A Nasal Inflammatory Cytokine Signature Is Associated with Early Graft-versus-Host Disease of the Lung after Allogeneic Hematopoietic Cell Transplantation: Proof of Concept

Abstract

Respiratory inflammation in bronchiolitis obliterans syndrome (BOS) after hematopoietic cell transplantation (HCT) is poorly understood. Clinical criteria for early-stage BOS (stage 0p) often capture HCT recipients without BOS. Measuring respiratory tract inflammation may help identify BOS, particularly early BOS. We conducted a prospective observational study in HCT recipients with new-onset BOS (n = 14), BOS stage 0p (n = 10), and recipients without lung impairment with (n = 3) or without (n = 8) chronic graft-versus-host disease and measured nasal inflammation using nasosorption at enrollment and then every 3 mo for 1 y. We divided BOS stage 0p into impairment that did not return to baseline values (preBOS, n = 6), or transient impairment (n = 4). We tested eluted nasal mucosal lining fluid from nasosorption matrices for inflammatory chemokines and cytokines using multiplex magnetic bead immunoassays. We analyzed between-group differences using the Kruskal–Wallis method, adjusting for multiple comparisons. We found increased nasal inflammation in preBOS and therefore directly compared patients with preBOS to those with transient impairment, as this would be of greatest diagnostic relevance. After adjusting for multiple corrections, we found significant increases in growth factors (FGF2, TGF-α, GM-CSF, VEGF), macrophage activation (CCL4, TNF-α, IL-6), neutrophil activation (CXCL2, IL-8), T cell activation (CD40 ligand, IL-2, IL-12p70, IL-15), type 2 inflammation (eotaxin, IL-4, IL-13), type 17 inflammation (IL-17A), dendritic maturation (FLT3 ligand, IL-7), and counterregulatory molecules (PD-L1, IL-1 receptor antagonist, IL-10) in preBOS patients compared to transient impairment. These differences waned over time. In conclusion, a transient multifaceted nasal inflammatory response is associated with preBOS. Our findings require validation in larger longitudinal cohorts.

Introduction

Allogeneic hematopoietic cell transplantation (allo-HCT) is a potentially life-saving procedure for high-risk hematologic malignancies. The long-term efficacy of allo-HCT is limited by relapse of the primary malignancy (1) and by chronic graft-versus-host disease (cGVHD) (2). Bronchiolitis obliterans syndrome (BOS), the most common form of lung cGVHD, is associated with high mortality, particularly when not diagnosed and treated promptly (3). However, BOS is challenging to diagnose early in its course because pulmonary screening is infrequent, and pulmonary symptoms are often subtle (4). BOS stage 0p, defined as a fall in forced expiratory volume in 1 s (FEV₁) of 10% or in forced expiratory flow at 25–75% of forced vital capacity (FVC) (FEF_25–75) of 25% from pre-HCT values, is only associated with subsequent BOS in ∼30% of cases (5). Newer biomarkers are necessary to accurately diagnose early BOS.

The “unified airway hypothesis” postulates that the upper and lower airways are homologous in disease and in health and dates back to observations by Galen, who noted that upper and lower airway symptoms often coexisted (6). Nasosorption is a minimally invasive technique that involves the placement of a thin synthetic absorptive matrix into the nares to obtain highly concentrated mucosal lining fluid (MLF) samples to measure inflammatory cytokines and chemokines without concern for dilution artifacts, as commonly occurs with saline lavage (7). Nasal and bronchial MLF inflammatory signatures show elevations in similar cytokines and chemokines in rhinovirus challenge studies of asthmatic patients, but this technique has not been well studied outside of human viral challenge models (8, 9).

