Lung function trajectories in a cohort of patients with moderate-to-severe asthma on mepolizumab, omalizumab, or dupilumab

Abstract

Background:

Lung function is an independent predictor of mortality. We evaluated the lung function trajectories of a cohort of patients with asthma receiving biologic therapy.

Methods:

We identified 229 monoclonal antibody-naïve adult patients with moderate-to-severe asthma who initiated omalizumab, mepolizumab, or dupilumab between 2010 and 2022 in a large healthcare system in Boston, MA. Generalized additive mixed models were used to estimate the lung function trajectories during the 156 weeks following biologic initiation. Response was defined as an improvement in FEV₁ or a decrease of ≤0.5% per year. The Kaplan-Meier estimator was used to assess time to no additional improvement in FEV₁ in responders. All models were adjusted for age, sex, body mass index, smoking status, baseline exacerbation rate, and baseline blood eosinophil count.

Results:

Eighty-eight patients initiated mepolizumab, 76 omalizumab, and 65 dupilumab. Baseline eosinophil count was highest in the mepolizumab group (405 cells/mcL) and lowest for omalizumab (250 cells/mcL). Both FEV₁ and FVC improved in the mepolizumab group (FEV₁ +20 mL/yr.; FVC +43 mL/yr.). For omalizumab, there was an initial improvement in the first year followed by decline with an overall FEV₁ loss of −44 mL/yr. and FVC −32 mL/yr. For dupilumab, both FEV₁ (+61 mL/yr.) and FVC (+74 mL/yr.) improved over time. Fifty percent of the mepolizumab group, 58% omalizumab, and 72% of dupilumab were responders. The median time to no additional FEV₁ improvement in responders was 24 weeks for omalizumab, 48 weeks for mepolizumab, and 57 weeks for dupilumab.

Conclusion:

In this clinical cohort, mepolizumab, omalizumab, and dupilumab had beneficial effects on FEV₁ and FVC with distinct post-initiation trajectories.

Graphical Abstract

1 |. INTRODUCTION

Lung function, especially the forced expiratory volume in one second (FEV₁), is an independent predictor of mortality in individuals with or without chronic lung diseases.^1–4 In the seminal randomized controlled trials to evaluate the efficacy of biologics in asthma, longitudinal trajectories of the pre-bronchodilator FEV₁,^5–10 or the change in FEV₁ from baseline to a specific time point over follow-up,^11–13 were the outcomes of interest. These studies showed modest to great improvement in the prebronchodilator FEV₁ in patients with moderate-to-severe asthma over relatively short follow-up periods. However, the effect estimates from these studies may not fully reflect the nuances of spirometry measurements in clinical practice where, for instance, individuals may be more likely to be non-adherent to their other asthma medications over time and lung function evaluation may be conducted at irregular intervals.¹⁴ Additionally, observational cohorts allow for the evaluation of FEV₁ trajectory over a longer period following therapy initiation.

Lung function is dynamic which makes longitudinal measurements especially useful. Maximum FEV₁ is attained in adolescence to early adulthood with a gradual decline over the adult lifetime of an individual.^15,16 These patterns are similar in patients with asthma although patients with chronic lung diseases (asthma and COPD) have greater decline in lung function over time compared to healthy subjects.^17–19 Additionally, the rate of lung function decline is greatest in patients who smoke,^15,17,20 have more asthma exacerbations,^{17–19,21,22} have eosinophilic airway inflammation,^23,24 or who are obese.²⁵ These features are not uncommon in patients who require biologics for asthma control in real-world practices. Thus, the evaluation of the lung function trajectories of patients from routine clinical practices will improve our understanding of the effectiveness of these therapies and in managing the expectations of both patients and providers.

In this study, we sought to evaluate the FEV₁ trajectories along with the forced vital capacity (FVC) and FEV1/FVC trajectories in individuals with moderate-to-severe asthma who initiated omalizumab, mepolizumab, or dupilumab in the asthma clinics of a large health system in Boston, MA.

