JNS-13913; No of Pages 11

Journal of the Neurological Sciences xxx (2015) xxx–xxx

Contents lists available at ScienceDirect

Journal of the Neurological Sciences

journal homepage: www.elsevier.com/locate/jns

Review article

The natural history of brain volume loss among patients with multiple
sclerosis: A systematic literature review and meta-analysis
Timothy Vollmer a, James Signorovitch b,⁎, Lynn Huynh b, Philip Galebach b, Caroline Kelley b,
Allitia DiBernardo c, Rahul Sasane c
a
University of Colorado Denver Department of Neurology, Denver, CO, USA
b
Analysis Group, Inc., Boston, MA, USA
c
Novartis Pharmaceuticals Corporation, East Hanover, NJ, USA

a r t i c l e i n f o a b s t r a c t

Article history: Background: Multiple sclerosis has been associated with progressive brain volume loss.
Received 29 December 2014 Objective: We aimed to systematically summarize reported rates of brain volume loss in multiple sclerosis and
Received in revised form 15 June 2015 explore associations between brain volume loss and markers of disease severity.
Accepted 9 July 2015 Methods: A systematic literature search (2003–2013) was conducted to identify studies with ≥12 months of
Available online xxxx
follow-up, reported brain volume measurement algorithms, and changes in brain volume. Meta-analysis
random-effects models were applied. Associations between brain volume change, changes in lesion volume
Keywords:
Multiple sclerosis
and disease duration were examined in pre-specified meta-regression models.
Systematic literature review Results: We identified 38 studies. For the meta-analysis, 12 studies that reported annualized percentage brain vol-
Meta-analysis ume change (PBVC), specified first-generation disease-modifying treatments (e.g., interferon-beta or glatiramer
Brain volume loss acetate) and used Structural Image Evaluation of Normalized Atrophy algorithm were analyzed. The annualized
Disease progression PBVC ranged from −1.34% to −0.46% per year. The pooled PBVC was −0.69% (95% CI = −0.87% to −0.50%)
Atrophy in study arms receiving first-generation disease-modifying treatments (N = 6 studies) and − 0.71%
Disease-modifying treatment (95% CI = −0.81% to −0.61%) in untreated study arms (N = 6 studies).
Conclusions: In this study, the average multiple sclerosis patient receiving first-generation disease-modifying
treatment or no disease-modifying treatment lost approximately 0.7% of brain volume/year, well above rates as-
sociated with normal aging (0.1%–0.3% of brain volume/year).
© 2015 Elsevier B.V. All rights reserved.

Contents

1. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
2. Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
2.1. Study selection criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
2.2. Search strategy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
2.3. Study selection process and data extraction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
2.4. Risk of bias assessment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
2.5. Statistical analyses . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
2.6. Data management and reporting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
3. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
3.1. Search . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
3.2. Overall study description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
3.3. Quality assessment of the included studies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
3.4. Brain volume measurements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
3.5. Meta-analyses and meta-regressions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
5. Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0

⁎ Corresponding author at: Analysis Group, Inc., 111 Huntington Avenue, Tenth Floor, Boston, MA 02199, USA.
E-mail address: james.signorovitch@analysisgroup.com (J. Signorovitch).

http://dx.doi.org/10.1016/j.jns.2015.07.014
0022-510X/© 2015 Elsevier B.V. All rights reserved.

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014
2 T. Vollmer et al. / Journal of the Neurological Sciences xxx (2015) xxx–xxx

Author contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
Declaration of conflicting interests . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
Appendix A. Supplementary material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0
References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 0

1. Introduction methodology. Each reviewer evaluated the data from the eligible stud-
ies and electronically entered the information into an Excel database de-
Multiple sclerosis (MS) is a chronic, progressive inflammatory auto- veloped specifically for the review with prepared fields. Disagreements
immune disease of the central nervous system that results in neurolog- between reviewers were resolved by consensus or by arbitration
ical dysfunction characterized by myelin destruction and axonal loss [1]. through a third party, referring to the original sources.
It affects over 400,000 Americans and more than 2.1 million people We collected information on the study design, population, compari-
worldwide [2]. sons or treatment groups, sample size, duration of follow-up, brain vol-
Magnetic resonance imaging (MRI) is an important tool for diagnos- ume algorithm, average baseline characteristics of patients, MS type, MS
ing and monitoring MS through the quantification of lesions in the brain disease duration, and reported changes in BVL over the specified time
[3]. The Structural Image Evaluation of Normalized Atrophy (SIENA) al- period (Table 1). For brain volume measures, we extracted information
gorithm is one of the common brain volume algorithms used in longitu- for percent brain volume change (PBVC), brain parenchymal fraction
dinal studies [4]. The rate of brain volume loss (BVL) in MS has been (BPF), white matter fraction (WMF), and grey matter fraction (GMF).
suggested as a potential marker of MS disease progression [3]. In pa- In addition, we recorded T1-hypointense lesion volume (LV) and T2-
tients with MS, one study reported that BVL occurs at a rate of 0.6% to hyperintense LV data. We extracted average patient baseline character-
1% per year [5]. In a study that examined a cohort of untreated MS pa- istics such as age, disease duration, and Kurtzke Expanded Disability
tients across subtypes for a median follow-up time of 14 months; it Status Scale (EDSS).
was found that BVL progressed relentlessly throughout the course of
the disease at a rate largely independent of subtype, after adjusting for 2.4. Risk of bias assessment
baseline brain volume [6].
Over the past decade, published studies in MS have reported BVL as a We developed a quality assessment form for each type of study de-
measurement of disease burden and discussed the possible factors that sign based on the Cochrane Handbook report of low, unclear, and high
influence the rate of BVL. To understand the natural history of BVL in MS risk of bias [8]. For studies in which participants were randomized, we
patients, this systematic literature review aimed to examine published assessed biases such as selection, performance, detection, attrition and
longitudinal studies (2003–2013) that reported BVL and to develop reporting using an assessment tool from the Cochrane Handbook [8].
consensus estimates of the annualized rate of BVL in MS stratified by For observational studies, we adapted the Newcastle-Ottawa Scale to
type of treatment. In addition, it is unclear whether BVL continues at a assess the following biases: selection, attrition, detection and informa-
constant rate throughout the disease course [7]. In particular, disease tion [9]. The form can be found in the online Supplemental materials
modifying treatments (DMTs), study population, study design, disease section.
duration, and imaging techniques may impact rates of observed BVL.
This study explored the associations between BVL and other features 2.5. Statistical analyses
that may impact measured rates of BVL.
When percentage brain volume measures were not provided, we
2. Methods calculated the percentage change by subtracting the baseline and
follow-up absolute brain volume measures. We divided the absolute
2.1. Study selection criteria difference by the baseline brain volume measures and multiplied by
100 to obtain the percentage change. For studies that did not annualize
We included clinical trials and longitudinal observational studies percentage brain volume measures, we annualized them by dividing the
that reported changes in brain volume measurements in MS patients, percentage brain volume measure change by the duration of the follow-
had at least 12 months of follow-up time, and specified the brain vol- up period. We estimated the standard error by dividing the standard de-
ume algorithm that was used. For clinical trials, we excluded studies if viation of the brain volume measures by the square root of the reported
a placebo or non-MS control group was not reported. sample size.
We computed a pooled estimate of the annual rate of BVL among
2.2. Search strategy studies that reported mean PBVC and used the SIENA algorithm, to facil-
itate comparability across studies. To better characterize the natural his-
We searched Medline, Embase, Cochrane Central Register of tory of brain volume loss in MS, we focused on studies that examined
Controlled Trials (CENTRAL), NHS Economic Evaluation Database patients treated with first-generation DMTs (e.g., interferon-beta [IFN]
(NHS EED) and the Cumulative Index to Nursing and Allied Health Liter- or glatiramer acetate [GA]). Heterogeneity across studies was assessed
ature (CINAHL) for studies published in English between January 2003 via the I2 statistic, which quantifies the degree of heterogeneity and de-
and September 2013. We hand-searched references of included reviews scribes the percentage of total variation across studies due to heteroge-
from 2013 and reviewed clinical trial registry to identify additional neity rather than chance [10]. We used meta-analysis with random-
studies that were not indexed in the electronic databases. The complete effects [11] to pool annualized PBVC across studies. In addition, we ex-
search strategies and results of the strategy are found in the online Sup- amined the pooled annualized PBVC by patients receiving first-
plemental materials. generation DMT and untreated patients, respectively.
To examine the relationship between study-level reported annual-
2.3. Study selection process and data extraction ized mean PBVC and the annualized mean changes in T1LV or T2LV,
we conducted random-effects meta-regression analysis to allow
Two independent reviewers applied the inclusion criteria and between-study variance not explained by the covariates by assuming
assessed the quality of the data collected using a standardized that the true effects follow a normal distribution around the linear