Unlike atopic diseases, no established link exists between nasal inflammation and bronchial inflammation in allo-HCT recipients, although respiratory viral infections, which typically progress from upper to lower airway inflammation, are common after allo-HCT (10) and associated with subsequent BOS (11). Rhinitis symptoms are common among cGVHD patients but have not been specifically associated with BOS (12). Ideally, direct sampling of bronchial MLF (“bronchosorption”) may identify bronchial inflammatory signatures that correlate with BOS, but bronchosorption requires bronchoscopy and sedation and bronchosorption synthetic absorptive matrices (SAMs) are not commercially available in the United States. Measurement of cytokines in bronchoalveolar lavage is limited by variable dilution (13). Therefore, in this proof-of-concept feasibility study, we sought to determine whether we could use nasosorption as a noninvasive technique to detect inflammatory cytokine and chemokine signatures in nasal MLF and discern between allo-HCT recipients with BOS and those without BOS. In addition, we sought to determine whether nasal cytokines and chemokines could distinguish between patients with BOS 0p who had ongoing impairment (preBOS) from those with transient impairment.

Materials and Methods

Cohort selection

We consented and enrolled allo-HCT recipients who were ≥18 of age and who had a new diagnosis of BOS meeting National Institutes of Health (NIH) criteria (14), BOS 0p (5), and patients without pulmonary impairment who had a new diagnosis of cGVHD between 2018 and 2020 to participate in a prospective observational biomarker study to identify markers of early BOS. Patients with BOS or BOS 0p were provided consent and enrolled within 1 wk of diagnosis after a pulmonary evaluation. We screened patients seen in a cGVHD specialty clinic and received consent and enrolled patients who had developed a new diagnosis of cGVHD within 3 mo. We chose this group in order to determine whether analytes were associated with cGVHD in the general sense, or were specific to BOS. Patients with extrapulmonary cGVHD also represent a group at high risk for future BOS, as the vast majority of patients who develop BOS have active extrapulmonary cGVHD (15). Finally, we enrolled allo-HCT recipients without cGVHD as an additional control group. The initial study planned to enroll 20 patients each with BOS and BOS stage 0p, as well as 10 control cGVHD patients, but we halted nasal MLF collection prematurely due to concerns for aerosol generation during the SARS-CoV-2 pandemic. As a result, this study included 14 patients with BOS, 10 with BOS stage 0p, and 11 HCT patients without pulmonary impairment (3 with cGVHD and 8 without cGVHD who were enrolled in 2022). We excluded patients with acute lower respiratory tract infection within 4 wk of enrollment and those who had uncontrolled asthma. We ruled out respiratory viral infection using nucleic acid Ag testing from the nares. All patients underwent chest computerized tomography (CT) at enrollment to exclude lower respiratory infections. The MD Anderson Institutional Review Board approved the study (approval no. 2018-0288).

Study definitions

BOS was diagnosed when patients met all NIH criteria: 1) forced expiratory volume at 1 second (FEV₁) <75% of predicted values with a decline of ≥10% over 2 y; 2) FEV₁/FVC ratio <0.7 or the fifth percentile of predicted values; 3) absence of respiratory infection on chest CT; and 4) evidence of air trapping by pulmonary function test (PFT) or CT, or the presence of cGVHD in another organ (14). BOS 0p was defined as a decline in FEV₁ of ≥10% or forced expiratory mid-flow rates (FEF_25–75) of ≥25% from pre-HCT values measured in the absence of acute pulmonary issues, such as pneumonia (5). We did not require a decline to be present on two consecutive PFTs, because we wanted to obtain biomarkers upon the first evidence of pulmonary impairment, where treatment may first be considered. Patients with BOS 0p who subsequently recovered FEV₁ within 10% of pre-HCT values within 3 mo were considered to have transient impairment, whereas those who did not were considered to have ongoing impairment (preBOS, Fig. 1). PFTs were usually repeated within 4–6 wk of enrollment. Within BOS, we defined steroid responsiveness as stability in FEV₁ without the need for second-line agents for immunosuppression, and steroid refractoriness as the requirement for second-line agents due to progressive impairment on high-dose inhaled and systemic corticosteroids.

FIGURE 1.

Hypothetical HCT recipient who develops pulmonary impairment at 12 mo post-HCT, with pre-HCT FEV₁ of 100% of predicted values.

