CONSENSUS Statement MAGNIMS consensus recommendations on the use of brain and spinal cord atrophy measures in clinical practice Jaume Sastre-​Garriga 1 ✉, Deborah Pareto2, Marco Battaglini3, Maria A. Rocca 4 , Olga Ciccarelli5,6, Christian Enzinger7, Jens Wuerfel8, Maria P. Sormani9,10, Frederik Barkhof6,11,12, Tarek A. Yousry5,13, Nicola De Stefano3, Mar Tintoré 1, Massimo Filippi 4,14, Claudio Gasperini15, Ludwig Kappos16, Jordi Río1, Jette Frederiksen17, Jackie Palace18, Hugo Vrenken11, Xavier Montalban1,19, Àlex Rovira2 ✉ and on behalf of the MAGNIMS study group Abstract | Early evaluation of treatment response and prediction of disease evolution are key issues in the management of people with multiple sclerosis (MS). In the past 20 years, MRI has become the most useful paraclinical tool in both situations and is used clinically to assess the inflammatory component of the disease, particularly the presence and evolution of focal lesions — the pathological hallmark of MS. However, diffuse neurodegenerative processes that are at least partly independent of inflammatory mechanisms can develop early in people with MS and are closely related to disability. The effects of these neurodegenerative processes at a macroscopic level can be quantified by estimation of brain and spinal cord atrophy with MRI. MRI measurements of atrophy in MS have also been proposed as a complementary approach to lesion assessment to facilitate the prediction of clinical outcomes and to assess treatment responses. In this Consensus statement, the Magnetic Resonance Imaging in MS (MAGNIMS) study group critically review the application of brain and spinal cord atrophy in clinical practice in the management of MS, considering the role of atrophy measures in prognosis and treatment monitoring and the barriers to clinical use of these measures. On the basis of this review, the group makes consensus statements and recommendations for future research. The inflammatory component of multiple sclerosis (MS) MRI-​based quantification of inflammatory activity in pathology can be focal or diffuse and is associated with MS — on the basis of lesion counts and lesion volumes neurodegenerative processes that ultimately lead to — is established as the main efficacy outcome in phase II irreversible tissue damage and neuronal loss1. Neuro­ clinical trials10. Currently, brain and spinal cord volume degeneration was originally thought to be a late-​stage measures have no role in the MS diagnostic criteria11,12 phenomenon with limited clinical relevance, but it is or disease course classification13, but a body of evidence now recognized as being associated with acute inflam­ that these measures are valuable for early evaluation of mation from the early stages of MS and as the main driver treatment responses and prediction of disease evolution of irreversible disability2–5. In parallel with improvements has been steadily growing alongside improvements in in our understanding of the mechanisms of neurodegen­ methodology that could facilitate widespread implemen­ eration, advances in imaging techniques have enabled tation of these measures in clinical practice14,15. A key in vivo assessment of brain and spinal cord area and vol­ difficulty arises in this implementation because transla­ umes using MRI. Although brain and spinal cord volume tion of group-​based results into actionable, patient-​level ✉e-​mail: jsastre-​garriga@ loss observed with MRI cannot be equated with atrophy6, information must be made with extreme caution. cem-​cat.org; alex.rovira.idi@ gencat.cat because the latter implies pathologically proven and irre­ In this Consensus Statement, we, on behalf of the https://doi.org/10.1038/ versible tissue loss, changes in these MRI measures are Magnetic Resonance in Imaging in MS (MAGNIMS) s41582-020-0314-x associated with atrophy7 and the level of disability in MS8,9. study group, provide specific recommendations for the NaTure RevIewS | NeuroLogy volume 16 | March 2020 | 171 C o n S e n S u S S tat e m e n t implementation of brain and spinal cord atrophy meas­ meeting was to review and discuss all published data ures in the clinical management of patients with MS on brain and spinal cord atrophy in MS and to consider and on the directions of future research to improve our whether the previously published recommendations16,17 knowledge in this field. The recommendations are based on its use for diagnosis, prognosis and monitoring of on a critical review of the literature and the personal patients with MS needed to be revised and updated in experience of MAGNIMS study group members. We dis­ view of technical advances and numerous clinical studies cuss the difficulties of translating group-​based data into of atrophy in MS. The panel agreed that updated recom­ clinical application and highlight where particular cau­ mendations were necessary. After this meeting, the panel tion is appropriate. We first discuss the role of atrophy members formulated specific recommendations in measures on prognosis, then treatment monitoring and, relation to the implementation of brain and spinal cord finally, the barriers to implementation in clinical prac­ atrophy measures in clinical practice. tice. Each of these three sections comprises a review of The authors of the Consensus statement are mem­ the available evidence and a set of consensus guidelines. bers of the MAGNIMS Study Group. The network is independent of any other organization and, at the Methods time of the workshop mentioned above, was run by a A multicentre international panel on the implementa­ Steering Committee whose members were À. Rovira tion of brain and spinal cord atrophy measures in clinical (Barcelona, co-​chair), C. Enzinger (Graz, co-chair), practice convened in Barcelona, Spain, under the aus­ F. Barkhof (Amsterdam), O. Ciccarelli (London), pices of MAGNIMS, an independent European network N. de Stefano (Siena), M. Filippi (Milan), J. Frederiksen of clinical research groups with a common interest in (Copenhagen), C. Gasperini (Rome), L. Kappos (Basel), the study of MS with MRI. The panel was made up of J. Palace (Oxford), M.A. Rocca (Milan), J. Sastre-​Garriga experts in the diagnosis and management of MS, includ­ (Barcelona), H. Vrenken (Amsterdam) and T. Yousry ing neuroradiologists, neurologists, physicists, imaging (London). The first draft of the recommendations was methodologists and statisticians, who were selected by written by the principal authors (J.S.-G., D.P. and M.B.) on the workshop organizers (with approval from all mem­ the basis of the panellists’ presentations and contributions bers of the Steering Committee) on the basis of their to discussions on specific topics, which were assigned to personal expertise, from MAGNIMS centres from seven individuals according to each member’s area of expertise. different countries. The purpose of this face-​to-face The initial draft was then circulated among all authors (who were all presenters and/or discussants at the meet­ ing). Modifications were made iteratively until consensus Author addresses was reached on all recommendations; all panel members 1 Multiple Sclerosis Centre of Catalonia (Cemcat), Department of Neurology/ agreed on the full contents of the final recommendations. Neuroimmunology, Hospital Universitari Vall d’Hebron, Universitat Autònoma de Barcelona, Barcelona, Spain. Defining and predicting MS severity 2 Section of Neuroradiology and Magnetic Resonance Unit, Department of Radiology, Evidence review Hospital Universitari Vall d’Hebron, Universitat Autònoma de Barcelona, Barcelona, Spain. Global brain volume measures to define and predict 3 Department of Medicine, Surgery and Neuroscience, University of Siena, Siena, Italy. MS severity. The initial studies to investigate clinical 4 Neuroimaging Research Unit, Institute of Experimental Neurology, Division of Neuroscience, IRCCS San Raffaele Scientific Institute, Milan, Italy. correlates of brain atrophy in MS focused on patients 5 NMR Research Unit, University College London Queen Square Institute of Neurology, with well-​established disease and severe clinical mani­ London, UK. festations, particularly in the cognitive sphere18–20, but 6 National Institute for Health Research Biomedical Research Centre, University College later studies included disability, as measured with the London Hospitals, London, UK. Expanded Disability Status Scale8. Evidence from these 7 Department of Neurology and Division of Neuroradiology, Vascular and Interventional studies made it clear that neurodegenerative proces­ Radiology, Department of Radiology, Medical University of Graz, Graz, Austria. ses occur in the earliest phases of MS21, even before the 8 Medical Image Analysis Center (MIAC AG) and Department of Biomedical Engineering, disease becomes symptomatic22. University of Basel, Basel, Switzerland. Yearly global brain volume loss in healthy ageing 9 Biostatistics Unit, Department of Health Sciences, University of Genoa, Genoa, Italy. individuals ranges from –0.05% at 20–30 years of age 10 IRCCS, Ospedale Policlinico San Martino, Genoa, Italy. 11 Amsterdam Neuroscience, MS Center Amsterdam, Department of Radiology and to –0.3% at 60–70 years of age23. A change of –0.4% per Nuclear Medicine, Amsterdam UMC, Amsterdam, Netherlands. year has been proposed as the cut-​off for pathological 12 Institutes of Neurology and Healthcare Engineering, University College London, brain atrophy in MS24 (Fig. 1), although care must be London, UK. taken before applying this threshold as a marker of ther­ 13 Lysholm Department of Neuroradiology, University College London Hospitals National apeutic efficacy owing to the phenomenon of pseudo­ Hospital for Neurology and Neurosurgery, University College London Institute of atrophy (see Brain volume as an outcome measure in Neurology, London, UK. randomized clinical trials)25,26. Multiple studies have 14 Vita-Salute San Raffaele University, Milan, Italy. shown that short-​term changes (over as little as 1 year) in 15 Multiple Sclerosis Center, Department of Neurosciences, San Camillo-​Forlanini brain volume are predictive of clinical status (diagnosis Hospital, Rome, Italy. of MS or disability status) at various follow-​up times in 16 Neurologic Clinic and Policlinic, Departments of Medicine, Clinical Research and Biomedical Engineering, University Hospital, University of Basel, Basel, Switzerland. clinically isolated syndromes27,28, relapsing–remitting 17 Department of Neurology, Rigshospitalet-​Glostrup and University of Copenhagen, MS (RRMS)29 and primary progressive MS30–32, either in Glostrup, Denmark. isolation or together with lesion-​related parameters33,34. 