Day-time variation of serum periostin in asthmatic adults treated with ICS/LABA and adults without asthma
This study shows that serum periostin levels vary throughout the day in adults with and without asthma, with higher levels in the morning in both groups. The magnitude of the variation was small, suggesting that the time that periostin levels are measured is unlikely to influence treatment decisions if predicting responsiveness to, or eligibility for monoclonal antibody therapy directed against IL-4Ra, IL-13, IgE or other components of type-2 inflammation in asthma.
The median baseline values of 51.7 and 49.4 ng/mL in the asthma and non-asthma groups respectively are similar to the median levels of 50.1 ng/mL in a population without asthma or COPD [12], 53.7 ng/mL in a random adult population with a diagnosis of asthma [13], and 50.2 ng/mL in adults with moderate to severe asthma inadequately controlled despite ICS therapy [9]. Interpretation of periostin levels might be confounded by the observation that initiation of ICS therapy may result in a modest reduction in serum periostin by a mean of 5.3 ng/mL [13]. Together, these findings suggest that serum periostin is not a measure which can differentiate patients with asthma across a range of severity from a population without asthma.
The daytime variation we have found with serum periostin levels is similar to that reported previously for the type 2-related biomarkers of sputum and blood eosinophils [14–18] and FeNO [19–21]. In allergic subjects with mild asthma, sputum eosinophils are about two-fold higher at 0700 h than at 1600 h [14]. Additionally, the early morning increase in sputum eosinophils correlated with enhanced airway obstruction and reversibility, suggesting that airway recruitment of eosinophils might contribute to circadian variations in lung function in patients with asthma [14].
In adults with mild asthma, blood eosinophil counts are about 25% higher at 0400 h than at 1600 h [15]. It has been observed that the circadian change in blood eosinophils and lung function appear to fall into a continuous range, suggesting that day/night variations in airways inflammation and lung function occur as a continuum, rather than as an all or nothing phenomenon [15]. In allergic subjects with moderately severe asthma, a circadian variation in blood eosinophil counts was also observed with peak values overnight [16]. In healthy subjects, blood eosinophil counts may also vary diurnally, being lowest in the morning and highest at night, correlating inversely to blood cortisol levels [17]. The demonstration of daytime variation in the non-asthma, but not the asthma group in our study is likely to reflect the lack of power, low sensitivity of the automated measurement of blood eosinophil levels to increments of 0.1 per 109/L, and possibly the effect of the maintenance treatment with ICS in all asthma participants.
In asthma, there is a variable degree of diurnal variation in FeNO, which is greatest in uncontrolled disease and serves as a predictor of risk of future exacerbations [19–21]. The mean diurnal FeNO variation, measured as the difference in morning (0700–1000 h) from evening (1800–2100 h) levels over a two-week period, was 15.6 ppb in uncontrolled asthma subjects compared with 8.2 ppb in stable controlled asthma, and 6.1 ppb in healthy subjects [21]. In another study of adults with asthma, morning FeNO levels were reported to be 14% higher than evening levels [19]. In our study we observed a mean difference in FeNO between 0800 and 1800 h on a single day of 4.4 ppb in controlled asthma and 3.8 ppb in non-asthma. Finally, the lesser variability, based on the ratio of geometric mean values compared to baseline, of serum periostin levels compared with FeNO levels in our study is consistent with previous observations that there is a lesser intra-patient variability in periostin levels compared with FeNO [9].
The clinical relevance of our findings is illustrated by the post hoc analysis in which we determined that only one in 16 asthma participants changed their periostin classification between ‘high periostin’ and ‘low periostin’, based on the 0800 and 1800 h levels, utilizing the proposed periostin cut point of 50 ng/mL, used to determine responsiveness to monoclonal antibody therapy directed against IL-13 [9], and IgE [10]. Thus daytime variation of periostin is unlikely to be an important consideration using the 50 ng/mL cut point. However, the validity of the cut-point as a predictor of responsiveness will need to be confirmed in future studies utilizing the same assay and a cut-off needs to be established if other assays are used. Although the used assay has a very good precision around the cut-point of 50 ng/mL (repeatability CV 0.9–1.5%, intermediate precision CV 1.2–1.7% and reproducibility CV 1.7–3.1%), a degree of error of misclassification cannot be excluded It should also be noted that the assay used has a reproducibility CV of 1.7–3.1% around the 50 ng/mL cut point of 1.7–3.1%, introducing a degree of reproducibility error when taking repeat samples [27].
There are a number of methodological limitations relevant to the interpretation of the study findings. No adjustment has been made for multiple statistical testing, thus the findings should be considered illustrative. The findings are generalizable to Caucasian adults with asthma on regular ICS and LABA treatment, representing GINA Step 3 and 4 therapy [28], but generalizability to other ethnicities is less certain. ICS reduces serum periostin levels [13], FeNO [29–31], and blood eosinophil counts [16], however the effect of ICS on the circadian rhythm of periostin is not known. Periostin levels are unlikely to have been influenced by the presence or absence of atopy in either the asthma or non-asthma group [12, 13], but It is possible that serum periostin levels may show a greater magnitude of circadian variability in a population in which there is a greater proportion with uncontrolled disease, similar to an effect shown with FeNO [21]. In this sample there were no participants on chronic oral corticosteroids however this is a group that might be more likely to be considered for monoclonal antibody therapy regardless of serum periostin levels. The use of LABA therapy prior to the baseline morning measurements and no subsequent medication use during the period of the study, is unlikely to have influenced the periostin or FeNO levels, but may have resulted in the gradual reduction in FEV1 during the study period as the bronchodilator effect of the LABA wore off.