Introduction
With a growing number of monoclonal antibodies (‘biologics’) approved for the treatment of moderate to severe asthma, patients and their providers are having to make an increasing number of therapeutic decisions on the optimal therapy for patients who are eligible for two or more of the six currently approved biologics. Given the high overlap in eligibility for these biologics [
1], and the limited information on distinguishing biomarkers associated with treatment response [
2,
3], research on biomarkers in this class of patients is sorely needed to optimize use of these expensive therapies.
Omalizumab, an anti-immunoglobulin E agent, and mepolizumab, an anti-interleukin-5 agent, were the two biologics first approved for the treatment of moderate-severe asthma and remain two of the most widely used biologics worldwide [
4‐
7]. Patients with eosinophilic asthma, the clinical phenotype mepolizumab is approved for, also often have allergic disease, the phenotype targeted by omalizumab [
1,
8]. Thus, identifying biomarkers associated with omalizumab and/or mepolizumab response will be helpful to clinicians as they seek to optimize treatment of their patient’s asthma.
Proteomics provides one such approach for identifying biomarkers associated with response to omalizumab and mepolizumab. A study of proteins and their interactions can provide a deeper understanding of the function of dynamic biologic systems beyond genomics which is further upstream and immutable [
9,
10]. With the increasing availability of proteomics assays, serum protein biomarkers associated with response to omalizumab and mepolizumab may be helpful. Proteomic profiling of nasal samples, exhaled breath, airway epithelium, urine, and serum have been used to probe various asthma-related phenotypes. These include evaluating the role of T-helper cells in pediatric obese asthma [
11], identifying proteins associated with airway inflammation [
12,
13], association of inflammasome-related proteins with susceptibility to rhinovirus infections in patients with asthma [
14], and plasma protein biomarkers of eosinophilic and neutrophilic asthma phenotypes [
15]. While there are multiple ongoing consortia and projects to study multi-omics of asthma in general with an evolving focus on response to biologic therapies, such as the Severe Asthma Research Program (SARP) [
16], the Unbiased Biomarkers Predictive of Respiratory Disease Outcomes (U-BIOPRED) [
17], and the Precision Medicine Intervention in Severe Asthma (PRISM) study [
18], there is a dearth of studies to identify protein biomarkers of response to any of the respiratory biologics, but prescriptions for these biologics continue to increase exponentially worldwide [
3,
5].
The goal of this study was to identify proteins associated with response to omalizumab and mepolizumab in a single site cohort of patients with asthma. In addition, we sought to identify single protein biomarker(s) that might be helpful in identifying a patient likely to respond to omalizumab or mepolizumab, and to identify biological pathways associated with treatment response.
Discussion
In this study, we measured 1437 proteins in plasma samples collected prior to therapy initiation from 51 patients with moderate to severe asthma who had received omalizumab or mepolizumab and sought to identify protein profiles at baseline associated with reductions in exacerbations over the twelve months of follow up. Higher levels of myostatin and LOXL2 were associated with a poor response to mepolizumab. The dichotomized levels of these proteins differentiated those with a reduction in exacerbations over follow up better than the baseline eosinophil count (≥ 300 vs < 300 cells per mcL). A higher level of LTB4R was associated with worse response to omalizumab while higher levels of CD9 antigen and MUC1 correlated with better response to omalizumab. The level of IgE, omalizumab’s target, was not significantly associated with the change in exacerbations. In GSEA, proteins associated with extracellular matrix organization were the most associated with mepolizumab response. For omalizumab, intracellular transport and metabolic processes involving phosphorylation and phosphate-containing compound processes were the most enriched. We evaluated the protein–protein interactions between the proteins associated with mepolizumab and omalizumab response to gain mechanistic insights into the functions of these proteins and the biological processes associated with response to these therapies. The TNF signaling, NF-kappa B signaling, and B cell receptor signaling pathways were enriched in the mepolizumab response module and included multiple receptor-interacting serine/threonine-protein kinases (RIPK). For omalizumab, similarly to the results of the GSEA, we found an enrichment of kinase pathways including SMAD proteins and mitogen-activated protein kinases (MAPKs), PRKC, and the CK2 protein kinase families, with the high affinity IgE receptor signaling pathway as the pathway with the strongest enrichment. To date, there has been limited information on plasma biomarkers associated with mepolizumab or omalizumab response beyond the currently available biomarkers like the peripheral eosinophil count. Thus, these results give fundamental information for potential biomarkers predictive of response to omalizumab and mepolizumab, two of the most used biologics worldwide.
