Pathogen and severity-dependent immune responses in bovine mastitis: highlight the dynamics of differential somatic cell count

Abstract

Immune responses in bovine clinical mastitis (CM) probably differ depending on the causative pathogen and disease severity. The observational study aimed to investigate whether both factors are associated with the dynamics of immune indicators, including somatic cell score (SCS), white blood cell count (WBC), serum albumin/globulin (A/G) ratio, and differential somatic cell count (DSCC). We collected blood and milk samples 0, 3, 5, 7, 14, and 21 days after CM occurred in 38 cows, and grouped the cases (n=49) by disease severity and pathogen. We analyzed data using a linear mixed model considering the effects of pathogens and severity, calculated estimated-marginal means for indicators at each time point, and compared the means between groups. The dynamics of WBC varied depending on both pathogen and severity. WBC changed drastically in either severe or coliform-caused CM, slightly elevated in streptococcal mastitis, but unchanged in staphylococcal mastitis. This possibly relates to the deficiency in innate immune response toward staphylococci. The A/G ratio also changed depending on severity, as it dropped sharply only in severe CM. We observed a non-linear relationship between DSCC and SCS, possibly due to mammary epithelial cells shedding in milk when CM occurred. When cows recovering from Streptococcus dysgalatiae mastitis, DSCC decreased while SCS remained high, suggesting a healing process requiring more macrophages. Our results demonstrate that both the severity and pathogen are associated with immune responses in CM, providing insights into mastitis pathogenesis.

Mastitis is one of the major concerns in the dairy industry, either in a view of economics or animal welfare. In the past decades, clinical mastitis (CM) in dairy cows has been controlled using antimicrobial agents, which accounts for a majority of antibiotic use on dairy farms. Nonetheless, outcomes were often not as expected [35]. We hence warranted measures for evaluating CM prognosis, to achieve an effective therapy and avoid unnecessary economic losses and pain in animals.

Mastitis is the inflammatory response toward intra-mammary infections (IMI), thus can be detected and monitored using immune indicators. As an example, somatic cell count (SCC) in milk has been widely used for monitoring mastitis [34]. The immune responses toward IMI can also be detected in the blood, as an association between the udder health status and the serum albumin/globulin (A/G) ratio has been reported [4]. Notably, the immune responses in bovine mastitis are pathogen- [37] and host-dependent [6]. Studies showed that changes in SCC [10] and milk cytokines [2] are pathogen-specific. A discrepancy can be found between immune responses toward gram-negative and gram-positive bacteria. Infections of gram-negative bacteria are often accompanied by the release of a large amount of pro-inflammatory cytokines, resulting in mastitis with severe symptoms. Moreover, the same pathogen can cause distinct immune responses, likely due to host factors. During coliform IMI, blood acute-phase protein concentration [27] and white blood cell count (WBC) [47] varied depending on disease severity. Such difference in immune responses leads to varied outcomes of CM. Each mastitis pathogen affects milk production to a different degree [19]. The impairment of milk yields [16] and reproductive performance [13] in mastitis proportions to the degree of inflammation. Consequently, the outcome after treatment depends on both the severity and the causative pathogen of mastitis [28]. Evidence shows that mastitis outcome is determined by both host and pathogen factors, while most studies considered either alone. Studies are thus required to describe general host-pathogen relationships in mastitis.

Instead of counting the total number of immune cells in milk (i.e., SCC), studies suggested that the subpopulation of milk cells provides additional information regarding mastitis, aiding us in identifying the causative pathogen and the stage of CM [24]. Techniques for differentiating immune cells in milk had been laborious and costly, by counting cells with either a microscope or a flow cytometer following immunostaining (reviewed by Halasa and Kirkeby [18]). Until recently, a high-throughput technique has been developed [9]. The analysis of somatic cell composition can thus be incorporated into the regular dairy herd improvement (DHI) program, allowing us to measure milk differential somatic cell count (DSCC) on a large scale. DSCC represents a combined proportion of polymorphonuclear leukocytes (PMNs) and lymphocytes in milk cells. Since lymphocytes generally maintain a low proportion among milk cells, changes in DSCC can be considered as a consequent change in PMNs. Studies showed that DSCC level varies depending on the causative pathogen [21, 30, 38] and stage [22] of mastitis. Specifically, DSCC increases more markedly in the early stage of major pathogen IMI (i.e., IMI caused by environmental streptococci, Staphylococcus aureus, and coliforms). For these reasons, along with SCC, DSCC has been proposed for predicting the udder health status of cows [40]. However, a limited number of studies discussed the effect of mastitis on the dynamics of DSCC, and in most studies, samplings were conducted monthly [22]. Investigations with a higher sampling frequency are supposed to provide a more comprehensive picture of changes in DSCC, helping us explore the potential use of DSCC.