2 |. METHODS

2.1 |. Study design and participants: The MATRIX Cohort

We used data from the Mapping Trajectories of Response to monoclonal antibodies In asthma (MATRIX) cohort. MATRIX is a growing electronic health record (EHR) based retrospective cohort from the Mass General Brigham (MGB) system in Boston, MA collating data from January 2010 to date. It includes biologic-naïve participants with physician-diagnosed moderate-to-severe persistent asthma who initiated biologic therapy for the treatment of their asthma in the MGB severe asthma clinics staffed by allergists or pulmonologists. MATRIX excludes patients who initiated therapy for alternate indications for these biologic therapies which during this study period included hypereosinophilic syndrome and eosinophilic granulomatosis with polyangiitis for mepolizumab; chronic spontaneous urticaria for omalizumab; atopic dermatitis and eosinophilic esophagitis for dupilumab. We did not exclude those with chronic rhinosinusitis or nasal polyposis which are components of aspirin-exacerbated respiratory disease, a subtype of asthma.

For this study, we limited the cohort to individuals aged 18 years or older who had initiated biologic therapy for asthma between 2010 and 2022, and who had at least 1 pulmonary function test (PFT) within 52 weeks before biologic therapy initiation. Mepolizumab, dupilumab, and omalizumab account for over 90% of the respiratory biologic prescriptions in our clinics.^26,27 All patients met the eligibility criteria for the biologic initiated including severe eosinophilic asthma, which we defined as asthma that was uncontrolled on maintenance medications and eosinophil counts ≥ 150 cells/μL at baseline, for mepolizumab, moderate-to-severe allergic asthma for omalizumab, and moderate-to-severe eosinophilic asthma or oral corticosteroids-dependent asthma for dupilumab. There were 1,566 moderate-to-severe biologic-naïve patients who initiated one of these biologics of interest for asthma during the study period and 299 patients met the eligibility criteria and were included in the MATRIX cohort. The study protocol was approved by the MGB Institutional Review Board.

2.2 |. Study outcomes

The primary outcome was the pre-bronchodilator FEV₁ and FEV₁ percent predicted, FVC, and the FEV₁/FVC ratio over 156 weeks (3 years) after biologic initiation. Secondary outcomes included the time to no additional FEV₁ improvement in responders to evaluate the time to observing maximal lung function benefits from these biologics and the sustainability of improvements. We also evaluated the effect on lung function over the 52 weeks following biologic initiation. Spirometry measurements were conducted by qualified personnel trained in the conduct of spirometry in routine clinical visit at steady state not during exacerbations. Interpretation was per ATS/ERS guidelines for spirometry measurements and as per standard procedures in the MGB severe asthma clinics.^14,28 For each patient, we estimated the rate of annual change in FEV₁ compared to the baseline using Huber regression, a regression technique robust to outliers,²⁹ and classified patients with an improvement in FEV₁ or a decline of less than or equal to 0.5% per year as responders.³⁰

2.3 |. Statistical analyses

In primary analyses, we limited our study to participants with a baseline PFT and at least one PFT within 156 weeks of therapy initiation. We assumed all patients who initiated and had an active prescription for these biologics in our clinics had continued to follow up with us. Patients were censored following a switch to another biologic, at 156 weeks following initiation of therapy, or the end of follow-up in December 2022.

We evaluated the overall FEV₁, FEV₁ %predicted, FVC, and FEV₁/FVC trajectories for each biologic by combining the individual PFT trajectories using Generalized Additive Mixed Models (GAMM), an extension of linear mixed effects models,³¹ that allows for modeling nonlinear changes in lung function over time.³² For each model, we first estimated the smooth function of time using a cubic regression spline along with other fixed covariates and obtained the optimal basis dimension from the model with the least root mean square error (RMSE) using maximum likelihood (ML) with 5-fold cross-validation.³³ Then, we refitted the optimized model with restricted maximum likelihood (REML) to minimize bias.³⁴ To account for the variability of individual’s PFT trajectory, a patient-specific random intercept and random slope were used and we assumed a continuous autoregressive correlation pattern.³⁵ All models included age, sex, pre-biologic annualized exacerbation rate, smoking status, body mass index (BMI), and log10 of peripheral eosinophil counts as fixed covariates. A Bayesian simulation-based approach was used to estimate 95% credible intervals (CIs).³² We evaluated the time to no additional improvement in FEV₁ in responders using the Kaplan-Meier estimator. We conducted GAMM analyses using mgcv package.³⁶ All analyses were conducted in R version 4.3.0 (R Foundation for Statistical Computing, Vienna).³⁷

2.4 |. Subgroup and sensitivity analyses

We conducted exploratory subgroup analyses by response status and by sex for each biologic. For mepolizumab and dupilumab, we also conducted subgroup analyses by baseline eosinophil counts using 300 cells/μL as cut-off. For omalizumab and dupilumab, subgroup analyses by baseline IgE using 100 IU/μL as cut-off were conducted. In sensitivity analyses, we included height in the model instead of BMI. Given that patient selection might differ based on the availability of alternative biologics, we evaluated the robustness of our results by limiting the study population to patients who initiated biologic on or after November 2018 when all 3 biologics had been approved for asthma.

