Haematological ndings in 158 dairy cows with toxic mastitis with a focus on the leukogram

Background: Acute toxic mastitis is characterised by typical clinical ndings and changes in the leukogram. The goal of our study was to compare the leukogram of 158 cows with toxic mastitis and 168 clinically healthy cows. The cows were examined clinically and underwent haematological and biochemical examination of blood and bacteriological culture of milk samples. Results: All cows with toxic mastitis were ill and had a poor appetite or anorexia, and 34 cows (21.5%) were recumbent. A single quarter was affected in 119 cows (75.3%), two quarters in 37 cows (23.4%) and three quarters in two cows (1.3%). Bacteriological culture showed gram-negative pathogens in 100 cows (63.3%), gram-positive in 15 (9.5%) and yeast in 4 (2.5%). The median total leukocyte count of ill cows was 4,300 cells/µl, which was signicantly lower than 8,000 cells/µl in controls. With the exception of band neutrophils and metamyelocytes, the counts of all components of the leukogram were lower in ill cows compared with controls. Signicantly more cows with toxic mastitis had leukopenia (60.1 vs. 4.1%) or leukocytosis (10.1 vs. 3.0%) than controls. Ill cows had signicantly lower segmented neutrophil counts than controls (860 vs. 2,598 cells/µl), and 69.5 and 17.3%, respectively, had counts below the reference interval. Ill cows had increased band (77.3%) and metamyelocyte (25.0%) counts compared with control cows (0.6 and 0%, respectively). In ill cows, eosinopenia occurred in 66.4% (controls, 1.8%), monocytopenia in 40.6% (controls, 4.2%) and lymphopenia in 60.2% (controls, 1.8%). Twenty-one ill cows (16.4%) had a regenerative and 57 (44.5%) had a degenerative left shift. The median neutrophil-to-lymphocyte ratio was 0.97 in ill cows and 0.63 in controls. Toxic changes in neutrophils including cytoplasmic basophilia and vacuolisation were seen in 101 (91.8%) of 110 blood smears of ill cows. Conclusions: Acute toxic mastitis results in severe changes in the leukogram particularly leukopenia, lymphopenia and degenerative left shift. The leukogram has signicant diagnostic value and may aid in the modication of treatment when needed.


