This work was performed at New Bolton Center, University of Pennsylvania School of Veterinary Medicine; University of New Hampshire; and Cummings School of Veterinary Medicine at Tufts University. Portions of these data were presented at the ACVIM Forum (2008) and the Annual Symposium of the Veterinary Comparative Respiratory (2000).
Reproducibility of Airway Responsiveness in Horses Using Flowmetric Plethysmography and Histamine Bronchoprovocation
Article first published online: 13 APR 2009
Copyright © 2009 by the American College of Veterinary Internal Medicine
Journal of Veterinary Internal Medicine
Volume 23, Issue 3, pages 631–635, May/June 2009
How to Cite
Nolen-Walston, R. D., Kuehn, H., Boston, R. C., Mazan, M. R., Wilkins, P. A., Bruns, S. and Hoffman, A. M. (2009), Reproducibility of Airway Responsiveness in Horses Using Flowmetric Plethysmography and Histamine Bronchoprovocation. Journal of Veterinary Internal Medicine, 23: 631–635. doi: 10.1111/j.1939-1676.2009.0307.x
- Issue published online: 19 MAY 2009
- Article first published online: 13 APR 2009
- Submitted December 9, 2008; Revised January 22, 2009; Accepted February 12, 2009.
- Inflammatory airway disease;
- Lung function test;
Background: Inflammatory airway disease has a high prevalence in horses, but is often a diagnostic challenge. Flowmetric plethysmography and histamine bronchoprovocation (FP/HBP) is a simple and effective tool for diagnosis, but reproducibility of these measurements made over time has not been established.
Hypothesis: We hypothesize that the measurement of airway responsiveness in horses using FP/HBP is consistent over both short and long periods of time.
Animals: Twenty-nine healthy adult horses from 2 university herds.
Methods: In this prospective experimental study, airway responsiveness was determined in each horse at day 0 (baseline [BL]) with FP/ HBP, using PC35 (provocative concentration of histamine needed to increase Δflow by 35%) as a measure of airway responsiveness. Each horse was re-tested 1–4 weeks after BL (short-term [ST]) and again at 3–12 months after BL (long-term [LT]).
Results: In the ST period, 23/27 (85%) of the horses had a PC35 that was within 1 doubling concentration of histamine of their BL value, with a mean change of 0.52 doubling concentrations (95% CI 0.26–0.79, range 0–2.06). For the LT data, 19/26 (73%) of horses were within 1 doubling concentration of their BL value, with a mean change of 0.81 doubling concentrations (95% CI 0.45–1.17, range 0.14–3.10). There was no significant difference in reproducibility between the 2 groups of subjects.
Conclusions and Clinical Importance: Repeated measurements of airway responsiveness obtained with FP/HBP show acceptable reproducibility over time periods up to a year. However, caution must be used when testing horses when ambient air temperature is low.
flowmetric plethysmography/histamine bronchoprovocation
inflammatory airway disease
provocative concentration (of histamine) needed to increase Δflow by 35%
recurrent airway obstruction
Inflammatory airway disease (IAD) is a common performance-limiting problem in both racehorses and sport horses, characterized by cough, mucoid airway secretions, or decreased performance.1 However, the clinical signs can be subtle and may be confused by other causes of exercise intolerance such as cardiac disease or upper airway obstruction. The diagnosis of IAD can be made on clinical signs combined with endoscopic evidence of tracheal mucus, abnormal bronchoalveolar lavage (BAL) cytology, or demonstration of airway obstruction or airway hyperresponsiveness.1 However, there is a degree of disjunction between BAL results and abnormalities in pulmonary function, with some horses showing abnormal lung function in the face of normal BAL cytology, and vice versa.1 Pulmonary function testing can allow the definitive diagnosis of IAD in a wide range of subjects, and also has the additional benefit of providing an objective method of monitoring the response to therapy. A variety of pulmonary function tests are available but only a few are noninvasive and therefore generally applicable to client-owned animals. One noninvasive approach is flowmetric plethysmography (FP), a technique that has been used not only in horses2–4 and humans,5 but also in sheep,5,6 llamas,7 rabbits,8 dogs,9 cats,10 and even roosters11 and Yucatan miniature pigs.12 In horses, FP (trade-named “Open Plethysmography”a) is conducted using a combination of external sensors placed on the body surface (respiratory inductance bands) and a measure of flow by a pneumotachograph.2 This method helps to characterize changes in lung mechanics associated with recurrent airway obstruction (RAO; heaves) and IAD.2,13
The clinical utility of bronchoprovocation as a method for both diagnosing and serially monitoring IAD, however, is dependent on its repeatability (agreement of measurements taken under identical conditions, where any variation can be assumed to be due to errors in the measurement process) as well as its reproducibility (the variation in measurements made on a subject under changing conditions, such as time).14 The objective of this study was to investigate the temporal agreement of airway responsiveness in adult horses by flowmetric plethysmography and histamine bronchoprovocation (FP/HBP) over both short and long time periods.
