Serum amyloid A (SAA) in synovial fluid has recently been used as a marker for septic arthritis in horses but the effects of repeated intra-articular (IA) administration of amikacin on synovial SAA concentrations are unknown.
To report the effect of repeated IA administration of amikacin on SAA, total protein (TP), nucleated cell count (NCC) and differential NCC in synovial fluid of healthy equine joints.
A controlled, 2 period crossover study was performed on 5 clinically healthy horses. Each intercarpal joint received one of 2 treatments every 48 h for 5 consecutive times: arthrocentesis alone (control group) or arthrocentesis combined with IA administration of 500 mg of amikacin (treatment group). Clinical and lameness examinations were performed daily. Serum SAA and synovial SAA, TP, NCC and differential NCC were measured and statistically compared. Significance level was set at P<0.05.
Horses remained healthy and nonlame throughout the study. Baseline values for all variables were not significantly different between groups. Values for TP in the treatment group were significantly higher than in the control group after the first sample (P<0.05). In both groups NCC increased significantly (P<0.05) after the first sample. No significant changes were identified in differential NCC. In both groups, all synovial and most serum SAA concentrations remained below the lower limit of quantification.
Repeated IA administration of amikacin caused increased values of TP and NCC in synovial fluid, with some TP concentrations falling within the range reported for septic arthritis. In contrast, synovial SAA concentrations did not increase in either group.
Synovial SAA could serve as a more reliable marker than TP and NCC when evaluating a joint previously sampled or treated with amikacin.
Septic arthritis is one of the most severe arthropathies affecting horses of all ages. Survival rates of affected equids range from 62% in foals to 85% in adult horses  and rates of return to previous athletic activity range from 48.3% to 65.8% [2, 3]. An early, aggressive multimodal approach is recommended for treatment of septic synovitis. Antimicrobials are frequently administered intra-articularly, as this is easy to perform and provides high synovial drug concentrations, thereby increasing antimicrobial efficacy . Amikacin is commonly the preferred antimicrobial owing to its broadest activity against the most common pathogens isolated from equine synovial infections .
Diagnosis of sepsis in a synovial structure can be challenging, as bacteria are not always identified on cytology  or isolated from bacterial cultures [1, 6, 7]. Therefore, practitioners usually use cytological examination of samples of synovial fluid and measurement of inflammatory markers in synovial fluid to make a diagnosis of septic synovitis [8-12]. The presence of septic synovitis is typically considered with synovial total protein (TP) concentrations >40.0 g/l and nucleated cell count (NCC) > 30 × 109 cell/l although lower values may be observed in some cases . A high percentage of neutrophils in the fluid >80% (normal range <10%) is also suggestive of sepsis [8, 14, 15]. Degenerative changes in neutrophils are less frequently observed in fluid obtained from septic joints than from other body cavities, owing to the lower concentration of bacterial toxins in synovial fluid than in other body tissues . Synovial TP and NCC determinations are usually repeated during the course of treatment and the results of these tests serve as a guideline for making adjustments to therapy and prognosis . Unfortunately, repeated arthrocenteses alone , as well as single IA administration of antimicrobials , can result in increased synovial TP concentrations and NCC values, which can confound the clinical interpretation of these parameters. Therefore, identification of more reliable markers for synovial sepsis is warranted.
Serum amyloid A (SAA) is an acute phase inflammatory protein which has various clinical applications in horses [19-22]. It is synthesised mainly by the liver in response to inflammation and infection [19, 23], but can also be synthesised by synoviocytes in response to synovial inflammation and sepsis . Serum amyloid A concentration in synovial fluid has been shown to serve as a good marker for septic arthritis and tenovaginitis, reflecting changes in inflammatory activity in the equine joint [17, 24]. When the effect of 8 repeated arthrocenteses performed after a single intra-articular injection of 2 ml of isotonic saline on inflammatory markers in the equine joint was evaluated , synovial TP concentration remained significantly increased over baseline values from the second to the last arthrocentesis. In contrast, concentrations of SAA in synovial fluid did not increase during the study . Although IA administration of antimicrobials is often repeated during treatment of horses with synovial sepsis, the effect of repeated IA administration of amikacin on synovial fluid concentrations of SAA, TP, NCC and differential NCC has not been evaluated in horses.
