J Neurol (2004) 251: 432-439
`DOI 10.1007/s00415-004-0349-8
`
`Andrea Paolillo
`Maria Cristina Piattella
`Patrizia Pantano
`Silvia Di Legge
`Francesca Caramia
`Pierluigi Russo
`Gian Luigi Lenzi
`Carlo Pozzilli
`
`The Relationship between inflammation
`and atrophy in clinically isolated
`syndromes suggestive of multiple sclerosis
`A monthly MRI study after
`triple-dose gadolinium-DTPA
`
`Abstract Objective To examine
`the relationship between inflam(cid:173)
`mation and brain atrophy in pa(cid:173)
`tients with a clinically isolated syn(cid:173)
`drome (CIS) suggestive of multiple
`sclerosis (MS). Methods Monthly
`triple-dose gadolinium
`(Gd/DTPA)-enhanced MRI scans
`over 6 months were obtained in 62
`consecutive CIS patients with an
`abnormal baseline MRI scan. Sub(cid:173)
`sequently MRI was performed at
`months 12 and 18. Patients who de-
`
`Received: 25 July 2003
`Received in revised form: 5 November 2003
`Accepted: 17 November 2003
`
`A. Paolillo, MD, PhD · M. C. Piattella, MD ·
`P. Pantano, MD · S. Di Legge, MD ·
`F. Caramia, MD· G.L. Lenzi, MD· C. Pozzilli,
`MD, PhD
`Dept. of Neurological Sciences
`University of Rome "La Sapienza"
`Rome, Italy
`
`P. Russo
`Dept. of Human Physiology
`and Pharmacology
`University of Rome "La Sapienza"
`Rome, Italy
`
`Carlo Pozzilli, MD, PhD (181)
`Dept. of Neurological Sciences
`2nd Faculty of Medicine
`University of Rome "La Sapienza"
`Viale dell'Universita, 30
`Rome, 00185, Italy
`Tel.: +39-6/49914716
`Fax: +39-6/4457705
`~ E-Mail: carlo.pozzilli@uniromal.it
`~
`z
`2
`
`veloped a clinically definite MS
`(i.e., a second clinical episode)
`ended the study at the time of the
`relapse. For each scan, the number
`and volume of newly active lesions
`(Gd-enhancement/new or newly
`enlarging T2 lesion that did not en(cid:173)
`hance), and the number and vol(cid:173)
`ume of T2 hyperintense lesions
`(T2-LL) and Tl-black holes (TI(cid:173)
`LL) were calculated. The percent(cid:173)
`age of brain volume changes
`(PBVC) was assessed using a fully
`automated technique (SIENA;
`Structural Image Evaluation using
`Normalization of Atrophy). Results
`Twenty-four (39%) developed clin(cid:173)
`ically definite MS by month 18.
`Thirty-eight (61 %) were relapse(cid:173)
`free and completed the MRI follow(cid:173)
`up. Relapse-free patients showed a
`progressive median increase be(cid:173)
`tween baseline and month 18 in
`Tl-LL (25 %, p < 0.001), but not in
`T2-LL (8.5 %, p = ns). PBVC de(cid:173)
`creased by 1.1%(p<0.001) in a
`time-dependent pattern (Kendall
`coefficient of concordance= 0.85).
`Exploratory subgroup analyses
`showed a trend towards progres(cid:173)
`sive decreases in brain volume in
`active patients (i.e., those with at
`least one newly active lesion during
`monthly MRI scanning; Spearman's
`R = -0.61; p < 0.001), but not among
`inactive patients (Spearman's
`R = -0.10; p = 0.53 ). Significant dif-
`
`f erences in median brain volume
`changes between the active and in(cid:173)
`active patient groups were found at
`months 12 and 18; the difference
`detected at month 6 was not signif(cid:173)
`icant. The cumulative number and
`volume of new Gd-enhancing le(cid:173)
`sions developed during the 6
`months of frequent MRI scanning
`were highly correlated with PBVC
`over the 18-month period (Spear(cid:173)
`man R values were 0.73 and 0.85,
`respectively). The strongest predic(cid:173)
`tor of PBVC at 18 months was the
`cumulative volume of newly active
`lesions during frequent MRI scan(cid:173)
`ning [B = -0. 83, standard error
`(SE)= 0.07, p < 0.001]. Conclusions
`This study shows that visible in(cid:173)
`flammation as detected by
`monthly, triple-dose Gd-enhanced
`MRI is an important factor in the
`pathogenesis of brain tissue loss in
`CIS patients. However, inflamma(cid:173)
`tion and brain atrophy do not pro(cid:173)
`ceed in parallel: atrophy
`appeared only after a delay of
`months following acute inflamma(cid:173)
`tion. Frequent MRI scanning allows
`for the detection of CIS patients
`who will develop brain atrophy in
`the short-term.
