Airflow trajectories in childhood asthma: what data-driven research can teach us

Asthma is one of the most common chronic childhood diseases globally (1). Its contribution to persistent and irreversible respiratory morbidity in adults is well defined and is likely driven by conversion of reversible airflow impairment to irreversible impairment (2). Despite this understanding, there has been a paucity of data to guide clinicians in identifying children at risk for possibly preventable permanent airflow limitation and its associated morbidities. Longitudinal cohort studies from different parts of the world have consistently reported on the variability in disease onset, presentation, and remission of childhood asthma (3,4). These studies provide an opportunity to define the life course implications of this highly heterogenous disease, which has episodic clinical presentation associated with subclinical to overt clinical evidence of airflow limitation.

In a recent study published in The Lancet Child & Adolescent Health (5), Ullah et al. took advantage of longitudinal data from six such longitudinal cohorts to elucidate the relationship of highest lung function during childhood, defined as ages 8–10 years, and adolescence, ages 15–18 years, with its impairment in early adulthood, defined as ages 20–26 years. The patterns of pulmonary function across these ages were examined in four discovery cohorts and validated in the other two cohorts (5). The discovery cohorts comprised of over 8,000 children with repeated spirometry assessments, including close to 20,000 measurements of forced expiratory volumes in one second (FEV₁) and forced vital capacity (FVC). To identify previously unknown trajectories for airflow limitation in childhood asthma, they employed an unsupervised data-driven methodology and identified four airflow trajectories over time with distinct temporal patterns of change in airflow limitation—these included (I) normal airflow which was present in 81% of the participants; (II) persistent airflow obstruction present in 16%; (III) worsening airflow obstruction, present in 2%; and (IV) improving airflow obstruction present in another 2% of the participants. While normal airflow and persistent airflow obstruction showed no significant changes over time, there was ~10% difference in both FEV₁ and FEV₁/FVC ratio between there groups, suggesting a substantive burden of lower lung function among those with persistent airflow obstruction. Change among those with worsening airflow obstruction and improving airflow obstruction were more dynamic and occurred to differing extents across ages. While transition from non-obstructive to obstructive in the worsening airflow trajectory group was most prominent from adolescence (15–18 years) into early adulthood (20–26 years), transition from obstructive to non-obstructive in the improving airflow trajectory group peaked between middle childhood (8–10 years) and adolescence.

The authors further explored the characteristics of children in these two categories of improving and worsening airflow limitation with the goal to improve risk stratification. Among the several known risk factors relevant to disease and to the trajectories, the authors included gestational age, birthweight, current wheeze, asthma diagnosis, current sensitisation, and body mass index (BMI) for the children whose lung function was included in the analysis. Given the role of parental factors to offspring respiratory health, they also included parental information such as presence of asthma, smoking, maternal age, as well as smoking during pregnancy. As expected, higher gestational age was associated with lower risk of persistent airflow obstruction. From the perspective of modifiable factors, lower birthweight in absence of maternal smoking was associated with improving lung function, whereas the presence of maternal smoking associated it with worsening obstruction.

There were additional intriguing and highly relevant associations of airflow obstruction with the child’s BMI that differed by their wheezing status. Higher BMI at first lung function test was associated with higher likelihood of persistent airflow obstruction, an association that was present in context of wheeze; asthma diagnosis was associated with worsening airflow obstruction. When the effects of varying combinations of these conditions were investigated, among wheezing children, higher BMI decreased the likelihood, while lower BMI increased the likelihood of improvement in airflow obstruction that corresponding to decrease in wheezing. This differed from children without wheeze, in whom higher BMI increased the likelihood of improvement of airflow obstruction. Similar interactions were observed between birthweight, BMI, and asthma diagnosis: low birthweight in conjunction with higher BMI and current asthma diagnosis reduced the likelihood of improvement of airflow obstruction. In contrast, children with low birthweight and higher BMI at first lung function assessment but no current asthma diagnosis had a greater chance of airflow improvement. This complex interplay between birth weight, BMI, wheezing status, and age of the child highlights the dynamic importance of natal factors contributing to birth weight, timing of weight gain during childhood, its association with wheeze, and asthma diagnosis. Together, these trajectories suggest that children born with low birth weight in absence of maternal smoking and absence of wheeze or with high BMI at time of first testing will do well in the long run. Even those with high BMI will do well as long as there is no wheeze or asthma diagnosis. The subset with high BMI and wheeze and asthma diagnosis are the ones that will be least likely to improve their airflow obstruction.