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014
 T. Vollmer et al. / Journal of the Neurological Sciences xxx (2015) xxx–xxx 3

Table 1
Study characteristics of included studies.

Ref. Author, year; study title Study design Participants Treatment Brain volume Sample Study
algorithm size duration
(months)

[14] Kappos, 2012; Long-term efficacy Randomized, RRMS patients Group 1, fingolimod 1.25 mg; SIENA 920 60
and safety of fingolimod (FTY720) double-blind, Group 2, fingolimod 0.5 mg;
in patients with placebo-controlled, Group 3, placebo;
relapsing–remitting multiple phase III study Group 4, placebo-Fingolimod
sclerosis.a 1.25 mg
[39] Radue, 2012; Impact of fingolimod Randomized, RRMS patients Group 1, fingolimod, 0.5 mg; SIENA 1272 24
therapy on magnetic resonance double-blind, Group 2, fingolimod, 1.25 mg;
imaging outcomes in patients with placebo-controlled Group 3, placebo
multiple sclerosis. study
[31] Kappos, 2010; A placebo-controlled Randomized, Patients diagnosed Group 1, fingolimod 1.25 mg; SIENA 1272 24
trial of oral fingolimod in relapsing double-blind, with MS according to Group 2, fingolimod 0.5 mg;
multiple sclerosis. placebo-controlled, the revised McDonald Group 3, placebo
phase III study criteria; a
relapsing–remitting
course
[27] Horakova, 2008; Evolution of Randomized, RRMS patients and Group 1, IFNb-1a 30 mg IM SIENA 174 up to 60
different MRI measures in patients double-blind, healthy controls once weekly plus two
with active relapsing–remitting placebo-controlled placebos;
multiple sclerosis over 2 and 5 Group 2, IFNb-1a 30 mg IM
years: a case–control study. once weekly plus Azathioprine
50 mg PO once a day plus
placebo;
Group 3, IFNb-1a 30 mg IM
once weekly plus Azathioprine
50 mg PO once a day plus
prednisone 10 mg PO every
other day
[50] Turner, 2003; Cerebral atrophy and Multicenter, RRMS, SPMS, and Interferon beta-1a Epidaure-Lib algorithms 122 48
disability in relapsing–remitting randomized, healthy controls
and secondary progressive multiple double-blind,
sclerosis over four years. placebo-controlled,
phase III
[20] Calabrese, 2012; Effect of Randomized, phase RRMS patients Interferon-beta-1a; Algorithm based on Fuzzy 191 24
disease-modifying drugs on cortical IV, pilot study Glatiramer acetate C-mean to get cortical
lesions and atrophy in lesions. GM and WM
relapsing–remitting multiple evaluated using SPM2
sclerosis.
[51] Zivadinov, 2007; Interferon beta-1a Open-label, RRMS patients Interferon beta-1a SPM99 54 36
slows progression of brain atrophy single-blind,
in relapsing–remitting multiple controlled study
sclerosis predominantly by
reducing gray matter atrophy.
[26] Hardmeier, 2005; Rate of brain Randomized, RRMS patients Group 1, interferon beta-1a 30 AutosegMS 386 36
atrophy in relapsing MS decreases non-controlled trial mg once weekly;
during treatment with IFNbeta-1a. Group 2, interferon beta-1a 60
mg once weekly
[45] Samann, 2012; Brain volume and Prospective RRMS, SPMS, and Almost all MS patients SIENA 98 12
diffusion markers as predictors of PPMS patients and received immunomodulatory
disability and short-term disease healthy controls therapy
evolution in multiple sclerosis.
[49] Tur, 2011; Complementary roles of Prospective PPMS patients and No treatment reported FSPGR, SPM2 65 60
grey matter MTR and T2 lesions in healthy controls
predicting progression in early
PPMS.
[17] Amato, 2010; Relevance of Prospective RRMS patients and Interferon beta SIENA 105 36
cognitive deterioration in early healthy controls
relapsing–remitting MS: a 3-year
follow-up study.
[18] Bendfeldt, 2010; Effect of Prospective RRMS patients Group 1, no SPM5 86 24
immunomodulatory medication on immunomodulatory
regional gray matter loss in medication;
relapsing–remitting multiple Group 2, interferon
sclerosis--a longitudinal MRI study. beta-1a/1b;
Group 3, glatiramer
[19] Calabrese, 2010; A 3-year magnetic Prospective RRMS and SPMS Most patients received DMT SPM2 and a semiautomatic 106 36
resonance imaging study of cortical patients during the study period thresholding technique
lesions in relapse-onset multiple based on the Fuzzy C-mean
sclerosis. algorithm and FLAIR
[25] Furby 2010; A longitudinal study of Prospective -- SPMS patients Placebo SIENA, SPM5 56 24
MRI-detected atrophy in secondary Placebo arm of a
progressive multiple sclerosis. randomized
controlled trial

(continued on next page)

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014
4 T. Vollmer et al. / Journal of the Neurological Sciences xxx (2015) xxx–xxx

Table 1 (continued)

Ref. Author, year; study title Study design Participants Treatment Brain volume Sample Study
algorithm size duration
(months)

