Air pollution exposure is associated with rhinitis in older US adults via specific immune mechanisms

Abstract

Background:

Pathophysiology of rhinitis in older adults is largely unknown. We tested whether air pollution is associated with this condition and how immune mechanisms may play a role in this relationship.

Methods:

We analyzed cross-sectional data from the National Social Life, Health, and Aging Project, a nationally representative study of older adults born 1920–47. PM_2.5 air pollution exposure estimates were generated using validated spatiotemporal models. Presence of rhinitis was defined based on medication use (≥1: intranasal steroids, antihistamines, lubricants, and/or decongestants). K-means cluster analysis (Jaccard method) was used to group 13 assayed peripheral blood cytokines into 3 clusters to facilitate functional determination. We fitted multivariate logistic regressions to correlate PM_2.5 exposure with presence of rhinitis, controlling for confounders, and then determined the role of cytokines in this relationship.

Results:

Long (but not short) term exposure to PM_2.5 was associated with presence of rhinitis: 3-year exposure window, OR=1.80, 95% CI 0.99, 3.42, per 5μg/m³ PM_2.5 increase. Inclusion of cytokine cluster in the model led to a modestly stronger effect of PM_2.5 exposure on rhinitis (OR=1.91, 95% CI 1.00, 3.69, 3-year exposure window). The particular immune profile responsible for this result was composed of elevated IL-3, IL-12, and IFN-γ (OR=6.20, 95% CI 1.01, 40.89, immune profile-PM_2.5 exposure interaction term).

Conclusion:

We show for the first time that IL-3, IL-12, and IFN-γ explain in part the relationship between PM_2.5 exposure on rhinitis in older US adults. If this mechanism is confirmed, these immune pathways may offer promise as therapeutic targets for this disorder.

1. Introduction

Air pollution is associated with increased prevalence of acute rhinitis in urban areas. For example, both short and long-term exposure to particulate matter ≤10μm (PM₁₀), particulate matter ≤2.5μm (PM_2.5), ozone (O₃), nitrogen dioxide (NO₂) and sulfur dioxide (SO₂) have linked with increased prevalence of allergic rhinitis^1,2. Interestingly, children and adolescents are susceptible to these hazardous environmental exposures². In general, these associations tend to be stronger in developing countries compared to developed ones¹ perhaps because extreme levels of ambient air pollution^3,4. Little is known about how pollution exposure may affect rhinitis in older adults, where non-allergic forms predominate.

One potential mechanism that may explain this relationship is immune responses generated by the deposition of microparticles on the airway epithelial surface^5,6. For example. oxidative stress modulates immune response to T-helper 2 (Th2) and T-helper 17 (Th17) pathways^7,8. Indeed, diesel exhaust particles are recognized by airway epithelial cells, detoxified by cytochrome P450 family 1-A1, and induce Nrf2 translocation to the nucleus, which increases antioxidant transcription. Failure in this detoxification process leads to production of pro-inflammatory cytokines (e.g., interleukins [IL] 8, IL-1, IL-6, C-C Motif Chemokine Ligand 20, tumor necrosis factor [TNF] α)^9,10. This results in chronic inflammation and subsequent airway remodeling in asthmatic individuals¹⁰. Other mechanisms are possible and those in the upper airway are poorly described, especially in the setting of aging.

How air pollution affects the upper airway of older adults is largely unknown. The immune pathophysiology by which PM_2.5 causes rhinitis in older adults is largely underexplored. Most of the pathophysiology studies focus on allergic rhinitis in young adults and children and these mechanisms may differ from the ones responsible for rhinitis (either non allergic or allergic) in older adults^10–12. Additionally, large scale populational studies on the mechanisms by which air pollution results in rhinitis are sparce.

To our knowledge, there are no studies demonstrating which specific cytokines are involved in the generation of rhinitis by outdoor pollution in a large-scale population of older adults. To address this gap, we studied how exposure to air pollution is associated with the use of rhinitis medications and whether systemic immune responses could be detected in nationally representative cohort of older adults.

