Dust and airborne exposure to allergens derived from cockroach (Blattella germanica) in low-cost public housing in Strasbourg (France) Frederic de Blay, MD," Jose Sanchez, MD, a Guy Hedelin, PhD, c Antonio Perez-lnfante, MD, a Ang~le Verot, BS, a Martin Chapman, PhD, b and Gabrielle Pauli, MD" Strasbourg, France, and Charlottesville, Va. Background: Although a strong association between allergy to cockroach (CR) and asthma has been observed in the United States and Asia, there are little data about the extent of exposure to CR allergen in Europe. Objective: To determine the levels of CR allergens in dust samples from apartments in Strasbourg and to determine the concentration and size of CR allergens in the air. Methods: Nine apartments in a public housing complex were chosen on the basis of visual evidence of CR infestation. Levels of CR allergens (Bla g 1 and Bla g 2) in kitchen and mattress dust samples were measured by immunoassay with the use of monoclonal antibodies. Air was sampled for 3 to 8 hours in the kitchen under undisturbed conditions, during artificial disturbance, and during normal domestic activity by using an impinger and a parallel glass fiber filter and at flow rates of 2 to 20 L/min. Airborne CR and mite allergens were measured concurrently in the bedroom of one apartment before, during, and after artificial disturbance. Results: High levels of Bla g 1 and Bla g 2 were found in kitchen dust from the nine apartments (geometric means of 3919 U/gm [range 530 to 14306 U/gm] and 497 U/gm [range 73 to 1946 U/gm], respectively). Under undisturbed conditions, airborne CR allergens were not detectable in any of the apartments. During vigorous artificial disturbance, Bla g 1 and Bla g 2 were detectable in air samples from seven apartments (geometric means of 4.5 U/m 3 [range O. 7 to 17.2 U/m 3] and 1.0 U/m s [range 0.4 to 3. 4 U/mS], respectively). Both allergens were predominantly collected on the first stage of the impinger, and 76% to 80% of the airborne allergen was associated with particles greater than 10 Ixrn in diameter. The levels were significantly higher than those collected on the second or third stages of the impinger (p < 0.001). A comparison of the levels of mite and CR allergens showed that the airborne properties of these allergens were similar, that is, measurable only during disturbance and not detectable 30 minutes after disturbance. Conclusion: Levels of CR allergen in low-cost public housing in Strasbourg can be as high as or higher than the levels measured in towns in the United States. CR allergens become airborne during disturbance and are primarily associated with particles greater than 10 ~ in diameter. Patients with asthma living in urban areas of Europe in housing prone to CR infestation should be evaluated for sensitization and exposure to CR allergens. (J Allergy Clin Immunol 1997;99:107-12.) Key words: Cockroach allergens, asthma, aeroallergen, environmental exposure, immediate hypersensitivity From aunit6 INSERM U425 Pavilion Laennec, H6pitaux Universitaires de Strasbourg; blnstitut d'epid6miologie, Facult6 de M6decine, Universit6 Louis Pasteur, Strasbourg; and ~Asthma and Allergic Diseases Center, Department of Medicine, University of Virginia, Charlottesville. Supported by a grant from Coca-Cola France S.A. (Prix Environnement et Sant6 1994) and by grants AI 32557 and AI 34607 from the National Institutes of Health. Received for publication July 31, 1995; revised June 12, 1996; accepted for publication June 21, 1996. Reprint requests: Fr6d6ric de Blay, Unit6 INSERM U425, Pavillon Laennec, H6pital Civil, H6pitaux Universitaires de Strasbourg, B.P. 426, 67091 Strasbourg Cedex. Copyright 1997 by Mosby-Year Book, Inc. 0091-6749/97 $5.00 + 0 1/1/76096 107
