Clinical, radiographic, endoscopic, bronchoalveolar lavage and lung biopsy findings in horses with exercise-induced pulmonary hemorrhage

Abstract

A clinical study was performed to determine whether clinical, endoscopic, radiographic, bronchoalveolar lavage (BAL) cytological, and pulmonary biopsy findings could be correlated in horses with exercise-induced pulmonary hemorrhage (EIPH) compared with controls. Racing standardbred horses were selected as either EIPH (n = 10) or control (n = 10), based on repeated postexertional endoscopy of the lower airways. Complete physical and respiratory examinations were performed and blood samples were submitted for arterial blood gas analysis, hematologic study, and fibrinogen determination. Bilateral chest radiographs were taken with the horse standing, and a BAL sample was obtained for cytological examination. Lung was biopsied transcutaneously. Weighted scores were calculated for clinical, radiographic, and pulmonary biopsy findings. The conclusion was that only routine physical examination may help the clinician when EIPH is suspected in horses, especially when there are abnormal findings on percussion of the caudodorsal areas of the chest.

Introduction

Exercise-induced pulmonary hemorrhage (EIPH) was first described in racehorses by Pascoe et al (1) in 1981, based on their endoscopic observations of blood in the lower airways following exertion. Since then, endoscopic field surveys have shown the prevalence of EIPH to be as high as 50% or more in most types of equine athletes throughout the world (2,3,4,5,6,7,8). Despite the worldwide distribution of EIPH and the high prevalence of the condition, there have been few field investigations of the clinical signs, radiographic findings, and pulmonary cellular populations in horses with EIPH. A survey of 26 racing Thoroughbreds in Hong Kong with a known history of EIPH revealed that only 4 horses had clinical signs referable to the respiratory system (9), although these findings were not compared with those in a control group of EIPH-negative horses. Another report concluded that horses with EIPH could not be distinguished from controls on the basis of physical examination alone (10); however, this statement was subjective and based on the past experience of the author rather than on controlled field trial results. Furthermore, hematologic and clinical biochemical findings from horses with EIPH showed statistically and clinically insignificant changes compared with those from EIPH-negative horses (10), although another study revealed that horses with EIPH demonstrated a trend towards lower platelet counts than did control horses (11).

A small number of investigators have attempted to characterize the radiographic changes that may accompany EIPH (12,13,14,15). Unfortunately, all of these reports have suggested that due to the large variability in findings combined with the inherent difficulty in assessing lesions in the caudodorsal areas of the lungs, standard radiographs were a poor means of identifying EIPH cases.

Microscopic pathological changes have been observed and described in horses with a history of EIPH (15,16,17,18,19). Although no control horses were used in these studies, it appears that EIPH can eventually lead to bronchiolitis, chronic inflammatory disease of small airways, and proliferation of bronchial arterial vessels around the affected airways. Also, some eosinophilic infiltrates and hemosiderin deposits were identified in advanced cases (15,16,17,18,19).

Until now, the detection of EIPH has been made best by endoscopic examination of the tracheobronchial airways, 30 to 90 min postexercise, and at least 3 such endoscopic examinations are required to rule out EIPH (10). However, although this method is quite specific for EIPH, there is some controversy as to its sensitivity, and many false negatives are thought to occur. The use of other clinical diagnostic methods may help to improve sensitivity in further defining EIPH-positive horses, clinically.

The objective of this study was to determine if EIPH-positive horses, detected by using postexertional endoscopy alone, could be further distinguished from control horses, based on clinical examination, routine clinical pathologic analyses, endoscopic and radiographic findings, bronchoalveolar lavage (BAL) cytologic examination, and pulmonary biopsy lesions.

Materials and methods

Experimental animals

A control horse was defined as a horse with no history of EIPH and no blood in the tracheobronchial airways on 3 consecutive postexercise endoscopic examinations. An EIPH-positive horse was defined as a horse that had shown blood in the tracheobronchial airways at least once on postexertional endoscopic examination, regardless of the volume of blood present. Subjects were selected among a population of currently racing or race-fit standardbred horses from 2 racetracks in southern Ontario. All horses were studied under the guidelines of the Canadian Council on Animal Care and study protocols were approved by the University of Guelph Animal Care Committee.