Materials and Methods
This prospective study was designed to examine the reproducibility of FP/HBP in adult horses. All animal testing was carried out with the approval of the Institutional Animal Care and Use Committee at each institution either performing the study (Tufts University, University of Pennsylvania) or providing the subjects (University of Pennsylvania, University of New Hampshire).
Twenty-nine adult horses from 2 university-owned herds (University of New Hampshire and University of Pennsylvania) were enrolled in the study. All horses had normal physical examination findings, and none had a history of RAO. Subjects from the University of New Hampshire (n = 20) were used for a collegiate riding program, and consisted of 4 Quarter horses, 6 Thoroughbreds, 6 Morgans, and 4 Warmbloods or Warmblood crosses. Of these, 14 were geldings and 6 mares, aged from 3 to 25 years. They were housed either in a stall bedded with wood shavings, and fed concentrate and dry grass hay, or turned out with a run-in shed bedded with wood shavings, and fed dry grass hay. Subjects from the University of Pennsylvania (n = 9) were mares from a reproduction-teaching herd consisting of 4 Thoroughbreds, 3 Thoroughbred crosses, and 2 Standardbreds, and aged from 5–19 years. They were housed in a large pasture and had access to a well ventilated run-in shed and dry grass hay. In neither herd did housing conditions or diet vary considerably over the study period. Horses were tested with plethysmography and histamine bronchoprovocation at day 0 (baseline, BL), and then a short-term (ST) repeat measurement was made between 1 and 4 weeks after BL, and a long-term (LT) measurement was taken from 3–12 months after BL.
Pulmonary function testing and histamine bronchoprovocation were performed in each animal with a commercial flowmetric plethysmography systema as described previously.2 Briefly, each horse was lightly sedated (xylazineb 0.5 mg/kg), and fitted with a mask, pneumotachograph, and abdominal and thoracic inductance bands. The system was calibrated as per the manufacturer's instructions. Measurement of airway obstruction was calculated by the software by subtracting the flow signal generated by thoracic volume change from the pneumotachograph flow at peak expiration. This parameter is termed Δflow, and increases with bronchoconstriction as gas is compressed in the lungs.2 After calibration, pre-test (PT) Δflow was measured for 3 minutes, and then again after 2 minutes of nebulization with 0.9% salinec (negative control). All nebulization was performed with the low dead-space, airtight mask associated with the commercial plethysmography systema and commercial portable air compressorsd,e and nebulizersf,g that generate particles with a mass median aerodynamic diameter (MMAD) of < 3 μm. Histamineh in salinec was then nebulized for 2 minutes at doubling concentrations (2, 4, 8, 16, and 32 mg/mL) until a ≥ 35% increase in Δflow was recorded, or the maximum concentration (32 mg/mL) was reached. Data collection were initiated immediately after nebulization of histamine with recording periods of 3 minutes, resulting in intervals of approximately 7 minutes between each concentration. A dose-response curve was generated, and the provocative concentration of histamine resulting in a 35% increase over the postsaline Δflow (PC35) obtained using linear interpolation.
Bronchoprovocation (PC35) data were transformed using log base 2, and these results as well as PT Δflow evaluated Tukey's ladder15 for normal distribution. Tukey's ladder either confirms normality or locates a normalizing transformation appropriate for the specific data set. As the horses used for this study were from 2 separate herds, they were separated based on this characteristic and compared with linear regression using an indicator variable to potentially distinguish between models for each data subset. By the normality assumption (above), confidence intervals of the geometric mean were calculated for BL, ST and LT PC35 and PT Δflow. The ST and LT reproducibility of PC35 was assessed by linear regression and concordance analysis, and the geometric mean and absolute variation from BL calculated. In addition, overall regression and time-restricted regressions (BL, ST, and LT) were run for PC35 as a function of PT Δflow to identify association between initial values of Δflow values and airway reactivity. Lin's concordance correlation coefficient16 was used to quantify the relationship between the 2 variables under analysis subject to neither being the sole source of variation in that relationship. In essence, Lin's concordance correlation coefficient quantifies the agreement of the 2 variables with an identity line that describes them. Whereas Pearson's correlation assesses the linear relationship between 2 variables, Lin's concordance correlation coefficient assesses the degree to which the 2 variables agree. All analysis was performed with a commercial statistics package,i and statistical significance was set at P < .05.
