A robust mRNA signature obtained via recursive ensemble feature selection predicts the responsiveness of omalizumab in moderate-to-severe asthma

Abstract

Background: Not being well controlled by therapy with inhaled corticosteroids and long-acting β2 agonist bronchodilators is a major concern for severe-asthma patients. The current treatment option for these patients is the use of biologicals such as anti-IgE treatment, omalizumab, as an add-on therapy. Despite the accepted use of omalizumab, patients do not always benefit from it. Therefore, there is a need to identify reliable biomarkers as predictors of omalizumab response.

Methods: Two novel computational algorithms, machine-learning based Recursive Ensemble Feature Selection (REFS) and rule-based algorithm Logic Explainable Networks (LEN), were used on open accessible mRNA expression data from moderate-to-severe asthma patients to identify genes as predictors of omalizumab response.

Results: With REFS, the number of features was reduced from 28,402 genes to 5 genes while obtaining a cross-validated accuracy of 0.975. The 5 responsiveness predictive genes encode the following proteins: Coiled-coil domain- containing protein 113 (CCDC113), Solute Carrier Family 26 Member 8 (SLC26A), Protein Phosphatase 1 Regulatory Subunit 3D (PPP1R3D), C-Type lectin Domain Family 4 member C (CLEC4C) and LOC100131780 (not annotated). The LEN algorithm found 4 identical genes with REFS: CCDC113, SLC26A8 PPP1R3D and LOC100131780. Literature research showed that the 4 identified responsiveness predicting genes are associated with mucosal immunity, cell metabolism, and airway remodeling.

Conclusion and clinical relevance: Both computational methods show 4 identical genes as predictors of omalizumab response in moderate-to-severe asthma patients. The obtained high accuracy indicates that our approach has potential in clinical settings. Future studies in relevant cohort data should validate our computational approach.

Keywords: anti-IgE; asthma; biomarker; machine-learning; omalizumab.

FIGURE 1

Heatmap of normalized gene expressions for the 5 selected genes in all samples identified by the Recursive Ensemble Feature Selection (REFS) algorithm (Section 2.3.1). The heatmap legend displays a color gradient range where −2 denotes the lowest gene expression and 4 the highest gene expression. Merely from visual inspection, samples can be differentiated into two groups: responders and non‐responders to omalizumab treatment.

FIGURE 2

Receiver Operating Characteristic (ROC) curve over all 8 classifiers from Recursive Ensemble Feature Selection (REFS) to validate the identified 5‐gene signature. The ROC curve shows the binarization threshold from 0 (all moderate‐to‐severe asthma patients as omalizumab responders and both the TPR and FPR = 1; upper right corner of ROC) to 1 (all moderate‐severe asthma patients classified as non‐responders to omalizumab and both TPR and FPR = 0; lower left corner of ROC). The area under the curve (AUC) is the area in the plot which stays under the ROC curve. The Passive aggressive classifier which produced the blue ROC curve shows the best predictive accuracy as it covers a larger area compared to the straight ROC curve with the random classifier (red dashed line).

FIGURE 3

Summarized results of the two algorithms predicting treatment responsiveness of omalizumab in moderate‐to‐severe asthma. Whole blood mRNA expression profiles in samples collected day 0 (1 week before the start of the treatment) were used for the Recursive Ensemble Feature Selection (REFS) analysis (Section 2.3.1). In total, 40 moderate‐to‐severe asthmatic patients, n = 30 responders, and n = 10 non‐responders were included. For LEN analysis (Section 2.4.1), n = 17 healthy controls were also included. With REFS, five independent responsiveness‐predictive genes were identified, whereas rule‐based LEN identified three gene groups that predicted responsiveness. Comparing both approaches, an overlap of four genes was found. The relationship between responder status (R/NR) is shown in the heatmap. The mRNA expression of responders compared with healthy controls is shown in the table.