Small airway function measured using forced expiratory flow between 25% and 75% of vital capacity and its relationship to airflow limitation in symptomatic ever-smokers: a cross-sectional study

Abstract

Background

Chronic obstructive pulmonary disease (COPD) is diagnosed and its severity graded by traditional spirometric parameters (forced expiratory volume in 1 s (FEV₁)/forced vital capacity (FVC) and FEV₁, respectively) but these parameters are considered insensitive for identifying early pathology. Measures of small airway function, including forced expiratory flow between 25% and 75% of vital capacity (FEF_25-75), may be more valuable in the earliest phases of COPD. This study aimed to determine the prevalence of low FEF_25-75 in ever-smokers with and without airflow limitation (AL) and to determine whether FEF_25-75 relates to AL severity.

Method

A retrospective analysis of lung function data of 1458 ever-smokers suspected clinically of having COPD. Low FEF_25-75 was defined by z-score<−0.8345 and AL was defined by FEV₁/FVC z-scores<−1.645. The severity of AL was evaluated using FEV₁ z-scores. Participants were placed into three groups: normal FEF_25-75/ no AL (normal FEF_25-75/AL−); low FEF_25-75/ no AL (low FEF_25-75/AL−) and low FEF_25-75/ AL (low FEF_25-75/AL+).

Results

Low FEF_25-75 was present in 99.9% of patients with AL, and 50% of those without AL. Patients in the low FEF_25-75/AL− group had lower spirometric measures (including FEV₁ FEF_25-75/FVC and FEV₃/FVC) than those in the normal FEF_25-75/AL− group. FEF_25-75 decreased with AL severity. A logistic regression model demonstrated that in the absence of AL, the presence of low FEF_25-75 was associated with lower FEV₁ and FEV₁/FVC even when smoking history was accounted for.

Conclusions

Low FEF_25-75 is a physiological trait in patients with conventional spirometric AL and likely reflects early evidence of impairment in the small airways when spirometry is within the ‘normal range’. FEF_25-75 likely identifies a group of patients with early evidence of pathological lung damage who warrant careful monitoring and reinforced early intervention to abrogate further lung injury.

WHAT IS ALREADY KNOWN ON THIS TOPIC

• Studies have demonstrated that small airway dysfunction is prevalent in chronic obstructive pulmonary disease (COPD) and can be seen in some susceptible individuals without airflow limitation (AL), but studies using forced expiratory flow between 25% and 75% of vital capacity (FEF_25-75) as measure of small airways function have generally included only a small number of patients with or at risk of developing COPD.

WHAT THIS STUDY ADDS

• Low FEF_25-75 is present in essentially in all of patients with AL and detected in 50% of those without AL, which was associated with lower lung function indices than those with normal spirometry. This highlights that low FEF_25-75 is a physiological feature in patients with AL and likely signifies earlier lung injury in the small airways before classical AL of COPD is present.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

• Low FEF_25-75 without AL might detect a group of patients at risk of developing COPD, where evidence-based preventative strategies could be emphasised and implemented, thereby avoiding progressive lung damage.

Introduction

Chronic obstructive pulmonary disease (COPD) is an inflammatory disease most commonly caused by significant exposure to noxious particles and, pathophysiologically, includes small airway disease and parenchymal destruction.^1–4 COPD is diagnosed based on subjective (respiratory symptoms, history of exposure to risk factors) and objective (physiologically by spirometry) assessments.⁵ The Global Initiative for Obstructive Lung Disease (GOLD), defines airflow limitation (AL) using a fixed forced expiratory volume in 1 s (FEV₁)/forced vital capacity (FVC) ratio and severity defined by FEV₁ % predicted.⁵ Other bodies recommend using the lower limit of normal (LLN) based on z-scores for the ratio to define AL and stratify the severity of the disease as this is thought to be less biased at the extremes of age.^{6 7}

COPD is a slowly progressive disease in most individuals⁸ and FEV₁/FVC and FEV₁ lack the diagnostic sensitivity to identify early lung pathology.^{9 10} As only a proportion of smokers develop COPD,¹¹ identifying individuals with early lung damage who are most at risk of developing overt COPD would enable a focused effort to prevent pathological progression.

The role of small airways in COPD has been explored in several studies.^{3 12–14} Small airways loss precedes the development of emphysema and AL in pathological studies investigated by microcomputed tomographic radiology.^{2 3 12} Further, in a longitudinal study of alpha-1 antitrypsin deficiency (AATD) patients using forced expiratory flow between 25% and 75% of vital capacity (FEF_25-75) as a measure of small airway,¹⁵ a reduced FEF_25-75 without AL was associated with worse health status and a faster subsequent decline in FEV₁ and appeared to precede AL defined by spirometry.¹⁵ This, and other studies, suggest that measures of small airways function (SAF; especially FEF_25-75) may be more sensitive to early damage than traditional spirometric measures.^16–20

We hypothesised that low FEF_25-75, would be ubiquitous in patients with AL, as this has been demonstrated to precede the development of AL.^{21 22} Furthermore, we hypothesised that patients with low FEF_25-75 but without AL would have physiological indicators of the risk of developing AL, even after the correction for potential confounders such as smoking history.