Background
Acute toxic mastitis is characterised by typical clinical ndings and changes in the leukogram [1] that include leukopenia, neutropenia, degenerative left shift and neutrophils with toxic changes. The clinical diagnosis of acute toxic mastitis is straightforward and therefore a leukogram is not required for an initial diagnosis. However, it aids in determining the severity of the disease. Cows with toxic mastitis commonly have leukopenia and neutropenia, which is in contrast to other domestic animal species such as the dog in which pronounced neutrophilia with a degenerative left shift occur in response to an in ammatory stimulus. This response typically occurs within six to eight hours after the insult and the degree of response is directly related to the severity of the in ammation [2]. A thorough knowledge of the process of leukopoiesis aids in the interpretation of changes in the leukogram of cattle with acute in ammation.
Leukopoiesis involves the proliferation, maturation and storage of white blood cells [3,4]. Pluripotent stem cells differentiate into myeloblasts, which mature through a series of stages that include promyelocytes, myelocytes, metamyelocytes, band cells and mature granulocytes; this so-called proliferative pool accounts for approximately 10-30% and the maturation and storage pools for 65-90% of the granulocytes in the bone marrow [3]. The maturation pool includes metamyelocytes and band neutrophils, which have lost their mitotic potential, and the storage pool comprises band and segmented neutrophils ready to be released into the circulation. The normal transit time for maturation of a myeloblast is about 6-7 days. In several animal species, but not in cattle, the transit time can be reduced to 2-3 days when leukocyte consumption is high. Therefore, the supply of mature leukocytes from the bone marrow rapidly becomes insu cient in cows with high consumption of neutrophils at the site of in ammation. When interpreting a bovine leukogram, one must consider that bone marrow response is slower with increased demand for neutrophils [5] and the granulocyte storage pool is smaller compared with other species [6]. This often results in neutropenia, rather than neutrophilia, in the rst 24-48 hours after the onset of acute in ammation because neutrophils move out of the circulation and into the tissues and the bone marrow cannot meet demand adequately. Approximately 24 hours after the start of the in ammatory process, immature leukocytes including band neutrophils, metamyelocytes and occasionally myelocytes are seen in blood smears. The neutrophils often have cytoplasmic basophilia due to staining of certain organelles that are present during early development [7]. Cytoplasmic basophilia is referred to as a toxic change, but this is an unfortunate term because the cells have not been damaged by the pathogen and in fact function normally.
The effects of intramammary infection on the total leukocyte and neutrophil counts in cows have been examined experimentally [8][9][10][11][12]. Gram-negative pathogens including Escherichia coli, Klebsiella pneumoniae and Serratia marcescens resulted in neutropenia within 16 hours, whereas neutropenia did not occur until 36 hours after infection with Streptococcus uberis and did not occur with Staphylococcus aureus infection. Transient neutropenia was observed 24 hours after infection with Pseudomonas aeruginosa [11] and from 48 to 168 hours period after infection with Mycoplasma bovis, which also included lymphopenia from 84 hours post infection until the end of the 10-day study [12]. Cows with acute mastitis caused by gram-negative bacteria had signi cantly lower leukocyte, segmented neutrophil, monocyte and lymphocyte counts than cows with mastitis caused by gram-positive bacteria [13]. Transient neutropenia, often followed by rebound neutrophilia with subsequent normalisation of the neutrophil count, is a good prognostic sign, whereas a continued decrease in neutrophil numbers on day 2 of infection or neutropenia that persists for more than 3-4 days is a poor sign indicating bone marrow suppression [5].
Lymphopenia and monocytopenia are also typical ndings in acute mastitis [1]. Stress and exogenous administration of corticosteroids are the most common causes of leukopenia in cattle [5], but infection with viruses, Anaplasma phagocytophilum, mycoplasma and other microorganisms as well as septicaemia are other possible causes [14]. Monocytopenia is of lesser signi cance [15] and cell counts normalise more quickly because the transit time is only 3 days compared with 6 days for neutrophils. Therefore monitoring monocyte numbers can be useful for predicting the recovery in cases with bone marrow suppression. The goal of this study was to compare the leukograms of 158 cows with toxic mastitis and 168 clinically healthy cows and to discuss the differences.

Animals
One hundred and sixty-eight healthy controls and 158 cows with acute toxic mastitis were used. The controls consisted of 182 healthy cows that were the offspring of cows with BSE and therefore referred by the Federal Veterinary O ce Switzerland (now Federal Food Safety and Veterinary O ce) to our clinic for examination. Clinical and laboratory ndings of these cows were published [16]; and the cows were previously used as controls [17]; all cows had a history of being healthy and the ndings of daily clinical examinations during three consecutive days were normal. The 158 cows with toxic mastitis included 151 dairy cows and 7 beef cows that were referred to the Clinic for Ruminants, Vetsuisse Faculty, University of Zurich from 2004 to 2016. The dairy cows comprised Swiss Braunvieh (n = 55), Swiss Fleckvieh (n = 50) and Holstein Friesian cows (n = 46), and the beef cows belonged to different breeds or were cross bred.
The cows were 2.0 to 15.1 years of age (mean ± sd = 6.4 ± 2.6 years), and 13 cows were dry. Of the remaining 145 cows, 91 were between 0.5 and 30 days postpartum and 54 were longer than 30 days postpartum or the lactation stage was not known. The duration of illness varied from 0.5 to 30 days (median = 2 days). One hundred and forty-four cows (91%) had received treatment from the herd veterinarian, but only 109 (69%) had been diagnosed as having mastitis. The reasons for referral included request for diagnostic workup (n = 62, 39.2%), downer cow syndrome (n = 29, 18.4%), tentative diagnosis of left or right displaced abomasum (n = 27, 17.1%), acute toxic mastitis (n = 12, 7.5%) and other disorders including fever, diarrhoea and suspicion of traumatic reticuloperitonitis, ileus, caecal dilatation, bronchopneumonia and endocarditis (n = 28, 17.7%).
Two blood smears were prepared for each cow and stained using a modi ed Wright's stain in an automated staining system (HemaTek, Siemens, Switzerland). A manual differential of 100 leukocytes per smear was done by two technicians each with ten years of experience in veterinary haematology. The absolute numbers of lymphocytes, monocytes and granulocytes was determined by multiplying the percentage obtained in the 200-cell differential count with the total leukocyte count from the haematology analyzer. Additionally, the morphology of erythrocytes, leukocytes and platelets was evaluated. The microbiological examination of milk was done as described [19].

