Collaboration between explainable artificial intelligence and pulmonologists improves the accuracy of pulmonary function test interpretation.

BACKGROUND: Few studies have investigated the collaborative potential between artificial intelligence (AI) and pulmonologists for diagnosing pulmonary disease. We hypothesised that the collaboration between a pulmonologist and AI with explanations (explainable AI (XAI)) is superior in diagnostic interpretation of pulmonary function tests (PFTs) than the pulmonologist without support. METHODS: The study was conducted in two phases, a monocentre study (phase 1) and a multicentre intervention study (phase 2). Each phase utilised two different sets of 24 PFT reports of patients with a clinically validated gold standard diagnosis. Each PFT was interpreted without (control) and with XAI's suggestions (intervention). Pulmonologists provided a differential diagnosis consisting of a preferential diagnosis and optionally up to three additional diagnoses. The primary end-point compared accuracy of preferential and additional diagnoses between control and intervention. Secondary end-points were the number of diagnoses in differential diagnosis, diagnostic confidence and inter-rater agreement. We also analysed how XAI influenced pulmonologists' decisions. RESULTS: In phase 1 (n=16 pulmonologists), mean preferential and differential diagnostic accuracy significantly increased by 10.4% and 9.4%, respectively, between control and intervention (p<0.001). Improvements were somewhat lower but highly significant (p<0.0001) in phase 2 (5.4% and 8.7%, respectively; n=62 pulmonologists). In both phases, the number of diagnoses in the differential diagnosis did not reduce, but diagnostic confidence and inter-rater agreement significantly increased during intervention. Pulmonologists updated their decisions with XAI's feedback and consistently improved their baseline performance if AI provided correct predictions. CONCLUSION: A collaboration between a pulmonologist and XAI is better at interpreting PFTs than individual pulmonologists reading without XAI support or XAI alone.

Comparative evaluation of PD-L1 expression in cytology imprints, circulating tumour cells and tumour tissue in non-small cell lung cancer patients.

Alternative sources of tumour information need to be explored in patients with non-small cell lung cancer (NSCLC). Here, we compared programmed cell death ligand 1 (PD-L1) expression on cytology imprints and circulating tumour cells (CTCs) with PD-L1 tumour proportion score (TPS) from immunohistochemistry staining of tumour tissue from patients with NSCLC. We evaluated PD-L1 expression using a PD-L1 antibody (28-8) in representative cytology imprints, and tissue samples from the same tumour. We report good agreement rates on PD-L1 positivity (TPS ≥ 1%) and high PD-L1 expression (TPS ≥ 50%). Considering high PD-L1 expression, cytology imprints showed a PPV of 64% and a NPV of 85%. CTCs were detected in 40% of the patients and 80% of them were PD-L1+ . Seven patients with PD-L1 expression of < 1% in tissue samples or cytology imprints had PD-L1+ CTCs. The addition of PD-L1 expression in CTCs to cytology imprints markedly improved the prediction capacity for PD-L1 positivity. A combined analysis of cytological imprints and CTCs provides information on the tumoural PD-L1 status in NSCLC patients, which might be used when no tumor tissue is available.

Minimal clinically important difference for impulse oscillometry in adults with asthma.

BACKGROUND: Impulse oscillometry (IOS) allows an effort-independent evaluation of small airway function in asthma. Unfortunately, well-determined minimal clinically important differences (MCIDs) for IOS measures are lacking. Here, we provide MCIDs for frequently used IOS measures, namely frequency dependence of resistance (FDR) and area of reactance (AX), in patients with asthma. METHODS: We performed IOS at baseline and 1 year later in adult patients with mild-to-severe asthma (n=235). In a two-step approach, we first applied a distribution-based method to statistically determine the MCID. Next, we validated the proposed MCID according to patient-reported outcome measures (PROMs): Asthma Quality of Life Questionnaire (AQLQ), Asthma Control Questionnaire-7 (ACQ-7) and Asthma Control Test (ACT). We used multivariable analyses to investigate the proposed MCIDs as predictors for improvements in PROMs compared with the established MCID of forced expiratory volume in 1 s (FEV1). RESULTS: The proposed MCID was a decline of ≥0.06 kPa·L-1·s-1 and ≥0.65 kPa·L-1 for FDR and AX, respectively. Patients who had changes beyond the MCIDs for both FDR and AX showed greater improvements in all PROMs than those who had not. The mean improvements in PROMs were beyond the established MCIDs for ACQ-7 and AQLQ, and approximated the MCID for ACT. Multivariable analyses demonstrated the MCIDs for both FDR and AX as independent predictors for the MCIDs of all PROMs. The MCID for FDR was a stronger predictor of all PROMs than the MCID for FEV1. CONCLUSIONS: This study provides MCIDs for IOS-derived measures in adult patients with asthma and emphasises that small airway function is a distinguished end-point beyond the conventional measure of FEV1.

