Clinical Implications of Having Reduced Mid Forced Expiratory Flow Rates (FEF25-75), Independently of FEV1, in Adult Patients with Asthma

Introduction FEF25-75 is one of the standard results provided in spirometry reports; however, in adult asthmatics there is limited information on how this physiological measure relates to clinical or biological outcomes independently of the FEV1 or the FEV1/FVC ratio. Purpose To determine the association between Hankinson’s percent-predicted FEF25-75 (FEF25-75%) levels with changes in healthcare utilization, respiratory symptom frequency, and biomarkers of distal airway inflammation. Methods In participants enrolled in the Severe Asthma Research Program 1–2, we compared outcomes across FEF25-75% quartiles. Multivariable analyses were done to avoid confounding by demographic characteristics, FEV1, and the FEV1/FVC ratio. In a sensitivity analysis, we also compared outcomes across participants with FEF25-75% below the lower limit of normal (LLN) and FEV1/FVC above LLN. Results Subjects in the lowest FEF25-75% quartile had greater rates of healthcare utilization and higher exhaled nitric oxide and sputum eosinophils. In multivariable analysis, being in the lowest FEF25-75% quartile remained significantly associated with nocturnal symptoms (OR 3.0 [95%CI 1.3–6.9]), persistent symptoms (OR 3.3 [95%CI 1–11], ICU admission for asthma (3.7 [1.3–10.8]) and blood eosinophil % (0.18 [0.07, 0.29]). In the sensitivity analysis, those with FEF25-75% <LLN had significantly more nocturnal and persistent symptoms, emergency room visits, higher serum eosinophil levels and increased methacholine responsiveness. Conclusions After controlling for demographic variables, FEV1 and FEV1/FVC, a reduced FEF25-75% is independently associated with previous ICU admission, persistent symptoms, nocturnal symptoms, blood eosinophilia and bronchial hyperreactivity. This suggests that in some asthmatics, a reduced FEF25-75% is an independent biomarker for more severe asthma.


Introduction
FEF 25-75 is one of the standard results provided in spirometry reports; however, in adult asthmatics there is limited information on how this physiological measure relates to clinical or biological outcomes independently of the FEV 1 or the FEV 1 /FVC ratio.

Purpose
To determine the association between Hankinson's percent-predicted FEF 25-75 (FEF 25-75 %) levels with changes in healthcare utilization, respiratory symptom frequency, and biomarkers of distal airway inflammation.

Methods
In participants enrolled in the Severe Asthma Research Program 1-2, we compared outcomes across FEF 25-75 % quartiles. Multivariable analyses were done to avoid confounding by demographic characteristics, FEV 1 , and the FEV 1 /FVC ratio. In a sensitivity analysis, we also compared outcomes across participants with FEF 25-75 % below the lower limit of normal (LLN) and FEV 1 /FVC above LLN.

Introduction
Both the diagnosis of asthma and the monitoring of disease severity rely on the use of pulmonary function testing (PFT) data. Though the forced expiratory volume in one second (FEV 1 ) and the FEV 1 to forced vital capacity (FVC) ratio are the most commonly used spirometric measurements to identify both the presence and degree of airflow obstruction, other values such as the forced expiratory flow between 25% and 75% of the FVC (FEF 25-75 ) are also commonly reported [1]. Although variation in the FEF 25-75 was previously thought to reflect changes exclusively in the small airways (< 2mm) [2][3][4], subsequent studies using other physiological or radiological parameters have shown that changes in FEF possibly reflect distal airflows that involve airways that have greater diameters. More importantly, there is limited information about its clinical usefulness, including the fact that current guidelines such as those from the Global Initiative for Asthma (GINA) and the Expert Panel Report 3 of the National Heart, Lung and Blood Institute do not provide specific recommendations for the use of FEF 25-75 in the evaluation or management of asthma [5,6]. In heterogeneous populations, FEF 25-75 is seldom discordant from FEV 1 and FEV 1 /FVC [7]. However, reduced FEF 25-75 in children with asthma has been shown to be associated with increased asthma severity, need for systemic steroid use and more frequent exacerbations in the setting of normal FEV 1 . It is not known whether these results are also applicable to adult asthmatics [8].
To answer this question, we sought to determine whether the percent predicted FEF 25-75 (FEF 25-75 %) is associated with clinical asthma outcomes among participants of the Severe Asthma Research Program (SARP). We hypothesized that having a reduced FEF 25-75 % would be associated with increased asthma morbidity independent of and beyond the severity implied by more traditional markers like FEV 1 %. We further hypothesized that FEF 25-75 % would be associated with biomarkers linked to more distal airway inflammation.

