If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Work exposures play a significant role in adult-onset asthma, but the mechanisms of work-related asthma are not fully elucidated.
We aimed to reveal the molecular mechanisms of work-related asthma associated with exposure to flour (flour asthma), isocyanate (isocyanate asthma), or welding fumes (welding asthma) and identify potential biomarkers that distinguish these groups from each other.
We used a combination of clinical tests, transcriptomic analysis, and associated pathway analyses to investigate the underlying disease mechanisms of the blood immune cells and the airway epithelium of 61 men.
Compared with the healthy controls, the welding asthma patients had more differentially expressed genes than the flour asthma and isocyanate asthma patients, both in the airway epithelia and in the blood immune cells. In the airway epithelia, active inflammation was detected only in welding asthma patients. In contrast, many differentially expressed genes were detected in blood cells in all 3 asthma groups. Disease-related immune functions in blood cells, including leukocyte migration and inflammatory responses, and decreased expression of upstream cytokines such as TNF and IFN-γ were suppressed in all the asthma groups. In transcriptome-phenotype correlations, hyperresponsiveness (R ∼ |0.6|) had the highest clinical relevance and was associated with a set of exposure group–specific genes. Finally, biomarker subsets of only 5 genes specifically distinguished each of the asthma exposure groups.
This study provides novel data on the molecular mechanisms underlying work-related asthma. We identified a set of 5 promising biomarkers in asthma related to flour, isocyanate, and welding fume exposure to be tested and clinically validated in future studies.
Asthma is a heterogeneous disease driven by interactions between the airway epithelium, the immune system, and environmental exposure. It can be divided into several phenotypes or endotypes based on clinical, functional, and inflammatory features.
Isocyanates are chemicals that are widely used in the manufacture of polyurethane foams, sealants, elastomers, adhesives, and coatings. They act mainly via IgE-independent mechanisms that are not fully delineated, and nonadaptive immune responses may be involved.
The welding process releases potentially hazardous gases and fumes composed mainly of metallic oxides, fluorides, and silicates. These fumes include fine and ultrafine particles that are able to reach terminal bronchioles and alveoli. Inflammation, lung defense suppression, oxidative stress, DNA damage, and genotoxic effects have been observed after exposure to welding fumes.
Linking pathobiologic mechanisms to clinical variables of asthma such as age at onset, symptom triggers, lung function, and inflammation type is increasingly being used to identify asthma endotypes. When asthma endotypes are defined by distinct pathophysiologic mechanisms, gene expression profiles of blood and nasal epithelia samples have shown promising results in identifying childhood asthma,
Asthma related to different well-characterized workplace exposures comprise an interesting model to investigate asthma endotypes.
We aimed to evaluate the mechanisms of WRA by assessing clinical parameters and transcriptomic profiles in airway epithelium (nasal mucosa) and blood immune cells (PBMCs). We assessed patients with WRA who were exposed to 1 of 3 items encountered in the workplace (flours, isocyanates, or welding fumes) that are known to act via different mechanisms without categorizing them into WRA subtypes with clinical tests. We sought to identify specific biomarkers for different exposure-related endotypes of WRA by combining clinical and transcriptomic findings.
Men aged 18 to 65 years were recruited. The study patients with WRA were patients from the tertiary occupational disease clinic at the Finish Institute of Occupational Health from 2015 to 2018. Clinical assessment was carefully performed by a respiratory physician (H.S. or I.L.). All subjects were interviewed, and they filled out a clinical questionnaire. The inclusion criteria were as follows: (1) diagnosis of asthma; (2) exposure to flour, isocyanates, or welding fumes while at work; (3) emergence of asthma symptoms during work exposure; and (4) aggravation of asthma symptoms during work periods and relief of symptoms during periods off work. Asthma diagnosis was verified from the participants' medical files; they all showed reversible airway obstruction or bronchial hyperresponsiveness. The WRA groups consisted of men with (1) flour asthma (FA), (2) isocyanate asthma (IA), and (3) welding asthma (WA).
