16 research outputs found
Explainable time-to-progression predictions in multiple sclerosis
No full textBackground: Prognostic machine learning research in multiple sclerosis has been mainly focusing on black-box models predicting whether a patients' disability will progress in a fixed number of years. However, as this is a binary yes/no question, it cannot take individual disease severity into account. Therefore, in this work we propose to model the time to disease progression instead. Additionally, we use explainable machine learning techniques to make the model outputs more interpretable. Methods: A preprocessed subset of 29,201 patients of the international data registry MSBase was used. Disability was assessed in terms of the Expanded Disability Status Scale (EDSS). We predict the time to significant and confirmed disability progression using random survival forests, a machine learning model for survival analysis. Performance is evaluated on a time-dependent area under the receiver operating characteristic and the precision-recall curves. Importantly, predictions are then explained using SHAP and Bellatrex, two explainability toolboxes, and lead to both global (population-wide) as well as local (patient visit-specific) insights. Results: On the task of predicting progression in 2 years, the random survival forest achieves state-of-the-art performance, comparable to previous work employing a random forest. However, here the random survival forest has the added advantage of being able to predict progression over a longer time horizon, with AUROC > 60% for the first 10 years after baseline. Explainability techniques further validated the model by extracting clinically valid insights from the predictions made by the model. For example, a clear decline in the per-visit probability of progression is observed in more recent years since 2012, likely reflecting globally increasing use of more effective MS therapies. Conclusion: The binary classification models found in the literature can be extended to a time-to-event setting without loss of performance, thus allowing a more comprehensive prediction of patient prognosis. Furthermore, explainability techniques proved to be key to reach a better understanding of the model and increase validation of its behaviour.Funding: This work was supported by Research Foundation Flanders, Belgium [grant number 1S38023N] and the Flemish government AI Research Program (FAIR), Belgium. Contributors: We acknowledge data contributions from the following MSBase principal investigators, ordered by the number of contributed patients (from high to low): Dana Horakova, Guillermo Izquierdo, Sara Eichau, Marc Girard, Pierre Duquette, Pierre Grammond, Francois Grand’Maison, Maria Pia Amato, Katherine Buzzard,Cavit Boz, Murat Terzi, Vahid Shaygannejad, Jeannette Lechner-Scott, Jens Kuhle, Bassem Yamout, Yolanda Blanco, Elisabetta Cartechini, Recai Turkoglu, Nevin John, Radek Ampapa, Davide Maimone, Cristina Ramo-Tello, Celia Oreja-Guevara, Maria Di Gregorio, Mark Slee, Aysun Soysal, Riadh Gouider, Richard Macdonell, Maria Edite Rio, Liesbeth Van Hijfte, Jiwon Oh, Tamara Castillo-Triviño, Michael Barnett, Ricardo Fernandez Bolaños, Marie D’hooghe, Justin Garber, Ayse Altintas, Cees Zwanikken, Eduardo Aguera-Morales, Magd Zakaria, Sarah Besora, Suzanne Hodgkinson, the late Yara Fragoso, Rana Karabudak, Edgardo Cristiano, Jose Antonio Cabrera-Gomez, Maria Laura Saladino, Leontien Den braber-Moerland, Bruce Taylor, Orla Gray, Shlomo Flechter, Fraser Moore, Claudio Gobbi, Chris McGuigan, Jennifer Massey, Jamie Campbell, Marzena Fabis-Pedrini, Nevin Shalaby, Mihaela Simu, Angel Perez sempere, Cameron Shaw, Jan Schepel, Steve Vucic, Jabir Alkhaboori, Magdolna Simo, Danny Decoo, Jose Andres Dominguez, Neil Shuey, Stella Hughes, Ilya Kister. Finally, we acknowledge for their aid in data acquisition: Dr Mark Marriott, Dr Trevor Kilpatrick, Dr John King, Dr Katherine Buzzard, Dr Ai-Lan Nguyen, Dr Chris Dwyer, Dr Mastura Monif, Dr Izanne Roos, Ms Lisa Taylor, Ms Josephine Baker, Prof Robert Zivadinov, Prof Ralph Benedict, Dr Marzena Fabis-Pedrini, Dr Clara Chisari, Dr Emanuele D’Amico, Dr Lo Fermo Salvatore, Dr Catherine Larochelle, Dr Raymond Hupperts, Dr Freek Verheul, Dr Krisztian Kasa
Contribution of different relapse phenotypes to disability in multiple sclerosis
