Factors associated with bovine respiratory disease case fatality in feedlot cattle

Abstract

Bovine respiratory disease (BRD) is the primary cause of morbidity and mortality in cattle feedlots. There is a need to understand what animal health and production factors are associated with increased mortality risk due to BRD. The aim of the present study was to explore factors associated with BRD case fatality in feedlot cattle. Four pens totaling 898 steers were monitored daily for visual signs of BRD such as difficult breathing and coughing, and animals exhibiting signs of BRD were taken to the hospital shed for further examination and clinical measures. Blood samples were obtained at feedlot entry and at time of first BRD pull from animals diagnosed with BRD (n = 121) and those that died due to BRD confirmed by postmortem examination (n = 16; 13.2% case fatality rate). Mixed-effects linear regression models were used to estimate differences in animal health and production factors and the relative concentrations of 34 identified blood metabolites between animals that survived versus those that died. Generalized linear mixed-effects models were used to obtain the odds of being seronegative (at both feedlot entry and first BRD pull) to 5 BRD viruses and having a positive nasal swab result at the time of first pull in died and survived animals. Animals that died from BRD had lower average daily gain (ADG), reduced weight at first BRD pull, higher visual BRD scores and received more treatments for BRD compared with animals that survived BRD (P < 0.05). The odds of being seronegative for bovine viral diarrhea virus 1 (BVDV-1) were 5.66 times higher for animals that died compared with those that survived (P = 0.013). The odds of having a positive bovine coronavirus nasal swab result were 13.73 times higher in animals that died versus those that survived (P = 0.007). Animals that died from BRD had higher blood concentrations of α glucose chain, β-hydroxybutyrate, leucine, phenylalanine, and pyruvate compared with those that survived (P < 0.05). Animals that died from BRD had lower concentrations of acetate, citrate, and glycine compared with animals that survived (P < 0.05). The results of the current study suggest that ADG to first BRD pull, weight at first BRD pull, visual BRD score, the number of BRD treatments, seronegativity to BVDV-1, virus positive to BCoV nasal swab, and that certain blood metabolites are associated with BRD case fatality risk. The ability of these measures to predict the risk of death due to BRD needs further research.

Introduction

Bovine respiratory disease (BRD) is the primary cause of morbidity and mortality in feedlots, with 60% to 90% of all morbidity and mortality in feedlots attributed to BRD (Smith, 1998; Hay et al., 2014; Baptista et al., 2017). Approximately 0.64% to 0.74% of cattle on feed in North America die due to BRD (Miles, 2009; Baptista et al., 2017). Economic losses associated with mortality have been estimated at USD$1173.54 per mortality, or USD$11.73 for every animal inducted at a 1% mortality rate (Blakebrough-Hall et al., 2020c). Efforts to reduce mortality associated with BRD would improve feedlot profitability and animal welfare.

Previously, most studies have predicted the likelihood of an animal dying due to BRD based on general animal characteristics at feedlot entry such as average weight at feedlot arrival, sex, month of feedlot arrival, and breed (Babcock et al., 2013b; Amrine et al., 2014). These characteristics have been shown to be associated with an increased risk of mortality and culling due to BRD (Cusack et al., 2007; Babcock et al., 2013a). Other factors such as longer distances travelled prior to feedlot entry, mixing of animals, and larger pen sizes have also been shown to increase mortality risk (Martin et al., 1982; Cernicchiaro et al., 2012b). While there is considerable research surrounding the relationship between cohort-level factors, disease severity, and risk of mortality, there are few studies exploring the association between individual animal health and production factors collected at the time of first BRD pull such as rectal temperature, visual BRD score, presence of BRD viruses, and the risk of mortality due to BRD (Theurer et al., 2014). The identification of animal factors collected when animal is first identified with BRD that are associated with increased mortality risk may allow for improved management and more targeted treatment programs for these higher risk animals. Additionally, an area which has not been well explored is the association between an animal’s metabolite profile at first BRD pull and their subsequent risk of dying due to BRD. Currently, there are few published studies identifying possible metabolite biomarkers for BRD mortality (Aich et al., 2009; Adkins et al., 2020), and so this is an area where further research is warranted. Examining differences in the metabolite profiles of animals that died compared with those that survived may identify additional biomarkers for BRD mortality which could eventually be used for more accurate and early identification of animals at risk of dying.

Previous literature on the relationship between the presence of viral pathogens and mortality due to BRD is also sparse. Viral pathogens have been found to contribute to fatal BRD in only 19% of animals, with bacterial BRD pathogens thought to play a much larger role than viral pathogens in BRD mortality (Gagea et al., 2006). This is most likely because viral pathogens are commonly the predisposing factor that leads to severe bacterial infection (Griffin et al., 2010). A previous study that inoculated steers with bovine alphaherpesvirus 1 (BoHV-1) found that there were significant differences in metabolite serum profiles between animals that survived and died prior to, and following experimental BoHV-1 infection (Aich et al., 2009). However, more research is needed to determine which metabolites may be associated with BRD mortality outcome in naturally occurring BRD.

