Biotinidase as a novel biomarker for pain assessment in dairy cattle

Supplemental Digital Content is Available in the Text.

Biotinidase, an amidohydrolase that catalyzes the cleavage of biotin from biocytin or biotinylated peptides, exhibits a robust response to diverse pain-inducing stimuli in nonhuman mammals.

Abstract

Introduction:

Freedom from pain is a key component of animal welfare. However, unlike in humans, objective methods for assessing and managing pain in farm animals remain limited. Pain assessment is particularly challenging in cattle, due to their inherently stoic nature.

Objective:

We investigated biotinidase, an enzyme with amino-exopeptidase activity on opioid neuropeptides, as a novel, short-term pain-biomarker in dairy cattle, in comparison with cortisol and substance P.

Methods:

Plasma levels of biotinidase, substance P, and cortisol were quantified by ELISA in Holstein cows subjected to 3 distinct pain stimuli: freeze branding (routine husbandry), topical application of capsaicin cream (nonclinical pain model), and spontaneous clinical mastitis.

Results:

Biotinidase levels increased significantly within 30 minutes after freeze branding compared to controls (1.36 difference of normalized values; P = 0.021), and after capsaicin application (from 18.5 ± 2.5 to 21.5 ± 2.7 ng/mL; P = 0.002). Neither cortisol nor substance P exhibited significant changes after freeze branding or capsaicin exposure. Biotinidase levels elevated significantly in mastitic cows (28.1 ± 8.2 ng/mL) compared to healthy controls (13.63 ± 5.62 ng/mL; U = 74.0, P = 0.038), whereas substance P and cortisol showed no significant differences. Receiver operating characteristic analysis suggests that biotinidase (AUC: 0.84, 0.52, 0.81) outperformed substance P and cortisol in distinguishing painful from painless states.

Conclusion:

These findings provide compelling new evidence that plasma biotinidase responds robustly to diverse pain stimuli in nonhuman mammals. Further studies should investigate its diagnostic performance individually, and alongside other biochemical, physiological, and behavioral indicators, to enhance pain assessment accuracy and of noninvasive pain detection protocols for farm animals.

1. Introduction

Pain is a subjective experience involving unpleasant sensory and emotional response to actual or potential tissue damage^7,14 and has important survival value in the animal kingdom.⁵² Accurate pain recognition is a fundamental prerequisite for its alleviation,¹⁸ but in farm animals, as dairy cattle—prey animals with a stoic nature to avoid attracting predators, this recognition is particularly challenging.^21,50 Yet, despite widespread agreement on the importance of addressing animal pain, no universally accepted gold standard for its assessment has been established, although it is fundamental to optimizing farm animal welfare.⁵³

Given the physiological mechanisms of nociceptive transduction, it is plausible that pain could be reflected in specific biochemical markers within body fluids. This is supported by evidence that pain can be modulated not only centrally but also through interactions between leukocyte-derived opioid peptides and opioid receptors on the peripheral terminals of primary afferent neurons.⁴⁸ Opioid neuropeptides, including encephalins, β-endorphins, and dynorphins, play major roles in pain modulation.^26,47 Nonopioid neuropeptides and neurotransmitters such as oxytocin, neurotensin, orexin A, and substance P (SP) have also been investigated as circulating biomarkers for pain.^12,33 Specifically, SP functions as a mediator of pain, stress, and anxiety. Studies in cattle, using circulating SP as an indicator for inflammation and nociception, have yielded inconsistent results.^{25,33,47,48,50} Generally, although neuropeptides rapid release after nociceptive stimuli may support their role as serum pain biomarkers, their clinical utility is limited by a short half-life.³² In contrast, cortisol, a steroid hormone, demonstrates a slower but more sustained release profile after stress or tissue injury.^8,35,43,46 However, cortisol levels vary across different interventions,³³ exhibit circadian variation,³⁸ and are highly responsive to general stressors,²⁵ factors that constrain its reliability as a specific pain biomarker.^25,33,38 The variability observed in both cortisol and SP levels has prompted studies to combine multiple factors to capture a broader range of physiological pain responses.³³ Incorporating additional candidates into serum biomarkers panel may significantly enhance objective pain detection.