2. Methods

2.1. Study population

In 2010–11 and 2015–16, professional interviewers from NORC at the University of Chicago conducted in-home interviews with 3377 older adults born 1920–47 and 4777 respondents born 1920–65, respectively, a nationally representative sample of the U.S. population of community-dwelling older adults. Measures collected included demographic, socioeconomic and comorbidity data, along with biological measures. Further details on data collection and analyses are provided elsewhere^13,14. For the present study, we used data obtained in 2010–11 (NHSAP Round 2), where cytokine data was collected using peripheral blood samples. Data on use of rhinitis medication was obtained in 2015–16 (5 year follow-up)¹⁵ (Figure 1).

Figure 1.

Study design.

Demographic data included age (in years), sex, race (standard NIH categories) and education (less than high school, high school or equivalent, some college, and bachelor’s degree or equivalent). Smoking cigarette status was analyzed as dichotomous variable (yes or no). Data on rhinitis related medication use were collected using a modified version of the Multum Lexicon Plus drug hierarchy; this included nasal steroids, nasal antihistamines, nasal lubricants, and nasal preparations or decongestants. A list of all medications included in these categories is described in Supplementary Table 1. Additional covariates included geographic region (West, Midwest, South, or Northeast; states included in each region are listed in Supplementary Table 2) and rural-urban commuting area (RUCA) codes, as defined by the U.S. Department of Agriculture (a scale from 1 to 10, where 1 corresponds to metropolitan areas` cores and 10 to rural areas).

The detailed process of collection, transportation and testing for peripheral blood samples is described elsewhere¹³. Briefly, samples were obtained using K2EDTA microtainers in the home by interviewers, stored at 3±7°C, transported to an interviewer`s field base and shipped cold overnight to the University of Chicago Flow Cytometry Facility. Cytokine levels were measured using Luminex bead array technology (Luminex 100 device; BioRad, München, Germany) at the University of Chicago Flow Cytometry Facility using standard protocols and BioPlex Manager Software (Version 5, BioRad). The measured cytokines included: interleukin (IL) 1rα, 1β, 2, 3, 4, 5, 6, 10, 12 and 13, vascular endothelial growth factor (VEGF), tumor necrosis factor (TNF) α, interferon (IFN) γ¹³.

2.2. Cluster analyses for cytokine levels

The 12 cytokines were categorized into empiric functional immune profiles. Each cytokine was divided into quintiles based on the actual measured concentration. Values below the detection limit were categorized as the lowest quintile and measurements above the detection limit were included in the highest quintile. We then dichotomized the categorized cytokines in highest quintile (≥80th percentile) vs. the remainder of the quintiles. This allowed us to differentiate respondents with elevated levels of cytokines (≥80th percentile) from the ones without elevated cytokines. Lastly, we used the K-means cluster analysis (Jaccard method) to divide the respondents into 3 functional immune groups^16,17. The number of clusters was determined by the Calinski-Harabasz index, a previously validated internal cluster validity method^18,19.

2.3. Exposure to air pollution

Ambient exposure to fine particles (PM_2.5) using the concentration outside each participant`s home address averaged for 30-, 60-, and 90-days and 1, 2 and 3 years prior to each participant’s date of NSHAP Round 2 data collection (2010–11). Previously well-validated GIS-based spatio-temporal models were used for this purpose and are described elsewhere²⁰. Briefly, PM_2.5 daily concentration were estimated for a 6 km grid for the conterminous U.S. Input model variables included PM_2.5 data from EPA, meteorological and geospatial data, and traffic-related PM point emission (estimated using a Gaussian dispersion model). The model predicted ambient PM_2.5 concentrations well, with a cross validation of R²=0.76, low bias and high precision. The mean distance from a model grid point to a participant’s residence was 2.23 km.