Methods
The study population consisted of participants ages 18 or older from the multi-center SARP study who met criteria for asthma and also had FEF measurements. Asthma diagnosis was based on having either a 12% increase in FEV 1 after short acting bronchodilator or a 20% drop in FEV 1 after inhalation of methacholine (PC 20 25 mg/ml). The SARP 1-2 study has been previously described in detail [9]. Briefly, the study population consisted of subjects recruited at SARP participating academic centers through the use of local advertisement and from their clinics who met eligibility criteria, including being a current nonsmoker with asthma and having less than 5 pack-years of tobacco use. Study participants were classified as having either severe or not severe asthma. According to the American Thoracic Society (ATS) definition, severe asthma was defined as: at least 1 major criteria: a) Use of high-dose inhaled steroids for > 50% of the preceding year or b) continuous or near-continuous oral steroids); and at least 2 minor criteria: a) daily controller medication in addition to inhaled steroids, b) beta agonist required daily or near-daily, c) persistent airway obstruction, d) one or more urgent care visits for asthma per year, e) 3 or more oral corticosteroid bursts/year, f) clinical deterioration with reduction in oral steroid dose, and g) near-fatal asthma event in the past. Non-severe asthmatics included those with moderate (pre bronchodilator FEV 1 < 80% with or without use of inhaled corticosteroids (CS)) or mild (FEV 1 ! 80% with or without use of inhaled CS) asthma.

Clinical Data
After signing informed consent, study participants provided demographic information, smoking history, past medical history and frequency of respiratory symptoms in the 3 months preceding enrollment, including cough, sputum production, chest tightness, nighttime asthma symptoms, wheezing, and shortness of breath. Subjects also completed the Juniper Asthma Quality of Life Questionnaire (AQLQ). As this specific project was performed as secondary analysis of the above data following de-identification, no further patient consent or institutional review board approval was required.

Allergy Skin Test
All participants underwent allergy skin testing for tree mix, grass mix, ragweed, weed mix, dogs, cats, molds, dust mites, and cockroach. To control for validity, diluting fluid and histamine were respectively used as negative and positive controls. Presence of atopy was defined as having at least one skin test reaction of !3 mm and greater than the saline control.

Lung Function testing
Spirometry was done following ATS guidelines[10]. Post bronchodilator FEV 1 was recorded as the maximum bronchodilator change between 4 and 8 puffs of albuterol. Patients with a baseline FEV 1 > 50% and FEV 1 ! 1.5L underwent methacholine challenge, following a 7-dose algorithm of incremental doses from 0.078mg/ml to maximum of 25 mg/ml. A provocation concentration (PC20) of <16 mg/ml was considered positive. This high value was chosen because of the high and prolonged steroid doses in the population. Because of FEV 1 criteria, methacholine challenges were only done to a subset of the subjects. FEF 25-75 % was calculated using Hankinson's regression models for each sex and by race [11]. Exhaled nitric oxide (eNO) was measured online following ATS and ERS standards [12].

Statistical Analysis
The FEF 25-75 % distribution was divided into quartiles to determine how the variability in this measure associates with changes in healthcare utilization, frequency of respiratory symptoms and biomarkers of distal airway inflammation. Healthcare utilization included emergency department (ED) visits, hospitalizations and intensive care unit (ICU) admissions for asthma. The frequency of respiratory symptoms was defined as a binary variable for having or not having symptoms at least twice or more per week. Continuous parametric and non-parametric data were respectively compared across FEF 25-75 % quartiles using one way ANOVA or Kruskal Wallis tests with Bonferroni pairwise comparisons. The Chi2 test was used to compare the distribution of categories.
To evaluate for possible confounders, we performed logistic regression analysis that included the following models: a) univariable, b) adjusted for demographic factors (age, gender, BMI, race), and c) full model adjusted for demographic variables + FEV 1 % predicted and the FEV 1 /FVC ratio. The covariable model selection was based a priori and whether their p value was < 0.1 or the model's estimate changed by ! 10%. Given the high degree of correlation between FEF and FEV 1 or the FEV 1 /FVC ratio, all models were evaluated for collinearity using the variance inflation factor (VIF). Due to the fact that in severe asthmatics, FVC accounts for most of the FEV 1 reversal after bronchodilation due to air trapping [13], a separate multivariate analysis was run with adjustment for FVC rather than FEV 1 (S3 Table); both could not be adjusted for in the same model due to collinearity. We also performed a sensitivity analysis to determine the study outcomes of discordant subjects with low FEF 25-75 % but no airway obstruction, defined as having FEF 25-75 <LLN with FEV 1 /FVC > the lower limit of normal (LLN). All statistics were done using Stata 13.0 (College Park, Tx).