Statistical analysis
The program SPSS Version 24 was used for statistical calculations and analysis of the total leukocyte count, segmented, band and metamyelocyte neutrophils, eosinophils, basophils, monocytes and lymphocytes of both groups. Frequency distributions were determined for both groups and the median and the 5th and 95th percentiles were calculated because the distributions were not normal based on the Shapiro-Wilk test. Differences between medians and frequency distributions of both groups were analysed using ANOVA and the Mann-Whitney U test. To calculate the neutrophil-to-lymphocyte ratio, the segmented, band and metamyelocyte neutrophil counts were divided by the lymphocyte count. The distributions of the ratios were not normal and differences between the medians of both groups were analysed using the Mann-Whitney U test. Pearson correlation coe cients were calculated to assess the relationships between variables.
For the analysis of the leukograms in relation to mastitis pathogens, the latter were divided into 3 groups including gram-negative (Escherichia coli, Klebsiella spp., Proteus spp.) and gram-positive bacteria (Staphylococcus aureus, other Staphylococcus spp., Streptococcus uberis, other Streptococcus spp., Trueperella pyogenes) and yeast. Differences between medians and frequency distributions of the leukograms of the three groups of pathogens were analysed using ANOVA and the Bonferroni post hoc test.
Frequency distributions were determined for clinical variables including changes in the udder and the milk of the diseased cows, and the median and the 5th and 95th percentiles were calculated for rectal temperature, heart rate and respiratory rate.
Udder and milk abnormalities One hundred and nineteen cows (75.3%) had toxic mastitis in a single quarter, 37 cows (23.4%) in two quarters and two cows (1.3%) in three quarters. The affected quarters were enlarged and rm in 157 (99.4%) cows. In seven cows (4.4%), the skin overlying the quarter was hyperaemic, and in 11 cows (7.0%) the teat was affected by the in ammation. The colour and consistency of the milk were markedly abnormal in 152 (96.2%) and 153 cows (96.8%), respectively, and clots were visible in the secretion of 128 cows (81.0%). A California mastitis test score of 3 positive (almost a solid gel) was seen in all affected quarters.

