RESEARCH ARTICLE OPEN ACCESS Association of NEDA-4 With No Long-term Disability Progression in Multiple Sclerosis and Comparison With NEDA-3 A Systematic Review and Meta-analysis Dalia Rotstein, MD, MPH, Jacqueline M. Solomon, MD, Maria Pia Sormani, PhD, Xavier Montalban, MD, PhD, Correspondence Dr. Rotstein Xiang Y. Ye, MSc, Dina Dababneh, MD, Alexandra Muccilli, MD, Georges Saab, MD, and Prakesh Shah, MD, MSc dalia.rotstein@unityhealth.to Neurol Neuroimmunol Neuroinflamm 2022;9:e200032. doi:10.1212/NXI.0000000000200032 Abstract Background and Objectives No evidence of disease activity (NEDA)-4 has been suggested as a treatment target for disease- modifying therapy (DMT) in relapsing-remitting multiple sclerosis (RRMS). However, the ability of NEDA-4 to discriminate long-term outcomes in MS and how its performance compares with NEDA-3 remain uncertain. We conducted a systematic review and meta-analysis to evaluate (1) the association between NEDA-4 and no long-term disability progression in MS and (2) the comparative performance of NEDA-3 and NEDA-4 in predicting no long-term disability progression. Methods English-language abstracts and manuscripts were systematically searched in MEDLINE, Embase, and the Cochrane databases from January 2006 to November 2021 and reviewed independently by 2 investigators. We selected studies that assessed NEDA-4 at 1 or 2 years after DMT start and had at least 4 years of follow-up for determination of no confirmed disability progression. We conducted a meta-analysis using random-effects model to determine the pooled odds ratio (OR) for no disability progression with NEDA-4 vs EDA-4. For the com- parative analysis, we selected studies that evaluated both NEDA-3 and NEDA-4 with at least 4 years of follow-up and examined the difference in the association of NEDA-3 and NEDA-4 with no disability progression. Results Five studies of 1,000 patients (3 interferon beta and 2 fingolimod) met inclusion criteria for both objectives. The median duration of follow-up was 6 years (interquartile range: 4–6 years). The prevalence of NEDA-4 ranged from 4.2% to 13.9% on interferon beta therapy and 24.9% to 25.1% on fingolimod therapy. The pooled OR for no long-term confirmed disability pro- gression with NEDA-4 vs EDA-4 was 2.14 (95% confidence interval: 1.36–3.37; I2 = 0). We did not observe any significant difference between NEDA-4 and NEDA-3 in the comparative analyses. Discussion In patients with RRMS, NEDA-4 at 1–2 years was associated with 2 times higher odds of no long-term disability progression, at 6 years compared with EDA-4, but offered no advantage over NEDA-3. From the Department of Medicine, (D.R., A.M., G.S.), University of Toronto, Ontario, Canada; St. Michael’s Hospital (D.R., A.M., G.S.), Toronto, Ontario, Canada; Department of Medicine, (J.M.S.), McMaster University, Hamilton, Ontario, Canada; Department of Health Sciences (M.P.S.), Section of Biostatistics, University of Genova, Italy; IRCCS Ospedale Policlinico San Martino (M.P.S.), Genova, Italy; Department of Neurology and Cemcat (X.M.), Hospital Universitari Vall d’Hebron, Universitat Autonoma de Barcelona, Barcelona; Department of Pediatrics (X.Y.Y., P.S.), Mount Sinai Hospital, Toronto, Canada; Columbia University Irving Medical Center (D.D.), Department of Neurology, New York City; York Presbyterian Hospital (NYP) (D.D.), New York City; and Institute of Health (P.S.), Policy, Management and Evaluation, University of Toronto, Canada. Go to Neurology.org/NN for full disclosures. Funding information is provided at the end of the article. This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND), which permits downloading and sharing the work provided it is properly cited. The work cannot be changed in any way or used commercially without permission from the journal. Copyright © 2022 The Author(s). Published by Wolters Kluwer Health, Inc. on behalf of the American Academy of Neurology. 1 Glossary BVL = brain volume loss; CI = confidence interval; DMT = disease-modifying therapie; EDA = evidence of disease activity; MS = multiple sclerosis; NEDA = no evidence of disease activity; NfL = neurofilament light chain; OR = odds ratio; PIRA = progression independent of relapse activity; RCT = randomized controlled trial; RRMS = relapsing-remitting MS; SPMS = secondary progressive MS. The use of composite outcome measures has become widespread included cohort and clinical trial extension studies. For the in recent years as early targets for evaluating response to multiple primary objective, studies were required to report the asso- sclerosis (MS) disease-modifying therapies (DMTs). No evi- ciation of NEDA-4 vs evidence of disease activity (EDA)-4 at dence of disease activity (NEDA) is one such measure. NEDA-3, the last follow-up, or for these data to be derivable. If data the most commonly used iteration of NEDA, is defined as no were not available or