2.4. Rhinitis medication variable

Rhinitis medication use was determined from data collected in 2015–16 and included nasal steroids, nasal antihistamines, nasal lubricants, and nasal preparations or decongestants. The mediations were classified according to a modified version of the Multum Lexicon Plus drug hierarchy. A dichotomized proxy variable for rhinitis was made by older adults that used or did not use these medications. We note that as prevalence of allergic rhinitis decreases with aging, non-allergic rhinitis is much more common in older adults^21,22 although we lack confirmatory testing (negative allergy testing) in this analyses. Thus, we focus here on rhinitis writ large (both allergic and non-allergic forms).

2.5. Statistical analyses

We restricted our analyses to the older adults in NSHAP with available measured cytokine levels and had rhinitis medication data available in 2015–16 (n=1839). We accounted for differential probabilities of non-response and selection due to over-sampling of African Americans, Latinos, men, and the oldest old using sampling weights and variance estimation by the linearization method. Statistical significance was set at p<0.05 and Stata version 17.0 (StataCorp LLC, College Station, TX) was used to perform statistical analyses.

To explore the relationship of rhinitis and PM_2.5 exposure, we fitted multivariate logistic regressions, adjusting for potential confounders. We then performed similar regressions with the inclusion of the cytokine cluster group to verify if and how the rhinitis-air pollution relationship changed. We also included an interaction term between PM_2.5 and cytokine cluster to test whether this immune function variable modified the PM_2.5-rhinitis relationship. Lastly, the results were reported as odds ratios (OR) and 95% confidence intervals (CI). The OR for PM_2.5 is in terms of 5 μg/m³ PM_2.5 increase.

3. Results

Of the 3377 and 4777 NSHAP respondents interviewed in 2010–11 and 2015–16, 1839 older adults had rhinitis medication data and had their cytokine levels measured, and therefore were included in our primary analysis according to the modular study design. The demographics of the analytic cohort can be found in Table 1. Of note, 55 respondents (3.0%) reported use of at least 1 rhinitis medication, with nasal preparations or decongestants being the most common (n=55), followed by nasal steroids (n=42). Short-term PM_2.5 exposures had higher variability (Standard Deviation [SD] 2.65–2.75, 30 to 90 days intervals) and lower means (8.21 to 8.78μg/m³) in contrast to long-term exposures which had higher means (8.69 to 9.17μg/m³, 1 to 3 years intervals) and lower variability (SD 2.29 to 2.40) (Table 2).

Table 1.

Demographic characteristics of the study sample.

	Frequency (%) or Mean ± Standard Deviation (SD)*
Rhinitis medication use (yes)	55 (3.0)
Nasal steroids	42 (2.3)
Nasal antihistamines	11 (0.6)
Nasal lubricants	2 (0.1)
Nasal preparations or decongestants	55 (3.0)
Age (years)	70.6 ± 7.5
Gender
Males	770 (41.9)
Females	1069 (58.1)
Race/ethnicity
White	1342 (73.2)
Black	248 (13.5)
Hispanic	204 (11.1)
Other	40 (2.2)
Schooling
<High School	314 (17.1)
High School	448 (24.4)
Some College	595 (32.3)
Bachelors or more	482 (26.2)
Cigarette Smoking (yes)	233 (12.7)

^*These estimates are unweighted.

Table 2.

Short- (30, 60 and 90 days) and long-term (1, 2 and 3 years) PM_2.5 exposures.

