Viral and Host-Related Factors in the Nasal Inflammatory Response to Critical Bronchiolitis in Infants and Children

Abstract

Objectives:

We aimed to evaluate the inflammatory cytokine response in children with critical bronchiolitis. Our secondary aims included: 1) understanding differences in inflammatory profiles between respiratory syncytial virus (RSV), rhinovirus (RV) and polyviral infection; and 2) examining the association between clinical risk factors and the host inflammatory response.

Design:

A single-center, prospective cohort study of PICU patients with critical bronchiolitis.

Setting:

Medium-sized academic PICU from October 2021 to July 2024.

Patients:

42 patients with critical bronchiolitis and 9 healthy control patients.

Measurements and Main Results:

Nasal aspirates were obtained and cytokines measured by ELISA (enzyme-linked immunosorbent assay). The influence of viral type and host factors was assessed by non-parametric testing. In an exploratory analysis, cells from the nasal aspirates of a select group of patients underwent next-generation RNA sequencing (RNASeq). The nasal samples from patients with critical bronchiolitis exhibited an increase in interferons, type 1, and type 2 immunity related cytokines. We did not identify differences in cytokine levels when comparing patients with rhinovirus to those with RSV infection. Patients with multiple (>1) bronchiolitis admissions exhibited lower levels of inflammatory cytokines compared to patients during their first bronchiolitis admission. Using RNASeq, a subgroup of patients with RV or RSV exhibited similar upregulation of gene pathways associated with leukocyte chemotaxis, interferon signaling and tumor necrosis factor superfamily cytokines compared to healthy controls. Pathways associated with ciliary function were downregulated compared to controls with RSV inducing a more pronounced downregulation than RV.

Conclusion:

These data demonstrate that patients with critical bronchiolitis exhibit a robust host inflammatory response independent of viral etiology. Our exploratory transcriptomic analysis showed upregulation of proinflammatory genes and downregulation of ciliary genes consistent with clinical characteristics of these patients at the bedside.

Approximately 25% of children hospitalized for viral bronchiolitis are cared for in a PICU, making it one of the leading causes of non-elective PICU admissions (1). Such children cared for in the PICU are now considered to have “critical bronchiolitis.” Hospitalization duration often spans multiple days, placing strain on hospital and PICU resources particularly during traditional respiratory viral seasons (2). The proportion of children with critical bronchiolitis, and the use of non-invasive mechanical ventilation to support them, has increased in recent years, particularly in the wake of the COVID-19 pandemic (3). But it is unclear whether this increase is due to a change in pathogen virulence, host immunologic response, or an evolving PICU patient population.

Bronchiolitis can be caused by many respiratory viruses (4). While the host immune response to respiratory syncytial virus (RSV) bronchiolitis is well understood, the inflammatory profile of patients with critical bronchiolitis as a general clinical phenotype remains under investigation (5-14). Although bronchiolitis is a disease of the lower airways, nasal epithelial cells can serve as effective surrogates for measuring markers of illness-induced inflammation (15). Studies have measured airway inflammation in invasively mechanically ventilated patients with bronchiolitis, however, nasal epithelial sampling has not been used to measure inflammation in the ever-increasing number of patients on non-invasive support (14).

Hence, our primary aim was to evaluate the inflammatory cytokine response in patients with critical bronchiolitis, inclusive of all viral triggers, in comparison with healthy controls. Secondarily, within the critical bronchiolitis cohort, we sought to investigate associations of: a) specific viral triggers; and b) host-related factors on the cytokine response. We hypothesized that, compared with healthy controls, patients with critical bronchiolitis would exhibit a predominant type 1 inflammatory response rather than a type 2 inflammatory response. We also hypothesized that a) the inflammatory response would not significantly differ based on individual virus or polyviral infection and b) the inflammatory response in critical bronchiolitis would differ in those patients experiencing their first bronchiolitis admission compared to those with multiple bronchiolitis admissions. Finally, we sought to complete an exploratory transcriptomic analysis in a subgroup of patients with critical bronchiolitis.

METHODS

In this single-center, prospective, cohort study we evaluated the inflammatory response in pediatric patients with critical bronchiolitis. The study was approved by the Institutional Review Board (IRB) at the University of Michigan (HUM00133414, 11/15/2017). All study procedures were conducted in accordance with the Helsinki Declaration of 1975. Informed consent was obtained from all patients enrolled.

