Impact of clinical factors and season on inflammatory cytokines in biologic-treated and untreated asthma

Abstract

Background

See More

Clinical features influence cytokine profiles and can inform biomarker studies.

Objectives

We assessed the impact of 13 preselected patient characteristics on the circulating levels of 15 Th-1/2/17 cytokines in moderate-to-severe asthma patients on omalizumab, anti-IL-5 (mepolizumab, benralizumab), or dupilumab (n = 76) versus controls (n = 162) not yet on biologics but meeting eligibility criteria for a T2-biologic.

Methods

Plasma cytokines (Olink) were analyzed for associations with these clinical/lifestyle factors using LASSO regression and observed variance explained estimated using generalized linear models. Differential expression analysis was conducted using limma.

Results

In controls, IL-6 had the highest variance explained by clinical/lifestyle factors (50% in non-allergic rhinitis patients, 22% in allergic rhinitis), with BMI and exacerbations contributing most to this. In T2-biologics users, eotaxin-1 had the highest explained variance (26.0%) and smoking was the most linked to Th1/17 cytokines. In omalizumab users: IFN-γ (51%) was most explained (exacerbations, smoking, age). In anti-IL-5 users, eotaxin-1 (58%; BMI, sex) and in dupilumab users, IL-4 (83%) was most explained (exacerbations, sex, BMI). The association between patient characteristics and cytokine levels differed by the season of sample collection. In non-biologic users, IL-6 was the cytokine with the most explained variance in the Winter (asthma admissions accounted for most of this variance) and IL-18 in the Spring/Summer/Fall. In T2-biologic users, TNF-α was the top cytokine in the Winter (smoking accounted for most of this variance); IL-4 (allergic rhinitis), IL-33 (IgE and eosinophil), and CXCL10 (allergic rhinitis and IgE) were the top cytokines in the Spring/Summer/Fall. In differential expression analyses, IL-1β was lower in biologics users than non-biologics users.

Conclusions

In moderate-to-severe asthma, multiple clinical features and season are associated with cytokine levels and might impact inference from proteomics studies. Smoking and BMI are the key proinflammatory factors in biologics-treated and untreated patients.

Introduction

Severe and/or uncontrolled asthma causes a major strain on healthcare systems in the United States and beyond [1]. While the recent approval of targeted therapies for specific subtypes of moderate-to-severe asthma is a breakthrough for those who have not responded to traditional options [2], not all patients benefit from these monoclonal antibody treatments [3]. This variability in response highlights the urgent need for biomarkers that can accurately predict who will or will not benefit from biologic therapies [4]. While previous untargeted and targeted omics approaches with small cohorts have yielded moderate success [5,6,7,8], there is a gap in our understanding of how potential biomarkers are associated with clinical and environmental factors co-existent with asthma. Understanding how clinical and lifestyle factors affect circulating cytokine levels, including in patients on monoclonal antibody therapy, may help guide future work toward identifying biomarkers that support more personalized asthma management.

Currently, there are only a handful of biomarkers that have been validated as predictive of asthma exacerbations and response to monoclonal antibody therapy. The proteome and circulating cytokine levels may be influenced by genes, clinical characteristics such as the body mass index (BMI), oral corticosteroid use, and asthma severity, and the environment, including pollution. Studies from the Severe Asthma Research Program (SARP) have demonstrated the strong association between the cytokine, IL-6, and asthma exacerbations as well as BMI [9,10,11]. Multiple cytokines and chemokines are involved with asthma pathogenesis, progression, and/or severity [12, 13]. Monoclonal antibodies targeting some of these cytokines including the Th2 cytokines, interleukin (IL)−4, IL-5, and IL-13 are now approved for the treatment of moderate-to-severe asthma. How these therapies might impact the relationships between cytokines and clinical features is also unclear.

In this study, we investigated the relationship between Th1/2/17 cytokine levels and clinical and lifestyle factors in biologics-naïve and biologics-treated patients with moderate-to-severe asthma. We replicated previous associations between IL-6 and clinical features and explored how these relationships may differ in patients receiving a respiratory biologic compared to those not on biologic therapy.

