Abbreviations ANCOVA analysis of covariance ANOVA analysis of variance BCG body condition score group BCS body condition score ELISA enzyme linked immunosorbent assay IL interleukin NEFA nonesterified fatty acids SAA serum amyloid A TG triglycerides TNF tumor necrosis factor In humans, obesity‐associated low‐grade inflammation increases the risk for metabolic disorders such as cardiovascular disease and diabetes. Obesity in horses similarly is associated with increased risk for the painful hoof disease, laminitis, and a role for increased inflammation, analogous to that observed in humans, is proposed. Circulating proinflammatory cytokine concentrations, including tumor necrosis factor (TNF)‐α, interleukin (IL)‐6, and IL‐1β, are increased with obesity in several species, including humans, mice, cats, and dogs. In horses, circulating concentrations of TNF correlated with body condition score (BCS) in BCS 4‐9 mares, whereas associations between BCS and circulating IL‐6 and IL‐1β have not previously been determined. Obesity in horses also is associated with increased circulating blood insulin and lipid concentrations, and it is possible these factors could influence, or be influenced by, cytokines. As we have previously demonstrated, increased insulin concentrations may directly influence circulating TNF and IL‐6 concentrations in horses. Lipids also are known to alter cytokine expression profiles, increasing IL‐6 and TNF in a mouse cell line. Thus, we also were interested in determining relationships between insulin, lipids, and cytokines. Increased production of the acute phase protein serum amyloid A (SAA) is associated with obesity in mice and humans. Furthermore, concentrations of SAA are used to predict risk of cardiovascular disease in humans. In horses, SAA increases rapidly during acute inflammatory stimulation or after natural infection. Although a relationship between obesity and SAA has not been established previously, SAA mRNA is increased in hoof tissue from laminitic horses. This investigation was initiated to identify correlations between adiposity and the inflammatory proteins, TNF, IL‐6, IL‐1β, and SAA, in horses. Identifying cytokines or acute phase proteins that have strong relationships with adiposity will advance research in the field of equine obesity. Furthermore, an increased understanding of obesity–inflammation interrelationships could lead to the development of a tool for predicting obesity‐associated metabolic disorders in horses, similar to the use of plasma SAA and C‐reactive protein to predict specific diseases in humans. Materials and Methods Samples were selected for this study from a previous investigation of a larger population of horses. This study was approved by the Institutional Animal Care and Use Committee of Virginia Tech. Horse Selection and Sample Collection The 110 plasma samples used for this investigation were selected from an original cohort of 300 samples. The original cohort was collected from horses randomly selected from a population of adult horses in southwest Virginia during the summer (June–July) of 2006. Age, body condition, body weight, and breed were recorded, and a blood sample was collected between 0600 and 1200 hours. All horses were held off concentrate feed, but access to forage before sampling varied. Horses that were maintained on pasture or that had a continuous source of hay (ie, round bale hay) were allowed to continue consuming forage (n = 95). Horses without access to a continual source of forage were given hay 10 hours before sampling, but not offered any additional hay after that point to prevent rapid consumption of feed before blood collection (n = 15). Body condition scores were determined as the average score of 2 independent assessors rounded to the integer whereas body weight was estimated based on girth circumference, body length, and height at the withers. Whole blood was collected by jugular venipuncture into EDTA and sodium heparin‐coated tubes, centrifuged within 15 minutes of collection, and plasma frozen in 1.5 mL aliquots at −80°C until analysis. Plasma insulin, glucose, triglyceride (TG), nonesterified fatty acids (NEFA), and leptin concentrations were determined as previously described. Samples (n = 110) were selected for analysis of plasma TNF and IL‐6. These samples were selected from the original 300 samples using stratification by BCS and plasma insulin concentration to ensure a normal distribution of variables (Table ). Samples were first stratified by plasma insulin concentration into groups that consisted of ≤20 (n = 245), 20.1–30 (n = 25), 30.1–40 (n = 9), and ≥40.1 (n = 21) mIU/L. All of the 55 samples with insulin concentrations ≥20.1 mIU/L were included whereas 55 samples from the ≤20 mIU/L group were chosen after stratifying by BCS. From this secondary stratification, samples were randomly selected from each BCS level to maximize equal representation of BCS. From these 