Plasma NMR metabolomics reveals a powerful multi-marker signature in canine inflammatory conditions

Plasma NMR metabolomics reveals a powerful multi-marker signature in canine inflammatory conditions | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Plasma NMR metabolomics reveals a powerful multi-marker signature in canine inflammatory conditions Claudia Elo, Mirja Kaimio, Essi Leminen, Hannes Lohi This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-6343599/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract BACKGROUND Identifying inflammatory states is crucial in veterinary diagnostic workups. While biomarkers for acute inflammation are widely utilized, there is a lack of reliable markers for chronic inflammation. OBJECTIVES This study aims to characterize metabolic alterations associated with common inflammatory diseases in dogs and evaluate the efficacy of metabolic markers in identifying inflammatory states. ANIMALS Plasma samples from 175 healthy dogs and 207 dogs diagnosed with specific acute and chronic diseases were collected during veterinary visits. Conditions studied included acute gastroenteritis, pyometra, neoplastic diseases, atopic dermatitis, periodontitis, urinary tract infections, canine infectious respiratory disease complex, and osteoarthritis. METHODS Samples were analyzed using a canine-validated 1H NMR spectroscopy platform. Logistic regression was employed to identify both general and disease-specific metabolic alterations. A multivariable metabolite model was developed, and its diagnostic performance was compared against C-reactive protein (CRP), albumin, glycoprotein acetyls (GlycA), and a combination of GlycA and albumin. RESULTS Metabolic changes were observed across all conditions studied, with some alterations specific to individual diseases and others common across conditions. The multivariable metabolite model demonstrated excellent overall diagnostic performance (AUC = 0.82; 95% CI: 0.78–0.86). Importantly, this model detected inflammation significantly better (p < 0.05) than CRP in all chronic diseases (AUC range: 0.68–0.89 vs. 0.46–0.60) and acute gastroenteritis (AUC: 0.86 vs. 0.71). Furthermore, it consistently showed higher AUC values compared to CRP in all diseases analyzed. CONCLUSIONS Metabolic profiling can effectively detect both acute and chronic inflammation in dogs. This approach appears superior to CRP, particularly for identifying chronic inflammatory conditions. Animal Physiology Cellular Metabolism metabolomics CRP inflammation GlycA Figures Figure 1 Figure 2 Figure 3 Figure 4 INTRODUCTION Inflammation is a common biological response observed in numerous diseases, aiming to restore homeostasis during and after tissue insult. Beyond conditions typically recognized as inflammatory—such as infectious or autoimmune diseases—many acute and chronic illnesses feature a component of systemic low-grade inflammation 1 . Chronic diseases often induce systemic inflammation; however, the connection between low-grade systemic inflammation and disease extends beyond this causal relationship. In humans, low-grade systemic inflammation can precede clinical disease onset by as much as a decade, and it contributes to various conditions ranging from diabetes mellitus to Alzheimer's disease 2 . Identifying an inflammatory state is crucial in clinical diagnostics. Inflammatory markers not only confirm the presence of inflammation but also help determine its severity, prognostic implications, and associated morbidity and mortality risks 3 . While practical biomarkers such as C-reactive protein (CRP) effectively identify acute inflammation, there is a lack of validated biomarkers for chronic low-grade inflammation in dogs 4 . Establishing reliable markers for chronic inflammation in dogs could facilitate earlier detection of chronic diseases, improving overall canine health and longevity. Glycoprotein acetyls (GlycA) represent a promising biomarker for chronic low-grade inflammation. GlycA combines proton nuclear magnetic resonance (1H NMR) spectroscopy signals from acute-phase proteins including α1-acid glycoprotein, haptoglobin, α1-antitrypsin, α1-antichymotrypsin, and transferrin, with minor contributions from glycosylated apolipoproteins 5 . Extensively studied in humans, GlycA is associated with chronic inflammation, chronic diseases, and increased morbidity and mortality risk 1 , 6 – 8 . Its advantages for detecting chronic inflammation include stability and consistent responsiveness to inflammatory stimuli 5 . Elevated GlycA levels have also been observed in dogs across multiple conditions analogous to human diseases, including liver shunts and hepatopathies 9 , Addison's and Cushing's diseases 10 , diabetes mellitus and hypothyroidism 11 , lipemia 12 , and during phenobarbital treatment 13 . Inflammation also influences metabolism, with metabolic changes potentially signaling inflammatory states. Detecting these metabolic alterations may not only enhance the identification of inflammatory conditions but also offer therapeutic insights. Recently, a validated 1 H NMR-based metabolomics platform for canine plasma and serum samples has been established 12 , enabling comprehensive evaluation of GlycA and over 120 additional metabolic biomarkers in canine inflammatory diseases. The objectives of this study were to: (i) identify universal metabolic alterations associated with several diseases commonly observed in veterinary practice; (ii) develop a metabolite-based multivariable model for detecting inflammation in dogs; (iii) characterize specific metabolic changes induced by each studied condition; and (iv) compare the discriminatory performance of GlycA, albumin, their combination, and the newly developed multivariable metabolic model with that of CRP in identifying inflammation in dogs. MATERIALS AND METHODS Ethical approval The blood sample collection from privately owned pet dogs is approved in the project license from the Animal Ethical Committee of the County Administrative Board for Southern Finland (ESAVI/16933/2021). Whenever feasible, blood samples were collected from the same needle puncture used for routine laboratory diagnostics performed for the benefit of the patient. Dog owners provided informed consent prior to inclusion in the study and retained the right to withdraw at any point. Clinical data for each dog were collected in a pseudonymized format, and no personal owner information was gathered during the study. Study population We conducted an observational, cross-sectional study involving dogs recruited during routine patient visits at four Finnish small animal veterinary hospitals: Eläinsairaala Evidensia Tammisto (Vantaa), Eläinsairaala Mevet (Helsinki), Lahden Eläinlääkäriasema (Lahti), and Espoon Eläinsairaala (Espoo). The recruitment period spanned from May 2022 to April 2023. A total of 209 healthy control dogs and 226 dogs diagnosed with predefined common canine diseases participated in the study. The investigated conditions included acute gastroenteritis, pyometra, cutaneous and noncutaneous neoplasia, atopic dermatitis, periodontal disease, urinary tract infection, canine infectious respiratory disease complex, and osteoarthritis. Dogs could present with multiple concurrent conditions, and both treatment-naïve and previously medicated patients were included due to many dogs already undergoing treatment at enrollment. Diagnoses were determined based on patient history, clinical signs, findings, and appropriate diagnostic tests tailored to each patient's condition. For neoplastic conditions, dogs initially enrolled based on clinical suspicion had their diagnoses confirmed through subsequent histological examination. Control group inclusion required a physical examination performed within the past three months, a medical history free of indications suggestive of illness, and clinical chemistry, hematology, and C-reactive protein (CRP) levels within normal ranges. In-house clinical chemistry analysis utilized the Catalyst Dx Chem15 analyzer, and hematology was assessed using ProCyte (both from IDEXX Laboratories, Inc., Maine, USA). For all participating dogs, detailed information on signalment, health status, medication usage, and fasting duration prior to blood sampling was collected. CRP was measured using an in-house Catalyst Dx CRP analyzer (IDEXX Laboratories, Inc., Maine, USA), with a reference interval of 0–10.0 mg/L. However, the analyzer detection limits varied among clinics, ranging from lower detection limits of either 1.0 mg/L or 10.0 mg/L, to upper limits of 170 mg/L or 240 mg/L. For metabolomic analysis, 2 ml of blood were drawn via cephalic venipuncture into Vacuette LH Lithium Heparin PREMIUM tubes (2 ml). Plasma volumes ranging from 110 to 500 µl were separated according to the manufacturer's recommendations and subsequently transferred into cryo tubes (Sarstedt 72.730.005). Samples were refrigerated at the veterinary clinics for up to one week, then transported within three hours under cool-pack conditions. After transportation, samples were stored at -80°C for a maximum of five months prior to metabolomic analysis performed by PetMeta Labs Ltd. (Helsinki, Finland) using proton nuclear magnetic resonance ( 1 H NMR) spectroscopy. 1 H NMR spectroscopy A targeted 1 H NMR spectroscopy method optimized and validated for dogs was utilized 12 . The method has been thoroughly described elsewhere 12,14,15 . The method is highly automated, from sample processing to the processing of the NMR spectra. The equipment used includes a sample changer SampleJet (Bruker Corp., Billerica, Massachusetts, USA), a PerkinElmer JANUS Automated Workstation with an 8-tip dispense arm with Varispan (PerkinElmer Inc., Waltham, Massachusetts, USA) and a Bruker AVANCE III HD 500 NMR spectrometer with a 5 mm triple-channel (1H, 13C, 15N) z-gradient Prodigy probe head 14 . Sample preparation included light mixing of the sample and removing the possible precipitate by centrifugation, sample transfer to individual NMR tubes, and mixing with sodium phosphate buffer 14,16 . Metabolite quantification in absolute units is achieved by processing the NMR spectra using scripts optimized for canine samples 12 based on regression modeling 15 . The proprietary software utilized has integrated quality control 15 . The method quantitates 123 measurands, including a detailed lipoprotein analysis, amino acids, fatty acids, triglycerides, cholesterols, glycolysis-related metabolites, albumin, creatinine, and GlycA 12 . Data pre-processing Laboratory and health data of all participating dogs underwent thorough review and evaluation. Dogs initially assigned to the control group were excluded if their CRP levels, clinical chemistry, hematology results, or general health information indicated either subclinical or clinical disease. Control group dogs displaying laboratory parameters outside reference ranges without clear disease indications were flagged for potential exclusion from subsequent analyses if metabolomics data identified them as outliers. Similarly, the disease status of case-group dogs was meticulously reassessed, resulting in exclusion of individuals with uncertain or incorrect diagnoses, or, specifically within neoplasia cases, those lacking histological confirmation. Metabolomics data were subjected to rigorous quality control. Dogs with extensive missing metabolite data were excluded. Individual metabolites were assessed for completeness, with those exhibiting greater than 20% missing observations removed. Specifically, metabolites such as very large very low-density lipoprotein phospholipids (XL-VLDL PL), esterified cholesterol (XL-VLDL CE), and free cholesterol (XL-VLDL FC)—known for characteristically low canine levels—were excluded. Additionally, metabolomics data were screened for outliers. When control group outliers were identified, their health and laboratory data were reevaluated. Dogs exhibiting significant deviations, including previously noted laboratory abnormalities unrelated to clear disease or moderate alterations in muscle mass or body condition, were excluded. Following these quality assurance steps, the final dataset comprised 175 control dogs and 207 case dogs. Remaining missing metabolite values were imputed using Random Forest imputation, employing the R packages missForest version 1.5 18 , dplyr version 1.1.2 19 and tidyverse 20 . The imputation's out-of-bag (OOB) error was 0.136, indicating adequate performance. A separate approach was applied for CRP value imputation: values below the detection limit were imputed as half the detection limit, while values exceeding the detection limit were imputed as detection limit plus 10 mg/L. One pyometra group dog lacking CRP measurements was excluded from analyses involving CRP. Statistical analyses and data preprocessing were conducted using the R programming environment 17 (R Core Team, 2023) and Microsoft Office Excel (Microsoft Corporation, Redmond, WA, USA). Statistical analyses Descriptive statistics were computed to characterize the study groups. The normality of metabolite variables was assessed using the Shapiro-Wilk test (stats package, R Core Team, 2023) 17 , indicating most metabolites were not normally distributed. P-value adjustments for metabolite analyses were conducted using Bonferroni correction based on the number of principal components (PCs) explaining >95% of data variance. Principal component analysis (PCA) was performed using FactoMineR and factoextra packages, with five PCs explaining 95% variance, resulting in a p-value threshold of 0.01 (0.05/5). Initially, general metabolic changes associated with disease were evaluated by comparing the combined case group (dogs diagnosed with one or more diseases) with the healthy control group. Before this comparison, physiological characteristics (sex, neutering status, fasting duration, age) between the groups were examined using the Wilcoxon rank-sum test (stats package) 17 , with significant differences identified only for age. Given age's known influence on metabolite concentrations 21 , age was included as a covariate in subsequent analyses. Additionally, the potential effect of medications administered within 24 hours prior to sampling on metabolite levels in the diseased group was assessed, revealing no significant differences. Metabolite concentrations between the case and control groups were compared using logistic regression (stats package) 17 , adjusted for age. The case-control status served as the dependent variable, and results were visualized using bubble plots created with bubbleHeatmap (version 0.1.1) 22 and grid 17 packages. The bubble plots reflected logistic regression p-values and median absolute deviation (MAD) differences from the dataset median. To confirm result robustness, the Wilcoxon rank-sum test (stats package) 17 was also performed, yielding