IgA+ memory B-cells are significantly increased in patients with asthma and small airway dysfunction.

BACKGROUND: Comprehensive studies investigated the role of T-cells in asthma which led to personalised treatment options targeting severe eosinophilic asthma. However, little is known about the contribution of B-cells to this chronic inflammatory disease. In this study we investigated the contribution of various B-cell populations to specific clinical features in asthma. METHODS: In the All Age Asthma Cohort (ALLIANCE), a subgroup of 154 adult asthma patients and 28 healthy controls were included for B-cell characterisation by flow cytometry. Questionnaires, lung function measurements, blood differential counts and allergy testing of participants were analysed together with comprehensive data on B-cells using association studies and multivariate linear models. RESULTS: Patients with severe asthma showed decreased immature B-cell populations while memory B-cells were significantly increased compared with both mild-moderate asthma patients and healthy controls. Furthermore, increased frequencies of IgA+ memory B-cells were associated with impaired lung function and specifically with parameters indicative for augmented resistance in the peripheral airways. Accordingly, asthma patients with small airway dysfunction (SAD) defined by impulse oscillometry showed increased frequencies of IgA+ memory B-cells, particularly in patients with mild-moderate asthma. Additionally, IgA+ memory B-cells significantly correlated with clinical features of SAD such as exacerbations. CONCLUSIONS: With this study we demonstrate for the first time a significant association of increased IgA+ memory B-cells with asthma and SAD, pointing towards future options for B-cell-directed strategies in preventing and treating asthma.

Longitudinal Impact of Sputum Inflammatory Phenotypes on Small Airway Dysfunction and Disease Outcomes in Asthma.

BACKGROUND: Little is known about the relationship between airway inflammatory phenotypes and some important asthma features such as small airway dysfunction (SAD). OBJECTIVE: To describe the longitudinal impact of airway inflammatory phenotypes on SAD and asthma outcomes. METHODS: We measured eosinophil and neutrophil counts in induced sputum at baseline and 1 year later to stratify 197 adult patients with asthma into 4 inflammatory phenotypes. We conducted a comprehensive assessment of lung function using spirometry, body plethysmography, impulse oscillometry, and inert gas single and multiple breath washouts. We compared lung function, asthma severity, exacerbation frequency, and symptom control between the phenotypes. We studied the longitudinal impact of persistent sputum inflammatory phenotypes and the change of sputum cell counts on lung function. RESULTS: Patients were stratified into eosinophilic (23%, n = 45), neutrophilic (33%, n = 62), mixed granulocytic (22%, n = 43), and paucigranulocytic (24%, n = 47) phenotypes. Patients with eosinophilic and mixed granulocytic asthma had higher rates of airflow obstruction and severe exacerbation as well as poorer symptom control than patients with paucigranulocytic asthma. All SAD measures were worse in patients with eosinophilic and mixed asthma than in those with paucigranulocytic asthma (all P values <.05). Eosinophilic asthma also indicated worse distal airflow obstruction, increased ventilation inhomogeneity (all P values <.05), and higher tendency for severe exacerbation (P = .07) than neutrophilic asthma. Longitudinally, persistent mixed granulocytic asthma was associated with the worst follow-up measures of SAD compared with persistent neutrophilic, persistent paucigranulocytic, or nonpersistent asthma phenotypes. In patients with stable forced expiratory volume in 1 second (FEV1), the mean increase in small airway resistance (R5-20) was greater in patients with persistent mixed granulocytic asthma (+103%) than in patients with persistent neutrophilic (+26%), P = .040, or persistent paucigranulocytic asthma (-41%), P = .028. Multivariate models adjusted for confounders and treatment with inhaled or oral corticosteroids or antieosinophilic biologics indicated that the change of sputum eosinophil rather than neutrophil counts is an independent predictor for the longitudinal change in FEV1, forced expiratory flow at 25% to 75% of forced vital capacity, specific effective airway resistance, residual lung volume, and lung clearance index. CONCLUSIONS: In asthma, airway eosinophilic inflammation is the main driver of lung function impairment and poor disease outcomes, which might also be aggravated by the coexistence of airway neutrophilia to confer a severe mixed granulocytic asthma phenotype. Persistent airway eosinophilia might be associated with dynamic SAD even in patients with stable FEV1.

