models were 0.37, 0.48, 0.71, and 0.46, respectively. ¹ Reference.

shows the relationship between HLC and blood metabolic biomarkers in the postpartum period. Dairy cows with severe HLC exhibited higher concentrations of NEFAs and a greater NEFA/cholesterol ratio compared to those with mild HLC. These metabolites remained elevated during week 1, showing a progressive decrease toward week 3 (p < 0.01). Regarding parity, PP cows displayed higher NEFA concentrations and a greater NEFA/cholesterol ratio than MP cows (p ≤ 0.01). A significant interaction between HLC and parity was observed for cholesterol concentrations. PP cows with severe HLC had higher cholesterol levels (3.8 ± 1.1 mmol/L) compared to MP cows with mild HLC (2.7 ± 0.7 mmol/L, p = 0.03). Cholesterol concentrations were also influenced by the month of calving (August vs. July; p = 0.02) and week of sampling (week 1 vs. weeks 2 and 3; p < 0.01). Additionally, BHB concentrations were elevated in cows with moderate and severe HLC compared to those with mild HLC (p < 0.01), with a significant interaction between HLC and the month of calving (p = 0.04), and a significant interaction between the week of sampling and HLC (p = 0.01).

Pairwise comparisons showed that the effect of HLC degree on BHB concentrations varied across weeks, with significant differences observed at weeks 1 and 3. At week 1, cows with severe HLC had higher BHB concentrations than those with mild HLC (p = 0.01). At week 3, cows with severe HLC had higher BHB concentrations than cows with both mild and moderate HLC (p < 0.01; Figure S1, Supplementary Material).

3.3. HLC and Postpartum Diseases

Many clinical conditions commonly associated with HLC were observed in this study (in order of decreasing prevalence): clinical metritis (32%), puerperal metritis (11.7%), retained fetal membranes (7.8%), severe lameness (7.8%), mastitis (2.9%), and clinical hypocalcemia (1.9%). Subclinical ketosis (47.6%), hypomagnesemia (43.7%), and hypocalcemia (32.4%) were also observed. Only subclinical ketosis was significantly related to HLC p < 0.001,

Table 5. Variables retained in the final logistic regression models describing the relationships between year, parity, month of calving, and HLC with risk of subclinical disease outcomes in postpartum grazing dairy cows.

Subclinical Diseases	Variable	Category	Estimate	OR	95% IC	p-Value
Hypomagnesemia	Parity	MP	Ref. 1
		PP	−1.9	0.15	0.04–0.44	<0.01
	Year	1	Ref.
		2	−1.5	0.23	0.09–0.56	<0.01
Ketosis	HLC	Mild	Ref.
		Moderate	1.9	6.6	2.2–21.8	<0.01
		Severe	3.2	23.9	6.3–125.4	<0.01
	Month of calving	July	Ref.
		August	−1.1	0.32	0.10–0.96	0.05
		September	−2.8	0.06	0.01–0.31	<0.01

¹ Reference.

). The final model showed no significant predictors related to clinical diseases or subclinical hypocalcemia.

3.4. HLC and Milk Production and Composition

Milk production was significantly related to HLC. Over the entire period, cows with severe HLC produced 422 L more compared to cows with mild HLC (p < 0.01;

Table 6. Milk production (mean ± SD) according to HLC category and lactation period.

HLC	n	Milk Production per Period (L)
		7 to 21 DIM	22 to 60 DIM	61 to 90 DIM	7 to 90 DIM
Mild (Ref. 1)	50	368 ± 15.1	1062 ± 40.3	798 ± 26.1	2227 ± 78.8
Moderate	25	386 ± 13.0	1096 ± 41.1	806 ± 43.5	2309 ± 89.2
Severe	21	438 ± 15.9	1257 ± 43.1	955 ± 32.6	2649 ± 81.6 ***

*** p < 0.01. ¹ Reference.

). Parity was also found to be a factor associated with milk production, with MP cows producing more than PP cows (p < 0.001). The R² value for the milk production model was 0.94. Parameters of Wilmink curves only indicated numerical variation and were not affected by any of the factors included in this study (p > 0.1; Table S1).

The models used for milk components are presented in

Table 7. Variables retained in the final linear mixed models * (mean ± SD and estimate) describing the relationships of month of calving, parity, sampling, ketosis, and HLC with milk components of grazing dairy cows during early lactation.

