Racing Alaskan Sled Dogs as a Model of "Ski Asthma"
Michael S. Davis,
Brendan McKiernan,
Sheila McCullough,
Stuart Nelson, Jr.,
Ronald E. Mandsager,
Michael Willard and
Karen Dorsey
Department of Physiological Sciences, Department of Veterinary Clinical Sciences, Department of Pathobiology, Oklahoma State University, Stillwater, Oklahoma; Denver Veterinary Specialists, Wheat Ridge, Colorado; Department of Veterinary Clinical Medicine, University of Illinois, Urbana, Illinois; Iditarod Trail Committee, Wasilla, Arkansas; and Department of Small Animal Clinical Sciences, Texas A&M University, College Station, Texas
Correspondence and requests for reprints should be addressed to Michael S. Davis, Department of Physiological Sciences, 264 McElroy Hall, Stillwater, OK 74078. E-mail: msdavis{at}okstate.edu
Athletes who play sports in cold weather, particularly skatersand cross-country skiers, have an increased prevalence of lowerairway disease that is hypothesized to result from repeatedpenetration of incompletely conditioned air into the lung periphery.In this study, we investigated the hypothesis that canine winterathletes also suffer from increased prevalence of lung diseasesecondary to hyperpnea with cold air. Bronchoscopy and bronchoalveolarlavage was conducted in elite racing sled dogs 24 to 48 hoursafter completion of a 1,100-mile endurance race. Bronchoscopicabnormalities were classified as none, mild, moderate, or severe,based on the quantity and distribution of intralumenal debris.Eighty-one percent of the dogs (48 of 59) examined had abnormalaccumulations of intralumenal debris, with 46% (27 of 59) classifiedas moderate or severe, indicating significant accumulation ofexudate. Bronchoalveolar lavage obtained from dogs after therace had significantly higher nucleated macrophage and eosinophilcounts compared with sedentary control dogs. Our findings supportthe hypothesis that strenuous exercise in cold environmentscan lead to lower airway disease and suggest that racing sleddogs may be a useful naturally occurring animal model of theanalogous human disease.
Epidemiologic surveys of human cold-weather athletes have founda high prevalence of airway inflammation and hyper-reactivity(14). In some studies, the prevalence of airway diseasewas statistically greater than that measured in sedentary controlsubjects (3) or control subjects who exercise in more temperateconditions (1). Cold-weather athletes have been found to havesignificantly more airway inflammation than their sedentarycounterparts (3), despite the absence of a comparable differencein atopy (2). These studies have resulted in the hypothesisthat repeated cold weather hyperpnea can predispose these athletesto chronic airway disease with similarities to asthma. In fact,these similarities have resulted in the term "ski asthma" todescribe the syndrome of nonatopic airway inflammation and hyper-reactivityin elite winter athletes (3).
Racing sled dogs also perform strenuous exercise (and thereforeincrease their minute ventilation) under frigid conditions.Sled dogs can sustain speeds as high as 25 km/hour (5), andendurance dogs can cover 200 km/day. Racing sled dogs expendnearly four times more weight-specific energy than cyclistscompeting in the Tour de France (6), placing a substantial thermalload on an animal incapable of sweating. Consequently, thesedogs must rely on respiratory heat exchange (both conductionand evaporation) to release approximately 60% of their metabolicheat excess (7), while at the same time maintaining an adequatelevel of alveolar ventilation to support strenuous aerobic exercise.(The actual increase in minute ventilation in racing sled dogshas not been reported, but a reasonable estimate is an 8- to10-fold increase over basal levels during exercise, based onthe measured total energy expenditure and the relative timespent racing.) Because of the similarities between the activitiesof human cold-weather athletes and racing sled dogs, we hypothesizedthat racing sled dogs might also suffer from an increased prevalenceof airway disease. To test this hypothesis, we conducted a bronchoscopicsurvey of elite racing sled dogs shortly after the completionof a 1,100-mile endurance race.
