Delineating asthma according to inflammation phenotypes with a focus on paucigranulocytic asthma

Abstract

Asthma is characterized by chronic airway inflammation and airway hyper-responsiveness. However, the differences in pathophysiology and phenotypic symptomology make a diagnosis of “asthma” too broad hindering individualized treatment. Four asthmatic inflammatory phenotypes have been identified based on inflammatory cell profiles in sputum: eosinophilic, neutrophilic, paucigranulocytic, and mixed-granulocytic. Paucigranulocytic asthma may be one of the most common phenotypes in stable asthmatic patients, yet it remains much less studied than the other inflammatory phenotypes. Understanding of paucigranulocytic asthma in terms of phenotypic discrimination, distribution, stability, surrogate biomarkers, underlying pathophysiology, clinical characteristics, and current therapies is fragmented, which impedes clinical management of patients. This review brings together existing knowledge and ongoing research about asthma phenotypes, with a focus on paucigranulocytic asthma, in order to present a comprehensive picture that may clarify specific inflammatory phenotypes and thus improve clinical diagnoses and disease management.

Introduction

Asthma is characterized by chronic airway inflammation and manifests clinically as airway hyper-responsiveness (AHR) and reversible airflow limitation. Despite extensive efforts to delineate this heterogeneous disease, universal guideline-based therapy seems to be ineffective in 5% to 10% of asthma sufferers.^[1] Thus, precise prevention and personalized treatment are urgently needed to reduce the disease's significant socioeconomic burden worldwide. One strategy to address this need is to establish clearly distinguishable asthma phenotypes and endotypes that may serve as a foundation for individualized treatments [Figure 1].

Figure 1: Asthma is an umbrella diagnosis and contains various phenotypes and underlying endotypes. Clinical phenotypes refer to observable characteristics of disease. Inflammatory phenotypes are based on profiles of inflammatory cells in peripheral blood or airway samples. EA: Eosinophilic asthma; MGA: Mixed-granulocytic asthma; NA: Neutrophilic asthma; PGA: Paucigranulocytic asthma.

Asthmatic phenotypes are observable disease-related properties formed by the interactions among genetics and environmental and immune factors. Phenotypic clusters can be recognized by demographic, clinical, and/or pathophysiological characteristics.^[2,3] By contrast, endotypes are defined as “subtypes” of a disease with distinct functional or pathophysiological mechanisms but overlapping clinical manifestations.^[4] It is not currently clear whether asthmatic phenotypes and endotypes are directly associated with each other, but focusing on the phenotypic stratification may prove more feasible.

However, appropriately classifying phenotypes is highly problematic due to the absence of uniform classification criteria. The Global Initiative for Asthma recognizes five common clinical phenotypes: allergic, non-allergic, adult-onset, asthma with persistent airflow limitation, and asthma with obesity. By contrast, cluster analysis of 1502 French adults with asthma identified a different set of five clinical phenotypes: early onset allergic asthma, obese asthma, late-onset asthma with severe obstructive syndrome, eosinophilic asthma, and aspirin-sensitive asthma.^[5] Still another phenotypic classification is based on pathogenesis of inflammation and the resulting immune response.^[6] Type 1, type 2, or type 3 cell-mediated immune responses drive the release of cytokines that regulate granulocytic infiltration into the airway. Airway inflammation of asthmatic patients can be assessed based on analysis of bronchial tissue, bronchial–alveolar lavage fluid, or induced sputum. Profiles of inflammatory cells, mainly eosinophils and neutrophils, isolated from induced sputum were used to segregate asthma into four inflammatory phenotypes: eosinophilic asthma (EA), which involves elevated eosinophil counts; neutrophilic asthma (NA), which involves elevated neutrophil counts; paucigranulocytic asthma (PGA), which involves no elevation of either cell counts; and mixed-granulocytic asthma (MGA), which involves concurrent increases in both cell counts. In that scheme, EA and MGA were further classified as type 2 (Th2) cell-mediated asthma (also known as Th2-high asthma), and NA and PGA as non-Th2 asthma (also known as Th2-low asthma). NA and PGA have also been termed non-EA.

Due to the considerable reflection of asthmatic endotypes and some ability to predict response to certain treatments, such as corticosteroids and biologics, the inflammatory phenotypes are widely accepted.^[1,7] Among these inflammatory phenotypes, EA has been most extensively studied, but interest in PGA is growing due to its wide prevalence.^[8] We have compiled this timely review to give researchers and clinicians the most up-to-date information about asthmatic inflammatory phenotypes, with an emphasis on the PGA phenotype, in order to guide future clinical practice.