The sample population consisted of 10 standardbred racehorses in each group. The control group, composed of 5 mares and 5 geldings, had a mean age of 3.3, s = 0.7 y with a range of 2.5 to 5 y. Horses in the EIPH group had a higher mean age of 4.5, s = 1.1 y, ranging from 3 to 6 y, and included 3 mares, 4 geldings, and 3 stallions, although these differences in age were not statistically significant. All of the control horses and most of the EIPH horses were pacers, while 2 EIPH horses were trotters. A complete medical history was obtained prior to selection, including whether there had been past respiratory problems, whether epistaxis had been noted, or whether a previous diagnosis of EIPH had been made. Husbandry information and the trainer's evaluation of the horse's recent performance were recorded before each examination, along with the horse's race time and the track conditions for that day. Endoscopic examinations were performed by using a 180-cm fiberoptic endoscope (Olympus CF-LB2; Olympus, Tokyo, Japan), equipped with a cold light source (Olympus CLK-3; Olympus), and inserted through the nares and passed down to the carina within 30 to 90 min of a race or a hard training session. The horses were restrained by using only a nose-twitch. The presence of blood in the tracheobronchial airways was noted, as well as any other abnormal endoscopic findings.

Performance data

Race lines were obtained retrospectively for each horse from the Ontario Racing Commission computerized records. In these records, both official and qualifying race starts are tabulated. The number of official race starts in the 6 mo prior to testing, as well as the total number of starts (including qualifying races) in the 2 mo prior to testing were noted for each individual in the study. The horse's best overall career time, as well as its most recent (within 2 mo of the investigation) best time and the average time in the last 5 starts previous to testing, was recorded.

Experimental protocol

Within 10 d of the last endoscopic examination, horses were admitted to the Ontario Veterinary College, Veterinary Teaching Hospital for day 1 of data collection. All subjects were kept in well-ventilated isolation stalls. At least 1 h after arrival, a physical examination was performed, venous blood samples were taken for hematologic examination and fibrinogen assessment, and an arterial blood sample was taken for blood gas analysis. Thoracic radiographs were then taken before the BAL sample was obtained. The next morning (day 2), pulmonary mechanics were measured and histamine inhalation challenges were performed, the results of which are in preparation for publication. At the end of day 2, the lung was biopsied percutaneously and the horse was returned home. All physical examinations, measurements of pulmonary mechanics, histamine inhalation challenges, BALs, and pulmonary biopsies were performed by the same investigator (MYD), who was unaware of the identity of the subjects.

Clinical examination

The thorough physical examination given on day 1 paid particular attention to the respiratory system. A rebreathing bag was placed over the horse's nose to favor deeper breaths, thereby facilitating auscultation of the lungs; then, the thorax was percussed. A weighted clinical score was derived for each subject, based on information from the history, physical examination, and endoscopic findings in the upper and lower airways at rest (Table 1). The clinical scores were calculated with and without historical data in order to compare the relative usefulness of each score in the detection of horses with EIPH. The maximum possible clinical score was 30 when all of the parameters presented in Table 1 were assessed and 23 when historical parameters were removed from the score.

Table 1.

Clinical pathology

Blood samples taken on day 1 were submitted for arterial blood gas analysis, hematologic study, and fibrinogen determination. Resting arterial blood gases were obtained from the carotid artery, collected in heparinized syringes, and analyzed using a Radiometer ABL3 blood gas instrument (Radiometer Corp., Copenhagen, Denmark). Complete blood cell (CBC) counts and differentials were performed on EDTA-blood samples taken from the jugular vein by using an automated cell counter (Coulter Counter ZM; Coulter Electronics, Burlington, Ontario). Sodium citrated samples were submitted for fibrinogen (FibroSystem; BBL, Cockeysville, Maryland, USA).