110 samples, 82 were randomly chosen for plasma SAA analysis and 98 were randomly chosen for plasma IL‐1β analysis using commercially available enzyme linked immunosorbent assays (ELISA). , Characteristics of horses used to investigate relationships between body condition and inflammation. Body Condition Score Group Variable 2–4 (n = 7) 5–6 (n = 33) 7 (n = 36) 8–9 (n = 34) Sex 5 m 10 m 16 m 24 m 1 g 21 g 18 g 10 g 1 s 2 s 2 s Age, years 4–20 4–18 5–20 4–20 11.9 ± 5.7 10.5 ± 4.7 10.9 ± 4.3 12.1 ± 4.8 BW, kg 327–565 411–695 417–708 342–785 474 ± 82 558 ± 76 538 ± 66 562 ± 96 Glucose, mg/dL 82–96 73–106 85–124 87–122 88.8 ± 4.7 91.7 ± 7.6 97.9 ± 8.4 98.8 ± 7.4 Insulin, mIU/L 1–21 1–50 3–108 5–108 7.6 ± 6.9 9.9 ± 11.1 24.5 ± 21.8 38.7 ± 26.7 TNF, pg/mL 11–14490 8–3848 15–7985 7–14001 (n = 7) (n = 31) (n = 40) (n = 27) 2191 ± 5426 660 ± 870 603 ± 1336 936 ± 2479 IL‐6, ng/mL 7–1427 2–618 4–944 6–1588 (n = 7) (n = 31) (n = 34) (n = 25) 295 ± 516 98 ± 129 152 ± 223 287 ± 413 IL‐1β, ng/mL 2–122 1–254 1–544 2–311 (n = 6) (n = 29) (n = 32) (n = 26) 47 ± 53 51 ± 70 63 ± 140 58 ± 84 SAA, ng/mL 27–678 4–821 14–3124 10–3845 (n = 2) (n = 24) (n = 31) (n = 22) 352 ± 461 180 ± 192 483 ± 741 600 ± 848 Values presented as ranges, the mean and standard deviation are included beneath the range. Horses grouped by body condition score, with number of horses in each group indicated. Total n = 105 out 110 samples originally assayed, values for 5 samples fell above the limit of detection of the assay. Total n = 97 out 110 samples originally assayed, values for 13 samples fell above the limit of detection of the assay. Total n = 93 out 105 samples originally assayed, values for 12 samples fell above the limit of detection of the assay. Total n = 79 out 82 samples originally assayed, values for 3 samples fell above the limit of detection of the assay. Abbreviations: BW, body weight; g, gelding; IL‐6, interleukin‐6; m, mare; SAA, serum amyloid A; s, stallion; TNF, tumor necrosis factor‐α. ELISA Validation Before sample analysis, dilutional parallelism and percent recovery were determined for the IL‐1β ELISA. Plasma samples from 3 healthy horses were diluted in parallel to the standard curve at dilutions of 0, 1 : 2, 1 : 4, 1 : 8, 1 : 16, 1 : 32, 1 : 64, 1 : 128, and 1 : 256, to determine dilutions within the range of the standard curve. Dilutions >1 : 16 were below the level of detection for the assay. Afterward, a sample of pooled plasma from healthy horses was used to reconstitute a vial of recombinant IL‐1β, and this spiked sample was assayed at dilutions of 1 : 4, 1 : 8, 1 : 16, and 1 : 32 to determine the percentage recovery of IL‐1β in equine plasma (Table S1). Statistical Analysis Simple correlations (r values) between variables (sex, age, BW, BCS, insulin, glucose, NEFA, and TG) and SAA, IL‐6, IL‐1β, and TNF were determined. Analysis of covariance (ANCOVA) was used to determine the relationship between BCS and insulin. For this analysis, animals were grouped into 3 BCS groups (BCG): normal (BCS 4‐6), overweight (BCS 7), and obese (BCS 8‐9). Horses with a BCS < 4 (n = 7) were excluded because there were inadequate observations for each sex and age category for statistical comparisons to horses in the normal, overweight, and obese BCS groups. The model included class variables of sex and BCG with insulin concentration and age included as covariates. Initially, interactions between the covariates and class variables were included, but were removed if not significant, indicating that the slope of the regression line was not different between levels of the class variable. In this instance, the slope of the regression line and differences between means of BCG were determined using orthogonal contrasts. If the interaction between a covariate and class variable was significant, indicating that the slope of the regression line was different for at least 1 level of the class variable, then slopes were determined for each level of the class variable and means comparisons between levels of the class variable were made at low, median, and high values of the covariate using Tukey tests. If the covariate of age was not significant then it was removed from the model. Where sex was significant, mare means were compared to the average of geldings and stallions using contrast statements. For all statistical analyses, TNF, IL‐6, IL‐1β, and SAA were log 10 transformed to improve normal distribution of residuals. Results Horses used for this study consisted of 55 mares, 50 geldings, and 5 stallions that ranged in age and body weight from 4 to 20 years and 327 to 785 kg (Table ). Ages were represented equally within each BCS whereas maximum body weights were higher in greater BCS horses. Breed representation consisted of mixed breed (n = 29), Quarter Horse (n = 20), Warmblood (n = 16) Thoroughbred (n = 11), Rocky Mountain Horse (n = 7), Tennessee Walking Horse, Arabian (both n = 6), Paint (n = 5), Appaloosa (n = 4), American Saddlebred (n = 2), Fjord, Andalusian, Morgan Horse, and Paso Fino (each n = 1). Serum TNF concentrations ranged from 7 to 14,490 pg/mL, excluding 5 samples that exceeded the limit of detection of the assay (final n = 105). Plasma IL‐6 concentrations ranged from 2 to 1,588 ng/mL, excluding 13 samples that exceeded the limit of detection of the assay (final n = 97). Plasma IL‐1β concentrations ranged from 1 to 544 ng/mL, excluding 12 samples that exceeded the limit of detection of the assay (final n = 93). Plasma SAA concentrations ranged from 4 to 3,845 ng/mL, excluding 3 samples that exceeded the limit of detection of the assay (final n = 79). As shown in Table , TNF was correlated with sex (r = −0.26, P = .011), with higher concentrations in mares. Tumor necrosis factor‐α also was negatively correlated with TG (r = −0.22, P = .031; data not shown). Interleukin‐6 was positively correlated with age (r = 0.32, P = .001) and correlated with sex (r = −0.24, P = .021), with higher concentrations in mares. Serum amyloid A was positively correlated with BCS (r = 0.27, P = .007) and insulin (r = 0.37, P < .001). Both IL‐1β (r = 0.19, P = .081) and SAA (r = 0.21, P = .066) were correlated, but not significantly, with leptin (data not shown). Significant ( P < .05) simple correlations between age, body condition score (BCS), sex, insulin, and inflammatory cytokines in horses Correlation Variables Cytokines Age BCS Sex Insulin TNF −0.26 IL‐6 0.32 −0.24 SAA 0.27 0.37 IL‐1β Analysis of covariance was used to determine relationships between the categorical variables, body condition and sex, and the continuous variables, plasma insulin concentration and age, for each of the inflammatory proteins measured. The plasma insulin by BCG interaction was not significant for any variable, indicating that the regression lines for insulin were parallel for each BCG, and so the interaction was removed from the models. For plasma TNF, neither age nor the BCG by age interaction was significant and thus both terms were removed from the model. In the next model, the sex by insulin interaction was not significant ( P > .09) and was dropped from the model. In the final model, there was no difference in TNF concentrations between BCG ( P > .4); however, there was a main effect of sex ( P = .009), with females having higher values than males ( P = .002; Fig ). Geometric mean (± 95% confidence intervals) plasma TNF concentrations in light‐breed mares (n = 46), geldings (n = 48), and stallions (n = 4). For plasma IL‐6, the insulin by sex interaction was not significant and was removed from the original model. In the final model, the slope of the regression lines for both covariates, insulin (slope = −0.007, P = .028) and age (slope = 0.047, P = .001) were significant; however, a main effect of either sex ( P > .2) or BCG ( P > .1) was not observed. This indicates that there was a negative relationship between insulin and IL‐6 (Fig a) and a positive relationship between age and IL‐6 (Fig b), neither of which was different among the 3 levels of BCG. Relationships between log 10 plasma IL ‐6 concentrations and (A) plasma insulin concentration or (B) age in horses grouped by body condition, when sex was included as a covariate. Slopes apply to all body condition score groups. Where body condition score normal = 5–6, overweight = 7, and obese = 8–9. For plasma IL‐1β and SAA, neither the insulin by sex interaction nor age was significant. The final model for IL‐1β included BCG, insulin, and sex, with no variable being significantly related to IL‐6 ( P > .5). For plasma SAA, the regression slope of insulin was significant (slope = 0.009, P = .007) (Fig c); however, there was no main effect of either BCG ( P > .3) or sex ( P > .2), indicating that although SAA and insulin were positively associated, there was no effect of BCG on this relationship. Simple correlations between serum amyloid A ( SAA ) and (A) insulin concentrations and (B) body condition score ( BCS ) in plasma of 79 mature light‐breed horses. Associations between log 10 plasma SAA concentrations and plasma insulin concentration in horses grouped by body condition, where normal = 5–6, overweight = 7, and obese = 8–9 (C). Slopes apply to all body condition score groups. Discussion In several species, obesity is associated with slight increases in circulating concentrations of proinflammatory cytokines that increase the risk of obesity‐associated metabolic diseases such as diabetes. In horses, laminitis is an obesity‐associated disease thought to have an inflammatory component, because ponies with a clinical diagnosis of laminitis or a history of laminitis have higher TNF concentrations than do healthy ponies. Furthermore, in the laminar tissue of horses subjected to acute carbohydrate‐induced laminitis, cytokine mRNA abundance is increased. Thus, it is possible that an obesity‐influenced increase in circulating cytokine