INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Microbial exposure in the indoor environment has long since been recognized as a potential cause of respiratory or other disorders. Reports of ill health related to microbial exposure have already appeared in the seventeenth and eighteenth centuries (1), and the first reference suggesting adverse health effects of microbial growth in the home environment may be found even in the Bible; in the book of Leviticus, home occupants are warned against what probably was mold growth in the house, described as "green- or red-like spots on and in the wall that may spread" (4). However, despite this early recognition, the causative agents and mechanisms involved in the relation between microbial exposure and the development and expression of respiratory symptoms are, apart from infectious diseases, only partially known and understood.

Indoor house dust mite and pet allergens have been suggested as causes of asthma in many areas of the world, particularly in children. In the past decade, there has been an increased interest in the possible role of indoor microorganisms. Epidemiologic studies have suggested that allergic and nonallergic inflammatory reactions to fungal components in the indoor environment might account for the frequently observed association between home dampness and prevalence and severity of respiratory symptoms (5). However, in most of the studies relating respiratory morbidity to fungal exposure, exposure to fungi was usually assessed by questionnaire, and it is not clear to what extent questionnaire reports of damp and mold spots correlate with exposure to relevant fungal components. We have previously reported a positive association between fungal extracellular polysaccharides (EPS) in house dust, as a marker of exposure to fungi, and respiratory symptoms in children (16). The associations were stronger for EPS measurements than for viable mold spore counts and self- reported mold spots in the home. Others have measured (1 3)--D-glucans as a marker of fungal exposure in the indoor environment (17). (1 3)--D-glucans originate from most fungi, some bacteria, and most plants and have the capacity to initiate a variety of inflammatory reactions in vertebrates (20), and they have been associated with adverse respiratory health effects in the indoor and occupational environment (17).

Other studies have suggested a role for bacteria and bacterial endotoxins (21). In a Swedish case-control study (21), asthma-related symptoms were significantly associated not only with house dust mite exposure but also with exposure to bacteria, assessed with a "nonviable" microscopic counting method, and studies of Michel and colleagues (22, 23) have demonstrated that levels of bacterial endotoxin in house dust are positively associated with the clinical severity of asthma. Endotoxins are proinflammatory components of gram-negative bacteria that are well known as causal agents in nonimmune mediated airway inflammation and associated respiratory disorders, particularly in the occupational environment (25).

At present no epidemiologic data have been reported showing a relation between objectively assessed indoor microbial exposure and lung function variability. In previous studies, peak expiratory flow (PEF) in asthmatic children was significantly associated with dust mite allergen levels in the home environment (28, 29). In the present study we have investigated the association between endotoxin and (1 3)--D-glucan levels in house dust and PEF variability in 148 school children in The Netherlands, 50% of whom had preexisting chronic respiratory symptoms.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Study Design

The data were collected in the winter of 1993/1994 as part of a study on acute respiratory effects of air pollution in children with or without chronic respiratory symptoms (30). One hundred fifty-nine school children, 7 to 11 yr of age and living in Amsterdam were selected with a parent-administered screening questionnaire (30), such that approximately 50% (n = 78) had chronic respiratory symptoms (recent wheeze, shortness of breath with wheezing, or dry cough, and/or doctor-diagnosed asthma ever in life), whereas the other 81 had no reported respiratory symptoms. Of the 159 children, 148 performed PEF measurements in the morning after waking up and in the evening before going to bed, for a period of 16 wk (13). Skin-prick tests (SPT) were performed to identify atopic subjects, using a panel of six common allergens: birch, timothy grass, cat fur, dog dander, house dust mite, and Cladosporium herbarum, all from ALK Laboratories (Copenhagen, Denmark) (13). A mean wheal diameter of more than 2 mm was regarded as a positive result. Subgroups of symptomatic children were defined on the basis of symptoms reported in the questionnaire ("asthma symptoms": including shortness of breath with wheezing or doctor-diagnosed asthma; "cough symptoms": including chronic cough and no asthma symptoms), whereas a subgroup of "atopics" consisted of all children with at least one positive SPT. Subjects who only reported recent wheezing and no other symptoms were not included in any of the sub-groups. Written consent was given by the parents of all children who participated in the study.

House Dust Samples

House dust samples were collected during the PEF monitoring period from floors (1 m²) of living rooms and bedrooms and from the children's mattresses (entire surface area), according to a standardized protocol, using a vacuum cleaner equipped with an ALK house dust sampling device (31, 32). No dust samples were taken in 11 of the 148 homes of children from whom PEF data were available, whereas the homes of the 11 children from whom we had no PEF data were all sampled. From both groups of 11 children we had complete data on other characteristics included in the data analyses such as presence of pets in the home, smoking in the home by parent(s) or other inhabitant(s), etc., and these data did not significantly differ from the corresponding data from the rest of the children included in the study.