108 de Blay et al. J ALLERGY CLIN IMMUNOL JANUARY 1997 Abbreviations used CR: Cockroach mab: monoclonal antibody Sensitization to cockroach (CR) allergens is considered a risk factor for asthma-related admissions to U.S. hospital emergency departments and is most prevalent among lower socioeconomic groups and minority populations. 1-3 Cockroaches have been reported to cause asthma in other parts of the world, especially in the Far East. 4-6 However, there are little data about the prevalence of allergy to CRs in Europe, and it has not been considered a significant clinical problem. A population survey in southern France (Marseilles) showed the prevalence of allergy to CR to be 4.7%, and our initial studies in Strasbourg identified six patients with asthma, who were sensitive to CRs and also exposed to CR allergens in their homes. 7, 8 Major allergens from Blattella germanica have been identified, and monoclonal antibody (mab)- based immunoassays for Bla g 1 and Bla g 2 have been used for assessing exposure to CR allergen. 9 In an earlier study, Swanson, Agarwal, and Reed 1 detected the presence of CR allergens in the air by using RAST inhibition. However, the properties of airborne CR allergens have not been studied in detail, and the conditions under which these allergens become airborne and initiate sensitization and asthma symptoms are poorly understood. In this study we compared exposure to CR allergens in nine apartments in a low-cost public housing complex in an urban area of Strasbourg. Four apartments were inhabited by patients with asthma who were allergic to CRs. We also analyzed the quantities of CR allergens (Bla g 1 and Bla g 2) that become airborne and determined their aerodynamic particle size. The results showed that high levels of CR allergens accumulated in these apartments in Strasbourg (at levels comparable to those found in housing in the United States) and that CR allergens are carried on particles greater than 10 /~m in diameter, which become airborne after disturbance. METHODS Design of environmental studies Nine apartments in low-cost public housing (Habitation h Loyer Mod6r6) in an urban area of Strasbourg were studied. All had visible cockroach infestation. In each apartment, cockroach specimens were captured, and the only species found was B. germanica (the identification was performed by Dr. Matter, Entomologist at the Museum of Zoology, Louis Pasteur University, Strasbourg). Four apartments were occupied by patients with asthma (two men and two women, with a mean age of 27.5 years [range, 22 to 36 years]). One patient had positive skin test responses to mite and cockroach extracts, whereas the other three patients had positive skin test responses only to cockroach extracts. All four patients had detectable serum IgE antibody to cockroach extracts, at levels greater than or equal to class II in the CAP system (Pharmacia, Uppsala, Sweden). House dust samples were obtained from cabinets and floors and from areas behind dishwashers, underneath sinks, and behind refrigerators in the kitchen of each apartment. Mattress dust samples were also obtained from apartments of the four patients with CR allergy. The dust was sieved and extracted as described previously.3, 9 Air samples were obtained under three different conditions. In the absence of disturbance. Sampling was performed with no one in the kitchen (except for the researcher, who monitored the air sampling) 1 hour before the beginning of sampling and during the airsampling period. Air was sampled in seven apartments by means of an impinger and a parallel glass fiber filter, which served as a control for allergen levels measured by the impinger. To confirm the results and to compare the duration of sampling, two vacant apartments (1 and 5) were sampled on two occasions for 3 hours by using only a glass fiber filter and on two occasions for 8 hours at night by using a glass fiber filter connected to a portable pump (sampling was at 2 L/rain.). The kitchen door and windows were always closed during air sampling. During artificial disturbance. For the assessment of the level of airborne CR allergens during peak exposure, air was sampled during