Obesity and its association with childhood asthma is well-documented, yet complex. Obesity has a negative impact via its mechanical effect, non-type 2 inflammation, and metabolic dysregulation on pulmonary function, which all increase the risk of asthma development (6). Ullah et al. suggest that higher BMI in children without asthma or wheeze is associated with improvement in airflow obstruction, which is most apparent between childhood and adolescence. In contrast higher BMI in the presence of wheeze or asthma decreases their likelihood of improvement in airflow obstruction, which is most apparent between adolescence and early adulthood. While these observations suggest a beneficial effect of high BMI on lung function among those without wheeze, they also identify timepoints in a child’s life when interventions such as weight loss would be effective. However, a limitation of the study is lack of clarity on whether higher BMI preceded wheeze and/or asthma or vice versa. Another potential limitation is the overlap between asthma diagnosis and reduced FEV1/FVC ratio. However, given that lung function is preserved for a long time after the diagnosis of asthma, asthma diagnosis is substantially dependent on clinical symptoms and treatment response, and therefore assessing its relationship with lung function trajectories remains valid and informative. Despite these limitations, their observations verify the reported beneficial effects of weight loss in children on asthma, including improvements in asthma control and improvements in asthma-related quality of life (7). In addition to the dynamic relationship with body weight, Ullah and colleagues assessed type 2 inflammation through allergic sensitisation and found an increased risk of persistent airflow obstruction with allergic sensitisation at middle childhood. While the findings are also highly relevant, the assessment of type 2 inflammation solely by allergic sensitization identifies a missed opportunity to assess the association with elevated blood eosinophils, fraction of exhaled nitric oxide (FeNO) and/or immunoglobulin E (IgE) levels that have all been associated with significant impairment of lung function in adults with asthma (8).

Put together, the study by Ullah et al. has several strengths. The findings align with the clinical experience of lung function variability throughout childhood and adolescence and shed much needed light on the specific patterns associated with adverse outcomes. Moreover, their study reinforces the importance of unsupervised data-driven methodologies. To analyse the data and uncover potential hidden patterns in lung function changes over time, Ullah et al. applied a data-driven approach using cluster analysis. This method gained impasse within the field of asthma in 2008 when Haldar et al. classified an adult population with asthma into clusters based on disease relevant characteristics such as age, sex, BMI, and pulmonary function (9). Data-driven approaches, in which a population is distributed into clusters based on disease relevant traits, have provided the scientific community with new insights into the heterogeneity of asthma in an otherwise complex and multidimensional disease (9,10).

While not using a data-driven approach, McGeachie et al. investigated whether childhood asthma affects long-term lung function, identifying four clinically defined FEV₁ trajectories based on visual inspection of lung function curves (11). Extending the relevance of findings reported by Ullah et al., McGeachie et al. had identified four distinct trajectories among which individuals with persistent asthma and reduced lung function in childhood had an increased risk of developing chronic obstructive pulmonary disease (COPD) in early adulthood (11). This complements the work by Bui et al., who identified six lung function trajectories from childhood to adulthood using latent class modeling and showed how early-life risk factors increased risk of developing COPD later in life (12). The detailed time course analysis by Ullah et al. complements the observations by McGeachie et al. and highlight specific age ranges in childhood when interventions on body weight as well as asthma management would have the greatest impact in mitigating persistence of airflow obstruction.

Given the large cohorts inclusive of populations from the United Kingdom, Netherlands, and Sweden, it was reassuring that the vast majority of children had normal lung function at three different crucial stages in life. However, Ullah et al. highlight that up to a sixth of the population lives with persistent airflow obstruction, and dynamism to airflow trajectories was only in less than 4% of the cohorts. This finding suggests that, for many individuals, lung function trajectories may be established well before the age of 8–10 years (“middle childhood”). This raises a critical question: how can children who follow the persistent airflow obstruction trajectory be identified in a timely manner? Currently, spirometry is considered the gold standard, however, spirometry is rarely employed in preschool-aged children due to challenges with cooperation and technique. Although not assessed in this study, we propose that presence of airflow obstruction at middle childhood that persists into early adulthood likely started in preschool years. Recent advancement with development of impulse oscillometry that measures airflow resistance and pressure across a range of frequencies using tidal volume breathing (13), offers a promising option to assess lung function in preschool-aged children, particularly among those that were born at low birthweight and/or have high BMI with concurrent wheeze and/or asthma in early years with the goal to identify airflow obstruction early and monitor its persistence into middle childhood and adolescence. Yet, impulse oscillometry remains sparsely investigated; future research on its contribution to defining dynamism of airflow obstruction is needed. We speculate that impulse oscillometry could play a role in mapping new lung function trajectories for preschool-aged children in the near future.

Furthermore, the findings by Ullah et al. highlight the dynamic nature of lung function trajectories from childhood into adulthood, and highlight that meaningful change is dependent on deeper understanding of each child’s individual trajectory in the context of his or her own co-existent risk factors. The knowledge gained from these trajectories will enable physicians to identify individuals who are at increased risk of “unhealthy aging” so that these children could benefit from early intervention. Two aspects that Ullah et al. emphasize is the importance of the diagnosis of asthma, and birthweight and BMI. As our understanding of modifiable traits increases, it is reasonable to assume that close monitoring of a child’s airflow trajectory could help improve asthma control, which is linked with alleviation of persistent wheezing (14). Enhancing asthma control by targeting modifiable traits will together reduce the impact of persistent wheezing.

In summary, the study by Ullah et al. highlights the value and necessity of unsupervised data-driven methodologies in understanding lung function trajectories followed by the importance of elucidating the role of known biological variables such as wheeze and body weight at different stages in life. As our knowledge of these trajectories improves, it will directly inform opportunities to improve long term lung health and preventative strategies for children with asthma.

Supplementary

The article’s supplementary files as

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.