Outcomes after treatment of nonsevere Gram-negative clinical mastitis with ceftiofur hydrochloride for 2 or 5 days compared to negative control

A study was conducted at 3 commercial dairies in California to compare outcomes of treating nonsevere (mild and moderate) Gram-negative ( GN ) clinical mastitis ( CM ) with intramammary ( IMM ) ceftiofur HCl (125 mg ceftiofur HCl per tube) in either 2-d ( SP2 ) or 5-d ( SP5 ) treatment programs compared with nontreatment ( CON ). In addition, we contrasted results from cases classified as mild and moderate. Four hundred 15 cases were included in the final data set, including 135 CON, 133 SP2, and 147 SP5. Milk from quarters with CM was sampled for on-farm culture ( OFC ) to differentiate Gram-positive ( GP ) and GN bacteria, with results known within 24 h. Those with GN infections were randomly assigned to experimental groups, while those with GP, mixed infections, and contaminated samples did not continue in the study and received standard farm therapy. For cows with GN infections, a sample was submitted for MALDI-TOF assay. Only nonsevere cases were enrolled, and all quarters yielded monocultures of GN species. Clinical scores ( CS ) were obtained 0, 1, 2, 3, 4, 5, 14, 21, and 28 ± 3 d relative to enrollment. Milk samples were collected from quarters 14, 21, and 28 ± 3 d after enrollment, and submitted for routine culture and, when appropriate, submitted to MALDI-TOF evaluation. For many response criteria, there were significant interactions between treatments and CM severity scores at the time of enrollment, with effectiveness of ceftiofur HCl treat-ment being more beneficial compared with CON as mastitis clinical severity increased. While most treat-ment responses were significant for animals with mild or moderate GN mastitis, the largest responses were noted among cows with moderate CM cases.


INTRODUCTION
Bovine mastitis is a common and costly disease that negatively impacts dairy farms due to production losses, increased health costs, decreased milk quality, negative effects on cow welfare, and increased culling and death rates (Cha et al., 2011).Clinical mastitis (CM) can be classified as mild, moderate, or severe according to its clinical presentation, which can vary from just abnormal milk to the presence of systemic signs (International Dairy Federation, 1999).Inflammation, characterized by redness, swelling, heat, and pain, is triggered by various factors, including pathogens.The degree and nature of immune responses are likely proportional to infection severity (Sharma, 2011).Although exacerbated in severe cases, cows with mild and moderate cases of mastitis can also experience pain (Leslie and Petersson-Wolfe, 2012), which impacts animal welfare.
Various pathogens, including environmental and contagious microorganisms, can cause mastitis.Gramnegative (GN) bacteria, especially coliforms (Escherichia coli, Klebsiella spp., and Enterobacter spp.), are responsible for about 40% of all intramammary infections (IMI; Schukken et al., 2012;Oliveira et al., 2013) and can be a risk to a herd's productivity.If left untreated, mild and moderate GN cases can become severe and toxic and have a negative impact on cows (Erskine et al.,1993).Milk somatic cell count (SCC) and microbiological cultures are used as diagnostic tools for mastitis and to guide treatment choices best suited to impacting bacteriological and clinical cures (Williamson et al., 2022).
Antibiotics are frequently used on lactating cows presenting mild and moderate cases, while additional support therapy is typically administered to severe cases.The intent of treating CM with antibiotics is to eliminate microorganisms from the infected quarter, preventing further damage to mammary tissue, avoiding a decrease in future milk production, reducing recovery time, and returning the quarter to healthy status (Tomazi et al., 2018).Furthermore, antimicrobial usage has significantly improved the well-being of animals, resulting in better health and higher productivity (Hao et al., 2014).Despite the benefits, there is criticism of antibiotic use due to the risk of promoting antimicrobial resistance (AMR), which has prompted government and public health officials to increase control of antimicrobials used for food animals (FAO, 2021).This has motivated researchers to provide justification for the use of antimicrobials to treat specific infections.Moreover, studies have suggested that selective therapy based on on-farm culture (OFC) results provides an opportunity to reduce antimicrobial use on dairy farms (Lago et al., 2011ab).
Various antimicrobials are available to treat IMI in dairy herds (Ruegg, 2017).Ceftiofur hydrochloride (ceftiofur HCl), a third-generation, β-lactamase-stable, broad-spectrum cephalosporin, has been used in veterinary medicine worldwide (Hornish and Kotarski, 2002) and is approved for treatment of E. coli infections (Schukken et al., 2011;Truchetti et al., 2014).A recent review (Ruegg, 2021) on the use of antimicrobials for the treatment of CM identified a limited number of field trials evaluating antimicrobial efficacy exclusively against GN infections involving nonsevere (Schukken et al., 2011;Fuenzalida and Ruegg, 2019) or acute (Suojala et al., 2010;Persson et al., 2015) cases.Other studies also evaluated the efficacy of intramammary (IMM) therapy with ceftiofur HCl against GN infections; however, the proportion of GN infections was low (9-25%) and did not include a negative (nontreated) control (Schukken et al., 2013;Truchetti et al., 2014;Cortinhas et al., 2016;Vasquez et al., 2016;Viveros et al., 2018).Research has shown an increased bacteriological cure (BC) rate when comparing treated vs. untreated quarters infected with E. coli (Schukken et al., 2011) and Klebsiella spp.(Fuenzalida and Ruegg, 2019).Although negative controls were not included in other studies, differences in BC were shown to be dependent on the number of d cows were treated with ceftiofur HCl (Truchetti et al., 2014); others did not find a significant difference when comparing different antimicrobials (Schukken et al., 2013;Vasquez et al., 2016;Viveros et al., 2018).Results for clinical cure (CC) and other outcomes also varied among the studies mentioned above.In addition, only a few studies evaluated long-term outcomes of GN infections.Hence, there is mixed evidence regarding the benefits of treating nonsevere GN mastitis cases with ceftiofur HCl.
Objectives of the present study were (1) to evaluate the outcomes of 2 IMM ceftiofur HCl therapy regimens compared with nontreatment of nonsevere CM cases caused by GN bacteria,(2) to contrast the outcomes for mild and moderate cases; and (3) to assess overall treatment outcomes according to severity at the time of diagnosis.