[22] Lukas, 2010; Early central atrophy Prospective MS patients stratified Some patients received DMT SIENA 54 66
rate predicts 5 year clinical outcome by whether the during the study period
in multiple sclerosis. patient had disease
progression
[34] Martola, 2010; A longitudinal Prospective RRMS, SPMS, and Most patients were treated HERMES semiautomatic 37 Mean
observational study of brain PPMS patients with interferon during the tool (range)
atrophy rate reflecting four decades follow-up 111
of multiple sclerosis: a comparison (87.6–120)
of serial 1D, 2D, and volumetric
measurements from MRI images.
[38] Petzold, 2010; Evidence for Acute Prospective SPMS patients and Bone marrow transplant SIENA, SIENAX 42 36
Neurotoxicity after Chemotherapy healthy controls
[21] Calabrese, 2009; Cortical lesions in Prospective PPMS patients and None of the patients were Semiautomatic 70 24
primary progressive multiple healthy controls treated with corticosteroids thresholding technique
sclerosis: A 2-year longitudinal MR within the 3 months before based on Fuzzy C-mean
study. magnetic resonance algorithm
acquisitions, but thirteen
patients were treated with
immunosuppressive drugs at
study entry
[32] Khaleeli, 2008; Magnetization Prospective PPMS patients and No patient was on DMT at the SPM2 65 36
transfer ratio in gray matter: a healthy controls time of study entry
potential surrogate marker for
progression in early primary
progressive multiple sclerosis.
[16] Amato, 2007; Association of Prospective RRMS patients were Interferon beta-1a SIENA 28 Mean
Neocortical Volume Changes With further stratified by 30
Cognitive Deterioration in stable/improving or
Relapsing–Remitting Multiple deteriorating
Sclerosis.
[30] Jasperse, 2007; Regional brain Prospective MS patients Some patients received DMT SIENA 79 26.4
atrophy development is related to during the study period
specific aspects of clinical
dysfunction in multiple sclerosis.
[29] Jasperse, 2007; Determinants of Prospective RRMS and PPMS Some patients received DMT SIENAX and SIENA 89 24
cerebral atrophy rate at the time of patients during the study period
diagnosis of multiple sclerosis.
[42] Rovaris, 2007; Long-term follow-up Prospective — RRMS patients who Received GA, beta-interferon SIENA and SIENAX 142 Mean
of patients treated with glatiramer extension of a participated in the or mitoxantrone. Some 69.6
acetate: a multicenter, multinational double-blind, European/Canadian patients did not receive any
extension of the European/Canadian placebo-controlled, GA study DMT
double-blind, placebo-controlled, MRI monitored trial
MRI-monitored trial.
[15] Agosta, 2006; Magnetization Prospective, RRMS, SPMS, and CIS 60% received unspecified DMT SPM2 89 Mean
transfer MRI metrics predict the long-term follow-up MS patients during the study 92.4
accumulation of disability 8 years study
later in patients with multiple
sclerosis.
[44] Rovaris, 2006; Grey matter damage Prospective PPMS patients At study entry, patients were SIENA 54 Median
predicts the evolution of primary not receiving any DMT, were (range)
progressive multiple sclerosis at 5 treated with azathioprine or 56 (35–63)
years. pulses of intravenous
mitoxantrone or methotrexate
[37] Oreja-Guevara, 2005; Progressive Prospective RRMS patients Treatment naïve SIENA, SIENAX 26 18
gray matter damage in patients
with relapsing–remitting multiple
sclerosis: a longitudinal diffusion
tensor magnetic resonance imaging
study.
[43] Rovaris, 2005; Short-term accrual of Prospective PPMS and SPMS Patients received azathioprine, SIENA, BET 84 15
gray matter pathology in patients patients and healthy mitoxantrone, or
with progressive multiple sclerosis: controls methotrexate. Some patients
an in vivo study using diffusion did not receive any
tensor MRI. disease-modifying treatment
[46] Sastre-Garriga, 2005; Grey and Prospective PPMS patients No treatment reported SPM-based segmentation 31 Mean (IQR)
white matter volume changes in method and SIENA v2.2 12.2
early and primary progressive (11.4–13.1)
multiple sclerosis.
[33] Khan, 2012; Effect of Retrospective RRMS patients who Glatiramer acetate; SIENA 309 60
disease-modifying therapies on had no more than 5 Interferon beta
brain volume in years of disease
relapsing–remitting multiple duration and had not
sclerosis: Results of a five-year received prior DMT
brain MRI study.

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014
 T. Vollmer et al. / Journal of the Neurological Sciences xxx (2015) xxx–xxx 5

Table 1 (continued)

Ref. Author, year; study title Study design Participants Treatment Brain volume Sample Study
algorithm size duration
(months)

[36] Moodie, 2012; Magnetic resonance Retrospective RRMS, SPMS, and Non-specified DMT Template driven 84 Mean
disease severity scale (MRDSS) for PPMS patients segmentation (TDS +) 38.4
patients with multiple sclerosis: a
longitudinal study.
[41] Rojas, 2012; Brain atrophy at onset Retrospective MS patients All patients had unspecified SIENA 26 111.6
and physical disability in multiple DMT during the study period
sclerosis.
[28] Horakova, 2009; Gray matter Retrospective RRMS patients Most patients were treated SIENA 181 60
atrophy and disability progression analysis of a stratified by with interferon beta-1 and
in patients with early double-blind, progression state azathioprine
relapsing–remitting multiple placebo-controlled
sclerosis: a 5-year longitudinal study
study.
[47] Summers, 2008; Cognitive Retrospective RRMS patients No patient was on DMT at the SIENA 30 60
impairment in relapsing–remitting time of study entry
multiple sclerosis can be predicted
by imaging performed several years
earlier.
[35] Mesaros, 2007; Clinical and Retrospective RRMS patients No treatment reported Measurements of brain 548 14
conventional MRI predictors of volumes were made using
disability and brain atrophy a seed-growing technique
accumulation in RRMS. A large
scale, short-term follow-up study.
[40] Richert, 2006; Relationship Retrospective RRMS patients and Interferon beta-1a following SPM99 22 48
between inflammatory lesions and healthy controls baseline scans
cerebral atrophy in multiple
sclerosis.
[48] Tiberio, 2005; Gray and white Retrospective RRMS patients and No treatment reported; No Tissue segmentation 31 24
matter volume changes in early healthy controls patients had DMT at the time derived from SPM99
RRMS: a 2-year longitudinal study. of the first scan
[23] Chard, 2004; Progressive grey Retrospective RRMS patients and No beta-interferon or Tissue segmentation 22 18
matter atrophy in clinically early healthy controls glatiramer prior to or during derived from SPM99
relapsing–remitting multiple the study
sclerosis.
[24] Chen, 2004; Relating neocortical Retrospective RRMS and SPMS No treatment reported SIENA 39 12
pathology to disability progression patients or healthy
in multiple sclerosis using MRI. controls. Patients
were stratified by
stable disability and
progressing disability

MS: multiple sclerosis; CIS: clinically isolated syndrome; RRMS: relapsing/remitting MS; SPMS: secondary progressive MS: PPMS: primary progressive MS; SPM: statistical parametric
mapping. SPM has several different versions such as SPM99 and SPM5; FLAIR: fluid attenuated inversion recovery; SIENA: Structural Image Evaluation of Normalized Atrophy; SIENAX:
cross-sectional version of SIENA; BET: brain extraction tool; FSPGR: fast spoiled gradient echo; DMT: disease modifying therapy; IFNb-1a: interferon beta-1a; GA: glatiramer acetate.
a
Additional data were extracted from ClinicalTrials.gov (Registry # NCT00662649).

