(1→3)-β-D-glucans and respiratory health: a review of the scientific evidence
Dr Jeroen Douwes
Private Box 756
Tel.: +64 4 380 0617
Fax: +64 4 380 0600
Exposure to indoor fungi has widely been recognized as a plausible cause of dampness-related respiratory morbidity (Peat et al., 1998; Zock et al., 2002). However, the potential mechanisms for fungal-related airway diseases caused by indoor exposures are not clear. Some studies have shown associations between fungal exposure, sensitization and asthma (Black et al., 2000; Halonen et al., 1997; Zureik et al., 2002), but the evidence that fungal allergens and IgE allergic responses play a major role in indoor related respiratory symptoms is still very limited (Douwes and Pearce, 2003). In addition to allergic mechanisms, non-allergic responses to fungal exposures have been reported mainly in relation to fungal (1→3)-β-D-glucans. The first reports suggesting a potential role for (1→3)-β-D-glucans in the development of indoor air-related health effects appeared in the late 1980s and since then, a limited number of population studies have been published, mainly because of the fact that commercially available methods to analyze (1→3)-β-D-glucan were not widely available until some years ago.
(1→3)-β-D-glucan are non-allergenic water-insoluble structural cell wall components of most fungi, some bacteria, most higher plants and many lower plants (Stone and Clark, 1992). Glucans may account for up to 60% of the dry weight of the cell wall of fungi, of which the major part is (1→3)-β-D-glucan (Klis, 1994). They consist of glucose polymers with variable molecular weight and degree of branching i.e. triple helix, single helix or random coil structures (Williams, 1997). In the fungal cell wall, (1→3)-β-D-glucans are linked to proteins, lipids and carbohydrates such as mannan and chitin, and they contain (1→6)-β-glucan side-branches which may connect with adjacent (1→3)-β-D-glucan polymers (Klis, 1994). The (1→3)-β-D-glucan content of fungal cell walls has been reported to be relatively independent of growth conditions (Foto et al., 2004; Rylander, 1997a).
(1→3)-β-D-glucans can initiate a wide range of biological responses in vertebrates including stimulation of the reticuloendothelial system (Di Luzio, 1979), activation of neutrophils (Zhang and Petty, 1994), macrophages (Adachi et al., 1994; Lebron et al., 2003) complement (Saito et al., 1992) and possibly eosinophils (Ramani et al., 1988), resulting in enhancement of host-mediated induced resistance to infections and antitumor activity (Stone and Clark, 1992). (1→3)-β-D-glucans thus have potent biological properties, some of which may also play a role in the adverse health effects associated with indoor mold exposure. These biological properties are not dependent on viability and (1→3)-β-D-glucans from dead organisms may thus be equally relevant in causing potential health effects.
In many studies reservoir dust from carpets or mattresses is collected and concentrations are usually expressed either in weight units per gram of sampled dust or per meter square. Although both measures are generally accepted, the latter measure may better reflect actual exposure [Institute of Medicine (IOM), 2004]. The advantage of settled dust sampling is the presumed time-integration that occurs in the deposition of fungal (1→3)-β-D-glucan on surfaces over time (IOM, 2000). Fungi may also proliferate in carpets provided there is sufficient access to water. However, surface sampling is a crude measure that is most likely only a poor surrogate for airborne concentrations.
Airborne sampling requires very sensitive analytical methods. In addition, for an accurate assessment potentially large numbers of samples need to be collected as temporal variation in airborne concentrations is most likely very high (IOM, 2004). Airborne sampling after agitation of settled dust – as has been employed in several studies (Rylander, 1997b; Rylander et al., 1992, 1998; Thorn and Rylander, 1998a) – may overcome this problem. In addition, it may have the advantage (over settled dust sampling) that the more appropriate dust fraction i.e. airborne or inhalable particles will be sampled. However, these assumptions have not been confirmed, and procedures have not been standardized and/or validated. Therefore, uncertainty in exposure assessment of (1→3)-β-D-glucan is generally large which may result in obscured exposure-response relationships in epidemiological studies.