Thoracic radiography

The bilateral standing thoracic radiographs on day 1 were taken according to conventional methods by using a 10-foot (3-m) film focal distance and an 18-inch (45.7 cm) air gap. The collimator was set at 140 kV and 35 mA. All radiographs were interpreted by a radiologist, who was unaware of the identity of the subjects. The following 6 lesions were assessed for each side of the thorax: bronchial wall thickening, generalized interstitial pattern, cranial and caudal focal interstitial pattern, and cranial and caudal focal air space filling (or alveolar pattern). For each type of lesion, a score was assigned as follows: unequivocal presence of the lesion (score of 2), unequivocal absence of the lesion (score of 0), or equivocal observation (not confirmed present or absent) of the lesion (score of 1). The radiographic score was then calculated by adding the scores for each lesion and for each side of the thorax. Since 6 types of lesion were assessed with a possible maximum score of 2 each, the highest possible radiographic score was 12 for each side of the thorax, so the maximum possible radiographic score for each horse was 24.

Bronchoalveolar lavage

Horses were sedated with xylazine (0.2 to 0.5 mg/kg bodyweight (BW) IV) and restrained in a stock. A 13-mm diameter, 180-cm fiberoptic endoscope (Olympus CF-LB2; Olympus) equipped with a cold light source (Olympus CLK-3; Olympus) was inserted through the nares and 60 mL of a 0.2% lidocaine solution was sprayed into the larynx and carina to reduce coughing. The endoscope was then passed into the left main stem bronchus and wedged into a 3rd or 4th generation caudodorsal bronchus. The BAL was performed with 2 consecutive 250-mL aliquots of warm (body temperature), sterile physiologic saline infused under pressure by using a hand pressure ball. The samples were immediately aspirated into a glass flask by using a suction pump (Medi-pump, Power Air Division, Sheboygan, Wisconsin, USA) with a pressure set between 8 and 10 cmH₂O. Total cell counts were obtained from an automated cell counter (Coulter Counter ZM; Coulter Electronics). Twenty milliliters of the sample were centrifuged at 1500 × g for 10 min, then the supernatant was removed and direct smears were made using a resuspension of the cell pellet in 0.5 mL of the supernatant. Cytocentrifuge slides were also prepared (Cytospin 2; Shandon Southern Products, Astmoor, United Kingdom). For differential cell counts, all slides were stained with Wright-Giemsa stain by using an automated stainer (Hema-Tek Slide Stainer; Miles Laboratories, Rexdale, Ontario). Differential cell counts were obtained by counting 500 cells/slide from direct and cytocentrifuge smears. The evaluator was unaware of the identity of the subjects when reading any of the BAL slides. A previously reported set of normal BAL cytology values from a similar population of horses was used as a “normal” reference for both groups (20). This historical control group was composed of 12 horses (6 females, 3 males, and 3 geldings), with a mean age of 3.4, s = 1.6 y in full training or racing condition.

Percutaneous pulmonary biopsy

On day 2, 3 to 4 pulmonary biopsies were obtained from the right lung of each subject using the technique described by Raphel and Gunson (21). Horses were sedated with xylazine (0.2 to 0.5 mg/kg BW, IV) and restrained in stocks. A small area in the 7th or 8th intercostal space, approximately 10 cm above the level of the point of the elbow, was clipped and prepared aseptically. The skin, intercostal space, and parietal pleura were anesthetized by using 10 mL of 2% lidocaine hydrochloride. An 11.4-cm, 14-gauge biopsy needle (Tru-cut; Travenol Laboratories, Deerfield, Illinois, USA) was introduced through the intercostal space and directed dorsally, medially, and ventrally in succession to obtain several samples of lung. The samples were fixed in 10% formalin and slides were prepared following standard histological techniques and stained with hematoxylin and eosin. A biopsy score based on a previously published histological scoring system was computed (22). Ten histologic parameters including airway contraction, goblet cell hyperplasia, smooth muscle hypertrophy, and inflammatory cell infiltration were assessed for their absence (0) or presence, as mild (1), moderate (2), or severe (3) changes. The total possible score was 30. A minimum of 3 bronchioles were used to establish each individual horse's score. The evaluator was unaware of the identity of the subject when reading the slides.