predictor [12]. We also assessed the association between the study-level screened the full-text reports for 518 studies. We excluded 480 studies
average MS disease duration in years with annualized mean PBVC. due to publication type (N = 13), study design (N = 28), population
(N = 178), and lack of suitable brain volume measures (N = 261). In
2.6. Data management and reporting total, we included 38 studies in this review [14–51]. Fig. 1 illustrates
the flow diagram of the identified studies.
We used Endnote version X5 to store the bibliographic citations
from the electronic search. For data entry and descriptive analyses, we
used Microsoft Excel 2010. We used STATA (STATA version 13.0, 3.2. Overall study description
StataCorp, College Station, TX, USA) to perform the meta-analyses and
meta-regression models. We followed the Preferred Reporting Items Of the 38 articles, 8 were clinical trials [14,20,26,27,31,39,50,51] and
for Systematic Reviews and Meta-Analysis (PRISMA) in reporting the 30 were observational studies [15–19,21–25,28–30,32–38,40–49]. Nine
systematic review and meta-analysis [13]. studies included a healthy control arm [15,21,23,27,32,40,48–50]. A ma-
jority of the studies (N = 20) recruited only relapsing–remitting (RR)
3. Results MS patients [14,16–18,20,23,26–28,31,33,35,37,39–42,47,48,51]. Six-
teen studies recruited primary progressive (PP) MS and secondary pro-
3.1. Search gressive (SP) MS patients [15,21,22,24,25,29,30,32,34,36,38,43–46,49].
The study durations ranged from 12 to 111 months. Average patient
We identified 2837 articles through the electronic search databases ages ranged from 30 to 51 years. Most studies had a higher proportion
(online Supplemental materials). Of the 2837 studies, after screening of females than males. Among the 34 studies reporting EDSS at baseline,
the titles and abstracts we excluded 2319 articles based on publication the mean scores ranged from 1.1 to 6.0 [15–33,35–51]. The reported
type (i.e., commentary, narrative report) (N = 425), study design brain volume algorithms were SIENA [14,16,17,22,24,25,27–31,33,
(e.g., follow-up was less than 12 months) (N = 506), population (e.g., 37–39,41–45,47] (N = 21), seed-growing technique [35] (N = 1),
non-adults) (N = 256), and lack of suitable brain volume measures Epidaure-Lib algorithms [50] (N = 1), algorithm based on Fuzzy
(e.g., reported only thalamic volume) (N = 1132). We retrieved and C-mean [21] (N = 1), statistical parametric mapping [15,18–20,23,32,

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014
6 T. Vollmer et al. / Journal of the Neurological Sciences xxx (2015) xxx–xxx

Fig. 1. Study flow diagram on the natural history of brain volume change in multiple sclerosis patients.

40,46,48,49,51] (N = 11), template driven segmentation [36] (N = 1), were retrospective studies [23,24,28,33,35,36,40,41,47,48]. In 20 (67%)
Hermes semi-automatic tool [34] (N = 1), and Autoseg [26] (N = 1). studies, authors reported that patients were receiving DMTs during
Among the 8 clinical trials, 7 (88%) studies enrolled only RRMS pa- the course of MS disease [15–19,21,22,28–30,33,34,36,38,40–45].
tients [14,20,26,27,31,39,51]. Three trials compared fingolimod and pla-
cebo [14,31,39], four trials compared IFN and placebo or healthy 3.3. Quality assessment of the included studies
controls [26,27,50,51], and one head-to-head trial examined IFN, GA,
and untreated MS patients [20]. Among the clinical trial studies, moderate to low risk of bias was
Of the 30 observational studies (N = 2748), 20 (67%) were prospec- observed (Table 2a). In a majority of the studies (N 75%), the method of
tive studies [15–19,21,22,25,29,30,32,34,37,38,42–46,49] and 10 (33%) randomization and allocation concealment, and presence of incomplete

Table 2a
Risk of bias assessment — clinical trial studies.

a
Low risk of bias +
Unclear risk of bias ?
High risk of bias -

Author, year Random sequence Allocation Blinding of participants Blinding of outcome Incomplete outcome data Selective reporting
generation (selection concealment (selection and personnel assessment (detection bias) (attrition bias) (reporting bias)
bias) bias) (performance bias)
Calabrese, 2012 + ? + + + +
Hardmeier, 2005 ? ? + + ? +
Horakova, 2008 ? ? + + ? +
Kappos, 2010 ? ? + + ? +
Kappos, 2012 ? ? + + ? +
Radue, 2012 ? ? + + ? +
Turner, 2003 ? ? + + ? +
Zivadinov, 2007 - - - + ? +
Each quality assessment category was coded according to the level of information available or addressed by the authors. Green indicates low risk of bias (i.e., plausible bias unlikely to
seriously alter the results). Yellow indicates unclear risk of bias (i.e., plausible bias that raises some doubt about the results). Red indicates high risk of bias (i.e., plausible bias that seriously
weakens confidence in the results).

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014
 T. Vollmer et al. / Journal of the Neurological Sciences xxx (2015) xxx–xxx 7

outcome data were not described by the authors and were marked as un- potential confounders. Of the 30 observational studies, seven studies
clear risk of bias. Of the eight clinical trials, high risk of bias for random- [19,36,40,41,43,44,46] did not account for potential confounders in the
ization, allocation concealment, and blinding was found in the Zivadinov analyses.
et al. study [51] because it was an open-label, single blind study.
Observational studies included prospective and retrospective stud- 3.4. Brain volume measurements
ies (Table 2b). Overall, there was moderate to low risk of bias. Some
studies were marked as high risk of bias because of the potential pres- Seventeen studies reported annualized mean PBVC [12,14,15,23,
ence of bias based on the study design and reporting of information in 27–29,31,35–37,39–41,43–45]. Of those 17 studies, 6 reported longitu-
the full-text. Eligibility criteria were explicitly described in most studies dinal PBVC measures [14,17,31,33,38,39]. The annualized mean PBVC
except for four studies [23,43,44,47] in which information about the in- ranged between − 2.1% and − 0.22%. The annualized mean PBVC
clusion criteria was not reported. For two studies [27,37], it appears that ranged between − 1.34% and − 0.22% for RRMS patients, between
the association of interest in the study population may be different from − 2.1% and − 0.59% for SPMS patients, and between − 0.94% and
the targeted population. There was high risk of bias for three studies [23, −0.63% for PPMS patients. Five studies reported annualized mean BPF
35,45] where some patients were excluded from the analysis. An imbal- [15,23,26,46,48]. The annualized mean percent changes in BPF ranged
ance of the number of patients excluded from the analysis in the various between −2.44% and −0.23%. Mean percent changes in BPF were re-
comparison groups was observed for two studies [33,45]. Two studies ported for non-MS patients and ranged between − 0.48% and 0%
did not report how the authors accounted for missing data [35,43]. De- among the studies that included a healthy control cohort [15,23,48].
tection bias was low for all observational studies except for two studies Tables 3a–3c report the average patient characteristics and changes in
[18,38] in which the participants were knowledgeable about the out- brain volume measurements and can be found in the online Supple-
come status. Low risk of bias whereby the same protocol was adminis- mental materials section.
tered to all comparator groups was observed across all observational Sixteen studies reported GM or WM volumes [15,16,18–21,23,25,27,
studies. Potential misclassification bias was present in one study [36]. 32,37,38,46,48,49,51]. High variability existed across studies reporting
In a majority of the studies, it was unclear how the authors addressed GMF and WMF with ranges between −2.36% and 0.01% for annualized

Table 2b
Risk of bias assessment — observational studies.