Study Design and Herd details
A randomized, negative-controlled field trial was conducted from December 2020 to August 2022 using dairy cattle at 3 commercial dairies in California operated under routine management practices according to 7 U.S.C. 54 and FASS (2010) following a protocol approved by the University of California Merced Institutional Animal Care and Use Committee (Protocol # AUP-20-0014).This protocol was replicated at each participating dairy, and data were pooled across sites for final analysis.Herd details are described in Table 1.To be eligible for the study, selected dairies were required to remain in compliance with the study protocol, routinely perform OFC, have good record keeping, trained personnel, individual animal identification, treatment facilities, appropriate drug storage facilities, and refrigeration and freezer capacity.The research team visited dairies each week throughout the study to perform animal observations, collect milk samples, and monitor record-keeping practices by farm personnel.All dairies utilized DairyComp 305 (DC305; Valley Agricultural Software, Tulare, CA) to maintain herd records.

Cow Enrollment and Treatment Groups
Lactating cows of all parities without prior cases of CM in the current lactation and without severe teat lesions were eligible for enrollment in the study.Before beginning the study, researchers trained farm personnel to identify CM, to aseptically collect milk samples, and to use a previously described clinical scoring (CS) system (Vasquez et al., 2016).Briefly, CM cases were classified as mild (CS = 1 -milk visually abnormal (watery, flakes, clots or bloody) and no other local or systemic signs of inflammation), moderate (CS = 2milk abnormal and udder had signs of inflammation: swelling, redness, heat, hardness, or pain), or severe (CS = 3 -milk grossly abnormal, udder had signs of inflammation and signs of systemic disease were present, e.g., fever, dehydration, or depression).Only nonsevere (mild and moderate) GN cases of CM were eligible for enrollment in this study.Severe cases were not included in the study and were treated according to farm Standard Operating Procedures (SOP).In addition, cows entering the postpartum period with evidence of injuries and illnesses other than mastitis that would adversely impact the likelihood of completing the study and any Bruno et al.: Treatment of Gram-negative mastitis cow undergoing systemic antibiotic therapy at the time of initiation of a mastitis event were not eligible.
Cows identified with CM were assigned an enrollment clinical score (ECS), and pre-enrollment milk samples were collected from the affected quarter(s) the day CM was identified (d 0) using standard aseptic techniques (NMC, 1999).Milk samples were subjected to OFC for differentiation of GP and GN organisms and used for pre-screening cows for enrollment.Two of the 3 dairies had a long history of using OFC before the study.The third dairy, not previously using OFC, was methodically trained before study implementation.Moreover, all dairies followed an OFC previously described (Lago et al., 2011a).Briefly, aseptically collected milk samples from the affected quarters were plated on a bi-plate (Minnesota Easy Culture System, University of Minnesota, St. Paul) at Dairies A and C, or on Accumast plates (FERA Diagnostics and Biologicals LCC, College Station, TX) at Dairy B. Plates were placed in on-farm incubators and incubated at approximately 37°C for 18 to 24 h.After inoculation onto OFC plates, milk samples were frozen at −18°C and later submitted to a diagnostic laboratory for bacteria confirmation by MALDI-TOF testing.Samples were also cultured for Mycoplasma spp.Cases of CM with OFC plates yielding no growth (NG), GP bacteria, or containing more than one bacterial species (mixed culture or contamination) were not enrolled in the study and were treated per dairy's SOP.Quarters were only enrolled once in the study during a lactation, but cows with CM identi-fied in more than one quarter in the same lactation were also eligible to participate in the study.In these cases, quarters were enrolled to the same experimental group.After enrollment, CM cases previously identified as GN by OFC but not confirmed by MALDI-TOF were excluded from the study and treated by the dairy following their SOP.For cows started on ceftiofur HCl that MALDI-TOF showed not to be infected by GN bacteria, the dairy modified the prescribed treatment in DC305 to follow an appropriate GP treatment protocol.If a cow was initially enrolled in the CON group and not confirmed to be a GN case, the dairy immediately started the protocol according to their SOP for that specific pathogen.The final determinant for the inclusion of a CM case in the study was based on the confirmation of GN infection in the pretreatment milk sample by MALDI-TOF.After enrollment, cows were evaluated daily for clinical signs of mastitis by trained farm personnel, and cows that developed severe symptoms were treated according to the dairy's SOP.These cows were included in final data and included in the overall treatment success analysis.
Randomization was performed using a previously prepared spreadsheet using the RAND function of Excel software (Microsoft Office Corporation, Redmond, WA).Weekly visits were made by research personnel, and the spreadsheet for treatment enrollment was inspected to ensure protocol compliance.Cows meeting enrollment criteria were randomly assigned to one of 3 treatment groups: SP2: IMM treatment with ceftiofur  , 1999), and cows treated with antibiotics were marked using a leg band for easier identification per farm SOP.Milk from all enrolled cows was discarded until it returned to normal (nontreated group) or until the end of the 72-h milk-withholding period after the end of antimicrobial therapy.Days in milk (DIM) and parity number (lactation ≤ 2 or ≥ 3) were documented.
Given the nature of the study and treatment structure, blinding was not possible, as individuals treating cows needed to know which cows received IMM therapy to follow appropriate milk and meat withholding times.However, laboratory personnel were blinded to treatment and details of the study, and the farm personnel were unaware of study outcomes until study completion.