The objective of the present study was to identify factors associated with BRD case fatality risk that could potentially be used to make more timely decisions regarding at-risk animals and to develop more informed disease management strategies, which has the potential to reduce the number of BRD mortalities in feedlots.

Materials and Methods

The study was approved by the Animal Ethics Committee of Research Integrity and Ethics Administration, The University of Sydney (Approval # 1118). All methods were carried out in accordance with the relevant guidelines and regulations.

Animals and management

The study was conducted at a commercial cattle feedlot in southern NSW, Australia. Mixed-breed, 1- to 2-year-old castrated male cattle (n = 898) were sourced from either saleyards (n = 788) or direct consignment from backgrounding farms (n = 110). Processing of trial animals into the feedlot was staggered over a 4-wk period based on cattle availability. Animals had initial body weight (BW) recorded (mean ± SD in-weight; 432 ± 51.2 kg) and were processed according to previously described feedlot entry protocols (Blakebrough-Hall et al., 2020a). This included vaccination for respiratory disease caused by Mannheimia haemolytica (Bovilis MH, Coopers Animal Health, NSW, Australia) and a live intranasal vaccine for BoHV-1 (Rhinogard, Zoetis Animal Health, NJ). Blood samples were obtained from the tail vein of all animals at feedlot entry to assess serological status for five viruses associated with BRD (BoHV-1; bovine viral diarrhea virus, BVDV; bovine respiratory syncytial virus, BRSV; bovine parainfluenza virus 3, BPI3; and bovine adenovirus 3; BAdV3) using methodology described previously (Blakebrough-Hall et al., 2020b).

Following feedlot entry, animals were assigned to four production pens and fed for an average of 114 d. Pen allocation was based on pen availability and stocking density remained similar between pens. Pen 1 housed 300 animals, pen 2 housed 266 animals, pen 3 housed 91 animals, and pen 4 housed 241 animals. Steers were fed to allow for ad libitum feed consumption and were transitioned through three starter rations to a steam-flaked barley-based finisher diet over an 18-d period. Animals remained on this ration until harvest or death, unless in hospital pens for disease treatment, in which case animals received a high roughage (lucerne and barley hay) steam-flaked barley starter diet. Details of ration formulation for the finisher diet are described in a previous study (Blakebrough-Hall et al., 2020a).

Bovine respiratory disease monitoring and clinical data collection

Animals were monitored daily for visual signs of BRD by trained feedlot staff from the day after the first group of animals entered the feedlot. Animals were scored for visual signs of BRD using a modified version of the Wisconsin calf scoring chart (McGuirk, 2008; Blakebrough-Hall et al., 2020a). The scoring system included assessment of 7 visual signs of BRD: lethargy (or slow to move in response to a stimulus), head carriage, labored breathing, cough, nasal discharge, ocular discharge, and rumen fill (Table 1). A score between 0 and 3 was assigned to each visual sign. Animals exhibiting a score > 0 for any of the visual signs specific to BRD (nasal or ocular discharge, labored breathing or cough; n = 137) were removed from their pens for blood sampling and clinical data collection. Individual animal data recorded at first BRD pull included date, visual identification number, electronic identification number, pen number, body weight, rectal temperature, and computer-assisted lung auscultation (CALA) score. Animals were treated for BRD according to treatment protocol described elsewhere (Blakebrough-Hall et al., 2020c).

Table 1.

Visual scoring system for identification of BRD in the pens by trained feedlot personnel

Score	0	1	2	3
Lethargy	Normal, active, alert	Mild lethargy	Moderate lethargy	Profound lethargy, slow movement
Head carriage	Normal head carriage	Poll of head level with withers	Poll of head dropping under withers often	Sustained low head carriage
Labored breathing	Normal	Mild audible signs of labored breathing	Moderate audible signs of labored breathing	Audible signs of labored breathing
Cough	Non	Dry non-productive cough	Moderate wet productive cough	Severe wet productive cough
Nasal discharge	Normal serous discharge	Small amount of unilateral cloudy discharge	Bilateral, cloudy or excessive mucus discharge	Copious bilateral mucopurulent discharge
Ocular discharge	No ocular discharge	Small amount of ocular discharge	Moderate amount of ocular discharge	Heavy ocular discharge
Rumen fill	Normal	Slight depression in rumen fill, no anorexia	Moderate depression in rumen fill, no anorexia	Severe depression in rumen fill, signs of anorexia