`
`Key words clinically isolated
`syndrome · MRI activity · brain
`atrophy · triple dose gadolinium
`DTPA · multiple sclerosis
`
`Page 1 of 8
`
`YEDA EXHIBIT NO. 2019
`MYLAN PHARM. v YEDA
`IPR2017-00195
`
`

`

`Introduction
`
`The assessment of tissue volume by serial magnetic res(cid:173)
`onance imaging (MRI) studies is currently being re(cid:173)
`garded as a new tool for monitoring disease progression
`in multiple sclerosis (MS) [17, 19-22, 31, 35, 42, 45, 46,
`52]. Loss of tissue volume (atrophy) in MS possibly orig(cid:173)
`inates from focal damage with loss of myelin and axons
`(lesions) and neurodegeneration ( wallerian degenera(cid:173)
`tion). A more generalized process including inflamma(cid:173)
`tion and edema, glial-cell proliferation or loss, and as(cid:173)
`trogliosis seems to be involved as well [14, 34].
`Recent studies have suggested that brain atrophy ap(cid:173)
`pears early in the course of MS [7, 32, 37, 40]. However,
`how early this process begins, and what is the relation(cid:173)
`ship between initial inflammation and the subsequent
`development of brain atrophy has not been established
`yet [21, 29, 30, 36, 41, 43, 45]. Few studies investigated
`brain atrophy in patients presenting with a clinically iso(cid:173)
`lated syndrome (CIS) suggestive of MS [4, 5, 9] and no
`studies have examined the relationship between brain
`atrophy with previous inflammatory activity as detected
`by frequent (monthly) MRI. Additionally, previous stud(cid:173)
`ies estimated brain atrophy in CIS patients by means of
`ventricular enlargement on two-dimensional (2D) MRI.
`Recent advance in computer-aided measurement tech(cid:173)
`niques have facilitated the development of new and reli(cid:173)
`able three-dimensional (3D) methods to quantify whole
`brain volume.
`The aim of the present study was to investigate the re(cid:173)
`lationship between acute inflammatory activity and the
`subsequent development of brain atrophy in CIS pa(cid:173)
`tients. We used a fully automated 3D technique for defin(cid:173)
`ing brain volume change between serially acquired im(cid:173)
`ages based on relative shifts in edge contour [ 4 7]. A
`triple dose of gadolinium (Gd)-DTPA was administered
`during each scan session to increase the sensitivity for
`detecting the new enhancing lesions across serial MRI
`evaluations [ 15, 16, 44].
`
`Patients and methods
`
`Sixty-two consecutive patients presenting with a CIS and referred to
`the MS Center of the University of Rome "La Sapienza" were enrolled
`in the study. Inclusion criteria were as follows: i) a single, clinical
`episode suggestive of MS; ii) positive baseline brain MRI scan ac(cid:173)
`cording to Fazekas' criteria [12]; iii) age between 18 and 50 years (the
`upper limit was set to reduce the likelihood of age-related non specific
`abnormalities on MRI); iv) no steroid treatment in the two months
`prior to the study entry; v) written informed consent. In all patients,
`the possibility of alternative diagnosis was taken into account, and ap(cid:173)
`propriate investigations were carried out as necessary.
`After a complete neurological examination, including rating of
`disability using the Expanded Disability Status Score (EDSS) [27] and
`baseline MRI scan, patients were subsequently examined and imaged
`with triple-dose Gd-enhanced MRI monthly for the first 6 months fol(cid:173)
`lowing the study entry. Subsequent MRI was performed at month 12
`and 18 of observation. For those patients who developed a clinically
`
`433
`
`definite MS (CDMS) [39], the study follow-up ended at the time of the
`relapse. A relapse was defined as the appearance of a new symptom
`or worsening of previous symptoms associated with significant
`changes in neurological signs lasting more than 24 h. These patients
`were given preventive treatment with disease-modifying therapies.
`
`MRI acquisition protocol
`
`MRI was performed with a 1.5 T magnet (Philips Gyroscan NT 15) in
`all patients. Proton density and T2WI conventional spin-echo (CSE)
`(TR= 2000 ms; TE= 20190 ms), fast fluid-attenuated inversion-recov(cid:173)
`ery (f-FLAIR) (TR= 6000 ms; TE= 150, TI= 1500 ms) and T 1 WI CSE
`(TR= 550 ms; TE= 12 ms) were acquired in the axial plane with 5-mm
`contiguous slices, and a field of view (FOV) = 240 mm and ma(cid:173)
`trix= 256 x 256. The Gd-enhanced Tl WI measurements were ob(cid:173)
`tained 5 minutes after injection of a triple dose of intravenous bolus
`(0.3 mmol/kg Gd-DTPA).
