Beneficial Plasma Exchange Response in Central Nervous System Inflammatory Demyelination

ORIGINAL CONTRIBUTION

ONLINE FIRST

Beneficial Plasma Exchange Response in Central Nervous System Inflammatory Demyelination Setty M. Magan˜a, BS; B. Mark Keegan, MD; Brian G. Weinshenker, MD; Bradley J. Erickson, MD, PhD; Sean J. Pittock, MD; Vanda A. Lennon, MD, PhD; Moses Rodriguez, MD; Kristine Thomsen, BA; Stephen Weigand, MS; Jay Mandrekar, PhD; Linda Linbo, RN; Claudia F. Lucchinetti, MD

Background: Plasma exchange (PLEX) is a beneficial rescue therapy for acute, steroid-refractory central nervous system inflammatory demyelinating disease (CNSIDD). Despite the approximately 45% PLEX response rate reported among patients with CNS-IDD, determinants of interindividual differences in PLEX response are not well characterized. Objective: To perform an exploratory analysis of clinical, radiographic, and serological features associated with beneficial PLEX response. Design: Historical cohort study. Setting: Neurology practice, Mayo Clinic College of

Medicine, Rochester, Minnesota. Patients: All Mayo Clinic patients treated with PLEX between January 5, 1999, and November 12, 2007, for a steroid-refractory CNS-IDD attack. Main Outcome Measure: The PLEX response in attackrelated, targeted neurological deficit(s) assessed within the 6-month period following PLEX.

Author Affiliations: Departments of Neurology (Mss Magan˜a and Linbo and Drs Keegan, Weinshenker, Pittock, Lennon, Rodriguez, and Lucchinetti), Radiology (Dr Erickson), Laboratory Medicine and Pathology (Drs Pittock and Lennon), Immunology (Dr Lennon), and Health Sciences Research (Ms Thomsen, Mr Weigand, and Dr Mandrekar), Mayo Clinic College of Medicine, Rochester, Minnesota.

C

Results: We identified 153 patients treated with PLEX for a steroid-refractory CNS-IDD, of whom 90 (59%) exhibited moderate to marked functional neurological improvement within 6 months following treatment. Pre-PLEX clinical features associated with a beneficial PLEX response were shorter disease duration (P=.02) and preserved deep tendon reflexes (P=.001); post-PLEX variables included a diagnosis of relapsing-remitting multiple sclerosis (P=.008) and a lower Expanded Disability Status Scale score (P⬍.001) at last follow-up. Plasma exchange was less effective for patients with multiple sclerosis who subsequently developed a progressive disease course (P=.046). Radiographic features associated with a beneficial PLEX response were presence of ring-enhancing lesions (odds ratio=4.00; P=.03) and/or mass effect (odds ratio=3.00; P=.02). No association was found between neuromyelitis optica–IgG serostatus and PLEX response. Conclusions: We have identified clinical and radiographic features that may aid in identifying patients with fulminant, steroid-refractory CNS-IDD attacks who are more likely to respond to PLEX.

Arch Neurol. 2011;68(7):870-878. Published online March 14, 2011. doi:10.1001/archneurol.2011.34

ENTRAL NERVOUS SYSTEM

inflammatory demyelinating diseases (CNSIDDs), of which multiple sclerosis (MS) is the prototype, represent a spectrum of disorders that differ in their clinical, radiographic, and serological findings. Although most acute attacks resolve spontaneously or are corticosteroid responsive, approximately 5% of CNS-IDD attacks are refractory to steroids, necessitating rescue therapy. We demonstrated the efficacy of plasma exchange (PLEX) in a randomized, sham-controlled, doublemasked clinical trial; other nonrandomized studies, both prospective and retrospective, have consistently supported the efficacy of PLEX expressly in the setting ARCH NEUROL / VOL 68 (NO. 7), JULY 2011 870

of steroid-refractory, fulminant CNSIDD attacks.1-10 Male sex, preserved reflexes, and early initiation of PLEX were associated with beneficial PLEX response.2,3 The efficacy of PLEX is likely due to the nonspecific removal of inflammatory and humoral factors that might contribute to reversible functional conduction block, before permanent axonal or neuronal damage has ensued.11-14 Immunopathological studies have demonstrated prominent humoral immunopathogenic mechanisms that may be responsive to PLEX in a subset of CNS-IDD cases, namely MS immunopattern II and neuromyelitis optica (NMO).15,16 Clinical, radiographic, and serological factors that reliably differentiate immunopathological patterns or disease mecha-

WWW.ARCHNEUROL.COM

Downloaded from www.archneurol.com at HINARI, on July 11, 2011 ©2011 American Medical Association. All rights reserved.

212 Confirmed PLEX for fulminant, steroid-refractory CNS-IDD attack

Inclusion criteria PLEX for fulminant steroidrefractory CNS-IDD attack 0- to 3-mo pre-PLEX MRI study and/or serum sample available Sufficient documentation regarding PLEX response

59 Exclusions 9 Insufficient documentation regarding PLEX response 46 No pre-PLEX MRI study or serum sample 4 Other†

153 PLEX + CNS-IDD and sufficient documentation

Subjects Available for Each Analysis, No. Analysis Clinical features Brain MRI Spine MRI Serology aResults

All Patients

Opticospinal

Brain

153a 114 85a 34

63 36 49 34a

90 78a 36 Not applicable

are reported separately for this group.