White blood cells, medians and frequency distributions
The leukocyte count was signi cantly lower in cows with mastitis compared with controls (4,300 versus 8,000 cell/µl; P < 0.01) ( Table 1); with the exception of band neutrophils and metamyelocytes, all components of the leukogram were signi cantly lower in the ill cows (P < 0.01) (Fig. 1). The frequency distributions also differed between the two groups (P < 0.01) (Fig. 2); cows with toxaemia had leukopenia (60.1%) or leucocytosis (10.1%) (P < 0.01) signi cantly more often than controls (4.1%, 3.0%, respectively). The number of segmented neutrophils was signi cantly lower in 69.5% of ill cows (860 cells/µl) compared with 17.3% of controls (2,598 cells/µl) (Fig. 3). Increased numbers of segmented neutrophils were seen in 16.4% of ill cows compared with 4.8% of controls. Band neutrophils were seen in 77.3% of ill cows (median, 530 cells/µl) and in 0.6% of controls (median, 0 cells/ µl) (P < 0.01) (Fig. 4). Metamyelocytes were seen in 25% of ill cows but did not occur in controls (Fig. 5). Eosinopenia was seen in 66.4% of ill cows and in 1.8% of controls (P < 0.01) (Fig. 6). Basophils were in the normal range (0 to 200 cells/µl) in all ill cows and in 97.0% of controls. Monocytopenia (< 100 cells/µl) was seen in 40.6% of ill cows (155 cells/µl) compared with 4.2% of controls (336 cells/µl) (P < 0.01) (Fig. 7). Lymphopenia occurred in 60.2% of ill cows (1,550 cells/µl) compared with 1.8% of controls (3,998 cells/µl, P < 0.01) (Fig. 8).  The medians and frequency distributions of all components of the leukogram differ signi cantly between the two groups (P < 0.01, ANOVA, Mann-Whitney U test) 1 Reference intervals for all variables: Wood and Quiroz-Rocha [18] Frequency distributions of leukocytes relative to the total leukocyte count and left shift Of the cows with leukopenia, 95.5% had decreased numbers of segmented neutrophils, 72.7% had increased numbers of band neutrophils (  Neutrophil-to-lymphocyte ratio The median neutrophil-to-lymphocyte (NL) ratio in controls was 0.63 and ranged from 0.28 to 1.53 (5th to 95th percentiles), which was lower than that of ill cows (0.97, P < 0.01, Mann-Whitney U test). The 5th percentile (0.13) was lower and the 95th percentile was higher (7.46) than in controls. The frequency distributions of the NL ratio differed between the two groups (P < 0.01, chi-square test, Fig. 9).
Toxic changes in neutrophils Toxic changes in neutrophils that included cytoplasmic basophilia and vacuolisation were detected in 101 (91.8%) of 110 blood smears. In 58 cows, the occurrence of toxic changes was quanti ed; 11 cows had changes in 5 to 10% of neutrophils, 9 had changes in 20 to 30% and 38 had changes in greater than 30% of the neutrophils.
Effect of gram-negative and gram-positive pathogens and yeast on the leukogram Differences between the medians and frequency distributions of the total leukocyte counts and the numbers of the components of the leukogram associated with gram-negative (n = 100), gram-positive (n = 15) and yeast infection (n = 4) were not signi cant (Table 3), but there were several signi cant differences between cows of the different pathogen groups and controls. Lymphocyte counts were signi cantly lower in all three groups, and cows with gram-negative infections had signi cantly lower total leukocyte, segmented neutrophil and monocyte counts and signi cantly higher band neutrophil counts than controls. Cows with gram-positive infections had signi cantly greater band neutrophil and signi cantly lower eosinophil, basophil and monocyte counts.  There were signi cant differences in total leukocyte and segmented neutrophil counts among cows with a duration of illness of one, two, three or more than three days (Table 4). Cows examined on the rst day of illness had signi cantly lower total leukocyte (3,250 vs. 9,200 cells/µl) and segmented neutrophil counts (610 vs. 6,170 cells/µl, P < 0.05) than cows that had been ill for more than three days. The counts of other components of the leukogram were not related to duration of illness.  Other laboratory variables Cows with toxic mastitis had a signi cantly higher haematocrit (34 vs. 33%) and signi cantly higher concentrations of brinogen (6.0 vs. 5.0 g/l), urea (7.9 vs. 4.5 mmol/l) and sodium (146 vs. 145 mmol/l) and a signi cantly lower total erythrocyte count (6.5 vs. 7.3 × 10 6 cells/µl) and lower concentrations of total protein (74 vs. 79 g/l), potassium (3.7 vs. 4.2 mmol/l) and calcium (2.3 vs. 2.4 mmol/l) than controls (Table 5).