3 |. RESULTS

3.1 |. Characteristics of patients

Two hundred and twenty-nine patients met the inclusion criteria. Compared to patients excluded from the analysis, patients included in the analysis had a higher usage of other asthma medications, higher baseline exacerbation rate in omalizumab group, and higher prevalence of other comorbid conditions such as chronic rhinosinusitis in mepolizumab group or allergic rhinitis in dupilumab group (Table S1). Of 229 patients, 88 initiated mepolizumab, 76 initiated omalizumab, and 65 initiated dupilumab (Figure 1). These patients accounted for 873 spirometry measurements. The mean age was 52.5 years, 63.8% were female, and mean BMI was 29.6 kg/m². Most of the patients identified as White and 18.3% were current or former smokers (Table 1). The median eosinophil count was highest in the mepolizumab group with a median of 405 cells/mL, while the median immunoglobulin E (IgE) was highest in the omalizumab group with a median of 164 IU/mL. The baseline exacerbation rate was lowest in the dupilumab group with a mean of 1.1 exacerbations in the prior year. The rate of FEV₁ decline prior to biologic initiation was 0.167 L/year in the mepolizumab group, 0.082 L/year in the dupilumab group, and 0.060 L/year in the omalizumab group (Table S2). Patients who initiated omalizumab had the highest number of follow-up PFTs with 78% having ≥ 2 PFTs post-treatment initiation while dupilumab users had the lowest number of measurements following therapy initiation with 52% of patients having ≥ 2 measurements (Table S3).

FIGURE 1.

Selection of the study population

TABLE 1.