Methods

Study population

We used plasma samples from 238 moderate-to-severe asthma patients enrolled in the Biobank of Mass General Brigham (MGB), a large US health system, and their blood discards program (Crimson Core) collected from May 2011 to June 2024. The Biobank and Crimson Core have been previously described [14, 15]. In brief, patients are prospectively consented for voluntary participation in the Biobank and samples are collected as per standard protocol from these patients [14]. We identified excess and discarded blood samples from our previously identified severe asthma cohort using the Crimson prospective collection system in the health system’s Clinical Laboratories [15]. After collection, tubes were anonymized and replaced with a unique subject ID replacing the patient MRN and a unique sample ID replacing the clinical lab accession number. We defined patients with recent use of omalizumab, mepolizumab, benralizumab, and dupilumab as T2 biologics users and patients without monoclonal antibody use within six months of sample collection as non-biologics users. We considered six months to be a sufficient washout period, which is about five to ten times the elimination half-life of these therapies [16]. Given that patients eligible for T2 biologic therapy could differ from patients not eligible, we limited the non-biologics users to those who met criteria for omalizumab, mepolizumab, benralizumab, and dupilumab at the time of sample collection. Furthermore, we subsequently matched the biologics-treated patients to non-biologics users. The matching algorithm included age, sex, BMI, smoking status, insurance status, baseline exacerbations, maximum eosinophil count, metabolic syndrome, allergic comorbidities, and LAMA use. Only one patient had used reslizumab and was excluded from the analysis. Patients with recent use of tezepelumab were also excluded from the analysis given the small sample size and the impact of TSLP on both Th1 and Th2 pathways. This study was a retrospective non-experimental study approved by the MGB Institutional Review Board (IRB; #2021P003536) and was performed in accordance with the Declaration of Helsinki [17]. Samples are from patients who had been preconsented for participation in the MGB Biobank and Crimson Core [14, 15].

Clinical and lifestyle factors

We evaluated the impact of thirteen clinical and lifestyle factors selected a priori on cytokine levels, including age, sex (male versus female), body mass index (BMI), smoking history (current or former versus never), baseline exacerbation counts (defined as asthma exacerbations treated with oral corticosteroids (OCS) but which did not require a hospitalization in the year prior to sample collection), baseline asthma admission counts in the year prior, concomitant comorbidities including allergic rhinitis, atopic dermatitis, and chronic rhinosinusitis with nasal polyps (CRSwNP), metabolic syndrome (defined as having diabetes mellitus, hypertension, or dyslipidemia; excluding obesity to avoid duplicating BMI’s effect), maximum blood eosinophil level within one year prior to sample collection, which we have found in prior studies to be more predictive and reflective of patient’s true asthma phenotype than the most recent eosinophil count [8], most recent serum immunoglobulin E (IgE) level, and insurance status (private versus public) as a proxy for unmeasured socioeconomic variables.

Cytokine profiling and pre-analytic processing

We conducted targeted profiling of 15 Th1, Th2, Th17-related cytokines using the Olink platform with a custom-designed Flex panel (Table S1). Details of the Olink platform and assay protocol have been previously reported [18, 19]. The proximity extension assay technology allows quantification of proteins with high sensitivity and specificity. Given that differences in the cytokine distributions between patients on biologics and patients not on biologics might be biologically meaningful [20, 21], we imputed cytokine levels which were below the lower limit of quantification (LOQ) using the half-minimum approach, rather than excluding cytokines with a high number of values below the LOQ, and all values were quantile normalized to reduce skewness and stabilize variance [22,23,24]. However, in sensitivity analyses, we excluded cytokines that had ≥ 80% missingness in both the biologics users and non-biologics users [25]. We used the ComBAT algorithm from the sva package in R to detect batch differences and removed batch effect before conducting statistical analysis [26].

Statistical analysis

Descriptive statistics for the characteristics of the study population were presented using mean with standard deviation or median with interquartile ranges for continuous factors and counts with proportion for categorical factors. For non-biologics and T2 biologics (anti-IgE, anti-IL5, and anti-IL4/13) groups, we sought to identify clinical and environmental factors associated with each cytokine. For the biologics’ users, we combined them in analyses due to sample size reasons. However, given their different mechanisms of action, we separated by each biologic in subgroup analyses. First, we conducted the least absolute shrinkage and selection operator (LASSO) regression to identify the set of factors that contributed meaningfully to each cytokine level. Then, we used generalized linear models with the cytokine level as the outcome and the factors selected from the LASSO model as predictors to get the contribution of these selected factors to the observed variance for each cytokine. Cytokine levels were compared between patients with and without the relevant clinical and lifestyle factors using either independent t-test or Wilcoxon rank-sum test as appropriate. Lastly, we conducted differential expression analyses using limma to compare patients on biologics to matched untreated controls to evaluate cytokine levels that were significantly different between both groups [27]. The matching algorithm included age, sex, BMI, smoking status, insurance status, baseline exacerbations, maximum eosinophil count, metabolic syndrome, allergic comorbidities, and LAMA use.

