Experimental mild pulmonary inflammation promotes the development of exercise-induced pulmonary haemorrhage



Reasons for performing study: Histological studies of exercise-induced pulmonary haemorrhage (EIPH) reveals inflammatory infiltrates within areas of lung that contain haemorrhage. This has resulted in the hypothesis that pulmonary inflammation could cause EIPH or contribute to an increased risk or severity of EIPH.

Objectives: To determine whether experimentally-induced pulmonary inflammation predisposes the lung to haemorrhage during exercise, by evaluating the bronchoalveolar lavage (BAL) cytology of normal and inflamed regions of lung following high speed treadmill exercise.

Materials and methods: Transendoscopic airway inoculations of 0.01% acetic acid were used to induce mild localised inflammation within bronchopulmonary segments. Horses underwent high speed exercise 24 h after inoculation. Following exercise, inoculated sites and corresponding segments in the opposite lung underwent BAL. The cytology results from inflamed and control bronchopulmonary segments were compared, using paired t tests.

Results: Erythrocytes were present in BAL samples from 12.5% (1/8) control segments compared with 75% (6/8) inoculated segments following exercise, indicating a significant increase (P = 0.04) in the relative risk of EIPH following the development of pulmonary inflammation. Samples from inoculated segments had significantly higher percentages and numbers of neutrophils (12.1 ± 1.0% and 601 ± 98 cells/μl) than control samples (4.3 ± 0.3% and 214 ± 52 cells/μl). Significantly higher erythrocyte numbers were observed in samples from inoculated segments (14,304 ± 6862 cells/μl) compared with control samples (3.5 ± 3.5 cells/μl).

Conclusions: The results showed inflammation increased the risk of developing pulmonary haemorrhage during exercise. These findings do not conflict with current theories on the common causes of EIPH, but suggest that care should be taken when recommending exercise in horses suspected to be suffering from pulmonary inflammatory disease. In addition, specific therapy to reduce pulmonary inflammation may benefit horses prone to the development of EIPH.


Modern theories regarding the aetiogenesis of exercise-induced pulmonary haemorrhage (EIPH) suggest that it may occur without the presence of pulmonary inflammation (West et al. 1991; Schroter et al. 1998; Williams et al. 2008). These theories are able to account for the high prevalence of EIPH amongst racehorses that do not have clinically apparent pulmonary inflammatory disease. However, if pulmonary inflammation is not a prerequisite for the development of EIPH, then it must be concluded that the inflammatory lesions, so consistently observed in association with areas of grossly evident EIPH, are largely the result of the inflammatory response of the lung to the presence of blood (McKane and Slocombe 1999).

Cook (1974) suggested that the 2% prevalence of epistaxis he observed was a result of continuing to race horses that had suffered previously from pulmonary inflammatory disease. This theory was supported by the extensive study of O'Callaghan et al. (1987), which concluded that localised areas of bronchiolitis and fibrosis predisposed areas of the lung to EIPH because of changes and damage caused by inflammatory lesions. However, their study did not prove a causative relationship between the inflammatory lesions and EIPH, and the authors could not be certain that the inflammatory lesions reported were not the result of injury caused by the previous presence of intrapulmonary haemorrhage.

It is possible that the challenges to pulmonary vascular integrity during exercise resulting in EIPH may act synergistically with those that occur with pulmonary inflammation, leading to increase the severity of bleeding from vessels that have altered blood flow and vascular wall permeability. If this synergism was to be the case, then pulmonary inflammation would not only be a result of an episode of EIPH, but it may also predispose to additional episodes, particularly in cases where the horse was not permitted to convalesce for an appropriate period.

Acetic acid solutions have been used previously to experimentally induce localised inflammatory lesions in the airways and intestines of a number of species (Slocombe et al. 1984; Rolandelli et al. 1985). The use of 20 ml of 5% acetic acid in calf airways has been shown histologically to produce a localised mucosal injury, dominated by neutrophil infiltration, resulting in submucosal oedema, epithelial sloughing and small airway obstruction with cellular debris (Slocombe et al. 1984). Being a weak organic acid that is readily diluted and metabolised as itdiffuses into surrounding tissues, acetic acid is very suited to producing localised epithelial injury and inflammation. This study examines whether a milder localised inflammatory lesion, induced by 20 ml of 0.01% acetic acid predisposes an area of the lung to haemorrhage more readily than would otherwise be expected in response to exercise.