Baseline characteristics of study cohort

	Total	Mepolizumab	Omalizumab	Dupilumab
Number of patients	229	88	76	65
Age in years, mean (SD)	52.52 (15.44)	53.85 (15.20)	48.34 (16.22)	55.60 (13.92)
Female, n (%)	146 (63.8)	55 (62.5)	53 (69.7)	38 (58.5)
Body mass index, kg/m2; mean (SD)	29.58 (7.06)	29.13 (6.57)	30.71 (8.42)	28.88 (5.82)
Height, m; mean (SD)	1.66 (0.10)	1.67 (0.10)	1.64 (0.11)	1.67 (0.09)
White race, n (%)	182 (79.5)	70 (79.5)	54 (71.1)	58 (89.2)
Current smoker, n (%)	4 (1.7)	1 (1.1)	3 (3.9)	0 (0.0)
Former smoker, n (%)	38 (16.6)	17 (19.3)	12 (15.8)	9 (13.8)
Baseline prebronchodilator FEV1 in liters, mean (SD)	2.21 (0.75)	2.22 (0.78)	2.13 (0.72)	2.30 (0.72)
Minimum-Maximum value, in L	0.74–4.52	0.92–4.52	0.74–3.84	0.87–4.15
Baseline prebronchodilator FEV1 % predicted, mean (SD)	75.5 (18.9)	74.5 (18.2)	74.4 (20.1)	78.1 (18.5)
% FEV1 reversibility prior to therapy initiation, mean (SD)	10.5 (13.3)	10.4 (15.5)	12.3 (14.5)	7.6 (7.6)
Baseline annualized exacerbation rate, mean (SD)	2.17 (2.29)	2.15 (2.16)	2.54 (2.64)	1.78 (1.97)
Baseline inhaled corticosteroid use, n (%)	229 (100.0)	88 (100.0)	76 (100.0)	65 (100.0)
Baseline combined inhaled corticosteroid and long-acting beta-agonist use, n (%)	198 (86.5)	75 (85.2)	66 (86.8)	57 (87.7)
Long-acting muscarinic antagonist use, n (%)	72 (31.4)	23 (26.1)	31 (40.8)	18 (27.7)
Leukotriene receptor antagonist use, n (%)	143 (62.4)	53 (60.2)	52 (68.4)	38 (58.5)
On oral corticosteroids for ≥6 months/year, n (%)	19 (8.3)	6 (6.8)	9 (11.8)	4 (6.2)
Eosinophil count
Blood eosinophil counts, cells/μL; median [IQR]	330 [160–660]	405 [184–740]	250 [126–433]	350 [190–660]
≥ 300 cells/μL, n (%)	124 (54.1)	54 (61.4)	33 (43.4)	37 (56.9)
Immunoglobulin E (IgE)
IgE, IU/μL; median [IQR]a	149 [66–363]	142 [62–368]	164 [69–420]	133 [65–288]
≥ 100 IU/μL, n (%)	129 (59.2)	47 (56.0)	48 (63.2)	34 (58.6)
< 100 IU/μL, n (%)	89 (40.8)	37 (44.0)	28 (36.8)	24 (41.4)
Asthma phenotype
Allergic (positive history of allergic rhinitis or allergen-specific sIgE)	187 (81.7)	61 (69.3)	76 (100.0)	50 (76.9)
Eosinophilic (eosinophil count≥ 150 cells/μL)	195 (85.2)	88 (100.0)	54 (71.1)	53 (81.5)
Combined allergic and eosinophilic asthma	156 (68.1)	61 (69.3)	54 (71.1)	41 (63.1)
Asthma severity limited to those who initiated after ICD-10 were introduced, n (%)b	184 (80.3)	68 (77.3)	51 (67.1)	65 (100.0)
Moderate, n (%)	34 (18.5)	10 (14.7)	7 (13.7)	17 (26.2)
Severe, n (%)	104 (56.5)	38 (55.9)	24 (47.1)	42 (64.6)
No explicit moderate or severe codes, n (%)	46 (25.0)	20 (29.4)	20 (39.2)	6 (9.2)
Allergic rhinitis, n (%)	187 (81.7)	61 (69.3)	68 (89.5)c	50 (76.9)
Atopic dermatitis, n (%)	5 (2.2)	3 (3.4)	2 (2.6)	0 (0.0)d
Chronic obstructive pulmonary disease, n (%)	13 (5.7)	4 (4.5)	4 (5.3)	5 (7.7)
Chronic sinusitis, n (%)	31 (13.5)	15 (17.0)	7 (9.2)	9 (13.8)
Nasal polyps, n (%)	16 (7.0)	4 (4.5)	3 (3.9)	9 (13.8)

FeNO, fractional exhaled nitric oxide; FEV₁, forced expiratory volume in one second; IQR, interquartile range; PFT, pulmonary function test; SD, standard deviation

^a4 mepolizumab and 7 dupilumab patients had missing immunoglobulin E levels

^bOnly patients who initiated biologic after ICD-10 was implemented (October 2016) were considered due to the availability of diagnosis code with asthma severity

^cThe eight patients on omalizumab with ‘no rhinitis’ had documented history of allergic rhinitis in the chart and had received omalizumab for asthma treatment

^dWe excluded patients on dupilumab with diagnosis of atopic dermatitis

3.2 |. Lung function trajectories in mepolizumab group

3.2.1 |. Overall trajectories

Baseline pre-bronchodilator FEV₁ was around 2.2 L for mepolizumab users with an estimated 20 mL gain/year (Figure 2a). In addition, mepolizumab steadily improved FEV₁ %predicted by 1.7 percentage points (pp)/year (Figure 2b). The FVC increased by an average of 43 mL/year gain and the FEV₁/FVC was relatively unchanged (Figures 2c, d).

FIGURE 2.

Overall PFT trajectories in mepolizumab group. a) FEV1 trajectory, b) FEV1 %predicted trajectory, c) FVC trajectory, d) FEV1/FVC trajectory. The solid red lines represent the estimated mean PFT from the generalized additive mixed model. The dashed black line represents the slope of the estimated mean. The width of the bands represents the 95% credible interval of estimated mean.

3.2.2 |. Trajectories by sex

FEV₁ in males was stable during the first 72 weeks and then declined with an overall 28 mL loss/year during the 3 years follow-up. However, there was a slight improvement in the FEV₁ %predicted. Meanwhile, females continuously gained FEV₁ over time with an estimated 2.5 pp FEV₁ %predicted improvement/year (Figure S1a, b). Both sexes had improvement in their FVC and consequently, FEV₁/FVC declined slightly over time for males and improved over time for females (Figure S1c, d).