Establishing the Inflammatory Phenotypes and Clinical Parameters for Diagnosis

In 1999, 34 severe asthmatic patients were divided pathologically according to the presence (+) or absence (−) of eosinophils in the airway. These two subtypes were distinguished by distinct pathological, physiological, and clinical characteristics.^[9] Because this landmark works, many studies have investigated differences between EA and non-EA in terms of their severity and populations affected. Significant differences between the two phenotypes were detected in the following characteristics: cytokines levels and types, serum immunoglobulin E (IgE) levels, blood and airway eosinophils, subepithelial fibrosis, airway mucin gene expression, and clinical characteristics, such as AHR and exacerbations.^[10-14] In 2006, the proportion of eosinophils and neutrophils in asthmatic-induced sputum was used to divide asthma further into the phenotypes EA, NA, MGA, and PGA.^[15] However, consensus is lacking about what cut-off proportions of cell types should be used to distinguish among inflammatory phenotypes. Cut-off values used for clinical diagnosis of EA have ranged between >1.01% and ≥4% of the sputum eosinophil percentage, and values for diagnosis of NA have varied from >40% to ≥76% of sputum neutrophil percentage.^[16-18] The most widely used diagnostic cut-off values in induced sputum are ≥3% for eosinophils and ≥61% for neutrophils.^[19] In recent years, Tanaka et al^[20] further investigated the value of granulocyte counts in spontaneous sputum to discriminate inflammatory phenotypes in 86 adult asthmatic patients and found that spontaneous sputum also helpful to identify airway inflammatory phenotypes in asthmatic patients based on the cut-offs as in induced sputum.

Sputum is the preferred source for determining cellular profiles in order to distinguish asthma inflammatory phenotypes because sampling it is non-invasive, and its composition of inflammatory cells correlates with that from bronchial mucosa tissue, serving as an index of airway inflammation.^[20] Unfortunately, collecting sputum, especially induced sputum, is not feasible in some clinical centers or large population-based studies due to the technical requirements. By comparison, blood samples are relatively easy to obtain from patients of all ages and asthma severities, and protocols for acquiring eosinophil and neutrophil counts are standardized across most clinical centers.^[21] Indeed, there is a theoretical basis for using blood granulocyte counts to identify inflammatory phenotypes, which is the infiltrating granulocytes in the airway of asthmatic patients are derived from the blood circulation.^[22] However, globally accepted cut-off values for the diagnosis of EA or non-EA in blood samples are lacking and highly variable across studies, ranging from ≥150 to ≥400 eosinophils/mm³.^[23,24] Typically, a cut-off of ≥300 eosinophils/mm³ in blood cell count is widely used as the clinical criterion for EA and non-EA, while a cut-off of ≥250 eosinophils/mm³ and ≥5000 neutrophils/mm³ are commonly adopted when differentiating among the four inflammatory phenotypes.^[25-27]

Some studies suggest there is only a weak association between blood eosinophil and neutrophil counts with sputum eosinophil and neutrophil counts.^[25,28-30] A study of 71 severe and 257 non-severe adult asthmatic patients in the Wake Forest Severe Asthma Research Program found that counts of eosinophils or neutrophils in sputum were not accurately predicted by counts of eosinophils or neutrophils in blood, regardless of whether the counts were considered on their own or in combination with other surrogates such as fractional exhaled nitric oxide (FeNO) or serum total IgE.^[28] Likewise, another study of 121 adult asthma patients found that counts of granulocytes in blood predicted inflammatory phenotypes worse than nasal lavage cytometry, based on granulocyte cut-off values determined from induced sputum.^[29] On the other hand, other studies have suggested that counts of eosinophils, but not neutrophils, in blood can provide a practical alternative to predict sputum eosinophilia and are a reliable biomarker for predicting response to biologics against severe asthma.^[31-33] These studies make clear that if eosinophil and neutrophil counts, especially the latter in blood, are to be used for accurate phenotypic diagnosis, more evidence and standardized cut-off values for appropriate tissues must be established. Meanwhile, further studies based on airway tissues will help to more accurately define inflammatory phenotypes including PGA.