Post-Treatment Assessments and Sample Collection
Post-enrollment milk samples were collected from enrolled quarters 14, 21, and 28 ± 3 d after enrollment into 2 vials: one for assessment of bacteriological outcomes and the second for assessment of milk components.Samples were identified with cow number, quarter, date, and dairy.Samples for bacterial culture and MALDI-TOF were chilled on ice or frozen at −18°C until sent to the laboratory for microbiological evaluation.Milk samples for quantification of components (milk composition (MC); fat, protein, lactose, solids-not-fat (SNF), milk urea nitrogen (MUN), and SCC were collected in vials containing Bronopol (Microtabs II, D&F Control Stems Inc., Norwood, MA) according to International Dairy Federation (1999), and evaluated by the Dairy Herd Improvement Laboratory within 24-48 h.
Clinical assessments were performed by farm personnel and/or the research team for the first 5 d after enrollment, then at 14, 21, and 28 ± 3 d after enrollment.Cows were examined during milking, and the same CS used at enrollment was used for post-treatment clinical assessments.A CS 0 was included to classify cows as normal (normal milk appearance and absence of signs of inflammation).Cows were monitored for up to 90 d after enrollment or until dried off or culled/ died.General cow behavior was also observed by the hospital barn manager.DairyComp 305 was utilized to document all clinical observations, treatments, and follow-up measurements.

Outcomes
Outcomes evaluated in this study included BC, CC, days to clinical cure (DCC), mastitis recurrence (MR), new intramammary infections (NIMI), milk composition, SCC, need for supplemental IMM or systemic therapy, culling/death, and overall treatment success.
A BC was achieved if milk samples collected at 14 ± 3 d post-enrollment were negative for the infecting GN species present at enrollment, and if there was no additional IMM or systemic antibiotic therapy.If GN bacteria, regardless of species, were isolated in the 14 ± 3 d samples, the case was considered a treatment failure.If no bacteria were recovered 14 d after enrollment but were recovered 21 or 28 d after enrollment along with CM signs, it was considered a treatment success, but incurred a NIMI.If a cow received supplemental IMM or systemic antibiotic therapy before 14 ± 3 d after enrollment, the 14 d sample was considered compromised, and bacteriological data were not included in the final analysis.If a cow received IMM or systemic antibiotic therapy after d 14 for mastitis due to a GP organism but had a NG culture result on d 14, it was considered a BC; this therapy may have affected d 21 or d 28 milk culture results but did not affect d 14 milk culture results.If cows died or were culled before 14 ± 3 d after enrollment due to GN mastitis-related reasons, data were excluded from BC analysis, but were accounted for in the culling and death analysis and considered a treatment failure.
Quarters were considered clinically cured when milk and udder appearance returned to normal within 14 d after enrollment.Days until clinical cure was defined as the number of d from first identification of CM until milk returned to normal, along with the absence of clinical signs.Any abnormality at follow-up evaluations within 14 ± 3 d was considered a treatment failure with respect to clinical assessment.Cases enrolled based on CS, treated (or not) according to the study protocol, declared CC, then returned to being milked with the herd but later observed with abnormal milk and or/ signs of inflammation within 14 d of enrollment with or without culture and re-treated (or not) or culled, were considered treatment failures.Mastitis recurrence was defined when CM was identified in the same quarter at least 14 d after enrollment.Information was documented in DC305, including the date of detection, OFC results when available, and treatment.
Other observations recorded included the date a cow was declared clinically well, date of reinfection after d 14, d between initial mastitis and reinfection, OFC results of the reinfecting organism (if performed/documented), remarks in the DC305 cowcard, supplemental IMM and/or systemic therapy, including drug names, Bruno et al.: Treatment of Gram-negative mastitis dates, and length of treatment.All this information was assessed as part of the process of determining whether a BC or CC was affected.Clinical outcome success was declared if CS improved from 1 to 0 or 2 to 0 within 14 d.
Milk composition was evaluated at the quarter level during post-treatment assessments 14, 21, and 28 ± 3 d after enrollment and at a composite (cow) level monthly in Dairies A and B up to 90 d after enrollment.Milk SCC were converted to linear scores (LSSCC) for statistical analysis by the formula: LSSCC = log 2 (SCC/100) + 3 (Schukken et al., 2003).In addition, data from DHIA test dates relative to when animals were diagnosed with GN mastitis, DIM on each test day, test day fat corrected milk (FCM) yield (kg/d), and LSSCC were extracted from DC305 for the test day preceding mastitis diagnosis and 4 test dates following mastitis treatment.
Data pertaining to death and culling were retrieved from DC305.Duration animals remained in the study was calculated based on the time elapsed between entrance into the study and exit from the study due to death or culling as previously described (Schukken et al., 2011) up to 90 d following enrollment.
Overall treatment success was declared for each cow that had BC at d 14, had CC by d 14, was not culled, and did not die during study evaluation (<28 ± 3 d after enrollment), had no MR, and did not receive other IMM or systemic antibiotic treatment before 14 d.Treatment failure was declared for cows that never improved clinically, went on to develop a severe case of mastitis, required antibiotic or supportive therapy other than experimental treatment, developed severe clinical signs, died or were culled due to the mastitis case, lost a teat due to the mastitis event, had another case of GN mastitis before 14 d after enrollment, or did not experience a BC at 14 d after enrollment.Cows that developed signs of severe mastitis during the study were considered treatment failures, returned to standard dairy management practices, and treated by dairy personnel according to established veterinary treatment protocols.