3.2.3 |. Trajectories by response status

There were 44 (50%) mepolizumab responders. Responders and non-responders had similar baseline PFTs but a distinct separation in trajectories post-mepolizumab. Responders showed an average gain in FEV₁ of 115 mL/year (Figure S2a). However, this was reflected as a continuous gain regarding the FEV₁ %predicted (Figure S2b). FVC and FEV1/FVC also improved in responders over the period of follow up (Figures S2c, d). In contrast, non-responders showed a steady decline in their FEV₁, FVC, and FEV₁/FVC throughout the follow-up period.

3.2.4 |. Sustained FEV₁ improvement in responders

At 48 weeks, half of the mepolizumab responders had reached their maximal FEV₁ on mepolizumab. Thus, the remaining half continued to show some ongoing FEV₁ improvement beyond 48 weeks (Figure 3a). Six (13.6%) responders showed additional lung function improvement after 2 years of mepolizumab. At 124 weeks after initiation, all mepolizumab responders had reached their maximal on-treatment FEV₁.

FIGURE 3.

Time to maximum FEV₁ improvement following initiation of mepolizumab in responders. a) Mepolizumab responders, b) Omalizumab responders, c) Dupilumab responders.

3.3 |. Lung function trajectories in omalizumab group

3.3.1 |. Overall trajectories

Omalizumab showed a beneficial effect on PFTs during the first year. However, both FEV₁ and FVC declined at the start of the second year and went back to the baseline level at the end of the second year (Figure 4a–c). The FEV₁/FVC followed a similar pattern but with a faster decline starting as soon as 24 weeks post-initiation (Figure 4d).

FIGURE 4.

Overall PFT trajectories in omalizumab group. a) FEV1 trajectory, b) FEV1 %predicted trajectory, c) FVC trajectory, d) FEV1/FVC trajectory. The solid purple lines represent the estimated mean PFT from the generalized additive mixed model. The dashed black line represents the slope of the estimated mean. The width of the bands represents the 95% credible interval of estimated mean.

3.3.2 |. Trajectories by sex

There was a discernable difference between PFT trajectories in males compared to female omalizumab users where males had an increase in their FEV₁ and FVC throughout the 3-year follow-up with an estimated FEV₁ gain of 68 mL/year, FEV₁ %predicted gain of 4.8 pp/year, and FVC gain of 35 mL/year. In contrast, females gained rapid improvement during the first year then the lung function declined substantially resulting in an overall lung function loss (Figure S3a–c). The FEV₁/FVC trajectories were relatively stable for both sex (Figure S3d).

3.3.3 |. Trajectories by response status

Of 76 omalizumab users, 44 (57.9%) were responders. Although responders had lower baseline FEV₁ and FVC compared to non-responders, the responders caught up during the initial 12 to 24 weeks on omalizumab. In omalizumab responders, FEV₁ improved over time though the greatest improvement was during the first 1.5-years. Non-responders had a steady decline in their lung function parameters (Figure S4).

3.3.4 |. Sustained FEV₁ improvement in responders

The median time to no additional FEV₁ improvement for omalizumab responders was 24 weeks. Nine (20.5%) omalizumab responders had not reached their maximal on-treatment FEV₁ improvement at 52 weeks of follow up and 5 (11.4%) by 104 weeks of follow-up. The longest time to reaching maximal on-treatment FEV₁ in this cohort was 149 weeks (Figure 3b).

3.4 |. Lung function trajectories in dupilumab group

3.4.1 |. Overall trajectories

During the 3-years of follow-up, individuals on dupilumab demonstrated beneficial effects on lung function including an average pre-bronchodilator FEV₁ gain of 61 mL/year, FEV₁ %predicted gain of 2.8 pp/year, and FVC gain of 74 mL/year (Figure 5a–c). The FEV₁/FVC trajectory remained flat throughout the follow-up (Figure 5d).

FIGURE 5.

Overall PFT trajectories in dupilumab group. a) FEV1 trajectory, b) FEV1 %predicted trajectory, c) FVC trajectory, d) FEV1/FVC trajectory. The solid green lines represent the estimated mean PFT from the generalized additive mixed model. The dashed black line represents the slope of the estimated mean. The width of the bands represents the 95% credible interval of estimated mean.