The (base 2) logarithmically transformed PC35 did not deviate significantly from normality (P= .116). PT Δflow data are transformed logarithmically to meet assumptions of normality. For the PC35 data, regression analysis was performed as described above, potentially allowing for differences in slope between the dependent (ST and LT) and independent (BL) variables for each dataset to evaluate if there was a significant difference between the 2 herds. When the data subsets separated by herd were compared, there was no significant difference between either the intercept (P= .823) or the slopes (P= .357), indicating that a single regression model would describe the relationship between these dependent and independent variables across the 2 populations concurrently (ie, there was no significant differences in either the ST or LT data's relation to BL values between the 2 herds). Twenty-nine BL, 27 ST, and 26 LT data points were obtained. The geometric mean of the BL PC35 was 5.29 mg/mL (95% confidence interval [CI] 3.44–8.22, range 0.39–48), the ST was 6.15 mg/mL (95% CI 4.09–9.25, range 1.33–35.29), and the LT was 4.51 mg/mL (95% CI 3.01–6.76, range 1.07–35.64). Geometric means of PT Δflow were 1.79 L/s (95% CI 1.49–2.15) at the BL time point, 1.87 L/s (95% CI 1.51–2.30) at the ST time point, and 1.58 L/s (95% CI 1.28–1.96) at the LT time point.
The PC35 values were log base 2 transformed. With log base 2, every unit of change in PC35 indicates a difference of 1 doubling concentration of histamine. In the ST period, 23/27 (85%) of the horses were within 1 doubling concentration of their BL value, with a mean difference of 0.52 doubling concentrations (95% CI 0.26–0.79, range 0–2.06). For the LT data, 19/26 (73%) of horses were within 1 doubling concentration of their BL value, with a mean difference of 0.81 doubling concentrations (95% CI 0.45–1.17, range 0.14–3.10). Linear regression analysis showed significant linear relationship (Fig 1A, B) between BL and both ST and LT data points (P < 0.001). For each 1 unit increase in log of BL, there is a corresponding 0.784 (±0.090) increase in log of ST, and a 0.641 (±0.121) increase in log of LT. There was evidence of fanning in the plots of the observation around the regression line at higher values of PC35 in the ST and especially the LT analyses (Fig 1B). By the concordance correlation coefficient, concordance with BL was 0.86 for the ST values (95% CI 0.76–0.96, P < .001), and 0.72 (95% CI 0.53–0.91, P < .001) for the LT data (Fig 1C, D). Overall regression and time-restricted regressions (at BL, ST, and LT time points) for PC35 as a function of PT values of Δflow showed no significant association between PT Δflow and subsequent airway responsiveness as measured by PC35 (overall regression, P= .139, smallest P-value by time P= .144).
This study demonstrates that measurements of bronchial responsiveness using FP/HBP in adult horses are consistent with the reproducibility of bronchoprovocation results in adult humans.17–22 We demonstrate that in 2 different populations of adult horses, PC35 is stable to within a doubling concentration of histamine in the majority of subjects. This reproducibility is greater in the ST group than the LT group, which may be explained by the fact that horses are less likely to have significant changes in environment, exposure to allergens or triggers (year to year, season to season), or pulmonary comorbidities. An additional factor that may have influenced the increased variation in the longer time period is that some of the horses were tested initially during the summer, and then retested 3–6 months later during the winter months. Two of the outliers which showed increased airway responsiveness at the LT period (Fig 1B, D) represent subjects that underwent testing on a day when ambient temperatures were 5°C. Cold-air–induced bronchoconstriction is well described in both humans and horses,23,24 and is related to responsiveness to nonspecific agents such as methacholine25–27 Both of these horse were initially hyperreactive to histamine during BL testing (PC35 < 8 mg/mL histamine), which may suggest cross-responsiveness to these stimuli exist in horses. This study was limited in its evaluation of horses that were extremely hyperresponsive to histamine (ie PC35 < 2 mg/mL histamine). As the lowest dose of histamine used was 2 mg/mL, reproducibility of PC35 below this level was not investigated in depth. Although this is of minimal importance clinically (as horses with PC35 <2 mg/mL can be considered to be extremely hyperresponsive regardless of the exact value), this may be relevant when evaluating changes in low values of PC35 in a research setting. The lack of association between PT Δflow and airway responsiveness is consistent with previous studies showing that BL measurements of airway obstruction such as pulmonary resistance or dynamic compliance are not associated with airway response to histamine.13,28–30 In part this may be because of the poor sensitivity of BL measurements in detecting disease in horses that are not clinically abnormal at rest, but may also suggest that the mechanisms that control BL airway tone and airway hyperreactivity are different.