could not be derived, they were evidence of (1) relapse, (2) disability progression, or (3) new or requested from the authors through at least 2 email com- enlarging T2 and/or T1 gadolinium-enhancing lesion measured munications. The cohort with a longer follow-up was selected over the first 1–2 years on therapy.1,2 However, NEDA-3 has where there were overlapping cohorts. For the secondary been criticized for being overly focused on inflammatory aspects objective, studies were required to report data on both of MS.3 Early4 and/or longitudinal5 brain volume loss (BVL) NEDA-4 and NEDA-3, or for these to be derivable, and to have been shown to independently predict long-term disability have a minimum of 4 years of follow-up with data on pro- progression in relapsing-remitting MS (RRMS). Subsequently, gression reported after a full follow-up. We excluded case NEDA-4 was introduced to include no significant annual BVL in reports, editorials, and review articles. addition to the NEDA-3 criteria.3 A threshold of BVL of >0.4% per year was defined as significant loss relative to the rate of BVL Population in healthy volunteers (0.1%–0.3%) vs untreated patients with We selected studies that included patients with RRMS. Pa- MS.3 A few studies6-10 have reported an association of NEDA-4 tients with progressive MS and those with clinically isolated with no long-term disability progression. The results have been syndrome were allowed to comprise up to 20% of the cohort; variable and have not been summarized. Moreover, it is unclear however, all the included studies enrolled only patients with how NEDA-4 and NEDA-3 compare in their association with no RRMS. disability progression in people with RRMS over a long-term follow-up. This understanding is essential to inform use of NEDA Literature Search as an outcome for treatment response in therapy trials and clinical Medline, Embase, the Cochrane Central Register of Con- practice, as suggested by American Academy of Neurology and trolled Trials, and the Cochrane Database of Systematic Re- European Academy of Neurology guideline committees.11,12 views were searched for English language studies published from January 1, 2006, until November 19, 2021. For MS, we The primary objective of this systematic review and meta-analysis searched using the MeSH “demyelinating disease” and the was to summarize the evidence on the association between Emtree subject heading “multiple sclerosis” and the key words NEDA-4, as measured in the first 1–2 years after DMT initiation, “multiple sclerosis.” Because NEDA is a codified concept, we and no long-term disability progression as assessed at a minimum used natural language: “disease-free status,” “disease-free ac- of 4 years from baseline. Our secondary objective was to compare tivity,” “no evidence of disease activity” or “NEDA” in the title. NEDA-4 and NEDA-3 for their associations with no long-term We also used reference lists of identified articles and sys- disability progression, where both were tested in the same cohort. tematic review studies as additional sources for primary studies. At least 2 investigators (D.R. and J.S., A.M., G.S., or D.D.) independently screened titles and abstracts for initial Methods eligibility and evaluated full-text articles in duplicate for final inclusion. Disagreements were resolved by discussion, which Standard Protocol Approvals, Registrations, included a third author (P.S.). and Patient Consents This study followed the Preferred Reporting Items for Systematic Exposure (Test) Reviews and Meta-analyses methodology.13 The systematic re- NEDA-4 and NEDA-3 assessment was required at 1 or 2 years view protocol was registered with the International Prospective from baseline after DMT initiation. For NEDA-4, the fol- Register of Systematic Reviews (CRD42020189316). This study lowing 4 criteria were mandatory: (1) no clinical relapse, (2) was exempt from ethics board review approval as a systematic no confirmed disability progression, (3) no new or enlarging review and meta-analysis. T2 lesion on brain MRI, and (4) no BVL ≥0.4% per year. For NEDA-3, only the first 3 former criteria were mandatory. Eligibility Criteria Often, absence of any T1 gadolinium-enhancing (gad+) le- Types of Studies sion was included in NEDA-3 and NEDA-4, but this was not We included studies of patients with a minimum of 4 years of required because some studies did not involve gad-enhanced follow-up after DMT initiation. Eligible study designs MRIs. 2 Neurology: Neuroimmunology & Neuroinflammation | Volume 9, Number 6 | November 2022 Neurology.org/NN Outcome the OR was not reported in the published manuscript, then it was The outcome was no confirmed disability progression over a calculated from the 2 × 2 table. Heterogeneity was evaluated using follow-up of a minimum of 4 years. Most often, this was iden- the I2 statistic. We also evaluated the pooled ORs separately by tified using 6-month confirmed disability progression (6m- DMT class. We performed a sensitivity analysis excluding 1 study6 CDP) as measured by EDSS change of 1.0. Some studies stip- due to differences in definition of the outcome and duration of ulated a change of at least 1.5 if baseline EDSS was zero and/or follow-up, which was much longer than the other included studies. 