Blood samples for metabolomics analysis were collected from the coccygeal vein of each animal using a 10-mL lithium heparin collection tube (BD Vacutainer Heparin, BD Vacutainer, Mississauga, ON). Samples were placed on ice until centrifugation (2,500× g for 20 min) within 1 h of collection. The plasma was transferred to 1.5 ml micro test tubes (Eppendorf safe-lock 0.5mL microtubes, Ependorf, Hamburg, Germany) and frozen at −20 °C for a maximum of 1 mo until all sampling was completed and then sent to the University of Sydney Metabolomics Laboratory for storage at −80 °C for 5 mo prior to analysis. Blood samples were obtained from the tail vein of all BRD animals for serology for antibodies to the 5 viruses associated with BRD at feedlot entry. Paired sera were identified from individual animals at the time of feedlot entry to test concurrently with blood samples obtained at first BRD pull (Blakebrough-Hall et al., 2020b). Nasal swabs were obtained from all BRD animals for real-time PCR assays for 5 BRD associated viruses: BoHV-1, BVDV-1, BRSV, BPI3, and BoCV. Nasal swab samples were stored dry at 4 °C in their collection vessel prior to analysis.

All mortalities associated with BRD were necropsied by trained feedlot personnel using a previously published methodology and reason of death was recorded (Sullivan and O’Brien, 2010).

Bovine respiratory disease mortality outcome

Animals that had visual signs of BRD and were treated for BRD but did not die due to BRD were considered to have survived (n = 121). Out of the 898 animals initially entered into the trial, 23 died, with 16 dying due to BRD and seven dying of causes other than BRD which were removed from the analysis. The final dataset for analysis therefore contained 137 animals.

Blood metabolomics analysis

Samples were prepared for metabolic profiling using methods from a previously published protocol (Dona et al., 2014). Samples were analyzed with a Bruker Avance III 600 MHz spectrometer equipped with a 5-mm TCI cryoprobe (Bruker, MA). An in-depth description of sample processing and analysis can be found in a previous study (Blakebrough-Hall et al., 2020a). The raw spectrums were imported into Matlab (Mathworks, Natick, MA), automatically phased, baseline corrected, and referenced to the α-C¹H glucose doublet occurring at 5.23 ppm (Dona et al., 2016). The water component was truncated from 4.30 to 5.10 ppm to reduce analytical variation and probabilistic quotient normalization of the spectrums was performed across all samples (Dieterle et al., 2006; Blakebrough-Hall et al., 2020a). The normalized spectrums were then processed using Standard Recoupling of Variables to calculate the start and end points of each component or component (Tzoulaki et al., 2014). The area under each component was calculated which represents the relative abundance or concentration of each component (Dona et al., 2016). Raw spectrums were then imported into Chenomx NMR Suite (Chenomx, Edmonton, Canada) to identify individual metabolites using reference libraries (Dona et al., 2016).

Statistical analysis

Mixed-effects linear regression models using the MIXED procedure in SAS (SAS Inst. Inc., Cary, NC) were used to estimate the effects of animal health and production factors on BRD mortality outcome. Animal health and production factors analyzed in the present study included feedlot entry weight (in-weight), days on feed (DOF) at first BRD pull, weight at first BRD pull, average daily gain (ADG) to first BRD pull, overall visual score (scores 0 to 21), rectal temperature, CALA score (scores 1 and 3), and the number of BRD treatments an animal received (0, 1, 2, 3, 4, or 5). Animal within pen was considered the experimental unit. Mortality outcome and breed were considered as fixed effects and pen as a random effect. In-weight was used as a covariate for weight and ADG to first pull.

Generalized linear mixed-effects models in the GLIMMIX procedure of SAS (SAS Inst. Inc.) were used to analyze the difference in the proportion of blood seronegative or swab positive animals that died and survived. The odds ratios and associated confidence intervals for animals to die compared to those that survived were determined from these models as well. Breed was included as a fixed effect. Where breed was found to be non-significant (P > 0.05) it was removed from the model. Pen number was included as a random effect. Main effects were considered significant at P ≤ 0.05 and tendencies discussed at 0.05 < P ≤ 0.10.

Results

Descriptive statistics for animal health and production factors are shown in Table 2. On average, animals were first identified with BRD at 21 d on feed and mortalities occurred at 27 d (range 3 to 77 d) after the first BRD pull (Table 2). Weight at first BRD pull showed a broad range of 300 to– 626 kg, and on average these animals lost 0.1 kg/d between feedlot entry and first BRD pull. Average visual BRD score was 9, with 18 being the most severe visual score observed.

Table 2.