For this analysis, we included a representative sample of patients admitted to the PICU with bronchiolitis. Eligible patients had an admission diagnosis of “bronchiolitis,” which was corroborated by the clinical team before enrollment. Critical bronchiolitis was defined as bronchiolitis with acute respiratory failure requiring support via high-flow nasal cannula (HFNC), non-invasive positive pressure ventilation (NIPPV) or invasive mechanical ventilation (IMV). Patients were excluded if they had a contraindication to nasal suctioning, immunodeficiency or malignancy, or were being treated for secondary bacterial infections. Healthy control patients were enrolled from children undergoing elective procedures. Enrollment occurred on weekdays from October 2021 to July 2024 when immediate sample processing was available.

Patient demographics, medical history and admission characteristics (length of stay [LOS], medications used, and level of respiratory support) were collected via chart review and stored in a REDCap (research electronic data capture) database.

Procedures

To minimize time off HFNC or NIPPV, nasal aspirates were obtained by a study team clinician at a timepoint when the clinical team confirmed safety of sampling and suctioning was deemed clinically necessary by bedside staff. Sample processing methodology is described in the Supplemental Digital Content (SDC), Appendix S1. For patients with tracheostomies, families were allowed to opt for a tracheal aspirate collection. Supernatant cytokines were analyzed by electrochemiluminescence biomarker assay (Meso Scale Discovery, Rockville MD). In total, 16 cytokines were measured (interferon-α2a [IFN-α2a], IFN-β, IFN-λ, IFN-γ, interleukin-1 β [IL-1β], IL-2, IL-4, IL-5, IL-8, IL-10, IL-13, IL-12p70, IL-17A, IL-25, IL-33, and tumor necrosis factor-α [TNF-α]). We excluded samples that failed to meet two of the following three criteria for concerns of poor quality: 1) nine or more measured cytokines above the lower limit of detection (LLD); 2) a live cell count > 400,000/mL; and 3) detectable total protein count. Cytokine measurements below the LLD were replaced with a value equal to the LLD divided by the square root of two (16). An exploratory analysis was completed in a subgroup of patients with the highest RNA integrity numbers to compare gene expression data in patients with critical bronchiolitis to healthy controls (see SDC Appendix S2).

Outcomes and Analyses

Our primary outcome variable was cytokine level. Our primary analysis compared cytokine levels in patients with critical bronchiolitis to healthy controls. Our a priori secondary analyses compared our primary outcome variable (cytokine levels) across multiple subgroups including: patients with critical bronchiolitis with rhinovirus (RV) versus RSV infections, single virus versus polyviral infection, systemic steroid use during their present illness versus no steroid use, and patients during their first bronchiolitis admission compared with those with a history of previous bronchiolitis admissions. Our secondary outcomes included PICU LOS and pulse oximetry oxygen saturation to fraction of inspired oxygen (S:F) ratios.

Data are summarized using median and interquartile range (IQR). Proportions are presented as percentages when the denominator is >50. However, the comparison between proportions is summarized with the difference in percentage and the 95% confidence interval (CI) of the difference. Continuous data were compared using the two-sample t-test. We used Mann-Whitney U (Wilcoxon rank-sum) testing to compare cytokine levels. No corrections are made for multiple comparisons. Chronic medical conditions analyzed included the need for prior home respiratory support, asthma or bronchopulmonary dysplasia (BPD), cardiac disease, neurologic disorders, or genetic syndromes. Data were analyzed using STATA (STATA S.E. v18.0, College Station, Tx) software.

RESULTS

Over the 34 months (October 2021 to July 2024) of study, 42 patients with critical bronchiolitis and 9 healthy controls were enrolled. The characteristics of the two groups are summarized in Table 1. Patients with critical bronchiolitis were younger than controls (median 5 [IQR 2-13] vs , 88 [IQR 78-103] months, p <0.001). The bronchiolitis cohort included one patient between 24 and 33 months of age; a sensitivity analysis excluding this patient showed no difference in primary outcome.