Statistical analysis

Population descriptors and clinical parameters represented by continuous variables such as race times, age, temperature, pulse, respiratory rate, and hematologic and arterial blood gas parameters were compared between groups by using a 2-sample t-test after verifying normality of the data with the Wilk-Shapiro statistic. Clinical, radiographic, biopsy, and hemosiderin scores, as well as BAL, fibrinogen, and blood differential cell counts were compared between groups by using the Wilcoxon rank sums test, a nonparametric analysis of variance. Associations between clinical, radiographic, and biopsy scores and other clinical parameters were assessed using the Spearman's rank correlation coefficient. Fisher's exact test was applied to determine whether radiographic lesions in the caudodorsal pulmonary regions were associated with percussion abnormalities in the same areas. A level of significance of P < 0.05 was used for all statistical tests (Statistix v. 4.0; Analytical Software, Tallahassee, Florida, USA).

Results

Sample population

Based on their race lines, the average best times for a one-mile race in the 2 mo prior to the investigation were mean 2:00.8, s = 0:05.6 min and mean 1:59.6, s = 0:06.5 min for the control and EIPH horses, respectively. There was no significant difference in any of the racing performance parameters between the 2 groups.

Clinical examination and endoscopic findings

The control group horses had a mean and range for temperature, heart rate, and respiratory rate of 37.8°C, s = 0.5 (37.2°C to 38.5°C), 38.4, s = 4.3 (32 to 48) beats/min, and 15.6, s = 4.4 (12 to 24) breaths/min, respectively. Horses in the EIPH group had a mean temperature of 38.1°C, s = 0.6 (range 37.1°C to 38.9°C), a mean heart rate of 40.4, s = 8.5 (range 28 to 56) beats/min, and mean respiratory rate of 18.4, s = 5.4 (range 12 to 28) breaths/min. There was no statistically significant difference between groups for any of these vital signs.

There were no detectable abnormalities on physical examination for the control group, except for 1 horse, where the auscultation and percussion findings were abnormal, whereas 5 (50%) horses in the EIPH group had abnormal findings on percussion of the caudodorsal areas of the thorax. Although most individuals in both groups with abnormal percussion findings had dullness of these regions on percussion of one or both sides of the chest, hyperresonance was also detected in 2 cases. Only 2 horses in the EIPH group and no horses in the control group had evidence of abnormal lung sounds on auscultation, even when a rebreathing bag was used. These abnormalities were limited to increased lung sounds on the right side of the chest for one horse and end expiratory wheezing for the other. Resting endoscopic examination of the upper and lower airways revealed various degrees of follicular lymphoid hyperplasia in 6 of the control subjects, while hyperemia of the airways was observed in 6 control and 4 EIPH horses. Mild to moderate mucous discharge was seen in 2 individuals from each group.

The EIPH group had a significantly higher mean weighted clinical score than did the control group when the score included historical data but not when historical information was removed from the score (Table 2).

Table 2.

Clinical pathologic findings

Mean group results for arterial blood gas analysis are given in Table 3. There was no significant difference between groups for any of the parameters reported. Table 3 also summarizes the mean values for hematologic parameters and fibrinogen levels for each group. Horses in the EIPH group had significantly lower platelet counts than did the control horses. There were no significant differences between groups for any of the other hematologic parameters.

Table 3.

Radiographic findings

Radiographs were not available for 2 horses in the study (1 control, 1 EIPH) due to equipment repairs at the time of the investigation. There was considerable intersubject variability in radiographic scores for both groups. Visible lesions in the caudodorsal areas of the lungs were not limited to the EIPH horses. The most commonly observed radiographic lesions for either group were bronchial thickening (mean scores for this lesion for both sides of the thorax combined (maximum possible score of 4) were 2.9/4 and 2.4/4 for control and EIPH horses, respectively) and generalized interstitial patterns (mean scores of 2.6/4 and 2.4/4, respectively, for control and EIPH groups). Focal lesions were noted only sporadically, and all the mean scores for both lungs combined were found to be less than 1.4/4 for both groups.

There was no significant difference between mean radiographic scores for both groups (Table 2), nor were there any significant differences between groups for the mean individual radiographic parameter scores.