The study had five main aims:

1. To investigate the prevalence of low FEF_25-75 in cigarette smokers with and without AL.

2. To assess whether low FEF_25-75 without AL was associated with lower lung function measurements within the normal range, which might reflect an increased risk for developing AL.

3. To assess the relationships between FEF_25-75 and other spirometric measures.

4. To assess the relationships between FEF_25-75 and AL severity in established COPD.

5. To determine whether the presence of low FEF_25-75 without AL was associated with lower lung function measurements, even after correction for potential confounders.

Methods

Study design and setting

This was a retrospective, cross-sectional study of anonymised data from patients known to have or suspected of having COPD who underwent routine pulmonary function test at University Hospitals Birmingham National Health Service Foundation Trust, UK. The study included data obtained between 1 January 2016 and 30 April 2021 and all patients who had lung function during this period were screened for inclusion.

Eligibility criteria

All participants attending for lung function within the study period with the following included:

1. Symptoms suggestive of COPD (breathlessness and/or a persistent cough).

2. Age 30 years or older.

3. >10 pack-years history of cigarette smoking.

4. Either a confirmed diagnosis or suspected of having COPD by a senior physician.

5. All traditional spirometric measures including FEF_25-75 were reported.

Participants were excluded if they had COPD related to AATD, a history/diagnosis of other chronic lung diseases or significant structural changes in the lung (such as bronchiectasis) defined radiologically. Patients with emphysema identified radiologically; however, were not excluded.

Study measures

Patients’ demographic data were collected. Smoking history included smoking status at the time of testing (ex-smoker or current smoker), pack-years history and years since quitting smoking. The smoking exposure was categorised into light (<20 pack-year), moderate (20–40 pack-years) and heavy (>40 pack-year).²³ Regular medication use was documented.

FEV₁, FVC, FEV₁/FVC, FEF_25-75, FEV in the first 3 s (FEV₃), and FEV₃/FVC were assessed. Corrected FEF_25-75 for lung volume (FEF_25-75/FVC) was also assessed.¹⁷ Lung function assessments used the Ultima PF Pulmonary Lung Function System (Medical Graphics UK, Tewkesbury, UK) and were performed in accordance with national guidelines.²⁴ In this study, predicted values for routine spirometric measures were derived from the European Community for Steel and Coal.²⁵ The z-score for the routine spirometric measures were calculated using the Global Lung Function Initiative 2012 formula.⁷

The z-scores for FEF_25-75 and FEV₁/FVC were used to define abnormality. A cut-off z-score for normality for FEF_25-75 was chosen of −0.8435 as this has shown to predict COPD development.²¹ The LLN (ie, z-score −1.645) was used for FEV₁/FVC to define AL, as recommended in the American Thoracic Society/European Respiratory Society guidelines.^{6 7} Using these thresholds, participants were grouped into three groups: normal FEF_25-75/ no AL (normal FEF_25-75/AL−); low FEF_25-75/ no AL (low FEF_25-75/AL−); and low FEF_25-75/ AL (FEF_25-75/AL+). AL severity was defined using FEV₁ z-score,²⁶ to classify five severity groups.

FEF_25-75 z-score was compared with z-scores of other physiological measures where available.

Statistical analysis

Statistical analysis was performed using IBM SPSS software (V.26). Data were not normally distributed, hence Kruskal-Wallis H tests were used throughout with the median and IQR reported. Where Kruskal-Wallis H tests were significant, a Mann-Whitney U test was conducted. For variables used in group definitions (FEF_25-75 and FEV₁/FVC), no statistical analysis was conducted, except where the definition did not cause the variable to differ. Here, Mann-Whitney U tests was performed to determine the differences. Categorical variables were assessed using χ² or Fisher’s exact test, with the count and percentage reported. The relationship of FEF_25-75 z-score with z-score of other physiological measures and whether smoking behaviours have impact on the relationships were assessed using weight least-square regression. Coefficient of determination (r²) was reported throughout. Curvilinear regression was used to determine the relationship between FEF_25-75 % predicted or FEF_25-75/FVC with % predicted or ratio of other physiological measures, with r² reported throughout.

Logistic regression was performed to identify factors associated with the presence of low FEF_25-75. χ² and Mann-Whitney U test were used to identify relevant univariable risk factors and significant variables were included in the univariate logistic regression and ORs with 95% CIs reported. Significant variables in univariate analyses were included in the subsequent multivariate analysis. Variables, which were associated with multicollinearity (defined by variable inflation factor (VIF) >10) with other variables, were not included in the multivariate logistic regression. A p<0.05 was considered statistically significant throughout. For group comparisons, p values were adjusted using the Benjamini-Hochberg method²⁷ with adjusted p value significance level set at p<0.05. No power calculations were conducted for this pragmatic study.