a vigorous 45-minute disturbance, which was achieved by running a vacuum cleaner (Hoover $3462; Creutzwald, France) with no filter. The vacuum cleaner was run in the kitchen, and dust was collected from areas throughout the kitchen (from behind the refrigerator and dishwasher, from cabinets, under the sink, and from the floor). The air sampler was in the middle of the kitchen, as far away as possible from the door and the window. The minimum distance between the sampler and the vacuum cleaner was 1 m. During a natural disturbance. To determine whether airborne CR allergen could be detected under normal living conditions, three patients with asthma who were allergic to CR carried a glass fiber filter attached to their lapel for 8 hours in apartments 2, 4, and 7. The inhabitants of these apartments consisted of one child and one adult (in apartment 2); one adult and three teenagers (in apartment 4); and three children and one adult (in apartment 7). Patients were free to move throughout the apartments and to carry on their usual domestic activities. All three apartments were carpeted, except
J ALLERGY CLIN IMMUNOL de Blay et al. 109 VOLUME 99, NUMBER 1, PART 1 TABLE I. CR allergen levels in dust and determination of particle size of airborne Bla g 1 and Bla g 2 allergens during an artificial disturbance Bla g 1 levels Bla g 2 levels In dust* In air (U/m3)t In dust In air (U/m3)t Particles Particles Particles On Particles Particles Particles On Apartment (U/gm} (>10 rtm} (1,5-10 Fm) (<5 i~m) Total PF (U/gm) (>10 i~m) (1.5-10 rim) (<5 i~m) Total PF 1 530 8.1 0.4 0.09 8.6 8.0 73 2.5 0.4 0.2 3.2 4.9 2 9316 0.7 0.2 0.04 0.9 0.9 1640 0.5 0.1 0.04 0.7 0.4 3 10446 1.6 0.2 0.08 1.9 1.9 256 0.4 0.03 0.02 0.4 0.4 4 12500 17.2 0.6 0.4 18.4 12.1 1946 3.4 0.2 0.06 3.7 2.7 5 3789 11.1 2.5 2.0 15.7 26.8 281 1.7 0.4 0.3 2.4 4.6 6 789 4.1 1.4 1.5 7.0 9.0 80 0.5 0.3 0.3 1.0 2.0 7 14306 5.2 0.4 0.1 5.8 3.0 1820 0.9 0.1 0.06 1.1 0.6 8:~ 3096 ND ND ND ND ND 762 ND ND ND ND ND 9:~ 2560 ND ND ND ND ND 1003 ND ND ND ND ND Geometric 3919.9 4.5 0.5 0.2 5.6 7.1 497.4 1.0 0.2 0.09 1.3 1.4 mean ND, Not done. *House dust samples were obtained at least 1 week before air sampling by using a vacuum cleaner powered through the main supply. tairborne allergen background was measured for 45 minutes before each artificial disturbance by means of an impinger and a parallel glass fiber filter; none of the seven apartments had detectable Bla g 1 or Bla g 2 levels in the air. During artificial disturbance, air was sampled by using an impinger and a parallel glass fiber filter (PF). For each apartment, air was sampled on three occasions, at least 1 week apart. ~:Two vacant apartments were not investigated for airborne measurements under artificial disturbance, because in the interim tenants had moved in and refused airborne samplings. in the kitchen and the bathroom, and were heated by radiators. Comparison of the aerodynamic characteristics of cockroach and mite allergen. Airborne CR allergens and group 1 and group 2 mite allergens were measured in the bedroom of apartment 2 by using a glass fiber filter before disturbance, during a 45-minute artificial disturbance, and 30 minutes after disturbance. Air sampling techniques Air sampling was carried out for 45 minutes by using an impinger (Research Glassware, U.K.) according to the manufacturer's recommendations and by using methods described previously. 11,12 According to the relative distribution of particles among all stages, all particles with an aerodynamic diameter greater than 10 p~m were collected on the first stage. 