`
`MRI analysis
`
`The hardcopy was analysed by two experienced readers working in
`pairs, who marked and outlined on an overlaid transparency every
`detectable lesion on the proton density scan. All MRI scans were
`archived onto electronic media and transferred to a SUN workstation
`(Sun Microsystems, Palo Alto, CA). The following outcomes were as(cid:173)
`sessed in each patient: i) the number and volume of newly active le(cid:173)
`sions during months 1 to 6. A newly active lesion was defined as an
`area of new Gd-enhancement or as a new or newly enlarging T2 lesion
`that did not enhance; ii) the percentage change in lesion volume of T2
`hyperintensities (T2-LL) and Tl-hypointensities (Tl-LL) at month 6,
`12 and 18, compared with baseline. Focal lesions were measured us(cid:173)
`ing a semiautomated contouring technique (Dispim, Dispimage) [23,
`33]; iii) the percentage brain volume change (PBCV), as determined
`by calculating the difference in brain volume values at months 6, 12
`and 18, compared with baseline. Changes in brain volume were mea(cid:173)
`sured on Tl weighted images using the SIENA (Structural Image
`Evaluation using Normalization of Atrophy) program [ 47]. SIENA al(cid:173)
`lows for two-time-point (longitudinal) analyses of brain change. The
`program estimates PBCV using two input images taken from the
`same subject at different points in time; it uses a series of image analy(cid:173)
`sis programs to strip the non-brain tissue from the two images, regis(cid:173)
`ter the two brains under the constraint that the skulls are used to hold
`the scaling constant during the registration, and analyse the brain
`change between the two points. To evaluate the accuracy of SIENA, 15
`healthy subjects (10 females and 5 males) were scanned during two
`separate sessions, each with an inter-session interval of 1 week. The
`median absolute error was 0.18 %, which is in accordance with pub(cid:173)
`lished data on the scan-rescan reproducibility of this method [ 47].
`Baseline brain volume was also assessed by SIENAX, a program
`closely related to the SIENA longitudinal method. Instead of using
`images from two different time points, SIENAX attempts to estimate
`normalised brain volume from a single image, using the skull to nor(cid:173)
`malise spatially, with respect to a standard image [ 48].
`
`Statistical analysis
`
`The effect of continuous variables is expressed both by mean values
`± standard deviations ( SDs) and by median and interquartile ranges
`(that is, from 25th to 75th percentile values). Comparisons between two
`independent samples (that is, patients who showed a relapse during
`the 18-month study period versus those who did not, and patients
`who showed at least one enhancing lesion versus those who did not)
`were performed using t-tests, or Mann-Whitney U-tests, or both.
`Friedman's one-way analysis of variance was used to test the statisti(cid:173)
`cal significance of differences in the median percentage changes of
`T2-LL, Tl-LL and PBCV from baseline to months 6, 12 and 18. The
`
`Page 2 of 8
`
`YEDA EXHIBIT NO. 2019
`MYLAN PHARM. v YEDA
`IPR2017-00195
`
`

`

`434
`
`univariate relationships between brain MRI related measures were es(cid:173)
`timated using Spearman rank correlation coefficients.
`The strength of the relationship between predictors (clinical and
`MRI-related variables) and PBCV at month 18 was measured using
`standardized regression coefficients or ~ weights (B), their corre(cid:173)
`sponding 95 % confidence intervals (95 % Cis), and p-values obtained
`by multiple stepwise regression analyses (forward selection model).
`In the first step of this model, both clinical (age, sex, baseline EDSS
`value) and MRI-related variables (baseline Gd-LL, T2-LL, Tl-LL,
`brain volume, and cumulative volume of newly active lesions during
`the first 6 months of frequent MRI scanning) were included. In each
`subsequent step, the regression equation was composed of those pre(cid:173)
`dictors reaching specific thresholds of F- and p-values (for predictor
`inclusion: Fbl and pb0.05; for exclusion: F < 1 and p > 0.05). All sta(cid:173)
`tistical analyses were performed using STATISTICA version 6.0 (Stat(cid:173)
`soft s. r. l., Italy). Confidence interval calculations were performed by
`Confidence Interval Analysis (CIA) for Windows [ l].