Phenotype Distribution

In the absence of standardized cut-off values of cell counts, the distributions of the four inflammatory phenotypes among asthmatic patients have varied greatly across studies. In the French population-based CONSTANCES cohort, cut-offs of ≥250 eosinophils/mm³ and ≥5000 neutrophils/mm³ in blood cell counts resulted in a distribution of 15,019 asthmatic adults into 57% with PGA, 6% NA, 33% EA, and 4% MGA.^[8] However, when the same blood cut-offs were used in a study of 421 female and 299 male adults with stable asthma, the proportion of patients with NA (22.5%) and MGA (8.2%) were more than doubled, while those of PGA (50.1%) and EA (19.2%) were notably reduced.^[34] The same cut-off values applied to samples from 474 adults with asthma from the Epidemiological Study on the Genetics and Environment of Asthma produced results falling between those stated above, with the exception of even greater MGA prevalence (8.9%).^[27]

Applying an eosinophil count of ≥3% and neutrophil count of ≥76% to induced sputum samples, one study divided 508 asthmatic patients from the University Asthma Clinic of Liege into 40% of PGA, 16% of NA, 42% of EA, and 3% of MGA^[35]; and these same values led to a very similar distribution in their other study in 833 asthmatic patients from the same medical center.^[36] However, when both cut-off values were lowered (≥1% eosinophils and ≥61% neutrophils), a study of 93 adult non-smokers with asthma showed frequencies of 31% for PGA, 20% for NA, 41% for EA, and 8% for MGA.^[15] When only the eosinophil cut-off was increased to ≥3% in a study of 232 Chinese asthmatic patients, proportion of patients with the PGA phenotype increased greatly to 52.1%, while that of NA fell to 4.3%.^[37] Dramatically decreasing the neutrophil cut-off value to ≥50% (with ≥2% eosinophils) separated 197 adult asthma patients into 24% PGA, 31% NA, 23% EA, and 22% MGA.^[38] Perhaps the greatest frequency of EA patients (52.5%) was reported in a study of 61 stable asthmatic patients based on cut-offs of ≥2.5% eosinophils and ≥64% neutrophils.^[39] More cut-off values for cell counts from induced sputum reported in recent years are shown in Table 1.^[40-44]

Table 1 - Relative frequencies of asthma inflammatory phenotypes in published studies.

			Frequency
Reference	Study population	Cut-offs (blood or sputum)	PGA (%)	NA (%)	EA (%)	MGA (%)
Tsiavia et al [8]	15,019 adults with asthma	≥250 EOS/mm3, ≥5000 NEU/mm3	57	6	33	4
Nadif et al [26]	381 adults with asthma	≥250 EOS/mm3, ≥5000 NEU/mm3	43.6	12.6	34.6	9.2
Nadif et al [27]	474 adults with asthma	≥250 EOS/mm3, ≥5000 NEU/mm3	48.9	10.6	31.6	8.9
Hsiao et al [34]	720 adults with stable asthma	≥250 EOS/mm3, ≥5000 NEU/mm3	50.1	22.5	19.2	8.2
Simpson et al [15]	93 adult non-smokers with asthma	≥1% EOS, ≥61% NEU	31	20	41	8
Schleich et al [35]	508 adults with asthma	≥3% EOS, ≥76% NEU	40	16	42	3
Demarche et al [36]	833 adults with asthma	≥3% EOS, ≥76% NEU	38	16	42	4
Shi et al [37]	232 adults with asthma	≥3% EOS, ≥61% NEU	52.1	4.3	38.4	5.2
Abdo et al [38]	197 adults with asthma	≥2% EOS, ≥50% NEU	24	31	23	22
Suárez-Cuartín et al [39]	61 patients with stable asthma	≥2.5% EOS, ≥64% NEU	34.4	9.8	52.5	3.3
Ntontsi et al [40]	240 adults with asthma	≥3% EOS, ≥60% NEU	47.9	5.4	40	6.7
Olgac et al [41]	116 non-smoking adult asthmatic patients	≥2% EOS, ≥61% NEU	62.9	7.8	22.4	6.9
Sun et al [42]	120 adults with stable asthma	>2.5% EOS, >61% NEU	41.67	13.33	38.33	6.67
Shi et al [43]	255 adults with asthma	≥3% EOS, ≥61% NEU	52.2	4.3	38.3	5.2
Zhang et al [44]	148 adults with asthma	≥3% EOS, ≥76% NEU	27	18.9	38.5	15.5
Wang et al [45]	29 adults with stable asthma	>3% EOS, >61% NEU	51.7	27.6	17.2	3.5
Wang et al [45]	22 adults with acute asthma	>3% EOS, >61% NEU	0	81.8	0	18.2
Wang et al [45]	49 children with stable asthma	>3% EOS, >61% NEU	49	20.4	28.6	2
Wang et al [45]	28 children with acute asthma	>3% EOS, >61% NEU	7.1	7.2	50	35.7
Rybka-Fraczek et al [46]	40 patients with cough variant asthma	≥3% EOS, ≥61% NEU	52.5	15	32.5	0