highly similar results. A multivariable logistic regression model to identify inflammatory markers was developed using metabolites significant (p<0.05) in univariable logistic regressions. The model, adjusted for age, was constructed through forward stepwise selection based on p-values using the stats package 17 . Variables were sequentially added until no further metabolites reached statistical significance (p<0.05). Akaike information criterion (AIC)-based selection yielded identical metabolites at each step. The multivariable model underwent diagnostic assessment to ensure logistic regression assumptions were met. Multicollinearity was evaluated using variance inflation factors (VIF; car package 23 ), revealing no multicollinearity (VIF range: 1.07–1.58). Influential observations were examined through standardized residual plots (broom package 24 ), showing no influential outliers (standardized residuals 0.05). Variable importance was assessed using the caret package. Model performance was evaluated using 10-fold cross-validation (cvAUC package version 1.1.4) 25 , with the area under the receiver operating characteristic curve (AUC) as the performance metric. Standard AUC interpretations were applied: >0.7 acceptable, >0.8 excellent, and >0.9 outstanding. Next, metabolic changes specific to each disease were investigated individually. Dogs with concurrent conditions were included in analyses for each relevant disease. Physiological characteristics were compared using the Wilcoxon rank-sum test (stats package) 17 , and results were summarized in Table 1. While some groups exhibited differences in sex or neutering status compared to controls, group sizes limited statistical adjustments. Considering sex and neutering status minimally impact metabolite concentrations 21 , only age was included as a covariate in logistic regression models. Effects of concurrent diseases and recent medications on metabolite concentrations were evaluated, identifying no significant effects, except for higher docosapentaenoic acid percentage in acute gastroenteritis dogs with concurrent illnesses. Disease-specific metabolite comparisons between the control group and each disease group were conducted using logistic regression (stats package) 17 , adjusting for age. Bubble plots visualizing logistic regression p-values and MAD differences from control medians were created using bubbleHeatmap (version 0.1.1) 22 and grid 17 packages. Wilcoxon rank-sum tests (stats package) 17 validated logistic regression findings, with only minor interpretation differences. Finally, the diagnostic performance of CRP, GlycA, albumin, the GlycA-albumin combination, and the multivariable metabolite model was compared using logistic regression (stats package) 17 . Performance was assessed and visualized with ROC curves (pROC package) 26 . AUC values were calculated, and differences in AUC between CRP and metabolic measures were statistically compared using the DeLong method. The dependent variable for each logistic regression model was disease presence versus control. Differences in CRP and metabolite concentrations across groups were illustrated through box plots using forcats (version 1.0.0) 27 and ggpubr (version 0.6.0) 28 packages, incorporating reference intervals for clarity. Final graphical adjustments were completed using Inkscape 29 . Results Group demography Demographic characteristics of the study groups are presented in Table 1. Dogs with pyometra, neoplastic diseases, periodontitis, and arthritis were significantly older compared to the control group, while those with canine infectious respiratory disease complex were notably younger. All study groups included diverse breeds and mixed-breed dogs. Among noncutaneous neoplasias, the most frequently observed were mammary tumors (n = 9), lymphoma (n = 8), leukemia (n = 3), and mast cell tumors (n = 3). The most common cutaneous neoplasias were cutaneous mast cell tumors (n = 9) and lipomas (n = 7). The prevalence of concurrent diseases varied among disease groups, ranging from 19% in dogs with atopic dermatitis to 52% in those with osteoarthritis. Administration of medication within 24 hours prior to sample collection was common, occurring in 51% of case group dogs. These medications included treatments specific to the diagnosed conditions as well as sedatives administered when blood sampling coincided with procedures requiring anesthesia, such as tumor removal or dental treatments. Table 1 . Demographical information of the study groups. *significant difference from the control group. %M/F: Percentage of males and females, %neut: percentage of neutered dogs, %drug24h: percentage of dogs that had received medication other than routine antiparasitic drugs within the last 24 hours prior to blood sampling, %concurrent dis: percentage of dogs having one or more concurrent diseases, All: all case group dogs with one or more of the studied diseases, AD: atopic dermatitis, AGE: acute gastroenteritis, CIRDC: canine infectious respiratory disease complex, OA: osteoarthritis, OT: noncutaneous neoplasia, Paro: parodontitis, Pyo: pyometra, ST: cutaneous neoplasia, UTI: urinary tract infection. Group Disease N Age median (range) %M/F %neut N breeds %drug24h %concurrent dis Control 175 5.6 (1.0 - 16.4) 36/64 33 64 Case All 207 8 (0.3 - 16.5)* 44/56 42 90 51 30 AD 27 4.5 (1.6 – 11.4) 61/39* 50 20 59 19 AGE 29 5.6 (0.6 – 12.7) 34/66 48 24 45 21 CIRDC 16 2.0 (0.4 – 7.4)* 50/50 6* 14 38 19 OA 21 9.9 (3.3 – 16.5)* 48/52 71* 15 62 52 OT 41 10.3 (3.5 – 14.4)* 44/56 52 16 56 39 Paro 33 9.5 (3.3 – 16.5)* 55/45* 45 25 48 30 Pyo 14 8.4 (4.9 – 11.3)* 0/100* 0* 13 42 42 ST 27 9.2 (2.4 – 16.1)* 48/52 52 19 44 44 UTI 19 7.8 (0.3 – 14.7) 37/63 37 14 32 26 Metabolite association with overall disease occurrence Several metabolites showed significant associations with overall disease occurrence when comparing the healthy control group to the combined case group, which included all dogs diagnosed with one or more of the studied conditions (Figure 1, Supplementary Tables O1 and O8). Among the strongest associations were the inflammatory markers GlycA and albumin. GlycA levels were elevated in the case group, while albumin, a negative acute-phase protein, was decreased. Notable alterations in amino acids included a significant decrease in tyrosine and glycine and an increase in the phenylalanine-to-tyrosine ratio. In metabolites related to sugar metabolism, pyruvate showed the most pronounced elevation in the case group. Total cholesterol, triglycerides, and very low-density lipoprotein (VLDL) levels did not differ significantly between groups, and changes in high-density lipoprotein (HDL) and low-density lipoprotein (LDL) cholesterol were minor. However, large and small HDL particle numbers (L-HDL and S-HDL), total lipids, cholesterol, and esterified cholesterol concentrations were significantly reduced in the case group, whereas large LDL (L-LDL) variables were notably increased, except for triglyceride levels. Detection of overall disease occurrence using the metabolic profile The forward stepwise selection process is shown in Supplementary Table O1-O7. The final multivariable logistic regression model included six metabolic measures: albumin, GlycA, pyruvate, alanine, glucose, and the ratio of phenylalanine to tyrosine. While the model was adjusted for age, age was not a significant predictor of disease in the model (Table 2 and Supplementary Table 1). The average AUC of the 10-fold cross-validated model was 0.82 (95% CI: 0.78 - 0.86), indicating excellent performance. Table 2 . Importance of each variable included in the final multivariable metabolite model. Predictor Variable importance Albumin 100 GlycA 91.54 Pyruvate 84.99 Ala 59.01 Glucose 53.94 Phe/Tyr 39.86 age 0.00 Metabolite associations with specific diseases While some changes in metabolite concentrations were universal across different diseases, multiple diseases also showed disease-specific changes (Figure 2, Supplementary Table D1). Universal changes across different conditions included decreased albumin and tyrosine and increased GlycA and pyruvate. An increase in phenylalanine and the phenylalanine/tyrosine ratio was mainly observed in acute diseases. Changes in other amino acids were relatively disease-specific, while all changes except phenylalanine were decreases in concentration. Multiple very low-density lipoprotein (VLDL) lipid variables and total triglycerides were increased in dogs with osteoarthritis, parodontitis, neoplastic diseases, and pyometra while being decreased in dogs with atopic dermatitis, acute gastroenteritis, and canine infectious respiratory disease complex. HDL and LDL variables showed different signatures. LDL variables, especially L-LDL variables, were increased in dogs with osteoarthritis, pyometra, and noncutaneous neoplasia with no or little changes in other diseases. Total HDL and total cholesterol remained unaltered in all diseases. Still, small and large HDL variables were relatively uniformly decreased across multiple conditions, including atopic dermatitis, acute gastroenteritis, canine infectious respiratory disease complex, noncutaneous neoplasia, and pyometra. In fatty acid concentrations, a decrease in oleic acid, docosapentaenoic acid, and omega-3 fatty acids and an increase in saturated fatty acids were observed in pyometra. Decreased omega-3 polyunsaturated fatty acid (PUFA) concentrations were observed in parodontitis, pyometra, cutaneous neoplasia, and urinary tract infections. Citrate was reduced in atopic dermatitis, acute gastroenteritis, and canine infectious respiratory disease complex, and acetate followed a similar pattern. Glucose was elevated in parodontitis and pyometra and slightly in the other studied acute conditions. Generally, dogs with atopic dermatitis had the fewest and slightest metabolic changes. Performance of CRP, GlycA, albumin, the combination of GlycA and albumin, and the multivariable metabolite model in classifying dogs as healthy and diseased The acute inflammatory marker CRP could not acceptably (AUC < 0.7) detect individuals with the studied chronic inflammatory conditions atopic dermatitis, osteoarthritis, cutaneous neoplasia, noncutaneous neoplasia, and parodontitis. In contrast, the acute conditions of acute gastroenteritis and urinary tract infection were acceptably (AUC 0.7-0.8) detected by CRP, canine infectious respiratory disease complex excellently (AUC 0.8-0.9), and pyometra outstandingly (AUC > 0.9) (Table 3). Table 3 . The ability of CRP, GlycA, albumin, the combination of GlycA and albumin, and the final multivariable metabolite model to correctly classify dogs as healthy or having one of the studied diseases. The classification is based on logistic regression and the classification ability is expressed as the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Generally, an AUC over 0.7 is considered acceptable, over 0.8 is considered excellent, and over 0.9 is considered outstanding. Final: The final multivariable metabolite model that included six metabolic measures: albumin, GlycA, pyruvate, alanine, glucose, and the ratio of phenylalanine to tyrosine. AD: atopic dermatitis, AGE: acute gastroenteritis, CIRDC: canine infectious respiratory disease complex, OA: osteoarthritis, OT: noncutaneous neoplasia, Paro: parodontitis, Pyo: pyometra, ST: cutaneous neoplasia, UTI: urinary tract infection. *: significant difference (p < 0.05) in AUC compared to CRP. CRP AUC (CI) GlycA AUC (CI) Albumin AUC (CI) GlycA + Albumin AUC (CI) Final AUC (CI) AD 0.58 (0.53-0.63) 0.54 (0.42-0.66) 0.67 (0.56-0.79) 0.69 (0.59-0.80)* 0.68 (0.58-0.78)* AGE 0.71 (0.63-0.79) 0.53 (0.41-0.64)* 0.75 (0.64-0.85) 0.76 (0.66-0.87) 0.86 (0.79-0.93)* CIRDC 0.84 (0.73-0.94) 0.69 (0.56-0.82) 0.62 (0.48-0.76)* 0.69 (0.55-0.83) 0.85 (0.75-0.95) OA 0.46 (0.40-0.52) 0.78 (0.66-0.90)* 0.63 (0.51-0.74)* 0.82 (0.73-0.92)* 0.75 (0.63-0.87)* OT 0.58 (0.53-0.64) 0.73 (0.63-0.83)* 0.70 (0.61-0.80)* 0.86 (0.79-0.92)* 0.89 (0.84-0.94)* Paro 0.60 (0.52-0.67) 0.65 (0.55-0.76) 0.62 (0.52-0.72) 0.72 (0.62-0.82) 0.84 (0.76-0.91)* Pyo 0.93 (0.84-1) 0.86 (0.70-1) 0.83 (0.72-0.93) 0.94 (0.85-1) 0.98 (0.94-1) ST 0.54 (0.48-0.61) 0.61 (0.47-0.76) 0.68 (0.57-0.78)* 0.78 (0.69-0.87)* 0.82 (0.73-0.91)* UTI 0.70 (0.60-0.80) 0.69 (0.55-0.83) 0.64 (0.52-0.77) 0.73 (0.59-0.87) 0.82 (0.71-0.93) The final multivariable metabolite model comprised albumin, GlycA, pyruvate, alanine, glucose, and the phenylalanine-to-tyrosine ratio. The model demonstrated acceptable performance (AUC 0.7–0.8) in osteoarthritis, excellent performance (AUC 0.8–0.9) in acute gastroenteritis, canine infectious respiratory disease complex, cutaneous neoplasia, noncutaneous neoplasia, periodontitis, and urinary tract infections, and outstanding performance (AUC > 0.9) in pyometra (Figure 3). The model significantly (p < 0.05) outperformed CRP in detecting all chronic conditions as well as acute gastroenteritis. Chronic inflammatory markers GlycA and/or albumin generally showed significantly better detection of chronic conditions compared to CRP (p < 0.05). Although the differences in AUC between the multivariable metabolite model and CRP did not achieve statistical significance in the acute conditions pyometra, urinary tract infection, and canine infectious respiratory disease complex, the multivariable metabolite model consistently yielded higher AUC values than CRP across all studied conditions. Additionally, the multivariable metabolite model displayed higher AUC values compared to albumin or GlycA individually in most conditions, except in atopic dermatitis and osteoarthritis, where the combination of GlycA and albumin slightly surpassed the performance of the metabolite model. GlycA, CRP, and albumin: reference interval-based result interpretation While the models classified most diseases well, the levels of both CRP and the single NMR metabolic measures, including GlycA and albumin, often remained within their reference intervals (Figure 4). Pyometra caused the most distinct changes in all inflammatory markers. Discussion This study, involving a large clinical cohort of over 380 dogs across multiple clinically relevant inflammatory conditions, yielded several significant findings: (i) both acute and chronic inflammatory diseases exhibit metabolic changes; (ii) systemic inflammation is confirmed in conditions previously considered locally inflammatory, such as osteoarthritis; (iii) each inflammatory condition displays a distinct metabolic profile; and (iv) a newly developed multi-metabolite model provides superior diagnostic performance compared to CRP, especially in chronic diseases. Universal metabolic alterations linked to inflammation can be clinically useful for detecting and evaluating inflammatory states. The multi-metabolite model developed here identified inflammation more effectively than CRP