concentrations could be involved in the pathogenesis of obesity‐associated laminitis. In addition to obesity, increased plasma insulin concentrations (hyperinsulinemia) are associated with an increased risk of laminitis. Hyperinsulinemia may directly influence the development of laminitis because an IV infusion of insulin that increased insulin concentrations 100‐fold for up to 72 hours induced laminitis in previously healthy, lean ponies. Alternatively, hyperinsulinemia may indirectly increase laminitis risk by stimulating increased proinflammatory cytokine concentrations. In humans, insulin stimulates pro‐inflammatory cytokine production in skeletal muscle and adipose tissue. Thus, it is possible that hyperinsulinemia and obesity influence the risk of laminitis independent of each other. Therefore, the main objective of this investigation was to determine associations between BCS, plasma insulin concentrations, and plasma concentrations of the proinflammatory cytokines TNF, IL‐1β, and IL‐6, and the acute phase protein SAA. A 2nd objective was to determine associations between plasma cytokine concentrations and factors that previously have been associated with inflammation, including age and plasma triglyceride, NEFA, and leptin concentrations. In this investigation, plasma insulin concentrations correlated positively with BCS, a relationship that has been previously reported in mares and a mixed‐sex group of horses and ponies. The hyperinsulinemia observed in obese horses most likely is a component of insulin resistance, because the pancreas increases insulin production to compensate for the decreased effectiveness of insulin that occurs in insulin‐resistant animals. Insulin resistance is positively correlated with BCS and is evident after diet‐induced obesity. The determination of insulin sensitivity requires dynamic, multisample testing, but, as fasting insulin concentrations correlate with insulin resistance, plasma insulin often is used to assess insulin resistance. Recently, a threshold concentration of 20 mIU/L was set by the American College of Veterinary Internal Medicine to differentiate insulin‐sensitive from insulin‐resistant horses. Other factors found to be associated with BCS were plasma leptin, triglyceride, and NEFA concentrations. Leptin is a hormone secreted from adipocytes, and in the present investigation, circulating concentrations correlated positively with BCS and insulin, findings that are in agreement with previous investigations. Furthermore, leptin concentrations increase during diet‐induced weight gain, and it is possible that adipose tissue leptin production is driven by insulin, as insulin infusion increases plasma leptin concentrations in horses. In addition to obesity, leptin concentrations were higher in mares than geldings. Hyperleptinemic mares are more insulin resistant than hyperleptinemic geldings, suggesting a relationship between leptin production and sex. Leptin concentrations also are higher in obese women than obese men, and this relationship potentially is because of sexually dimorphic regional fat distribution, because leptin concentrations depend more on SC fat deposition, which is higher in females, than visceral adipose deposition, which is higher in males. In addition to leptin, we noted a positive correlation between both plasma triglycerides and NEFA, and body condition. These findings are similar to previous reports that fasting plasma triglyceride and NEFA concentrations are greater in higher BCS horses. Intriguingly, however, a 16‐week period of diet‐induced weight gain did not influence either triglyceride or NEFA concentrations despite a 71% reduction in insulin sensitivity and the development of hyperinsulinemia. This may suggest that chronic insulin resistance and hyperinsulinemia are more essential components of altered lipid metabolism than body condition. Plasma triglycerides also were positively correlated with insulin and leptin concentrations, but because BCS, insulin, and leptin are all positively associated, it is impossible to discern precise relationships among these factors. Our measurement of TNF, IL‐1β, and IL‐6 plasma concentrations did not identify any direct correlations with either BCS or insulin in this population of horses. Our results are similar to those of a recent investigation comparing obese to normal condition mares and geldings that indicated that TNF concentrations were not altered by obesity. In contrast to the present investigation, plasma TNF concentrations positively correlate with both BCS and plasma insulin in mares between a BCS of 4‐9. The exclusion of geldings from that study could have eliminated a source of variation, allowing detection of a relationship between obesity and TNF. This is an interesting possibility in light of our finding that TNF concentrations were higher in mares and could indicate the potential for sex‐specific effects. In contrast to TNF, neither IL‐6 