18 Nuffield Department of Clinical Neurosciences, University of Oxford, Oxford, UK. The findings above are group-​based results, and 19 Division of Neurology, St Michael’s Hospital, University of Toronto, Toronto, Canada. translation of these findings to the individual level is not 172 | March 2020 | volume 16 www.nature.com/nrneurol C o n S e n S u S S tat e m e n t Baseline 1 year 2 years 3 years 4 years a T2LL = 1,030 mm3 T2LL = 730 mm3 T2LL = 673 mm3 T2LL = 661 mm3 T2LL = 2,423 mm3 b BPF = 82% BPF = 80.9% BPF = 79.8% BPF = 79.7% BPF = 77.2% c 100 82% BPF (%) 79.7% aPBVC: –0.089% 77.2% aPBVC: –3.8% 0 1 2 3 4 Fig. 1 | Lesion load and brain atrophy in relapsing–remitting multiple sclerosis. a | Transverse T2-weighted fluid attenuation inversion recovery images from a patient with highly active relapsing–remitting multiple sclerosis (MS) who started a disease-​modifying therapy at baseline. The T2 lesion load (T2LL) is stable during the first 3 years of treatment while the patient remained clinically stable (no relapses and no disability worsening), but markedly increases at the fourth year after treatment discontinuation associated with clinical activity (the rebound effect). b | Contrast-​enhanced T1-weighted images from the same patient showing the change in brain parenchymal fraction (BPF) over time. The decrease in global brain volume in the first 3 years is mild (annualized percentage of brain volume change (aPBVC) −0.089%), but the volume loss at the fourth year is severe (aPBVC −3.8%), matching the change in T2LL and clinical evolution. The severe loss observed in year 4 is well beyond the −0.4% suggested as a pathological cut-​off for brain volume loss in MS24. c | Graphical representation of the changes in BPF over time, emphasizing the dramatic loss of volume in year 4. straightforward. In a study published in 2017, Sormani an important marker of disability status in MS9. Further et al.35 made the first attempt to define individual cut-​off studies demonstrated that spinal cord area and volume values for brain volume changes according to patients’ are affected differently in different MS subtypes, with the baseline characteristics. Pooled baseline data from the most profound atrophy in cross-​sectional studies being placebo arms of two large international clinical trials that seen in patients with progressive MS36–39. Since 2015, involved a total of 2,342 patients with RRMS showed an association between reduced cervical cord area and that expected normalized brain volumes can be calcu­ increased disability and motor dysfunction, indepen­ lated from demographic (age and sex), clinical (Expan­ dent of brain atrophy, has been confirmed40–43. An asso­ ded Disability Status Scale score and disease duration) ciation between cord atrophy and reduced peripapillary and neuroradiological (T2-weighted lesion volume) retinal nerve fibre layer thickness has been identified, parameters for individuals. Deviation of the true brain indicating that cervical cord atrophy reflects, at least in volume from this expected value enabled classification part, global pathological processes and not only specific of individuals with MS as having low, medium or high damage of long tracts41. Most studies of spinal cord area brain volume. Patients with low brain volume had a have focused on global cervical cord area measurements, 2.4-fold higher risk of disability progression over the but some work has highlighted that damage in particular next 2 years than patients with high brain volume. locations in the spinal cord, such as cervical grey mat­ ter44, the thoracolumbar segment45 and the posterior and Spinal cord atrophy measures to define and predict MS lateral cord segments46, are also relevant to disability. severity. Early, seminal studies of cervical cord atrophy Longitudinal studies indicate that atrophy rates in the in MS already suggested that cervical spinal cord area is spinal cord are higher than those in the brain and higher NaTure RevIewS | NeuroLogy volume 16 | March 2020 | 173 C o n S e n S u S S tat e m e n t in progressive MS than in established RRMS47,48. Higher whether the pathological involvement of these areas rates of cervical cord area loss have been associated with is relevant in clinical practice remains unclear. We disability progression, independent of other clinical recommend further research to determine the clinical and MRI parameters30,47 including spinal cord lesions49. relevance of atrophy in these regions. However, as for brain atrophy, use of such group-​level evidence to inform clinical decisions at the individual Monitoring therapeutic effect level is not easy. Results that can be used at the individ­ Evidence review ual level are slowly emerging; for example, Tsagkas et al.43 Brain volume as an outcome measure in randomized have shown that a 1% increase in the annual rate of clinical trials. Many trials of disease-​modifying thera­ spinal cord atrophy increases the risk of disability pro­ pies for MS have included brain atrophy as an outcome gression by 28%, reinforcing the notion that spinal cord measure (Table 1). Most early studies of interferon-β atrophy is a reliable and independent tool for monitoring (IFNβ) and glatiramer acetate did not include preplanned disease progression. brain volume measures as secondary MRI outcomes. Those that did include a sound comparison of brain Regional and tissue-​specific brain volumetry measures volume changes between intervention arms or between to define and predict MS severity. Early cross-​sectional intervention and placebo arms produced mixed results67. studies of brain white matter and grey matter changes The only study of IFNβ that provided evidence for in patients with MS indicated that both white mat­ a positive effect of treatment of brain atrophy was the ter and grey matter loss occurred early in the disease ETOMS trial68. In this study, accrual of atrophy was course, regardless of disease phenotype50–53. Evidence reduced by 30% in patients with clinically isolated syn­ also indicates that grey matter damage can occur before dromes who received low-​dose subcutaneous IFNβ1a white matter atrophy and can occur independently compared with patients who received placebo68. In of white matter lesions54–56. Results of further longitudinal several trials — particularly the trial of intramuscular studies have identified larger decreases in grey matter IFNβ1a in RRMS69,70 — negative results were at least volumes than in white matter volumes57–59 and that grey partly attributed to a pseudoatrophy effect, caused by matter damage is more relevant than white matter injury brain volume loss linked to the presumed treatment-​ to clinical outcomes, both concurrent and forthcom­ associated resolution of inflammatory activity and ing56,60–62. Two studies — one in which cortical thickness oedema. In the RRMS intramuscular IFNβ1a trial, was estimated63 and one meta-​analysis of voxel-​based significant differences that favoured treatment with morphometry studies64 — have revealed statistically IFNβ1a were only observed in the second year69,70. A post- significant associations between disability end points hoc analysis of grey matter and white matter atrophy and grey matter atrophy65, which occurs bilaterally, pre­ during the 2 years of the trial confirmed this find­ dominantly in the cingulate, pre-​central and/or post-​ ing and indicated that pseudoatrophy of white matter central gyri and the thalami and basal ganglia. Despite contributed most to the observed effect71. The same these results, global brain volume changes seem to be effect has been described in observational studies more strongly associated with clinical outcomes than of patients taking natalizumab72 or IFNβ73, although are regional changes. This observation is unexpected more research is needed to confirm these findings. because grey matter loss is thought to underlie disability Results with glatiramer acetate were also mixed, though accumulation. Associations between grey matter volume some nonprimary analyses have suggested a positive change and disability accumulation might be masked effect of the treatment in patients who received glatir­ by the effects of high variability of regional segmenta­ amer acetate from the beginning of the trial when tions, which makes clinical application of these regional compared with those who received the treatment later74. measures inadvisable at present62,66. Trials of IFNβ and of glatiramer acetate in progres­ sive MS have been negative75 or have also suggested a Statements and recommendations pseudoatrophy effect76. 1. We recommend measurement of global brain volume Trials of natalizumab provided a clear demonstration to better gauge global disease burden in patients with of pseudoatrophy. In the AFFIRM trial77, brain volume MS because brain volume loss is associated with and decreases among patients who received natalizumab predicts disability in all clinical MS phenotypes, were larger in the first year than among patients who including the earliest stages of the condition. received placebo, but the observation was reversed in 2. We recommend measurement of cervical cord area the second year. Subsequent clinical trials of newer loss because this measure is associated with and pre­ drugs (including fingolimod, dimethyl fumarate, ter­ dicts disability in all clinical MS phenotypes, including iflunomide, ocrelizumab and alemtuzumab) have all the earliest stages of the condition. incorporated brain volume measures as secondary or 3. Grey matter volume changes in the brain are more tertiary outcomes, and results have been positive over­ pronounced and clinically relevant than white matter all78, although studies are not readily comparable. Of volume changes, even in the earliest stages of MS, but note, in studies of powerful anti-​inflammatory drugs their exact relevance in clinical practice is unclear. We against active comparators, the trial drugs have been recommend further research to clarify this relevance. superior at decreasing accrual of atrophy79–81, indicating 4. Some cerebral grey matter regions (including the that the pseudoatrophy effect can be overcome by the thalami, basal ganglia and specific cortical areas) are beneficial effects of anti-​inflammatory drugs on neuro­ affected particularly strongly by atrophy in MS, but degeneration in MS. Strategies to minimize the effect of 174 | March 2020 | volume 16 www.nature.com/nrneurol C o n S e n S u S S tat e m e n t Table 1 | Brain atrophy outcomes in pivotal trials of approved disease-​modifying drugs Drug Clinical trial Phenotype Comparator Time frame Software Treatment favoured IFNβ1a (SC) ETOMS68 CIS Placebo 0–24 months SIENA113 IFNβ1a (SC) REFLEX 155 CIS Placebo 0–24 months SIENA 113 None PRISMS156 RRMS Placeboa 0–6 years Kappos et al.156 None IFNβ1b (SC) BENEFIT 157 CIS Placeboa 0–36 months SIENA 113 None EUSPMS158 SPMS Placebo 0–36 months Losseff et al.8 None Montalban et al. 76 PPMS Placebo 0–24 months SPM 51 None IFNβ1a (IM) Rudick et al.69 RRMS Placebo 0–24 months Rudick et al.70 None Leary et al. 90 PPMS Placebo 0–24 months Fox et al. 140 None