Biotinidase (BTD), best known for its amidase activity on biocytin (biotinyl-lysine) in humans,⁴¹ also functions as an amino-exopeptidase, cleaving opioid neuropeptides in serum.⁴⁰ Recent clinical evidence links elevated BTD activity with pain severity in patients with arthritis.⁵¹ Insights from infectious disease models, notably bacterial meningitis, further suggest a functional link between BTD and pain modulation, through rapid increase in the enzyme's activity during acute inflammation, coinciding with elevated body temperature and CNS-enkephalin concentrations.¹⁵ This suggests that CNS-BTD may hydrolyze opioid neuropeptides during inflammatory episodes, modulating nociception.^15,27 In the context of animal physiology, BTD is secreted primarily by the liver,^23,42 and the plasma demonstrates the highest specific activity among biological samples examined.⁵⁴ Supported by these findings, we aimed to evaluate BTD as a novel plasma biomarker for pain in cattle.

To advance objective pain recognition, we examined the dynamics of plasma BTD, SP, and cortisol in dairy cattle, in response to 3 distinct pain-inducing stimuli: freeze branding (FB), a routine husbandry practice, topical application of capsaicin cream—representing a nonclinical pain model, and spontaneous manifestation of mastitis.

2. Material and methods

2.1. Experiment 1

2.1.1. Animals, housing, management, and preparation

The experiment was conducted at the Volcani Center Experimental Dairy Farm in Beit Dagan, Israel. Protocol was approved by the Volcani Center Animal Care and Use Committee (IL 884/20). Freeze branding, a procedure routinely employed on Israeli dairy farms to mark young cattle, was applied to 3-month-old healthy Holstein heifers (N = 18). Heifers were grouped-housed in a shaded loose-housing pen, with ad-libitum food and water. A single clinician handled all animals. To allow repeated blood sampling with minimal disturbance, an indwelling angio-catheter (16 GA, 3.25 inches) was inserted into the jugular vein, flushed with heparinized saline, connected to a 20-cm extension tube, and secured to the skin with 2 stitches.

2.1.2. Experimental procedure

The heifers were randomly assigned to 3 groups: FB without analgesics, FB with analgesics, and a control group of naïve heifers not subjected to FB. The analgesic group received 20 mg of meloxicam per 40 kg body weight (Loxicom, Norbrook Laboratories, United Kingdom) subcutaneously with a G21 needle, 3 hours before the procedure. The FB procedure was performed by an expert. Heifers were tied during branding and released thereafter. Blood samples (4 mL per time point, 32 mL total) were drawn into lithium heparin vacutainer tubes at −120, 0, 10, 20, 30, 60, 180, and 360 minutes relative to the branding (Fig. 1). Samples were kept on ice (≤30 minutes) and then centrifugated at 3,000 G for 15 minutes. Plasma was harvested into Eppendorf tubes and stored at −80°C until analysis. Breathing rate was assessed by counting flank movements for 15 seconds per minute over 3 minutes before each blood sampling. Heart rate and pedometer data were collected as described by Brosh et al.⁴ and Ben Meir et al.⁵

Figure 1.

Procedure for assessing heifer response to freeze branding (FB) in experiment 1. Each heifer was equipped with a leg sensor (Pedometer Plus; S.A.E. Afikim, Israel) on the metatarsus continuously recording step count, lying bouts, and lying duration at 15-minute intervals, and with a custom heart rate belt (Pegasus, Eli-ad, Israel) containing a Polar transmitter (T51H) and watch (S610i; Polar Electro Oy, Finland) placed around the thorax to log heart rate at 1-minute intervals. Inset: 2 heifers fitted with a green no-chew band securing the catheter.