Results
The characteristics of the study population are shown in Table 1, which consisted of 829 participants, of whom the majority was Caucasian, female and overweight. With decreasing FEF 25-75 % quartiles, there was a greater proportion of African American race, larger BMI, longer duration of asthma and a higher proportion of moderate to severe asthmatics. There was also a strong association between FEF 25-75 % quartiles with increased medication use health care utilization, eNO, IgE and sputum eosinophil proportions. Similarly, participants in the lower quartile distribution were more likely to have required an oral corticosteroid burst, to be seen in the emergency room or to be hospitalized for asthma. Those in the lowest FEF 25-75 % quartile had the highest likelihood of having been previously intubated for asthma. With progressive FEF 25-75 % decline, there was also a significant reduction in FVC, FEV 1 and the FEV 1 /FVC ratio as shown in Table 2. Adjustment for FVC rather than FEV 1 did not significantly change the results of the multivariate analysis (S3 Table).

Multivariable analyses
Respiratory symptoms. In the univariable analysis (S1 Table), compared to highest referent FEF 25-75 % quartile (median 92.5 (IQR [80.1, 101.1]) category, the odds of having persistent respiratory symptoms increased in association with lower FEF 25-75 % quartiles and remained largely significant after adjustment for age, BMI, sex, duration of asthma and ever smoking; however, after adjustment for FEV 1 percent predicted and the FEV 1 /FVC ratio, only nocturnal symptoms and persistent symptoms remained significantly associated with the lowest FEF 25-75 % quartile (median 25.6 [IQR 20.4, 32.2]) as shown in Table 3 and Fig 1A. Healthcare utilization. In the univariable analysis, compared to the referent category, the odds of ever requiring hospitalizations, ED visits or ICU care increased in relation to lower FEF 25-75 % quartiles (S1 Table). After adjusting for demographics, these associations were attenuated but remained significant. After adjusting for FEV 1 % and the FEV 1 /FVC ratio, only those in the lowest FEF 25-75 % quartile had a significantly increased odds ratio (OR) of ever having an ICU admission for asthma (Table 3, Fig 1B).
Biomarkers of allergic, airway inflammation and bronchial hyperactivity. In the univariable analysis, lower FEF 25-75 % quartiles were significantly associated with log-transformed eNO, IgE, PC20, blood and sputum eosinophils (S1 Table). After adjustment for demographics, these remained significant with the exception of blood eosinophilia. In the fully adjusted model, the 3 rd FEF 25-75 % quartile was associated with IgE, eNO, PC20, and marginally with sputum eosinophils. In the lowest quartile, only blood eosinophils and PC20 remained significant (Table 3 and Fig 1C), though a trend towards increasing IgE, eNO and sputum eosinophilia remained.
Sensitivity analysis on patients with discordant FEF 25-75 and FEV1/FVC. Compared to asthmatics with an FEF 25-75 >LLN and FEV 1 /FVC >LLN, those with a reduced FEF 25-75 <LLN but FEV 1 /FVC >LLN, had significantly increased nocturnal symptoms and persistent symptoms. They were also more likely to visit the emergency room, to have serum eosinophilia and bronchial hyperreactivity (S2 Table).