`
`Results
`
`From a consecutive series of patients with CIS enrolled
`between August 1998 and June 2000, we considered, for
`the present study, 62 patients who had at least 18 months
`of follow-up. They were 24 males and 38 females, aged
`
`Table 1 Baseline characteristics of patients accord-
`ing to the development of CDMS
`
`Characteristics
`
`between 18 and 50 years, mean (SD) age was 30.3 years
`(7.4), and mean time between the clinical presentation
`and the inclusion in the study was 4.4 (0.9) months. The
`clinical onset involved the following systems: visual
`(n = 21), brainstem-cerebellar (n = 13), spinal cord
`(n = 13), and hemispheric (n = 15).
`Of the sixty-two patients included in the study, 24
`(39 %) developed CDMS by month 18, while 38 (61 %) re(cid:173)
`mained relapse-free. Table 1 shows demographic and
`baseline clinical and MRI characteristics of CDMS and
`relapse-free patients. Baseline Gd-LL was greater in pa(cid:173)
`tients who developed CDMS compared to those who
`were relapse-free at month 18. The presence of Gd-en(cid:173)
`hancing lesions at baseline predicted the development of
`CDMS, since 75 % of patients who reached this outcome
`had at least one enhancing lesion at baseline, compared
`to 45 % of those who did not develop CDMS [a 30 % dif(cid:173)
`ference (95% CI: from 50% to 5%), chi-square=5.5,
`p = 0.019]. Patients who did not develop CDMS by
`month 18 (n = 38) had follow-up MRI data at all time
`points and, thus, were considered for each following
`MRI analysis. Table 2 shows the cumulative number and
`
`Sex M/F (%)
`
`Age (years)
`Mean (SD)
`Median (interquartile range)
`
`Time since first attack (months)
`Mean (SD)
`Median (interquartile range)
`
`EDSS
`Mean (SD)
`Median (interquartile range)
`
`Number of Gd enhancing lesions
`Mean (SD)
`Median (interquartile range)
`
`Gd-LL (cm3)
`Mean (SD)
`Median (interquartile range)
`
`Number ofT2 hyperintense lesions
`Mean (SD)
`Median (interquartile range)
`
`T2-LL (cm 3)
`Mean (SD)
`Median (interquartile range)
`
`T1-LL (cm3)
`Mean (SD)
`Median (interquartile range)
`
`Brain volume (ml)
`Mean (SD)
`Median (interquartile range)
`
`Development of CDMS
`Yes (n = 24)
`
`41 %
`
`30.5 (5.9)
`30 (20-40)
`
`4.2 (0.9)
`4(2-10)
`
`1.4 (0.66)
`1.5 (1-2.5)
`
`2.0 (3.2)
`1 (0-2.5)
`
`No (n = 38)
`
`38%
`
`30.1 (6.2)
`30 (22-38)
`
`4.5 (0.8)
`4(3-9)
`
`1.3 (0.66)
`1.5 (1-2)
`
`0.9 (1.4)
`0 (0-1)
`
`0.5 (0.46)
`0.54 (0-1.5)
`
`0.3 (0.4)
`0.4 (0-1.8)
`
`32.1 (23.6)
`27 (15.8-40)
`
`27.6 (28.3)
`15 (10-30)
`
`6.2 (4.7)
`4.7 (3.4-7.6)
`
`5.6 (4.9)
`4.1 (3.1-6.8)
`
`0.96 (1.1)
`0.81 (0.5-1.4)
`
`0.87 (1.2)
`0.76 (0.4-1.2)
`
`1563.4 (56.2)
`1566 (1512-1608)
`
`1571 (76.1)
`1572 (1546-1616)
`
`P-value
`
`0.76
`
`0.80
`0.88
`
`0.17
`0.30
`
`0.50
`0.56
`
`0.04
`0.05
`
`0.07
`0.03
`
`0.51
`0.07
`
`0.54
`0.63
`
`0.10
`0.22
`
`0.12
`0.18
`
`CDMS clinically definite multiple sclerosis; SD standard deviation; Gd-LL baseline Gd-enhancing lesion load; T2-LL
`baselineT2-hyperintense lesion load; Tl-LL baseline Tl hypo intense lesion load
`
`Page 3 of 8
`
`YEDA EXHIBIT NO. 2019
`MYLAN PHARM. v YEDA
`IPR2017-00195
`
`

`

`volume of new Gd-enhancing lesions, new or newly en(cid:173)
`larging T2 lesions and newly active lesions during the
`first six monthly MRI evaluations. Table 3 shows the
`changes in T2-LL, Tl-LL and brain volume at months 6,
`12 and 18, compared with baseline scan. Patients showed
`progressive median increases from baseline to month 18
`in Tl-LL (25 %, p < 0.001), but not in T2-LL (8.5 %,
`p = ns). The median percent change in brain volume at
`month 18 was -1.1%(p<0.001), with a time-dependent
`pattern (Kendall coefficient of concordance= 0.85). Ex(cid:173)
`ploratory subgroup analyses were performed by strati(cid:173)
`fying patients into those without (inactive patients,
`n = 13) and those with at least one newly active lesion
`during the first six monthly MRI scans (active patients,
`n = 25) (Fig 1). A significant trend towards progressive
`decreases
`in brain volume (Spearman's R = -0.61;
`p < 0.001) was found in active patients; however, no such
`significant trend was noted in inactive patients (Spear(cid:173)
`man's R = -0.10; p = 0.53). In addition, significant differ(cid:173)
`ences in median brain volume changes between the ac(cid:173)
`tive and inactive patient groups were found at month 12
`[-0.47% (95% Cl: from -0.89% to -0.15%), p=0.009]
`and at month 18 [-1.23% (95%CI: from -1.71 % to
`-0.82 %), p < 0.001 ].