0.5 if baseline EDSS was >5.0. Other permitted definitions in- cluded not reaching EDSS milestone of 6.0 or secondary pro- For the secondary objective, we included studies that reported gressive MS (SPMS) or death. For the randomized controlled both NEDA-4 and NEDA-3. To assess the difference between trial (RCT) extension studies, where possible, we evaluated only the degree of association of NEDA-4 with the outcome and the treatment arms of approved doses as indicated. NEDA-3 with the outcome, we estimated the difference be- tween the log OR of outcome (NEDA-4 vs EDA-4) and log Data Extraction and Covariates OR of outcome (NEDA-3 vs EDA-3). Then, we converted Two authors (D.R. and J.S.) used a predesigned data collec- back to the ratio of ORs to assess whether ORs were signifi- tion form to independently extract study details that were cantly different between NEDA-4 vs EDA-4 and NEDA-3 vs then checked together. Baseline cohort characteristics were EDA-3. Data management and all statistical analyses were collected based on published description of the intention-to- performed using SAS 9.4 (SAS Institute, Inc., Cary, NC) and treat clinical trial population or observational cohort. “meta” “mada” R package in R 4.0.0 (r-project.org). Risk of Bias Assessment Data Availability We used the Quality in Prognosis Studies tool to evaluate the risk Data not published within this article will be made available on of bias.14 Six domains are considered in the risk of bias assess- reasonable request to the corresponding author from any qual- ment (eTable 1, links.lww.com/NXI/A745). Domains were each ified investigator. rated as a low, medium, or high risk of bias. Studies were classified as overall low risk of bias if all domains were rated as low risk, and high risk of bias if one or more domains were rated as high risk. Results Analysis Five studies including 1,000 participants met inclusion criteria For the primary objective, we used a random-effects model to for the primary objective (Figure 1). The same 5 studies were synthesize the pooled odds ratio (OR) with 95% confidence in- selected for analysis of the secondary objective because all terval (CI) for no disability progression with NEDA-4 vs EDA-4. If reported data for NEDA-3 in addition to NEDA-4. The Figure 1 PRISMA Flow Diagram for Study Selection PRISMA flow diagram showing study selection and reasons for omission of excluded studies. PRISMA = Preferred Reporting Items for Systematic Reviews and Meta-Analyses. Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 9, Number 6 | November 2022 3 Table 1 Baseline Characteristics of Included Studies Study design Baseline Overall risk of (name of Number mean bias rating Source, study or % on enrolled/ Duration Baseline Baseline disease ARR (reason, if location (years cohort, if Therapies given number of follow- mean Female median duration before medium or of recruitment) given) studied therapy followed up (y) age (y) (%) EDSS (y) study higher risk) Cohen et al.,9,10 RCT Extension FTYa 100 429/232 6 36.7 65.4 2.0 7.5 1.5 High International (TRANSFORMS), (attrition) (2006–2007) multicenter Goodin et al.,6 RCT Extension IFNB-1b 68.6 376/215 16 35.2 69.0 2.5 7.7 1.7 High 2019 USA, (IFNB Multiple Placebo 31.4 (attrition) Canada (1988) Sclerosis Study Group), multicenter Kappos RCT Extension FTYa 100 425/341 6 36.6 69.6 2.0 8.0 1.5 Low et al.,10,15 2015 (FREEDOMS), International multicenter (2006–2007) Perez-Miralles Retrospective IFN 100 124/101 4 34 80.2 1.5 3.0 1.5 Low et al.,8 Spain single center (2001 onwards) cohort (Barcelona cohort) Srpova et al.,7 Prospective IFN 100 220/166 4 29 66.9 1.5 0.2 N/A Medium Czech Republic multicenter (attrition) (2005–2009) cohort (SET) ARR = annualized relapse rate; EDSS = expanded disability status scale; FTY = fingolimod; IFN = interferon beta; IFNB-1b = interferon beta-1b; RCT Ext = Randomized Controlled Trial Extension. a 0.5 mg treatment arm only. median duration of follow-up was 6.0 years (range: 4–16 subgroup analysis stratified by DMT class, NEDA-4 was as- years) (Table 1). Three of the included studies investigated sociated with no disability progression for both interferon participants on interferon beta therapy or placebo.6-8 Of these, beta (OR: 1.81; 95% CI: 0.72–4.56; I2 = 0) and fingolimod 2 were observational studies,7,8 and one was an RCT exten- (OR: 2.20; 95% CI: 0.88–5.53; I2 = 68%) therapies, but these sion study.6 The other 2 included studies were the extension associations did not meet statistical significance. cohorts