Figure 1. Ascertainment of plasma exchange (PLEX) cohort and sample size for each analysis. †Exclusions included receiving fewer than 2 exchanges (n=2) and receiving PLEX for an attack related to a coexisting condition and not for a central nervous system inflammatory demyelinating disease (CNS-IDD) attack (diabetic neuropathy, n = 1; polycythemia vera, n=1). MRI indicates magnetic resonance imaging.

nisms remain elusive. Despite the approximately 45% PLEX response rate reported among patients with CNSIDD, determinants of interindividual differences in PLEX response as well as the mechanism(s) underlying the rapid, all-or-none response are not well characterized. This study represents the largest clinical-radiographic-serological series to date of acute, steroid-refractory attacks of CNSIDD treated with PLEX and identifies clinical and radiographic factors associated with beneficial PLEX response. These findings may identify patients more likely to exhibit a favorable PLEX response and provide further insights into the underlying pathobiology of steroidrefractory CNS-IDD.

Apheresis Unit. We identified 212 patients, of whom 59 did not meet inclusion criteria and were excluded (Figure 1). The remaining 153 patients met inclusion criteria and were assigned to the following cohorts for subgroup analysis: clinical cohort (n=153), brain attack cohort (n=78), opticospinal attack cohort (n=85), and serology cohort (n=34). The clinical cohort consisted of all patients with sufficient documentation to ascertain PLEX response. In addition, patients with 0- to 3-month pre-PLEX brain and/or spine MRI available for analysis composed the brain attack and opticospinal attack cohorts, respectively. Owing to the established disease specificity of NMOIgG, the serological cohort was restricted exclusively to patients with NMO spectrum disorders with 0- to 3-month pre-PLEX serum samples available for NMO-IgG analysis.17

CLINICAL DATA

METHODS This study, approved by Mayo Clinic Institutional Review Board 120-06, analyzed clinical, radiographic, and serological findings among patients treated with PLEX for steroid-refractory CNS-IDD attacks between January 5, 1999, and November 12, 2007. Inclusion criteria were the following: (1) CNS-IDD; (2) PLEX for a steroid-refractory CNS-IDD attack; (3) at least one 0- to 3-month pre-PLEX magnetic resonance imaging (MRI) study and/or serum sample; and (4) sufficient clinical documentation regarding PLEX response. Exclusion criteria included the following: (1) clinical, radiographic, or pathological findings suggesting diseases other than CNS-IDD; and (2) receiving fewer than 2 PLEX treatments.

PATIENT ASCERTAINMENT All Mayo Clinic patients receiving PLEX between January 5, 1999, and November 12, 2007, for a CNS-IDD were identified. For the period prior to June 1, 2000, patients were ascertained from our previously reported PLEX cohort (n=37).3 For the period of December 1, 2005, to December 31, 2007, patients were prospectively ascertained from the Mayo Clinic

Medical record review provided sufficient clinical documentation for study inclusion in 153 patients. The clinical course was classified as monophasic, relapsing-remitting, secondary progressive, or primary progressive.18 Diagnoses were based on published criteria: probable or definite MS,19,20 NMO spectrum disorders (ie, NMO, longitudinally extensive transverse myelitis [ⱖ3 vertebral segments], and recurrent optic neuritis),21 monophasic optic neuritis, acute disseminated encephalomyelitis,22 short transverse myelitis (⬍3 vertebral segments),23 and clinically isolated syndrome24 in patients with a single neurological episode at last follow-up. The index attack was defined as the constellation of neurological symptoms leading to initiation of PLEX. Disease duration at PLEX was defined as the time from the patient’s first-ever documented neurological deficits to the time of PLEX initiation; disease duration at last follow-up was defined as the interval between the patient’s first-ever documented neurological deficits and his or her last documented follow-up.

RADIOGRAPHIC MATERIAL Analysis by MRI, performed by 1 evaluator blinded to clinical data (S.M.M.), was restricted to T1-weighted, T2-weighted, fluid-

ARCH NEUROL / VOL 68 (NO. 7), JULY 2011 871

WWW.ARCHNEUROL.COM

Downloaded from www.archneurol.com at HINARI, on July 11, 2011 ©2011 American Medical Association. All rights reserved.

Table 1. Clinical Spectrum of Plasma Exchange Cohort at Time of Plasma Exchange All Subjects (N=153)

PLEX Responders (n = 90)

PLEX Nonresponders (n = 63)

1.4 y (18 d to 38.5 y)

1.1 y (18 d to 32.9 y)

2.3 y (29 d to 38.5 y)

Clinical Feature Disease duration at PLEX, median (range) a Clinical course prior to PLEX, No. (%) Monophasic Relapsing Progressive b Diagnosis at PLEX, No. (%) Definite MS Probable MS NMO LETM Recurrent ON TM ADEM CIS EDSS score at index attack, median (IQR)

43 (28) 104 (68) 5 (3) 55 (36) 18 (12) 26 (17) 36 (24) 1 (1) 9 (6) 3 (2) 5 (3) 8 (6.5-8.5)

PLEX Response Rate, %

27 (30) 60 (67) 2 (2)

16 (25) 44 (70) 3 (5)

63 58 40

34 (38) 10 (11) 11 (12) 25 (28) 0 6 (7) 0 4 (4) 8 (6.5-8.5)

21 (33) 8 (13) 15 (24) 11 (17) 1 (2) 3 (5) 3 (5) 1 (2) 8 (6-8.5)

62 56 42 69 0 67 0 80

Abbreviations: ADEM, acute disseminated encephalomyelitis; CIS, clinically isolated syndrome; EDSS, Expanded Disability Status Scale; IQR, interquartile range; LETM, longitudinally extensive transverse myelitis; MS, multiple sclerosis; NMO, neuromyelitis optica; ON, optic neuritis; PLEX, plasma exchange; TM, transverse myelitis. a P ⱕ .05. b Four patients had a progressive course with superimposed attacks; 1 patient had a progressive course without superimposed attacks.

attenuated inversion recovery, and T1-weighted gadoliniumenhanced sequences. When more than 1 lesion was observed, the largest lesion was characterized independently and analyzed for the same radiographic features. The radiographic features of interest were initially defined by consensus among study investigators (S.M.M, B.J.E., C.F.L.) during training sessions prior to data collection. Questions regarding radiographic interpretations were adjudicated by a certified neuroradiologist (B.J.E.). Radiographic features of interest were lesion location, T2-weighted lesion load, T2-weighted lesion size, presence and grade of mass effect, presence and degree of edema, and presence of a T2weighted hypointense rim. Enhancement patterns were defined as homogeneous, ringlike, or heterogeneous. Complex lesions had more than 1 enhancement pattern. The MRI studies were evaluated for multifocality and fulfillment of Barkhof criteria.25

SEROLOGICAL MATERIAL

graphic features, and NMO-IgG serostatus. The unit of analysis was based on individual patients. Descriptive statistics, including medians, ranges, and interquartile ranges (IQRs), were used to summarize continuous variables. Comparisons between categorical and continuous variables were performed using Fisher exact test and Wilcoxon rank sum test, respectively. Logistic regression modeling was used for secondary analysis to further summarize several univariate associations. We report unadjusted odds ratios (ORs) and ORs adjusted for sex, days to PLEX, and preservation of deep tendon reflexes, which are 3 background covariates that could a priori be considered potential confounders. Our analysis looked at a large number of potential predictors of a beneficial PLEX response. Although increasing the number of tests could increase our chance of 1 or more false-positive findings, unadjusted (or percomparison) P values are reported because single-inference questions were asked and multiple-comparison tests would greatly inflate false-negative errors in this context.26-29

Pre-PLEX serum samples were tested for NMO-IgG (by indirect immunofluorescence assay)17 by 2 independent and blinded observers (S.J.P. and V.A.L.). Samples were scored as positive (titer ⱖ120) or negative (negative at a titer ⬍120).

RESULTS

PRIMARY OUTCOME

PATIENT DEMOGRAPHIC AND CLINICAL CHARACTERISTICS

The primary outcome was PLEX response as defined in our randomized trial.2 Improvement in attack-related, targeted neurological deficit(s) was graded as the following: no improvement if there was no gain in neurological function; mild improvement if there was improvement in neurological status without impacting function; moderate improvement if there was definite improvement in function; and marked improvement if there was major functional improvement. Treatment success requiring moderate or marked functional improvement in at least 1 targeted neurological deficit was assessed in the 6-month period following PLEX treatment.

STATISTICAL ANALYSIS We performed an exploratory statistical analysis to evaluate associations between PLEX response and clinical factors, radio-

One hundred four patients were female (68%). The median age at disease onset was 38 years (IQR, 28-50 years; range, 7-75 years), with the median age at the time of PLEX being 44 years (IQR, 34-55 years; range, 6-76 years). Table 1 and Table 2 summarize PLEX response rate by pre- and post-PLEX clinical features, respectively. The median disease duration at the time of PLEX was 1.4 years (range, 18 days to 38.5 years). The most common diagnoses at the time of PLEX were definite or probable MS (n=73 [48%]), longitudinally extensive transverse myelitis (n=36 [24%]), and NMO (n=26 [17%]). At last follow-up, the most common diagnoses were definite or probable MS (n=73 [49%]), NMO (n=34 [23%]), and longitudinally extensive transverse myelitis (n=27 [18%]).

ARCH NEUROL / VOL 68 (NO. 7), JULY 2011 872

WWW.ARCHNEUROL.COM

Downloaded from www.archneurol.com at HINARI, on July 11, 2011 ©2011 American Medical Association. All rights reserved.

Table 2. Clinical Spectrum of Plasma Exchange Cohort at Time of Last Follow-up

Clinical Feature Disease duration at last follow-up, median (range) Clinical course at last follow-up, No. (%) a Monophasic Relapsing Progressive b Diagnosis at last follow-up, No. % Definite MS b Probable MS NMO LETM Recurrent ON TM ADEM CIS EDSS score at last follow-up, median (IQR) c

All Subjects (N = 153)

PLEX Responders (n = 90)

PLEX Nonresponders (n = 63)

5.2 y (83 d to 39 y)

4.8 y (83 d to 38 y)

6.4 y (109 d to 39 y)

PLEX Response Rate, %

49 (35) 83 (58) 10 (7)

30 (34) 54 (62) 3 (3)

19 (35) 29 (53) 7 (13)

61 65 30

64 (43) 9 (6) 34 (23) 27 (18) 1 (1) 9 (6) 2 (1) 4 (3) 6.5 (3.5-8)

42 (47) 4 (4) 18 (20) 17 (19) 0 6 (7) 0 2 (2) 4 (2-7)

22 (36) 5 (8) 16 (26) 10 (16) 1 (2) 3 (5) 2 (3) 2 (3) 8 (6.5-8.5)

66 44 53 63 0 67 0 50

Abbreviations: ADEM, acute disseminated encephalomyelitis; CIS, clinically isolated syndrome; EDSS, Expanded Disability Status Scale; IQR, interquartile range; LETM, longitudinally extensive transverse myelitis; MS, multiple sclerosis; NMO, neuromyelitis optica; ON, optic neuritis; PLEX, plasma exchange; TM, transverse myelitis. a At the last follow-up, 6 responders and 17 nonresponders were deceased. The cause of death was non–MS related in 1 responder and unknown in 5 responders; the cause of death was MS related in 8 nonresponders and unknown in 9 nonresponders. b P ⱕ .05. c P ⬍ .001.