Subgroup analyses

To evaluate how the characteristic-to-cytokine relationship might differ between subgroups, in non-biologics users, we conducted subgroup analyses by the presence of allergic rhinitis or not, obesity or not, and by baseline blood eosinophil counts (≥ 300, 150–299, < 150 cells/µL). For T2 biologic users, we conducted subgroup analysis by each monoclonal antibody. Given the smaller sample size when divided by the biologic received, we used a parsimonious model focused on age, sex, BMI, smoking history, baseline OCS counts, baseline asthma admission counts, and maximum blood eosinophil levels. Finally, since there could be differences in cytokine levels by season of blood collection, given seasonal variations in viral infections and in seasonal allergic rhinitis symptoms, we conducted subgroup analyses by season (Winter vs. Fall/Spring/Summer). All analyses were conducted in R version 4.3.0.

Results

Baseline clinical characteristics of participants

Of 238 patients with moderate-to-severe asthma, 162 were not receiving biologic therapy but met criteria for a biologic and 76 were on omalizumab, mepolizumab, benralizumab, or dupilumab. Both groups had similar mean age, female proportion, and BMI (Table 1). Patients not on biologic therapy at the time of sample collection were like those on biologic therapy on clinical and demographic characteristics. However, they had a lower exacerbation burden in the year prior to sample collection (1.2 vs. 2.2 exacerbations, p = 0.002) than T2 biologics users. A lower proportion of the non-biologic users also had CRSwNP, though that was not statistically significant (13.6% vs. 23.7%, p = 0.08). Maximum blood eosinophil counts were similar in both groups (Table 1). Following matching, we had 59 biologics-treated and 59 untreated controls who met eligibility for biologics at the time of sample collection. Both groups had similar clinical characteristics, including age, sex, smoking status, comorbidities, and baseline exacerbations (Table 1; Figure S1).

Distribution of cytokines

All values were above the LOQ for eight of the 15 cytokines in both biologic users and patients not on biologics (Table S2). Four cytokines (IL-17 A, IL-33, IL-4, and TSLP) had 80% or more of the assays below the LOQ in the non-biologics group. In biologic users, three of these four cytokines (IL-17 A, IL-33, and TSLP) had ≥ 80% of values below the LOQ. The median and interquartile ranges (IQR) of the raw and half-minimum imputed values were similar (Table S2). Raw values were mostly right-skewed (Figure S2, S3) in both groups. All values were rank normalized to have a mean of 0 and standard deviation of 1. The values of IL-17 A, IL-33, IL-4, TSLP, and IL-13 did not follow this standardized distribution (Figure S4, S5).

Benchmarking against previously established findings in the NHLBI Severe Asthma Research Program (SARP) in patients not on biologics

Of 15 cytokines, IL-6 was the cytokine with the highest observed variance explained by clinical and lifestyle factors at 27.6% (Fig. 1A). BMI had the greatest contribution to this observed variance (7.8%), followed by 6.4% from baseline exacerbations requiring hospitalization. We evaluated the relationship between IL-6 levels and BMI as well as baseline asthma exacerbations since these were previously reported relationships. We found a weak yet significant correlation between BMI and IL-6 levels (r = + 0.29, p < 0.001): the higher the BMI, the higher the IL-6 level (Fig. 1B). Violin plots demonstrated significantly higher IL-6 levels in obese than in non-obese patients (p = 0.005) (Fig. 1C). While there were slightly higher IL-6 levels in patients with higher baseline asthma exacerbation counts (including exacerbations leading to hospitalizations and those not requiring hospitalization), these differences were not statistically significant when we grouped patients by their baseline asthma exacerbation (0 vs. ≥ 1; p = 0.14).