Most studies use the glucan-specific limulus amebocyte lysate (LAL) assay [available from Associates of Cape Cod, Falmouth, MA, USA (a subsidiary of Seikagaku Corporation, Tokyo, Japan)] which is based on the same principles as the LAL assay described for endotoxin measurements (Aketagawa et al., 1993). However, rather than activating factor C, glucans activate factor G leading to a series of enzymatic reactions resulting in a color or turbidimetric response. The glucan-specific LAL assay does not cross-react with endotoxin as factor C has either been removed or disabled from the LAL preparation. Recently a modification of the conventional LAL assay (containing both factor C and G) has been described for the analysis of (1→3)-β-D-glucan (Foto et al., 2004). The modification comprises pre-treatment of samples with 0.5 N NaOH which is known to destroy endotoxin and increase the response of factor G to (1→3)-β-D-glucans. Only little experience is available with the modified LAL assay and a comparison with the glucan-specific assay has not been conducted. In addition, NaOH treatment does not destroy all endotoxin in complex samples containing a mixture of both endotoxin and (1→3)-β-D-glucan (Foto et al., 2004). False positive results can therefore not be excluded.
(1→3)-β-D-glucan-specific immunoassays have also been developed (Douwes et al., 1996; Milton et al., 2001). In the inhibition enzyme-linked immunoassay (ELISA), (1→3)-β-D-glucans in the test sample inhibit the binding of affinity-purified rabbit anti-glucan antibodies to the (1→3)-β-D-glucans coated to the microtiter plate (Douwes et al., 1996). Quantification is achieved by labeling the rabbit antibodies with an enzyme-linked anti-rabbit antibody. In the sandwich ELISA, galactosyl ceramide is used as the capture reagent and a monoclonal antibody specific for (1→3)-ß-D-glucans is used as the detector reagent (Milton et al., 2001). The ELISA methods are considerably cheaper than the LAL assays but are (as yet) not commercially available, and extensive experience in the indoor environment is only available for the inhibition ELISA. In addition, the ELISA methods are significantly less sensitive than LAL methods. No comparison studies using environmental samples have been conducted and results obtained using different analytical methods may therefore not be directly comparable.
Most (1→3)-β-D-glucans present in nature are not water-soluble at room temperature. Therefore, two alternative methods of extraction of environmental samples have been described in the literature: (i) alkaline extraction using either 0.3 m NaOH (Rylander et al., 1992) or 0.5 N NaOH (Foto et al., 2004); and (ii) heat extraction by autoclaving samples at 120°C (1 bar) for 1 h (Douwes et al., 1996). The extraction efficiency of both methods is not known. Alkaline extraction is most often employed when samples are analyzed using the LAL assay, whereas heat extraction is commonly used in combination with the ELISA method(s). No differences in (1→3)-β-D-glucan content were seen between both extraction methods with the inhibition ELISA method (Douwes et al., 1996). However, large differences are likely for the LAL assay i.e. heat extraction will most likely underestimate the glucan concentration as it favors the formation of triple helix structures and these structures are severely underestimated in the LAL assay. Alkaline treatment on the other hand, transforms triple helix to single helix or random coil formations which are associated with increased factor G activation. Therefore, alkaline extraction may be the most optimal extraction method to be used in combination with the LAL assay. Apart from one small comparison (Douwes et al., 1996), no validation work has been conducted and large differences in exposure assessment because of differences in extraction procedures may thus occur.
An overview of population studies published in the scientific literature is presented in Table 1. A large number of health effects has been evaluated including lung function [forced expiratory volume in 1 s (FEV1) and peak flow (PEF) variability], nasal congestion, airway hyperreactivity, atopy, symptoms (upper and lower respiratory symptoms, eye irritations, head ache, fatigue/tiredness, joint pains, skin symptoms, flu-like symptoms, nausea, gastro-intestinal symptoms), inflammation characterized by inflammatory cells (T-lymphocytes, neutrophils, eosinophils, macrophages), and cytokines and other inflammatory markers [interleukin (IL)-1ß, IL-4, IL-6, IL-8, IL-10, Interferon (INF)-γ, Tumour necrosis factor (TNF)-α, Eosinophil cationic protein (ECP), Myeloperokidase (MPO), C-reactive protein (CRP), albumin] in blood, sputum and nasal lavage. Most studies included a range of these outcomes generally resulting in only a few significant associations (see Table 1). Although the focus has been on these positive findings (Rylander and Lin, 2000) it is clear that the results were not always consistent (see below).
