To the Editor,
Severe eosinophilic asthma is characterized by increased blood eosinophil levels and
recurrent exacerbations, and often associated with nasal polyposis.
1
Eosinophil cationic protein (ECP) and eosinophil‐derived neurotoxin (EDN), granule
proteins released by eosinophils, are markers of eosinophil activation and have been
identified as potential biomarkers of type 2 eosinophilic disease in patients with
asthma.
2
,
3
,
4
The anti‐interleukin (IL)‐5 monoclonal antibody mepolizumab has been shown to reduce
peripheral blood eosinophil counts (PBEC) and asthma exacerbations versus placebo
in clinical studies.
5
,
6
,
7
Elevated PBEC and frequent exacerbations are key determinants for predicting patients
with severe asthma who are most likely to respond to mepolizumab;
6
,
8
however, identifying additional biomarkers may enhance patient selection. This post
hoc analysis of data from the Phase III MENSA study (GSK ID: 115588; NCT01691521)
7
investigated the relationship between baseline type 2 biomarkers and clinical outcomes
in patients with severe asthma receiving mepolizumab.
MENSA was a randomized, double‐blind trial in patients aged ≥12 years with severe
eosinophilic asthma.
7
Patients were randomized (1:1:1) to receive mepolizumab 75mg intravenously (IV) or
100 mg subcutaneously (SC), or placebo every 4 weeks for 32 weeks plus standard of
care (see Appendix S1 for further details). Levels of the biomarkers EDN, ECP, chemokines
(CCL‐13, CCL‐17, CCL‐22 and eotaxin‐1), periostin and IL‐13 were determined using
serum samples taken at Weeks 0 (randomization) and 32 (exit).
Endpoints included the following: annualized rate of clinically significant exacerbations
(defined as worsening of asthma requiring systemic corticosteroids for ≥3 days and/or
hospitalization/emergency department visit); ratio to baseline of PBEC; change from
baseline in prebronchodilator forced expiratory volume in 1 second (FEV1), asthma
control questionnaire (ACQ)‐5 score, St George's Respiratory Questionnaire (SGRQ)
total score (all at Week 32). Post hoc analysis assessments included baseline biomarker
levels by PBEC subgroups (<150, 150‐<300, 300‐<500, ≥500 cells/μL) and presence of
nasal polyposis at screening, correlation between biomarker levels and PBEC at baseline,
and change from baseline in biomarker levels at Week 32. Additionally, the ratio of
PBEC to baseline, changes from baseline in FEV1, ACQ‐5 and SGRQ scores, and annualized
exacerbations (all at Week 32) were assessed in high (>median) and low (≤median) EDN
(median = 57.6 μg/L) or ECP (median = 21.06 μg/L) subgroups at baseline (initial results
determined which biomarkers were analysed further [EDN and ECP]). The annualized exacerbation
rate was also assessed in high/low EDN and ECP subgroups further stratified by PBEC;
baseline EDN level and baseline PBEC as predictors of response to mepolizumab treatment
were also evaluated. Statistical analyses are described in Appendix S1. The MENSA
trial was conducted in accordance with all applicable country‐specific regulatory
requirements, and all patients provided written informed consent. Ethical approval
was not required for this post hoc analysis.
In MENSA, 194 patients received mepolizumab 100 mg SC, 191 received mepolizumab 75
mg IV and 191 received placebo. At baseline, levels of most biomarkers increased with
increasing PBEC, and the biomarkers EDN, ECP and IL‐13 showed a moderately positive
correlation with PBEC (Table S1). Patients with nasal polyposis also had numerically
higher baseline PBEC, EDN, ECP, CCL‐13, CCL‑17 and IL‐13 levels than those without
nasal polyposis (data not shown). EDN was reduced by 70% from baseline to Week 32
with mepolizumab versus placebo, with a similar trend noted for ECP (58% reduction)
(Figure S1A). In contrast, there was a trend for an increase in the levels of CCL‐13,
CCL‐17, CCL‐22 and eotaxin‐1 between baseline and Week 32 with mepolizumab versus
placebo (Figure S1B).
Owing to the positive correlation of EDN and ECP with PBEC, and their use as markers
of eosinophil activation, we investigated clinical outcomes following mepolizumab
treatment in patients with differing baseline EDN and ECP levels. Baseline PBEC was
higher in patients with high versus low EDN or ECP levels at baseline and mepolizumab
significantly reduced PBEC between baseline and Week 32 to a similar extent in both
high and low EDN and ECP subgroups versus placebo. Mepolizumab‐induced improvements
in other clinical outcomes were also numerically greater in high versus low baseline
EDN or ECP subgroups (Table S2); as such, EDN and ECP may predict improvements in
FEV1.