nor IL‐1β concentrations have been investigated as indicators of inflammation in equine obesity; however, white blood cell mRNA levels of IL‐1β increase with obesity in mares. The disagreement in results of the present study and those of Vick et al suggests that several other factors influence circulating cytokine concentrations. In contrast to BCS and insulin, a direct correlation among age, sex, and triglycerides was noted with cytokine concentrations, suggesting that these factors might have a larger influence on cytokine production and secretion than either BCS or insulin. Mares from this investigation had higher TNF and IL‐6 concentrations than did geldings and stallions, indicating a sex effect on the relationship between BCS and inflammation. In addition, there was a positive relationship between age and IL‐6 concentrations. Increased inflammation is associated with aging in humans, and horses aged 25 years old have higher white blood cell production of TNF than horses aged 4 years. Interestingly, because obese aged horses have greater production of these cytokines than their lean counterparts, an interaction between obesity and age might exist. Because none of the horses included in this investigation were older than 20 years, increased inflammation may occur at an even younger age than previously described. Because plasma leptin, triglyceride, and NEFA concentrations are associated with obesity and insulin resistance, we hypothesized that they might also relate to inflammation. Given that triglycerides are increased in horses with a history of obesity‐associated insulin resistance, it was surprising that concentrations were negatively correlated with plasma TNF. Triglycerides may be influenced by other factors, because short‐term weight gain that decreased insulin sensitivity did not cause an increase in triglyceride concentrations. This could indicate that chronic obesity or insulin resistance is required to alter fasting triglyceride concentrations. In addition, triglycerides can be altered by diet and recent weight loss. In contrast to the triglyceride findings, plasma IL‐1β was positively correlated with leptin. Because BCS and insulin are highly correlated with leptin but not IL‐1β, it is likely that other factors influence the relationship between IL‐1β and adiposity. In addition to proinflammatory cytokines, acute phase protein activation is associated with obesity in several species. The acute phase response includes several proteins that are primarily produced by the liver as part of the innate defense system response to infection. These proteins include C‐reactive protein, serum amyloid A, and fibrinogen, with some indication that activation is species specific. In humans, an increase in plasma C‐reactive protein is used as a predictive indicator of cardiovascular disease risk. In horses, however, the acute phase protein, SAA, appears to be the principal acute phase protein, and has been repeatedly shown to be reactive to an inflammatory stimulus, increasing 236‐fold above basal measurements after administration of lipopolysaccharide. Our observation of a relationship between SAA and both insulin and adiposity was similarly reported in rodents. Body condition and insulin may have a greater influence on SAA concentrations than on any of the measured cytokines. This could indicate that SAA concentration is a better marker of obesity‐associated inflammation and laminitis risk than are TNF, IL‐6, and IL‐1β. This investigation was initiated to examine relationships between circulating concentrations of inflammatory proteins, obesity, and insulin resistance. Several factors, including sex and age, seem to influence circulating TNF, IL‐6, and IL‐1β concentrations, whereas serum SAA is correlated with both BCS and fasting insulin concentration. Thus, SAA concentrations may indicate obesity‐associated insulin resistance. Acknowledgments This study was supported by the Virginia Horse Industry Board, the estate of Paul Mellon, and the Macromolecular Interfaces with Life Sciences (MILES) Integrative Graduate Education and Research Traineeship (IGERT) of the National Science Foundation under Agreement No. DGE‐0333378. The authors also thank Louisa Gay, Kimberly A. Negrin, and Julie Franklin for assistance with data collection and leptin, glucose, insulin, triglyceride, and NEFA assays. Conflict of Interest : Dr Geor is currently an Associate Editor with the Journal of Veterinary Internal Medicine . Footnotes BD Vacutainer, Beckton, Dickson, and Company, Franklin Lakes, NJ Equine TNFα Screening Set, Endogen, Rockford, IL Phase Range Serum Amyloid A Assay, Tridelta Diagnostics, Ltd., Kildare, Ireland Equine IL‐1 beta VetSet, Kingfisher Biotech, Inc, St. Paul, MN SAS Enterprise Guide v 4.2, SAS v.9.2, SAS Systems, Cary, NC
Journal of Veterinary Internal Medicine – Wiley
Published: Jan 1, 2013
Keywords: ; ; ;
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera