Twitter Facebook LinkedIn
Center for Global Research Data

Associations between steroid treatment, functional outcomes and disease milestones among non-ambulatory patients with Duchenne muscular dystrophy (DMD)

Lead Investigator: James Signorovitch, Analysis Group Inc.
Title of Proposal Research: Associations between steroid treatment, functional outcomes and disease milestones among non-ambulatory patients with Duchenne muscular dystrophy (DMD)
Vivli Data Request: 7159, 6634
Funding Source: PTC Therapeutics will fund the proposed research.  PTC Therapeutics will fund Analysis Group to provide consulting services for the proposed research.
Potential Conflicts of Interest: Employee of Analysis Group, a research consultancy which will receive funding from PTC Therapeutics for conducting this research. All the analyses will be carried out by the Analysis Group team.

Summary of the Proposed Research:

Duchenne muscular dystrophy (DMD) is an X-linked recessive disorder characterized by progressive deterioration in skeletal and cardiac muscles due to mutations in the dystrophin gene. Globally, DMD affects approximately one in every 3,500 newborn males. Initial manifestations of the disorder include progressive declines in motor function during childhood that typically culminate in complete loss of independent walking ability by adolescence.[3] Additional impacts include worsening heart and lung function, as well as a deterioration in bone health; over time, these disabling declines collectively contribute to mortality by early adulthood. The median survival for boys with DMD is 21.8 years.

Corticosteroids are the current standard of care for patients with DMD, and have modified the disease by slowing progression of motor and lung functional decline and extending survival. The glucocorticoids in use have been prednisone/prednisolone and deflazacort, an oxazoline derivative of prednisolone. Several comparative studies exist that examine prednisone vs. deflazacort treatment among patients with DMD. However, additional understanding is needed about whether the comparative effectiveness of these steroids is different among subgroups of DMD patients. For instance, older vs. younger patients, patients who have been on steroids longer, and those who have more severe symptoms. This study will examine the association between steroid treatment and selected functional outcomes, body weight, and disease milestones among relevant subgroups of DMD patients. The results of this study could help inform clinical decision making around steroid options for individual patients.

Statistical Analysis Plan

The following analyses will be conducted using the requested DEMAND III data:

Summary of patient characteristics
The following baseline characteristics will summarized for the deflazacort and prednisone/prednisolone groups: age, frequency of use, latest steroid dose, duration of steroid use (considering all historical steroid use), and baseline values of the outcomes. Continuous variables will be summarized using means and standard deviations, while categorical variables will be summarized using proportions.

Mixed model repeated measures analysis
An MMRM analysis will be conducted for each of the functional measures. The mean of the change from baseline score for the functional measure at trial visits over time will be modeled as a function of:
– time (continuous variable in weeks)
– steroid group (deflazacort vs. prednisone/prednisolone)
– prior steroid use duration (<1 year, 1-3 years, > 3 years)
– age (<8 years versus ≥8 years)
– baseline rise time category (≥5 seconds vs <5 seconds)
– baseline value of the functional measure serving as the outcome (continuous)
– interactions between time and each of the other characteristics

Random intercepts and slopes will be included at the patient level. Model specifications will be assessed by evaluating Akaike information criterion (AIC) and Bayesian information criterion (BIC) statistics.

Additionally, figures displaying the average trajectories over time (i.e., from baseline to 48 weeks), stratified by steroid type, will be generated for each outcome.

Based on the fitted model, predicted means (least-squares means) will be obtained, and an estimate of the mean and standard error of the difference in 48-week change between the deflazacort and prednisone/prednisolone groups will be generated for use in the meta-analysis.

The MMRM analyses will be based on all available data for all patients in each of the three trials. No data imputations will be conducted beyond those specified above for patients who have lost the ability to complete functional tests.

Subgroup analyses
Baseline characteristics, as described above, will be summarized by subgroup.

Separate MMRM models will be used to estimate differences in outcomes by steroid group among each subgroup. For each subgroup of interest, interaction terms will be included in the models between the subgroup, steroid group, and time.

The subgroups examined may include:
– Age:
o 4-8 years
o 8+ years
– Steroid duration at baseline
o <1 year
o 1-3 years
o > 3 years
– Rise time at baseline
o ≥ 5 seconds
o < 5 seconds

————————————————————————————-
The following analyses will be conducted using the requested PRO-DMD-01 data:

Summary of patient characteristics
Patient characteristics will be summarized among all patients included in any analysis. Characteristics will be summarized in aggregate for all patients and stratified by steroid type at index visit. Described characteristics will include demographics (e.g., age at first clinic visit, age at index visit, height, weight, BMI, age at DMD diagnosis), steroid-related characteristics (e.g., age at steroid initiation, duration of use, steroid type, steroid regimen), and markers of function (e.g., timed function tests, NSAA), pulmonary (e.g., FVC) and/or cardiac status (e.g., LVEF).