Patient and public involvement

Patients and/or the public did not take part in the development, conduct, reporting or dissemination of this study.

Results

Participant’s selection

On initial screening, the dataset included 2258 records. After assessing for eligibility, 1458 ever-smokers were included (see figure 1 for a flow chart including reasons for exclusion). These participants were placed into the three groups based on the predefined criteria: normal FEF_25-75/AL− (n=316); low FEF_25-75/AL− (n=335) and low FEF_25-75/AL+ (n=806). One participant did not meet any of the grouping criteria and was therefore excluded from the final analysis.

Figure 1.

Flow chart of the retrospective study. This figure shows the selection of patients according to eligibility criteria. One participant did not meet any of the group definition and was therefore not included in the grouping analysis. AL, airflow limitation; COPD, chronic obstructive pulmonary disease; FEF_25-75, forced expiratory flow between 25% and 75% of vital capacity; FEV₁, forced expiratory volume in 1 s; FVC, forced vital capacity; PPY, pack per year.

Prevalence of low FEF_25-75

All but one participant with AL had low FEF_25-75 (806/807; 99.9%). Of those without AL, 51.4% (335/650) had low FEF_25-75.

Demographics and clinical characteristics

Baseline demographics for the eligible participants and groups are shown in table 1. The average age was higher in low FEF_25-75/AL+ group (median 65 years; IQR 58–73) vs both normal FEF_25-75/AL− group (median 63 years (IQR 54.75–72); p=0.012) and low FEF_25-75/AL− group (median 63 years (IQR 54.75–72); p=0.025). There were no differences in sex across groups. Body mass index (BMI) was lower (p<0.001) in low FEF_25-75/AL+ group than both normal FEF_25-75/AL− group (median BMI 25.67; IQR 21.88–29.82 vs 30.20; IQR 25.34–34.71) and low FEF_25-75/AL− group (median BMI 28.94; IQR 25.33–34.071).

Table 1.

Baseline demographics of the included participants

Variable	Total n=1458	Normal FEF25-75/AL− n=316	Low FEF25-75/AL− n=335	Low FEF25-75/AL+ n=806
Age (years)	64 (56.75–72)	63 (54.75–72)	63 (54.75–72)	65 (58–73)*†
Sex (male: female) n	744: 714	168: 148	150: 184	425: 382
Race (n, %)
Caucasian	1382 (94.8)	286 (90.5)†‡	321 (96.1)	774 (94.8)
Black	22 (1.5)	9 (2.8)†	1 (0.3)	12 (1.5)
Asian	49 (3.4)	19 (6.0)‡	11 (3.3)	19 (2.3)
Others	5 (0.3)	2 (0.6)	1 (0.3)	2 (0.2)
Smoking status (n, %)
Current smokers	842 (57.8)	163 (51.6)‡	197 (59)	482 (59.7)
Ex-smokers	616 (42.2)	153 (48.4)‡	137 (41)	325 (40.3)
Smoking exposure (n, %)
Light	216 (14.8)	73 (23.1)†‡	43 (12.8)	100 (12.4)
Moderate	568 (39)	138 (43.7)	133 (39.7)	297 (36.8)
Heavy	673 (46.2)	105 (33.2)†‡	159 (47.5)	409 (50.7)
Pack-year	40 (25–55)	31 (20–45)†‡	40 (26–55)	41 (28–59)
Years quit§	10 (3–20)	11 (4–24.50)‡	10 (4–20)	8 (3–15)
BMI (kg/m2)	27.32 (23.09–31.95)	30.20 (25.34–34.71)	28.94 (25.33–34.07)	25.67 (21.88–29.82)*†

Data are presented as median and IQR, unless otherwise stated.

In the group comparisons, the significance level for adjusted p value was set at 0.05.

*Significantly different from low FEF_25-75/AL−.

†Significantly different from low FEF_25-75/AL+.

‡Significantly different from normal FEF_25-75/AL−.

§Only assessed in ex-smokers.

AL, airflow limitation; BMI, body mass index; FEF_25-75, forced expiratory flow between 25% and 75% of vital capacity; FEV₁, forced expiratory volume in 1 s; FVC, forced vital capacity.

Participants in normal FEF_25-75/AL− group had generally smoked less (less heavy smokers and a lower pack-year history) compared with low FEF_25-75/AL− group and low FEF_25-75/AL+ group, with no differences between the latter 2.

Expectedly, patients in low FEF_25-75/AL+ group used more COPD-associated medications than those in normal FEF_25-75/AL− group or low FEF_25-75/AL− group, including short-acting beta-2 agonists (SABA), inhaled corticosteroids (ICS)/long-acting beta-2 agonists (LABA) and long-acting muscarinic antagonists (LAMA) (p<0.001 for all). Interestingly, participants in low FEF_25-75/AL− group used more COPD medications (including SABA and ICS/LABA) than normal FEF_25-75/AL− group (p<0.001 for all). Details of the medications used across groups are provided in online supplemental table E1.