T2-high asthma phenotypes across lifespan.

RATIONALE: In adults, personalised asthma treatment targets patients with type 2 (T2)-high and eosinophilic asthma phenotypes. It is unclear whether such classification is achievable in children. OBJECTIVES: To define T2-high asthma with easily accessible biomarkers and compare resulting phenotypes across all ages. METHODS: In the multicentre clinical All Age Asthma Cohort (ALLIANCE), 1125 participants (n=776 asthmatics, n=349 controls) were recruited and followed for 2 years (1 year in adults). Extensive clinical characterisation (questionnaires, blood differential count, allergy testing, lung function and sputum induction (in adults)) was performed at baseline and follow-ups. Interleukin (IL)-4, IL-5 and IL-13 were measured after stimulation of whole blood with lipopolysaccharide (LPS) or anti-CD3/CD28. MEASUREMENTS AND MAIN RESULTS: Based on blood eosinophil counts and allergen-specific serum IgE antibodies, patients were categorised into four mutually exclusive phenotypes: "atopy-only", "eosinophils-only", "T2-high" (eosinophilia + atopy) and "T2-low" (neither eosinophilia nor atopy). The T2-high phenotype was found across all ages, even in very young children in whom it persisted to a large degree even after 2 years of follow-up. T2-high asthma in adults was associated with childhood onset, suggesting early origins of this asthma phenotype. In both children and adults, the T2-high phenotype was characterised by excessive production of specific IgE to allergens (p<0.0001) and, from school age onwards, by increased production of IL-5 after anti-CD3/CD28 stimulation of whole blood. CONCLUSIONS: Using easily accessible biomarkers, patients with T2-high asthma can be identified across all ages delineating a distinct phenotype. These patients may benefit from therapy with biologicals even at a younger age.

Impact of imposed social isolation and use of face masks on asthma course and mental health in pediatric and adult patients with recurrent wheeze and asthma.

BACKGROUND: There is currently a dramatic increase in the number of COVID-19 cases worldwide, and further drastic restrictions in our daily life will be necessary to contain this pandemic. The implications of restrictive measures like social-distancing and mouth-nose protection on patients with chronic respiratory diseases have hardly been investigated. METHODS: Our survey, was conducted within the All Age Asthma Cohort (ALLIANCE), a multicenter longitudinal observational study. We assessed the effects of COVID-19 imposed social isolation and use of facial masks, on asthma course and mental health in patients with asthma and wheezing. RESULTS: We observed a high rate of problems associated with using facemasks and a significant reduction in the use of routine medical care. In addition to unsettling impacts, such as an increase in depression symptoms in adults, an astonishing and pleasing effect was striking: preschool children experienced an improvement in disease condition during the lockdown. This improvement can be attributed to a significant reduction in exposure to viral infections. CONCLUSION: Long-term observation of this side effect may help improve our understanding of the influence of viral infections on asthma in early childhood.

The Relevance of Small Airway Dysfunction in Asthma with Nocturnal Symptoms.

RATIONALE: Small airway dysfunction (SAD) is a frequent feature of asthma that has been linked to disease severity and poor symptom control. However, little is known about the role of SAD in nocturnal asthma. OBJECTIVE: To study the association between the severity of SAD and frequency of nocturnal symptoms compared to conventional lung function testing. METHODS: We assessed the frequency of self-reported nocturnal symptoms through the asthma control test. We studied the impact of nocturnal asthma using the Asthma Quality of Life Questionnaire (AQLQ) and the Multidimensional Fatigue Inventory (MFI-20). We assessed the lung function using spirometry, body plethysmography, impulse oscillometry, single and multiple inert gas washout and measured markers of T2-inflammation (blood and sputum eosinophils; fractional exhaled nitric oxide (FeNo)). We stratified the patients according to the presence and frequency of nocturnal asthma. RESULTS: A total of 166 asthma patients were enrolled in the analysis. Eighty-seven patients (52%) reported to have nocturnal symptoms at least once in the last four weeks. The odds ratio of nocturnal asthma correlated with the severity of all non-spirometric measures of SAD, yet neither with airflow obstruction (FEV1 and FEV/FVC) nor with large airway resistance (R20). Patients with frequent nocturnal asthma (n = 29) had a numerical increase of T2 markers and more severe SAD, as indicated by all non-spirometric measures of SAD (all p-values < 0.05), worse overall asthma control, increased fatigue and reduced quality of life (all p-values < 0.01) compared to patients with infrequent nocturnal asthma (n = 58) or patients without nocturnal asthma (n = 79). We identified 63 patients without airflow obstruction, nearly 43% of them (n = 27) had nocturnal asthma. In this subgroup, only markers of air trapping and ventilation heterogeneity were significantly elevated and correlated with the frequency of nocturnal symptoms: LCI (Spearman's coefficient = -0.42, p < 0.001), RV% (-0.32, p = 0.02). CONCLUSION: SAD is closely associated to asthma with nocturnal symptoms. Spirometry might underestimate the broad spectrum of distal lung function impairments in this population of patients.