For these reasons, we investigated the dynamics of immune indicators in naturally occurring CM, where the effects of causative pathogens and severity were taken into concern. Investigated traits included WBC, A/G ratio, SCC, and DSCC. We highlight the dynamics of DSCC because it has yet to be studied after CM.

MATERIALS AND METHODS

This study had been originally conducted to evaluate the effect of orally administrated chitosan on mastitis, whose effect is statistically insignificant on all evaluated traits and the data has thus been utilized for this study.

Study herd

This study was conducted at the Field Center of Animal Science and Agriculture belonging to Obihiro University of Agriculture and Veterinary Medicine in Hokkaido, Japan (42° 52′ 34″ N, 143° 10′ 23″ E), with all procedures approved by the Animal Care and Use Committee of the university (Permission number: 21-156). The investigation continued from June 2021 to February 2022. During the period, the herd consisted of around 70 lactating Holstein cows. Cows were kept in a free-stall barn, in which straw was used as bedding material. Cows were fed a total mixed ration, milked twice a day in a parlor, and produced 11,000 kg of milk per year on average. All milking cows participated in monthly DHI testing, and bulk tank SCC had been lower than 150,000 cells/mL throughout the study period.

Inclusion criteria and sampling procedures

Cows showing clinical signs of mastitis, including reduced milk yield, abnormal milk, and/or swollen udder, were first identified by farm staff and later confirmed by study personnel (a veterinarian) within 6 hr. To confirm the mastitis, a thorough physical examination and a California Mastitis Test were performed, and milk samples were collected aseptically from the affected quarter for bacterial culture. Cows confirmed with CM were considered for inclusion in the study, but cows concurrently receiving antibiotics or anti-inflammatory medications for other diseases (e.g., lameness) or with mastitis caused by the same pathogen as the previous occurrence (i.e., recurrent mastitis) were excluded. Mastitis treatment was allocated based on milk culture results [23]. Specifically, mastitis caused by major pathogens had been treated with antibiotics once a day for 5 consecutive days (i.e., d 1–5 after mastitis occurrence) or otherwise left untreated. To evaluate immune responses, blood samples and quarter milk samples were collected immediately (d 0) as well as 3, 5, 7, 14, and 21 days after CM occurred.

Collection and analyses of blood samples: Peripheral blood samples were collected from the coccygeal veins of cows. Samples anticoagulated with Ethylenediaminetetraacetic acid (EDTA) were used for WBC measurement, performed immediately after sampling with an automated hematology analyzer (Celltac-α^®; Nihon Kohden Co., Tokyo, Japan). Samples collected in plain serum tubes were allowed to clot for 10 min at 37°C, followed by centrifugation at 1,600 g, 4°C for 15 min. Isolated serum was stored at −30°C until analysis. The serum was analyzed with an automatic photometer analyzer (TBA-120FR^®, Toshiba Co., Tokyo, Japan) for total protein and albumin. Globulin concentration was calculated by subtracting the concentration of albumin from serum total protein, and the A/G ratio was subsequently determined.

Collection and analyses of milk samples: Quarter milk samples were collected after disinfecting the udder and forestripping. Except on the day mastitis identified, samples were collected right before afternoon milking. A 30~50 mL of milk was collected for SCC and DSCC, preserved at 4°C, and examined with CombiFoss 7 (FOSS Analytical A/S, Hillerød, Denmark) in less than 4 days. About 5 mL of milk was collected aseptically for bacterial culture, conducted right after collection. Milk samples with abundant clots could not be submitted for the SCC test, and DSCC values were unreliable in samples with ≤50,000 cells/mL [9]. These records were excluded.

Definition of severity

The severity of CM was evaluated based on a scoring system described in Pinzón-Sánchez and Ruegg (2011) [32], where CM was defined as mild when only the milk was abnormal (e.g., containing clots or flakes), as moderate when the affected quarter was abnormal (e.g., swollen or redness) but without systemic signs, or as severe when cow showed systemic signs such as hyperthermia (>39.5°C), dehydration, anorexia or depression.