Physiological assessment of lung function

Table 2 shows the baseline spirometric measures for the three groups. All spirometric measures were lower in low FEF_25-75/AL− group than normal FEF_25-75/AL− group (p<0.001).

Table 2.

Baseline spirometric measures of the included participants

Variable	Total n=1458	Normal FEF25-75/AL− n=316	Low FEF25-75/AL− n=335	Low FEF25-75/AL+ n=806
FEV1
z-score	−2.09 (−3.16 to −1.11)	−0.44 (−1.00 to 0.20)	−1.67 (−2.26 to −1.18)*	−2.97 (−3.70 to −2.12)*†
% predicted	67.05 (47.65 to 84.12)	93.68 (85.60 to 103.92)	74.05 (64.50 to 82.43)*	50.91 (37.25 to 66.36)*†
FVC
z-score	−0.50 (−1.15 to 0.20)	−0.50 (−1.15 to 0.20)†‡	−1.34 (−2.02 to −0.63)	−1.19 (−2.02 to −0.35)
% Predicted	84.43 (71.96 to 96.75)	93.16 (83.37 to 103.37)†‡	80.44 (70.60 to 91.37)	82.05 (69.22 to 95.38)
FEV1/FVC ratio§
z-score	−1.93 (−3.34 to −0.63)	0.09 (−0.30 to 0.46)	−1.00 (−1.33 to −0.60)*	−3.18 (−4.11 to −2.34)
%	63 (48 to 74)	79 (76 to 83)	71 (68 to 75)*	50 (39 to 59)
FEF25-75§
z-score	−1.96 (−2.80 to −1.01)	−0.16 (−0.49 to 0.25)	−1.37 (−1.67 to −1.11)	−2.72 (−3.2 to −2.18)†
% Predicted	40.56 (21.08 to 67.58)	95.25 (83 to 110.51)	56.02 (47.76 to 63.70)	22.74 (14.80 to 33.76)†
FEF25-75/FVC ratio	48.28 (27.25 to 79.86)	104.80 (90.10 to 122.99)	68.02 (59.16 to 79.86)*	28.73 (19.90 to 40.47)*†
FEV3/FVC ratio	85.44 (74.22 to 92.26)	94.65 (92.10 to 96.83)	90.75 (87.73 to 93.46)*	75.13 (64.70 to 82.67)*†

Data are presented as median and IQR. In the groups’ comparisons, the significance level for adjusted p value was set at 0.05.

*Significantly different from normal FEF_25-75/AL−.

†Significantly different from low FEF_25-75/AL−.

‡Significantly different from low FEF_25-75/AL+.

§Statistical test was only undertaken for differences between groups where a definition did cause the variable to differ.

AL, airflow limitation; FEF25-75, forced expiratory flow between 25% and 75% of vital capacity; FEV1, forced expiratory volume in 1 s; FEV3, FEV in 3 s; FVC, forced vital capacity.

Participants in low FEF_25-75/AL+ group had lower lung function (p<0.001 for all comparisons) than both low FEF_25-75/AL− group and normal FEF_25-75/AL− group. FVC z-score and FVC % predicted did not differ between low FEF_25-75/AL+ group and low FEF_25-75/AL− group. The distribution of FEF_25-75 z-score, FEV₁ z-score, FEV₁/FVC z-score and FVC z-score across groups are shown graphically in figure 2. The distribution of FEF_25-75% predicted, FEF_25-75/FVC, FEV₁ % predicted, FVC % predicted, FEV₁/FVC ratio and FEV₃/FVC ratio across groups are shown in online supplemental figure E1.

Figure 2.

Distribution of spirometric measures across study groups. A box plot demonstrating the distribution of z-scores of spirometric measures across groups. The plot shows median, IQR, minimum and maximum. (A) The distribution of FEF_25-75 z-score across groups. (B) The distribution of FEV₁/FVC z-score across groups. (C) The distribution of FEV₁ z-score across groups. (D) The distribution of FVC z-score across groups. For figures (A, D), statistical test was only done for differences between groups where a definition did cause the variable to differ, and the reported p values are for the Mann-Whitney U test. For figures (B, C), the presented p values are for Mann-Whitney U test, and the Kruskal Wallis tests p values for both figures were<0.001. AL, airflow limitation; FEF_25-75, forced expiratory flow between 25% and 75% of vital capacity; FEV₁, forced expiratory volume in 1 s; FVC, forced vital capacity; NS, not significant.

The relationship of FEF_25-75 with AL severity

Participants with AL were grouped according to AL severity. Table 3 summarises baseline demographics and measures of small airways of these participants. In this cohort, patients with very severe disease were younger than those with lesser severity (p<0.001 for all comparisons). There were no differences between subgroups for sex or ethnicity, although BMI was lower in patients with very severe disease compared with moderately severe patients (median BMI 23.43 (IQR 19.62–28.73) vs 26.99 (IQR 22.85–30.36), p=0.01). Of note, smoking status and pack-year history did not differ across severity groups but those with the most severe disease had stopped smoking later than the other groups.