Glatiramer acetate PRECISE159 CIS Placebo 0–36 months SIENA113 None Sormani et al.160 RRMS Placeboa 0–18 months SIENA113 Glatiramer acetate PROMISE75 PPMS Placebo 0–36 months Wolinsky et al.75 None Teriflunomide TOPIC 161 CIS Placebo 0–24 months Miller et al. 161 None TEMSO162 RRMS Placebo 0–24 months Wolinsky et al.163 None TEMSO 164,b RRMS Placebo 0–24 months SIENA 113 Teriflunomide Dimethyl fumarate DEFINE82 RRMS Placebo 6–24 months SIENA113 Dimethyl fumarate CONFIRM165 RRMS Placebo 0–24 months SIENA113 None Natalizumab AFFIRM77 RRMS Placebo 0–24 months Rudick et al.70 Noned SENTINEL 166 RRMS Placeboc 0–24 months Rudick et al. 70 Noned ASCEND167 SPMS Placebo 24–96 weeks SIENAX113 None Fingolimod FREEDOMS 1 (ref. ) 168 RRMS Placebo 0–24 months SIENA 113 Fingolimod FREEDOMS 2 (ref.169) RRMS Placebo 0–24 months SIENA113 Fingolimod TRANSFORMS 81 RRMS IFNβ1a (IM) 0–12 months SIENA 113 Fingolimod INFORMS91 PPMS Placebo 0–36/60 months SIENA113 None Alemtuzumab CARE-​MS 1 (ref. ) 79 RRMS IFNβ1a (SC) 0–24 months Rudick et al. 70 Alemtuzumab CARE-​MS 2 (ref.80) RRMS IFNβ1a (SC) 0–24 months Rudick et al.70 Alemtuzumab Ocrelizumab OPERA 1 (ref. ) 170 RRMS IFNβ1a (SC) 24–96 weeks SIENA 113 Ocrelizumab OPERA 2 (ref.170) RRMS IFNβ1a (SC) 24–96 weeks SIENA113 Ocrelizumab ORATORIO 171 PPMS Placebo 24–120 weeks SIENA 113 Ocrelizumab Cladribine ORACLE172 CIS Placebo 0–24 months SIENA113 None CLARITY 83 RRMS Placebo 6–24 months SIENA 113 Cladribine CIS, clinically isolated syndrome; IFN, interferon; IM, intramuscular ; PPMS, primary progressive multiple sclerosis; RRMS, relapsing–remitting multiple sclerosis; SC, subcutaneous; SIENA , Structural Image Evaluation, using Normalization, of Atrophy ; SPMS, secondary progressive multiple sclerosis. aIncludes a period receiving the active drug. bReanalysis of TEMSO trial data using SIENA. cAs an add-​on to IFNβ1a (IM). dResults favoured natalizumab in the 12–24-month period. pseudoatrophy on clinical measures include, but are not trials has driven the adoption of brain volume change restricted to, obtaining baseline measurements once the as a primary outcome in phase II trials in cohorts of anti-​inflammatory effect is well established (for example, patients with progressive MS85,86. re-​baseline with MRI at 6 or 12 months after treatment initiation)82,83. Spinal cord atrophy as an outcome in randomized Further support for the clinical relevance of brain vol­ clinical trials. Despite the relevance of spinal cord atro­ ume outcomes in trials of treatment for RRMS comes phy to long-term disability, this measure has scarcely from a meta-​analysis that included >13,500 patients from been used as an outcome in clinical trials87; when it has been 13 different clinical trials84. The conclusion of the analy­ used, the results have been negative. For example, spinal sis was that the effect of a given therapy on changes in cord atrophy was an outcome measure in an investigator-​ brain volume over 2 years is associated with the effect of initiated study of lamotrigine for neuroprotection in the drug on disability outcomes and that this association secondary progressive MS, but no differences were seen is, at least in part, independent of its anti-inflammatory between the treatment and placebo arms88. Spinal cord effect on active MRI lesions84. This close association atrophy measures have been used in several other studies between brain atrophy and disability outcomes in clinical in progressive MS89 but the atrophy and clinical results NaTure RevIewS | NeuroLogy volume 16 | March 2020 | 175 C o n S e n S u S S tat e m e n t have either been negative76,90 or were not published with confirm these associations before brain volume can be the rest of the trial91. considered for use as a treatment-​monitoring tool. 5. Use of spinal cord atrophy as a treatment-​monitoring Brain volume and spinal cord atrophy to monitor tool in clinical trials and in clinical practice has been clinical treatment response. The relevance of brain scarce, but the rate of spinal cord atrophy is faster than volume measures to the evolution of disability in MS that of brain atrophy and methodological advances clinical trials is beyond any doubt 84. The evidence could improve reproducibility and reliability, so we from trials is complemented by that from studies of recommend further research to establish the role of individual-​level data from clinical trials92,93 and from spinal cord atrophy for treatment monitoring. observational studies of real-​world cohorts25,94, which confirm a close association between brain volume Barriers to clinical implementation changes with therapy and concurrent95 or subsequent96 Evidence review: technical barriers disability progression. These studies also indicate that Several technical aspects of image acquisition and quan­ the association between brain volume loss and dis­ tification can affect the measurement of brain and spinal ability progression is independent of clinical and MRI cord volumes and thereby affect the accuracy of estimated inflammatory markers. values. These technical barriers are discussed below. Most models for the prediction of disability progres­ sion have included brain volume change combined with Acquisition protocols. The choice of the acquisition either the appearance of new T2 lesions or the presence parameters (usually repetition time, echo time, inversion of clinical relapses25,92–94. Brain volume changes have time or flip angle) is usually based on the image con­ also been proposed as an addition to the ‘no evidence of trast, as assessed visually by an expert neuroradiologist. disease activity’97,98 outcome measure so as to enable Changes in scan parameters, which tend to happen in a assessment of neurodegenerative processes as well as clinical environment, affect quantification and hamper inflammatory processes, with the aim of achieving full reliable cross-​sectional and longitudinal comparisons. remission that includes an absence of disease-​specific Image contrast also depends greatly on the age of the neurodegeneration; the proposed cut-​off for this mea­ population that undergoes MRI. The Alzheimer’s Disease sure is –0.4% change in volume per year24. In a poten­ Neuroimaging Initiative100 has made a large effort to tially more realistic ‘minimal evidence of disease activity’ homogenize acquisition protocols across vendors. approach99, a less stringent cut-​off has been suggested that would allow for pseudoatrophy-​driven brain vol­ Gradient distortion. By design, the gradients applied ume loss25. However, all these data need confirmation, to the magnetic field in MRI are generally not uni­ and different cut-​offs might be needed for different form, which affects the geometry of the image. Small calculation methods and for different drugs or groups of displacements of the patient’s head in the z axis have a drugs according to different temporal patterns of brain notable effect on the estimated brain volume change101. volume effects of each drug6,78. Positioning of the patient identically across scanning sessions can minimize this effect, but this is time-​ Statements and recommendations consuming and difficult; a better solution is to apply 1. We recommend the use of whole brain atrophy over approaches developed by MRI scanner manufacturers a minimum period of 12 months as a secondary for 3D correction for the gradient nonlinearity effect102. end point in clinical trials in MS and even as a pri­ mary outcome measure in trials in the progressive Intrascanner variability. Any MRI-​derived measure is forms of MS to show the effects of the drug on the inherently variable, even when technical and physio­ neurodegenerative component of the disease. logical conditions are controlled103–108. Global estimates, 2. Ongoing and forthcoming trials are expected to such as that of the whole brain volume, are the least include grey matter volume loss as an outcome mea­ variable (<1%)106, whereas measures of smaller struc­ sure, as atrophy in the grey matter compartment is tures, such as the amygdala, are much more variable more substantial and more clinically relevant than (~5%)104,105. Such variability must be taken into account atrophy in the white matter and is likely to be affected because changes that are smaller than the estimated less by pseudoatrophy; however, data on pseudoatro­ variability cannot be reliably detected. This limitation phy remain discordant and we recommend further is highly relevant to small grey matter structures and research to clarify the contribution of grey matter when follow-​up periods are short because the expected atrophy. change is small23. 3. Pseudoatrophy effects mostly occur within the first 6–12 months from treatment initiation with any anti-​ Movement. Movement of the patient during image inflammatory therapy, so we recommend re-​baseline acquisition generates characteristic artefacts that affect MRI at 6–12 months after initiation of any therapy image quality; as a result, estimated volumes are substan­ to mitigate the impact of pseudoatrophy on outcome tially decreased109. Visual verification of image quality measures. is important because the problem is resolved when the 4. Associations between treatment effects on brain only images included in an analysis are those that an volume and disability have been demonstrated in expert considers artefact free109. Various approaches have clinical trials and indicated by evidence at the indi­ been developed to correct for movement, but an accurate vidual level, but we recommend further research to method is still not available110. 