Descriptive statistics for animal health and production factors of feedlot steers affected by bovine respiratory disease

Variable	N	Minimum	Mean	Maximum	SD
In-weight, kg/animal	137	294.0	422.5	554.0	52.02
DOF at first BRD pull	137	2	21	42	9.2
Days first BRD pull to death	16	3	27	77	23.4
ADG to first BRD pull, kg/animal/day	137	−10.0	−0.1	3.5	2.13
Weight at first BRD pull, kg/animal	137	300.0	434.5	626.0	58.51
Overall visual BRD score, 0 to 21	137	1	9	18	3.2
Rectal temperature, oC	137	38.0	40.0	41.8	0.74
CALA score, 1 to 3	137	1	2	3	0.45
Number of BRD treatments, 0 to 5	137	1	2	5	1.55

The number and proportion of animals that were seronegative (naïve) at feedlot entry and first BRD pull, and positive nasal swab for BRD viruses when first identified with BRD are shown in Table 3. Most animals (97.7%) were seronegative for BoHV-1 at feedlot entry and this proportion decreased to 64.0% of animals at first BRD pull following vaccination for BoHV-1. Most animals were seropositive for BRSV, BPI3, and BAdV3 at feedlot entry, and the proportion of seronegative animals decreased by more than half at first BRD pull. Most animals had negative nasal swab results for all BRD viruses at first BRD pull, with the most common virus present in swabs being BoHV-1 (Table 3).

Table 3.

Proportion of animals with positive and negative serology results for viruses frequently associated with BRD¹

BRD virus	Serostatus at feedlot entry		Serostatus at first BRD pull		Combined serostatus2		PCR swab result at first BRD pull
	Negative number (%)	Positive number (%)	Negative number (%)	Positive number (%)	Negative number (%)	Positive number (%)	Negative number (%)	Positive number (%)
BoHV-1	126 (97.7)	3 (2.3)	87 (64.0)	49 (36.0)	84 (65.1)	45 (34.9)	112 (81.8)	25 (18.3)
BVDV-1	78 (60.5)	51 (39.5)	56 (41.2)	80 (58.8)	53 (41.1)	76 (58.9)	131 (95.6)	6 (4.4)
BRSV	25 (19.4)	104 (80.6)	13 (9.6)	123 (90.4)	5 (3.9)	124 (96.1)	131 (95.6)	6 (4.4)
BPI3	37 (28.7)	92 (71.3)	18 (13.2)	118 (86.8)	16 (12.4)	113 (87.6)	136 (99.3)	1 (0.7)
BAdV	48 (37.2)	81 (62.8)	21 (15.4)	115 (84.6)	25 (18.3)	112 (81.8)	—	—
BCoV	—	—	—	—	—	—	132 (96.4)	5 (3.7)

¹BRD-associated viruses: bovine alphaherpesvirus-1 (BoHV-1), bovine viral diarrhea 1 (BVDV-1), bovine respiratory syncytial virus (BRSV), bovine paraInfluenza-3 (BPI3), and bovine adenovirus (BAdV) and positive or negative nasal swab results for BoHV-1, BVDV-1, BRSV, BPI3, and bovine coronavirus (BCoV).

²Combined serostatus: negative or positive at both feedlot entry and first BRD pull.

Animal health and production factors associated with mortality due to BRD

Mean values for animal health and production factors in animals that survived or died from BRD are displayed in Table 4. Entry weight, rectal temperature, and CALA score did not differ between animals that died and those that survived (P > 0.05). Animals were first identified with BRD at between 2 and 42 DOF. Animals that died were first detected with BRD an average of 7 d earlier than those that survived BRD (P = 0.011), with all deaths receiving at least one treatment for BRD before death. Animals that died gained 0.92 kg/d less (P = 0.044) and were 22.8 kg lighter (P < 0.001) than those that survived. Additionally, animals that died also had more severe visual BRD scores (P = 0.029) and received more treatments for BRD (P < 0.001) compared with those that survived.

Table 4.

Evaluation of BRD mortality outcome using least squares means (±SE) of production performance and clinical measures of animals with nonfatal (survived) and fatal (died) disease

	Survived (n = 121)	Died (n = 16)	P-value
In-weight, kg/animal	421.5 ± 4.74	430.1 ± 13.04	0.535
DOF at first BRD pull	22 ± 0.8	15 ± 2.3	0.011
ADG to first BRD pull, kg/animal/d	0.3 ± 0.15	−0.6 ± 0.44	0.044
Weight at first BRD pull, kg/animal1	436.5 ± 2.93	413.7 ± 8.04	<0.001
Overall visual score, score 0 to 21	9.0 ± 0.29	10.9 ± 0.79	0.029
Rectal temperature, oC	40.0 ± 0.07	40.0 ± 0.19	0.915
CALA score, score 1 to 3	2.0 ± 0.04	2.1 ± 0.11	0.302
Number of BRD treatments	2.0 ± 0.14	3.4 ± 0.37	<0.001

¹In-weight was used as a covariate.