Milk Component	Variable	Category	Mean ± SD	Estimate	p-Value
Milk fat	Month of calving	July	3.7 ± 0.1	Ref. 1
		August	3.5 ± 0.1	−0.27	0.03
		September	3.4 ± 0.1	−0.22	>0.1
	Parity	MP	3.7 ± 0.1	Ref.
		PP	3.3 ± 0.1	−0.45	<0.01
	Sampling	30 DIM	3.7 ± 0.1	Ref.
		60 DIM	3.6 ± 0.1	−0.07	>0.1
		90 DIM	3.4 ± 0.1	−0.30	<0.01
	Ketosis	No	3.6 ± 0.1	Ref.
		Yes	3.6 ± 0.1	−0.20	0.08
Milk protein	HLC	Mild	3.3 ± 0.0	Ref.
		Moderate	3.3 ± 0.0	−0.01	>0.1
		Severe	3.2 ± 0.0	−0.11	0.04
	Sampling	30 DIM	3.2 ± 0.0	Ref.
		60 DIM	3.3 ± 0.0	0.09	<0.01
		90 DIM	3.3 ± 0.0	0.15	<0.01

* R² values for milk fat and milk protein models were 0.20 and 0.65, respectively. ¹ Reference.

. Milk fat percentage remained consistent across each degree of HLC; however, it was influenced by the month of calving, parity, and sampling. A reduced milk fat percentage was observed for cows that calved in August (p = 0.03), in PP cows (p < 0.01), and at the third sampling at 90 DIM (p < 0.01). None of the interactions included in the model were found to influence milk fat (p > 0.1). Nonetheless, the milk protein percentage was affected by HLC and sampling, although the interaction between these factors was not significant. Severe HLC reduced milk protein percentage (p = 0.04), and the highest values were observed in the second and third samplings compared with the first (p < 0.01).

4. Discussion

Previous studies on the relationship between HLC in dairy cows and various clinical, laboratory, or production outcomes, in both housed and grazing systems, have used samples obtained via conventional liver biopsy and quantified using chemical laboratory methods [6,8,15,30]. The present study differed methodologically in both sample acquisition FNB and HLC quantification (cytological), thus being a faster, less invasive, less expensive, technically simpler approach that could be incorporated easily into routine herd management [16,31,32]. Sensitivity (73%), specificity (85%), positive predictive value (90%), and negative predictive value (63%) have been reported for fine needle aspiration cytology [31]. In this way, it is relevant to improve the understanding of how cytologically determined HLC in grazing dairy cattle may be related to variations in certain blood metabolites and milk production results. Although cytology has moderate to high specificity but moderate sensitivity, it is a technique that enables early diagnosis of hepatic lipidosis, especially when the percentage of hepatic triacylglycerols ranges from 2% to 5%.

This study’s null hypothesis was rejected based on several findings: in both the pre-partum and postpartum periods, PP cows with severe HLC had higher plasma cholesterol concentrations than MP cows with mild HLC. Moreover, in the postpartum period, cows with severe HLC had higher serum NEFAs and NEFA/cholesterol ratios than cows with mild HLC; whereas cows with moderate and severe HLC had higher plasma BHB than cows with mild HLC. Finally, cows with moderate and severe HLC had an increased risk of subclinical ketosis compared to cows with mild HLC, and cows with severe HLC had higher milk production and lower milk protein content than those with mild HLC.

A recent work by Arshad and Santos [8], using liver samples obtained via surgical biopsy from housed university herds in the USA (in Florida and California), found an increased risk of hyperketonemia, hypocalcemia, metritis, diagnosis of multiple diseases postpartum, and decreased survival in the herd by 300 days postpartum in cows with 7.5% hepatic triacylglycerol compared to 2.5%. Thus, the fact that both this study and the previously cited study found that increased HLC in the early postpartum period was associated with negative outcomes—despite using cows in different management systems and different methods for obtaining liver samples and quantifying HLC—supports a potential role for FNB liver cytology to help inform herd management decisions.