No full textThis study evaluated the effect of relapse phenotype on disability accumulation in multiple sclerosis. Analysis of prospectively collected data was conducted in 19,504 patients with relapse-onset multiple sclerosis and minimum 1-year prospective follow-up from the MSBase cohort study. Multivariable linear regression models assessed associations between relapse incidence, phenotype and changes in disability (quantified with Expanded Disability Status Scale and its Functional System scores). Sensitivity analyses were conducted. In 34,858 relapses recorded during 136,462 patient-years (median follow-up 5.9 years), higher relapse incidence was associated with greater disability accumulation (β = 0.16, p < 0.001). Relapses of all phenotypes promoted disability accumulation, with the most pronounced increase associated with pyramidal (β = 0.27 (0.25–0.29)), cerebellar (β = 0.35 (0.30–0.39)) and bowel/bladder (β = 0.42 (0.35–0.49)) phenotypes (mean (95% confidence interval)). Higher incidence of each relapse phenotype was associated with an increase in disability in the corresponding neurological domain, as well as anatomically related domains. Relapses are associated with accumulation of neurological disability. Relapses in pyramidal, cerebellar and bowel/bladder systems have the greatest association with disability change. Therefore, prevention of these relapses is an important objective of disease-modifying therapy. The differential impact of relapse phenotypes on disability outcomes could influence management of treatment failure in multiple sclerosis
Comparison Between Dimethyl Fumarate, Fingolimod, and Ocrelizumab After Natalizumab Cessation.
No full textIMPORTANCE: Natalizumab cessation is associated with a risk of rebound disease activity. It is important to identify the optimal switch disease-modifying therapy strategy after natalizumab to limit the risk of severe relapses.
OBJECTIVES: To compare the effectiveness and persistence of dimethyl fumarate, fingolimod, and ocrelizumab among patients with relapsing-remitting multiple sclerosis (RRMS) who discontinued natalizumab.
DESIGN, SETTING, AND PARTICIPANTS: In this observational cohort study, patient data were collected from the MSBase registry between June 15, 2010, and July 6, 2021. The median follow-up was 2.7 years. This was a multicenter study that included patients with RRMS who had used natalizumab for 6 months or longer and then were switched to dimethyl fumarate, fingolimod, or ocrelizumab within 3 months after natalizumab discontinuation. Patients without baseline data were excluded from the analysis. Data were analyzed from May 24, 2022, to January 9, 2023.
EXPOSURES: Dimethyl fumarate, fingolimod, and ocrelizumab.
MAIN OUTCOMES AND MEASURES: Primary outcomes were annualized relapse rate (ARR) and time to first relapse. Secondary outcomes were confirmed disability accumulation, disability improvement, and subsequent treatment discontinuation, with the comparisons for the first 2 limited to fingolimod and ocrelizumab due to the small number of patients taking dimethyl fumarate. The associations were analyzed after balancing covariates using an inverse probability of treatment weighting method.
RESULTS: Among 66 840 patients with RRMS, 1744 had used natalizumab for 6 months or longer and were switched to dimethyl fumarate, fingolimod, or ocrelizumab within 3 months of natalizumab discontinuation. After excluding 358 patients without baseline data, a total of 1386 patients (mean [SD] age, 41.3 [10.6] years; 990 female [71%]) switched to dimethyl fumarate (138 [9.9%]), fingolimod (823 [59.4%]), or ocrelizumab (425 [30.7%]) after natalizumab. The ARR for each medication was as follows: ocrelizumab, 0.06 (95% CI, 0.04-0.08); fingolimod, 0.26 (95% CI, 0.12-0.48); and dimethyl fumarate, 0.27 (95% CI, 0.12-0.56). The ARR ratio of fingolimod to ocrelizumab was 4.33 (95% CI, 3.12-6.01) and of dimethyl fumarate to ocrelizumab was 4.50 (95% CI, 2.89-7.03). Compared with ocrelizumab, the hazard ratio (HR) of time to first relapse was 4.02 (95% CI, 2.83-5.70) for fingolimod and 3.70 (95% CI, 2.35-5.84) for dimethyl fumarate. The HR of treatment discontinuation was 2.57 (95% CI, 1.74-3.80) for fingolimod and 4.26 (95% CI, 2.65-6.84) for dimethyl fumarate. Fingolimod use was associated with a 49% higher risk for disability accumulation compared with ocrelizumab. There was no significant difference in disability improvement rates between fingolimod and ocrelizumab.