across all studied diseases, particularly chronic conditions. Levels of CRP and individual metabolites included in the model frequently remained within reference intervals, indicating that multi-marker models may better identify subtle disease-related changes compared to reference interval-based diagnostics. Validation of this model in real-world settings is recommended as a next step. GlycA and albumin were confirmed as universal markers of inflammation. GlycA aggregates signals from various acute-phase proteins 5 and reliably detects both acute and chronic inflammation in humans 1,6–8 and dogs 9–13 . Albumin, a negative inflammatory marker, decreases during inflammatory episodes. Both markers were generally superior to CRP in detecting chronic inflammatory states. Osteoarthritis, traditionally considered locally inflammatory, exhibited metabolic changes indicative of systemic low-grade inflammation 30 . These systemic alterations, undetectable by routine diagnostic methods, may impact disease management. Similarly, acute gastroenteritis showed metabolic similarities with chronic enteropathy 31 , such as elevated phenylalanine and reduced glycine and fatty acids, though differences were not noted in citrate and acetate levels. Atopic dermatitis displayed minimal metabolic changes, and CRP did not effectively detect the disease. Anti-inflammatory treatments potentially masked systemic inflammation. While the medication status within the last 24 hours did not significantly affect metabolite concentrations, several dogs had ongoing long-term treatments not administered within 24 hours. Furthermore, a systemic inflammatory response has only been observed in human patients with moderate to severe atopic dermatitis, whereas mild disease causes local responses 32 . While the severity of the disease was not documented in this study, dogs without any prior treatment probably had mild disease. Therefore, it could be possible that dogs with long-term treatment would have low systemic inflammation due to treatments, and dogs without treatment would have low systemic inflammation due to low disease severity. Further investigation into systemic inflammation and disease severity in canine atopic dermatitis is warranted. Reduced omega-3 PUFA levels were noted in periodontal disease, pyometra, cutaneous neoplasia, and urinary tract infections, with pyometra also showing increased saturated fatty acids and decreased monounsaturated oleic acid. In humans, decreased PUFA and increased monounsaturated and saturated fatty acids are among the most potent predictors of present disease and disease risk in many diseases 1 . Supplementation with omega-3 fatty acids is popular in dogs and is considered to have anti-inflammatory and immunomodulatory properties 33 . However, the association of serum fatty acid concentrations and diseases has been scarcely studied in dogs. Given the potential anti-inflammatory benefits of omega-3 supplementation, further exploration of fatty acid metabolism and disease susceptibility in dogs is justified. Elevated glucose levels in acute conditions, particularly pyometra, reflect stress-induced hyperglycemia. Elevated glucose observed in periodontal disease, alongside increases in pyruvate, suggests alterations in cellular energy metabolism. The changes in citrate and lactate did not go hand in hand with this change and showed only disease-specific changes. Lactate elevation correlated with disease severity and is clinically relevant for conditions like pyometra. Amino acid concentrations generally decreased, except for phenylalanine, a pattern previously linked to muscle catabolism and disease severity 35-38 . Similar changes have been previously reported in inflammatory and neoplastic diseases, and the magnitude of the amino acid changes has been linked with disease severity 35–38 . Given amino acids' therapeutic potential, supplementation merits further study 36,38 . Inflammation has been shown to increase VLDL and LDL levels and to decrease HDL levels in humans 39 , and HDL has also been suggested as a negative inflammatory marker in vector-borne diseases in dogs 40,41 . This study observed VLDL and LDL variables changes in diseases common in aging individuals, such as osteoarthritis, parodontitis, neoplastic diseases, and pyometra. While the changes in VLDL and LDL might be truly inflammation-related, these groups might also have more overweight/obese individuals, contributing to the increased concentration of these lipoproteins 42 . Small and large HDL variables were the lipoprotein variables most uniformly associated with inflammatory diseases in this study, whereas the concentrations and compositions of very large HDL particles remained unaltered. This information adds to our understanding of HDL metabolism during inflammation and its properties as a possible inflammatory marker. Limitations of this study include that there were differences in sex and neutering status between the individual disease groups and the control group that could not be taken into account in statistical analyses. However, the studied metabolites are only minimally affected by sex and neutering status 21 . All groups consisted of dogs of a large variety of breeds, and while breed affects metabolite concentrations 21 , no single breed or breed type dominated any of the groups. Therefore, the findings of this study cannot be explained by only group differences in these demographical factors. While multiple dogs had concurrent diseases and medications given during the last 24 hours, these medications or concurrent diseases did not cause the disease-specific metabolic changes observed in this study. This finding suggests that diseases generally affect metabolism more greatly than treatments. The negligible effects of comorbidities in this study probably stem from the fact that most comorbidities were less severe than the studied disease. A limitation of the CRP analysis was that the lower detection limit varied between clinics. The discriminatory capacity of CRP might have been better if the detection limit had been 1.0 mg/l in all clinics. It must be noted that since CRP was used as an exclusion criterion for the healthy control group, the discriminatory capacity of CRP might also be worse in a typical population of clinically healthy dogs. Overall, this study provides valuable insights into the metabolic effects of several common canine diseases. The newly identified metabolic profiles offer enhanced diagnostic capabilities, particularly in chronic inflammatory conditions, laying the groundwork for improved diagnostic methods for chronic low-grade inflammation in veterinary medicine. Declarations Acknowledgements Dr. Jenni Puurunen is thanked for contributing to the study design. Heta Forsström, Nightingale Health Ltd. is providing the practical needs of the study. Heli Julkunen, Nightingale Health Ltd. is thanked for statistical advice and for revising the statistical methods. Laura Parikka and Camilla Salmi, Evidensia Eläinlääkäripalvelut Ltd. are thanked for patient recruitment and collection of study data. Evidensia Eläinlääkäripalvelut Ltd. is thanked for enabling sample collection in their veterinary clinics and for scientific collaboration. The owners of the recruited dogs are thanked for the possibility of utilizing their pet dogs to advance research and dog health. The veterinarians of Eläinsairaala Evidensia Tammisto, Espoon Eläinsairaala, Eläinsairaala Mevet, and Lahden Eläinlääkäriasema are thanked for their help in patient recruitment and diagnostics. The laboratory personnel of these clinics are thanked for running in-house laboratory analyses and handling the samples intended for NMR metabolomics. PetMeta Labs Ltd. is thanked for enabling this study. Conflicts of interest The study was funded by PetMeta Labs Ltd. CO is an employee, and HL the board director and shareholder of PetMeta Labs Ltd. This company provides metabolomics testing for dogs. Author contribution Claudia Ottka: Conceptualization and study design. Conduction of statistical analyses. Interpreting the results and drafting the manuscript. Mirja Kaimio: Supervising and practical planning of the study in Evidensia premises. Commenting and approving the manuscript. Essi Leminen: Inclusion of study participants, collecting samples and data for the study. Commenting and approving the manuscript. Hannes Lohi: Conceptualization and supervision of the study. Editing the manuscript. References Julkunen H, Cichońska A, Tiainen M, Koskela H, Nybo K, Mäkelä V, et al. Atlas of plasma NMR biomarkers for health and disease in 118,461 individuals from the UK Biobank. Nat Commun. 2023 Feb 3;14(1):604. doi: 10.1038/s41467-023-36231-7. PMID: 36737450; PMCID: PMC9898515. Tabas I, Glass CK. Anti-inflammatory therapy in chronic disease: challenges and opportunities. Science. 2013 Jan;339(6116):166–72. Ceron JJ, Eckersall PD, Martýnez-Subiela S. Acute phase proteins in dogs and cats: current knowledge and future perspectives. Vet Clin Pathol. 2005 Jun;34(2):85–99. Nakamura M, Takahashi M, Ohno K, Koshino A, Nakashima K, Setoguchi A, et al. C-reactive protein concentration in dogs with various diseases. J Vet Med Sci. 2008 Feb;70(2):127–31. Otvos JD, Shalaurova I, Wolak-Dinsmore J, Connelly MA, Mackey RH, Stein JH, et al. GlycA: A Composite Nuclear Magnetic Resonance Biomarker of Systemic Inflammation. Clin Chem. 2015 May;61(5):714–23. Connelly MA, Otvos JD, Shalaurova I, Playford MP, Mehta NN. GlycA, a novel biomarker of systemic inflammation and cardiovascular disease risk. J Transl Med. 2017;15(1):219. Fischer K, Kettunen J, Wurtz P, Haller T, Havulinna AS, Kangas AJ, et al. Biomarker profiling by nuclear magnetic resonance spectroscopy for the prediction of all-cause mortality: an observational study of 17,345 persons. PLoS Med. 2014 Feb;11(2):e1001606. Ritchie SC, Wurtz P, Nath AP, Abraham G, Havulinna AS, Fearnley LG, et al. The Biomarker GlycA Is Associated with Chronic Inflammation and Predicts Long-Term Risk of Severe Infection. Cell Syst. 2015 Oct;1(4):293–301. Imbery CA, Dieterle F, Ottka C, Weber C, Schlotterbeck G, Müller E, et al. Metabolomic serum abnormalities in dogs with hepatopathies. Sci Rep. 2022 Mar;12(1):5329. Imbery CA, Dieterle F, Ottka C, Weber C, Schlotterbeck G, Müller E, et al. Metabolomic Abnormalities in Serum from Untreated and Treated Dogs with Hyper- and Hypoadrenocorticism. Metabolites [Internet]. 2022;12(4). Available from: https://www.mdpi.com/2218-1989/12/4/339 Ottka C, Puurunen J, Müller E, Weber C, Klein R, Lohi H. Metabolic changes associated with two endocrine abnormalities in dogs: elevated fructosamine and low thyroxine. Metabolomics. 2022 Jul;18(8):58. Ottka C, Vapalahti K, Puurunen J, Vahtera L, Lohi H. A novel canine nuclear magnetic resonance spectroscopy-based metabolomics platform: Validation and sample handling. Vet Clin Pathol [Internet]. 2021;n/a(50):410–26. Available from: https://onlinelibrary.wiley.com/doi/abs/10.1111/vcp.12954 Ottka C, Weber C, Müller E, Lohi H. Serum NMR metabolomics uncovers multiple metabolic changes in phenobarbital-treated dogs. Metabolomics. 2021 Jun;17(6):54. Soininen P, Kangas AJ, Würtz P, Suna T, Ala-Korpela M. Quantitative serum nuclear magnetic resonance metabolomics in cardiovascular epidemiology and genetics. Circ Cardiovasc Genet. 2015 Feb;8(1):192–206. Würtz P, Kangas AJ, Soininen P, Lawlor DA, Davey Smith G, Ala-Korpela M. Quantitative Serum Nuclear Magnetic Resonance Metabolomics in Large-Scale Epidemiology: A Primer on -Omic Technologies. Am J Epidemiol [Internet]. 2017 May 10;186(9):1084–96. Available from: https://doi.org/10.1093/aje/kwx016 Soininen P, Kangas AJ, Wurtz P, Tukiainen T, Tynkkynen T, Laatikainen R, et al. High-throughput serum NMR metabonomics for cost-effective holistic studies on systemic metabolism. Analyst. 2009 Sep;134(9):1781–5. R Core Team. R: A language and environment for statistical computing. [Internet]. Vienna, Austria: R Foundation for Statistical Computing; 2023. Available from: https://www.r-project.org/ Stekhoven DJ. missForest: Nonparametric Missing Value Imputation using Random Forest. R package version 1.4. 2013. Wickham H, Francois R, Henry L, Müller K.de: A Grammar of Data Manipulation. R package version 1.0.1. 2020. Wickham H, Averick M, Bryan J, Chang W, McGowan L, François R, et al. Welcome to the Tidyverse. J Open Source Softw. 2019;4:1686. Puurunen J, Ottka C, Salonen M, Niskanen JE, Lohi H. Age, breed, sex and diet influence serum metabolite profiles of 2000 pet dogs. R Soc Open Sci [Internet]. 2022;9(2):211642. Available from: https://royalsocietypublishing.org/doi/abs/10.1098/rsos.211642 Ruth B. bubbleHeatmap: Produces “bubbleHeatmap” Plots for Visualising Metabolomics Data [Internet]. 2023. Available from: https://cran.r-project.org/package=bubbleHeatmap John F and Sanford W. An R Companion to Applied Regression [Internet]. Third. Sage, Thousand Oaks CA; 2019. Available from: https://socialsciences.mcmaster.ca/jfox/Books/Companion/ Robinson D, Hayes A, Couch S. broom: Convert Statistical Objects into Tidy Tibbles. R package version 0.7.0. 2020. LeDell E, Petersen M, van der Laan M. cvAUC: Cross-Validated Area Under the ROC Curve Confidence Intervals_. R package version 1.1.4. 2022. Robin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez J-C, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics [Internet]. 2011;12(1):77. Available from: https://doi.org/10.1186/1471-2105-12-77 Wickham H. forcats: Tools for Working with Categorical Variables (Factors)_. R package version 1.0.0 [Internet]. 2023. Available from: https://cran.r-project.org/package=forcats Kassambara A. ggpubr: “ggplot2” Based Publication Ready Plots. R package version 0.6.0. 2023. Inkscape Project. Inkscape [Internet]. 2020. Available from: https://inkscape.org Berenbaum F. Osteoarthritis as an inflammatory disease (osteoarthritis is not osteoarthrosis!). Osteoarthr Cartil. 2013 Jan;21(1):16–21. Walker HK, Boag AM, Ottka C, Lohi H, Handel I, Gow AG, et al. Serum metabolomic profiles in dogs with chronic enteropathy. J Vet Intern Med [Internet]. 2022;36(5):1752–9. Available from: https://onlinelibrary.wiley.com/doi/abs/10.1111/jvim.16419 He H, Del Duca E, Diaz A, Kim HJ, Gay-Mimbrera J, Zhang N, et al. Mild atopic dermatitis lacks systemic inflammation and shows reduced nonlesional skin abnormalities. J Allergy Clin Immunol. 2021 Apr;147(4):1369–80. Magalhães TR, Lourenço AL, Gregório H, Queiroga FL. Therapeutic Effect of EPA/DHA Supplementation in Neoplastic and Non-neoplastic Companion Animal Diseases: A Systematic Review. In Vivo. 2021;35(3):1419–36. Genco RJ, Graziani F, Hasturk H. Effects of periodontal disease on glycemic control, complications, and incidence of diabetes mellitus. Periodontol 2000. 