Bacterial culture

Bacterial culture was performed following the National Mastitis Council guidelines [26] with some modifications. Fifty microliters of milk were plated onto a 5% sheep blood agar (TSA II^®; BD Japan) and incubated at 37°C. After 24 hr and 48 hr of incubation, bacteria were categorized into either staphylococci, streptococci, Gram-negative rods (GNR), or Gram-positive rods based on morphology, catalase reaction, and Gram staining. Staphylococcus aureus (SA) was differentiated from coagulase-negative staphylococci (CNS) based on hemolysis patterns and a tube coagulase test. A presumptive diagnosis of streptococcal isolates was made by culturing bacteria on a chromogenic agar (CHROMagar™ Orientation, BD Japan) [15], a sorbitol-containing andrade peptone broth, and an SF broth (Streptococcus uberis diagnosis kit; Kanto Chemical, Tokyo, Japan). Isolates presumed to be Streptococcus uberis (SU) and Streptococcus dysgalactiae (SD) were confirmed using API 20 Strep^® (BioMérieux Japan Ltd., Tokyo, Japan), and the other streptococci-like isolates were categorized into other streptococci and related genera (OS). We did not detect CM arising from Gram-positive rods throughout the study period. Samples were considered contaminated when three or more distinct colonies were identified. Plates with less than 60 cfu/mL of microbes were considered no significant growth (NG).

Statistical analyses

Analyses were performed with the statistical software R version 4.0.5 [33]. Data were first explored and transformed with the package “tidyverse” [17]. For modeling purposes, SCC was transformed to somatic cell score (SCS), equivalent to log₂ (SCC/100,000) +3, to fit a normal distribution [49]. Days in milk (DIM) was scaled in a range from 0 to 1 with min-max normalization and transformed back to the original scale after modeling for interpretation purposes. The statistical analyses aimed to clarify how mastitis pathogens and severity affected immune responses over time.

Statistical analyses of SCS, WBC, and A/G ratio: We fitted data in linear mixed models using the ‘lmer’ function within the ‘lme4’ package [12]. Models were of the form:

$Y_{ijklm}={\beta }_0+{\beta }_{1}Pat{h}_k+{\beta }_{2}Se{v}_m+{\beta }_{3}Tim{e}_l+{\beta }_{4}Pat{h}_k\times Tim{e}_l+{\beta }_{5}Se{v}_m\times Tim{e}_l+{\beta }_{6}DI{M}_i+{\beta }_{7}DI{M}_i^2+{\beta }_{8}Parit{y}_{ij}+r_j+{\epsilon }_{ijklm}$ , (1)

where Y_ijklm is the response variable, SCS, WBC, and A/G ratio (with albumin and globulin concentrations in supplementary analysis), for the i^th observation in the j^th mastitis event. Path_k, Sev_m and Time_l are fixed effects of mastitis pathogen (7 levels: NG, CNS, SA, SU, SD, OS, and GNR), severity (3 levels: mild, moderate, and severe), and days after mastitis (6 levels: d 0, 3, 5, 7, 14 and 21), respectively. Interactions between Path_k, Sev_m and Time_l were included to evaluate the effects of the pathogen and severity at each time point. DIM_i and $DI{M}_i^2$ are the linear and quadratic terms of days in milk, Parity_ij is the term of parity, and γ_j is the random intercept for each mastitis case. The random effect of cow was also tested but later excluded due to its insignificance in all evaluated traits.

Statistical analyses of DSCC: Previous studies suggested that DSCC values are strongly associated with SCS [9, 29]. Consequently, analyzing DSCC alone probably reaches a conclusion similar to what we can draw from analyzing SCS [38]. As DSCC represents the proportion of PMNs and lymphocytes in SCC, the deviations from this relationship may have a specific meaning, for instance, an indication of the mastitis stage. To test this hypothesis, we included both linear and quadratic terms of SCS in the model:

$DSC{C}_{ijklm}={\beta }_0+{\beta }_{1}SC{S}_{ijklm}+{\beta }_{2}SC{S}_{ijklm}^2+{\beta }_{3}Pat{h}_k+{\beta }_{4}Se{v}_m+{\beta }_{5}Tim{e}_l+{\beta }_{6}Pat{h}_k\times Tim{e}_l+{\beta }_{7}Se{v}_m\times Tim{e}_l+{\beta }_{8}DI{M}_i+{\beta }_{9}DI{M}_i^2+{\beta }_{10}Parit{y}_{ij}+r_j+{\epsilon }_{ijklm}$ , (2)

where DSCC_ijklm is the observed DSCC, SCS_ijklm and $SC{S}_{ijklm}^2$ are linear and quadratic terms of SCS. In the preliminary analysis, we found that the relationship between SCS and DSCC was not linear by checking the residual plot, and thus the quadratic term was included. The other terms remain as they were in equation (1).