11 The airflow was adjusted at the inlet of the impinger to 20 L/min, corresponding to an airflow of 17.5 L/min measured at the suction pressure outlet. A glass fiber filter (Millipore, St Quentin, France) was connected to a stationary pump (IP20; Bioblock SA, Strasbourg, France) or to a portable pump. When the glass fiber filter was connected to a stationary pump, the airflow was adjusted to 20 L/min at the inlet of the cassette carrying the filter. When the glass fiber filter was connected to a portable pump, the duration of sampling was 8 hours at 2 L/min, as recommended by the manufacturer. All air sampling with stationary devices was performed by the same re- searcher. The impinger and the glass fiber filter were located 1.5 m above the floor, in the middle of the room. Before each sampling with an impinger or a glass fiber filter, the airflow was calibrated using a bubble airflow meter (Gillibrator; Panametrics, France) according to the French regulations for sampling airborne allergens. 13 mab-based allergens ELISA for cockroach and mite The assay for Bla g 1 and Bla g 2 has been previously described. 9 In brief, mab 10A6 (anti-bla g 1) or mab 8F4 (anti-bla g 2) was coated on ELISA plates (1 ixg/well) in 0.05 mol/l carbonate/bicarbonate buffer, ph 9.6, overnight at 4 C. A control curve, from 0.01 to 5.0 U/ml, was obtained with doubling dilutions of B. germanica reference UVA 89/01 (which contained 5000 U/ml of Bla g 1 and 3000 U/ml of Bla g 2) and was used to quantitate the ELISA. The detection limits for Bla g 1 and Bla g 2 were 0.01 U/ml and 0.02 U/ml, respectively. These are equivalent to detection limits of 0.01 U/m 3 and 0.02 U/m 3 for Bla g 1 and Bla g 2, respectively, for air sampling for 1 hour at 20 L/min. Group 1 and Group 2 mite allergens were assayed by ELISA with mabs as previously describedj 4, 15 Statistical analysis Statistical analysis was conducted according to methods developed and described by Aitchison on the statistical analysis of compositional dataj 6
110 de Blay et al. J ALLERGY CLIN IMMUNOL JANUARY 1997 TABLE II. Airborne levels of Bla g 1, Bla g 2, and mite allergens before, during, and after disturbance in a mite- and cockroachinfested bedroom of apartment 2* Bla g 1 (U/m 3) Bla g 2 (U/m 3 ) Group 1 mite allergen (ng/m 3) Group 2 mite allergen (ng/m 3 ) Before During After disturbance disturbance disturbance (45 min)t (30-45 rain) (30 min) < 0.01 0.4 (0.3-0.6):~ 0.015 (n=6) (n=6) (n=4) < 0.02 0.1 (< 0.02-0.13) < 0.02 (n=6) (n =6) (n =4) < 0.2 150.4 (123-171.5) < 0.2 (n =6) (n=6) (n=4) < 0.2 49 (27.7-82.6) < 0.3 (n=6) (n=6) (n=4) *Air was sampled at 20 L/min in the bedroom of apartment 2, which was infested by mites. The disturbance was performed from 30 to 45 minutes using a vacuum cleaner with no filter. Bla g 1 and Bla g 2 levels in the mattress dust were 11.3 U/gin and 5 U/gm, respectively. Group 1 and group 2 mite allergen levels were 55.3 p~g/gm and 9.9 Ixg/gm, respectively. Median values are shown. tvalues in parentheses indicate the duration of sampling. ~:Range of airborne cockroach and mite allergens during disturbance. RESULTS Cockroach and mite allergen levels in house dust samples High levels of CR allergens were detected in dust samples from nine apartments in Strasbourg. The highest levels were found in the kitchen, with geometric means of 3919 U/g and 497 U/g for Bla g 1 and Bla g 2, respectively (Table I). Both Bla g 1 and Bla g 2 were detectable in four mattress dust samples that were tested, although at much lower levels (geometric means, 24.2 U/g and 3.0 U/g, respectively). In contrast, group 1 and group 2 mite allergen levels were lowest in kitchen dust (less than 2 ~g/g), but present at high levels (greater than 10 p~g/g) in two out of four mattress dust samples. Concentration and particle size of cockroach allergens in the air None of the apartments had detectable levels of Bla g 1 or Bla g 2 in the air under undisturbed conditions, with any of the sampling conditions tested. When the air in the kitchen was artificially disturbed by running a vacuum cleaner for 45 minutes without a filter, both Bla g 1 and Bla g 2 were measurable in air samples from seven apartments, which were sampled on three separate occasions (two apartments were not available for air sampling). Although there was significant variation in the amount of allergen