from the fingolimod clinical trials.9,10,15 Two studies had a low risk of bias,8,15 one had a medium risk of bias,7 and 2 In the sensitivity analysis, excluding the beta interferon-1b ex- had a high risk of bias,6,9 with a higher risk of bias primarily tension study, which differed in its definition of the outcome due to attrition. All but one8 were multicenter studies. and had a much longer follow-up than the other included studies, the pooled OR for no disability progression with Baseline cohort characteristics varied across studies including NEDA-4 vs EDA-4 was 2.12 (95% CI: 1.12–3.40; I2 = 21%). mean age (29–36.7 years), percentage of female participants (66%–80%), median or mean EDSS (1.5–2.5), disease duration Secondary Outcome: No Disability Progression (0.2–8 years), and annualized relapse rate (1.5–1.7) (Table 1). With NEDA-4 vs EDA-4 and NEDA-3 vs EDA-3 Four studies evaluated NEDA at 2 years after initiating therapy The ratio of OR for NEDA-4 vs EDA-4 and OR for NEDA-3 and 1 study8 evaluated NEDA at 1 year (Table 2). Four studies vs EDA-3 (OR: 0.71, 95% CI: 0.39–1.30) was not statistically used no 6m-CDP as their outcome at the end of follow-up. One significantly different in the meta-analysis and when evaluat- study6 used conversion to SPMS or reaching an EDSS ≥6.0 or ing interferon beta (OR: 0.88, 95% CI: 0.29–2.71) and fin- death to evaluate disability progression at the end of follow-up. golimod (OR: 0.65, 95% CI: 0.32–1.33) therapy studies separately (Figure 3). NEDA-4 was met by 4.2%–13.9% on interferon beta therapy and 24.9%–25.1% on fingolimod therapy (Table 2). NEDA-3 was met by 8.8%–34.7% of those on interferon beta therapy Discussion and 34.1%–37.2% of those on fingolimod therapy. In this systematic review and meta-analysis, we identified an Primary Outcome: No Disability Progression association between NEDA-4 and no long-term disability With NEDA-4 progression in RRMS. NEDA-4 at a median of 6 years of The pooled OR for no disability progression with NEDA-4 vs follow-up was associated with a 2 times higher odds of no EDA-4 was 2.14 (95% CI: 1.36–3.37; I2 = 0; Figure 2). For the disability progression than EDA-4. As expected, with the 4 Neurology: Neuroimmunology & Neuroinflammation | Volume 9, Number 6 | November 2022 Neurology.org/NN Table 2 Definitions and Data Concerning NEDA-4 and NEDA-3 From Included Studies Definition of Number with Number with Number with Number with long-term NEDA and no NEDA and EDA and no EDA and Definition disability NEDA-4 Number (%) disability disability disability disability Source of NEDA progression or NEDA-3 with NEDA progression progression progression progression Cohen et al.,9,10 2 y, gad+ lesions 6m-CDPa NEDA-4 58 (25.1%) 48 10 135 38 2016 included NEDA-3 79 (34.1%) 68 11 116 37 Goodin et al.,6 2y SPMS or EDSS NEDA-4 7 (4.2%) 5 2 83 78 2019 ≥6.0 or death NEDA-3 19 (8.8%) 12 7 91 105 Kappos et al.,10,15 2 y, gad+ lesions 6m-CDPa NEDA-4 83 (24.9%) 73 10 170 81 2015 included NEDA-3 127 (37.2%) 115 12 135 79 a c Perez-Miralles 1 y, gad+ lesions 6m-CDP NEDA-4 11 (10.9%) N/A N/A N/A N/A et al.,8 2020 included NEDA-3 35 (34.7%) N/Ac N/A N/A N/A Srpova et al.,7 2021 2 y, gad+ lesions 6m-CDPb NEDA-4 23 (13.9%) 21 2 118 25 included NEDA-3 39 (23.5%) 34 5 105 22 CDP = confirmed disability progression; EDA = evidence of disease activity; EDSS = expanded disability status scale; NEDA = no evidence of disease activity; SPMS = secondary progressive multiple sclerosis. a 6m-CDP = Change in EDSS of at least 1.0 if baseline EDSS 0–5.0, 0.5 if baseline EDSS >5.0. b Change in EDSS of at least 1.0 if baseline EDSS 1.0–5.0, 1.5 if baseline EDSS 0, and 0.5 if baseline EDSS >5.0. c Odds ratios were reported in the paper but not numbers for NEDA/EDA vs disability progression. addition of a brain atrophy measure on top of the criteria re- which can accurately predict long-term prognosis in RRMS. quired for NEDA-3, a lower proportion in each study met The historic tradition of using relapse rate as a primary out- NEDA-4 vs NEDA-3. Comparing NEDA-4 with NEDA-3, there come for DMT efficacy is arguably of less relevance in the was no significant difference between these 2 composite out- modern era of early MS diagnosis and treatment when an- come measures in predicting disability progression, although nualized relapse rates in clinical trials are typically less than power was limited by the small number of studies (n = 5) and 0.417 and as low as 0.10.18 Composite outcome measures are relatively small number of participants (n = 1,000) across studies. more sensitive to different aspects of disease activity. They also reflect our ability to aim for a higher standard of disease In a larger systemic review and meta-analysis of 27 studies that control with second-generation, high-efficacy DMT. NEDA- we previously conducted investigating NEDA-3, we found 3, however, has been criticized for overly emphasizing focal