If of interest, the number of patients who do not initiate steroids, patients who switch from steroids to no steroids, and patients who switch from one type of steroid to another will be summarized . Additionally, trajectory plots of steroid dose (mg/kg) by age during patients’ non-ambulatory follow-up period, stratified by index steroid type and color-coded to reflect any switch, will be generated.

Associations between steroid type and longitudinal patient outcomes:
This analysis will characterize how steroid type is associated with outcome levels over time. For this analysis, all patients who meet the sample selection criteria will be included as long as they have at least one visit with the outcome of interest measured. All visits will be included as long as these visits have a measured value for the outcome of interest.

Mixed-effects models will be used to model levels of each outcome, adjusting for covariates recorded at patients’ index visit, tentatively including steroid type (potentially time-varying in the case of switching), pre-index steroid duration, and age at DMD diagnosis. Additionally, for the non-ambulatory models, a flag will be included that indicates whether a patient entered the database already non-ambulatory or whether they lost ambulation during the study. Both this ambulatory status flag and the steroid type variable will be interacted with patients’ age in the regression models. Random intercepts and slopes will be tentatively included at the patient level.

Model specifications will be assessed by evaluating Akaike information criterion (AIC) and Bayesian information criterion (BIC) statistics.

Additionally, figures displaying the average trajectories over time (i.e., patients’ age), stratified by steroid type, will be generated for each outcome.

Associations between steroid type and disease milestones:
This analysis will help to capture longer-term changes in outcomes and differences in disease milestone occurrence between patients on different steroid treatments. For this analysis, all patients who meet the sample selection criteria will be included as long as they have at least one visit with the outcome of interest measured. All visits will be included as long as these visits have a measured value for the outcome of interest.

Mixed-effects logistic models will be used to model the probability of achieving each binary disease milestone, adjusting for covariates recorded at patients’ index visit, including steroid type (potentially time-varying in the case of switching), pre-index steroid duration, and age at DMD diagnosis. Additionally, for non-ambulatory models, a flag will be included that indicates whether a patient entered the database already non-ambulatory or whether they lost ambulation during the study. Both this ambulatory status flag and the steroid type variable will be interacted with patients’ age in the regression models. Random intercepts and slopes will be tentatively included at the patient level.

Model specifications will be assessed by evaluating Akaike information criterion (AIC) and Bayesian information criterion (BIC) statistics.

Additionally, figures displaying adjusted risk over time, stratified by steroid type, will be generated for each outcome.

Requested Studies:

A Prospective Natural History Study of Progression of Physical Impairment, Activity Limitation and Quality of Life in Duchenne Muscular Dystrophy.
Data Contributor: Cure Duchenne
Study ID: NCT01753804
Sponsor ID: PRO-DMD-01

(Note: Additional study added as part of Data Request 7159)

A Phase III, Randomized, Double Blind, Placebo-controlled Clinical Study to Assess the Efficacy and Safety of GSK2402968 in Subjects With Duchenne Muscular Dystrophy
Data Contributor: Cure Duchenne
Study ID: NCT01254019
Sponsor ID: 114044

Public Disclosures:

  1.  Associations Between Steroid Treatment and Clinical Outcomes Among Non-ambulatory Patients with Duchenne Muscular Dystrophy (DMD) (P1-1.Virtual). Craig McDonald, Oscar Mayer, Kan Hor, Jessica Marden, Jonathan Freimark, Henry Lane, Adina Zhang, Molly Frean, Claudio Santos, Richard Able, James Signorovitch. Neurology May 2022, 98 (18 Supplement) 1348; https://n.neurology.org/content/98/18_Supplement/1348
  2.  Associations Between Deflazacort Versus Prednisone/Prednisolone and Markers of Disease Progression in Clinically Important Subgroups of Patients with Duchenne Muscular Dystrophy (P1-1.Virtual). Craig McDonald, Jessica Marden, Henry Lane, Adina Zhang, Ha Nguyen, Claudio Santos, Richard Able, James Signorovitch. Neurology May 2022, 98 (18 Supplement) 1437; https://n.neurology.org/content/98/18_Supplement/1437