`Finally, significant differences between active and in(cid:173)
`active patients on median T2-LL changes [16.3% (95%
`CI: from 0.8% to 37.7%),p=0.037] and Tl-LL changes
`[22.2% (95%CI: from 5% to 38.5%), p=0.013] were
`found at month 18, but not at months 6 and 12.
`
`435
`
`Month 6
`
`Month 12
`
`Month 18
`
`-0.23
`
`- 0.32
`
`- 0.48
`
`- 0.85
`
`- 1.71
`
`D Active patien ts
`CJ Inac tive pat ients
`
`- 0.60
`
`...
`..
`g:i -0.40
`..c::
`e -= - 0.80
`v
`~ .s -l.00
`..
`li - 1.20
`'#-
`~ -1 .4 0
`;a
`~ -1.60
`
`- 1.80
`
`- 2.00
`
`Fig. 1 Median % brain volume change in patients without (inactive patients,
`n = 13) and in those with at least one new active lesion {active patients, n = 25)
`during the first six monthly MRI scans.
`A significant trend towards progressive decreases in brain volume {Spearman's
`R = -0.61; p < 0.001) was found in active patients; no significant trend was found
`in inactive patients {Spearman's R = -0.1 O; p = 0.53). Significant differences in
`median brain volume changes between the active and inactive patient groups were
`found at month 12 [-0.47 % {95 % Cl: from -0.89 % to -0.15 %), p = 0.009] and
`at month 18 [- 1.23 % {95 % Cl: from -1.71 % to -0.82 %), p < 0.001]
`
`Predictors of brain atrophy
`
`Univariate analysis revealed a significant relationship
`(p < 0.01) between PBVC over the 18-month period and
`the cumulative number and volume of new Gd-enhanc(cid:173)
`ing lesions that developed during the 6 months of fre(cid:173)
`quent MRI scanning (Spearman R values were 0.73 and
`
`Table 2 Evaluation of monthly scans during the first 6 months in patients without CDMS {n = 38)
`
`Lesion type
`
`Number of lesions
`
`New Gd-enhancing lesions
`
`New or newly enlarging T2 lesions
`
`Newly active lesions
`
`Mean (SD)
`Median (interquartile range)
`
`Mean (SD)
`Median (interquartile range)
`
`Mean (SD)
`Median (interquartile range)
`
`4.8 (6.47)
`2 (0-7)
`
`2.2 (2.90)
`1.5 {0-3)
`
`6.9 (7.83)
`4.5 (0-14)
`
`CDMS clinically definite multiple sclerosis; SD standard deviation
`
`Volume of lesions {ml)
`
`Mean (SD)
`Median (interquartile range)
`
`Mean {SD)
`Median (interquartile range)
`
`Mean (SD)
`Median (interquartile range)
`
`1.0(1.94)
`0.19(0-1)
`
`0.2 (0.26)
`0.114 (0-0.3)
`
`1.2 (1.99)
`0.43 (0-2)
`
`Table 3 Median percentage changes in T2-LL, T1-LL
`and brain volume during the 18-month study period
`in patients without CDMS (n = 38)
`
`Percent change
`from baseline
`
`T2-LL
`Median
`(interquartile range)
`
`T1-LL
`Median
`(interquantile range)
`
`Brain volume
`Median
`(interquantile range)
`
`Month 6
`
`Month 12
`
`Month 18
`
`P-value
`
`0.66
`(from -12.0to19.9)
`
`5.0
`(from -12.4 to 25.5)
`
`8.5
`(from -6.4 to 31.9)
`
`0.549
`
`12.5
`(from 2.1 to 25.6)
`
`20.3
`(from 4.1 to 46.0)
`
`25.1
`(from 10.9 to 56.1)
`
`< 0.001
`
`-0.16
`(from -0.32 to -0.06)
`
`-0.58
`from -1.02 to -0.24)
`
`-1.1
`(from -1.91 to -0.67)
`
`< 0.001
`
`T2-LL T2-hyperintense lesion load; Tl-LL T1 -hypointense lesion load; CDMS clinically definite multiple sclerosis
`
`Page 4 of 8
`
`YEDA EXHIBIT NO. 2019
`MYLAN PHARM. v YEDA
`IPR2017-00195
`
`

`

`R = -0.75
`p < 0.01
`
`0
`
`0
`
`0
`
`0
`
`0
`
`0
`
`0
`
`'-Q
`0
`
`0
`
`0
`
`0
`
`0
`
`0
`
`10
`
`IS
`
`20
`
`25
`
`30
`
`35
`
`Number of newly active lesions
`
`standard error (SE)= 0.07, p < 0.001]. All other baseline
`clinical/demographic and MRI-related variables did not
`reach statistical significance.