Statistical Analysis
Data analysis was facilitated using a recognized biometrics program (SAS Release 9.4, SAS Institute, Cary, NC), and quarter was the experimental unit.Descriptive statistics were performed using PROC FREQ.Binomially distributed categorical data were analyzed by PROC GLIMMIX, and continuous data were analyzed by PROC MIXED to evaluate fixed effects of treatment (CON, SP2, SP5), parity (lactation ≤ 2 or lactation ≥ 3), treatment by parity interaction, ECS (1 or 2), ECS by parity interaction, mastitis DIM; and random effects of farm (A, B, C), cow breed (Holstein, Jersey), and season of the year (spring, summer, fall, winter).ECS was used as a covariate.Treatment differences were evaluated with a 2-tailed t-test to determine whether responses by cows treated with SP2 or SP5 differed from CON, and whether SP2 and SP5 were different from each other.Treatment effects were considered significant if P ≤ 0.05; if 0.05 > P ≤ 0.10, the treatment effect was considered a tendency.Adjusted odds ratios and respective 95% confidence intervals were calculated.
Because they are ordinal data with a multinomial distribution, CS were analyzed using PROC GLIMMIX to evaluate fixed effects of treatment, parity, treatment by parity, ECS, and treatment by ECS; and random effects of dairy, breed, and season of the year.Data were fit to a cumulative logit proportional-odds model, treatment effects were assessed, and odds ratios were computed to provide an interpretation of effects.The LIFETEST procedure of SAS was used to calculate survival probabilities for time-to-event outcomes, where time was defined as the number of d until the event of interest (culling or death due to GN CM) occurred.Censoring was used for cows that completed the study through 90 d post-enrollment (maximum survival was 90 d).GraphPad Prism 8 software (GraphPad Software Inc., LaJolla, CA) was used to create a Kaplan-Meier survival plot to visualize time cow remained in the study in each treatment group.

Descriptive Data
A total of 423 quarters that exhibited signs of mild to moderate CM and were infected with GN organisms were randomized to treatments between December 2, 2020 and July 3, 2022.Of these 423 quarters (135 CON, 133 SP2, 155 SP5), 8 quarters from SP5 cows were excluded from the final analysis because of deviations from the treatment protocol.Final enrollment in data analysis included 135 CON, 133 SP2, and 147 SP5 quarters.Details of enrollments to treatments at each study site are summarized in Table 2.Because Site A was a Jersey herd and Sites B and C were Holstein herds there is confounding of breed and study site, but these factors were considered random site effects in the statistical model and are appropriately accounted for.Across farms, animals in treatment and control groups did not differ significantly in parity, DIM, milk production, or SCC at the time of enrollment.No significant difference was present among groups at the onset of treatments.Table 3 summarizes the herd-wide distri- The number of quarters infected with each genus of GN bacteria was relatively uniformly distributed across treatments (Table 2).Escherichia coli was the most common GN organism identified in CM cases (362/415; CON = 117, SP2 = 116, SP5 = 129) followed by Klebsiella spp.(K.oxytoca, K. pneumonia, and K. aerogenes) (26/415; CON = 11; SP2 = 8, SP5 = 7).Other isolates found in low numbers were combined into an "others" category (27/415; CON = 7, SP2 = 9, SP5 = 11).Importantly, a single species was cultured from each of the quarters enrolled, and no mixed cultures were noted.Mycoplasma spp.were not isolated from any milk sample.

Bacteriological Cures.
The overall effect of treatments on BC rates is represented in Table 4. Due to the number of cases enrolled in the study, there was adequate replication within treatment by ECS combinations for meaningful analysis.Thirty-seven cows (CON = 18; SP2 = 15; SP5 = 4) received supplemental IMM and/or support therapy before 14 ± 3 d after enrollment, were removed from BC analysis, and were considered treatment failures.Unfortunately, OFC was not performed for all cases  (CON = 7: 3 GN and 4 NG; SP2 = 4: 2 NG,1 GP and 1 mixed; SP5 = 4: 2 NG, 1 GP, 1 mixed).
Results showed significant effects of treatment (P = 0.0008) and parity (P = 0.0333), and a tendency for ECS (P = 0.0937).Bacteriological cure rate was higher for treated groups compared with controls (P < 0.05), and higher (P < 0.05) for quarters of older (or lactation ≥ 3) than for quarters of younger cows (lactation ≤ 2).There was also a BC response difference between ECS1 and ECS2 quarters within the same group.For ECS 1 quarters there were no differences between treatments.Among ECS 2 quarters, however, BC rates for both SP2 and SP5 were higher (P < 0.05) than CON and not different from each other.Figure 1 presents the treatment by ECS interaction on BC rates.Table 7 summarizes main effects of treatment on BC rates across ECS.Quarters treated with SP2 or SP5 responded with higher BC than CON (P < 0.05).