2020 Jun;83(1):59–65. Wannemacher RWJ, Klainer AS, Dinterman RE, Beisel WR. The significance and mechanism of an increased serum phenylalanine-tyrosine ratio during infection. Am J Clin Nutr. 1976 Sep;29(9):997–1006. Chan DL, Rozanski EA, Freeman LM. Relationship among plasma amino acids, C-reactive protein, illness severity, and outcome in critically ill dogs. J Vet Intern Med. 2009;23(3):559–63. Azuma K, Osaki T, Tsuka T, Imagawa T, Minami S, Okamoto Y. Plasma free amino acid profiles of canine mammary gland tumors. J Vet Sci. 2012 Dec;13(4):433–6. Cruzat V, Macedo Rogero M, Noel Keane K, Curi R, Newsholme P. Glutamine: Metabolism and Immune Function, Supplementation and Clinical Translation. Nutrients. 2018 Nov;10(11). Khovidhunkit W, Kim M-S, Memon RA, Shigenaga JK, Moser AH, Feingold KR, et al. Effects of infection and inflammation on lipid and lipoprotein metabolism: mechanisms and consequences to the host. J Lipid Res. 2004 Jul;45(7):1169–96. Ibba F, Rossi G, Meazzi S, Giordano A, Paltrinieri S. Serum concentration of high density lipoproteins (HDLs) in leishmaniotic dogs. Res Vet Sci. 2015 Feb;98:89–91. Nieto CG, Barrera R, Habela MA, Navarrete I, Molina C, Jimenez A, et al. Changes in the plasma concentrations of lipids and lipoprotein fractions in dogs infected with Leishmania infantum. Vet Parasitol. 1992 Oct;44(3–4):175–82. Jeusette IC, Lhoest ET, Istasse LP, Diez MO. Influence of obesity on plasma lipid and lipoprotein concentrations in dogs. Am J Vet Res. 2005 Jan;66(1):81–6. Additional Declarations The authors declare potential competing interests as follows: The study was funded by PetMeta Labs Ltd. CO is an employee, and HL the board director and shareholder of PetMeta Labs Ltd. This company provides metabolomics testing for dogs. Supplementary Files Supplementarytable.xlsx Supplemental Table 1 Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-6343599","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":436277032,"identity":"97db47ac-c2cc-4c37-a209-a4c74d7df57b","order_by":0,"name":"Claudia Elo","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0003-4837-6309","institution":"Petmeta Labs Ltd","correspondingAuthor":true,"prefix":"","firstName":"Claudia","middleName":"","lastName":"Elo","suffix":""},{"id":436277033,"identity":"41734512-af06-4b9d-9fb1-01e8cc4be874","order_by":1,"name":"Mirja Kaimio","email":"","orcid":"https://orcid.org/0000-0003-1189-7557","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Mirja","middleName":"","lastName":"Kaimio","suffix":""},{"id":436277034,"identity":"f3cedc35-3a70-4dbf-a319-59e641b2aa86","order_by":2,"name":"Essi Leminen","email":"","orcid":"","institution":"","correspondingAuthor":false,"prefix":"","firstName":"Essi","middleName":"","lastName":"Leminen","suffix":""},{"id":436277035,"identity":"f6243cf0-a2c3-4257-bd3d-e901a0f95f06","order_by":3,"name":"Hannes Lohi","email":"data:image/png;base64,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","orcid":"https://orcid.org/0000-0003-1087-5532","institution":"Petmeta Labs Ltd","correspondingAuthor":true,"prefix":"","firstName":"Hannes","middleName":"","lastName":"Lohi","suffix":""}],"badges":[],"createdAt":"2025-03-31 09:44:13","currentVersionCode":1,"declarations":{"humanSubjects":false,"vertebrateSubjects":true,"conflictsOfInterestStatement":true,"humanSubjectEthicalGuidelines":false,"humanSubjectConsent":false,"humanSubjectClinicalTrial":false,"humanSubjectCaseReport":false,"vertebrateSubjectEthicalGuidelines":true},"doi":"10.21203/rs.3.rs-6343599/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-6343599/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":79748028,"identity":"0cf4566d-4479-491a-91fe-8a1c5aa22473","added_by":"auto","created_at":"2025-04-02 09:04:28","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":442261,"visible":true,"origin":"","legend":"\u003cp\u003eMetabolites associated with overall disease burden. The case group includes all dogs suffering from one or more of the studied diseases. negLog10p: the absolute value of the log10 transformed logistic regression p-value adjusted for age, MAD from median: the difference of the group median from the median of the full dataset as median absolute deviations of the full dataset. FA: fatty acids, PUFA: polyunsaturated fatty acids, SFA: saturated fatty acids, PalA: palmitic acid, SteA: stearic acid, OleA: oleic acid, LA: linoleic acid, AA: arachidonic acid, DPA: docosapentaenoic acid, DHA: docosahexaenoic acid, %: relative concentration of fatty acid to total fatty acids, C: cholesterol, GlycA: glycoprotein acetyls, Ala: alanine, Gln: glutamine, Gly: glycine, His: histidine, BCAA: branched-chain amino acids, Ile: isoleucine, Leu: leucine, Val: valine, Phe: phenylalanine, Tyr: tyrosine, TG: triglycerides, VLDL: very low-density lipoprotein, LDL: low-density lipoprotein, HDL: high-density lipoprotein, L: lipids, P: particles, PL: phospholipids, CE: esterified cholesterol, FC: free cholesterol, XL-: very large, L-: large, S-: small.\u003c/p\u003e","description":"","filename":"floatimage1.png","url":"https://assets-eu.researchsquare.com/files/rs-6343599/v1/cb2021cc3fcc2b181dd4fb4d.png"},{"id":79748029,"identity":"f4365221-46de-464b-844b-bb491e725670","added_by":"auto","created_at":"2025-04-02 09:04:28","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":539700,"visible":true,"origin":"","legend":"\u003cp\u003eMetabolite association with specific disease states. negLog10p: the absolute value of the log10 transformed logistic regression p-value adjusted for age, MAD from median: difference of the group median from the median of the control group as median absolute deviations of the control group. AD: atopic dermatitis, AGE: acute gastroenteritis, CIRDC: canine infectious respiratory disease complex, OA: osteoarthritis, OT: noncutaneous neoplasia, Paro: parodontitis, Pyo: pyometra, ST: cutaneous neoplasia, UTI: urinary tract infection, FA: fatty acids, PUFA: polyunsaturated fatty acids, SFA: saturated fatty acids, PalA: palmitic acid, SteA: stearic acid, OleA: oleic acid, LA: linoleic acid, AA: arachidonic acid, DPA: docosapentaenoic acid, DHA: docosahexaenoic acid, %: relative concentration of fatty acid to total fatty acids, HDL: high-density lipoprotein, P: particles, L: lipids, PL: phospholipids, C: cholesterol, CE: esterified cholesterol, FC: free cholesterol, TG: triglycerides, XL-: very large, L-: large, S-: small, Ala: alanine, Gln: glutamine, Gly: glycine, His: histidine, BCAA: branched-chain amino acids, Ile: isoleucine, Leu: leucine, Val: valine, Phe: phenylalanine, Tyr: tyrosine, GlycA: glycoprotein acetyls, LDL: low-density lipoprotein, VLDL: very low-density lipoprotein.\u003c/p\u003e","description":"","filename":"floatimage2.png","url":"https://assets-eu.researchsquare.com/files/rs-6343599/v1/1b3c87ec2c566d3e0ede613a.png"},{"id":79748034,"identity":"d4c6a2aa-7eb8-48af-b7a7-8ee52e4a88f2","added_by":"auto","created_at":"2025-04-02 09:04:28","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":191682,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves and AUC of the metabolite-based multivariable logistic regression model by disease. The final multivariable metabolite model included six metabolic measures: albumin, GlycA, pyruvate, alanine, glucose, and the ratio of phenylalanine to tyrosine. Generally, an AUC over 0.7 is considered acceptable, over 0.8 is considered excellent, and over 0.9 is considered outstanding. Pyo: pyometra, OT: noncutaneous neoplasia, AGE: acute gastroenteritis, CIRDC: canine infectious respiratory disease complex, ST: cutaneous neoplasia, UTI: urinary tract infection, Paro: parodontitis, OA: osteoarthritis, AD: atopic dermatitis.\u003c/p\u003e","description":"","filename":"floatimage3.png","url":"https://assets-eu.researchsquare.com/files/rs-6343599/v1/067b5c4a273999eff8985b1b.png"},{"id":79748036,"identity":"ffde33cd-ec99-4302-aac0-b9a60b1fb721","added_by":"auto","created_at":"2025-04-02 09:04:28","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":157619,"visible":true,"origin":"","legend":"\u003cp\u003eBox plots of Albumin, GlycA, and CRP concentrations in all individual studied conditions. The red lines indicate the reference intervals, CRP reference intervals stem from the used IDEXX CRP analysis, and albumin and GlycA reference intervals are those of the NMR method \u003csup\u003e12\u003c/sup\u003e. = log10 of p-value under 2, * log10p 2-4, ** log10p4-6, *** log10p 6-8, **** log10p 8-10. AD: atopic dermatitis, AGE: acute gastroenteritis, CIRDC: canine infectious respiratory disease complex, OA: osteoarthritis, OT: noncutaneous neoplasia, Paro: parodontitis, Pyo: pyometra, ST: cutaneous neoplasia, UTI: urinary tract infection, Ctrl: control group.\u003c/p\u003e","description":"","filename":"floatimage4.png","url":"https://assets-eu.researchsquare.com/files/rs-6343599/v1/c6d74b7d341931307073ef9d.png"},{"id":79749732,"identity":"e8836d43-af9b-484e-ac62-4aed375677db","added_by":"auto","created_at":"2025-04-02 09:12:34","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":1904194,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-6343599/v1/7f660476-7ff8-4fde-b345-9cb7e660bf80.pdf"},{"id":79749727,"identity":"cca64e02-6204-452e-9a72-17e79d83965f","added_by":"auto","created_at":"2025-04-02 09:12:28","extension":"xlsx","order_by":1,"title":"","display":"","copyAsset":false,"role":"supplement","size":208272,"visible":true,"origin":"","legend":"\u003cp\u003eSupplemental Table 1\u003c/p\u003e","description":"","filename":"Supplementarytable.xlsx","url":"https://assets-eu.researchsquare.com/files/rs-6343599/v1/fc70d42893f735905a7cc02c.xlsx"}],"financialInterests":"The authors declare potential competing interests as follows: The study was funded by PetMeta Labs Ltd. CO is an employee, and HL the board director and shareholder of PetMeta Labs Ltd. This company provides metabolomics testing for dogs. ","formattedTitle":"\u003cp\u003e\u003cstrong\u003ePlasma NMR metabolomics reveals a powerful multi-marker signature in canine inflammatory conditions\u003c/strong\u003e\u003c/p\u003e","fulltext":[{"header":"INTRODUCTION","content":"\u003cp\u003eInflammation is a common biological response observed in numerous diseases, aiming to restore homeostasis during and after tissue insult. Beyond conditions typically recognized as inflammatory\u0026mdash;such as infectious or autoimmune diseases\u0026mdash;many acute and chronic illnesses feature a component of systemic low-grade inflammation \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003e. Chronic diseases often induce systemic inflammation; however, the connection between low-grade systemic inflammation and disease extends beyond this causal relationship. In humans, low-grade systemic inflammation can precede clinical disease onset by as much as a decade, and it contributes to various conditions ranging from diabetes mellitus to Alzheimer's disease \u003csup\u003e\u003cspan citationid=\"CR2\" class=\"CitationRef\"\u003e2\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eIdentifying an inflammatory state is crucial in clinical diagnostics. Inflammatory markers not only confirm the presence of inflammation but also help determine its severity, prognostic implications, and associated morbidity and mortality risks \u003csup\u003e\u003cspan citationid=\"CR3\" class=\"CitationRef\"\u003e3\u003c/span\u003e\u003c/sup\u003e. While practical biomarkers such as C-reactive protein (CRP) effectively identify acute inflammation, there is a lack of validated biomarkers for chronic low-grade inflammation in dogs \u003csup\u003e\u003cspan citationid=\"CR4\" class=\"CitationRef\"\u003e4\u003c/span\u003e\u003c/sup\u003e. Establishing reliable markers for chronic inflammation in dogs could facilitate earlier detection of chronic diseases, improving overall canine health and longevity.\u003c/p\u003e \u003cp\u003eGlycoprotein acetyls (GlycA) represent a promising biomarker for chronic low-grade inflammation. GlycA combines proton nuclear magnetic resonance (1H NMR) spectroscopy signals from acute-phase proteins including α1-acid glycoprotein, haptoglobin, α1-antitrypsin, α1-antichymotrypsin, and transferrin, with minor contributions from glycosylated apolipoproteins \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. Extensively studied in humans, GlycA is associated with chronic inflammation, chronic diseases, and increased morbidity and mortality risk \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e,\u003cspan additionalcitationids=\"CR7\" citationid=\"CR6\" class=\"CitationRef\"\u003e6\u003c/span\u003e\u0026ndash;\u003cspan citationid=\"CR8\" class=\"CitationRef\"\u003e8\u003c/span\u003e\u003c/sup\u003e. Its advantages for detecting chronic inflammation include stability and consistent responsiveness to inflammatory stimuli \u003csup\u003e\u003cspan citationid=\"CR5\" class=\"CitationRef\"\u003e5\u003c/span\u003e\u003c/sup\u003e. Elevated GlycA levels have also been observed in dogs across multiple conditions analogous to human diseases, including liver shunts and hepatopathies \u003csup\u003e\u003cspan citationid=\"CR9\" class=\"CitationRef\"\u003e9\u003c/span\u003e\u003c/sup\u003e, Addison's and Cushing's diseases \u003csup\u003e\u003cspan citationid=\"CR10\" class=\"CitationRef\"\u003e10\u003c/span\u003e\u003c/sup\u003e, diabetes mellitus and hypothyroidism \u003csup\u003e\u003cspan citationid=\"CR11\" class=\"CitationRef\"\u003e11\u003c/span\u003e\u003c/sup\u003e, lipemia \u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e, and during phenobarbital treatment \u003csup\u003e\u003cspan citationid=\"CR13\" class=\"CitationRef\"\u003e13\u003c/span\u003e\u003c/sup\u003e.\u003c/p\u003e \u003cp\u003eInflammation also influences metabolism, with metabolic changes potentially signaling inflammatory states. Detecting these metabolic alterations may not only enhance the identification of inflammatory conditions but also offer therapeutic insights. Recently, a validated \u003csup\u003e\u003cspan citationid=\"CR1\" class=\"CitationRef\"\u003e1\u003c/span\u003e\u003c/sup\u003eH NMR-based metabolomics platform for canine plasma and serum samples has been established \u003csup\u003e\u003cspan citationid=\"CR12\" class=\"CitationRef\"\u003e12\u003c/span\u003e\u003c/sup\u003e, enabling comprehensive evaluation of GlycA and over 120 additional metabolic biomarkers in canine inflammatory diseases.\u003c/p\u003e \u003cp\u003eThe objectives of this study were to: (i) identify universal metabolic alterations associated with several diseases commonly observed in veterinary practice; (ii) develop a metabolite-based multivariable model for detecting inflammation in dogs; (iii) characterize specific metabolic changes induced by each studied condition; and (iv) compare the discriminatory performance of GlycA, albumin, their combination, and the newly developed multivariable metabolic model with that of CRP in identifying inflammation in dogs.