Table 3.

Baseline demographics and FEF_25-75 across AL severity

Variable	Mild n=177	Moderate n=111	Moderately severe n=120	Severe n=263	Very severe n=135
Age (years)	65 (57–75)	67 (60–75)	67 (58.50–74)	69 (61–73)	59 (53–64)*†‡§
Sex (male: female) n	92: 86	50: 61	72: 48	138: 125	73: 62
Smoking status (n, %)
Current smokers	113 (63.5)	59 (53.2)	72 (60)	159 (60.5)	79 (58.5)
Ex-smokers	65 (36.5)	52 (46.8)	48 (40)	104 (39.5)	56 (41.5)
Pack-year	40 (26.75–55)	41 (25–53)	43 (29–60)	44 (30–62)	38 (23–63)
Years quit	12 (3–21.50)	9 (3–16)	9 (2.25–19.50)	7 (3–14)	5 (2–10)*
BMI (kg/m2)	25.78 (22.96–28.68)	26.17 (21.29–30.42)	26.99 (22.85–30.36)	25.52 (21.92–30.66)	23.43 (19.62–28.73)‡
FEF25-75
z-score	−1.94 (−2.18 to −1.69)	−2.28 (−2.57 to −2.07)*	−2.56 (−2.82 to −2.32)*†	−3.01 (−3.26 to −2.78)*†‡	−3.77 (−4.11 to −3.52)*†‡§
% Predicted	40.50 (33.74 to 48.48)	32.50 (26.49 to 38.56)*	25.76 (21.40 to 29.61)*†	17.60 (13.95 to 21.62)*†‡	10.32 (8.76 to 13.67)*†‡§
FEF25-75/FVC	41.93 (30.95 to 48.58)	38.11 (29.23 to 47.09)	31.61 (24.04 to 40.27)*†	23.28 (18.08 to 31.43)*†‡	15.68 (13.26 to 22.33)*†‡§

Data are presented as median and IQR unless otherwise stated. Severity of AL are stratified using FEV₁ z-score.

In the groups’ comparisons, the significance level for adjusted p value was set at 0.05.

*Significantly different from mild.

†Significantly different from moderate.

‡Significantly different from moderately severe.

§Significantly different from severe.

AL, airflow limitation; BMI, body mass index; FEF25-75, forced expiratory flow between 25% and 75% of vital capacity; FEV3, forced expiratory volume in 3 s; FVC, forced vital capacity.

FEF_25-75 z-score worsened in a stepwise manner as the severity of AL increased (p<0.001; see figure 3). Of note, even in mild AL, FEF_25-75 % predicted was substantially impaired (median 40.50% (IQR 33.74–48.48) and 41.93% (IQR 30.95–48.58) for FEF_25-75 /FVC; see online supplemental figure E2).

Figure 3.

Distribution of FEF_25-75 z-score across AL severity. A box plot demonstrating the distribution of FEF_25-75 z-score across AL severity. The plot shows median, IQR, minimum and maximum. AL severity was assessed using FEV₁ z-score. The presented p values are for Mann-Whitney U test, and the Kruskal Wallis tests p value was <0.001 for all. AL, airflow limitation; FEF_25-75, forced expiratory flow between 25% and 75% of vital capacity.

The relationship of FEF_25-75 with other lung function parameters

Including all participants (n=1458), FEF_25-75 z-score demonstrated a strong relationship to FEV₁ (r²=0.90, p<0.001; see figure 4) and FEV₁/FVC z-score (r²=0.86, p<0.001; see figure 5), but a weaker relationship to FVC z-score (r²=0.17, p<0.001). FEF_25-75 z-score also demonstrated strong relationship to FEV₃/FVC % (r²=0.69, p<0.001; see figure 6).

Figure 4.

FEV₁ z-score plotted against FEF_25-75 z-score. A scatter plot showing the relationship between FEF_25-75 z-score and FEV₁ z-score. The coefficient of determination (r²) for the WLS regression is shown in the figure along with its p value. FEF_25-75, forced expiratory flow between 25% and 75% of vital capacity; FEV₁, forced expired volume in 1 s; WLS, weight-least square.

Figure 5.

FEV₁/FVC z-score plotted against FEF_25-75 z-score. A scatter plot showing the relationship between FEF_25-75 z-score and FEV₁/FVC z-score. The plot is divided according to groups definition. The coefficient of determination (r²) for the WLS regression is shown in the figure along with its p value. AL, airflow limitation; FEF_25-75, forced expiratory flow between 25% and 75% of vital capacity; FEV₁, forced expired volume in 1 s; FVC, forced vital capacity; WLS, weight-least square.

Figure 6.