Fig. 1

Clinical/lifestyle factors and cytokine levels in patients not on monoclonal antibody therapy who met the eligibility criteria for T2 biologics at the time of sample collection. A Observed variance explained by the clinical/lifestyle factors selected by LASSO regression for each factor (upper) and combined factors (lower) for 15 cytokine levels. B IL-6 levels in relation to body mass index (BMI). The red line shows the positive association between BMI and IL-6 level with 95% confidence intervals presented as grey area. C Comparison of IL-6 levels by obesity and exacerbation counts (including exacerbations leading to hospitalizations and those not requiring hospitalization). Statistical tests for difference between groups are conducted using independent t-test.

BMI was also associated with five other cytokines though it contributed less to their observed variance: CXCL10 (5.3%), IL-17 C (4.3%), IL-18 (4.0%), TNF-α (2.1%), and IL-1β (2.0%). (Tables 1 and 2). Asthma admissions and IgE levels contributed to many cytokines, but with lower variance explained. IL-6 (28% variance explained; mostly by BMI and exacerbations) and CCL11 (20%; mostly age and maximum eosinophil count) were the top cytokines that could be explained by clinical features in the non-biologics group. In subgroup analyses, while BMI showed more prominent effects on multiple cytokine levels in patients with concomitant allergic rhinitis while baseline asthma admission explained more observed variance in cytokine levels in patients without allergic rhinitis (Tables S3 and S4). Patients with and those without obesity had distinct patterns of association between clinical/lifestyle factors and cytokine levels (Table S5 and S6). Baseline asthma admission contributed more to cytokine levels than BMI in patients with blood sample collected in winter but less to the observed variance than BMI in those with blood sample collected in spring, summer or fall (Figures S6 and S7, and Tables S7 and S8). IL-6 and IL-17 C were the top cytokines explained in the Winter while IL-18 and CXCL10 were top in the Spring/Summer/Fall.

Table 1 Baseline characteristics at sample collection

Table 2 Association between clinical/lifestyle factors and cytokine levels

T2 biologics users

For patients with current T2 biologics use, smoking status was associated with the highest number of cytokines (Fig. 2), contributing to the observed variance of CXCL10, TNF-α, IL-18, IFN-γ, IL-6, and IL-1β (Fig. 2A and Table 3). Comparing the levels of these six cytokine levels in smokers and non-smokers, smokers had higher levels with significant differences in CXCL10 (p = 0.002), TNF-α (p = 0.02), IL-18 (p = 0.004), IFN-γ (p = 0.01), and IL-6 (p = 0.04) levels, but not IL-1β (p = 0.15) (Fig. 2B). Female sex was positively associated with CXCL10 and IFN-γ but negatively associated with IL-18 levels. Metabolic syndrome was also positively associated with multiple cytokines with its highest contribution to IL-6 and IL-1β.

Fig. 2

Clinical/lifestyle factors and cytokine levels in T2 monoclonal antibody users. A Observed variance explained by the clinical/lifestyle factors selected by LASSO regression for each factor (upper) and combined factors (lower) for 15 cytokine levels. B Comparison of 6 cytokine levels by history of smoking. Statistical tests for difference between groups are conducted using Wilcoxon rank-sum test.

Table 3 Association between clinical/lifestyle factors and cytokine levels in patients not on biologics who met the eligibility criteria for T2 biologics at the time of sample collection

Table 4 Association between clinical/lifestyle factors and cytokine levels in T2 biologics users

About a quarter or more of the observed variance of CCL11, CXCL10, and IL-18 could be explained by clinical and lifestyle factors; 20.3% for IFN-γ, 20.1% for IL-4, and 14.1% for IL-6 levels (Table 3). BMI contributed the most to the observed variance in CCL11 level, followed by IgE and public insurance. TNF-α and IL-10 were the top cytokines explained in the Winter while IL-4 and IL-33 were the top cytokines in the Spring/Summer/Fall (Tables S9 and S10).