The proportion of animals with a seronegative blood or virus positive nasal swab result for animals that survived and those that died are presented in Table 5. Being seronegative at feedlot entry or first BRD pull for any of the 5 BRD viruses did not affect the likelihood of an animal dying due to BRD (P > 0.05). However, animals that died tended (P = 0.08) to be more likely to be seronegative for BVDV-1 either at feedlot entry or pulling. As a result, a larger proportion (76.9%) of animals that died were seronegative (did not seroconvert), for BVDV-1 at both feedlot entry and first BRD pull, with the likelihood of animals being seronegative for BVDV-1 5.66 times greater in animals that died compared with those that survived (P < 0.05). Additionally, animals that were naïve to BPI3 at both feedlot entry and first BRD pull were more likely to die (P = 0.048). Nasal shedding of viruses indicated that animals that were shedding BCoV at first BRD pull were 13.73 times more likely to die compared to those that survived (P = 0.007). However, no differences were found between animals that died and survived for the remaining BRD viruses detected in nasal samples (P > 0.05).

Table 5.

Proportion of blood samples seronegative or nasal swabs positive results to BRD viruses in animals that survived (n = 121) or died from BRD (n = 16) and associated odds ratios^1,2

BRD viral status	Survived	Died	Odds ratio	95% CI	P-value
Seronegative at feedlot entry
BHV1	97.4 ± 0.01	100.0 ± 0.00	>999.99	(<0.001 to )	0.977
BVDV	57.8 ± 0.05	84.6 ± 0.10	4.02	(0.840 to 19.256)	0.081
BRSV	19.8 ± 0.04	15.4 ± 0.10	0.74	(0.150 to 3.603)	0.702
BPI3	27.6 ± 0.04	38.5 ± 0.14	1.64	(0.494 to 5.451)	0.416
BAdV	35.3 ± 0.04	53.9 ± 0.138	2.13	(0.665 to 6.849)	0.201
Seronegative at first BRD pull
BHV1	62.5 ± 0.04	75.0 ± 0.11	1.80	(0.541 to 5.984)	0.335
BVDV	38.3 ± 0.04	62.5 ± 0.12	2.68	(0.904 to 7.949)	0.075
BRSV	8.3 ± 0.03	18.8 ± 0.10	2.54	(0.610 to 10.558)	0.198
BPI3	11.7 ± 0.03	25.0 ± 0.11	2.52	(0.707 to 9.013)	0.153
BAdV	14.2 ± 0.03	25.0 ± 0.11	2.02	(0.576 to 7.076)	0.270
Seronegative feedlot entry and first BRD pull
BHV1	63.8 ± 0.05	76.9 ± 0.12	1.89	(0.487 to 7.353)	0.355
BVDV	37.1 ± 0.05	76.9 ± 0.12	5.66	(1.457 to 21.983)	0.013
BRSV	3.5 ± 0.02	7.7 ± 0.07	2.33	(0.236 to 23.101)	0.466
BPI3	10.3 ± 0.03	30.8 ± 0.13	3.85	(1.015 to 14.614)	0.048
BAdV	10.3 ± 0.03	15.4 ± 0.10	1.58	(0.307 to 8.094)	0.583
Positive nasal swab
BHV1	20.7 ± 0.04	4.2 ± 0.02	2.61	(<0.001 to >999)	0.969
BVDV	3.3 ± 0.02	12.5 ± 0.08	4.18	(0.690 to 25.323)	0.119
BRSV	5.0 ± 0.02	0.003 ± 0.01	3.21	(<0.001 to >999)	0.973
BPI3	0.8 ± 0.00	0.0004 ± 0.13	2.49	(<0.001 to >999)	0.977
BCoV	1.7 ± 0.01	18.8 ± 0.10	13.73	(2.063 to 91.395)	0.007

¹BRD-associated viruses: bovine alphaherpesvirus-1 (BoHV-1), bovine viral diarrhea 1 (BVDV-1), bovine respiratory syncytial virus (BRSV), bovine paraInfluenza-3 (BPI3), and bovine adenovirus (BAdV) and positive or negative nasal swab results for BoHV-1, BVDV-1, BRSV, BPI3, and bovine coronavirus (BCoV).

²Values in bold were significantly different.

Relative concentrations of metabolites

Animals that died from BRD had higher concentrations of α glucose chain, hydroxybutyrate, leucine, phenylalanine, and pyruvate compared with animals that survived (P < 0.05; Table 6). In contrast, animals that died from BRD had lower concentrations of acetate, citrate, and glycine compared with animals that survived (P < 0.05). The relative concentration of the rest of the metabolites identified did not differ between animals that died and those that survived (P > 0.05).

Table 6.