Clinical Assessments
Clinical assessment of quarters from the time of enrollment through 28 ± 3 d post-enrollment is summarized in Table 4.With respect to mean clinical scores at individual observation points, significant treatment effects were noted at observation points on d 2, 5, and 14 (P < 0.05) but not on d 21 or 28.Similarly, significant effects of ECS were noted at observation points on d 2, 5, and 14 but not on d 21 or 28.The interaction of treatment and ECS was significant only on d 14. Figure 2 is a graphic presentation of these mean clinical scores by treatment and ECS.Table 5 presents treatment comparisons based on odds ratios derived from analysis of the multinomial CS data, including exponentiated odds ratio estimates for treatment and ECS comparisons.An example of how to interpret data follows.At CS2, the comparison of CON to SP2 had an odds ratio of 2.32, which indicates that the odds of SP2 quarters having lower CS than CON quarters was increased by 2.32-fold, and this effect was significant (P = 0.0161).Examination of all comparisons of CON to SP2 showed increased odds of lower scores for SP2 at d 5 but not for d 14, 21, and 28.Odds that SP5 quarters would have lower CS than CON quarters were significant at 2, 5, and 14 d after the start of treatment.There was no significant likelihood of lower scores by SP5 quarters compared with SP2 quarters.
Odds ratios comparing ECS1 to ECS2 are also shown in Table 5.The odds that quarters with ECS of 2 have lower CS than quarters with ECS of 1 were significant on d 2, 5, and 14 (P < 0.05).Overall, ECS scores accounted for a significant proportion of the variation in data through 14 d following the start of treatment.Thereafter, CS progressed lower across both ECS levels, and the impact that ECS had on CS was negligible.
Perhaps a better indicator of clinical treatment outcomes is the success rate on CS associated with treatment, defined as the proportion of quarters in each group that experienced a reduction in CS from 2 to 0 or 1 to 0 by d 14 (Table 4).By this standard, there was a significant effect of treatment (P < 0.0001) and ECS (P < 0.0001), and the treatment by ECS interaction was not significant.Thus, treatments behaved similarly among quarters with ECS of either 1 or 2. As with other measures of treatment success, the response to treatments was more robust when ECS was 2 rather than 1.Regarding clinical outcome success, CON quarters with ECS of 1 had a lower success rate than SP2 and SP5 (P < 0.05).Among quarters with ECS of 2, CON cows had a treatment success rate inferior (P < 0.05) to the success rate among quarters treated with SP2 or SP5; SP5 was numerically higher than SP2 but was not significantly different.Table 7 summarizes clinical outcome success data across parity groups and ECS.In all instances, quarters treated with SP2 or SP5 responded with higher clinical outcome success than CON (P < 0.05).
Average DCC was significantly impacted by treatment (P < 0.0001) and ECS (P = 0.0127), but there was no treatment by ECS interaction (Table 4).Summarized across ECS (Table 7), DCC observed by SP2 quarters was not different from CON quarters, but both treatments were less than the average for SP5 quarters (P < 0.05).Milk from cows treated with ceftiofur HCl must be withheld for 72 h following the last treatment, while milk for untreated cows must be withheld as nonsaleable until clinical signs, including abnormal milk, have resolved.Given that clinical signs for both CON and SP2 quarters resolved after d 5 from study enrollment, there is essentially no difference regarding the required milk discard duration between these treatments.Although clinical signs among SP5 quarters were not resolved until just over 7 d from enrollment, milk withholding was necessary for 8 d to meet product label compliance.Categorical data were analyzed by PROC GLIMMIX and continuous data were analyzed by PROC MIXED to evaluate fixed effects of treatment, parity, treatment by parity interaction, ECS, ECS by parity interaction, mastitis DIM, and random effects of farm, breed, and season of year.

3
CS of 0 = normal milk and no local or systemic signs; 1 = abnormal milk and no local or systemic signs; 2 = abnormal milk, abnormal udder with signs of inflammation, and one clinical sign (rectal temperature ≥ 39.5°C, moderate to marked enophthalmos, or marked depression defined as inappetence or not able to stand, or both); 3 = milk grossly abnormal, abnormal behavior, and at least 2 systemic signs.

4
Proportion of quarters that experienced bacteriological cure at 14 ± 3 d.

7
Clinical outcome success declared if clinical score improved from 1 to 0 or 2 to 0 within 14 ± 3 d. 8 Defined as a cow with bacteriological and clinical cures without additional antimicrobial intervention or support therapy, and no culling or death as consequence of the Gramnegative clinical mastitis.
ab Within ECS = 1, means in the same row with different superscripts are different (P < 0.05).
xy Within ECS = 2, means in the same row with different superscripts are different (P < 0.05).

Mastitis Recurrence and New Intramammary Infections
Proportions of quarters with MR were calculated, and effects of both treatment and ECS were significant (P < 0.0224; Table 6), but there was no treatment by ECS interaction; as such, the main effects of treatment are presented in Table 7.The proportion of quarters with MR among CON was greater (P < 0.05) than either SP2 or SP5; SP2 and SP5 were not different from each other.Treatment differences were not observed for quarters that incurred a NIMI > 14 d after enrollment (P = 0.7346).
The proportion of quarters that required supplemental IMM antibiotic intervention within 28 ± 3 d from the onset of experimental treatments was significantly affected by treatment (P = 0.0068; Table 6) but not by parity or ECS.Examining mean comparisons of main effects of treatments across ECS levels (Table 7), the proportion of CON quarters that required supple-mental IMM antibiotic treatment showed a tendency to be higher (P = 0.0998) than SP2 quarters but was significantly higher (P = 0.0025) than SP5 quarters.The difference between SP2 and SP5 quarters approached significance (P = 0.0557).

Culling and Death
Proportions of cows that left herds due to death or culling were an important measure of treatment effectiveness in this study.No trial animals were euthanatized during the study.Forty-nine cows were culled or died due to GN CM before 14 ± 3 d (CON: ECS 1 = 4, ECS 2 = 24; SP2: ECS 1 = 0, ECS 2 = 10; SP5: ECS 1 = 0, ECS 2 = 11).Animals that were culled or died due to GN CM, expressed as a proportion of all culled or dead cows in each treatment, are shown in Table 6 (by treatment and ECS) and Table 7 (by treatment).For this metric, the overall effect of treatment was significant (P = 0.0009), as was ECS (P = 0.0326), but there was no treatment by ECS interaction.Main effects of treatment showed that the proportion of dead and cull cows that left the study for reasons related to mastitis among CON cows was higher (P < 0.0007) than for either SP2 or SP5.
Proportions of cows in each treatment that left the herds due to GN CM-related death or culling through 30, 60, and 90 d post-enrollment were also summarized as a percent of cows enrolled in each treatment group; in each case, the effect of treatment was significant (P < 0.0131), as was the effect of ECS (P < 0.0001).The proportion of CON cows that left the herd for reasons related to GN CM was higher than proportions of SP2 or SP5 cows that left by 30 d, 60 d and 90 d postenrollment (Table 7).The average d cows remained  in the study was evaluated after right-censoring d on study to a maximum fixed value of 90 d (any cow that remained in the study through 90 d was assigned a value of 90 d).Main effects of both treatment and ECS were significant (P < 0.0001), and the interaction of treatment and ECS was also significant (P < 0.05).For quarters with an ECS of 1, treatment means for CON (80.0 d), SP2 (85.5 d), and SP5 (80.7 d) were not different.However, among quarters with ECS of 2, the number of d cows remained in the study was 45.5 for CON, which was less (P < 0.0001) than either SP2 (73.0) or SP5 (75.1). Figure 3 is a graphic representation of the cumulative proportion of cows that did not depart the herd through culling or death due to mastitis for each treatment over the 90 d after enrollment for ECS 1 and 2.