The control group was recruited by using advertisements and social media. The control group members did not have allergic symptoms related to common environmental allergens, asthma, chronic cough, or dyspnea; nor did they have occupational exposure to flours, isocyanates, or welding fumes.
The study was approved by the ethics committee of Helsinki University Central Hospital (approval no. 291/13/03/00/2014). All participants signed forms indicating their informed consent.
Nasal biopsy samples and PBMCs were obtained from all subjects. The aim was to sample patients with WRA within 24 hours after relevant exposure. None of the study participants had respiratory infection within 2 weeks before sampling, and they were advised to stop using nasal corticosteroids 1 month earlier when possible.
ICS course for healthy controls
For safety reasons, inhaled corticosteroids (ICSs) were not withdrawn from the asthma patients before sample collection. To control the systemic effect of ICSs in the analysis, we assessed the effect ICS use in the healthy controls, who used a high-dose ICS (budesonide administered via a dry powder inhaler at a dose of 800 μg per day) for 5 to 6 weeks; PBMC samples were taken before and after the treatment.
Fig E1 (available in this article's Online Repository at www.jacionline.org) shows a workflow outlining the allocation of the patient groups, sampling, and analyses. The methods of clinical tests and transcriptomic analysis are described in the Methods section of the Online Repository (available at www.jacionline.org).
SPSS, version 25.0, software (IBM Corporation, Armonk, NY) was used for analysis of demographic and clinical parameters. Medians and interquartile ranges were used to summarize continuous measures, whereas categoric variables were defined by their frequencies and proportions. Between-group differences were analyzed by using the Kruskal-Wallis test for continuous variables and chi-square tests for categoric variables. The eUTOPIA bioinformatic package
by using age, body mass index (BMI), smoking status, and ICS use as covariates where applicable. A fold change of 1.5 or greater and a Benjamini-Hochberg adjusted P value less than .05 were considered indicative of significantly differentially expressed genes (DEGs).
Clusters and heatmaps were generated by using the Perseus program.
K-means clustering and principal component analysis (PCA) were used to visualize clustering of the gene profiles. The physiologic implications of the DEGs were predicted via pathway enrichment analyses by using Ingenuity Pathway Analysis tool (QIAGEN, Hilden, Germany).
Demographic and clinical characteristics
The study population comprised 61 men: 46 patients with WRA who were exposed to flours (the FA group; n = 15), isocyanates (the IA group; n = 20), or welding fumes (the WA group; n = 11) and 15 heathy control subjects. In the IA group, 11 subjects were exposed to methylene diphenyl diisocyanate, 7 were exposed to hexamethylene diisocyanate, and 4 were exposed to toluene diisocyanate. In the WA group, all participants were exposed to stainless steel fumes in their workplace. A total of 15 participants were diagnosed with occupational asthma. We did not detect statistically significant differences between the study groups in terms of age, smoking status, BMI, or atopy (Table I
). Duration and frequency of work exposure, symptomatic work exposure, and asthma control did not differ significantly between the asthma groups. Work-related rhinitis was more common in the asthma groups than in the control group (P < .001). Blood eosinophil count did not differ significantly between the study groups. Total IgE level (P = .004) and fractional exhaled nitric oxide (Feno) level (P = .011) were highest in the FA group. During sampling, 30 patients with asthma (65%) used an ICS. Lung function did not differ significantly between the groups, and the patients with asthma had more bronchial hyperresponsiveness than the controls did (P = .004). Objective tests showed association between asthma and work exposure (IgE-mediated sensitization to occupational allergen, suggestive workplace peak expiratory flow measurement, or positive specific inhalation challenge) in 39 patients with WRA (85%) (Table I and see Table E1 [available in this article's Online Repository at www.jacionline.org]).