Materials and methods

Pilot study protocol

A pilot study was requested by the University of Melbourne animal ethics committee to provide evidence of the safety of acetic acid inoculation into horse airways. Following a 3 month rest from exercise, 2 Standardbred horses each had one segmental bronchus inoculated with 20 ml of sterile, 0.01% acetic acid solution (pH = 4.31) and a second segmental bronchus inoculated with 20 ml of sterile, 0.05% acetic acid solution (pH = 3.70). These dilute acetic acid solutions were inoculated in order to provoke localised pulmonary inflammation within the lower airways of the respiratory tract. The inoculations were performed using a 250 cm long catheter1 passed down the biopsy channel of a fibreoptic endoscope (Colonoscope LB3R-CF)2 to permit positive visual identification of the sites inoculated. As the endoscope was passed into the lungs, the tracheal bifurcation and junctions of the first lateral segmental bronchus of the caudal lobe of each lung were sprayed with 2 ml of 2% Mepivicaine3 to reduce the coughing reflex during bronchoscopy. The sites inoculated in each horse were the caudal branch of the first lateral segmental bronchus of the caudal lobe of each lung (Fig 1).

Figure 1.

Schematic diagram of the equine bronchial tree. L1.1 = cranial branch of first lateral segmental bronchus of caudal lobe; L1.2 = caudal branch of first lateral segmental bronchus of caudal lobe; L2 = second lateral segmental bronchus of caudal lobe.

Twenty-four hours later the horses were sedated with 300 mg of xylazine (Xylazil-100)4 administered i.v. and again had the tracheal bifurcation and junctions of the first lateral bronchi, as well as the second lateral bronchus of one lung (Fig 1), anaesthetised with 2 ml of 2% mepivicaine. These 3 bronchial segments were then each lavaged with 180 ml of sterile, 5% glucose solution infused via the biopsy channel of the endoscope and aspirated using 3–5 kPa continuous suction, until 60–100 ml of fluid was recovered. Smears were prepared by cytocentrifugation (Cytospin 2)5 at 200 g of 200 μl of each of the 3 samples from each horse and stained using a rapid Giemsa like stain (Diff Quik)6, before performing differential cell counts to determine the effects of inoculation with 0.01 and 0.05% acetic acid solutions on bronchoalveolar cell populations, compared with the sample from the uninoculated (control) segment.

The control segments (second lateral segmental bronchus) from both horses returned low neutrophil percentages (4.3 and 5.1%), which are consistent with normal bronchoalveolar lavage neutrophil percentages (Mair et al. 1987; McKane et al. 1994). The samples from the 4 inoculated segments contained neutrophil counts of 12.2 and 13.1% for the 0.01% acetic acid inoculation segments, and 15.4 and 16.9% for the 0.05% acetic acid inoculations. The macrophage and to a lesser extent the lymphocyte percentages were observed to decrease reciprocally to the increased neutrophil percentage. There were no erythrocytes seen grossly or during the cytological examination of any of the samples in this pilot study, demonstrating that neither the 0.01 nor 0.05% acetic acid inoculations resulted in pulmonary haemorrhage in unexercised horses.

Based on the results of these lavages it was decided that 20 ml of a 0.01% acetic acid solution would safely induce the desired localised pulmonary inflammation similar to the changes in BAL cytology seen 24 h after intrapulmonary blood inoculation (McKane and Slocombe 1999).

Main study protocol

Eight Standardbred horses ranging from 5–8 years of age that formed part of a research herd, which had all been rested from exercise for a period of 6 months, underwent clinical examination including thoracic auscultation during the application of a rebreathing bag and a tracheal wash to establish the absence of recent pulmonary haemorrhage and their fitness to take part in this study. All 8 horses had the second lateral segmental bronchus of the caudal lobe of one lung inoculated with 20 ml of 0.01% acetic acid, as described above. Which lung (right or left) to be inoculated was randomly assigned to each horse, so that in 4 subjects the right lung was inoculated and in the other 4 subjects, the left was inoculated with acetic acid. The inoculations were tolerated well by the horses with no adverse clinical effects observed, based on rectal temperature, resting heart rate and respiratory rate and effort.