Raised sputum extracellular DNA confers lung function impairment and poor symptom control in an exacerbation-susceptible phenotype of neutrophilic asthma.

BACKGROUND: Extracellular DNA (e-DNA) and neutrophil extracellular traps (NETs) are linked to asthmatics airway inflammation. However, data demonstrating the characterization of airway inflammation associated with excessive e-DNA production and its impact on asthma outcomes are limited. OBJECTIVE: To characterize the airway inflammation associated with excessive e-DNA production and its association with asthma control, severe exacerbations and pulmonary function, particularly, air trapping and small airway dysfunction. METHODS: We measured e-DNA concentrations in induced sputum from 134 asthma patients and 28 healthy controls. We studied the correlation of e-DNA concentrations with sputum neutrophils, eosinophils and macrophages and the fractional exhaled nitric oxide (FeNO). Lung function was evaluated using spirometry, body plethysmography, impulse oscillometry and inert gas multiple breath washout. We stratified patients with asthma into low-DNA and high-DNA to compare lung function impairments and asthma outcomes. RESULTS: Patients with severe asthma had higher e-DNA concentration (54.2 ± 42.4 ng/µl) than patients with mild-moderate asthma (41.0 ± 44.1 ng/µl) or healthy controls (26.1 ± 16.5 ng/µl), (all p values < 0.05). E-DNA concentrations correlated directly with sputum neutrophils (R = 0.49, p < 0.0001) and negatively with sputum macrophages (R = - 0.36, p < 0.0001), but neither with sputum eosinophils (R = 0.10, p = 0.26), nor with FeNO (R = - 0.10, p = 0.22). We found that 29% of asthma patients (n = 39) had high e-DNA concentrations above the upper 95th percentile value in healthy controls (55.6 ng /µl). High-DNA was associated with broad lung function impairments including: airflow obstruction of the large (FEV1) and small airways (FEF50%, FEF25-75), increased air trapping (RV, RV/TLC), increased small airway resistance (R5-20, sReff), decreased lung elasticity (X5Hz) and increased ventilation heterogeneity (LCI), (all P values < 0.05). We also found that high e-DNA was associated with nearly three-fold greater risk of severe exacerbations (OR 2·93 [95% CI 1.2-7.5]; p = 0·012), worse asthma control test (p = 0.03), worse asthma control questionnaire scores (p = 0.01) and higher doses of inhaled corticosteroids (p = 0.026). CONCLUSION: Increased production of extracellular DNA in the airway characterizes a subset of neutrophilic asthma patients who have broad lung function impairments, poor symptom control and increased risk of severe exacerbations.

Small Airway Dysfunction Links Asthma Severity with Physical Activity and Symptom Control.