Bronchoalveolar lavage cytology

Fluid recovery from the BAL procedure ranged from 100 to 300 mL for all the horses in the study with a mean of 205, s = 41.7 mL and of 214, s = 68.3 mL for control and EIPH groups, respectively. Total cell counts were not statistically different between the 2 groups and remained within normal range. Based on differential counts obtained from direct smear preparations, EIPH horses had a significantly higher percentage of eosinophils in their BAL fluid than did the control horses (Table 4). There was also a tendency for EIPH horses to have higher epithelial cell counts than did control horses, although this difference was not significant (P = 0.07). There were no differences for the other cell types between EIPH and control groups when overall group means were compared. There were, however, several individuals with abnormal differential cell counts in both groups of horses.

Table 4.

Compared with a population of standardbred racehorses of similar age, husbandry, and fitness level (n = 11) (20), most horses (90% of controls and 70% of EIPH horses) had increased lymphocyte counts and lower macrophage levels. Two control and 2 EIPH subjects had elevated neutrophil counts. Mast cell counts were higher than normal in 5 control and 4 EIPH horses and eosinophils were elevated in 1 control and in 3 EIPH horses. Epithelial cells were increased in 2 control animals and in 6 EIPH cases.

Pulmonary biopsy findings

Due to lack of owner's consent, pulmonary biopsies were not obtained for 1 control and 2 EIPH horses. Mean total biopsy scores for each group are given in Table 2. Peribronchiolar and perivascular eosinophilic infiltration was noted in 3 control horses and in 3 EIPH horses.

The total biopsy scores were not statistically different between groups, but there was considerable intersubject variability in both EIPH and control groups. Mean scores for airway contraction were found to be significantly greater in EIPH than in control horses, while mean scores for goblet cell hyperplasia, smooth muscle hypertrophy, and inflammatory cell infiltration tended to be higher in the EIPH group as well, although these were not significant (P = 0.12, P = 0.40 and P = 0.30, respectively). None of the other histological parameters were found to be significantly different between groups.

Correlations between clinical, radiographic, and biopsy scores and other findings

There was no significant correlation between the weighted clinical score and either the overall radiographic or the biopsy scores in both groups. Also, the weighted clinical score did not correlate significantly with the individual radiographic findings in either group of horses. There was also no significant correlation between clinical scores and arterial blood gas, CBC counts, fibrinogen levels, or BAL parameters in either group.

Radiographic scores were not significantly associated with biopsy scores or with any of the BAL cell counts in either group, except for mast cells in the control group, which were significantly correlated with radiographic scores (Spearman's rank correlation coefficient = 0.76; P = 0.02). Biopsy scores were not significantly correlated with BAL cytologic results in either group.

Finally, abnormal findings on percussion of the caudodorsal areas of the chest were not significantly associated with focal radiographic changes in the same areas in horses with EIPH or in control horses.

Discussion

The population of subjects used in this study was composed of a representative sample of young standardbred racehorses. Differences in age, gender, and gait were not statistically significant, and it is not known whether certain trends affected the outcome of the results. Based on race line information, the majority of subjects could be considered to be at the top performance level in Ontario. Although there was no statistically significant difference between the 2 groups of horses, the race line results seemed to indicate that the horses in the EIPH group tended to race more frequently in the period preceding the investigation and to have faster times than did controls. The EIPH horses tended to be older than were the control horses as a group and this may explain their more frequent use. Some previous studies have noted that EIPH incidence increases with age (1,2,3,4,7,23,24), while others failed to show such a relationship (7,24). The horses in the EIPH group may have developed this condition due to the inability of the lung to adequately repair regions damaged by continuous racing and training, and to the increased risk of small airway disease with increasing age due to chronic exposure to poor environmental conditions at the race track. However, 2 horses in the EIPH group, a 6-year-old and a 3-year-old, which had not been racing regularly but which bled repeatedly upon exertion, and 3 horses from the control group aged 5, 3, and 4 y, respectively, which had more than 14 starts in the 6 mo prior to testing but who did not show EIPH, perhaps provide evidence that factors other than frequent racing and increasing age need to be considered as predisposing factors to EIPH.