The green zone represents what would generally be considered the range for normal variation of lung function, whereas the yellow zone represents BOS 0p for this hypothetical patient, and the red zone represents BOS, when other NIH criteria were met. Hypothetical lung function trajectories are shown for transient impairment (green line), preBOS progressing to BOS without effective treatment (yellow to red line), and preBOS progressing to BOS, but halted after treatment (yellow to orange line).

Study procedures

All patients underwent nasosorption (Mucosal Diagnostics/Hunt Developments, Midhurst, U.K.) at enrollment, and then every 3 mo for 1 y, performed as previously described (7). Briefly, we inserted the nasosorption SAM into the lumen of the right nostril and then pressed against the external nares for 60 s. This process was repeated for the left nostril. Samples were immediately stored on ice after collection and stored in a −80°C freezer until the time of analysis. Nasal MLF samples were removed from the −80° C freezer and allowed to thaw on ice for 15 min. SAMs were then inserted into a microcentrifuge tube containing 300 μl of extraction buffer (1× PBS, 1% BSA, 0.05% Tween 20) and vortexed for 30 s. Following this, the SAMs were inserted into a spin-filter mini-column and inserted into microcentrifuge tubes, which were centrifuged for 20 min at 16,000 × g in a minicentrifuge cooled to 4°C. The SAM was removed using sterile forceps.

All BOS patients were treated with high-dose inhaled steroids (16) and systemic immunosuppression as directed by the HCT team, which typically consisted of 6–12 wk of high-dose systemic corticosteroids and additional immunosuppressive therapies (e.g., ruxolitinib or belumosudil) when indicated. BOS 0p patients were treated with high-dose inhaled steroids but not systemic immunosuppression; therapy was discontinued at the discretion of the treating team when impairment was transient.

Evaluation of nasal cytokines and chemokines

We used a 45-plex magnetic Ab-coated bead microplate (human XL cytokine magnetic Luminex performance assay, R&D Biosystems, Minneapolis, MN) to measure cytokines, chemokines, and growth factors on the Luminex MAGPIX multiplex platform (Luminex, Austin, TX).

Statistical analysis

Initial analyses were performed in R version 4.1.0 using packages plyr, reshape2, ggplot2, cowplot, pheatmap, dunn.test, pROC, and ClassComparison (17–24). Pairwise comparisons were performed using a Student t test or Mann–Whitney U test depending on normality of the data, with Benjamini–Hochberg false discovery rate correction for multiple comparisons (25). Normality was assessed using a Shapiro–Wilk test. For more than two independent groups, ANOVA with Tukey post hoc testing or Kruskal–Wallis one-way ANOVA with Dunn post hoc testing was used. For multiplex cytokine measurements between preBOS and transient impairment, given small sample sizes, a Mann–Whitney U test with a beta-uniform mixture model correction for multiple comparisons was employed (22, 25). Longitudinal analyses were conducted using linear mixed models with a fixed effect for time and random intercept for each individual to adjust for intrasubject correlation.

Results

Characteristics of the study cohort

Table I shows differences between patients with BOS, preBOS, and transient impairment and cGVHD controls with no pulmonary impairment. We found no differences in the proportion of patients with atopy or prior respiratory viral infections. As expected, patients with BOS had substantial airflow obstruction upon presentation. PreBOS patients had significantly greater declines in FEV₁ measurements from pre-HCT values compared to those with transient impairment (median decline, 17 versus 13%, p = 0.03), but not in FEF_25–75 (median decline, 37 versus 17%, p = 0.2) or FVC (median decline, 16 versus 12%, p = 0.59). Baseline clinical and spirometric characteristics were not significantly different between steroid-responsive (n = 5) and steroid-refractory BOS (n = 9).

Table I. Characteristics of the study cohort.