detected in the air at each sampling, all seven apartments had detectable airborne allergen. The majority of the Bla g 1 or Bla g 2 was collected on the first stage of the impinger, that is, on particles greater than 10 ~m in diameter, with much smaller amounts of allergen being collected on the second and third stages (Table I). The geometric means of Bla g 1 and Bla g 2 concentrations were 5.6 U/m 3 (range, 0.9 to 18.4 U/m 3) and 1.3 U/m 3 (range, 0.4 to 3.7 U/m3), respectively. Particle sizing experiments carried out with the impinger showed that 80% of airborne Bla g 1 and 74% of Bla g 2 was carried on particles greater than 10 ~m in diameter (Table I). In all of the experiments, less than 10% of airborne Bla g 1 or airborne Bla g 2 was detected on the last stage of the impinger or on the final filter (which trapped particles less than 5 ~m in diameter). Statistical analysis confirmed that the allergen levels measured on the first stage of the impinger were significantly higher than those on the second or third stages (p < 0.001). Both Bla g 1 and Bla g 2 showed similar particle size distributions, and no statistical differences were found between them (p = 0.7). Comparison of airborne cockroach and mite allergens One apartment (2) contained high levels of CR allergens in the kitchen and high levels of mite allergens in mattress dust (Table II). This apartment was selected for further study in order to directly compare the airborne properties of CR and mite allergens sampled before, during, and after artificial disturbance. Before disturbance, neither CR nor mite allergens were detected in the air. After artificial disturbance by using a vacuum cleaner without a filter, airborne Bla g 1, Bla g 2, and group 1 and group 2 mite allergens were all detectable; 30 minutes postdisturbance, CR and mite allergens were again detected at low levels or were undetectable (Table II). These results are further supported by the results of personal sampler experiments, in which three patients with asthma carried a lapel filter for 8 hours during their normal domestic activities in apartments 2, 4, and 7. In none of these cases (a total of six measurements) was Bla g 1 or Bla g 2 detected on the filters.
J ALLERGY CLIN IMMUNOL de Blay et al. 111 VOLUME 99, NUMBER 1, PART 1 DISCUSSION Data on the prevalence of sensitization to CR allergens have been reported from several parts of the United States and the Far East, whereas in Europe only a few surveys have been performed. Analysis of skin test data from our allergy and asthma clinics in Strasbourg has shown that of 986 outpatients who were skin tested with CR extract, 9.8% had positive skin-test responses to CR allergens and 25% of those patients had a single positive skin test to CR allergens (unpublished data). Unconfirmed reports from other European countries showed that 6.3% to 15% of patients with asthma were sensitized to CR allergens (i.e., they had positive skin test responses and serum IgE antibody). 17, 18 The present study is the first full report from Europe describing measurement of CR allergens in dust and air samples from CR-infested housing. Some of the Bla g 1 and Bla g 2 levels in the kitchen dust from Strasbourg apartments were as high as or higher than those found in Wilmington (Delaware) or Atlanta (Georgia) in the United States, with use of the same assay?, 3 Swanson et al. 1 first reported semiquantitative measurements of airborne CR allergens, but they did not analyze the particle size distribution. In the present study airborne CR allergens were not detected under undisturbed conditions. It is difficult to establish whether the CR allergen particles detected in the air following artificial disturbance correspond to those inhaled during natural exposure, and this continues to be a problem for the interpretation of experiments on aeroallergens. Vigorous disturbance may create an artificial particle size distribution with a predominance of larger