FEV₃/FVC plotted against FEF_25-75 z-score. A scatter plot showing the relationship between FEF_25-75 z-score and FEV₃/FVC. The coefficient of determination (r²) for the WLS regression is shown in the figure along with its p value. FEF_25-75, forced expiratory flow between 25% and 75% of vital capacity; FEV₁, forced expired volume in 1 s; FVC, forced vital capacity; WLS, weight-least square.

In the multiple regression analyses (n=1458; accounting for pack-year history), FEF_25-75 z-score showed strong relationship to FEV₁ z-score (r²=0.90, p<0.001) and FEV₁/FVC (r²=0.86, p<0.001), and weak relationship to FVC z-score (r²=0.18, p<0.001). Pack-years was not a statistically significant predictor in any of the regression models.

In the multiple regression analysis for ex-smokers (n=616; accounting for years since quitting smoking), FEF_25-75 z-score showed a strong relationship to FEV₁ z-score (r²=0.89, p<0.001) and FEV₁/FVC z-score (r²=0.88, p<0.001), and a weak relationship to FVC z-score (r²=0.22, p<0.001). Years since quitting was a significant predictor in all models (p<0.001 for all models, except in the model for FEV₁ z-score p=0.017).

FEF_25-75 % predicted and FEF_25-75/FVC ratio also showed strong relationship to other spirometric measures, presented in online supplemental table E3. The relationships of FEF_25-75 % predicted and FEF_25-75/FVC with FEV₁/FVC and FEV₁ % predicted are graphically shown in online supplemental figures E3 and E4.

The association of the presence of low FEF_25-75 with low lung function measurements

A regression model was built to assess whether the presence of low FEF_25-75 without AL was associated with lower lung function measurements (see table 4). In the univariate analysis, pack-years, sex, FEV₁ z-score, FVC z-score and FEV₁/FVC z-score were significant factors related to the presence of low FEF_25-75. All significant variables were included in the multivariate analysis except FVC z-score because of multicollinearity with other spirometric measures (VIF=30.94). The multivariate analysis demonstrated that females had a 33.22 times higher OR of having low FEF_25-75 compared with males (95% CI, 8.19 to 134.72). The multivariate analysis also showed that the presence of low FEF_25-75 was associated with a lower FEV₁ z-score and FEV₁/FVC z-score even when in the normal range. Of the significant factors in univariate analysis, pack-years was no longer significant in the multivariate analysis.

Table 4.

Logistic regression of the association of the presence of low FEF_25-75 with low lung function measurement in participants without AL

Variable	Univariate			Multivariate*
	OR	95% CI	P value	OR	95% CI	P value
Age	1.004	0.991 to 1.018	0.55
Pack-years	1.009	1.003 to 1.015	0.002	0.988	0.971 to 1.005	0.168
Smoking status†
Current smokers	1.340	0.983 to 1.827	0.064
Sex‡
Female	1.383	1.016 to 1.883	0.039	33.225	8.194 to 134.723	<0.001
FEV1 z-score	0.136	0.100 to 0.185	<0.001	0.001	0.00008 to 0.005	<0.001
FEV1/FVC z-score	0.043	0.027 to 0.068	<0.001	0.00001	0.000001 to 0.0003	<0.001
FVC z-score	0.449	0.377 to 0.536	<0.001

This tables demonstrate the logistic regression of the association of the presence of low FEF_25-75 with low lung function measurements in participants without AL (n=651 (those normal FEF_25-75 n=316 vs those with low FEF_25-75 n=335)).

Low FEF_25-75 was defined by z-score<−0.8435.

Statistically significant p values are written in bold.

*The multivariate regression model showed a Nagelkerke R²=0.942 and Hosmer-Lemeshow p value=0.999.

†The reference category was ex-smokers.

‡The reference category was male.

AL, airflow limitation; BMI, body mass index; FEF25-75, forced expiratory flow between 25% and 75% of vital capacity; FEV1, forced expiratory volume in 1 s; FVC, forced vital capacity.

Discussion

This cross-sectional study of commonly measure of SAF (FEF_25-75) in smokers suspected of having COPD highlights four important points.

First, low FEF_25-75 (considered indicative of impairment in the small airways) is a constant feature of those who have developed AL, with and without correction for FVC.

Second, there was a significant reduction in FEF_25-75 even in mild AL, suggesting a substantial disruption of SAF prior to crossing the AL diagnostic criteria. Indeed, once AL is established, there is a strong association between FEF_25-75 z-score across AL severity.

Third, evidence of low FEF_25-75 is common (51.4%) in symptomatic ever-smokers even without AL and is associated with lower lung function parameters (even while in the normal range) compared with those with normal FEF_25-75 and normal FEV₁/FVC. This suggests that even when routine spirometry appears ‘normal’, those with low FEF_25-75 may have physiological evidence suggesting decline compared with health. This group of patients likely have early lung injury reflecting small airway impairment. Our data support the notion that such patients may form a cohort that would benefit from close monitoring, to ascertain progression potentially leading to COPD and support to mitigate such an outcome.