pooled ORs of 2.32 (95% CI: 1.58–3.42; I2 = 73%) on low- inflammatory activity, with the CDP criterion being the sole efficacy therapy and 3.19 (95% CI: 1.86–5.47; I2 = 86%) on proxy for neurodegenerative aspects of MS.19 In some real- high-efficacy therapy for the association of NEDA-3 with no world cohorts, a considerable share of those with NEDA-3 at long-term disability progression.16 In this review, we per- 2 years demonstrated disability progression at a long-term formed subgroup analyses by DMT type (interferon beta and follow-up.6,20 For example, in the extension study of the fingolimod), with pooled ORs in similar ranges, although the randomized trial of interferon beta-1b, which was included as results did not meet statistical significance, which was likely part of this meta-analysis, 37% of those with NEDA-3 at 2 due to power limitations. The ORs for the association be- years had negative disability outcomes at 16 years of follow- tween NEDA-4 and no long-term disability progression were up, although numbers were small (19 with NEDA-3). In an greater than 1 in each of the included studies, but results Italian cohort study of patients with RRMS who were started reached statistical significance in only 1 study. Where rela- on interferon beta or glatiramer acetate, which did not qualify tively low proportions of individuals achieve a composite for this review because of lack of NEDA-4 data, 26% (58/224) outcome measure such as NEDA-4, larger studies and reviews of those with NEDA-3 over 2 years had subsequent disability such as this may be necessary to better understand predictive accrual by a median of 12 years of follow-up.20 Notably, both capabilities. these studies involved a follow-up of more than 10 years. In the subgroup of patients who demonstrated long-term dis- There is a great unmet need for evidence-based composite ability progression independent of relapse activity (PIRA), outcome measures assessed in the first years on therapy, MRI activity was also uncommon (11%), suggesting the need Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 9, Number 6 | November 2022 5 Figure 2 Association of NEDA-4 vs EDA-4 With No Long-term Disability Progression ORs were assessed for each study for the association of NEDA-4 vs EDA-4 with no long-term disability progression. NEDA-4 was defined by (1) no clinical relapse, (2) no confirmed disability progression, (3) no new or enlarging T2 lesion on brain MRI, and (4) no brain volume loss ≥0.4% per year. Pooled ORs were evaluated for each therapy group (interferon beta and fingolimod) and across all studies. EDA = evidence of disease activity; NEDA = no evidence of disease activity; OR = odds ratio. for other early biomarkers to capture this group of RRMS who demonstrate progression long-term, which is often PIRA, patients with poorer prognosis.20 and at this point, it is unclear whether a switch in therapy could ameliorate long-term outcomes in these patients. NEDA-4 was introduced in 2016 based on post hoc analyses of the fingolimod treatment trials.3 The promise of NEDA-4 Ongoing research is investigating other candidate measures that was that inclusion of BVL, a marker of neurodegeneration, could be used in combination with NEDA or in an evidence- might better predict the long-term disability trajectory on a based scoring system to evaluate optimal responders over the given therapy. BVL is known to occur in the earliest stages of first years on DMT. These include serum or CSF biomarkers MS,21 and gray matter atrophy has been associated with such as neurofilament light chain (NfL) and glial fibrillary acid clinical worsening over both the short-term22 and long- protein, cognitive measures, MRI assessment of smouldering or term.4,5 Challenges of incorporating BVL into an early com- cortical lesions, and optical coherence tomography measures posite outcome measure include limited availability of brain such as retinal nerve fiber layer or ganglion cell/inner plexiform volume assessment at many clinical centres; limited re- layer thinning. One study found that a composite measure of producibility of exact BVL measurements across centres and NfL levels over 2 years, clinical and MRI measures, led to greater methods; vulnerability of brain volume changes to other prognostic value (as measured by area under the receiver oper- factors, including hydration status and use of corticosteroids; ating curve) for long-term outcomes at 7 years compared with variable rates of BVL depending on age at MS onset and MS models with NfL, clinical and MRI predictors alone or in any disease duration; and a pseudoatrophy effect in the first year other combination.26 Another shortcoming of current defini- on certain DMT, such as natalizumab.23-25 The findings in this tions of NEDA is that many of those with EDA do well over meta-analysis suggest that NEDA-4 does not offer superior long-term follow-up, and