b
Low risk of bias +
Unclear risk of bias ?
High risk of bias -

Author, year Eligibility Selection of Exclusion of Proportion of Accounting Ascertainment Ascertainment Methods for Controlled for
criteria eligible participants excluded for missing of exposure of exposure ascertainment of confounders
explicitly population from participants data without using the same exposure and
described from the analysis of were similar (attrition participant’s measurement outcome was
(selection target the outcome between bias) knowledge tool for all susceptible to
bias) population (selection exposed an (detection participants misclassification
(selection bias) unexposed bias) (detection (information bias)
bias) groups bias)
(selection bias)
Khan, 2012 + + ? - ? + + + +
Moodie, 2012 + + ? ? ? + + - -
Rojas, 2012 ? + ? + ? + + + -
Horakova, 2009 + - ? + ? + + + +
Summers, 2008 - ? + ? + + + + +
Mesaros, 2007 + + - ? - + + + ?
Richert, 2006 + + + + + + + + -
Tiberio, 2005 ? ? + + + + + ? +
Chard, 2004 - ? - ? ? + + + ?
Chen, 2004 + + ? ? ? + + + +
Agosta, 2006 + + ? + + + + ? +
Calabrese, 2010 ? + + + + + + + -
Furby, 20101 + + + + + + + ? +
Oreja-Guevara, 2005 + - + + + + + + +
Rovaris, 2005 - ? + + - + ? + -
Rovaris, 2006 - ? ? ? ? + + ? -
Samann, 2012 + + - - ? + + ? ?
Sastre-Garriga, 2005 ? + ? ? ? + + ? -
Tur, 2011 ? + ? ? + + + + +
Lukas, 2010 + + + + ? + + + ?
Martola, 2010 ? ? + + ? + + ? ?
Petzold, 2010 ? ? + + + - + ? ?
Calabrese, 2009 ? + + + ? + + + +
Khaleeli, 2008 + + + + + + + + +
Amato, 2007 + + + + + + + + ?
Jasperse, 20072 + + + + + + + + ?
Jasperse, 20073 + + + + + + + + +
Rovaris, 20071 + + + + + + + + ?
Amato, 2010 + + + + + + + + +
Bendfeldt, 2010 ? ? ? ? ? - + + +
[1] The study was an extension/follow-up of a trial cohort.
[2] Jasperse, Bas, et al. “Regional brain atrophy development is related to specific aspects of clinical dysfunction in multiple sclerosis.” Neuroimage 38.3 (2007): 529–537.
[3] Jasperse, Bas, et al. “Determinants of cerebral atrophy rate at the time of diagnosis of multiple sclerosis.” Archives of neurology 64.2 (2007): 190–194.
Each quality assessment category was coded according to the level of information available or addressed by the authors. Green indicates low risk of bias (i.e., plausible bias unlikely to
seriously alter the results). Yellow indicates unclear risk of bias (i.e., plausible bias that raises some doubt about the results). Red indicates high risk of bias (i.e., plausible bias that seriously
weakens confidence in the results).

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014
8 T. Vollmer et al. / Journal of the Neurological Sciences xxx (2015) xxx–xxx

mean percent changes in GMF and − 1.88% to 1.87% for annualized three times a week or every other day [55]. In addition, they found
mean percent changes in WMF. Seventeen studies reported T1LV or that in trials of daily subcutaneous GA, a reduction in inflammatory le-
T2LV and the ranges varied greatly [16,23,25–28,31,36–39,41–44,48, sions was observed but no difference in BVL between the intervention
51]. The annualized mean percent change ranged between − 6.67% and placebo arms was reported [55]. The “pseudoatrophy” phenome-
and 25.35% in T1LV and between 1.36% and 22.61% in T2LV. Three stud- non is a possible explanation for the uncertainty of the effect of first gen-
ies reported the correlations between brain volume and T1LV or T2LV eration DMTs on BVL because the anti- inflammatory role of DMTs may
[27,48,51]. The correlations between BPF and T2LV were − 0.58 and appear to decrease BV even as it might help reduce future BVL.
−0.51; correlation between BPF and T1LV was −0.68; and the partial This study did not assess the effect of contemporary DMTs on BVL in
correlation analysis between PBVC (0–12 months) and T2LV (0–60 a meta-analysis because of the limited brain volume data available at
months) was −0.36 adjusting for age at baseline MRI scan. the time the systematic literature review was conducted. Among the
DMTs that were included in the analysis, fingolimod was the only con-
3.5. Meta-analyses and meta-regressions temporary DMT study that reported BVL over time. Since then, volumet-
ric data for delayed-release dimethyl fumarate (DMF) has become
Among the 12 studies (N = 767 MS patients) that included untreat- available. Arnold et al. found that the median percentage change in
ed patients or patients treated with first-generation DMT with IFN or brain volume from baseline to 24 months was − 0.64% and − 0.77%
GA, reported PBVC rates and acquired BV with the SIENA algorithm, for delayed-release DMF twice daily and thrice daily, respectively [56].
the annualized PBVC ranged from −1.34% to −0.46% per year [16,17, The DMF results show that there is a significant 21% reduction in the
25,29,30,37,41–43,45–47]. The pooled annualized PBVC was − 0.69% PBVC among patients treated with DMF twice daily compared with pa-
(95% CI = −0.87% to −0.50%, I2 = 97%) in study arms receiving first- tients in the placebo arm (p = 0.04). In the FREEDOMS I and II trials,
generation DMT (N = 6 studies) [16,17,41–43,45] and − 0.71% (95% fingolimod 1.25 mg and 0.5 mg showed similarly significant (p b 0.05
CI = −0.81% to −0.61%, I2 = 0%) in untreated study arms (N = 6 stud- in both trials) reductions in PBVC from baseline to 24 months compared
ies) [25,29,30,37,46,47]. High heterogeneity in the study arms receiving with patients in the placebo arms (fingolimod 1.25 mg: − 0.89% and
first-generation DMT was driven by a wide spread of annualized PBVC −0.60%, respectively; fingolimod 0.5 mg: −0.84% and −0.86%, respec-
estimates. Fig. 2a–b provides a schematic diagram of the meta- tively) [31,57].
analysis for first-generation DMT and untreated patients, respectively. We reported data on changes in regional BVL (i.e., GMF and WMF)
We did not observe an apparent trend (p = 0.696) when investigat- and lesions and observed that the annual changes in regional BVL and
ing the association between annualized mean T1LV and annualized lesions were variable across studies. The variability may be due to the
mean PBVC among the four study arms that reported both annualized differences in brain volume algorithm used to measure regional brain
mean PBVC and T1LV [37,42,43]. Similarly, no trend was found when volume and lesion volumes as well as MS disease course. In cross-
we examined the association between annualized mean PBVC and an- sectional studies and clinical trials, BVL has been shown to be associated
nualized mean T2LV (p = 0.162) [37,42,43]. At the study level, longer with T2LV, which suggests a relationship between inflammatory lesion
average MS disease duration was numerically associated with a 0.02% rate and BVL [7,58]. In our systematic literature review that examines
(p = 0.262) greater decline in annualized mean PBVC. longitudinal studies, we did not observe this relationship. This may be
due to the limited number of studies that reported both annualized
4. Discussion PBVC and annualized changes in T1LV or T2LV. Hence, associations be-
tween average annualized PBVC and annualized changes in T1LV or
This systematic literature review and meta-analysis found that BVL T2LV were not observed in the meta-regression analysis. Consistent
over the MS disease course was frequently reported in both clinical with long-term follow-up studies that showed poor relationship be-
and observational studies in RRMS, PPMS, and SPMS patients. We tween disease duration and rate of BVL, we did not observe an associa-
found that reported rates of BVL in MS are above rates expected for tion between annualized PBVC and disease duration at the study level
normal aging and consistent with rates seen in prodromal states of [59]. The lack of association may be due to the higher rates of BVL at
Alzheimer's Disease [52]. In De Stefano et al.'s review, the authors re- MS onset reported in some studies while other studies reported higher
ported that the annual rate of BVL ranged between 0.5% and 1.35% per BVL at the more advanced RRMS phase [59]. However, we did observe in
year in patients with RRMS compared with the normal age-related de- some studies that the rate of BVL slowed down over the course of MS
terioration in healthy individuals of 0.1% to 0.3% per year [53]. BVL has [17,22,33,38,60,61].
been observed from the earliest phase of disease in patients with clini- This study has a few notable limitations. Firstly, we focused on pub-
cally isolated syndrome (CIS) [54]. lished papers from 2003 to 2013 to minimize bias due to the advance-
The present study reports measurements of the annualized rate of ments in imaging technology and brain volume algorithms. This
BVL over time, characterizing each population with a single rate of inclusion criterion limits the generalizability for studies published be-
change, and did not quantify potential differences in the rate of BVL fore 2003. Secondly, we reviewed only studies in English, which exclud-
over time. However, we informally observed that in studies that report- ed studies published in other languages that may have provided
ed multiple brain volume measurements over time, the annualized rate additional insights on the natural history of BVL. Because of the exclu-
of BVL was almost always steeper in the first year compared with subse- sion of non-English studies, the applicability of our findings to non-
quent measurements (Table 3a in the online Supplemental material English speaking countries is limited. Thirdly, we performed the meta-
file) [17,22,33,38]. This warrants further quantification. A steeper de- analysis on studies that reported SIENA algorithm to avoid the potential
cline in brain volume observed in the earlier measurements could be introduction of heterogeneity; thus, our meta-analysis findings might
due to greater levels of inflammatory disease in the early phase of MS not be generalizable to studies that used different algorithms. Fourthly,
or it may be due to, at least in part, the “pseudoatrophy” phenomenon our assessment that there was no apparent difference in BVL between
in which the anti-inflammatory properties of DMTs significantly de- untreated MS patients and MS patients with first-generation DMTs
crease BV within the first 6 months to 1 year of treatment [53]. was based on the overlap of the 95% confidence intervals for the pooled
The extent of BVL over time was, on average, similar among patients annualized PBVC of first-generation DMTs and untreated MS patients,
treated with first-generation DMTs and untreated MS patients. This respectively. We did not perform a meta-analysis of pairwise differ-
finding was consistent with a previous review that did not find a clear ences between treated and untreated groups for each trial because we
effect of first generation DMTs on BVL. In a review by Bermel and Bakshi, did not have any studies of first generation DMTs with both a treated
they reported that the effect of treatment on the rate of BVL has been and untreated arm. Furthermore, because the meta-analysis focused
unclear in randomized clinical trials where IFN was administered on first-generation DMTs, results from the FREEDOMS trials were not

