Ventilation defects on hyperpolarized helium-3 MRI in asthma are predictive of 2-year exacerbation frequency

Abstract

Background: There is an unmet need for an objective biomarker to predict asthma exacerbations.

Objective: Our aim was to assess the ventilation defect percent (VDP) on hyperpolarized helium-3 magnetic resonance imaging as a predictor of exacerbation frequency following imaging.

Methods: Subjects underwent hyperpolarized helium-3 and conventional clinical measurements, including pulmonary function tests, during a period of disease stability, and exacerbations were recorded prospectively over the following 2 years. We used a Poisson regression tree model to estimate an optimal VDP threshold for classifying subjects into high- versus low-exacerbation groups and then used statistical regression to compare this VDP threshold against conventional clinical measures as predictors of exacerbations.

Results: A total of 67 individuals with asthma (27 males and 40 females, 28 with mild-to-moderate asthma and 39 with severe asthma) had a median VDP of 3.75% (1.2% [first quartile]-7.9% [third quartile]). An optimal VDP threshold of 4.28% was selected on the basis of the maximum likelihood estimation of the regression tree model. Subjects with a VDP greater than 4.28% (n = 32) had a median of 1.5 exacerbations versus 0.0 for subjects with a VDP less than 4.28% (n = 35). In a stepwise multivariate regression model, a VDP greater than 4.28% was associated with an exacerbation incidence rate ratio of 2.5 (95% CI = 1.3-4.7) versus a VDP less than or equal to 4.28%. However, once individual medical history was included in the model, VDP was no longer significant. Nonetheless, VDP may provide an objective and complementary quantitative marker of individual exacerbation risk that is useful for monitoring individual change in disease status, selecting patients for therapy, and assessing treatment response.

Conclusion: VDP measured with magnetic resonance imaging shows promise as a biomarker of prospective asthma exacerbations.

Keywords: Asthma; airway obstruction; functional; magnetic resonance imaging; patient outcomes assessment.

Figure 1.

Example ³He MRI images from subjects with a range of clinical patterns, VDP, and exacerbations in the two years following baseline.

Figure 2.

VDP in subjects with zero (left) and one or more (right) two-year exacerbations. Median and quartile [1Q – 3Q] range of VDP is 5.5% [3.0% – 9.8%] in severe subjects vs. 2.2% [0.9% – 4.9%] in mild/moderate (p = 0.010). Five outlier points not shown.

Figure 3.

Number of exacerbations predicted by regression tree model of VDP vs. exacerbations (solid line) and 95% confidence interval (shaded area). The maximum likelihood estimate for a VDP threshold (4.28%) stratifying low (N = 35) vs. high (N = 32) exacerbators is indicated by the solid vertical line.

Figure 4:

Box and whisker plot of 2-year exacerbation frequency for subjects above and below the VDP threshold. Subjects above the threshold experienced a median of 1.5 exacerbations vs. zero exacerbations for subjects below the threshold (p < 0.001).

Figure 5.

Exacerbation incidence rate ratio (IRR) and 95% confidence intervals for model predictors. Within each predictor, the top line indicates the IRR in the univariate (individual) models, the second line the inclusive multivariate model, and the third line (if present) the parsimonious multivariate model.