\u003c/p\u003e"},{"header":"MATERIALS AND METHODS","content":"\u003cp\u003e\u003cstrong\u003eEthical approval\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;The blood sample collection from privately owned pet dogs is approved in the project license from the Animal Ethical Committee of the County Administrative Board for Southern Finland (ESAVI/16933/2021). Whenever feasible, blood samples were collected from the same needle puncture used for routine laboratory diagnostics performed for the benefit of the patient. Dog owners provided informed consent prior to inclusion in the study and retained the right to withdraw at any point. Clinical data for each dog were collected in a pseudonymized format, and no personal owner information was gathered during the study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eStudy population\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWe conducted an observational, cross-sectional study involving dogs recruited during routine patient visits at four Finnish small animal veterinary hospitals: El\u0026auml;insairaala Evidensia Tammisto (Vantaa), El\u0026auml;insairaala Mevet (Helsinki), Lahden El\u0026auml;inl\u0026auml;\u0026auml;k\u0026auml;riasema (Lahti), and Espoon El\u0026auml;insairaala (Espoo). The recruitment period spanned from May 2022 to April 2023. A total of 209 healthy control dogs and 226 dogs diagnosed with predefined common canine diseases participated in the study.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe investigated conditions included acute gastroenteritis, pyometra, cutaneous and noncutaneous neoplasia, atopic dermatitis, periodontal disease, urinary tract infection, canine infectious respiratory disease complex, and osteoarthritis. Dogs could present with multiple concurrent conditions, and both treatment-na\u0026iuml;ve and previously medicated patients were included due to many dogs already undergoing treatment at enrollment. Diagnoses were determined based on patient history, clinical signs, findings, and appropriate diagnostic tests tailored to each patient\u0026apos;s condition. For neoplastic conditions, dogs initially enrolled based on clinical suspicion had their diagnoses confirmed through subsequent histological examination.\u003c/p\u003e\n\u003cp\u003eControl group inclusion required a physical examination performed within the past three months, a medical history free of indications suggestive of illness, and clinical chemistry, hematology, and C-reactive protein (CRP) levels within normal ranges. In-house clinical chemistry analysis utilized the Catalyst Dx Chem15 analyzer, and hematology was assessed using ProCyte (both from IDEXX Laboratories, Inc., Maine, USA).\u003c/p\u003e\n\u003cp\u003eFor all participating dogs, detailed information on signalment, health status, medication usage, and fasting duration prior to blood sampling was collected. CRP was measured using an in-house Catalyst Dx CRP analyzer (IDEXX Laboratories, Inc., Maine, USA), with a reference interval of 0\u0026ndash;10.0 mg/L. However, the analyzer detection limits varied among clinics, ranging from lower detection limits of either 1.0 mg/L or 10.0 mg/L, to upper limits of 170 mg/L or 240 mg/L.\u003c/p\u003e\n\u003cp\u003eFor metabolomic analysis, 2 ml of blood were drawn via cephalic venipuncture into Vacuette LH Lithium Heparin PREMIUM tubes (2 ml). Plasma volumes ranging from 110 to 500 \u0026micro;l were separated according to the manufacturer\u0026apos;s recommendations and subsequently transferred into cryo tubes (Sarstedt 72.730.005). Samples were refrigerated at the veterinary clinics for up to one week, then transported within three hours under cool-pack conditions. After transportation, samples were stored at -80\u0026deg;C for a maximum of five months prior to metabolomic analysis performed by PetMeta Labs Ltd. (Helsinki, Finland) using proton nuclear magnetic resonance (\u003csup\u003e1\u003c/sup\u003eH NMR) spectroscopy.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003e\u003csup\u003e1\u003c/sup\u003e\u003c/strong\u003e\u003cstrong\u003eH NMR spectroscopy\u003cbr\u003e\u003c/strong\u003eA targeted \u003csup\u003e1\u003c/sup\u003eH NMR spectroscopy method optimized and validated for dogs was utilized \u003csup\u003e12\u003c/sup\u003e. The method has been thoroughly described elsewhere \u003csup\u003e12,14,15\u003c/sup\u003e. The method is highly automated, from sample processing to the processing of the NMR spectra. The equipment used includes a sample changer SampleJet (Bruker Corp., Billerica, Massachusetts, USA), a PerkinElmer JANUS Automated Workstation with an 8-tip dispense arm with Varispan (PerkinElmer Inc., Waltham, Massachusetts, USA) and a Bruker AVANCE III HD 500 NMR spectrometer with a 5 mm triple-channel (1H, 13C, 15N) z-gradient Prodigy probe head \u003csup\u003e14\u003c/sup\u003e. Sample preparation included light mixing of the sample and removing the possible precipitate by centrifugation, sample transfer to individual NMR tubes, and mixing with sodium phosphate buffer \u003csup\u003e14,16\u003c/sup\u003e. Metabolite quantification in absolute units is achieved by processing the NMR spectra using scripts optimized for canine samples \u003csup\u003e12\u003c/sup\u003e based on regression modeling \u003csup\u003e15\u003c/sup\u003e. The proprietary software utilized has integrated quality control \u003csup\u003e15\u003c/sup\u003e. The method quantitates 123 measurands, including a detailed lipoprotein analysis, amino acids, fatty acids, triglycerides, cholesterols, glycolysis-related metabolites, albumin, creatinine, and GlycA \u003csup\u003e12\u003c/sup\u003e. \u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eData pre-processing\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eLaboratory and health data of all participating dogs underwent thorough review and evaluation. Dogs initially assigned to the control group were excluded if their CRP levels, clinical chemistry, hematology results, or general health information indicated either subclinical or clinical disease. Control group dogs displaying laboratory parameters outside reference ranges without clear disease indications were flagged for potential exclusion from subsequent analyses if metabolomics data identified them as outliers. Similarly, the disease status of case-group dogs was meticulously reassessed, resulting in exclusion of individuals with uncertain or incorrect diagnoses, or, specifically within neoplasia cases, those lacking histological confirmation.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Metabolomics data were subjected to rigorous quality control. Dogs with extensive missing metabolite data were excluded. Individual metabolites were assessed for completeness, with those exhibiting greater than 20% missing observations removed. Specifically, metabolites such as very large very low-density lipoprotein phospholipids (XL-VLDL PL), esterified cholesterol (XL-VLDL CE), and free cholesterol (XL-VLDL FC)\u0026mdash;known for characteristically low canine levels\u0026mdash;were excluded. Additionally, metabolomics data were screened for outliers. When control group outliers were identified, their health and laboratory data were reevaluated. Dogs exhibiting significant deviations, including previously noted laboratory abnormalities unrelated to clear disease or moderate alterations in muscle mass or body condition, were excluded. Following these quality assurance steps, the final dataset comprised 175 control dogs and 207 case dogs.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;Remaining missing metabolite values were imputed using Random Forest imputation, employing the R packages missForest version 1.5 \u003csup\u003e18\u003c/sup\u003e, dplyr version 1.1.2 \u003csup\u003e19\u003c/sup\u003e and tidyverse \u003csup\u003e20\u003c/sup\u003e. The imputation\u0026apos;s out-of-bag (OOB) error was 0.136, indicating adequate performance. A separate approach was applied for CRP value imputation: values below the detection limit were imputed as half the detection limit, while values exceeding the detection limit were imputed as detection limit plus 10 mg/L. One pyometra group dog lacking CRP measurements was excluded from analyses involving CRP. Statistical analyses and data preprocessing were conducted using the R programming environment \u003csup\u003e17\u003c/sup\u003e (R Core Team, 2023) and Microsoft Office Excel (Microsoft Corporation, Redmond, WA, USA).\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eStatistical analyses\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDescriptive statistics were computed to characterize the study groups. The normality of metabolite variables was assessed using the Shapiro-Wilk test (stats package, R Core Team, 2023) \u003csup\u003e17\u003c/sup\u003e, indicating most metabolites were not normally distributed. P-value adjustments for metabolite analyses were conducted using Bonferroni correction based on the number of principal components (PCs) explaining \u0026gt;95% of data variance. Principal component analysis (PCA) was performed using FactoMineR and factoextra packages, with five PCs explaining 95% variance, resulting in a p-value threshold of 0.01 (0.05/5).\u003c/p\u003e\n\u003cp\u003eInitially, general metabolic changes associated with disease were evaluated by comparing the combined case group (dogs diagnosed with one or more diseases) with the healthy control group. Before this comparison, physiological characteristics (sex, neutering status, fasting duration, age) between the groups were examined using the Wilcoxon rank-sum test (stats package) \u003csup\u003e17\u003c/sup\u003e, with significant differences identified only for age. Given age\u0026apos;s known influence on metabolite concentrations \u003csup\u003e21\u003c/sup\u003e, age was included as a covariate in subsequent analyses. Additionally, the potential effect of medications administered within 24 hours prior to sampling on metabolite levels in the diseased group was assessed, revealing no significant differences.\u003c/p\u003e\n\u003cp\u003eMetabolite concentrations between the case and control groups were compared using logistic regression (stats package) \u003csup\u003e17\u003c/sup\u003e, adjusted for age. The case-control status served as the dependent variable, and results were visualized using bubble plots created with bubbleHeatmap (version 0.1.1) \u003csup\u003e22\u003c/sup\u003e and grid \u003csup\u003e17\u003c/sup\u003e packages. The bubble plots reflected logistic regression p-values and median absolute deviation (MAD) differences from the dataset median. To confirm result robustness, the Wilcoxon rank-sum test (stats package) \u003csup\u003e17\u003c/sup\u003e was also performed, yielding highly similar results.\u003c/p\u003e\n\u003cp\u003eA multivariable logistic regression model to identify inflammatory markers was developed using metabolites significant (p\u0026lt;0.05) in univariable logistic regressions. The model, adjusted for age, was constructed through forward stepwise selection based on p-values using the stats package \u003csup\u003e17\u003c/sup\u003e. Variables were sequentially added until no further metabolites reached statistical significance (p\u0026lt;0.05). Akaike information criterion (AIC)-based selection yielded identical metabolites at each step.\u003c/p\u003e\n\u003cp\u003eThe multivariable model underwent diagnostic assessment to ensure logistic regression assumptions were met. Multicollinearity was evaluated using variance inflation factors (VIF; car package \u003csup\u003e23\u003c/sup\u003e), revealing no multicollinearity (VIF range: 1.07\u0026ndash;1.58). Influential observations were examined through standardized residual plots (broom package \u003csup\u003e24\u003c/sup\u003e), showing no influential outliers (standardized residuals \u0026lt;2.6). Linearity of the logit assumption was tested by adding interaction terms between each variable and its natural logarithm; no significant interaction terms were found (p\u0026gt;0.05). Variable importance was assessed using the caret package. Model performance was evaluated using 10-fold cross-validation (cvAUC package version 1.1.4) \u003csup\u003e25\u003c/sup\u003e, with the area under the receiver operating characteristic curve (AUC) as the performance metric. Standard AUC interpretations were applied: \u0026gt;0.7 acceptable, \u0026gt;0.8 excellent, and \u0026gt;0.9 outstanding.\u003c/p\u003e\n\u003cp\u003eNext, metabolic changes specific to each disease were investigated individually. Dogs with concurrent conditions were included in analyses for each relevant disease. Physiological characteristics were compared using the Wilcoxon rank-sum test (stats package) \u003csup\u003e17\u003c/sup\u003e, and results were summarized in Table 1. While some groups exhibited differences in sex or neutering status compared to controls, group sizes limited statistical adjustments. Considering sex and neutering status minimally impact metabolite concentrations \u003csup\u003e21\u003c/sup\u003e, only age was included as a covariate in logistic regression models. Effects of concurrent diseases and recent medications on metabolite concentrations were evaluated, identifying no significant effects, except for higher docosapentaenoic acid percentage in acute gastroenteritis dogs with concurrent illnesses.\u003c/p\u003e\n\u003cp\u003eDisease-specific metabolite comparisons between the control group and each disease group were conducted using logistic regression (stats package) \u003csup\u003e17\u003c/sup\u003e, adjusting for age. Bubble plots visualizing logistic regression p-values and MAD differences from control medians were created using bubbleHeatmap (version 0.1.1) \u003csup\u003e22\u003c/sup\u003e and grid \u003csup\u003e17\u003c/sup\u003e packages. Wilcoxon rank-sum tests (stats package) \u003csup\u003e17\u003c/sup\u003e validated logistic regression findings, with only minor interpretation differences.\u003c/p\u003e\n\u003cp\u003eFinally, the diagnostic performance of CRP, GlycA, albumin, the GlycA-albumin combination, and the multivariable metabolite model was compared using logistic regression (stats package) \u003csup\u003e17\u003c/sup\u003e. Performance was assessed and visualized with ROC curves (pROC package) \u003csup\u003e26\u003c/sup\u003e. AUC values were calculated, and differences in AUC between CRP and metabolic measures were statistically compared using the DeLong method. The dependent variable for each logistic regression model was disease presence versus control.\u003c/p\u003e\n\u003cp\u003eDifferences in CRP and metabolite concentrations across groups were illustrated through box plots using forcats (version 1.0.0) \u003csup\u003e27\u003c/sup\u003e and ggpubr (version 0.6.0) \u003csup\u003e28\u003c/sup\u003e packages, incorporating reference intervals for clarity. Final graphical adjustments were completed using Inkscape \u003csup\u003e29\u003c/sup\u003e.