4.1. HLC and Blood Metabolic Biomarkers

The associations between metabolic markers (NEFAs, BHB), lipid mobilization, and TAG have been well established [8,9,33], demonstrating that periparturient energy balance and hepatic TAG content are interrelated [34,35]. The results of this study showed that cows with severe HLC had higher serum NEFA concentrations, while those with moderate or severe HLC had elevated BHB concentrations compared to those with mild HLC during the postpartum period. However, no differences in pre-partum NEFA concentrations or pre-partum NEFA/cholesterol ratios were observed in cows with different HLC categories at postpartum. Previous studies have shown that HLC tends to remain low during the pre-partum period and increases significantly after calving [6,36], suggesting that pre-partum NEFA concentrations may not be sufficient to induce or predict a significant increase in hepatic lipid content. Moreover, since NEFAs are mobilized as a source of energy, especially for muscle contraction [37], it is possible that concentrations are lower in dairy cows in pastoral systems due to their higher daily physical activity [38].

We observed that PP cows with severe HLC had higher cholesterol concentrations before and after calving compared to MP cows with mild HLC. This finding may be associated with differences in both the metabolic and functional capacity of the liver between cows of different parity. In this sense, Osada et al. [39] reported that during the postpartum period, MP cows had significantly higher concentrations of NEFAs and BHB than PP cows, suggesting a more severe NEB due to the greater milk production of MP cows and lower concentrations of very-low-density lipoprotein (VLDL) and triacylglycerols. These findings indicate that MP cows have poor triglyceride secretion from the liver and become more susceptible to HLC. Therefore, the relatively higher cholesterol concentrations observed in PP cows with severe HLC may be explained by their greater capacity to secrete VLDL. Finally, the current analysis indicated that an increased NEFA/cholesterol ratio was significantly associated with severe HLC compared with mild HLC during the postpartum period. Similarly, Mostafavi et al. [33] reported that cows with a NEFA/cholesterol ratio > 0.2 were 9.9 times more likely to be affected by HL (total lipid > 10%). On the one hand, elevated NEFA concentrations are associated with the potential to impair liver function due to fatty acid re-esterification and triglyceride accumulation in the hepatocyte cytoplasm [3]. On the other hand, a low total cholesterol value indicates conditions in which VLDL production is limited and fatty infiltration is likely [40]. Consequently, elevated NEFA/cholesterol ratio values may be highly suggestive of HLC.

4.2. HLC and Postpartum Diseases

In the present study, HLC was merely associated with the occurrence of subclinical ketosis. The severity of lipid accumulation was associated with a higher risk of the disease, as cows with moderate and severe HLC were more prone to this metabolic alteration than those with mild HLC. These data agree with those reported by Arshad and Santos [8], who found that a change in hepatic triacylglycerol from 2.5 to 7.5% was associated with a linear increase in the relative risk of hyperketonemia by 2.5 times (15.2 vs. 37.5%). The relationship between TAG and ketosis has been described in the literature. In addition to conversion to TAG, NEFA in hepatocytes enters the β-oxidation pathway. However, under conditions of decreased dry matter intake, this pathway is intensified [34]. Acetyl-coenzyme A, the product of β-oxidation, is further oxidized in the tricarboxylic acid (TCA) cycle by binding with oxaloacetic acid, an intermediate of the TCA cycle and an obligatory link between the cycle and the gluconeogenesis pathway [34]. In cows during early lactation, the demand for gluconeogenesis increases markedly for milk lactose synthesis. Oxaloacetic acid is depleted for gluconeogenesis and in the mitochondria. As a result, acetyl-coenzyme A cannot enter the TCA cycle and is instead directed into the ketogenesis pathway. The excessive formation of ketone bodies, such as BHB, leads to ketosis [41]. Subclinical ketosis negatively impacts dairy herds during the transition period, resulting in reduced milk production [42,43] and decreased reproductive performance [44,45] compared with cows without this metabolic condition. In addition, affected cows are more likely to develop diseases such as displaced abomasum, clinical ketosis, and metritis [46], therefore being more likely to be discarded during early lactation [47].

Unlike our findings, Arshad and Santos [8] reported that an increase in hepatic triacylglycerol content from 2.5% to 7.5% was significantly associated with linear increases in the relative risk of hypocalcemia by 1.7 times (30.3 vs. 52.4%), metritis by 2.1 times (12.5 vs. 25.7%), and diagnosis of multiple diseases postpartum by 2.4 times (8.7 vs. 21.1%). However, extrapolating these findings to those of Arshad and Santos [8] should be limited due to differences in system type, given that grazing cows have a lower productive magnitude than housed cows [48,49]. Future research is recommended to further investigate the relationship between HLC and disease incidence in grazing dairy cows during the transition period.