Studies to determine the prevalence and nature of postexerciseairway inflammation were conducted at the conclusion of the2001 and 2002 Iditarod, a 1,100-mile long race conducted eachMarch stretching from Anchorage to Nome, Alaska. In the earlyportions of the race, ambient temperatures are typically between-10°C and 0°C. However, once over the Alaskan rangeand into the interior, temperatures can remain between -40°Cand -10°C throughout the remainder of the race. In addition,many mushers prefer to run the dogs at night when the temperaturesare lower, as this strategy reduces the risk of hyperthermiain the dogs. Thus, dogs will experience intermittent bouts ofcold-air hyperpnea for 9 to 14 days while completing this race.For all studies, informed consent was obtained from the musheror authorized agent before the race and again at the conclusionof the race before initiating the preanesthetic fast. Dogs wereallowed to rest with ad libitum food and water for a minimumof 12 hours after completing the race, followed by a 12-hourfast before bronchoscopy. Immediately before anesthetic induction,each dog underwent physical examination, including measurementof pulse and respiratory rates, rectal temperature, thoracicauscultation, palpation of the trachea and larynx, and coupageof the thorax (thoracic percussion using an open palm). No systemicillness or contraindications to anesthesia were found in anyof the dogs examined. Dogs were anesthetized with an intravenousbolus of propofol (7 mg/kg; Abbott Laboratories, Abbott Park,IL) and were maintained with intermittent doses as needed toachieve suitable relaxation. An endotracheal tube was placed,and a 5-mm OD fiberoptic bronchoscope equipped with a videocamera attached to the eyepiece was advanced into the lowerairways. All large bronchi were individually inspected, andthe presence and volume of mucus/exudates were scored usinga scale of 03 (seeFigure 1
for score definitions andexamples).
Figure 1. Examples of bronchoscopic scoring grades: (A) bronchoscopy score = 0, no visible abnormalities; (B) bronchoscopy score = 1, mild accumulations of intraluminal material (note the crescent of mucus/exudates on the upper left); (C) bronchoscopy score = 2, moderate accumulations of intraluminal material throughout the airways; (D) bronchoscopy score = 3, extensive accumulation of intraluminal material throughout the airways.
In the first study (conducted in 2001), 59 dogs from eight teamswere examined 24 hours after completion of the race to estimatethe prevalence of airway disease secondary to cold-weather strenuousexercise. In the second study (conducted in 2002), endoscopicand bronchoalveolar lavage findings in postexercise sled dogs(dogs examined 48 hours of completing the race, n = 10) werecompared with a group of control sled dogs (dogs that had beentrained throughout the winter but had not been exercised forat least 2 weeks at the time of examination, n = 12). Afterendoscopic examination and scoring (as described previouslyhere), bronchoalveolar lavage fluid was obtained from a randomlyselected sublobar airway from all dogs in this study. A sterilelength of polyethylene tubing was advanced through the biopsychannel of the endoscope, and 20 ml of sterile Hank's phosphate-bufferedsaline was infused and immediately aspirated by hand. A totalnucleated cell count and aerobic bacterial culture were performedwithin 24 hours of collection, and a differential cell countwas performed on a cytocentrifuged slide preparation (200 µl,1,000 rpm, 4 minutes) stained with a modified Wright-Giemsastain. As our data were not normally distributed in all cases,endoscopic scores and differential nucleated cell concentrationswere statistically compared using the Mann-Whitney Rank Sumtest, with p < 0.05 considered significant. Subjective cytologicalevaluation was provided by an investigator (K.D.), who was blindedas to the physical examination, endoscopic findings, and sampledifferential cell concentrations.
All dogs participating in this study had normal rectal temperature,pulse rate, and respiratory rate. Despite normal vital signs,thoracic coupage elicited coughs in 36% of the dogs. Thirty-sixpercent coughed during tracheal palpation, and 22% had abnormallung sounds (crackles and/or wheezes) heard during thoracicauscultation. Of the dogs showing abnormal signs during physicalexamination, most had more than one abnormality. The mean endoscopicscore was 1.41 ± 0.21 SEM. An abnormal accumulation ofmucus or exudate was present in the lower airways of 48 outof 59 dogs (81%): 36% were classified as mild (endoscopic score1), 32% were classified as moderate (endoscopic score 2), and14% were considered severe (endoscopic score 3).
