After an initial appearance in childhood, asthma may remain inactive for an extended period or reappear later in life. A number of recent studies have also described the onset of asthma during adulthood (1). Thus, early-onset asthma (EOA) and late-onset asthma (LOA) can be viewed as two distinct phenotypes, based on the categorization of disease entities according to underlying mechanisms or endotypes, such as risk factors, remission rates, co-morbidities, and gene expression profiles (2–4). The age at diagnosis determining the early vs. late-onset asthma varies from 12 or 18 years (4–6) of age to 40 (7) years of age. However, emerging evidence indicates that environmental factors also play a critical role in the development of asthma. The risk factors of EOA include single nucleotide polymorphisms on chromosome 17q21, atopic status, rhinovirus infection, and exposure to traffic-related air pollution (TRAP) (8–12). TRAP is one of the major environmental impacts of urbanization and previous research has shown the long-term effects on asthma onset in children (5, 13). Among children, proximity to traffic has been linked to an elevated risk of asthma (12, 14–16).

Despite the non-negligible incidence of adult-onset asthma, the causes have not been extensively investigated. Identifying the risk factors of LOA is crucial to understanding the underlying mechanisms and LOA was identified as the most significant independent risk factor for non-remission in patients with asthma (17). It should be noted however that the risk factors for LOA are more complex than those for EOA. Specific characteristics have been mentioned for their different influence on EOA and LOA. A population-based study in Finland showed the incidence of allergic asthma decreases with advancing age and after the age of 40, new asthma cases are almost non-atopic (18). The other study in Finland discovered the influence of the family history of asthma is higher on EOA than LOA (risk ratio (RR): 4.10 vs. 1.44) (9). A European survey discovered patients with chronic rhinosinusitis reported less EOA (RR: 0.45) but more LOA (RR: 3.09) (19). In the cluster study of asthma phenotypes, obese female is one of the major cluster of LOA (20). One large-scale genome-wide association study (GWAS) suggested that non-genetic risk factors play a more important role in LOA than in EOA. This is a clear indication that environmental factors are worthy of further consideration (21). The important role of TRAP in adult-onset asthma has also been emphasized (22). The association between traffic proximity and LOA has been identified by a certain amount of traffic volume and patients in risk (23–25). One study of our adult asthma cohort reported that higher IL-17A expression in the epithelium of patients among those living within 1,000 m from heavy traffic roads (HTRs) than among those living more than 1,000 m from HTRs (26).

In the current study, we were interested in whether the traffic proximity was different in EOA and LOA. We sought to determine whether proximity or density of HTRs is associated with the LOA in the asthma cohort of an urban medical center. We also examined patient characteristics in order to identify factors significantly associated with traffic proximity in LOA.

Our analysis revealed that patients with LOA tended to be older than patients with EOA, to have had asthma for a shorter duration, were less likely to have a family history of asthma, were less exposed to fumes or dust in the home, were less likely to be atopic, were less sensitive to common inhaled allergens, and were more likely to have poor lung function. In urban areas, more patients with LOA had multiple HTRs (≥ 2 HTRs) within a distance of 900 m. Patients living within 900 m from multiple HTRs had a higher total IgE level, a more atopic status, and a higher sensitivity to HDM, compared to patients with ≥ 1 HTRs at a comparable distance. Among patients with LOA, the minimum distance to an HTR was positively associated with the age of onset and BMI and was negatively associated with atopy and mood status. Among patients with LOA, a higher proportion of atopic patients were living within 900 m from HTRs, a higher proportion of patients with anxiety or depression were living within 900 m from HTRs and a trend of a higher proportion of obese patients were living beyond 900 m from HTRs. To the best of our knowledge, this is the first study demonstrating the different traffic density between EOA and LOA and different phenotypes in LOA by geolocation, which showed the novel relation between asthma phenotypes and the urban environment.

Correlation between TRAP and asthma is usually assessed in terms of pollutant concentration, such as nitrogen oxide and particulate matter (PM), or the distance to HTRs. Numerous cohort studies have demonstrated a positive association between exposure to TRAP and the risk of asthma (11, 23, 29, 30); however, studies on the link between proximity to HTRs and the risk of asthma have been inconsistent. A birth cohort study in southwestern British Columbia failed to observe a significant correlation between proximity to highways and major roads (<150 m) and development of childhood asthma (15). A birth cohort study in New York City reported a significant correlation between proximity to heavy traffic (<250 m) and childhood asthma among patients without a history of moving prior to the age of 5 (16). In a cohort study of children with asthma attending elementary school in an urban area of the northeastern United States, the incidence of asthma symptoms was shown to increase inversely with the distance to major roads (12). One cohort study conducted in three cities in Sweden reported a positive association between adult-onset asthma and proximity (<50 m) to major roads (≥8,000 vehicles/day) (23). Two cross-sectional studies examined the links between subgroups of asthma and proximity to major roads. One study on adults in southern Sweden discovered that proximity (>100 m) to a major road (>10 cars/min) was associated with allergic asthma but not with non-allergic asthma (24). One study on adults in Tasmania, Australia, (i.e., an area with low air pollution levels) reported that proximity (<200 m) to a major road was associated with an elevated risk of asthma; however, this was only among carriers of glutathione S-transferase theta-1 (25). The different distances to HTR for asthma risk in the studies above may result from the different traffic volumes, the population in risks, and the local environment of individual cities. In the current study, HTRs were defined by daily mean PCU exceeding 36,329, which was much higher than that of other studies (23, 24). The range of 1,000 m to HTR was selected initially because the relevant reports included one study from our asthma patients (26, 28). By defining HTRs with the top-ten high traffic volume from the official database, we determined that living within 900 m of multiple HTRs was associated with an elevated risk of LOA compared to EOA. The current study provides new evidence of the greater impact of heavy traffic exposure for LOA. In addition to the direct distance to HTRs, exposure to the density of HTRs is also an important determinant for asthma control.