Table 1. Overview of epidemiological studies, a case study and human challenge studies regarding ß(1→3)-D-glucan exposure and associated health effects
|Rylander et al., 1992||46||LAL||Problem buildings: 0.2–0.55 ng/m3||39||Dry cough ↑; skin rashes ↑a||Nose and eye irritations; chest tightness; head ache; tiredness; joint pains, etc.|
| Schools, post office, day care||36||Control building: <0.1 ng/m3||405|
|Rylander, 1997a†||24||LAL||Before renovation: 11.4 ng/m3||11||Airway hyperreactivity ↑a||Lung function; symptoms|
| Day care center||13||After renovation: 1.2 ng/m3|
|Thorn and Rylander, 1998a||75||LAL||0–19 ng/m3||129||Atopy ↑b; serum MPO ↑; FEV1↓b||Atopyc; Airway hyperreactivity; ECP; C-reactive protein; FEV; symptoms|
| Row houses|
|Rylander et al., 1998†||6||LAL||Problem school: 15.3 ng/m3||65||Cough ↑; cough with phlegm ↑, hoarseness ↑||Atopy|
| Schools||11||Control school: 2.9 ng/m3||141|
|Wan and Li, 1999||?||LAL||Day care centers: 5.7 ng/m3||40||Lethargy/fatigue ↑||Eye and nose irritations, skin and respiratory symptoms|
| Day care, Offices, homes||Office buildings: 3.2 ng/m3||69|
| ||Homes: 3.7 ng/m3||22|
|Douwes et al., 2000a||69||ELISA||Non-symptomatic children: 126 ng/m2||69||PEF variability in symptomatic children ↑d||No other health effects were studied|
| Homes||74||Symptomatic children: 169 ng/m2||74|
|Beijer et al., 2003* Row houses||17||LAL||High exposed: 6 ng/m3||17 high exp||Cytotoxic CD8+ T-cells ↑; IFN-γ/IL-4 ratio after in vitro stimulation of BMNCs ↑b||BMC secretion of IL-10 and Il-1β, serum ECP, MPO, IFN-γ and IL-4; differential cell counts in blood; symptoms|
| ||18||Low exposed: 0.9 ng/m3||18 low exp|
|Mandryk et al., 1999||54||LAL||Sawmill: 1.4 ng/m3||168||Base line lung function ↑; cross-shift lung function ↑↓;||No other health effects were studied|
| Sawmill, wood chipping, joineries||39||Joineries: 0.6 ng/m3|
|Mandryk et al., 2000||36||LAL||Green mills: 3 ng/m3||87||Base line lung function ↓; cross-shift lung function; respiratory symptoms ↑; these effects were observed only in green mill workerse||No other health effects were studied|
| Saw mills||18||Dry mills: ∼0.3 ng/m3|
|Rylander et al., 1999||20||LAL||Paper mill: 2.0–97.7 ng/m3||83||Throat and nose irritation ↑ (not significant); cough with phlegm ↑ (ns); joint pains↑; tiredness ↑; flu-like symptoms ↑; airway hyperreactivity (AH)↑, ECP in blood ↑; results for AH and ECP were not shown in the paper f||Lung function|
| Paper industry||8||Controls: 0.1 ng/m3||44|
|Thorn and Rylander, 1998b House hold waste collectors||20||LAL||Compostable waste: 19.1 ng/m3||25 collectors||Blood lymphocytes ↑||Respiratory and other symptoms; lung function; airway hyperreactivity; inflammatory markers in induced sputum and serum/blood|
| ||Unsorted waste: 9.2 ng/m3||20 controls|
| ||Controls: 1.1 ng/m3|| |
|Douwes et al., 2000b||60||ELISA||First survey: 0.54–4.85 μg/m3||14||No significant associations were found||Inflammatory markers in nasal lavage|
| Compost industry||43||Second survey: 0.36–4.44 μg/m3||15|
|Wouters et al., 2002||118||ELISA||1.3 μg/m3||47||No significant associations were found||Inflammatory markers in nasal lavage|
| Waste collectors|