3.4.2 |. Trajectories by sex

Both sexes demonstrated lung function improvement on dupilumab. The PFT trajectories demonstrated ongoing improvement in males at the end of follow-up while the improvement in females reached the plateau stage after approximately 72 weeks (Figure S5a–c). The FEV₁/FVC trajectory in males was slightly improved over time while the trajectory in females was stable (Figure S5d).

3.4.3 |. Trajectories by response status

45 (72.3%) dupilumab users were responders. Responders had slightly lower FEV₁ and FVC at baseline compared to non-responders. For responders, FEV₁ and FVC continuously increased during the three years of dupilumab while lung function declined for non-responders (Figure S6a–c). For responders, the FEV₁/FVC remained stable while the FEV₁/FVC trajectory in the non-responder group decreased over time (Figure S6d).

3.4.4 |. Sustained FEV₁ improvement in responders

Half of the dupilumab responders had reached their maximal FEV₁ within 57 weeks of therapy. Fourteen (29.8%) dupilumab responders continued to show FEV1 improvement at week 104 of follow-up. The longest time to reaching maximal on-treatment FEV₁ in this cohort was 154 weeks (Figure 3c).

3.5 |. Trajectory by eosinophil counts and IgE.

Mepolizumab users with higher eosinophils counts experienced more lung function gain over the first year of follow-up (Figure S7) but had slightly less total lung function improvement over 156 weeks of follow-up (Figure S8). For dupilumab group, the rate of FEV₁ improvement was higher in those with lower baseline eosinophil counts while FVC change was similar between two groups (Figure S9).

Omalizumab users with lower IgE experienced better FEV₁ improvement in both short-term and long-term follow-up while FVC change was similar between two groups (Figures S10 and S11). For dupilumab group, both FEV₁ and FVC improvement were higher in patients with higher baseline IgE (Figures S12a–c). However, the FEV₁/FVC change was not different between two groups (Figure S12d).

3.6 |. Short-term lung function trajectories

All three biologics showed an improvement in prebronchodilator FEV1 and FVC in the first year after therapy. For mepolizumab, FEV₁ increased to a greater extent than FVC but FEV₁/FVC increased initially and then slightly declined in the first year (Figure S13a–d). FEV₁ and FVC improved in the first year of omalizumab (Figure S14a–c) and FEV₁/FVC improved during the first 24 weeks followed by a decline (Figure S14d). For dupilumab, both FEV₁ and FVC improved over 52 weeks, with FEV₁/FVC remaining stable over the 1-year period (Figure S15a–d).

3.7 |. Sensitivity analyses

When BMI was replaced with height in the model, the lung function trajectories were generally consistent with the primary analyses (Figure S16). Similarly, analyses limited to patients who initiated biologics after dupilumab approval showed similar patterns for each biologic (Figure S17).

4 |. DISCUSSION

This single health system study demonstrated that mepolizumab, omalizumab, and dupilumab were associated with improvements in FEV₁ and FVC in a large proportion of patients with moderate-to-severe asthma. The population was predominantly female and White. Patients with mepolizumab had the highest eosinophil count at baseline and greater lung function decline prior to therapy initiation. In general, the gains in lung function were largest in the first year following therapy initiation but more than half of the patients who responded to mepolizumab or dupilumab had additional gains in their FEV₁ after 48 weeks on treatment. Three-quarters of patients who responded to omalizumab had their greatest gains in the first year of therapy with a quarter of patients showing additional benefits after the first year. Females obtained greater benefits than males from mepolizumab while males obtained greater lung function improvements from omalizumab and dupilumab. Our conclusions remained unchanged in sensitivity analyses. The FEV₁/FVC trajectory was relatively unchanged over time on the three biologics, due to improvements in both FEV₁ and FVC and might suggest some improvement in air trapping and small airways disease thus attenuating the “pseudorestriction” (decreased FVC) associated with air trapping in asthma or COPD.³⁸