A

B Motor Sensory

Bowel/Bladder/Sexual Dysfunction Cerebellar Brainstem Cognitive Paroxysmal Symptoms Optic Neuritis Visual Field Defect Apraxia Respiratory Failure Movement Disorder Cortical Blindness Cortical Sensory Loss 0

20

40

60

80

100

% With Presenting Symptom (Among All Patients)

0

20

40

60

80

100

% With Moderate or Marked Response (Among Patients With Feature)

Figure 2. Frequency of presenting symptoms at the index attack (A) and plasma exchange response rate by presenting symptom among all patients with that symptom (B). Most patients had polysymptomatic attacks and thus could have more than 1 presenting symptom.

INDEX ATTACK LEADING TO PLEX The median time from the index attack to PLEX was 23 days (range, 0-186 days), with a median of 7 exchanges (range, 2-24 exchanges). Most patients were severely disabled in at least 1 clinical domain, with a median Expanded Disability Status Scale30 (EDSS) score of 8 (range, 3-9.5), and presented with a relapsing clinical course (n = 104 [68%]) with typically polysymptomatic index attack symptoms (Figure 2). The most common presenting symptoms in decreasing frequency and grouped according to systems were motor (84%), sensory (71%), sphincter (65%), cerebellar (48%), brainstem (38%), and

cognitive (25%). Of interest to CNS-IDDs, 14% of patients presented at the time of PLEX with either monocular or binocular optic neuritis. Symptoms associated with the highest PLEX response rates were brainstem (64%), cerebellar (63%), motor (59%), and cognitive (54%). OUTCOME OF PLEX AND CLINICAL FEATURES ASSOCIATED WITH TREATMENT SUCCESS Ninety patients (59%; 95% confidence interval [CI], 51%66%) exhibited moderate to marked functional improvement within the 6 months following PLEX. Table 3 summarizes logistic regression modeling of clinical features

ARCH NEUROL / VOL 68 (NO. 7), JULY 2011 873

WWW.ARCHNEUROL.COM

Downloaded from www.archneurol.com at HINARI, on July 11, 2011 ©2011 American Medical Association. All rights reserved.

Table 3. Unadjusted Logistic Regression Models of Clinical Features Associated With Plasma Exchange Response Among All 153 Patients

Feature Sex Female Male Time from index attack to PLEX, d ⱕ20 21-60 ⬎60 EDSS score at index attack ⬍8 ⱖ8 Deep tendon reflexes Flaccid or absent Brisk or normal

PLEX Response Rate, %

OR (95% CI)

58 61

1 [Reference] 1.16 (0.58-2.32)

P Value .68

.89 60 60 55

sponding to PLEX compared with areflexic patients (OR=4.28; 95% CI, 1.78-10.26; P=.001). Sex, EDSS score at the index attack, and time from the index attack to initiation of PLEX were not associated with PLEX response. The PLEX responders had a shorter median disease duration at the time of PLEX compared with PLEX nonresponders (1.1 vs 2.3 years, respectively; P = .02) (Table 1). The overall PLEX response rate among our serological cohort was 74%, with PLEX responders having a shorter median disease duration than PLEX nonresponders (345 days vs 4 years, respectively; P=.04).

1 [Reference] 1.03 (0.51-2.07) 0.81 (0.31-2.14)

TEMPORAL COURSE OF CLINICAL RESPONSE TO PLEX AND FOLLOW-UP PERIOD .98

59 58

1 [Reference] 0.99 (0.51-1.94)

31 66

1 [Reference] 4.28 (1.78-10.26)

.001

Abbreviations: CI, confidence interval; EDSS, Expanded Disability Status Scale; OR, odds ratio; PLEX, plasma exchange.

Table 4. Brain and Spine Magnetic Resonance Imaging Features of Plasma Exchange Cohort No. (%) Brain Scans a (n=78)

Feature T2-weighted lesion load b No lesions observed Unifocal Multifocal T1-weighted gadolinium-enhanced lesions c No enhancing lesions observed Unifocal Multifocal Mass effect on brain MRI (n=72) Mild Moderate or marked Edema on brain MRI (n=71) d Mild Moderate or marked Mass effect or edema on spine MRI (n=63) Mild Moderate or marked Largest lesion, cm (n=65) ⱕ5 ⬎5 Median (range) Vertebral segments involved (n=70) ⱕ3 ⬎3 Scans/patient, median (range), No.