In the second study, the control dogs had a mean ± SEMendoscopic score of 0.67 ± 0.15, compared with a meanscore of 1.8 ± 0.21 for the postexercise dogs (p = 0.0017).Postexercise dogs had greater bronchoalveolar lavage fluid (BALF)-nucleatedcell concentrations (53.0 ± 7.6 x 104 cells/µlversus 28.8 ± 8.0 x 104 cells/µl in control dogs),with statistically significant increases in macrophages, lymphocytes,and eosinophils (Table 1)
. There was a strong nonsignificanttrend (p = 0.055) toward increased neutrophils in the BALF recoveredfrom postexercise dogs. The majority of the control BALF sampleswere considered subjectively normal, although multinucleatedgiant cells were observed in five of the nine BALF samples fromcontrol dogs, suggesting chronic inflammatory stimulus (we didnot recover sufficient volume of BALF from three of the controldogs to permit analysis). Of the 10 BALF samples obtained frompostexercise dogs, 4 were judged to represent neutrophilic ormixed inflammation, with the rest of the postexercise BALF samplesconsidered normal. Aerobic bacterial cultures of the recoveredBALF were uniformly negative in both groups of dogs.
Human athletes who routinely experience hyperpnea in cold conditionshave been reported to also have a high prevalence of peripheralairway inflammation and hyper-reactivity, leading some investigatorsto suggest that repeated penetration of incompletely conditionedair (air that has been incompletely warmed and humidified) intohuman peripheral airways can predispose these athletes to chronicairway disease similar to asthma (14, 8). In fact, thissyndrome has been termed "ski asthma" in recognition of thesimilar phenotype and the most frequently studied human populationaffected by this syndrome. Studies using a canine laboratorymodel of peripheral airway exposure to incompletely conditionedair have reported the development of persistent airway obstruction,airway hyper-reactivity, impaired bronchodilation, airway inflammation(9, 10), and remodeling of the airway mucosa and lamina propria(11). In this study, we provide evidence that airway inflammationoccurs under natural conditions in canine athletes. These observationsprovide additional support for the contention that repeatedhyperpnea with cold air can injure peripheral airways and alsoidentify a potential naturally occurring animal model of theanalogous human condition.
During inhalation, heat and water vapor are transferred fromthe airway mucosa to the inspired air until the inspired airis warmed to body temperature and fully humidified. At restand in temperate conditions, this process is completed in theupper airways so that there is minimal heat and water transferthat occurs in the intrapulmonary airways (12). However, studiesin humans and horses have demonstrated that during hyperpnea,air that is not fully warmed (and, therefore, not completelyhumidified) penetrates into the intrapulmonary airways, particularlywhen the inspired air is cold (13, 14). As a result, heat andwater is lost from the surface of the lower airways, resultingin airway mucosal cooling and probably desiccation (15, 16).
The complete pathogenesis of ski asthma is not known, but keyfeatures of this process have been suggested in human and laboratoryanimal experiments. Hyperosmolarity of the airway surface liningfluid has been demonstrated in dogs after peripheral airwayexposure to cool, dry air (17) and has been suggested to alsooccur in humans during hyperpnea with cold, dry air (15, 16).Airway epithelial cells have been shown to release interleukin-8(a key chemokine for neutrophils) in vitro in response to localhyperosmolarity (18), and mast cells have been shown to degranulatewhen exposed to hyperosmotic stimuli in vitro (19, 20). However,hyperosmolarity of the airway surface lining fluid may not besufficient to produce inflammatory cell recruitment in vivo,as nebulization of hypertonic saline into canine peripheralairways failed to produce inflammatory cell influx (21). Thus,coincident airway cooling may be required for the developmentof ski asthma.
We do not believe that the airway pathology found in this studyis the result of an infectious agent shared among the animals.All of the dogs were afebrile at the time of examination andas competing athletes were prohibited from receiving any medicationsthat would have masked signs of illness. The dogs in the firststudy came from eight different racing teams that were separatedby as many as 4 days, and the dogs in the second study camefrom three different teams, including one in which (comprisingthe entire control group) the dogs were geographically separatedfrom each other. Thus, it is unlikely that the dogs had directcontact with each other before examination. Bacteria were observedin only one sample in the first study, and none of the samplesfrom the second study yielded positive bacterial cultures. Wecannot completely exclude the possibility of aeroallergens orother inhaled irritants as a factor in the respiratory diseasein these dogs; however, in the case of aeroallergens, theseparticles are often observed in airway secretions that are examinedmicroscopically. No foreign particles were noted in the samplesobtained from these dogs. Thus, we believe that the airway diseaseidentified in this population of racing sled dogs is a resultof peripheral airway cooling and desiccation, resulting in airwayinjury and activation of local inflammatory pathways.