|Heldal et al., 2003a||93||LAL‡||40 ng/m3||31||IL-8 in nasal lavage ↑; nasal congestion ↑||Neutrophils; MPO and ECP in nasal lavage|
| Waste handlers|
|Heldal et al., 2003b**||25||LAL||52 ng/m3||25||IL-8 in sputum ↑; analyses were not adjusted for endotoxin whereas endotoxin was associated with IL-8 and other inflammatory markers||Neutrophils; MPO and ECP in sputum|
| Waste handlers|
|Gladding et al., 2003 Waste recycling||156||LAL||4.8–40.1 ng/m3||159||Cough with phlegm ↑; hoarse/parched throat ↑; stomach problems ↑; other respiratory symptoms ↑ (ns); skin rash ↑ (ns); nausea ↑ (ns); analyses were not adjusted for endotoxin whereas endotoxin was associated with symptoms.||Blood lymphocytes, neutrophils, eosinophils and serum IgE|
|Blood monocytes and erythrocyte sedimentation rates were significantly lower in subjects working in one plant with elevated glucan levels compared with two other plants with lower glucan levelsf|
|Eduard et al., 2001||90||ELISA||0.82 μg/m3||106||No significant associations were found||Respiratory, eye and nose symptoms|
|Rylander et al., 1994||?||LAL||41.9 ng/m3||2||Airway inflammation ↑; cough ↑; wheeze ↑; tiredness ↑; symptoms disappeared after moving out of the mold infested house||No other health effects were studied|
|Human challenge studies|
|Rylander, 1996||–||–||210 ng/m3 (aerosolized and particulate curdlan)||26||Nose and throat irritations ↑; FEV1↓b; this was only observed for particulate curdlan||FEV; Airway hyperreactivity|
|Thorn et al., 2001||–||–||210 ng/m3 (grifolan suspended in saline)||21||Blood eosinophil levels ↓; TNF-α secretion by stimulated BMNCs ↓; FEV1↑||Sputum ECP, MPO, TNF-α, IL-8, IL-10, eosinophil, lymphocyte, macrophage, and neutrophils; Blood ECP, TNF-α, IL-10, eosinophil, lymphocyte, monocyte and neutrophils; lung function parameters other than FEV1|
|Beijer et al., 2002***||–||–||28 ng/m3 (grifolan suspended in saline)||17 high exp||TNF-α secretion by stimulated BMNCs ↓; blood lymphocytes ↑; These associations were only significant for the 17 subjects living in high exposure environments||IL-1β, IL-4, IFN-γ secretion by stimulated BMNCs; serum ECP and MPO|
|18 low exp|
|Sigsgaard et al., 2000||–||–||1 mg/ml (nasal instillation of solubilized curdlan)||5||Albumin and IL-1β in nasal lavage||TNF-α, IL-1β, IL-6 and IL-8 in nasal lavage|
Upper airway irritations, lung function and airway responsiveness
Several studies reported associations between (1→3)-β-D-glucan exposure and upper airway irritations and fatigue/tiredness (Gladding et al., 2003; Heldal et al., 2003a; Mandryk et al., 2000; Wan and Li, 1999), however, these associations were not confirmed in other studies (Rylander, 1997b; Rylander et al., 1992; Thorn and Rylander, 1998b). Also, no clear associations with lung function were found i.e. some studies reported adverse effects on lung function (Douwes et al., 2000a; Mandryk et al., 2000; Rylander, 1996) whereas others found no (Thorn and Rylander, 1998a,b) or even an opposite association (Mandryk et al., 1999; Thorn et al., 2001). Similarly, some studies reported significant effects on airway responsiveness (Rylander, 1997b) whereas others failed to demonstrate such an association (Thorn and Rylander, 1998a,b).