BACKGROUND: Little is known about the role of small airway dysfunction (SAD) and its complex relation with asthma control and physical activity (PA). OBJECTIVE: To investigate the interrelations among SAD, risk factors for asthma severity, symptom control, and PA. METHODS: We assessed SAD by impulse oscillometry and other sophisticated lung function measures including inert gas washout in adults with asthma (mild to moderate, n = 140; severe, n = 128) and 69 healthy controls from the All Age Asthma Cohort. We evaluated SAD prevalence and its interrelation with risk factors for asthma severity (older age, obesity, and smoking), type 2 inflammation (sputum and blood eosinophils, fractional exhaled nitric oxide), systemic inflammation (high-sensitivity C-reactive protein), asthma control (AC), and PA (accelerometer for 1 week). We applied a clinical model based on structural equation modeling that integrated causal pathways among these clinical variables. RESULTS: The prevalence of SAD ranged from 75% to 90% in patients with severe asthma and from 53% to 64% in mild to moderate asthma. Severe SAD was associated with poor AC and low PA. Structural equation modeling indicated that age, obesity, obesity-related systemic inflammation, T2 inflammation, and smoking are independent predictors of SAD. Small airway dysfunction was the main determinant factor of AC, which in turn affected PA. Obesity affected AC directly and through its contribution to SAD and low PA. In addition, PA had bidirectional associations with obesity, SAD, and AC. Structural equation modeling also indicated interrelations among distal airflow limitation, air trapping, and ventilation heterogeneity. CONCLUSIONS: Small airway dysfunction is a highly prevalent key feature of asthma that interrelates a spectrum of distal lung function abnormalities with risk factors for asthma severity, asthma control, and physical activity.

Persistent Uncontrolled Asthma: Long-Term Impact on Physical Activity and Body Composition.

RATIONALE: Asthma, obesity and physical activity (PA) are interrelated. However, longitudinal data with objective PA measures and direct assessment of body composition are still lacking. OBJECTIVE: To study the impact of symptom control on PA and body composition. METHODS: In a longitudinal cohort study of the German Center for Lung Research, we assessed the body composition of 233 asthma patients and 84 healthy controls using bioelectrical impedance analysis. PA (ie average daily steps and time of at least moderate activity, steps/min) was measured by accelerometry for one week. Asthma control was assessed by ACT score, ACQ-5 score and history of severe exacerbations. After two years of follow-up, we studied changes in physical activity and body composition in relation to asthma control. RESULTS: Patients with uncontrolled asthma had increased fat mass and decreased muscle mass compared to patients with controlled asthma or healthy controls. Both fat mass and muscle mass correlated better with asthma control than the body mass index (BMI). In multivariate regressions adjusted for age and sex, asthma control and physical activity were independent predictors of body composition (R2 = 0.61, p < 0.001). Persistent uncontrolled asthma patients (n=64) had lower physical activity at both baseline (6614 steps/118 min) and follow-up (6195/115). Despite having stable BMI, they also had significant muscle loss (-1.2%, -0.88 kg, p<0.01) and fat accumulation (+1%, +1.1 kg, p<0.01). By contrast, temporarily uncontrolled or controlled asthma patients had higher physical activity at baseline (8670/156) and follow -up (9058/153) with almost unchanged body composition. CONCLUSION: Persistent uncontrolled asthma is associated with sustained physical inactivity and adverse changes in body composition that might be overlooked by relying solely on BMI. Physical activity is an independent predictor of body composition and reliable long-term marker of symptom control.

Influence of Cell Quality on Inflammatory Biomarkers in COPD Sputum Supernatant.

PURPOSE: We recently introduced a sputum cell quality score to rate how cell morphology, cellular debris and squamous cell contamination influence inflammatory cell identification during microscopic evaluation. However, sputum cell quality is generally not considered for the interpretation of sputum fluid phase biomarkers. Therefore, we compared the soluble protein concentrations between sputum samples with different cell quality. The impact of cell quality was compared to other factors potentially affecting soluble biomarker concentrations. METHODS: A comprehensive sputum dataset from 154 clinically stable COPD patients was used to analyse the differences and the variability of sputum supernatant concentrations for 23 proteins between low, medium, and high sputum cell quality samples. A model was developed and tested to compare the impact of different factors on sputum supernatant protein levels. RESULTS: Mean percentages of sputum macrophages, neutrophils, eosinophils, monocytes and lymphocytes showed no significant differences between low, medium and high cell quality levels. The mean percentage of squamous cells were lower, while total cell count/mL sputum and cell viability were significantly higher in sputum samples with higher cell quality. The concentrations of Interleukin-6, Interleukin-8 and Tumor Necrosis Factor Receptor 2 were significantly increased in sputum samples of higher cell quality. The variability of most protein concentrations declined with increasing cell quality levels. Sixteen proteins showed significantly negative correlations with the percentage of squamous cells. For 14 proteins we observed a positive correlation with cell number/mL sputum. Multiple regression analysis shows that generally less than 30% of the protein variability can be explained by the included factors. CONCLUSION: Sputum cell quality has a significant impact on some soluble biomarker concentrations in sputum supernatant. Sputum samples with low sputum cell quality show a higher variability of fluid phase proteins in comparison to medium and high sputum cell quality levels.