Milk Composition and Milk Production
Results of milk samples harvested at 14, 21, and 28 ± 3 d following enrollment showed no effects of treatment, parity, ECS, or interactions on MC treatment means, except for percent protein on d 21 which is likely a spurious statistical observation and not a biological effect of treatment with ceftiofur HCl.The DHIA data collected to evaluate pre-and post-infection MC and milk production was inconsistent and offered a limited opportunity to reliably assess the impact of GN CM on milk yield or milk quality; therefore, no further analysis was possible.

Overall Treatment Outcomes
The definitive measure of treatment success reflects responses in terms of both bacteriological and clinical outcomes.To be considered an overall treatment success, in addition to BC and CC, cows could not receive supplemental IMM in the same quarter, systemic antibiotic treatments, and could not be the subject of either culling or death due to mastitis.Overall treatment outcome data are presented in Table 4. Main effects of treatment, parity, and ECS were all significant (P < 0.0215), and there were no interactions.As before, while treatment responses were significant for animals with ECS of both 1 and 2, the largest responses were noted among cows with ECS of 2. The overall treatment success rate for the CON group was inferior (P < 0.05) to that of the SP2 and SP5 groups.Odds ratios were computed to demonstrate the impact of data related to overall treatment success rates, and data are summarized in Table 4 for treatment by ECS combinations.Quarters treated with SP2 and SP5 were more likely to experience overall treatment success than CON quarters in mild (ECS 1) cases.The effect of treatment was even more pronounced for quarters that were enrolled with an ECS of 2. Table 7 summarizes overall treatment success across ECS scores.The overall treatment success rate for CON was inferior to both SP2 and SP5 (P < 0.05); and SP2 and SP5 were not different from each other.Computation of odds ratios revealed that SP2 and SP5 cases were more likely to experience an overall treatment success than CON cases.