Exposure interval	Mean (μg/m3)	Standard Deviation (SD)	Median (μg/m3)	Interquartile range (IQR)
30 days	8.21	2.75	8.05	6.37–9.80
60 days	8.51	2.67	8.41	6.69–10.22
90 days	8.78	2.65	8.82	6.93–10.47
1 year	8.69	2.36	8.92	7.10–10.37
2 years	8.74	2.29	8.77	7.32–10.37
3 years	9.17	2.40	9.20	7.74–10.91

3.1. Cytokine levels and clusters

Cluster one contained the largest number of individuals and were those mostly without elevated cytokine levels (less than 15% of respondents had elevated TNF-α, VEGF, and IL-1β). In contrast, cluster 2 was primarily composed of older adults with high levels IL-1rα, IL-2, IL-4, IL-5, IL-6, IL-10, and IL-13; these are mostly T helper (T helper) 2 cytokines. Cluster 3 contained respondents with elevated levels of IL-3 (61.5%), IL-12 (48.1%), and IFN-γ (36%); interestingly, these are mostly Th1 responses (Table 3).

Table 3.

Characteristics of the cytokine cluster groups^*.

Cytokine	Cluster 1 (n=677**)	Cluster 2(n=723)	Cluster 3(n=439)
IFN-γ	0	199 (27.5)	158 (36.0)
TNF-α	79 (11.7***)	178 (24.6)	55 (12.5)
VEGF	101 (14.9)	170 (23.5)	87 (19.8)
IL-1β	88 (13.0)	194 (26.8)	78 (17.8)
IL-1rα	0	289 (40.0) ****	69 (15.7)
IL-2	0	360 (49.8)	15 (3.4)
IL-3	0	93 (12.9)	270 (61.5)
IL-4	0	356 (49.2)	16 (3.6)
IL-5	0	292 (40.4)	76 (17.3)
IL-6	0	349 (48.3)	15 (3.4)
IL-10	0	361 (49.9)	5 (1.1)
IL-12	0	159 (22.0)	211 (48.1)
IL-13	0	374 (51.7)	7 (1.6)

^*Count of older adults with elevated levels of each cytokine (80th percentile or more) are presented for each cluster. Respondents can have more than one cytokine increased and thus may be counted more than once or have no elevated cytokines and not be counted at all.

^**Total number of respondents within the cluster;

^***Percentage respondents with elevated cytokine within the cluster;

^****The key distinguishing cytokines for each cluster are shaded (30% or more of older adults within the cluster have elevated levels [≥80^th percentile] of the specific cytokine).

3.2. Multivariate logistic regressions

We found suggestive evidence that older US adults exposed to higher long-term levels of PM_2.5 had increased odds of rhinitis (OR=1.74 per 5μg/m³ increase in PM_2.5, 95% CI 0.90, 3.45, p=0.09, 2 years window; OR=1.80, 95% CI 0.99, 3.42, p=0.06, 3 years), accounting for age, sex, race/ethnicity, education, cigarette smoking, BMI, country region and urbanicity. When the cytokine cluster variable was included in the regression, the strength of this relationship was increased (OR=1.84, 95% CI 0.95, 3.73, p=0.07, 2 years; OR=1.91, 95% CI 1.00, 3.69, p=0.05, 3 years) (Table 4).

Table 4.

Logistic regressions for 3 years PM_2.5 exposure-rhinitis relationship, adjusting for potential confounders. Model 1 without cytokine cluster group, Model 2 with cytokine cluster group.