Archaeal hyperthermostable mannitol dehydrogenases: A promising industrial enzymes for d-mannitol synthesis.

Recently, the term healthy lifestyle connected to low-calorie diets, although it is not possible to get rid of added sugars as a source of energy, despite the close relation of added sugars to some diseases such as obesity, diabetes, etc. As a result, the sweetener market has flourished, which has led to increased demand for natural sweeteners such as polyols, including d-mannitol. Various methods have been developed to produce d-mannitol to achieve high productivity and low cost. In particular, metabolic engineering for d-mannitol considers one of the most promising approaches for d-mannitol production on the industrial scale. To date, the chemical process is not ideal for large-scale production because of its multistep mechanism involving hydrogenation and high cost. In this review, we highlight and present a comparative evaluation of the biochemical parameters that affecting d-mannitol synthesis from Thermotoga neapolitana and Thermotoga maritima mannitol dehydrogenase (MtDH) as a potential contribution for d-mannitol bio-synthesis. These species were selected because purified mannitol dehydrogenases from both strains have been reported to produce d-mannitol with no sorbitol formation under temperatures (90-120 °C).

Small airway dysfunction as predictor and marker for clinical response to biological therapy in severe eosinophilic asthma: a longitudinal observational study.

BACKGROUND: Anti-T2 biological therapies have proven to effectively reduce acute exacerbations and daily doses of oral steroids in severe eosinophilic asthma. Despite the remarkable clinical efficacy, there are usually only moderate improvements in airflow limitation, suggesting that other measures of lung function like small airway dysfunction (SAD) might better reflect the clinical response. We aimed to investigate if measures of small airway function would predict and correlate with the clinical response to anti-T2 therapy. METHODS: We studied data of patients who were previously included in the German prospective longitudinal All Age Asthma Cohort (ALLIANCE) that recruits asthma patients of all severity grades and inflammatory phenotypes. The selection criteria for this analysis were adult patients with severe eosinophilic asthma under treatment with anti-T2 biological agents. Asthma control was assessed by asthma control test (ACT) and number of severe exacerbations. Small airway function was assessed by the frequency dependence of resistance (FDR, R5-20)) derived from impulse oscillometry (IOS) and the mean forced expiratory flow between 25 and 75% of the forced vital capacity (FEF25-75). We also studied air trapping (RV and RV/TLC), blood eosinophils and FeNO. Patients were classified into responders and partial or non-responders. Clinical response was defined as at least 50% reduction in annualized severe exacerbations and daily oral steroid doses accompanied with a minimum increase of 3 points in the ACT score. We used a Receiver Operator Characteristic (ROC) to study the capacity of FDR in predicting clinical response compared to other clinical variable like blood eosinophils. We studied the correlation between FDR measures and clinical response, represented by the ACT score and number of exacerbations, using linear regressions. RESULTS: 20 patients were included (mean age, 59 ± 9 years; 60% female; mean body mass index (BMI), 27.6 ± 5.4 kg/m2; mean absolute blood eosinophils, 570 ± 389/µl; mean number of severe exacerbations 12 months prior to initiating the biological therapy, 5.0 ± 3; mean predicted FEV1, 76 ± 21%; mean predicted FDR, 224 ± 140%; mean daily prednisolone dose, 6.4 ± 4.9 mg; mean ACT score, 15 ± 5). Responders had significantly higher baseline FDR compared to partial or non-responders but similar FEV1, FEF25-75, RV and RV/TLC. ROC analysis showed that the combination of FDR and blood eosinophils had the best predictive capacity of the clinical response among all tested clinical markers (FeNO, FEV1, FDR, blood eosinophils) with an AUC of 85% [67-100%], (CI = 0.95, p = 0.01). Linear regressions indicated better associations between improvements in FDR and ACT score (R2 = 0.42, p = 0.001) than with FEV1 and ACT score (R2 = 0.25, p = 0.013). Likewise, we observed better associations between improvements in FDR and reduction of exacerbations (R2 = 0.41, p = 0.001) than with FEV1 (R2 = 0.20, p = 0.025). CONCLUSION: Our data suggest that severe SAD may represent a distinct phenotype of eosinophilic asthma that substantially improves under anti-T2 biological therapy. Measures of small airway function might be useful in selecting appropriate patients qualifying for anti-T2 biological therapy in addition to blood eosinophil count.