Table ICharacteristics of the 61 men participating in the study
Gene expression is significantly stronger, especially in nasal biopsy samples, in WA patients than in FA and IA patients
Microarray-based transcriptomic profiling was carried out on RNA extracted from both nasal biopsy samples and PBMCs. The top 2 principal components, explaining 45% (in the case of nasal biopsy samples) and 59% (in the case of PBMCs) of the variation between the samples, are shown in Fig E2 (available in this article's Online Repository at www.jacionline.org). In the nasal biopsy samples, only individuals with WA clustered separately from the controls (Fig E2, A). All 3 WRA categories were visibly distinct from the controls in the PBMC samples (Fig E2, B). To identify disease-relevant genes for each exposure group, the nasal biopsy sample and blood cell gene expression profiles of the control subjects were compared with those of each of the FA, IA, and WA patients. Differential gene analysis was adjusted for age, BMI, smoking status, and ICS use. The number of DEGs was clearly highest in the WA patients (Fig 1, A). In the nasal biopsy samples, only 1 DEG was identified in the FA group, 24 DEGs were identified in the IA group, and 664 DEGs were identified in the WA group compared with in the controls (Fig 1, B). In their PBMCs, the FA patients exhibited 692 DEGs, the IA patients exhibited 585 DEGs, and the WA group exhibited 1544 DEGs compared with the controls (Fig 1, C). Taken together, these data indicate that much larger numbers of DEGs were detected in the PBMCs from each WRA exposure group than in their nasal biopsy samples.
Disease-related immune functions are suppressed in the PBMCs from all asthma exposure groups but enhanced in the nasal biopsy samples from WA patients
To gain further insight into the underlying mechanisms of disease, all of the identified DEGs were subjected to functional enrichment analysis. Although Venn comparisons of significantly enriched diseases and biofunctions revealed unique disease mechanisms for each asthma exposure group (Fig 2, A and B), the most significant gene functions were shared by the members of the IA and WA groups (nasal biopsy DEGs) and between the FA, IA and WA groups (PBMC DEGs). In the nasal biopsy samples, the top significant functions common to the IA and WA categories were quantity of antigen presenting cells and quantity of leukocytes (Fig 2, A). In the blood cells, genes involved in neutrophil degranulation, leukocyte migration, and general inflammatory response were identified as the top enriched functions common to all asthma groups (Fig 2, B). A heatmap based on significantly enriched pathways separates the nasal biopsy sample groups from the blood cell sample groups and further substantiates WA as the most pathologic asthma group in this cohort (Fig 2, C). Similar to the gene expression profiles, the significance of enriched disease or biofunctions was greater in the PBMCs than in the nasal biopsy samples (Fig 2, C).
Ranking of the top 20 enriched functions and a hierarchic cluster of their predicted activation status are shown in Fig 3, A and B, respectively. Genes involved in leukocyte migration, degranulation of neutrophils, and general inflammatory response were highly represented (a false discovery rate of 1E–58.1 to 1E–15.6) in the DEGs from the peripheral blood cells of the FA, IA, and WA groups (Fig 3, A). The DEGs in the nasal biopsy samples of the WA patients represent the same diseases and biofunctions as the DEGs identified in the blood cells of the FA, IA, and WA patients. However, the underlying pathways are predicted to be activated in the nasal biopsy sample gene profiles of the FA, IA, and WA patients, the same pathways are predicted to be inactivated in the PBMCs of WA patients (Fig 3, B).
We next sought to answer the question of whether some of these DEGs are under the control of the upstream inflammatory regulators that play important roles in immune system disorders. The top 20 upstream regulators predicted in the Ingenuity Pathway Analysis are shown in Fig E3, A (available in this article's Online Repository at www.jacionline.org). These regulators include the proinflammatory cytokines TNF, IFN-γ, TGF-β1, IL-1β, and CSF2, as well as anti-inflammatory cytokines such as IL-13, IL-4, IL-6, and IL-10. Yet again, the nasal biopsy sample gene expression profile was consistent with activation of these cytokines, the PBMC gene expression profile revealed suppression or inactivation of these same cytokines (see Fig E3, B). The trend in expression of selected upstream regulators is in line with the activation states predicted from the expression profiles of their downstream targets (see Fig E4 [available in this article's Online Repository at www.jacionline.org]).