	Model 1			Model 2
Variable	OR	p-value	95% CI	OR	p-value	95% CI
3 years PM2.5 (per 5μg/m3 increase)	1.80	0.06	0.99–3.42	1.91	0.05	1.00–3.69
Cluster of cytokines (vs cluster 1)
2				2.16	0.11	0.83–5.61
3				2.03	0.12	0.83–4.96
Age (per year)	0.99	0.87	0.95–1.04	0.99	0.81	0.95–1.04
Sex (vs. male)	0.92	0.79	0.52–1.64	0.95	0.86	0.54–1.69
Race/ethnicity (vs. Caucasian)
African American	0.72	0.49	0.27–1.88	0.74	0.52	0.29–1.87
Hispanic	0.96	0.95	0.27–3.44	0.97	0.96	0.28–3.32
Education (vs. <High School)
High School	0.61	0.46	0.16–2.28	0.61	0.43	0.17–2.15
Some College	1.50	0.49	0.46–4.86	1.49	0.48	0.48–4.60
Bachelors or more	1.38	0.58	0.44–4.31	1.36	0.58	0.45–4.15
Smoke cigarettes (vs no)	0.85	0.85	0.15–4.89	0.88	0.88	0.15–5.04
BMI	1.05	0.03	1.00–1.09	1.05	0.04	1.00–1.09
Urbanicity (RUCA codes)*	1.13	0.08	0.98–1.29	1.14	0.06	0.99–1.31
Country region (vs West)
Midwest	1.60	0.52	0.37–7.02	1.50	0.58	0.34–6.61
South	2.70	0.12	0.76–9.61	2.57	0.15	0.70–9.38
Northeast	3.02	0.06	0.96–9.43	3.15	0.05	1.02–9.69

^*RUCA codes were treated as a continuous variable in a 1 to 10 scale, in which 1 was the most urbanized areas (metropolitan cores) and 10 was the least ones (rural areas).

Our analyses suggested that cytokine cluster modifies the PM_2.5-rhinitis relationship (Wald test, interaction p=0.02, 2 years; interaction p=0.02, 3 years). In postestimation analyses, respondents in cluster 3 exposed to higher PM_2.5 had the strongest association with rhinitis (OR=13.24, 95% CI 2.99, 58.66, 2 years window; OR=10.94, 95% CI 2.82, 42.37, 3 years) when compared to other clusters (Table 5 and Figure 2). Full details about the regressions for 1- and 2-year windows of PM_2.5 exposure can be found in Supplementary Tables 3 through 6.

Table 5.

Logistic regressions for 3 years PM_2.5 exposure-rhinitis relationship, adjusting for potential confounders. Model 3 with cytokine cluster and inclusion of an interaction term between cytokine cluster and PM_2.5.

	Model 3
Variable	OR	p-value	95% CI
3 years PM2.5 (per 5μg/m3 increase)	1.76	0.38	0.49–6.32
Cluster of cytokines (vs cluster 1)
2	2.04	0.13	0.80–5.21
3	1.63	0.30	0.63–4.18
Cluster of cytokines and 3 years PM2.5 interaction term
2	0.55	0.55	0.07–4.04
3	6.20	0.05	1.04–40.89
Age (per year)	0.99	0.90	0.96–1.04
Sex (vs. male)	0.99	0.99	0.54–1.82
Race/ethnicity (vs. Caucasian)
African American	0.66	0.35	0.27–1.61
Hispanic	0.93	0.91	0.27–3.25
Education (vs. <High School)
High School	0.60	0.41	0.17–2.07
Some College	1.53	0.44	0.51–4.61
Bachelors or more	1.37	0.54	0.50–3.77
Smoke cigarettes (vs no)	0.86	0.86	0.15–4.81
BMI	1.05	0.03	1.01–1.10
Urbanicity (RUCA codes)*	1.13	0.06	0.99–1.29
Country region (vs West)
Midwest	1.53	0.57	0.34–6.82
South	3.03	0.11	0.78–11.82
Northeast	3.27	0.04	1.05–10.16

^*RUCA codes were treated as a continuous variable in a 1 to 10 scale, in which 1 was the most urbanized areas (metropolitan cores) and 10 was the least ones (rural areas).

Figure 2.

Adjusted predictions of rhinitis medication use among the three cytokine clusters with 95% confidence interval (CI) when exposed to increasing particulate matter ≤2.5 μm (PM2.5) (3 years average minus median).

Interestingly, older adults exposed to short-term PM_2.5 were not more likely to have rhinitis: 30 days (OR=1.31, 95% CI 0.92, 1.87, per 5μg/m³ increase) to 90 days (OR=1.49, 95% CI 0.87, 2.53) (Table 6). Full regression for 30-days PM_2.5 exposure can be found in Table 6 and for 60- and 90-days exposures in Supplementary Tables 7 and 8.