The significance of each term (P<0.05) in these models was tested by type III ANOVA using the package “lmerTest” [1], and Kenward-Roger’s approximation was used to calculate denominator degrees of freedom. The effect of numerical variables was visualized using the package “visreg” [5]. To explore the effect of categorical variables, estimated-marginal means (EMMs) were computed, compared, and plotted using the “emmeans” package [25]. P-values were adjusted when multiple comparisons were conducted. Specifically, for pairwise comparisons, Tukey’s method was used, and for many-to-one comparisons, Dunnett’s method was used. In the following text, the number following “± “ represents the standard error (SE) of the estimate if not specified otherwise.

RESULTS

Characterization of CM cases

We followed 49 cases of CM occurring in 38 cows and obtained 294 records during the investigation. The number of cases within each pathogen or severity group is detailed in Table 1. The most recovered pathogen was SU (n=15; 30.6%), followed by OS (n=8; 16.3%), and CNS (n=7; 14.3%). The number of cases with mild, moderate, and severe symptoms was 32 (65.3%), 9 (18.4%), and 8 (16.3%), respectively. Except for mastitis caused by GNR, cases mostly presented mild clinical signs, accounting for 50% (SD) to 85.7% (CNS) of cases within each pathogen group. Table 1 shows the average of parity and DIM at the occurrence of each pathogen-specific CM. The mean DIM was lower in CM caused by SA and SD, but no distinct differences in the parity existed between each group.

Table 1. The number of cases within each pathogen and severity group, as well as the averages of parity and days in milk (DIM) of cows when cases were identified.

Pathogens*	Severity				DIM	Parity
	Mild	Moderate	Severe	Total
	Number of cases (%)				Mean ± SD
CNS	6 (85.7)	1 (14.3)	0	7 (14.3)	87.6 ± 66.0	2.6 ± 1.0
GNR	0	1 (50.0)	1 (50.0)	2 (4.1)	154.0 ± 4.2	3.5 ± 2.1
NG	4 (57.1)	3 (42.9)	0	7 (14.3)	170.4 ± 98.8	3.1 ± 1.3
OS	5 (62.5)	1 (12.5)	2 (25.0)	8 (16.3)	187.6 ± 104.1	3.2 ± 1.5
SA	4 (66.7)	1 (16.7)	1 (16.7)	6 (12.2)	39.0 ± 48.5	2.2 ± 1.0
SD	2 (50.0)	0	2 (50.0)	4 (8.2)	43.8 ± 74.6	2.2 ± 0.5
SU	11 (73.3)	2 (13.3)	2 (13.3)	15 (30.6)	110.7 ± 70.6	2.4 ± 0.9
Total (%)	32 (65.3)	9 (18.4)	8 (16.3)	49 (100.0)	116.0 ± 89.7	2.7 ± 1.1

*NG=no significant growth; CNS=couagulase-negative staphylococci; GNR=Gram negative rods; OS=other streptococci; SA=Staphylococcus aureus; SD=Streptococcus dysgalactiae; SU=Streptococcus uberis.

Factors affecting the dynamics of SCS, WBC, and A/G ratio

Factors affecting Somatic Cell Score (SCS): SCS in four samples could not be measured due to abundant clots in these samples. Table 2 shows that sampling time affected SCS most significantly. EMMs of SCS at d 0, 3, 5, 7, 14, and 21 were 10.34 ± 0.50, 7.67 ± 0.43, 6.54 ± 0.43, 5.74 ± 0.42, 4.48 ± 0.42, and 4.8 ± 0.42, respectively. The pathogen and severity of CM also affected SCS. Although SCS varied on average depending on the causative pathogen or the severity (in pairwise comparisons, NG vs. SU= −2.3 ± 0.68, P=0.023; Mild vs. Severe= −1.5 ± 0.63, P=0.05), a similar downward trend was observed in all groups (Fig. 1 and Supplementary Table 1).

Table 2. ANOVA for the effect of time, mastitis pathogen, and severity on somatic cell score, white blood cell count, and serum albumin/globulin (A/G) ratio.