CONCLUSION AND RELEVANCE: Study results show that among patients with RRMS who switched from natalizumab to dimethyl fumarate, fingolimod, or ocrelizumab, ocrelizumab use was associated with the lowest ARR and discontinuation rates, and the longest time to first relapse
A non-inferiority study of rituximab versus ocrelizumab in relapsing-remitting multiple sclerosis
No full textEarly non-disabling relapses are important predictors of disability accumulation in people with relapsing-remitting multiple sclerosis
Get PDFAbstract: BACKGROUND The prognostic significance of non-disabling relapses in people with relapsing-remitting multiple sclerosis (RRMS) is unclear. OBJECTIVE To determine whether early non-disabling relapses predict disability accumulation in RRMS. METHODS We redefined mild relapses in MSBase as 'non-disabling', and moderate or severe relapses as 'disabling'. We used mixed-effects Cox models to compare 90-day confirmed disability accumulation events in people with exclusively non-disabling relapses within 2\u2009years of RRMS diagnosis to those with no early relapses; and any early disabling relapses. Analyses were stratified by disease-modifying therapy (DMT) efficacy during follow-up. RESULTS People who experienced non-disabling relapses within 2\u2009years of RRMS diagnosis accumulated more disability than those with no early relapses if they were untreated (n\u2009=\u2009285 vs 4717; hazard ratio (HR)\u2009=\u20091.29, 95% confidence interval (CI)\u2009=\u20091.00-1.68) or given platform DMTs (n\u2009=\u20091074 vs 7262; HR\u2009=\u20091.33, 95% CI\u2009=\u20091.15-1.54), but not if given high-efficacy DMTs (n\u2009=\u2009572 vs 3534; HR\u2009=\u20090.90, 95% CI\u2009=\u20090.71-1.13) during follow-up. Differences in disability accumulation between those with early non-disabling relapses and those with early disabling relapses were not confirmed statistically. CONCLUSION This study suggests that early non-disabling relapses are associated with a higher risk of disability accumulation than no early relapses in RRMS. This risk may be mitigated by high-efficacy DMTs. Therefore, non-disabling relapses should be considered when making treatment decisions
Predictors of treatment switching in the Big Multiple Sclerosis Data Network
Get PDFBackgroundTreatment switching is a common challenge and opportunity in real-world clinical practice. Increasing diversity in disease-modifying treatments (DMTs) has generated interest in the identification of reliable and robust predictors of treatment switching across different countries, DMTs, and time periods.ObjectiveThe objective of this retrospective, observational study was to identify independent predictors of treatment switching in a population of relapsing-remitting MS (RRMS) patients in the Big Multiple Sclerosis Data Network of national clinical registries, including the Italian MS registry, the OFSEP of France, the Danish MS registry, the Swedish national MS registry, and the international MSBase Registry.MethodsIn this cohort study, we merged information on 269,822 treatment episodes in 110,326 patients from 1997 to 2018 from five clinical registries. Patients were included in the final pooled analysis set if they had initiated at least one DMT during the relapsing-remitting MS (RRMS) stage. Patients not diagnosed with RRMS or RRMS patients not initiating DMT therapy during the RRMS phase were excluded from the analysis. The primary study outcome was treatment switching. A multilevel mixed-effects shared frailty time-to-event model was used to identify independent predictors of treatment switching. The contributing MS registry was included in the pooled analysis as a random effect.ResultsEvery one-point increase in the Expanded Disability Status Scale (EDSS) score at treatment start was associated with 1.08 times the rate of subsequent switching, adjusting for age, sex, and calendar year (adjusted hazard ratio [aHR] 1.08; 95% CI 1.07-1.08). Women were associated with 1.11 times the rate of switching relative to men (95% CI 1.08-1.14), whilst older age was also associated with an increased rate of treatment switching. DMTs started between 2007 and 2012 were associated with 2.48 times the rate of switching relative to DMTs that began between 1996 and 2006 (aHR 2.48; 95% CI 2.48-2.56). DMTs started from 2013 onwards were more likely to switch relative to the earlier treatment epoch (aHR 8.09; 95% CI 7.79-8.41; reference = 1996-2006).ConclusionSwitching between DMTs is associated with female sex, age, and disability at baseline and has increased in frequency considerably in recent years as more treatment options have become available. Consideration of a patient's individual risk and tolerance profile needs to be taken into account when selecting the most appropriate switch therapy from an expanding array of treatment choices
Effectiveness of cladribine compared to fingolimod, natalizumab, ocrelizumab and alemtuzumab in relapsing-remitting multiple sclerosis
Get PDFBackground:
Comparisons between cladribine and other potent immunotherapies for multiple sclerosis (MS) are lacking.