176 | March 2020 | volume 16 www.nature.com/nrneurol C o n S e n S u S S tat e m e n t Scanner system upgrades and interscanner variability. with a higher rate of brain atrophy over a 6-year period115 Scanner upgrades are unavoidable, particularly during and with a specific pattern of regional involvement of the the course of longitudinal studies, and can affect the white matter and grey matter125. A similar effect has been image contrast even if the same acquisition parameters described for cigarette smoking and substance abuse (for are used. Previous studies have shown that the system example, marijuana use)115,126. Many systemic conditions, upgrade should be included as a variable in the statis­ such as diabetes, chronic kidney disease, hypertension, tical analysis103,111,112. Quantification methods based on obesity and vascular conditions can also accelerate brain the subtraction of images, rather than on differences in atrophy115,127,128. brain parenchymal fraction between two time points, seem to be more sensitive to system upgrades113, although no The MS brain. All confounding factors previously studies have been performed to confirm this observation. discussed can interact with features of MS and affect Reliable quantification of longitudinal changes in MS estimates of brain atrophy in patients with the disease; requires scans to be acquired with the same magnet and these interactions can also affect comparisons between exactly the same sequence protocol. Variability between groups. For instance, more severe brain atrophy has been different scanners is higher than all the factors above observed in patients with MS who have one or more together108. If data acquired in different scanners need cardio­vascular risk factors129, although their impact on to be merged, a variable that accounts for the scanner longitudinal assessments might be limited, as vascu­ should be taken into consideration. lar risk factors were not associated with greater brain volume loss during 3.5 years of follow-​up in the same Evidence review: confounding factors study129. In addition, white matter lesions in MS influ­ Numerous factors can have confounding effects on the ence the accuracy of most available software for estima­ quantification of brain volume (and its changes) and tion of atrophy because they alter the image intensity thereby cause overestimation or underestimation114. histogram and influence the detection of intensity bor­ These factors are discussed below. ders between grey matter, white matter and cerebrospi­ nal fluid (CSF). This effect can be minimized by use of Age, sex and brain size. Several physiological factors lesion filling techniques130,131, which enable replacement influence brain volume estimations in healthy indi­ of lesions in the image with voxels that have intensities viduals. Studies of healthy elderly individuals have that closely resemble normal-​appearing white matter. demonstrated ongoing brain volume loss, which tends to accelerate with age115. This age-​related effect is par­ Pseudoatrophy. As discussed above, studies of the cor­ ticularly pronounced for specific CNS structures, such relation between inflammatory disease activity (new T2 as the hippocampus116. and/or gadolinium enhancing lesions) and brain volume Sex is another key factor in brain volume changes. have shown that inflammation can cause a transient Sex differences in global brain size in humans are well increase in brain volume. This increase can dramat­ established; on average, the total volume of men’s brains ically resolve following treatment with steroids132 or is ~10% larger than that of women’s brains117. Differential other disease-​modifying drugs, and the resultant reduc­ patterns of age-​related brain volume loss118 and sex-​ tion in brain volume can be erroneously interpreted as specific differences in brain morphology have also been atrophy133. demonstrated119,120. Global and regional volumetric stud­ ies have suggested that hormonal status can contribute Evidence review: volumetry tools to these sex-​related differences121. Several free-​to-use online libraries of software for neuro­ imaging analyses include fully automated pipelines for Diurnal fluctuations and hydration state. Studies of quantification of brain volume (Table 2). On the basis of healthy individuals have shown that estimations of brain the current literature that relates to this software, these volume fluctuate with the time of scanning and the software tools can be classified into two broad catego­ hydra­tion state of the individuals. Analysis of MRI data ries. The first are ‘segmentation-​based’ tools, which use from patients with MS (n = 755, 3,269 scans) and from a priori localization-​related and intensity information participants in the Alzheimer’s Disease Neuroimaging to classify the brain voxels of each MRI without using Initiative (n = 834, 6,114 scans) revealed that time of day information from brain MRI images taken at different had a notable effect on estimates of the brain parenchy­ time points. These tools do not enable direct evaluation Brain parenchymal fraction The percentage of the mal fraction in both groups. Brain volumes were sub­ of volumetric changes over time. This type of software is intracranial volume that is stantially larger in the morning122, and the effect size mostly used in cross-​sectional analyses. The second are occupied by brain parenchyma, was comparable to the yearly rate of brain atrophy in the ‘registration-​based’ tools, which enable comparison calculated as the total MS and in healthy elderly people122. Similarly, in studies of brain MRI images from the same individual acquired intracranial volume minus the in which hydration status was manipulated by overnight over time and are based on an initial registration step; volume of cerebrospinal fluid divided by the total intracranial thirsting and subsequent drinking of water, hydration-​ this type of software is used in longitudinal analyses134. volume. related changes in brain volume were as large as –0.55% Most segmentation-​based software packages provide on dehydration and +0.72% on rehydration123. measures of total brain volume, grey matter volume and Partial volume estimation white matter volume based on the partial volume estimation (PVE). The proportions of white matter, grey matter and Lifestyle and risk factors. Many lifestyle factors, includ­ (PVE) of each tissue in each voxel. The initial step is cerebrospinal fluid present ing physical activity124, influence estimates of brain vol­ assignment of the PVE to a given brain voxel on the in each voxel of an MRI image. ume. A higher level of alcohol intake has been associated basis of its intensity and the intensities of the surrounding NaTure RevIewS | NeuroLogy volume 16 | March 2020 | 177 C o n S e n S u S S tat e m e n t Table 2 | Available brain and spinal cord volumetry tools Tool Freely Measures Major limitationsa available? SIENAX Yes Global and regional brain volumes for cross-​ Segmentations are affected by the presence of sectional comparisons brain lesions SPM/VBM Yesb Global and regional brain volumes, pixel-​ Segmentations are affected by the presence of to-pixel statistical comparisons between two brain lesions groups or time points GIF Yes Regional brain volumes for cross-​sectional Time consuming; data analysed remotely comparisons Atropos Yes Regional brain volumes for cross-​sectional Limited information about the method as it has not and longitudinal comparisons been used extensively FreeSurfer Yes Cortical thickness, global and regional grey Time-​consuming; requires manual correction of the matter and white matter volumes segmented surfaces; segmentations are affected by the presence of lesions CIVET No Cortical thickness Software not freely available SIENA Yes Percentage brain volume change between Only provides global measures that include grey two time points matter and white matter SIENA-​XL No Grey matter and white matter volumes for Software not freely available longitudinal comparisons SIENAX-​MTP No Grey matter and white matter volumes for Software not freely available longitudinal comparisons BBSI Yes Percentage brain volume change between Only provides global measures that include grey two time points matter and white matter CLADA No Cortical thickness Software not freely available NeuroQuant (FDA clearance No Global and regional grey matter volumes Validation of results is only external; segmentations and CE mark received) affected by the presence of brain lesions; only images from the scanner can be analysed (that is, filled images cannot be used) Icometrix (FDA clearance and No Global and regional grey matter volumes Whole verification of the results is not direct; data CE mark received) analysed remotely Biometrica (CE mark received) No Global and regional grey matter volumes Whole verification of the results is not direct Quantib (FDA clearance and No Global and regional grey matter volumes Whole verification of the results is not direct CE mark received) Cordial Yes Spinal cord volume Limited information about the method as it has not been used extensively Spinal Cord Toolbox Yes Spinal cord area, volume and length Regions of interest should be edited and manually corrected JIM No Spinal cord area, volume and length Needs several reference marks for accurate estimates BBSI, Brain boundary shift integral; CE, Conformité Européenne; CL ADA , cortical longitudinal atrophy detection; GIF, Geodesical information flows; JIM, Jacobian integration method; SIENA , Structural Image Evaluation, using Normalization, of Atrophy; SPM, statistical parametric mapping; VBM, voxel-​based morphometry. a Not exhaustive, only major limitations are included. bSPM itself is free but a MATL AB licence is needed. voxels113. To improve the segmentation, the a priori spa­ These methods tend to be more susceptible than some of tial information for each voxel can be included, thereby the previously mentioned methods to the low-​intensity increasing the probability that each voxel belongs to spe­ contrast between tissues because they heavily rely on the cific tissue type on the basis of its location135, although gradient intensities between tissue interfaces138,139. the accuracy of this step strongly depends on the ana­ Registration-​based software packages provide mea­ tomical similarity between the MRI image and the sures of total brain volume, grey matter volume and white a priori tissue maps used. To avoid problems due to an matter volume changes by comparison of serially acquired anatomical mismatch with the atlas, only MRI images MRI images from the same individual. A common prelim­ with high anatomical similarity should be used to pro­ inary step in most of these procedures is registration of all vide the voxel location information136. Use of different MRI images from the same subject on the same virtual anatomical maps, such as probability maps of tissues or space. The first such software packages that were used structure labelling maps, can also offer improvements137. in longitudinal analyses113,140 involved registration of two Other approaches that do not depend on the PVE can MRI images of the same individual and measurement of provide a measure of cortical thickness by calculating whole brain volume change by analysing the shift of the the distances between pairs of voxels at the grey matter– parenchyma–CSF border over time. Newer approaches white matter and grey matter–CSF interfaces perpendic­ apply different methods to enable assessment of grey ular to the grey matter–white matter surface interface. matter and white matter volume changes. In one, for each 178 | March 2020 | volume 16 www.nature.com/nrneurol C o n S e n S u S S tat e m e n t voxel, the intensity information from neighbouring voxels services, Quantib Brain can be run locally or on a vendor at each time point is used141. In another, a new intensity console (General Electric), and NeuroQuant can be a harmonization scheme is applied to all MRI images from remote analysis service or local installation. All packages one individual, with the aim of assigning similar inten­ have the CE mark and, with the exception of Biometrica, sity to