The objective of this study was to evaluate the synovial inflammatory markers SAA, TP and NCC in healthy horses undergoing repeated IA administration of amikacin. We hypothesised that synovial SAA would not increase in response to repeated arthrocentesis with or without IA administration of amikacin, whereas synovial TP and NCC would increase.
Materials and methods
This study was approved by the Institutional Animal Care and Use Committee. A controlled, 2 period crossover study was performed on 6 Nooitgedacht mares, with a mean ± s.d. age of 3.8 ± 0.76 years (range 2–6 years) and mean ± s.d. bwt of 400 ± 58.4 kg (range 302–450 kg). Horses had not received any medical treatment for at least 8 weeks prior to the study and were determined to be healthy and free of musculoskeletal disorders based on physical and lameness examinations as well as complete blood cell count, fibrinogen and systemic SAA measurement. Horses with abnormal results were rejected.
Each horse underwent 2 study trials (control and treatment group) randomly distributed between the front limbs with a 20 day washout period in between.
Control group: Arthrocentesis of the intercarpal joint was performed every 48 h for a total of 5 times. A synovial fluid sample (∼1.5 ml) was collected each time. Amikacin sulphate (500 mg; 2 ml) [Amikacin-Fresenius]1 was injected into the intercarpal joint after the last synovial fluid collection.
Treatment group: Arthrocentesis and synovial fluid collection were performed as for the control group. Amikacin sulphate (500 mg; 2 ml) was administered into the intercarpal joint after each synovial fluid collection.
Horses were sedated with romifidine (0.02–0.03 mg/kg bwt i.v.) and the carpal area clipped and aseptically prepared. Arthrocentesis was performed on the dorsolateral synovial pouch with the carpus in semiflexion and using a 22 gauge needle . Blood was collected from the jugular vein at each time point. The same investigator (A.F.S.T.) performed all sample collections.
Physical and lameness examinations were performed in all horses every 24 h during and for 2 days after each study period. Lameness was graded in a scale from 0 to 5  by one investigator (L.R.M.) who was unaware of group distribution.
An aliquot of each synovial fluid sample was used to measure TP concentration with a standard refractometer (Protein Refractometer)2. The remaining synovial fluid was transferred into an EDTA tube and used for NCC analysis using an automated haematology analyser (Cell-Dyn 3700 System)3. The remaining sample was centrifuged for 5 min at 1341 g (Rotofix 32A)4. Supernatant was stored at -80°C for SAA determination. A smear of the precipitate was stained with haematoxylin and eosin (Rapidiff 1)5 and used for calculation of differential NCC and percentage of neutrophils by an experienced pathologist unaware of group distribution.
Venous blood samples were centrifuged for 5 min at 1341 g (Universal 320)4 and serum was stored at -80°C for SAA determination.
An automated chemistry analyser (COBAS INTEGRA 400 plus)6 with a human SAA turbidometric immunoassay (Eiken SAA TIA)7 previously validated for equine use  was used for SAA concentration determination. In house performance of the assay was assessed using synovial fluid and serum samples from clinical cases with increased SAA concentrations. Three independent, 5 serial dilutions for each sample type (synovial fluid and serum) were prepared using the diluents provided with the SAA measurement kit.
Descriptive and comparative statistic analyses were performed using the Statistical Analysis System8. Normality of data was assessed by Shapiro-Wilk and Kolmogorov-Smirnov statistical tests (level of significance P<0.05). For dependent variables (TP, NCC, neutrophil percentage), a repeated samples scheme based on the area under the curve (AUC) was applied . The AUC for each dependent variable was estimated following the trapezoidal rule  with the following equation :
Where ‘t’ was sampling time and ‘y’ the observed outcome.
Crossover analysis of variance (ANOVA) was implemented for the statistical comparison of AUC of dependent variables between the groups. In addition, Mann-Whitney U and unpaired Student's t tests were applied to compare dependent variables between treatment groups at each sampling time and each sampling time vs. baseline within each group. Significance level was set at P<0.05.
All horses remained free of lameness throughout the study. Five horses remained clinically healthy, but one mare was excluded from the study as she developed respiratory disease unrelated to the study. This horse had hyperfibrinogenaemia (26.5 µmol/l) and increased serum SAA concentration (86.1 mg/l).