One study suggested that (1→3)-β-D-glucan was associated with an increased risk of atopy (Thorn and Rylander, 1998a), similar as has been observed in some animal studies (see below). However, this was not confirmed in a smaller study (Rylander et al., 1998). Also, an association between (1→3)-β-D-glucan exposure and an increased T helper 1 cell (Th1) immune response was suggested (Beijer et al., 2003) which appears to be contradictive with the finding of a higher prevalence of atopy in subjects with high (1→3)-β-D-glucan exposure (NB atopy is a Th2 driven immune response). An up-regulation of Th1 immune activity was, however, only shown in non-atopic subjects. Finally, some studies suggested that associations between (1→3)-β-D-glucan exposure and symptoms and lung function changes were stronger in atopics than in non-atopics (Douwes et al., 2000a; Rylander et al., 1998).
The data with regard to the potential effects of glucan on airway inflammation is also mixed. In vitro studies have demonstrated the potential of (1→3)-β-D-glucans to induce inflammatory responses (see below). In addition, several studies have shown that (1→3)-β-D-glucan may cause airway inflammation in laboratory animals (see below). However, in most population studies (and human challenge studies; see below) no such association was found (Table 1). Heldal et al. (2003a,b), found a significant association between (1→3)-β-D-glucan exposure and IL-8 in sputum and nasal lavage in waste handlers but that may have also been caused by highly correlated endotoxin exposures. Endotoxin levels in this study were, however, relatively low (7–180 EU/m3).
Some observational studies also focused on inflammatory mediators in blood; one study showed an association between (1→3)-β-D-glucan and MPO (Thorn and Rylander, 1998a) but this was not confirmed in another study (Rylander et al., 1999). Also, no association between (1→3)-β-D-glucan and MPO in sputum was found (Heldal et al., 2003a; Thorn et al., 2001). A positive association was demonstrated between (1→3)-β-D-glucan exposure and lymphocytes in blood (Thorn and Rylander, 1998b). Another study showed an inverse association between (1→3)-β-D-glucan exposure and the number of cytotoxic CD8+ T-cells (Beijer et al., 2003) and Phytohemagglutinin (PHA) and Lipopolysaccharide (or endotoxin) (LPS)-induced TNF-α production by blood mononuclear cells (BMNCs) (Beijer et al., 2002; Thorn et al., 2001). An inhibited endotoxin-induced TNF-α production by alveolar macrophages has also been demonstrated in a study in rats exposed to (1→3)-β-D-glucan (see below). The meaning of these findings is not clear and results require further confirmation from both human and animal studies.
Limitations of the epidemiological studies
The population studies described above (and summarized in Table 1) suggest an association between exposure and a wide range of health effects, however, only a small number of all tested variables showed significant associations and results were often inconsistent. The lack of consistency between studies may largely be because of the relatively small sample size of most of the reported studies; some studies in fact involved a comparison of only two exposure situations (Rylander, 1997b; Rylander et al., 1998). Also, often only relatively few exposure measurements were taken. Therefore, uncertainty in exposure assessment was large which may have resulted in obscured exposure-response relationships. Another limitation in many of the reported field studies was the lack of control for potential confounders such as other common bioaerosol exposures that may cause similar health effects (Douwes et al., 2003). A number of studies showed a strong correlation between endotoxin and (1→3)-β-D-glucan levels (Douwes et al., 2000a,b; Gladding et al., 2003; Rylander et al., 1999) suggesting that some of the findings may potentially have been caused by endotoxin. However, in some of the studies where analyses were adjusted for allergen and endotoxin no major changes in the relationship between (1→3)-β-D-glucan and health effects were observed after adjustment (Douwes et al., 2000a; Mandryk et al., 2000). Another potential weakness is that in several studies (Gladding et al., 2003; Rylander et al., 1999; Thorn and Rylander, 1998a) exposure groups were constructed using cut-off points that were not based on objective criteria, but in stead the reasons for choosing cut-off points were not clear and did not appear to be driven on an a priori hypothesis precluding a straight forward interpretation. Similarly, in some studies (Beijer et al., 2003; Thorn and Rylander, 1998a) associations were only found when part of the study population was excluded whereas reasons for exclusion were often unclear. For example, one study in the general population (Thorn and Rylander, 1998a) showed that baseline FEV1 values were inversely related to (1→3)-β-D-glucans, but only when analyses were restricted to those younger than 65 yr and with glucan exposures above 1 ng/m3 (exclusion criteria were not motivated). When the data were analyzed using three instead of two exposure groups (criteria for cut-off points were unclear) no association with FEV1 was found. As criteria for subgroup analyses and exposure grouping were not clearly described results from these studies should be interpreted with caution.