The accumulation of intraluminal material in the postrace dogsis likely due to a combination of both increased productionof mucus and exudate, as well as decreased clearance of theintraluminal material. Both airway cooling and hyperosmolaritycan provoke release of mucus by goblet cells and mucus glands(22). Goblet cells and mucous glands are found throughout theconducting airways in dogs (23), and thus, dogs are capableof considerable mucus production caused by airway cooling anddesiccation, even in the peripheral airways. Airway coolingand desiccation also causes airway mucosal damage in humans(24), horses (25), and the canine laboratory model (26). Furthermore,repeated airway cooling and desiccation has been shown to produceextensive loss of ciliated mucosa, with the development of squamousmetaplasia after only four daily challenges (11). These morphologicchanges can be expected to impair mucociliary clearance, resultingin additional accumulation of intraluminal material. Althoughincreased intraluminal mucus may help protect the airway fromfurther desiccation, this theoretical benefit may be offsetby the increase in airway resistance due to lumenal obstruction.
The absolute and differential nucleated cell concentrationsin the BALF obtained from control dogs are comparable to thosereported by our laboratory for untrained mongrel dogs (10),although the cytology of some of the control dogs suggestedthe presence of low-grade chronic inflammation. This inflammationmay reflect previous training and/or racing injury that hadnot completely resolved in the 2 weeks of rest before examination.Increased BALF macrophage and eosinophil concentrations foundin the postexercise dogs are in agreement with BALF obtainedfrom dogs that received repeated airway cooling and desiccation(10). Furthermore, our results are in general agreement withsimilar studies in human winter athletes. Increased macrophage,lymphocyte, and eosinophil concentrations have been reportedin endobronchial biopsies obtained from human winter athletes(27), and increased concentrations of BALF macrophages havealso been reported in this population of humans (3). However,neutrophil concentrations were not significantly different fromcontrols in this study, in contrast to previous studies usingthe canine laboratory model, which found increased neutrophilconcentrations 24 hours after repeated peripheral airway challengewith inadequately conditioned air (10). It is important to notethat in contrast to the laboratory studies, BALF in the secondstudy was collected from the dogs in this study 4850hours after completion of the race. (The study was originallyplanned for examination of the dogs within 24 hours, but difficultiesin moving personnel and equipment to Nome forced a delay.) Wehave previously shown that even after 4 days of severe laboratorychallenge, resolution of airway neutrophilia takes less than7 days (11). In the absence of continued airway challenge withinadequately conditioned air, the stimulus for release of neutrophilchemotactic factors undoubtedly waned and normal neutrophilturnover resulted in the loss of identifiable cells in the airwaylumen. Thus, it is possible that the inadvertent delay in obtainingBALF from the postexercise dogs resulted in sufficient resolutionof neutrophilia to prevent the detection of a statisticallysignificant difference. This may also explain the lack of strikingsubjective abnormalities in the cytological evaluation of thesamples.
Our demonstration of airway inflammation and intraluminal accumulationof mucus and exudate is in agreement with the hypothesis thatsome cases of ski asthma are the result of airway mucosal injurysecondary to hyperpnea-induced airway hyperosmolarity. As recentlysummarized by Anderson and Holzer (28), excessive water lossby peripheral airways during hyperpnea can trigger a seriesof events leading to cough, mucus production, and airway obstruction.These signs (particularly cough) are commonly described afterexercise in elite athletes regardless of whether they demonstrateairway hyper-reactivity consistent with exercise-induced asthma(29). Our findings thus add to the body of evidence suggestingthat signs and symptoms consistent with exercise-induced asthmaare phenomena resulting from exercise-induced airway injuryand inflammation and can be provoked in otherwise healthy subjectsby strenuous exercise.
Racing sled dogs are potentially useful animal models of skiasthma. Similar to cross-country skiers, sled dogs typicallyexercise below their maximum aerobic capacity for extended periodsof time and in frigid conditions. In addition, typical sleddog teams are relatively uniform in their size, age, husbandry,environmental history, and genetic background. Thus, less intersubjectvariability would be expected compared with human athletes.Compared with humans, dogs can be subjected to more invasivetechniques in the course of an investigation, and the canineperipheral airway responses to incompletely conditioned airmeasured in laboratory studies have remarkable fidelity to thosereported in humans. Based on the results of these studies, thesesimilarities extend to the naturally occurring disease thatis shared by human and canine elite winter athletes.