We also found mepolizumab reduced the exacerbation rate versus placebo in both baseline
EDN subgroups, with a greater reduction in patients with high versus low EDN (Figure 1).
To further investigate whether this effect could be explained in terms of confounding
by baseline PBEC, subgroups were stratified by baseline PBEC. In patients with baseline
PBEC ≥ 300 cells/µL, there was a larger reduction in exacerbations with mepolizumab
in the high (75%) versus low (28%) baseline EDN subgroups. A similar trend was noted
between the two < 300 cells/µL EDN subgroups. However, no difference was observed
in the ECP subgroups. Furthermore, predicted rate ratio modelling of exacerbations
demonstrated that a predictive model based on baseline PBEC was marginally better
than that based on baseline EDN in terms of precision and model fit (Figure 2).
Figure 1
Ratio of the annual rate of clinically significant exacerbations with mepolizumab
100 mg SC versus placebo by baseline biomarker and PBEC subgroup. CI, confidence interval;
ECP, eosinophil cationic protein; EDN, eosinophil‐derived neurotoxin; PBEC, peripheral
blood eosinophil count; SC, subcutaneous
Figure 2
Predicted rate ratio (95% CI) of mepolizumab 100 mg SC versus placebo for clinically
significant exacerbations per year versus (A) baseline PBEC, (B) baseline EDN concentration
and (C) baseline EDN concentration adjusted for baseline PBEC and treatment interaction.
†Akaike information criterion; lower values indicate a better model fit. Shading indicates
95% CI; wider bands indicate lower precision in predicting exacerbation rate. All
analyses performed using a negative binomial regression model with covariates: treatment,
maintenance corticosteroid use, exacerbation number in previous year. Additional covariates:
(A) baseline PBEC (square transformation) including additional term for treatment
interaction; (B) baseline EDN concentration (square‐root transformation) with term
for treatment interaction; (C) baseline EDN concentration (square‐root transformation)
with term for treatment interaction and baseline PBEC (log transformation) with treatment
interaction. CI, confidence interval; EDN, eosinophil‐derived neurotoxin; PBEC, peripheral
blood eosinophil count; SC, subcutaneous
Overall, we found baseline PBEC correlated with baseline EDN levels, as reported elsewhere,9
with a more pronounced reduction in the placebo‐adjusted annualized exacerbation rate
with mepolizumab in patients with high versus low baseline EDN levels (trend not seen
with ECP). Modelling analysis results demonstrated EDN levels and the combination
of EDN levels, PBEC and treatment interaction did not show improved predictive power
versus that of PBEC alone for treatment response to mepolizumab regarding exacerbation
reduction. This result indicates that the predictive power of EDN is largely due to
its correlation with baseline PBEC and highlights that PBEC remains an important and
clinically relevant biomarker for identifying patients with severe asthma who are
likely to respond to mepolizumab treatment.
In conclusion, our results demonstrate that in general, patients with higher versus
lower EDN levels are likely to have improved responses to mepolizumab, although further
research is needed on the role of EDN as a potential biomarker for treatment response
to mepolizumab. However, baseline PBEC has greater precision as a predictive biomarker
of treatment response to mepolizumab than EDN and is more widely assessed in clinical
practice. Our data provide further evidence that PBEC is the most suitable biomarker
identified to date for identifying patients likely to respond to mepolizumab, although
the identification of additional biomarkers may further aid patient selection in the
future.
CONFLICTS OF INTEREST
PH, SM, SB, SY and NK are employees of GlaxoSmithKline (GSK) and hold stocks/shares.
FCA is a former employee of GSK and holds stocks/shares. SQ has served as a consultant
to AstraZeneca, Novartis, Sanofi, Genentech, TEVA, ALK, Mundipharma and GSK; and has
received lecture fees from Chiesi, Novartis, GSK, Leti, AstraZeneca and Mundipharma.
AP reports grants, personal fees, nonfinancial support and other from Chiesi, AstraZeneca,
Boehringer Ingelheim, Mundipharma and TEVA; personal fees and nonfinancial support
from Menarini, Novartis and Zambon; and personal fees from Sanofi, all outside the
submitted work. EI has served as a consultant to and received personal fees from AstraZeneca,
Equillium, GSK, Novartis, 4D Pharma, Pneuma Respiratory, Regeneron Pharmaceuticals,
Sanofi, Sienna Biopharmaceuticals and TEVA; reports nonfinancial support from TEVA,
and other from Vorso Corp; and has received clinical research grants from AstraZeneca,
Genentech, Novartis and Sanofi.
TRIAL REGISTRATION
Data from this post hoc analysis are from the MENSA trial, which is registered on
ClinicalTrials.gov (GSK ID: MEA115588; NCT01691521).
Supporting information
Appendix S1
Click here for additional data file.
Fig S1
Click here for additional data file.