therefore, EDA alone seems in- prognostic value compared with NEDA-3 and thus may not sufficient to justify changes in therapy.16 A recent study reported warrant the extra effort and cost required for its routine that low serum NfL could predict those with EDA-3 who would evaluation in clinical practice, although these findings need to be free of disease activity in the subsequent year.27 It could also be confirmed in other cohorts with extended longitudinal predict those with NEDA-3 who would have disease activity follow-up. In the meantime, NEDA-4 may be most relevant as and/or brain atrophy in the following year, although high NfL an outcome measure in the clinical trial context, in which BVL levels were rare (9.4%) in NEDA-3.27 A key consideration in measurement is more readily supported. Overall, NEDA, applying NfL or any other biomarker to a predictive algorithm whether NEDA-3 or NEDA-4, can be taken as an indication with NEDA should be its ability to flag the subgroup of people that a patient is a good responder to therapy. However, it is with RRMS who have indolent progression over the long-term important to note that there are some patients with NEDA but are stable for NEDA-3 in the first years on therapy. 6 Neurology: Neuroimmunology & Neuroinflammation | Volume 9, Number 6 | November 2022 Neurology.org/NN Figure 3 Ratio of ORs for No Disability Progression With NEDA-4 vs EDA-4 Compared With NEDA-3 vs EDA-3 The ratio of ORs was evaluated for each study for the association of no long-term progression with NEDA-4 vs EDA-4 compared with NEDA-3 vs EDA-3. NEDA-3 was defined by (1) no clinical relapse, (2) no confirmed disability progression, and (3) no new or enlarging T2 lesion on brain MRI. For NEDA-4, no brain volume loss ≥0.4% per year was required in addition to the former criteria for NEDA-3. Pooled ratios were evaluated for each therapy group (interferon beta and fingolimod) and across all studies. A ratio of greater than 1 would imply a stronger association of NEDA-4 vs EDA-4 with no long-term disability progression compared with NEDA-3 vs EDA-3. EDA = evidence of disease activity; NEDA = no evidence of disease activity; OR = odds ratio. Limitations of this review include the small number of studies In this systematic review and meta-analysis, NEDA-4, com- assessing NEDA-4 (n = 5). As mentioned earlier, NEDA-4 is not pared with EDA-4, was associated with twice the odds of no assessed routinely in observational studies because of the re- disability progression after a median of 6 years from DMT quirement of BVL measurement, and the minimum follow-up of initiation in RRMS. We did not observe any difference be- 4 years reduced the number of qualifying studies, although tween NEDA-4 and NEDA-3 in their association with no necessary for the determination of long-term progression. long-term disability progression. However, due to the small Follow-up was 6 years or less in 4 of the 5 included studies, but number of studies and limited duration of follow-up, risk of an extended longitudinal follow-up of 10 years or more may be bias, and heterogeneity in definitions, further studies are necessary for accurate long-term prognostication in RRMS.28,29 necessary. In addition, other early composite measures in- Thus, further studies of NEDA-4, NEDA-3, and other iterations corporating novel biomarkers may be able to improve on the of NEDA with an extended follow-up would be helpful. Study prognostic value of NEDA-4 and NEDA-3 in early RRMS and quality varied, with a higher risk of bias, notably due to attrition warrant further investigation. over time. In addition, there may be selection bias in clinical trial extension studies with preferential retainment of participants Acknowledgment with favorable outcomes. Some heterogeneity was observed in The authors would like to acknowledge David Lightfoot who the baseline cohort characteristics, although mean age, pro- helped to perform the systematic literature search. The portion of female participants, and baseline median EDSS were authors would also like to acknowledge the investigators from quite similar across studies. Similarly, some heterogeneity was qualifying studies who gave generously of their time to observed in definitions of NEDA-4, NEDA-3, and the outcome provide data necessary for this work: Daniela Piani Meier measure of no disability progression. One study6 excluded gad+ (Novartis), Rolf Meinert (Novartis), Francisco Carlos Pérez lesions from the MRI component of NEDA and had a different Miralles, Chaitali Pisal (Novartis), Jordi Rio, Shannon Ritter definition of the outcome, but a sensitivity analysis removing this (Novartis), and Barbora Srpova. study did not change the results. While 6m-CDP was the most commonly used outcome measure across studies, sustained Study Funding disability progression or attainment of certain disability mile- The authors report no targeted