2.2. Experiment 2

The study was conducted at the Volcani Center Experimental Dairy Farm (Beit Dagan, Israel). The Volcani Center Animal Care and Use Committee approved the protocol (IL 774/18). Twenty-five Holstein cows participated. Full experimental details are in Salzer et al.,⁴⁵ and Figure 2. Briefly, blood was drawn by venipuncture from the caudal vein before evenly spreading a ∼5 g of cream to a 10 × 10-cm tonsured patch of the cow's rump, and again 30 minutes after application. A vegetable oil was then used to remove the cream.²²

Figure 2.

Procedure for assessing the effect of capsaicin cream application on plasma biomarkers in lactating Holstein cows in experiment 2. The heart and breathing rate readings were taken before and immediately after or 30 minutes after the cream application. Blood was sampled before and 30 minutes after the cream application. The cream was applied on a tonsured patch on the left or right side of the cow's rump, using a 10 cm × 10 cm stencil. The control (neutral) and capsaicin treatments took place on different days. For general details, see Salzer et al. (2021).⁴⁵

Capsaicin (8-methyl-N-vanillyl-6-nonenamide) is a chili pepper irritant that induces a burning sensation upon contact with bodily tissues⁹ and causes pain at high dosage (10%) in humans³ and animals.^16,22,45 On day 1, an amphiphilic neutral base cream (Deutscher Arzneimittel-Codex) was applied (control); on day 2, 10% capsaicin (Affix Scientific) mixed with the base was used. Blood samples handled as earlier described.

2.3. Experiment 3

The experiment was conducted on a commercial farm located in the Hefer Valley region, central Israel. The Volcani Center Animal Care and Use Committee approved the protocols (IL 852/20). We hypothesized that mastitic cows perceive milking as a noxious event, reflected in elevated plasma pain biomarker.

2.3.1. Animals, housing, and management

Forty dairy cows participated (mean lactation 3.7 ± 1.7; age 4.61 ± 2 years; days in milk 206 ± 125; daily milk yield 38.2 ± 18.2 L). Cows were milked 3 times daily and routinely cooled during the summer months. Housing was in open ventilated sheds with free access to water and feed.

2.3.2. Preparation

A cow suspected of clinical mastitis—based on visible changes in milk (eg, lumps, discoloration, or viscosity) and/or udder (eg, swelling or sensitivity) during routine morning milking—was reported to the National Service for Udder Health & Milk Quality (MALE). MALE reviewed cows' data (age, days since calving, udder infection history) and selected 3 additional herd-mates—matched for age, calving date, and preferably days in milk—to represent healthy, subclinical, and chronic mastitis categories. This process was repeated on 10 days, yielding 40 cows (10 per group). Data from all 4 groups were collected on the same days, ensuring similar weather conditions, barn care interface, and other uncontrollable factors during the experiment.

Clinical mastitis (IDF 448/2011) was defined as a laboratory-confirmed udder infection accompanied by visible changes in the milk and/or udder.¹¹ Subclinical mastitis was absence of visible signs, but with a somatic cell count (SCC) exceeding 400,000 cells/mL milk and at least 2 positive diagnoses after the initial diagnosis. Chronic mastitis SCC above 400,000 cells/mL in 3 consecutive monthly tests, with a pathogenic diagnosis in the most recent test, and no visible signs. Healthy (control) cows had no clinical, subclinical, or chronic signs, SCC below 150,000 cells/mL in the last audit, and no detected pathogens.

On the trial day, the 4 cows were moved to the treatment yard with access to water, food, shade, and ventilation (Fig. 3). The dairy manager emptied the infected udder but withheld treatment until the trial ended. Treatment began within 6 hours of mastitis detection, provided the cow showed no signs of life-threatening illness, per IDF 498/2019 guidelines.

Figure 3.