The patient profiles revealed by the subgroup analysis of patients living with multiple HTRs provided one of the possible mechanisms of asthma inception of LOA (Table 3). Patients living within 900 m from multiple HTRs presented higher total IgE levels, were more likely to be atopic and were more likely to present sensitivity to HDM, compared to patients with fewer than two HTRs at a comparable distance. We also found that asthma onset was earlier among patients with LOA living near HTRs. This suggests that an interaction between TRAP and allergic sensitization may be the force driving asthma inception at an early age among LOA living near Figures 3, 4. Similar to our results, a cross-sectional study revealed associations between exposure to high road density and the prevalence of allergic sensitization and small airway function in subjects with a family history of asthma (31). Previous animal studies on TRAP exposure and allergic sensitization support our results. In adult and neonatal mice, co-exposure to diesel exhaust particles (DEPs) and house dust mites was shown to promote the persistence of TH2/TH17 effector/memory cells in the lungs (32). In studies in an adult mouse model, DEP and HDM co-exposure has also been shown to enhance airway hyper-responsiveness and generate a mixed TH2 and TH17 response or the number of type 2 innate lymphoid cells (33, 34). Co-exposure to HDM and benzo(a)pyrene has also been shown to enhance IL-33 and TSLP production in an asthma mouse model (35). Environmental factors (e.g., ambient air polyaromatic hydrocarbons, PM, and DEP) have been linked to epigenetic changes that modify the gene expression of T regulatory cells or innate response, further promoting the Th2 response (36).

Anxiety and depression contribute to asthma symptoms and stressful life events have been shown as the risk factor of onset of asthma (37, 38). Some studies have shown psychiatric stress enhances allergic inflammation (39, 40). We found the risk of anxiety or depression increased in patients living within 900 m to HTRs (Figure 4B). This finding further demonstrated the complex interactions between the psychiatric status, traffic exposure in patients with LOA, and further studies for better analysis are required.

The incidence of obesity in patients with LOA was higher among those living >900 m from HTRs than among those living <900 m from HTRs (Figure 4C). This conflicts with a number of previous studies suggesting that air pollution plays a role in the incidence of asthma among the obese (41). By contrast, the incidence of atopic asthma in patients with LOA was lower among those living far from HTRs (Figures 3, 4). We surmise that exposure to TRAP may play a more important role in the pathogenesis of atopic asthmatics than it does in obese asthmatics. Future studies on obesity-related LOA and TRAP in urban areas are required.

The study has several limitations. First, the cohort had a smaller patient number of EOA compared to LOA. This is because we recruited adult patients consecutively without selection in the pay-for-performance program to prevent selection bias. The higher ratio of LOA to EOA in our cohort is possible due to the remission rate of EOA being much higher than LOA and LOA is suggested to be more severe than EOA (2, 42). Therefore, similar to other cohorts (43), patients with LOA were referred from local clinics to our center more frequently than patients with EOA. The present results from a single center will have to be confirmed further in subsequent longitudinal studies in a larger population. Second, self-reports pertaining to the age of the patient at the time of diagnosis by a physician were subject to recall bias. Note that we were unable to obtain documented medical records related to asthma diagnosis; however, the patient characteristics in the current study (e.g., family history of lung function) were comparable with those obtained in large-cohort studies and major review articles related to LOA (2, 3, 6, 9, 18, 44). Third, covariates of exposure to fume or dust or comorbidities were defined by questionnaires or medical records, not by real inspection of patients' environments or strict medical diagnostic criteria. The results of a negative association between exposure and comorbidities and HTR proximity in the current study are required further studies to confirm. Fourth, we were unable to obtain information related to TRAP concentrations; therefore, our findings are not necessarily generalizable to all HTRs. Our findings could have been affected by local climatic conditions, occupational exposure, and indoor air pollution in the individual urban environment.

In conclusion, the characteristics of patients with LOA were distinct from those of patients with EOA. By geolocation in urban centers, we discovered the correlation between asthma phenotypes and the urban environment in LOA. patients with LOA tended to live in areas of higher HTR density, which was associated with an elevated incidence of atopic symptoms, sensitivity to HDM, and mood disorder. Proximity to HTRs and obesity may be factors contributing to uncontrolled asthma in cases of LOA. The novel evidence of patient-environment interaction provides further explanation for asthma persistence in the modern world.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by Chang Gung Medical Foundation Institutional Review Board (No. 201900211B0). The patients/participants provided their written informed consent to participate in this study.

T-YL: conceptualization, data curation, investigation, formal analysis, and writing of the original draft. Y-SL: methodology, software, formal analysis, and visualization. H-CL: data curation, resources, and formal analysis. S-ML: data curation, resources supervision, formal analysis, and writing with review and editing. Y-LL, C-HW, P-JC, and C-YL: data curation and resources. All authors contributed to the article and approved the submitted version.

Y-SL is employed by BalDr Strategic Consulting (Hong Kong) Ltd. The remaining authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.