Twenty-four hours after inoculation the horses were each subjected to treadmill7 exercise at 10 m/s on a 10% (6°) slope until fatigue, indicated by a failure to keep pace with the treadmill speed. After the cessation of exercise the horses were allowed a recovery period of approximately 90 min and then sedated and subjected to bronchoscopy and lavage of the inoculated segment and the corresponding second lateral bronchus of the caudal lobe of the opposite lung (control), as described for the horses in the pilot study. A recovery period of 90 min was chosen as it provided sufficient time for the horses to settle after exercise and is consistent with the protocols of other studies involving post exercise endoscopy of horses (Hinchcliff et al. 2005). At the same time as the lavage samples were obtained, a venous blood sample was also obtained and stored in a chilled tube containing lithium heparin to enable harvesting of plasma for urea nitrogen determination.

Sample handling

Total leucocyte counts were determined from each sample using a standard haemocytometer (Levy Double Neubauer counting chamber)8 and differential cell counts determined from cytocentrifuge smears prepared in the same manner as described for the pilot study. A 10 ml aliquot of lavage fluid was also centrifuged at 600 g to remove cells and mucus before analysis, to determine the lavage urea nitrogen concentration, and the plasma/lavage urea nitrogen ratio determined to provide the dilution factor of epithelial lining fluid (ELF) with the 5% glucose lavage solution.

Statistical analysis

The mean ± s.e. for the differential cell counts and the absolute cell counts/μl of ELF are presented in Tables 1 and 2, respectively. The data for each cell count was compared between inoculated and control segments using 2 sample t tests, conducted at the 95% confidence level. An analysis of the relative risk of developing EIPH following provocation of pulmonary inflammation with 0.01% acetic acid was also conducted using Fisher's exact test. A commercial statistical package9 was used for all analyses with the P value set at 0.05. All procedures described in this report were approved by the University of Melbourne animal ethics and research committee prior to the start of the study.

Table 1. Differential cell count of bronchoalveolar lavages obtained from 0.01% acetic acid inoculated and control segmental bronchi following strenuous exercise
Cell type %Acetic acidControlP value
  • a

    Percentage of leucocyte population.

  • b

    b Percentage of macrophage population containing phagocytosed erythrocytes.

  • c

    c Percentage of all cells present.

  • *

    *Mean is different to control mean.

Macrophagea59.6 ± 1.5*67.3 ± 1.20.0001
Lymphocytea27.8 ± 0.927.5 ± 1.00.9
Neutrophila12.1 ± 1.0*4.3 ± 0.30.0001
Eosinophila0.4 ± 0.10.6 ± 0.10.1
Mast cella0.3 ± 0.10.5 ± 0.10.3
Erythrophageb1.4 ± 0.4*0.0 ± 0.00.003
Erythrocytec53.6 ± 13.0*0.25 ± 0.250.001
Table 2. Absolute cell count of bronchoalveolar lavages obtained from 0.01% acetic acid inoculated and control segmental bronchi following strenuous exercise
Cells/μl ELFAcetic acidControlP value
  • *

    Mean is different to control mean.

Total leucocyte5032 ± 8554706 ± 9130.8
Macrophage3040 ± 5713162 ± 6240.9
Lymphocyte1369 ± 2071286 ± 2370.8
Neutrophil601 ± 98*214 ± 520.004
Eosinophil19.7 ± 6.531.0 ± 9.20.3
Mast cell14.8 ± 5.224.1 ± 6.50.28
Erythrophage40.8 ± 14.0*0.0 ± 0.00.01
Erythrocyte14,304 ± 6862*3.5 ± 3.50.05


Overall the subjects of the study tolerated the interventions very well and none of the inoculation procedures conducted in this study provoked clinically detectable changes in the respiratory systems of any of the subjects, with respect to fever, coughing, nasal discharge or noticeably increased respiratory effort. Only 12.5% (1/8) of the uninoculated bronchopulmonary segments lavaged following exercise yielded a BAL sample that contained any erythrocytes (2.9% of cells recovered). However, 75% (6/8) of acetic acid inoculated segments were found to contain substantially more erythrocytes (>39% of cells recovered), following the same episode of exercise. Under these experimental conditions, pre-existing airway inflammation of this type was 6 times more likely to promote the development of EIPH during intense exercise, significant at the 95% confidence level (P = 0.04). Grossly there was no visible blood in the trachea of any of the 8 horses. The gross appearance of the BAL samples from the 6 acetic acid inoculated regions that bled during exercise was that of obvious red colour; however, the one control segment that was cytologically positive for erythrocytes had a grossly normal appearance, reflecting the much lower erythrocyte content.