Objectives:
To compare the effectiveness of cladribine against fingolimod, natalizumab, ocrelizumab and alemtuzumab in relapsing-remitting MS.
Methods:
Patients with relapsing-remitting MS treated with cladribine, fingolimod, natalizumab, ocrelizumab or alemtuzumab were identified in the global MSBase cohort and two additional UK centres. Patients were followed for ⩾6/12 and had ⩾3 in-person disability assessments. Patients were matched using propensity score. Four pairwise analyses compared annualised relapse rates (ARRs) and disability outcomes.
Results:
The eligible cohorts consisted of 853 (fingolimod), 464 (natalizumab), 1131 (ocrelizumab), 123 (alemtuzumab) or 493 (cladribine) patients. Cladribine was associated with a lower ARR than fingolimod (0.07 vs. 0.12, p = 0.006) and a higher ARR than natalizumab (0.10 vs. 0.06, p = 0.03), ocrelizumab (0.09 vs. 0.05, p = 0.008) and alemtuzumab (0.17 vs. 0.04, p < 0.001). Compared to cladribine, the risk of disability worsening did not differ in patients treated with fingolimod (hazard ratio (HR) 1.08, 95% confidence interval (CI) 0.47–2.47) or alemtuzumab (HR 0.73, 95% CI 0.26–2.07), but was lower for patients treated with natalizumab (HR 0.35, 95% CI 0.13–0.94) and ocrelizumab (HR 0.45, 95% CI 0.26–0.78). There was no evidence for a difference in disability improvement.
Conclusion:
Cladribine is an effective therapy that can be viewed as a step up in effectiveness from fingolimod, but is less effective than the most potent intravenous MS therapies
Towards personalized therapy for multiple sclerosis: prediction of individual treatment response
Get PDFTimely initiation of effective therapy is crucial for preventing disability in multiple sclerosis; however, treatment response varies greatly among patients. Comprehensive predictive models of individual treatment response are lacking. Our aims were: (i) to develop predictive algorithms for individual treatment response using demographic, clinical and paraclinical predictors in patients with multiple sclerosis; and (ii) to evaluate accuracy, and internal and external validity of these algorithms. This study evaluated 27 demographic, clinical and paraclinical predictors of individual response to seven disease-modifying therapies in MSBase, a large global cohort study. Treatment response was analysed separately for disability progression, disability regression, relapse frequency, conversion to secondary progressive disease, change in the cumulative disease burden, and the probability of treatment discontinuation. Multivariable survival and generalized linear models were used, together with the principal component analysis to reduce model dimensionality and prevent overparameterization. Accuracy of the individual prediction was tested and its internal validity was evaluated in a separate, non-overlapping cohort. External validity was evaluated in a geographically distinct cohort, the Swedish Multiple Sclerosis Registry. In the training cohort (n = 8513), the most prominent modifiers of treatment response comprised age, disease duration, disease course, previous relapse activity, disability, predominant relapse phenotype and previous therapy. Importantly, the magnitude and direction of the associations varied among therapies and disease outcomes. Higher probability of disability progression during treatment with injectable therapies was predominantly associated with a greater disability at treatment start and the previous therapy. For fingolimod, natalizumab or mitoxantrone, it was mainly associated with lower pretreatment relapse activity. The probability of disability regression was predominantly associated with pre-baseline disability, therapy and relapse activity. Relapse incidence was associated with pretreatment relapse activity, age and relapsing disease course, with the strength of these associations varying among therapies. Accuracy and internal validity (n = 1196) of the resulting predictive models was high (>80%) for relapse incidence during the first year and for disability outcomes, moderate for relapse incidence in Years 2-4 and for the change in the cumulative disease burden, and low for conversion to secondary progressive disease and treatment discontinuation. External validation showed similar results, demonstrating high external validity for disability and relapse outcomes, moderate external validity for cumulative disease burden and low external validity for conversion to secondary progressive disease and treatment discontinuation. We conclude that demographic, clinical and paraclinical information helps predict individual response to disease-modifying therapies at the time of their commencement