Variable	BOS (n = 14)	0p without Recovery (preBOS) (n = 6)	0p with Recovery (transient impairment) (n = 4)	cGVHD Controls (n = 3)
Median age at enrollment (y)	50	59	30	59
Female sex (%)	86	67	0	33
Recipient race (n, %)
White	8 (57)	6 (100)	1 (25)	3 (100)
Black	1 (7)	0	0	0
Hispanic	4 (29)	0	1 (25)	0
Asian	1 (7)	0	2 (50)	0
Underlying malignancy (n, %)
AML/MDS	6 (43)	3 (50)	0	0
ALL	5 (36)	0	3 (75)	1 (33)
Chronic lymphoid malignancies	3 (21)	3 (50)	1 (25)	2 (67)
Donor relationship (%)
Matched unrelated	5 (36)	5 (83)	3 (75)	1 (33)
Matched related	7 (50)	1 (17)	1 (25)	2 (67)
Haploidentical	2 (14)	0	0	0
Cell source (%)
Peripheral blood	10 (71)	5 (83)	1 (25)	2 (66)
Bone marrow	3 (21)	1 (17)	3 (75)	1 (33)
Unknown	1 (7)	0	0	0
CMV status (%)
D−/R−	0	0	0	0
D−/R+	7 (50)	3 (50)	0	1 (33)
D+/R−	0	1 (17)	1 (25)	1 (33)
D+/R+	6 (43)	2 (33)	3 (75)	0
Unknown	1 (7)	0	0	1 (33)
Myeloablative conditioning (%)	12 (86)	4 (67)	4 (100)	2 (66)
Acute GVHD prior to enrollment (%)	8 (57)	5 (83)	3 (75)	2 (66)
Active cGVHD at enrollment (other than lung (%)	11 (79)	4 (67)	1 (25)	3 (100)
Positive history of atopy or seasonal allergies (%)	10 (71)	3 (50)	3 (75)	2 (66)
RVI within 6 mo (%)	6 (43)	1 (17)	1 (25)	1 (33)
Median Pulmonary Function
FEV1 (% predicted)	60	82	85	102
FVC (% predicted)	74	85	83	95
FEV1/FVC ratio	0.62	0.77	0.83	0.81
FEF25–75 (% predicted)	31	69	81	108

ALL, acute lymphoblastic leukemia; AML, acute myeloblastic leukemia; BOS, bronchiolitis obliterans syndrome; cGVHD: chronic graft-versus-host disease; FEF_25–75, forced expiratory flow at 25–75% of forced vital capacity; FEV₁, forced expiratory volume in 1 s; FVC, forced vital capacity; MDS, myelodysplastic syndrome; RVI, respiratory viral infection.

Differences in cytokines and chemokines between BOS, preBOS, transient impairment, and controls

Tables II and III compare median values for chemokines and cytokines between BOS, preBOS, transient impairment, and cGVHD. We observed significant differences in growth factors (FGF2, Flt3 ligand, TGF-α, VEGF, IL-3), T cell stimulation (IL-15, PD-L1), inflammatory cytokines (IL-4, IL-8), and TRAIL when comparing all groups, but these were not significant after correction for multiple testing. Characteristics of allo-HCT controls without cGVHD are reported in Supplemental Table I; we found no differences between controls with or without cGVHD (data not shown). After correction for multiple testing, we found no significant differences in analytes between steroid-responsive and steroid-refractory BOS patients.

Table II. Chemokine, growth factor, and IFN levels by group.

