Predicting Disability Following Clinically Isolated Syndrome

BERLIN—Smoking and low vitamin D levels are key modifiable prognostic factors that show a significant interaction with risk of Expanded Disability Status Scale (EDSS) progression in patients with clinically isolated syndrome (CIS), according to a study presented at ECTRIMS 2018. This finding suggests potential preventive strategies for avoiding disability, said Susana Otero-Romero, MD, PhD, an epidemiologist at the Multiple Sclerosis Center of Catalonia (CEMCAT) at University Hospital Vall d’Hebron in Barcelona, and her research colleagues.

A dynamic model for predicting long-term prognosis incorporating age, sex, topography of CIS, oligoclonal bands, and number of T2 lesions was designed by Dr. Otero-Romero and colleagues. They then sought to include modifiable environmental factors such as vitamin D and cotinine serum levels (as a surrogate for smoking status) to their modeling strategy for predicting long-term prognosis.

From 1995 to 2016, the researchers prospectively recruited 1,088 patients with CIS for clinical assessment and brain MRI follow-up. Baseline hazard for EDSS 3.0 was calculated after fitting several parametric models (Weibull, generalized gamma) for a linear combination of baseline domains, including: age, sex, topography, oligoclonal bands, and baseline T2 lesions. A recursive partitioning regression tree (RPART) method was used to stratify patients in risk groups according to their predicted median time to EDSS 3.0. This prognostic model was further updated by adding vitamin D deficiency (n = 472; cutoff, 8 ng/mL), smoking (n = 435; cotinine cutoff, 14 ng/mL), and their interactions with risk groups. Harrell C statistic was used to evaluate the performance of these models.

The predictive model is based on 1,062 patients with more than one year of follow-up. After fitting several models, a final Weibull (proportional hazard) model was selected. RPART allowed patients to be stratified in three groups: low risk (n = 442; hazard ratio [HR], 1.0 reference), medium risk (n = 561; HR, 3.0), and high risk (n = 51; HR, 9.6) with a Harrell C of 0.65. An interaction was observed for vitamin D deficiency and risk group. Vitamin D deficiency increased the risk for EDSS progression. Patients with vitamin D deficiency with low risk moved to the medium-risk group, and patients with medium risk moved to the high-risk group. In patients with high risk at baseline, the impact of vitamin D deficiency was minor. Overall Harrell C increased from 0.65 to 0.69.

A similar interaction was observed for smoking and risk group. High cotinine levels increased the risk for EDSS progression. Patients with low risk moved to the medium-risk group, and patients with medium risk moved to the high-risk group. For the high-risk group, adding smoking status raised hazard ratios from 13.8 to 64.2. Harrell C improved from 0.68 to 0.73.

This study was financed with grants from the Spanish Ministry of Economy and Competitiveness, the Fondo de Investigación Sanitaria, and the Instituto de Salud Carlos III. It also was supported by a grant from Genzyme.

BERLIN—Smoking and low vitamin D levels are key modifiable prognostic factors that show a significant interaction with risk of Expanded Disability Status Scale (EDSS) progression in patients with clinically isolated syndrome (CIS), according to a study presented at ECTRIMS 2018. This finding suggests potential preventive strategies for avoiding disability, said Susana Otero-Romero, MD, PhD, an epidemiologist at the Multiple Sclerosis Center of Catalonia (CEMCAT) at University Hospital Vall d’Hebron in Barcelona, and her research colleagues.

A dynamic model for predicting long-term prognosis incorporating age, sex, topography of CIS, oligoclonal bands, and number of T2 lesions was designed by Dr. Otero-Romero and colleagues. They then sought to include modifiable environmental factors such as vitamin D and cotinine serum levels (as a surrogate for smoking status) to their modeling strategy for predicting long-term prognosis.

From 1995 to 2016, the researchers prospectively recruited 1,088 patients with CIS for clinical assessment and brain MRI follow-up. Baseline hazard for EDSS 3.0 was calculated after fitting several parametric models (Weibull, generalized gamma) for a linear combination of baseline domains, including: age, sex, topography, oligoclonal bands, and baseline T2 lesions. A recursive partitioning regression tree (RPART) method was used to stratify patients in risk groups according to their predicted median time to EDSS 3.0. This prognostic model was further updated by adding vitamin D deficiency (n = 472; cutoff, 8 ng/mL), smoking (n = 435; cotinine cutoff, 14 ng/mL), and their interactions with risk groups. Harrell C statistic was used to evaluate the performance of these models.