Gene-phenotype correlations reveal genes with high clinical relevance
Hierarchic clustering based on the Pearson correlation coefficient revealed 2 distinct clusters of clinically relevant disease features (see Fig E5 [available in this article's Online Repository at www.jacionline.org]). Clinical measures of lung function, eosinophilic inflammation, bronchial hyperresponsiveness, atopy, and asthma control make up cluster 1, whereas demographic parameters and features related to disease severity and exposure (age, smoking, BMI, ICS dose, frequency of exposure, duration of exposure, and duration of symptoms) make up cluster 2 (see Fig E5).
We next sought to answer the question of whether the expression of any disease-relevant genes identified from PBMC transcriptome profiling correlates with specific diagnostically relevant clinical features in cluster 1. In all 3 asthma groups, genes with a Pearson correlation coefficient (R) of –0.6 to 0.6 to at least 1 clinical feature were included. A heatmap illustrating correlations of DEGs to diagnostically relevant clinical criteria in each of the patient groups is shown in Fig 4. The correlated genes and their corresponding correlation coefficients are provided in Table E2 (available in this article's Online Repository at www.jacionline.org). PCA analysis showed that the expression profile of these highly correlating genes clustered the patients separately from the controls in all of the asthma groups (Fig 4, A-C [lower panel]). Bronchial hyperresponsiveness had the strongest correlation to gene expression for all of the asthma groups (Fig 4, A-C). The gene correlations with all other clinical parameters were weak or nonexistent, except for specific IgE level and total IgE level, which had the strongest correlation with gene expression in the FA group (Fig 4, A).
Similarly, we identified several DEGs in the PBMCs that were associated (–0.6 ≤ R ≥ 0.6) with at least 1 parameter from cluster 2. A heatmap of the correlated genes for each WRA exposure group is depicted in Fig E6, A (available in this article's Online Repository at www.jacionline.org). The strongest modulators of gene expression among the WRA patients were ICS use in all WRA subtypes; exposure frequency in the IA subtype; smoking in the IA and WA subtypes; and age, exposure duration, symptom duration, and BMI in the FA subtype. The top enriched biologic processes represented by patient demographics–associated DEGs in all WRA subtypes was regulation of cytokine production (see Fig E6, B).
Biomarker sets with only 5 genes accurately identify WRA exposure–related endotypes
We next sought to identify endotype biomarkers that could facilitate diagnosis by identification of group-specific DEGs that are associated with clinical features. Fig 4 shows that 302, 189, and 67 genes were associated with at least 1 clinical feature in the FA, IA, and WA patients, respectively. A Venn comparison of these clinical feature–associated DEGs is shown in Fig E7, A (available in this article's Online Repository at www.jacionline.org), and the most significant pathways represented by these genes are shown in Fig E7, B. The top enriched disease or functions identified by these genes are quantity of myeloid cells and cellular infiltration by leukocytes in FA patients, degranulation of immune cells in IA patients, and quantity of immune cells in WA patients. To further refine the diagnostic potential, Venn comparisons were used to identify unique DEGs that are associated with diagnostically relevant clinical features. This way we end up with genes that are both associated with at least 1 clinical feature and uniquely differentially expressed in that particular asthma group. The unique DEGs (Fig 1, C) comprised 42, 36, and 734 genes in the FA, IA, and WA endotypes, respectively. The expression profile of these asthma exposure group–specific DEGs in PBMCs separated all of the patients from the controls (see Fig E7, C). The separation of patient samples from control samples was most distinct in the FA and WA groups. In total, 24, 9, and 49 of such genes were identified from the FA, IA, and WA groups, respectively (see Fig E7, D). We then calculated the intraclass correlation coefficients for each of these genes (see Table E3 in this article's Online Repository at www.jacionline.org).