Analyte	BOS (n = 14)	preBOS (n = 6)	Transient Impairment (n = 4)	cGVHD Controls (n = 3)	p Valuea
Chemokines
CCL2 (MCP-1)	101 ± 251	144 ± 165	112 ± 96	40 ± 25	0.24
CCL3 (MIP-1α)	14 ± 3	14 ± 2	17 ± 3	14 ± 2	0.78
CCL4 (MIP-1β)	13 ± 6	16 ± 11	9 ± 2	11 ± 2	0.01 b
CCL5 (RANTES)	5 ± 5	9.03 ± 0	4 ± 2	9 ± 3	0.03
CCL11 (Eotaxin)	14 ± 2	17 ± 20	12 ± 4	14 ± 2	0.02 b
CCL19 (MIP-3β)	18 ± 45	38 ± 34	10 ± 1	10 ± 8	0.06
CCL20 (MIP-3α)	11 ± 30	14 ± 149	11 ± 2	11 ± 4	0.14
CXCL1 (GROα)	1129 ± 1993	3478 ± 5306	307 ± 205	133 ± 298	0.06
CXCL2 (GROβ/MIP-2)	81 ± 156	107 ± 596	26 ± 28	25 ± 36	0.02 b
CXCL10 (IP-10)	386 ± 2457	1706 ± 5433	180 ± 158	85 ± 1606	0.06
Growth factors
EGF	483 ± 616	622 ± 1269	433 ± 540	214 ± 168	0.13
FGF2	10 ± 5	21 ± 11	7 ± 1	7 ± 1	0.02 b
FLT3 ligand	31 ± 42	55 ± 42	21 ± 4	19 ± 10	<0.01 b
G-CSF	44 ± 80	73 ± 842	31 ± 14	25 ± 68	0.30
GM-CSF	15 ± 20	24 ± 28	6 ± 1	7 ± 10	0.02 b
PDGF-AA	905 ± 884	1340 ± 696	747 ± 690	414 ± 368	0.06
PDGF-AB/BB	13 ± 1	13 ± 0	11 ± 3	13 ± 2	0.03
TGF-α	14 ± 13	28 ± 26	9 ± 1	9 ± 3	0.01 b
VEGF	691 ± 1336	1452 ± 502	479 ± 345	365 ± 258	0.02 b
IFNs
IFN-α	4 ± 0	4 ± 0	4 ± 2	4 ± 0	0.23
IFN-β	6 ± 3	9 ± 1	5 ± 1	6 ± 0	0.05
IFN-γ	16 ± 17	21 ± 24	18 ± 8	11 ± 5	0.18

^aBold p values denote significant differences between preBOS and transient impairment.

^bValue is significant after multiple comparison testing using a beta-uniform mixture model.

BOS, bronchiolitis obliterans syndrome; cGVHD, chronic graft-versus-host disease; EGF, epidermal growth factor; FGF, fibroblast growth factor; PDGF, platelet-derived growth factor; VEGF, vascular endothelial growth factor.

Table III. IL and other analyte levels by group.

Analyte	BOS (n = 14)	preBOS (n = 6)	Transient Impairment (n = 4)	cGVHD Controls (n = 3)	p Valuea
ILs
IL-1α	15 ± 11	34 ± 30	31 ± 24	16 ± 4	0.46
IL-1β	32 ± 103	62 ± 878	60 ± 55	25 ± 2	0.38
IL-1Rα	13,335 ± 11,192	19,486 ± 0	12,035 ± 3,160	17,124 ± 5,160	0.02 b
IL-2	25 ± 37	42 ± 47	15 ± 6	12 ± 7	<0.01 b
IL-3	5 ± 3	7 ± 3	4 ± 3	4 ± 0	0.05
IL-4	4 ± 4	8 ± 4	3 ± 1	6 ± 2	0.01 b
IL-5	7 ± 2	8 ± 2	7 ± 0	8 ± 0	0.06
IL-6	20 ± 69	81 ± 66	9 ± 2	32 ± 15	<0.01 b
IL-7	25 ± 24	52 ± 46	9 ± 6	11 ± 11	0.02 b
IL-8/CXCL8	6,297 ± 20,234	19,902 ± 19,276	2,919 ± 2145	1,565 ± 3,373	<0.01 b
IL-10	13 ± 23	29 ± 29	9 ± 2	9 ± 12	0.02 b
IL-12 p70	8 ± 5	11 ± 6	5 ± 4	7 ± 3	0.01 b
IL-13	12 ± 4	16 ± 8	9 ± 3	12 ± 3	0.02 b
IL-15	10 ± 10	17 ± 8	5 ± 3	7 ± 4	<0.01 b
IL-17A	5 ± 1	6 ± 6	3 ± 1	4 ± 1	0.01 b
IL-17E/IL-25	6 ± 2	6 ± 0	6 ± 1	6 ± 0	0.86
IL-33	10 ± 16	20 ± 17	7 ± 2	7 ± 1	0.09
Other analytes
CD40 ligand	10 ± 11	16 ± 20	8 ± 2	9 ± 0	0.01 b
PD-L1/B7-H1	7 ± 5	11 ± 5	4 ± 1	7 ± 2	<0.01 b
Granzyme B	18 ± 39	78 ± 143	17 ± 11	12 ± 2	0.13
TRAIL/TNFSF	203 ± 243	398 ± 519	180 ± 181	20 ± 49	0.18
TNF-α	6 ± 5	7 ± 17	4 ± 1	5 ± 1	0.02 b