Fourth, the relationship between FEF_25-75 and FEV₁ and FEV₁/FVC is maintained even following adjustment for smoking history, indicating it is independent of cigarette load. Further, the logistic regression demonstrated that the presence of low FEF_25-75 was associated with lower FEV₁ and FEV₁/FVC, after correcting for smoking status. This suggests there are a group of smokers who are pathophysiologically different, consistent with a ‘susceptible’ cohort. Further study is needed to understand the mechanisms underpinning this potential susceptibility.

In the regression model, sex was related to low FEF_25-75 in the absence of AL, with females 33 times more likely to have low FEF_25-75, although with a wide 95% CI. In the AATD study by Stockley et al there was also a higher proportion of females with low FEF_25-75 than males compared with those with normal spirometry and AL.¹⁵ This study and the AATD study highlight that females have a greater likelihood of a low FEF_25-75 in the absence of AL. Given that females with COPD have greater small airway impairments than males²⁸ and females are at higher risk of developing COPD than males with similar smoking histories,²⁹ our finding and those of Stockley et al indicate that low FEF_25-75 (which is likely suggestive of impairment in the small airways) is likely to be greater in females before developing overt AL. Studies have reported that females have small tracheal cross-sectional area compared with males.^{30 31} This may be similar throughout the bronchial tree explaining why females are most likely to have low FEF_25-75 without AL than males. However, confirming this will require more comprehensive studies.

In the current study, age was higher in the low FEF_25-75/AL+ group than the normal FEF_25-75/AL− group and low FEF_25-75/AL− group, but was reduced in those with very severe AL compared with all other severities of AL. In a complex disease such as COPD, decline rates are variable. Age (as a surrogate of time) might account for some of the differences in baseline lung function between the low FEF_25-75/AL− group and low FEF_25-75/AL+ group. However, age was not a significant factor accounting for the presence of low FEF_25-75 in multivariate regression modelling. The contribution of ageing on the presence of low FEF_25-75 can only be confirmed by longitudinal follow-up, which would also enhance our understanding of the relationship between small and large airways function in COPD and might support new monitoring and treatment strategies.

Smoking exposure was similar between low FEF_25-75/AL− group and low FEF_25-75/AL+ group and did not differ across increasing AL severity (as grouped by FEV₁ z-score) nor was associated with low FEF_25-75 in multivariate analysis. These results suggest that smoking exposure alone cannot explain the physiological differences between groups. Tsushima et al reported similar findings, demonstrating that smokers with COPD had similar pack-year history compared with those designated at-risk of COPD,¹⁶ although Mirsadraee et al suggested this reflected a lower smoke exposure.¹⁷ This latter study used GOLD criteria and % predicted to define groups while our study used the z-scores to define abnormality in FEV₁/FVC and FEF_25-75. The physiological criteria used may account for some differences in study findings.

The FEV₃/FVC has been used to detect mild lung injury in the absence of AL.³² Morris et al reported that, compared with those with normal FEV₃/FVC, patients with a lower ratio had lower FEV₁, higher residual volume (RV)/total lung capacity (TLC), higher RV, higher TLC and lower transfer factor for carbon monoxide (TLco), potentially highlighting the presence of early physiological impairment including air trapping and impaired gas exchange.³² Our study demonstrated that FEV₃/FVC was lower (although within normal range) in the low FEF_25-75/AL− group than in normal FEF_25-75/AL− group and was strongly associated with the FEF_25-75 z-score, providing further support that the FEF_25-75 z-score is likely detecting early lung pathology in this group. FEF_25-75/FVC has also been used in the early detection of COPD¹⁷ and again this measure was also lower in low FEF_25-75/AL− group, further supporting the FEF_25-75 z-score.

This study found that the use of ICS/LABA was as high in the low FEF_25-75/AL− group as in the low FEF_25-75/AL+ group, despite no AL in former group. This contradicts the recommendation by NICE guidelines that the use of ICS/LABA should be for those with spirometrically confirmed AL.³³ Therefore, the absence of AL in low FEF_25-75/AL− group raises concern regarding the reason for prescribing such high levels of ICS/LABA. ICS/LABA combinations contains high dose of ICS characterised by high potency, and adverse effects, including community-acquired pneumonia, glucose dysregulation and adrenal suppression.³⁴ There are two possible reasons why patients in low FEF_25-75/AL− group are prescribed ICS/LABA. First, the current study used the LLN to define AL, whereas the fixed 70% cut-off is still widely used in clinical practice. This could explain that some patients were given ICS/LABA following the confirmation of AL using the fixed ratio cut-off. Second, given the lack of evidence on how to treat patients with symptoms of COPD despite no AL, the patients might have experienced worse respiratory symptoms, requiring physicians to escalate therapy, by the addition of ICS. Whether using COPD medications (and especially ICS) to treat patients without AL is of benefit in the patients described here, requires appropriate randomised control trials. An RCT by Han et al is ongoing, which evaluates using LABA/LAMA in patients with COPD symptoms but no AL to determine whether such medication is effective in such patients³⁵ and the same should be done with ICS.