TABLE 1:

Demographic and baseline medical characteristics of the bronchiolitis and control cohorts

Characteristics	Total n=51	Control n=9	Bronchiolitis n=42
Age, mo, median (IQR)	10 (3-21)	88 (78-103)	5 (2-13)
Female, n (%)	17 (33%)	3/9	14/42
Race/ethnicity, n (%)
Asian	1 (2%)	0	1/42
Black	5 (10%)	1/9	4/42
White	42 (83%)	7/9	35/42
Other	3 (6%)	1/9	2/42
Smoke exposure, n (%)
Yes	9 (17%)	3/9	7/42
No	26 (51%)	2/9	24/42
Unknown	16 (31%)	4/9	11/42
Baseline medical characteristics, n (%)
Supplemental oxygen use	9 (17%)	0	9/42
Tracheostomy use	4 (8%)	0	4/42
Asthma	6 (12%)	1/9	5/42
Bronchopulmonary dysplasia	5 (10%)	0	5/42
Airway malacia	9 (17%)	0	8/42
Congenital heart disease	6 (12%)	0	6/42
Neurologic disorder	7 (13%)	4/9	4/42
Cerebral palsy	2 (4%)	2/9	0
Genetic syndrome	9 (17%)	0	8/42
No medical history	25 (50%)	5/9	21/42

Abbreviations: IQR = Interquartile range.

In the 42 critical bronchiolitis patients, the most common virus isolated was RV in 18, followed by RSV in 10, and polyviral infections in another 10 (Table 2). All polyviral infections included RV. In the 42 critical bronchiolitis patients, 31 were experiencing their first admission for bronchiolitis; 21/42 patients with critical bronchiolitis had at least one chronic medical condition (Table 1). None of the patients required endotracheal intubation or met criteria for the pediatric acute respiratory distress syndrome (PARDS) (17). The median S:F ratio in the bronchiolitis cohort was 384 (IQR 330-448), and the PICU median LOS was 3 (IQR 2-4) days (Table 2). There were no significant differences in cytokine levels between patients sampled on day 1 or 2 of admission compared to those sampled on day 3 or 4 (see SDC Figure S1).

TABLE 2:

Clinical Course and Admission Characteristics for Bronchiolitis Cohort as whole, as well as First-Time Bronchiolitis Admission and Multiple Bronchiolitis Admission Cohorts

Clinical Characteristics	Total n=42	First-Time Bronchiolitis Admission n=31	Multiple Bronchiolitis Admissions n=11
Hospital Day Patient Sampled On, n/N
Days 1-2	17/42	13/31	4/11
Days 3-4	23/42	17/31	6/11
Days 5 +	2/42	1/31	1/11
Virus Isolated on PCR, n/N
Rhinovirus	18/42	10/31	8/11
RSV	10/42	9/31	1/11
Poly-viral Infection	10/42	9/31	1/11
Enterovirus	1/42	1/31	0
Human Metapneumovirus	1/42	0	1/11
Coronavirus (non-COVID 19)	1/42	1/31	0
Parainfluenza	1/42	1/31	0
Type of Respiratory Support, n/N
Nasal NIPPV or HFNC	24/42	16/31	8/11
Oronasal NIPPV	18/42	15/31	3/11
Scheduled Albuterol Used During Admission, n/N
Yes	18/42	9/31	9/11
Scheduled Systemic Steroids Used During Admission, n/N
Yes	6/42	4/31	2/11
Single or Poly-virus infection, n/N
Single Virus Infection	32/42	22/31	10/11
Poly-viral Infection	10/42	9/31	1/11
S:F Ratio, median (IQR)	386 (330 - 452)	330 (317 - 388)	388 (365 - 457)
PICU LOS, median (IQR) days	3 (2 - 4)	3 (1 – 4)	3 (2 – 5)

Abbreviations: IQR = Interquartile range, PCR = Polymerase Chain Reaction, RSV = Respiratory syncytial virus, NIPPV = non-invasive positive pressure ventilation, HFNC = High-flow nasal cannula, LOS = length of stay, S:F = oxygen saturation to fraction of inspired oxygen.