EA: Eosinophilic asthma; EOS: Eosinophil; MGA:, Mixed-granulocytic asthma; NA: Neutrophilic asthma; NEU: Neutrophil; PGA: Paucigranulocytic asthma.

Nearly all of the above studies focused on the identification of inflammatory phenotypes in stable adult asthma, while data are scarce for childhood and acute asthma. Applying the induced sputum cut-offs of >3% eosinophils and >61% neutrophils to 49 children and 29 adults with stable asthma, one study determined the most common phenotype as PGA (49% in children and 51.7% in adults), while MGA was the least common phenotype (2% and 3.5%, respectively).^[45] By contrast, among 28 children and 22 adults with acute asthma in the same study, the most common phenotypes were EA in children (50%) and NA in adults (81.8%), while the least common phenotype was PGA for both children (7.1%) and adults (0%). Although the study examined a relatively small sample, the differences in the distribution of inflammatory phenotypes in stable and acute asthma cohorts may have important clinical implications. Real-world studies with larger samples may provide insights into the profile of inflammatory phenotypes throughout the asthma cycle and thus provide valuable references for the differential management of asthma in its stable and acute stages.

Interestingly, the distributions of inflammatory phenotypes in cough-variant asthma have also been investigated. Based on the cut-offs of sputum eosinophils ≥3% and neutrophils ≥61%, 40 patients with confirmed cough variant asthma were classified into 52.5% PGA, 15% NA, and 32.5% EA.^[46] Given the relatively small sample, the absence of MGA in patients with cough-variant asthma should be confirmed in larger studies.

Taken together, the available evidence suggests that PGA and EA are the most common inflammatory phenotypes and MGA is the least common in stable asthmatic cohorts, but this distribution may be biased by cut-off values and asthma stage. They may also depend on ethnicity and geographic location.

Phenotype Stability

Asthmatic inflammatory phenotypes are not immutable and may change with the inflammatory profile over time, even over the course of 1 day.^[47] Some factors influencing inflammatory phenotypic variability include, but are not limited to, age, time of sputum or blood sampling, and corticosteroid treatment. It is well known that corticosteroid influences the eosinophil and neutrophil profiles in peripheral blood and sputum.^[48] Inflammatory phenotypes are purportedly only stable in one-third of asthmatic patients receiving corticosteroids therapy.^[49] A small study of 54 subjects suggested that classification of asthma based on inflammatory phenotype was unstable and fluctuated with time, leading the investigators to conclude that classifying inflammatory phenotypes based on a single induced sputum sample was unreliable.^[50] In another study, inflammatory phenotypes changed in 23.6% of 169 adult patients with varying severities of asthma and in 42.3% of 93 patients with severe asthma after 1-year follow-up, leading the investigators to conclude that asthma diagnosis based on phenotype may be less stable than diagnosis based on physiological variables, such as lung function.^[51] Furthermore, in a retrospective study of 61 asthmatic patients, the proportion of change reached 50.8%, which was hypothesized to be associated with smoking and recent asthma exacerbations.^[39] In addition, monitoring of 175 adult patients >4 weeks found that 47% of initially NA patients and 43% of initially EA patients changed to PGA, while 24% of initially PGA patients changed to NA.^[52] By contrast, 70% of initially PGA patients remained stable. Meanwhile, monitoring of 51 children with severe asthma and 28 children with mild to moderate asthma during 1 year found that 63% of children showed a different inflammatory phenotype during follow-up compared with at baseline, and this variability was not associated with that changes in inhaled corticosteroid (ICS) dose, level of asthma control, or FeNO.^[53] Collectively, these studies show that the variability of the inflammatory phenotype is affected not only by age but also by disease severity. Longitudinal assessment of phenotypic stability and studies with larger samples are needed to understand how these phenotypes behave over time. At the same time, the general instability of the inflammatory phenotype further demonstrates the heterogeneity of asthma and the inadequacy of inflammatory cell profiles alone for inflammatory phenotyping. These considerations challenge us to individualize asthma treatment based on inflammatory phenotypes.