Most of the evidence on the benefits of biologics on lung function has been from randomized controlled trials. However, the follow-up periods, ranging mostly from 24 to 56 weeks, are insufficient to evaluate the durable effects of these therapies. Prior studies have also shown that the trial populations may not be representative of real-world cohorts, particularly cohorts with a high proportion of smokers, obese patients, or underrepresented racial/ethnic groups.³⁹ Furthermore, very little is known about the time taken to reach peak FEV₁ improvement after initiation of these biologics, but this information will be helpful in managing expectations of response or non-response to therapy. We found that the median time to no additional improvement in FEV1 were different for each biologic group. This reflects both the speed and the sustainability of the response. It is likely that patients who have not demonstrated any gains in FEV₁ within the median time have less propensity to have meaningful improvements. This might be helpful to providers as they evaluate if a patient is demonstrating an appropriate response to these therapies. Moreover, as we move towards a focus on asthma remission and disease-modifying therapies,⁴⁰ providers and patients may value sustainable benefits over bursts of improvement followed by decline.

Our results are consistent with prior studies that mepolizumab is associated with modest improvements in prebronchodilator FEV₁ at six,⁴¹ and 12 months after initiation,⁴² but less than the FEV₁ improvement seen in ‘super responders’ to mepolizumab.^42–44 In the OLE study, COSMEX, there was a slight improvement of FEV1 over 3 years follow-up which was consistent with our study.⁴⁵ However, there was a different pattern of FEV₁ improvement observed in the OLE, COLUMBA, where most improvement was during the first 24 weeks of mepolizumab followed by a gradual decline.⁴⁶ The different pattern between the COLUMBA study and our results might be due to the difference in study population where the mepolizumab group in this study had higher baseline eosinophil counts and prebronchodilator FEV₁. In addition, differences between the groups in the rate of lung function decline prior to biologic initiation could influence the change in lung function after therapy initiation. Despite the different patterns of improvement, this study and previous OLE studies consistently demonstrated moderate yet sustainable lung function improvement during three years of mepolizumab.

Omalizumab improved lung function rapidly during the first year then followed by a decline. The rapid improvement within the first year, particularly within 16 weeks of initiation, was consistent with the meta-analysis of real-world evidence of omalizumab in severe allergic asthma.⁴⁷ Also, in the pooled analyses of the EXTRA and INNOVATE trials, omalizumab was associated with improvements in the FEV₁ trajectory compared to placebo during 28 weeks of follow-up.⁴⁸ Notably, however, there was sustained, but no additional FEV₁ improvement after 4 weeks of omalizumab treatment to the end of follow-up. The differences between that study and our results may be due to differences in the study population. The EXTRA/INNOVATE study populations had higher median IgE and eosinophil counts than this real-world study population. The waning benefits of omalizumab over time might be partly explainable by decreased adherence to omalizumab over time or discontinuation of other asthma medications following the improvement in lung function in the first year of omalizumab.

The FEV₁ improvement in responders to dupilumab is consistent with the meta-analyses of the dupilumab trials,^49,50 and previous studies which included patients who switched to dupilumab.^51–53 While the lung function improvement was sustained, dupilumab’s impact however attenuated over time with the rapid improvement during the first year followed by a plateau. This might be due to a true plateauing of the benefit of dupilumab but could be related to other factors including adherence to dupilumab which is self-administered at home and the discontinuation of inhaled corticosteroids. Moreover, the reduced effect of dupilumab over time might be related to the development of anti-dupilumab antibody in which the incidence was 7.6% in the trial-based meta-analysis.⁵⁴ In the OLE, TRAVERSE, the effect of dupilumab on FEV₁ from the phase 2/3 trials were sustained over the 96 weeks of follow-up but largest in the first year, especially the first four to eight weeks, which was consistent with our results.^55,56 Our results may also be influenced by the number of PFTs per patient which differed slightly by biologic cohort though we accounted for autocorrelation and used a modeling method robust to outliers and single point evaluations.

Consistent with a prior study,⁴² and with prior evidence that female sex hormones may be associated with airway eosinophilic inflammation,⁵⁷ females in this cohort showed a better FEV₁ response trajectory to mepolizumab. Other studies have however shown that sex was not predictive of response to mepolizumab.^58,59 Though response in these studies was defined differently using thresholds that were greater than the average improvement from mepolizumab clinical trials.^27,49,60,61 For dupilumab, males had larger improvements in their FEV₁ following initiation. This is contrary to prior studies in mice showing that eosinophilic and allergic airway inflammation and IL-13 expression are more pronounced in females following OVA challenge which might suggest females had higher propensity to respond to IL-13-targeted therapies.⁵⁷ In the omalizumab group, females had lower long-term lung function improvement. A prior observational study in severe allergic asthma showed that females had a shorter time to omalizumab response.⁶³ This is consistent with our results considering the onset of response to omalizumab, but not with the sustainability of improvement. However, the effect modification of sex on lung function change are based on exploratory analyses and warrant further research.