^aBold p values denote significant differences between preBOS and transient impairment.

^bValue that is significant after multiple comparison testing using a beta-uniform mixture model.

BOS, bronchiolitis obliterans syndrome; cGVHD, chronic graft-versus-host disease.

Because most analytes were found in the highest concentrations in preBOS, we performed a post hoc analysis comparing preBOS to transient impairment. Analytes that differed between preBOS and transient impairment after multiple testing correction are presented in Fig. 2. After adjusting for multiple corrections, we found significant increases in chemokines (CCL3/MIP-1β, CCL11/eotaxin, CXCL2/GRO-β, TNF-α), growth factors (FGF2, FLT3 ligand, GM-CSF, TGF-α, VEGF), T cell activation (IL-2, IL-12p70, IL-15, PD-L1), type 2 (IL-4, IL-13) and type 17 inflammation (IL-17A), neutrophilic inflammation (IL-6, IL-8), lymphoid maturation (IL-7, CD40 ligand), and counterregulatory cytokines (IL-1 receptor antagonist, IL-10). Similarly, we found no differences between preBOS and BOS, or between BOS and transient impairment or cGVHD controls. Analytes clustered in expected ways (e.g., type 1 inflammation versus type 2 inflammation), as highlighted in the correlation heatmap depicted in Fig. 3. When plotting the Spearman correlations of cytokine values from all patients, clear clusters were noted containing IL-1α, and IL-1β, MIP-1α, and MIP-1β and another a large cluster of cytokines including IL-4, IL-13, IL-12p70, IL-2, IFN-γ, IFN-α, and IL-17A.

FIGURE 2.

Boxplots showing levels of analytes by group (BOS, preBOS, or combined controls) for those analytes that showed differences between preBOS and controls with Dunn post hoc testing.

Median values are represented by the horizontal bar within the box, and the boundaries of the box represent the interquartile range. Whiskers extend from the boundaries of the box (interquartile ranges) to 1.5 × IQR. Solid black dots represent datapoints outside of the whiskers (24).

FIGURE 3.

Heatmap showing correlation and anticorrelation of all measured analytes in all patients.

Clear clusters of type 1, type 2, and type 17 inflammation are evident, indicating the expected association (or lack thereof) between analytes. More red indicates a correlation; more blue indicates an anticorrelation.

Supplemental Table II shows that several analytes were able to discern preBOS from transient impairment, as indicated by areas under the receiver operator characteristic curves. Despite differences in midflow expiratory rates and change in FEV₁ from pre-HCT values between preBOS and transient impairment, inflammatory biomarkers had superior diagnostic performance. We found no differences between patients with BOS, preBOS, transient impairment, or cGVHD at 3 or 6 mo after baseline, suggesting that the presence of nasal inflammation in preBOS is short-lived. We found a decreasing magnitude for several measured analytes over time, but fewer patients provided samples after baseline. We found no statistical differences between patients who were or were not observed at 3 or 6 months, however. Therefore, the lack of differences between groups in longitudinal analyses may be due to diminished inflammation in preBOS over time, or a lack of statistical power due to the smaller sample size.