Cytokine levels in airway aspirates from children with critical bronchiolitis exhibited a prominent type 1 inflammatory response compared to healthy control patients, including significant elevations in IFN-γ, IL-1β, IL-2, IL-12 and TNF-α (Table 3 and SDC Figure S2). The critical bronchiolitis group also showed statistically significant elevations in the type 2 inflammatory cytokines (IL-4, IL-10, and IL-13) compared to healthy control patients. The type I interferons IFN-α and IFN-β, and the type III IFN-λ, also significantly increased (Table 3).

TABLE 3:

Quantitative Cytokine Values for Patients with Bronchiolitis Compared to Healthy Controls

Cytokine (units)	Bronchiolitis Patients n=42 Median (IQR)	Control Patients n=9 Median (IQR)	p-value
IFN-α2 (fg/mL)	49.1 (8.8 – 179.7)	3.5 (2.4 – 49.2)	<0.01
IFN-β (fg/mL)	120.5 (18.4 – 358.0)	12.0 (12.0 – 12.0)	<0.01
IFN- γ (fg/mL)	1,163.4 (318.8 – 1,896.7)	127.1 (30.0 – 227.3)	<0.01
IFN-λ1 (pg/mL)	12.3 (2.6 – 32.6)	0.9 (0.8 – 4.9)	<0.01
IL-1β (pg/mL)	64.8 (21.7 – 228.6)	15.0 (4.1 – 15.1)	<0.01
IL-2 (fg/mL)	569.4 (173.9 – 1,718.0)	8.5 (8.5 – 68.3)	<0.01
IL-4 (fg/mL)	55.9 (13.5 – 167.9)	5.7 (5.7 – 12.1)	0.02
IL-5 (fg/mL)	649.7 (63.0 – 1,877.0)	80.9 (18.4 – 187.7)	0.2
IL-8 (fg/mL)	1,077.7 (1,006.1 – 1,124.7)	844.8 (495.4 – 1,055.5)	0.05
IL-10 (fg/mL)	17,155.1 (6,213.3 – 59,007.9)	411.8 (58.8 – 1,033.2)	<0.01
IL-12p70 (fg/mL)	424.1 (89.3 – 977.0)	30.4 (30.4 – 30.4)	<0.01
IL-13 (pg/mL)	32.6 (20.7 – 39.1)	16.5 (15.0 – 17.9)	<0.01
IL-17A (fg/mL)*	2,895.6 (896.1 – 8,131.9)	1,596.1 (1,100.9 – 3,202.4)	0.7
IL-25 (pg/mL)	0.39 (0.19 – 0.41)	0.41 (0.34 – 0.41)	0.3
IL-33 (pg/mL)	2.8 (0.7 – 27.2)	57.6 (19.3 – 156.7)	<0.01
TNF-α (fg/mL)	35,000 (7,059.1 – 62,601.1)	612.8 (126.3 – 1,894.6)	<0.01

Abbreviations: IQR = Interquartile range, IFN = interferon, IL = interleukin, TNF = tumor necrosis factor

Significance was determine using Wilcoxon rank sum testing.

^{* =}Bronchiolitis patient n = 14 and Control patient n = 4

We then compared cytokine levels in critical bronchiolitis subgroups with different viral triggers. Inflammatory cytokine levels in patients with bronchiolitis secondary to RSV were not significantly different when compared to those with RV, except for IL-12p70 (p=0.01) and TNF-α (p=0.001) which were both elevated in those infected with RSV (see SDC Table S1). We found no significant differences in cytokine levels between patients with a single viral infection compared to those with polyviral infections (data not shown). There were no differences in PICU LOS or S:F ratios between RSV and RV or single virus versus polyviral infection (data not shown).

We next compared cytokine levels between critical bronchiolitis subgroups based on host-related factors. Patients who received systemic steroids (n=6) had lower levels of IL-4 (p=0.03) and IL-5 (p<0.01) compared to those who did not (n=35) (see SDC Figure S3). When comparing patients with bronchiolitis admissions prior to enrollment (n=11) to those sampled on their first admission for bronchiolitis (n=31), patients with multiple bronchiolitis admissions experienced a relatively blunted type 1 and 2 cytokine response (Table 4 and see SDC Figure S4). Demographics of these two groups were similar (see SDC Table S2), except that those patients with multiple bronchiolitis admissions versus those during their first admission, had more medical comorbidities (10/11 vs 10/31; mean difference 58.6% [95%CI of difference 25.0 to 74.2%], p<0.001). There were no differences in S:F ratio (p=0.14), use of systemic steroids (p=0.67), or PICU LOS (p=0.6) between patients with multiple bronchiolitis admissions prior to enrollment and patients on their first admission for bronchiolitis (Table 2).