Early non-disabling relapses are important predictors of disability accumulation in people with relapsing-remitting multiple sclerosis
No full textBackground: The prognostic significance of non-disabling relapses in people with relapsing-remitting multiple sclerosis (RRMS) is unclear. Objective: To determine whether early non-disabling relapses predict disability accumulation in RRMS. Methods: We redefined mild relapses in MSBase as 'non-disabling', and moderate or severe relapses as 'disabling'. We used mixed-effects Cox models to compare 90-day confirmed disability accumulation events in people with exclusively non-disabling relapses within 2 years of RRMS diagnosis to those with no early relapses; and any early disabling relapses. Analyses were stratified by disease-modifying therapy (DMT) efficacy during follow-up. Results: People who experienced non-disabling relapses within 2 years of RRMS diagnosis accumulated more disability than those with no early relapses if they were untreated (n = 285 vs 4717; hazard ratio (HR) = 1.29, 95% confidence interval (CI) = 1.00-1.68) or given platform DMTs (n = 1074 vs 7262; HR = 1.33, 95% CI = 1.15-1.54), but not if given high-efficacy DMTs (n = 572 vs 3534; HR = 0.90, 95% CI = 0.71-1.13) during follow-up. Differences in disability accumulation between those with early non-disabling relapses and those with early disabling relapses were not confirmed statistically. Conclusion: This study suggests that early non-disabling relapses are associated with a higher risk of disability accumulation than no early relapses in RRMS. This risk may be mitigated by high-efficacy DMTs. Therefore, non-disabling relapses should be considered when making treatment decisions
Predictors of treatment switching in the Big Multiple Sclerosis Data Network
No full textBackgroundTreatment switching is a common challenge and opportunity in real-world clinical practice. Increasing diversity in disease-modifying treatments (DMTs) has generated interest in the identification of reliable and robust predictors of treatment switching across different countries, DMTs, and time periods.ObjectiveThe objective of this retrospective, observational study was to identify independent predictors of treatment switching in a population of relapsing-remitting MS (RRMS) patients in the Big Multiple Sclerosis Data Network of national clinical registries, including the Italian MS registry, the OFSEP of France, the Danish MS registry, the Swedish national MS registry, and the international MSBase Registry.MethodsIn this cohort study, we merged information on 269,822 treatment episodes in 110,326 patients from 1997 to 2018 from five clinical registries. Patients were included in the final pooled analysis set if they had initiated at least one DMT during the relapsing-remitting MS (RRMS) stage. Patients not diagnosed with RRMS or RRMS patients not initiating DMT therapy during the RRMS phase were excluded from the analysis. The primary study outcome was treatment switching. A multilevel mixed-effects shared frailty time-to-event model was used to identify independent predictors of treatment switching. The contributing MS registry was included in the pooled analysis as a random effect.ResultsEvery one-point increase in the Expanded Disability Status Scale (EDSS) score at treatment start was associated with 1.08 times the rate of subsequent switching, adjusting for age, sex, and calendar year (adjusted hazard ratio aHR 1.08; 95% CI 1.07-1.08). Women were associated with 1.11 times the rate of switching relative to men (95% CI 1.08-1.14), whilst older age was also associated with an increased rate of treatment switching. DMTs started between 2007 and 2012 were associated with 2.48 times the rate of switching relative to DMTs that began between 1996 and 2006 (aHR 2.48; 95% CI 2.48-2.56). DMTs started from 2013 onwards were more likely to switch relative to the earlier treatment epoch (aHR 8.09; 95% CI 7.79-8.41; reference = 1996-2006).ConclusionSwitching between DMTs is associated with female sex, age, and disability at baseline and has increased in frequency considerably in recent years as more treatment options have become available. Consideration of a patient's individual risk and tolerance profile needs to be taken into account when selecting the most appropriate switch therapy from an expanding array of treatment choices