Table 6.

Logistic regressions for 30 days PM_2.5 exposure-rhinitis relationship, adjusting for potential confounders.

Variable	OR	p-value	95% CI
30-days PM2.5 (per 5μg/m3 increase)	1.31	0.14	0.92–1.87
Age (per year)	0.99	0.93	0.95–1.04
Sex (vs. male)	0.93	0.79	0.53–1.64
Race/ethnicity (vs. Caucasian)
African American	0.76	0.56	0.29–1.99
Hispanic	1.05	0.94	0.30–3.60
Education (vs. <High School)
High School	0.64	0.48	0.18–2.28
Some College	1.53	0.46	0.48–4.86
Bachelors or more	1.41	0.54	0.46–4.35
Smoke cigarettes (vs no)	0.84	0.85	0.14–4.93
BMI	1.05	0.03	1.00–1.09
Urbanicity (RUCA codes)*	1.09	0.15	0.97–1.23
Country region (vs West)
Midwest	1.79	0.43	0.41–7.87
South	2.68	0.13	0.75–9.62
Northeast	2.80	0.07	0.91–8.65

^*RUCA codes were treated as a continuous variable in a 1 to 10 scale, in which 1 was the most urbanized areas (metropolitan cores) and 10 was the least ones (rural areas).

4. Discussion

We demonstrate here that long-term PM_2.5 exposure appears to be associated with rhinitis as defined by medication use in a nationally representative cohort of older US adults. Additionally, we showed this relationship is strongest among those with elevated IFN- γ, IL-3 and IL-12 levels. If confirmed, this could represent a potential immune mechanism by which PM_2.5 induces rhinitis, providing a clue to the complex causal pathway between air pollution exposure and upper airway inflammatory disease.

Analyzing the patterns of cytokines that were elevated in respondents with rhinitis could help to understand how these substances may lead to the development of an inflammatory process in the nose. For instance, INF-γ is released by T lymphocytes and NK cells and induces MHC expression on phagocytes, promotes cell growth and differentiation, and specifically activates macrophages²³. More specifically, IFN-γ induces the expression of interferon-stimulated genes that act through Janus kinase (JAK)-signal transducer and activator of transcription (STAT) signaling pathway to promote defense against microbial infections. When dysregulated, this mechanism is involved in autoimmune diseases and chronic inflammation²⁴. Regarding respiratory diseases, CD4⁺ and CD8⁺ T helper cells exposed to increased PM_2.5 express higher quantities of IFN-γ mRNA and protein, along with IL-10, IL-17 and IL21. This induces a macrophage dependent Th1 response, which ultimately leads changes in function of respiratory epithelial cells²⁵. A similar mechanism involving IFN-γ is also involved in the pathogenesis of a type of chronic rhinosinusitis, which in some forms involves dysregulation of Th1 responses²⁶. Thus, our data are broadly consistent with the concept that airborne particulate matter induces macrophage-dependent cytotoxic T cell response.

IL-3 also exerts a pivotal function in chronic inflammation. This cytokine is part of the β common chain (βc) cytokine family, which also include granulocyte-macrophage colony-stimulating factor (GM-CSF) and IL-5²⁷. IL-3 binds to IL-3 receptor α (IL-3Rα/CD123) and associates with βc subunit, which signals through JAK2-STAT5A/B. Though this signaling pathway, IL-3 promotes basophil growth and differentiation, which are involved in allergic rhinitis^27,28. Additionally, IL-3 has strong hematopoietic effects, promoting the production of granulocytes, monocytes, mast cells, basophils, and macrophages, which could also promote airway inflammation²⁹ in response to air pollution.