Procedure for assessing the effect of mastitis manifestation on plasma biomarkers in lactating Holstein cows in experiment 3. Cows were moved to the head gate in the treatment yard for blood sampling, and for mounting of Polar S610i watch and T51H transmitter (Polar Electro Oy, Finland) behind the forelegs with an elastic belt (Pegasus, Eli-ad, Israel) to record heart rate every second. Milk was sampled at the milking parlor. Breathing rate was monitored for about 5 minutes before, during, and after milking. AfiAct-I leg sensors (Afimilk, Afikim, Israel) continually log cow activity. Elastic belts were removed before cows were returned to their home groups.

2.3.3. Experimental procedure

After routine morning milking (approximately 5–7 am), the 4 cows—healthy, chronic, subclinical, and clinical mastitis—were moved to the treatment yard. Blood samples were collected via venepuncture from the caudal vein into lithium heparin tubes and processed as previously described.⁴⁵ To evaluate the effect of milking as a potential pain stimulus in mastitis cows, an off-schedule milking was conducted at ∼11 am. Once at the milking parlor, milk samples were collected separately from each quarter of the udder for laboratory analysis, in accordance with MALE guidelines. After milking, cows returned to the treatment yard, where a second blood sample was drawn. Breathing rate was measured by counting flank movements for 15 seconds every minute over 5 minutes: first in the treatment yard with free-moving cows, then during milking, and finally postmilking back in the yard. Clinical mastitis cows were treated immediately after blood collection and then all 4 returned to their group.

Milk was sampled again on the following 2 days to confirm the mastitis-causing pathogen. All milk samples were stored at −15°C and transported the next day to MALE laboratory. Samples were plated on lamb blood agar, incubated at 32°C for 48 hours, and examined for bacterial pathogens per the National Mastitis Council (USA) protocols.⁶

2.4. Plasma biomarker analysis

Plasma samples for experiments 1, 2, and 3 were analyzed for BTD, SP, and cortisol levels. Plasma BTD concentration was measured using a Bovine BTD ELISA Kit (#EK3741; SAB Signalway Antibody), following the manufacturer's protocol. Standards, blank, or plasma samples were added to a 96-well plate and incubated. Detection reagents were then added for 1 hour. After repeated washing, a substrate solution was applied, and the reaction was stopped with a stop solution. Optical density was measured at 450 nm using a microplate reader, Multiskan FC (Thermo Scientific). BTD concentrations were calculated against a standard curve prepared by serial dilutions spanning 15.6 to 1000 pg/mL (see Table 1, supplemental digital content, http://links.lww.com/PR9/A346) and a blank (1000, 500, 250, 125, 62.5, 31.2, 15.6, 0 pg/mL). Plasma SP and cortisol concentration were measured using Bovine SP ELISA Kit (#MBS2609790; My BioSource) and cortisol ELISA assay (EIA-1887 DRG Diagnostics) following the manufacturers' protocols (see Table 1, supplemental digital content, http://links.lww.com/PR9/A346). In experiment 1, plasma concentrations of BTD, SP, and cortisol were measured at 0, 10, 20, 30, 60, 180, and 360 minutes relative to the time of branding and normalized to each heifer's individual baseline value at −120 minutes.

2.5. Statistical analysis

Statistical analyses were conducted using SPSS 28, and complimentary Python using SciPy.Stats and custom functions. Data are presented as Mean ± SE. In experiment 1, a mixed-model ANOVA assessed the effects of time (within-subjects), group (between-subjects), and their interaction on each biochemical marker. Mauchly test indicated violations of sphericity for the within-subjects factor in all ANOVAs; thus, degrees of freedom were adjusted using the Huynh–Feldt correction. All reported F-statistics and P-values reflect this adjustment. Significant group effects were further examined with Bonferroni post-hoc tests. In experiment 2, comparisons of pre- and postapplication (+30 minutes) of neutral and capsaicin creams were performed using paired sample t-tests. Effect sizes were estimated using Cohen d. In experiment 3, the Shapiro–Wilk test rejected normality; prespecified group comparisons for each biomarker (BTD, SP, cortisol) were performed using one-sided Mann–Whitney U tests—clinical vs healthy, subclinical vs chronic, healthy vs subclinical and chronic (pooled), and clinical vs subclinical and chronic (pooled). To correct for multiple testing across 4 comparisons, significance was defined as P < 0.0125 (Bonferroni-adjusted alpha). Effect sizes were estimated using pooled Cohen d.