`Significant predictors of earlier brain volume
`changes were also assessed by including PBVC from
`baseline to month 12 as the dependent variable in the re(cid:173)
`gression model. Similar to the findings of the previous
`model, the volume of newly active lesions [~ = -0.57,
`SE= 0.19, p = 0.005] was the only significant predictor of
`brain volume change [multiple R2 = 0.40 for the model,
`F [8, 29]= 2.4; p < 0.041].
`In line with previous hypotheses, a further analysis
`was performed using the PBVC from baseline to month
`6 as the dependent variable in the regression model. No
`clinical or MRI-related variables, including the cumula(cid:173)
`tive volume of newly active lesions during the first 6
`months, reached the threshold values for inclusion in the
`regression equation.
`
`R = - 0.85
`p < 0.01
`
`Discussion
`
`436
`
`a)
`
`0.5
`
`0.0
`
`-0.5
`
`- 1.0
`
`-1.5
`
`-2.0
`
`-2.5
`
`l
`..c:
`c:
`~
`;;;
`u
`~
`
`- 3.0
`-5
`
`b)
`
`00
`
`..c: ;:: s
`"' ~
`~
`
`0.5
`
`0.0
`
`-0.5
`
`-1.0
`
`- 1.5
`
`-2.0
`
`-2.5 1
`
`§
`8
`
`0
`
`0
`
`r:
`
`This study indicates that the development of brain atro(cid:173)
`phy in CIS patients is influenced by previously detected
`inflammatory activity in the brain. We found a strong re(cid:173)
`lationship between the cumulative volume of newly ac(cid:173)
`tive lesions measured on monthly MRI over 6 months
`and the development of brain atrophy over the study pe(cid:173)
`riod. Acute inflammation and brain atrophy do not pro(cid:173)
`ceed in parallel. Rather, atrophy appears only after a de(cid:173)
`lay of months following inflammation. This observation
`was addressed by the multiple stepwise regression
`analyses (which showed no relationships between the
`cumulative volume of newly active lesions during the
`first 6 months and PBVC from baseline to month 6), as
`well as by stratification of subjects as either active or in(cid:173)
`active. No significant difference in brain atrophy be(cid:173)
`tween active and inactive patients was found during the
`first 6 months, whilst a clear difference emerged in the
`subsequent 6 months and, particularly, in the final 6
`months. The decrease in brain volume over the 18-
`month period was three to four times greater in CIS pa(cid:173)
`tients with newly active lesions than in those without ac(cid:173)
`tive inflammatory lesions (-1.71 % vs -0.48%). Our
`study, however, cannot completely rule out the hypothe(cid:173)
`sis that brain volume loss is the consequence only of ab(cid:173)
`normalities observed during the 6-month of frequent
`MRI interval. Rather, this loss may have resulted from
`the constant accumulation of subclinical damage that
`occurred over the entire study period. The short-term
`changes (1-2 years) in the brain volume of our CIS pa(cid:173)
`tients with acute inflammation are of the same magni(cid:173)
`tude as those observed in previous studies on the nat(cid:173)
`ural history of MS, showing a rate of brain atrophy
`varying from -0.7% to -1.8% [17, 19, 20, 22, 24, 31, 35,
`41-43, 45, 46]. In contrast, the brain volume changes of
`
`-3.0 ~-- -----------~--....__.