TABLE 4:

Quantitative Cytokine Values for First Bronchiolitis Admission Patients compared to Multiple Bronchiolitis Admission Patients

Cytokine (units)	First Admission Patients n=31 Median (IQR)	Multiple Admission Patients n=11 Median (IQR)	p-value
IFN-α2 (fg/mL)	39.2 (8.8 – 160.7)	63.5 (3.4 – 238.3)	0.8
IFN-β (fg/mL)	98.3 (18.4 – 307.2)	128.2 (12.0 – 439.2)	0.8
IFN- γ (fg/mL)	1,276.9 (412.6 – 1,896.7)	1,068.5 (124.4 – 3,520.3)	0.6
IFN-λ1 (pg/mL)	12.5 (2.2 – 32.1)	12.0 (2.7 – 100.3)	0.8
IL-1β (pg/mL)	95.2 (50.9 – 300.7)	18.6 (4.6 – 51.9)	<0.01
IL-2 (fg/mL)	800.8 (326.3 – 2,399.3)	131.9 (27.2 – 431.3)	<0.01
IL-4 (fg/mL)	60.2 (32.1 – 262.9)	6.6 (5.7 – 58.6)	<0.01
IL-5 (fg/mL)	816.8 (99.3 – 3,218.4)	103.9 (27.4 – 1,321.5)	0.08
IL-8 (fg/mL)	1,079.8 (1,047.9 – 1,116.5)	1,065.4 (850.4 – 1,768.2)	0.8
IL-10 (fg/mL)	21,996.5 (9,650.2 – 63,761.3)	3,708.4 (1,006.6 – 16,980.2)	0.02
IL-12p70 (fg/mL)	474.2 (227.0 – 2,053.9)	48.3 (30.4 – 425.9)	<0.01
IL-13 (pg/mL)	35.3 (24.5 – 40.9)	22.6 (16.1 – 35.7)	0.03
IL-17A (fg/mL)*	8,122.8 (3,888.3 – 11,276.4)	1,184.1 (287.6 – 1,902.8)	0.1
IL-25 (pg/mL)	0.31 (0.15 – 0.41)	0.41 (0.32 – 0.41)	0.2
IL-33 (pg/mL)	3.3 (1.4 – 27.2)	0.5 (0.4 – 56.9)	0.3
TNF-α (fg/mL)	35,000 (17,849,8.1 – 124,054.6)	3,252.2 (2,158.9 – 38,927.3)	<0.01

Abbreviations: IQR = Interquartile range, IFN = interferon, IL = interleukin, TNF = tumor necrosis factor

Significance was determine using Wilcoxon rank sum testing.

^{* =}Bronchiolitis patient n = 7 and Control patient n = 7

We further explored the transcriptomic profiles of nasal aspirate cells comparing 12 patients with critical bronchiolitis and 4 healthy controls. Analysis showed a total of 1,496 differentially expressed genes (DEGs), with 1,307 upregulated genes and 189 downregulated genes (see SDC Figure S5). Upregulated families of genes included those involved in leukocyte migration, antiviral and antibacterial responses (Figure 1A). Downregulated families of genes included transcripts encoding ciliary proteins (Figure 1B). Analysis of individual DEGs revealed an upregulation of many transcripts encoding type 1 pro-inflammatory cytokines including IL1A, IL1B, and TNFA (see SDC Table S3) as well as transcripts encoding type I and type III IFNs, IFN-stimulated genes, TNF-stimulated genes, receptors for RV and RSV, and Toll-like receptor signaling. A subgroup analysis comparing patients with RSV to RV bronchiolitis showed similar expression of genes associated with interferon signaling, leukocyte chemotaxis, and TNF superfamily cytokines, but a more pronounced downregulation of cilia genes in those with RSV (Figure 2).