Transcriptomic Definition of Inflammatory Phenotypes

To some extent, phenotypic instability poses a challenge to individualized treatment based on the asthmatic inflammatory phenotypes, which need to be complemented by more defined molecular phenotypes of asthma. Totally, asthma can be classified into molecular phenotypes of Th2, to which EA and MGA belong, and non-Th2, to which NA and PGA belong. Transcriptomic analysis facilitates identification of more specific molecular pathways and tends to find associations between the transcriptome profile and inflammatory phenotypes. A study of the Unbiased Biomarkers for the Prediction of Respiratory Disease Outcomes (U-BIOPRED) cohort revealed three transcriptomic-associated clusters (TAC), TAC1-3 in induced sputum. TAC1 was enriched for interleukin (IL)-13/Th2 and innate lymphoid cell type 2 (ILC2) molecular signatures; TAC2, for damage-associated molecular patterns and upregulation of interferon (IFN) and tumor necrosis factor (TNF) super-family; and TAC3, for metabolic, ubiquitinating enzymes and mitochondrial function. Thus, TAC1 was defined as a Th2 molecular phenotype, while TAC2 and TAC3 were defined as non-Th2 phenotypes. In that cohort, EA, NA, and PGA were present mainly in TAC1, TAC2, and TAC3, respectively, but MGA was composed of TAC1 and TAC2.^[54] Given the different clinical characteristics and distribution of inflammatory phenotypes among the three molecular phenotypes, such a phenotypic approach may complement inflammatory phenotypes to personalize asthma treatment, despite the molecular phenotypic instability remains a concern.^[55]

In a similar methodology, gene expression profiling of induced sputum from 59 stable asthmatic patients and 13 healthy subjects identified three transcriptional asthma phenotypes that could be assigned to EA, NA, and PGA.^[56] Although further analysis found 92% of the differentially expressed genes overlapped among the three inflammatory phenotypes, six signature genes with discriminatory power were identified: CharcotLeydon crystal; carboxypeptidase A3; deoxyribonuclease I-like 3; interleukin-1β; alkaline phosphatase, tissue-non-specific isozyme; and chemokine (C-X-C motif) receptor 2.^[57] Further studies showed that this 6 gene expression signature discriminated inflammatory phenotypes in individuals with more severe asthma and in 132 individuals with stable COPD.^[58,59] In another transcriptomic analysis, six hub genes central to the PGA phenotype were identified from among 449 differentially expressed genes and were validated in a separate dataset: ADCY2, CXCL1, FPRL1, GPR109B, GPR109A, and ADCY3.^[60] Nearly all of these hub genes are involved in the regulation of immune and inflammatory processes, so their further study may help to explain the weak inflammatory response in PGA. In addition, in the U-BIOPRED severe asthma cohort study, the expression of IL18R1, IL1R2, IRAK3, and NLRP3 was lower in PGA than in the other inflammatory phenotypes; conversely, the expression of NLRP1, NLRC4, and NOD2 was higher in NA and MGA than in EA and PGA.^[61]

Other Approaches to Define Inflammatory Phenotypes

There are many limitations associated with inflammatory cell profiles in blood or sputum for inflammatory phenotype discrimination; therefore, the search is underway for more reliable surrogates to discriminate among phenotypes. Other than transcriptomics, proteomics has also been employed to compare inflammatory phenotypes. Protein microarray data from induced sputum belonging to 242 asthmatic patients identified differences in inflammatory protein expression among phenotypes. For example, patients with NA showed the greatest elevation in levels of inflammatory mediators.^[62] Serum levels of periostin, eosinophil-derived neurotoxin (EDN), S100A9, and folliculin have been examined for their ability to discriminate inflammatory phenotypes in 421 adult asthmatic patients: periostin and EDN levels were highest in EA, S100A9 was elevated only in MGA and NA, and folliculin levels were highest in PGA.^[63] Proteins involved in oxidative stress pathways, particularly glutaredoxin isoform 1 (Grx1) and S-glutathionylated proteins (PSSG), have been proposed as potential candidates to differentiate inflammatory phenotypes. Grx1 localizes to the cytosol and can remove glutathione, the main pulmonary anti-oxidant, which restores the function of proteins targeted by PSSG. In one study, levels of Grx1 protein in sputum were significantly higher in EA or PGA patients than in NA patients, while PSSG levels were significantly lower in EA or NA patients than in PGA patients.^[64]