\u003c/p\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cstrong\u003eGroup demography\u003c/strong\u003e\u003cbr\u003e\u0026nbsp;Demographic characteristics of the study groups are presented in Table 1. Dogs with pyometra, neoplastic diseases, periodontitis, and arthritis were significantly older compared to the control group, while those with canine infectious respiratory disease complex were notably younger. All study groups included diverse breeds and mixed-breed dogs. Among noncutaneous neoplasias, the most frequently observed were mammary tumors (n = 9), lymphoma (n = 8), leukemia (n = 3), and mast cell tumors (n = 3). The most common cutaneous neoplasias were cutaneous mast cell tumors (n = 9) and lipomas (n = 7). The prevalence of concurrent diseases varied among disease groups, ranging from 19% in dogs with atopic dermatitis to 52% in those with osteoarthritis. Administration of medication within 24 hours prior to sample collection was common, occurring in 51% of case group dogs. These medications included treatments specific to the diagnosed conditions as well as sedatives administered when blood sampling coincided with procedures requiring anesthesia, such as tumor removal or dental treatments.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e1\u003c/strong\u003e. Demographical information of the study groups. *significant difference from the control group. %M/F: Percentage of males and females, %neut: percentage of neutered dogs, %drug24h: percentage of dogs that had received medication other than routine antiparasitic drugs within the last 24 hours prior to blood sampling, %concurrent dis: percentage of dogs having one or more concurrent diseases, All: all case group dogs with one or more of the studied diseases, AD: atopic dermatitis, AGE: acute gastroenteritis, CIRDC: canine infectious respiratory disease complex, OA: osteoarthritis, OT: noncutaneous neoplasia, Paro: parodontitis, Pyo: pyometra, ST: cutaneous neoplasia, UTI: urinary tract infection.\u0026nbsp;\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"623\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 56px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eGroup\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eDisease\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eN\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge median (range)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e%M/F\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e%neut\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e\u003cstrong\u003eN breeds\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e%drug24h\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u003cstrong\u003e%concurrent dis\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 56px;\"\u003e\n \u003cp\u003eControl\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e175\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e5.6 (1.0 - 16.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e36/64\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e64\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd rowspan=\"10\" valign=\"top\" style=\"width: 56px;\"\u003e\n \u003cp\u003eCase\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eAll\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e207\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e8 (0.3 - 16.5)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e44/56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e90\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e51\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e30\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eAD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e4.5 (1.6 \u0026ndash; 11.4)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e61/39*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e20\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e59\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eAGE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e29\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e5.6 (0.6 \u0026ndash; 12.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e34/66\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e24\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eCIRDC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e2.0 (0.4 \u0026ndash; 7.4)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e50/50\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e6*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e38\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eOA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e21\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e9.9 (3.3 \u0026ndash; 16.5)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e48/52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e71*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e15\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e62\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e52\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eOT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e41\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e10.3 (3.5 \u0026ndash; 14.4)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e44/56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e16\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e56\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e39\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eParo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e33\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e9.5 (3.3 \u0026ndash; 16.5)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e55/45*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e45\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e25\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e48\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e30\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003ePyo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e8.4 (4.9 \u0026ndash; 11.3)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e0/100*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e0*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e13\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e42\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e42\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eST\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e27\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e9.2 (2.4 \u0026ndash; 16.1)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e48/52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e52\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e44\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e44\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eUTI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 38px;\"\u003e\n \u003cp\u003e19\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 123px;\"\u003e\n \u003cp\u003e7.8 (0.3 \u0026ndash; 14.7)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e37/63\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e37\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 66px;\"\u003e\n \u003cp\u003e14\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 76px;\"\u003e\n \u003cp\u003e32\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 85px;\"\u003e\n \u003cp\u003e26\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cstrong\u003eMetabolite association with overall disease occurrence\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eSeveral metabolites showed significant associations with overall disease occurrence when comparing the healthy control group to the combined case group, which included all dogs diagnosed with one or more of the studied conditions (Figure 1, Supplementary Tables O1 and O8). Among the strongest associations were the inflammatory markers GlycA and albumin. GlycA levels were elevated in the case group, while albumin, a negative acute-phase protein, was decreased. Notable alterations in amino acids included a significant decrease in tyrosine and glycine and an increase in the phenylalanine-to-tyrosine ratio. In metabolites related to sugar metabolism, pyruvate showed the most pronounced elevation in the case group. Total cholesterol, triglycerides, and very low-density lipoprotein (VLDL) levels did not differ significantly between groups, and changes in high-density lipoprotein (HDL) and low-density lipoprotein (LDL) cholesterol were minor. However, large and small HDL particle numbers (L-HDL and S-HDL), total lipids, cholesterol, and esterified cholesterol concentrations were significantly reduced in the case group, whereas large LDL (L-LDL) variables were notably increased, except for triglyceride levels.\u003c/p\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eDetection of overall disease occurrence using the metabolic profile\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe forward stepwise selection process is shown\u0026nbsp;in Supplementary Table O1-O7. The final multivariable logistic regression model included six metabolic measures: albumin, GlycA, pyruvate, alanine, glucose, and the ratio of phenylalanine to tyrosine. While the model was adjusted for age, age was not a significant predictor of disease in the model (Table 2 and Supplementary Table 1).\u0026nbsp;The average AUC of the 10-fold cross-validated model was 0.82 (95% CI: 0.78 - 0.86), indicating excellent performance.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e2\u003c/strong\u003e. Importance of each variable included in the final multivariable metabolite model.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003ePredictor\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eVariable importance\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eAlbumin\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003e100\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eGlycA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003e91.54\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003ePyruvate\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003e84.99\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eAla\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003e59.01\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eGlucose\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003e53.94\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003ePhe/Tyr\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003e39.86\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003eage\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 150px;\"\u003e\n \u003cp\u003e0.00\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u0026nbsp;\u003cstrong\u003eMetabolite associations with specific diseases\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWhile some changes in metabolite concentrations were universal across different diseases, multiple diseases also showed disease-specific changes (Figure 2, Supplementary Table D1). Universal changes across different conditions included decreased albumin and tyrosine and increased GlycA and pyruvate. An increase in phenylalanine and the phenylalanine/tyrosine ratio was mainly observed in acute diseases. Changes in other amino acids were relatively disease-specific, while all changes except phenylalanine were decreases in concentration. Multiple very low-density lipoprotein (VLDL) lipid variables and total triglycerides were increased in dogs with osteoarthritis, parodontitis, neoplastic diseases, and pyometra while being decreased in dogs with atopic dermatitis, acute gastroenteritis, and canine infectious respiratory disease complex. HDL and LDL variables showed different signatures. LDL variables, especially L-LDL variables, were increased in dogs with osteoarthritis, pyometra, and noncutaneous neoplasia with no or little changes in other diseases. Total HDL and total cholesterol remained unaltered in all diseases. Still, small and large HDL variables were relatively uniformly decreased across multiple conditions, including atopic dermatitis, acute gastroenteritis, canine infectious respiratory disease complex, noncutaneous neoplasia, and pyometra. In fatty acid concentrations, a decrease in oleic acid, docosapentaenoic acid, and omega-3 fatty acids and an increase in saturated fatty acids were observed in pyometra. Decreased omega-3 polyunsaturated fatty acid (PUFA) concentrations were observed in parodontitis, pyometra, cutaneous neoplasia, and urinary tract infections. Citrate was reduced in atopic dermatitis, acute gastroenteritis, and canine infectious respiratory disease complex, and acetate followed a similar pattern. Glucose was elevated in parodontitis and pyometra and slightly in the other studied acute conditions. \u0026nbsp;Generally, dogs with atopic dermatitis had the fewest and slightest metabolic changes.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003ePerformance of CRP, GlycA, albumin, the combination of GlycA and albumin, and the multivariable metabolite model in classifying dogs as healthy and diseased\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe acute inflammatory marker CRP could not acceptably (AUC \u0026lt; 0.7) detect individuals with the studied chronic inflammatory conditions atopic dermatitis, osteoarthritis, cutaneous neoplasia, noncutaneous neoplasia, and parodontitis. In contrast, the acute conditions of acute gastroenteritis and urinary tract infection were acceptably (AUC 0.7-0.8) detected by CRP, canine infectious respiratory disease complex excellently (AUC 0.8-0.9), and pyometra outstandingly (AUC \u0026gt; 0.9) (Table 3).\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eTable\u0026nbsp;\u003c/strong\u003e\u003cstrong\u003e3\u003c/strong\u003e. The ability of CRP, GlycA, albumin, the combination of GlycA and albumin, and the final multivariable metabolite model to correctly classify dogs as healthy or having one of the studied diseases. The classification is based on logistic regression and the classification ability is expressed as the area under the curve (AUC) of the receiver operating characteristic (ROC) curve. Generally, an AUC over 0.7 is considered acceptable, over 0.8 is considered excellent, and over 0.9 is considered outstanding. Final: The final multivariable metabolite model that included six metabolic measures: albumin, GlycA, pyruvate, alanine, glucose, and the ratio of phenylalanine to tyrosine. AD: atopic dermatitis, AGE: acute gastroenteritis, CIRDC: canine infectious respiratory disease complex, OA: osteoarthritis, OT: noncutaneous neoplasia, Paro: parodontitis, Pyo: pyometra, ST: cutaneous neoplasia, UTI: urinary tract infection. *: significant difference (p \u0026lt; 0.05) in AUC compared to CRP.