Figure 1.:

Dot plots displaying gene ontology (GO) groups containing significantly upregulated (A) and downregulated (B) differentially expressed genes (DEGs) in bronchiolitis patients relative to control patients. DEGs with adjusted p<0.05 and fold change >1.5 or <0.67 were considered significant. The color of the circle indicates the adjusted p values of the degree of enrichment. The size of the circle represents gene count of the GO group. The y-axis represents the GeneRatio.

Figure 2.:

Heat maps of representative differentially expressed genes for A) Motile Cilium Assembly B) Leukocyte Chemotaxis C) Interferons and D) TNF superfamily. Gene expression is separated by RSV bronchiolitis (n=5), RV bronchiolitis (n=7) and healthy controls (n=4). Red boxes denote positive Z-scores (upregulation of genes) and blue boxes denote negative Z-scores (downregulation of genes).

DISCUSSION

This investigation utilized noninvasive nasal aspiration to gather cytokine protein and RNASeq data in PICU patients with critical bronchiolitis. Our cohort represented diverse viral etiologies for critical bronchiolitis obtained after the COVID-19 pandemic. Our data demonstrated that critically ill patients with bronchiolitis experience a strong type 1 inflammatory response alongside an increase in type 2 cytokines. We identified a differential response in patients with multiple bronchiolitis hospitalizations compared to first-time critical bronchiolitis admissions. Finally, our exploratory RNA expression data in a subgroup of patients with critical bronchiolitis showed an upregulation in genes associated with immune system activation, while also identifying downregulation of ciliary gene function.

Nasal sampling allowed us to recruit a cohort of patients requiring NIPPV or HFNC; a group often excluded from similar studies given inherent risk of instability with removal of support. As bronchiolitis patients are increasingly managed with HFNC and NIPPV, these data show that nasal sampling can be used to study airway inflammation in this evolving critically-ill patient population(18). Nasal airway samples likely represent an optimal method of measuring respiratory system inflammation in these patients with single-organ system disease that are not endotracheally intubated (19).

In the post COVID-19 pandemic era and with increasing use of RSV monoclonal antibody treatment, it is imperative to study PICU-specific cohorts with multiple viral etiologies beyond RSV. RV is present in up to 35% of all virus-positive cases of children hospitalized with bronchiolitis (20). Our data describe a robust antiviral response in patients with critical bronchiolitis including a type I IFN response and type 1 inflammation in both nasal cytokines and exploratory transcriptomic analyses. IFN-α and IFN-β mRNA and protein expression each increased approximately 10-fold. IL-1α, IL-1β, IL-2, IL-10 and TNF-α mRNA and protein expression also significantly increased. Upregulated gene pathways included activation of interferons, type 1 cytokines and TNF-α activation. This response was present in infants with RV as well as those with RSV, as described by others (7, 9). The existing literature on the role of interferon signaling in viral bronchiolitis is mixed, with some evidence associating higher levels with more severe illness (21).

Interestingly, our patients with critical bronchiolitis exhibited a type 2 inflammatory response, which may represent a normal healing or reparative response (22). Vu et al. described a prominent type 2 inflammatory response in 93 infants with RSV bronchiolitis that became more pronounced in the 30 patients admitted to the PICU compared to general care wards (5). Similarly, Legg et al, found that the IL-4/IFN-γ ratio for 9 infants with acute bronchiolitis was elevated in nasal lavage compared to 19 infants with upper respiratory tract infection alone (23). Given the association between early-life viral infections and asthma (24-27), it is tempting to speculate that viral-induced type 2 inflammation has a role in asthma development.

Six of 42 (14.3% [95%CI 6.7 to 27.8%]) bronchiolitis patients in our cohort received systemic steroids. We identified a blunted type 2 inflammatory response in the steroids group with no difference in the type 1 inflammatory or interferon response noted. These data align with findings from Fahy et al. showing that systemic steroids suppress type 2, but not type 1, inflammatory signaling in adult asthmatics, with T1-high patients often experiencing viral infection induced asthma (28). These data also support anecdotal bedside experience suggesting that a subgroup of bronchiolitis patients experience a traditional asthma-like presentation.