Receiver operating characteristic (ROC) curve analysis was used to assess the ability of BTD, SP, and cortisol to discriminate between painful and painless conditions: control vs FB groups (experiment 1), capsaicin vs neutral cream application (experiment 2), and healthy vs clinical cows (experiment 3). The area under the ROC curve (AUC) quantified overall test performance. Sensitivity and specificity were determined at the optimal threshold, defined by the maximum Youden index (sensitivity + specificity − 1). Analyses were conducted in Python using scikit-learn.

Analyses and results of breathing rate (BR), heart rate (HR), and activity metrics are detailed in the Supplementary Materials section.

3. Results

3.1. Experiment 1

Significant main effects of time (F₆,₇₈ = 6.16, P < 0.001) and group (F₂,₁₃ = 5.03, P = 0.024) were found for plasma BTD (Fig. 4A). No significant group × time interaction was observed (F₁₂,₇₈ = 0.94, P > 0.05). Bonferroni-corrected post-hoc tests showed lower BTD in controls (0.57 ± 0.05) compared to the FB groups (with analgesics 1.55 ± 0.2; without analgesics 1.4 ± 0.12; P < 0.05), but not between FB groups. Pairwise t-tests with Holm correction showed significantly higher BTD levels in the heifers subjected to FB without analgesics group vs controls at 30 minutes (P = 0.0211) and 60 minutes (P = 0.0041), and in the heifers subjected to FB with-analgesics group vs controls at 180 minutes (P = 0.0017). Other comparisons were not significant. Mixed-model ANOVA showed no significant effects of time (F₆,₇₈ = 0.73, P > 0.05), group (F₂,₁₃ = 0.73, P > 0.05), or their interaction (F₁₂,₇₈ = 0.48, P > 0.05) on plasma SP levels (Fig. 4B). Similarly, no significant effects of time (F₆,₉₀ = 2.37, P > 0.05), group (F₂,₁₅ = 0.63, P > 0.05), or interaction (F₁₂,₉₀ = 0.66, P > 0.05) were found for cortisol (Fig. 4C).

Figure 4.

Kinetics of plasma biomarkers after freeze branding (FB) in experiment 1. Holstein heifers (N = 18) were subjected to freeze branding. Plasma concentrations of (A) biotinidase, (B) substance P, and (C) cortisol are shown for control (light gray), FB with analgesics (medium gray), and FB without analgesics (black) groups at 0, 10, 20, 30, 60, 180, and 360 minutes relative to branding, normalized to baseline at −120 minutes. Error bars represent SEM. An asterisk indicates a time point with significant group differences (P < 0.05).

3.2. Experiment 2

Plasma BTD concentrations increased significantly 30 minutes after capsaicin treatment (P = 0.002; d = 73), rising from 18.5 ± 2.5 to 21.5 ± 2.7 ng/mL, but not after neutral cream application (P = 0.26; Fig. 5A). Plasma SP concentration increased significantly after both treatments, rising from 57.5 ± 6.8 to 70.2 ± 8.1 pg/mL with neutral cream (P = 0.001; d = 86), and from 44.3 ± 6.1 to 52.5 ± 6.7 pg/mL with capsaicin (P = 0.001; d = 89; Fig. 5B). Plasma cortisol levels remained unchanged after neutral cream, but decreased significantly after capsaicin cream from 22.6 ± 5.2 to 12.0 ± 3.9 ng/mL (P = 0.002; d = 81; Fig. 5C).