An interesting attempt at identifying a surrogate applied an electronic nose breath analyzer to 52 patients with persistent asthma to detect volatile organic compounds in their exhaled gas. This cross-sectional study identified unique volatile organic compound breath-prints for each inflammatory phenotype, implying that this could be a robust method for phenotypic discrimination.^[65] Indeed, this non-invasive method may be easy to incorporate into everyday clinical use.

Other potential biomarkers may be useful for discriminating asthmatic inflammatory phenotypes, such as microRNAs (miRNAs), exosomes, and some cell surface antibodies. Exosomes and miRNAs can be obtained from induced sputum, and they participate in many cellular processes in many diseases, including asthma.^[66,67] Therefore, they have the potential to serve as biomarkers for inflammatory phenotypes. Additionally, in view of each approach with respective advantage and limitation, the combination of multiple approaches may be even more effective for discrimination of inflammatory phenotypes. For example, the combination of sputum transcriptomics and serum proteomics identified eight clusters of T2-low asthma (NA and PGA) with different clinical and inflammatory characteristics, indicating considerably greater heterogeneity than T2-high asthma. This finding highlights the need to identify specific biomarkers that can differentiate T2-low asthma phenotypes.^[68]

Totally, despite extensive efforts, no specific surrogate approaches are currently available in the clinic for diagnosing or phenotyping PGA. Devoting to find specific surrogate biomarkers to identify airway inflammation phenotypes are very helpful for making the optimal treatment decision.

Pathophysiology of PGA

Inflammatory phenotypes reflect the diversity of pathological mechanisms behind asthma. Although ILC2 is also involved, EA is driven mainly by the Th2 mechanism.^[69] Th2 mechanism is characterized by eosinophil infiltration into airways and upregulation of the Th2-dependent cytokines IL-4, IL-5, and IL-13.^[70] By contrast, NA is classified as a non-T2 asthma which is driven predominantly by non-T2 immunologic pathways and involves Th1 and Th17 cells; IL-6, IL-8, IL-17, and IL-22; and several epithelial cell-derived cytokines.^[71,72] The mechanism driving the non-T2 asthma phenotype PGA remains largely unknown. PGA appears to involve upregulation of genes involved in metabolism and mitochondrial function, yet the high co-expression relationship of the gene signatures and the corresponding protein signatures seem to be absence in PGA.^[54] This apparent paradox suggests that PGA may be driven by alterations in post-translational modifications and metabolic enzyme activity, rather than by classical cellular activation mechanisms. The abnormal expression of genes involved in mitochondrial function may induce mitochondrial dysfunction, but how mitochondrial oxidative stress may drive PGA is unclear. Thus, potential contributions by other airway cells such as macrophages and mast cells should be explored. Indeed, PGA presents with neither obviously increased eosinophilic nor neutrophilic inflammation in the airway, but with elevated counts of macrophages and mast cells in sputum.^[41,73-75] One early study speculated that the absence of increased granulocytes in airways of patients with PGA may be associated with exhaustion of the inflammatory cell pool as a result of past episodes of intense inflammation.^[76] Future studies should explore this hypothesis.

Thus far, the pathogenesis of PGA remains a black box, though animal models have suggested some potential pathways of pathogenesis [Figure 2]. One potential factor is uncoupling of airway obstruction from airway inflammation, which could be due to several mechanisms but largely ascribed to airway structural changes, such as hypertrophy of airway smooth muscle (ASM).^[17] Although most studies have shown different degrees of interactions between airway inflammation and airway obstruction, at least one preclinical study supports the formation of AHR and airway obstruction in the absence of airway inflammation. Disorder of contractile function in ASM can induce AHR in chronic airway diseases.^[77] There are several possible mechanisms that may explain how AHR can occur in the absence of airway inflammation in asthma. Based on neurogenic mechanism involved in the pathophysiology of AHR, the association of nerve growth factor with AHR is among them. In an allergen-sensitized mouse model, nerve growth factor induced an alternate neuronal pathway to control ASM contractility, activating AHR in the absence of airway inflammation.^[78] Consistent with this work, densities of cholinergic nerve innervations in bronchial biopsies are higher in asthma patients than in healthy subjects, which may contribute to AHR independently of eosinophilic inflammation.^[79]