\u003c/p\u003e\n\u003ctable border=\"1\" cellspacing=\"0\" cellpadding=\"0\" width=\"624\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003e\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eCRP AUC (CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eGlycA AUC (CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eAlbumin AUC (CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eGlycA + Albumin AUC (CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003eFinal AUC (CI)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eAD\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.58 (0.53-0.63)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.54 (0.42-0.66)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.67 (0.56-0.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.69 (0.59-0.80)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.68 (0.58-0.78)*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eAGE\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.71 (0.63-0.79)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.53 (0.41-0.64)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.75 (0.64-0.85)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.76 (0.66-0.87)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.86 (0.79-0.93)*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eCIRDC\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.84 (0.73-0.94)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.69 (0.56-0.82)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.62 (0.48-0.76)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.69 (0.55-0.83)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.85 (0.75-0.95)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eOA\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.46 (0.40-0.52)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.78 (0.66-0.90)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.63 (0.51-0.74)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.82 (0.73-0.92)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.75 (0.63-0.87)*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eOT\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.58 (0.53-0.64)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.73 (0.63-0.83)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.70 (0.61-0.80)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.86 (0.79-0.92)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.89 (0.84-0.94)*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eParo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.60 (0.52-0.67)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.65 (0.55-0.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.62 (0.52-0.72)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.72 (0.62-0.82)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.84 (0.76-0.91)*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003ePyo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.93 (0.84-1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.86 (0.70-1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.83 (0.72-0.93)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.94 (0.85-1)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.98 (0.94-1)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eST\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.54 (0.48-0.61)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.61 (0.47-0.76)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.68 (0.57-0.78)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.78 (0.69-0.87)*\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.82 (0.73-0.91)*\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd valign=\"top\" style=\"width: 57px;\"\u003e\n \u003cp\u003eUTI\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.70 (0.60-0.80)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.69 (0.55-0.83)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.64 (0.52-0.77)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.73 (0.59-0.87)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd valign=\"top\" style=\"width: 113px;\"\u003e\n \u003cp\u003e0.82 (0.71-0.93)\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eThe final multivariable metabolite model comprised albumin, GlycA, pyruvate, alanine, glucose, and the phenylalanine-to-tyrosine ratio. The model demonstrated acceptable performance (AUC 0.7\u0026ndash;0.8) in osteoarthritis, excellent performance (AUC 0.8\u0026ndash;0.9) in acute gastroenteritis, canine infectious respiratory disease complex, cutaneous neoplasia, noncutaneous neoplasia, periodontitis, and urinary tract infections, and outstanding performance (AUC \u0026gt; 0.9) in pyometra (Figure 3). The model significantly (p \u0026lt; 0.05) outperformed CRP in detecting all chronic conditions as well as acute gastroenteritis. Chronic inflammatory markers GlycA and/or albumin generally showed significantly better detection of chronic conditions compared to CRP (p \u0026lt; 0.05). Although the differences in AUC between the multivariable metabolite model and CRP did not achieve statistical significance in the acute conditions pyometra, urinary tract infection, and canine infectious respiratory disease complex, the multivariable metabolite model consistently yielded higher AUC values than CRP across all studied conditions. Additionally, the multivariable metabolite model displayed higher AUC values compared to albumin or GlycA individually in most conditions, except in atopic dermatitis and osteoarthritis, where the combination of GlycA and albumin slightly surpassed the performance of the metabolite model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eGlycA, CRP, and albumin: reference interval-based result interpretation\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eWhile the models classified most diseases well, the levels of both CRP and the single NMR metabolic measures, including GlycA and albumin, often remained within their reference intervals (Figure 4). Pyometra caused the most distinct changes in all inflammatory markers.\u0026nbsp;\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study, involving a large clinical cohort of over 380 dogs across multiple clinically relevant inflammatory conditions, yielded several significant findings: (i) both acute and chronic inflammatory diseases exhibit metabolic changes; (ii) systemic inflammation is confirmed in conditions previously considered locally inflammatory, such as osteoarthritis; (iii) each inflammatory condition displays a distinct metabolic profile; and (iv) a newly developed multi-metabolite model provides superior diagnostic performance compared to CRP, especially in chronic diseases.\u003c/p\u003e\n\u003cp\u003eUniversal metabolic alterations linked to inflammation can be clinically useful for detecting and evaluating inflammatory states. The multi-metabolite model developed here identified inflammation more effectively than CRP across all studied diseases, particularly chronic conditions. Levels of CRP and individual metabolites included in the model frequently remained within reference intervals, indicating that multi-marker models may better identify subtle disease-related changes compared to reference interval-based diagnostics. Validation of this model in real-world settings is recommended as a next step.\u003c/p\u003e\n\u003cp\u003eGlycA and albumin were confirmed as universal markers of inflammation. GlycA aggregates signals from various acute-phase proteins \u003csup\u003e5\u003c/sup\u003e and reliably detects both acute and chronic inflammation in humans \u003csup\u003e1,6\u0026ndash;8\u003c/sup\u003e\u0026nbsp; and dogs \u003csup\u003e9\u0026ndash;13\u003c/sup\u003e. Albumin, a negative inflammatory marker, decreases during inflammatory episodes. Both markers were generally superior to CRP in detecting chronic inflammatory states.\u003c/p\u003e\n\u003cp\u003eOsteoarthritis, traditionally considered locally inflammatory, exhibited metabolic changes indicative of systemic low-grade inflammation \u003csup\u003e30\u003c/sup\u003e. These systemic alterations, undetectable by routine diagnostic methods, may impact disease management. Similarly, acute gastroenteritis showed metabolic similarities with chronic enteropathy \u003csup\u003e31\u003c/sup\u003e, such as elevated phenylalanine and reduced glycine and fatty acids, though differences were not noted in citrate and acetate levels.\u003c/p\u003e\n\u003cp\u003eAtopic dermatitis displayed minimal metabolic changes, and CRP did not effectively detect the disease. Anti-inflammatory treatments potentially masked systemic inflammation. While the medication status within the last 24 hours did not significantly affect metabolite concentrations, several dogs had ongoing long-term treatments not administered within 24 hours. Furthermore, a systemic inflammatory response has only been observed in human patients with moderate to severe atopic dermatitis, whereas mild disease causes local responses \u003csup\u003e32\u003c/sup\u003e. While the severity of the disease was not documented in this study, dogs without any prior treatment probably had mild disease. Therefore, it could be possible that dogs with long-term treatment would have low systemic inflammation due to treatments, and dogs without treatment would have low systemic inflammation due to low disease severity. Further investigation into systemic inflammation and disease severity in canine atopic dermatitis is warranted.\u003c/p\u003e\n\u003cp\u003eReduced omega-3 PUFA levels were noted in periodontal disease, pyometra, cutaneous neoplasia, and urinary tract infections, with pyometra also showing increased saturated fatty acids and decreased monounsaturated oleic acid. In humans, decreased PUFA and increased monounsaturated and saturated fatty acids are among the most potent predictors of present disease and disease risk in many diseases \u003csup\u003e1\u003c/sup\u003e. Supplementation with omega-3 fatty acids is popular in dogs and is considered to have anti-inflammatory and immunomodulatory properties \u003csup\u003e33\u003c/sup\u003e. However, the association of serum fatty acid concentrations and diseases has been scarcely studied in dogs. Given the potential anti-inflammatory benefits of omega-3 supplementation, further exploration of fatty acid metabolism and disease susceptibility in dogs is justified.\u003c/p\u003e\n\u003cp\u003eElevated glucose levels in acute conditions, particularly pyometra, reflect stress-induced hyperglycemia. Elevated glucose observed in periodontal disease, alongside increases in pyruvate, suggests alterations in cellular energy metabolism. The changes in citrate and lactate did not go hand in hand with this change and showed only disease-specific changes. Lactate elevation correlated with disease severity and is clinically relevant for conditions like pyometra.\u003c/p\u003e\n\u003cp\u003eAmino acid concentrations generally decreased, except for phenylalanine, a pattern previously linked to muscle catabolism and disease severity \u003csup\u003e35-38\u003c/sup\u003e. Similar changes have been previously reported in inflammatory and neoplastic diseases, and the magnitude of the amino acid changes has been linked with disease severity \u003csup\u003e35\u0026ndash;38\u003c/sup\u003e.\u0026nbsp;Given amino acids\u0026apos; therapeutic potential, supplementation merits further study\u0026nbsp;\u003csup\u003e36,38\u003c/sup\u003e.\u003c/p\u003e\n\u003cp\u003eInflammation has been shown to increase VLDL and LDL levels and to decrease HDL levels in humans \u003csup\u003e39\u003c/sup\u003e, and HDL has also been suggested as a negative inflammatory marker in vector-borne diseases in dogs \u003csup\u003e40,41\u003c/sup\u003e. This study observed VLDL and LDL variables changes in diseases common in aging individuals, such as osteoarthritis, parodontitis, neoplastic diseases, and pyometra. While the changes in VLDL and LDL might be truly inflammation-related, these groups might also have more overweight/obese individuals, contributing to the increased concentration of these lipoproteins \u003csup\u003e42\u003c/sup\u003e. Small and large HDL variables were the lipoprotein variables most uniformly associated with inflammatory diseases in this study, whereas the concentrations and compositions of very large HDL particles remained unaltered. This information adds to our understanding of HDL metabolism during inflammation and its properties as a possible inflammatory marker.\u003c/p\u003e\n\u003cp\u003eLimitations of this study include that there were differences in sex and neutering status between the individual disease groups and the control group that could not be taken into account in statistical analyses. However, the studied metabolites are only minimally affected by sex and neutering status \u003csup\u003e21\u003c/sup\u003e. All groups consisted of dogs of a large variety of breeds, and while breed affects metabolite concentrations \u003csup\u003e21\u003c/sup\u003e, no single breed or breed type dominated any of the groups. Therefore, the findings of this study cannot be explained by only group differences in these demographical factors. While multiple dogs had concurrent diseases and medications given during the last 24 hours, these medications or concurrent diseases did not cause the disease-specific metabolic changes observed in this study. This finding suggests that diseases generally affect metabolism more greatly than treatments. The negligible effects of comorbidities in this study probably stem from the fact that most comorbidities were less severe than the studied disease. A limitation of the CRP analysis was that the lower detection limit varied between clinics. The discriminatory capacity of CRP might have been better if the detection limit had been 1.0 mg/l in all clinics. It must be noted that since CRP was used as an exclusion criterion for the healthy control group, the discriminatory capacity of CRP might also be worse in a typical population of clinically healthy dogs.\u003c/p\u003e\n\u003cp\u003eOverall, this study provides valuable insights into the metabolic effects of several common canine diseases. The newly identified metabolic profiles offer enhanced diagnostic capabilities, particularly in chronic inflammatory conditions, laying the groundwork for improved diagnostic methods for chronic low-grade inflammation in veterinary medicine.