Only 2 horses in the EIPH group had abnormal sounds on auscultation of the lungs indicating, as others have found (9,10), that auscultation is not a useful means of detecting horses with EIPH. However, 50% of the horses in this group had abnormal percussion findings in the caudodorsal regions of the thorax compared with only 1 horse in the control group, thus suggesting that percussion of the thorax could be a useful diagnostic technique. In fact, it appeared that clinical scores (with and without historical data) observed in EIPH horses, but not in controls, were related to abnormal findings on percussion of the caudodorsal areas of the thorax (Spearman's rank correlation coefficient = 0.54; P = 0.11). Although this association was not statistically significant, it suggests that 54% of the difference between clinical scores is due to these percussion findings. Furthermore, reliability and repeatability of the percussion technique of the thorax was not assessed in this study. Further studies are needed to assess these parameters before this method can be used routinely to detect horses with EIPH. Hyperresonance in the caudodorsal areas of the lung field, which was found in 3 horses with EIPH, could be attributed to air trapping as a consequence of small airway obstruction. Dullness in this region, detected in 3 EIPH horses and 1 control horse, could be caused by inflammation of the pulmonary tissue again due to small airway disease or to increased density in pulmonary tissue caused by blood vessel proliferation and parenchymal fibrosis (18). It could not be determined from this study whether the small airway disease was primary or secondary to EIPH. Finally, the detection of abnormal findings on percussion was not associated with abnormal radiographic findings in the same regions of the chest. This suggests that pulmonary lesions responsible for abnormal percussion findings are not necessarily visible on chest radiographs in these horses.

Resting endoscopic changes observed in the airways of horses from both groups did not reveal significant abnormalities that could be attributed to EIPH. Follicular lymphoid hyperplasia, airway hyperemia, and mucoid discharge may all be secondary to persistent airway inflammatory processes, caused possibly by poor environmental conditions and frequent exposure to respiratory viruses at the racetracks, although these assessments remain highly speculative.

The weighted clinical score appeared to allow a better distinction between control and EIPH horses only when historical information was included in the score. This is not surprising, since all horses in the EIPH group had shown endoscopic evidence of postexertional bleeding and, by design, the results of this study were biased regarding the historical data. Prospective studies are needed to assess the value of the weighted clinical score, with and without historical data, on a larger population of horses where the EIPH status is unknown prior to scoring.

The normal arterial blood gas and CBC count parameters noted in the 2 groups of horses indicate that there was no serious underlying respiratory disease in any of the sample subjects. The lack of significant difference in these blood tests between the 2 groups supports the observations of other authors who have stated that hematological and clinical biochemistry findings were not useful in detecting horses with EIPH (10). To our knowledge, arterial blood gas findings for EIPH horses compared with controls have not previously been reported. O'Callaghan et al (25) suggested that horses with EIPH had ventilation-perfusion deficits that correlated with radiographic opacities in the affected areas of the lung. Unfortunately, their study did not include a control group of EIPH-negative horses. The clinical significance of a potentially lower PaO₂ and ventilation-perfusion deficits in such a localized area, as well as their impact on performance, is unknown.

Blood platelet counts were significantly lower in EIPH than in control horses in this study, while fibrinogen levels were slightly lower, though this difference was not statistically significant and may be due to the fact that 50% (5/10) of the control horses, but only 20% (2/10) of the EIPH horses, had fibrinogen levels above the normal range for our laboratory. All platelet counts were found to be within the normal range in both groups. It could not be determined from this study whether these differences were related to a chronic source of inflammation in some control horses, or to an increased fibrinolysis and consumption of platelets in the damaged areas of the lungs, and to the possibility that horses with EIPH are predisposed to bleeding due to low platelet numbers and increased fibrinolysis. Johnstone et al (11) also reported a slight decrease in platelet counts in horses with EIPH compared with control horses, although the difference was not significant. Rather, their findings clearly indicated that there was no detectable enhanced fibrinolysis in horses with EIPH and that impaired platelet function rather than low platelet numbers was the most likely reason for this predisposition to bleeding in EIPH. Since most values were within the normal range, the results of this study suggest that the possibility of pathologic changes related to hemostasis is unlikely and that the significant difference in platelet levels may not be clinically relevant.