Discussion
This study has shown that the total leukocyte count in cows with toxic mastitis is considerably lower than in healthy cows. Approximately 60.1% of the cows with mastitis had leukopenia and only 10.1% had leukocytosis. Segmented neutrophil counts were similar and were decreased in about 69.5% and increased in about 16.4% of the cows. Acute in ammation is generally accompanied by leukocytosis rather than leukopenia, but our ndings are in agreement with published reports of clinical [1,13,19] as well as experimental acute mastitis [8][9][10][11][12]. The reason for the leukopenia is the low bone marrow reserve of neutrophils in cattle as opposed to other domestic animal species such as the dog [6] and the fact that the transit time from myeloblast to complete maturation is not reduced in the face of high neutrophil demand [5]. In the dog the transit time can be reduced from 6 to 2 to 3 days with increased demand of segmented neutrophils but not in calves [3] and in all likelihood not in adult cattle either. With acute mastitis, large numbers of segmented neutrophils are moved from the circulation into the infected mammary tissue causing leukopenia. This is followed by an increase in leukocyte numbers after about 48 hours, often associated with a rebound neutrophilia [5]. For these reasons, the leukocyte count and particularly the segmented neutrophil count were lower in the cows that were examined on the rst day of illness compared with cows that were referred and sampled at a later stage of illness. Severe leukocytosis with cell counts exceeding 20,000 cells/µl occurred in only two cows (1.3%), which was in agreement with the observation that cattle as well as sheep and goats with acute in ammation have much lower leukocyte peaks than other domestic animal species [14]. A leukocyte count between 20,000 and 30,000 cells/µl is considered extremely high in cattle [14] and only 16.4% of ill cows had neutrophilia in the present study. Stress is another cause of neutrophilia. Stress leukograms caused by endogenous or exogenous corticosteroids are common in cattle [14] and are characterised by neutrophilia, lymphopenia, eosinopenia and monocytosis [14,20]. However, a stress leukogram is not associated with a left shift and therefore band neutrophils and metamyelocytes are not seen. In the present study, neutrophilia was always accompanied by a regenerative left shift in affected cows, which means that neutrophilia was due to in ammation and not stress. Degenerative left shift, characterised by an abundance of band neutrophils and metamyelocytes relative to segmented neutrophils, was more common than regenerative left shift and occurred in 44.5% of ill cows. Degenerative left shift with increased numbers of metamyelocytes is an alarming haematological nding and a poor prognostic sign when it persists for more than 3 to 4 days [5].
Lymphopenia was detected in 60.4% of all cows with toxic mastitis. Lymphopenia in cattle is most commonly caused by endogenous or exogenous corticosteroids [5], and in the present study was due to toxic mastitis. This was also observed after experimental intramammary infection with Pseudomonas aeruginosa [11] and Mycoplasma bovis [12]. Interestingly, lymphopenia occurred as early as 24 hours after Pseudomonas aeruginosa inoculation but only 84 hours after Mycoplasma bovis inoculation. Lymphopenia is also seen in viral infections and diseases caused by Anaplasma spp. and other bacteria [14].
Metamyelocytes were not detected in blood smears from healthy cows, which was in agreement with the veterinary literature [21]. The occurrence of metamyelocytes should therefore be considered pathological and indicative of a massive demand for leukocytes.
In 104 cows (62%), toxic mastitis occurred during late pregnancy or during the puerperal period, which most likely was related to impaired neutrophil function in the periparturient period [22,23] described previously [24,25]. It is believed that this is related to increased stress and thus increased glucocorticoid concentrations at this lactation stage [26]. Furthermore, recruitment of neutrophils into the mammary gland and neutrophil function are altered in the periparturient period, which may also contribute to the occurrence of mastitis in early-lactation cows [22,24,27]. This is supported by the observation that experimental intramammary infusion of endotoxin results in more severe clinical signs in early lactation compared with late lactation in cows [28].
In contrast to other domestic animal species, cows have more lymphocytes than neutrophils in the circulation and therefore a relatively low NL ratio of about 0.5 [6,14]; in the present study, the controls had a median NL ratio of 0.63. The signi cantly larger NL ratio of 0.97 in the cows with toxic mastitis was due primarily to lymphopenia (in 60.2%) rather than neutrophilia (only 16.4%). Similarly, cows with sole ulcer had a NL ratio of 1.04, which was signi cantly greater than that of healthy cows even though neutrophil and lymphocyte counts did not differ signi cantly between the groups [29]. It was therefore recommended to include the NL ratio in the assessment of a blood cell count to emphasise the difference in absolute neutrophil and lymphocyte numbers [5].
Cytoplasmic basophilia combined with vacuolisation in all but 8.2% of the examined blood smears was indicative of toxic change [5]. This con rms that the cows with mastitis had toxaemia because the changes are generated when maturation of neutrophils in the bone marrow is hindered because of acute and severe in ammation [14].
Cows with mastitis caused by gram-negative and gram-positive bacteria did not differ signi cantly with respect to total leukocyte, segmented and band neutrophil, monocyte and lymphocyte counts. Differences in cell counts between cows with gram-negative and gram-positive infections were signi cant in an earlier study but the number of cows was larger in that study [13]. The observation that cows with gram-positive infection tended to have lower total leukocyte and segmented neutrophil counts and signi cantly lower eosinophil, basophil, monocyte and lymphocyte counts than controls indicates that mastitis caused by gram-positive pathogens constitutes a severe strain on the bone marrow. It should be remembered that differentiation of mastitis caused by gram-negative and gram-positive pathogens requires bacteriological examination and is not possible based on clinical signs alone [1].
Increased haematocrit and azotaemia in cows with toxic mastitis was in all likelihood attributable to prerenal causes. The increased brinogen concentration points to the role of brinogen as a positive acute-phase protein [18,30]. Similar to haptoglobin, it increases rapidly in response to in ammation and was therefore higher in cows that were ill longer than three days compared with cows on the rst day of illness. In cattle, hyper brinogenaemia is considered as good as or better than the neutrophil count for determining in ammation [18]. As a general rule, the degree of increase in concentration of positive acutephase proteins parallels the severity of in ammation. Fibrinogen is preferred by many because it is easy to measure.
We were surprised that only about 69% of all toxic mastitis cases were diagnosed by the primary care veterinarian. This may have been due to other clinical signs, such as poor general health status, poor rumen motility, low faecal output and positive percussion and/or ballottement and simultaneous auscultation, which were considered diagnostic of other disorders. This emphasises the importance of a comprehensive clinical examination that includes the assessment of the udder and the milk.