Terms	NumDFc	SCSa			WBCb			A/G ratio
		DenDFd	F value	P value	DenDF	F value	P value	DenDF	F value	P value
Time	5	198.9	36.0	3.6E-26	201.4	8.6	2.0E-07	203.5	4.0	1.8E-03
Pathogen	6	37.6	2.4	0.047	37.1	2.3	0.06	37.3	2.6	0.032
Severity	2	37.6	3.4	0.046	37.2	0.8	0.45	37.3	3.7	0.034
Parity	1	37.1	0.2	0.70	37.2	10.2	2.9E-03	37.4	14.6	4.8E-04
DIMe	1	42.3	1.8	0.18	40.9	2.9	0.10	47.3	1.7	0.19
DIM2	1	42.8	0.3	0.62	41.5	5.1	0.029	49.1	3.0	0.09
Time × Pathogen	30	196.7	0.9	0.61	200.2	1.7	0.015	200.4	0.9	0.66
Time × Severity	10	196.5	1.2	0.31	199.9	1.9	0.05	199.8	2.6	5.0E-03

^aSomatic cell score; ^bWhite Blood Cell count; ^cNumDF=numerator degrees of freedom; ^dDenDf=adjusted denominator degrees of freedom; ^eDIM=days in milk.

Fig. 1.

Estimated-marginal means of somatic cell score over time in each pathogen (A) and severity (B) group. NG=no significant growth; CNS=coagulase-negative staphylococci; GNR=Gram-negative rods; OS=other streptococci; SA=Staphylococcus aureus; SD=Streptococcus dysgalactiae; SU=Streptococcus uberis.

Factors affecting White Blood Cell count (WBC): Time, parity, DIM, the interaction between time and pathogen as well as the interaction between time and severity significantly affected the WBC level (Table 2). Figure 2 shows that WBC decreased as parity increased, and a non-linear relationship existed between DIM and WBC, as WBC tended to be lower at the lactation peak. Both the causative pathogen and CM severity affected the dynamic of WBC. On the day GNR mastitis occurred, the WBC was lower than usual (Fig. 3A and Supplementary Table 1), while it increased dramatically at d 3 and decreased gradually in the following days. In streptococcal mastitis (i.e., OS, SD & SU), WBC rose from d 5 and peaked at d 7. On the contrary, in mastitis caused by the other pathogens (i.e., NG, CNS & SA), WBC remained at the same level as on d 0 (Fig. 3A). We confirmed this finding by computing polynomial contrasts as a function of time for each pathogen. Results show that WBC in GNR, OS, SD, and SU mastitis had significant linear, quadratic, and/or cubic trends, while no significant trend of WBC was observed in NG, CNS, or SA mastitis (Table 3). The same analysis was conducted to compare the dynamic of WBC in each severity group. There was a substantial difference between the contrasts for each group (Table 3). To verify this difference, we further compared polynomial contrasts between groups. Results show a statistical difference in the quadratic trend between the severe group and the other groups (Mild vs. Moderate= −57.4 ± 73.1, P=0.43; Mild vs. Severe=192.5 ± 79.1, P=0.016; Moderate vs. Severe=249.9 ± 96.6, P=0.01), showing that the WBC level changed more aggressively in cows with severe mastitis (Fig. 3B).

Fig. 2.

The effect of parity (A) and days in milk (DIM; B) on white blood cell (WBC) count. The solid lines represent fitted regression lines, and the dots represent partial residuals.

Fig. 3.

Estimated-marginal means (EMMs; black dot) and 95% confidence interval (grey bar) of white blood cell count over time in each pathogen (A) or severity (B) group. Corresponding polynomial contrasts for each group are reported in Table 3. NG=no significant growth; CNS=coagulase-negative staphylococci; GNR=Gram-negative rods; OS=other streptococci; SA=Staphylococcus aureus; SD=Streptococcus dysgalactiae; SU=Streptococcus uberis.

Table 3. Trends of white blood cell counts (WBC) over time in each pathogen and severity group.