Acknowledgments
The authors acknowledge the logistical contributions of LynnettePerrine, R.V.T. (chief veterinary technician, Iditarod TrailCommittee) and the generosity of Karl Storz Veterinary Endoscopyand Abbott Laboratories for their generous donations of equipmentand supplies.
FOOTNOTES
Supported by the Oklahoma Center for the Advancement of Scienceand Technology (HR 99001), the Iditarod Trail Committee,the International Sleddog Veterinary Medical Association, theInternational Federation of Sleddog Sports, and the generousdonation of materials and equipment by Abbott Laboratories,Inc., and Karl Storz Veterinary Endoscopy, Inc.
Received in original form December 12, 2001;accepted in final form June 11, 2002
Leuppi JD, Kuhn M, Comminot C, Reinhart WH. High prevalence of bronchial hyperresponsiveness and asthma in ice hockey players. Eur Respir J 1998;12:1316.[Abstract]
Sue-Chu M, Karjalainen E-M, Altraja A, Laitinen A, Laitinen LA, Noess AB, Larsson L, Bjermer L. Lymphoid aggregates in endobronchial biopsies from young elite cross-country skiers. Am J Respir Crit Care Med 1998;158:597601.[Abstract/Free Full Text]
Sue-Chu M, Larsson L, Moen T, Rennard SI, Bjermer L. Bronchscopy and bronchoalveolar lavage findings in cross-country skiers with and without "ski asthma." Eur Respir J 1999;13:626632.[Abstract]
Provost-Craig MA, Arbour KS, Sestili DC, Chabalko JJ, Ekinci E. The incidence of exercise-induced bronchospasm in competitive figure skaters. J Asthma 1996;33:6771.[Medline]
Van Citters RL, Franklin DL. Cardiovascular performance of Alaska sled dogs during exercise. Circ Res 1969;24:3342.[Abstract/Free Full Text]
Hinchcliff KW, Reinhart GA, Burr JR, Schreier CJ, Swenson RA. Metabolizable energy intake and sustained energy expenditure of Alaskan sled dogs during heavy exertion in the cold. Am J Vet Res 1997;58:14571462.[Medline]
Phillips CJ, Coppinger RP, Schimel DS. Hyperthermia in running sled dogs. J Appl Physiol Respir Environ Exercise Physiol 1981;51:135142.[Abstract/Free Full Text]
Larsson K, Ohlsen P, Larsson L, Malmberg P, Rydstrom PO, Ulriksen H. High prevalence of asthma in cross country skiers. BMJ 1993;307:13261329.
Davis MS, Freed AN. Repetitive hyperventilation causes peripheral airway obstruction and eosinophilia. Eur Respir J 1999;14:5762.[Abstract]
Davis MS, Freed AN. Repeated hyperventilation causes peripheral airways inflammation, hyperreactivity, and impaired bronchodilation in dogs. Am J Respir Crit Care Med 2001;164:785789.[Abstract/Free Full Text]
Davis MS, Freed AN. Repeated dry air hyperventilation causes airway inflammation, edema, and mucosal squamous metaplasia [abstract]. Am J Respir Crit Care Med 1998;157:A672.
McFadden ER Jr, Pichurko BM, Bowman HF, Ingenito E, Burns S, Dowling N, Solway J. Thermal mapping of the airways in humans. J Appl Physiol 1985;58:564570.[Abstract/Free Full Text]
McFadden ER Jr, Pichurko BM. Intraairway thermal profiles during exercise and hyperventilation in normal man. J Clin Invest 1985;76:10071010.