The PCA plots based on the expression profile of the top 5 genes for each of the asthma exposure groups are shown in Fig 5. We propose these top 5 genes as biomarker panel candidates for FA (Fig 5, A), IA (Fig 5, B), and WA (Fig 5, C). These biomarkers are as follows: downregulation of MYOM2, CDK5R1, ZDBF2, and VCPIP1 together with upregulation of ARPC5L to FA; downregulation of TST, TMEM55A, SPHAR, and DNAJA1 together with upregulation of ANO6 to IA; and downregulation of ITGA5 together with upregulation of HERC2P4, DHRS3, NUCB2, and DSC1 to WA.
The mechanisms of WRA are not fully elucidated, and disease endotype–specific biomarkers are not currently known. We used a combination of clinical tests and transcriptomic analysis of blood immune cells and airway epithelium to investigate underlying disease mechanisms and identify endotype-specific biomarkers. Elaborating on pathologic mechanisms and identifying disease-associated biomarkers in WRA related to different exposures have the potential to improve diagnosis and facilitate the development of targeted treatment.
We studied WRA patients with exposure to 3 major workplace agents (ie, flour, isocyanates, and welding fumes) associated with different disease mechanisms. WRA can be related to various other workplace exposures, and our results cannot be generalized to WRA related to other exposure types.
The WRA patients were recruited from a tertiary occupational medicine clinic in which the patient population was preselected to have a high probability of occupational disease. The WRA patients were not categorized into clinical subgroups of WRA (occupational asthma, work-exacerbated asthma). Clinical tests targeted to diagnose occupational asthma are typically aimed at detection of sensitizer-induced asthma, whereas welding fumes act via different mechanisms.
We studied WRA without preselection based on clinical testing and used the same inclusion criteria in all exposure groups (asthma emergence and aggravation of symptoms following exposure to specific occupational agents). Nevertheless, we detected biomarkers that identify each WRA exposure group, supporting homogeneity and similar background mechanisms among the participants in each group. Overall, our results shed light on mechanisms of asthma related to certain work exposures rather than on mechanisms of WRA categories (occupational or work-exacerbated asthma), and the results cannot be applied in the differential diagnosis of them. All of the WRA patients fulfilled the diagnostic criteria of asthma before sampling. At the time of sampling, most of the patients were using an ICS, and their clinical test results showed mainly well or partly controlled asthma.
To diminish heterogeneity of the study population, all of the participants were male and all of the WRA patients were currently exposed to the relevant occupational agent. Demographic variables did not differ between the groups. Because ICS treatment had a great effect on transcriptome in previous studies,
to avoid bias related to ICS, steroid effect was examined in healthy controls and taken into account in the analysis. Similar to in previous studies, most of the WRA patients reported work-related rhinitis.
Similar to in previous studies, asthma related to high-molecular-weight allergens (flour) showed IgE-mediated sensitization to the occupational agent and had a high Feno level reflecting T2-derived inflammation
Localized (nasal biopsy samples) and systemic (PBMCs) gene expression was profiled after exposure to either flour, isocyanates, or welding fumes. The several times higher number of DEGs in blood immune cells relative to the number in nasal biopsy samples suggests that the pathomolecular changes existed predominantly in the systemic circulation at the time of sampling. Our results also suggest that ongoing inflammation in the airway epithelium in patients with FA or IA was virtually nonexistent. It is possible, however, that the use of ICSs may partially explain why inflammation in the airway epithelium could not be identified in these WRA groups. In contrast, exposure to welding fumes was related to changes in gene expression that were detectable both in the airway epithelium and in PBMCs, which is indicative of a hazardous inflammatory process in the airways. On the basis of an extensive literature search, it appears that comparable comprehensive transcriptomics studies in human subjects are practically nonexistemt,
With regard to the most robust results derived from the pathway enrichment analysis, all of the asthma groups exhibited significant suppression of immune functions related to cell migration and cell movement in blood immune cells. Only WA patients demonstrated enrichment of cell migration–related immune functions, but contradictory to what was observed in PBMCs, those pathways were enhanced in the airway epithelia. Similar phenomena could also be seen with predicted upstream regulators in WA patients: several inflammatory cytokines (eg, TNF, IFN-γ, IL-1β, and TGF-β1) were suppressed or inactivated in the blood immune cells but stimulated or activated in the nasal epithelia. This suggests that the changes in gene expression observed in WA patients, both in their nasal epithelia and in their peripheral blood cells, are controlled by a central pathologic mechanism. It can be speculated that these contradictory gene expression profiles are due to differences in the inflammatory cell pools in the nasal epithelium and in the circulation. In other words, the inflammatory cells that are likely the main source of the exposure-related gene expression are recruited to the site of the inflammation (ie, the airway epithelia) and are therefore missing from the circulation. In the nasal biopsy samples, because of the limited number of identified DEGs, it is unsurprising that very few to no significantly enriched immune functions were identified from the FA and IA groups.
