Cluster analysis in phenotyping a Portuguese population

A Systematic Review of Asthma Phenotypes Derived by Data-Driven Methods

Classification of asthma phenotypes has a potentially relevant impact on the clinical management of the disease. Methods for statistical classification without a priori assumptions (data-driven approaches) may contribute to developing a better comprehension of trait heterogeneity in disease phenotyping. This study aimed to summarize and characterize asthma phenotypes derived by data-driven methods. We performed a systematic review using three scientific databases, following Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) criteria. We included studies reporting adult asthma phenotypes derived by data-driven methods using easily accessible variables in clinical practice. Two independent reviewers assessed studies. The methodological quality of included primary studies was assessed using the ROBINS-I tool. We retrieved 7446 results and included 68 studies of which 65% (n = 44) used data from specialized centers and 53% (n = 36) evaluated the consistency of phenotypes. The most frequent data-driven method was hierarchical cluster analysis (n = 19). Three major asthma-related domains of easily measurable clinical variables used for phenotyping were identified: personal (n = 49), functional (n = 48) and clinical (n = 47). The identified asthma phenotypes varied according to the sample’s characteristics, variables included in the model, and data availability. Overall, the most frequent phenotypes were related to atopy, gender, and severe disease. This review shows a large variability of asthma phenotypes derived from data-driven methods. Further research should include more population-based samples and assess longitudinal consistency of data-driven phenotypes.

Challenges of Clustering Multimodal Clinical Data: Review of Applications in Asthma Subtyping

Background

In the current era of personalized medicine, there is increasing interest in understanding the heterogeneity in disease populations. Cluster analysis is a method commonly used to identify subtypes in heterogeneous disease populations. The clinical data used in such applications are typically multimodal, which can make the application of traditional cluster analysis methods challenging.

Objective

This study aimed to review the research literature on the application of clustering multimodal clinical data to identify asthma subtypes. We assessed common problems and shortcomings in the application of cluster analysis methods in determining asthma subtypes, such that they can be brought to the attention of the research community and avoided in future studies.

Methods

We searched PubMed and Scopus bibliographic databases with terms related to cluster analysis and asthma to identify studies that applied dissimilarity-based cluster analysis methods. We recorded the analytic methods used in each study at each step of the cluster analysis process.

Results

Our literature search identified 63 studies that applied cluster analysis to multimodal clinical data to identify asthma subtypes. The features fed into the cluster algorithms were of a mixed type in 47 (75%) studies and continuous in 12 (19%), and the feature type was unclear in the remaining 4 (6%) studies. A total of 23 (37%) studies used hierarchical clustering with Ward linkage, and 22 (35%) studies used k-means clustering. Of these 45 studies, 39 had mixed-type features, but only 5 specified dissimilarity measures that could handle mixed-type features. A further 9 (14%) studies used a preclustering step to create small clusters to feed on a hierarchical method. The original sample sizes in these 9 studies ranged from 84 to 349. The remaining studies used hierarchical clustering with other linkages (n=3), medoid-based methods (n=3), spectral clustering (n=1), and multiple kernel k-means clustering (n=1), and in 1 study, the methods were unclear. Of 63 studies, 54 (86%) explained the methods used to determine the number of clusters, 24 (38%) studies tested the quality of their cluster solution, and 11 (17%) studies tested the stability of their solution. Reporting of the cluster analysis was generally poor in terms of the methods employed and their justification.

Conclusions

This review highlights common issues in the application of cluster analysis to multimodal clinical data to identify asthma subtypes. Some of these issues were related to the multimodal nature of the data, but many were more general issues in the application of cluster analysis. Although cluster analysis may be a useful tool for investigating disease subtypes, we recommend that future studies carefully consider the implications of clustering multimodal data, the cluster analysis process itself, and the reporting of methods to facilitate replication and interpretation of findings.

Development and validation of exhaled breath condensate microRNAs to identify and endotype asthma in children

Detection and quantification of microRNAs (miRNAs) in exhaled breath condensate (EBC) has been poorly explored. Therefore we aimed to assess miRNAs in EBC as potential biomarkers to diagnose and endotype asthma in school aged children. In a cross sectional, nested case control study, all the asthmatic children (n = 71) and a random sample of controls (n = 115), aged 7 to 12 years, attending 71 classrooms from 20 local schools were selected and arbitrarily allocated to the development or validation set. Participants underwent skin-prick testing, spirometry with bronchodilation, had exhaled level of nitric oxide determined and EBC collected. Based on previous studies eleven miRNAs were chosen and analyzed in EBC by reverse transcription-quantitative real-time PCR. Principal component analysis was applied to identify miRNAs profiles and associations were estimated using regression models. In the development set (n = 89) two clusters of miRNAs were identified. After adjustments, cluster 1 and three of its clustered miRNAs, miR-126-3p, miR-133a-3p and miR-145-5p were positively associated with asthma. Moreover miR-21-5p was negatively associated with symptomatic asthma and positively associated with positive bronchodilation without symptoms. An association was also found between miR-126-3p, cluster 2 and one of its clustered miRNA, miR-146-5p, with higher FEF25-75 reversibility. These findings were confirmed in the validation set (n = 97) where two identical clusters of miRNAs were identified. Additional significant associations were observed between miR-155-5p with symptomatic asthma, negative bronchodilation with symptoms and positive bronchodilation without symptoms. We showed that microRNAs can be measured in EBC of children and may be used as potential biomarkers of asthma, assisting asthma endotype establishment.