Spine Scans (n = 85)

5 (6) 2 (3) 70 (91)

12 (15) 53 (65) 17 (21)

22 (31) 16 (23) 32 (46) 36 (46) 30 (42) 6 (8) 34 (48) 23 (33) 9 (13)

14 (21) 47 (70) 6 (9)

41 (65) 8 (13) 33 (52) 46 (71) 19 (29) 3.4 (0.3-15.0)

1 (1-5)

20 (29) 50 (71) 1 (1-5)

Abbreviation: MRI, magnetic resonance imaging. a Brain MRI cohort excludes patients with neuromyelitis optica. b Sample sizes: brain scans, n=77; spine scans, n=82. c Sample sizes: brain scans, n=70; spine scans, n=67. d Severity is unknown for 2 subjects.

that appear to be associated with PLEX response in our clinical cohort (n=153). Patients with preserved deep tendon reflexes had a 4-fold increase in their odds of re-

Patients who exhibited functional improvement did so within a median of 4 days (range, 1-100 days) and by a median of the third exchange (range, 1-12 exchanges). However, a small subset of patients (n=4 [6%]) exhibited a delayed response to PLEX (60-100 days). Patients were followed up for a median of 1.9 years (IQR, 8 months to 4.4 years). The PLEX responders were significantly less impaired at the time of last follow-up (median EDSS score, 4; IQR, 2-7) compared with nonresponders (median EDSS score, 8; IQR, 6.5-8.5) (P⬍.001). Patients with relapsingremitting MS at last follow-up had the highest PLEX response rate (75%) compared with patients with all other CNS-IDDs (52%) (P=.008). Patients who developed a progressive disease course by last follow-up had a lower PLEX response rate than patients with monophasic or relapsing clinical courses (30% vs 64%, respectively; P=.046) (Table 2). RADIOGRAPHIC CHARACTERISTICS OF THE COHORT Table 4 summarizes separately the radiographic spectrum for non-NMO patients with at least 1 brain scan (n = 78) and for all subjects with at least 1 spine MRI (n=85). Multifocal brain lesions were present in 70 patients (91%), with Barkhof criteria fulfilled in 47 patients (65%). The most common lesion locations were periventricular white matter (n = 64 [89%]), hemispheric or subcortical white matter (n=64 [89%]), and juxtacortical white matter (n = 61 [85%]), followed by brainstem (n=38 [53%]), corpus callosum (n=38 [53%], of whom 6 had a butterfly configuration), deep gray matter (n=32 [44%]), cerebellar (n=23 [32%]), and cortical (n=10 [14%]). The median largest lesion size on brain MRI was 3.4 cm (range, 0.3-15.0 cm), with nearly half of the cohort (46%) having a largest lesion between 2.1 and 5.0 cm. Perilesional edema and/or mass effect on brain lesions were noted. Representative enhancement patterns are illustrated in Figure 3. The most common enhancement patterns were ring or arc (n=25 [52%]), diffuse or patchy (n=32 [46%]), fluffy or cotton ball (n=21 [30%]), and nodular (n=19 [27%]). Less frequent enhancement patterns included homogeneous (n = 12 [17%]), heterogeneous (n=6 [9%]), punctate (n=5 [7%]), and other miscellaneous patterns (n = 4 [6%]). T2weighted hypointense rims were observed in 18 patients (25%), and colocalization of ring enhancement and

ARCH NEUROL / VOL 68 (NO. 7), JULY 2011 874

WWW.ARCHNEUROL.COM

Downloaded from www.archneurol.com at HINARI, on July 11, 2011 ©2011 American Medical Association. All rights reserved.

A

B

C

D

E

F

G

H

I

Figure 3. Representative gadolinium enhancement patterns, including homogeneous (A), heterogeneous (B), diffuse or patchy (C), fluffy or cotton ball (D), punctate (⬍2 mm) (E), nodular (⬎2 mm) (F), open ring to cortex (G), and arc with colocalization of the ring-enhancing lesion (arrow) (H), and T2-weighted hypointense rim (arrow) colocalizing with ring enhancement (I). A-H, T1-weighted gadolinium-enhanced sequences.

T2-weighted hypointense rims was observed in 11 of these patients (66%). RADIOGRAPHIC BRAIN FEATURES ASSOCIATED WITH PLEX RESPONSE The PLEX response rates were significantly higher among patients with ring enhancement of the largest lesion (82%) compared with patients without ring

enhancement (54%) and in patients with mass effect (75%) compared with patients without mass effect (50%) (Figure 4). Patients with edema on the largest lesion also had higher PLEX response rates compared with patients without edema (73% vs 54%, respectively). Expressed another way, patients with ringenhancing lesions (RELs) of their largest lesion on prePLEX MRI appeared to have 4-fold odds of exhibiting a beneficial PLEX response compared with patients with-

ARCH NEUROL / VOL 68 (NO. 7), JULY 2011 875

WWW.ARCHNEUROL.COM

Downloaded from www.archneurol.com at HINARI, on July 11, 2011 ©2011 American Medical Association. All rights reserved.

A

B

Feature absent Feature present P = .03

Crude OR Adjusted OR

P = .10

P = .02

100

Patients Responding, %

RELs 80

Mass Effect

60

40 Edema 20

0 RELs

Edema

Mass Effect

0

2

4

6

8

OR

Figure 4. Plasma exchange response rate with 95% confidence interval (A) and odds ratio (OR) with 95% confidence interval (B) by brain magnetic resonance imaging feature on the largest lesion. A, Bars indicate the plasma exchange response rate; lines, 95% confidence interval. P values are based on unadjusted logistic regression models. B, The adjusted model was adjusted for days to plasma exchange on the log-transformed scale, Expanded Disability Status Scale score at the time of plasma exchange, and deep tendon reflexes. RELs indicates ring-enhancing lesions.