DISCUSSION
Mastitis caused by GN bacteria has been a subject of discussion leading to multiple clinical trials to investigate the need to treat mild and moderate cases with antimicrobials.The present study was designed to determine if there was a benefit to treating mild to moderate GN CM with ceftiofur HCl and included treatment outcomes from 415 CM cases from 3 commercial dairies in California.Although pre-study power calculations were based on data from previously published studies, the sample size per treatment was greater than in prior negative-controlled randomized clinical trials intended to assess IMM treatment impact on nonsevere GN mastitis (Schukken et al., 2011;Fuenzalida and Ruegg, 2019).It was suggested that 40% of CM cases in dairy herds are caused by GN bacteria, and case severity is  equally distributed (Oliveira et al., 2013).However, the proportion of GN cases in herds enrolled in this study was lower, varying between 8.4% and 12.1%.Although OFC systems are reliable and widely used to identify mastitis pathogens, they are not 100% accurate and can result in false negatives (Ferreira et al., 2018).This study relied on OFC for pre-screening cases for study enrollment, and we acknowledge that there could have been some samples that were false negatives for GN infections.Escherichia coli was the dominant species cultured, followed by Klebsiella spp.This is not unexpected, as cows on all 3 dairies were housed in free stalls bedded with recycled manure solids, which is a known source of E. coli and Klebsiella spp.(Rowbotham and Ruegg, 2016).Previous studies evaluating the efficacy of ceftiofur HCl therapy on GN infections found a proportion of Klebsiella spp.infections (Schukken et al., 2011;Fuenzalida and Ruegg, 2019), and this could be the result of differences in the prevalence of mastitis pathogens in different geographies.
Previous studies that evaluated outcomes of GN mastitis therapy have unequal severity distributions on the enrolled cases, which could have contributed to no better or more favorable outcomes when comparing treated and untreated quarters.For instance, the distribution of mild and moderate cases was different in a study comparing outcomes of ceftiofur HCl treatment, which may have affected the outcomes (Fuenzalida and Ruegg, 2019).In contrast, our study had mild and moderate cases equally distributed among groups.This is important because our results showed differences in outcomes according to case severity at the time of enrollment.Although our study was not designed to evaluate efficacy of GN core vaccine and its association with mastitis severity, it is important to highlight that immunization is a tool to control mastitis in dairy cows and minimize disease severity and duration (Wilson et al., 2007).However, vaccination does not appear to be a factor affecting severity at enrollment in this study as Herds A and B had similar proportions of severe and nonsevere cases, even though Herd A does not vaccinate, and Herd B does.Herd C, which uses GN core vaccine, had a lower proportion of severe cases.Moreover, because of the possible confounding of the use of the vaccine and its effect on the outcomes evaluated, herd factor was accounted for in the statistical analysis, and results showed no difference in outcomes among herds.
Several researchers have proposed that BC should be the goal of antimicrobial therapy, but most mild and moderate GN cases do not require antimicrobial treatment as they can be quickly cleared by the cow's immune system (Leininger et al., 2003;Lago et al., 2011a;Fuenzalida and Ruegg, 2019).In fact, Leininger et al. (2003) showed that 85% of CM infected with E. coli experienced spontaneous BC within 7 d of CM onset.In our study, the proportion of quarters experiencing spontaneous cure by 14 d after mild or moderate CM onset was lower (70.1%)than BC on quarters treated for either 2 or 5 d with ceftiofur HCl (88.3 and 84.0%, respectively).However, when we separate results by severity, mild cases had a higher rate of spontaneous cure (77.5%) compared with moderate cases (68.3%).These results intuitively make sense insofar as mild cases of GN mastitis appear to largely cure spontaneously, but in moderate cases, the application of IMM antibiotic in the form of ceftiofur HCl significantly improved health outcomes.
Researchers discourage antimicrobial therapy for GN infections due to the lack of improvement in BC rates, especially E. coli infections (Pyörälä et al., 1994;Suojala et al., 2010).However, these studies did not consider case severity when making recommendations.Results of the present study showed that across herds, severity scores (ECS1-mild or ECS2 -moderate), and bacterial species, BC was higher for treatment groups compared with the CON group.These results agree with previous studies showing higher BC rates for quarters treated for 2, 5, and 8 d with ceftiofur HCl (Schukken et al., 2011;Fuenzalida and Ruegg, 2019).However, when evaluating mild and moderate cases separately, we found that the proportions of quarters with BC in mild cases in both treatment groups were similar to the CON group.For moderate cases, on the other hand, results showed Clinical outcome success declared if clinical score improved from 1 to 0, 2 to 1, or 2 to 0 within 14 d. 5 Defined as case that had a combination results: bacteriological and clinical cures without additional antimicrobial intervention or support therapy, no culling or death as consequence of the Gram-negative clinical mastitis.
only half of the CON quarters experienced BC, while over 70% and 80% of cows treated with either 2-d or 5-d courses of ceftiofur HCl, respectively, experienced BC.Although other studies (Todhunter et al., 1991;Gröhn et al., 2004Gröhn et al., , 2005;;Fuenzalida and Ruegg, 2019) suggested more negative outcomes in quarters infected with Klebsiella spp., it was not possible to evaluate outcomes by etiology in the present study because most of the cases were due to E. coli infection.Even though we were not able to statistically evaluate outcomes, our data showed that among the 26 quarters enrolled with Klebsiella spp.,3 experienced BC, and 12 experienced CC.Hence, no conclusions can be drawn specific to Klebsiella spp.infections, and more studies are needed.
Clinical outcomes are widely used by farmers and researchers as a visual indication of the success of IMM therapy (Swinkels, 2014) but should not be an objective determinant of treatment efficacy because it may take time for milk to resume normality and for inflammation to resolve (Ruegg, 2021).In our study, we found that the majority of mild cases across all 3 groups returned to normal CS (CS = 0) by d 2, while for moderate cases, it took an average of 5 d for treated quarters and 14 d for nontreated quarters.It is important to mention that CS is a subjective measure and could be interpreted differently by each person, even when following a defined CS system.
Previous studies also investigated DCC and reported no differences in DCC in treated and nontreated quarters (Fuenzalida and Ruegg, 2019;Schukken et al., 2011) or were higher in treated quarters (Morin et al., 1998).The average DCC for the present study was similar to values reported in previous studies (Fuenzalida and Ruegg, 2019;Schukken et al., 2011) but higher than the results reported by Lago et al. (2011a).No explanation for the higher average DCC for SP5 was identified, but we speculate that it could be attributed to a longer duration of treatment and, therefore, more prolonged stimulation of the local epithelia by syringes used to administer the IMM antibiotic.Some earlier in vitro studies showed that antimicrobials might disturb phagocytosis when given IMM (Nickerson et al., 1986), leading to the appearance of clinical signs.However, the clinical relevance of this finding is unknown.
The proportion of quarters with MR differed between the 3 treatment groups and was more frequent in the CON group.Our results do not agree with those of Fuenzalida and Ruegg (2019), who found no significant differences between treated and untreated groups.Because of the higher proportion of MR in the CON group, we attributed treatment differences to the efficacy of ceftiofur HCl.Identification of CM in the work by Fuenzalida and Ruegg (2019) and in the present study was done by farm personnel and documented in herd recordkeeping software.Therefore, differences in results between the present study and the abovementioned study could be due to failure of accurate CM identification by milkers and/or failure to accurately enter information in herd records.
Monitoring milk SCC is considered a valuable indicator of mastitis.We evaluated quarter SCC on d 14, 21, and 28 following enrollment but found no significant difference in quarter SCC between groups; these results agree with previous research (Schukken et al., 2011;Fuenzalida and Ruegg, 2019).In addition, there were no significant treatment differences in MC between groups and no interaction with ECS, parity, or other effects on outcomes.However, Williamson et al. (2022) studied associations between pretreatment SCC values and BC rates following antibiotic therapy for mastitis and showed that quarters with lower SCC before antibiotic treatment are more likely to successfully cure than quarters with high SCC before treatment.
Research evaluating pathogen-specific production losses following a CM case found that infections with E. coli diagnosed before peak lactation caused losses of 10.6% of the 305-d milk yield (3.5 kg/d;Heikkila et al., 2018).Whereas our data did not offer an opportunity to assess the impact of GN CM on milk yield before, during, or after a GN CM diagnosis, previous studies have shown a milk production decrease associated with a case of GN mastitis, but did not show persistent significant differences in milk production decreases or recovery times between treated and control groups (Schukken et al., 2011;Fuenzalida and Ruegg, 2019).
While CM is associated with an increased risk of culling (Pinzón-Sánchez et al., 2011), IMM antibiotic treatment positively affects cow survival and decreases mastitis-related milk losses (Roberson et al., 2004;Schukken et al., 2009).When comparing treated and untreated quarters, Schukken et al. (2011) found that cows treated for 5 d with ceftiofur HCl left the study through culling, death, or farmer interference less frequently (18%) compared with control animals (27%).In the present study, cows in the CON group represented the majority of cows that died or were culled by 30 d from enrollment.Although we evaluated up to 90 d post-enrollment, the proportion of cows culled in this study was higher than those noted in previous studies (Schukken et al., 2011;Fuenzalida and Ruegg, 2019).We hypothesize that dairies were more aggressive in eliminating problem cows that would not cure or recover productivity.In addition, others showed that CM caused by Klebsiella spp.leads to more significant milk losses (Gröhn et al., 2004) and a higher risk of culling (Gröhn et al., 2004;Fuenzalida and Ruegg, 2019) compared with E. coli infections.However, the limited number of Klebsiella spp.cases in our study did not al- Bruno et al.: Treatment of Gram-negative mastitis low comparisons regarding outcomes.Moreover, in this study early treatment of nonsevere GN CM infections with IMM antibiotics proved significantly beneficial compared with not treating cases and simply assuming that cows would self-cure without consequence.
Clinical mastitis therapy should aim to promote animal welfare.Denying or delaying therapy due to etiology could lead to worsening health, death, or animal discard.There are limited data evaluating GN CM bacteria on vitro susceptibility to ceftiofur HCl (Sheedy et al., 2021), but AMR monitoring should continue.Results of this study show the benefits of treating GN CM, with better outcomes for both treated groups compared with the nontreated control group.In addition, this paper highlights significant interactions between treatment outcomes and severity of CM at the time of enrollment, with effectiveness of the treatment being more pronounced as clinical severity of GN mastitis increased.Furthermore, our results suggest that moderate cases may benefit more from treatment than mild cases.Successful evaluation of treatment outcomes by severity scores (mild x moderate) was possible because research personnel were thoroughly involved to ensure protocol compliance.However, in a commercial dairy, it is unlikely that farm personnel will differentiate mild from moderate cases consistently and effectively.Therefore, developing recommendations for the practical treatment of GN mastitis in a commercial dairy based on the ability to accurately assess severity may not be feasible for every dairy, and in this case, a blanket procedure for treatment of all nonsevere cases is an appropriate recommendation.It seems important to emphasize that OFC was used in this study, but not all farms have this capability and would therefore have to rely on decision-making based on culture results from external laboratories or use blanket therapy.