Discussion

In this study, we show the utility of nasal MLF sampling in post-HCT patients who develop new pulmonary impairment. Nasal MLF analytes were most useful to identify patients with preBOS, who exhibited elevations in fibrogenic growth factors and displayed T cell activation with Th17-mediated and neutrophilic inflammation as compared to those with transient impairment of lung function. Our work suggests that the underlying pathologic mechanisms leading to BOS are a result of complex, dynamic immune mechanisms that can be easily measured in nasal MLF early in the course of BOS, potentially in time to modify clinical outcomes. However, our work requires validation in larger prospective cohorts in which patients are identified early in the course of impairment in a more systematic fashion.

Studies of airway inflammation in BOS after allo-HCT are limited. Murine models of BOS suggest a causal role of IL-13, IL-17, macrophage infiltration, and lymphoid maturation driven in part by IL-21 and CD40 ligand (26–28). HCT recipients with idiopathic pneumonia syndrome or BOS may have higher levels of TNF-α and lower levels of IL-18 in bronchoalveolar lavage compared to healthy controls (29). In patients with noninfectious pulmonary impairment (primarily BOS), plasma TGF-β levels were higher in patients with pulmonary impairment compared to controls (30) Our study offers insights into inflammatory pathways that may be implicated in the pathogenesis of BOS, assuming that the unified airway hypothesis applies. We found a robust nasal immunological response predominantly in patients with preBOS, with signals pointing towards T cell and macrophage activation, and perhaps skewed toward type 2 inflammation. This response included signals showing increased secretion of growth factors (FGF2, TGF-α, GM-CSF, VEGF) macrophage activation (CCL4, TNF-α, IL-6), neutrophil activation (CXCL2, IL-8), T cell activation (CD40 ligand, IL2, IL-12p70, IL-15), dendritic maturation (FLT3 ligand, IL-7), type 2 inflammation (eotaxin, IL-4, IL-13), type 17 inflammation (IL-17A), and counterregulatory cytokines (PD-L1, IL-1 receptor antagonist, IL-10). The complexity of the nasal immune response should temper expectations that a single pathway may be modified to adequately treat BOS and is consistent with the notion that systemic cGVHD is characterized by a multifaceted alloimmune response involving multiple pathways and immune lineages (31). However, although our work does not identify a single pathway that may be targeted to treat BOS, therapies that exert broad anti-inflammatory effects, such as corticosteroids or ruxolitinib, may still be efficacious (32, 33).

Further work is necessary to elucidate patterns of inflammation that may identify clusters of inflammatory phenotypes in early BOS. Nasal MLF had marginal utility in patients who had already developed BOS, whether to distinguish incident BOS from controls, or to distinguish steroid-responsive from steroid-refractory BOS. Furthermore, chemokine and cytokine signals no longer distinguished between groups at 3 or 6 mo, potentially suggesting a short-lived nasal inflammatory response. The transient nature of the inflammatory response may also explain why the degree of inflammation in preBOS was not as markedly elevated as in highly controlled experiments with viral challenge models (9), suggesting that an optimal method to evaluate the utility of nasal MLF would be in scenarios where impairment was detected early, such as with the use of home spirometry (34), and variations in the timing of MLF collection were minimal. However, until we can validate our findings in a larger prospective study, nasal MLF remains an exploratory biomarker and this work should be considered a proof of concept. Nevertheless, the ability of nasal MLF to distinguish preBOS and transient impairment was superior to pulmonary function testing at the time of impairment.