Human challenge studies
In addition to the population studies discussed above several experimental exposure studies have also been conducted (Table 1). The first study showed a small increase in the severity of symptoms of nose and throat irritations after a challenge of aerosolized (1→3)-β-D-glucan (Rylander, 1996). No effects on FEV1 or airway responsiveness were found. Exposure to particulate (1→3)-β-D-glucan resulted in a significant but very small decrease in FEV1 directly and 3 days after exposure; no significant association was found with airway responsiveness. Another study failed to show airway inflammation after a (1→3)-β-D-glucan challenge (Thorn et al., 2001). Seventy-two hours after the challenge both blood eosinophil levels (borderline significant, P < 0.06) and TNF-α secretion by PHA stimulated BMNC were lower. No significant differences in lung function values were observed at 24 h, however, the FEV1 value was significantly higher 72 h after glucan challenge. A third study showed that (1→3)-β-D-glucan exposure decreased TNF-α production of BMNCs in subjects with previous high glucan exposures but not in those with low indoor exposures (Beijer et al., 2002). No significant differences were found in BMNC secretion of the other studied cytokines (Table 1). Also no differences in ECP and MPO in serum were observed.
Sigsgaard et al. (2000) demonstrated an increase in albumin and a slight increase in IL-1ß in the nasal mucosa in a small group of Danish garbage workers (n = 5) and controls (n = 5) who were exposed to solubilized (1→3)-β-D-glucan by nasal instillation. No increase in the other tested cytokines was observed (Table 1). However, in a whole blood assay measuring cytokine release after ‘ex vivo’ exposure to high concentrations of (1→3)-β-D-glucans, a significant increase in all measured cytokines (TNF-α, IL-1ß, IL-6 and IL8) was found. This was confirmed in several other studies in which blood was collected from healthy volunteers (Wouters et al., 2002), atopics and non-atopics (Kruger et al., 2004), and municipality and fish factory workers (Bønløkke et al., 2004).
Results of the whole blood experiments demonstrate the pro-inflammatory potential of glucans at very high levels. However, in vivo challenge studies as described above show only minor inflammatory effects and no clear effects on lung function. However, these mainly negative results should be interpreted with caution as only a few are available and experiments were conducted with only two types of (1→3)-β-D-glucan (curdlan and grifolan) which may not necessarily be the most biologically active.
Studies in rats have demonstrated the inflammatory potency of (1→3)-β-D-glucans using high concentrations of Zymosan. Intratracheal installation of Zymosan induced a dose-dependent neutrophil influx into the airways and an increase in pulmonary albumin, breathing frequency and nitric oxide production by alveolar macrophages (Young et al., 2001). Exposure to particulate (1→3)-β-D-glucan induced greater pulmonary toxicity than soluble (1→3)-β-D-glucans (Young et al., 2003a). Also, a partially opened triple helix conformation of (1→3)-β-D-glucans was more active than the closed formation (Young et al., 2003b). The inflammatory potency thus appears strongly dependent on the type and conformation of (1→3)-β-D-glucans. However, there is reason for caution as some of the inflammatory activity may be because of other components as Zymosan is a rather crude mixture of glucans and other yeast cell wall components. In addition, other studies in guinea pigs did not show an acute inflammatory effect of (1→3)-β-D-glucan exposure (Fogelmark et al., 1992, 1994). Repeated exposures on the other hand did result in an increase in inflammatory cells in the airways (Fogelmark et al., 1992, 1994), particularly eosinophils (Fogelmark et al., 2001). This was not confirmed in mice, in which lung specimens were normal (i.e. no signs of inflammation or fibrosis) at histopathological examination, even after repeated exposures to aerosolized (1→3)-β-D-glucan (Korpi et al., 2003).