Figure 5.

Effect of topical capsaicin application on plasma biomarkers in experiment 2. Plasma concentrations of (A and B) biotinidase, (C, D) substance P, and (E and F) cortisol measured in Holstein cows (N = 24) before and 30 minutes after control (left panels) and 10% capsaicin (right panels) cream application. Each dot represents an individual cow. Boxplots indicate median, interquartile range, and range. An asterisk indicates a significant difference (P < 0.05).

3.3. Experiment 3

The pathogen analysis confirmed the absence of bacterial growth in healthy cows. Among the subclinical mastitis group, 6 cows exhibited pathogen growth: Enterococcus (1 cow), Streptococcus uberis (1), non-aureus Staphylococcus (3), T. pyogenes (1), whereas 4 cows had no detectable pathogens. In the chronic mastitis group, isolates included: non-aureus Staphylococcus (5), Candida sp. (1), Streptococcus uberis (3), and one mixed infection involving Candida and other streptococci (1); one cow showed no growth. In the clinical mastitis group, pathogens identified were non-aureus Staphylococcus (3), Streptococcus uberis (1), E. coli (3), Bacillus cereus (1), and no detectable growth in one cow. Data from the 6 cows with no detectable pathogens across the clinical, chronic, and subclinical groups were excluded from further analysis.

Healthy cows (13.69 ± 5.62 ng/mL) exhibited significantly lower plasma BTD concentrations compared to clinical (31.19 ± 7.49 ng/mL; U = 73, P = 0.012, d = 0.80) and the subclinical and chronic mastitis groups (17.33 ± 5.71 and 15.88 ± 3.29 ng/mL, respectively; U = 83, P < 0.001, d = 0.99; Fig. 6A). No significant differences were observed between the clinical and the subclinical and chronic groups (U = 19, P = 0.97), nor between subclinical and chronic cows (U = 9, P = 0.98).

Figure 6.

The effect of mastitis manifestation on plasma biomarkers in lactating Holstein cows in experiment 3. Plasma concentrations of (A) biotinidase, (B) substance P, and (C) cortisol in lactating Holstein cows (N = 34) allocated to 4 groups based on mastitis manifestation: healthy, chronic, subclinical, and clinical. Each point represents an individual cow. Groups not sharing a letter differ significantly (P < 0.05). Boxplots indicate the median, interquartile range, and range, with individual data points overlaid.

For SP, all comparisons were nonsignificant (P > 0.05; Fig. 6B). For cortisol, healthy cows (19.10 ± 5.78) had exhibited significantly lower levels compared to the subclinical and chronic groups (16.67 ± 3.98 and 8.59 ± 3.89 ng/mL; U = 56, P = 0.0012, d = 0.8; Fig. 6C). All other pairwise comparisons were not statistically significant.

3.4. Sensitivity and specificity

Biotinidase showed the highest AUC values in experiments 1 and 3 (0.84 and 0.81, respectively), outperforming SP (0.36 and 0.55) and cortisol (0.71 and 0.41). In experiment 2, all biomarkers had AUCs of 0.52 or below (BTD: 0.52, SP: 0.37, cortisol: 0.50), indicating performance no better than chance. Optimal sensitivity and specificity for each biomarker in each experiment are detailed in Figure 7.

Figure 7.

Receiver operating characteristic (ROC) curves for biotinidase, substance P, and cortisol classification performance across 3 independent experiments. Each panel displays the true positive rate (sensitivity) versus false positive rate (1-specificity) with corresponding area under the curve (AUC) values. Red dots indicate optimal threshold points determined by maximizing Youden index. Inserts: sensitivity and specificity values at optimal thresholds. The diagonal dashed line represents random classification performance (AUC = 0.5).

4. Discussion

Effective and timely pain recognition is essential for safeguarding animal health and welfare, enabling early therapeutic intervention before clinical conditions deteriorate.