particles carrying progressively higher amounts of allergen. The extent to which such particles become airborne during normal domestic activity is not known. Our results show that under similar conditions of disturbance, CR allergens behave very similarly to mite allergens in that they are present predominantly on particles greater than 10 ixm in size that remain in the air for only 20 to 30 minutes. These properties contrast with those of cat and dog allergens, which are readily detectable in the air under undisturbed conditions and in which a greater proportion of the allergen appears to be present on particles less than 5 ixm in diameter. 12, 19-22 It has recently been calculated that 1 U of Bla g 2 is equivalent to 40 ng Bla g 2 protein (L. K. Arruda and M. D. Chapman, unpublished data). Using this value, the absolute quantities of Bla g 2 detected in the air of seven apartments ranged from approximately 12 to 150 ng/m 3 (0.2 to 3.7 U/m3), with a geometric mean of 40 to 50 ng/m 3 (Table I). These concentrations of airborne CR allergen during an artificial disturbance are comparable to those previously reported for mite allergens. It is possible that patients are exposed to very low doses of airborne CR allergens associated with small particles, without being aware of acute bronchospasm at the time of exposure, as has been shown for other allergens by Ihre and Zetterstrom. 23 It has also been shown in 24 nonsmoking healthy males that 21% of a monodispersed fluorocarbon resin aerosol with an aerodynamic diameter of 13.7 Ixm could deposit in the lungs. 24 Thus particles greater than 10 Ixm may make an important contribution to the amount of inhaled allergen, even if only a small fraction is deposited in the lungsy, 26 The precise nature and source of the CR allergen particles is unclear. Nonetheless, extracts of CR feces have been shown to have potent allergenic activity by RAST inhibition, and immunofluorescence studies have localized IgE antibody binding to proteins present in intestinal epithelial cells and Malpighian tubules (CR excretory organs). 27,28 The complementary DNA encoding Bla g 2 has recently been cloned, and the allergen has been shown to be an aspartic protease. 29 This suggests that the allergen is a digestive enzyme secreted by CR with the feces. In conclusion, our studies have shown that the levels of CR allergen in low-cost public apartments of patients with asthma living in Strasbourg were comparable to those found in housing in the United States and that CR allergens are carried on particles greater than 10 ~m in diameter, which become airborne following disturbance. These studies suggest that patients with asthma living in urban areas of Europe in housing prone to cockroach infestation should be evaluated for sensitization and exposure to CR allergens. We thank Mrs. B. Oesterl6 and J. Clodong for preparing the manuscript. REFERENCES 1. Call RS, Smith TF, Morris E, Chapman MD, Platts-Mills TAE. Risk factors for asthma in inner city children. J Pediatr 1992;121:862-6. 2. Pollart SM, Chapman MD, Fiocco GP, Rose G, Platts-Mills T. Epidemiology of acute asthma: IgE antibodies to common inhalant allergens as a risk factor for emergency room visits. J Allergy Clin Immunol 1989;83:875-82. 3. Gelber LE, Seltzer L, Bouzoukis JK, Pollart SM, Chapman
112 de Blayet al. J ALLERGY CLIN IMMUNOL JANUARY 1997 MD, Platts-Mills TAE. Sensitization and exposure to indoor allergens (dust mite, cat and cockroach) as risk factors for asthma among patients presenting to hospital. Am Rev Resp Dis 1993;147:573-8. 4. Lan JL, Lee DT, Wu CH, Chang CP, Yeh CL. Cockroach hypersensitivity: preliminary study of allergic cockroach asthma in Taiwan. J Allergy Clin Immunol 1988;82:736-40. 5. Tandon N, Maitra SB, Saha GK, Modak A, Hati AK. Role of cockroaches in allergy to house dust in Calcutta, India. Ann Allergy 1990;64:155-7. 6. Choovivathanavanich P, Suwanprateep P, Kanthavichitra N. Cockroach sensitivity in allergic Thais. Lancet 1970;ii: 1362-3. 