voxels with similar content of PVE142. Another FDA clearance. These certifications guarantee standard­ approach, known as the Jacobian integration method143,144, ization of procedures and results, meaning the software is based on local assessment of relative volumetric differ­ can be used as medical devices. ences between two MRI images of the same individual, Importantly, the companies must provide the magni­ one of which is usually the baseline image; the net sum tude of the error in their results, and health care profes­ of all local volumetric changes provides an estimate of sionals should use this information to validate or discard total volume changes over time. Finally, cortical thickness findings of analyses. All four commercial software pack­ changes can be detected by the use of a within-subject tem­ ages have been evaluated scientifically to some extent but plate (an MRI image created by merging all MRI images not completely. To our knowledge, only MSmetrix has from one individual) to improve cortical thickness esti­ been validated by an independent group in the context of mation at each time point, or by fitting a subject-specific MS154. Furthermore, the real-​world clinical value of these cortical deformable model at each time point145,146. software packages has not yet been assessed, and the pro­ Assessment of spinal cord atrophy is more difficult cedures are not widely reimbursed (with a few exceptions, than brain segmentation owing to particular anatomical such as in the USA). Although promising, these analyt­ (higher mobility and smaller dimensions than the brain) ical approaches should therefore be more extensively and imaging (lower tissue contrast) features of the spinal validated by expert groups in the field of MRI preprocess­ cord. Semiautomated (Cordial)147 and automated (Spinal ing, especially in the context of MS134, before they can be Cord Toolbox)148 tools have now been developed, based considered for use in the routine clinical setting. on deformable models. These promising new software tools still need to be extensively validated on independent Statements and recommendations datasets before they can even be considered for use in 1. We recommend appropriate management of several clinical practice. scanner-​related factors (including, but not limited to, Academic software packages have important advan­ variation in acquisition protocols, different scanner tages over commercial software packages, such as the systems and upgrades, movement artefacts and gra­ fact that they have been validated in many studies under dient distortions) to ensure reliability of brain volume a plethora of different MRI conditions over the past dec­ estimates, particularly at an individual patient level. ade. However, they have the severe limitation of being 2. We recommend appropriate management of physio­ highly technically demanding and their use is therefore logical and MS-​related factors (including, but not limited to centres that are specialized in MRI processing. limited to, age, sex, hydration status, time of day, In addition, clinical application of software to support steroid use and MS-​related parenchymal alterations). diagnosis or care is only permitted with products that 3. Brain volume measures are software-​dependent so have received the “Conformité Européenne” (CE) mark the use of software that has been approved as a med­ in Europe or FDA clearance in the USA. For this reason, ical device and independently evaluated in MS is a translation of imaging analysis software tools to clini­ prerequisite; we recommend further research to vali­ cal practice is challenging and almost unfeasible for date existing software tools in MS and assess their academic neuroimaging laboratories. clinical value. In the past 10 years, several companies have proposed centralized MRI reading services, often using their in-​ Conclusions and future directions house software for quantification of atrophy (Table 2). Based on the evidence reviewed, the idea that brain vol­ Four software packages have been approved for use in ume changes and, to a lesser extent, spinal cord atrophy Europe and three of these have also received FDA clear­ are helpful predictors of the evolution of MS before ini­ ance in the USA. The IcoBrain MS (Icometrix, previ­ tiation of therapy is undisputed, so these measures could ously MSmetrix)149 quantifies cross-​sectional volumes be valid treatment-​decision tools. The evidence reviewed with software based on Nifty Seg and quantifies longitu­ also supports the idea that brain volume measures have dinal changes in grey matter and white matter with soft­ value in monitoring the effects of MS drugs as part of ware that implements Jacobian integration. NeuroQuant the no evidence of disease activity outcome measure or (CorTechs Labs)150 provides both cross-​sectional and minimal evidence of disease activity outcome measure. longitudinal quantification of atrophy151, building on However, several potential sources of substantial error Deformable model approaches already developed by previous methods138. remain, including, but not limited to, differential effects of A mathematical construct Biometrica MS (Jung Diagnostics) builds on develop­ drugs on brain volume measures, confounding physio­ capable of representing a ments of Statistical Parametric Mapping, a software logical and technical factors and the performance and broad range of shapes that can be used in MRI image library for neuroimaging analysis, for atrophy measure­ value of volumetric tools. To make implementation of registration or segmentation. ment and of Lesion Segmentation Tool software for auto­ volume measurements in clinical practice feasible, these matic lesion segmentation152,153. Quantib Brain (Quantib) potential sources of error need to be accounted for and Nifty Seg is a platform that is integrated into the General Electric appropriately managed, and further research is needed to An open-​source image analysis tool developed at University MRI scanner and can assess cross-​sectional brain vol­ ensure the accuracy and reliability of the measurements. College London to perform umes and longitudinal changes in volume. IcoBrain segmentations of MRI images. MS and Biometrica MS are offered as remote analysis Published online 24 February 2020 NaTure RevIewS | NeuroLogy volume 16 | March 2020 | 179 C o n S e n S u S S tat e m e n t 1. Kawachi, I. & Lassmann, H. Neurodegeneration in 27. Perez-​Miralles, F. et al. Clinical impact of early brain 53. Quarantelli, M. et al. Brain tissue volume changes in multiple sclerosis and neuromyelitis optica. J. Neurol. atrophy in clinically isolated syndromes. Mult. Scler. relapsing-​remitting multiple sclerosis: correlation with Neurosurg. Psychiatry 88, 137–145 (2017). 19, 1878–1886 (2013). lesion load. Neuroimage 18, 360–366 (2003). 2. Ferguson, B., Matyszak, M. K., Esiri, M. M. & 28. Di Filippo, M. et al. Brain atrophy and lesion load 54. Fisher, E., Lee, J. C., Nakamura, K. & Rudick, R. A. Perry, V. H. Axonal damage in acute multiple sclerosis measures over 1 year relate to clinical status after Gray matter atrophy in multiple sclerosis: lesions. Brain 120, 393–399 (1997). 6 years in patients with clinically isolated syndromes. a longitudinal study. Ann. Neurol. 64, 255–265 3. Trapp, B. D. et al. Axonal transection in the lesions J. Neurol. Neurosurg. Psychiatry 81, 204–208 (2008). of multiple sclerosis. N. Engl. J. Med. 338, 278–285 (2010). 55. Roosendaal, S. D. et al. Grey matter volume in a large (1998). 29. Minneboo, A. et al. Predicting short-​term disability cohort of MS patients: relation to MRI parameters 4. Kuhlmann, T., Lingfeld, G., Bitsch, A., Schuchardt, J. & progression in early multiple sclerosis: added value and disability. Mult. Scler. 17, 1098–1106 (2011). Bruck, W. Acute axonal damage in multiple sclerosis is of MRI parameters. J. Neurol. Neurosurg. Psychiatry 56. Dalton, C. M. et al. Early development of multiple most extensive in early disease stages and decreases 79, 917–923 (2008). sclerosis is associated with progressive grey matter over time. Brain 125, 2202–2212 (2002). 30. Sastre-​Garriga, J. et al. Long-​term clinical outcome atrophy in patients presenting with clinically isolated 5. Bjartmar, C., Kidd, G., Mork, S., Rudick, R. & of primary progressive MS: predictive value of clinical syndromes. Brain 127, 1101–1107 (2004). Trapp, B. D. Neurological disability correlates with and MRI data. Neurology 65, 633–635 (2005). 57. Valsasina, P. et al. Evidence for progressive gray spinal cord axonal loss and reduced N-​acetyl 31. Khaleeli, Z. et al. Predicting progression in primary matter loss in patients with relapsing-​remitting MS. aspartate in chronic multiple sclerosis patients. progressive multiple sclerosis: a 10-year multicenter Neurology 65, 1126–1128 (2005). Ann. Neurol. 48, 893–901 (2000). study. Ann. Neurol. 63, 790–793 (2008). 58. Tiberio, M. et al. Gray and white matter volume 6. Sastre-​Garriga, J. et al. Brain atrophy in natalizumab-​ 32. Rocca, M. A. et al. Long-​term disability progression changes in early RRMS: a 2-year longitudinal study. treated patients: a 3-year follow-​up. Mult. Scler. 21, in primary progressive multiple sclerosis: a 15-year Neurology 64, 1001–1007 (2005). 749–756 (2015). study. Brain 140, 2814–2819 (2017). 59. Sastre-​Garriga, J. et al. Grey and white matter 7. Popescu, V. et al. Postmortem validation of MRI 33. Popescu, V. et al. Brain atrophy and lesion load volume changes in early primary progressive cortical volume measurements in MS. Hum. Brain predict long term disability in multiple sclerosis. multiple sclerosis: a longitudinal study. Brain 128, Mapp. 37, 2223–2233 (2016). J. Neurol. Neurosurg. Psychiatry 84, 1082–1091 1454–1460 (2005). 8. Losseff, N. A. et al. Progressive cerebral atrophy (2013). 60. Fisniku, L. K. et al. Gray matter atrophy is related to in multiple sclerosis. A serial MRI study. Brain 119, 34. Miller, D. H. et al. Brain atrophy and disability long-​term disability in multiple sclerosis. Ann. Neurol. 2009–2019 (1996). worsening in primary progressive multiple sclerosis: 64, 247–254 (2008). 9. Losseff, N. A. et al. Spinal cord atrophy and disability insights from the INFORMS study. Ann. Clin. Transl. 61. Filippi, M. et al. Gray matter damage predicts the in multiple sclerosis. A new reproducible and sensitive Neurol. 5, 346–356 (2018). accumulation of disability 13 years later in MS. MRI method with potential to monitor disease 35. Sormani, M. P. et al. Defining brain volume cutoffs Neurology 81, 1759–1767 (2013). progression. Brain 119, 701–708 (1996). to identify clinically relevant atrophy in RRMS. 62. Jacobsen, C. et al. Brain atrophy and disability 10. Miller, D. H. et al. Guidelines for the use of magnetic Mult. Scler. 