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014
 T. Vollmer et al. / Journal of the Neurological Sciences xxx (2015) xxx–xxx 9

a) Meta-analysis of annualized PBVC in treated MS patients

Author, Year N Annualized PBVC (95% CI) % Weight

Amato, 2010a 49 -0.72 (-0.99, -0.45) 10.17

Amato, 2010b 49 -0.16 (-0.26, -0.06) 12.31

Samann, 2012 44 -0.46 (-0.82, -0.10) 8.69

Amato, 2007a 16 -0.52 (-0.77, -0.27) 10.37

Amato, 2007b 12 -0.84 (-1.11, -0.57) 10.11

Rovaris, 2007 142 -0.87 (-0.93, -0.80) 12.54

Rovaris, 2005a 54 -0.99 (-1.03, -0.96) 12.69

Rovaris, 2005b 22 -0.94 (-1.01, -0.88) 12.57

Rojas, 2012 26 -0.57 (-0.81, -0.33) 10.56

Overall (I-squared = 97.0%, p = 0.000) -0.68 (-0.87, -0.49) 100.00

-1.25 -.75 -.25 0

b) Meta-analysis of annualized PBVC in untreated MS patients

Author, Year N Annualized PBVC (95% CI) % Weight

Furby, 2010 56 -0.59 (-0.79, -0.39) 25.57

Jasper, 2007.1b 15 -0.90 (-1.20, -0.60) 10.66

Jasper, 2007.2a 56 -0.80 (-1.01, -0.59) 22.38

Oreja, 2005 26 -0.61 (-0.96, -0.25) 7.89

Sastre, 2005 31 -0.63 (-1.00, -0.26) 7.19

Summers, 2007 30 -0.73 (-0.92, -0.53) 26.31

Overall (I-squared = 0.0%, p = 0.523) -0.71 (-0.81, -0.61) 100.00

-1.25 -.75 -.25 0

Fig. 2. a. Meta-analysis of annualized PBVC in treated MS patients. b. Meta-analysis of annualized PBVC in untreated MS patients.

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014
10 T. Vollmer et al. / Journal of the Neurological Sciences xxx (2015) xxx–xxx

