Treatment approaches: personalising therapy using biomarkers
The first serious attempt to personalise therapy was the INFANT study, using peripheral blood eosinophil count and aeroallergen sensitization, both readily available in the clinic [49]. 300 children age between 12 and 59 months prescribed step two treatment were recruited from 18 sites in the USA. They received in a blinded, three-way crossover trial in random order either daily ICS, daily montelukast or as needed ICS and short acting β-2 agonist. Each treatment period was 4 months, with the first two weeks of data in each treatment arm being discarded in lieu of a washout period which was thought to be unethical. The primary endpoint was a composite outcome of asthma control days and time to attack requiring oral corticosteroids. They prespecified that aeroallergen sensitization, gender and wheeze attacks would predict a differential treatment response; the use of blood eosinophil count was apost-hoc analysis. Sixty of 300 improved spontaneously, and unsurprisingly there was no differential response to treatment; whatever they received they did well. 170 children showed a differential response, and in this group as a whole, regular ICS was the best option, and montelukast the least good. When they divided the group by aeroallergen sensitization, the non-sensitized patients (n=130) did equally well (or badly) irrespective of treatment, whereas those allergen sensitized (n=100) did best in the regular ICS arm. They then carried out a post-hoc analysis, dividing the groups at the semi-arbitrary cut point of a blood eosinophil count of 300 cells/μλ. Below this level, the treatment results were the same in all three arms (n=82). Those with a count of 300 and above (n=71) were the group that did best on regular ICS. Those who were both aeroallergen sensitized and with a blood eosinophil count of at least 300 (n=64) were the group who did best when prescribed regular ICS; in all others, treatment effects were the same.
This study has opened the door to personalising treatment using two simple biomarkers, but a note of caution must be sounded. The blood eosinophil analyses were post-hoc , and thus hypothesis generating and requiring confirmation in a second study. The stability of blood eosinophil count was not measured; at least in school age children with asthma, sputum inflammatory biomarkers are not stable [50]. The cut-off level of blood eosinophils needs thought; 300 cells/μλ is the upper limit of normal for adults and used as an indicator for Type 2 biologics, [51] but the upper limit of normal in children is much higher [52]. Furthermore, elevated blood eosinophil count may be caused by eczema or other atopic disease, or parasitic infections. There are limited paediatric data showing bronchoalveolar lavage and peripheral blood eosinophil counts correlate [48] but probably the safest interpretation is that if blood eosinophil count is normal, airway eosinophilia is unlikely; if high, then one possible explanation is that the treatable trait of airway eosinophilia may be present.
The use of biomarkers was further explored in a meta-analysis [53] of three previously reported randomised controlled trials in 1074 children age 12-71 months (Table 6) [27. 35, 40]. Blood eosinophil count and aeroallergen sensitisation were determined at the start of the trial. The investigators determined the predictive value of different blood eosinophil counts from > 150 to 350 cells/μλ. Unsurprisingly, patients with eczema had higher blood eosinophil counts. The risk of an exacerbation increased with increasing blood eosinophil count, but the predictive value of a blood eosinophil count was low. Prediction was improved if allergen sensitization was added to the model, such that at any level of eosinophil count, allergen sensitization was present. In children prescribed ICS, the predictive effect of the two biomarkers was not clinically significant. Perhaps it is unsurprising that these three studies did not give clearcut answers; the treatments were randomised, not clinically prescribed, and this may well have affected the findings.
Future work, in addition to validating the original INFANT observations, will include optimising the eosinophil cut-off, including in areas of high parasite burden, and exploring whether the addition of exhaled nitric oxide (FeNO), as in adults [54], will improve risk assessment and personalising medicine. At the present time, it seems reasonable at least in secondary care to measure both biomarkers and use them to guide whether ICS are indicated – specifically, if neither blood eosinophilia nor aeroallergen sensitisation is present, it is probably right to withhold ICS.
Whatever the biomarker status, if an N-of-1 trial of ICS is contemplated, a three step protocol is advocated, to prevent transient symptoms being interpreted as chronic. The steps are:
  1. Commence ICS through an age-appropriate spacer; dose is arbitrary, but I would use a relatively high dose, beclomethasone 200 mcg twice daily on the basis that if the child does not respond, then a steroid sensitive airway disease is unlikely. The family is told that the treatment will be reviewed and discontinued after six weeks (again, an arbitrary time period). Ideally adherence should be monitored electronically
  2. Review the child at six weeks. If there has been no response, then the treatable trait of airway eosinophilia is not present, and alternative diagnoses and management strategies should be sought. If the child is symptomatically improved, it is not clear whether this was spontaneous, or treatment related. This is resolved by a period off treatment.
  3. Review again after 6-8 weeks. If the child is asymptomatic, no further action is needed. If symptoms have recurred, then ICS are re-started and titrated down to the lowest dose needed to control symptoms