BACKGROUND: Gadolinium enhancement is often used in randomized clinical trials to evaluate the efficacy of new drugs in multiple sclerosis (MS). Knowledge about predictors of enhancement status is important for the selection of patients for MRI monitored trials. METHODS: Data from 17 trials were available in anonymized format through the Sylvia Lawry Centre for MS Research. In an open part containing 1,328 (non primary progressive) patients, two logistic regression analyses were explored, including demographic, clinical, and MRI predictors. The authors examined the area under the curve (AUC) and the increase in positive predictive value (PPV). The final selection of models was validated in a closed part of 848 comparable patients. RESULTS: Age at onset, disease duration, and disease course (CIS/RR/SP) were important predictors from the multivariate models. Further, a multivariate model including T2 burden of disease was more predictive than one with only clinical predictors (AUC 0.719 vs 0.625, p < 0.001). For the model with T2 burden of disease, the PPV was 66.8%, compared to 58.5% for the model without (a priori chance 46.4%). These findings were unequivocally confirmed in the closed part of the database. CONCLUSION: Gadolinium status can be predicted by a set of baseline variables, certainly when T2 burden of disease is included. These findings may benefit the design and statistical power of future randomized clinical trials.
BACKGROUND: Gadolinium enhancement is often used in randomized clinical trials to evaluate the efficacy of new drugs in multiple sclerosis (MS). Knowledge about predictors of enhancement status is important for the selection of patients for MRI monitored trials. METHODS: Data from 17 trials were available in anonymized format through the Sylvia Lawry Centre for MS Research. In an open part containing 1,328 (non primary progressive) patients, two logistic regression analyses were explored, including demographic, clinical, and MRI predictors. The authors examined the area under the curve (AUC) and the increase in positive predictive value (PPV). The final selection of models was validated in a closed part of 848 comparable patients. RESULTS: Age at onset, disease duration, and disease course (CIS/RR/SP) were important predictors from the multivariate models. Further, a multivariate model including T2 burden of disease was more predictive than one with only clinical predictors (AUC 0.719 vs 0.625, p < 0.001). For the model with T2 burden of disease, the PPV was 66.8%, compared to 58.5% for the model without (a priori chance 46.4%). These findings were unequivocally confirmed in the closed part of the database. CONCLUSION: Gadolinium status can be predicted by a set of baseline variables, certainly when T2 burden of disease is included. These findings may benefit the design and statistical power of future randomized clinical trials.
Authors: Frederik Barkhof; Jack H Simon; Franz Fazekas; Marco Rovaris; Ludwig Kappos; Nicola de Stefano; Chris H Polman; John Petkau; Ernst W Radue; Maria P Sormani; David K Li; Paul O'Connor; Xavier Montalban; David H Miller; Massimo Filippi Journal: Nat Rev Neurol Date: 2011-12-06 Impact factor: 42.937
Authors: Marcus W Koch; Jop Mostert; Yinan Zhang; Jerry S Wolinsky; Fred D Lublin; Eva Strijbis; Gary Cutter Journal: Neurology Date: 2021-08-10 Impact factor: 11.800
Authors: Ilana R Leppert; S Narayanan; D Araújo; P S Giacomini; Y Lapierre; D L Arnold; G B Pike Journal: J Neurol Date: 2014-02-26 Impact factor: 4.849
Authors: Stefan M Gold; Anne Willing; Frank Leypoldt; Friedemann Paul; Manuel A Friese Journal: Semin Immunopathol Date: 2018-10-25 Impact factor: 9.623
Authors: Jan-Patrick Stellmann; Anneke Neuhaus; Lena Herich; Sven Schippling; Matthias Roeckel; Martin Daumer; Roland Martin; Christoph Heesen Journal: PLoS One Date: 2012-11-29 Impact factor: 3.240