Comparison of the least squares means (±SE) of the relative concentration (arbitrary units) of blood metabolites for animals that survived and died from BRD¹

Metabolite	Survived (n = 121)	Died (n = 16)	P-value
1-Methyl-histidine	1.11 ± 0.017	1.17 ± 0.046	0.195
3-Hydroxyisobutyrate	12.49 ± 0.243	13.73 ± 0.665	0.081
α Glucose chain	1.22 ± 0.030	1.47 ± 0.083	0.004
Acetate	7.58 ± 0.258	5.73 ± 0.700	0.015
Acetone	2.96 ± 0.063	3.12 ± 0.183	0.416
Alanine	6.71 ± 0.075	7.01 ± 0.213	0.184
Choline	5.88 ± 0.090	5.81 ± 0.249	0.773
Citrate	1.65 ± 0.021	1.51 ± 0.058	0.029
Creatine	11.47 ± 0.244	12.74 ± 0.670	0.077
Creatinine	11.47 ± 0.230	11.79 ± 0.647	0.634
Dimethyl sulfone	2.55 ± 0.033	2.44 ± 0.091	0.242
Ethanol	28.60 ± 0.244	28.08 ± 0.665	0.465
Formate	0.45 ± 0.012	0.43 ± 0.032	0.467
Glucose	17.65 ± 0.299	18.20 ± 0.623	0.409
Glutamate	3.61 ± 0.046	3.58 ± 0.125	0.811
Glutamine	2.18 ± 0.035	2.15 ± 0.097	0.796
Glycine	12.98 ± 0.198	11.60 ± 0.541	0.018
Glycoprotein acetyls	13.05 ± 0.222	13.12 ± 0.605	0.912
Hydroxybutyrate	6.98 ± 0.068	7.58 ± 0.193	0.004
Hydroxyisobutyrate	2.09 ± 0.027	2.04 ± 0.074	0.505
Isobutyrate	2.69 ± 0.025	2.67 ± 0.069	0.774
Isoleucine	14.51 ± 0.196	15.42 ± 0.537	0.114
Isopropanol	13.90 ± 0.137	13.56 ± 0.376	0.406
Lactate	17.08 ± 0.402	16.89 ± 1.141	0.888
Leucine	7.22 ± 0.092	8.04 ± 0.253	0.003
Low Density Lipid	17.07 ± 0.345	15.40 ± 0.940	0.099
Mannose	1.00 ± 0.022	1.09 ± 0.061	0.185
Methanol	10.10 ± 0.099	9.57 ± 0.275	0.068
Phenylalanine	0.47 ± 0.010	0.54 ± 0.028	0.015
Pyruvate	6.31 ± 0.129	7.28 ± 0.349	0.010
Succinate	1.80 ± 0.031	1.75 ± 0.091	0.579
Tyrosine	0.47 ± 0.010	0.47 ± 0.027	0.912
Unsaturated lipid	3.47 ± 0.091	3.33 ± 0.251	0.616
Valine	5.39 ± 0.087	5.58 ± 0.237	0.458

¹Metabolites in bold were significantly different in concentration between animals that survived and those that died.

Discussion

The present study aimed to explore factors associated with BRD case fatality in feedlot cattle. Only animals that were diagnosed with BRD were included in the current study, with healthy animals excluded from analysis. This approach is different from previous studies which have focused on the relationship between cohort-level feedlot arrival data and BRD mortality outcome (Cusack, et al., 2007; Babcock et al., 2013a, b). The animal health and production factors evaluated in the present study comprised of data collected at the time of first BRD pull including ADG to first BRD pull, overall visual BRD score, rectal temperature, and viral BRD status, as well as blood metabolite profiles. To the authors’ knowledge, this study is the first to evaluate the relationship between such a wide range of animal performance, health, and physiological measures and mortality due to BRD. It was hypothesized that these factors could characterize and identify animals at risk of dying after the first BRD treatment. This information could then be used to reduce the negative economic effects of BRD mortality in feedlots by identifying severe animals earlier and developing targeted treatment protocols for these animals (Blakebrough-Hall et al., 2020c).

There was a large difference in ADG and weight at first pull between animals that died compared with those that survived. Decreased ADG is often seen in animals affected by BRD (Waggoner et al., 2007; Blakebrough-Hall et al., 2020c), although there seems to be limited research on the relationship between ADG and mortality due to BRD. Animals that died were identified with visual signs of BRD 7 d earlier than those that survived. This could have contributed to the large difference in weight at first BRD pull between groups as they had less time to gain weight. The finding that animals that subsequently died were identified with BRD earlier for BRD is consistent with results from previous studies, where cattle treated in the first 20 d following feedlot arrival had higher post treatment failure and an increased risk of mortality compared with those that were treated more than 20 d after arrival (Babcock et al., 2010; Avra et al., 2017). This may indicate that there are adaptational effects following feedlot arrival that are related to more severe BRD outcomes (Babcock et al., 2010). The magnitude and timing of BRD development can be influenced by several factors such as animal age, level of stress, previous pathogen exposure (Hay et al. 2016), animal source, and transport times (Hay et al. 2014). It is possible that animals which required treatment earlier suffered a more severe infection because they were immunosuppressed after the stress of transport and feedlot arrival, and did not have enough time to recover adequate immune function prior to exposure to BRD-associated pathogens. However, further research which evaluates immune functionality over time are required to confirm this. The apparent influence of BRD severity and risk of mortality on ADGs that was found in the current study could offer an opportunity to focus on and, strengthen health and monitoring protocols for animals identified with BRD earlier in the feeding phase. Identifying at-risk individuals and treating these animals earlier may provide opportunities for feedlots to reduce rates of BRD mortality. It is also apparent from the results of the present study as well as previously published companion studies that ADG is a useful indicator of BRD severity and outcome (Blakebrough-Hall et al., 2020b).