23, 656–664 (2017). progression in multiple sclerosis patients: a 10-year resonance techniques in monitoring the treatment of 36. Daams, M. et al. Mean upper cervical cord area follow-​up study. J. Neurol. Neurosurg. Psychiatry 85, multiple sclerosis. US National MS Society task force. (MUCCA) measurement in long-​standing multiple 1109–1115 (2014). Ann. Neurol. 39, 6–16 (1996). sclerosis: relation to brain findings and clinical 63. Narayana, P. A. et al. Regional cortical thickness in 11. Thompson, A. J. et al. Diagnosis of multiple sclerosis: disability. Mult. Scler. 20, 1860–1865 (2014). relapsing remitting multiple sclerosis: a multi-​center 2017 revisions of the McDonald criteria. Lancet 37. Bieniek, M. et al. Cord atrophy separates early study. Neuroimage Clin. 2, 120–131 (2012). Neurol. 17, 162–173 (2017). primary progressive and relapsing remitting multiple 64. Lansley, J., Mataix-​Cols, D., Grau, M., Radua, J. 12. McDonald, W. I. et al. Recommended diagnostic sclerosis. J. Neurol. Neurosurg. Psychiatry 77, & Sastre-​Garriga, J. Localized grey matter atrophy in criteria for multiple sclerosis: guidelines from the 1036–1039 (2006). multiple sclerosis: a meta-​analysis of voxel-​based international panel on the diagnosis of multiple 38. Lukas, C. et al. Relevance of spinal cord abnormalities morphometry studies and associations with functional sclerosis. Ann. Neurol. 50, 121–127 (2001). to clinical disability in multiple sclerosis: MR imaging disability. Neurosci. Biobehav. Rev. 37, 819–830 13. Lublin, F. D. et al. Defining the clinical course of findings in a large cohort of patients. Radiology 269, (2013). multiple sclerosis: the 2013 revisions. Neurology 83, 542–552 (2013). 65. Eshaghi, A. et al. Deep gray matter volume loss drives 278–286 (2014). 39. Rocca, M. A. et al. A multicenter assessment of disability worsening in multiple sclerosis. Ann. Neurol. 14. Miller, D. H., Barkhof, F., Frank, J. A., Parker, G. J. & cervical cord atrophy among MS clinical phenotypes. 83, 210–222 (2018). Thompson, A. J. Measurement of atrophy in multiple Neurology 76, 2096–2102 (2011). 66. Pichler, A. et al. Combined analysis of global and sclerosis: pathological basis, methodological aspects 40. Daams, M. et al. Unraveling the neuroimaging compartmental brain volume changes in early multiple and clinical relevance. Brain 125, 1676–1695 predictors for motor dysfunction in long-​standing sclerosis in clinical practice. Mult. Scler. 22, 340–346 (2002). multiple sclerosis. Neurology 85, 248–255 (2015). (2016). 15. Rocca, M. A. et al. Brain MRI atrophy quantification 41. Oh, J. et al. Relationships between quantitative spinal 67. Vidal-​Jordana, A., Sastre-​Garriga, J., Rovira, A. & in MS: from methods to clinical application. Neurology cord MRI and retinal layers in multiple sclerosis. Montalban, X. Treating relapsing-​remitting multiple 88, 403–413 (2017). Neurology 84, 720–728 (2015). sclerosis: therapy effects on brain atrophy. J. Neurol. 16. Wattjes, M. P. et al. Evidence-​based guidelines: 42. Kearney, H. et al. Magnetic resonance imaging 262, 2617–2626 (2015). MAGNIMS consensus guidelines on the use of MRI correlates of physical disability in relapse onset 68. Filippi, M. et al. Interferon beta-1a for brain tissue in multiple sclerosis — establishing disease prognosis multiple sclerosis of long disease duration. Mult. Scler. loss in patients at presentation with syndromes and monitoring patients. Nat. Rev. Neurol. 11, 20, 72–80 (2014). suggestive of multiple sclerosis: a randomised, 597–606 (2015). 43. Tsagkas, C. et al. Spinal cord volume loss: a marker of double-​blind, placebo-​controlled trial. Lancet 364, 17. Rovira, A. et al. Evidence-​based guidelines: MAGNIMS disease progression in multiple sclerosis. Neurology 1489–1496 (2004). consensus guidelines on the use of MRI in multiple 91, e349–e358 (2018). 69. Rudick, R. A., Fisher, E., Lee, J. C., Duda, J. T. & sclerosis — clinical implementation in the diagnostic 44. Schlaeger, R. et al. Spinal cord gray matter Simon, J. Brain atrophy in relapsing multiple sclerosis: process. Nat. Rev. Neurol. 11, 471–482 (2015). atrophy correlates with multiple sclerosis disability. relationship to relapses, EDSS, and treatment with 18. Rao, S. M. et al. Chronic progressive multiple Ann. Neurol. 76, 568–580 (2014). interferon beta-1a. Mult. Scler. 6, 365–372 (2000). sclerosis. Relationship between cerebral ventricular 45. Yiannakas, M. C., Kakar, P., Hoy, L. R., Miller, D. H. & 70. Rudick, R. A., Fisher, E., Lee, J. C., Simon, J. & size and neuropsychological impairment. Arch. Neurol. Wheeler-​Kingshott, C. A. The use of the lumbosacral Jacobs, L. Use of the brain parenchymal fraction to 42, 678–682 (1985). enlargement as an intrinsic imaging biomarker: measure whole brain atrophy in relapsing-​remitting 19. Comi, G. et al. Brain magnetic resonance imaging feasibility of grey matter and white matter cross-​ MS. Multiple Sclerosis Collaborative Research Group. correlates of cognitive impairment in multiple sectional area measurements using MRI at 3 T. Neurology 53, 1698–1704 (1999). sclerosis. J. Neurol. Sci. 115, S66–S73 (1993). PLoS One 9, e105544 (2014). 71. Fisher, E. et al. Effect of intramuscular interferon 20. Huber, S. J. et al. Magnetic resonance imaging 46. Rocca, M. A. et al. Voxel-​wise mapping of cervical cord beta-1a on gray matter atrophy in relapsing-​remitting correlates of dementia in multiple sclerosis. damage in multiple sclerosis patients with different multiple sclerosis: a retrospective analysis. Mult. Scler. Arch. Neurol. 44, 732–736 (1987). clinical phenotypes. J. Neurol. Neurosurg. Psychiatry 22, 668–676 (2016). 21. De Stefano, N. et al. Assessing brain atrophy rates 84, 35–41 (2013). 72. Vidal-​Jordana, A. et al. Early brain pseudoatrophy in a large population of untreated multiple sclerosis 47. Lukas, C. et al. Cervical spinal cord volume loss is while on natalizumab therapy is due to white matter subtypes. Neurology 74, 1868–1876 (2010). related to clinical disability progression in multiple volume changes. Mult. Scler. 19, 1175–1181 (2013). 22. Amato, M. P. et al. Association of MRI metrics and sclerosis. J. Neurol. Neurosurg. Psychiatry 86, 73. Vidal-​Jordana, A. et al. Brain volume loss during the cognitive impairment in radiologically isolated 410–418 (2015). first year of interferon-​beta treatment in multiple syndromes. Neurology 78, 309–314 (2012). 48. Tsagkas, C. et al. Preferential spinal cord volume loss sclerosis: baseline inflammation and regional brain 23. Battaglini, M. et al. Lifespan normative data on rates in primary progressive multiple sclerosis. Mult. Scler. volume dynamics. J. Neuroimaging 26, 532–538 of brain volume changes. Neurobiol. Aging 81, 30–37 25, 947–957 (2019). (2016). (2019). 49. Kearney, H. et al. Cervical cord lesion load is associated 74. Comi, G. et al. Effects of early treatment with 24. De Stefano, N. et al. Establishing pathological cut-​offs with disability independently from atrophy in MS. glatiramer acetate in patients with clinically isolated of brain atrophy rates in multiple sclerosis. J. Neurol. Neurology 84, 367–373 (2015). syndrome. Mult. Scler. 19, 1074–1083 (2013). Neurosurg. Psychiatry 87, 93–99 (2016). 50. Chard, D. T. et al. Brain atrophy in clinically early 75. Wolinsky, J. S. et al. Glatiramer acetate in primary 25. Perez-​Miralles, F. C. et al. Predictive value of early relapsing-​remitting multiple sclerosis. Brain 125, progressive multiple sclerosis: results of a brain atrophy on response in patients treated with 327–337 (2002). multinational, multicenter, double-​blind, placebo-​ interferon beta. Neurol. Neuroimmunol. 51. Sastre-​Garriga, J. et al. Grey and white matter atrophy controlled trial. Ann. Neurol. 61, 14–24 (2007). Neuroinflamm. 2, e132 (2015). in early clinical stages of primary progressive multiple 76. Montalban, X. et al. A single-​center, randomized, 26. Andorra, M. et al. Assessing biological and sclerosis. Neuroimage 22, 353–359 (2004). double-​blind, placebo-​controlled study of interferon methodological aspects of brain volume loss in 52. De Stefano, N. et al. Evidence of early cortical beta-1b on primary progressive and transitional multiple sclerosis. JAMA Neurol. 75, 1246–1255 atrophy in MS: relevance to white matter changes multiple sclerosis. Mult. Scler. 15, 1195–1205 (2018). and disability. Neurology 60, 1157–1162 (2003). (2009). 180 | March 2020 | volume 16 www.nature.com/nrneurol C o n S e n S u S S tat e m e n t 77. Miller, D. H. et al. MRI outcomes in a placebo-​ phantom and human data. Neuroimage 30, 436–443 128. Hooshmand, B. et al. Association of vitamin B12, controlled trial of natalizumab in relapsing MS. (2006). folate, and sulfur amino acids with brain magnetic Neurology 68, 1390–1401 (2007). 103. Han, X. et al. Reliability of MRI-​derived measurements resonance imaging measures in older adults: a 78. Branger, P., Parienti, J. J., Sormani, M. P. & Defer, G. of human cerebral cortical thickness: the effects of longitudinal population-​based study. JAMA Psychiatry The effect of disease-​modifying drugs on brain atrophy field strength, scanner upgrade and manufacturer. 73, 606–613 (2016). in relapsing-​remitting multiple sclerosis: a meta-​ Neuroimage 32, 180–194 (2006). 129. Pichler, A. et al. The impact of vascular risk factors on analysis. PLoS One 11, e0149685 (2016). 104. Maclaren, J., Han, Z., Vos, S. B., Fischbein, N. & brain volume and lesion load in patients with early 79. Cohen, J. A. et al. Alemtuzumab versus interferon Bammer, R. Reliability of brain volume measurements: multiple sclerosis. Mult. Scler. 25, 48–54 (2019). beta 1a as first-​line treatment for patients with a test–retest dataset. Sci. Data 1, 140037 (2014). 130. Chard, D. T., Jackson, J. S., Miller, D. H. & Wheeler-​ relapsing-​remitting multiple sclerosis: a randomised 105. Morey, R. A. et al. Scan–rescan reliability of Kingshott, C. A. Reducing the impact of white matter controlled phase 3 trial. Lancet 380, 1819–1828 subcortical brain volumes derived from automated lesions on automated measures of brain gray and (2012). segmentation. Hum. Brain Mapp. 31, 1751–1762 white matter volumes. J. Magn. Reson. Imaging 32, 80. Coles, A. J. et al. Alemtuzumab for patients with (2010). 223–228 (2010). relapsing multiple sclerosis after disease-​modifying 106. Sampat, M. P. et al. Disease modeling in multiple 131. Battaglini, M., Jenkinson, M. & De Stefano, N. therapy: a randomised controlled phase 3 trial. Lancet sclerosis: assessment and quantification of sources of Evaluating and reducing the impact of white matter 380, 1829–1839 (2012). variability in brain parenchymal fraction lesions on brain volume measurements. Hum. Brain 81. Cohen, J. A. et al. Oral fingolimod or intramuscular measurements. Neuroimage 52, 1367–1373 (2010). Mapp. 33, 2062–2071 (2012). interferon for relapsing multiple sclerosis. N. Engl. J. 107. Yang, C. Y. et al. Reproducibility of brain morphometry 132. Cheriyan, J., Kim, S., Wolansky, L. J., Cook, S. D. & Med. 362, 402–415 (2010). from short-​term repeat clinical MRI examinations: a Cadavid, D. Impact of inflammation on brain volume 82. Arnold, D. L. et al. Effects of delayed-​release dimethyl retrospective study. PLoS One 11, e0146913 (2016). in multiple sclerosis. Arch. Neurol. 