Protein concentrations in synovial fluid
Synovial fluid TP (mean ± s.d.; g/l) for both groups is included in Figure 1 and Table 1. Synovial fluid TP baseline values were not significantly different between groups. Protein AUC0–192 (g x h/dl) for the control group (262 ± 79.8) was significantly lower (P<0.01) than for the treatment group (621 ± 91.2).
Table 1. Mean ± s.d. synovial total protein (TP; g/l), nucleated cell count values (NCC; x 109 cells/l), percentage of neutrophils (PNEU; %) and serum amyloid A (SAA; mg/l) in synovial fluid and serum samples from 5 horses in response to repeated arthrocentesis (CG) or repeated arthrocentesis and administration of amikacin (TG) into the intercarpal joint at each sampling time (hours)
NCC (x109 cells/l)
Synovial SAA (mg/l)
SAA in serum (mg/l)
10.2 ± 1.30 (9.00–12.0)
0.06 ± 0.04 (0.02–0.15)
5.60 ± 9.53 (0–22)
31.0 ± 9.80 (15.0–39.0)
0.63 ± 0.37 (0.21–1.12)
5.80 ± 4.60 (1–13)
31.2 ± 5.80 (23.0–39.0)
2.49 ± 2.38 (0.46–6.31)
8.80 ± 3.35 (14–41)
40.4 ± 7.20 (32.0–52.0)
1.85 ± 2.21 (0.38–5.69)
10.4 ± 9.07 (2–24)
43.2 ± 9.50 (36.0–60.0)
1.68 ± 1.14 (0.65–3.24)
15.2 ± 10.9 (3–32)
10.6 ± 1.80 (8.00–13.0)
0.06 ± 0.03 (0.04–0.12)
11.8 ± 14.8 (3–38)
13.3 ± 4.80 (8.00–21.0)
0.54 ± 0.46 (0.05–1.07)
24.2 ± 10.2 (14–41)
19.2 ± 7.80 (11.0–31.0)
1.33 ± 1.20 (0.16–3.36)
32.0 ± 29.3 (6–79)
12.0 ± 4.30 (8.00–19.0)
0.68 ± 0.42 (0.43–1.22)
8.00 ± 6.32 (0–16)
10.2 ± 0.80 (9.00–11.0)
0.45 ± 0.22 (0.25–0.76)
9.80 ± 7.29 (2–18)
Synovial fluid TP in the control group ranged from 8.00 to 31.0 g/l during the study (13.0 ± 5.4 g/l) and values were significantly higher than baseline only at 96 h (P = 0.03) (Fig 1). In the treatment group, synovial fluid TP ranged from 9.00 to 60.0 g/l during the study (31.2 ± 13.6 g/l) and values were significantly increased compared with baseline after the first injection of amikacin and remained increased thereafter (P<0.01) (Fig 1). Except for baseline, TP concentrations were significantly different (P≤0.03) between groups at all collection time points (Fig 1 and Table 1).
Nucleated cell count in synovial fluid
Values of NCC (mean ± s.d.) for both groups are shown in Table 1. Baseline NCC values did not differ significantly between groups. Nucleated cell count AUC0–192 in the control group (136 ± 95.9 × 10 x h/l) and the treatment group (281 ± 252 × 10 x h/l) did not differ significantly.
Values of NCC in the control group ranged from 0.04 to 3.36 × 109 cells/l (0.61 ± 0.70 × 109 cells/l). In the treatment group, NCC values ranged from 0.02 to 6.31 × 109 cells/l (1.34 ± 1.67 × 109 cells/l). Compared with baseline, NCC was significantly higher at all sampling times (P<0.05) in both groups (Fig 2). When values for the 2 groups were compared, NCC was significantly different only at 192 h (P<0.05). Values of NCC peaked at 96 h in both groups (Fig 2).
Percentage of neutrophils in synovial fluid
The percentages of neutrophils in the synovial fluid for both groups are included in Table 1. Baseline values and AUC0–192 did not differ significantly between groups.