included in the meta-analysis; however, we observed that the annual- [6] N. De Stefano, A. Giorgio, M. Battaglini, et al., Assessing brain atrophy rates in a large
population of untreated multiple sclerosis subtypes, Neurology 74 (23) (2010)
ized PBVC in the untreated arms of the FREEDOMS I trial (− 0.65%) 1868–1876.
[31] and FREEDOMS II trial (−0.64%) [57] were similar to the annual- [7] R. Zivadinov, R. Bakshi, Role of MRI in multiple sclerosis II: brain and spinal cord at-
ized PBVC of the pooled untreated patients in the meta-analysis rophy, Front. Biosci. 9 (2004) 647–664.
[8] Cochrane handbook for systematic reviews of interventions version 5.1.0 [updated
(−0.71% [95% CI = −0.81% to −0.61%]). In addition, our computation March 2011], http://www.cochrane-handbook.org/, 2011 (Accessed August 6, 2012).
of PBVC is based on the mean brain volume measures as opposed to [9] G. Wells, B. Shea, D. O'Connell, J. Peterson, V. Welch, M. Losos, P. Tugwell, The
patient-specific difference in brain volume change for 7 out of 14 stud- Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised studies
in meta-analysis, http://www.ohri.ca/programs/clinical_epidemiology/oxford.asp
ies. This computation may have a nominal effect on the reported per- (Accessed August 6, 2012).
centage change. Finally, we included observational studies with [10] J.P. Higgins, S.G. Thompson, J.J. Deeks, D.G. Altman, Measuring inconsistency in
different eligibility criteria and study objectives in the meta-analysis meta-analyses, BMJ 327 (7414) (2003) 557–560.
[11] R. DerSimonian, N. Laird, Meta-analysis in clinical trials, Control. Clin. Trials 7 (3)
and meta-regression analysis. We examined the influence of study de-
(Sep 1986) 177–188.
sign on the analysis of annualized rate of PBVC across studies and [12] S.G. Thompson, J.P. Higgins, How should meta-regression analyses be undertaken
found that the pooled estimates were similar. and interpreted? Stat. Med. 21 (11) (2002) 1559–1573.
[13] A. Liberati, D.G. Altman, J. Tetzlaff, et al., The PRISMA statement for reporting sys-
tematic reviews and meta-analyses of studies that evaluate healthcare interven-
5. Conclusion tions: explanation and elaboration, BMJ 339 (2009) (2009-07-21 11:46:49).
[14] Kappos, Long-term efficacy and safety of fingolimod (FTY720) in patients with
relapsing–remitting multiple sclerosis, http://www.clinicaltrials.gov/ct2/show/
In this systematic review, the average MS patient receiving either NCT00662649?term=Long-term+Efficacy+and+Safety+of+Fingolimod+%
first-generation DMT or no DMT lost approximately 0.7% of BV per 28FTY720%29+in+Patients+With+Relapsing-remitting+Multiple&rank=1,
year, well above rates associated with normal aging (0.1% to 0.3% of 2012.
[15] F. Agosta, M. Rovaris, E. Pagani, M.P. Sormani, G. Comi, M. Filippi, Magnetization
BV per year) [5,53]. There was no apparent difference in the rate of transfer MRI metrics predict the accumulation of disability 8 years later in patients
BVL between untreated MS patients and MS patients treated with inter- with multiple sclerosis, Brain 129 (Pt 10) (Oct 2006) 2620–2627.
ferons or glatiramer acetate. [16] M.P. Amato, E. Portaccio, B. Goretti, et al., Association of neocortical volume changes
with cognitive deterioration in relapsing–remitting multiple sclerosis, Arch. Neurol.
64 (8) (Aug 2007) 1157–1161.
[17] M.P. Amato, E. Portaccio, B. Goretti, et al., Relevance of cognitive deterioration in
Author contributions
early relapsing–remitting MS: a 3-year follow-up study, Mult. Scler. 16 (12)
(Dec 2010) 1474–1482.
Lynn Huynh, Philip Galebach, and Caroline Kelley contributed to the [18] K. Bendfeldt, H. Egger, T.E. Nichols, et al., Effect of immunomodulatory medication
collection, management, analysis and interpretation of the data, manu- on regional gray matter loss in relapsing–remitting multiple sclerosis—a longitudi-
nal MRI study, Brain Res. 1325 (2010) 174–182.
script development, and approval of the final submitted version. James [19] M. Calabrese, M.A. Rocca, M. Atzori, et al., A 3-year magnetic resonance imaging
Signorovitch, Rahul Sasane, Allitia DiBernardo, and Timothy Vollmer study of cortical lesions in relapse-onset multiple sclerosis, Ann. Neurol. 67 (3)
conceived and planned the work that led to the manuscript, provided (April 2010).
[20] M. Calabrese, V. Bernardi, M. Atzori, et al., Effect of disease-modifying drugs on
substantive suggestions for revisions, and approved the final submitted cortical lesions and atrophy in relapsing–remitting multiple sclerosis, Mult. Scler.
version. 18 (4) (Apr 2012) 418–424.
[21] M. Calabrese, M.A. Rocca, M. Atzori, et al., Cortical lesions in primary progressive
multiple sclerosis: a 2-year longitudinal MR study, Neurology 72 (15) (2009)
Declaration of conflicting interests 1330–1336.
[22] C. Lukas, A. Minneboo, V. De Groot, B. Moraal, D. Knol, C. Polman, F. Barkhof, H.
Vrenken, Early central atrophy rate predicts 5 year clinical outcome in multiple scle-
James Signorovitch, Lynn Huynh, Philip Galebach, and Caroline rosis, Journal of Neurology, Neurosurgery & Psychiatry (2010) jnnp-2009.
Kelley are employees of Analysis Group, Inc., a consulting company [23] D.T. Chard, C.M. Griffin, W. Rashid, et al., Progressive grey matter atrophy in clinical-
that has received research grants from Novartis Pharmaceuticals Corpo- ly early relapsing–remitting multiple sclerosis, Mult. Scler. 10 (4) (2004) 387–391.
[24] J.T. Chen, S. Narayanan, D.L. Collins, S.M. Smith, P.M. Matthews, D.L. Arnold, Relating
ration. Rahul Sasane is an employee of Novartis Pharmaceuticals neocortical pathology to disability progression in multiple sclerosis using MRI,
Corporation. NeuroImage 23 (3) (Nov 2004) 1168–1175.
[25] J. Furby, T. Hayton, D. Altmann, et al., A longitudinal study of MRI-detected atrophy
in secondary progressive multiple sclerosis, J. Neurol. 257 (9) (Sep 2010)
Acknowledgments 1508–1516.
[26] M. Hardmeier, S. Wagenpfeil, P. Freitag, et al., Rate of brain atrophy in relapsing MS
decreases during treatment with IFNbeta-1a, Neurology 64 (2) (2005) 236–240.
This work was financially supported by Novartis Pharmaceuticals [27] D. Horakova, J.L. Cox, E. Havrdova, et al., Evolution of different MRI measures in
Corporation. patients with active relapsing–remitting multiple sclerosis over 2 and 5 years: a
case–control study, J. Neurol. Neurosurg. Psychiatry 79 (4) (Apr 2008) 407–414.
[28] D. Horakova, M.G. Dwyer, E. Havrdova, et al., Gray matter atrophy and disability
Appendix A. Supplementary material progression in patients with early relapsing–remitting multiple sclerosis: a 5-year
longitudinal study, J. Neurol. Sci. 282 (1–2) (2009) 112–119.
[29] B. Jasperse, A. Minneboo, V. de Groot, et al., Determinants of cerebral atrophy rate at
Supplementary material to this article can be found online at http:// the time of diagnosis of multiple sclerosis, Arch. Neurol. 64 (2) (Feb 2007) 190–194.
dx.doi.org/10.1016/j.jns.2015.07.014. [30] B. Jasperse, H. Vrenken, E. Sanz-Arigita, et al., Regional brain atrophy development is
related to specific aspects of clinical dysfunction in multiple sclerosis, NeuroImage
38 (3) (2007) 529–537.
References [31] L. Kappos, E.W. Radue, P. O'Connor, et al., A placebo-controlled trial of oral
fingolimod in relapsing multiple sclerosis, N. Engl. J. Med. 362 (5) (2010) 387–401.
[1] K. Noyes, B. Weinstock-Guttman, Impact of diagnosis and early treatment on [32] Z. Khaleeli, D.R. Altmann, M. Cercignani, O. Ciccarelli, D.H. Miller, A.J. Thompson,
the course of multiple sclerosis, Am. J. Manag. Care 19 (17 Suppl.) (Nov 2013) Magnetization transfer ratio in gray matter: a potential surrogate marker for
s321–s331. progression in early primary progressive multiple sclerosis, Arch. Neurol. 65 (11)
[2] Novartis gains FDA approval for Gilenya(TM), a novel first-line multiple sclerosis (Nov 2008) 1454–1459.
treatment shown to significantly reduce relapses and delay disability progression, [33] O. Khan, F. Bao, M. Shah, et al., Effect of disease-modifying therapies on brain vol-
http://www.novartis.com/newsroom/media-releases/en/2010/1445917.shtml ume in relapsing–remitting multiple sclerosis: results of a five-year brain MRI
(Accessed April 18, 2014). study, J. Neurol. Sci. 312 (1–2) (2012) 7–12.
[3] N.C. Fox, R. Jenkins, S.M. Leary, et al., Progressive cerebral atrophy in MS: a serial [34] J. Martola, J. Bergstrom, S. Fredrikson, et al., A longitudinal observational study of
study using registered, volumetric MRI, Neurology 54 (4) (2000) 807–812. brain atrophy rate reflecting four decades of multiple sclerosis: a comparison of se-
[4] D.H. Miller, F. Barkhof, J.A. Frank, G.J. Parker, A.J. Thompson, Measurement of atro- rial 1D, 2D, and volumetric measurements from MRI images, Neuroradiology 52 (2)
phy in multiple sclerosis: pathological basis, methodological aspects and clinical rel- (Feb 2010) 109–117.
evance, Brain 125 (Pt 8) (Aug 2002) 1676–1695. [35] S. Mesaros, M.A. Rocca, M.P. Sormani, A. Charil, G. Comi, M. Filippi, Clinical and con-
[5] C.S. Raine, H.F. McFarland, R. Hohlfeld, Multiple Sclerosis: A Comprehensive Text, ventional MRI predictors of disability and brain atrophy accumulation in RRMS. A
Saunders/Elsevier, 2008. large scale, short-term follow-up study, J. Neurol. 255 (9) (Sep 2008) 1378–1383.