COMMENT

Table 5. Plasma Exchange Response by Neuromyelitis Optica–IgG Serostatus in 34 Patients

CLINICAL FEATURES AND PLEX RESPONSE PLEX Response Responders No. (%) 95% CI, % Nonresponders No. (%) 95% CI, %

NMO-IgG Seropositive

NMO-IgG Seronegative

15 (68) 47-84

10 (83) 55-95

7 (32) 16-53

2 (17) 5-45

Abbreviations: CI, confidence interval; NMO, neuromyelitis optica; PLEX, plasma exchange.

out RELs (unadjusted OR = 4.00; 95% CI, 1.03-15.60; P=.03). Presence of mass effect and/or edema was associated with increased likelihood of PLEX response (mass effect: unadjusted OR=3.00; 95% CI, 1.12-8.05; P = .02; edema: unadjusted OR = 2.33; 95% CI, 0.846.48; P = .10). There were no significant associations between PLEX response and lesion load, location, size, other enhancement patterns, multifocality, or fulfillment of Barkhof criteria. SEROLOGICAL PREDICTORS OF PLEX Table 5 summarizes PLEX response by NMO-IgG serostatus. Thirty-four patients had at least 1 prePLEX serum sample available for NMO-IgG immunofluorescence assay. The median time from blood draw to PLEX was 6 days (IQR, 2-15 days). We detected NMO-IgG in 22 patients (65%), with a median titer value of 1920 (range, 120-61 440). The PLEX response rate did not differ significantly between patients who were NMO-IgG seropositive (68%; 95% CI, 47%-84%) and those who were NMO-IgG seronegative (83%; 95% CI, 55%-95%) (P= .44).

Our study represents the largest series of patients receiving PLEX therapy for a steroid-refractory attack of CNSIDD. Functional improvement within 6 months was observed in 59% of patients, which agrees with previously published PLEX response rates ranging between 40% and 63%.1-10 The higher PLEX response rate in our series is comparable to the 63% response rate observed in the series reported by Llufriu et al10 and may be owing to both studies assessing neurological improvement beyond the acute response phase. Previously reported clinical factors predictive of PLEX response include male sex, early initiation of PLEX, preserved deep tendon reflexes, and early improvement.3,10 Similar to the series by Llufriu et al,10 we found no association between male sex and PLEX response. However, patients with preserved deep tendon reflexes appeared to have a 4-fold increased likelihood of PLEX response compared with areflexic patients. Areflexia may be the consequence of irreversible anterior horn cell damage less amenable to PLEX rescue therapy. Additionally, PLEX responders in our clinical and serological cohorts had shorter pre-PLEX disease duration than PLEX nonresponders (1.1 vs 2.3 years, respectively, in the clinical cohort; 345 days vs 4.0 years, respectively, in the serological cohort). Because pre-PLEX disability was not determined in our study, the lower PLEX response rate may be related to greater preexisting disability. At the time of PLEX, both PLEX responders and PLEX nonresponders had a median EDSS score of 8. Although this degree of disability is more typical of progressive MS, the high EDSS score reflects the fulminant nature of CNSIDD attacks within our predominantly relapsing cohort. At the time of last follow-up, PLEX responders had a median EDSS score of 4, further indicating that these were not fixed, irreversible deficits at the time of PLEX.

ARCH NEUROL / VOL 68 (NO. 7), JULY 2011 876

WWW.ARCHNEUROL.COM

Downloaded from www.archneurol.com at HINARI, on July 11, 2011 ©2011 American Medical Association. All rights reserved.

Two clinically relevant temporal courses must be taken into account when considering the management of patients with fulminant CNS-IDD attacks. The first is the time from attack onset to PLEX initiation, and the second is the time from PLEX initiation to PLEX response. Early initiation of PLEX may portend a favorable response. In our previous PLEX series, the highest PLEX response rate was observed among patients receiving PLEX 20 days or sooner from attack onset.3 Although there may be a decreased likelihood of prolonged inflammation and subsequent axonal damage among patients in whom PLEX is initiated early, this observation has not been consistently replicated. Patients in our series receiving PLEX more than 60, 80, and 90 days from attack onset had similar PLEX response rates as those receiving PLEX within 20 days of attack onset (55%, 63%, and 60% vs 60%, respectively). Indeed, previous reports have also found PLEX to be beneficial even when delayed after attack onset.1,3,4,9 Although it may be appropriate to consider PLEX in patients with persistent symptoms beyond 90 days in duration, other clinical or radiographic indicators of ongoing disease activity may help guide this decision. The second clinically relevant temporal course relates to the interval between PLEX initiation and PLEX response. Consistent with previous observations, PLEX responders in our cohort responded rapidly (median of 4 days and by the third exchange). Although defining the therapeutic window of PLEX was not the subject of this study, it is noteworthy that a subset of patients (n=4 [6% of all responders]) were delayed PLEX responders (ie, 60-100 days following PLEX) as observed in the series by Llufriu et al.10 Although the possibility of spontaneous improvement cannot be excluded, these studies suggest that neurological improvement may continue beyond the initial PLEX treatment phase and is relevant when discussing functional neurological prognosis with patients and their families. RADIOGRAPHIC FEATURES AND PLEX RESPONSE To our knowledge, this is the first report to identify a reliable radiographic surrogate marker predictive of a beneficial PLEX response in patients with steroidrefractory CNS-IDD attacks. In this series, 91% of patients presented with multiple lesions at the time of PLEX, which may account for the polysymptomatic presentation observed among the cohort. Large (⬎2 cm) tumefactive lesions were common, with nearly 53% of patients demonstrating either mass effect or edema, and 52% exhibiting RELs on brain MRI. An analysis investigating potential clinical confounders driving PLEX response among patients with RELs did not reveal any significant differences in terms of diagnosis, clinical course, EDSS score, or other clinical variables; therefore, the association between PLEX response and RELs is driven not by underlying clinical variables but rather by the underlying pathobiology of the disease. Ring-enhancing lesions are more common among patients with MS immunopattern II, where a humoral pathology dominates and PLEX response has been reported.15 Therefore, the exquisite PLEX response observed among patients with RELs in our series