Matched analysis between T2 biologics users and non-biologics users

Results were generally consistent when focusing on the matched cohort (n = 118). As observed in the full cohort of controls, the matched controls exhibited significant associations between BMI and IL-6 and baseline burden of asthma admissions contributed significantly to IL-6 level (Fig. 3A and Table S11). In the matched cohort of T2 biologics users, as in the full-cohort analyses, smoking contributed to multiple cytokine levels including CXCL10 (29.4%), IFN-γ (22.7%), IL-10 (21.5%), TNF-α (13.7%), IL-18 (13.4%), and IL-1β (8.1%). In this T2 biologics cohort, BMI was associated with only two cytokines: including positive association with IL-4 (9.3%) and negative association with CCL11 (6.3%). Smoking in conjunction with maximum eosinophil counts explained over 30% of the observed variance in CXCL10 levels (Fig. 3B and Table S12). Smoking remained the major factor contributing to cytokine levels in patients with blood sample collected in winter but not in those with blood sample collected in spring, summer or fall where the presence of allergic rhinitis, BMI, and metabolic syndrome were the main contributors to the observed variance in cytokine levels (Figure S8 and S9). TNF-α and IL-10 were the top cytokines in the Winter while IL-4 and IL-33 were top in the Spring/Summer/Fall.

Fig. 3

Clinical and lifestyle/factors and cytokine levels in matched analysis. A Observed variance explained by the clinical/lifestyle factors selected by LASSO regression for each factor (upper) and combined factors (lower) for 15 cytokine levels in matched control group. B Observed variance explained by the clinical/lifestyle factors selected by LASSO regression for each factor (upper) and combined factors (lower) for 15 cytokine levels in matched T2 monoclonal antibody group. C Differential expression analysis between matched T2 monoclonal antibody group versus matched control group using limma.

Subdividing by most recent eosinophil count (measured in cells/mcl), IL-6 and TNF-α were the top cytokines with variance explained in the BEC ≥ 150 group (Tables S13). Only 19 matched non-biologics user had BEC < 150 and so analyses not conducted in this group. In the matched biologics user group, TNF-α, IL-1β, and IL-33 were the top cytokines explained in the BEC ≥ 150 group with smoking the leading clinical feature associated with TNF-α and IL-1β (Table S14). However, in the BEC < 150 group, the leading cytokines associated with clinical features were CCL11, IL-4, and CXCL8, with smoking making little to no contribution (Table S15).

Differential expression analysis showed a higher level of IL-1β in matched controls not on biologics compared to the T2 biologics users (p = 0.04) (Fig. 3C and Table S16). IL-1β was only associated with allergic rhinitis in matched controls and associated with smoking (8.1%) in matched T2 biologics users. IL-1β remained the top cytokine in differential expression analyses after excluding cytokines with ≥ 80% of values below the detection limit (TSLP, IL-17 A, and IL-33), p = 0.055 (Figure S10 and Table S17).

Subgroup analyses in biologic user group by biologic used

Patients on anti-IL5 were older compared to those on omalizumab or dupilumab, had more asthma admissions in the prior year, and higher OCS use (Table S18). IFN-gamma (51%), IL-10 (50%) and IL-6 (40%) were the top cytokines that could be explained by clinical features in omalizumab users with exacerbations and smoking being the predominant clinical features. (Table S19). For the anti-IL5 group, the effects of maximum eosinophil count and smoking history on cytokine levels were more prominent with less contribution from baseline exacerbations (Table S20). Clinical features explained 64% of the variance in CCL11, 58% in IL-18, and 40% in IL-4 levels. Baseline asthma admission contributed the most in the dupilumab group with less impact from smoking history (Table S21). IL-4 (83%), CXCL10 (63%), IL-6 (45%) and CCL11 (45%) were the top cytokines that could be explained by clinical features. When split by biologic, IL1β was lower in the combined cohort and significantly lower in the anti-IL5 group (Fig. 4).

Fig. 4

Comparison of IL-1β levels between matched T2 monoclonal antibody group versus matched control group, by biologic. Statistical tests for difference between groups are conducted using independent t-test for omalizumab, anti-IL5, and dupilumab subgroups.

Discussion

This study examined the relationship between clinical and lifestyle factors, such as BMI and smoking, and Th1/2/17 cytokine levels in 238 patients with moderate-to-severe asthma. We found that in patients not receiving biologic therapy but who met criteria for an anti-IgE, anti-IL5, or anti-IL4/13 at the time of sample collection, BMI was strongly associated with levels of several proinflammatory cytokines, including IL-6, confirming previous research. In patients receiving T2 biologics, this association weakened, and smoking emerged as the primary lifestyle factor linked to proinflammatory cytokines like CXCL10 and TNF-α. These conclusions were consistent in analyses matching biologic-treated patients to controls. Interestingly, biologic treatment also altered the relationships between different cytokines, with CXCL8 showing a stronger correlation with IL-1β and TNF-α. Notably, IL-1β levels were lower in patients receiving any of the T2 biologics. These findings suggest that biologic therapy may shift the inflammatory landscape in asthma and could be leveraged for identifying new biomarkers that can predict treatment response.