Animals that died displayed more severe overt clinical signs of the disease compared with those that survived, reflected in higher visual BRD scores. Additionally, animals that died also received a greater number of BRD treatments compared with those that survived. These findings support the importance of treatment protocols that are based on the severity of visual signs. Heightened surveillance of animals after an initial BRD treatment may also prevent further mortality losses, and therefore, relocation of sick animals into hospital pens is desirable for closer monitoring, although it needs to be considered that hospital pens concentrate clinically ill animals and therefore may increase the risk of severe BRD and mortality. Previous literature on the relationship between animal health data at the time of BRD pull and risk of BRD mortality (compared with morbidity) appears to be lacking, and therefore more research is needed to confirm these findings. Nevertheless, the findings from the present study suggest that the severity of visual signs and ADG to first BRD pull are useful indicators of BRD mortality risk that can be used to improve the management of clinically severe animals for improved health outcomes and reduced mortality rates.

A larger proportion of animals that died due to BRD were seronegative (naïve) for BVDV-1 at both feedlot entry and first BRD pull compared with animals that survived, and therefore at greater risk of BVDV-1 infection if exposed to the virus. It has been demonstrated that BVDV-1 is an important pathogen in animals that die with gross postmortem lesions of pneumonia, with 37% of animals that died due to BRD observed to have BVDV-1 present in their lung tissue at death (Haines et al., 2004). However, it is important to note that the detection of BVDV in nasal swab samples can also be strain dependent and therefore identification of BVDV through nasal swab sampling alone may not always indicate infection with the virus (Ambrose et al. 2018). Animals that died were also nearly four times more likely to be naïve for BPI3 compared with those that survived; however, there was no increase in the likelihood of being naïve for the remaining 3 BRD viruses in animals that died compared with those that survived.

The odds of having a positive BCoV swab result were 13.73 times greater in animals that died compared with those that survived. These results seem to indicate that BCoV is a significant pathogen associated with mortality risk in feedlot cattle and has previously been shown to be one of the main causes of serious BRD outbreaks in an Australian study (Hick et al., 2012). However, little is known at present about the role of BCoV in naturally occurring BRD outbreaks or in BRD mortality risk, and therefore further research is needed to explore the role of BCoV in BRD morbidity and mortality. Most of the previous research has focused on the presence of BRD pathogens and their relationship with clinical BRD outbreaks rather than mortality specifically (Francoz et al., 2015; Hay et al., 2016). In one study, neither the presence of BCoV or BRSV was found to be associated with higher odds of clinical signs of BRD, with only the bacterial pathogen Mycoplasma bovis showing a relationship to clinical signs of BRD (Francoz et al., 2015). In agreement with these findings, bacterial BRD pathogens seem to play a larger role in BRD mortality than viral pathogens (Gagea et al., 2006). Unfortunately, a limitation of the present study was that data on bacterial BRD pathogens was not collected due to resource constraints. Additionally, the small number of BRD mortalities in the present study limits the ability to draw statistically sound and conclusive inferences about the relationship between viral pathogens and mortality. Nevertheless, results from the present study suggest that laboratory tests for pathogens such as PCR panels that detect multiple viral BRD pathogens may help in the prediction of BRD severity and mortality outcome. If incorporated into feedlot BRD management systems, these tests may contribute to significantly decreasing the number of severe BRD cases and therefore mortalities. To enable this however, costs would need to be reduced and accessibility to the industry on a large-scale facilitated.