Our use of a nasal inflammatory signature is supported by recent work highlighting the link between upper and lower respiratory tract inflammation. Nasal and bronchial epithelial type 2 gene expression levels are moderately concordant in atopic adults (35) and children (36). Cytokines from nasosorption are concordant with those measured from bronchial MLF (bronchosorption) (8) and sputosorption samples, supporting the notion of upper and lower respiratory concordance (37). However, this signal is transient in studies of respiratory syncytial virus infection, with cytokine levels waning within 7 d (38). Ideally, nasal MLF would be collected within days of detection of any new pulmonary impairment. Our study differs from earlier nasosorption studies because BOS is a disease process that is not typically linked to atopic disease or upper airway inflammation (14), outside of antecedent respiratory viral infection (11). Our results suggest that early pulmonary impairment, or preBOS, is associated with respiratory tract inflammation detectable by noninvasive nasal MLF sampling via nasosorption, but would otherwise not be clinically distinguishable from transient impairment, even with close scrutiny of PFT results. If validated, this would represent a new biomarker to identify early BOS (preBOS).

Our results are noteworthy because, although we expected to see escalating levels of inflammation in steroid-sensitive and steroid-refractory BOS, our findings show that nasal inflammation is attenuated or absent by the time patients present with NIH-defined BOS. In contrast, we found substantially higher inflammation in preBOS compared to patients with transient impairment. This finding suggests that investigators who seek to understand the inflammatory drivers of BOS might shift their focus to preBOS, which may be identified systematically through the use of home spirometry (34, 39–41). In contrast, focusing on NIH-defined BOS may reveal less inflammation and more fibrosis (42). The need to study inflammatory markers earlier in the course of was recently highlighted in the NIH Consensus Development Project regarding highly morbid forms of cGVHD (43). Finally, we note that our patients with preBOS did not progress to BOS when treated with early anti-inflammatory therapy, whereas more than half of patients with NIH-defined BOS required additional therapy beyond inhaled and systemic corticosteroids. Although this finding does not prove utility of early anti-inflammatory therapy in preBOS, it may suggest that early identification of preBOS may open the window for appropriate therapeutic trials to improve BOS outcomes (3, 44). Conversely, the optimal window for efficacy may well be closed for a substantial proportion of patients who present with NIH-defined BOS.

Our study has several strengths and limitations. Among the strengths, our study was prospective and biospecimens were collected immediately upon diagnosis of pulmonary impairment. We show a strong immune response in an understudied subset of allo-HCT recipients with pulmonary impairment using a simple and robust assay. Our study’s primary limitation is its small sample size, particularly HCT recipients without pulmonary impairment, which was due to the inability to continue our work during the SARS-CoV-2 pandemic. Furthermore, our study was designed to focus on BOS, but our primary findings were in those with preBOS, and therefore this study is not optimally designed to detect differences between preBOS and transient impairment. For these reasons, our study should be considered a proof-of-concept feasibility study that is hypothesis generating but not definitive, and our results require validation in a cohort of patients at risk for BOS with nasosorption ideally performed prior to pulmonary impairment. Furthermore, our definition of preBOS needs validation (5, 45). We were unable to perform bronchoscopy of preBOS patients to measure correlation between nasal and bronchial MLF, which would be important to confirm the validity of the nasosorption approach to identify a unified airway signature for BOS. In addition, our sample size was too small to determine whether there were distinct clusters of inflammatory signatures in preBOS patients. Finally, our multiplex assay was prone to batch effect and sensitive to freeze-thaw cycles and storage time before analysis, which limited our ability to evaluate differences between samples collected before COVID-19 restriction and those collected after (HCT patients without cGVHD or pulmonary impairment). Although the most crucial goal should be to identify preBOS among patients with new impairment, ideally this biomarker should also differentiate preBOS from those with no impairment.

In conclusion, we show as a proof of concept that nasal MLF may distinguish patients with BOS stage 0p who developed preBOS versus those who developed transient impairment. Our results show that measurement of nasal inflammation is most informative early in the course of BOS. Our work requires replication and validation in a larger study focused on BOS stage 0p, but it illustrates the possibility of a combined strategy of early detection of impairment, prognostic validation using inflammatory biomarkers early in the course of BOS, and prompt institution of broad anti-inflammatory therapies.

Disclosures

S.P. holds a patent on “Biomarkers and assays to detect chronic graft versus host disease” (U.S. patent 10,571,478 B2). The other authors have no financial conflicts of interest.