Samples, i.e., filters plus all the dust collected on the filter, were extracted at room temperature with pyrogen-free distilled water containing 0.05% Tween-20 (31), and after centrifugation and collection of the supernatant, redissolved in an identical volume of the same medium (H₂O-Tween) and extracted at 120° C to solubilize (1 3)--D-glucans (32). The first supernatant was used to measure endotoxin in the chromogenic kinetic Limulus Amoebocyte Lysate (LAL) test, using one single lot of the LAL reagent (31). (1 3)--D-glucan was measured in the second supernatant with a (1 3)--D-glucan-specific inhibition enzyme immunoassay (EIA) (32), using affinity-purified polyclonal rabbit anti-(1 3)--D-glucan antibodies, and laminarin [a linear (1 3)-- D-glucan] as the coated antigen and calibration standard. The major house dust mite allergen Dermatophagoides pteronyssinus (Der p1) was measured in the first extract, with a Der p1-specific, monoclonal antibody-based, sandwich EIA (33). Because H₂O-Tween is not commonly used for extraction of Der p1, control experiments were performed with a series of duplicate house dust samples that were extracted with either H₂O-Tween or with buffered saline. The results showed that recovery of Der p1 was approximately 50% in H₂O-Tween extracts compared with buffered saline, but that this factor was constant over a wide range of concentrations, and thus did not change the relative Der p1 content in each sample compared with the mean of the whole population. Because of analytical failures some house dust analysis data were missing; an exact overview of the total number of exposure measurements is given in Tables 2 and 3 (Table 2 includes dust samples from homes of children from whom we did have data on respiratory symptoms but not PEF data). Concentrations were expressed both per square meter of sampled area and per gram of sampled dust. Samples with nondetectable concentrations were given a value of ²/₃ of the detection limit when expressed per square meter and when expressed per gram dust, a value of ²/₃ of the lowest concentration found in the samples with levels above the detection limit.

Statistical Analysis

Statistical analyses were performed using SAS version 6.12 (SAS Institute, Cary, NC). Because exposure data followed a log-normal distribution, endotoxin, (1 3)--D-glucan and Der p1 concentrations were ln-transformed. A chi-square test was used to compare prevalences, and Student's t test for comparisons of exposure values in subgroups and at different locations. To calculate PEF variability (Ampl%mean), the absolute difference between morning (AM) and evening (PM) PEF of each day was divided by the mean PEF of that day, and subsequently this relative daily PEF-amplitude was averaged over the whole study period (34). PEF data from the first 2 d of each subject were excluded from the analyses because of a possible learning effect. Ampl%mean was log-normally distributed, and regression analyses on the relation with exposure were therefore performed on ln-transformed data. The regression coefficients for each relation are for presentation transformed to the relative increase in Ampl%mean associated with an increase in the exposure variable of two geometric standard deviations (GSDs). Stratified analyses were performed for groups of children with or without respiratory symptoms, and for subgroups with symptoms of asthma, and subgroups with or without "atopy" defined as described. After unadjusted analyses, multiple regression models were used to relate PEF variability to either endotoxin or (1 3)--D-glucan, adjusting for Der p1 levels, pets in the home and type of floor cover (carpet versus smooth). Because of the high correlation between endotoxin and (1 3)--D-glucan levels, these were analyzed separately in the multiple regression analysis. Analyses were performed using endotoxin, (1 3)--D-glucan and Der p1 levels both when expressed per square meter of sampled area and per gram sampled dust.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

An overview of the various groups and subgroups is given in Table 1, including only the children for whom PEF-data were available. Sex and age showed no significant differences, except that the group with asthmatic symptoms contained more boys. Parental smoking and presence of pets were less prevalent in the group with asthmatic symptoms than in the non-symptomatic children. Atopy, especially positive SPTs to mites or pets, was much more prevalent in children with respiratory symptoms, and particularly in children with asthmatic symptoms. PEF was lower and PEF variability was higher in symptomatic children, and this also was most pronounced in the group with asthmatic symptoms.

Endotoxin and (1 3)--D-glucan levels did not differ between non-symptomatic and symptomatic subjects, with the exception of mattress endotoxin, which was higher in subjects with chronic cough (Table 2). Mite allergen levels were relatively low, compared with previous studies in Dutch homes, with lowest concentrations in homes of symptomatic children. Highest Der p1 levels were measured on mattresses. Endotoxin, (1 3)- -D-glucan and Der p1 levels were approximately 10-fold higher on carpets than on smooth floors (data not shown). Unadjusted regression analyses showed that Ampl%mean in symptomatic children was significantly associated with endotoxin, (1 3)-- D-glucan and Der p1 levels on living room floors when expressed per square meter, and also with the presence of pets in the home and type of floor cover (Table 3). The associations were strongest in atopic and/or asthmatic children. Environmental tobacco smoke (ETS) was not associated with PEF variability, and PEF variability in nonsymptomatic children and in children with chronic cough showed no significant associations with any of the investigated exposure parameters. No significant associations were found when endotoxin and glucan levels were expressed per gram dust, whereas for Der p1 the association was very similar for both measures of exposure (data not shown). Also, there was no association between endotoxin, (1 3)--D-glucan and mite allergen levels in bedroom floors and mattresses (either expressed per square meter or per gram dust) and Ampl%mean of either symptomatic or nonsymptomatic children (data not shown).