Figure 2: Possible pathophysiology of PGA. The pathological features including eosinophil (−), neutrophil (−), epithelial damage (+), mucus (+), RBM thickening (±), ASM mass (+), extracellular matrix (+), and mast cell (+). Ach: Acetylcholine; ASM: Airway smooth muscle; IL: Interleukin; MCP-1: Monocyte chemoattractant protein 1; MR: Muscarinic receptor; PGA: Paucigranulocytic asthma; RBM: Reticular basement membrane; TGF-β: Transforming growth factor-beta.

In mice, marked increases in oxidized lipids after ozone exposure can also induce AHR independently of airway inflammation, and this AHR does not respond to steroids or cyclooxygenase inhibitors.^[80] Several signaling molecules can also promote AHR in the absence of airway inflammation. For instance, depletion of membrane-bound caveolin-1 in aging mice increases the thickness of the subepithelial layer in the airways.^[81] Meanwhile, the other studies found the function dysregulation of G protein-coupled receptor (GPCR) signaling pathway in ASM of mice, which was induced by the pathway regulators of G protein signaling-2,4,5 (RGS2, RGS4, and RGS5), could mediate AHR without airway inflammation.^[82-84] The mechanism appears to involve enhanced contractility due to GPCR-induced calcium influx and myosin light chain phosphorylation in the ASM.

In mice, overexpression of the genes orosomucoid-like 3 (ORMDL3) and gasdermin B (GSDMB) on chromosome 17q21 is strongly linked to asthma and potentially also to increased AHR without airway inflammation. Increased expression of ORMDL3 and GSDMB induces hypertrophy in ASM and increases subepithelial fibrosis and mucus in airways by elevating levels of mediators such as transforming growth factor beta 1 (TGF-β1) and 5-lipoxygenase.^[85-87] Further analysis of asthma-associated 17q21 genotypes in children found that non-T2 asthma was associated with lower concentrations of sphingomyelins in the blood than T2 asthmatic or healthy children.^[88] Overexpression of ORMDL3 decreases sphingolipid synthesis, further supporting its association with AHR in the pathogenesis of PGA.

Although these animal studies provide insights into the pathobiology of PGA, the findings should be extrapolated to humans with caution. Clinical evidence linking PGA to abnormal ASM in the absence of airway inflammation is still lacking. In addition, ASM alterations in asthma are not the only characteristic of PGA, and the absence of elevated inflammation in sputum does not exclude the possibility of inflammation in the airway submucosa. Thus, direct studies with PGA patients are needed to further understand this asthma phenotype and provide evidence for the concept of uncoupling of AHR and remodeling from airway inflammation in PGA.

Clinical Characteristics of PGA

In 62 adult non-smoking asthma patients, FeNO levels were lower in patients with PGA than in patients with EA, higher than in patients with NA but comparable to levels in patients with MGA.^[89] That study also found the highest levels of AHR in patients with EA, followed by those with MGA, PGA, and NA. By contrast, a retrospective study conducted on 508 adult asthmatics found that those with PGA had lower FeNO and IgE levels than patients with the other three phenotypes, yet patients with PGA tended to have optimal lung function, based on forced expiratory volume in one second (FEV1)% predicted and FEV1/forced vital capacity (FVC).^[35] This may explain why PGA patients typically require lower ICS doses. Small airway dysfunction is a frequent feature of asthma, and it impairs distal lung function, which usually precedes abnormal FEV1 and FEV1/FVC measures. A longitudinal study in 197 adult patients found all small airway dysfunction indicators to be worse in patients with EA or MGA than in those with PGA.^[38] Furthermore, patients with the PGA phenotype also show the highest macrophage counts in sputum among the four inflammation phenotypes, as well as higher counts of lymphocytes and basophils in blood than patients with EA.^[37,41]