funding. stones (e.g., EDSS 4.0 or 6.0) may be more meaningful indica- tors of long-term disability evolution.19 We could not evaluate Disclosure characteristics of those with NEDA-4 vs EDA-4 because these D.L. Rotstein has received research support from the MS data were not available for most of the studies. Society of Canada, Consortium of MS Centers (CMSC), and Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 9, Number 6 | November 2022 7 Roche Canada. She has received speaker or consultant fees from Alexion, Biogen, EMD Serono, Novartis, Roche, and Appendix (continued) Sanofi Aventis. J.M. Solomon has nothing to disclose. Maria Name Location Contribution Pia Sormani received consulting fees from Biogen, Merck, Roche, Novartis, Sanofi, Immunic, and GeNeuro. X. Mon- Georges Department of Medicine, Drafting/revision of the Saab, MD University of Toronto, article for content, including talban has received speaking honoraria and travel expenses for Ontario, Canada; St. medical writing for content; scientific meetings and has been a steering committee mem- Michael’s Hospital, Toronto, major role in the acquisition Ontario, Canada of data ber of clinical trials or participated in advisory boards of clinical trials in the past 3 years with Biogen Idec, Merck Prakesh Department of Pediatrics, Drafting/revision of the Shah, MD, Mount Sinai Hospital, article for content, including Serono, Genentech, Genzyme, Novartis, Sanofi-Aventis, Teva MSc Toronto, Canada; Institute of medical writing for content; Pharmaceuticals, Roche, Celgene, Actelion, Mylan and BMS. Health, Policy, Management study concept or design; and and Evaluation, University of analysis or interpretation of X. Ye and D. Dababneh have nothing to disclose. A. Muccilli Toronto, Toronto, Canada data has received speaker or consultant fees from Biogen, EMD Serono, and Novartis. P. Shah has nothing to disclose. Go to Neurology.org/NN for full disclosure. References 1. Rotstein DL, Healy BC, Malik MT, Chitnis T, Weiner HL. Evaluation of no evidence of disease activity in a 7-year longitudinal multiple sclerosis cohort. JAMA Neurol. Publication History 2015;72(2):152-158. Received by Neurology: Neuroimmunology & Neuroinflammation 2. Giovannoni G, Turner B, Gnanapavan S, Offiah C, Schmierer K, Marta M. Is it time to March 22, 2022. Accepted in final form August 2, 2022. Submitted target no evident disease activity (NEDA) in multiple sclerosis? Mult Scler Relat Disord. 2015;4:329-333. and externally peer reviewed. The handling editor was Friedemann Paul, 3. Kappos L, De Stefano N, Freedman MS, et al. Inclusion of brain volume loss in a MD. revised measure of ’no evidence of disease activity’ (NEDA-4) in relapsing-remitting multiple sclerosis. Mult Scler. 2016;22(10):1297-1305. 4. Radue EW, Barkhof F, Kappos L, et al. Correlation between brain volume loss and clinical and MRI outcomes in multiple sclerosis. Neurology. 2015;84(8):784-793. 5. University of California San Francisco MS-EPIC Team, Cree BAC, Hollenbach JA, Kirkish G, et al. Silent progression in disease activity-free relapsing multiple sclerosis. Ann Neurol. 2019;85(5):653-666. Appendix Authors 6. Goodin DS, Reder AT, Traboulsee AL, et al, IFNB Multiple Sclerosis Study Group and the 16- and 21-Year LTF Investigators. Predictive validity of NEDA in the 16- and 21-year Name Location Contribution follow-up from the pivotal trial of interferon beta-1b. Mult Scler. 2019;25(6):837-847. 7. Srpova B, Uher T, Hrnciarova T, et al. Serum neurofilament light chain reflects Dalia Department of Medicine, Drafting/revision of the inflammation-driven neurodegeneration and predicts delayed brain volume loss in Rotstein, University of Toronto, article for content, including early stage of multiple sclerosis. Mult Scler. 2021;27(1):52-60. MD, MPH Ontario, Canada; St. medical writing for content; 8. Perez-Miralles FC, Rio J, Pareto D, et al. Adding brain volume measures into response Michael’s Hospital, Toronto, major role in the acquisition criteria in multiple sclerosis: the Rio-4 score. Neuroradiology. 2020;63(7):1031-41. Ontario, Canada of data; study concept or 9. Cohen JA, Khatri B, Barkhof F, et al. Long-term (up to 4.5 years) treatment with design; and analysis or fingolimod in multiple sclerosis: results from the extension of the randomised interpretation of data TRANSFORMS study. J Neurol Neurosurg Psychiatry. 2016;87(5):468-475. 10. Cohen JA, Tenenbaum N, Bhatt A, Zhang Y, Kappos L. Extended treatment with Jacqueline Department of Medicine, Drafting/revision of the article fingolimod for relapsing multiple sclerosis: the 14-year LONGTERMS study results. M. McMaster University, for content, including medical Ther Adv Neurol Disord. 2019;12:1756286419878324. Solomon, Hamilton, Ontario, Canada writing for content; major role 11. Corboy JR, Weinshenker BG, Wingerchuk DM. Comment on 2018 American MD in the acquisition of data Academy of Neurology guidelines on disease-modifying therapies in MS. Neurology. 