Bronchoalveolar lavage cytological findings indicated significantly higher eosinophil counts in horses with EIPH than in control horses. Eosinophilic infiltration has been reported in histologic preparations of caudodorsal lung lobes of horses with EIPH (18,19), although no explanation for these findings has been offered to date. Gunson et al (19) suggested that metalloproteins produced by these eosinophils could play a role in the degradation of collagen fibers, leading to damage of the blood vessels, but this assumption has not been verified in horses. Indirectly, eosinophils may be a marker of underlying allergic lung disease, or they could be secondary inflammatory cells migrating to the sites of bleeding and causing further damage via their mediators. Since neither the control nor the EIPH horses were tested initially for “normality” of the respiratory system and were chosen based solely on their EIPH status, a reference group of “normal” horses was chosen from a previous study to evaluate the BAL findings. Cell differentials in all cases in this study showed higher lymphocyte and lower macrophage percentages than in a group of “normal” horses used as a reference (20). These results, combined with the finding that several individuals from both groups showed abnormal percentages of mast cells, eosinophils, and epithelial cells, strongly suggest the presence of airway inflammation compatible with small airway disease. This again may be associated with exposure to barn dust and respiratory viruses, which are common in the racing environment and do not appear to be different between both groups of horses.

In this study, radiographic scores were too variable to allow significant differences to be found between EIPH and control groups. Radiographic densities in the caudodorsal regions of the lungs were seen in horses that bled, as well as in control horses, and there was no significant correlation between abnormal findings on percussion and focal abnormalities in the same areas on radiographs in either group, indicating poor sensitivity to detect EIPH cases. This study confirms findings by others (10,12,14,15) that interpretation of chest radiographs is a poor diagnostic tool for defining EIPH. Additionally, radiographic scores were not correlated with clinical or biopsy scores. The significant association between radiographic scores and mast cell counts on BAL in horses from the control group may have been a chance finding that has no significant clinical implication. Although high mast cell counts are an indication of type I allergic small airway disease and a generalized interstitial pattern could be present in cases of small airway disease, only 3 horses in the control group actually had mast cell counts above the reference normal values (20) and this does not constitute a large enough group of subjects for sound scientific conclusions to be made.

Finally, pulmonary biopsies did not reveal histological abnormalities in the lungs of horses with EIPH that were significantly different and would distinguish EIPH horses from control horses. Horses in both groups showed varying degrees of small airway disease, yet no subject had apparent severe or irreversible changes. Unfortunately, safety precautions involved with the percutaneous pulmonary biopsy procedure on standing horses did not allow sampling in the caudodorsal areas of the lungs, where previous histological investigations have been undertaken (18). Therefore, it comes as no surprise that previously published histological findings for horses with EIPH were not noted in this study. Nonetheless, in keeping with this study's objective, the pulmonary biopsies were performed as they would be in routine clinical investigations. Also, pulmonary biopsies obtained in the cranial portions of the lungs may represent the overall health status of the small airways and lung parenchyma as a whole more adequately. Based on these assumptions, localized small airway disease, secondary to EIPH cannot be ruled out in this study; however, it is interesting to note that horses in the EIPH group had more consistently elevated lung biopsy scores, whereas horses in the control group had either very low or very high scores. This suggests that horses with EIPH may consistently have moderate small airway disease lesions, while horses that do not bleed may or may not have small airway disease, depending on other triggering factors from the environment.

In conclusion, routine blood analyses, thoracic radiographs, BAL cell differentials, hemosiderin scores, and percutaneous lung biopsies did not clearly differentiate EIPH-positive from EIPH-negative horses in this study. Several parameters may have been more discriminatory had a more selective population of control horses been used, and it is not known whether age, gender, or gait may have had an impact on results. Nonetheless, this study suggests that routine physical examinations may be helpful as an initial method for screening suspected EIPH in horses, especially when abnormal findings on percussion of the caudodorsal areas of the thorax are noted. Prospective studies are needed to evaluate the diagnostic value of the weighted clinical scores presented in this report, and the repeatability or reliability of the chest percussion method in horses.