Conclusions
Acute toxic mastitis is accompanied by severe changes in the leukogram, chie y leukopenia and degenerative left shift. The leukogram has signi cant diagnostic value and may aid in the modi cation of treatment when needed. Leukograms similar to those presented in this study strongly suggest toxic mastitis or toxaemia of another aetiology when mastitis can be ruled out clinically.

Declarations
Ethics approval and consent to participate The study was not submitted to our institutional ethics committee, because it was a retrospective study using medical records of cows that had been referred to our clinic for examination and treatment. No experimental animals were used. Under these circumstances, ethical approval is not required in Switzerland. Written consent was obtained from all owners to participate.

Consent for publication
All owners signed a consent form allowing us to use the animals and all the associated medical data for scienti c analysis and publication.
Availability of data and materials The datasets used and analysed for this study are available from the corresponding author on reasonable request.

Competing interests
The authors declare that they have no competing interests.

Fundings
Not applicable since it was a retrospective analysis of medical records.
Author's contributions UBr initiated, planned and supervised the study and prepared the manuscript. CG, SB and UBl were involved in the clinical examination and treatment of the cows. BR was responsible for the haematologic examination and SC for the bacteriologic examination of milk. All authors made contributions to acquisition and interpretation of data and were involved in drafting the manuscript. All authors read and approved the nal manuscript.