Myostatin levels positively correlate with levels of the neutrophil chemoattractant, CCL20, a Th17 cells chemoattractant to inflammatory sites [
27]. Glucocorticoids, a main component of treatment of persistent asthma, promote CCL20 expression in keratinocytes [
28] and in the airway epithelium [
29]. In the airway epithelium, CCL20 is associated with mucus hypersecretion in asthma and promotes Th17-mediated neutrophilic airway inflammation, an asthma endotype that is associated with decreased response to glucocorticoids [
29,
30]. CCL20 is also higher in individuals with overweight/obese asthma and may explain the corticosteroid resistance observed in these patients [
29,
31]. Our finding that higher myostatin levels are associated with poorer response to mepolizumab might imply that these patients met criteria for eosinophilic asthma based on their peripheral eosinophil counts but that the underlying mechanism driving their asthma is neutrophilic or mixed granulocytic [
13,
32]. A recent study of nasal proteomics in patients within the endotype of non-eosinophilic asthma (ENDANA) study, replicated in U-BIOPRED, found that many pathways enriched in type 2 ‘T2’ asthma are also enriched in neutrophilic non-T2 asthma suggesting that there are overlapping pathways within patients of eosinophilic and neutrophilic asthma. Multiple other studies of nasal, salivary, and plasma proteomes in children and adults with asthma have shown that protein profiles differ by the presence or absence of asthma and by the degree of asthma control [
13,
33,
34]. In one such study in which findings from the next gen proteomics approach was validated by enzyme-linked immunosorbent assay (ELISA), SERPINA3 was found to be higher in patients with asthma and was particularly higher in patients with an acute exacerbation of their asthma [
34]. In enrichment analyses, the pathways most enriched and associated with mepolizumab response were those pertaining to the extracellular matrix organization and structure. LOXL2, which we found to be associated with poor response to mepolizumab, oxidizes lysine to allysine which crosslinks extracellular matrix (ECM) proteins [
35]. A prior murine study showed that administration of the lysyl oxidase inhibitor β-aminoproprionitrile (BAPN) to TGF-β1-treated mice blocked airway collagen deposition, thereby showing the importance of this enzyme in ECM deposition, a fundamental component of airway remodeling in asthma [
36].
Higher levels of CD9 antigen and MUC1 correlated with better response to omalizumab. CD9 is an extracellular vesicle (EV) associated protein demonstrable in multiple biological fluids, including the serum, plasma, urine, and airway [
37,
38]. EVs are associated with many physiological processes and diseases, including lung diseases like asthma [
37‐
39]. In a recent cohort study, EV miRNA expression profiles distinguished patients with obese type 2-low asthma from non-obese patients with type 2 low asthma [
40], and in another study using the sputum lipidome from 211 patients with asthma and 41 healthy controls from U-BIOPRED [
41], EV secretion by eosinophils and neutrophils was associated with increased lipid in the epithelial lining of the asthma patients. MUC1, which was also associated with better response to omalizumab, is important in allergen response as well as corticosteroid response in asthma patients [
42,
43]. MUC1 knock-out mice are resistant to the anti-inflammatory effects of corticosteroids [
42], and in a study of patients with asthma sensitized to pigeon, MUC1 was upregulated when these patients were exposed to the pigeon allergen [
43]. Though the presence of perennial allergen sensitivity is one of the eligibility criteria for omalizumab, those with high plasma CD9 antigen and MUC1 levels may represent a subgroup of these patients that are continuously exposed to their triggers or who have clinically important perennial allergen sensitivity, and thus with better response to omalizumab. However, we found that LTB4R, also known as BLT1, was associated with worse response to omalizumab. In murine studies, blocking LTB4R led to the inhibition of both early-phases [
44] and late-phases [
45] of allergen-induced airway inflammation. These mixed effects warrant further investigation.
In unadjusted analyses, MALT1 was found to be associated with a poorer response to both mepolizumab and omalizumab. These effects met the FDR-corrected threshold of < 0.20 but not the stringent threshold of < 0.05. In PPI analyses, MALT1 and caspase activation and recruitment domain-9 (CARD9) formed the fifth cluster associated with the mepolizumab response module, and MALT1 was a main component of the enriched NF-kB and the B cell receptor signaling pathways. Multiple MAPKs (1, 3, 8, and 14) were associated with the omalizumab response module. CARD9 is expressed in neutrophils and antigen presenting cells including dendritic cells and macrophages and is involved in immune responses to the environment [
46]. As part of the CARD-BCL10-MALT1 (CBM) complex, CARD activates the MAPK and the NF-kB pathways downstream [
46], and favors a Th1/Th17 polarization [
47,
48]. MAPKs are at the intersection of multiple signaling and inflammatory pathways in the airways and many studies in mice and humans have shown that MAPK inhibitors can restore corticosteroid sensitivity in patients with asthma [
49,
50]. MALT1 levels, on the other hand, have been shown to be higher in children with a higher frequency of asthma exacerbations and to correlate positively with T2 asthma but negatively with T2 low asthma [
51]. Taken together, if replicated in future studies, these results suggest that MALT1 or the CBM complex, may offer important clues towards the identification of a single biomarker that might reflect if an individual who meets criteria for both an anti-IL5 and anti-IgE is likely to respond to mepolizumab and omalizumab or not.
Our results should be interpreted with caution. This was a small single center study. Thus, statistical power was limited, and our results may not be generalizable to other cohorts. For instance, our population is predominantly White and might not be generalizable to more diverse populations. Our results need to be validated in larger and more diverse cohorts. However, many of the findings are biologically plausible and supported by prior asthma literature. Secondly, asthma is a heterogeneous disease and some of our findings may be related to other factors such as the underlying disease severity, other concomitant medications, and/or medication adherence. To limit this variability, we limited our cohort to patients with moderate to severe asthma, on maintenance therapy, and excluded patients with other indications which may or may not be more likely to be compliant with their medications. Third, we evaluated only the baseline levels of these proteins. However, the proteome is dynamic and longitudinal changes to the proteome might be more powerful in identifying a useful biomarker. Fourth, we did not confirm our findings with a direct immunoassay method such as ELISA and some of the associations found may be off target. Furthermore, we did not have a control group and despite our best efforts there is concern for confounding by indication in which these protein associations may be associated with the presence of asthma or not, or with asthma severity, rather than solely with therapeutic response. Nonetheless, this study lays foundational work that can be built open by larger studies and/or studies with more sophisticated study designs.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.