further supports a role for humoral pathogenic mechanisms underlying the observed favorable PLEX response. NMO-IgG SEROSTATUS AND PLEX RESPONSE The discovery of the disease-specific serum biomarker NMO-IgG in patients with NMO spectrum disorders and elucidation of its antigenic target aquaporin 4 have enhanced the early diagnosis of NMO. The overall PLEX response rate in our serological cohort was 74%, with NMO-IgG detected in 65% of patients (Table 5). Two studies have examined NMO-IgG serostatus and its association with PLEX response, with conflicting results.5,7 The PLEX response was not associated with NMO-IgG serostatus (P=.44) in our cohort and was actually higher in NMO-IgG–seronegative patients (83%) compared with NMO-IgG–seropositive patients (68%). The beneficial PLEX response observed in patients with NMO irrespective of NMO-IgG serostatus suggests that a unified humoral pathomechanism may be operative in both NMOIgG–seronegative and NMO-IgG–seropositive patients.31 Future in vitro and in vivo studies are needed to better characterize this unique subset of patients. Accepted for Publication: January 12, 2011. Published Online: March 14, 2011. doi:10.1001 /archneurol.2011.34 Correspondence: Claudia F. Lucchinetti, MD, Department of Neurology, Mayo Clinic College of Medicine, 200 First St SW, Rochester, MN 55905 (lucchinetti.claudia @mayo.edu). Author Contributions: Study concept and design: Magan˜a, Keegan, Weinshenker, Erickson, Rodriguez, and Lucchinetti. Acquisition of data: Magan˜a, Keegan, Weinshenker, Erickson, Pittock, Lennon, Rodriguez, Linbo, and Lucchinetti. Analysis and interpretation of data: Magan˜a, Keegan, Weinshenker, Erickson, Pittock, Rodriguez, Thomsen, Weigand, Mandrekar, and Lucchinetti. Drafting of the manuscript: Magan˜a and Lucchinetti. Critical revision of the manuscript for important intellectual content: Magan˜a, Keegan, Weinshenker, Erickson, Pittock, Lennon, Rodriguez, Thomsen, Weigand, Mandrekar, Linbo, and Lucchinetti. Statistical analysis: Magan˜a, Thomsen, Weigand, Mandrekar, and Lucchinetti. Obtained funding: Pittock and Lucchinetti. Administrative, technical, and material support: Magan˜a, Erickson, and Rodriguez. Study supervision: Magan˜a, Keegan, and Lucchinetti. Financial Disclosure: Drs Lennon, Lucchinetti, and Weinshenker stand to receive royalties for intellectual property related to the aquaporin 4 autoantigen. Dr Keegan is compensated as section editor for Neurology and as a chief editor for eMedicine. Dr Pittock and Mayo Clinic have a financial interest in the technology entitled “Aquaporin-4 Autoantibody as a Cancer Marker.” This technology has been licensed to a commercial entity but no royalties have been received. Dr Pittock is an inventor of technology entitled “Aquaporin-4 Binding Autoantibodies in Patients With Neuromyelitis Optica Impair Glutamate Transport by Down-Regulating EAAT2.” Mayo Clinic has filed a nonprovisional patent application for

ARCH NEUROL / VOL 68 (NO. 7), JULY 2011 877

WWW.ARCHNEUROL.COM

Downloaded from www.archneurol.com at HINARI, on July 11, 2011 ©2011 American Medical Association. All rights reserved.

this technology. Dr Pittock has received a research grant from Alexion Pharmaceuticals for his investigatorinitiated study entitled “An Open Label Study of Eculizumab in NMO.” Dr Pittock is a consultant in the Department of Laboratory Medicine and Pathology. In his role as codirector of the Neuroimmunology Laboratory, Dr Pittock has no additional intellectual property related to any tests performed on a service basis in the laboratory. Dr Pittock receives no royalties from the sale of these tests when used for patients. Mayo Collaborative Services, Inc, does receive revenue for conducting these tests. Funding/Support: This work was funded by Mayo Foundation, Clinical Translational Science grant 1 TL1 RR024152-01 from the National Institutes of Health (Ms Magan˜a), grant RG-3185-B-3 from the National MS Society (Dr Lucchinetti), grant NS-49577-01 from the National Institute of Neurological Disorders and Stroke (Dr Lucchinetti), and a generous gift from the GuthyJackson Charitable Foundation. Role of the Sponsors: The sponsors had no role in the study design, data collection, data analysis, data interpretation, or writing of the manuscript and were not involved in the decision to submit the manuscript for publication. Additional Contributions: Jeffrey Winters, MD, and the Mayo Clinic Apheresis Unit assisted in patient ascertainment and Lea Dacy, AB, assisted with manuscript preparation. REFERENCES 1. Rodriguez M, Karnes WE, Bartleson JD, Pineda AA. Plasmapheresis in acute episodes of fulminant CNS inflammatory demyelination. Neurology. 1993;43(6): 1100-1104. 2. Weinshenker BG, O’Brien PC, Petterson TM, et al. A randomized trial of plasma exchange in acute central nervous system inflammatory demyelinating disease. Ann Neurol. 1999;46(6):878-886. 3. Keegan M, Pineda AA, McClelland RL, Darby CH, Rodriguez M, Weinshenker BG. Plasma exchange for severe attacks of CNS demyelination: predictors of response. Neurology. 2002;58(1):143-146. 4. Bennetto L, Totham A, Healy P, Massey E, Scolding N. Plasma exchange in episodes of severe inflammatory demyelination of the central nervous system: a report of six cases. J Neurol. 2004;251(12):1515-1521. 5. Watanabe S, Nakashima I, Misu T, et al. Therapeutic efficacy of plasma exchange in NMO-IgG-positive patients with neuromyelitis optica. Mult Scler. 2007; 13(1):128-132. 6. Watanabe S, Nakashima I, Miyazawa I, et al. Successful treatment of a hypothalamic lesion in neuromyelitis optica by plasma exchange. J Neurol. 2007;254 (5):670-671. 7. Bonnan M, Valentino R, Olindo S, Mehdaoui H, Smadja D, Cabre P. Plasma exchange in severe spinal attacks associated with neuromyelitis optica spectrum disorder. Mult Scler. 2009;15(4):487-492.