Group		Polynomial terms in trends of WBC
		Linear		Quadratic		Cubic
		Estimate ± SE	P value	Estimate ± SE	P value	Estimate ± SE	P value
Pathogens*
	NG	73.6 ± 67.6	0.28	−106.0 ± 73.9	0.15	−78.9 ± 108.1	0.47
	CNS	−14.7 ± 69.8	0.83	−63.5 ± 76.3	0.41	−24.2 ± 111.6	0.83
	GNR	254.8 ± 119.5	0.034	−543.4 ± 130.9	4.9E-05	521.2 ± 191.6	7.1E-03
	OS	20.9 ± 61.5	0.73	−183.7 ± 67.1	6.7E-03	−98.4 ± 98.2	0.32
	SA	−36.3 ± 71.0	0.61	−102.2 ± 77.3	0.19	−19.3 ± 113.1	0.86
	SD	196.0 ± 87.0	0.025	−161.8 ± 94.9	0.09	−44.6 ± 139.0	0.75
	SU	193.7 ± 48.4	8.7E-05	−151.9 ± 52.8	4.5E-03	−270.1 ± 77.3	5.8E-04
Severity
	Mild	95.2 ± 36.9	0.011	−142.5 ± 40.2	4.8E-04	29.2 ± 58.8	0.62
	Moderate	−28.4 ± 59.1	0.63	−85.1 ± 64.6	0.19	−28.8 ± 94.5	0.76
	Severe	228.1 ± 62.2	3.1E-04	−335.0 ± 67.9	1.7E-06	−6.6 ± 99.4	0.95

Estimated coefficients (± SE) for polynomial contrasts and their significance are shown. *NG=no significant growth; CNS=coagulase-negative staphylococci; GNR=Gram-negative rods; OS=other streptococci; SA=Staphylococcus aureus; SD=Streptococcus dysgalactiae; SU=Streptococcus uberis.

Factors affecting the Albumin/Globulin (A/G) ratio: Time, parity, the severity and pathogen of mastitis, and the interaction between time and the severity significantly affected the A/G ratio (Table 2). A strong negative relationship existed between parity and A/G ratio (−0.053 ± 0.014; P=5.2E-04). Although the A/G ratio varied on average depending on the causative pathogen, the trends were similar: the A/G ratio dropped slightly from d 3 in most pathogen groups except for SD mastitis (Supplementary Table 1). In contrast, CM severity had a more pronounced effect on the dynamics of the A/G ratio (Fig. 4 and Supplementary Table 1). The A/G ratio in severe mastitis dropped significantly from d 3 till the end of sampling (d 21) but remained stable in both mild and moderate CM. To elucidate the causes of variations in the A/G ratio, we analyzed the concentration of albumin and globulin using the same model (Model 1). Figure 4 shows that, in severe mastitis, the albumin declined significantly at d 3 and gradually increased afterward while the globulin rose significantly from d 3 until d 21.

Fig. 4.

Estimated-marginal means (EMMs; black dot) and 95% confidence interval (grey bar) of serum albumin/globulin (A/G) ratio, albumin, and globulin over time in each severity group. EMMs were compared with EMM on d 0 in each group, and P-values adjusted by Dunnett’s method are shown on the top of each corresponding time point.

Factors affecting the dynamics of DSCC

We excluded 34 records where DSCC measurements were unreliable (in samples with SCC ≤50,000 cells/mL) or unmeasurable (in samples with abundant clots), resulting in 260 observations. We found a strong non-linear relationship between DSCC and SCS (Table 4 and Fig. 5A). Also, parity significantly affected DSCC. Figure 5B shows that as parity increased, DSCC values tended to be higher. Interestingly, the dynamic of DSCC differed in mastitis caused by SD compared to mastitis caused by other pathogens. This was confirmed by comparing polynomial contrasts in NG mastitis (as a reference) with those in the other types of mastitis, in which a statistical difference was found in the quadratic trend between NG mastitis and SD mastitis (−212.3 ± 55.6, P=1.9E-04). To visualize this effect, we plotted model-based predictions of DSCC as a function of SCS by pathogen group (Fig. 6). The figure shows that, for SD mastitis, DSCC decreased while SCS remained high.

Table 4. ANOVA for the effect of time, mastitis pathogen, and severity on differential somatic cell count (DSCC).

Terms	Diffieretial somatic cell count
	NumDFa	DenDFb	F value	P value
Somatic cell score	1	191.5	81.0	2.3E-16
Somatic cell score2	1	186.6	55.7	3.1E-12
Time	5	173.5	2.2	0.06
Pathogen	6	39.4	1.1	0.36
Severity	2	39.7	2.1	0.14
DIMc	1	40.0	3.4	0.07
DIM2	1	38.7	1.7	0.20
Parity	1	37.6	4.6	0.038
Time × Pathogen	30	168.4	1.6	0.029
Time × Severity	10	168.1	0.6	0.82

^aNumDF=numerator degrees of freedom; DenDf=adjusted denominator degrees of freedom; ^cDIM=days in milk.