The percentage of neutrophils in the control samples (4.4 ± 0.3%) was consistent with normal values for BAL samples, and was significantly different to the elevated neutrophil percentage (12.3 ± 1.1%) in samples from inoculated segments (Table 1). This relative elevation in the numbers of neutrophils within the epithelial lining fluid was associated with an increase in the absolute cell count of neutrophils within the ELF also, indicating a true influx of neutrophils into the airways in response to the damage induced by the presence of dilute acetic acid, rather than simply an alteration in the volume of ELF present. This inflammatory response appeared to be well localised within the lung and certainly inoculation of a segment of one lung with 20 ml of either 0.01 or 0.05% of acetic acid solution did not produce an apparent change in the neutrophil percentage of control sites in either the same or contralateral lung in the subjects used in the pilot portion of this study.

The increase in neutrophil percentage 24 h following acetic acid inoculation was reciprocated by a significant (P<0.0001) reduction in the percentage of macrophages in inoculated vs. uninoculated segments. However, this was not associated with a decrease in the absolute numbers of macrophages in the ELF between the inoculated and control segments, indicating no net change in the ELF concentration of macrophages but rather a relative decrease in macrophage numbers compared to the increased numbers of neutrophils in inoculated segments. Percentages and absolute cell counts of lymphocytes, eosinophils and mast cells did not differ significantly between treatments (Tables 1, 2). These data support the conclusion that the change in neutrophil numbers in inoculated segments is due to an influx in cells and not simply to an alteration in the resident volume of ELF in these bronchopulmonary segments.


The use of acetic acid to induce epithelial damage and inflammation is well documented in the airways and intestines of a number of species (Slocombe et al. 1984; Rolandelli et al. 1985). In this study it was used primarily because of one of the investigators had previous experience with safe use and efficacy of acetic acid to produce a localised lesion. Unfortunately, economic and ethical constraints meant that histological examination of the inoculated lung segments was not possible in this study. However, the histological characteristics of acetic acid injury to the airways have been described previously (Slocombe et al. 1984). Notably, the descriptions of acetic acid injury indicate that it does not produce significant haemorrhage at the site of injury. Also in the 4 lung segments inoculated as part of the pilot study in this report, there were no erythrocytes recovered in the BAL samples from any segment, even when a concentration of 0.05% acetic acid was administered. In contrast, 6 of the 8 lung segments inoculated with 0.01% acetic acid solution, in horses that were then exercised 24 h later, haemorrhage was evident on gross appearance of the BAL samples, despite these being from regions of the lung not expected to haemorrhage during exercise in horses undergoing spontaneous EIPH.

In this study, the control was BAL samples taken from segments of the lung that had not been inoculated. Reports in the literature demonstrate neutrophil influx into the small airways in response to the inoculation of a number of liquids, including 0.9% saline and autologous blood (Sweeney et al. 1994; McKane and Slocombe 1999). It was therefore decided that the best control in this study were lung segments that had not been inoculated with anything. It is also interesting to note that in this study there was no corresponding increase in the neutrophil percentage in the contralateral lung as demonstrated in the study by Sweeney et al. (1994), despite the apparently greater intensity of inflammation induced by 0.01% acetic acid than by 0.9% saline. Indeed, it has been the authors' experience in this and other studies evaluating the inoculation of bronchopulmonary segments in horse lungs (McKane and Slocombe 1999, 2002) that the response to inoculation is more reflective of isolated units rather than a coordinated global response across both lungs.

The results of this study show that there is an association between the development of pulmonary haemorrhage during exercise and the existence of prior airway inflammation. Of the 6 inoculated bronchial segments that haemorrhaged, the minimum percentage of erythrocytes observed was 39%; however, in the one control lavage that contained erythrocytes, this was only 2.9% of the total cell count and it may be that this haemorrhage arose as a consequence solely attributable to the lavage procedure and unrelated to haemorrhage associated with exercise.