Our data showed only a few differences in cytokine levels between RSV bronchiolitis and RV bronchiolitis patients. Specifically, RV patients had decreased IL-12p70 and TNF-α levels compared to RSV patients. A similar lower level of type 1 inflammation in RV compared to RSV was noted by Lee et al., who saw lower levels of INF-γ and TNF-α (10). No significant differences were seen in cytokines associated with type 2 inflammation between RSV and RV in our cohort. In contrast, Hasegawa et al. showed RV bronchiolitis induced higher expression of some type 2 inflammatory cytokines (IL-10 and IL-13) compared to RSV bronchiolitis (11). Finally, our transcriptomic analysis showed little difference in inflammatory signaling between RSV and RV bronchiolitis patients. Given the burden of critical illness seen in viruses beyond RSV, particularly with the advent of RSV vaccinations, a larger, multicenter PICU cohort will be needed, ideally including a wider spread of viral etiologies and across multiple respiratory viral seasons.

We also noted downregulation of gene pathways associated with airway epithelial cell ciliary protein production and function, providing a biologic basis for the impaired mucus clearance clinically experienced in patients with bronchiolitis. Impairment in these pathways may also lead to improper epithelial healing post-infection, a process which has been implicated in airway remodeling and the development of chronic airway disease (29, 30). Future studies will need to further explore differences in repair mechanisms that may contribute to the increased risk of chronic lung disease, repeated infections and admissions.

Eleven of 42 (26.2% [95%CI 15.3 to 41.1%]) patients in the bronchiolitis cohort had a history of prior bronchiolitis admission at enrollment. There is emerging literature that children are at risk for subsequent lower respiratory tract infection (LRTI) after an episode of viral bronchiolitis, and that childhood LRTI is associated with premature adult death from respiratory disease (31, 32). Our data showed that children with multiple bronchiolitis admissions exhibited lower levels of type 1 and 2 inflammatory cytokines compared to children on their first admission for bronchiolitis, providing a biochemical basis for a difference in these two groups. Our cohort with multiple admissions was also noted to have higher rates of chronic comorbidities. Prospective observational studies examining both a) the relationship between chronic comorbid conditions and b) measuring longitudinal changes in cytokine levels will be needed to better understand the group of patients at risk for recurrent hospitalizations.

Our study had limitations. Our cohort was relatively small and from a single center, thus limiting generalizability. Identification of healthy controls, a comparator group not often examined in prior bronchiolitis investigations, was challenging and led to a control group that was older than the experimental group (13, 33, 34). Despite attempts at serial sampling, our patients were able to be sampled only once thereby obviating temporal changes in cytokine levels. Date of first sampling was dependent on the primary medical team assent of clinical safety, leading to variability. A comparison of patients sampled in the first 48-hours of their hospitalization compared to those sampled after 48-hours and saw no differences in cytokine levels (see SDC Figure S1). To obtain a “pure” viral cohort, we excluded patients being treated for secondary bacterial infections; thus, limiting intubated patients in this study meeting PARDS criteria (19). We did not measure viral load or examine the host microbiome, limiting our knowledge of how these factors influenced cytokine levels.

In conclusion, our findings show that children with critical bronchiolitis exhibit both a robust type 1 and partial type 2 inflammatory response to viral infection. Our findings also help to confirm a feasible noninvasive method for obtaining airway samples in critically ill infants requiring non-invasive modes of respiratory support. Larger, multicenter studies will be needed to link inflammatory cytokine levels, their changes over time, and their relationship to more expansive multiomic methodologies to patient outcomes including length of stay, need for mechanical ventilation, mortality and the transition to an asthma phenotype.

Research in Context.

• The host immune response to RSV bronchiolitis is well understood.

• In contrast, the inflammatory profile of bronchiolitis as a general clinical phenotype, including those with rhinovirus (RV) or polyviral bronchiolitis, remains under investigation.

• We therefore sought to identify the inflammatory profile of patients with critical bronchiolitis as a general phenotype, as well as compare cytokines across a variety of clinically relevant subgroups.

At the Bedside.

• Critical bronchiolitis induces both a type 1 and a type 2 cytokine response, the latter typically seen in allergic asthma.

• Patients with critical bronchiolitis exhibit downregulation of gene pathways associated with ciliary function, thereby effecting mucus clearance and expectoration.

• Patients with multiple admissions for bronchiolitis appear to exhibit a blunted type 1 and type 2 immune signaling response when compared to patients being admitted for bronchiolitis for the first time.