Gilbert IA, Fouke JM, McFadden ER Jr. Heat and water flux in the intrathoracic airways and exercise- induced asthma. J Appl Physiol 1987;63:16811691.[Abstract/Free Full Text]
Daviskas E, Gonda I, Anderson SD. Local airway heat and vapour losses. Respir Physiol 1991;84:115132.[CrossRef][Medline]
Daviskas E, Gonda I, Anderson SD. Mathematical modeling of heat and water transport in human respiratory tract. J Appl Physiol 1990;69:362372.[Abstract/Free Full Text]
Freed AN, Davis MS. Effects of hyperventilation on peripheral airway surface fluid and airway resistance. Am J Respir Crit Care Med 1999;159:11011107.[Abstract/Free Full Text]
Hashimoto S, Matsumoto K, Gon Y, Nakayama T, Takeshita I, Horie T. Hyperosmolarity-induced interleukin-8 expression in human bronchial epithelial cells through p38 mitogen-activated protein kinase. Am J Respir Crit Care Med 1999;159:634640.[Abstract/Free Full Text]
Eggelston PA, Kagey-Sobotka A, Lichtenstein LM. A comparison of the osmotic activation of basophils and human lung mast cells. Am Rev Respir Dis 1987;135:10431048.[Medline]
Eggleston PA, Kagey-Sobotka A, Schleimer RP, Lichtenstein LM. Interaction between hyperosmolar and IgE-mediated histamine release from basophils and mast cells. Am Rev Respir Dis 1984;130:8691.[Medline]
Suzuki R, Freed AN. Hypertonic saline aerosol increases airway reactivity in the canine lung periphery. J Appl Physiol 2000;89:21392146.[Abstract/Free Full Text]
Dwyer TM, Farley JM. Mucus glycoconjugate secretion in cool and hypertonic solutions. Am J Physiol 1997;272:L1121L1125.[Abstract/Free Full Text]
Pliss LB, Ingenito EP, Ingram RH, Pichurko B. Assessment of bronchoalveolar cell and mediator response to isocapnic hyperpnea in asthma. Am Rev Respir Dis 1990;142:7378.[Medline]
Davis MS, Lockard AJ, Marlin DJ, Freed AN. Airway cooling and mucosal injury during cold weather exercise. Equine Vet J Suppl (In press)
Omori C, Schofield BH, Mitzner W, Freed AN. Hyperpnea with dry air causes time-dependent alterations in mucosal morphology and bronchovascular permeability. J Appl Physiol 1995;78:10431051.[Abstract/Free Full Text]
Karjalainen EM, Laitinen A, Sue-Chu M, Altraja A, Bjermer L, Laitinen LA. Evidence of airway inflammation and remodeling in ski athletes with and without bronchial hyperresponsiveness to methacholine. Am J Respir Crit Care Med 2000;161:20862091.[Abstract/Free Full Text]
Anderson SD, Holzer K. Exercise-induced asthma: is it the right diagnosis in elite athletes? J Allergy Clin Immunol 2000;106:419428.[CrossRef][Medline]
Rundell KW, Im J, Mayers LB, Wilber RL, Szmedra L, Schmitz HR. Self-reported symptoms and exercise-induced asthma in the elite athlete. Med Sci Sports Exerc 2001;33:208213.[CrossRef][Medline]
This article has been cited by other articles:
F. Drobnic and T. Haahtela The role of the environment and climate in relation to outdoor and indoor sports
Diagnosis, Prevention and Treatment of Exercise-Related Asthma, Respiratory and Allergic Disorders in Sports,
June 28, 2010;
35 - 48.
[Abstract][Fulltext][PDF]
K-H. Carlsen Asthma, airway inflammation and epithelial damage in elite athletes
Eur. Respir. J.,
April 1, 2009;
33(4):
713 - 714.
[Full Text][PDF]
L. Chimenti, G. Morici, A. Paterno, A. Bonanno, L. Siena, A. Licciardi, M. Veca, W. Guccione, F. Macaluso, G. Bonsignore, et al. Endurance Training Damages Small Airway Epithelium in Mice
Am. J. Respir. Crit. Care Med.,
March 1, 2007;
175(5):
442 - 449.
[Abstract][Full Text][PDF]
T. D. Mickleborough, R. L. Murray, A. A. Ionescu, and M. R. Lindley Fish Oil Supplementation Reduces Severity of Exercise-induced Bronchoconstriction in Elite Athletes
Am. J. Respir. Crit. Care Med.,
November 15, 2003;
168(10):
1181 - 1189.
[Abstract][Full Text][PDF]
M. J. Tobin Asthma, Airway Biology, and Nasal Disorders in AJRCCM 2002
Am. J. Respir. Crit. Care Med.,
February 1, 2003;
167(3):
319 - 332.
[Full Text][PDF]