Discussion
In this cross sectional study of SARP participants, reduced FEF 25-75 % was associated with increased frequency of respiratory symptoms, greater healthcare utilization and higher levels of biomarkers of distal airway inflammation. Although many of these associations were attenuated by adjusting for FEV 1 % and the FEV 1 /FVC ratio, having a reduced FEF 25-75 % was independently related to more frequent nocturnal and persistent symptoms, ICU admission for asthma, higher eNO, greater bronchial hyperresponsiveness and higher sputum eosinophil percentage. Further, among patients with an FEV 1 /FVC ratio above the lower limit of normal, having an FEF 25-75 below the LLN was associated with increased symptom burden, healthcare utilization, serum eosinophilia and bronchial hyperreactivity. Together, these results show for the first time in adults with asthma, that having a low FEF 25-75 % identifies a group of patients with higher morbidity and elevated biomarkers of distal airway inflammation. According to the European Respiratory Society (ERS) and the American Thoracic Society (ATS) task force for standardization of lung function testing, the FEF 25-75 is defined as the mean forced expiratory flow between the 25% and 75% of the FVC [14,15], which some have interpreted as a quantitative measure of small airways (<2mm) obstruction [4]. Indeed, since the 1970s FEF 25-75 % rates were proposed to be a marker of small airway obstruction and a more sensitive way to detect early stages of obstructive airway disease. However, others have argued that FEF 25-75 % is highly variable and neither sufficiently sensitive nor specific to diagnose obstructive lung disease [16]. Moreover, FEF 25-75 % has not been shown to correlate with other physiologic or histologic measures of distal lung inflammation [16]. Using computed tomography airway morphometric analysis, FEF 25-75 % has been shown to be moderately and inversely correlated with the bronchial wall area (WA) and WA corrected for body surface area, though not exclusively in the small airways [17]. Although it is possible that FEF 25-75 % predominantly reflects flows derived from more distal airways, there is insufficient data to support the concept that variability in this measure is specific to small airway changes. Despite these limitations, FEF 25-75 % continues to be part of the standard spirometry report. More importantly, there are no recommendations as to how reductions in this measure should be taken into consideration for asthma treatment or for risk stratification.
Unlike Quanjer et al [7] who have argued that FEF 25-75 % does not aid clinical decision making, our results support a different conclusion. Although we also found the percentage of discordant cases (low FEF25-75% with a normal FEV1/FVC) is relatively small, FEF was independently associated with several clinical, inflammatory and healthcare outcomes. Further, the study by Quanjer et al was unable to examine these associations, as their study did not include any non-physiological clinical outcomes.
In children, a low FEF 25-75 % has been associated with greater odds for systemic steroids and ED visits, despite having a normal FEV 1 [18]. Similarly, our study found that even when controlling for FEV 1 and the FEV 1 /FVC ratio, asthmatics with lower FEF 25-75 % have greater odds of having been admitted to the ICU and to have persistent respiratory symptoms. In addition, we have shown that lower FEF 25-75 % is independently associated with eNO. As shown by Dweik et al [19], eNO is associated with increased asthma morbidity when levels are above 35 ppb. When taken into consideration with increased persistent and nocturnal symptoms, we speculate that FEF 25-75 % is indeed related to more distal airway inflammation either not fully captured or not yet evident by FEV 1 ; however, further work must be done to evaluate this hypothesis. If proven in prospective investigation, the clinical implications would include both prognosis and identification of an at-risk asthma population for more intensive therapy.
There are significant limitations that need to be considered when evaluating the results from this study. First, given the cross-sectional nature of this study, no causal implications can be made between FEF 25-75 % and asthma severity. This question will need to be answered in the ongoing longitudinal SARP 3 study. Also, reliance on questionnaire responses may contribute to recall bias, though this should be a non-differential bias with regards to the FEF 25-75 % distributions. Additionally, our results may lack external validity, as the SARP study population is enriched with a higher proportion of participants with severe asthma and is not representative of the general adult asthma population. Though this limits the generalization of these findings, it should be noted that SARP still represents a broader spectrum of asthma than most studies, which may help with distinguishing effects.

Conclusions
Independent of FEV 1 and FEV 1 /FVC, FEF 25-75 % predicted identifies a population of adult asthmatics with more severe symptoms, greater health care utilization and elevated biomarkers of distal airway inflammation. Further research is needed to determine if this information can be used clinically to guide treatment decisions or for prognostic evaluation; these questions are currently undergoing longitudinal analysis within SARP 3.
Supporting Information S1