69, 82–88 (2012). fumarate on MRI measures in the phase 3 DEFINE 108. Biberacher, V. et al. Intra- and interscanner variability 133. Zivadinov, R. et al. Mechanisms of action of disease-​ study. J. Neurol. 261, 1794–1802 (2014). of magnetic resonance imaging based volumetry in modifying agents and brain volume changes in 83. De Stefano, N. et al. Reduced brain atrophy rates are multiple sclerosis. Neuroimage 142, 188–197 multiple sclerosis. Neurology 71, 136–144 (2008). associated with lower risk of disability progression in (2016). 134. Storelli, L. et al. Measurement of whole-​brain and gray patients with relapsing multiple sclerosis treated with 109. Reuter, M. et al. Head motion during MRI acquisition matter atrophy in multiple sclerosis: assessment with cladribine tablets. Mult. Scler. 24, 222–226 (2018). reduces gray matter volume and thickness estimates. MR imaging. Radiology 288, 554–564 (2018). 84. Sormani, M. P., Arnold, D. L. & De Stefano, N. Neuroimage 107, 107–115 (2015). 135. Ashburner, J. & Friston, K. J. Unified segmentation. Treatment effect on brain atrophy correlates with 110. Godenschweger, F. et al. Motion correction in MRI Neuroimage 26, 839–851 (2005). treatment effect on disability in multiple sclerosis. of the brain. Phys. Med. Biol. 61, R32–R56 (2016). 136. Cardoso, M. J. et al. Geodesic information flows: Ann. Neurol. 75, 43–49 (2014). 111. Jovicich, J. et al. MRI-​derived measurements of spatially-​variant graphs and their application to 85. Chataway, J. et al. Effect of high-​dose simvastatin on human subcortical, ventricular and intracranial brain segmentation and fusion. IEEE Trans. Med. Imaging brain atrophy and disability in secondary progressive volumes: reliability effects of scan sessions, acquisition 34, 1976–1988 (2015). multiple sclerosis (MS-​STAT): a randomised, placebo-​ sequences, data analyses, scanner upgrade, 137. Avants, B. B., Tustison, N. J., Wu, J., Cook, P. A. & controlled, phase 2 trial. Lancet 383, 2213–2221 scanner vendors and field strengths. Neuroimage 46, Gee, J. C. An open source multivariate framework for (2014). 177–192 (2009). n-​tissue segmentation with evaluation on public data. 86. Fox, R. J. et al. Phase 2 trial of ibudilast in progressive 112. Barnes, J. et al. Head size, age and gender adjustment Neuroinformatics 9, 381–400 (2011). multiple sclerosis. N. Engl. J. Med. 379, 846–855 in MRI studies: a necessary nuisance? Neuroimage 138. Fischl, B., Sereno, M. I. & Dale, A. M. Cortical (2018). 53, 1244–1255 (2010). surface-​based analysis. II: inflation, flattening, and 87. Kearney, H., Miller, D. H. & Ciccarelli, O. Spinal cord 113. Smith, S. M. et al. Accurate, robust, and automated a surface-​based coordinate system. Neuroimage 9, MRI in multiple sclerosis — diagnostic, prognostic and longitudinal and cross-​sectional brain change analysis. 195–207 (1999). clinical value. Nat. Rev. Neurol. 11, 327–338 (2015). Neuroimage 17, 479–489 (2002). 139. MacDonald, D., Kabani, N., Avis, D. & Evans, A. C. 88. Kapoor, R. et al. Lamotrigine for neuroprotection 114. Amiri, H. et al. Urgent challenges in quantification Automated 3-D extraction of inner and outer surfaces in secondary progressive multiple sclerosis: a and interpretation of brain grey matter atrophy in of cerebral cortex from MRI. Neuroimage 12, randomised, double-​blind, placebo-​controlled, parallel-​ individual MS patients using MRI. Neuroimage Clin. 340–356 (2000). group trial. Lancet Neurol. 9, 681–688 (2010). 19, 466–475 (2018). 140. Fox, N. C. & Freeborough, P. A. Brain atrophy 89. Cawley, N. et al. Spinal cord atrophy as a primary 115. Enzinger, C. et al. Risk factors for progression of brain progression measured from registered serial MRI: outcome measure in phase II trials of progressive atrophy in aging: six-​year follow-​up of normal subjects. validation and application to Alzheimer’s disease. multiple sclerosis. Mult. Scler. 24, 932–941 (2018). Neurology 64, 1704–1711 (2005). J. Magn. Reson. Imaging 7, 1069–1075 (1997). 90. Leary, S. M. et al. Interferon beta-1a in primary 116. Fraser, M. A., Shaw, M. E. & Cherbuin, N. A systematic 141. Dwyer, M. G., Bergsland, N. & Zivadinov, R. Improved progressive MS: an exploratory, randomized, review and meta-​analysis of longitudinal hippocampal longitudinal gray and white matter atrophy assessment controlled trial. Neurology 60, 44–51 (2003). atrophy in healthy human ageing. Neuroimage 112, via application of a 4-dimensional hidden Markov 91. Lublin, F. et al. Oral fingolimod in primary progressive 364–374 (2015). random field model. Neuroimage 90, 207–217 multiple sclerosis (INFORMS): a phase 3, randomised, 117. Ruigrok, A. N. et al. A meta-​analysis of sex differences (2014). double-​blind, placebo-​controlled trial. Lancet 387, in human brain structure. Neurosci. Biobehav. Rev. 142. Battaglini, M., Jenkinson, M. & De Stefano, N. 1075–1084 (2016). 39, 34–50 (2014). SIENA-​XL for improving the assessment of gray 92. Sormani, M. P. et al. Fingolimod effect on brain 118. Kiraly, A. et al. Male brain ages faster: the age and white matter volume changes on brain MRI. volume loss independently contributes to its effect and gender dependence of subcortical volumes. Hum. Brain Mapp. 39, 1063–1077 (2018). on disability. Mult. Scler. 21, 916–924 (2015). Brain Imaging Behav. 10, 901–910 (2016). 143. Ashburner, J. & Ridgway, G. R. Symmetric 93. Fisher, E. et al. Eight-​year follow-​up study of 119. Good, C. D. et al. Cerebral asymmetry and the effects diffeomorphic modeling of longitudinal structural MRI. brain atrophy in patients with MS. Neurology 59, of sex and handedness on brain structure: a voxel-​ Front. Neurosci. 6, 197 (2012). 1412–1420 (2002). based morphometric analysis of 465 normal adult 144. Nakamura, K. et al. Jacobian integration method 94. Rojas, J. I., Patrucco, L., Miguez, J., Besada, C. human brains. Neuroimage 14, 685–700 (2001). increases the statistical power to measure gray matter & Cristiano, E. Brain atrophy as a non-​response 120. Luders, E. et al. Gender effects on cortical thickness atrophy in multiple sclerosis. Neuroimage Clin. 4, predictor to interferon-​beta in relapsing-​remitting and the influence of scaling. Hum. Brain Mapp. 27, 10–17 (2014). multiple sclerosis. Neurol. Res. 36, 615–618 (2014). 314–324 (2006). 145. Reuter, M. & Fischl, B. Avoiding asymmetry-​induced 95. Radue, E. W. et al. Correlation between brain volume 121. De Bondt, T., Pullens, P., Van Hecke, W., Jacquemyn, Y. bias in longitudinal image processing. Neuroimage loss and clinical and MRI outcomes in multiple & Parizel, P. M. Reproducibility of hormone-​driven 57, 19–21 (2011). sclerosis. Neurology 84, 784–793 (2015). regional grey matter volume changes in women 146. Nakamura, K., Fox, R. & Fisher, E. CLADA: cortical 96. Jeffery, D. R. et al. The relationship between the using SPM8 and SPM12. Brain Struct. Funct. 221, longitudinal atrophy detection algorithm. Neuroimage rate of brain volume loss during first 24 months and 4631–4641 (2016). 54, 278–289 (2011). disability progression over 24 and 48 months in 122. Nakamura, K. et al. Diurnal fluctuations in brain 147. Amann, M. et al. Reliable volumetry of the cervical relapsing MS. J. Neurol. 263, 299–305 (2016). volume: statistical analyses of MRI from large spinal cord in MS patient follow-​up data with cord 97. Kappos, L. et al. Inclusion of brain volume loss in a populations. Neuroimage 118, 126–132 (2015). image analyzer (Cordial). J. Neurol. 263, 1364–1374 revised measure of ‘no evidence of disease activity’ 123. Duning, T. et al. Dehydration confounds the (2016). (NEDA-4) in relapsing-​remitting multiple sclerosis. assessment of brain atrophy. Neurology 64, 548–550 148. De Leener, B. et al. SCT: spinal cord toolbox, an open-​ Mult. Scler. 22, 1297–1305 (2016). (2005). source software for processing spinal cord MRI data. 98. Gasperini, C. et al. Unraveling treatment response 124. Klaren, R. E. et al. Objectively measured physical Neuroimage 145, 24–43 (2017). in multiple sclerosis: a clinical and MRI challenge. activity is associated with brain volumetric 149. Jain, S. et al. Automatic segmentation and volumetry Neurology 92, 180–192 (2019). measurements in multiple sclerosis. Behav. Neurol. of multiple sclerosis brain lesions from MR images. 99. Rio, J. et al. Disability progression markers over 2015, 482536 (2015). Neuroimage Clin. 8, 367–375 (2015). 6–12 years in interferon-​beta-treated multiple 125. Mechtcheriakov, S. et al. A widespread distinct pattern 150. Ross, D. E., Ochs, A. L., Seabaugh, J. M. & sclerosis patients. Mult. Scler. 24, 322–330 (2018). of cerebral atrophy in patients with alcohol addiction Shrader, C. R. Man versus machine: comparison 100. Mueller, S. G. et al. Ways toward an early diagnosis revealed by voxel-​based morphometry. J. Neurol. of radiologists’ interpretations and NeuroQuant® in Alzheimer’s disease: the Alzheimer’s Disease Neurosurg. Psychiatry 78, 610–614 (2007). volumetric analyses of brain MRIs in patients with Neuroimaging Initiative (ADNI). Alzheimers Dement. 126. Mashhoon, Y., Sava, S., Sneider, J. T., Nickerson, L. D. traumatic brain injury. J. Neuropsychiatry Clin. 1, 55–66 (2005). & Silveri, M. M. Cortical thinness and volume Neurosci. 25, 32–39 (2013). 101. Caramanos, Z. et al. Gradient distortions in MRI: differences associated with marijuana abuse in 151. Brewer, J. B. Fully-​automated volumetric MRI with characterizing and correcting for their effects on emerging adults. Drug Alcohol Depend. 155, normative ranges: translation to clinical practice. SIENA-​generated measures of brain volume change. 275–283 (2015). Behav. Neurol. 21, 21–28 (2009). Neuroimage 49, 1601–1611 (2010). 127. Beauchet, O. et al. Blood pressure levels and brain 152. Opfer, R. et al. Atlas based brain volumetry: how to 102. Jovicich, J. et al. Reliability in multi-​site structural MRI volume reduction: a systematic review and meta-​ distinguish regional volume changes due to biological studies: effects of gradient non-​linearity correction on analysis. J. Hypertens. 31, 1502–1516 (2013). or physiological effects from inherent noise of the NaTure RevIewS | NeuroLogy volume 16 | March 2020 | 181 C o n S e n S u S S tat e m e n t methodology. Magn. Reson. Imaging 34, 455–461 171. Montalban, X. et al. Ocrelizumab versus placebo in Roche and Teva, all unrelated to this Consensus Statement. (2016). primary progressive multiple sclerosis. N. Engl. J. M.T. declares personal fees from Almirall, Bayer Healthcare, 153. Spies, L. et al. Fully automatic detection of deep white Med. 376, 209–220 (2017). Biogen Idec, Merck Serono, Novartis, Roche, Sanofi-​Genzyme matter T1 hypointense lesions in multiple sclerosis. 