An additional notable feature from these results is that the data demonstrate increased heteroskedasticity (unequal variance) as PC35 values increase (Fig 1B), especially in the LT group. It is questionable that this is of significant clinical relevance as PC35 values of > 8 mg/mL histamine are not associated with an IAD/RAO phenotype,31 for the degree to which PC35 is greater than this value is unlikely to alter diagnosis or therapy. However, despite being free from clinical signs associated with respiratory disease or a history of RAO, many of the horses in this study would be considered hyperreactive by the previous criteria. Recent exposure to respiratory allergens, irritants or pathogens may have occurred either systematically or randomly during this study leading to inconsistency in PC35 at any time point. Importantly, it is evident that airway hyperresponsiveness is a stable phenotype over the ST or LT, suggesting that structural features of the airways (airway wall thickness, epithelial hyperplasia) rather than labile features of airway bronchomotor tone likely control this feature.
Bronchial responsiveness in humans has been found to be reproducible to within 1 doubling concentration of histamine over ST periods of < 1 week17,18 and within ∼2 doubling doses of histamine over longer time periods up to 30 months.19–22 In the equine literature, Klein et al28 found excellent ST (1–4 days) reproducibility (r= 0.97) of histamine bronchoprovocation in 10 horses on dynamic compliance using an esophageal balloon technique; others have noted similar findings in the control groups horses involved in RAO studies, but have not performed specific analyses of repeatability or reproducibility.29,30 These findings in both the human and veterinary literature are very consistent with the results obtained in this experiment. As discussed previously, the greater variability seen when testing is repeated at distant time points may be because of a variety of intra-or extra subject factors that are more likely to fluctuate or trend as a function of time. However, a specific concern for performing multiple HBP within a short period is the potential of repeated histamine administration to cause a change in responsiveness, either downwards due to tachyphylaxis (a rapid decrease in response with repeated exposure to a drug), or upwards due to increased sensitization to the agent. Studies in humans confirm that tachyphylaxis to inhaled histamine occurs if HBP is repeated at 1- or 3-hour intervals, but that bronchial responsiveness is repeatable to within 1 doubling concentration if separated by 6–48 hours.18,22 Persistent hyperresponsiveness is considered likely to be a genetic phenotype, and thus stable over time although elements such as ozone exposure or recent infection can increase responsiveness transiently.32 It is unclear exactly the exact degree to which genetic polymorphism in horses are associated with an RAO phenotype, but to date multiple genes have been implicated.33–35
To be clinically valuable, pulmonary function testing should be accurate, precise, repeatable, reproducible, and user friendly. In veterinary medicine, meeting these criteria in horses has been challenging. Initial measurements of lung physiology were obtained using conventional lung mechanics where changes in pleural pressure are measured with an esophageal balloon,36 and this is still considered the gold standard for research purposes. However, it is relatively invasive, and is generally performed in a hospital setting. Forced oscillometry,37 impulse oscillometry,38 and forced expiratory maneuvers39 have also been validated in the horse, but involve extensive laboratory space, expertise, or financial outlay. With these concerns in mind, and the prevalence and clinical importance of IAD and RAO in the horse,40–42 the use of FP/HBP is increasingly appealing. The data in this study support the use of flowmetric plethysmography from the standpoint of reproducibility, in addition to the known portability and accuracy for measurement of airway responsiveness to histamine.2
a Open Pleth, Ambulatory Monitoring Inc, Ardsley, NY
b AnaSed, Akorn Inc, Decatur, IL
c 0.9% preservative-free saline, Hospira Inc, Lake Forest, IL
d ProNeb Turbo, Model 38B0201 (Tufts/U New Hampshire herd), Pari, Midlothian, VA
e DeVilbiss Pulmo-Mate Compressor, Model 4650D (UPenn herd), Somerset, PA
f Pari LC Plus, Part #22F81 (Tufts/U New Hampshire herd), Pari
g Micro-Mist nebulizer, Part #1882 (UPenn herd), Hudson RCI, Temecula, CA
h Histamine diphosphate monohydrate, MP Biomed, Solon, OH
i Stata 10.0, StataCorp, College Station, TX
Conflict of interest: AH and Tufts University hold the patent for Open Pleth technology, which was licensed by Ambulatory Monitoring Inc, where AH serves as a consultant (<$2500 in 2008). These studies were not supported by any granting agency.
- 4Salbutamol Inhalation does Not Modify Airway Hyper-Reactivity in the Horse. 51 Annual Convention of the American Association of Equine Practitioners – AAEP, 2005 Seattle, WA, USA. Lexington, KY: American Association of Equine Practitioners; 2005, Internet Publisher: International Veterinary Information Service, Ithaca, NY. Available at: http://www.ivis.org. Accessed December 7, 2005, P2609.1205., , , et al.
- 15Exploratory Data Analysis, 1st ed. White Plains, NY: Addison-Wesley; 1977.