Several studies have explored the potential modifying effects of (1→3)-β-D-glucan on endotoxin toxicity. In guinea pigs it was shown that the inflammatory response to endotoxin was reduced by simultaneous (1→3)-β-D-glucans exposure (Fogelmark et al., 1992, 1994). An inhibitory effect of (1→3)-β-D-glucan (Zymosan) on pulmonary responsiveness has also been demonstrated in rats, but only when they were pre-treated with (1→3)-β-D-glucan, and only for certain indicators of airway inflammation such as lactate dehydrogenase activity, albumin level, and pulmonary TNF-α, and alveolar macrophage production of reactive oxidant species. (Young et al., 2002). The endotoxin-induced increased breathing rate and neutrophil infiltration were not affected. In contrast, other studies have suggested that (1→3)-β-D-glucan may enhance the toxicity of endotoxin (Bower et al., 1986; Cook et al., 1980). Also, in guinea pigs, repeated exposures to a combination of (1→3)-β-D-glucans and endotoxin resulted in a larger increase in inflammatory cells than for the two components separately, suggesting an enhanced rather than an inhibitory effect (Fogelmark et al., 1992, 1994).
Finally, several studies assessed the effects of (1→3)-β-D-glucan on the development of specific IgE sensitization. One study in mice showed that pre-exposure to inhaled (1→3)-β-D-glucans enhanced ovalbumin-induced IgE antibody responses and airway antigen-specific allergic reactions including eosinophil infiltration (Wan and Li, 1999). In addition, it appeared that spleen cells derived from (1→3)-β-D-glucan exposed mice produced significantly higher amounts of IL-4 (IL-4 stimulates B-cells to IgE production and naïve Th0 cells to differentiate toward an atopic Th2 state) (Wan and Li 1999, unpublished results). This suggests that (1→3)-β-D-glucans may enhance atopic Th2 immune responses (Rylander and Lin, 2000). A study in guinea pigs that were exposed daily to an aerosol of (1→3)-β-D-glucans for a period of 5 weeks showed an increase in eosinophil numbers in the airways of exposed animals (Fogelmark et al., 2001) confirming the previous study to a certain extent. In addition, two other studies demonstrated that (1→3)-β-D-glucans isolated from barley, the fungus Sclerotinia sclerotiorum, and baker's yeast significantly increased the IgE immune response in mice (Instanes et al., 2004; Ormstad et al., 2000). In contrast, another study in guinea pigs showed that (1→3)-β-D-glucan exposure caused a decrease in ovalbumin induced eosinophil numbers (Rylander and Holt, 1998).
Thus, animal experiments suggest that (1→3)-β-D-glucan has the potential (in high concentrations) to (i) induce neutrophilic and possibly eosinophilic airway inflammation; (ii) modify endotoxin-induced airway inflammation; and (iii) modify the atopic immune response. However, there is considerable reason for caution as the results were not always consistent. This may be because of differences in solubility and conformation of the (1→3)-β-D-glucans used in these experiments (Young et al., 2003a,b). Differences in administration and timing of (1→3)-β-D-glucan exposure may be another reason.
If environmental exposures to (1→3)-β-D-glucans cause health effects as suggested then control is essential. Specific measures to control exposure have not been developed but several studies investigated the association between exposure and housing characteristics and occupant behavior. These studies may help to develop procedures to manage exposure in the home environment.
Gehring et al. (2001) showed that (1→3)-β-D-glucan levels in house dust were strongly associated with the amount of dust on the floor. Presence of carpets, mold growth and pets (particularly dogs) in the home, number of occupants, time spent indoors by the occupants, and low frequency of cleaning were also associated with higher levels of dust and (1→3)-β-D-glucan (Douwes et al., 1998; Gehring et al., 2001; Rylander et al., 1994, 1998). A (weak) positive association between indoor relative humidity and (1→3)-β-D-glucan levels in the home has also been demonstrated (Gehring et al., 2001). Also, one study reported that homes in which organic waste was separated and stored indoors in kitchen compost bins (as is becoming increasingly common in many European countries) had substantially higher levels of (1→3)-β-D-glucan indoors (Wouters et al., 2000). Therefore, (1→3)-β-D-glucan levels may be lowered by measures to reduce indoor dust, mold growth and humidity. In addition, avoiding the use of indoor compost bins for organic waste storage may also lower indoor (1→3)-β-D-glucan levels.