The percentage of neutrophils in the control samples was consistent with normal values for BAL samples (Mair et al. 1987; McKane et al. 1994), and was significantly different to the elevated neutrophil percentage (12.3 ± 1.1%) in samples from inoculated segments. This inflammatory response, 24 h following inoculation with 0.01% acetic acid, was similar to the rise in neutrophil percentage observed in the pilot study horses and in samples collected 24 h after intrapulmonary blood inoculation in other studies (McKane and Slocombe 1999). The increase in neutrophils suggests that 0.01% acetic acid solution provokes a mild inflammatory reaction within the lower airways of the lung, similar to that caused by the presence of blood. The fact that this level of inflammation was sufficient to predispose the lung to EIPH has important implications for horse management. Neutrophil percentages of approximately 10% are not uncommon in horses undergoing training (McKane et al. 1993; McKane and Rose 1995), and lavages from horses with clinically mild respiratory disease may have neutrophil percentages as high as 25–30% (Fogarty and Buckley 1991). If these clinically silent levels of pulmonary inflammatory disease have the same ability to predispose the equine lung to EIPH as the inflammation caused by the inoculation of 0.01% acetic acid, then the role of inflammatory disease in the aetiology of EIPH has been greatly underestimated in recent years. Even if these situations are not directly comparable, there is obvious reason for equine trainers and veterinarians to reconsider the practise of continuing to exercise a horse that has evidence of respiratory disease.

There is some evidence in the literature that might be seen to conflict with the findings of this study in the declining prevalence of inflammatory airway disease (IAD) with age (Wood et al. 2005) as opposed to the increasing prevalence of EIPH in older horses (Mason et al. 1983). The present study would suggest that IAD should be important in the development of EIPH in young racehorses. However, that does not mean that other forms of inflammation are not also important in older horses and certainly it is not suggested that inflammation is prerequisite for the development of EIPH in all horses. The results of this study support the idea that the inflammatory response provoked by previous episodes of EIPH could play a role in the development of new and perhaps more severe episodes of haemorrhage. This would be particularly so if horses are strenuously exercised within 3–7 days of an episode of EIPH. The ability of previous episodes of EIPH to increase the severity of future EIPH agrees with the epidemiological data, which indicate an increase in prevalence and severity of EIPH with the increasing age of racehorses (Mason et al. 1983; McKane et al. 1993).

In this study, the segments inoculated were located in the cranioventral regions of the caudal lobe, which are generally considered not to be involved in EIPH. These segments were chosen partly because they were expected to be free from the effects of previous EIPH and because they are more easily accessible to bronchoscopic examination. This study proves that under circumstances where inflammatory disease exists in the cranioventral pulmonary segments, then sufficient forces must be present to induce haemorrhage during strenuous exercise. Racehorses may haemorrhage into the cranioventral lung segments during exercise, although the prevalence of this must be low enough that the distribution of lesions commonly ascribed to EIPH (O'Callaghan et al. 1987) are far more subtle and infrequent than those observed in the caudodorsal regions of the lung. Whether the distribution of inflammatory lesions, their severity and rates of resolution have regional differences in equine lungs remains unknown, but it would seem that the principle determinant of haemorrhage in the caudal lung with EIPH is related to factors other than low grade pulmonary inflammatory disease.

In conclusion, this study demonstrates that at least certain forms of inflammation can promote pulmonary haemorrhage in response to exercise in horses. It is not clear whether this haemorrhage is necessarily from the same vessels as are typically involved in EIPH and unfortunately sampling of the lungs of these horses for histological sectioning was not possible. Small volume (40 ml) inoculation of bronchopulmonary segments with autologous blood is known to fill dependent alveolar spaces with erythrocytes within 30 min of inoculation (McKane and Slocombe 2002) and so it is therefore likely that the inoculated acetic acid solution also rapidly reached the bronchioles and alveolar spaces. However, even if inflammation produces bleeding from a different location (for example bronchiolar and not alveolar sites), it would add to the volume of haemorrhage present in the lower airways, and this is important as previous studies have shown that the adverse affects of EIPH on athletic performance are proportionate to the volume of blood in the airways (Kingston et al. 2002; Hinchcliff et al. 2005; McKane et al. 2008). Hence, there may be important implications for horses that undergo intense exercise whilst they are affected by inflammatory disease affecting distal bronchpulmonary segments.

Manufacturers' addresses

1 Critchley Electrical Products, Auburn, New South Wales, Australia.

2 Olympus, New Hyde Park, New York, USA.

3 Nature Vet, Agnes Banks, New South Wales, Australia.

4 Troy Laboratories, Smithfield, New South Wales, Australia.

5 Shandon Southern Products, Astmore, Cheshire, UK.

6 Lab-Aids, Narrabeen, New South Wales, Australia.

7 Beltalong, Euroa, Victoria, Australia.

8 Becton Dickinson, Parsippany, New Jersey, USA.

9 Sigma Stat, San Jose, California, USA.

Conflicts of interest

The authors have not declared any potential conflicts.