A framework for individualized prediction of the rate and severity of asthma exacerbations

Proposal
1630

Title of Proposed Research

Lead Researcher

Mohsen Sadatsafavi

Affiliation

Faculty of Pharmaceutical Sciences and Faculty of Medicine, University of British Columbia

Funding Source

The investigators of the proposed study are Theme Leaders of the Health Economics Platform of CRRN with a secured share of the CRRN budget.

Potential Conflicts of Interest

None

Data Sharing Agreement Date

12 Jun 2017

Lay Summary

Asthma is a chronic disease of the lungs that affects more than 300 million people across the world. Daily symptoms such as cough and shortness of breath are common features of asthma. In addition, periods of high disease activity, referred to as asthma exacerbations (or ‘lung attacks') exert a high toll on patients as well as healthcare resources. Exacerbations are the main reasons for asthma-related visits to emergency departments, hospitalizations, or death.

Understanding the differences among asthma patients in their pattern of exacerbations will be important in many ways. Such understanding not only improve our understanding of the disease mechanisms, but also help us design ‘individualized' asthma management plans that take into account each patient's unique features.
Asthma patients might be different in at least two important ways about exacerbations. First, some patients frequently exacerbate, while the others do not. Second, there might be some patients in whom exacerbations are mostly mild, whereas some other patients might have a tendency towards experiencing severe exacerbations. The overall burden of exacerbations in an asthma patient is indeed affected by how often they experience them, and how severe they are when they occur.

Unfortunately, there is currently little information about how the occurrence and severity of exacerbations are different among different patients. Our research group has recently developed a method that can be used for the following purposes:

• Identifying the full spectrum of variability in rate and severity of exacerbations

• Testing whether those who exacerbate more often also tend to experience more severe exacerbations

We proposed to use the high quality data of the DREAM and MENSA studies to address these questions for the first time in asthma. This study builds on our success in addressing similar questions in patients with chronic obstructive pulmonary disease (COPD).

This study will provide important insights into how patients with asthma differ from each other in important aspects of their disease (especially in terms of exacerbations). Answer to the question “whether those who exacerbate a lot tend to have more severe exacerbation” is also important in our understanding of asthma. Results from this study can be used to make individualized prediction models of asthma exacerbations.
Findings from this study will be presented at academic conferences and will be published in peer-reviewed journals. Once this project is finished, we will consider using other data to validate the mathematical equations from this work that can predict the pattern of future exacerbations in asthma patients.

Overall, this study has the potential to significantly contribute to the understanding of asthma. Also, it has high potential to impact patient care by enabling care providers to make patient-specific predictions about the future patterns of asthma exacerbations.

Study Data Provided

[{ "PostingID": 3792, "Title": "GSK-MEA112997", "Description": "A multicenter, randomized, double-blind, placebo-controlled, parallel group, dose ranging study to determine the effect of mepolizumab on exacerbation rates in subjects with severe uncontrolled refractory asthma" },{ "PostingID": 3793, "Title": "GSK-MEA115588", "Description": "MEA115588 A randomised, double-blind, double-dummy, placebo-controlled, parallel-group, multi-centre study of the efficacy and safety of mepolizumab adjunctive therapy in subjects with severe uncontrolled refractory asthma" },{ "PostingID": 3794, "Title": "GSK-MEA115575", "Description": "MEA115575: A Randomised, Double-Blind, Placebo-Controlled, Parallel-Group, Multicenter Study of Mepolizumab Adjunctive Therapy to Reduce Steroid Use in Subjects with Severe Refractory Asthma" },{ "PostingID": 4514, "Title": "GSK-MEA115661", "Description": "MEA115661: A Multi-centre, Open-label, Long-term Safety Study of Mepolizumab in Asthmatic Subjects who participated in the MEA115588 or MEA115575 trials" }]