`-0.S
`0.0
`0.5
`2.0
`2.5
`3.0
`3.5
`L.O
`l.5
`4.0
`
`Volume of newly active lesions
`
`Fig. 2 Correlation between percentage brain volume change (PBVC) over the
`18-month period and number (a) and volume (b) of newly active lesions during the
`first 6 months
`
`0.85, respectively). With regard to the other lesion mea(cid:173)
`sures obtained during the monthly MRI scans, signifi(cid:173)
`cant correlations between PBVC and the cumulative
`number (r = -0.63) and volume (r = -0.65) of new or
`newly enlarging T2 lesions, and with the cumulative
`number (r = -0.75) and volume (r = -0.85) of newly ac(cid:173)
`tive lesions were also found (Fig. 2).
`The forward stepwise regression model was set to
`identify significant predictors of PBVC at month 18.
`Since MRI-related measures addressing volumetric di(cid:173)
`mensions provided a more accurate evaluation than the
`number of lesions, they were preferred as potential pre(cid:173)
`dictors in the model. After two steps, the model ex(cid:173)
`plained 69 % of the variance in PBVC data at month 18
`[F (1, 36]= 80; p < 0.001]. The only significant predictor
`of PBVC at 18 months was the cumulative volume of
`newly active lesions during the first 6 months [B = -0.83,
`
`Page 5 of 8
`
`YEDA EXHIBIT NO. 2019
`MYLAN PHARM. v YEDA
`IPR2017-00195
`
`

`

`CIS patients without MRI activity are comparable to
`those reported in studies of healthy control subjects,
`varying from 0.1 % to 0.3 % per year [8, 20, 38].
`Although the correlation between acute inflamma(cid:173)
`tory activity (as measured by lesion enhancement) and
`the subsequent development of brain atrophy has been
`investigated in various MS subtypes [21, 29, 30, 32, 36, 43,
`45], few studies have been performed on CIS patients
`and, of those that have been performed, none used fre(cid:173)
`quent (monthly) MRI [4, 5, 9].
`Dalton et al. [9] studied ventricular enlargement over
`1 year in 55 CIS patients. They reported a more substan(cid:173)
`tial development of ventricular enlargement in a small
`group of patients with persistent enhancing lesions.
`However, while enhancing lesions contributed to atro(cid:173)
`phy, the magnitude of the correlation was modest
`(r = 0.38), suggesting that other mechanisms (i.e., a
`more diffused process involving normal-appearing tis(cid:173)
`sue) may also play a role in the development of atrophy.
`In this regard, recent MR studies using more advanced
`techniques (i. e., spectroscopic imaging, diffusion tensor
`imaging, and magnetization transfer ratio) indicate that
`the development of brain atrophy depends more on the
`extent of damage in normal-appearing tissue than on fo(cid:173)
`cal lesion damage [ 11, 25].
`Compared with the study by Dalton etal. [9], a
`stronger relationship between brain atrophy and prior
`inflammation was observed in our cohort (r = 0.85). Sev(cid:173)
`eral factors may contribute to the latter. First, this study
`supports the concept that serial MRI examinations at
`frequent intervals play a crucial role in finding a
`stronger relationship between brain atrophy and in(cid:173)
`flammation. For example, in previous studies in which
`patients were imaged monthly, the number of enhancing
`lesions correlated well with increases in ventricular size
`[32, 29]. However, in cohorts that were imaged only once
`per year, [32, 43] enhancing lesions were not predictive
`of cerebral atrophy. Furthermore, the results of our
`study indicate that the number and volume of enhanc(cid:173)
`ing lesions at baseline are not predictive of brain atrophy
`development.
`Secondly, the strong relationship between active in(cid:173)
`flammation and brain atrophy observed in this study
`could be explained by the use of a triple-dose of Gd(cid:173)
`DTPA. Serial administration of triple-dose Gd allows for
`the identification of more active lesions/patients, partic(cid:173)
`ularly in the subgroup of patients whose blood-brain
`barrier permeability to Gd may be low and whose leak(cid:173)
`age space is small [16].
`Thirdly, the method used to measure brain atrophy in
`previous studies on CIS patients [ 4, 9], albeit method(cid:173)
`ologically simple, has some limitations. For instance,
`whether changes in ventricular structures are represen(cid:173)
`tative of whole brain changes has not yet been clearly de(cid:173)
`fined. Changes in ventricular width are mostly deter(cid:173)
`mined by lesions located periventricularly. Therefore,
`
`437
`
`the role played by cortical-subcortical enhancing lesions
`in determining brain tissue loss may be underestimated.
`The method used in the present analysis is reliable and
`accurate, with a brain volume change error of 0.18 %.
`The major advantageous features of this method are its
`robustness (it is possible to apply this technique to the
`analysis of images acquired with a range of pulse se(cid:173)
`quences and it is relatively insensitive to differences in
`slice thickness), and its sensitivity to subtle parenchyma
`changes [47,48].