172. Leist, T. P. et al. Effect of oral cladribine on time to and Teva, grants from Sanofi-​Genzyme, and other competing Phys. Med. Biol. 58, 8323–8337 (2013). conversion to clinically definite multiple sclerosis in interests related to Biogen Idec, all unrelated to this 154. Steenwijk, M. D. et al. Agreement of MSmetrix with patients with a first demyelinating event (ORACLE Consensus Statement. M.F. declares personal fees from established methods for measuring cross-​sectional MS): a phase 3 randomised trial. Lancet Neurol. 13, Biogen Idec, Merck-​Serono, Novartis and Teva, and grants and longitudinal brain atrophy. Neuroimage Clin. 15, 257–267 (2014). from Biogen Idec, Merck-​Serono, Novartis, Roche and Teva, 843–853 (2017). all unrelated to this Consensus Statement. He is the Editor-​ 155. De Stefano, N. et al. Efficacy of subcutaneous Acknowledgements in-Chief of the Journal of Neurology. C.G. declares advisory interferon beta-1a on MRI outcomes in a randomised O.C., F.B. and T.A.Y. receive support from the National board membership for Biogen, Genzyme, Merck-​Serono, controlled trial of patients with clinically isolated Institute for Health Research University College London Roche and Teva, and honoraria for speaking or consultation syndromes. J. Neurol. Neurosurg. Psychiatry 85, Hospitals Biomedical Research Centre. fees from Almirall, Bayer Schering, Biogen Idec, Genzyme, 647–653 (2014). Merck Serono, Novartis, Roche and Teva. L.K. declares com- 156. Kappos, L. et al. Long-​term subcutaneous interferon Author contributions peting interests related to Actelion, Almirall, Alkermes, Bayer, beta-1a therapy in patients with relapsing-​remitting J.S.-G. and A.R. developed the idea for the manuscript and Biogen, df-​mp, The European Union, Excemed, GeNeuro, MS. Neurology 67, 944–953 (2006). created the structure. J.S.-G., D.P. and M.B. each wrote spe- Genzyme, Japan Tobacco, Merck, Minoryx, Mitsubishi 157. Kappos, L. et al. Effect of early versus delayed cific sections of the manuscript. All authors participated in the Pharma, Neurostatus, Novartis, Receptos–Celgene, Roche, interferon beta-1b treatment on disability after a first workshop, commented on and revised the manuscript and Roche Research Foundations, Sanofi-​Aventis, Santhera, the clinical event suggestive of multiple sclerosis: a 3-year approved the final version. Swiss Multiple Sclerosis Society, Swiss National Research follow-​up analysis of the BENEFIT study. Lancet 370, Foundation, Teva and Vianex, all unrelated to this Consensus 389–397 (2007). Competing interests Statement. J.R. declares personal fees from Biogen, Merck, 158. Molyneux, P. D. et al. The effect of interferon beta-1b J.S.-G. declares grants and personal fees from Genzyme, per- Novartis and Roche, and other competing interests related to treatment on MRI measures of cerebral atrophy in sonal fees from Almirall, Biogen, Celgene, Merck, Novartis, Merck and Mylan, all unrelated to this Consensus Statement. secondary progressive multiple sclerosis. European Roche and Teva and is a member of the Editorial Committee J.F. declares personal fees from Biogen Idec, Merck Serono Study Group on Interferon beta-1b in secondary of Multiple Sclerosis Journal and Director of the Scientific and Sanofi-​Aventis, participation in scientific advisory boards progressive multiple sclerosis. Brain 123, Committee of Revista de Neurologia, all unrelated to this for Almiral, Genzyme and Novartis, personal fees (speaker 2256–2263 (2000). Consensus Statement. D.P. declares personal fees from honoraria) from Biogen Idec, Merck Serono, Santhera 159. Comi, G. et al. Effect of glatiramer acetate on Novartis and Sanofi-​Genzyme, unrelated to this Consensus and Teva, and participation as an advisor on preclinical conversion to clinically definite multiple sclerosis in Statement. M.A.R. declares personal fees from Biogen Idec, development for Takeda, all unrelated to this Consensus patients with clinically isolated syndrome (PreCISe Genzyme, Merck Serono, Novartis, Roche, Sanofi-​Aventis and Statement. J.P. declares grants and personal fees from study): a randomised, double-​blind, placebo-​controlled Teva, all unrelated to this Consensus Statement. O.C. declares Alexion, Biogen Idec, Chugai, and Merck Serono, personal trial. Lancet 374, 1503–1511 (2009). grants from the Spinal Cord Research Foundation, the fees from ABIDE, ARGENX, Med Day, Novartis, Roche and 160. Sormani, M. P. et al. Measurement error of two Rosetree Trust, the Progressive MS Alliance, Bioclinica, Teva, and grants from ABIDE and MedImmun, all unrelated different techniques for brain atrophy assessment GE Neuro, EU-​H2020, the UK MS Society, the National MS to this Consensus Statement. She also has a patent (patent in multiple sclerosis. Neurology 62, 1432–1434 Society, and the National Institute for Health Research number 13704627.2-1408) with Isis: Diagnosing Multiple (2004). Biomedical Research Centre, University College London Sclerosis. H.V. declares grants from Merck Serono, Novartis 161. Miller, A. E. et al. Oral teriflunomide for patients Hospitals, and declares personal fees from Merck, Neurology, and Teva, and other competing interests related to Merck and with a first clinical episode suggestive of multiple Novartis, Roche and Teva, all unrelated to this Consensus Novartis, all unrelated to this Consensus Statement. He also sclerosis (TOPIC): a randomised, double-​blind, Statement. O.C. also declares grants from the UK MS Society, has a patent related to brain atrophy measurement pending. placebo-​controlled, phase 3 trial. Lancet Neurol. 13, the National Institute for Health Research Biomedical X.M. declares personal fees from Actelion, Biogen, Celgene, 977–986 (2014). Research Centre, University College London Hospitals, and Merck, Novartis, Roche, Sanofi-​Genzyme and Teva, all unre- 162. O’Connor, P. et al. Randomized trial of oral the National MS Society during the course of this Consensus lated to this Consensus Statement. À.R. declares personal teriflunomide for relapsing multiple sclerosis. N. Engl. Statement. C.E. declares personal fees from Bayer, Biogen, fees from Bayer, Biogen, Icometrix, Merck Serono, Novartis, J. Med. 365, 1293–1303 (2011). Celgene, Genzyme, Merck, Novartis, Roche, Sanofi-​Aventis, Sanofi-​Genzyme, Roche and Teva, and other competing inter- 163. Wolinsky, J. S. et al. Magnetic resonance imaging Shire and Teva, all unrelated to this Consensus Statement. ests related to Bayer, Novartis, OLEA Medical, Sanofi-​ outcomes from a phase III trial of teriflunomide. J.W. declares employment with Medical Image Analysis Genzyme and Synthetic MRI, all unrelated to this Consensus Mult. Scler. 19, 1310–1319 (2013). Center, scientific advisory board membership for Biogen, Statement. M.B. declares no competing interests. 164. Radue, E. W. et al. Teriflunomide slows BVL in relapsing Genzyme, Novartis, TEVA and Roche, personal fees from MS: a reanalysis of the TEMSO MRI data set using Bayer, Novartis and Roche, and grants from the European Peer review information SIENA. Neurol. Neuroimmunol. Neuroinflamm. 4, Union (Horizon2020), German Federal Ministries of Nature Reviews Neurology thanks N. Bergsland, D. e390 (2017). Education and Research (BMBF), and Economic Affairs and Ontaneda and A. Traboulsee for their contribution to the peer 165. Miller, D. H. et al. Effects of delayed-​release dimethyl Energy (BMWI), all unrelated to this Consensus Statement. review of this work. fumarate on MRI measures in the phase 3 CONFIRM M.P.S. declares personal fees from Actelion, Biogen, study. Neurology 84, 1145–1152 (2015). Genzyme, Merck, Novartis, Roche, Serono, Synthon and Teva, Open Access This article is licensed under a 166. Radue, E. W. et al. Natalizumab plus interferon all unrelated to this Consensus Statement. F.B. acts as a con- Creative Commons Attribution 4.0 Inter­ beta-1a reduces lesion formation in relapsing multiple sultant for Apitope, Bayer-​Schering, Biogen-​Idec, GeNeuro, national License, which permits use, sharing, sclerosis. J. Neurol. Sci. 292, 28–35 (2010). Sanofi-​Genzyme, Ixico, Janssen Research, Merck-​Serono, adaptation, distribution and reproduction in any medium or 167. Kapoor, R. et al. Effect of natalizumab on disease Novartis, Roche and TEVA. He has received grants, or grants format, as long as you give appropriate credit to the original progression in secondary progressive multiple are pending, from the Amyloid Imaging to Prevent Alzheimer’s author(s) and the source, provide a link to the Creative sclerosis (ASCEND): a phase 3, randomised, Disease (AMYPAD) initiative, the Biomedical Research Centre Commons licence, and indicate if changes were made. The double-​blind, placebo-​controlled trial with an at University College London Hospitals, the Dutch MS Society, images or other third party material in this article are open-​label extension. Lancet Neurol. 17, 405–415 ECTRIMS–MAGNIMS, EU-​H 2020, the Dutch Research included in the article’s Creative Commons licence, unless (2018). Council (NWO), the UK MS Society, and the National Institute indicated otherwise in a credit line to the material. If material 168. Radue, E. W. et al. Impact of fingolimod therapy on for Health Research, University College London. He has is not included in the article’s Creative Commons licence and magnetic resonance imaging outcomes in patients received payments for the development of educational pres- your intended use is not permitted by statutory regulation or with multiple sclerosis. Arch. Neurol. 69, 1259–1269 entations from Ixico and to his institution from Biogen-​Idec. exceeds the permitted use, you will need to obtain permission (2012). He is on the editorial board of Radiology, Brain, European directly from the copyright holder. To view a copy of this 169. Calabresi, P. A. et al. Safety and efficacy of fingolimod Radiology, Multiple Sclerosis Journal and Neurology. T.A.Y. licence, visit http://creativecommons.org/licenses/by/4.0/. in patients with relapsing-​remitting multiple sclerosis declares research support from Biogen, GlaxoSmithKline, (FREEDOMS II): a double-​blind, randomised, placebo-​ Novartis and Schering, and honoraria from Bayer Schering, Related links controlled, phase 3 trial. Lancet Neurol. 13, 545–556 Biogen, Hikma and Novartis. N.D.S. declares personal magnetic Resonance in Imaging in mS (maGnImS) study (2014). fees from Biogen-​Idec, Celgene, Sanofi-​Genzyme, Merck group: www.magnims.eu 170. Hauser, S. L. et al. Ocrelizumab versus interferon Serono, Novartis, Roche and Teva, grants from Merck Serono, beta-1a in relapsing multiple sclerosis. N. Engl. J. Novartis and the Italian MS Society, and nonfinancial support Med. 376, 221–234 (2017). from Biogen-​Idec, Sanofi-​Genzyme, Merck Serono, Novartis, © The Author(s) 2020 182 | March 2020 | volume 16 www.nature.com/nrneurol