Statistical Analysis Plan

This study will be a secondary analysis of the MENSA and DREAM studies. Our objectives are as follows:1) To quantify between-individual variability (heterogeneity) in the rate and severity of asthma exacerbationsHypothesis: individuals vary widely in both the frequency of experiencing severe exacerbations (rate) and tendency to experience severe (versus milder) exacerbations (severity).2) To examine whether there is a correlation between the rate and severity of exacerbations.Hypothesis: individuals who tend to experience more exacerbations also tend to have a higher ratio of severe to total exacerbationsThis study builds on the novel methodology that our team has developed (Sadatsafavi et. al., American Journal of Epidemiology, 2016, doi:10.1093/aje/kww085). In this work, using the data from the MACRO study (Albert et al. N Engl J Med. 2011;365(8):689-698), we have used this methodology to confirm that there is tremendous heterogeneity in rate of chronic obstructive pulmonary disease (COPD) exacerbations, to show, for the first time, that individual COPD patients also differ in their tendency towards experiencing severe (v. mild) exacerbations, and have tested the hypothesis that individuals who experience more frequent exacerbations tend to have higher ratio of severe to total exacerbations. Using the requested data, we plan to perform the same analyses in asthma.To achieve this objective, we will merge the data from the MENSA and DREAM studies to create a unified dataset containing the following variables: anonymized patient id, demographic variables at baseline (e.g., age and sex and history of asthma, lung function metrics, quality of life and functional impairment levels at baseline), follow-up time (in years or days), as well as timing (from the randomization date) and severity of each asthma exacerbation.We will specify a joint frailty-logistic model to simultaneously estimate the association between clinical features and the rate and severity of exacerbations. The rate component will be an Accelerated Failure Time model with appropriate survival function (tested based on goodness-of-fit). The severity component will be a random-effects ordinal regression variable. Details of the model structure and the likelihood function for the joint model can be found in our open-access publication (Sadatsafavi et. al. American Journal of Epidemiology, 2016, doi:10.1093/aje/kww085: available from http://aje.oxfordjournals.org/content/early/2016/10/13/aje.kww085.full). Both the frailty and ordinal model incorporate zero-mean, normally distributed random-effects terms. The random-effect term for the former captures heterogeneity (over and beyond what is explained by observable characteristics) in the background rate of exacerbations; the latter captures heterogeneity in the ratio of severe to total exacerbations.The model will be fitted using non-linear mixed model optimizers (SAS PROC NLMIXED). We have fully developed and validated the code and have made it available alongside the published paper as supplementary material. We will adapt the code for the present context and will test the convergence of the optimization routine through examining different starting values for model parameters as well as testing the routine on synthetic data with known true parameter values. The model provides maximum likelihood estimates (and their covariance matrix) for all regression coefficients, as well as the joint distribution of the two random-effect terms governing heterogeneity in rate and severity of asthma exacerbations.To examine the extent of heterogeneity (objective 1), we will calculate the Coefficient of Variation (CV), defined as the ratio of the standard deviation to the mean, for model-estimated individualized exacerbation rate as well as the proportion of severe to total exacerbations. We will also determine the lower and upper bounds on these quantities that contained 95% of the sample. This framework will also be used to test the hypothesis outlined in objective 2. The two random-effect terms are governed by three parameters (two variance and one covariance parameters). A positive (negative) covariance indicates that individuals with a higher rate of exacerbation tend to have a higher (lower) risk of their exacerbations being severe. Combined, the DREAM and MENSA studies will provide us with data on more than 1,200 patients with more than 1,200 exacerbations. These numbers are in the range of the numbers from the MACRO study that we used for fitting a similar model for COPD exacerbations (1,600 exacerbations from 1,100 individuals). There, our estimation of the effect of the treatment variable on the rate was HR (hazard ratio)=0.77, with 95% CI 0.67 - 0.89. For the effect of treatment assignment on severity, we estimated an OR (odds ratio)=0.93 with 95% CI 0.62 - 1.38. We considered the level of sampling variability around these and other estimates of model parameters to be acceptable. The correlation coefficient for the random-effect terms for rate and severity was -0.18 (95% CI -0.40 - 0.03, P=0.099), again providing a relatively robust estimate of the likely range of this value.

Publication Citation

https://www.atsjournals.org/doi/abs/10.1164/ajrccm-conference.2018.197.1_MeetingAbstracts.A7412