CONCLUSION
Results of this study highlight that for many of the critical effectiveness outcomes, there was a benefit to treating nonsevere CM caused by GN bacteria, and that the magnitude of the therapeutic response was more prominent and statistically relevant in quarters diagnosed with moderate than with mild GN mastitis.The effectiveness of 2 versus 5 d treatments with ceftiofur HCl was similar and better than not treating.Treated cases had better rates of BC, CC, less MR, less mastitis-related death and culling, all leading to better overall treatment success compared with nontreated treated cases.Based on these results, there is a benefit to investing in the development of treatment protocols to treat nonsevere cases of GN mastitis, and the benefits outweigh the risks of not treating.
Figure 1.Bacteriological cure rates for cows that received no supplemental IMM or systemic antibiotic treatments.

Figure 2 .
Figure 2. Clinical scores between Day 0 to Day 28 Bruno et al.: Treatment of Gram-negative mastitis Bruno et al.: Treatment of Gram-negative mastitis Table 6.Outcomes as influenced by treatment and Enrollment Clinical Score (ECS) 1 (least squares means ± standard error) analyzed by PROC GLIMMIX and continuous data were analyzed by PROC MIXED to evaluate fixed effects of treatment, parity, treatment by parity interaction, ECS, ECS by parity interaction, mastitis DIM, and random effects of farm, breed, and season of the year. 2 CON = non-treated negative controls; SP2 = ceftiofur HCl once daily for 2 d; SP5 = ceftiofur HCl once daily for 5 d. 3 Qualified as clinically cured based on clinical score, but then developed clinical mastitis again before 14 d after the start of treatment in the same quarter.4 Defined as new infections identified at d 21 and d 28 after the quarter was considered BC at d 14. 5 Treatment by ECS interaction is significant (P < 0.05).ab Within ECS = 1, means in the same row with different superscripts are different (P < 0.05).xy Within ECS = 2, means in the same row with different superscripts are different (P < 0.05).

Figure 3 .
Figure 3. Survival curve of culling and death events related to mastitis through 90 d post-treatment

negative mastitis Table 1. Herd description upon entering the study
Bruno et al.: Treatment of Gram- Bruno et al.:Treatment of Gram-negative mastitis bution of CM cases (nonsevere vs. severe) occurring during the study period at each site.Not all cases of GN CM were enrolled in the study because they did not present as nonsevere.

Table 2 : Summary of enrollments to treatments at each study site 1
Bruno et al.: Treatment of Gram-negative mastitis 1 Control = non-treated negative controls; SP2 = ceftiofur HCl once daily for 2 d; SP5 = ceftiofur HCl once daily for 5 d.

Table 4 .
Bruno et al.: Treatment of Gram-negative mastitis Bruno et al.: Treatment of Gram-negative mastitis Results by treatment group and Enrollment Clinical Score (ECS) groups 1

Table 7 .
Bruno et al.:Treatment of Gram-negative mastitis Results by treatment group across parity and Enrollment Clinical Score groups 1 (least squares means ± standard error) Categorical data were analyzed by PROC GLIMMIX and continuous data were analyzed by PROC MIXED to evaluate fixed effects of treatment, parity, treatment by parity interaction, ECS, ECS by parity interaction, mastitis DIM, and random effects of farm, breed, and season of year.