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003e\u003cstrong\u003eAcknowledgements\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eDr. Jenni Puurunen is thanked for contributing to the study design. Heta Forsstr\u0026ouml;m, Nightingale Health Ltd. is providing the practical needs of the study. Heli Julkunen, Nightingale Health Ltd. is thanked for statistical advice and for revising the statistical methods. Laura Parikka and Camilla Salmi, Evidensia El\u0026auml;inl\u0026auml;\u0026auml;k\u0026auml;ripalvelut Ltd. are thanked for patient recruitment and collection of study data. Evidensia El\u0026auml;inl\u0026auml;\u0026auml;k\u0026auml;ripalvelut Ltd. is thanked for enabling sample collection in their veterinary clinics and for scientific collaboration. The owners of the recruited dogs are thanked for the possibility of utilizing their pet dogs to advance research and dog health. The veterinarians of El\u0026auml;insairaala Evidensia Tammisto, Espoon El\u0026auml;insairaala, El\u0026auml;insairaala Mevet, and Lahden El\u0026auml;inl\u0026auml;\u0026auml;k\u0026auml;riasema are thanked for their help in patient recruitment and diagnostics. The laboratory personnel of these clinics are thanked for running in-house laboratory analyses and handling the samples intended for NMR metabolomics. PetMeta Labs Ltd. is thanked for enabling this study.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConflicts of interest\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eThe study was funded by PetMeta Labs Ltd. CO is an employee, and HL the board director and shareholder of PetMeta Labs Ltd. This company provides metabolomics testing for dogs.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eAuthor contribution\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eClaudia Ottka: Conceptualization and study design. Conduction of statistical analyses. Interpreting the results and drafting the manuscript.\u003c/p\u003e\n\u003cp\u003eMirja Kaimio: Supervising and practical planning of the study in Evidensia premises.\u0026nbsp;Commenting and approving the manuscript.\u003c/p\u003e\n\u003cp\u003eEssi Leminen: Inclusion of study participants, collecting samples and data for the study.\u0026nbsp;Commenting and approving the manuscript.\u003cbr\u003eHannes Lohi: Conceptualization and supervision of the study. Editing the manuscript.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n\u003cli\u003eJulkunen H, Cichońska A, Tiainen M, Koskela H, Nybo K, M\u0026auml;kel\u0026auml; V, et al. Atlas of plasma NMR biomarkers for health and disease in 118,461 individuals from the UK Biobank. Nat Commun. 2023 Feb 3;14(1):604. doi: 10.1038/s41467-023-36231-7. PMID: 36737450; PMCID: PMC9898515.\u003c/li\u003e\n\u003cli\u003eTabas I, Glass CK. Anti-inflammatory therapy in chronic disease: challenges and opportunities. Science. 2013 Jan;339(6116):166\u0026ndash;72. \u003c/li\u003e\n\u003cli\u003eCeron JJ, Eckersall PD, Mart\u0026yacute;nez-Subiela S. Acute phase proteins in dogs and cats: current knowledge and future perspectives. Vet Clin Pathol. 2005 Jun;34(2):85\u0026ndash;99. \u003c/li\u003e\n\u003cli\u003eNakamura M, Takahashi M, Ohno K, Koshino A, Nakashima K, Setoguchi A, et al. C-reactive protein concentration in dogs with various diseases. J Vet Med Sci. 2008 Feb;70(2):127\u0026ndash;31. \u003c/li\u003e\n\u003cli\u003eOtvos JD, Shalaurova I, Wolak-Dinsmore J, Connelly MA, Mackey RH, Stein JH, et al. GlycA: A Composite Nuclear Magnetic Resonance Biomarker of Systemic Inflammation. Clin Chem. 2015 May;61(5):714\u0026ndash;23. \u003c/li\u003e\n\u003cli\u003eConnelly MA, Otvos JD, Shalaurova I, Playford MP, Mehta NN. GlycA, a novel biomarker of systemic inflammation and cardiovascular disease risk. J Transl Med. 2017;15(1):219. \u003c/li\u003e\n\u003cli\u003eFischer K, Kettunen J, Wurtz P, Haller T, Havulinna AS, Kangas AJ, et al. Biomarker profiling by nuclear magnetic resonance spectroscopy for the prediction of all-cause mortality: an observational study of 17,345 persons. PLoS Med. 2014 Feb;11(2):e1001606. \u003c/li\u003e\n\u003cli\u003eRitchie SC, Wurtz P, Nath AP, Abraham G, Havulinna AS, Fearnley LG, et al. The Biomarker GlycA Is Associated with Chronic Inflammation and Predicts Long-Term Risk of Severe Infection. Cell Syst. 2015 Oct;1(4):293\u0026ndash;301. \u003c/li\u003e\n\u003cli\u003eImbery CA, Dieterle F, Ottka C, Weber C, Schlotterbeck G, M\u0026uuml;ller E, et al. Metabolomic serum abnormalities in dogs with hepatopathies. Sci Rep. 2022 Mar;12(1):5329. \u003c/li\u003e\n\u003cli\u003eImbery CA, Dieterle F, Ottka C, Weber C, Schlotterbeck G, M\u0026uuml;ller E, et al. Metabolomic Abnormalities in Serum from Untreated and Treated Dogs with Hyper- and Hypoadrenocorticism. Metabolites [Internet]. 2022;12(4). Available from: https://www.mdpi.com/2218-1989/12/4/339\u003c/li\u003e\n\u003cli\u003eOttka C, Puurunen J, M\u0026uuml;ller E, Weber C, Klein R, Lohi H. Metabolic changes associated with two endocrine abnormalities in dogs: elevated fructosamine and low thyroxine. Metabolomics. 2022 Jul;18(8):58. \u003c/li\u003e\n\u003cli\u003eOttka C, Vapalahti K, Puurunen J, Vahtera L, Lohi H. A novel canine nuclear magnetic resonance spectroscopy-based metabolomics platform: Validation and sample handling. Vet Clin Pathol [Internet]. 2021;n/a(50):410\u0026ndash;26. Available from: https://onlinelibrary.wiley.com/doi/abs/10.1111/vcp.12954\u003c/li\u003e\n\u003cli\u003eOttka C, Weber C, M\u0026uuml;ller E, Lohi H. Serum NMR metabolomics uncovers multiple metabolic changes in phenobarbital-treated dogs. Metabolomics. 2021 Jun;17(6):54. \u003c/li\u003e\n\u003cli\u003eSoininen P, Kangas AJ, W\u0026uuml;rtz P, Suna T, Ala-Korpela M. Quantitative serum nuclear magnetic resonance metabolomics in cardiovascular epidemiology and genetics. Circ Cardiovasc Genet. 2015 Feb;8(1):192\u0026ndash;206. \u003c/li\u003e\n\u003cli\u003eW\u0026uuml;rtz P, Kangas AJ, Soininen P, Lawlor DA, Davey Smith G, Ala-Korpela M. Quantitative Serum Nuclear Magnetic Resonance Metabolomics in Large-Scale Epidemiology: A Primer on -Omic Technologies. Am J Epidemiol [Internet]. 2017 May 10;186(9):1084\u0026ndash;96. Available from: https://doi.org/10.1093/aje/kwx016\u003c/li\u003e\n\u003cli\u003eSoininen P, Kangas AJ, Wurtz P, Tukiainen T, Tynkkynen T, Laatikainen R, et al. High-throughput serum NMR metabonomics for cost-effective holistic studies on systemic metabolism. Analyst. 2009 Sep;134(9):1781\u0026ndash;5. \u003c/li\u003e\n\u003cli\u003eR Core Team. R: A language and environment for statistical computing. [Internet]. Vienna, Austria: R Foundation for Statistical Computing; 2023. Available from: https://www.r-project.org/\u003c/li\u003e\n\u003cli\u003eStekhoven DJ. missForest: Nonparametric Missing Value Imputation using Random Forest. R package version 1.4. 2013. \u003c/li\u003e\n\u003cli\u003eWickham H, Francois R, Henry L, M\u0026uuml;ller K.de: A Grammar of Data Manipulation. R package version 1.0.1. 2020. \u003c/li\u003e\n\u003cli\u003eWickham H, Averick M, Bryan J, Chang W, McGowan L, Fran\u0026ccedil;ois R, et al. Welcome to the Tidyverse. J Open Source Softw. 2019;4:1686. \u003c/li\u003e\n\u003cli\u003ePuurunen J, Ottka C, Salonen M, Niskanen JE, Lohi H. Age, breed, sex and diet influence serum metabolite profiles of 2000 pet dogs. R Soc Open Sci [Internet]. 2022;9(2):211642. Available from: https://royalsocietypublishing.org/doi/abs/10.1098/rsos.211642\u003c/li\u003e\n\u003cli\u003eRuth B. bubbleHeatmap: Produces \u0026ldquo;bubbleHeatmap\u0026rdquo; Plots for Visualising Metabolomics Data [Internet]. 2023. Available from: https://cran.r-project.org/package=bubbleHeatmap\u003c/li\u003e\n\u003cli\u003eJohn F and Sanford W. An R Companion to Applied Regression [Internet]. Third. Sage, Thousand Oaks CA; 2019. Available from: https://socialsciences.mcmaster.ca/jfox/Books/Companion/\u003c/li\u003e\n\u003cli\u003eRobinson D, Hayes A, Couch S. broom: Convert Statistical Objects into Tidy Tibbles. R package version 0.7.0. 2020. \u003c/li\u003e\n\u003cli\u003eLeDell E, Petersen M, van der Laan M. cvAUC: Cross-Validated Area Under the ROC Curve Confidence Intervals_. R package version 1.1.4. 2022. \u003c/li\u003e\n\u003cli\u003eRobin X, Turck N, Hainard A, Tiberti N, Lisacek F, Sanchez J-C, et al. pROC: an open-source package for R and S+ to analyze and compare ROC curves. BMC Bioinformatics [Internet]. 2011;12(1):77. Available from: https://doi.org/10.1186/1471-2105-12-77\u003c/li\u003e\n\u003cli\u003eWickham H. forcats: Tools for Working with Categorical Variables (Factors)_. R package version 1.0.0 [Internet]. 2023. Available from: https://cran.r-project.org/package=forcats\u003c/li\u003e\n\u003cli\u003eKassambara A. ggpubr: \u0026ldquo;ggplot2\u0026rdquo; Based Publication Ready Plots. R package version 0.6.0. 2023. \u003c/li\u003e\n\u003cli\u003eInkscape Project. Inkscape [Internet]. 2020. Available from: https://inkscape.org\u003c/li\u003e\n\u003cli\u003eBerenbaum F. Osteoarthritis as an inflammatory disease (osteoarthritis is not osteoarthrosis!). Osteoarthr Cartil. 2013 Jan;21(1):16\u0026ndash;21. \u003c/li\u003e\n\u003cli\u003eWalker HK, Boag AM, Ottka C, Lohi H, Handel I, Gow AG, et al. Serum metabolomic profiles in dogs with chronic enteropathy. J Vet Intern Med [Internet]. 2022;36(5):1752\u0026ndash;9. Available from: https://onlinelibrary.wiley.com/doi/abs/10.1111/jvim.16419\u003c/li\u003e\n\u003cli\u003eHe H, Del Duca E, Diaz A, Kim HJ, Gay-Mimbrera J, Zhang N, et al. Mild atopic dermatitis lacks systemic inflammation and shows reduced nonlesional skin abnormalities. J Allergy Clin Immunol. 2021 Apr;147(4):1369\u0026ndash;80. \u003c/li\u003e\n\u003cli\u003eMagalh\u0026atilde;es TR, Louren\u0026ccedil;o AL, Greg\u0026oacute;rio H, Queiroga FL. Therapeutic Effect of EPA/DHA Supplementation in Neoplastic and Non-neoplastic Companion Animal Diseases: A Systematic Review. In Vivo. 2021;35(3):1419\u0026ndash;36. \u003c/li\u003e\n\u003cli\u003eGenco RJ, Graziani F, Hasturk H. Effects of periodontal disease on glycemic control, complications, and incidence of diabetes mellitus. Periodontol 2000. 2020 Jun;83(1):59\u0026ndash;65. \u003c/li\u003e\n\u003cli\u003eWannemacher RWJ, Klainer AS, Dinterman RE, Beisel WR. The significance and mechanism of an increased serum phenylalanine-tyrosine ratio during infection. Am J Clin Nutr. 1976 Sep;29(9):997\u0026ndash;1006. \u003c/li\u003e\n\u003cli\u003eChan DL, Rozanski EA, Freeman LM. Relationship among plasma amino acids, C-reactive protein, illness severity, and outcome in critically ill dogs. J Vet Intern Med. 2009;23(3):559\u0026ndash;63. \u003c/li\u003e\n\u003cli\u003eAzuma K, Osaki T, Tsuka T, Imagawa T, Minami S, Okamoto Y. Plasma free amino acid profiles of canine mammary gland tumors. J Vet Sci. 2012 Dec;13(4):433\u0026ndash;6. \u003c/li\u003e\n\u003cli\u003eCruzat V, Macedo Rogero M, Noel Keane K, Curi R, Newsholme P. Glutamine: Metabolism and Immune Function, Supplementation and Clinical Translation. Nutrients. 2018 Nov;10(11). \u003c/li\u003e\n\u003cli\u003eKhovidhunkit W, Kim M-S, Memon RA, Shigenaga JK, Moser AH, Feingold KR, et al. Effects of infection and inflammation on lipid and lipoprotein metabolism: mechanisms and consequences to the host. J Lipid Res. 2004 Jul;45(7):1169\u0026ndash;96. \u003c/li\u003e\n\u003cli\u003eIbba F, Rossi G, Meazzi S, Giordano A, Paltrinieri S. Serum concentration of high density lipoproteins (HDLs) in leishmaniotic dogs. Res Vet Sci. 2015 Feb;98:89\u0026ndash;91. \u003c/li\u003e\n\u003cli\u003eNieto CG, Barrera R, Habela MA, Navarrete I, Molina C, Jimenez A, et al. Changes in the plasma concentrations of lipids and lipoprotein fractions in dogs infected with Leishmania infantum. Vet Parasitol. 1992 Oct;44(3\u0026ndash;4):175\u0026ndash;82. \u003c/li\u003e\n\u003cli\u003eJeusette IC, Lhoest ET, Istasse LP, Diez MO. Influence of obesity on plasma lipid and lipoprotein concentrations in dogs. Am J Vet Res. 2005 Jan;66(1):81\u0026ndash;6. \u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":true,"hideJournal":true,"highlight":"","institution":"Petmeta Labs Ltd","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"metabolomics, CRP, inflammation, GlycA","lastPublishedDoi":"10.21203/rs.3.rs-6343599/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-6343599/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003eBACKGROUND Identifying inflammatory states is crucial in veterinary diagnostic workups. While biomarkers for acute inflammation are widely utilized, there is a lack of reliable markers for chronic inflammation.\u003c/p\u003e \u003cp\u003eOBJECTIVES This study aims to characterize metabolic alterations associated with common inflammatory diseases in dogs and evaluate the efficacy of metabolic markers in identifying inflammatory states.\u003c/p\u003e \u003cp\u003eANIMALS Plasma samples from 175 healthy dogs and 207 dogs diagnosed with specific acute and chronic diseases were collected during veterinary visits. Conditions studied included acute gastroenteritis, pyometra, neoplastic diseases, atopic dermatitis, periodontitis, urinary tract infections, canine infectious respiratory disease complex, and osteoarthritis.\u003c/p\u003e \u003cp\u003eMETHODS Samples were analyzed using a canine-validated 1H NMR spectroscopy platform. Logistic regression was employed to identify both general and disease-specific metabolic alterations. A multivariable metabolite model was developed, and its diagnostic performance was compared against C-reactive protein (CRP), albumin, glycoprotein acetyls (GlycA), and a combination of GlycA and albumin.\u003c/p\u003e \u003cp\u003eRESULTS Metabolic changes were observed across all conditions studied, with some alterations specific to individual diseases and others common across conditions. The multivariable metabolite model demonstrated excellent overall diagnostic performance (AUC\u0026thinsp;=\u0026thinsp;0.82; 95% CI: 0.78\u0026ndash;0.86). Importantly, this model detected inflammation significantly better (p\u0026thinsp;\u0026lt;\u0026thinsp;0.05) than CRP in all chronic diseases (AUC range: 0.68\u0026ndash;0.89 vs. 0.46\u0026ndash;0.60) and acute gastroenteritis (AUC: 0.86 vs. 0.71). Furthermore, it consistently showed higher AUC values compared to CRP in all diseases analyzed.\u003c/p\u003e \u003cp\u003eCONCLUSIONS Metabolic profiling can effectively detect both acute and chronic inflammation in dogs. This approach appears superior to CRP, particularly for identifying chronic inflammatory conditions.\u003c/p\u003e","manuscriptTitle":"Plasma NMR metabolomics reveals a powerful multi-marker signature in canine inflammatory conditions","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2025-04-02 09:04:23","doi":"10.21203/rs.3.rs-6343599/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"6d001810-79b0-4217-b5d7-a1b9e1f36068","owner":[],"postedDate":"April 2nd, 2025","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[{"id":46446734,"name":"Animal Physiology"},{"id":46446735,"name":"Cellular Metabolism"}],"tags":[],"updatedAt":"2025-04-02T09:04:23+00:00","versionOfRecord":[],"versionCreatedAt":"2025-04-02 09:04:23","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-6343599","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-6343599","identity":"rs-6343599","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}