Discovering Pediatric Asthma Phenotypes on the Basis of Response to Controller Medication Using Machine Learning

RATIONALE

Pediatric asthma has variable underlying inflammation and symptom control. Approaches to addressing this heterogeneity, such as clustering methods to find phenotypes and predict outcomes, have been investigated. However, clustering based on the relationship between treatment and clinical outcome has not been performed, and machine learning approaches for long-term outcome prediction in pediatric asthma have not been studied in depth.

OBJECTIVES

Our objectives were to use our novel machine learning algorithm, predictor pursuit (PP), to discover pediatric asthma phenotypes on the basis of asthma control in response to controller medications, to predict longitudinal asthma control among children with asthma, and to identify features associated with asthma control within each discovered pediatric phenotype.

METHODS

We applied PP to the Childhood Asthma Management Program study data (n = 1,019) to discover phenotypes on the basis of asthma control between assigned controller therapy groups (budesonide vs. nedocromil). We confirmed PP's ability to discover phenotypes using the Asthma Clinical Research Network/Childhood Asthma Research and Education network data. We next predicted children's asthma control over time and compared PP's performance with that of traditional prediction methods. Last, we identified clinical features most correlated with asthma control in the discovered phenotypes.

RESULTS

Four phenotypes were discovered in both datasets: allergic not obese (A⁺/O^-), obese not allergic (A^-/O⁺), allergic and obese (A⁺/O⁺), and not allergic not obese (A^-/O^-). Of the children with well-controlled asthma in the Childhood Asthma Management Program dataset, we found more nonobese children treated with budesonide than with nedocromil (P = 0.015) and more obese children treated with nedocromil than with budesonide (P = 0.008). Within the obese group, more A⁺/O⁺ children's asthma was well controlled with nedocromil than with budesonide (P = 0.022) or with placebo (P = 0.011). The PP algorithm performed significantly better (P < 0.001) than traditional machine learning algorithms for both short- and long-term asthma control prediction. Asthma control and bronchodilator response were the features most predictive of short-term asthma control, regardless of type of controller medication or phenotype. Bronchodilator response and serum eosinophils were the most predictive features of asthma control, regardless of type of controller medication or phenotype.

CONCLUSIONS

Advanced statistical machine learning approaches can be powerful tools for discovery of phenotypes based on treatment response and can aid in asthma control prediction in complex medical conditions such as asthma.

Omalizumab for Severe Asthma: Beyond Allergic Asthma

Different subsets of asthma patients may be recognized according to the exposure trigger and the frequency and severity of clinical signs and symptoms. Regarding the exposure trigger, generally asthma can be classified as allergic (or atopic) and nonallergic (or nonatopic). Allergic and nonallergic asthma are distinguished by the presence or absence of clinical allergic reaction and in vitro IgE response to specific aeroallergens. The mechanisms of allergic asthma have been extensively studied with major advances in the last two decades. Nonallergic asthma is characterized by its apparent independence from allergen exposure and sensitization and a higher degree of severity, but little is known regarding the underlying mechanisms. Clinically, allergic and nonallergic asthma are virtually indistinguishable in exacerbations, although exacerbation following allergen exposure is typical of allergic asthma. Although they both show several distinct clinical phenotypes and different biomarkers, there are no ideal biomarkers to stratify asthma phenotypes and guide therapy in clinical practice. Nevertheless, some biomarkers may be helpful to select subsets of atopic patients which might benefit from biologic agents, such as omalizumab. Patients with severe asthma, uncontrolled besides optimal treatment, notwithstanding nonatopic, may also benefit from omalizumab therapy, although currently there are no randomized double-blind placebo controlled clinical trials to support this suggestion. However, omalizumab discontinuation according to each patient's response to therapy and pharmacoeconomical analysis are questions that remain to be answered.

Identification of Asthma Subtypes Using Clustering Methodologies

Asthma is a heterogeneous disease comprising a number of subtypes which may be caused by different pathophysiologic mechanisms (sometimes referred to as endotypes) but may share similar observed characteristics (phenotypes). The use of unsupervised clustering in adult and paediatric populations has identified subtypes of asthma based on observable characteristics such as symptoms, lung function, atopy, eosinophilia, obesity, and age of onset. Here we describe different clustering methods and demonstrate their contributions to our understanding of the spectrum of asthma syndrome. Precise identification of asthma subtypes and their pathophysiological mechanisms may lead to stratification of patients, thus enabling more precise therapeutic and prevention approaches.