The percentages of neutrophils in the control group ranged from 0.00% to 79.0% (17.2 ± 17.4%). In the treatment group, the values ranged from 0.00% to 24.0% (9.16 ± 8.20%). No significant differences were detected in the percentages of neutrophils when compared against the baseline values or between groups at any time point.
In house validation of SAA measurement technique
For SAA in synovial fluid, the lower limit of quantification was 0.05 mg/l, the coefficient of determination was >0.90 and the intra-assay variability ranged from 1% to 10%. For SAA in serum, the lower limit of quantification was 0.21 mg/l, the coefficient of determination was >0.90 and the intra-assay variability ranged from 0.8% to 16%.
Serum amyloid A in synovial fluid and serum samples
Synovial SAA values remained below the lower limit of quantification in both groups (Table 1). Systemic SAA concentrations remained ≤18.42 mg/l and most of the samples remained below the lower limit of quantification (Table 1). Owing to the small number of samples with values of SAA above the lower limit of quantification, SAA results were not compared statistically.
The results of previous studies indicate that synovial TP increases either in the presence of synovitis or after intra-articular injections . In the present study, synovial TP remained within the normal range (<25.0 g/l)  after repeated arthrocentesis, with the exception of one sample (96 h). These findings are similar to those previously reported for samples obtained from the digital flexor tendon sheath, when samples were collected sequentially after intrathecal administration of lactated Ringer's solution . In contrast, Jacobsen et al. (2006) reported a significant increase in synovial TP concentration after repeated arthrocentesis combined with intra-articular administration of saline solution . The use of different sampling protocols, along with injection of different solutions [17, 18], might have resulted in varying grades of inflammatory reaction and thereby account for the different findings among studies.
In the present study, repeated intra-articular administration of amikacin resulted in TP concentrations that exceeded the normal range in all horses when the samples were obtained after either the second (4 horses) or third (1 horse) arthrocentesis; these concentrations remained increased thereafter. Some of these samples had TP ≥40.0 g/l, values typically associated with septic synovitis [8, 15]. In previous studies, single intra-synovial administration of amikacin (250 mg into the digital flexor tendon sheath or 500 mg intra-articularly) produced a mild increase in synovial TP concentrations, which remained below values typically associated with septic arthritis [4, 18]. Differences in the frequency of amikacin administration (single dose in previous studies vs. 5 doses in our study), the response to synoviocentesis (joint vs. tendon sheath), antimicrobial pharmaceutical preparations, as well as total antimicrobial dose per horse (larger horses receiving lower net doses than smaller horses) might explain different results among studies.
Nucleated cell count can increase in the presence of inflammation, secondary to arthrocentesis or after injection of different substances . In our study, increased NCC was observed in both groups compared to baseline and the only difference between groups was at 192 h, when NCC in the treatment group exceeded that in the control group. In both groups NCC remained ≤6.31 × 109 cells/l, values that are similar to those reported in previous studies after arthrocentesis or intra-articular injection of local anaesthetics . In contrast, Dykgraaf et al. (2007) reported increases in synovial NCC values that were consistent with septic synovitis (>30 × 109 cells/l) after a single administration of lactate Ringer's solution or amikacin into the digital flexor tendon sheath . Differences between studies might be attributed to a greater difficulty in accessing a tendon sheath, with a correspondingly higher risk of blood contamination of the sample, and greater sensitivity of sheaths to centesis than with joints .
Blood contamination during sampling may have been responsible for the increased synovial TP concentration or NCC values in our study. However, arthrocentesis of the intercarpal joint via a dorsal approach has low degree of difficulty , all arthrocenteses were performed by the same operator (AST) and needle repositioning was necessary only in 3 occasions. Owing to the reduced number of times the needle was repositioned, it was not possible to evaluate statistically any potential association between repositioning of the needle and either TP or NCC in the synovial fluid samples. In one of the 3 horses that required needle repositioning TP and NCC remained within the normal reference range, in the second horse both were already above the reference range before the needle was repositioned, and in the third both parameters increased to values above the reference range 144 h after the time of needle repositioning.
Modest increases in the percentages of neutrophils in synovial fluid have been observed in the presence of mild synovitis, after arthrocentesis, and after intra-articular injection of different substances . An overall mild increase in the percentages of neutrophils that lacked cytological evidence of toxic changes was observed in the present study. These findings would not be consistent with sepsis.