8. Mao-Draayer Y, Braff S, Pendlebury W, Panitch H. Treatment of steroidunresponsive tumefactive demyelinating disease with plasma exchange. Neurology. 2002;59(7):1074-1077. 9. Paus S, Prömse A, Schmidt S, Klockgether T. Treatment of steroid-unresponsive tumefactive demyelinating disease with plasma exchange. Neurology. 2003;61 (7):1022. 10. Llufriu S, Castillo J, Blanco Y, et al. Plasma exchange for acute attacks of CNS demyelination: predictors of improvement at 6 months. Neurology. 2009;73 (12):949-953. 11. Dau PC. Plasmapheresis in acute multiple sclerosis: rationale and results. J Clin Apher. 1991;6(4):200-204. 12. Khatri BO. Therapeutic apheresis in neurological disorders. Ther Apher. 1999;3 (2):161-171. 13. Gold R, Hartung HP. Towards individualised multiple-sclerosis therapy. Lancet Neurol. 2005;4(11):693-694. 14. Archelos JJ, Storch MK, Hartung HP. The role of B cells and autoantibodies in multiple sclerosis. Ann Neurol. 2000;47(6):694-706. 15. Keegan M, König F, McClelland R, et al. Relation between humoral pathological changes in multiple sclerosis and response to therapeutic plasma exchange. Lancet. 2005;366(9485):579-582. 16. Lucchinetti CF, Mandler RN, McGavern D, et al. A role for humoral mechanisms in the pathogenesis of Devic’s neuromyelitis optica. Brain. 2002;125(pt 7): 1450-1461. 17. Lennon VA, Wingerchuk DM, Kryzer TJ, et al. A serum autoantibody marker of neuromyelitis optica: distinction from multiple sclerosis. Lancet. 2004;364 (9451):2106-2112. 18. Lublin FD, Reingold SC; National Multiple Sclerosis Society (USA) Advisory Committee on Clinical Trials of New Agents in Multiple Sclerosis. Defining the clinical course of multiple sclerosis: results of an international survey. Neurology. 1996;46(4):907-911. 19. Poser CM, Paty DW, Scheinberg L, et al. New diagnostic criteria for multiple sclerosis: guidelines for research protocols. Ann Neurol. 1983;13(3):227-231. 20. McDonald WI, Compston A, Edan G, et al. Recommended diagnostic criteria for multiple sclerosis: guidelines from the International Panel on the diagnosis of multiple sclerosis. Ann Neurol. 2001;50(1):121-127. 21. Wingerchuk DM, Lennon VA, Lucchinetti CF, Pittock SJ, Weinshenker BG. The spectrum of neuromyelitis optica. Lancet Neurol. 2007;6(9):805-815. 22. Wingerchuk DM, Lucchinetti CF. Comparative immunopathogenesis of acute disseminated encephalomyelitis, neuromyelitis optica, and multiple sclerosis. Curr Opin Neurol. 2007;20(3):343-350. 23. Transverse Myelitis Consortium Working Group. Proposed diagnostic criteria and nosology of acute transverse myelitis. Neurology. 2002;59(4):499-505. 24. Tintore´ M, Rovira A, Martı´nez MJ, et al. Isolated demyelinating syndromes: comparison of different MR imaging criteria to predict conversion to clinically definite multiple sclerosis. AJNR Am J Neuroradiol. 2000;21(4):702-706. 25. Barkhof F, Filippi M, Miller DH, et al. Comparison of MRI criteria at first presentation to predict conversion to clinically definite multiple sclerosis. Brain. 1997; 120(pt 11):2059-2069. 26. O’Brien PC. The appropriateness of analysis of variance and multiplecomparison procedures. Biometrics. 1983;39(3):787-794. 27. Perneger TV. What’s wrong with Bonferroni adjustments. BMJ. 1998;316(7139): 1236-1238. 28. Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology. 1990;1(1):43-46. 29. Rothman KJ, Greenland S, Lash TL. Modern Epidemiology. 3rd ed. Philadelphia, PA: Lippincott Williams & Wilkins; 2008. 30. Kurtzke JF. Rating neurologic impairment in multiple sclerosis: an Expanded Disability Status Scale (EDSS). Neurology. 1983;33(11):1444-1452. 31. Amiry-Moghaddam M, Ottersen OP. The molecular basis of water transport in the brain. Nat Rev Neurosci. 2003;4(12):991-1001.

ARCH NEUROL / VOL 68 (NO. 7), JULY 2011 878

WWW.ARCHNEUROL.COM

Downloaded from www.archneurol.com at HINARI, on July 11, 2011 ©2011 American Medical Association. All rights reserved.