The predictive model is based on 1,062 patients with more than one year of follow-up. After fitting several models, a final Weibull (proportional hazard) model was selected. RPART allowed patients to be stratified in three groups: low risk (n = 442; hazard ratio [HR], 1.0 reference), medium risk (n = 561; HR, 3.0), and high risk (n = 51; HR, 9.6) with a Harrell C of 0.65. An interaction was observed for vitamin D deficiency and risk group. Vitamin D deficiency increased the risk for EDSS progression. Patients with vitamin D deficiency with low risk moved to the medium-risk group, and patients with medium risk moved to the high-risk group. In patients with high risk at baseline, the impact of vitamin D deficiency was minor. Overall Harrell C increased from 0.65 to 0.69.

A similar interaction was observed for smoking and risk group. High cotinine levels increased the risk for EDSS progression. Patients with low risk moved to the medium-risk group, and patients with medium risk moved to the high-risk group. For the high-risk group, adding smoking status raised hazard ratios from 13.8 to 64.2. Harrell C improved from 0.68 to 0.73.

This study was financed with grants from the Spanish Ministry of Economy and Competitiveness, the Fondo de Investigación Sanitaria, and the Instituto de Salud Carlos III. It also was supported by a grant from Genzyme.

BERLIN—Smoking and low vitamin D levels are key modifiable prognostic factors that show a significant interaction with risk of Expanded Disability Status Scale (EDSS) progression in patients with clinically isolated syndrome (CIS), according to a study presented at ECTRIMS 2018. This finding suggests potential preventive strategies for avoiding disability, said Susana Otero-Romero, MD, PhD, an epidemiologist at the Multiple Sclerosis Center of Catalonia (CEMCAT) at University Hospital Vall d’Hebron in Barcelona, and her research colleagues.

A dynamic model for predicting long-term prognosis incorporating age, sex, topography of CIS, oligoclonal bands, and number of T2 lesions was designed by Dr. Otero-Romero and colleagues. They then sought to include modifiable environmental factors such as vitamin D and cotinine serum levels (as a surrogate for smoking status) to their modeling strategy for predicting long-term prognosis.

From 1995 to 2016, the researchers prospectively recruited 1,088 patients with CIS for clinical assessment and brain MRI follow-up. Baseline hazard for EDSS 3.0 was calculated after fitting several parametric models (Weibull, generalized gamma) for a linear combination of baseline domains, including: age, sex, topography, oligoclonal bands, and baseline T2 lesions. A recursive partitioning regression tree (RPART) method was used to stratify patients in risk groups according to their predicted median time to EDSS 3.0. This prognostic model was further updated by adding vitamin D deficiency (n = 472; cutoff, 8 ng/mL), smoking (n = 435; cotinine cutoff, 14 ng/mL), and their interactions with risk groups. Harrell C statistic was used to evaluate the performance of these models.

The predictive model is based on 1,062 patients with more than one year of follow-up. After fitting several models, a final Weibull (proportional hazard) model was selected. RPART allowed patients to be stratified in three groups: low risk (n = 442; hazard ratio [HR], 1.0 reference), medium risk (n = 561; HR, 3.0), and high risk (n = 51; HR, 9.6) with a Harrell C of 0.65. An interaction was observed for vitamin D deficiency and risk group. Vitamin D deficiency increased the risk for EDSS progression. Patients with vitamin D deficiency with low risk moved to the medium-risk group, and patients with medium risk moved to the high-risk group. In patients with high risk at baseline, the impact of vitamin D deficiency was minor. Overall Harrell C increased from 0.65 to 0.69.

A similar interaction was observed for smoking and risk group. High cotinine levels increased the risk for EDSS progression. Patients with low risk moved to the medium-risk group, and patients with medium risk moved to the high-risk group. For the high-risk group, adding smoking status raised hazard ratios from 13.8 to 64.2. Harrell C improved from 0.68 to 0.73.

This study was financed with grants from the Spanish Ministry of Economy and Competitiveness, the Fondo de Investigación Sanitaria, and the Instituto de Salud Carlos III. It also was supported by a grant from Genzyme.