Several studies have assessed FEF_25-75 in COPD. FEF_25-75 % predicted was lower (though not necessarily abnormal) in patients at risk of developing COPD.¹⁶ Correction of FEF_25-75 for FVC also identifies early pathological changes prior to COPD development¹⁷ and expiratory flow rates (including FEF_25-75) detected abnormality in those with normal FEV₁/FVC.³⁶ Our findings, together with other studies strengthen FEF_25-75 (expressed as either % predicted or z-score) as a valuable marker of impairment in the small airways before classically defined AL is present.15–17 21 36

Concerns about the use of FEF_25-75 in clinical management have been raised, for example, in a large cross-sectional study using FEF_25-75 z-score.³⁷ That study concluded that FEF_25-75 did not provide additional information to current spirometric measures used in clinical practice, which contrasts with the close relationship demonstrated in our study. However, the study by Quanjer et al³⁷ included a large and mixed population of participants including a variety of lung diseases. The lack of utility of a test in a general population does not negate its use in a selected one, a concept supported in the study of a highly selected population (AATD), where low FEF_25-75 % predicted in the absence of AL was associated with a reduced health status and a subsequent faster decline in lung function.¹⁵ In addition, that study suggested that low FEF_25-75 preceded the development of macroscopic emphysema, a classic component of the PiZZ genetic variant.

A 10-year longitudinal study demonstrated that non-AATD patients with low FEF_25-75 z-score had a higher incidence rate of developing COPD than those with normal FEF_25-75 z-score (41.8% vs 7.4%, p<0.001).²¹ The authors used the same normality cut-off for as used in the current study.²¹ Considering that small airways dysfunction seems to precede AL¹⁵ and the fact that loss of >70% of small airways has to occur before COPD becomes detectable by FEV₁/FVC,¹² patients with FEF_25-75 z-score<−0.8435 described by Kwon et al possibly had impairment in their small airways that would have worsened over time due to the continual exposure to risk factors, leading to the development of AL.²¹

Our study provides evidence to support the use of FEF_25-75 (expressed as z-score) as an assessment tool in patients potentially at risk of developing COPD. We suggest that patients with FEF_25-75 <-0.8453 should be considered a phenotypic group that likely reflects early impairment in the small airways. This group of patients should be monitored and early preventive measures (most importantly, smoking cessation) should be objectively supported and encouraged especially when there is progression. In this group, the reduction of environmental-related exposure (ie, pollution, work related exposure and biomass fuel exposure) may also be beneficial in stabilising progression to COPD. Moreover, pharmacological treatments such as extra-fine particles inhalers may be of particular use in this group, as they achieve higher deposition in the small airways.^38–40 However, this concept clearly requires further research to determine whether such treatments are of value for this group. Other measures of small airways have also demonstrated value in the early detection of COPD.^{18 20 41} In this study, we chose FEF_25-75 because of its availability already in routine physiological assessment.

Our study has limitations. It was a cross-sectional study but the value of FEF_25-75 as a monitoring tool has also been demonstrated longitudinally^{15 21 22} and our study provided a larger sample confirming the prevalence of low FEF_25-75 in smokers with and without AL. FEF_25-75 is a highly variable spirometric measure but we used FEF_25-75 z-score to optimise the interpretive accuracy. This was also a retrospective study, meaning that available data were limited to routine lung function tests, although this is more representative of the real-world approach to such strategies. Studies have shown that RV/TLC is also a potential marker for SAF.^42–45 However, the data analysed in this study was limited to spirometric measures and did not include lung volumes such as RV and TLC. Therefore, further studies should evaluate whether low FEF_25-75 is associated with low RV/TLC. We pragmatically used<−0.8453 z-score cut-off to define low FEF_25-75, which is different from the LLN for other lung function parameters. The ERS/ATS guidelines highlight that no satisfactory outcome-based thresholds for lung function have been defined and that further research is needed to establish a comprehensive disease-specific clinical approach to interpretation.⁴⁶ The chosen cut-off for our study has also been used by others and shown to significantly predict COPD development,²¹ indicating it likely reflects early impairment in the small airways.

In conclusion, low FEF_25-75 z-score is a physiological feature present in patients with AL and also in symptomatic patients in the absence of AL. These findings highlight the potential importance of FEF_25-75 as marker of small airways impairment, and importantly, in the detection of early pathological features of COPD. FEF_25-75 is part of routine lung function assessment, and therefore, closely monitoring patients with low FEF_25-75 and considering early interventions may be central to improving health and prognosis.

Data availability statement

Data are available on reasonable request. The lung function data used and evaluated during this study can be made available from the corresponding author, NYA, on reasonable request.

Ethics statements

Patient consent for publication

Not applicable.

Ethics approval

The data study was approved by the Health Research Authority (HRA – project number 274729) and the South Birmingham Research Ethics Committee (Reference number 20/WM/0024).