2018;90(24):1106-1112. Maria Pia Department of Health Drafting/revision of the 12. Montalban X, Gold R, Thompson AJ, et al. ECTRIMS/EAN guideline on the phar- Sormani, Sciences, Section of article for content, including macological treatment of people with multiple sclerosis. Mult Scler. 2018;24(2):96-120. PhD Biostatistics, University of medical writing for content; 13. McInnes MDF, Moher D, Thombs BD, et al. Preferred reporting Items for a sys- Genova, Italy; IRCCS study concept or design; and tematic review and meta-analysis of diagnostic test accuracy studies: the PRISMA- Ospedale Policlinico San analysis or interpretation of DTA statement. JAMA. 2018;319(4):388-396. Martino, Genova, Italy data 14. Hayden JA, van der Windt DA, Cartwright JL, Cote P, Bombardier C. Assessing bias in studies of prognostic factors. Ann Intern Med. 2013;158(4):280-286. Xavier Department of Neurology Drafting/revision of the article 15. Kappos L, O’Connor P, Radue EW, et al. Long-term effects of fingolimod in multiple Montalban, and Cemcat, Hospital for content, including medical sclerosis: the randomized FREEDOMS extension trial. Neurology. 2015;84(15):1582-1591. MD, PhD Universitari Vall d’Hebron, writing for content; study 16. Rotstein DL, Solomon JM, Sormani MP, et al. Association of “No Evidence of Disease Universitat Autonoma de concept or design; and analysis Activity” (NEDA) with no long-term disability progression in multiple sclerosis: a Barcelona, Barcelona or interpretation of data systematic review and meta-analysis. Neurology. 2022;99(2):e209-e220. 17. Zhang Y, Salter A, Wallstrom E, Cutter G, Stuve O. Evolution of clinical trials in Xiang Y. Ye, Department of Pediatrics, Drafting/revision of the article multiple sclerosis. Ther Adv Neurol Disord. 2019;12:1756286419826547. MSc Mount Sinai Hospital, for content, including medical 18. Hauser SL, Bar-Or A, Cohen JA, et al, ASCLEPIOS I and ASCLEPIOS II Trial Toronto, Canada writing for content; analysis or Groups. Ofatumumab versus teriflunomide in multiple sclerosis. N Engl J Med. 2020; interpretation of data 383(6):546-557. 19. Gasperini C, Prosperini L, Tintore M, et al, the MAGNIMS Study Group. Unraveling Dina Columbia University Irving Drafting/revision of the treatment response in multiple sclerosis: a clinical and MRI challenge. Neurology. Dababneh, Medical Center, Department article for content, including 2019;92(4):180-192. MD of Neurology, New York City; medical writing for content; 20. Prosperini L, Ruggieri S, Haggiag S, Tortorella C, Pozzilli C, Gasperini C. Prognostic New York Presbyterian major role in the acquisition accuracy of NEDA-3 in long-term outcomes of multiple sclerosis. Neurol Neuro- Hospital (NYP), New York City of data immunol Neuroinflamm. 2021;8(6):e1059. 21. Azevedo CJ, Overton E, Khadka S, et al. Early CNS neurodegeneration in radiolog- Alexandra Department of Medicine, Drafting/revision of the ically isolated syndrome. Neurol Neuroimmunol Neuroinflamm. 2015;2(3):e102. Muccilli, MD University of Toronto, article for content, including 22. Rocca MA, Valsasina P, Meani A, et al, MAGNIMS Study Group. Association of gray Ontario, Canada; St. medical writing for content; matter atrophy patterns with clinical phenotype and progression in multiple sclerosis. Michael’s Hospital, Toronto, major role in the acquisition Neurology. 2021;96(11):e1561-e1573. Ontario, Canada of data 23. Rocca MA, Battaglini M, Benedict RHB, et al. Brain MRI atrophy quantification in MS: from methods to clinical application. Neurology. 2017;88(4):403-413. 8 Neurology: Neuroimmunology & Neuroinflammation | Volume 9, Number 6 | November 2022 Neurology.org/NN 24. Zivadinov R, Reder AT, Filippi M, et al. Mechanisms of action of disease-modifying 27. Uher T, Havrdova EK, Benkert P, et al. Measurement of neurofilaments improves agents and brain volume changes in multiple sclerosis. Neurology. 2008;71(2): stratification of future disease activity in early multiple sclerosis. Mult Scler. 2021; 136-144. 27(13):2001-2013. 25. Andorra M, Nakamura K, Lampert EJ, et al. Assessing biological and methodological 28. Sayao AL, Devonshire V, Tremlett H. Longitudinal follow-up of “benign” multiple aspects of brain volume loss in multiple sclerosis. JAMA Neurol. 2018;75(10):1246-1255. sclerosis at 20 years. Neurology. 2007;68(7):496-500. 26. Haring DA, Kropshofer H, Kappos L, et al. Long-term prognostic value of longitudinal 29. Beiki O, Frumento P, Bottai M, Manouchehrinia A, Hillert J. Changes in the risk of measurements of blood neurofilament levels. Neurol Neuroimmunol Neuroinflamm. reaching multiple sclerosis disability milestones in recent decades: a nationwide 2020;7(5):e856. population-based cohort study in Sweden. JAMA Neurol. 2019;76(6):665-671. Neurology.org/NN Neurology: Neuroimmunology & Neuroinflammation | Volume 9, Number 6 | November 2022 9