Substantial heterogeneity exists even among patients with PGA. Through cluster analysis, three subgroups were identified among 145 patients with PGA: cluster 1 (75.9% of the study population) was termed mild PGA with good lung function; cluster 2 (13.8%), PGA with psychological dysfunction and other allergic disorders; and cluster 3 (10.3%), smoking-associated PGA. During a 1-year period, patients in cluster 3 experienced more numerous severe exacerbations of asthma, they visited the emergency room more, and they were admitted to hospital more than the other two clusters.^[52]

Although it appears that patients with PGA suffer from milder clinical symptoms than patients with other inflammatory phenotypes, they are still at moderate risk of severe or refractory asthma. Indeed, the frequency of severe refractory asthma was 21.7% among 115 patients with PGA, although this was lower than the frequencies among other phenotypes: 41.6% of EA, 43.7% of MGA, and 25% of NA.^[40] However, the asthma was not well controlled in 14.8% of the 115 PGA patients in that study. With this caveat in mind, we cautiously conclude that PGA may be a milder or benign inflammatory phenotype. Further investigation should clarify whether this is due to intrinsic differences or to a therapy-induced shift from another inflammatory phenotype such as EA.

Current Treatments for PGA

There are currently no specific treatments against PGA. In contrast to T2 asthma, non-T2 asthma is marked by an absence of airway and systemic eosinophilia, as well as lack of response to corticosteroids or other inhibitors of T2 inflammation.^[16] Real world research into the step-down use of ICS to treat non-T2 asthma phenotype is rare. One prospective study completely withdrew or reduced ICS dose in two-thirds of non-T2 asthmatic patients, regardless of the asthma control level at baseline. This study concluded that withdrawing or reducing the ICS dose is feasible in most of non-T2 asthmatic patients, including PGA.^[90] The potential explanations for less need of ICS to PGA include symptom was well controlled, primarily driven by intrinsic alteration in structural cell function and the mildness of airway inflammation. However, cautions need to be taken to reach the conclusion that PGA has no response to ICS. First, eosinophil count in sputum, which is the primary indicator of ICS use, is higher in PGA patients than that in healthy subjects.^[36] Furthermore, less frequent high-dose ICS leads to adequate asthma control in a greater proportion of PGA patients than patients with other inflammatory phenotypes.^[40] In addition, inflammatory phenotypes are variable and will behave over time. Thus, more researches around these issues are needed and will guide the use of ICS in patients with PGA.

Biologics are available in many countries or regions, but nearly all of them specifically target T2 asthma phenotypes, especially uncontrollable severe EA patients. Thus, until the underlying pathogenesis can be elucidated, the biologics such as anti-IgE or anti-IL-5 therapies may not be warranted for patients with PGA theoretically.^[16] In addition, although carrying risk of adverse effects and bacterial resistance, macrolides show promise against NA, especially refractory NA, but potential benefits of macrolides for PGA patients have not yet been tested.^[91] However, although patients with PGA may not benefit from anti-inflammatory treatments, based on their symptoms may largely be induced by AHR which was driven by ASM dysfunction, ASM targeted treatments were recommended, such as additional bronchodilators, long-acting antimuscarinic antagonists, and even bronchial thermoplasty in a handful patients with severe, persistent AHR and frequent exacerbations.^[19] Nevertheless, about 15% of patients with PGA remain uncontrolled despite optimal treatment in clinical practice.^[16] For these patients, the potential targeted therapies that focus on airway structural cells may be a direction for future research.

Conclusions

An understanding of the asthmatic inflammatory phenotypes including PGA has important clinical significance. The prevalence of each inflammatory phenotype varies in different study populations, which may reflect the inherent heterogeneity of asthma as well as the lack of consensus granulocyte cut-offs. Finding surrogate or complementary signature biomarkers for the accurate discrimination of asthmatic inflammatory phenotypes is a significant direction for future research and may help clarify the instability of inflammatory phenotypes. Several underlying mechanisms support the uncoupling of AHR and remodeling from airway inflammation in PGA, but the pathogenic pathway remains unclear. In addition, whether PGA only represents a subgroup of asthmatic patients with good response to treatment or a mild asthma inflammatory phenotype needs to be determined. Such studies should cover the entire asthmatic cycle including the acute stage, which is helpful to understand how phenotypes behave over time. Lastly, the identification of new therapeutic targets may improve the individualized treatment of PGA and improve the prognosis of patients with refractory asthma.

Acknowledgments

None.

Funding

This work was supported by a grant from the Sichuan Science and Technology Project (No. 2022YFS0105).

Conflicts of interest

None.