The strengths of this study include the relatively longer follow-up time for PFT trajectories up to 156 weeks post-biologic, allowing us to evaluate the sustainability of beneficial effects and provide additional information compared to the clinical trials of these therapies and recent observational studies. These are coupled with the generalized additive mixed model which reflects non-linear dynamics of FEV₁ change and more aptly captures the irregularities in spirometry assessment in clinical practices.

Our results should however be interpreted with caution. First, spirometry was assessed at irregular intervals and the measurement may be subject to systematic error due to differences in the quality, technique, and equipment used.¹⁴ For instance, while usually requested, patients may not have uniformly withheld long-acting beta-agonists (LABA) before PFT. This however reflects real-world practice of routine lung function measurement. We accounted for in-person variation and limited our analysis to patients seen in the specialty allergy and pulmonology clinics to limit measurement error. Second, there might be a potential selection bias where we capture more severe patients who required more spirometry evaluation post-biologic initiation. Thus, our results may not be generalizable to all asthma patients receiving these biologics. Third, our sample size was small, and results may not be generalizable to pediatric populations, cohorts with mostly publicly insured individuals, or of underrepresented racial and ethnic groups. Fourth, we focused on prebronchodilator PFT, the main lung function outcome in the efficacy trials of these therapies. However, the postbronchodilator FEV₁ might also give additional insights into response to these therapies. A previous study showed that anti IL-5 biologics worked similarly regardless of bronchodilator reversibility, though the study did not focus on lung function change.⁶⁴ Moreover, we observed slight differences in the rate of lung function decline prior to biologic initiation with the mepolizumab group experiencing the most rapid decline in FEV₁. However, these analyses were limited by low sample sizes and further research is needed to disentangle the connection between the lung function trajectory before therapy initiation and response to therapy. Additionally, despite our best efforts, this is an observational study and there might be important unmeasured confounders influencing results. For instance, we could not sufficiently capture the changes to other asthma medications, and patients whose FEV₁ continued to decline may be patients who discontinued their other maintenance medications such as inhaled corticosteroids. Our results were however robust to sensitivity analyses, but these results need to be replicated in other cohorts. Other important confounders or important covariates not captured sufficiently were the asthma control score, the duration of asthma, and the fractional exhaled nitric oxide (FeNO). Patients with longer duration of asthma might experience more airway remodeling and poorer lung function improvement on biologics. To mitigate this limitation, we adjusted for age, but this would only partially account for duration of asthma and does not allow us to look at this potentially important clinical question. Similarly, FeNO is now a well-recognized predictor of dupilumab and omalizumab’s response.^65,66 Additionally, patients may have stopped following within the MGB system though MGB accounts for the overwhelming majority of prescriptions of these biologics in Northeastern, US. Lastly, these biologics were approved for different patient populations and the timing of regulatory approval of these biologics might have influenced prescribing patterns. Thus, we have made comparisons within and not across biologics. Sensitivity analyses limited to patients who initiated biologics during the period when all 3 biologics were approved showed consistent results.

In conclusion, mepolizumab, omalizumab, and dupilumab improved the rate of FEV₁ decline in 50–72% of patients with moderate-to-severe persistent asthma over 156 weeks post-initiation. Half of the patients who responded to omalizumab, mepolizumab, and dupilumab continued to experience an ongoing improvement in FEV₁ up to and beyond 6-, 12-, and 14-months respectively.

• This study evaluates the lung function trajectories of a cohort of patients with asthma receiving biologic therapy.

• Mepolizumab, omalizumab, and dupilumab improved the rate of FEV1 decline in 50–72% of patients with moderate-to-severe persistent asthma over 3 years post-initiation.

• Half of the patients who responded to omalizumab, mepolizumab, and dupilumab continued to experience an ongoing improvement in FEV1 up to and beyond 6-, 12-, and 14-months respectively.

Abbreviations:

BD

bronchodilator

FEV1

forced expiratory volume in one second

L

liter