The observational and experimental studies described above suggest some association between (1→3)-β-D-glucan exposure, airway inflammation and symptoms, however, results are mixed and specific symptoms associated with exposure can at this stage not be identified. Based on subgroup analyses, it has been speculated that atopics and/or subjects with pre-existing symptoms may be more susceptible (Douwes et al., 2000a; Rylander et al., 1998) but this requires further study. Ex vivo observations and animal experiments indicate that (1→3)-β-D-glucan at high concentrations have the potential to initiate airway inflammation, however, in inhalation experiments in human (at much lower doses) these effects could not consistently be reproduced. This may be because of the fact that inhalation experiments were conducted with soluble (1→3)-β-D-glucan that may not represent the most potent (1→3)-β-D-glucan fraction. Currently, a clear interpretation of the potential health effects of (1→3)-β-D-glucan is hampered because of the fact that (i) only a relatively small number of observational studies are available, most of which lacked statistical power; (ii) some of these studies have not appropriately controlled the analyses for other potential causal exposures and/or had weaknesses in the design and/or statistical analyses; and (iii) different methods to assess exposure were used and only relatively few samples were collected, potentially resulting in substantial exposure misclassification obscuring exposure – response relationships. Thus, the currently available epidemiological data do not permit conclusions to be drawn regarding the presence (or absence!) of an association between environmental glucan exposure and specific adverse health effects, nor is it clear from the currently available evidence which specific inflammatory mechanisms underlie the presumed health effects.
Recommendations for future research
In addition to more and larger observational studies there is a need for human challenge studies testing various conformations of (1→3)-β-D-glucans (triple, single helix and random coil) as well as soluble and non-soluble forms of glucan. Moreover, there is a clear need for validation and further development of existing methods to measure environmental (1→3)-β-D-glucan. This should include a comparison of extraction methods as well as a comparison between the currently commonly applied assays to measure (1→3)-β-D-glucan i.e. the LAL and the inhibition ELISA assay. Other important areas that require further research include: (i) issues regarding individual susceptibility for (1→3)-β-D-glucans; (ii) the potential interaction effects between (1→3)-β-D-glucans and other exposures such as allergens and endotoxin; (iii) more research into other health effects such as skin conditions, neurological and gastro-intestinal symptoms, all of which have been speculated to be related to (1→3)-β-D-glucan exposure; (iv) the assessment of the biological properties of non-fungal (1→3)-β-D-glucan; and (v) the assessment of determinants of exposure to allow more specific control measures to be developed.
The author would like to thank Professor Neil Pearce and Dr Wijnand Eduard for their valuable comments on the draft manuscript. Jeroen Douwes is supported by a sir Charles Hercus Research Fellowship from the Health Research Council (HRC) of New Zealand. The Centre for Public Health Research is supported by a Program Grant from the HRC of New Zealand.
(1→3)-β-D-glucan are non-allergenic structural cell wall components of most fungi that have been suggested to play a causal role in the development of respiratory symptoms associated with indoor fungal exposure. This review describes the currently available epidemiological literature on health effects of (1→3)-β-D-glucan, focusing on atopy, airway inflammation and symptoms, asthma, and lung function. In addition to population studies, studies in human volunteers experimentally exposed to (1→3)-β-D-glucan are described as well as relevant animal studies. Furthermore, the review discusses exposure assessment methods, the potential for exposure control and it concludes with identifying research needs. The observational and experimental studies reviewed suggested some association between (1→3)-β-D-glucan exposure, airway inflammation and symptoms, however, results were mixed and specific symptoms and potential underlying inflammatory mechanisms associated with exposure could not be identified. Large observational studies using well validated exposure assessment methods are needed to further our knowledge regarding the potential health effects of indoor (1→3)-β-D-glucan exposure.