However, pain detection in cattle remains challenging due to their inherently stoic nature.^21,50 Although severe pain is accompanied by overt behavioral and physiological sings, mild pain often goes unnoticed. Despite the importance of accurate pain assessment to ensure animal welfare, a definitive validated biomarker for pain in cattle has yet to be established.⁵⁰

Biotinidase is primarily known for its amidase activity on biocytin,⁴⁴ but also demonstrates amino-exopeptidase activity, enabling the cleavage of opioid neuropeptides in serum.⁴³ This enzymatic function may play a direct role in nociceptive modulation by altering endogenous opioid signaling. In support of this, studies in bacterial meningitis models have reported rapid increases in BTD activity during acute inflammation, which coincide with elevated enkephalin levels in the central nervous system and fever,^19,30 suggesting a role for CNS BTD in hydrolyzing analgesic peptides under inflammatory conditions. In addition, recent clinical findings in humans have reported an association between elevated BTD activity and arthritis-related pain severity.⁵⁴ Biotinidase is primarily secreted by the liver^27,45 and presents the highest specific activity in the plasma [57]. Driven by the above findings, we hypothesized that BTD may serve as a novel plasma biomarker for pain in cattle.

Collectively, our findings across 3 distinct pain inducing conditions support the hypothesis that, beyond its established role in biotin metabolism,^28,31 plasma BTD may serve as a pain biomarker in cattle. This is evidenced by the elevated levels of BTD in response to both experimentally induced and naturally occurring pain stimuli, reinforcing its involvement in nociceptive processes. A fundamental criterion for a reliable pain biomarker is its ability to respond rapidly to a pain event, thereby enabling timely clinical intervention. In this context, both the freeze branding and capsaicin application trials demonstrated the potential of plasma BTD as a real-time indicator of pain. Notably, a significant increase in BTD levels was detected within 30 minutes post-FB relative to controls, as well as topical capsaicin exposure. Furthermore, elevated plasma BTD concentrations in mastitic cows support broader applicability as a pain biomarker under both, controlled and naturally occurring conditions.

Circulating SP has been employed in numerous studies to evaluate inflammation and nociception in cattle subjected to painful routine procedures such as castration, disbudding, dehorning, and umbilical surgery.^{2,12,13,17,29,36,37,44,49} However, although some studies reported elevated SP concentrations in response to painful stimuli, others observed no measurable change. Such variability hinders the establishment of SP as a reliable biochemical marker for pain in cattle.⁵⁰ Moreover, although a few studies have reported a short-term or immediate elevation in plasma SP after pain induction,¹² others have described a delayed increase 8 to 24 hours after lameness,⁸ a decrease 96 hours after dehorning,¹³ or no change 24 to 96 hours after disbudding.^29,49 Similarly, no uniform SP response was observed across the 3 pain-inducing stimuli examined in the present study; SP levels were unaffected by FB or spontaneous mastitis, but increased after both neutral and capsaicin cream applications, suggesting a lack of specificity to nociceptive stimulation.

Several inherent biological factors limit the reliability of plasma SP as a pain biomarker. Interindividual variability and age-related differences complicate the comparison of true differences in SP concentrations across individuals.^12,17,50 Moreover, SP is involved in immune activation, resulting in fluctuations in its circulating levels, driven not only by nociceptive stimuli but also by general stress and inflammatory states.^19,34 From methodological perspective, accurate quantification of SP in plasma is challenging, as it necessitates the use of protease inhibitors and immediate sample processing¹⁷ to mitigate rapid peptide degradation.³⁹ Collectively, current evidence, including our findings, suggests that SP responses may differ according to the nature of the stimulus, indicating that its utility as a biomarker is likely to be context-dependent.