Fig. 5.

The relationship between somatic cell score and differential somatic cell count (DSCC; A), and the effect of parity on DSCC (B). The solid lines represent fitted regression lines, and the dots represent partial residuals.

Fig. 6.

Model-based predictions of differential somatic cell count (DSCC) according to somatic cell score for each pathogen. NG=no significant growth; CNS=coagulase-negative staphylococci; GNR=Gram-negative rods; OS=other streptococci; SA=Staphylococcus aureus; SD=Streptococcus dysgalactiae; SU=Streptococcus uberis.

DISCUSSION

Characterization of CM cases

Most enrolled CM cases were caused by gram-positive bacteria, dissimilar from studies conducted in either China [14], Belgium [45], or the US [28]. In those studies, gram-negative bacteria were more commonly isolated. The high prevalence of SU in the present study is attributable to the use of straw as bedding [44]. Previous studies showed that cows identified with CM mostly exhibited mild symptoms [28, 32, 45]. However, we identified severe symptoms in a relatively large proportion of cows with mastitis caused by gram-positive bacteria. In this study, clinical signs were evaluated by a trained investigator (a veterinarian). Consequently, the sensitivity for detecting systemic clinical signs may be higher than that in other studies, where clinical signs were recorded by producers. In addition, mastitis caused by SD and SA was prone to occur in early lactation, in agreement with a previous report [10].

Pathogen-specific immune responses

The similar patterns of SCS in CM caused by major and minor pathogens (i.e., CNS) might be surprising (Fig. 1A), due to the well-known differences between them in virulence [36] and in the ability to elevate SCS [11]. However, we collected only cases with clinical signs, and in such a case, minor pathogens can cause a comparable elevation of SCS [10] and milk yield loss [19] as major pathogens. Although the level of SCS slightly varied depending on the pathogen (Fig. 1A), the difference was trivial, hence challenging to use for differentiating the causative pathogen of CM. On the other hand, we observed a more specific effect of mastitis pathogens on the dynamics of WBC. CM caused by GNR triggered the most immediate and dramatic immune responses (leukopenia), and streptococcal infections (i.e., OS, SD, and SU) induced delayed, less intensive, but significant responses (leukocytosis). Contrastingly, immune responses toward SA and CNS infections were the most obscure (Fig. 3A). These findings correspond to the change in milk cytokines [2] and Toll-like receptors gene expression in mammary epithelial cells toward IMI [31], likely related to the impairment of NF-κB activation in staphylococcal infections [50]. Leukogram (i.e., segmented neutrophil count, monocyte count, etc.) has been proposed as a diagnostic tool for differentiating Gram-positive and Gram-negative mastitis, with an overall accuracy of 91% [42]. In addition to that, our results suggest the potential of WBC count for differentiating streptococcal and staphylococcal mastitis. As an example, we can use the difference between WBC from d 0 to d 7 to differentiate streptococcal mastitis from staphylococcal mastitis. This achieved a specificity of 1, a sensitivity of 0.41, and an accuracy of 0.6 when the threshold was set to 3,900 cells/µL. This indicates that, when the difference between d 7 WBC and d 0 WBC is higher than 3,900 cells/µL, the causative pathogen is more likely to be Streptococcus rather than Staphylococcus.

Using the A/G ratio to evaluate the severity of CM

Our results show that mastitis severity affects the dynamics of the A/G ratio, as the A/G ratio dropped sharply only in severe mastitis (Fig. 4). Bobbo et al. (2017) [3] reported a negative association between the A/G ratio and SCS. They concluded that the decreased A/G ratio can be attributed to an increase in globulin production due to inflammation and an albumin loss due to the damaged blood-milk barrier. Similarly, we found a significant reduction in albumin and a significant elevation after the outbreak of severe mastitis (Fig. 4). As higher concentrations of cytokine were detected in cows with severe mastitis [48], an increase in globulin production was expected. Also, the damage to the blood-milk barrier was supposedly serious in severe mastitis, thus our observations were a result of both mechanisms. Our results show that mastitis severity can be revealed by the A/G ratio (Fig. 4) rather than SCS (Fig. 1B), suggesting that compared to SCS, the A/G ratio provides information on the patient’s general health status.