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014
 T. Vollmer et al. / Journal of the Neurological Sciences xxx (2015) xxx–xxx 11

[36] J. Moodie, B.C. Healy, G.J. Buckle, et al., Magnetic resonance disease severity scale [49] C. Tur, Z. Khaleeli, O. Ciccarelli, et al., Complementary roles of grey matter MTR and
(MRDSS) for patients with multiple sclerosis: a longitudinal study, J. Neurol. Sci. T2 lesions in predicting progression in early PPMS, J. Neurol. Neurosurg. Psychiatry
315 (1–2) (2012) 49–54. 82 (4) (Apr 2011) 423–428.
[37] C. Oreja-Guevara, M. Rovaris, G. Iannucci, et al., Progressive gray matter damage in [50] B. Turner, X. Lin, G. Calmon, N. Roberts, L.D. Blumhardt, Cerebral atrophy and disabil-
patients with relapsing–remitting multiple sclerosis: a longitudinal diffusion tensor ity in relapsing–remitting and secondary progressive multiple sclerosis over four
magnetic resonance imaging study, Arch. Neurol. 62 (4) (Apr 2005) 578–584. years, Mult. Scler. 9 (1) (2003) 21–27.
[38] A. Petzold, T. Mondria, J. Kuhle, et al., Evidence for acute neurotoxicity after chemo- [51] R. Zivadinov, L. Locatelli, D. Cookfair, et al., Interferon beta-1a slows progression of
therapy, Ann. Neurol. 68 (6) (Dec 2010) 806–815. brain atrophy in relapsing–remitting multiple sclerosis predominantly by reducing
[39] E.W. Radue, P. O'Connor, C.H. Polman, et al., Impact of fingolimod therapy on mag- gray matter atrophy, Mult. Scler. 13 (4) (May 2007) 490–501.
netic resonance imaging outcomes in patients with multiple sclerosis, Arch. Neurol. [52] C. Davatzikos, Y. Fan, X. Wu, D. Shen, S.M. Resnick, Detection of prodromal
69 (10) (Oct 2012) 1259–1269. Alzheimer's disease via pattern classification of MRI, Neurobiol. Aging 29 (4)
[40] N.D. Richert, T. Howard, J.A. Frank, et al., Relationship between inflammatory lesions (2008) 514–523.
and cerebral atrophy in multiple sclerosis, Neurology 66 (4) (2006) 551–556. [53] N. De Stefano, L. Airas, N. Grigoriadis, et al., Clinical relevance of brain volume mea-
[41] J.I. Rojas, L. Patrucco, C. Besada, L. Bengolea, E. Cristiano, Brain atrophy at onset and sures in multiple sclerosis, CNS Drugs 28 (2) (Feb 2014) 147–156.
physical disability in multiple sclerosis, Arq. Neuropsiquiatr. 70 (10) (Oct 2012) [54] E. Fisher, J.C. Lee, K. Nakamura, R.A. Rudick, Gray matter atrophy in multiple sclero-
765–768. sis: a longitudinal study, Ann. Neurol. 64 (3) (Sep 2008) 255–265.
[42] M. Rovaris, G. Comi, M.A. Rocca, et al., Long-term follow-up of patients treated with [55] R.A. Bermel, R. Bakshi, The measurement and clinical relevance of brain atrophy in
glatiramer acetate: a multicentre, multinational extension of the European/Canadian multiple sclerosis, Lancet Neurol. 5 (2) (Feb 2006) 158–170.
double-blind, placebo-controlled, MRI-monitored trial, Mult. Scler. 13 (4) (May [56] D.L. Arnold, R. Gold, L. Kappos, et al., Effects of delayed-release dimethyl fumarate on
2007) 502–508. MRI measures in the phase 3 DEFINE study, J. Neurol. 261 (9) (Sep 2014) 1794–1802.
[43] M. Rovaris, A. Gallo, P. Valsasina, et al., Short-term accrual of gray matter pathology [57] P.A. Calabresi, E.W. Radue, D. Goodin, et al., Safety and efficacy of fingolimod in pa-
in patients with progressive multiple sclerosis: an in vivo study using diffusion ten- tients with relapsing–remitting multiple sclerosis (FREEDOMS II): a double-blind,
sor MRI, NeuroImage 24 (4) (2005) 1139–1146. randomised, placebo-controlled, phase 3 trial, Lancet Neurol. 13 (6) (Jun 2014)
[44] M. Rovaris, E. Judica, A. Gallo, et al., Grey matter damage predicts the evolution of 545–556.
primary progressive multiple sclerosis at 5 years, Brain 129 (Pt 10) (Oct 2006) [58] F. Barkhof, J.A. Cohen, E.W. Radue, et al., Brain volume changes, on-study correla-
2628–2634. tions and the link to disability in three fingolimod phase 3 studies, Paper Presented
[45] P.G. Samann, M. Knop, E. Golgor, S. Messler, M. Czisch, F. Weber, Brain volume and at: ECTRIMS 2013; Copenhagen, Denmark, 2013.
diffusion markers as predictors of disability and short-term disease evolution in [59] R. Zivadinov, R. Bakshi, Central nervous system atrophy and clinical status in multi-
multiple sclerosis, AJNR Am. J. Neuroradiol. 33 (7) (Aug 2012) 1356–1362. ple sclerosis, J. Neuroimaging 14 (3 Suppl.) (Jul 2004) 27S–35S.
[46] J. Sastre-Garriga, G.T. Ingle, D.T. Chard, et al., Grey and white matter volume changes [60] D.S. Meier, H.L. Weiner, C.R. Guttmann, Time-series modeling of multiple sclerosis
in early primary progressive multiple sclerosis: a longitudinal study, Brain 128 disease activity: a promising window on disease progression and repair potential?
(Pt 6) (Jun 2005) 1454–1460. Neurotherapeutics 4 (3) (Jul 2007) 485–498.
[47] M. Summers, L. Fisniku, V. Anderson, D. Miller, L. Cipolotti, M. Ron, Cognitive impair- [61] C. Tortorella, A. Bellacosa, D. Paolicelli, et al., Age-related gadolinium-enhancement
ment in relapsing–remitting multiple sclerosis can be predicted by imaging per- of MRI brain lesions in multiple sclerosis, J. Neurol. Sci. 239 (1) (2005) 95–99.
formed several years earlier, Mult. Scler. 14 (2) (Mar 2008) 197–204.
[48] M. Tiberio, D.T. Chard, D.R. Altmann, et al., Gray and white matter volume changes in
early RRMS: a 2-year longitudinal study, Neurology 64 (6) (2005) 1001–1007.

Please cite this article as: T. Vollmer, et al., The natural history of brain volume loss among patients with multiple sclerosis: A systematic literature
review and meta-analysis, J Neurol Sci (2015), http://dx.doi.org/10.1016/j.jns.2015.07.014