To identify genes with high clinical relevance, we carried out correlation analysis between DEGs identified in blood cells and several clinical parameters. Bronchial hyperresponsiveness showed the strongest correlation with gene expression in all of the asthma groups. Interestingly, Feno level and specific and total IgE levels demonstrated moderate correlations with gene expression in the FA group but not in the other groups. This is likely due to the more prominent IgE-associated and T2-type immunity in the FA patients than in the IA and WA groups, which are more non–T2-associated asthma types. Identification of genes correlating with important clinical function may facilitate development of a molecular diagnostic tool for more accurate clinical diagnosis.
The effects of smoking status, subject age, BMI, and ICS dose were adjusted during between-group (patient vs control) differential expression analysis. Within each WRA exposure group, however, in addition to exposure duration and frequency and symptom duration, these factors were also found to affect expression of disease-related genes. The involvement of these genes in regulation of cytokine production indicates that these factors may modulate disease-related features by affecting the extent of proinflammatory and/or anti-inflammatory signaling. This finding is consistent with the findings of a previous report indicating that the severity of asthma is associated with older age, higher BMI, and higher rate of tobacco smoking
However, a larger cohort will be needed to fully unravel the mechanisms via which these factors influence disease features.
Despite the overlapping gene expression profiles between the WRA groups in the blood immune cells, which is indicative of similarities in the pathologic mechanisms, exposure-specific differences were also found. To identify specific exposure-related biomarkers, we combined and analyzed the information on the unique DEGs identified in PBMCs, as well as information on the DEGs that are associated with the clinical parameters. Our prioritization analysis revealed specific sets of 5 candidate genes for each exposure-related endotype that are capable of distinguishing it as a disease subgroup. Further studies in unrelated patient materials are needed to confirm and validate the predictability of our biomarkers.
This study sheds light on asthma mechanisms related to 3 different types of exposures present at workplace. Active inflammation in airway epithelia could be detected from the gene expression only in WA. In contrast, large numbers of DEGs were detected in the blood immune cells in all of the patient groups. Gene-phenotype correlations revealed genes with high clinical relevance: bronchial hyperresponsiveness was the strongest clinical feature correlating with gene expression. Our study identified biomarker sets capable of accurately identifying asthma related to flour, isocyanates, or welding. Our results could provide a basis for future attempts to create diagnostic biomarkers for clinical settings.
Many DEGs were detected in blood immune cells of patients with work-related asthma. Disease-related immune functions, including leukocyte migration and inflammatory responses, and decreased expression of upstream cytokines such as TNF and IFN-γ were suppressed in the study patients.
In transcriptome-phenotype correlations, hyperresponsiveness had the highest clinical relevance and was associated with a set of genes specific to asthma related to flour, isocyanate, or welding fume exposure.
Biomarker sets containing 5 genes capable of accurate identification of the endotypes have been identified for each of the work-related asthma exposure groups.
We thank Sirpa Hyttinen, Tanja Katovich, and the study nurses for their assistance in this study, as well as Terhi Vesa, MSc, Varpu Alenius, and Mikko Asiala for their excellent help in processing cell samples. Additionally, we thank Professor Dario Greco for valuable discussions on the study design.