`Finally, measurement of the cumulative volume of
`newly active lesions may have contributed to the
`stronger relationship between brain atrophy and prior
`inflammation noted in our study [28, 49]. As previously
`suggested, the volume of new T2 lesions reflects both
`disease activity (similar to Gd enhancement) and dis(cid:173)
`ease progression (net T2 lesion volume), and consists
`more of enlarging pre-existing lesions with low levels of
`inflammatory activity than of new discrete lesions [ 49].
`Interestingly, Stevenson et al. [ 49] reported a correlation
`between total new T2 lesion volume and the progression
`of cerebral atrophy, suggesting that the total new T2 le(cid:173)
`sion volume may, ultimately, be the most useful measure
`of disease activity in MS.
`Our results confirm previous studies showing that
`the presence of Gd-enhancing lesions on baseline scans
`predict the development of CDMS over a relatively short
`follow-up period (i.e., 18 months) [2, 3, 6, 9]. The pres(cid:173)
`ence of enhancement on MRI scans represents a selec(cid:173)
`tion criterion for identifying patients with very active
`disease who are more likely to have a clinical eloquent
`lesion than patients with no inflammatory activity.
`This study also supports the notion that a significant
`number of newly active lesions (as assessed through fre(cid:173)
`quent MRI scanning) accrue even in patients who do not
`develop CDMS. Although clinically silent, this underly(cid:173)
`ing process leads to irreversible axonal damage, as re(cid:173)
`flected by the significant development of brain atrophy
`and increases in Tl black holes over the 18-month pe(cid:173)
`riod. This issue has recently been addressed in quantita(cid:173)
`tive MRI studies that showed early, progressive brain at(cid:173)
`rophy and widespread axonal damage in patients with
`CIS [ 4, 10, 18]. This observation is also consistent with
`post-mortem data showing that acute axonal damage is
`extensive in the early stages of the disease and correlates
`with inflammation [ 13, 26, 51].
`Compared with the significant Tl black holes and
`brain atrophy that we observed in this cohort of CIS pa(cid:173)
`tients, the slight increase of T2 lesion load ( 8.5 % ) is
`likely to be the result of great variability of T2 measure(cid:173)
`ment within and between subjects. Substantial month to
`month fluctuations in T2 lesion load have been identi(cid:173)
`fied [SO] and they could reflect a range of pathological
`changes from acute inflammation to irreversible tissue
`damage.
`In conclusion, this study shows that visible inflam-
`
`Page 6 of 8
`
`YEDA EXHIBIT NO. 2019
`MYLAN PHARM. v YEDA
`IPR2017-00195
`
`

`

`438
`
`mation, as detected by monthly, triple-dose Gd-en(cid:173)
`hanced MRI, is a significant factor in the pathogenesis of
`whole brain atrophy in CIS patients. The results of this
`
`study support early anti-inflammatory treatment to
`prevent important injury in CIS patients with ongoing
`inflammatory activity.
`
`References
`
`1. Altman DG, Machin D, Bryant TN,
`Gardner MJ (2000) Statistics with con-
`fidence. BMJ Books, Bristol England
`2. Barkhof F, Filippi M, Miller DH, Schel-
`tens P, Campi A, Polman CH, Cami G,
`Ader HJ, Losseff N, Valk J (1997) Com-
`parison of MRI criteria at first presen-
`tation to predict conversion to clini-
`cally definite multiple sclerosis. Brain
`120:2059-2069
`3. Brex PA, O' Riordan JI, Miskiel KA,
`Moseley IF, Thompson AJ, Plant GT,
`Miller DH (1999) Multisequence MRI
`in clinically isolated syndromes and
`the early development of MS. Neural-
`ogy 53:1184-1190
`4. Brex PA, Jenkins R, Fox NC, Crum WR,
`O'Riordan JI, Plant GT, Miller DH
`(2000) Detection of ventricular en-
`largement in patients at the earliest
`clinical stage of MS. Neurology 541:
`689-691
`5. Brex PA, Leary SM, O'Riordan JI,
`Miszkiel KA, Plant GT, Thompson AJ,
`Miller DH (2001) Measurement of
`spinal cord area in clinically isolated
`syndrome suggestive of MS. J Neural
`Neurosurg Psychiatry 70:544-547
`6. CHAMPS Study Group (2002) MRI
`predictors of early conversion to clini-
`cally definite MS in the CHAMPS
`placebo group. Neurology 59:998-1005
`7. Chard DT, Griffin CM, Parker GJ,
`Kapoor R, Thompson AJ, Miller DH
`(2002) Brain atrophy in clinically ear