Given their rapid release, neuropeptides hold potential as serum biomarkers for pain. However, because of their short half-life,³² they are often studied alongside longer-lasting indicators like cortisol,^8,35,43,46 to capture a broad range of response kinetics.³³ Nevertheless, the interpretation of cortisol as a serum biomarker for pain is confounded by its susceptibility to various non-nociceptive influences, including stress, circadian rhythm, handling, human presence, or physical restraint. The extent of cortisol elevation is further modulated by the nature of stressor and individual temperament.^25,33,38 In the present study, neither the capsaicin-induced challenge nor naturally occurring mastitis elicited significant changes in plasma cortisol concentrations, thereby undermining its reliability as a specific indicator of pain in these contexts.

In the first experiment, plasma cortisol of heifers subjected to FB without analgesics followed closely the kinetic pattern of BTD. In heifers treated with analgesics, plasma cortisol levels remained consistently low as anticipated, whereas BTD did not. Several plausible options may underlie this divergence. First, the attenuated cortisol response likely reflects the pharmacological effects of the analgesics, which are known to inhibit hypothalamic–pituitary–adrenal (HPA) axis activation and subsequently reduce cortisol secretion.²⁰ Second, differential patterns of cortisol and BTD responses suggest that the combined use of these biomarkers could help disentangle the stress-related and nociceptive components of the procedure of FB.¹ Supporting this approach, Kleinhenz and colleagues evaluated the effect of flunixin meglumine, a nonsteroidal anti-inflammatory drug, on plasma cortisol and SP concentrations, in calves undergoing surgical castration.³⁰ Although SP levels remained unchanged, flunixin administration led to a reduction in cortisol levels compared to placebo, demonstrating its effectiveness in moderating the physiological stress response associated with the procedure. Third, a fast-responding pain biomarker is expected to increase (or decrease) after a noxious stimulus and decrease (or increase) after the administration of effective analgesia.¹⁰ However, if the analgesic treatment is insufficient or ineffective, the biomarker may serve as an indicator of ongoing pain. In this context, the sustained elevation of BTD response may suggest that the analgesic regimen used during FB may not have fully alleviated pain. Therefore, simultaneous measurement of plasma BTD and cortisol could provide complementary insights into the efficacy of different analgesic protocols and varying pain conditions, warranting further investigation.

Biotinidase shows potential as a diagnostic biomarker for identifying pain in animals. However, to be established as a standardized pain biomarker, it must meet key criteria: clinical validity, sensitivity, specificity, reproducibility, and robustness.²⁴ The high AUC values of BTD in experiments 1 and 3 indicate its promise as a pain marker for discrimination between painful and painless events. The low AUC for substance P suggests that its classification was random or even inversely predictive. Cortisol's AUC was above chance in experiment 1 but at chance level in experiment 3. Notably, experiments 1 and 3 involved real-life farm events, whereas experiment 2 used an artificial pain model. Thus, BTD limited ability to discriminate pain in experiment 2 may reflect the capsaicin challenge being less painful than freeze branding or mastitis.

A fundamental limitation of this study, shared by all animal-based pain research, is the difficulty in establishing biomarker sensitivity in the absence of subjective feedback, such as self-reported pain scores. Even if BTD is responsive to subtle nociceptive changes, definitive validation remains difficult. Nonetheless, the reproducibility of our findings across multiple time points, farms, and animal cohorts supports the reliability and potential applicability of the proposed protocol. Independent validation is required to enhance confidence in BTD diagnostic utility. From a practical standpoint, the current reliance on invasive plasma sampling limits its field applicability. Future research should investigate alternative biological matrices such as saliva or milk. Finally, like cortisol and SP, BTD is implicated in multiple physiological pathways and is not exclusively pain-specific. Therefore, we propose that BTD be integrated with complementary biomarkers, as well as physiological, and behavioral indicators, to form a robust, multimodal approach for pain detection.

Disclosures

The authors have no conflict of interest to declare.

Supplemental digital content

Supplemental digital content associated with this article can be found online at http://links.lww.com/PR9/A346.