The severity of CM affected its outcomes, including the number of treatments [28], the culling rate [47], and the reproduction performance [13] of cows, thus evaluating the severity of CM is essential for herd management. In this context, the A/G ratio can be used to determine the disease’s severity afterward, as the decrease in A/G ratios remains detectable for at least 3 wk (Fig. 4). Cattaneo et al. (2021) [8] demonstrated that cows with a low A/G ratio before dry-off perform worse compared to their pen mates in the next lactation, both in terms of milk yield and fertility. Because a pronounced decrease in the A/G ratio was observed in severe mastitis (Fig. 4), we can thus presume that severe CM occurring in the late lactation would deteriorate cows’ milk yield and fertility in the subsequent lactation. This highlights the importance of mastitis severity evaluation in dairy herd management, and such evaluation can be facilitated by measuring the A/G ratio. To demonstrate the clinical usefulness of the A/G ratio, future investigation is necessary into the relationship between the A/G ratio and survival rate in cows suffering from severe CM.

The dynamics of DSCC after the occurrence of CM

This is the first study to describe the dynamics of DSCC after CM occurred, making it difficult to compare the results with other studies. Surprisingly, we identified a non-linear relationship between SCS and DSCC (Fig. 5A). Since PMNs are the first cells to arrive at the site of inflammation, we expected proportions of PMNs, and thus DSCC would increase as SCS increased. Most studies suggested a linear dependence between SCS and DSCC [9, 29], whereas we can recognize an indistinct non-linear pattern between them in the figures presented by Kirkeby et al. (Fig. 2 of [21]). The non-linear relationship was revealed since we collected samples immediately after CM occurred, and consequently, some of our samples contained extremely large numbers of somatic cells. The cause for this non-linear relationship may involve mammary epithelial cells shedding into milk due to mastitis damage. Epithelial cells are counted as SCC but not included in DSCC [9]. In one study, the proportion of epithelial cells in SCC could be up to 44% [24]. Both explained why, during the outbreak of CM, DSCC decreased when SCS was extremely high.

We applied an alternative strategy to analyze the dynamics of DSCC to avoid obtaining a result similar to that from the analysis of SCS, due to the high association between these two traits. With this strategy, we found that the relationship between DSCC and SCS differed in the healing process of SD mastitis, in which DSCC decreased while SCS remained high (Fig. 6). In the healing process of mastitis, macrophages are recruited to remove apoptotic neutrophils [41] and to repair damaged tissue [43]. SD releases hyaluronidase and fibrinolysin and invades mammary epithelial cells [7] during its infection, explaining how it causes stronger damage to the mammary gland, and thus the healing process may require more macrophages. Similarly, Pegolo et al. (2022) [30] described a pathogen-specific effect on DSCC, in which Prototheca spp. caused lower DSCC values compared to Streptococcus agalactiae. Interestingly, Prototheca spp. was reported to cause irreversible damage to the mammary gland and induce a chronic granulomatous inflammation (infiltration of macrophages and other immune cells) following the failure of healing [46]. Furthermore, the low DSCC and high SCS situation described here can be linked to economic loss and mastitis prognosis. Studies showed that cows with high SCS and low DSCC produced the least milk and were more likely to be culled [20, 39, 40], which is attributable to the above-mentioned irreversible mammary tissue damage.

Even though in the present study we were unable to find a significant effect of CM severity on the dynamics of DSCC, we refrained from concluding that severity has no impact on DSCC. In this investigation, most severe mastitis cases had yet to enter the healing process at the end of follow-up, as shown by their high SCS values on d 21 in Fig. 1B. The sampling scheme applied here was insufficient to describe the dynamics of DSCC in severe mastitis, especially regarding the healing process. To carefully quantify this effect, an extended follow-up is required in the future.

To conclude, we investigated the dynamics of several immunological indices after CM occurred. We observed significant differences in the dynamics of WBC between mastitis caused by coliforms, streptococci, and staphylococci and the A/G ratio dropped dramatically only in severe mastitis. This study provides a novel insight: when analyzed together with SCS, DSCC can provide additional information regarding the udder’s health status. We suggest that future works on DSCC better consider the information provided by SCS in their analysis. Overall, the results indicate that immune responses in CM varied depending on both the severity and causative pathogen, supporting the importance of severity evaluation and pathogen identification in mastitis control. The results also help elucidate the pathogenesis of impaired productivity due to mastitis. For instance, we can understand why cows with low A/G ratios and cows with high SCS/ low DSCC tend to perform worse. Eventually, these findings can contribute to preventing cows from entering such a status.

CONFLICT OF INTEREST

The authors have nothing to disclose.