To determine whether the associations found for living room floor levels expressed per square meter were independent of other known risk factors, two multiple regression models were applied: one with (1 3)--D-glucan (Figure 1A) and one with endotoxin (Figure 1B), both adjusted for mite allergen exposure, type of floor cover, and presence of pets. As shown in Figure 1A, the association between (1 3)--D-glucan and Ampl%mean remained in symptomatic subjects, particularly in atopic asthmatics. The associations with pets, and to a lesser extent, with Der p1 levels, remained as well. In contrast, the association with endotoxin (Figure 1B) did not remain after adjustment for pets, mite allergen, and floor cover.

View larger version (14K): [in this window] [in a new window]   View larger version (13K): [in this window] [in a new window]   Figure 1.   Multiple regression between ln-transformed Ampl%mean and ln-transformed living room floor (1 3)--D-glucan (A) and endotoxin (B) concentration, adjusted for Der p1 concentration, pets in the home, and type of floor cover. Results are presented as relative increases (ratio) in Ampl%mean associated with an increase of two GSDs in exposure or with the presence of wall-to-wall carpet or pets in the home.

Because asthma medication (taken during the study by 16 symptomatic children who mainly used bronchodilators and/ or maintenance medication) might have attenuated the effects of exposure on PEF variability, the analyses were repeated, excluding children who used asthma medication. Regression coefficients in both unadjusted and in multiple regression analyses were not significantly different from those found in the whole group (data not shown).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

In this study, PEF variability in children with chronic respiratory symptoms was shown to be strongly associated with (1 3)--D-glucan levels in dust from living room floors when expressed in microgram per square meter. The association was strongest for atopic children with asthmatic symptoms. In non-symptomatic children and children with chronic cough only, no significant associations were found. Reporting bias seems unlikely since assessment of exposure and PEF variability were based on objective measurements. Confounding by mite allergen and pets was demonstrated for endotoxin but not for (1 3)--D-glucan. ETS was not associated with PEF variability and could thus not have biased the observed associations. Asthma medication could have biased these relations, but excluding the children who regularly received medicine during the study did not lead to significantly different results.

Animal and in vitro studies have shown that (1 3)--D-glucans have profound proinflammatory effects, including activation of neutrophils, macrophages, complement, and possibly eosinophils (20). Furthermore, (1 3)--D-glucans have been suggested to be involved in bio-aerosol-induced inflammatory responses and resulting respiratory symptoms and adverse lung function effects (17, 35). This study would be the first to show a relation between indoor (1 3)--D-glucan exposure and PEF variability as an objective marker of respiratory health. This association was found primarily in atopic, asthmatic children who thus seem to be more susceptible to (1 3)--D-glucan (or to fungi, if glucan was mainly a marker of fungi). The relation is probably not based on type I allergic reactions since glucans are considered nonimmunogenicor at least nonallergenic-in humans (20). Presumably, (1 3)-- D-glucan exposure may increase PEF variability in atopic asthmatics by enhancing preexisting allergic or nonallergic inflammation. Alternatively, (1 3)--D-glucan may activate an independent pathway of airway inflammation that is most strongly expressed in hyperresponsive atopic asthmatics. In the atopic asthmatics an increase in living room floor (1 3)- -D-glucan levels by 2 standard deviations (in our data an approximately 18-fold increase) was associated with a 1.6-fold increase in PEF-variability. PEF-variability was also approximately 1.7 times higher in subjects with pets in their homes. This is a substantial increase, considering the fact that the Ampl%mean in the asthmatic children was only 20% higher than in the nonsymptomatic children. A previous study among European children with chronic respiratory symptoms showed that in atopic children fungal growth in the home was associated with a 15% increase in PEF variability (13).