In the present study, there was a difference in the concentration of 8 metabolites (α glucose chain, hydroxybutyrate, leucine, phenylalanine, pyruvate, acetate, citrate, and glycine) between animals that died compared with those that survived BRD. This is the first study to report significant associations between metabolite levels and BRD mortality in naturally occurring BRD in a commercial feedlot. This is an important finding as given the number of factors that may have contributed to the cattle developing fatal BRD, it is far less likely that associations may be detected. A previous study that identified biomarkers to predict risk of mortality in experimental induced BRD reported multiple metabolites associated with BRD mortality, with valine higher in animals that died compared with animals that survived (Aich et al., 2009). However, concentrations of valine did not differ in the current study. Phenylalanine on the other hand is one metabolite that appears to be consistently higher in BRD-affected animals (Basoglu et al., 2016; Blakebrough-Hall et al., 2020a), as well as in human patients with pneumonia and sepsis (Mickiewicz et al., 2013; Antcliffe et al., 2018). Results from the present study expanded on these findings, indicating the potential application of phenylalanine to detect animals with a higher risk of mortality due to BRD. A possible explanation for increased concentrations of phenylalanine in pneumonia patients and BRD-affected animals is a reduction in the conversion of phenylalanine to tyrosine due to oxidative stress as a result of the immune response to disease (Ploder et al., 2008). In an earlier study by the same authors, the metabolites α glucose chain, hydroxybutyrate, leucine, and pyruvate were also higher in BRD-affected animals compared with healthy animals (Blakebrough-Hall et al., 2020a). These metabolites are associated with energy metabolism and may reflect the negative energy status of animals with more severe BRD and the consequent need to mobilize body reserves (Baticz et al., 2002; Kanikarla-Marie and Jain, 2016). These findings suggest that there are metabolites associated with the severity of BRD and mortality due to BRD; however, further validation is needed to determine if those metabolites can accurately differentiate animals at risk of dying versus those that will survive compared with just those that may be at risk of developing clinic BRD.

Animals that died from BRD had lower concentrations of acetate, citrate, and glycine compared to those that survived. These 3 metabolites were also found to be lower in BRD-affected animals compared with healthy animals in the aforementioned companion study (Blakebrough-Hall et al., 2020a), with citrate selected by the machine learning models as a biomarker for BRD. The tricarboxylic acid (TCA) cycle produces energy under such conditions using acetate converted to acetyl-CoA, which then enters the TCA to combine with citrate (Williams and O’Neill, 2018). The TCA consumes both acetate and citrate to produce chemical energy and this may explain the reduction in the relative concentration reported in the present study. It is likely negative energy balance is greater in severe animals that end up dying than the less severe animals, which would also explain the difference in weight gains between the 2 groups. In the only other study that compared differences in metabolite concentrations between animals that died and survived, lactate and glucose were lower in animals that died (Aich et al., 2009), which was not confirmed in the present study. The lack of consistency between studies suggest that more research is needed to find stable patterns in metabolite concentrations associated with BRD severity across independent datasets. Nevertheless, the metabolites that were higher in animals that died compared with those that survived are involved in energy metabolism and the increase in blood circulation of animals under negative energy balance or during periods of high demand for energy (Baticz et al., 2002; Kanikarla-Marie and Jain, 2016). The application of metabolomics for clinical diagnosis in humans is progressing (Ford et al., 2020; Morduant et al., 2020), thus the availability for veterinary applications is plausible although needs further research and validation.

In conclusion, the current study found that there were large differences in animal health and production measures between animals that died compared with those that survived BRD. These measures have the potential to be used as indicators to predict the risk of mortality after the first BRD diagnosis. Animals with lower ADG and live weights, and higher visual scores at first BRD pull, as well as those that received a greater number of treatments for BRD were at a greater risk of dying. Serological naivety for the viruses BVDV and BPI3 and nasal shedding of BCoV at the time of first BRD pull were also associated with a higher mortality risk. Additionally, the concentrations of the blood metabolites α glucose chain, hydroxybutyrate, leucine, phenylalanine and pyruvate, acetate, citrate and glycine differed between the 2 groups. These metabolites may be useful biomarkers for BRD mortality risk. Identifying potential animal risk factors that are associated with mortality at the time an animal is first identified with BRD could aid in earlier and more effective management decisions such as separating severe animals into isolated pens at first treatment for BRD and developing treatment protocols targeted at severe BRD infection. Identifying these animals earlier and implementing practices such as these has the potential to decrease the rates of BRD mortality in feedlots. Novel approaches such as quantifying circulating metabolite concentrations have also shown great potential to be used as biomarkers for mortality risk and could be used to improve the management of the disease pending further research.

Glossary

Abbreviations

ADG

average daily gain

BAdV3

bovine adenovirus 3

BoHV-1

bovine alphaherpesvirus 1

BoCV

bovine coronavirus

BPI3

bovine parainfluenza virus 3

BRD

bovine respiratory disease

BRSV

bovine respiratory syncytial virus

BVDV-1

bovine viral diarrhea virus 1

BW

body weight

CALA

computer-assisted lung auscultation

DOF

days on feed

¹H NMR

proton nuclear magnetic resonance

IBR

infectious bovine rhinotracheitis

NMR

nuclear magnetic resonance

PCR

polymerase chain reaction

RT-qPCR

quantitative reverse transcription polymerase chain reaction

Conflict of Interest Statement

The authors declare no real or perceived conflicts of interest.