Finally, IL-12, is part of a family of cytokines (IL-23, IL-27, and IL-35) which mediate many complex immune functions. IL-12 is produced by dendritic cells, macrophages and B cells and drives a positive feedback loop with IFN-γ (produced by T cells and also elevated in our analyses) to promote inflammation³⁰. Interestingly, IL-12 is produced by dendritic cells localized close to mast cells at the mucosal interface, showing a potential role of IL-12 in atopic airway processes. To this end, increased levels of IL-12 have been already associated with allergic rhinitis in children³¹. Segboear et all have also pointed that IL-12 could also be associated with non-allergic rhinitis to a lesser degree³².

Our results are consistent across long-term windows of PM_2.5 exposure. However, short-term exposures were not significantly associated with rhinitis. We interpret this finding as pollution exposure may have a cumulative effect on rhinitis development in older adults (as occurs for many other diseases such as Chronic Obstructive Pulmonary Disease, asthma, and lung cancer)^33–36. Reducing exposure to air pollution, therefore, may decrease other rhinitis-related burdens faced by older adults, which includes decreased productivity, diminished good quality sleep, increased rates of depression, and decreased quality of life^37–39. Additionally, decreasing the rates of rhinitis in the geriatric population could potentially mitigate polypharmacy, which is independently associated with adverse events, morbidity, and even mortality in the elderly^40,41.

Our outcomes were robust to the addition of age, sex, race/ethnicity, education, cigarette smoking, and BMI. We did not find any association of rhinitis with any of these variables, except for BMI, which was associated with rhinitis (older adults with increased BMI were more likely to have rhinitis). Obesity may produce a proinflammatory state and has been associated with asthma and atopy in children, although this is a topic of debate in the literature^42–45. Perhaps by a similar mechanism, obese adults are more likely to have an increased prevalence of non-allergic rhinitis⁴⁶. Nevertheless, our findings of increased BMI and rhinitis in older adults add to the complex body of evidence on this topic.

There were several limitations of our study. Our design is cross-sectional, so we cannot identify causal relationships. Longitudinal studies to define causal pathways are needed. Secondly, we lack the ability to determine actual rhinitis symptoms and atopy status as these data were not included in NSHAP. Direct information on symptoms and allergy testing to distinguish allergic and non-allergic rhinitis are needed to determine specific relationships with air pollution by these forms of the disease. Our study involved in-home interviews of older adults, but we had no access to electronic medical records for clinical information nor to diagnostic testing (e.g., allergy testing, rhinoscopy, nasal endoscopy). Thus, we cannot distinguish other indications for these medications (chronic rhinosinusitis, nasal polyposis, etc.). However, we note that rhinitis diagnosis is made mainly by clinical information on symptoms (nasal congestion, sneezing, nasal pruritus, excessive mucus production) and the medications commonly used for those symptoms were present in our analyses^47,48. Moreover, allergic rhinitis prevalence decreases with age, which makes the diagnosis of non-allergic rhinitis in older adults more likely^21,22. The k-means clustering method was used to describe the immune function by identifying cytokine patterns by standard methods. Such immune profiles are crude, but more complex immunophenotyping (B and T cell function, etc.) was not possible in the omnibus study performed in homes across the country. More sophisticated methods to measure immunologic metrics are possible in smaller scale clinical studies to followup these findings. Finally, we lack information on nasal physiology; nasal sampling would be useful in subsequent studies to examine the differences between local effects in the airway vs. systemic effects in the peripheral blood.

5. Conclusion

To our knowledge, we are one of the first studies to propose that specific immune functions (through IL-3, IL-12, and IFN-γ) connect ambient PM_2.5 exposure to rhinitis in a nationally representative sample of older US adults. Further work is necessary to precisely characterize the immunologic processes that connect air pollution to rhinitis and delineate causal mechanisms. Identifying the underlying molecular processes that are involved will allow to develop new precision therapies to reduce the burden of environmentally mediated rhinitis in older adults.