Other studies have shown an association between endotoxin in house dust and clinical severity of asthma in asthmatic adults and children (22). In those studies relationships were not adjusted for the presence of pets in the home. In our study, presence of pets was associated with higher endotoxin levels in living room floor samples (3,074 versus 1,569 EU/m²; p < 0.05), and after adjustment for pets (and mite allergens) there was no association between endotoxin and PEF variability, while the effect of pets remained significant. Pets are well known indoor sources of allergens, and may thus have contributed to PEF variability through allergic mechanisms. This may seem plausible considering the high prevalence of dog and cat positive SPTs among the symptomatic children (as much as 50% in the asthmatics and 81% in the group with the most pronounced effects; the atopic children with symptoms of asthma). However, the associations were not stronger when the analyses were limited to those children with a positive SPT to pets (data not shown), which suggests that the effect of pets may not be due only to type I antipet allergy. It even may be that the strong effect of pet exposure was mainly an effect of higher endotoxin exposure, and that the association of endotoxin with PEF variability was obscured by overadjustment in a model that included "pets" as a more robust variable for endotoxin exposure. We therefore also performed stratified multivariate analyses in all subgroups of children ("nonsymptomatic," "symptomatic," "symptomatic and atopic," "asthmatic," "asthmatic and atopic") with and without pets at home and found a positive association between endotoxin exposure and PEF variability only in the atopic symptomatic (general and asthma) children without pet exposure at home (only significant for the group "asthma and atopy"; relative increase in Ampl%mean associated with two GSDs = 2.2, p < 0.05; n = 13), whereas no or marginally negative (but not statistically significant) associations were found for symptomatic atopic children with indoor pets (data not shown). These results, however, should be interpreted with caution because of the very low numbers of children included in the stratified analyses, and also because multivariate regression analyses for the other subgroups ("symptomatic" and "asthmatic") showed no significant associations, with relative increases in PEF variability of around 1.

Freezing and thawing of dust extracts is known to affect the levels of endotoxins measured in house dust (31, 36), with one study showing a considerable 86% decline in endotoxin concentration after house dust extracts were frozen and thawed once (36). It may be that this could have biased the relation between PEF variability and house dust endotoxin levels in our study. However, this seems unlikely since (1) this would probably also have attenuated the relation in the unadjusted analysis, where a clear and significant association with PEF variability was found; (2) although all extracts were stored frozen, and freezing and thawing may result in a considerable loss of endotoxin activity, we previously demonstrated (31) a strong correlation (r² = 0.8) between endotoxin concentrations found in the same samples after one or two freeze-thaw cycles. In this study all samples were frozen and thawed only once, and therefore the risk of misclassification caused by this factor is expected to be low.

No associations were found between PEF variability and endotoxin and (1 3)--D-glucan levels in the bedroom. This was also true for Der p1. Allergen-avoidance measures, taken particularly in bedrooms of children with more severe asthma, may have biased these results, as shown previously in another Dutch study on the relationship between Der p1 levels and chronic respiratory health parameters (37). In our study it was not possible to adjust for allergen avoidance measures because of the relatively small number of atopic asthmatic children.

Der p1 levels in this study were generally not, or only weakly and nonsignificantly, associated with PEF variability. Also, when restricting the analyses to symptomatic subjects with a positive SPT to house dust mites, associations did not become stronger, which further indicates that in this study mite allergens did not contribute to an increased PEF variability. The absence of such a relation may, apart from allergen avoidance measures (see above), also be due to the levels of Der p1 that were relatively low (Table 2), also when adjusted for the 50% reduction in allergen levels because of a nonoptimal extraction method (see METHODS). Alternatively, positive associations with house dust mite allergen levels found in other studies may be (partially) explained by confounding effects of microbial exposures that in previous studies have not been appropriately adjusted for.

At present it is not known whether concentrations expressed either per gram dust or per square meter of the sampled area better reflect the actual exposure levels to allergens or other dust components. In our study (1 3)--D-glucan and endotoxin levels were associated with PEF variability only when expressed per square meter, whereas the (weak) associations of house dust mite allergens levels with PEF variability were independent of how levels were expressed. Dust mite levels per gram dust and per square meter were, however, strongly correlated (r² = 0.5; p < 0.05), whereas glucan and endotoxin levels expressed relative to dust weight were only moderately correlated with levels expressed per square meter (r² < 0.3; p < 0.5). Strong correlation between both measures for Der p1 has been shown previously by van Strien and colleagues (33) and they suggested, given this high correlation that both measures could be used for exposure assessment. Coloff and colleagues (38) suggested that concentrations of dust mite allergens expressed per square meter would give a better estimate of the potential exposure. Given these findings, both measures per square meter and per gram dust should be explored for microbial agents in future epidemiologic investigations.

In conclusion, exposure to house-dust-associated (1 3)--D-glucan in the home environment may increase peak flow variability in asthmatic children. Although (1 3)--D-glucans can be derived from both plant and fungal material, and thus can not be used as a highly specific marker for mold growth, the here-reported association may be one of the major mechanisms accounting for the relation between dampness and fungal growth in the home environment, and respiratory morbidity.