Contributions
Abstract: P871
Type: Poster
Abstract Category: Clinical aspects of MS - 4 Natural course
Background: In randomized clinical trials (RCTs) variables collected at baseline may predict the risk of disability progression in MS patients, but the strength of this prediction is typically weak, in part due to short follow up time and low event rate. Therefore, we explored whether prognostic models that incorporate time-dependent changes in key covariates would more strongly predict disability progression.
Objectives: To assess the combination of baseline and time-dependent prognostic variables as predictors of EDSS disability progression in a pooled sample of RCT placebo arms.
Methods: A pooled dataset (n=1,582) of placebo arms from various phase 3 RCTs in MS was used, including ADVANCE (peginterferon beta-1a), AFFIRM (natalizumab), DEFINE and CONFIRM (dimethyl fumarate). A time-dependent Cox proportional hazards regression was used to predict EDSS disability progression confirmed at 24 weeks over 2 years of follow-up. EDSS progression was defined as ≥1.0 point increase in patients with baseline EDSS ≥1.0 or a ≥1.5 point increase from a baseline EDSS of 0. Baseline covariates were age, sex, years since symptom onset, number of relapses prior to study, months since pre-study relapse, prior MS treatment, MSFC-4, SF-36 PCS and MCS, T1 and T2 lesion volume, Gd+ lesions, and brain volume z-score. MSFC-4 progression over time and changes in SF-36 PCS and MCS during follow-up were considered time-dependent predictors. Patients' follow-up time was censored either at EDSS progression or at last follow-up visit.
Results: Overall 188 (11.8%) of 1,582 patients revealed EDSS progression confirmed at 24 weeks in these 2-year studies. Strongest baseline predictors were (HR, 95%CI, p) relapses in 3 years prior to study (1.19, 1.06-1.33, < 0.0001), SF-36 PCS (0.83, 0.75-0.91, < 0.0001), EDSS (0.74, 0.63-0.87, < 0.0001), PASAT (0.92, 0.86-0.99, 0.017), Gd+ lesions (1.03, 1.00-1.06, 0.027), and T25FW (1.06, 1.00-1.13, 0.061); while time-dependent predictors were T25FW (3.17, 2.13-4.72, < 0.0001), 5 point-change in SF-36 PCS (0.67, 0.60-0.75, < 0.0001) and MCS (0.88, 0.81-0.96, 0.005).
Conclusions: We demonstrate the impact of some key predictors over time. Further model building attempts and validation are necessary to better understand the role of changes over time as for the appropriate definition and validation of thresholds for progression and the generalization of flexible prognostic and/or treatment response scoring systems.
Disclosure: Supported by: Biogen, Inc.
Author disclosures:
Bernd C. Kieseier, Ulrich Freudensprung, Robert Hyde, Carl de Moor and Fabio Pellegrini are Biogen employees and hold stock/stock options in the company.
Abstract: P871
Type: Poster
Abstract Category: Clinical aspects of MS - 4 Natural course
Background: In randomized clinical trials (RCTs) variables collected at baseline may predict the risk of disability progression in MS patients, but the strength of this prediction is typically weak, in part due to short follow up time and low event rate. Therefore, we explored whether prognostic models that incorporate time-dependent changes in key covariates would more strongly predict disability progression.
Objectives: To assess the combination of baseline and time-dependent prognostic variables as predictors of EDSS disability progression in a pooled sample of RCT placebo arms.
Methods: A pooled dataset (n=1,582) of placebo arms from various phase 3 RCTs in MS was used, including ADVANCE (peginterferon beta-1a), AFFIRM (natalizumab), DEFINE and CONFIRM (dimethyl fumarate). A time-dependent Cox proportional hazards regression was used to predict EDSS disability progression confirmed at 24 weeks over 2 years of follow-up. EDSS progression was defined as ≥1.0 point increase in patients with baseline EDSS ≥1.0 or a ≥1.5 point increase from a baseline EDSS of 0. Baseline covariates were age, sex, years since symptom onset, number of relapses prior to study, months since pre-study relapse, prior MS treatment, MSFC-4, SF-36 PCS and MCS, T1 and T2 lesion volume, Gd+ lesions, and brain volume z-score. MSFC-4 progression over time and changes in SF-36 PCS and MCS during follow-up were considered time-dependent predictors. Patients' follow-up time was censored either at EDSS progression or at last follow-up visit.
Results: Overall 188 (11.8%) of 1,582 patients revealed EDSS progression confirmed at 24 weeks in these 2-year studies. Strongest baseline predictors were (HR, 95%CI, p) relapses in 3 years prior to study (1.19, 1.06-1.33, < 0.0001), SF-36 PCS (0.83, 0.75-0.91, < 0.0001), EDSS (0.74, 0.63-0.87, < 0.0001), PASAT (0.92, 0.86-0.99, 0.017), Gd+ lesions (1.03, 1.00-1.06, 0.027), and T25FW (1.06, 1.00-1.13, 0.061); while time-dependent predictors were T25FW (3.17, 2.13-4.72, < 0.0001), 5 point-change in SF-36 PCS (0.67, 0.60-0.75, < 0.0001) and MCS (0.88, 0.81-0.96, 0.005).
Conclusions: We demonstrate the impact of some key predictors over time. Further model building attempts and validation are necessary to better understand the role of changes over time as for the appropriate definition and validation of thresholds for progression and the generalization of flexible prognostic and/or treatment response scoring systems.
Disclosure: Supported by: Biogen, Inc.
Author disclosures:
Bernd C. Kieseier, Ulrich Freudensprung, Robert Hyde, Carl de Moor and Fabio Pellegrini are Biogen employees and hold stock/stock options in the company.