Molecular Heterogeneity, Biomarker Discovery, and Targeted Therapy in Asthma - Genomic Biomarkers for Pharmaceutical Development - page 83

were comparable in intrapatient variability to FeNO with a CV of 21.3% (95% CI 18.7-24.0%)
[72] . In a subsequent analysis, we found that while baseline serum periostin and FeNO lev-
els were significantly positively correlated among patients in the trial, there were subsets of
patients that were above the median for one, but not the other biomarker, while others were
either below or above the median for both biomarkers. Of these four subsets, the one with
both serum periostin and FeNO greater than the median exhibited the greatest magnitude
of FEV1 improvement relative to placebo [75] . While the size and duration of the study were
not adequately powered to show benefit in terms of severe exacerbations within the 20 week
treatment period, pharmacodynamic effects of lebrikizumab were observed for as long as
32 weeks. Considering a 32 week observation period, there was a significant reduction in the
rate of severe exacerbations in the lebrikizumab arm as compared to the placebo arm; while
there was a greater magnitude of exacerbation reduction in 'diagnostic-positive' patients, this
effect was not statistically significant [76] .
Crucial mechanistic insights about the relationships between IL13 and non-invasive bio-
markers came from pharmacodynamic analyses. Lebrikizumab-treated patients exhib-
ited a small increase in peripheral blood eosinophil counts, which may be a consequence
of decreased levels of IL13-induced eosinophil-attracting chemokines, such as CCL13 [72] .
While it remains to be formally determined whether IL13 blockade reduces the levels of
eosinophils in bronchial mucosal tissue, the increased blood eosinophil counts and decreased
CCL13 levels are consistent with this hypothesis. Lebrikizumab treatment induced sus-
tained reductions in the mean levels of FeNO to below 20 ppb in the study, which is within
the range of nonasthmatic healthy control patients [55] . Interestingly, lebrikizumab-treated
patients with serum periostin levels above the median at baseline exhibited a significant
decrease in serum periostin levels, while there was no significant change in serum periostin
levels in patients whose baseline serum periostin levels were below the median [77] . Taken
together, these pharmacodynamic effects in which FeNO and serum periostin are 'normal-
ized' (but not reduced to undetectable levels) suggest that IL13 is a primary driver of excess
FeNO in asthma patients and that the excess serum periostin levels observed in 'periostin-
high' patients with asthma are due to the effects of IL13.
By assessing the effects of lebrikizumab in a stratified population comprising patients with
greater and lesser degrees of activity of the targeted pathway, we have tested the hypothesis
that 'diagnostic-positive' patients are more likely to exhibit clinical benefit from IL13 block-
ade than 'diagnostic-negative' patients. Given the size of the study, the width of the confi-
dence intervals for the clinical outcome measures, and the intrapatient variability in FEV1
and predictive biomarkers, it is difficult from these data to more precisely define biomarker
cutoffs to differentiate between patients most likely to show clinical benefit from IL13 block-
ade and those less likely to benefit. Clearly, these encouraging initial findings will need to be
replicated and refined in larger cohorts.
4.4 COMPANION DIAGNOSTIC DEVELOPMENT
As detailed above, the tasks of discovering candidate biomarkers and devising prototype
assays to detect those biomarkers early in the drug development process is difficult and com-
plex. However, these basic and translational research questions can be pursued according to