The association observed between BMI and IL-6 levels in severe asthma patients has been previously noted in the well-characterized severe asthma cohort, SARP [9, 11], and our finding also suggested a positive association between asthma exacerbations and IL-6 levels, though this did not reach statistical significance [9]. This is also consistent with findings from U-BIOPRED and BIOAIR, which also pointed to an association between IL-6 levels and asthma severity [28], increased exacerbations [29], and hospital admissions [30]. Our findings reinforce the potential benefit of anti-IL6 therapy in severe asthma, which is one of the targets of the Precision Interventions for Severe and/or Exacerbation-Prone Asthma Network (PrecISE) [31, 32]. Though the asthma subgroup likely to benefit the most from this therapy is still unclear, with initial results from PrecISE suggesting no benefit from the anti-IL6, clazakizumab, in predominantly T2-low asthma patients [Oral abstract presentation at the 2025 American Thoracic Society, San Francisco]. Of note, we noted in our study that the use of monoclonal antibody therapy for asthma seemed to disrupt the association between BMI and IL-6 levels while making the relationship between obesity-independent metabolic syndrome and IL-6 levels more prominent.

The relationship between eosinophilia and IL-6 levels appears to be more complex. In our cohort, we find a positive association between the maximum blood eosinophil count (BEC) and IL-6 levels in patients who had not received biologics. These findings should be contrasted to studies where no association between IL-6 levels and most recent BEC were noted [11], and compared to positive associations previously found between IL-6 and non-allergic asthma in obese patients [33]. The conflicting findings are potentially explained by difference between the choice of BEC used: we used the maximum BEC, but the SARP study used most recent BEC. We have found that the maximum BEC and most recent BEC may correlate with various processes differentially in patients with severe asthma who often require eosinophil-depleting corticosteroid therapy. For instance, we found the maximum BEC to be a better predictor of response to the anti-eosinophilic agent, mepolizumab compared to most recent BEC [8]. Subgroup analysis showed a more prominent effect of BMI on cytokine levels in patients with concomitant allergic rhinitis and those with blood sample collected during the pollen allergy seasons. Importantly, we observed that cytokine levels varied not only with patient features such as BMI and biologic use, but also with the season of sample collection. Seasonality has not previously been considered as an important covariate in major biomarker initiatives in asthma, yet our findings suggest it may represent an important factor to incorporate into predictive models. While Th1 cytokines were the top cytokines with variance explained in the winter, alarmins and Th2 cytokines were the top cytokines in the Spring/Fall/Summer. These suggests that season of the year might alter the proteome and this needs to be considered in biomarker studies when patients might be recruited at different times of the year, or the same patient might be followed longitudinally across different seasons.

The patterns of association differed slightly by the biologic received and related largely to their mechanisms of action. For instance, in patients on anti-IL5s which are anti-eosinophilic agents, CCL11 (eotaxin-1), a key chemokine in recruiting eosinophils to the lungs that has been shown to be associated with response to anti-IL5 therapy [34, 35], was the top cytokine that could be explained by clinical features while in users of the anti-IL-4/IL-13 agent, dupilumab, IL-4 (83%) was the top cytokine with variance explained by clinical features. These suggest that the patterns of cytokine relationships seen here may reflect cytokines affected either directly or indirectly by these biologics and should be considered as possible response biomarkers. Importantly, since the human body is in a state of equilibrium, when one inflammatory pathway is blocked, another pathway might become hyperactive via a compensatory mechanism. For example, T2-blockage might lead to overly active Th1 and Th17 pathways. This is consistent with our finding that smoking, which promotes Th1 inflammation, was the leading lifestyle factor associated with Th1 cytokine levels in patients on T2 biologics but was not associated with T2 cytokine levels or alarmins. In murine models, cigarette smoke exposure has been shown to change asthma endotypes and worsen airway remodeling, potentially leading to difficult-to-control disease with refractoriness to conventional anti-inflammatory therapy [36]. The effect of smoking was more prominent in patients with blood sample collected in winter. In our matched analysis, smoking was positively associated with IL-10 and TNF-α which is consistent with the previous study demonstrating higher IL-10 and TNF-α production by airway macrophages in smokers [37]. In our study, asthma patients with a smoking history under Th2 blockage had higher TNF-α levels. Advising smoking cessation remains important even in patients receiving biologic therapy given that blockage of T2 pathways may leave proinflammatory Th1 cytokines unchecked [38]. However, there is no evidence to date of the development of T2-low asthma related to increased Th1 or Th17 activity in human studies of the anti-IL-4/13 monoclonal antibody use [39, 40].