7. Birnbaum J, Guilloux L, Charpin D, Vervloet D. Cockroach allergens. Lancet 1991;337, 249. 8. de Blay F, Kassel O, Chapman M, Ott M, Verot A, Pauli G. Mise en 6vidence des allerg~nes majeurs des blattes par test ELISA dans la poussi~re domestique. La Presse M6dicale 1992;21:1685. 9. Pollart SM, Smith T, Morris E, Gelber L, Platts-Mills TAE, Chapman M. Environmental exposure to cockroach allergens: analysis with monoclonal antibody-based enzyme immunoassays. J Allergy Clin Immunol 1991;87:505-10. 10. Swanson MC, Agarwal MK, Reed CE. An immunochemical approach to indoor aeroallergen quantification with a new volumetric air sampler: studies with mite, roach, cat, mouse, and guinea-pig antigens. J Allergy Clin Immunol 1985;76:724-9. 11. May KR. Multistage liquid impinger. Bacteriol Rev 1966; 30:559-70. 12. Luczynska CM, Li Y, Chapman MD, Platts-Mills TAE. Airborne concentrations and particle size distribution of allergen derived from domestic cats (Felis domesticus). Am Rev Respir Dis 1990;141:361-7. 13. NF AFNOR X 43 404:1993 Qualit6 de l'air. Air de l'habitat domestique et des locaux collectifs. Pr616vement a6rien et analyse des allerg6nes de l'environnement int6rieur. 14. Luczynska CM, Arruda LK, Platts-Mills TAE, Miller JD, Lopez M, Chapman MD. A two-site monoclonal antibody ELISA for the quantification of the major Dermatophagoides spp. allergens, Der p I and Der f I. J Immunol Methods 1989;118:227-35. 15. de Blay F, Heymann PW, Chapman MD, Platts-Mills TALE. Airborne dust mite allergens: comparison of group 2 allergens with group 1 mite allergen and cat-allergen Fel d 1. J Allergy Clin Immunol 1991;88:919-26. 16. Aitchison J. In: The statistical analysis of compositional data. New York: Chapman and Hall Ed, 1986. 17. Pola J, Valdivieso R, Zapata C, Moneo I, Duce F, Larrad L, Losada E. Cockroach hypersensitivity in asthmatic patients. Allergol Immunopathol (Madr.) 1988;16:105-7. 18. Mosimann M, Peitrequin R, Blanc C, Pecoud A. Allergie aux blattes (cafards) dans une population suisse souffrant d'asthme et de rhinite chronique. Schweiz Med Wochenschr 1992;122:1245-8. 19. Custovic A, Green R, Fletcher A, Smith A, Pickering CAC, Chapman MD, Woodcock A. Aerodynamic properties of the major dog allergen, Can f 1: distribution in homes, concentration and particle size of allergen in the air. Am J Crit Care Med 1996. In press. 20. Sakaguchi M, Inouye S, Irie T, Miyazawa H, Watanabe M, Yasueada H, et al. Airborne cat (Fel d 1), dog (Can f 1), and mite (Der 1 and Der 2) allergen levels in the homes of Japan. J Allergy Clin Immunol 1993;92:797-802. 21. Wood R, Laheri A, Eggleston P. The aerodynamic characteristics of cat allergen. Clin Exp Allergy 1993;23:733-9. 22. de Blay F, Chapman MD, Platts-Mills TAE. Airborne cat allergen (Fel d 1): environmental control with the cat in situ. Am Rev Respir Dis 1991;143:1334-9. 23. Ihre E, Zetterstrom O. Increase in non-specific bronchial responsiveness after repeated inhalation of low doses of allergen. Clin Exp Allergy 1993;23:298-305. 24. Swartengren M, Falk R, Linnman L, Philipson K, Camner P. Deposition of large particles in human lung. Exp Lung Res 1987;12:75-88. 25. Brain J, Valberg P. Deposition of aerosol in the respiratory tract. Am Rev Resp Dis 1979;120:1325-73. 26. Task Group on Lung Dynamics. Deposition and retention models for internal dosimetry of the human respiration tract. Health Physics 1966;12:173-207. 27. Zwick H, Popp W, Sertl R, Rauscher H, Wanke T. Allergenic structures in cockroach hypersensitivity. J Allergy Clin Immunol 1991;87:626-30. 28. Lehrer S, Horner WE, Menon P, Stankus RP. Comparison of cockroach allergenic activity in whole body and fecal extracts. J Allergy Clin Immunol 1991;87:574-80. 29. Arruda LK, Vailes LD, Mann B J, Shannon J, Fox JW, Vedvick TS, et al. Molecular cloning of a major cockroach (Blattella germanica) allergen, Bla g 2: sequence homology to the aspartic proteases. J Biol Chem 1995; 270:19563-8.