Repeated arthrocentesis every 48 h did not cause a detectable increase in SAA concentrations in either serum or synovial fluid, a finding that is consistent with those of previous reports . Repeated arthrocentesis with or without the intra-articular administration of amikacin might not be a sufficiently strong inflammatory stimulus to induce the synthesis of SAA. In humans, serum SAA concentrations increased more in response to bacterial stimuli than to viral infection or other kinds of inflammatory processes . Synovial and serum SAA in horses with septic arthritis have previously been reported to be >1000 mg/l . In our equine hospital, concentrations of SAA in synovial fluid collected from septic joints typically are >800 mg/l (unpublished data). In addition, SAA has a very short half-life in serum; in mice, the half-life is 75–80 min and 95% is eliminated from plasma 6 h after the synthesis has ceased . Therefore, SAA concentrations may have increased and returned to normal values before the next sample was collected 48 h later. More frequent arthrocenteses might have allowed us to detect increased synovial SAA concentrations if they had occurred. To the authors' knowledge, the half-life of SAA in equine synovial fluid has not been reported, but repeated arthrocentesis (4 h apart) combined with intra-articular injection of saline solution did not increase synovial SAA concentrations in horses in a previous report . The fact that SAA concentrations were not increased after repeated arthrocentesis with or without repeated intra-articular administration of amikacin every 48 h, and that synovial SAA concentrations reached very high levels during septic arthritis [17, 24], suggest that measurement of synovial SAA might be valuable when monitoring clinical cases with septic synovitis.
This study was designed to resemble a clinical scenario when veterinarians evaluate and/or treat horses with septic arthritis. Amikacin was selected for its efficacy against common bacterial isolates , and administration of 500 mg of amikacin was repeated every 48 h based on previous studies . Intra-articular injection of amikacin produced significant changes on the synovial inflammatory markers, mainly TP and NCC. Whether these changes were due to a chemical effect of amikacin or to the act of introducing foreign fluid into the synovial structure cannot be concluded from this study. In previous studies a single intra-articular injection of amikacin did not produce cytological or clinical evidence of chemical synovitis , but amikacin exerted mild toxic effects on equine chondrocytes in vitro . An additional control group undergoing repeated arthrocentesis combined with injection of a placebo substance (lactated Ringer's or saline solution) could have been included to determine whether the changes observed in the treatment group could be related to the act of injecting a substance rather than just to amikacin. This was not done in the study owing to economic limitations, and the inclusion of a control group without injection of a sham solution was considered preferable, as this is routinely performed by equine practitioners to collect samples of synovial fluid and evaluate response to treatment.
The human SAA turbidometric immunoassay used in this study has been previously validated for equine samples  and has been used by other investigators [17, 24] and diagnostic laboratories  to measure SAA in equine serum and synovial fluid. The in house evaluation of assay performance revealed lower limits of quantification values and intra-assay variability similar to those reported for previous studies .
In conclusion, repeated arthrocentesis and intra-articular administration of amikacin every 48 h increases synovial TP concentration and NCC values in some horses to values that are consistent with septic arthritis. However, synovial SAA concentrations were not affected and therefore could serve as a better marker for synovial sepsis when evaluating a joint previously sampled or treated intra-articularly with amikacin.
Authors' declaration of interests
None of the contributing authors have received grants or speaker fees from a commercial body within the past two years.
Source of funding
This study was supported by the Research Developmental Fund, University of Pretoria, South Africa.
Total protein and SAA concentration measurements and cytological analyses were conducted at the Section of Clinical Pathology, Faculty of Veterinary Science, University of Pretoria, Onderstepoort, South Africa.
All listed authors made substantial contributions to study design, interpretation of data, drafting/revising the article and gave final approval of the version to be published. A.S.T., L.R.M. and M.G.S. performed data collection. N.F.V. performed statistical analysis on the data.
Intramed, South Africa.
A.S.T. Inc., Japan.
Hettich Zentrifugen, UK.
Clinical Sciences Diagnostics, South Africa.
Roche Laboratories, Switzerland.
LZ test SAA, Eiken Chemical Co.
SAS Institute, version 9.2, Cary, North Carolina, USA.