In differential expression analysis, we demonstrated lower IL-1β in the T2 biologics group compared to controls. High IL-1β levels have been shown to be associated with neutrophilic airway inflammation, asthma severity, and corticosteroid resistance [41]. Endoplasmic reticulum (ER) stress in asthma also correlates with both Th2 cytokines and IL-1β [42]. Furthermore, IL-1β increases mucin production and decreases airway epithelial barrier integrity [43]. A recent study showed that IL-5 and IL-13 correlate with both IL-1β mRNA expression and NOD-, LRR- and pyrin domain-containing protein 3 (NLRP3) inflammasome in obesity-associated asthma [44]. Thus, IL-5 or IL-4 targeted therapies may reduce NLRP3 inflammasome, leading to decreased caspase 1-dependent release of IL-1β [45]. While there has been some interest in IL-1β as a target for asthma treatment [46], with a prior study showing promising in vitro [47], and in vivo results of the IL-1β receptor antagonist [48], anakinra, reducing neutrophilic airway inflammation, this is still far from translation into clinical practice. As Th2 cytokines have been extensively explored as potential predictors for T2 monoclonal antibody response without great success, shifting our focus to the dynamics of proinflammatory Th1 cytokines and how these might change with clinical and environmental factors might hold a key in identifying predictive biomarkers and warrants further studies.

Our results should be interpreted with caution. First, this study is cross-sectional and cytokine levels fluctuate with clinical status, treatment, and environmental exposures. Additionally, for some cytokines, notably the alarmins (TSLP and IL-33), a majority of samples had quantification below the limit of detection for the assays used. Thus, no temporal explanation can be made based on these results and these results need to be validated in larger prospective cohorts with repeated sampling of the same individuals before and after biologics’ use and using other assays. Secondly, treated patients and controls may differ in ways beyond what we have captured with risk of residual confounding bias. Also, despite evaluating for and adjusting for batch effects, the samples were collected at various times in the disease trajectory of patients, and this might have influenced our findings. We sought to mitigate these limitations by using samples from a well-characterized Biobank with samples drawn using standard and evidence-based protocols; sought to replicate a previously established association between IL-6 and clinical features; and matched treated patients to controls on key characteristics. Thirdly, we used broad categorizations in subgroup analyses, but these might be insufficiently granular to explain findings or underlying mechanisms. For instance, the impact of season of sample collection on Th-2 cytokines might differ in patients with seasonal allergic rhinitis from patients without. Furthermore, our study involved multiple comparisons which could inflate our statistical error. To handle this, we used a regularized-based method which did not rely on statistical significance. Lastly, age-related changes in immune function could contribute to variance in cytokine levels and our cohort had a mean age of about 56.5 years, thus generalizability to younger patients might be limited.

In summary, our study provides exploratory insights into how circulating Th1/2/17 cytokine levels in severe asthma are influenced by clinical characteristics, treatment status, and season of sample collection. We evaluated the impact of clinical and environmental factors on the proteome of moderate-to-severe asthma patients, focusing on Th-1/2/17 cytokines and assessed how biologic therapy may modify these associations. Consistent with prior reports, we replicated the association of IL-6 with BMI and asthma admissions. In patients not on biologics, BMI emerged as the predominant clinical variable influencing multiple cytokine levels, whereas in those receiving omalizumab or other Th2-targeted biologics (anti-IL5 or anti-IL4/13), smoking explained the greatest variance in several proinflammatory cytokines. We also observed lower IL-1β level in patients on biologic therapy, suggesting this cytokine may warrant further study as a potential biomarker of biologic response especially in patients on anti-IL5 therapy. While these findings require replication in larger, longitudinal studies, they highlight novel considerations, particularly seasonality and treatment context, that may inform future biomarker development in severe asthma.