4.3. HLC and Milk Production and Composition

Both PP and MP cows diagnosed with severe HLC increased milk production throughout early lactation. Although dairy cows with high production levels are described as more likely to develop HL [4,15], the results found by Spaans et al. [6], together with the prevalence shown in this study for moderately productive dairy cows from pastoral systems, showed similarities to those reported for stalled systems. This suggests that HL is a significant metabolic disorder in dairy herds that responds to the NEB during the postpartum transition period. The mobilization of body reserves through lipolysis to meet energy requirements for lactation onset [50] is a physiological process in the cow [51]; i.e., HLC may be inherent to the productive cycle of a dairy cow, despite its productive performance [52,53].

The relationship between HLC and the milk performance in dairy cows has not been fully elucidated [8]. Following calving, increased energy demands initiate lipolysis to ensure adequate resource availability and support milk synthesis [42,50,54]. This metabolic adaptation leads to an elevated influx of NEFAs and other lipids to the liver, where they are used as substrates for milk production. Cows with higher milk yields mobilize greater energy reserves due to their elevated nutritional requirements [35,55,56]. Arshad and Santos [8] reported that increased liver TAG concentrations were associated with higher milk production (2.1 kg/d). Conversely, Jorritsma et al. [15] found that higher TAG liver infiltration, in conjunction with other factors, was associated with reduced milk production (R² = 0.22). In the present study, the coefficients of lactation curves did not differ across HLC categories, nor did the days or litters at peak lactation. Breed characteristics and genetic factors, as highlighted by Vargas and Ulloa [57], significantly influence curve coefficients and peak production. To our knowledge, this study is the first to examine HLC’s impact on lactation curves. However, as with milk production, further research is needed to clarify the relationships among HLC, milk yield, and lactation curves during the transition period in dairy cows.

The findings indicate that the milk fat percentage was not influenced by HLC; however, the month of calving, parity, and sampling did significantly affect this parameter. Similarly, Arshad and Santos [8] found no linear relationship between milk fat percentage and hepatic TAG, with values similar to those reported in our study (3.74%, 3.77%, and 3.80% milk fat for hepatic TAG levels of 2.5%, 5.0%, and 7.5%, respectively). Milk fat is the most variable milk component, influenced by several factors, including genetics, nutrition, and physiological status [58,59]. During periods of NEB, mobilized fatty acids provide additional substrates for milk fat synthesis in the mammary gland. Nevertheless, other variables—such as those previously mentioned and the presence of metabolic disorders—may also modulate milk fat levels. For instance, Yang et al. [60] observed an increase in milk fat percentage in stalled dairy cows diagnosed with subclinical ketosis, whereas the present study found no such increase.

This study determined that milk protein percentage decreases in cows diagnosed with severe HLC. No significant effect of hepatic TAG content on milk protein was reported by Jorritsma et al. [15]. However, the results are consistent with more recent findings. Arshad and Santos [8] found that when hepatic TAG content exceeds 7%, milk protein content declines quadratically. We cannot entirely exclude the possibility that the reduced milk protein observed in the cows diagnosed with severe HLC was due to a dilution effect, considering the higher production observed in those cows. However, as excessive lipid content in the liver is linked to exacerbated catabolism [50], it has diverse pathological consequences and may affect both cell function and metabolism, thereby directly and indirectly influencing milk protein synthesis [61,62,63].

5. Conclusions

This study in grazing dairy cows examined the relationship between the degree of early postpartum HLC, determined cytologically from an FNB sample, and metabolic status, health, and production during the transition and early lactation periods. This study’s findings show an association between increased HLC in the early postpartum period and subsequent negative clinical outcomes that are potentially economically relevant. This study was conducted in a single grazing dairy herd, underscoring the need to replicate these findings in other herds. In this way, it will be possible to obtain sufficient evidence to demonstrate the usefulness of HLC, determined via liver cytology using FNB, as an aid in herd management decision making. Further research could provide an economic justification for medically intervening or culling cows with HLC above a certain cytological threshold.